Multiplexed Intracellular Single-Cell Sensing Strategies of Microorganisms Integrated for Cultivation in Microfluidic Devices”

“Multiplexed Intracellular Single-Cell Sensing Strategies of Microorganisms Integrated for Cultivation in Microfluidic Devices”

“Intrazelluläre Multiplex-Messmethoden für Einzelzellen von mikrofluidisch kultivierten Mikroorganismen”

Von der Fakultät für Maschinenwesen der Rheinisch-Westfälischen Technischen Hochschule Aachen zur Erlangung des akademischen Grades einer Doktorin der Naturwissenschaften genehmigte Dissertation

vorgelegt von

Christina Erna Maria Bucher-Krämer, geborene Krämer

Berichter Berichter: Univ.-Prof. Dr. rer. nat. Wolfgang Wiechert Univ.-Prof. Dr.-Ing. Lars Blank

Tag der mündlichen Prüfung: 13.03.2019

III

Gewidmet meinem verstorbenen Vater und meinem verstorbenen Großvater, die zu Beginn meines Promotionsvorhabens an meiner Seite waren. Gewidmet meiner Mutter und meiner Schwester. Gewidmet meinem Mann Sebastian, der meinen Lebensweg mit mir geht.

V Die vorgelegte Promotionsarbeit „Multiplexed intracellular single-cell sensing strategies of microorganisms integrated for cultivation in microfluidic devives“ wurde in Auszügen in Fachzeitschriften publiziert und auf Fachkonferenzen vorgestellt: Publications: [8] Krämer, CE, Singh, A, Wiechert, W, Dietrich, D: Microfluidic Screening of Temporal Antibiotic Tolerant Bacteria by Non-Invasive Imaging. (Research Article, submission in progress) [7] Krämer, CE, Wiechert, W, Kohlheyer, D (2016): Time- resolved, single-cell analysis of induced and programmed cell death via non-invasive propidium iodide and counterstain perfusion. Sci Rep 6: 32104 (Research Article) (doi: 10.1038/srep32104) [6] Krämer, C, Wiechert, W, Kohlheyer, D (2016): Artificial Fluorogenic Substrates in Microfluidic Devices for Bacterial Diagnostics in Biotechnology. J Flow Chem 6(1): 3-7 (Invited Perspective Article) (doi: 10.1556/1846.2015.00035) [5] Krämer, C, and Kohlheyer, D (2016): Mikrobiologische Einzelzell-Phänotypencharakterisierung im Mikrochip. BIOspektrum 22(1): 48-50 (Invited Perspective Article) (doi: 10.1007/s12268-016-0658-3) [4] Krämer, C, Singh, A, Probst, C, Kohlheyer, D (2015): Antibiotic Sensitivity Screening of Bacteria on Chip. 19th International Conference on Miniaturized Systems for Chemistry and Life Sciences, Gyeongiy, Korea, 25: 921-24 (Proceeding) [3] Krämer, CE, Singh, A, Helfrich, S, Grünberger, A, Wiechert, W, Nöh, K, Kohlheyer, D (2015): Non-Invasive Microbial Metabolic Activity Sensing at Single Cell Level by Perfusion of Calcein Acetoxymethyl Ester. PLoS One 10(10):e0141768 (Research Article) (doi: 10.1371/journal.pone.0141768) [2] Helfrich, S, Pfeifer, E, Krämer, C, Sachs, CC, Wiechert, W, Kohlheyer, D, Nöh, K, Frunzke, J (2015): Live cell imaging of SOS and prophage dynamics in isogenic bacterial VI

populations. Mol Microbiol 98(4):636-50 (Research Article) (doi: 10.1111/mmi.13147) [1] Nanda, AM, Heyer, A, Krämer, C, Grünberger, A, Kohlheyer, D, Frunzke, J (2014): Analysis of SOS-Induced Spontaneous Prophage Induction in Corynebacterium glutamicum at the Single-Cell Level. J Bacteriol 196(1):180- 8 (Research Article) (doi: 10.1128/JB.01018-13)

Conference Talks: [2] Krämer, C: Instantaneous microfluidic metabolic activity sensing for non-invasive single cell physiology analysis of a prokaryote and its descendants. 3rd International Conference Implementation of Microreactor Technology in Biotechnology, Opatija, Croatia, 10 May – 13 May 2015 [1] Nanda, AM, Krämer, C: Analysis of spontaneous prophage induction in a subpopulation of Corynebacterium glutamicum cultures and its physiological consequence. DFG SPP1617 Progress Meeting, Bad Staffelstein, Germany, 28 Nov 2013

Poster: [14] Pfeifer, E, Helfrich, S, Krämer, C, et. al.: A spatiotemporal analysis of SOS and prophage dynamics in Corynbacterium glutamicum. 30. Jahrestagung der Vereinigung für Allgemeine und Angewandte Mikrobiologie, VAAM, Marburg, Germany, 1 Mar – 4 Mar 2015 [13] Probst, C, Grünberger, A, Krämer, C, et al.: Genetically encoded sensor analysis with spatial and temporal Resolution by time-lapse imaging. 1st Evaluation Symposium, Molecular Interaction Engineering, MIE- Project, Karlsruhe, Germany, 23 Nov – 24 Nov 2015 [12] Krämer, C, Singh, A, Probst, C, et al.: Antibiotic Sensitivity Screening of Bacteria on Chip. 19th International Conference on Miniaturized Systems for Chemistry and Life Sciences, Gyeongjy, Korea, 25 Oct – 29 Oct 2015 [11] Krämer, C, Helfrich, S, Azzouzi, CE, et al.: Microfluidics as platform for the spatio-temporal analysis of bacterial populations at single-cell level. Phenotypic heterogeneity VII and sociobiology of bacterial populations, Kloster Irsee, Kaufbeuren, Germany, 15 Oct – 17 Oct 2014 [10] Pfeifer, E, Helfrich, S, Krämer, C, et al.: Spatiotemporal analysis of SOS and prophage dynamics in Corynebacterium glutamicum. Phenotypic heterogeneity and sociobiology of bacterial populations, Kloster Irsee, Kaufbeuren, Germany, 15 Oct – 17 Oct 2014 [9] Helfrich, S, Azzouzi, CE, Krämer, C, et al.: Microbial Populations on the Microfluidic Analyst’s Couch: Approaching Cellular Heterogeneity. Bioimage Informatics 2014, Leuven, Belgien, 8 Oct – 10 Oct 2014 [8] Krämer, C, Pfeifer, E, Nanda, A, et al.: Bacterial Stess Studies in Controlled Microfluidic Environments: Enabling a Phenotypical Insight on Bacteria Tackling for Survival. 30th symposium “Mechanisms of Gene Regulations” of the VAAM study group “Regulation & Signal Transduction in Prokaryotes”, Düsseldorf, Germany, 1 Oct – 3 Oct 2014 [7] Helfrich, S, Krämer, C, Grünberger, A, et al.: Microfluidics as platform for the spatiotemporal analysis of bacterial populations at single-cell level. 30th symposium “Mechanisms of Gene Regulations” of the VAAM study group “Regulation & Signal Transduction in Prokaryotes”, Düsseldorf, Germany, 1 Oct – 3 Oct 2014 [6] Pfeifer, E, Helfrich, S, Krämer, C, et al.: Spatiotemporal analysis of SOS and prophage dynamics in Corynebacterium glutamicum populations. 30th symposium “Mechanisms of Gene Regulations” of the VAAM study group “Regulation & Signal Transduction in Prokaryotes”, Düsseldorf, Germany, 1 Oct – 3 Oct 2014 [5] Krämer, C, Pfeifer, E, Nanda, A, et al.: Bacterial Stess Studies in Controlled Microfluidic Environments: Enabling a Phenotypical Insight on Bacteria Tackling for Survival. EMBL Conference – Microfluidics 2014, Heidelberg, Germany, 23 Jul – 25 Jul 2014 [4] Krämer, C, Grünberger, A, Probst, C, et al.: Continuous Non-Invasive Real-Time Viability Measurement of Bacteria on Single Cell Level in Microfluidic Chip Cultures. EMBL Conference – New Approaches and Concepts in Microbiology, Heidelberg, Germany, 14 Oct – 16 Oct 2013

VIII

[3] Krämer, C, Grünberger, A, Probst, C, et al.: Continuous Non-Invasive Real-Time Viability Measurement of Bacteria on Single Cell Level in Microfluidic Chip Cultures. Nikon Symposium on Advanced Imaging in Cell- and Microbiology Technology and Applications, Jülich, Germany, 10 Oct 2013 (Poster Prize Awarded) [2] Nanda, A, Heyer, A, Krämer, C et al.: SOS-induced spontaneous prophage induction in Corynebacterium glutamicum. Nikon Symposium on Advanced Imaging in Cell- and Microbiology Technology and Applications, Jülich, Germany, 10 Oct 2013 [1] Krämer, C, Grünberger, A, Probst, C, et al.: Continuous Viability Fluorescent Staining of Growing Corynebacterium glutamicum Colonies Without Sample Preparation in an Automated Microfluidic Measurement System. “How Dead is Dead III: Life cycles”, Berlin, Germany, 6 Jun – 7 Jun 2013 (Poster Prize Awarded)

IX Abstract

Abstract

Intracellular sensing in microorganisms is challenging due to their fast response to environmental changes and the possibilities of unbiased intracellular probes for reliable live single-cell analysis. The submitted PhD thesis presents implementation of multiplexed fluorescence real time imaging concepts using conventional fluorescent fusion protein expression and novel non-toxic fluorescence in situ stainings (FISS) in a controlled microfluidic environment. Stressful growth conditions with subsequent recovery phases were simulated in microfluidic cultivation devices to study behaviour of cells and their descendants by time and single-cell resolved microscopy. Firstly, the sporadic promotor activities of the SOS cascade gene recA and of the induction of prophage CGP3 were determined under non-stressful conditions and under nutrient depletion. The involved promotor inductions have been analysed by spontaneous induced fusion protein fluorescences in a bacterium. Phenotypic minority formation and individual cell fates were visualized using eyfp, e2-crimson, and venus. Second, novel non-invasive, instantaneous sensing methods were developed without the requirement of genetic modification, to distinguish heterogeneous cell states of growing wild type bacteria using fluorogenic molecules. An innovative complementary approach to conventional cultivation-based methods is utilization of fluorogenic substrates with designed enzyme-labile moieties for live cell analysis. Hence, acetoxymethyl ester (AM) has been discovered as shuttle molecule moiety of aromatic compounds for C. glutamicum. This has been demonstrated for multiplexed imaging with six fluorogenic probe molecules coupled to AM to sense intracellular parameters (metabolic activity, dormancy, aging, cell lysis, apparent antibiotic tolerance, chemotoxicity, phototoxicity, spontaneous intracellular radical formation, and intracellular pH). Further, the protective function of the cell wall or cell membranes of native wild types have been analysed by novel live-cell sensing based on dynamic FISS with propidium iodide and PO-PRO-1. Loss of function of the cellular barriers, programmed cell death, survivor cells with evolving non-gene based antibiotic tolerance and resistance to gene based lytic toxin production were defined phenotypically by diffusive intrusion of probe molecules due to their specific phenotype development.

X Zusammenfassung

Zusammenfassung

Intrazelluläre Messung in Mikroorganismen ist herausgefordernd durch deren schnelle Reaktion auf Umweltveränderungen und den Möglichkeiten von unbeeinträchtigenden Intrazellulärsonden für verlässliche Lebendeinzelzell- analyse. Die vorgelegte Doktorarbeit zeigt die Umsetzung von Multiplex- fluoreszenzkonzepten zur Echtzeitbildgebung mit konventioneller Expres- sion von Fluoreszenzfusionsproteinen und neuartiger, non-toxischer Fluores- zenz in situ Färbung (FISS) in kontrollierter mikrofluidischen Umgebung. Stressende Wachstumsbedingungen mit nachfolgenden Erholungsphasen wurden in mikrofluidischen Kultivierungssystemen simuliert um Verhalten von Zellen und deren Tochterzellen durch Zeit- und Einzelzell-aufgelöste Mikroskopie zu beobachten. Zunächst wurde die sporadische Promotorakti- vität des SOS-Kaskadengens recA und die Induktion des Prophagen CGP3 unter stressfreien Bedingungen und unter Nährstoffarmut bestimmt. Die be- teiligten Promotorinduktionen wurden durch spontan induzierte Induktion von Fusionsproteinfluoreszenzen in einem Baterium analysiert. Phänotypi- sche Minderheitbildung und individuelle Zellschicksale wurden durch eyfp, e2-crimson und venus visualsiert. Desweiteren wurden neuartige non-invasive, unmittelbare Messmethoden ohne Anspruch auf genetische Modification entwickelt um heterogene Zell- zustände wachsender Wildtypbakterien durch fluorogene Moleküle zu unter- scheiden. Eine innovative Herangehensweise ergänzend zu konventionellen Kultur-basierten Methoden ist die Verwendung fluorogener Substrate mit En- zym-labilen funktionellen Gruppen für die Einzelzellanalyse. Folglich wurde Acetoxymethylester (AM) als Vektormolekülrest von aromatischen Verbin- dungen für C. glutamicum entdeckt. Dies wurde gezeigt durch Multiplexfluo- reszenzbildgebung mit sechs fluorogenen AM-gekoppelten Sondenmolekü- len zur Messung intrazellulärer Parameter (Metabolische Aktivität, Dormanz, Alterung, Zelllyse, apparenter Antibiotiktoleranz, Chemotoxizität, Phototo- xizität, spontaner intrazellulärer Radikalbildung und intrazellulärem pH). Zudem wurde die Schutzfunktion der Zellwand oder der Zellmembran nati- ver Wildtypen durch neue Lebendzellmessung auf der Basis von dynamischer FISS mit Propidiumiodid und PO-PRO-1 untersucht. Der Funktionsverlust der Zellaußenbarrieren, programmierter Zelltod, Überlebendzellen mit ent- wickelnder non-Gen-basierter Antibiotikatoleranz und Resistenz gegenüber Gen-basierter Produktion eines lytischen Toxins wurden phänotypisch defi- niert durch diffusives Eindringen von Sondenmolekülen bei spezifischer Phä- notypenentwicklung. XI Preface

Preface

Die vorliegende Dissertation fasst die Ergebnisse meiner eigenständigen Arbeit am Forschungszentrum Jülich GmbH zusammen. Mein Dank gilt Prof. Dr. rer. nat Wiechert, dass ich meine Promotion am Institut für Bio- und Geowissenschaften erstellen durfte. Ich danke Herrn Prof. Dr.-Ing. Büchs, für sein Interesse an meiner Arbeit und dafür, dass er sich als Doktorvater bereit erklärt hat. Ich danke Jun.-Prof. Dr. Kohlheyer für seine Unterstützung für Layout bei Publikationen und Präsentationsfolien, sowie für die Bereitstellung von Chemikalien und Materialien für diese Doktorarbeit. Mein Dank gilt auch den Sektretärinnen, insbesondere Frau Marianne Heß, Frau Fittig, und Frau Dahmen-Göbbels bei der Koordination bürokratischer Fragen zum Promotionsverfahren.

Ich danke auch Horst Kiehl für seine stete Hilfe bei Fragen zur Datenlagerung und -sicherung und der IT des IBG-1. Ich möchte sehr herzlich Frau Prof. Dr. Martina Pohl für die wertvollen Gespräche, ihre erfahrenen Ratschläge und Ihrer fachlichen Meinung von Fluoreszenz-basierten biokatalytischen Messungen und die Diskussion im Zusammenhang mit der Veröffentlichung Non-invasive metabolic activity sensing danken. Ich bedanke mich für die enge Zusammenarbeit mit den Arbeitsgruppen von Frau Dr. Nöh und Frau Jun.-Prof. Dr. Frunzke. Hierbei gilt mein XII Preface

Dank insbesondere Dr. Stefan Helfrich für die Unterstützung der aufwändigen Bildanalyse, als auch für die Bereitstellung der Softwaretools zur Bildauswertung, Mitdoktorand Christian Carsten Sachs für die Grafik zur anteiligen Datenspeicherung auf dem Bilddatenserver und Arun Nanda, sowie den Doktoranden Eugen Pfeifer und Rafael von Boeselager fürs Vorbeibringen der unzähligen Reporterstämme. Mein Dank geht auch an die Kolleginnen und Kollegen der Arbeitsgruppe von Herrn Jun.-Prof. Dr. Kohlheyer, die an der technischen Entwicklung der mikrofluidischen Plattformtechnologie beteiligt sind und waren. Meine Danksagung gilt insbesondere Dr. Alexander Grünberger der maßgeblich am mikrofluidischem Design und der Systementwicklung beteiligt war. Für die Maskenproduktion danke ich besonders Nadja Braun als auch der Helmholtz Nanoelectronic Facility. Ich danke Meike Zimmermann, Laura Beust, Lina Hollmann, Agnes Müller-Schröer, und für die aktive Assisstenz bei der Herstellung von Medien und Herstellung von Chips. Für die tatkräftige und motivierte Unterstützung bei der Untersuchung von Antibiotikatoleranzen im mikrofluidischen Chip danke ich meinem Masterstudenten Abhjeet Singh und wünsche ihm von Herzen Erfolg bei seiner begonnenen Doktorarbeit in Schweden. Ich danke auch der Helmholtz-Gemeinschaft, dem Bundesministerium für Bildung und Forschung und der Deutschen Forschugsgesellschaft für die Finanzierung. Die Arbeiten wurden finanziert durch die Helmholtz-

XIII Preface

Gemeinschaft (VH-NG-1029), das BMBF (FK 031A095A) und die DFG (KO 4537/1-2). Ich möchte auch den netten Mitdoktorandinnen und Mitdoktoranden am IBG-1 danken. Speziell Frau Dr. Viktoria Steffen und Herrn Dr. Daniel Jussen. Mein besonderer Dank gilt auch Frau Dr. Abigail Koch-Koerfges, die mich nicht nur in ihre Forschungsarbeit über das Respirationssystem von C. glutamicum miteinbezog, sondern auch mit wertvollen Tipps, viel Rat, und dem ein oder anderen Telefonat zur Seite stand. Ich danke ihr vor allem auch für das Korrekturlesen der molekularbiologischen Inhalte dieser Arbeit. Vielen Dank. Ich möchte auch Frau Dr. Pia Müller und ihren Kollegen von Nikon danken, dass sie eine Lösung für die vielen Abstürze von Multiplexing Fluoreszenzzeitrafferaufnahmen gefunden haben und mir damit die Forschungsarbeit ungemein erleichtert haben. Mein Dank geht auch an Herrn Lechhart und Herrn Dr. Thiele der PolyAn GmbH, Berlin, für die Lösungs-orientierte individuelle Beadfertigung und die vielen kostenlosen Beadproben. Ich danke Anton Meinhart und Andrea Rocker von der Abteilung für Biomolekulare Mechanismen des Max-Planck-Instituts für Medizinische Forschung in Heidelberg für die wertvolle Diskussion und die zur Verfügung gestellten chemisch kompetenten E. coli BL21(DE3) CodonPlus(DE3)-RIL (Stratagene) und der Plasmide, pET28b(pezTΔC242), pET28b(pezTΔC242(D66T)), und pET28b(pezA/pezT), welche für die Transformation genutzt wurden.

XIV Preface

Insbesondere möchte ich meiner Familie und meinen Freunden danken für ihre kontinuierliche Unterstützung. Insbesondere meiner Mutter Rosemarie, meiner Schwester Michaela und meiner langjährigen Freundin Janet Götze. Ich möchte auch der Familie meines (zukünftigen) Mannes für ihre Unterstützung danken. Meinem Sebastian, den ich über alles liebe, danke ich, dass er mich als Mensch und Forscherin so liebt wie ich bin, mich moralisch und tatkräftig unterstützt und die Unstetigkeit einer Forscherkariere akzeptieren kann.

XV Abbreviations

Abbreviations

ABC ATP-binding cassette AM acetoxymethyl ester (moiety) AMP ampicellin ATCC American tissue and cell culture collection ATP adenosine triphosphate AU arbitrary unit BCECF-AM bis(carboxyethyl)-carboxyfluorescein-tetra acetoxymethyl ester BCECF bis(carboxyethyl)-carboxyfluorescein-tetra BHI brain heart infusion CAM calcein acetoxymethyl ester CbAM blue fluorogenic calcein acetoxymethyl ester CCD charge-coupled device CgAM green fluorogenic calcein acetoxymethyl ester CGP3 Corynebacterium glutamicum prophage 3 CHL Chloramphenicol CTC 5-cyano-2,3-di-(p-tolyl)tetrazolium chloride CFP cyan florescent protein CvAM violet fluorogenic calcein acetoxymethyl ester CAL calcein

XVI Abbreviations

CALb blue fluorescent calcein CALg green fluorescent calcein CALox green fluorescent calcein upon oxidation CALv violet fluorescent calcein CFD computational fluid dynamics CFDA carboxyfluorescein diacetate CFDA-AM Carboxyfluorescein diacetate acetoxymethyl ester CFDASE carboxyfluorescein diacetate succinimidyl ester CLSM confocal laser scanning microscopy CSNARF1-AM carboxy-seminaphtorhodafluor acetoxymethyl ester CSNARF4 carboxy-seminaphtorhodafluor CMFDA 5-chloromethylfluorescein diacetate

CM-H2DCFDA chlormethyl-2,7-dichlorodihydrofluorescein diacetate CTC 5-cyano- 2,3-ditolyl tetrazolium chloride DAPI 4,6-diamidino-2-phenylindole DCF 2-,7-dichlorodihydrofluorescein DHCAM green fluorogenic dihydrocalcein acetoxymethyl ester DMSO dimethyl sulfoxide DNA deoxyribonucleic acid dsDNA double stranded DNA

XVII Abbreviations

DSMZ Deutsche Sammlung von Mikroorganismen und Zellkulturen EMB ethambutol ERC extrachromosomal ribosomal DNA circles eYFP enhanced yellow fluorescent protein FACS fluorescence assisted cell sorting FCM flow cytometry FDA fluorescein diacetate FISH fluorescence in situ hybridisation FISS fluorescence in situ staining FLIM fluorescence lifetime microscopy GFP green fluorescent protein GLC Glucose Int2 Phage integrase IPTG isopropyl β-D-1-thiogalactopyranoside KAN kanamycin LB Luria-Bertani medium or lysogeny broth Lys Lysin (= phage lyase) MGC microfluidic cultivation chambers MMC mitomycin C MPTP membrane permeability transition pores nDEP negative dielctrophoresis driven non-contact cell traps NIG nigiricin

XVIII Abbreviations

PALM photoactivated localization microscopy PCA protocatechuate PCD programmed cell death PDMS polydimethylsiloxane PFA para-formaldehyde PLBR picoliter bioreactor pHex extracellular pH pHint internal pH PI propidium iodide PMMA Polymethyl methacrylate PO-PRO-1 Benzoxazolium,3-methyl-2-[[1-[3- iodide (trimethylammonio)propyl]-4(1H)- pyridinylidene]methyl]-, diiodide ROI region of interest ROS reactive oxygen species Rpf resuscitation promoting factor S/B signal-to-background ratio SFDA 5- (and 6-) sulfofluorescein diacetate S/N signal-to-noise ratio ssDNA single stranded DNA STED stimulation emission depletion STORM stochastic optical reconstruction microscopy STR streptomycin TBARS assay thiobarbituric acid reactive substances assay

XIX Abbreviations

VAL valinomycin V-ATPase vacuolar ATPase WT wilde type YPD yeast extract peptone dextrose medium

XX Table of Contents

Table of Contents

Abstract ...... X

Zusammenfassung ...... XI

Preface ...... XII

Abbreviations ...... XVI

Table of Contents ...... 1

Figures, Tables, and Videos ...... 8

1. Introduction ...... 16

2. Motivation ...... 24

3. Technical Background ...... 26

3.1 The Microfluidic Cultivation Platform of the IBG-1, Forschungszentrum Jülich ...... 26

3.2 Single-Cell Analyses with the In-House Developed Microfluidic Devices ...... 28

3.3 Customer-Made Software-Aided Imaging Data Analysis . 30

3.4 Fluorescence-Based Single-Cell Analyses in Microenvironments ...... 33

4. Material and Methods ...... 39

1 Table of Contents

4.1 Cultivation and Media ...... 39

4.1.1 C. glutamicum ATCC 13032 for Studies of SOS Response and Prophage Induction...... 39

4.1.2 C. glutamicum for Dynamic Metabolic Activity Measurement ...... 42

4.1.3 C. glutamicum for Intracellular Reactive Oxygen Species Measurement ...... 43

4.1.4 Continuous in Vivo Viability Staining and Antibiotic Sensitivity Testing ...... 44

4.2 Microfluidic Device ...... 46

4.3 Fluorescence Time-Lapse Imaging ...... 47

4.4 Metabolic Activity Sensing Method and Experimental Validation ...... 49

4.5 Internal Radical Oxygen Species Determination in C. glutamicum Strains ...... 50

4.6 Internal pH Measurement in C. glutamicum ...... 51

4.7 Dynamic Viability Staining ...... 52

4.8 Additional Staining Methods ...... 53

4.9 Plate Count Test for Antibiotic Sensitivity Testing...... 53

4.10 Data Analysis ...... 54

4.11 Calculations ...... 56 2 Table of Contents

4.12 Cell Classification Criteria ...... 56

5. Results ...... 59

5.1 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum ...... 61

5.1.1 Phenotype Characterization of Living Cells – an Introduction ...... 63

5.1.2 Development of a Temporal Resolved Single-Cell Imaging Approach to Analyse Spontaneous Gene Induction in C. glutamicum ...... 71

5.1.3 Spontaneous SOS Response and Sporadic Prophage Induction ...... 84

5.1.4 Spontaneous SOS Response and Sporadic Phage Induction under Nutritive Stress ...... 89

5.1.5 Discussion and Conclusions ...... 103

5.2 Artificial Fluorogenic Substrates in Microfluidic Devices for Bacterial Diagnostics ...... 110

5.2.1 Introduction ...... 110

5.2.2 Relevance of Novel Nontoxic Fluorogenic Substrates and Recent Application Approaches ...... 115

5.2.3 Conversion of a Calcein Acetoxymethyl Ester for in vivo Single-Cell Analysis ...... 119

3 Table of Contents

5.2.4 Conclusion and Outlook ...... 123

5.3 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamicum ...... 126

5.3.1 Introduction ...... 127

5.3.2 Bacterial Analyses with Calcein Derivatives ...... 130

5.3.3 Real-Time Oxygen Reactive Species Sensing in Living Bacteria ...... 136

5.3.4 A Novel Integration of AM-Bound pH Sensitive Probes in C. glutamicum Cells ...... 150

5.3.5 Conclusion and Outlook ...... 158

5.4 Non-invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of Calcein Acetoxymethyl Ester ...... 162

5.4.1 Introduction ...... 164

5.4.2 CAM Uptake and Calcein Fluorescence Formation . 168

5.4.3 CvAM Concentration Optimization ...... 173

5.4.4 Experimental Validation ...... 177

5.4.5 Metabolic Activity Sensing under Intermittent Nutrient Limitation ...... 181

5.4.6 Temporary Growth Inhibited C. glutamicum Cells due to Antibiotic Exposure ...... 189

4 Table of Contents

5.4.7 Discussion of the Novel Non-Invasive Metabolic Activity Sensing ...... 192

5.4.8 Application of the Metabolic Activity Sensing Method ...... 196

5.5 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion ...... 202

5.5.1 Introduction ...... 203

5.5.2 PI Concentration Optimization and Continuous Viability Staining Validation ...... 206

5.5.3 Instantaneous Cell Death Monitoring and Non-Toxic Counterstaining ...... 216

5.5.4 Prokaryotic Cell Death and Apparent Antibiotic Tolerance Following the Addition of Antibiotics...... 222

5.5.8 Programmed Cell Death (PCD) of E. coli ...... 233

5.5.5 Temporally Resolved Programmed Cell Death (PCD) in Yeast ...... 236

5.5.11 Discussion ...... 247

5.6 Real-Time Antibiotic Susceptibility Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device ...... 252

5 Table of Contents

5.6.1 Introduction to Current Antibiotic Sensitivity Research ...... 253

5.6.2 Microfluidic Antibiotic Susceptibility and Survivor Assay ...... 257

5.6.3 Microfluidic Antibiotic Susceptibility Testing Compared to Conventional Plate Count Assay ...... 263

5.6.4 Determination of Time-Resolved Antibiotic Sensitivity and Antibiotic Tolerance ...... 265

5.6.5 Discussion and Conclusion ...... 270

6. Discussion and Conclusion ...... 272

6.1 Time-Resolved Visualisation of Heterogeneous Promotor Induction Events ...... 279

6.2 Acetoxymethylester (AM)-Enabled Intracellular Fluorescence Sensing ...... 285

6.3 Time-Resolved Single-Cell Survival Sensing Using Fluorescent In Situ Staining ...... 294

7. Outlook ...... 303

8. Appendix ...... 315

8.1 Supplementary Material and Videos of SOS Response and Phage Induction Studies ...... 315

6 Table of Contents

8.2 Supplementary Videos and Supplementary Text of ROS Visualization in Living Bacteria ...... 317

8.3 Supplementary Videos of Metabolic Activity Sensing ... 318

8.4 Supplementary Videos of Dynamic Viability Staining ... 320

8.5 Summary of previous findings of A. Koch-Koerfges...... 323

9. References ...... 329

7 Tables, Figures, and Videos

Figures, Tables, and Videos

Figure 1: Common difficulties of continuous fluorescence staining for time-lapse imaging of growing bacteria...... 20 Figure 2: Microfluidic device fabrication from computer-aided design to the final microcultivation device ...... 27 Figure 3: Single-cell image data processing and visualization ...... 32 Figure 4: Single-cell analysis methods...... 37 Figure 5: A theoretical case of time-dependent single-cell phenotype differentiation ...... 38 Figure 6: Microfluidic device and its submicrostructures ...... 47 Figure 7: Single-cell analysis by fluorescence time-lapse imaging .. 69 Figure 8: Dynamic single-cell behaviour can be analysed under ..... 70 Figure 9: Regulatory scheme of the SOS response, prophage CGP3, and a putative trigger dtxR of prophage induction...... 73 Figure 10: Radical oxygen species determination in a SOS reporter strain expressing e2-crimson under the control of PRecA ...... 74 Figure 11: Relative SOS+ and relative phage+ cells over time determined with different fluorescent reporters ...... 80 Figure 12: Control experiments of fluorophore ripening using two control strains bearing a plasmid each with the IPTG inducible ptac promotor fused to venus or e2-crimson ...... 81 Figure 13: Fluorescence validation and correction of systematic fluorescence fluctuation and photobleaching initiated by the microscopic light source. (A) ...... 83 Figure 14: Phenotypes of SOS+ and phage+ cells of all constructed reporter strains ...... 85

8 Tables, Figures, and Videos

Figure 15: Spatio-temporal resolution of SOS response and prophage induction distribution by lineage tree analysis ...... 87 Figure 16: Spatial resolution of SOS response and prophage induction in the microfluidic device after 8.5 h ...... 89 Figure 17: SOS response and prophage induction distribution of C. glutamicum ATCC 13032::PrecA-venus/pJC1-Plys-e2-crimson...... 91 Figure 18: Physiological live cell analysis of C. glutamicum ATCC13032::PrecA-eyfp/pJC1-Plys-e2-crimson ...... 93 Figure 19: Temporal resolved scatter plot of mean single-cell fluorescence of SOS response, phage induction, and metabolic activity under nutrient depletion in a C. glutamicum ATCC13032::PrecA- eyfp/pJC1-Plys-e2-crimson colony ...... 95 Figure 20: Micrographs of the dual reporter strain C. glutamicum ATCC 13032::PrecA-eyfp/pJC1-Plys-e2-crimson colonies and a representative section of the supply channel under nutrient deprivation and CvAM addition ...... 97 Figure 21: Influence of nutrient depletion on growth and metabolic fitness of C. glutamicum ATCC13032::PrecA-eyfp/pJC1-Plys-e2- crimson ...... 99 Figure 22: Relative SOS response and relative phage induction of C. glutamicum ATCC13032::PrecA-eyfp/pJC1-Plys-e2-crimson under nutrient depletion ...... 101 Figure 23: Influence of iron homeostasis on prophage induction .. 102 Figure 24: Fast, reliable and high throughput bacterial detection for putative pathogens and characterization of production strains is of high relevance for clinical diagnostic, pharmaceutical drug development and biotechnological production ...... 114 Figure 25: Fluorogenic substrate conversions for microbial single-cell analysis combined with microfluidic cultivation and time-lapse imaging ...... 121 Figure 26: Conversion of a fluorogenic substrate by C. glutamicum ATCC 13032 ...... 122

9 Tables, Figures, and Videos

Figure 27: Fluorogenic acetoxymethly esters and their corresponding fluorescent carboxylic acids after hydrolysis by intracellular esterases ...... 132 Figure 28: Phylogenic tree of prokaryotes tested for calcein acetoxymethyl ester derivative staining ...... 134 Figure 29: Scheme of the branched respiratory chain of C. glutamicum ...... 140 Figure 30: Normalized growth rate, hydrogen production, singlet oxygen radical formation of C. glutamicum WT, and the mutants ΔctaD and Δqcr in presence and absensce of thiamine ...... 143 1 Figure 31: O2 formation monitored by single-cell CALox fluorescence traces of C. glutamicum Δqcr and ΔctaD in the absence or presence of thiamine ...... 145 Figure 32: Metabolic activity sensing and temporal resolved cell death determination of C. glutamicum ATCC 13032 and the mutants Δqcr and ΔctaD ...... 149 Figure 33: Intracellular fluorescence of pHrodo Red and pHrodo Green and intracellular fluorescence ratios over time in buffer with VAL and NIG ...... 155 Figure 34: Intracellular fluorescence and ratio of intracellular fluorescence of pHrodo Red and pHrodo Green with a calibration between pH 6.6 and 7.4...... 156 Figure 35: Cellular CvAM metabolism model of a C. glutamicum cell ...... 170 Figure 36: Metabolic activity sensing of growing and dividing cells ...... 171 Figure 37: Comparison of mean single-cell fluorescence and apparent growth rate at different extracellular CvAM concentrations ...... 176 Figure 38: Mean CvAM conversion rate constant and mean CALv efflux rate constant ...... 176 Figure 39: Experimental validation of the metabolic activity sensing method ...... 179 10 Tables, Figures, and Videos

Figure 40: Metabolic activity sensing of C. glutamicum ATCC 13032 at intermittent nutrient limitation in minimal medium CGXII ...... 183 Figure 41: Metabolic activity sensing under iron limitation at single- cell level...... 187 Figure 42: Metabolic activity sensing under carbon limitation at single-cell level ...... 188 Figure 43: Metabolic activity sensing of cells exposed to unviable antibiotic concentrations and metabolic activity changes of descendants after antibiotic stress...... 191 Figure 44: Single-cell distribution and fluorescence of PI and CALv in antibiotics-treated C. glutamicum ATCC 13032 cells in the microfluidic chamber array ...... 211 Figure 45: Determination of optimal propidium iodide concentration ...... 212 Figure 46: Toxicity test with C. glutamicum ATCC 13032 and phenol ...... 215 Figure 47: Validation of dynamic PI staining ...... 216 Figure 48: Comparison of bacterial growth under reference conditions in growth medium, growth medium with PI, and growth medium with PI and counterstain ...... 218 Figure 49: Phototoxicity and photobleaching analysis ...... 220 Figure 50: Comparison of yeast growth under reference conditions with growth medium, growth medium with PI, and growth medium with PI and counterstain ...... 221 Figure 51: Continuous monitoring of cell wall integrity and metabolic activity of wild-type C. glutamicum ATCC 13032 colonies treated with different antibiotics using PI ...... 224 Figure 52: Fractions of different cell states following continuous antibiotic treatment of C. glutami-cum ATCC 13032 cells at 25 µg/mL for 12 h and at 50 µg/mL for 16.6 h ...... 225

11 Tables, Figures, and Videos

Figure 53: Antibiotic-induced cell death and antibiotic tolerance of wild-type C. glutamicum ATCC 13032 ...... 228 Figure 54: Validation of the dynamic dual counterstaining with PI and CvAM by endpoint staining of C. glutamicum cells treated 48 h with 50 µg/ml CHL with the lipid membrane indicator nile red or total DNA staining with DAPI ...... 232 Figure 55: PI to determine the bacterial survival rate following toxin- antitoxin module expression in E. coli BL21CodonPlus(DE3)-RIL ...... 235 Figure 56: Budding and cell death ...... 238 Figure 57: PI combined with PO-PRO-1 to indicate apoptosis in yeast cells ...... 241 Figure 58: PI combined with CALg to indicate apoptosis in yeast cells ...... 242 Figure 59: Stress-triggered cell rescue via membrane permeability transmission pore formation ...... 245 Figure 60: Microfluidic test principle of antibiotic susceptibility and tolerance testing ...... 259 Figure 61: Cell categorisation of the fluorescence in situ staining of antibiotic impaired ...... 261 Figure 62: Antibiotic susceptibility of (A) Bacillus subtilis 168 and (B) Corynebacterium glutamicum ATCC 13032 for AMP, CHL, KAN and STR ...... 264 Figure 63: Antibiotic sensitivity and tolerance of Bacillus subtilis 168 for AMP, CHL, KAN and STR...... 267 Figure 64: Antibiotic sensitivity and tolerance of Corynebacterium glutamicum ATCC 13403 for AMP, CHL, KAN and STR...... 269 Figure 65: Optical sensor integration in microfluidic devices for measurement of microenvironmental parameters ...... 308 Figure 66: Tremendous demand of data storage and big data analysis of multiplexed fluorescence imaging data ...... 313

12 Tables, Figures, and Videos

Figure S1: SOS response and prophage induction distribution of C. glutamicum ATCC 13032::PrecA-venus/pJC1-Plys-e2-crimson ... 317 Figure S2: Schematic overview of the respiratory chain and oxidative phosphorylation of C. glutamicum ...... 325 Figure S3: Scheme of radical formation in C. glutamicum ΔctaD due to the presence and activity of the two cytochrome bc1 complex subunits QcrA and QcrB ...... 327 Figure S4: Comparison of C. glutamicum wild type and its ΔctaD, Δqcr and ΔqcrA mutants with respect to growth (A, D, G), glucose consumption (B), lactate consumption (E), dissolved oxygen (C), pyruvate formation (F), and pH of the supernatant ...... 328

Table 1: Reporter strains, control strains, and imaging parameters.. 41 Table 2: Selection of fluorogenic substrates and their use in bacterial diagnostics in static and perfused analysis systems ...... 116 Table 3: LogP values of fluorescence indicators and inducer ...... 208 Table S1. Growth parameters of C. glutamicum ATCC 13032 wild type, ΔctaD, and Δqcr during cultivation in shaking flasks in CGXII + 222 mM GLC...... 324 Table S2. TBARs formation in C. glutamicum ATCC13032 wild type, and its ΔctaD, Δqcr, and ΔqcrA mutants...... 326

Video 1: A dual reporter strain colony with SOS+/phage- cell that proceeded cell division ...... 315 Video 2: A dual reporter strain colony with SOS+/phage- minority with flickering and constant Venus fluorescence ...... 315 Video 3: A dual reporter strain colony with SOS-/phage+, SOS+/phage+, and elongated putative divS inhibited SOS+ cell minority ...... 315

13 Tables, Figures, and Videos

Video 4: Cells of the SOS reporter strain exhibited SOS+ phenotype with proceeded cell division ...... 316 Video 5: Cells of the SOS reporter strain exhibited SOS+ phenotype with reduced cell division rate ...... 316 Video 6: Putative divS inhibited Cell of the SOS reporter strain exhibited SOS+ phenotype with elongated cell shape ...... 316 Video 7: A dual reporter strain colony with appearance of a SOS+/phage- under starvation condition under intermittent carbon supply ...... 316 Video 8: Simultaneous ROS formation in a ΔctaD colony after 14.5 h of cultivation ...... 318 Video 9: Simultaneous ROS formation in a ΔctaD colony after 12.7 h of cultivation ...... 318 Video 10: Absence of ROS formation in a growing ΔctaD population ...... 318 Video 11: Unsteady ROS formation visualised by flickering CALox fluorescence in ΔctaD cells ...... 318 Video 12 Prevented ROS formation in ΔctaD cells in presence of 0.2 µg/L thiamine ...... 318 Video 13 Prevented ROS formation in ΔctaD cells in presence of 200 µg/L thiamine ...... 318 Video 14: Absence of ROS formation in a growing Δqcr population ...... 318 Video 15: Disappearence of initial ROS in a single Δqcr cell and proceeded cell division in presence of 0.2 µg/L thiamine ...... 318 Video 16: Disappearence of initial ROS in a single Δqcr cell and proceeded cell division in presence of 200 µg/L thiamine ...... 318 Video 17: Metabolic activity sensing under reference conditions. . 319 Video 18: Metabolic activity sensing at different media pH...... 319 Video 19: Intermittent iron supply ...... 320

14 Tables, Figures, and Videos

Video 20: Elongated cells after iron depletion ...... 320 Video 21: Intermittent carbon supply ...... 320 Video 22: Non-growing cells after carbon depletion ...... 320 Video 23: Intermittent iron chelator supply ...... 320 Video 24: Intermittent iron supply with single cell events ...... 320 Video 25: Bursting iron depleted cells ...... 320 Video 26: Spontaneous non-growing cell after iron depletion ...... 320 Video 27: Short term growth impairment by AMP ...... 320 Video 28: Short term growth impairment by CHL ...... 320 Video 29: Cell death and antibiotic tolerance of Corynebacterium glutamicum cells after the addition of antibiotics...... 320 Video 30: Programmed cell death of Escherichia coli...... 321 Video 31: Macroautophagy in Saccharomyces cerevisiae...... 321 Video 32: Ageing in Saccharomyces cerevisiae...... 322 Video 33: Dying zygote...... 322 Video 34: Shmoo-mediated mating of aged cells...... 322 Video 35: Fluctuation in membrane potential followed by cell recovery...... 322

15 Introduction

1. Introduction

Since Antony van Leeuwenhoek paved the way for modern microscopic single-cell analyses with his first observations of bacteria and protozoa, imaging was a permanent essential element in modern life sciences 1,2. Visualisation of microorganisms actually founded what is today defined as microbiology. Furthermore, microscopy is recommended for microstructured device validation of the relatively young field of microsystems engineering. Amazingly, time-lapse microscopy was invented more than a century ago as analog microcinematography with the same concerns as exist nowadays 3. These solicitudes during microscopic longtime observations are now and then about viability and life-sustaining measures, e.g. continuous nutrient supply, of the living observation subjects as well as technical hurdles as to maintain the focus, to avoid vibrations during observation, image data storage and processing 3,4. In parallel to evolving microcinematography, the impulse of biology interested engineers developed to construct microscope-aided single- cell investigation devices 5. The modern researchers, involved in single-cell studies, has the possibilities to choose between several well-established integrated microcultivation chambers that are mounted on the microscope stage. Bacterial single-cell studies have been shown in commercially available micro-cultivation platforms as

16 Introduction the microfluidic CellASIC® ONIX system 6, using self-assembled agar pads with a comparable low demand on technical skills 7,8, or with the help of a self-developed, custom-made microfluidic cultivation device 9. However, an influence on cellular growth by the chosen cultivation device for single-cell analysis has to be taken into account before it can be denied by evaluation experiments 10. Furthermore, also microscopy equipment and microscopic techniques improved since Leeuwenhoek’s century tremendously. Of high importance are that technical advances of microscopic cell analyses need to be non-invasive, with high resolution, and ideally implementable in cultivation processes. Therefore, the technological progress of microscopy for biological in vivo imaging notably advanced with fluorescence microscopy. Both Nobel prize winning discoveries of heterologous expressible green fluorescent protein (GFP) and stimulated emission depletion (STED) microscopy brought spatio-temporal resolution microscopy globally in reach for researchers even at molecular resolution 11–13. Spatial super-resolution microscopy like stimulation emission depletion (STED), photoactivated localisation microscopy (PALM) or stochastic optical reconstruction microscopy (STORM) is not limited by light diffraction as defined by the law of Abbé 12. Thus, cellular structures can be resolved in nanoscale under in vivo conditions due to advanced lateral and axial resolution in comparison to diffraction-limited microscopy 14.

17 Introduction

Whereas, even diffraction-limited microscopy helps to determine single-cell events in temporal and spatial manner in multi-cellular matrices. Both, confocal laser scanning microscopy (CLSM) as well as wide-field microscopy enable to observe stochastic single-cell events in isogenic cell populations, if equipped with an automated imaging platform for multi-dimensional image stacks in x-y-z-axis (spational 3D) 15, x-y-time axis (3D, spational 2D) 16,17 and even x-y- z-time axis (4D, spational 3D) 18,19. Temporal resolution microscopy visualize ultra-fast, molecular processes in cells and bacterial populations (e.g. conformational changes of enzymes, diffusion in nanosecond to millisecond range) by fluorescence lifetime microscopy (FLIM) 18,20, fast cellular movements (e.g., chemotaxis) by slow motion image capture happening in millisecond to second range 21, and cellular events evolving over hours to days by fluorescence time-lapse imaging 17,22. Especially, fluorescence time-lapse microscopy had become essential for single-cell imaging, whereas several fluorescence signals can be combined for multiplexed fluorescence time-lapse imaging. For time-resolved in vivo imaging of bacteria to perform subsequent single-cell analyses, the established techniques are i.) phase contrast based 10, ii.) use fluorescent protein (subsequentially termed as fluorophore) expression 22,23, or iii.) involve bioluminescence 24,25. As phase contrast imaging is limited to cell counting and visualization of the cell silhouettes, intracellular changes (e.g., promotor activity,

18 Introduction increase/decrease of metabolic activity, loss of viability) remain undiscovered. On the other side, heterologous protein expression for fluorescence or bioluminescence imaging require genetic modification prior to fluorescence in vivo imaging of bacteria. Other fluorescence single-cell analysis presented in literature use introduction of intracellular fluorescence with probe molecules that are mostly considerable as toxic to living cells. For example, taxonomic fluorescent probes for fluorescence in situ hybridization (FISH) are highly invasive due to their application protocols, that involve fixation, cell permeabilisation, and probe binding to biomolecules necessary for cell function 26,27. In difference to these high molecular sized FISH probes, continuous staining with low molecular weight fluorescent chemicals (subsequentially termed as fluorochromes), which are capable to pass the highly selective and protective microbial cell wall, are a novel strategy for in vivo fluorescence imaging. For this approach, the fluorochromes have to be nontoxic in relevant concentration during microscopic observation, else fluorescence staining is an endpoint measurement with possible requirement of chemical cell fixation. Unfortunately, microbial cell analysis approaches using continuous fluorescence staining has been established rarely for microorganisms up to now 28. I define the approach as non-invasive in situ fluorescence staining, if growth and cell division is not impaired, and chemotoxicity as well as

19 Introduction

Figure 1: Common difficulties of continuous fluorescence staining for time-lapse imaging of growing bacteria. (A) Escherichia coli MG1655 perfused with 4- methylumbelliferyl β-D-galactopyranoside. The fluorogenic substrate is converted to its blue florescent extracellularly observed by perfusion termination and diffusional limitations as in crowded cell environments. (B) Bacillus subtilis 168 perfused with green fluorogenic calcein acetoxymethyl ester. Only obviously dead cells appeared fluorescent. (C) Shewanella oneidensis MR-1 stained with LAURDAN. Marginal blue fluorescence required comparable intensive excitation with ultraviolet light, that stopped bacterial motility during light (figure legend continues on next page▼)

20 Introduction

(▲continued figure legend of Fig. 1) exposure. (D) E. coli BL21CodonPlus(DE3)- Ril/pET28b(pezTΔC242) expressing a truncated cell wall destructive toxin perfused with 5-chlormethylfluoresceindiacetate. Lysed cell walls are brightly green fluorescent, cell wall-deficient L-form cells remain unstained, the fast-growing control E. coli MG1655 cells showed diffuse staining of the cell wall with a bias due to cell age. (E) E. coli MG1655 perfused with 5-cyano-2,3-ditolyl-tetrazolium chloride. Under physiological concentration no intracellular fluorescence was observed. Concentration increase to result undistributed dot-like intracellular fluorescence was toxic to cells. phototoxicity is absent, and cell stress or cell death has been proofed to be not induced by the single-cell analysis approach in presence of fluorescence indicators at defined low concentration. Common difficulties of the continuous addition of fluorochromes are illustrated with examples in Fig. 1. Fluorogenic substrates are converted extracellularly or are not taken up by intact living bacteria as found with E. coli MG1655 and Bacillus subtilis 168 with perfusion of the fluorogenic substrates 4-methylumbelliferyl β-D- galactopyranoside or a calcein acetoxymethyl ester derivative (Fig. 1 A and B). Or only low fluorescence signals are achieved and/or phtototoxicity is given due to required fluorochrome excitation, that terminate cell growth, division, motility or even kill cells. This was the case with Shewanella oneidensis MR-1 stained with the membrane polarity indicator 2-dimethylamino-6-lauroylna- phthalene (LAURDAN) (Fig. 1 C). Another complication of staining living bacteria can be strong and time-dependent shifting fluorescence and fluorescence background, although bright fluorescence is given as with cell wall linked 5- chlormethlfluoresceindiacetate, that allow to distinguish between

21 Introduction lysed E. coli cells fragments, and the rare L-form phenotype that is cell wall-deficient and remained unlysed by cell wall toxic toxin expression (Fig. 1 D) 17,29. At least, chemotoxicity of molecular fluorescence probes at reasonable concentration can be given as it was observed with the fluorogenic tetrazolium 5-cyano- 2,3-ditolyl tetrazolium chloride (CTC) built from the corresponding formzan in E. coli cells (Fig. 1 E) or with cell permeant total DNA stains. In fact, microscopic long-time observation with single-cell resolution have to be noninvasive to the living cells of interest, especially, if physiology experiments under controlled stress implementing cultivation conditions are planned. As stress cannot be measured directly and described by SU units, as well as multiple impact on cell function and phenotype differentiation have to expected as relevant to be observed in different temporal consequence, and in addition, stressors (stress inducing parameters) are sensed individually by biological organisms, single-cell stress analyses are complex multiparameter system to determine. Furthermore, accidental induction of additional stress has to be completely avoided for single- cell stress studies and absence of triggered systematic stress by single- cell fluorescence imaging or single-cell cultivation has to be evaluated accurately. The presented PhD thesis gives an impression on possibilities of fluorescence single-cell imaging strategies in a previously well- established, custom-made microfluidic device for microbial

22 Introduction diagnostic for the future. For this purpose, conventionally established fluorescent protein expression is shown for determination of phenotypic heterogeneity of isogenic C. glutamicum ATCC 13032 cells after genetic modification to elucidate spontaneous induction of SOS stress response genes and non-SOS genes of an innate prophage of the species. Alternatively, non-toxic dynamic staining strategies with fluorescent or fluorogenic chemicals were developed for single-cell analysis of living microorganisms, which need no prior genetic modification. According to the term FISH, I define the novel approach of continuous staining with fluorochromes as fluorescent in situ staining (FISS), that is in difference to FISH a live cell imaging method without cell permeabilisation or cell fixation and implementable for fluorescence time-lapse microscopy. Additionally, combinatory strategies of fluorescent protein expression and dynamic staining strategies for multiplexing imaging of living cells under controlled microfluidic environments are shown.

23 Motivation

2. Motivation

For scientists, it is always the driving force to understand the inner core of existing matter, whereas engineers design and built-up technical devices to reconstruct existing matter or to define existing matter. If it comes to microorganisms the matter in focus is defined by very small dimensions. The motivation to look inside the precise machinery of production bugs or native wild type microorganism is tremendous high. However, sensing in single cells is very challenging, especially when it comes to bacteria. The engineer has to tackle the small dimensions of a bacterial cell, cell-to-cell heterogeneity, biocompatibility, and at least the integration of the sensor with preferably no impact on single cell viability and metabolism. The motivation of this scientific work was to analyse and sense living single bacteria with established genetically integrated biosensors 22,23,30 and novel integrative nontoxic sensing principles, which had to be developed, 16,17,28 for microbial fluorescence based in vivo studies without the investment in additional technical devices and mainly the manpower of one scientific person. Therefore, the conventional design of a biosensor was simplified to the replicating microbial cell, whereas the integration of molecule sized fluorescence probes sensing intracellular changes and a conventional epifluorescence microscope equipped with several optic filters to perform multiplexed

24 Motivation fluorescence real-time imaging analyses of single cells and their descendants. The sampled images at different wavelength resulted at a composite cell image of phasecontrast and up to three fluorescence settings for multichanelled cell information.

25 Technical Background

3. Technical Background

3.1 The Microfluidic Cultivation Platform of the IBG-1, Forschungszentrum Jülich

The fabrication protocol of the microfluidic devices was developed previously to this PhD thesis by A. Grünberger with assistance of C. Probst, A. Müller-Schröer and N. Braun and is described in detail in the PhD thesis of A. Grünberger 31 and in 9. The microfluidic device fabrication protocol is summarized in Fig. 2. The fabrication process was partly performed at the Helmholtz Nano-Facility on the Forschungszentrum Jülich campus. The microfluidic cultivation devices were fabricated as PDMS-glass sandwich, whereas the design and realization of microstructures imprinted in the flexible elastomer layer is flexible for testing. The preliminary work from scratch paper to final microfluidic cultivation devices is described in detail in the PhD thesis of A. Grünberger 31. The developed microstructure designs contain single-cell traps to catch a single cell 32, a mother machine design 33 for a mother cell and a few daughter cells per cultivation track, a microfluidic picoliter bioreactor (PBLR) 34,35 within a several hundred cells, and cultivation chamber arrays 36 within a few hundred microcultivation chambers for single-colony growth. For spatial

26 Technical Background control over single cells and to perform single-cell seeding, microfluidic growth chamber (MGC) designs were additionally equipped with optical tweezers 37.

Figure 2: Microfluidic device fabrication from computer-aided design to the final microcultivation device. (A) Microstructure design realization as master mold silicon wafer (B) Rapid elastomer molding with the master mold (C) Fabrication of ready-to-use microfluidic devices of parallelized molded PDMS chips (adapted from 9)

Grünberger et al. (2015) demonstrated the predominantly diffusional microscopic flow profile with commercially available bright fluorescent microbeads of different size and computational fluid dynamics computation 36. Furthermore, Westerwalbesloh et al. described the theoretical nutrient supply and diffusional mass transport for a cultivation chamber based design 38.

27 Technical Background

3.2 Single-Cell Analyses with the In-House Developed Microfluidic Devices

Prior single-cell analyses of wildtype bacteria focused on cellular growth determined by cell number count per frame allowing the calculation of the growth rate 39,40, the determination of elongation rate 32,36,37, morphological characteristics like division asymmetry of cell division 10,36, and unusual cell shapes 36. The viability of E. coli was validated cell growth based after laser treatment with optical tweezers 37. Non-growing cells, which appeared to be intact by phase contrast imaging, were assumed as dormant without detection of remaining cell activity, cell integrity, or regrowth 36,41 since there was no additional single-cell sensing method used to differentiate cell death and dormancy. Furthermore, the faster growth of C. glutamicum ATCC 13032 in continuously with medium perfused PBLRs and MGCs were compared to the growth in batch cultivation systems containing 1, 50, and 600 mL working volume 39,40. The results, that culture dilution in a batch system accelerate cell growth and perfusion with batch culture supernatant decreased the growth rate in PBLRs, led to the conclusion, that culture by-products mainly influence the growth of the production microorganism C. glutamicum 39. Additionally, a biphasic growth of C. glutamicum ATCC 13032 on the iron chelator protocatechuate as hidden carbon source in the minimal medium CGXII was revealed 40.

28 Technical Background

Furthermore, C. glutamicum ATCC 13032 growth in MGCs was systematically compared to other microenvironments on agarose pads and in negative dielctrophoresis driven non-contact cell traps (nDEP) elucidating a growth bias in growth rate, cell length, division symmetry, and division angle 10. Further, comparative cell growth analysis in microfluidic cultivation was performed for C. glutamicum ATCC 13032 and a prophage cured variant with no remarkable result 42 . Additionally, growth determinations were accompanied with intracellular fluorescence determination, if fluorophore expressing production strains were observed in the microcultivation structures. Thus, the cell integrity of the non-growing whole cell catalyst E. coli BL21(DE3) expressing fluorescent fusion protein nFbFP-AtHNL in a micro-aqueous reaction environment could be demonstrated 43. Single-cell resolution elucidated cell-to-cell variation of gene expression for C. glutamicum biosensor strains for amino acid production 23,34,44,45 and for chemically induced expression systems in the host organisms E. coli 41,46 using relative fluorescence comparison. In addition, the fluorescent cell percentage of isogenic C. glutamicum colonies with an early plasmid-based SOS-reporting system was shown as morphological single-cell analysis that is not directly detectable with phase contrast 36.

29 Technical Background

3.3 Customer-Made Software-Aided Imaging Data Analysis

The raw data from time-lapse imaging with all observed positions of the microfluidic device and snap shot time points in one matrix-like file containing all frames, that has the commercial format (.nd) of the microscopy software (NIS-Elements AR, version 3.x and higher) (Fig. 3 A). The .nd-file is imported in the software Fiji (ImageJ) 47 and exported to .tiff- files consisting of an image stack with all frames in chronological order of each position. Time-lapse image stacks were first pre-processed, then cells were recognized and segmented with a customized Fiji (ImageJ) plugin called MASTERPLUGIN programmed by S. Helfrich 47,48. Therefore, the region of interest (ROI) consisting of the growth chamber area with the growing cells was extracted as described by S. Helfrich (2016) 48. During that pre-processing, the growth area of the colony and its surrounding was marked, excised, and the thermal drift of the automised stage during imaging was corrected in x-y- direction by the MASTERPLUGIN 48 (Fig. 3 B). Subsequently, cells were recognized and segmented in the tiff-stacks. After the processing of the segmentation, the detected cells and erroneous artefacts (e.g. particles) were marked with a yellow overlay (Fig. 3 C), that had to be manually corrected in a time-consuming post- processing frame-by-frame and for every position to remove artefacts and to mend linked cells or interrupted overlays (Fig. 3 D). Averaged cell quantities (e.g., mean fluorescence, mean cell size) and sums (e.g., 30 Technical Background colony area, cell numbers) could be extracted automised as .txt file. Else, the single-cell quantities without mother-daughter cell relation can be extracted as .cvd file with the custom-developed Fiji (ImageJ) plugin JUNGLE (by S. Helfrich) implemented in the particle tracking plugin TRACKMATE in Fiji (ImageJ) (Fig. 3 E). Further, the mother cells could be linked in a time-consuming manual software-aided connecting process for later lineage tree analyses and exported as .xml files also using the JUNGLE plugin 48,49 (Fig. 3 F). The generated .xml files can be imported in the Java based software tool Vizardous, that had been developed by S. Helfrich and C. Azzouzi for lineage tree analyses (Fig. 3 G) 50. Vizardous enable visualization, analysis, and meta data extraction of prior created lineage trees of microscopic time-lapse imaging data, what was limited at the moment to three visual channels (phase contrast and two fluorescence channels) and a finite data size, what restrict multiplexed fluorescence single-cell analysis. For graphic visualization with e.g. Excel, meta data of lineage trees could be copied from Vizardous, where they were displayed, and pasted in another program, given that the export functions of the visualization tool are still under construction and no adaption of graphical parameters like font or plot style were available in the visualization tool. 48,50 (Fig. 3 H).

31 Technical Background

Figure 3: Single-cell image data processing and visualization. (A) Image acquisition of chosen positions in the microfluidic device over time with the commercial microscopy software (B) Export of the raw image data, spatial correction and excision of the growth sites with Fiji (ImageJ, MASTERPLUGIN by S. Helfrich) (C) Processing of cell segmentation and marking with overlay with Fiji (ImageJ, MASTERPLUGIN by S. Helfrich) (D) Manual overlay correction and artefact removal (E) Export of averaged cell quantities as .txt file by the MASTERPLUGIN or temporal resolved single-cell quantities of all cells as .cvd file by the Fiji plugin JUNGLE in TRACKMATE/Fiji (ImageJ) (F) Software-aided single-cell tracking and manual linkage of mother-daughter cell relations with JUNGLE and export as linage tree as .xml files (G) Visualisation single-cell data and extraction of lineage tree meta data as .cvd file with the software tool Vizardous (by S. Helfrich and co-worker) (H) Result plotting for visualization and publication with commercial software.

32 Technical Background

3.4 Fluorescence-Based Single-Cell Analyses in Microenvironments

A tremendous demand for microfluidic devices combined with fluorescence time-lapse imaging is ruled by the awareness of population heterogeneities, which appear even in isogenic cultures, and spontaneous phenotype differentiation due to environmental stress, genetic switches, spontaneous mutations or combinational triggers. The influence of population heterogeneity has often manifold and individual reasons in every particular cell study case of interest. A short excursus in published single-cell research should give only an impression of current “hot spots” that had given inspirations for this PhD thesis. However, the impacts of these dynamically occurring phenotype minorities are of high interest for individual cellular survival strategies, e.g., antibiotic resistance, stress tolerance, disease development, and not at least cell-to-cell variation of gene expression or metabolite production. Induction of fluorophore fusion proteins expression is per se heterogeneous in isogenic cultures due to differences of inducer binding or uptake bias, promotor activity, cell size, impacts on heterologous gene integration as rare codon usage or position effects, that can lead to gene expression noise or bias 41,51–55. Predictable gene expression on a single-cell basis in isogenic production strain cultures is in focus for systems biology as well as synthetic biology 56,57. 33 Technical Background

Fluorescence detection methods of living cells are conventionally based on previously introduced genes for fluorescent reporter protein expression, or fluorochrome-linked coupling molecule binding as antigene-antibody attachment or streptavidin-biotin coupling that have been reviewed previously 11,58–60. Else, bioluminescence is introduced in microorganisms by introduction of heterologous genes of luciferases and their substrates to distinguish particular species in cellular microenvironments for in vitro as well as in vivo infection models, to analyse quorum sensing of bacterial communities, or for imaging based bioassays 24,25,61. Fluorescent cells are routinely analysed by fluorescence-assisted cell sorting (FACS) systems. Therefore, cell samples of cultures are provided to the cytometric analysis system and measured cell by cell, data can be visualized as scatter plots, e.g. for principle component analysis, and sorted automatically according to the set-up of gating parameters for re-cultivation after measurement as shown schematically in Fig. 4 A. In difference, fluorescence time-lapse microscopy enables highly time-resolved analysis of single-cell gene expression dynamics and cell division analysis, while cell cultivation can be performed in parallel under controlled microenvironmental conditions as depicted in Fig. 4 B 62,63. Although, FACS is advantageous for automised identification of subpopulations instead of population averages by measuring several 10.000 cells up to millions per analysis run, cytometric methods have

34 Technical Background several drawbacks limiting dynamic single-cell analysis of individual cells and their fate 64,65: Almost always manual cell sampling is performed for FACS analysis. Thus, the sampling procedure can influence dynamic subpopulation formation by time of performance, heterogeneous sampling performance, sample handling, and transport.

External cultivation is required in conventional cultivation systems, which have less control of microenvironments than microfluidic devices. Loss of environmental control directly before and during analysis is provoked due to required use of FACS buffers for analysis, absence of pH control or supply of nutrients and oxygen and change of ambient temperature of sampled cells.

FACS analysis requires time for performance of sequential cell analysis, temperature control of the specimen holder, cell sample storage, system validation and maintenance by experienced performers.

i.) Fluorescence excitation by lasers and convection provide undefined stress situation prior re-cultivation. ii.) Subpopulation ratios are determined by FACS, but no resolution of distinct progenitor cell dynamics by environmental impacts or intrinsic triggers can be pursued.

35 Technical Background

iii.) Analysis parameters for gating essentially influence FACS data assessment and cell sorting for re- cultivation. However, gating parameters are rarely given for published cell studies based on FACS analyses.

Especially, if several intracellular events, that are expressed as phenotype, are linked to each other, time-dependent development of single-cell events cannot be resolved with snapshot determination as with FACS. Development of different phenotypes (Fig. 5 A) can be coincidental or interact in a causal manner (e.g., if phenotype A established, phenotype B follows) or temporal manner (e.g., when phenotype A can be found initially, phenotype B follows independently of A after a time span). Phenotype differentiation is almost always an energy-dependent process. Cells have to decide to express proteins, to divide whether or not, to actively communicate via chemical secretion or perform other action 66–68. As a matter of fact, phenotype development take time to evolve over intermittent cell states and to appear over time, what makes time resolved single-cell studies so valuable.

36 Technical Background

Figure 4: Single-cell analysis methods. (A) FACS with separated cultivation, sequential cell sampling, measurement (n ≈ 10.000 cells), and cell sorting with the possibility of subsequent re-cultivation, or (B) fluorescence time-lapse imaging of dividing cells (n ≈ 10 – 100 initial cells with approximately 100 – 5000 descendants in last frame) combined with cultivation in a microfluidic device for microenvironmental control.

Therefore, limited temporal resolution and lack of spatial resolution limit FACS analyses to distributional results of subpopulations (Fig. 5 B) that tend to obscure single-cell profiles and their fates (Fig. 5 C). Observation by fluorescence time-lapse microscopy helps to determine single-cell fluorescence traces over time. Thus, phenotypic differentiation and gene expression profiles can be defined and

37 Technical Background distinguished to be heterogeneous in level (Fig. 5 C, i.), iii.), iv.)), and in stability (Fig. 5 C, i.), ii.), v.)) 53,54,56,69.

Figure 5: A theoretical case of time-dependent single-cell phenotype differentiation. (A) Time-dependent single-cell phenotype change in an intermediate state and a second state. (B) Possible histograms of a cell population determined with FACS analysis at five different sampling time points. Interaction or dependencies of the two different phenotypes are indeterminable. (C) Possible single-cell trace profiles of the exemplarily in (B) shown FACS analysis that a challenging to determine or can be overseen with insufficient temporal resolution. Cells can perform i.) full signal development. ii.) Signal decay or loss and iii.) reduced single-cell signal development in cells can happen partly in the population. iv.) A population fraction produces no signal and remain in the initial phenotype state. v.) Rare single-cell phenomena are described as reversible, repeatedly switching or even regularly oscillating.

38 Material and Methods

4. Material and Methods

Chemicals were provided by Carl Roth, Karlsruhe, Germany, Merck, Darmstadt, Germany, or Sigma-Aldrich, München, Germany, if not mentioned else. The materials for media preparation were provided by Carl Roth, Karlsruhe, Germany, unless otherwise stated.

4.1 Cultivation and Media

4.1.1 C. glutamicum ATCC 13032 for Studies of SOS Response and Prophage Induction

C. glutamicum ATCC 13032 reporter strains of Tab.1 were pre- cultivated for SOS response and phage induction visualization in a 20 mL shaking culture with brain heart infusion (BHI, BD, Heidelberg, Germany) and with addition of 25 μg/mL kanamycin (KAN) at 30 °C overnight. Consecutively, a second shaking culture of the same volume with minimal medium CGXII, which is described by Keilhauer et al. (1993), containing 4 % (w/v) glucose (GLC) and 25 μg/mL KAN (defined as standard cultivation condition for fluorophore expression) was incubated with the initial pre-culture 70. For microfluidic cultivation, the microfluidic device was infused

39 Material and Methods continuously with medium CGXII containing 25 μg/mL KAN as indicated at a rate of 300 nL/min with a high-precision syringe pump (neMESYS, Cetoni GmbH, Korbussen, Germany) to ensure stable and constant environmental conditions. A constant cultivation temperature of 30 °C was ensured by an incubation chamber (PeCon GmbH, Erbach, Germany). For initiation of a starvation phase the medium was switched to minimal medium lacking glucose and protocatechuate (no carbon source) for 24 h after an initial growth phase of ~8 h. Cells were exposed to carbon limitation as described in the results section. Then growth was continued under standard growth condition with GLC. Additionally, minimal medium without phosphate, iron or nitrogen was tested with addition of CvAM as described in 4.1.2. For SOS-response, fluorophore expression was either under control of the plasmid encoded promotor of RecA fused to e2-Crimson or in the bacterial chromosome integrated espression of RecA-eYFP and RecA-Venus, respectively. For phage induction, fluorophore expression was controlled either by the promotor of CGP3 integrase

30 (Pint-eyfp) or by the promotor of the putative lyase lysin (Plys-eyfp ,

22 and the dual reporter strain with Plys-e2-crimson , respectively). For more information according to reporter strain construction the interested reader is referred to 22,30,71. The parameters of imaging can be found in Tab. 1. For fluorophore excitation bias, custom-made green-yellow reduced fluorescent reference beads (PolyAN, Berlin,

40 Material and Methods

Germany) were used for development of a quantitative expression dynamic measurement method.

Table 1: Reporter strains, control strains, and imaging parameters C. glutamicum ATCC Interval Exposure Camera 13032 strains times/intensity and references of phase contrast/YFP channel/Crimson channel

ATCC 13032/pJC1-PrecA-e2- 20 min 50 ms (3 %)/ CCD Andor crimson 71 -/ Clara-DR- 300 ms (~ 3 %) 3041 71 ATCC 13032::PrecA-eyfp 10 min 50 ms (3 %)/ CCD Andor 200 ms (~ 3 %)/- Clara-DR- 3041

ATCC 13032::PrecA- 8 min 100 ms (3 %)/ CCD Andor 30 eyfp/pJC1-Pint2-e2-crimson 300 ms (~ 0.4 %)/ Clara-DR- 400 ms (~ 0.4 %) 3041

ATCC 13032::PrecA- 8 min 100 ms (3 %)/ CCD Andor 30 eyfp/pJC1-Plys-e2-crimson 300 ms (~ 0.4 %)/ Clara-DR- 400 ms (~ 0.4 %) 3041

ATCC 13032::PrecA- 8 min 100 ms (3 %)/ CMOS Andor

venus/pJC1-Plys-e2-crimson 400 ms (~ 0.4 %)/ Neo SCC- 22 600 ms (~ 0.4 %) 01462

ATCC 13032/pJC1-Ptac-venus 8 min/ 100 ms (8 %)/ CMOS Andor 22 continuous 400 ms (~ 0.4 %)/ Neo SCC- 600 ms (~ 0.4 %) 01462

41 Material and Methods

C. glutamicum ATCC Interval Exposure Camera 13032 strains times/intensity and references of phase contrast/YFP channel/Crimson channel

ATCC 13032/pJC1-Ptac-e2- 8 min/ 100 ms (8 %)/ CMOS Andor crimson 22 continuous 400 ms (~ 0.4 %)/ Neo SCC- 600 ms (~ 0.4 %) 01462

ATCC 13032/pJC1-Ptac-eyfp continuous - CMOS Andor (provided by R. Freih. v. Neo SCC- Boeselager, RG Frunzke) 01462

72 ATCC 13032::Plysin-eyfp 8 min 100 ms (8 %)/ CMOS Andor 400 ms (~ 0.4 %)/ Neo SCC- 600 ms (~ 0.4 %) 01462

ATCC 13032 ΔdtxR::Plysin- 8 min 100 ms (8 %)/ CMOS Andor eyfp 72,73 400 ms (~ 0.4 %)/ Neo SCC- 600 ms (~ 0.4 %) 01462

4.1.2 C. glutamicum for Dynamic Metabolic Activity Measurement

C. glutamicum ATCC 13032 was pre-cultivated in a 20 mL shaking culture with BHI at 30 °C overnight. For microfluidic cultivation using complex medium BHI, another shaking flask culture was inoculated with the pre-culture. Exponentially growing cells of the shaking flask culture were transferred into the microfluidic device. Else two further 42 Material and Methods pre-cultures were performed in minimal medium CGXII + 4 % (w/v) GLC 70. Media with violet fluorogenic calcein acetoxymethyl ester (CvAM) for microfluidic cultivations were modified as indicated by pH adjustment, omitted addition of carbon, protocatechuate (PCA), and iron, respectively, or addition of 10 µg/mL ampicillin (AMP) or 10 µg/mL chloramphenicol (CHL), respectively. A constant cultivation temperature at 30 °C and medium perfusion were ensured for microfluidic cultures as described in 4.1.1.

4.1.3 C. glutamicum for Intracellular Reactive Oxygen Species Measurement

For reactive oxygen species (ROS) development in C. glutamicum ATCC 13032, and its knockout mutants C. glutamicum ATCC 13032 ΔctaD and C. glutamicum ATCC 13032 Δqcr were pre-cultivated in a 20 mL shaking flask culture with BHI. If not successively cultivated in the microfluidic device, dense bacterial cultures were diluted 1:20 in CGXII + 4 % GLC consecutively three times when stationary phase was reached. The fourth pre-culture was considered as absent of an influence of the BHI medium and exponentially growing cells were inoculated in the microfluidic device. Additionally, pro-longed precultivated C. glutamicum ATCC 13032, and the mutants C. glutamicum ATCC 13032 ΔctaD and C. glutamicum ATCC 13032 Δqcr were additionally

43 Material and Methods tested in temporal resolved viability (described in 4.1.4) and metabolic activity (4.1.2). Phtototoxicity determination of C. glutamicum strains under imaging condition was performed with the same pre-cultivation and time-lapse imaging parameters as used for the microfluidic cultivation condition that is of interest.

4.1.4 Continuous in Vivo Viability Staining and Antibiotic Sensitivity Testing

A constant cultivation temperature in the microfluidic device was ensured by an incubation chamber (PeCon GmbH, Erbach, Germany). C. glutamicum ATCC 13032 cultures were pre-cultured with BHI at 30 °C overnight. Then, the primary shaking culture was inoculated in BHI medium, or additional pre-culture was performed using minimal medium (CGXII containing 4 % GLC (w/v) to inoculate the primary shaking culture with the same minimal medium 70. Cells from shaking flask cultures were injected into the microfluidic device for single-cell cultivation and perfused with medium as indicated. B. subtilis 168 was cultivated in BHI at 37 °C in shaking flasks and in the microfluidic device. V. harveyi DSMZ 6904 was grown in marine broth 2216 (BD, Heidelberg, Germany) at 30 °C (shaking cultures) and 28 °C (microfluidic cultivation). M. luteus DSMZ 14234 shaking cultures were grown in a nutrient solution consisting of 5 g/L peptone and 3 g/L meat extract, adjusted to pH 7, and incubated at 30 °C before

44 Material and Methods cultivation at 28 °C in the microfluidic device. E. coli MG1655 was cultivated in lysogeny broth (LB) containing 5 g/L yeast extract, 10 g/L peptone, and 10 g/L NaCl at 37 °C in shaking cultures and during microfluidic cultivation. Chemically competent E. coli BL21CodonPlus(DE3)-RIL cells were transformed with pET28b(pezTΔC242), pET28b(pezTΔC242(D66T)), or pET28b(pezA/pezT) (kindly provided by A. Rocker and A. Meinhart, MPI Heidelberg, Germany). E. coli BL21CodonPlus(DE3)-RIL shares high genomic similarity with E. coli MG1655 and was used in lieu of MG1655 due to its improved codon usage and induction properties for heterologous protein expression74. After transformation, mutants were cultivated at 37 °C overnight on LB media agar plates with 50 µg/mL KAN and 34 µg/mL CHL. The next day, single colonies were selected and transferred to LB medium containing 50 µg/mL KAN and 34 µg/mL CHL. Before cell seeding in the microfluidic device, shaking cultures were incubated at 37 °C until an optical density higher than 0.1 was reached. Protein expression during microfluidic cultivation was induced by changing the LB perfusion medium containing 50 µg/mL KAN and 34 µg/mL CHL to an LB medium containing 100 µM IPTG in addition to both antibiotics.75 Commercially available compressed baker’s yeast (S. cerevisiae) (UNIFERM GmbH & Co.KG, Werne, Germany) from the supermarket was dissolved in YPD medium (20 g/L peptone, 10 g/L yeast extract, and 20 g/L glucose) and pre-cultivated 48 h prior to the

45 Material and Methods primary cultivation for microfluidic device inoculation. Starved cells were cultivated prior to microfluidic cultivation with fresh YPD medium for 48 h in YPD medium or 0.9 % NaCl solution (w/v).

4.2 Microfluidic Device

A microfluidic device platform was used for microbial cultivation and single cell analysis, that is described in detail by Kohlheyer and coworkers 36. The microfluidic device as depicted in Fig. 6 A is based on common polydimethylsiloxane micro-moulding as described in full detail by Grünberger et al. (2013) and (2015) 35,36, and designed by A. Grünberger. The picoliter sized micro-structured cultivation chambers (Fig. 6 C-D) have a height of approximately 1 µm facilitating the monolayer growth of isogenic microcolonies. Several hundred cultivation chambers are arranged in parallel in four arrays for each of the four main channels (Fig. 6 B). The different media used were infused with a high-precision syringe pump (neMESYS, Cetoni GmbH, Korbussen, Germany) at a rate of 300 nL/min. A constant cultivation temperature was ensured by an incubation chamber (PeCon GmbH, Erbach, Germany).

46 Material and Methods

Figure 6: Microfluidic device and its submicrostructures. (A) Assembled microfluidic device with connected inlet and outlet tubing. (B) Parallel cultivation chamber arrays with branched main channels that subdivide in smaller channels for media perfusion. A single device contains four cultivation chamber arrays with 352 chambers each. (C) Design I used for cultivations of less than 8 hours with a single entrance on each side connected to the perfusion channel for diffusive nutrient supply and initial inoculation with ideally one cell. Design II for long time cultivation with shifting media supply has 12 smaller entrances and an increased cultivation surface in comparison to Design I. (D) SEM micrograph of a cultivation chamber of Design I and Design II after fabrication. (adapted from 16)

4.3 Fluorescence Time-Lapse Imaging

The microfluidic device was installed on an inverted epifluorescence microscope (TI-Eclipse, Nikon GmbH, Düsseldorf, Germany), which was equipped with a motorized stage (Nikon GmbH, Düsseldorf, Germany), a high-speed charge-coupled device (CCD) camera (Clara DR-3041, Andor Technology Plc., Belfast, United Kingdom), a digital

47 Material and Methods

CMOS camera (Neo sCMOS, Andor Technology Plc., Belfast, United Kingdom), the Nikon Perfect Focus System (PFS, Nikon GmbH, Düsseldorf, Germany) for thermal drift compensation, and a Plan Apo 100 Oil Ph3 DM objective (Nikon GmbH, Düsseldorf, Germany). For phase contrast images, a cooled LED light source (at 3-8 % of total intensity) was used. Epifluorescence illumination was performed with a mercury light source (Intensilight, Nikon GmbH, Düsseldorf, Germany, set to 1/32 of total intensity and additionally reduced 1/8 by filter settings). Fluorescent time-lapse imaging was carried out based on the growth rate and photosensitivity of the species, described as follows: every 8 min for C. glutamicum, every 5 min for E. coli, B. subtilis, and V. harveyi, every 30 min for S. cerevisiae, and every 60 min for M. luteus. An APC filterset (EX 600/37 nm, 630 nm DM, BA 675/67 nm, AHF Analysentechnik AG, Tübingen, Germany) was installed for e2- Crimson, and for eYFP and Venus, a HQ YFP filterset (EX 520/30 nm, 510 nm DM, BA 540/20 nm, AHF Analysentechnik AG, Tübingen, Germany)was used. A TRITC filter (EX 540/25 nm, DM 565 nm, BA 605/55 nm, Nikon GmbH, Düsseldorf, Germany) and a Texas Red filter (EX 540-580 nm, DM 595 nm, BA 600-660 nm, AHF Analysentechnik AG, Tübingen, Germany) were employed for PI signal imaging. CvAM and CbAM were analysed with a DAPI filter (EX 340-380 nm, DM 400 nm, BA 435-485 nm, Nikon GmbH, Düsseldorf, Germany). CgAM was excited with a FITC filter (EX

48 Material and Methods

465-495 nm, DM 505 nm, BA 515-555 nm, Nikon GmbH, Düsseldorf, Germany). A CFP HC filter set (EX 438/24 nm, DM 458 nm, BA 483/32 nm, Nikon GmbH, Düsseldorf, Germany) was used for the PO- PRO1 dye. Dihydrocalcein acetoxymethyl ester (DHCAM) was excited and imaged every 6th frame using a filter set (EX 465 – 495 nm, DM 505 nm, BA 515 – 555 nm). The transient state of initial CAM uptake and conversion of freshly seeded bacterial cells during the experimental set-up phase were imaged at high temporal resolution (frame interval 2.4 sec). Elsewise, an imaging frame interval of 8 min was kept for C. glutamicum cultivations. For, single-cell studies of respiratory chain gene deficient strains, fluorescence time-lapse imaging was performed with an interval of 8 min for C. glutamicum ATCC 13032, and an interval of 10 min for C. glutamicum ΔctaD and C. glutamicum Δqcr, respectively.

4.4 Metabolic Activity Sensing Method and Experimental Validation

The perfusion media for microfluidic cultivation contained 46.3 µM violet fluorogenic calcein acetoxymethyl ester (CvAM) (Thermo Fischer Scientific, Darmstadt, Germany), if not indicated else, and were infused at a rate of 300 nL/min with a high-precision syringe pump (neMESYS, Cetoni GmbH, Korbussen, Germany) through the microfluidic device during time lapse imaging. The rate of 49 Material and Methods photobleaching was determined with growth inhibited cells under glucose free conditions by constant illumination with the same light intensity chosen for all experiments. Cells were grown overnight and perfused with CGXII + 4 % GLC and CvAM. The medium perfusion was stopped for several hours until the calcein fluorescence was constant before photobleaching was measured on 85 individual single cells in a fluorescence range from 110 AU to 700 AU. The intensity loss by photobleaching during light exposure was determined as percentage from the initial single cell mean fluorescence per frame. The presence of phototoxicity was determined by detection of radical oxygen species (ROS) with dihydroxycalcein acetoxymethyl ester (DHCAM). The influence of co-metabolisation of the ester groups of CAM was analysed using the CvAM surrogate substrate methyl methoxyacetate. Three mol methyl methoxyacetate corresponded to one mol CvAM.

4.5 Internal Radical Oxygen Species Determination in C. glutamicum Strains

Medium with 0.2, 2.0 or 200 µg/L thiamine (Carl Roth, Germany) or no thiamine was infused at a flow rate of 300 nL/min with a high- precision syringe pump (neMESYS, Cetoni GmbH, Korbussen, Germany). The microfluidic biochip was perfused with minimal medium CGXII + 4 % GLC (w/v). Radical oxygen species (ROS) formation was induced by mild phototoxic stress (doubled 50 Material and Methods fluorescence excitation compared to phototoxicity evaluation of other imaging analyses) 76. Reduced dihydroxycalcein-acetoxymethyl ester (DHCAM, Life Technologies GmbH, Darmstadt, Germany) was used to detect especially energetic singlet oxygen radicals. OH⸱ were sensed using chlormethyl-2,7-dichlorodihydrofluorescein diacetate

(CM-H2DCFDA, Life Technologies GmbH, Darmstadt, Germany) For additionally real-time viability analysis and metabolic activity sensing, propidium iodide (PI) (Carl Roth, Karlsruhe, Germany) and calcein acetoxymethylester (CvAM) (Life Technologies GmbH, Darmstadt, Germany) was added to the perfusion medium as described in chapter 4.7, in chapter 4.4 as well as in 16,17.

4.6 Internal pH Measurement in C. glutamicum

C. glutamicum ATCC 13032 was pre-cultivated in a 20 mL shaking culture with BHI and two consecutive pre-cultivation cultures with CGXII 70 + 4 % GLC to perform internal pH calibration and measurement with the pHrodo AM variety pack (Thermo Fischer Scientific, Darmstadt, Germany) in microfluidic cultivation devices. For calibration cells were chemically permeabilised with nigircin (NIG) and/or valinomycin (VAL) as discussed in the results section and described else in 77. The internal pH measurement was performed using ratiometric single-cell fluorescence determination of pH sensitive pHrodo RED and pHrodo GREEN (Thermo Fischer Scientific, Darmstadt, Germany) in C. glutamicum ATCC 13032. A 51 Material and Methods calibration was performed by perfusion of the kit buffer solutions with equimolar NIG and VAL (without power load solution) at pH 4.5, 5.5, 6.5, and 7.5. A second calibration was performed with medium containing NIG at pH 6.6, 7.0, and 7.4.

4.7 Dynamic Viability Staining

PI (Carl Roth, Karlsruhe, Germany) stock solution (20 mM) was prepared and dissolved in sterile water. PI was added to the media perfused through the microfluidic device. An end concentration of 0.1 µM was used for dynamic viability staining in the microfluidic device unless otherwise stated. Counterstaining with CvAM (Life Technologies GmbH, Darmstadt, Germany) was performed for C. glutamicum ATCC 13032 as described in detail elsewhere 16. CgAM (Life Technologies GmbH, Darmstadt, Germany) and calcein blue acetoxymethyl ester (CbAM, Life Technologies GmbH, Darmstadt, Germany) were used as sequestering agents for the dynamic staining of S. cerevisiae cells. All calcein acetoxymethyl esters (CAMs) were prepared as fresh 2.5 mg/mL stock solutions in water-free DMSO (Carl Roth, Karlsruhe, Germany) directly before use. In addition to 0.1 µM PI, 46 µM CvAM, 60 µM CbAM, or 28 µM CgAM was dissolved in the indicated perfusion media. Alternatively, 2 µM PO-PRO1 (Life Technologies GmbH, Darmstadt, Germany) was added to the PI- containing perfusion medium to indicate cells without membrane potential. Radical oxygen species (ROS) formation is an indicator of 52 Material and Methods phototoxicity 76. Reduced dihydroxycalcein-acetoxymethyl ester (DHCAM, Life Technologies GmbH, Darmstadt, Germany) was used to detect ROS. Dye adsorption or absorption by PDMS was not observed as determined by LogP values and molecular size.

4.8 Additional Staining Methods

Additional endpoint staining of cells treated continuously with 50 µg/mL CHL was performed after 24 hours. Dynamic PI staining during antibiotic addition was combined with 10 µg/mL DAPI staining to visualize total DNA. Dynamic CvAM conversion due to CHL was endpoint-stained using 2 µg/mL Nile red (a membrane indicator). For Nile red staining, 10 µg/mL BSA was added to the perfusion medium for passivation. Both endpoint staining methods were performed with perfusion of 0.15 % PFA for mild cell fixation.

4.9 Plate Count Test for Antibiotic Sensitivity Testing

Antibiotics used were ampicillin (AMP), Kanamycin (KAN) (both from Carl Roth, Germany), Chloramphenicol (CHL), Streptomycin (STR) (both from Merck, Germany), and Mitomycin C (MMC) (Sigma Aldrich, Germany). McFarland standard solution with an absorbance of 0.5 consisted of 5 % barium chloride dihydrate solution (Merck, Germany) at a concentration of 11.75 g/L and 99.5 % sulfuric acid solution (1 % (v/v) (Carl Roth, Germany).

53 Material and Methods

BHI agar plates with 0 mg/L (reference), 0.1 mg/L, 1 mg/L and 10 mg/L AMP, KAN, CHL, or STR were prepared. A bacterial solution, sampled of exponentially growing shaking cultures, was adjusted with 9 g/L NaCL (Carl Roth GmbH, Germany) to approximately 108 CFU/mL using the McFarland standard. A serial dilution (1:10) was performed of this bacterial solution and diluted bacterial solutions (104 – 105) were plated in triplicate on reference agar plates and plates of three different antibiotic concentrations.

4.10 Data Analysis

Time lapse image data was analysed with a customized workflow implemented as an ImageJ/Fiji plugin 47. Cell identification was performed using a segmentation procedure specialized to detect individual bacteria in crowded populations. The segmentation overlay was corrected manually frame by frame for every experiment. A selection of detected cells was subsequently tracked throughout image sequences using an adapted single particle tracking approach as implemented in TrackMate 49. The image analysis was used for extraction of area and fluorescence of individual cells, as well as derived quantities (i.e., mean fluorescence of all cells of a colony). The analysis and visualization software Vizardous 50 assisted with analysis and interpretation tasks of single cell data in an interactive, configurable and visual way by augmenting lineage trees with time- resolved cellular characteristics. 54 Material and Methods

Time-lapse imaging data from antimicrobial-treated C. glutamicum ATCC 13032 cells (Fig. 49 and Fig. 55), yeast cells (Fig. 48), and E. coli BL21(DE3) CodonPlus(DE3)-RIL cells transformed with pET28b(pezTΔC242), pET28b(pezTΔC242(D66T), or pET28b(pezA/pezT) (Fig. 58) were analysed using the counting option of the NIS-Elements software program to enumerate total cell numbers and PI+ cells with identical LUT settings. The analysis of prokaryotic imaging data, shown in Fig. 48, Fig. 50-52, Fig. 54 and Fig. 56, was facilitated by a user-specialized workflow run constructed as an ImageJ/Fiji plugin 47. Cell identification was performed using a segmentation procedure tailored to detect individual bacterial cells in crowded populations. All frames were checked manually to remove artefacts as well as to identify segmentation and cell identification failures. Yeast cells were segmented manually using ImageJ. Identified cells were subsequently linked throughout image sequences by implementing an adapted single-particle tracking approach with TrackMate 49. Image analysis permitted the extraction of measurable quantities of individual cells, (i.e., mean fluorescence) as shown in Fig. 56, Fig. 62, and Fig. 54. Finally, data sets derived from C. glutamicum ATCC 13032 treated with antibiotics were processed to generate individual single-cell traces over time using the analysis and visualization software Vizardous, as recently described in detail 50.

55 Material and Methods

4.11 Calculations

The apparent growth rate µapp was calculated by the increase in the sum of cell sizes of a colony during 10 frames (80 min) in minimal medium CGXII and 6 frames (48 min) in complex medium BHI. Single cell reaction rate constants of conversion or efflux were determined by the course of the mean fluorescence directly after shift of media condition to carbon free medium (CGXII – GLC – PCA) and after backshift to CGXII + 4 % GLC, respectively.

4.12 Cell Classification Criteria

SOS positive cells were defined as cells that expressed e2-crimson, eyfp or venus. Phage induction in C. glutamicum ATCC 13032 cells was indicated by expression of eyfp or e2-crimson. For both auto- induced expression, a threshold was defined upon basal cell fluorescence and fluorescence maximum. Calcein single-cell fluorescence showed differ fluorescence intensities according the cell fitness. Efflux inhibited, non-growing cells showed maximal fluorescence signals whereas metabolically active, growing and dividing cells exhibited medium fluorescence. Non-growing dead or lysed cells showed no fluorescence signal. PI intrusion of cell wall-compromised cells was indicated by red fluorescence. PI+ cells were considered dead. Cells were defined as PI+ if the PI fluorescence increased drastically between two frames or if

56 Material and Methods the initial basal fluorescence signal was increased by more than 5 % and demonstrated an increasing trend in subsequent time-lapse images until equilibrium was achieved. CAM conversion to fluorescent CAL indicated enzymatic activity in cells or cell organelles (yeast). Uninhibited growing C. glutamicum ATCC 13032 cells exhibited moderate fluorescence (CAL+) due to the partial efflux of CAL, whereas cells with reduced metabolic activity exhibited increased CAL fluorescence (CAL++) in comparison to CAL+ cells 16. Yeast cells with reduced V-ATPase activity retained CAL in their vacuoles. A reduction in cellular enzymatic activity was a criterion for reduced cell activity and cell survival. Cells that lost their membrane potential or had cell membrane injuries underwent PO-PRO-1 intrusion, indicated by blue fluorescence. PO-PRO-1+ cells were considered dying or dead, but recovery was observed for yeast cells. Cells were defined as PO- PRO-1+ if the PO-PRO-1 fluorescence increased drastically between two frames. Lysed cells were non-fluorescent due to the loss of molecule retention, observed as pale cells in phase contrast images, or cell debris disintegration. A positive control of dynamic PI staining was performed. After an initial growth phase under optimal growth conditions for C. glutamicum, M. luteus, B. subtilis, E. coli, V. harveyi, and S. cerevisiae, cells were continuously perfused with 50 mg/mL sodium cyanide and PO-PRO-1 to visualize the loss of membrane potential due to intoxication. After a substantial number of cells lost their

57 Material and Methods membrane potential, dynamic PI staining was initiated. Signal-to- noise ratios were calculated as difference between the mean fluorescence and the mean background fluorescence divided by the standard deviation of the background fluorescence. The signal-to- background ratio was calculated as the ratio of the mean signal and the mean background fluorescence.

58 Results

5. Results

The following text has been adapted partly from Krämer, C. and Kohlheyer, D. (2016), BIOspektrum 22 (1): 48-50 as described at the end of the text section. The text has been written and published originally in German and was translated into English. The reader is referred to Krämer, C. and Kohlheyer, D. (2016), BIOspektrum 22 (1): 48-50.

Microorganisms adapt to environmental changes by DNA embedded mechanisms (e.g. resistance, genetic switches) or non-DNA based phenotypical differentiation (e.g. antibiotic tolerance). As described in Krämer et al. (2016), single-cell fluorescence imaging combined with environmental control by microfluidic devices helps to unravel phenotypical changes of single microbial cells in isogenic cell populations. Therefore, the intracellular implementation of fluorescent molecules by genetic modification or non-invasive fluorescence in situ staining (FISS) is essential for non-invasive live-cell imaging. Microorganisms have developed strategies to react rapidly to environmental changes. Thereby, the survival strategies are versatile developed and adapted to the particular extracellular condition. The analysis of heterogeneous phenotypes on a single-cell level (e.g., persistence, tolerance) has growing relevance in fundamental research

59 Results

68,78. However, the triggers of heterogeneous single-cell differentiations are not totally defined, yet. The spontaneous development of subpopulations in isogenic colonies is partly due to mutations, partly stochastic distributions of macromolecules during cell division metastability of genomic switches 53,78,79. Although, there are still unanswered questions how intracellular processes generate these population minorities and how some mechanisms (e.g. altruistic cell death) are controlled to establish advantages for the survival of the species to overcome evolutionary selection. A temporal and spatial resolved phenotype characterization of single microbes is recommended for the analysis of heterogeneous cell reactions. Microscopic fluorescence time lapse images of colonies in microfluidic cultivation devices enable development of versatile single-cell analysis 80. To visualize changes in living cells by intracellular fluorescence signals, expression cassettes are introduced in the genome or in bacterial plasmids. These expression cassettes consist of an inducible promoter, that regulates the translation of the target protein, which is followed by a subsequent gene sequence of a fluorescence protein 22,30,80. A further innovative method is the continuous staining during cultivation by fluorescence indicators that are not influencing cell physiology 16,17,28,80. Conventional fluorophore expression and innovative novel staining approaches are presented subsequently for multiplexed fluorescence based single-cell analysis of microbial stress and survival.

60 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

5.1 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

The following text has been adapted partly from Krämer, C. and Kohlheyer, D. (2016), BIOspektrum 22 (1): 48-50. Results were partly published in Nanda, A. M., Heyer, A., Krämer, C., Grünberger, A., Kohlheyer, D., Frunzke, J. (2014), J Bacteriol 196 (1): 180-188 and in Helfrich, S., Pfeifer, E. Krämer, C., Sachs, C. C., Wiechert, W., Kohlheyer, D., Nöh, K., Frunzke, J. (2015), Mol Microbiol. 98 (4): 636-650. Additional contributions are described at the end of this chapter. The chapter text has been partly published in German and was translated into English. Figures 7, 8, and 18 have been additionally layouted before publishing by D. Kohlheyer. The reader is referred to Krämer and Kohlheyer (2016), BIOspektrum 22 (1): 48-50. C. glutamicum strains have been received from the research group of J. Frunzke as described in the Material and Methods section to obtain the presented results. Results from coworkers have been cited as published in Nanda, A. M., Heyer, A., Krämer, C., Grünberger, A., Kohlheyer, D., Frunzke, J. (2014), J Bacteriol 196 (1): 180-188 and in Helfrich, S., Pfeifer, E. Krämer, C., Sachs, C. C., Wiechert, W., Kohlheyer, D., Nöh, K., Frunzke, J. (2015), Mol Microbiol. 98 (4): 636-650.

61 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Chapter Abstract: C. glutamicum ATCC 13032 has been genetically modified to resulting in reporter strains for SOS response and the gene induction of CGP3 that is an inherent prophage in the organism genome. Plasmid-based and chromosome-integrated fluorophore fused protein expression have been advanced to dual reporter strains for spontaneous SOS response, indicated by enhanced yellow fluorescent protein (eYFP) derivatives, and sporadic prophage induction, visualized by red fluorescent e2-crimson, under non- stressful cultivation conditions using perfusion of minimal medium CGXII + 4 % GLC in a microfluidic device combined with multiplexed fluorescence time-lapse imaging. A phenotypic minority evolved with all tested reporter strains under non-stressful conditions, which were non-growing SOS positive (SOS+) cells with different cell fate. One fraction of SOS+ cells resuscitated and lost fluorescence with following cell divisions, another fraction remained continuously in SOS+ state, and a third fraction proceeded to induce subsequent prophage induction (phage+). However, a remarkable number of phage+ cells appeared independent of SOS response. Absence of phototoxic stress has been tested by a novel developed sensing method using a homeopathic addition of the fluorogenic co- substrate dihydrocalcein acetoxymethyl ester (DHCAM) that is taken up by the bacterium of interest, converted to a green fluorescent oxidized calcein derivative (CALox) in presence of intracellular esterase activity and photo-induced radical oxygen species (ROS) as

62 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum described in chapter 5.3. SOS+ cells and phage+ cells appeared randomly distributed in the microfluidic cultivation device. The temporal fluorophore expression bias after induction was tested by IPTG inducible control stains and was determined to be approximately 12 min. The induction of SOS response and prophage induction was studied under nutrient deprivation conditions without carbon, iron, phosphate and nitrogen. In addition, metabolic activity of C. glutamicum ATCC 13032 has been tested by perfusion of the violet fluorogenic substrate calcein acetoxymethyl ester (CvAM) that is transported prior internal esterase hydrolysis to the product violet fluorescent calcein (CALv), which is secreted energy-dependent as shown in chapter 5.4. A direct trigger of phage induction by iron could not be concluded. Under carbon free hunger condition, SOS and phage gene induction decreased and also CALv efflux reduction revealed CALv accumulating cells considered as dormant, existed in non-fluorescent dead state, SOS+ state, and SOS+ state transforming to SOS+/phage+ state with reducing phage+ state indicating fluorescence.

5.1.1 Phenotype Characterization of Living Cells – an Introduction

Genetic modification of cells is well established in biological research to elucidate functions of DNA sequences. Fused, truncated, or 63 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum modified genes can be introduced by different strategies in the host organism. The expression of these genes is mostly indicated if a fluorophore (fluorescent protein molecule) is co-transcribed and translated. Therefore, a plethora of fluorophores was developed for genetic studies 81. Fluorophore expression is considered as non- invasive approach for fluorescence sensing, also it is reported that modification of the host genome can lead to alterations of the metabolic phenotype, inclusion body formation, cause limitations of rare tRNAses, and show expression bias on single-cell level 82–84. However, fluorophore expression is a frequent practised method for precise analysis of gene functions in living cells without intended molecule attachment or avital intercalation during transcription as it is the case with FISH probes or DNA dyes. Non-invasive staining strategies can be developed for living bacteria and their descendants as it is the case for eukaryotes 16,17,85. Unfortunately, the repertoire of ready available fluorochromes (fluorescent non-protein molecules) reported for dynamic bacterial in vivo analysis is rarely given at the moment, although there is an upward trend in the near future, as reviewed by Krämer et al. (2016) recently 28. A multiplexing single-cell analysis with combined spontaneous fluorophore expression and dynamic, intracellular, and novel fluorochrome sensor integration is presented in this chapter with calcein derivatives to visualize the state of phototoxic stress and metabolic activity as described in chapters 5.2 – 5.6 in detail 16.

64 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Nevertheless, conventional multiplexing single-cell analysis use genetically modification strategies for temporal resolved imaging. Schlüter et al. (2015) used a functional triple reporter strain approach with expression of three fluorophore fusion proteins for phenotypic time-lapse imaging 67. Accordingly, an approach of dual fluorophore expression was integrated in a microfluidic cultivation device to analyse the correlation of prophage genes and genes activated during cell stress due to erroneous replication in C. glutamicum ATCC 13032. Therefore, a protein fusion of GFP derivatives (eYFP, Venus) and e2- Crimson to promoters of the bacterial SOS cascade and of cryptic enzyme activities of the CGP3 prophage were investigated using dynamic phenotype development analysis by sensitive fluorescence time-lapse imaging. There were multiple requirements to perform fluorochrome sensor integration for multiplexed metabolic activity sensing of a dual reporter strain for single-cell imaging analysis: i.) Excitation and emission spectra overlap with expressed fluorophores had to be avoided, ii.) small Stokes-shift of the sensor was preferred, iii.) use of a present filter set to avoid extra costs, iv.) absence of chemotoxicity, v.) cell permeation under cultivation conditions, vi.) low background fluorescence, vii.) no interaction with the microfluidic device material, and viii.) sufficient intracellular fluorescence output. These criteria were all solved by the use of brightly violet fluorescent CALv, that

65 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum has been found to be a non-invasive metabolic activity sensor for actinobacteria 16 as shown in chapters 5.3 and 5.4. There is a broad range of conventionally performed biological phenotype analysis, but the remaining challenge is to analyse living bacteria without unwanted impact on their viability or gene expression 80. The reaction of bacteria upon environmental triggers is evolutionary optimized and it is a still ongoing process to ensure the survival of the species. Microscopic fluorescence time-lapse imaging in combination with microfluidic cultivation devices enable the continuous observation of single cells and their descendants during periods of constant conditions or controlled changing cell environments. For long time cultivation and observation of microorganisms, our research group developed microfluidic devices with integrated cultivation structures that trap cells for cultivation and for observation of cell growth (Fig. 7 A, B, 9,36,80). The supply channels of the microfluidic devices are perfused continuously with defined medium. Thus, the cultivation chambers in between the channels are diffused with nutrients. Cultivation chambers are selected for microscopic observation in defined intervals (Fig. 7 C, 80). Afterwards, the image analysis reveals single-cell growth and remarkable evolving phenotypes. Previously, fluorescent biosensor strategies were developed for implementation in living cells (Fig. 7 D, 80). One approach of in vivo

66 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum fluorescence integration can be performed by genetic modification to determine fluorescent protein expression of cells. In expression cassettes, a fluorophore is built additionally to the target protein by the cell. Therefore, fluorophores are integrated by molecular biological methods either plasmid-based or chromosome-integrated 22,71. Differences in copy numbers of the plasmid-borne fluorescent protein directly influence the fluorescence signal of the cell, whereas chromosomal-integrated fluorophores normally yield reduced intracellular fluorescence compared to a plasmid-integrated approach with comparable higher copy numbers. Another approach to achieve intracellular fluorescence is continuous staining by fluorochromes (fluorescent dye molecules). A selection of fluorescent dyes have to be validated carefully by toxicity tests, before they are applied for long time analysis of living cells 16,17,86. Fluorochromes can be coupled to functional moieties that result in florescence quenching and that are considered as substrates of the cell to facilitate molecular uptake. The conversion of such fluorogenic substrates to fluorescent products gives insight in enzymatic activities of single living cells 16,28,87. The combination of multiple fluorescence signals reveals interaction of intracellular processes 17,22,67. Simultaneous imaging of multiple fluorescence signals by multiplexing facilitates approaching cell models and enables their validation 22. We implemented different artificial fluorescence sensors in the chosen model organism

67 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

C. glutamicum ATCC 13032 as shown in Fig. 7 D and combined them for analysis of the bacterial stress response, the induction of genome integrated prophage, the single-cell metabolic activity, and internal phototoxicity determination of C. glutamicum ATCC 13032. If double-stranded DNA (dsDNA) breaking occurs, bacteria switch on their intracellular SOS response that is indicated by expression of the fluorophore eYFP or Venus under the control of the chromosomal integrated recA promotor. The induction of the genome integrated prophage CGP3 of C. glutamicum ATCC 13032 is linked to the fluorescent protein e2-Crimson (red fluorescence) by plasmid-based expression of a lytic prophage enzyme 22,30. In parallel, the metabolic activity of the bacterium was determined as described in 16 and shown in chapter 5.4. Thus, the violet fluorogenic substrate calcein acetoxymethyl ester (CvAM) was added in traces to the perfusion medium. As described in chapter 5.3, acetoxymethyl ester (AM) bound molecular probes are taken up by actinobacteria, metabolized by intracellular esterases and converted to their fluorescent products as the corresponding calcein of CvAM (CALv, violet fluorescence) that is effluxed energy-dependent by putative ATPases of C. glutamicum ATCC 13032 16. Dormant cells showed because of their reduced metabolism a remarkable increased CALv fluorescence as shown in detail in chapters 5.4 and 5.5, and in 16,28.

68 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 7: Single-cell analysis by fluorescence time-lapse imaging. (A) The microfluidic single-cell analysis device consists of a glass slide that is bond with a polydimethylsiloxane (PDMS) chip with imbedded microstructures. (B) A section of the microstructures shows the supply channels and cells in the microcultivation chambers, which are perfused with nutrients. (C) Micrographs are taken automised to analyse growth and fluorescence as time-lapse sequence. (D) The single-cell fluorescence analysis is performed using different fluorescence strategies by 1. chromosome integrated fluorophore expression, 2. plasmid-based fluorophore expression, and 3. conversion of non-toxic fluorogenic substrates.80

69 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 8: Dynamic single-cell behaviour can be analysed under (A, a) constant cellular microenvironment to characterize (B, a) spontaneous phenotypic cell responses, and under (A, b) discontinuous microenvironments induced by precise media supply profiles to (B, b) trigger cell responses and analyse cell reactions on environmental stressors. 80

For analysis of dynamic phenotypic single-cell behaviour, controlled environmental conditions as given in microfluidic devices are of high relevance. Continuous cell cultivation conditions help to analyse the spontaneous and non-induced cell response of single cells (Fig. 8 A, a) and b)). In addition, the induction of intracellular responses by controlled environmental changes shows the distribution in an isogenic population and duration of a specific cell response (Fig. 8 B, a) and b)). This gain attention to reveal the triggers of phenotypic cell 70 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum differentiation of isogenic cells as in our example of SOS response and induction of a genome-inherent prophage in C. glutamicum ATCC 13032.

5.1.2 Development of a Temporal Resolved Single-Cell Imaging Approach to Analyse Spontaneous Gene Induction in C. glutamicum

The sporadic induction of CGP3 and the spontaneous SOS response of C. glutamicum ATCC 13032 were analysed in parallel to determine a potential correlation. First, SOS response reporter strain construction and prophage induction reporter strain development were done in parallel prior implementation of both signals in the dual reporter strain that has been observed for signal correlation. We approached a dual reporter strain time-lapse analysis by stepwise integration of fluorophores into the bacterial host 22,30 and optimization of the subsequential microscopic imaging process 80. The bacterial SOS response is usually initiated if dsDNA breakage is induced by environmental stressors or occurs due to stalled replication forks. The RecA protein binds to single-stranded DNA (ssDNA) and turns to its activated form RecA*. Reduction of available unbound RecA activates the central control switch of the SOS gene cascade lexA that negatively controls the promotor PrecA (Fig. 9 A). SOS genes are highly conserved in bacteria and are activated when the replication 71 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum fork stops during replication due to double strand breakages or induced DNA damage and initiate a gene cascade that decide the fate of a specific cell 22,88,89. The SOS gene cascade is a well-controlled feedback loop mechanism shown schematic in Fig 9 A. SOS response sensor strains were constructed by fluorophore expression of e2- Crimson, eYFP, and Venus that have been set under the control of the recA promotor as depicted in Fig. 9 A 22,30,89,90. The fluorophores e2-Crimson (Fig. 9 A, i.), plasmid-based), eYFP (Fig. 9 A, ii.), chromosome-integrated), and Venus (Fig. 9 A, iii.), chromosome-integrated) to PrecA to visualize SOS response, respectively. EYFP and the brighter Venus were implemented and tested for dual reporter strains for visualization of SOS response and prophage induction. C. glutamicum ATCC 13032 harbours three prophages (CGP1 - 3) in the genome. The biggest of those, CGP3, comprises 6 % of the whole genome of its host 73. The about 187 kbp sized prophage DNA sequence is shown schematically in Fig. 9 B. The function of all prophage genes are recently under intensive studies 22,30,42,72,91. For reporter strains of CGP3 prophage induction, fluorophore fusion with alpA (fused to cfp), lys (fused to eyfp or venus), and int2 (fused to eyfp) has been tested with time lapse imaging. Due to rapid bleaching of CFP, time resolved imaging of alpA induction was not pursued with time-lapse imaging. The spontaneous prophage induction was

72 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum determined by e2-Crimson fluorescence under the control of a putative lyase (lysin) (Fig. 9 B, iv.) and a phage-type integrase (Fig. 9 B, v.).

Figure 9: Regulatory scheme of the SOS response, prophage CGP3, and a putative trigger dtxR of prophage induction. The analysed reporter strain variants that differ in their fluorophore fusion proteins are shown. (A) LexA is the central control switch of SOS cascade induction. LexA represses self-expression and RecA expression. RecA indicates dsDNA breakage and is the initial one of the SOS cascade genes. SOS response is indicated by fusion of (i.) e2-crimson (plasmid), (ii.) eyfp (chromosome), or (iii.) venus (chromosome) to the recA promotor. (B) The genes of 187 kbp long prophage CGP3 are given. For indication of prophage induction, the fluorophore e2-crimson was fused to (iv.) the promotor lys of a putative phage lyase (plasmid) or (v.) the promotor int2 (plasmid). (C) DtxR is a central switch of iron regulation that induces genes of iron storage and controls genes of iron uptake and another central regulation node (ripA) that controls expression of iron containing protein and the resuscitation protein factor rpf2.

73 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

The triggers of CGP3 prophage induction were under investigation since its lysogenic remaining in the host genome is involved in a beneficial selection of a weakened minority for population fitness. The influence of nutrient depletion, metabolic activity and the iron metabolism on prophage induction were tested. Thus, the central node of the iron metabolism dtxR was knocked out in a reporter strain of prophage induction. DtxR is a central switch of iron regulation that induces genes of iron storage and controls genes of iron uptake and another central regulation node (ripA) that controls expression of iron containing protein and the resuscitation protein factor rpf2 that controls resuscitation from the dormant state (Fig. 9 C) 92,93.

Figure 10: Radical oxygen species determination in a SOS reporter strain expressing e2-crimson under the control of PRecA. (A) Correlation of mean single- cell e2-crimson fluorescence (SOS response, red box) and mean single cell CALox fluorescence (ROS/phototoxicity, green box) under excitation with 6.3 % of total intensity. A minority of phototoxic stressed cells with SOS response (SOS+/ROS+) is marked yellow. (B) Micrographs show all colonies with ROS forming SOS+ cells indicating phototoxic stress (yellow circle). Other ROS+ cells remained SOS- as well as other SOS+ cells exhibited no ROS formation.

74 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

In an analytical approach, the spontaneous SOS response and the sporadic induction of the prophage were analysed individually and subsequently in dual reporter strains. All reporter strains used in this work are described in detail by Nanda et al. (2014) and Helfrich et al. (2015) 22,30. The light excitation parameters of imaging are given in Tab.1., chapter 4.2.1 of the material and methods section. The challenge of spontaneous SOS response and phage induction analysis of C. glutamicum ATCC 13032 by time-lapse imaging was to optimize light excitation and the imaging interval while avoiding any disturbance triggering the SOS response or phage induction by phototoxic stress. This could increase the ratio of SOS+ or phage+ cells artificially as both events were reported to be triggered by intensive light excitation 94–97. Phototoxic stress induces formation of reactive oxygen species (ROS) such as singlet oxygen, what causes light-dependent destruction of biomolecules like DNA, lipids, and proteins 76. Therefore, the absence of phototoxic stress triggering SOS response was verified by a developed intracellular, in vivo ROS measurement sensing method using the green fluorescent fluorogenic substrate DHCAM as sensor in the first constructed reporter strain, C. glutamicum ATCC

13032/pJC1-PrecA-e2-crimson. DHCAM is taken up by C. glutamicum cells, converted by intracellular esterases and oxidized by singlet oxygen radicals to green fluorescent CALox (as described in detail in chapter 5.3).

75 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

The maximal light intensity used for fluorescence excitation (~ 6 % of maximal intensity) rarely show ROS reaction with DHCAM in growing SOS reporter strains. The SOS response of a population minority (SOS+) and the cells that appeared to be in a photo-oxidative stress state (ROS+) showed to be non-correlated (Fig. 10 A). Only a small fraction of these minorities (n = 3 cells in total of several hundred cells) were SOS+/ROS+ (Fig. 10 A and B). Thus, phototoxicity was not verified as a significant trigger for spontaneous SOS response. Additionally, a critical parameter for count of positive cells was the threshold value of fluorescence that defined the lower limit of intracellular single-cell mean fluorescence of a cell that defined when and if a cell was considered as SOS+ or phage+. The ratio of positive cells at the lowest meaningful threshold are given in Fig. 11 for all tested reporter strains of Tab.1 (Material and Methods) that are shown schematically in Fig. 9 A and 9 B. A dynamic variation over time could be observed for all time-lapse experiments, whereas an increase of relative SOS+ and relative phage+ cells with the course of experimental time and increase of total cell number could be seen. The first approach of a SOS reporter strain tested in microfluidic cultivation was a reporter strain which, expressed plasmid-based e2-

36,71 + crimson under the control of PrecA . The ratio of SOS to total cell number remained almost constant with a broad threshold band between 700 – 900 AU under constant supply with CGXII + 4 % GLC.

76 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

A threshold criterium of 600 AU almost doubled the SOS+ cells indicated by the error bars (Fig. 11 A). In the following, SOS reporter strains were constructed with genome-intregrated fluorescence protein promotor fusion 22,30. The test of two different clones, both expressing eyfp PrecA controlled, showed a remarkable increase of spontaneous stress response after two hours of perfusion cultivation with CGXII + 4 % GLC (Fig. 11 B and C). While one clone showed initially increased SOS response, differences in relative SOS response were in the range of variation of a threshold decrease of only 1 AU. In a next step, the reporter strain of SOS response shown in Fig. 11 D and E was combined with a plasmid-based expression of e2-crimson under the control of the lyase promotor Plys (Fig. 11 D) or under the control of a phage-type integrase promotor Pint2 (Fig. 11 E) to indicate if the prophage CGP3 was sporadically induced 30. The correlation of SOS response and phage induction was analysed and shown in Nanda et al. (2014) for PInt2 and Plys in shaking flask cultivation analysed by FACS and found to be in correlation with SOS gene initiation, which was not verified to be true for one third of SOS positives by constant conditions of microfluidic cultivation combined with temporal resolved time-lapse imaging of individual cells and their daughter cells 22,30. However, the spontaneous SOS response is with both reporter strains higher than the phage induction with perfusion of CGXII + 4 % GLC at 30 °C that was considered as non-stressful standard cultivation

77 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum condition. Minor differences of both reporter strains are not remarkable since a threefold technical replicate of a reporter strain with enhanced fluorescence by expressed venus compared to the eyfp fused reporter showed intrinsic variation of SOS response and phage induction in ratio and in temporal distribution (Fig. 11 F – H). For reasons of low signal to background ratio of eYFP, it was replaced by Venus that is an advanced derivative of YFP due to faster maturation and reduced sensitivity to environmental intracellular influences 98. Sporadic prophage induction is not completely linked with spontaneous SOS response as proposed by A. Nanda previously 30. However, time-lapse imaging analysis revealed a permanent non- induced spontaneous SOS response and spontaneous CGP3 induction

+ was observed in a population minority (≥ 8 % SOS indicated by PrecA-

+ fluorophore fusion, ≥ 3 % phage visualized by Plys-e2-crimson fusion,

+ and ≥ 1 % phage shown by Pint2-e2-crimson fusion) in absence of external stress stimuli. Differences in fluorophore signalling due to expression and ripening was tested by control strains expressing venus or e2-crimson upon IPTG induction under control of the tac promoter (Fig. 12). Thus, different IPTG concentrations were tested for best continuous induction by perfusion (Fig. 12 A). For 100 µM IPTG a difference in signal response was found to be approximately 12 min (Fig. 12 B). Hence, simultaneous co-induction of spontaneous SOS response and

78 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum sporadic prophage expression should happen within three imaging time points (frames). Furthermore, the impact on imaging results by systematic error due to technical settings, e.g. light excitation, lamp aging, focus drifting and others, was determined with reference beads used as internal standard during fluorescence time-lapse imaging with a yfp filter set. Therefore, tailor-made green/yellow fluorescent beads were used as internal standards. These beads were made of PMMA (Dr. Thiele, PolyAN, Germany) with two different fluorescence intensities according to the cell fluorescence to avoid hot pixels during time-lapse imaging. One bead batch, the surrogate cell beads, was set to show resembling fluorescence as C. glutamicum ATCC 13032 cells expressing eYFP, the other bead batch, the reference beads, fluoresced ~ 10-fold more at conventional exposure time and light intensity as the surrogate beads. The beads self-attached to the supply channel walls, considered as biocompatible, and remained during cultivation mainly at the same position. Rarely, spatial drift reduced the measured mean bead fluorescence of single imaging positions. The mean bead fluorescence of both bead batches is shown at different exposure times in Fig. 13 A and the normalized mean bead fluorescence at conventional fluorescence intensity is shown for 103 beads over time in Fig. 13 B.

79 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 11: Relative SOS+ and relative phage+ cells over time determined with different fluorescent reporters. (A) SOS reporter strain expressing plasmid-based + e2-crimson under the control of PrecA (i.). Error bars indicate ratio of SOS at a threshold decrease of 300 AU (B) and (C) Result of a biological duplicate of two identic mutants. The SOS reporter strains both expressing genome-integrated eyfp + under the control of PrecA (ii.). Error bars indicate ratio of SOS at a threshold decrease of 1 AU. (D) Dual reporter strain indicating SOS response (green) and prophage induction (red). Eyfp is expressed under the control of PrecA (ii.) and e2-crimson is expressed plasmid-based under the control of Plys (iv.). Error bars indicate ratio of phage+ and SOS+ at a threshold decrease of 4 AU and 2 AU, respectively. (E) Dual reporter strain indicating SOS response (green) and prophage induction (red). Eyfp is expressed under the control of PrecA (ii.) and e2-crimson is expressed plasmid-based + + under the control of Pint2 (v.). Error bars indicate ratio of phage and SOS at a threshold decrease of 8 AU and 4 AU, (figure legend continues on next page▼)

80 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

(▲continued figure legend of Fig. 11) respectively. (F) – (H) The result of a technical triplicate with the same dual reporter strain indicating SOS response (green) and prophage induction (red) is given. Venus is expressed under the control of PrecA (iii.) and e2-crimson is expressed plasmid-based under the control of Plys (iv.). Error bars indicate ratio of phage+ and SOS+ at a threshold decrease of 11 AU for both.

Figure 12: Control experiments of fluorophore ripening using two control strains bearing a plasmid each with the IPTG inducible ptac promotor fused to venus or e2-crimson. (A) Resulted mean Venus flurescence (green), and mean e2-Crimson fluorescence (red) are given at different IPTG concentrations. (B) Mean fluorescence distribution and standard deviation at 100 µM IPTG are shown of Venus (green) and e2-Crimson (red) over time.

As fluorescence beads randomly attached in the main channel, the measured mean single-cell fluorescence of self-inducing cells that express venus under the control of the RecA promotor (Fig. 13 C) were corrected exemplarily for the systematic error due to illumination imprecision by a factor (small diagram in Fig. 13 D) of 7 beads imaged in the same image frames as the bacterial cells and normalized to their initial fluorescence (Fig. 13 D). The systematic influence on mean

81 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum single-cell fluorescence showed to be almost negligible but non-linear after 8 h. Additionally, the loss of fluorescence (photobleaching) per imaging frame was determined for all fluorophores at the excitation optima for the performed time-lapse single-cell analyses. The fluorescence loss due to photobleaching was 5.8 ± 0.2 % for Venus, 8.4 ± 0.5 % for eYFP, and 2.4 – 2.6 ± 0.1 % for e2-Crimson. Exemplarily, the bleaching correction of mean single-cell fluorescence was performed for the single-cell data shown in Fig. 13 D that had been corrected for the systematic fluorescence fluctuation previously and is shown in Fig. 13 E. The course of bleaching is given for mutants expressing their fluorophore upon IPTG induction (Fig. 13 F). After fluorophore expression and a mild fixation, the fluorophore determination was performed under constant light excitation. In combination, quantitative single-cell fluorescence time-lapse imaging of gene induction dynamics can be performed, because influence of excitation bias by illumination heterogeneity can be compensated using the tailor-made PMMA beads of constant fluorescence, if remaining in their initial position. The beads of considerable constant fluorescence were embedded in the same image frames as the observed cell population of variable fluorescence. Both were excited by light of an oscillating and aging UV lamp. A bleaching correction of every imaging snapshot was demonstrated using the determined mean loss of fluorescence due to illumination per frame.

82 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 13: Fluorescence validation and correction of systematic fluorescence fluctuation and photobleaching initiated by the microscopic light source. (A) Mean bead fluorescence of medium fluorescent beads (reference beads) and comparably fluorescent beads to intracellular fluorescence of chromosomal-integrated Venus expressed by the reporter strain C. glutamicum ATCC 13032::PrecA- venus/pJC1-Plys-e2-crimson at different exposure times (surrogate cell beads). (B) Normalized mean fluorescence of different beads distributed in the main channels of one microfluidic device and imaged over twice the experimental time of bacterial cultivations. (n = 103 beads) (figure legend continues on next page▼) 83 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

(▲continued figure legend of Fig. 13) (C) Mean single-cell venus fluorescence of the dual reporter strain (n = 725 cells). (D) Mean single cell venus fluorescence of the dual reporter after systematic fluorescence fluctuation correction by imaged beads. The determined correction factor over time is shown in the integrated diagram. (E) Mean single-cell venus florescence of the dual reporter strain after correction of the photobleaching due to light excitation. (F) Photobleaching over time of Venus (n = 101 cells), eYFP (n = 99 cells) and e2-Crimson (n = 100 cells).

This could be even improved by determination of bleaching in dependency of cell size and mean single-cell fluorescence per frame. An reliable trigger resulting in induction of SOS response or the prophage in majority of population would allow in future to calculate the promotor activity as increment of the increasing single-cell fluorescence over time normalized by cell size as described by Amir and Kobiler (2007) not only of the sporadic population minority under standard cultivation condition and nutrient deprivation conditions to perform the impressive quantification of gene expression and expression noise as described by Young et al. (2012) in future 53,62,95.

5.1.3 Spontaneous SOS Response and Sporadic Prophage Induction

Single-cell analysis by time-lapse imaging of SOS reporter strains as well as phage induction reporter strains in microfluidic cultures revealed development of phenotypic population minorities in C. glutamicum ATCC 13032 initiating spontaneous SOS response or/and phage induction in absence of an external trigger (Fig. 11 and

84 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 14: Phenotypes of SOS+ and phage+ cells of all constructed reporter strains. After switching on the SOS response, some SOS+ cells resuscitated and reinitialised cell growth and inherited the SOS+ state to their descendants or reduced the intracellular fluorescence over the generations. Another phenotype of stressed cell showed cell division inhibition, meanwhile their cells proceeded branched cell growth and elongation. C. glutamicum ATCC 13032/pJC1-PrecA-e2-crimson cells, that switched on SOS response (red fluorescent) (A) remained in SOS+ state after cell division, or (B) they were cell division inhibited, or (C) they stayed in a non-growing state. These phenotypes of stressed cells (figure legend continues on next page▼)

85 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

(▲continued figure legend of Fig. 14) were also found in the reporter strain + C. glutamicum ATCC 13032::PrecA-eyfp. SOS cells that built eYFP (yellow fluorescent) performed (D – F) resuscitation, (G) branched cell growth, and (H) growth inhibition. (I – J) The dual reporter stains C. glutamicum ATCC 13032::PrecA- eyfp/pJC1-Pint2-e2-crimson and C. glutamicum ATCC 13032::PrecA-eyfp/pJC1-Plys-e2- crimson (K, L, and M) revealed that cells switched on (I, K) SOS+, (K, M) SOS+/phage+, and (J, L) phage+, respectively. The dual reporter C. glutamicum ATCC 13032::PrecA-venus/pJC1-Plys-e2-crimson confirmed by Venus (yellow fluorescent) (N) dividing SOS+, (O) elongated and non-growing stressed cells. (O) Further, non- growing phage+ and SOS+/phage+ cells were observed.

Fig. 12). The observed phenotypes were comparable for all SOS reporter strains (Videos 1 – 6). Most of the population minority of SOS+ cells showed constant fluorescence and growth inhibition with a plasmid-based promotor-e2-crimson construct (Fig. 14 C), promotor::eypf fusion (Fig. 14 H, I, K, M), as well as promotor::venus fusion (Fig. 14 O). Rare SOS+ events showed cell division inhibition with ongoing branched cell elongation (Fig. 14 B, G, M, O) as the cell division involved gene divS is reported to be inhibited by SOS gene induction 30,99,100 (Video 3 and 6) or recovered after spontaneous SOS response and reinitiated cell division (Fig. 14 A, D, E, F, N) 22,30,96 (Videos 1, 2,

+ and 5). Phage cells did not perform resuscitation neither with Pint2-e2- crimson fusion (Fig. 14 J), nor with Plys-e2-crimson fusion (Fig. 14 K, L, M, O) (Videos 1 – 3). In comparison to FACS analyses, time-lapse microscopy revealed additionally to time-resolved ratios of population minorities the single-cell fates evolving over time. The single-cell fates, developed

86 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 15: Spatio-temporal resolution of SOS response and prophage induction distribution by lineage tree analysis. (A) Distributional tracks (red, yellow, blue and light blue) of descendants derived of four (figure legend continues on next page▼)

87 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

(▲continued figure legend of Fig. 15) initially seeded cells. (B) – (E) The micrographs show the colony development and subpopulation formation over time. (F – I) Individual linage trees of the presented colony. SOS response and prophage induction influence the number of descendants in the finally recorded filial generation and thus, lineage tree width and symmetry. (F) Lineage tree with SOS+ and SOS+/phage+ cells occurring with 8th generation. (G) Lineage tree with almost absence of any induction (single SOS+ cell in the last generation) and highest count of descendants in last frame. (H) SOS response happened in 2nd generation and was passed on to following generations with development of all phenotypes (SOS-/phage- , SOS+ with recovery, constant SOS+, SOS+/phage+, and phage+. (I) Lineage tree with a phage+ cell and late occurrence of SOS+ and SOS+/phage+ cells. in the dual reporter strain, were analysed by Helfrich et al. (2015) lineage tree-based in detail to quantify single-cell growth rates of SOS+ and their ratio of resuscitation, cell division, subsequent switch on of the phage+ state and cell lysis of phage+ cells 22,50. In Fig. 15, a colony that harboured four lineage trees is given exemplarily. Initially four cells were seeded in the growth chamber and their daughter cells evolved in the colony with only minor mixing as shown in Fig. 15 A. Additionally, the colony growth and phenotype development is depicted in Fig. 15 B – E. The four lineage trees developed differently, whereas those with stressed cells or phage+ siblings had a minor width than those without phenotype heterogeneity (Fig. 15 F – I). Although, the most positive events were counted here in the centre of the colony. A spatial correlation of SOS+ cells or phage+ cells could not be stated along the main channel in the microfluidic device (Fig. 16). Phage+ and SOS+ cells distributed randomly in the growth chamber array.

88 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 16: Spatial resolution of SOS response and prophage induction in the microfluidic device after 8.5 h. Green dots represent SOS+ cells and red dots represent phage+ cells.

5.1.4 Spontaneous SOS Response and Sporadic Phage Induction under Nutritive Stress

Nutrient starvation experiments with intermittent hunger phases were performed with both dual reporter strains constructed and described in detail by Nanda et al. (2014) and Helfrich et al. (2015) 22,30. Starvation stress leaded evolutionary to regulated networks in C. glutamicum ATCC 13032 to overcome hunger phases for survival of the species. Thus, we performed intermittent carbon supply experiments in microfluidic devices and analysed the impact on SOS response and

89 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum phage induction in C. glutamicum ATCC 13032 (Fig. 17 A, and additionally Fig. S1 given in chapter 8.1). An interruption of the carbon supply (CGXII – procatechuate) for 10 h did not lead to a remarkable decrease of SOS response and no phage induction before the back switch to CGXII + 4 % GLC (n = 10 colonies, Fig. S1, see chapter 8.1) since C. glutamicum is reported to perform remarkable carbon storage using glycogen formation 101,102. An extended hunger phase of 24 h revealed a remarkable extent of intermittent SOS+ cells with flickering venus fluorescence (n = 25 colonies). Thus, in difference to reference colonies with constant carbon supply (n = 20 colonies), the number of total SOS+ cells consisted cells with intermittent SOS response and cells with constant SOS response, which showed permanently increased venus fluorescence. During carbon starvation, constant SOS+ cells stagnated in difference to intermittent SOS+ cells (Video 7). After re-supply of carbon, the increase of constant SOS+ cells were comparable to that of SOS+ cells of reference colonies. An evolving colony is depicted in Fig. 17 B – D to visualize the adaption of the population to the environmental changes induced by the medium supply. At the time point of medium change to carbon- free condition, previously non-starved cells were mostly non- fluorescent despite a single SOS+ cell (Fig. 17 B). After adapting to hunger condition by reduction of cell size and inhibited cell division, rare SOS+ and phage+ cells appeared (Fig. 17 C).

90 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 17: SOS response and prophage induction distribution of C. glutamicum ATCC 13032::PrecA-venus/pJC1-Plys-e2-crimson under intermittent hunger condition. (n = 25 colonies with 24 h intermittent hunger condition, n = 10 colonies for reference and 10 h intermittent hunger phase) (A) Micrographs of a colony (B) at the beginning of the hunger phase, (C) at the end of hunger phase, and (D) after re- induced exponential cell growth under full nutrient supply.

After 24 h hunger phase, the re-supply of CGXII + 4 % GLC re- induced cell growth of the colonies and re-initialized the phage induction, which was stopped after ~ 13.5 h, and a comparable increase of phage+ cells per time as for the undisturbed reference colonies was reached after ~ 4 h of carbon re-supply (Fig. 17 D). This

91 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum re-initialisation of phage induction after an extended hunger phase in correlation with re-growth and cell division gave a hint to a relation of phage induction to energy metabolism or the divisome. Thus, the metabolic activity was determined instantaneously with SOS response and prophage induction during extended phases of nutrient deprivation by addition of violet fluoregenic CvAM. The conversion of CvAM to fluorescent CALv and the subsequent comparison of intracellular and extracellular mean CALv fluorescence is reported for non-invasive metabolic activity sensing of living C. glutamicum cells as shown by Krämer et al. (2015) and in chapter 5.4 16. Hence, the differentiation of phenotypical heterogeneity in an isogenic colony is shown that was cultivated for 6.3 h in minimal medium CGXII + 4 % GLC with addition CvAM followed by a continuous limitation of the carbon source (Fig. 18). The development of the single phenotypes is shown dynamically and correlatively by the single-cell fluorescence profiles over time (Fig. 18 A and B). The spontaneous cell differentiation had its origin in a single initial ancestor cell (Fig. 18 C). The course of mean single cell fluorescence is shown for eYFP (green), e2-Crimson (red), and CALv (blue) exemplarily for one colony with a nutrient starvation phase without carbon (Fig. 19 A), without iron (Fig. 19 B), without nitrogen (Fig. 19 C), and without phosphate (Fig. 19 D). The micrograph of these colonies is shown in Fig. 20. The fluorophore fluorescence of eYFP and e2-Crimson was

92 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 18: Physiological live cell analysis of C. glutamicum ATCC13032::PrecA- eyfp/pJC1-Plys-e2-crimson. Initially, the colony had been supplied with CGXII + 4 % GLC. After 6.3 h, the cultivation conditions were changed to carbon starvation. (A) Micrographs show subpopulation formation in an isogenic colony by expression of the fluorescence proteins eypf (SOS response, yellow fluorescent cell 1 and 2), e2- crimson (phage induction, red florescent cell 2), and the additional enzymatic conversion of the fluorogenic substrate calcein acetoxymethyl ester (CvAM) to the violet fluorescent calcein (CALv, ambient level: metabolically active cell, high level: dormant cell, blue fluorescent cell 3). (B) Intracellular fluorescence of the cells given in A, SOS+ cell: green line, phage+ cell: red line, dormant cell blue line), their ancestor cell (black line), and sibling cells (grey lines). (C) Lineage tree of the single cells given in A and B.80

93 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum significantly lower than the fluorochrome fluorescence of CALv (Fig. 18 B). Mean single-cell CALv fluorescence maximum was ~ 5.5 fold and ~ 65 fold higher than the fluorescence maximum of plasmid- integrated expressed e2-Crimson and chromosome-integrated expressed eYFP, respectively. Although, the mean single-cell CALv fluorescence was maximal under carbon depletion (Fig. 19 and 20), the switch to nutrient depleted medium was directly followed by an immediate change in intracellular CALv for all four lacking media components. Whereas lack of carbon (Fig. 19 A, Fig. 20 A, E, I) as well as absence of iron (Fig. 19 B, Fig. 20 B, F, J) resulted in a positive impulse on intracellular CALv fluorescence, a switch to perfusion with CGXII – nitrogen (Fig. 19 C, Fig. 20 C, G, K) and CGXII – phosphate (Fig. 19 D, Fig. 20 D, H, L) was followed by decrease of intracellular CALv fluorescence. Immediately before nutrient depletion, all cells showed CALv basis fluorescence under all four starvation conditions (Fig. 19 and Fig. 20 A – D). However, nitrogen depletion reduced extracellular CALv fluorescence (Fig. 19 C and Fig 20 G and K). During the extended starvation phase, mean single-cell fluorescence was remarkably increased without carbon and slightly increased without iron (Fig. 20 E, F, I, J). The efflux of CALv was significantly reduced after ~ 10 h absence of carbon in comparison to lack of iron, nitrogen or phosphate (Fig. 20 E – H). After 25 h cultivation in the microfluidic device with a starvation phase of ~ 19 h, carbon depleted cells exhibited highest

94 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum intracellular fluorescence at meanwhile lowest extracellular CALv fluorescence due to product efflux inhibition (Fig. 20 I) in comparison to carbon supply combined with depletion of iron (Fig. 20 J), nitrogen (Fig. 20 K), or phosphate (Fig. 20 L).

Figure 19: Temporal resolved scatter plot of mean single-cell fluorescence of SOS response, phage induction, and metabolic activity under nutrient depletion in a C. glutamicum ATCC13032::PrecA-eyfp/pJC1-Plys-e2-crimson colony. Fluorescence of eYFP (green), e2-Crimson (red), and CALv (blue) in absence of (A) carbon, (B) iron, (C) nitrogen, and (D) phosphate after full nutrient supply for ~ 6.5 h.

Lack of iron and phosphate impaired cell division, whereas cell division slowed down in absence of carbon significantly, and cell

95 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum growth stopped completely in CGXII – nitrogen (Fig. 21 A). CvAM conversion was observed under all nutrient deprivation conditions, whereby a single-cell fluorescence increase of the hydrolysis product CALv stated presence of esterase activity and an increase of extracellular CALv fluorescence, which indicated the energy- dependent CALv efflux performed by a putative ATPase 16. The different impact on growth by deprivation of carbon, iron, nitrogen, and phosphate attended a different influence on C. glutamicum metabolism, SOS response, and phage induction. The metabolic activity of SOS+ cells (Fig. 21 B), phage+ cells (Fig. 21 C), and SOS+/phage+ cells (Fig. 21 D) was indicated by their single-cell CALv fluorescence. Cells limited of iron, nitrogen, and phosphate, that expressed a SOS+ response, showed comparably low metabolic activity, whereas carbon starved SOS+ cells strongly varied in CALv fluorescence (Fig. 21 B). Cells with very low CALv fluorescence, that were considered as dead, constant increased CALv fluorescence, that were defined as dormant, and cells of intermittent increased single-cell CALv fluorescence, that were supposed to change from non-growing state to dead (Fig. 21 B). Phage+ and SOS+/phage+ cells exhibited low single-cell CALv fluorescence (dead), constant moderate single-cell CALv fluorescence, and intermittent high single-cell CALv fluorescence of changing metabolic activity.

96 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 20: Micrographs of the dual reporter strain C. glutamicum ATCC 13032::PrecA-eyfp/pJC1-Plys-e2-crimson colonies and a representative section of the supply channel under nutrient deprivation and CvAM addition. Red fluorescent cells (e2-Crimson) indicate phage+ cells, yellow fluorescent cells (eYFP) indicate SOS+ cells and blue fluorescent cells (CALv) indicate cells with increased intracellular CALv fluorescence after CvAM conversion and reduced ATP-dependent CALv efflux. A qualitative impression on ATP-dependent CALv secretion is given by the extracellular fluorescence shown in the micrographs wihin every panel on the left. These micrographs depict a representative section of the main channel at the same position and LUT settings. A colony and a section of the main channel is shown at the beginning of nutrient deprivation (6.6 h) (A – E), for almost 10 h nutrient deprivation (16.0 h) (F – H), and at the end of cultivation after more than 18 h of nutrient deprivation (25 h) (I – L) (A, E, I) without glucose, (B, F, J) without iron, (C, G, K) without nitrogen, and (D, H, L) without phosphate.

The single-cell e2-Crimson fluorescence and the single-cell eYFP fluorescence of all non-lysed carbon-starved cells with a single-cell

97 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

CALv fluorescence threshold of 500 AU (Fig. 21 E). Notably, carbon- starved cells with a single-cell CALv fluorescence greater 1000 AU (Fig. 21 F) revealed despite of the already defined states the categories non-growing but metabolically (still) active cells in non-induced SOS+/phage+ state, in phage+ state, in SOS+ state, in SOS+ state changing to SOS+/phage+, and in SOS+ state with moderate e2- crimson fluorescence (~ 550 AU). The relative SOS response (Fig. 22, green line), the relative phage induction (Fig. 22, red line) and the ratio of cells that expressed both fluorescence signals (Fig. 22, orange line) are shown for all four experimental nutrient limitation conditions. Therefore, the induction of CGP3 is considered to be energy-dependent, since carbon starvation resulted an almost constant high relative phage+ state (Fig. 22 A), iron showed a decreasing relative phage+ state (Fig. 22 B), nitrogen depletion reached a constant low relative phage+ state (Fig. 22 C), and phosphate starvation exhibited the highest final relative phage+ state with increasing trend (Fig. 22 D). Almost similar results were observed for relative SOS+ state and relative SOS+/phage+ state. Iron deprivation triggered at initial nutrient depletion phase and at the end of experiment a bimodal relative SOS+ maximum. The phage+ cells seemed to be overgrown during iron starvation.

98 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 21: Influence of nutrient depletion on growth and metabolic fitness of C. glutamicum ATCC13032::PrecA-eyfp/pJC1-Plys-e2-crimson. (n = 10 colonies for each condition) (A) Cell growth before and after nutrient depletion. (B) Temporal resolved scatter plot of the mean single-cell CALv fluorescence of SOS+ cells under depletion of carbon (red), iron (green), (figure legend continues on next page▼)

99 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

(▲continued figure legend of Fig. 21) nitrogen (lilac), or phosphate (orange). (C) Temporal resolved scatter plot of the mean single-cell CALv fluorescence of phage+ cells under depletion of carbon (red), iron (green), nitrogen (lilac), or phosphate (orange). (D) Temporal resolved scatter plot of the mean single-cell CALv fluorescence of SOS+ and phage+ cells under depletion of carbon (red), iron (green), nitrogen (lilac), or phosphate (orange). (E) Correlation of mean single-cell e2- Crimson fluorescence and mean single-cell eYFP fluorescence of all carbon depleted cells. (F) Correlation of mean single-cell e2-Crimson fluorescence and mean single- cell eYFP fluorescence of carbon depleted cells with a mean single-cell CALv fluorescence greater than 1000 AU.

Additionally, the impact of iron homeostasis was further investigated by another microfluidic analysis approach with stationary cells cultivated for 3 days in CGXII + 4 GLC with 36 µM iron or without iron. The behaviour of C. glutamicum ATCC 13032 and a wild type mutant with knockout of the central control switch gene of iron homeostasis, dtxR, was analysed. WT recovered only in minimal medium with 36 µM iron (Fig. 23 A). In three experiments, with 90 chambers imaged in total, no regrowth of stationary cells could be initiated in medium lacking iron in microfluidic perfusion culture. In spite of the WT (Fig. 23 A, E), WT ΔdtxR resuscitated in minimal medium CGXII with iron and in the absence of iron with an elongated phenotype (Fig. 23 A, F, G). In addition, the WT ΔdtxR mutant showed a partly cell division inhibited, elongated phenotype that exhibited bifurcated cell shapes, asymmetric cell division, maldistribution of fluorescence, frequent cell lysis and lysis of cell halves.

100 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 22: Relative SOS response and relative phage induction of C. glutamicum ATCC13032::PrecA-eyfp/pJC1-Plys-e2-crimson under nutrient depletion. (n = 10 colonies for each condition) SOS+ cells (green line), relative phage+ cells (red line), and relative SOS+ and phage+ cells (orange line) (A) under carbon starvation, (B) under iron depletion, (C) under nitrogen starvation, or (D) under phosphate depletion.

The relative phage induction compared to the total cell number is shown in Fig. 23 B – D, which was observed to be remarkably higher for all three cases with regrown stationary cells than with the previous nutrient depletion experiments with seeded cells of shaking flask cultures with CGXII + 4 % GLC in early exponential phase. The ratio of phage+ cells increased with raising cell density in the colony of WT (Fig. 18 B), while the ratio of phage+ cells among WT ΔdtxR

101 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

Figure 23: Influence of iron homeostasis on prophage induction. C. glutamicum 13032 WT and the knockout mutant WT ΔdtxR after three days of iron starvation during pre-cultivation in shaking flask cultures with CGXII + 4 % GLC containing 36 µM iron or no iron. (A) Growth of WT and ΔdtxR mutant in microfluidic cultivation with iron and without iron. (B) – (D) Relative phage+ cells in five microcolonies of WT resuscitated with iron, ΔdtxR mutant resuscitated with and without iron. Micrographs of (E) WT cultivated with iron, (F) ΔdtxR mutant cultivated with iron, and (G) ΔdtxR mutant perfused with minimal medium CGXII lacking iron after 14 h. populations was initially very high, decreased to a non-detectable level before increasing to stagnation plateau after 25 h with iron and without 102 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum iron. Under a lack of iron (Fig. 23 C), the relative phage+ cell level nearly doubled in comparison to WT ΔdtxR cells grown in CGXII + 4 % GLC (Fig. 23 C).

5.1.5 Discussion and Conclusions

Fluorescence time-lapse imaging has been shown to be an excellent and precise tool to analyse and even quantify expression dynamics of stressed population minorities of living bacteria 53,95,103, especially if it is combined with environmental control as it is given in microfluidic devices 53,62,69. In difference to FACS, that help to analyse population fractions of SOS+ cells at different cultivation time points and overall regrowth after sampling with the help of several samples 30,96, single- cell time-lapse imaging in microfluidic devices reveals the individual cell fate and help to determine impact on subsequent generations of distinct SOS+ cells 22. The multiplexed fluorescence time-lapse imaging method has been developed with a comparative, iterative approach using different SOS and/or phage C. glutamicum ATCC 13032 strains provided by coworkers as stated in the material and methods section and non-invasive integration of fluorogenic substrates for indication of metabolic activity and presence of phototoxicity. The importance to establish a fluorescence threshold individually to the specific fluorophore fusion protein construct of the reporter has been shown for the tested SOS 103 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum response reporter strains, the prophage induction reporter stains, and the dual reporter strains. The evaluation of reporter fluorescence delay of venus (SOS response) and e2-crimson (phage induction) has been determined to be approximately 12 min (Fig. 12). Induction of sporadic prophage induction or genetic stress response due to phototoxicity could be denied by development of an imaging in vivo method of ROS determination with the fluorogenic substrate DHCAM. An advancement of the time-lapse imaging of SOS response visualized by venus expression has been achieved for future quantitative expression dynamic analysis of single-cells. Therefore, systematic fluorescence bias was followed by tailor-made PMMA fluorescent reference beads ordered upon demand by PolyAN and photobleaching was determined for all fluorophores used for dual reporter constructions for demonstration of compensation calculations. If a direct trigger of prophage gene induction is found, the promotor activities could be determined of a significant cell number for quantitative expression dynamic analysis in the future. Time-lapse imaging of C. glutamicum ATCC 13032 cultivated under constant conditions in microfluidic devices revealed that spontaneous SOS response and prophage induction were switched on independently in absence of extracellular stimuli in a population minority as previously described 22,36. Although, a significant fraction of SOS+ cells were additionally phage+, the phage+ state always ended

104 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum in a non-growing state that stopped cell division or recovery of the cell. SOS response in C. glutamicum ATCC 13032 was reported to end in a non-growing state or in cell recovery as it was previously stated as SOS recovery by DNA repair or pulsing SOS response in E. coli 96, in B. subtilis 53 and in C. glutamicum ATCC 13032 22,30. Several cells remained in SOS+/non-dividing state with branched cell elongation as reported by SOS-induced gene activation of divS 30,100. Non-growing cells that spontaneously induced SOS response were tested by others to exclude the cell death indicator propidium iodide (PI) and defined as senescence-like SOS+ cells 96,104. An accidental trigger of SOS stress genes or prophage induction by time-lapse imaging excitation could be excluded by in vivo ROS measurement. Although, spontaneous SOS response is followed remarkably often by phage induction and cell division arrest, a distinct trigger of the prophage element could not be defined so far that allows also SOS-independent prophage induction 22. Thus, the impact of CGP3 on its host fitness is still not cleared in detail and how its genes survive evolutionary selection. Knockout mutants without CGP3 are reported to show no special phenotype 42. The role of remaining prophage elements such as CGP3 in C. glutamicum ATCC 13032 and their influence on host fitness are of interest since it’s reported that bacteriophages are capable to control pathogenesis and metabolism in corynebacteria 105.

105 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

The SOS-independent prophage induction is under further investigation in the future. Although, the prophage genes of CGP3 are under intensive examination, a prophage inducer is not defined, yet 22,30,42,72,73,91. Hence, the prophage CGP3 might be an altruistic intracellular control of mutant formation due to SOS-induced DNA repair mechanisms that went wrong, to control population fitness under environmental nutrient limitations and energy-depletion conditions. The nucleotide-associated protein CgpS was reported to possess xenogenetic silencer function to CGP3 genes, which could harbour an internal trigger of altruistic activation of lethal prophage induction 72. The role of prophage elements and remaining bacteriophages in bacterial genomes are mostly not fully understood, although, they are very abundant 106. In difference, environmentally triggered SOS gene-associated phage induction is reported of lambdoid phages in E. coli 107,108. In addition, a beneficial behaviour of phage elements in E. coli is reported for λ to protect the host from other bacteriophages 109. Furthermore, SOS genes are known to induce lethal toxin-antitoxin module function in E. coli 104,110 and they trigger SOS response related mutagenesis for environmental stress adaption 88,111–113. C. glutamicum ATCC 13032 and a prophage-cleared mutant grown under phosphate starvation (0.65 mM), iron limitation (0 mM), or with different gluconeoglycogene substrates exhibited no growth difference under batch cultivation conditions 42. Similar nutrient

106 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum depletions of carbon, iron, phosphate and in addition nitrogen under extended microfluidic perfusion conditions showed indirect influence of iron and phosphate. After a cultivation phase with full nutrient- containing minimal medium CGXII + 4 % GLC, hunger phase was initiated by a rapid medium switch. Nitrogen supply is elementary for cell metabolism, protein synthesis, maintenance and cell division 114. Therefore, nitrogen limitation is thought to suppress the upregulation of intracellular esterases. The prior to nitrogen limitation built esterases are considered to remain catalytic activity at least until cell decay. In addition, fluorophore and protein expression is inhibited under nitrogen starvation. Thus, nitrogen limitation showed only marginal metabolic activity, SOS response and phage induction as expected. The absence of glucose reduced the ATP level and was supposed to be compensated for several hours by the storage molecule glycogen 115. Glycogen is reported as carbon source of C. glutamcium to overcome growth periods of carbon limitation and was considered to influence the beginning of carbon starvation phases 102,116,117. Further, phosphate limitation was applied, which is essential for DNA synthesis, membrane formation, phosphorylation reactions, sugar transport and highly uptake regulated under limitation conditions 114,118,119. Furthermore, phosphate metabolism and polyphosphate was reported to trigger lytic phage induction in bacterial populations 120, and it is found to induce SOS response in mycobacteria 121.

107 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

However, Donovan et al. (2015) have reported nucleotide hydrolysis by the CGP3 protein AlpC in vitro with a preference to ATP in comparison to GTP and proofed the protein to be involved in egoistic phage transport to the cell membrane 91. This may lead to the hypothesis that C. glutamicum ATCC 13032 missuses the prophage induction as an apoptotic-like altruistic cell death pathway to remove ATP poor, damaged, and stressed cells in the population with a lytic strategy. During nutrient starvation periods, rare nutrition elements would be released, since phage remnant transport would be inhibited due to a lack of required nucleotides or inhibition of phosphate- dependent nucleotide recycling. Li et al. (2007) mentioned increased induction of a temperate λ phage in the absence of polyphosphate building polyphosphate kinase 1, higher sensitivity to external stressors, that could explain the SOS+/phage+ state, and a phage promotor inhibition by guanosine 3’, 5’- bidiphosphate (ppGpp) of E. coli 120. The induction of phage induction of C. glutamicum ATCC 13032 in absence of iron and with the C. glutamicum ATCC 13032 knockout mutant without the mayor iron uptake regulatior DtxR under extended pre-cultivation in stationary phase with and without iron revealed no direct prophage induction mechanism. Iron limitation is known to be an essential transition molecule in oxygen transfer and catalytic processes 92,114,115. Iron homeostasis is controlled by the regulator DtxR to control a non-toxic intracellular iron concentration 122,123.

108 Single-Cell Analysis of Rare Spontaneous Cell Responses of C. glutamicum

C. glutamicum ATCC 13032 ΔdtxR mutant has shown increased relative phage+ induction as upregulation of CGP3 genes has been reported previously of this mutant strain 73. Further, dtxR is known to be involved in corynephage associated toxin production in the relative strain C. diphteriae 114. Hence, an indirect phage induction mechanism or a more complex role of the genetic regulation node dtxR might be given in C. gltuamicum, which has not been revealed, yet.

109 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

5.2 Artificial Fluorogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

The following text has been published as Krämer, C, Wiechert, W, Kohlheyer, D (2016): J Flow Chem 6(1): 3-7.D. Kohlheyer and W. Wiechert proofread the manuscript.

Chapter Abstract: Providing new fluorogenic substrates with designed enzyme-labile moieties for microfluidic live cell analysis is an innovative complementary approach to conventional cultivation based methods for bacterial diagnostics. The advance of their integrated application in microfluidic devices is presented in comparison to established approaches. A comprehensive insight on recent implementation is given and highlighted with a commercially available example.

5.2.1 Introduction

Fluorogenic substrates are non-fluorescent due to quenching moieties until those are cleaved off enzymatically. Those artificial substrates can be generated by conventional click chemistry 124 or are commercially provided in a broad range. Fluorogenic substrates are

110 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics enzymatically converted to fluorescent products and can be designed according to distinguishing metabolic features of pathogens or bacterial classes. Appearance of fluorescence reveals information about uptake, conversion and efflux of those chemicals and their products. Their innovative implementation in microfluidic analytics of microorganisms was demonstrated recently 16,125. Novel applications with those chemical compounds for bacterial detection 126–128, metabolic characterization 125,129,130, drug development 131, pre-clinical studies 132, and screening for antibiotic resistance 133 and antibiotic tolerance 17 are reported. Important scaffolds are from the fluorochrome family of xanthene dyes as fluorescein and rhodamine derivatives or the family of coumarins as 4-methylumbelliferone, resorufin, 7-hydroxycoumarin as reviewed elsewhere 134. Also other fluorescent scaffolds as, e.g. oxy-luciferin 132 or artificial fluorochromes as Cy3, or Cy5.5 133. The application of commercially available fluorogenic or cholorogenic substrates and their conversion in conventional bacterial diagnostics is typically performed by time consuming and cumbersome media or gel based assays. These methods enable detection and characterization of microorganisms by their characteristic enzyme activities only after days of incubation and if colony growth is given 135. Unfortunately, not all bacteria of relevance are in a cultureable state.

111 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Selective media for microbiological diagnostics relying on biocatalytical conversion are described in literature and they are commercially available since a long time 136. Unfortunately, extended cultivation and incubation times often lead to long delay times of days up to weeks until data evaluation can be performed 137. The results of agar plate count methods are typically a whether or not finding, depending on a positive or negative conversion of the selective media components after days of incubation. Therefore, for every sample, many agar plates have to be prepared, plated and analysed one after another due to mandatory serial dilutions to generate separated superficial colony forming units of bacterial suspensions with unknown cell number (Fig. 24 A). Especially pharmaceutical and clinical diagnostics depend on reliable high throughput measurements for fast and precise analysis of high sample loads. For this purpose, fluorescence measurements with automated plate readers 138 or flow cytometry (FCM) 139 have improved sample throughput (Fig. 24 B and D). Further, microtiter well plates facilitate parallelisation and automation of sample analysis by the use of plate readers and robotic liquid handling stations. The fluorogenic substrate is provided in the liquid medium of the cavities and fluorescence measurements can be performed dynamically with different substrate concentrations or as endpoint analysis of several bacteria isolates.

112 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

FCM allows snapshot analysis of single bacterial cells and their viability or cell functionality 84,137. However, bacteria have to be cultivated, sampled and intracellular fluorescence has to be generated by genetic modification for introduction of fluorophore expression or bioluminescence, by labelling with fluorescent dyes or previous cultivation in shaking cultures with fluorogenic substrates prior analysis. Nevertheless, FCM still requires offline sample preparation and storage before results can be acquired. Obviously, sample preparation and delay in analysis can influence measurements, especially in a dynamic enzymatic reaction system prevailing in living bacteria. Although, the analysis mode is fast and automated, samples have to be analysed successively and temporal resolution is very limited. A great advantage of FCM is the analysis of heterogeneities in bacterial populations. These single-cell resolved analyses can also be performed in microfluidic cultivation devices (Fig. 24 C), which integrate cultivation and spatio-temporal analyses. Further, microfluidic devices offer the opportunities of volume reduction, reduction of expensive reaction reagents as fluorogenic substrates, parallelization of sample measurements and flexibility in device design. The trapping of single cells is sufficient without the requirement of high cell numbers as it is the case for plate count methods, plate reader based assays, FCM analysis or gel based diagnostics.

113 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Figure 24: Fast, reliable and high throughput bacterial detection for putative pathogens and characterization of production strains is of high relevance for clinical diagnostic, pharmaceutical drug development and biotechnological production. (A) Conventional bacterial diagnostics utilizing fluorogenic substrates is still mostly a two-step approach separating conventional plate cultivation and subsequent analysis. (B) Miniaturization and parallelisation is realized in multi well plate assays without single-cell resolution. (C) Microfluidic cultivation with a continuous supply of fluorogenic substrates can integrate and parallelize sample analyses. In combination with time-lapse microscopy, microfluidic cultivation approaches facilitate spatial and temporal resolution of bioconversion at the single cell level. Thus, comparable low cell numbers and short cultivation phases suffice for bacterial analyses. (D) Flow cytometry facilitates fast single cell snap shot analysis at tremendous throughput. However, conventional pre-cultivation and sample preparation steps are necessary.

Therefore, we would like to draw the attention on recent advances on synthesis and implementation of fluorogenic substrates for analysis of native bacterial enzymes of crude extracts or living cells. The

114 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics importance of non-toxic fluorogenic compounds is shown with a commercially available calcein acetoxymethyl ester derivative.

5.2.2 Relevance of Novel Nontoxic Fluorogenic Substrates and Recent Application Approaches

The synthesis of fluorogenic substrates and their application for conversion by specific enzymes of pathogens and non-pathogenic bacteria is of interest to detect bacteria prior to drug therapy of patients 127, to define pathogenicity 129 or to find possible drug targets to tackle bacterial resistance by screening of lead component libraries of production strains 124. We summarized recent analytical approaches using fluorogenic substrates in Tab. 2 that have the potential to be advanced to microfluidic bacterial diagnostics methods or already using dynamic analysis systems. Thus, microfluidic approaches will gain importance in future due to their advances of sample analyses with spatio-temporal resolution and the potential of sample free measurement automation with lowered reagent usage. However, prior synthesis of fluorogenic compounds with enzyme-labile moieties is required before bacterial diagnostics of enzymatic activities can be established. A recent application of previously synthesised artificial β-lactamase substrates containing green fluorescent Tokyo Green has been presented for Mycobacterium tuberculosis detection in microfluidic droplets 128. 115 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Table 2: Selection of fluorogenic substrates and their use in bacterial diagnostics in static and perfused analysis systems Fluorogenic Enzyme activity Analysis system Reference substrate

7-hydroxy-3h- metabolic phenoxazin-3-one activity of spectrophotometer Ishiguro et al. 10-odxide periodontopathic (2015) 126 (resazurin) bacteria 4- cell wall- methylumbelliferyl anchored well plate reader Lun and butyrate, carboxylesterase Bishai (2007) 4- (M. tuberculosis) 129 methylumpelliferyl heptanoate,

4- methylumbelliferyl oleate N-acetyl tripeptide- protease aminomethylcoum (M. tuberculosis) well plate reader Akopian et al.

131

Static analysis systems analysis Static arins (2015) well plate reader and 5-nitrofuryl caged bacterial pre-clinical in vivo Vorobyeva et luciferin nitroreducta-ses imaging of mice al. (2015) 132 7-hydroxy-9h-(1,3- dichloro-9,9- mycobacterial native PAGE Tallman and dimethylacridin-2- esterases Beatty (2015) one- 130 acetoxymethylester resorufin- acetoxymethylester

116 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Fluorogenic Enzyme activity Analysis system Reference substrate

carboxyfluorescein diacetate and bacterial flow cytometry Hoefel et al. carboxyfluorescein esterases (2003) 127 diacetate succinimidyl ester fluorescein alkaline time-lapse imaging Shim et al. diphosphate phosphatase (2009) 125

LRBL1-3 (β- β-lactamase FRET based live lactamase- cell imaging and Shao et al. responsive flow cytometry (2013) 133 bacterial labelling) β-D-cellobioside- cellobiohydrolas microfluidic FCM Najah et al. 6,8-difluoro-7- e (2014) 140

Dynamic analysis systems analysis Dynamic hydroxycoumarin- 4-methanesulfonate CDG-1 and CDG- β-lactamase microfluidic FCM Lyu et al. OMe 2015 128 fluorescein lipases time-lapse imaging Hosokawa et dicapylate al. (2015) 141 CvAM bacterial time-lapse imaging Krämer et al. esterases (2015) 16 CvAM bacterial FACS Hendon-Dunn esterases et al. (2016) 87 CvAM bacterial FACS Hendon-Dunn esterases et al. (2018) 142

117 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Smart substrate design enables diagnostic concepts with coupled enzymatic reactions as shown for a caged luciferin that requires conversion by nitroreductases prior reaction with luciferase for metabolic characterization of living bacteria 132. Furthermore, enzymatic cleavage of quencher moieties can be used for analytical FRET biosensor construction 133. In addition, enzyme activity screening for fine chemical production bears a high motivation to synthesise tailored fluorogenic substrates as demonstrated by microfluidic micro droplet biotransformation of bacterial libraries with cellolytic and lipolytic enzyme activities 140,141. These hydrolases are of relevance for fine chemical production or biofuel production 140,141. Although, studies with fluorogenic substrates are still performed nowadays partly in static analysis systems without single-cell resolved results, they have a high potential to be advanced in combination with microfluidic devices for microbial cultivation 16,17. The future potential of these microfluidic devices for in-flow bioanalysis can be demonstrated with calcein acetoxymethyl esters (CAM), which are used conventionally for non-invasive mammalian cell analysis 85,143. CAMs are quenched by covalently bound ester groups belonging to the fluorochrome family of xanthene and coumarin derivatives, respectively. After cellular uptake the ester groups are cleaved off the fluorescent calcein, followed by equimolar ethanol formation 85.

118 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

5.2.3 Conversion of a Calcein Acetoxymethyl Ester for in vivo Single-Cell Analysis

A disadvantage of conventional bacterial diagnostic methods is that non-growing bacteria are easily overseen. Dormancy of bacteria is characterized as a non-replicating state with reduced enzyme activities. Thus, methods relying on enzymatic activity have to be very sensitive towards low cell numbers and decreased substrate conversion. Notably, dormant bacteria are of relevance in pharmaceutical research and clinical treatment, because they bear an increased potential for antibiotic tolerance, hidden infections and bacterial persistence. Especially non-sporulating actinobacteria are reported to develop dormant phenotypes that can exist as latent (dormant) infections in patients without symptoms of the disease, endure medication, become drug resistant, and initiate acute infections in patients 144. Gengenbacher and Kaufmann (2012) summarized the important facts of dormancy related to Mycobacterium tuberculosis that is one of the most severe infection diseases of the world 144. Commercially available CvAM is highlighted here for bacterial diagnostics to give a perspective for future microfluidic approaches of microbiological assay performance. Experimental procedures are described in detail in Krämer et al. (2015) 16. CvAM concentrations from 23.1 µM to 138.9 µM were dissolved in the perfusion medium

119 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics of the microfluidic cultivation device, to test if increasing CvAM concentrations influence the cell growth of the organism C. glutamicum ATCC 13032. The microorganism is not only relevant for fine chemical production, but it is also a non-pathogenic relative of the human pathogen M. tuberculosis and other human diseases. The nontoxic, intracellular conversion of CvAM, commercially available as Cyto Calcein 450 or CellTraceTM calcein violet AM, by growing bacterial microcolonies of C. glutamicum ATCC 13032 is indicated by the exponentially increasing colony fluorescence within the first 5 h (Fig. 25 A). During the experiment, intermediate energy driven efflux of the well retained fluorochrome CALv was found to be inhibited by cultivation in the absence of glucose. A lack of carbon caused increasing mean single-cell fluorescence, whereas the mean colony fluorescence remained constant due to stagnation of total cell number. Especially dormant cells, which decrease their metabolic activity and have high relevance in pharmaceutical and clinical diagnostics, can be distinguished due to significantly increased single- cell fluorescence in comparison to the average cell (Fig. 25 B). Notably, the apparent conversion of CvAM by the growing C. glutamicum ATCC 13032 colony depended on the increasing total cell number (Fig. 26 A – B). According to differences in cell fitness, cell size, and cell cycle, individual variations in intracellular fluorescence increased especially during intermediate starvation stress without carbon feed and afterwards. The slope of the mean colony

120 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Figure 25: Fluorogenic substrate conversions for microbial single-cell analysis combined with microfluidic cultivation and time-lapse imaging. (A) Apparent conversion of CvAM by a microbial colony increases exponentially with growing cell numbers. Heterogeneity of intracellular fluorescence of single-cells can be observed due to fluctuations of metabolic activity of individuals or as reaction to environmental conditions (as shown in the micrographs). (B) An exemplary microcolony lineage of C. glutamicum ATCC 13032 grown on continuous supply of CGXII medium with an intermediate famine phase (3 h – 15 h) with medium lacking a carbon source. Despite the artificially introduced growth arrest by nutrient deprivation real-time mean single- cell fluorescence provided information about metabolic activity of every single-cell and heterogeneity in the colony. Reduction of metabolic activity due to a dormant state decreases CALv efflux. Dormant cells (highlighted in red) were detected by an increased coefficient of variation of their mean intracellular fluorescence in comparison to the mean fluorescence of all siblings in the colony. fluorescence due to exponential cell growth could be used to qualify cultivation conditions in comparison to the reference condition in

121 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics complex medium BHI at pH 7.0. The apparent mean single-cell reaction rate constant has been determined for apparent CvAM conversion from the slope of the mean colony fluorescence over time. This allows evaluating the physiologic impact of different cultivation conditions and media (Fig. 26 C – D).

Figure 26: Conversion of a fluorogenic substrate by C. glutamicum ATCC 13032. C. glutamicum ATCC 13032 was cultivated in a microfluidic cultivation device with microarrays of picoliter sized cultivation chambers. The bacteria were supplied with nutrients and dissolved CvAM by medium perfusion as described previously 16,36. (A) Apparent total colony fluorescence of bacterial colonies fed with 23.1 µM CvAM (black lines), 46.3 µM CvAM (violet lines), 92.6 µM CvAM (green lines), or 138.0 µM CvAM (red lines) supplemented to the perfusion medium (CGXII + 4 % GLC (w/v)). The intermediate phase of glucose free cultivation condition is indicated. For every CvAM concentration five colonies were analysed to determine if an increase of CvAM concentration reduces the mean colony cell number. (B) The corresponding mean total cell number of panel A is shown over time. (C) Apparent mean fluorescence of five C. glutamicum ATCC 13032 colonies cultivated with complex medium at different conditions. The reference cultivation condition was BHI medium at pH 7.0 without antibiotic addition (black line). Other conditions were: pH 7.4 (blue 122 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics line), pH 6.6 (red line), and with addition of 10 mg/mL chloramphenicol (green line) or 10 mg/mL ampicillin (orange line) for 1 h (as indicated), respectively. (D) The apparent reaction rate constant, derived from the slope of the mean colony fluorescence, was normalized to the reference conditions to validate the impact of stress conditions on the metabolic activity of the model organism C. glutamicum ATCC 13032 given in panel C.

Results can be exploited for drug-development, pre-clinical analyses and personalized medicine to treat infections with the appropriate drug in efficient concentration. The impact of a short (1 h) perfusion of bacterial colonies with 10 µg/mL chloramphenicol (CHL) or ampicillin (AMP) demonstrate, that the microfluidic approach gives results after several hours instead of waiting 1-3 days, revealing insight about remaining metabolically active bacteria and can be used for drug therapy optimization in future.

5.2.4 Conclusion and Outlook

The interdisciplinary approach of chemical design of new fluorogenic substrate moieties, their bioanalytical application, and engineering microfluidic cultivation devices has a broad perspective for microbiological characterization, biocatalytic activity screening and bacterial diagnostics by enzymatic reactions. The possibilities of analytical flow chemistry are not fully exploited by far for bacterial analyses, especially for low cell number diagnostics. However, enzyme activities of single mammalian cells have already been discovered with continuous or intermittent fluorogenic substrate feed

123 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics in microfluidic cultivation devices 85,143. A broad innovative use of fluorogenic substrates lies in approaches for rapid future bacterial diagnostics or for rapid enzyme activity screening of bacterial libraries in hours instead of days with the possibility of cell sorting integrated in microfluidic devices 141. Especially chemical synthesis of organic molecules, appropriate for bacterial uptake and relevant enzymatic activity specificity, has a great future perspective in flow biochemistry. The conversion of fluorogenic compounds during cultivation answers, if bacteria are viable under the condition of interest 16,126,132. The role of enzymes in virulence and bacterial survival is of high interest for basic research as well as future oriented therapeutic strategies 129,131,133. Enzyme expression in expression hosts bear the risk of accumulation of inclusion bodies and enzyme denaturation 129. Furthermore, the expression of cell wall anchored enzymes and their purification are quite sophisticated and can be circumvented by analysis of the unmodified microorganism in controlled microfluidic environments. The design and use of fluorogenic compounds with fluorescent scaffolds emitting red or blue to violet fluorescence 16,130 opens up the combined use of GFP expression after genetic modification of bacteria 145. These would give the possibility to determine protein interaction in vivo by multiplexing fluorescence measurements. Therefore, a variation of novel nontoxic fluorescent scaffolds with enzyme-labile moieties will be of rising interest in future for bacterial diagnostics.

124 Artificial Flurogenic Substrates in Microfluidic Devices for Bacterial Diagnostics

Therefore, nontoxic fluorescent scaffolds with emission variety of minor Stokes shift will be of rising interest in future. Microfluidic approaches will facilitate screening and diagnostic use of new potential fluorogenic substrates to replace conventional incubation until colony growth is given 146.

125 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

5.3 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamicum

Knockout strains C. glutamicum ATCC 13032 Δ ctaD and C. glutamicum ATCC 13032 Δ qcr were provided by A. Koch-Koerfges (Research group of Prof. Dr. Michael Bott).

Chapter Abstract: Bacteria are well-protected by their cell wall and cell membrane from diffusive intrusion of external chemicals. Therefore, the transport of therapeutic compounds or probe molecules into bacteria is challenging due to three major defense mechanisms: i.) Uptake of fluorogenic substrates as well as antibiotics can be blocked by size excluding diffusive barriers in the bacterial cell wall 147, ii.) bacteria can actively perform efflux of chemical compounds, e.g., with the help of ATPases or ion transport channels 148,149, iii.) bacteria can destroy taken up molecules by metabolizing these to untoxic products e.g. by cracking aromatic ring structures 150–152. Acetoxymethyl ester (AM) is an ideal molecule to shuttle complex aromatic molecules not only into mammalian cells, but also into actinobacteria as C. glutamicum ATCC 13032. This has been tested in this PhD thesis for multiplexed imaging with six fluorogenic probe molecules coupled to AM moieties to sense

126 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm metabolic activity, dormancy, cell lysis, chemotoxicity, phototoxicity, spontaneous radical formation, and intracellular pH of the model organism C. glutamicum ATCC 13032. The results of these single-cell studies proved successful sensing molecule uptake and subsequent intracellular enzymatic cleavage of the AM moieties. Uptake was affirmed by fluorescence of calcein, dihydrocalcein, pHrodo Red, and pHrodo Green in C. glutamicum ATCC 13032, although other bacterial species remained unfluorescent. These open up novel intracellular sensing strategies of specific molecules in actinobacteria, which have high relevance for pharmaceutical research and biotechnological production.

5.3.1 Introduction

Intracellular actions and reactions of living cells are challenging to sense in bacteria, especially on single-cell level. Mostly, phenotypic analyses, metabolomics and gene modification are used to perform comparative analyses of different experimental conditions or different mutants. Conventional probe integration for single-cell analyses of bacteria rely on genetically implemented expression of fluorescence or bioluminescence generating proteins. The Nobel prize winner Roger Y. Tsien is famous for his research according of GFP from the jellyfish Aequorea victoria and its derivatives, which are nowadays well-established in molecular 127 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm biology and live cell imaging 13,81. The successful implementation of fluorescent proteins in biological research almost overshadowed Roger Y. Tsien’s and his coworkers’ development of advanced fluorogenic Ca2+ indicators quenched by esterification with AM moieties for cellular uptake and intracellular measurements in living cells 153,154. The quenching of AM moieties alter charged fluorescence sensors to neutral charged molecules with minor lipophilic character 154. Intracellular carboxylesterases remove the quenching ester groups which result in trapped sensing molecules freely and unbound in the cytoplasm of the living cell 154. These properties were discovered for rapid transport of complex aromatic compounds into eukaryotic cells to enable intracellular molecule sensing (e.g. Ca2+) in the cytosol for fluorescence cell analyses 153,155–157. Commercially available esterified fluorogenic probes established for mammalian cells can be chosen from a broad catalogue. For applications with bacteria, applied analysis methods are almost absent in comparison to mammalian cell application of protocols using fluorogenic chemicals and cell analysis kits 28. However, FACS analysis of bacteria with esterified fluorescein derivatives have been shown in 77,127,139,158–160. For example, fluorescein diacetate (FDA) has been used for measurement of total microbial activity of soil samples 161,162. Carboxy fluorescein diacetate (CFDA) has been shown as viability indicator for FACS analysis of bacteria summarized subsequentially 127,158,160,162,163.

128 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Additionally, 5- (and 6-) sulfofluorescein diacetate (SFDA) was tested and compared with FDA and CFDA for microscopic detection of diverse microorganisms in soil in 162. 5-chloromethylfluorescein diacetate (CMFDA) was determined as indicator of the hydrolytic enzyme activity of planktonic bacteria 164. Carboxyfluorescein diacetate succinimidyl ester (CFDASE) was analysed as bacterial esterase activity indicator 127. Furthermore, microscopic viability assays with FDA and CFDA were shown for dental biofilms 165. Bis(carboxyethyl)-carboxyfluorescein-tetra acetoxymethyl ester (BCECF-AM) and CFDA-AM have been developed for improved intracellular retention compared to FDA 166. Although, detergents were used to improve cell permeability for cell staining, Jepras et al. (1995) reported inconsistent results with FACS due to insufficient dye loading of FDA, CFDA, and CFDA-AM in E. coli and P. aeruginosa 167. Additionally, insignificant staining was found with the fluorogenic esters CFDA, BCECF-AM, and calcein green AM (CgAM) for diverse bacteria and water sample microorganisms analysed by FACS 139,163. Thus, fluorogenic substrate uptake and conversion by living bacteria is highly specific and has to be tailor-made for the species of interest. In this PhD thesis, several analogues of CAM and an AM-coupled pH indicator were tested for multiplexed real-time imaging of the bacterium C. glutamicum ATCC 13032 that is reported to possess metabolic routes for transport and conversion of several aromatic

129 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm compounds 168–171. Direct measurement principles of intracellular changes in C. glutamicum ATCC 13032 are presented with different fluorescent molecules quenched by AM ester moieties and compared to literature. AM ester moieties enable the cell permeation of C. glutamicum ATCC 13032 cells as shown with CvAM in the next chapter 5.4 and by Krämer et al. (2015) 16. Therefore, dyes can be used that are developed for mammalian cells and tissue 156,166,172,173. The metabolic functionality, formation of oxygen radicals and the internal pH of living bacteria are of high relevance to understand metabolite production, basic equilibrium reactions, survival mechanisms, and stress response of microorganisms. Therefore, single-cell analyses were developed based on uptake and intracellular conversion of CgAM (Fig. 27 A, left side), blue fluorogenic calcein AM (CbAM) (Fig. 27 B, left side), violet fluorogenic calcein AM (CvAM) (unrevealed structure), green fluorogenic dihydrocalcein AM (DHCAM) (Fig. 27 C, left side), AM-coupled pHrodo Red (structure suggested by Ogawa et al. (2010) 156 shown in Fig. 27 D, left side) and pHrodo Green (unrevealed structure) to their corresponding fluorescent carboxylic acid form (Fig. 27 A - D, right side).

5.3.2 Bacterial Analyses with Calcein Derivatives

Uptake of different derivatives of calcein acetoxymethyl esters (CAM) have been tested with the facultative anaerobic actinobacterium 130 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Corynebacterium glutamicum that is reported to possess metabolic routes for transport and conversion of several aromatic compounds 168– 171. As part of this PhD thesis, continuous bacterial calcein staining has been tested in C. glutamicum ATCC 13032, E. coli MG1655, B. subtilis 168 and Micrococcus luteus DSMZ 14234 with CgAM, CbAM, and CvAM in a microfluidic cultivation device as described by Krämer et al. (2015) 16. Addition of CAM derivatives to growth medium failed to result in homogenous fluorescent cell populations of B. subtilis, E. coli, and M. luteus (data not shown). Whereas E. coli showed no fluorescence, B. subtilis showed only rare fluorescent cells, which were supposed as dead or lysed (Fig.1 B), and M. luteus showed heterogeneous CALg and CALb fluorescence. C. glutamicum showed intense and homogenous CALv fluorescence. Noteably, addition of CALg and CALb to the perfusion medium did not result in comparable bright intracellular fluorescence in C. glutamicum ATCC 13032 cells (data not shown). The AM-bound fluorogenic substrates were taken up by C. glutamicum ATCC 13032 and hydrolysed by intracellular esterases to the corresponding fluorescent products green fluorescent calcein (CALg) (Fig. 27 A, right side), blue fluorescent calcein (CALb) (Fig. 27 B, right side), violet fluorescent calcein (CALv), and also the later discussed sensing molecules, oxidised green fluorescent calcein (CALox) in case of singlet oxygen radical presence and oxidation

131 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Figure 27: Fluorogenic acetoxymethly esters and their corresponding fluorescent carboxylic acids after hydrolysis by intracellular esterases. (A) calcein green acetoxymethylester (CgAM) and (figure legend continues on next page▼)

132 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

(▲continued figure legend of Fig. 27) calcein green (CALg). (B) calcein blue acetoxymethylester (CbAM) and calcein blue (CALb). (C) dihydrocalcein acetoxymethyl ester (DHCAM) and the corresponding green fluorescent oxidized and hydrolysed calcein (CALox). (D) pHrodo Red AM and red fluorescent pH sensitive pHrodo Red as proposed by Ogawa et al. (2010). reaction (Fig. 27 C, right side), and red fluorescent pHrodo Red (Fig. 27 D, right side) or green fluorescent pHrodo Green. The fluorescent products were found to undergo partly energy-dependent efflux as proved by carbon starvation experiments and described in detail in the chapter 5.4 and by Krämer et al. (2015) 16. These findings led to the hypothesis of taxonomic dependency of successful uptake of AM-coupled sensing molecules by bacteria in difference to mammalian cells. Hence, own results were thouroughly compared to literature. Bacterial studies with calcein derivatives were summarized in a phylogenic tree with own results in Fig. 28 (adapted from 29). However, microbiological analysis with CbAM and CvAM were almost not found for bacteria. Hendon-Dunn et al. (2016 and 2018) showed a cell growth independent viability assay of the actinobacterium Mycobacterium tuberculosis cultivated and treated with antibiotics in a chemostat and staining with CvAM prior FACS analysis 87,142. In the past, CgAM was tested together with PI or Sytox Red as microscopic viability assay of dental biofilms and rated as inappropriate because of difficulties as rapid fading 165. In addition, CgAM was implemented successfully together with PI to FACS

133 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Figure 28: Phylogenic tree of prokaryotes tested for calcein acetoxymethyl ester derivative staining. Modified according to 16,29,87,139,142,158,174–178. Positively stained strains are marked green for CALg and violet for CALv. Results of this work are marked with *. M. luteus additionally was stained successfully by CbAM. ** indicates that aerobic S. aureus and B. subtilis were CALg positive after prolonged static staining for 1h at unknown CgAM concentration. However, B. subtilis remained unstained under CgAM perfusion in own experiments despite of rare apparent unviables (adapted from 29). analysis of Comamonas testosteroni TK102 175 and the actinobacterium Clavibacter michiganensis 158. Nevertheless, Diaper and Edwards (1994) reported that CgAM failed to stain γ- proteobacteria (Erwina herbicola, Klebsiella pneumoniae, 134 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Salmonella pullorum, Salmonella thyphimurium, Vibrio natriegens, Vibrio vulnificus, Pseudomonas aeruginosa, Pseudomonas fluorescens, Pseudomonas putida), firmicutes (Bacillus cereus, Bacillus thuringiensis, and the actinobacterium Arthrobacter globiformis, and Porter et al. (1995) described CgAM as inappropriate for FACS analysis of environmental water bacteria 139,163. However, Diaper and Edwards (1994) achieved CALg-fluorescent Staphylococcus aureus and B. subtilis at prolonged staining time for 1 h at unstated CgAM concentration and staining conditions 139. In summary, positive calcein staining was demonstrated for viability measurement with CgAM of the actinobacterium Clavibacter michiganensis 158, the β-proteobacterium C. testosteroni 175 and with CvAM for the actinobacteria C. glutamicum ATCC 13032 16 and M. tuberculosis 87,142. In addition, intercellular molecule exchange of CALg has been shown for the cyanobacterium Trichodesmium erythraeum 177,179,180, the firmicute Clostridium acetobutylicum as well as the δ-proteobacterium Desulfovibrio vulgaris 176. Microscopic analysis of the archaeal consortium Nanoarchaeum equitans and Ignicoccus hospitalis revealed a separated periplasm of the non- fluorescent host and its by CALg fluorescent epibiont 178. Thus, a phylogenetic connectivity according to key cell envelope features lays on hand, if AM-coupled molecular fluorescent probes are taken up by a prokaryotic species and converted 29. The non-existence of internal AM-moieties hydrolysing carboxylesterases would be a

135 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm logical explanation of absent intracellular fluorescence. However, carboxylesterases are highly conserved in the tree of life and can be found in all kind of prokaryotes 181. In difference, the bacterial cell wall is highly species specific in structure and content 29,152. There is no evidence found that state the role of membrane fluidity, membrane lipid divergence, or membrane-embedded transport proteins for AM- bound molecules 182–184. Nevertheless, for gram-negative bacteria, the composition of the bilayered outer membrane decides passage of aromatic molecules such as antibiotics by alteration of the lipidpolysaccharide 151.

5.3.3 Real-Time Oxygen Reactive Species Sensing in Living Bacteria

Another AM-coupled calcein derivative is the green fluorogenic DHCAM (Fig. 27 C). This probe molecule has not only to be hydrolysed by an internal esterase, additionally, it has to be oxidised

1 17 by high-energetic singlet oxygen ( O2) to evolve fluorescence .

1 DHCAM is reported to indicate selectively O2 since it was reported to fluoresce induced by visible light and in the presence of a hydroxyl

1 185 radical (OH•) source that is able to induce O2 . Thus, DHCAM is extracellular non-fluorescent, permeates in cells, where it is trapped after hydrolysis of the AM-moieties, and turn to bright fluorescence

136 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm of the converted to oxidised reduced green fluorescent calcein

1 186 (CALox), if O2 is generated in the cytosol . DHCAM conversion to CALox was used for evaluation of phototoxicity absence of the fluorescence time-lapse imaging methods developed and presented in this PhD thesis as described more in details in chapters 5.1, 5.4 and 5.5 as although in Krämer et al. (2015), and in Krämer et al. (2016) 16,17. Whereas, C. glutamicum ATCC 13032 showed absence of ROS at different multiplexed fluorescence imaging conditions inspite of the positive controls.

Bacteria that grow on the presence of oxygen (O2) are continuously exposed to oxidative stress, which is mainly caused by ROS that are byproducts of the respiratory chain. ROS appear in case of incomplete

- O2 reduction. As a consequence, superperoxide radicals (O2 ), OH•,

2- hydrogen peroxide (H2O2) that induces peroxides (O2 ) and other oxygen radicals, especially if iron is present to catalyse the Fenton’s reaction 187–189. It is noteworthy, that ROS formation in the cell can cause severe damages to biomolecules such as proteins and the DNA 76,189. Oxidative stress is reported to be involved in antibiotic toxicity 190, aging 191, protein degradation 192, lipid peroxidation 193, as also in invasive light dosage of bacteria 186. Furthermore, oxidative stress decrease industrial production yield of chemicals as L-methionine 194. In actinobacteria, redox-buffering low-molecular-weight mycothiol is oxidised by ROS to mycothiol disulphide and recovered by NADPH- dependent mycothiol disulfide reductase (Mtr) 195,196. Several bacteria

137 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm perform oxidative stress sensing and regulation by the SoxR(S) system

- (a sensor for O2 ) and the transcriptional regulator OxyR (a sensor for

188,193 H2O2) . H2O2 and peroxides are degraded by highly conserved

- peroxidases such as catalase, whereas O2 moieties are metabolised by superperoxide dismutases (SOD), if oxidative protection mechanisms are induced 188,193. However, soxRS is absent in C. glutamicum 197. Expression of fluorescent fusion proteins under the control of known oxidative regulator gene promotors is an indirect study of ROS impact. The approach has its shortcomings due to the fact that C. glutamicum is considered that it has lost partly genes as soxRS during evolution, whereas multiple genes are upregulated and still involved in ROS presence 197. Further, ROS disappear rapidly due to radical-radical coupling, presence of scavenger molecules like vitamins, biocatalytic reactions, or due to destructive radical reactions with biomolecules 192,196,198. Therefore, temporal resolved intracellular ROS sensing has high relevance for single-cell analysis, since they can occur spontaneously and sporadic in cells in discontinuous concentration 188. The implementation of an intracellular ROS measurement technique means integration of the sensor into the cell itself. Fluorescent nanoparticle sensing in eukaryotic cell species, which take them up by phagocytosis, is no option for by magnitudes minor sized bacteria, that resist integration of sensing nanostructures by active uptake of foreign bodies 199. Oxidative stress in bacteria is determined by comparative

138 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm growth and expression level analysis 187,200, knockout mutant strain studies and their alteration of metabolism 201,202, as well as direct intracellular measurement of ROS 198,203.

1 The implementation of fluorogenic DHCAM as O2 sensor of living C. glutamicum ATCC 13032 cells is a direct and highly sensitive sensing of radical presence with the advantage of in situ measurement, where the radical formation takes place with temporal resolution to know when the radical appearance and disappearance happen. The environmental control in the microfluidic device allowed defined condition and avoid artefacts by sample preparation. Hence, shortcomings of indirect measurements of promotor activity or radical product measurement are avoided resulting in a real-time sensing method or ROS.

1 O2 molecules are induced, if oxygen molecules are pushed in an energetically higher state 186. Intracellularly, this is especially of relevance in optic measurement systems of living cells. Hence, the temporal resolved imaging was proved for absence of phototoxicity under optimized imaging conditions for single-cell studies of C. glutamicum ATCC 13032, that was found as non-phototoxic for expression dynamic studies of stress and innate prophage genes in absence of stress and under nutritive stress (see chapter 5.1.2) and multiplexed viability studies (see chapters 5.4 and 5.5). However, not only absence of ROS is of interest in single-cell studies of C. glutamicum ATCC 13032. Furthermore, it can be of importance

139 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm to verify ROS accumulation in knockout strains with deficiencies in the respiratory chain, the major source for ROS in aerobic organism.

The cytochrome bc1-aa3 supercomplex is the primary terminal oxidase of the aerobic branched respiratory chain of C. glutamicum (Fig. 29 and Fig. S1, chapter 8.5).

Figure 29: Scheme of the branched respiratory chain of C. glutamicum. The respiratory system consitst of several primary dehydrogenases (DH, and oxidoreductase (OR)) that transfer electrons towards menaquinone (MK to MKH2). The respiratory chain is terminated by two terminal oxidases, cytochrome bd oxidase and the chytrochrome bc1-aa3 supercomplex (with O2 as final electron acceptor), menaquinones and nitrate reductase, that operates under anaerobic conditions with nitrate as final electron acceptor. The scheme has been adapted from Niebisch and Bott (2003) 204.

The C. glutamicum ATCC 13032 knockout mutants C. glutamicum ATCC 13032 ΔctaD (ΔctaD) and C. glutamicum ATCC 13032 Δqcr (Δqcr) were either deficient of the gene ctaD encoding for the subunit

205 1 of the cytochrome aa3 oxidase or devoid of the genes qcrC, qcrA,

205,206 and qcrB of the bc1-aa3 supercomplex to build bc1 . Previously, 140 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Koch-Koerfges (2011) found that mutants that lack either the cytochrome bc1 complex (Δqcr) or the cytochrome aa3 oxidase (ΔctaD) showed different growth on minimal medium CGXII, although both mutants have a non-functional bc1-aa3 supercomplex 207. The presence of the Fe-S cluster containing Rieske protein QcrA in ΔctaD was suggested to cause this growth difference, because of all

205,207 subunits of the bc1-aa3 supercomplex are missing in the Δqcr .

The resuidal bc1 complex subunits may catalyse the Fenton’s reaction (see supplementary Fig. S2, chapter 8.5) 207. In previous studies, it was shown that both supercomplex mutants showed a by 50 % impaired growth compared to the wild type on complex medium BHI 201,205. However, on CGXII + 4 % (w/v) GLC as sole carbon and energy source, growth of ΔctaD was nearly absent in static cultivation conditions, while Δqcr showed 50% of wild type again 207. Growth of ΔctaD could be reserved by addition of thiamine, which has been describes as an inhibitor of oxidative stress and lipid peroxidation (see supplementary Fig. S2 and Tab. S1, chapter 8.5) 207. The mutants ΔctaD and Δqcr were growth impaired in microfluidic cultivation, too, as already described for shaking flask cultures in CGXII + 4 % GLC in 201,207. However, the addition of thiamine did not remarkably improve cell growth of both mutants in microfluidic perfusion cultivation (Fig. 30 A). Hence, the influence of thiamine addition was determined on the intracellular state of C. glutamicum strains ΔctaD and Δqcr. Thus, the intracellular oxidative stress, the

141 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm metabolic activity and cell death were analysed on single-cell resolution as described subsequently in chapters 5.4 and 5.5. Koch-Koerfges (2011) showed increased ROS production in both mutants by the indirect ROS detection of lipid peroxidation byproducts using the thiobarbituric acid reactive substances (TBARS) assay (supplementary Tab. S2, chapter 8.5) 207. Therefore, the overall

ROS indicator CM-H2DCFDA was tested as general ROS indicator to

- - detect lipid peroxidation causing ROS like •O2 molecules and O2 in

190 both mutants . As H2DCFDA, CM-H2DCFDA is cell permeant, non-fluorescent and hydrolysed by internal esterases to the chloromethylform of 2-,7-dichlorodihydrofluorescein (CM- H2DCF) that is considered as retained inside the cell.

CM- H2DCF is rapidly oxidised to the fluorescent chloromethylform

208 of 2-,7-dichlorofluorescein (CM-DCF) by ROS . CM-H2DCFDA is hydrolysed intracellularly after uptake and converted in presence of

- - H2O2, •O2 , HO•, peroxyl radicals (ROO ), nitric oxide (•NO), peroxynitrite (ONOO-) to the green fluorescent fluorescein derivative CM-DCF that is supposed to bind to cytoplasmic proteins 190. Although, CM-DCF showed no favourable retention in the cells, the mean single-cell CM-DCF fluorescence normalised by total cell area increased continuously for both knockout mutants indicating an increase of oxidative stress in both mutants (Fig. 30 B).

142 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Figure 30: Normalized growth rate, hydrogen production, singlet oxygen radical formation of C. glutamicum WT, and the mutants ΔctaD and Δqcr in presence and absensce of thiamine. (A) Growth rates of both mutants were compared and normalised to C. glutamicum ATCC 13032 growth under equal prolonged pre- cultivation conditions in CGXII + 4 % (w/v) GLC (grey) and in with CGXII + 4 % (w/v) GLC + 2 µg/mL thiamine. Normalized growth rate is shown from cultivation in CGXII + 4 % (w/v) GLC in the presence of CM-H2DCFDA (green), DHCAM (red), DHCAM and 0.2 µg/mL thiamine (orange), and DHCAM and 200 µg/mL thiamine (yellow). (B) Increase of mean single-cell (figure legend continues on next page▼)

143 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

(▲continued figure legend of Fig. 30) fluorescence of DCF that have been normalized by cell area of both mutants. (C) Total cell number of ΔctaD (red lines) and Δqcr (green lines) in CGXII + 4 % (w/v) GLC with addition of DHCAM, DHCAM and 0.2 µg/mL thiamine, and DHCAM and 200 µg/mL thiamine. (D) Mean single-cell CALox fluorescence of ΔctaD (red) and Δqcr (green) normalized by total cell area without thiamine, with 0.2 µg/mL thiamine, and with 200 µg/mL thiamine. (E-J) Phenotypic heterogeneity of oxygen radical appearance in absence of thiamine in ΔctaD with (E) spontaneous oxidative stress in all cells of the colony, (F) non-fluorescent cells without radical formation, and (G) – (J) intermittent single-cell CALox fluorescence indicating oxygen radical fluctuations.

With addition of DHCAM, division of both mutants proceeded (Fig. 30 C), whereas the mean single-cell CALox fluorescence normalised by total cell area differed in both mutants in absence of thiamine addition as shown in Fig. 30 D. Thereby, the mutant Δqcr showed

1 remarkable different phenotypes of O2 appearance in CGXII + 4 % GLC without thiamine addition (Fig. 30 E – J). It happened that i.) the whole colony showed simultaneously CALox fluorescence and stopped cell division (Video 8 and 9, chapter 8.2), ii.) the colony remained non-fluorescent without cell growth arrest (Video 10, chapter 8.2), or iii.) that cells exhibited flickering CALox fluorescence and growth inhibition (Video 11, chapter 8.2). This intermittent CALox fluorescence is shown for a small colony at the beginning of the cultivation (Fig. 30 G), after ~7 h (Fig. 30 H), after ~13 h (Fig. 30 I) and after almost 20 h (Fig. 30 J) of cultivation. The mutant ΔctaD showed only in single cells initially CALox fluorescence that disappeared before delayed cell division started of these cells (Videos 12 – 14, chapter 8.2).

144 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

1 Figure 31: O2 formation monitored by single-cell CALox fluorescence traces of C. glutamicum Δqcr and ΔctaD in the absence or presence of thiamine. DHCAM 1 is converted to fluorescent CALox after hydrolysis and oxidation with O2. The single- cell CALox fluorescence traces of Δqcr colonies are shown (A) with absence of thiamine, (B) with 0.2 µg/mL thiamine, and (C) with 200 µg/mL thiamine. Single-cell CALox fluorescence traces of the ΔctaD mutant are shown (D, G, and H) without thiamine addition, (E) with 0.2 µg/mL thiamine, and (F) with 200 µg/mL thiamine. Without thiamine, oxygen radical (figure legend continues on next page▼)

145 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

(▲continued figure legend of Fig. 31) appearance in ΔctaD mutant cells (D) was absent in a colony, (G) varied alternating over the colony, or (B) increased linearly after a certain time for the whole colony. (I) A schematic overview shows the aerobic respiratory chain, that is embedded in the cytoplasmic membrane of C. glutamicum, with its cytochrome bd branch (green transparent underlaid) and its cytochrome bc1- aa3 supercomplex branch (yellow/red transparent underlaid). The genetic difference of the cytrochrome supercomplex bc1-aa3 mutants ΔctaD and Δqcr does not matter in prescence of thiamine under light induced condition (h ν). If no thiamine is present 1 only the ΔctaD tend to produce O2 radicals. The scheme has been adapted from 204,209,210.

Although, the addition of thiamine resulted in no difference for growth, metabolic activity or viability of C. glutamicum ATCC 13032

1 and its ΔctaD and Δqcr knockout mutants, O2 sensing with DHCAM exhibited a difference in the ΔctaD mutant according the thiamine addition. The other Δqcr mutant showed neither CALox fluorescence without (Fig. 31 A), nor with a low concentration at 0.2 µg/mL (Fig. 31 B) or a high concentration at 200 µg/mL (Fig. 31 C).

1 In difference, thiamine had a scavenger function of high energetic O2 radicals for the mutant ΔctaD. The single-cell CALox fluorescence

1 traces of a ΔctaD mutant showed no O2 presence in some ΔctaD colonies without thiamine (Fig. 31 D), for all ΔctaD colonies with 0.2 µg/mL thiamine (Fig. 31 E), and with 200 µg/mL thiamine (Fig. 31 F). Partly, the ΔctaD colonies without thiamine addition exhibited

1 1 sporadic O2 formation with increase followed by spontaneous O2 decrease assumed by radical-radical reaction (Fig. 31 G) and constant increase (Fig. 31 H). The increase of single-cell CALox fluorescence was always followed by growth inhibition.

146 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

In Fig. 31 I, a schematic overview shows the part of the aerobic respiratory chain, that is embedded in the cytoplasmic membrane of C. glutamicum ATCC 13032, and is evolved in menaquinone (MK) oxidation for electron transfer. Part of this MK oxidizing route is the cytochrome bd branch and the more substantial cytochrome bc1-aa3

204–206 supercomplex branch . The cytochrome supercomplex bc1-aa3 is embedded in the cytoplasmic membrane of C. glutamicum ATCC 13032 that is covered by an arabinogalactan-peptidoglycan layer, the mycolic acid lipid layer, and an outer polysaccharide pattern layer 204,209,210. Under external light induced condition (h ν), the Δqcr mutant with the heme and copper containing subunits of the cytochrome supercomplex

1 part aas did not produce O2 radicals independent of thiamine presence.

1 In contrary, the ΔctaD mutant tend to develop O2 radicals in absence of thiamine in the heme and Fe-S cluster containing subunits of the

1 cytochrome supercomplex part bc1, whereas O2 is O2 in an energetically higher state induced by light exposure or H2O2. The

1 oxidative stress triggered by O2 can cause destruction of proteins, DNA damage or lipid peroxidation that impairs the cell severely 193,203. The extended and sequential pre-cultivation of C. glutamicum ATCC 13032 and its mutants with deficiencies of the respiratory system was analysed if an impact on the metabolic activity was given or an increased case of cell death was triggered. The CvAM conversion to fluorescent CAL was observed over time with and without addition of

147 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

2 µg/mL thiamine in the perfusion medium CGXII + 4 % GLC (Fig. 32). Independent of thiamine addition, both mutants, ΔctaD and Δqcr, showed lowered CAL fluorescence compared to the WT (Fig. 32 A – F). The lowered mean single-cell CAL fluorescence of ΔctaD and a slight increase of Δqcr overtime could not be reasonably explained by increased efflux. Thus, the hydrolytic conversion of CvAM seemed to be lowered either by decreased esterase activity or due to by-product inhibition of ethanol that has to be metabolized by alcohol dehydrogenase or alcohol oxidase. The partial deficiencies of the respiratory system of C. glutamicum ATCC 13032 led to growth decrease of the mutant strains ΔctaD and Δqcr in comparison with the WT without remarkable cell death (Fig. 32 G – L) with and without 2µg/mL thiamine. In summary, the knockout mutants were not more prone to cell death as the wild type. However, the mutants were remarkably reduced in growth and metabolic activity, that has been found by static cultivation (Supplementary Fig. S1, chapter 8.5) and found for microfluidic cultivation (Fig. 30 A, and Fig. 32). Previously, indirect ROS measurement with the TBARS assay (supplementary Tab. S2, chapter 8.5) showed that increased lipid peroxidation was found in ΔctaD compaired to wild type or Δqcr and was significantly reduced with thiamine.

148 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Figure 32: Metabolic activity sensing and temporal resolved cell death determination of C. glutamicum ATCC 13032 and the mutants Δqcr and ΔctaD. Metabolic activity of C. glutamicum ATCC 13032 colonies determined by CALv fluorescence traces of (A) WT, (B) ΔctaD, and (C) Δqcr cultivated in CGXII + 4 % GLC without thiamine and (D) WT, (E) ΔctaD, and (F) Δqcr cultivated in CGXII + 4 % GLC with 2 µg/mL thiamine. Single-cell death determination of C. glutamicum ATCC 13032 colonies determined by PI fluorescence traces of (G) WT, (H) ΔctaD, and (I) Δqcr cultivated in CGXII + 4 % GLC (figure legend continues on next page▼)

149 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

(▲continued figure legend of Fig. 32) without thiamine and (J) WT, (K) ΔctaD, and (L) Δqcr cultivated in CGXII + 4 % GLC with 2 µg/mL thiamine.

Direct single-cell ROS analysis showed comparable presence and

- - increase of H2O2, •O2 , HO•, ROO , •NO, ONOO• over time, which has been indicated by intracellular hydrolysis and oxidation of CM-

190 H2DCFDA with high temporal resolution (Fig. 30 B) . Especially,

1 the internal detection of O2 with DHCAM, which is hydrolysed and

1 oxidised to CALox, revealed heterogenous and sporadic O2 presence in ΔctaD without thiamine addition that immediately stopped cell growth. This gave the indication for difference of ΔctaD growth with thiamine in shaking flasks and in microfluidic perfusion cultivation, since thiamine stopped spontaneous growth inhibition by intracellular ROS formation on the single-cell level (see supplementary Videos 8 – 13, chapter 8.2). This effect was obscured in population growth determination based of OD600 measurement that growth improvement by thiamine (Fig. 30, Fig. 31, and supplementary Fig. S3, and Tab. S1, chapter 8.5).

5.3.4 A Novel Integration of AM-Bound pH Sensitive Probes in C. glutamicum Cells

The internal pH (pHint) is essential for every cell to maintain enzyme functionality, protein stability, and at least viability. PHint is regulated by pH homeostasis to ensure vital conditions and efficient cell 150 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

211 metabolism . External pH (pHex) of C. glutamicum cell cultures was measured instead or set partly by researchers and bacterial growth, gene expression, or proton motive force was measured at known

202,211 external pH . However, sensing of pHint in single-cellular living and growing organisms has its challenges and pitfalls 212,213.

Single-cell pHint is already measured in bacteria by expression of heterogeneous, pH-sensitive fluorescent GFP derivative pHluorin

213,214 . Prior pHint mesurements in bacteria used active uptake of 5(and 6-) carboxyfluorescein succinimidyl ester (CFSE) or unesterified 2’,7’-bis-(2-carboxyethyl)-5 (and 6-)-carboxyfluorescein (BCECF) for pHint measurement that has been reported by Breeuwer et al. (1996) for the firmicutes Lactobacillus lactis, Listeria innocua, B. subtilis and mixed cultures of Lactobacillus delbrueckii and Listeria innocua 77,215. Best intracellular retention of the hydrolysed fluorescent product carboxyfluorescein and the carboxylic acid BCECF were achieved in absence of a major carbon source than in presence of glucose or lactose. Similar findings of probe molecule retention were found with the actinobacterium C. glutamicum and CALv used for metabolic activity sensing as shown in detail by Krämer et al. (2015) and in the following chapter 16,77. The internalization of sensing particles in bacteria has not been demonstrated to our knowledge so far. For actinobacteriaceae, the fluorogenic substrate BCECF-AM was demonstrated to be suitbale for pHint sensing in the anaerobic microorganism Propionibacterium acnes 216. Another pH-sensitive

151 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm fluorescent probe is the fluorogenic esterase substrate carboxy- seminaphtorhodafluor acetoxymethyl ester (CSNARF1-AM) that has been successfully demonstrated for ratiometric pHint measurement of

217 C. glutamicum . A similar approach for pHint measurement was shown for gram-negative Pseudomonas aeruginosa biofilms using the carboxylic acid CSNARF4 218. Ratiometric fluorescence measurement is widely established for cellular pHint sensing after calibration with permeabilised cells 77,213,214,216–218. For calibration, the cell wall has to be made chemically permeable for all of these intracellular pH measurement protocols for bacteria to equalise pHex of the cellular environment and pHint of the cytoplasm, where the fluorescence-based optical pH sensing is performed.

Similarly, to the successfully for actinobacteria tested pHint probes

BCECF-AM and CSNARF1-AM, a pHint single-cell sensing method was tested in C. glutamicum using the commercially available fluorescent pH probes pHrodo Red and pHrodo Green which are linked with AM for improved uptake by eukaryotic cells and tissue 156. The pH homeostasis in C. glutamicum is linked to oxidative stress, iron associated metabolism, and valid amino acid production 211. C. glutamicum maintains its pHint at 7.5 ± 0.5 at pHex range of 6 -9 as found by Follman et. al (2009) with the help of low acidic or basic radioactive probes and their intracellular accumulation 211. Similar results were found by Leyval et al. who stated a constant cytoplasmic

152 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm pH for pHex 6.6 – 8.3 and a pHint variation of 7.7 – 8.3 at pHex 7.3 using the probe CSNARF1-AM 217. Comparable to CSNARF1-AM, pHrodo Red-AM and pHrodo Green- AM were taken up by C. glutamicum ATCC 13032 immediately prior calibration was started. Prior calibration start, a medium switch was performed from BHI medium to pH adjusted buffer or pH adjusted BHI medium with antibtiotic addition for cell permeabilization. The

pH adjusted media perfusion subsequently changed pHex (Fig. 33, after 2 h, and Fig. 34, from the beginning of the experiment). The presence of pHrodo Red-AM and pHrodo Green-AM was kept at constant concentration. An increase of the fluorescence background could not be observed in the cellular environment that indicated good sensor molecule retention and no remarkable efflux of both hydrolysed pH sensitive sensing molecules. Furthermore, the invasive calibration with the ionophoric antibiotics NIG and VAL with buffer at pH 4.5, 5.5, 6.5, and 7.5 immediately stopped cell growth and cell shape adaption. However, permeabilisation of gram-positive C. glutamicum ATCC 13032 for calibration of pHint measurement was not a trivial process and could not be performed as described in the manufacturer’s protocol for adherent mammalian cells. After initiation of calibration after ~ 2 h at pH 4.5 to 7.5, some bacterial cells appeared to be more antibiotic tolerant to the ionophoric NIG and VAL than other cells in the isogenic colonies (Fig. 33 A-D),

153 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm although, the pH sensitivity of the molecular pH probes with decreasing pHex could be observed due to the increase of single-cell fluorescence (Fig. 33 A-D) and the fluorescence ratio (Fig. 33 E). Nevertheless, several cells lost intracellular fluorescence, that indicated antibtiotic induced cell lysis (Fig. 33 A, C, D). The resulting fluorescence ratios are shown in Fig. 33 E at pH 4.5 – 7.5. At pH 5.5, 6.5, and 7.5 the fluorescence ratios stayed rather constant for more than ~16 h with minor standard deviation. The slight drift was supposed to be caused by concentration difference of pHrodo Red compared to pHrodo Green, because it could not be clearly identified as cause of photobleaching. In addition, the fluorescence ratio at pH 4.5 significantly differed from the linearity of the other fluorescence ratios at higher buffer pH. Notably, the standard deviation of the fluorescence ratio at pH 4.5 was rather small. Therefore, the resulted calibration was not satisfying and had to be improved.

Breeuwer et al. (1996) described improvement of pHint measurement with CFSE over time using the same antibiotics at the same concentration for permeabilisation of several gram-positive firmicutes and E. coli and with addition of the nutrient lactose 77. The importance of a carbon source presence could be considered to fuel the metabolic activity of bacteria that improved antibiotic treatment for cell wall permeabilisation. Hence, the calibration protocol was modified and performed in medium BHI with VAL only instead of the

154 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm manufacturers buffer with undefined ingredients, which is concepted for mammalian cells. After initiation of the second calibration in a new

Figure 33: Intracellular fluorescence of pHrodo Red and pHrodo Green and intracellular fluorescence ratios over time in buffer with VAL and NIG. Intracellular fluorescence of pHrodo (red) and pHrodo Green (green) (A) at pHex 7.5, (B) at pHex 6.5, (C) at pHex 5.5, and (D) at pHex 4.5. (E) Fluorescence ratio of both internal pH fluorescence sensors at different pHex.

155 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Figure 34: Intracellular fluorescence and ratio of intracellular fluorescence of pHrodo Red and pHrodo Green with a calibration between pH 6.6 and 7.4. Mean intracellular fluorescence of C. glutamicum ATCC 13032 cells permeabilised with VAL in BHI medium at (A) pH 7.4 (red), (B) pH 7.0 (green), and (C) pH 6.6 (blue), and (D) in CGXII medium + 4% GLC at pH 7.0. (E) Fluorescence ratio of both internal pH fluorescence sensors at different external pH in BHI medium with valinomycin at pH 7.0 (green), 6.6 (red), and 7.4 (blue). CGXII medium with 4 % GLC and without VAL was set to pH 7.0 (black).

156 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm microfluidic culture, the pHint immediately increased in a fluorescence peak for all pHex (Fig. 34 A-C) in BHI, before a constant single-cell fluorescence was reached for both pH sensitive molecular probes after ~5 h and kept constant for the following ~15 h. For proof of principle, one channel was supplied with minimal medium CGXII + 4 % GLC with addition of pHrodo Red and pHrodo Green (Fig. 34 D). Living C. glutamicum ATCC 13032 cells showed full intracellular fluorescence of both molecular pHint probes after ~1 h to start measurement. In comparison to the fluorescence ratios of permeabilised cells in BHI at different pH, the non-permeabilised cells in CGXII + 4 % GLC tended to the same fluorescence ratio as cells adjusted to neutral pH with remarkable oscillation ~12 h after dye addition (Fig. 34 E). Thus, the pHint of C. glutamicum ATCC 13032 cultivated in CGXII +4 % GLC was determined to be lower as with other measurement methods for pHint measurements using radioactive probes or CSNARF1-AM 211,217 maybe due to ongoing cell metabolism 212 or disconvergece due to differences in growth phase of the cells 213. It was demonstrated that both AM-coupled pH probes were taken up prior permeabilisation with antibiotics. However, the antibiotic treatment influenced the calibration of the pHint due to the duration of several hours until the cell wall was permeable to equalize pHint with pHex. In addition, antibiotic treatment showed to be more homogenous with VAL in presence of BHI medium as with NIG and VAL added

157 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm to perfusion buffer without nutrients. The perturbation of pHint by metabolism was revised in 212, whereas the cytoplasmic buffer capacity was not immediately overcome with BHI adjusted to pH 6.6 or pH 7.4 by mild phosphoric acid addition or KOH addition.

5.3.5 Conclusion and Outlook

The AM moiety had been discovered as a shuttle of complex aromatic compounds from the cell environment in particular into the production strain C. glutamicum ATCC 13032. This shuttle function of AM if coupled to aromatic chemicals does not exclusively exist in C. glutamicum ATCC 13032. Furthermore, it has been reported also for other actinobacteria and other bacteria. Successful implementation of fluorescence assays with AM-coupled dyes were reported for actionobacteria as follows: BCECF-AM with P. acnes, CSNARF1-AM with C. glutamicum 217, CvAM with C. glutamicum ATCC 13032 16 and M. tuberculosis 87,142, and CgAM with Clavibacter michiganensis 158. Nevertheless, the proteobacteria C. testosteroni and Desulfovibrio vulgaris have been stained with CgAM 175,176. And in addition, intercellular exchange have been demonstrated with the fluorescent product of CgAM in the cyanobacterium Trichodesmium erythraeum 179 and in Clostridium acetobutylicum 176. Although, a hypothesis of phylogenetic similarity of the bacterial cell wall have been raised for the uptake shuttle 158 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm potency of AM moieties for aromatic compounds into bacterial cytoplasm, the transport mechanism of AM-bound molecules into bacteria has to be unravelled in detail. Novel intracellular sensing strategies based on AM-coupled molecular probes have been developed for C. glutamicum ATCC 13032 to measure metabolic activity, oxygen radical formation, and internal pH. Thus, temporal resolved single-cell studies with AM-bound indicators showed to be a novel methodical approach for studying the metabolism of living production strains such as C. glutamicum ATCC 13032. Direct single-cell ROS analysis was performed with the general ROS indicator CM-H2DCFDA, which indicated presence and increase of

- - - oxidantive molecules like H2O2, •O2 , HO•, ROO , •NO, ONOO over

190 1 time and the more selective DHCAM that indicated O2 appearance

1 in the cytoplasm. Especially, the internal detection of O2 with DHCAM showed better cell retention, more specificity, and revealed

1 heterogenous and sporadic O2 presence in the mutant ΔctaD in

1 absence of thiamine in the perfusion medium. O2 appearence immediately stopped cell growth, whereby the mode of action of the scavenger thiamine could be revealed to be a prevention of growth arrest by ROS formation instead of improvement of growth as suggested by population originated growth determination by OD600 measurement 207.

159 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm

Both AM-coupled pH probes, pHrodo Red and pHrodo Green, that have been tested, were taken up by C. gltuamicum ATCC 13032 cells prior permeabilisation with antibiotics. However, the antibiotic treatment influenced the calibration of the pHint due to the duration to equalize pHint with pHex. In addition, antibiotic treatment showed to be more homogenous with VAL in presence of BHI medium than with NIG and VAL added to calibration buffers. To reveal the perturbation of pHint of C. gltuamicum ATCC 13032 by changes fo pHex, metabolism and growth has to be elucidated more in detail in future with further improvement of the method calibration 212. Since, there are many more AM-bound molecular probes commercially available, the possibilities to establish further molecular sensing methods for actinobacteria has not been exhausted, yet. This includes e.g. the internal Ca2+ indicators quin2-AM154, Fura2-AM 157,219, BABTA-AM or BTC-AM 220, the internal K+ indicator PBFI- AM 157, the internal H+ indicators BCECF-AM 157,166 and CSNARF4 218, Na+ indicator CoroNA Green AM 220, Mg2+ and Zn2+ indicator MAG-Fura 2-AM 220, or the iron indicator Quin2-AM 220. Furthermore, esterification of molecular probes is in use for dye uptake of mammalian cells what can be extended by a selection of bacteria. The ability to take up AM-coupled compounds is a distinguishing feature of bacteria to differentiate bacterial consortia and bacterial metabolism. In addition to intracellular sensing, the esterification of compound could be advantageous to overcome

160 Acetoxymethyl Ester – a Shuttle to Cross the Cell Wall Barrier of C. glutamiucm antibiotic therapy shortcomings due to resistance or tolerance of severe pathogens especially their persisters. Hence, esterification of e.g. fluoroquinolones could revolutionize tuberculosis treatment and help to treat e.g. M. tuberculosis persister cells or antibiotic resistance spreading in future.

161 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

5.4 Non-invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of Calcein Acetoxymethyl Ester

The following text has been published in Krämer, C. E. M., Singh, A., Helfrich, S., Grünberger, A., Wiechert, W., Nöh, K., Kohlheyer, D. (2015). PLoS ONE 10(10): e0141768 as described at the end of the text section and adapted to this PhD thesis. The text has been published previously in Krämer, C. E. M., Singh, A., Helfrich, S., Grünberger, A., Wiechert, W., Nöh, K., Kohlheyer, D. (2015). PLoS ONE 10(10): e0141768. A. Singh contributed experimental results to Fig. 43. S. Helfrich implemented extended software to facilitate my single-cell analysis. A. Grünberger contributed to Fig. 6 to explain his microfluidic device that have been used for metabolic sensing development. D. Kohlheyer designed the layout of Fig. 35. D. Kohlheyer, A. Grünberger, and W. Wiechert proofread the manuscript.

Chapter Abstract: Phase contrast microscopy cannot give sufficient information on bacterial metabolic activity, or if a cell is dead, it has the fate to die or it is in a viable but non-growing state. Thus, a reliable sensing of the metabolic activity helps to distinguish different categories of viability. We present a non-invasive, 162 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM instantaneous sensing method using a fluorogenic substrate for online monitoring of esterase activity and calcein efflux changes in growing wild type bacteria. The violet fluorescent conversion product of calcein acetoxymethyl ester (CvAM) and its efflux indicates the metabolic activity of cells grown under different conditions at real- time. The dynamic conversion of CvAM and the active efflux of violet fluorescent calcein (CALv) were analysed by combining microfluidic single cell cultivation technology and fluorescence time-lapse microscopy. Thus, an instantaneous and non-invasive sensing method for apparent esterase activity was created without the requirement of genetic modification or harmful procedures. The metabolic activity sensing method consisting of esterase activity and CALv secretion was demonstrated in two applications. Firstly, growing colonies of our model organism C. glutamicum ATCC 13032 were confronted with intermittent nutrient starvation by interrupting the supply of iron and carbon, respectively. Secondly, bacteria were exposed for one hour to fatal concentrations of antibiotics. Bacteria could be distinguished in growing and non-growing cells with metabolic activity as well as non- growing and non-fluorescent cells with no detectable esterase activity. Microfluidic single cell cultivation combined with high temporal resolution time-lapse microscopy facilitated monitoring metabolic activity of stressed cells and analysing their descendants in the subsequent recovery phase. Results clearly show that the combination of CvAM with a sampling free microfluidic approach is a powerful

163 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM tool to gain insights in the metabolic activity of growing and non- growing bacteria.

5.4.1 Introduction

Metabolic activity is a very important parameter when analysing prokaryotes, under starvation stress, exposed to antimicrobials or fed with alternative carbon sources for fermentation processes 221–225. Nutrient limitations or dynamically changing environments provoke evolutionary optimised adaption of the bacterial metabolism to ensure survival of the species. Therefore, bacteria are capable of rapidly sensing important intrinsic and extrinsic parameters affecting their survival, growth, and reproduction. However, the term “metabolic activity” is a resulting sum parameter of many enzymatic reactions. Generally, metabolic activity can be determined by measuring a specific substrate conversion, a detectable enzyme activity or a metabolite 226. In contrast to invasive metabolic activity measurements using, e.g. quenching, cell lysis, or harmful environmental changes, common strategies for non-invasive measurement of metabolic activity use fluorescence to resolve intracellular, metabolic pathways. Since the metabolic pathways vary between organisms as well as individuals, metabolic activity measurements have to be tailored to metabolic specificities of microorganisms. The sensing-regulation function of 164 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM cells can be exploited by using biosensor constructs for which the host has to be genetically modified beforehand. An overview of biosensor applications is given in 60. However, this is not a suitable approach if genetic modification is problematic, e.g. for anaerobic and environmental studies or fermentative production validation 227. The differentiation in cofactor-dependent metabolic processes like pump activity and co-factor independent enzymatic conversion by hydrolases and dehydrogenases is stressed in 228. Thus, fluorescence can also be introduced by fluorogenic substrates inside cells. An overview about those substrates used for bacterial differentiation is given in 134,136. Enzymatic hydrolysis of fluorogenic esters has been reported already for five decades to study living cells 229. The commercially available green fluorogenic CgAM is conventionally used, besides other fluorochromes, to validate viability of mammalian cells 85,230,231. Intracellular carboxylesterase hydrolyses the ester groups of CAM to the corresponding carboxyl groups and alcohol is cleaved off equimolar 231. The fluorescent product calcein is retained inside the cell due to its ionized carboxyl groups 173. A recent study on cancer cells reported that the fluorochrome is transported out in an energy-dependent mechanism in dependency of environmental conditions of the cell 85. In general, cancer cells use ATP-binding cassette (ABC) transporters, e.g. MRP-1, which is related to the potential of multidrug resistance, to shuttle calcein through the membrane. The hydrolysis of CAM and the efflux of calcein is already

165 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM commercialized for mammalian viability testing or used for testing of multidrug resistance potential of tumor cell lines 85,230. Nevertheless, in the past, green fluorescent calcein (CALg) has also been presented for microbial analysis with flow cytometry devices 139,232. Hitherto, calcein fluorescence signals were not reported for all bacterial strains 139. However, gram-positive Staphylococcus aureus cells analysed by FACS exposed to heating and antimicrobials after 1 h incubation at optimal growth temperature showed significant differences of fluorescence compared to control measurements 233. FACS systems are widely established in all fields of microbiology for high-throughput single-cell analysis 234. However, cells have to be sampled from their environment of interest and can only be measured once, resulting in a snapshot view. As a complementary technique, time-lapse microscopy offers a high temporal resolution long time analysis of single cells. CALg was used to demonstrate nutritional stress induced exchange of intracellular material between gram-positive C. acetobutylicum and gram-negative D. vulgaris. Nevertheless, controlled experimental conditions according to nutrient supply or stress conditions and analysis of statistically relevant cell numbers can be challenging during live cell microscopy 232,235. Microfluidic devices developed in recent years offer well controlled environmental conditions that can be easily combined with fluorescence time-lapse microscopy 69,236. Thus, combination of microfluidic devices with automated time-lapse

166 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM imaging opens the gate for single-cell observations with high temporal and spatial resolution of statistically significant cell numbers. For sensing the dynamics of single-cell esterase activity and fluorochrome pump activity of the gram-positive representative microorganism C. glutamicum ATCC 13032, a microfluidic method is presented. The intracellular hydrolysis of a CvAM derivative and subsequent calcein efflux were unravelled in isogenic prokaryotes and their descendants of several generations. C. glutamicum ATCC 13032 is a relevant model organism to study dormancy as well as antibiotic tolerance or resistance, since it is related to pathogens like Mycobacterium tuberculosis or Corynebacterium diphtheriae 237,238. Dormant cells are characterized by reduced metabolism, no cellular growth, and the absence of cell division. These bacteria have to be distinguished from cells with the fate to die by proving metabolic activity or resuscitation 144. Continuously perfused CvAM converted by C. glutamicum ATCC 13032 has been applied to distinguish non-viable cells from metabolically active but non-growing cells. For the first time to our knowledge, violet fluorescent calcein efflux by a prokaryote was determined. This was realized by using our microfluidic cultivation technology in which several hundred microbial microcolonies can be cultivated in cellular monolayers under constant environmental conditions 36,38. This setup facilitates the single-cell analysis of dynamical intracellular heterogeneities of bacterial physiology in

167 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM combination with high temporal resolution given by the combination of microfluidics and fluorescence time-lapse microscopy 36.

5.4.2 CAM Uptake and Calcein Fluorescence Formation

A dynamic, non-invasive metabolic activity sensing method was set up successfully using the biotechnologically relevant model prokaryote C. glutamicum ATCC 13032. The esterase substrate CvAM is taken up by the bacterium and converted intracellularly to violet CALv. The mean single-cell fluorescence of bacterial cells, which is the averaged fluorescence signal of all pixels belonging to an imaged cell, described the metabolic activity considered of esterase activity and calcein efflux of bacteria on single cell level. The average single-cell fluorescence of all cells of a microcolony was designated as mean fluorescence. The mechanism how CvAM passed the approximately 32 nm thick cell wall of four different layers and the cell membrane of the gram- positive prokaryote is not unravelled yet 239. However, the metabolic route of CvAM and its hydrolysis product CALv is illustrated schematically in Fig. 35. CvAM is hydrolysed intracellularly to its corresponding carboxylic acid and all ester groups are cleaved off and the fluorochrome accumulates in living cells. The acidic CALv is secreted partially according to the cells fitness and reaction to the environmental conditions. CvAM was fed in excess continuously 168 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM together with complex medium (BHI) or minimal medium (CGXII) into a supply channel to feed single bacterial cells seeded into sequential cultivation chambers, followed by growth and division. Four cultivation conditions could be compared during every microfluidic cultivation in parallel. Video 17 (see Appendix 8.3 and Supplemented CD) displays metabolic activity sensing of growing bacteria under reference conditions at pH 7. Generally, the mean single-cell fluorescence of growing bacteria increased linearly over time until cell division. After cell division smaller daughter cells showed reduced mean single cell fluorescence compared to their mother cell. Non-growing but viable cells exhibited higher mean single-cell fluorescence than growing bacteria under comparable cultivation conditions. Cells considered as non-viable showed no significant mean single-cell fluorescence. The schematic categories according to growth and metabolic activity are depicted in Fig. 36 A. The half time t50 of fluorescence signal formation in freshly seeded C. glutamicum ATCC 13032 cells following CvAM uptake and subsequent enzymatic conversion was determined to be 10.6 ± 1.0 min in CGXII + 4 % GLC.

169 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 35: Cellular CvAM metabolism model of a C. glutamicum cell. Gram- positive C. glutamicum has a cell wall of different layers, which has to be passed by the fluorogenic substrate CAM. Once the cell wall is transversed by a presumably active diffusive mechanism, CAM is converted to the fluorophore calcein and ethanol by present intracellular carboxylesterases. The acidic calcein is assumed to be secreted by an energy dependent transport mechanism with a putative ATPase.

This was measured immediately after the addition of CvAM to the medium (Fig. 36 B). The difference of endpoint mean single-cell fluorescence of individual cells indicated that the CvAM uptake was not strictly by passive diffusion (Fig. 36 B-C) as it is supposed for CgAM and mammalian cells 143.

170 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 36: Metabolic activity sensing of growing and dividing cells. (A) Non-growing cells can be distinguished in non-viable non-fluorescent cells and non-growing but metabolically active bacteria, which showed highest mean single-cell fluorescence in comparison. Growing bacteria showed moderate to medium fluorescence (B-C) Mean single-cell fluorescence of freshly seeded bacteria is given during the transient condition after start of perfusion medium supply with CvAM during the experimental set-up phase. (B) Increase of mean single-cell fluorescence of six individual cells over time. (C) Increase of mean single-cell fluorescence in dependency of the cell size revealed marginal differences of the final equilibrium mean fluorescence. (D-F) Metabolic activity sensing of cells grown in complex medium BHI at pH 6.6 (red), pH 7.0 (green), and pH 7.4 (blue), respectively. (D) Average values of 10 cultivation chambers are presented. No significant difference in CALv fluorescence (solid lines) and growth represented by total cell area (dashed lines) could be observed for cultivation of C. glutamicum at pH 7.0 and pH 7.4. At pH 6.6, however, increased mean CALv fluorescence as well as a reduced total cell area was observed. (E) The mean single-cell area indicated a tendency (figure legend continues on next page▼)

171 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

(▲continued figure legend of Fig. 36) of cell size reduction in average over time. (F) Mean fluorescence correlates positively with the mean single-cell area at all three pH values.

Metabolic activity sensing was performed with complex medium BHI at media pH at 7.0 ± 0.4 to test the method under non-toxic cultivation conditions. C. glutamicum ATCC 13032 is reported to maintain its internal pH at 7.5 ± 0.5 in an external pH range of 6 to 9 211. Hence, the external pH was not supposed to alter the intracellular pH drastically circumventing detectable influence on CALv fluorescence. We determined that an external pH of 7.0 and higher had neither significant impact on the maximal growth rate (0.97 ± 0.03 h-1 at pH 7.0 and 1.02 ± 0.02 h-1 at pH 7.4, respectively) nor on the mean fluorescence over time. In contrast, at external pH 6.6 the mean fluorescence signal was clearly increased. Furthermore, a 20 %

-1 decrease of maximal growth rate µmax to 0.78 ± 0.02 h due to a significant reduction of total cell area over time was observed (Fig. 36 D, Video 18, see Appendix 8.3 and Supplemented CD). The mean fluorescence of all three growth conditions showed an initial decline within the two hours (Fig. 36 D). This happened simultaneously with a decrease in the mean single-cell area over time (Fig. 36 E). Due to the correlation of mean single-cell area and mean fluorescence (Fig. 36 F), consequently the mean fluorescence decreased over time.

172 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM 5.4.3 CvAM Concentration Optimization

The CvAM concentration in the perfusion medium was optimized to yield the highest fluorescence signal for online metabolic activity monitoring characterized by a hydrolysis reaction and a homeostasis like product efflux. Five different extracellular CvAM concentrations (12 µM, 23 µM, 46 µM, 93 µM, or 139 µM) CvAM in CGXII + 4 % GLC (feast condition) and in despite of CvAM carbon free CGXII – PCA (famine condition) were prepared, respectively (Fig. 37). The perfusion of C. glutamicum ATCC 13032 cells with these CvAM concentrations showed a significant increase in the mean single-cell fluorescence during a famine phase of 10 h. For CvAM concentrations higher than the optimal concentration at 46 µM no further increase in the mean single-cell fluorescence could be observed. Instead, the mean single-cell fluorescence decreased for CvAM concentrations above 93 µM. The increase in the mean single- cell fluorescence within the indicated starvation phase was followed by a sudden decrease after re-providing carbon in the perfusion medium. The intracellular CALv fluorescence transiently changed because of the active fluorochrome efflux (Fig. 37). Media shift experiments were performed to discriminate and determine the influence of CALv efflux. Mean single-cell reaction rate constants of CvAM conversion to CALv in dependency of the extracellular substrate concentration were determined immediately

173 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM after the initiation of carbon depletion (Fig. 38 A). After 10 h of famine condition the medium supply was switched back to CGXII + 4 % GLC. After the re-supply of glucose, cells responded with an immediate and significant reduction of intracellular mean single-cell fluorescence under feast conditions (Fig. 37, Fig. 38 B). The efflux rate constants of CALv secretion were determined in the initial 50 min of the re-established feast condition (Fig. 38 A). The maximal mean single-cell reaction rate constant of the CALv efflux at 0.005 min-1 was found to be twice as high as the maximal mean single- cell reaction rate constant of the CvAM conversion at approximately 0.0025 min-1. However, C. glutamicum ATCC 13032 cells always showed remaining mean calcein fluorescence depending on their relevant enzyme activity, cell size, and environmental cultivation conditions. Despite of the changes in mean single-cell fluorescence, the resulting apparent growth rates showed no significant differences at all five CvAM concentrations. The apparent growth rates rapidly decreased after initiating carbon depletion and comparably recovered with an increased standard deviation after starvation stress (Fig. 37).

174 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

175 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 37: Comparison of mean single-cell fluorescence and apparent growth rate at different extracellular CvAM concentrations. Mean CALv fluorescence for five extracellular CvAM concentrations are shown for five cultivation chambers each (every chamber is indicated with a separate colour). Perfusion medium was switched from CGXII + 4 % GLC to carbon free CGXII medium (CGXII – PCA) supply for 10 h (indicated with red frame) to inhibit the energy-dependent calcein efflux. CvAM conversion by intracellular esterase activity and subsequent CALv fluorescence showed a non-linear fluorescence increase except for concentrations higher than the optimal extracellular CvAM concentration of 46 µM. The corresponding apparent growth rates changed according to the carbon supply and not because of an increase CvAM concentration.

Figure 38: Mean CvAM conversion rate constant and mean CALv efflux rate constant. (A) A mean maximal single-cell reaction rate constant at 0.0025 min-1 was determined for CvAM conversion under inhibition of CALv efflux due to carbon limitation. A maximal mean single-cell CALv efflux rate constant was determined to be approximately twice as high at 0.005 min-1 (n = 5 colonies for each mean maximal single-cell reaction rate constant). (B) Limitations in carbon supply inhibited cell division and energy driven transport of CALv out of the cell. Thus, mean single-cell

176 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM traces increase linearly over time until carbon supply was continued (every coloured dotted line represents one individual cell, n = 15 cells).

5.4.4 Experimental Validation

The presented metabolic activity sensing was validated regarding non- invasiveness to bacterial growth and fluorescence signal stability. As already indicated by Fig. 37, a measurable impact on growth by CvAM addition could not be found for concentrations up to 139 µM. Furthermore, a possible influence on growth by metabolized ester groups attached to CvAM was analysed by supplying the CvAM- surrogate substrate methyl methoxyacetate. It is supposed to be taken up comparable to CvAM and it is then enzymatically converted to ethanol and acetic acid. The methyl methoxyacetate concentrations were varied from 5 µM to 500 µM in CGXII + 4 % GLC, corresponding to 1.7 µM to 166.7 µM CvAM. No remarkable growth rate differences were observed between microcolonies grown in CGXII + 4 % GLC with various methyl methoxyacetate concentrations or with addition of 46 µM CvAM (corresponding to 138 µM methyl methoxyacetate) (Fig. 39 A). Furthermore, phototoxicity due to fluorochrome exposure was analysed by replacing CvAM with the reduced DHCAM. DHCAM was taken up and was hydrolysed comparable to CvAM. In contrast, DHCAM requires oxidation by intracellular radical oxygen species

177 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

(ROS) typically induced under photo-oxidative stress, to generate the green fluorescent calcein derivative CALox. As described in literature, photo-oxidative stress by ROS, which were introduced by energetic light exposure, is responsible for intracellular damage of DNA, proteins, and cell membranes and triggers stress responses of microorganisms 76. ROS were detected supplying DHCAM during the cultivation of C. glutamicum ATCC 13032 with CGXII + 4 % GLC. For positive control, inoculated cells were additionally exposed (> 1 sec) to the highest illumination intensity before time-lapse imaging to generate abundant intracellular ROS. Fig. 39 B shows that phototoxic stress was clearly indicated for the positive control by increased mean single cell fluorescence (pink scatter plot) and a stagnating total cell number (red dotted line). In contrast, the typical experimental illumination intensity during phase contrast and fluorescence imaging of CALv resulted in basal mean single-cell fluorescence of CALox (green scatter plot) and cell growth (black dotted line). Concurrently, CALv had a low vulnerability to photobleaching under standard experimental conditions with low light illumination. Photobleaching of CALv was determined with cells in a growth arrest phase due to carbon starvation. It was determined to reduce the mean single-cell fluorescence by less than 0.2 % at every fluorescence time- lapse imaging snapshot (Fig. 39 C).

178 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 39: Experimental validation of the metabolic activity sensing method. (A) The CvAM-surrogate methyl methoxyacetate was added to CGXII + 4 % GLC and infused in three different concentrations in three separate supply channels of the microfluidic device to analyse the impact on growth by the intracellular digestion of the acetoxymethyl ester groups of CvAM. The average maximal growth rates of five chambers cultivated with CGXII + 4 % GLC (figure legend continues on next page▼)

179 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

(▲continued figure legend of Fig. 39) with addition of three different methyl methoxyacetate concentrations are compared to a reference without addition (n = 5 colonies). The tested methyl methoxyacetate concentrations at 5 µM, 50 µM, and 500 µM corresponded to a CvAM concentration of 1.7 µM, 16.7 µM, and 166.7 µM, respectively. In comparison the result for bacterial growth in CGXII + 4 % GLC + 46 µM CvAM is given. (B) An influence of the excitation light exposure (phototoxicity) during the fluorescence time-lapse imaging was investigated by infusing DHCAM, which is converted to green fluorescent CALox if light induced oxygen radical species are present. Typical experimental light exposure resulted in exponential cell growth (dashed black line, n = 5 colonies) and basal mean single-cell fluorescence (green scatter plot, only every 6th frame was measured, n = 5 colonies). In contrast, cells of the positive control experiment were initially exposed (> 1 sec) to maximal light intensity before starting time-lapse imaging. Control cells (n = 271 cells) displayed immediate increase in the mean single-cell fluorescence (pink scatter plot) and a stagnating total cell number (red dashed line) due to photo-oxidative stress. (C) Photobleaching was determined to be marginal as plotted as percentage of signal loss over mean single cell fluorescence (n = 85 cells).

Besides, the excellent biological compatibility, CALv also performed no recognizable photobleeding. As evident from Fig. 38 B, CALv was contained intracellularly until the initiation of active transport by resupply of glucose. Moreover, CALv fluorescence showed a signal- to-noise ratio of 15:1 and higher compared to the extracellular medium. Nevertheless, significant increase of mean fluorescence of a colony resulted not of a mere growth reducing cultivation condition. However, it depends if bacterial growth is influenced due to reduction of the internal energy level due to increased ATPase activity (e.g. secretion of internal H+ excess) 211 or ATP depletion by substrate limitation 40.

180 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM 5.4.5 Metabolic Activity Sensing under Intermittent Nutrient Limitation

Hence, starvation stress was induced to experimentally manipulate the CALv fluorescence in C. glutamicum ATCC 13032. Therefore, cells were pre-cultivated for 4 h under normal conditions (CGXII + 4 % GLC). Then an intermittent supply of the iron chelator procatechuate (PCA) (Fig. 40 B), iron (Fig. 40 C) or glucose and PCA (carbon limitation) (Fig. 40 D) was established for 12 h. Full nutrient supply was only re-provided to iron depleted and carbon starved cells after 12 h. Subsequently, recovery and growth of bacterial cells was initiated again (Fig. 40 C-D). Cells exposed to iron and carbon depletion needed approximately 24 h to reach comparable cell numbers compared to 12 h during the reference cultivation under continuous supply of CGXII + 4 % GLC (Fig. 40 A). A late depletion phase of PCA after 4 h showed only minor impact on mean fluorescence or growth (Fig. 40 B). In contrast, the switch to limitation of iron (see Video 19 and Video 20, Appendix 8.3 and Supplemented CD) and carbon (see Video 21 and Video 22, Appendix 8.3 and Supplemented CD) was accompanied by cell growth stagnation, respectively. Furthermore, elongated cells were found after iron re-supply (see Video 20, Appendix 8.3 and Supplemented CD) and under PCA limitation (Video 23, Appendix 8.3 and Supplemented CD).

181 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

The apparent growth rate revealed that iron depletion provoked a considerable growth reduction and cell division was inhibited. C. glutamicum ATCC 13032 cells with a deficiency of carbon exhibited a remarkable decline of cell growth, but continued cell division resulting in smaller descendants. Although the growth reduction was comparable during the starvation phase of iron limitation as well as carbon deprivation, the mean fluorescence differed significantly according to the missing medium component. The mean fluorescence homogeneously increased steadily for all cells in the absence of the major carbon source glucose and metabolizable PCA, whereas the mean fluorescence remained unchanged without iron supply (Fig. 40 C – D). Further on, the extracellular CALv fluorescence of the perfusion medium was compared to the intracellular fluorescence. This extracellular fluorescence was measured at three positions in various replicates: i.) inside the supply channels, ii.) at the connection channels inside the chamber and iii) in close proximity to the cells. The fluorescence measurement inside the supply channel gave a sum signal of all chambers connected upstream resulting in a global signal of CALv efflux. Supply channel measurements revealed higher fluorescence than inside the cultivation chambers due to the tenfold medium height compared to the chamber height. The extracellular mean fluorescence of the reference and the cultivation condition with intermittent iron chelator supply as well as

182 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 40: Metabolic activity sensing of C. glutamicum ATCC 13032 at intermittent nutrient limitation in minimal medium CGXII. Bacterial cells were cultivated in minimal medium CGXII with 4 % glucose (CGXII + 4 % GLC) at pH 7 before an indicated shift to a depletion phase by switch of perfusion medium supply. (A) Reference cultivation under continuous supply of CGXII + 4 % GLC. (B-D) After 4 h pre-cultivation, the microchambers were perfused with minimal medium (n = 10 colonies) (B) without iron chelator (figure legend continues on next page▼)

183 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

(▲continued figure legend of Fig. 40) protocatechuate (PCA) (CGXII + 4 % GLC – PCA, marked with violet boxes, n = 10 colonies), (C) without iron (CGXII + 4 % GLC – iron, marked with black boxes, n = 10 colonies), and (D) with carbon limitation (CGXII– PCA, marked with red boxes, n = 10 colonies), respectively. For carbon and iron depletion conditions the medium was switched back to initial medium after 15 h. Microcolony images of every cultivation conditions are shown at different experimental time points. The mean fluorescence of the colony and apparent growth rate over time are shown in comparison to the extracellular mean fluorescence of the perfusion medium in the supply channel (10 µm fluid height), inside the cultivation chamber entrances (1 µm fluid height) and in the direct cell proximity (1 µm fluid height). iron deprivation, reflected our findings of the intracellular CALv continuously released into the surrounding medium (Fig. 40 A and 40 B, respectively). However, under absence of glucose the course of the significantly lowered extracellular mean CALv fluorescence was contrasted to the continuously increasing intracellular mean fluorescence, proving CALv efflux inhibition (Fig. 40 D). Although, the intracellular mean fluorescence remained rather constant during 12 h of iron starvation, calcein was continuously secreted and consequently CvAM uptake and conversion proceeded steadily (Fig. 40 C). On a single-cell basis, iron starvation caused heterogeneous phenotypes at the end of iron limitation and during the first five hours after iron re-supply as illustrated in Fig. 41 and Video 24 (see Appendix 8.3 and Supplemented CD). The mean single cell traces of all growing and metabolically active descendants of one progenitor cells are shown in comparison to a spontaneously non-viable cell that lost its metabolic activity after lysis (Fig. 41 A). Bacteria showed an increase in their mean single-cell fluorescence initially after media 184 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM switch to CGXII + 4 % GLC – iron and to CGXII + 4 % GLC, respectively. The bacterial cells that were growth inhibited due to iron limitation showed comparable mean single-cell fluorescence as their descendants after recovery during re-supply of iron. As illustrated in Fig. 36 A, non-growing cells due to iron induced stress can be distinguished in i.) metabolically active fluorescent cells and ii.) non-viable cells with reduced mean single-cell fluorescence, with experimental examples shown in Fig. 41. In contrast to non- growing but active cells, we define non-viable cells as cells without the ability to restart growth and show no CvAM conversion with resulting CALv fluorescence. The mean single cell fluorescence shortly increased after the medium switch before iron depletion and before the recovery phase, respectively. Cells with higher intracellular fluorescence than their siblings produced descendants with increased mean single cell fluorescence in comparison (Fig. 41 A, light blue lines). Although bacteria recovered and continued to grow after iron starvation (Fig. 41 A), a minority of cells exhibited a non-viable lysing phenotype (Fig. 41 B – D, Video 25, Appendix 8.3 and Supplemented CD) or despite of full nutrient availability, a non-growing but metabolically active state with remarkably increased mean single-cell fluorescence (Fig. 41 B, Fig. 41 E, and Video 26, Appendix 8.3 and Supplemented CD). After returning to full carbon supply, few cells stayed in a non- growing and non-dividing, but metabolically active state. This rare

185 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM phenotype was present in every analysed cultivation chamber (n = 10 cultivation chambers). We observed that these dormant cells could be identified by their much higher mean single-cell CALv fluorescence compared to their siblings during the metabolic activity sensing. The appearance and evolution of a considerably high number of those non- growing C. glutamicum ATCC 13032 cells in a colony is depicted in Fig. 42 and shown in Video 22 (see Appendix 8.3 and Supplemented CD). If carbon supply was changed from famine (Fig. 42 A, end of starvation phase) to feast (Fig. 42 B, end of cultivation) condition, most non-growing cells (blue and black lines) reacted with an immediate drastic decrease in the mean single-cell fluorescence. This behaviour was similar compared to readily growing cells with continued cell division. Some non-growing cells reacted with a delay of mean single-cell fluorescence decrease after carbon re-supply. Using the single-cell mean fluorescence traces, the non-growing cells could be subdivided in different phenotypes as shown in Fig. 42 C. Almost every fifth cell of these dormant bacteria exhibited a slowed down decrease in the mean single-cell fluorescence after re-supply of carbon (Fig. 42, indicated red). The other two categories showed similar mean single-cell fluorescence during the starvation phase, but differed after the recovery phase of the colony. On the one hand there were non-growing cells with unremarkable course of mean single-cell fluorescence compared to growing cells (Fig. 42, indicated black), on

186 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM the other hand the majority of non-growing cells revealed an increase in the mean single-cell fluorescence after some hours (Fig. 42, indicated blue).

Figure 41: Metabolic activity sensing under iron limitation at single-cell level. (A) Mean single-cell fluorescence traces of all descendants of a progenitor cell are shown under intermittent iron supply. A daughter cell with higher CALv fluorescence than its siblings (light blue arrow) generated growing descendants with increased mean single-cell fluorescence traces (light blue lines) in comparison to other descendants of the initial progenitor cell (grey mean single-cell fluorescence traces). A single spontaneously non-growing cell changed from a dividing state to a non-growing state loosing esterase activity and intracellular CALv due to lysis (indicated by black arrow, lysed cell shown in (C) and (D), respectively). (B) Mean single-cell fluorescence traces of spontaneously non-growing cells of a microcolony are shown. Lysing cells (red lines) lost fluorescence spontaneously after lysis (red arrows). However, they were still detectable as apparently intact cells (end of recognition marked with red asterisks). The mean single-cell fluorescence traces of a spontaneously non-growing but metabolically active cell (blue line) are shown in comparison. Mean single-cell fluorescence increased shortly after cell birth (blue arrow). (C) A lysed but apparently intact cell (marked with black dashed line) and a cell directly before performing lysis (marked with red dashed line) are depicted. (D) Lysed cells still appear to be intact cells (red dashed line and black dashed line, respectively) after lysis. These non- growing cells showed no CALv fluorescence and were considered to be metabolically non-active. (E) A non-growing but metabolically active cell after re-supply of iron with elevated mean single-cell fluorescence (marked by blue dashed line).

187 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 42: Metabolic activity sensing under carbon limitation at single-cell level. (A) Microcolony image at the end of the cultivation phase under carbon limitation. Cells that performed no cell division after (figure legend continues on next page▼)

188 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM (▲continued figure legend of Fig. 42) re-supply of carbon are marked with arrows and are framed with dotted lines coloured according to their mean single-cell fluorescence traces in (C). (B) Microcolony image after regrowth under full nutrient supply at the end of cultivation. Cells that performed no cell division after carbon re- supply, are marked with arrows and are framed with dotted lines coloured according to their mean single-cell fluorescence traces in (C). (C) Mean single-cell fluorescence traces of growing and non-growing bacteria before, during and after carbon depletion. Cells, which exhibit cell growth and division (grey lines), are compared to three phenotypes of non-growing cells with alternating mean single-cell fluorescence during famine phase and reduced rate constants of CALv efflux under feast condition (red lines), increased mean single-cell fluorescence during famine phase and after two hours following carbon re-supply (blue lines) and average mean single-cell fluorescence in comparison to normally dividing cells (black lines), respectively.

5.4.6 Temporary Growth Inhibited C. glutamicum Cells due to Antibiotic Exposure

It is actively discussed for severe diseases like tuberculosis and multidrug resistant pathogens, that there is an interfering relation between a transient state of reduced metabolic activity and antimicrobial tolerance or inheritable antibiotic resistance of cells 144,240–243. The metabolic activity sensing based on CvAM was used to compare the bacteriostatic antibiotic chloramphenicol (CHL) and the bactericidal ampicillin (AMP) in their impact on bacterial growth and metabolic activity. CHL inhibits the protein synthesis that impacts among other protein expression the bioneogenesis of enzymes such as esterases, whereas AMP targets the bacterial cell wall synthesis. Thus, the cell wall structure is altered impairing, e.g. cell growth or molecular containment.

189 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Changes in CALv fluorescence of growing bacteria and their subsequently evolving descendants differed according to the applied antibiotic (Fig. 43 A-B, Video 27 and Video 28, Appendix 8.3 and Supplemented CD). Exponentially growing C. glutamicum ATCC 13032 cells were exposed to fatal concentration at 10 µg/mL of CHL and AMP, respectively, for one hour, as shown in Fig. 43. Both antibiotic exposures were followed by an immediate arrest of cell growth (Fig. 43 C). The antibiotic CHL caused a faster change in mean single-cell fluorescence compared to cells exposed to AMP addition. CHL impaired the growth and homogeneity of mean single-cell CALv fluorescence of C. glutamicum ATCC 13032 cells and their descendants less than AMP at identical conditions (Fig. 43 A-B). After cells recovered from CHL stress, the mean single-cell fluorescence increased simultaneously for all bacteria of the colony. AMP caused an immediate increase in the mean single-cell fluorescence. In addition, the growth totally stagnated during the antibiotic treatment of one hour and for the next subsequent two hours (Fig. 43 A, 43 C). In comparison, cells recovered faster from the treatment with bacteriostatic CHL than from contact with bactericidal AMP (Fig. 43 C). Both antibiotics induced heterogeneous mean single-cell fluorescence of subsequent populations after growth disturbance by these antibiotics. AMP additionally caused elongated as well as bifurcated cells.

190 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Figure 43: Metabolic activity sensing of cells exposed to unviable antibiotic concentrations and metabolic activity changes of descendants after antibiotic stress. After an initial growth phase in (figure legend continues on next page▼)

191 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

(▲continued figure legend of Fig. 43) complex medium BHI, cells were exposed to (A) 10 µg/mL ampicillin (cell wall inhibition, n = 5 colonies) and (B) 10 µg/mL chloramphenicol (inhibition of protein synthesis, n = 5 colonies), respectively, for one hour. The antibiotics were added to the perfusion medium during the exposure time and after an hour perfusion with antibiotic free BHI was continued. CALv mean single-cell fluorescence revealed how the antibiotics change the bacterial fitness during antibiotic exposure. (C) The apparent growth rate was determined for five microcolonies treated for one hour with AMP or CHL.

Further, asymmetric cell division happened frequently. Following the increased noise of mean single-cell CALv fluorescence, the esterase activity and CALv efflux were not decreased significantly in average after AMP treatment, but increased in cell to cell variation. In contrast to CHL, some cells reduced their mean single-cell fluorescence during AMP presence remarkably and entered a non-growing state.

5.4.7 Discussion of the Novel Non-Invasive Metabolic Activity Sensing

Intracellular CALv fluorescence is a universal indicator that the cell has active esterases and the energetic capability to perform transport mechanisms to secrete CALv. Therefore, comparative metabolic activity sensing relying on co-factor independent hydrolysis and fluorochrome trafficking could be demonstrated with intermittent deprivation of carbon, iron and the iron chelator PCA. Fluorochrome incorporation, ester conversion, and light exposure were found to be non-invasive for the bacterial growth. The fluorescence signal was not found to be prone to photobleaching or 192 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM photobleading. Both were not the reason for reduction of the mean single-cell fluorescence following carbon re-supply after a starvation phase of 10 – 12 h. The revealed energy dependent CALv efflux of the actinobacteria representative applied in the present work gives rise to the assumption that comparable ATPase activities are present to those reported for cancer cells 85,244. The mechanism of CvAM uptake or CALv secretion by prokaryotes is not completely understood yet, but it is of interest to use CvAM turnover for toxicity assays or multidrug resistance screening of actinobacteria relevant in biotechnological or pharmaceutical fine chemical production as it is already done with cancer cells efflux measurement 85,244 or recently with M. tuberculosis for antibiotic response tests 87,142. We performed a growth perturbation test with the antibiotics AMP and CHL, differing in mode of action, to demonstrate a temporal resolved insight on adaption of bacteria to abrupt antimicrobial exposure. After one hour of antibiotic addition the recovery and regrowth of these stressed bacteria and the impact on their descendants could be shown. The enhanced heterogeneity after AMP contact was supposed to be induced by disturbance of CvAM uptake and CALv transport mechanism or CALv containment due to the antibiotic impact on the cell wall synthesis. CHL impaired mean single-cell fluorescence especially during exposure and shortly after. This can be explained by rather reduced esterase activity than influence of CALv leakage.

193 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Although, there are approaches with CgAM reported for physiology analysis of prokaryotes using FACS in literature 139,233, a comparable time-resolved approach under required environmental stability as presented here and recommended for mammalian cells 85 has not been reported to our knowledge so far. Indeed, a growth-independent antibiotic susceptibility screening of non-culturable surviving M. tuberculosis, incubated in chemostats, was performed by Hendon- Dunn et al. (2016) to analyse once a day the viability using FACS and dual labelling with CvAM and Sytox green 87,142. Nevertheless, there are many methodologies to measure the metabolic activity of prokaryotes. It has to be distinguished wether a quantitative knowledge of a bacterial population is required or comparative information of single cells is necessary. Invasive methods require quenching and/or cell lysis, e.g. intracellular enzyme activity measurement 245 or metabolic profiling 242. These are analyses that provide snapshot insights in dependency of sampling time points. Moreover, these assays are generally very specific for enzymatic reactions and reveal precise information about metabolic pathways in detail. However, these analyses are often sophisticated and labour intensive. For measurements many cells have to be sampled, prepared and lysed. Thus, no information about dynamic changes of intracellular enzyme activity is provided as demonstrated with our comparative fluorescence method.

194 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM Another technical method to test presence of metabolic activity in environmental samples is isothermal microcalorimetry, that is a precise measurement of generated heat by cellular reactions of metabolism 246 and 16s RNA (rRNA) detection by surface plasmon resonance imaging 247. For both methods, it can be stressed that these are real-time measurements, although the result interpretation can be challenging for non-experts 246,247. Non-invasive metabolic activity sensing using bacterial substrates with stable isotopes is also of interest for diagnostic of human pathogens presence in patience. Approaches of stable-isotope breath tests using detection of isotopes of carbon, nitrogen or oxygen as evidence of infections are reviewed in 248. In spite of the requirement, that the metabolic activity analysis has to be non-invasive to the host, the use of stable isotopes has to be performed by specialized scientist and has to be devoted with the relevant resources. However, fluorescence based real-time measurement of metabolic activity enable scientists to determine physiological changes of bacteria in unviable cultivation conditions or metabolic activity changes due to substrate fluctuations. Metabolic key reactions used as detection mechanism are mostly based on redox reaction, respiratory activity, presence of rRNA, cellular energy pool or other enzymatic reactions as reviewed by Hammes et al. (2011) 249. These metabolic activity measurements are either time resolved signals from bacterial suspensions or single-cell resolved snapshots of endpoint reactions.

195 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM

Hence, the ideal prospective method of molecular resolved quantitative metabolic activity of viable and growing prokaryotes has still to be developed. Even though, fluorophore expression systems, which are often based on GFP, have a high potential to be advanced to single-cell real-time monitoring of the metabolic activity, they share the disadvantages of expression heterogeneity and dependency of molecular oxygen for molecule ripening 25,240. Furthermore, the species of interest has to be genetically modified beforehand. Otherwise the use of dyes to prove enzymatic reactions require that the cellular uptake is given and neither product nor fluorogenic substrate cause toxicity for long term observations 250. Further, differential cellular uptake of a fluorogenic substrate can contribute to heterogeneous fluorescence output and has to be considered. Nevertheless, finding an appropriate substrate for microbial metabolic activity sensing can be the needle in the haystack, but as examined with CvAM and C. glutamicum ATCC 13032, it can help to answer the key question active or not in many experimental conditions.

5.4.8 Application of the Metabolic Activity Sensing Method

We applied the temporal resolved metabolic activity sensing based on intracellular fluorescence modulating enzyme reactions with microfluidic cultivation with lack of nutrient components and addition 196 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM of antibiotics to demonstrate the applicability of the method with increasing impact on metabolic activity. Growth of C. glutamicum ATCC 13032 is reported by Liebl et al. (1989) to be stimulated by iron chelators as PCA in batch cultures with minimal medium. 40,251 Siderophore-mediated iron transport in C. glutamicum ATCC 13032 was postulated in literature, but the mechanism and function is not fully stated, yet 252. Metabolic activity sensing and single-cell growth determination showed no impact, if PCA is omitted after 4 hours pre- cultivation with full nutrient supply. PCA might be an initial growth stimulating factor. The influence of iron depletion is more difficult to elucidate 92. For our model organism C. glutamicum ATCC 13032, it is reported, that the regulator of iron homeostasis dtxR controls a range of genes directly or indirectly. These genes are related to, e.g., i.) TCA cycle enzymes (aconitase, succinate dehydrogenase), ii.) hydrogen peroxide decomposing catalase, iii.) iron uptake, iv.) and iron storage 92,123. Iron is an important nutrient for growth and cell division, which were inhibited as long as iron was omitted in the perfusion medium. Nevertheless, CvAM uptake, conversion and CALv efflux were not impaired by iron limitation. Metabolic activity sensing during intermittent iron starvation demonstrated that spontaneously non- growing cells can be distinguished due to intracellular CALv fluorescence in active cells and non-viables. Also, cells appeared to be intact according phase contrast images, disintegrated cells can be

197 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM identified by means of rapid mean single-cell fluorescence loss due to cell lysis. Non-growing cells that arose after intermittent carbon limitation proved to be metabolically active. These non-growing but metabolically active cells appeared in every analysed microcolony under lack of carbon and differed phenotypically according to their mean single-cell fluorescence after carbon re-supply. The resuscitation promoting factor Rpf2 is reported to trigger the regrowth of non-growing starved C. glutamicum ATCC 13032 cells after a switch from famine to feast condition. Rpf2 expression is controlled by GlxR, RamA, RamB in dependency of presence or absence of glucose. GlxR alone is involved in the control of 150 genes of carbon anabolism, catabolism and respiration of C. glutamicum ATCC 13032 93,253. The complex regulation of Rpf2 may explain the heterogeneity of dormant cells observed after carbon limitation. This convinced us to consider further research using metabolic activity sensing on non- growing and recovering cells. Clearly, studies of non-dividing bacteria are challenging due to overgrowth by readily growing viable cells 254. Therefore, we consider the presented method as highly relevant for other bacteria reported developing dormant phenotypes as it is the case for M. tuberculosis that is closely related to C. glutamicum ATCC 13032 87,142,144,238,240. A tremendous increase in the mean fluorescence under carbon limitation supported the conclusion that CALv is secreted energy

198 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM dependent, because in the absence of glucose intracellular ATP was reduced. Another hypothesis for the drastic change in mean fluorescence can be induction of esterase activity and enhanced calcein production. However, increasing extracellular CvAM concentrations did not result in a constant increase of CALv fluorescence. Therefore, a constitutive intracellular esterase activity is applicable. Nevertheless, the CvAM esterase activity of C. glutamicum ATCC 13032 has not been characterized yet and requires more experimental insights e.g. by gene expression analysis. The fluorogenic substrate CgAM is conventionally used to validate viability of mammalian cells 85,231. For cancer cells, an energy- dependent transport of CALg by ATP-binding cassette (ABC) transporters like MRP-1 is reported in literature. These ATPases are related to multidrug resistance of tumour cells to several structurally unrelated cancer therapeutic drugs and were analysed by their potential to secrete CALg 85,244,255. The hydrolysis of CgAM and the efflux of CALg of tumor cells is already demonstrated in literature and commercialized for mammalian viability testing 230 or multidrug resistance potential of tumor cell lines 85. C. glutamicum ATCC 13032 harbours ABC-type multidrug transport systems 256 and bears ABC- type multidrug transporter genes involved in homeostasis 211. Nevertheless, revealing the molecular mechanism of CALv efflux has importance to further develop the metabolic activity sensing with CvAM to a method of bacterial multidrug resistance screening as

199 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM already established for tumor cells 85,244. Thus, drug related bacterial ABC transport mechanism can be considered as a further important application of our metabolic activity sensing method in future. Since effects induced by antibiotics are already demonstrated in relation to actinobacterial ATP metabolism of non-growing and growing phenotypes to elucidate evolving antimicrobial resistance or tolerance 17,87,142,257. The use of metabolic activity sensing with CvAM for antimicrobial testing was demonstrated with an exemplary use of antibiotics AMP and CHL. The antibiotics differ in mode of action and they generated different mean single-cell fluorescence traces of descendants after antibiotic exposure. The increased heterogeneity of mean single-cell CALv fluorescence under AMP addition is partly explained by the impaired cell wall integrity due to bacterial cell wall synthesis inhibition. Furthermore, resistance of C. glutamicum ATCC 13032 to AMP is influenced in contrast to CHL by the expression level of the potential multidrug resistance gene cepA encoding an efflux pump like protein 258. Spatial and temporal resolution as given by the single-cell microfluidic cultivation approach was advanced with non-invasive instantaneous fluorescence imaging. Thus, otherwise hidden changes of metabolic activity after nutrient depletion or exposure to antimicrobials could be made visible. Although, efflux of CALv is not fully understood, yet, the conversion from CvAM to CALv in

200 Non-Invasive Microbial Metabolic Activity Sensing at Single-Cell Level by Perfusion of CvAM combination with a sampling free microfluidic approach a powerful tool to gain new insights in the metabolic activity of growing and non- growing bacteria. Non-growing, dormant or resistant cells exhibit a large potential for bacterial survival of antimicrobial substances such as antibiotics.

201 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

5.5 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

The following text has been published in Krämer, C., Wiechert, W. and Kohlheyer, D. (2016), Sci Rep 6: 32104. D. Kohlheyer and W. Wiechert proofread the manuscript.

Chapter Abstract: Conventional propidium iodide (PI) staining requires the execution of multiple steps prior to analysis, potentially affecting assay results as well as cell vitality. In this study, this multistep analysis method has been transformed into a single-step, non-toxic, real-time method via live-cell imaging during perfusion with 0.1 µM PI inside a microfluidic cultivation device. Dynamic PI staining was an effective live/dead analytical tool and demonstrated consistent results for single-cell death initiated by direct or indirect triggers. Application of this method for the first time revealed the apparent antibiotic tolerance of C. glutamicum ATCC 13032 cells, as indicated by the conversion of CvAM. Additional implementation of this method provided insight into the induced cell lysis of Escherichia coli cells expressing a lytic toxin-antitoxin module, providing

202 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion evidence for non-lytic cell death and cell resistance to toxin production. Finally, our dynamic PI staining method distinguished necrotic-like and apoptotic-like cell death phenotypes in Saccharomyces cerevisiae among predisposed descendants of nutrient-deprived ancestor cells using PO-PRO-1 or CgAM as counterstains. The combination of single-cell cultivation, fluorescent time-lapse imaging, and PI perfusion facilitates spatiotemporally resolved observations that deliver new insights into the dynamics of cellular behaviour.

5.5.1 Introduction

“Alive or dead?”, “How dead is dead?” or “How red is dead?” are pivotal questions posed during cellular live/dead determination, particularly when in vivo staining is performed with PI. Although PI is a common cell death indicator, a gold standard protocol for its use does not exist, and inconsistent staining results and pitfalls have been reported in the literature 259–264. PI is a versatile indicator dye for dead cells that acts by intercalating with cellular DNA and emitting red fluorescence. Vital staining with PI is dependent on the impermeability of an intact cell membrane to this molecule. Live/dead staining with PI is commonly implemented to evaluate the viability of bacteria sampled from food products,

203 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion clinical samples, and environmental or fermentation processes and to characterize vitality in eukaryotic cells 231,259,265. This staining procedure has been employed for bacteria 260,261, biofilms 266, yeasts 259, and a variety of mammalian cells 267. However, the toxicities of fluorescence indicators or certain concentrations are rarely considered. Microscopic imaging approaches employing microfluidic devices containing cells prestained with PI and cell-wall permeant SYTO 9 have been reported for the live/dead quantification of bacterial cells 268–270, sperm cells 271, and yeast 272 and are, in principle, comparable to studies using FACS. Conventional staining protocols using PI concentrations higher than 1 µM intended for sorting 262,271, confirmation of cell lysis 273, or cellular analytics 19,86,274,275 have been described for prokaryotes and eukaryotes. PI staining is generally performed as an endpoint measurement, frequently after cell fixation 19,275,276. PI is often, but not exclusively, used in combination with SYTO 9 as a counterstain 260,262,263,277. PI is also combined with other cell- permeable DNA dyes, such as other SYTO dyes (e.g., SYTO 15 and SYTO 13) 19,278, acridine orange 275,279, SYBR green 264 and SYTOX dyes (e.g., Sytox Red and Sytox Green) 86,280 to facilitate total cell staining. Alternatively, prokaryotic viability can be accessed via bacterial GFP expression prior to PI staining in lieu of a total cell stain 19,260,266. Furthermore, PI is used in combination with the monomeric

204 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion cyanine dyes PO-PRO-1 and YO-PRO-1 86, the green fluorogenic esterase substrate CgAM 231,274,276,281, a fluorescent caspase inhibitor 282, annexin V 283, or specific overall stains (e.g., ConA-Alexa Fluor 488 284, FITC-dextran 267, and CMFDA 276). In addition, cancer cells have been continuously perfused with PI to demonstrate the efficiency of live cell trapping in a media stream or to indicate the need for cytotoxicity testing with microfluidic devices 86,285. We present a dynamic and non-invasive cell viability staining method employing a low PI concentration in combination with non-toxic counterstains, provided continuously to bacteria or yeast in a microfluidic growth chamber. Staining experiments utilized a microfluidic PDMS-glass device designed for single-cell studies of C. glutamicum and Escherichia coli 16,36,46. This cultivation approach, in combination with time-lapse fluorescence microscopy, confers spatiotemporal insight into the phenotypic variations and dynamics of cellular death at the single-cell level. In particular, the dynamic heterogeneity of cell phenotypes, such as differences in tolerance, resistance, or epigenetic predisposition among isogenic cultures, can be analysed during well-controlled perturbation studies. Exogenous and endogenous cell death triggers were applied to test our ability to perform temporally resolved cell death analysis of C. glutamicum, E. coli and Saccharomyces cerevisiae, emphasizing the generic analytical approach. PI invasion of fungal and bacterial cells in

205 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion combination with non-toxic counterstains CvAM, CgAM, and PO- PRO-1) is indicative of both sudden and prolonged cell death. Thus, the immediately apparent fluorescence responses of this novel and dynamic staining approach facilitate the elucidation of survival strategies in small cell populations, such as antibiotic tolerance, temporal resistance, resuscitation after membrane potential loss, or the formation of membrane permeability transition pores (MPTPs). The technical approach presented here permits the temporally resolved observation of phenotypic variations at the single-cell level. The spatial resolution of intracellular fluorescence distribution enables the visualization of specific details regarding cell death, such as the partial death of cell poles after antibiotic contact.

5.5.2 PI Concentration Optimization and Continuous Viability Staining Validation

In the present study, the conventional viability assay employing PI and SYTO 9, a multi-step method 277, was transformed into a single-cell, one-step method resolved in real-time with microbial cultivation occurring inside a microfluidic device. This microfluidic perfusion system ensures the continuous presence of extracellular PI at a specific concentration for all cells during experiments.

206 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion The dynamic staining of living microorganisms under different stress situations was analysed in a microfluidic perfusion device. This microfluidics approach advanced conventional staining through the use of a reduced PI concentration that did not affect single-cell growth. Simultaneous validation utilizing non-toxic counterstaining and spatiotemporal resolution permitted us to analyse the emergence of lethal cell wall damage, cell lysis with DNA discharge, or the bipolarity of cell-division–inhibited cells. Data analysis has not yet been fully automatized. The microfluidic device, employed for our dynamic staining studies, permits control of the bacterial microenvironment as demonstrated experimentally and through CFD simulations in previous studies 36,38– 40. The cultivation chamber arrays lack multiple channel inlets for sink and supply settings and it has no gradient mixer in the upstream region, similar to previous designs for concentration gradient simulation in large-scale cultivation reactors 9,34. Continuous medium supply and nutrient renewal are present in the described microstructures. The microstructures and the environmental control of the microfluidic perfusion system used in this work were described in detail previously 36,38,40. A continuous nutrient supply and waste-product removal were implemented in studies examining the roles of glucose, a major carbon source, as well as a minor carbon

207 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion source (the iron chelator PCA) during C. glutamicum cultivation 39,40. Table 3: LogP values of fluorescence indicators and inducer Substance LogP Source calcein (blue) -1.8 (pred.) PubChem 286

calcein (green) -4.04 (pred.) ChemSpider 287 calcein (violet) Not given -

calcein acetoxymethyl 1.1 (pred.) PubChem 286 ester (blue)

calcein acetoxymethyl 2.9 (pred.) PubChem 286 ester (green)

IPTG (Isopropyl β-D-1- -1.26 (exp.), ChemSpider 287 thiogalactopyranoside) -1.25 (pred.)

PO-PRO-1 -6.29 (pred.) ChemSpider 287 propidium iodide -4.64 (pred.) ChemSpider 287

For our dynamic staining studies, chemicals and dyes were selected based on their hydrophilic characteristics, water solubility, and LogP

208 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion values (LogP < 2) (Tab. 3). The LogP is the partition coefficient and describes the logarithmic concentration ratio of a solute dissolved in water and octanol. The cultivation chambers were randomly chosen and analysed. Clustering of fluorescence data was not observed in the transverse direction of the microchamber arrays for multiple parallel channels, in the flow direction of the microchambers, or over time for any experiment. Final single-cell fluorescence values of antibiotic-treated C. glutamicum cells are shown in a directional distribution for the microchamber array (Fig. 44). The gram-positive bacterium C. glutamicum was cultivated with minimal medium (CGXII + 4 % glucose (w/v) without PI) and used as the reference for three different PI concentrations (0.1 µM, 1 µM, and 10 µM). C. glutamicum growth was impaired by 10 µM PI. PI permeated and slightly stained intact cells, but these bacteria continued to grow, although at a reduced rate. Bacterial growth was unimpaired by concentrations of 0.1 or 1 µM PI (Fig. 45 A). However, positively stained cells (PI+) were observed at frequencies of < 0.01 % for all three PI concentrations due to spontaneous single-cell death. Based on these data, a PI concentration of 0.1 µM was employed for our microfluidic analyses and validated by the addition of phenol during C. glutamicum cultivation as described consecutively (Fig. 46). Furthermore, 0.1 µM PI was found to be non-toxic and universally

209 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion applicable, as revealed by testing a wide range of microorganisms cultivated in different complex media, including Micrococcus luteus (1.78 % PI+), Bacillus subtilis (0.09 % PI+), E. coli (< 0.01 % PI+), Vibrio harveyi (< 0.01 % PI+) and the yeast S. cerevisiae (2.72 % PI+) (Fig. 45 B). The tested microbes were selected for their diverse cell-wall structures and taxonomic variations. Independent of cell-wall structure, membrane disintegration was instantaneously observable during cultivation. Compared to reference cultures, microorganismal growth was not influenced by the addition of 0.1 µM PI. However, a negligible fraction of cells was PI+ directly following inoculation and at the end of cultivation when the cultivation chambers were nearly or completely filled with cells (Fig. 45 C-I). Classification criteria based on fluorescence signals are described in detail in the Material and Methods section. An M. luteus tetrad with a dead coccus and a dead E. coli cell obtained from the pre-culture directly after seeding are shown in Fig. 45 E and 45 G, respectively. PI+ cells randomly distributed in culture at the end of cultivation are shown for C. glutamicum (Fig. 45 C), M. luteus (Fig. 45 D), B. subtilis (Fig. 45 F), V. harveyi (Fig. 45 H), and S. cerevisiae (Fig. 45 I).

210 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 44: Single-cell distribution and fluorescence of PI and CALv in antibiotics-treated C. glutamicum ATCC 13032 cells in the microfluidic chamber array. (A) Microfluidic channel network and growth chamber array. (A*) Enlarged view of the growth chamber array. Final distributions of single-cell fluorescence in the flow direction for (B) PI and (D) CALv and in the transverse direction for (C) PI and (E) CALv.

211 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 45: Determination of optimal propidium iodide concentration. (A) The model organism Corynebacterium glutamicum was stained continuously with 0.1 µM PI, 1 µM PI, and 10 µM PI, and bacterial (figure legend continues on next page▼)

212 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion (▲continued figure legend of Fig. 45) growth was normalized to the growth rate without PI addition. Total cell numbers are indicated with N. PI+ dead cells are marked by white arrows. (B) A PI concentration of 0.1 µM was used with gram-positive bacteria (C. glutamicum ATCC 13032, Micrococcus luteus DSMZ 14234 and Bacillus subtilis 168), gram-negative bacteria (Escherichia coli MG1655 and Vibrio harveyi ATCC 33867), and a small eukaryote (Saccharomyces cerevisiae). The growth rates of all microorganisms cultivated with 0.1 µM PI were normalized and compared to reference colonies grown without PI. Total cell numbers are given by N. (C) C. glutamicum ATCC 13032 colony with a single PI+ cell. (D) M. luteus DSMZ 14234 colony in the late exponential phase with distributed PI+ cocci. (E) M. luteus DSMZ 14234 tetrad with the early appearance of a PI+ cell. (F) Densely grown B. subtilis 168 cell colony with the late appearance of a PI+ cell. (G) Early appearance of a PI+ E. coli MG1655 cell. (H) Segmented V. harveyi ATCC 33867 PI+ phenotype in a cell-packed region. (I) Dense S. cerevisiae colony with PI+ yeast cells.

Additionally, dynamic PI staining was validated by adding a disinfectant to the perfusion medium. Conventionally, PI viability analyses are validated with cell suspensions treated with isopropanol or ethanol 260,264. Both isopropanol and ethanol are able to penetrate the device material PDMS and diffuse from channels containing high organic solvent concentrations to neighbouring solvent-free channels through the wall of the microfluidic device 288. Therefore, we used phenol at concentrations ranging from 0.64 µM to 64 µM in CGXII + 4 % GLC, rather than ethanol or isopropanol, to avoid material interactions. A phenol concentration of 6.4 µM or less was sub-lethal, whereas 64 µM was lethal for all imaged cells (N = 130 cells) after 10 h. Only two cells exhibited marginal cell growth before cell division stopped. After 10 h of treatment with 64 µM phenol, all C. glutamicum cells exhibited red PI fluorescence (exemplary cell shown in Fig. 46).

213 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion A positive control involving additional PO-PRO-1 staining during cyanide intoxication confirmed PI as rapid and precise cell death detection system during cultivation (Fig. 47). A “gold standard” for the non-disintegrative cell death of a broad variety of microorganism is difficult to define. For aerobic and facultative anaerobic microorganisms, we tested the addition of a lethal sodium cyanide concentration (50 mg/mL) to inhibit oxidative metabolic reactions in combination with the membrane potential loss indicator PO-PRO-1 in the cultivation media for C. glutamicum, M. luteus, B. subtilis, E. coli, V. harveyi, and S. cerevisiae. After membrane potential loss was indicated for a substantial number of cells, dynamic PI staining was initiated. Gram-positive bacteria had a signal-to-background ratio (S/B) between 1.78 and 1.84. Gram-negative bacteria demonstrated an S/B ranging from 2.36 to 2.62, and the analysed yeast strain had an S/B of 4.73 (Fig. 47 A). The signal-to-noise ratio (S/N) varied by type (cell size, cell wall thickness, DNA content), ranging from 37.4-fold (E. coli) up to 901.5- fold (S. cerevisiae) (Fig. 47 A). The cyanide incubation time reflected the cyanide resistance of individual cells (Fig. 47 B). For the tested gram-positive bacteria and the yeast strain, the dead-cell ratios were identical. E. coli and V. harveyi achieved 84.0 % and 110.7 % dead cells (PI+), and some cells lost their membrane potential due to substantial cell lysis accompanied by DNA loss (E. coli), cell

214 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion disintegration caused cell loss (V. harveyi) and ongoing cell death (V. harveyi) (Fig. 47 B).

Figure 46: Toxicity test with C. glutamicum ATCC 13032 and phenol. Three phenol concentrations (0.64 µM, 6.4 µM, and 64.0 µM) were continuously perfused along with the medium (CGXII + 4 % GLC). (A) Images of the reference and all tested phenol concentrations are shown 2.2-2.4 h after the start of the experiment and after 11.8-12.0 h for comparison. (B) The growth rate, normalized to the untreated reference, was substantially reduced with 6.4 µM phenol, but 64.0 µM phenol almost completely halted cell growth and resulted in PI+ cells after more than 8 h. Final total cell numbers are given by N. Standard deviation is indicated by error bars.

215 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 47: Validation of dynamic PI staining. Final total cell numbers are given by N. (A) Signal-to-noise ratio over signal-to-background ratio. (B) Ratio of PI+ cells to PO-PRO-1+ cells during incubation with a lethal cyanide concentration required for substantial cell death.

5.5.3 Instantaneous Cell Death Monitoring and Non-Toxic Counterstaining

A counterstain was employed to ensure that non-growing, PI-negative (PI-) cells were intact and viable cells exhibiting growth inhibition. A non-invasive counterstain for microfluidic cultivation must meet several criteria, such as a sufficient signal-to-noise ratio, no inhibition of growth and division (toxicity), cellular uptake and fluorochrome 216 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion containment, no adhesion or penetration of the channel wall material, stability during experiments, and no overlap of excitation or emission spectra with PI. SYTO 9, which is used in the BacLight™ assay, stains the DNA of all cells. Given that SYTO 9 impairs cell growth during microfluidic cultivation and has been reported to be somewhat toxic 289, it was not used. Acridine orange is considered harmful because it induced a 40 % growth rate reduction in C. glutamicum due to DNA-dye interactions. Furthermore, CTC was tested with E. coli MG1655 and found to inhibit cell growth (Fig. 1 E), which is consistent with findings reported in the literature 250. The CvAM met all criteria for C. glutamicum, as shown recently by Krämer et al. (2015) 16, as well as for eukaryotic cells 85. Violet fluorogenic CvAM (CvAM) was taken up by C. glutamicum cells and converted by intracellular esterases to violet fluorescent calcein, which was secreted in an energy-dependent manner. Energy limitation is assumed to cause an increase in CALv fluorescence due to intracellular accumulation 16.

217 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 48: Comparison of bacterial growth under reference conditions in growth medium, growth medium with PI, and growth medium with PI and counterstain. (A) The mean growth rates of at least (figure legend continues on next page▼)

218 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion (▲continued figure legend of Fig. 48) 10 colonies cultivated with minimal medium (CGXII + 4 % glucose (w/v)) (reference), minimal medium with 0.1 µM PI, and minimal medium with 0.1 µM PI and 46 µM CvAM demonstrate that the addition of PI and CvAM is not detrimental to C. glutamicum cell growth. Final total cell numbers are given by N. Standard deviation is indicated by error bars. (B) The mean fluorescence values of PI (red) and the fluorogenic substrate CvAM, used as a counterstain (blue), are shown along with the total cell numbers of all analysed cultivation chambers over time. The PI fluorescence did not change considerably over time because PI-positive cells only appear rarely under reference conditions. CvAM was taken up by all viable cells and converted to fluorescent CALv. Standard deviation is indicated by error bars. (C) The mean growth rates of at least 3 colonies cultivated with the complex medium BHI (reference), BHI with 0.1 µM PI, and BHI with 0.1 µM PI and PO-PRO-1 demonstrate that the addition of PI and CvAM is not detrimental to B. subtilis cell growth. Final total cell numbers are given by N. Standard deviation is indicated by error bars.

C. glutamicum growth was not impaired by the addition of PI or the simultaneous addition of PI and CvAM (Fig. 48 A). The mean CvAM and PI fluorescence profiles of C. glutamicum colonies are shown in Fig. 48 B. The formation of intracellular oxygen radical species due to phototoxicity during multiplexed fluorescent time-lapse imaging was tested with DHCAM as described previously 16. DHCAM addition indicated no phototoxicity under excitation conditions for imaging (Fig. 49 A). The loss of fluorescence upon repeated light exposure was found to be marginal for PI (Fig. 49 B), which was shown previously for CALv 16.

219 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 49: Phototoxicity and photobleaching analysis. (A) The phototoxicity of time-lapse imaging was tested with C. glutamicum and DHCAM. DHCAM is taken up by C. glutamicum cells as CvAM and converted intracellularly to DHCALox. DHCALox is oxidized to green fluorescent CALox if light-induced oxygen radicals are present in bacterial cells. The phototoxicities of phase contrast time-lapse imaging and fluorescent time-lapse imaging (phase-contrast imaging and excitation of PI and CAM) were analysed. The total cell number is indicated by the grey (phase contrast imaging) and black lines (phase-contrast and fluorescence imaging). The CALox fluorescence of every measured cell is indicated by green (phase contrast light exposure) and red-blue dots (phase contrast and fluorescence light exposure). After 10 h of fluorescent time-lapse imaging, one cell out of more than 500 exhibited an increase in mean single-cell CALox fluorescence. (B) The photobleaching ability of PI was evaluated using C. glutamicum cells pre-treated with ampicillin. The mean loss of single-cell PI fluorescence was found to be 2.3 ± 0.6 % per time-lapse imaging time point.

220 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion No intracellular calcein fluorescence of E. coli MG1655 or B. subtilis 168 was observable with continuous CvAM or green fluorogenic CgAM perfusion during microfluidic cultivation (data not shown). Therefore, PO-PRO-1 was tested with B. subtilis 168 and found to be a non-invasive counterstain for PI validation (Fig. 48 C). For S. cerevisiae, multiplexed fluorescent time-lapse imaging with PI, PO- PRO-1 and calcein derivatives CgAM and CbAM was satisfactory for growth (Fig. 50).

Figure 50: Comparison of yeast growth under reference conditions with growth medium, growth medium with PI, and growth medium with PI and counterstain. CALg and CALb were tested as counterstains. The mean growth rates of at least 10 colonies were analysed for each condition. Final total cell numbers are given by N. Standard deviation is indicated by error bars.

221 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion 5.5.4 Prokaryotic Cell Death and Apparent Antibiotic Tolerance Following the Addition of Antibiotics

On-line viability monitoring with PI was performed for the non- pathogenic organism C. glutamicum, which is related to various human pathogens (e.g., Mycobacterium tuberculosis and Corynebacterium diphtheriae) 256, in the continuous presence of environmental antibiotic concentrations. Dynamic PI staining was validated by the addition of CvAM as a non-invasive counterstain, which is converted intracellularly to CALv, as described in the previous chapter and recently in 16. Viable cells exhibited moderate CALv fluorescence, whereas tolerant cells with reduced metabolism were intensely fluorescent 16. Dual staining had no impact on growth (Fig. 48), and additional excitation during multiplexed fluorescent time-lapse imaging was tested and shown to be non-phototoxic (Fig. 49). C. glutamicum was cultivated during the perfusion of constant concentrations of ethambutol (EMB), ampicillin (AMP), kanamycin (KAN), streptomycin (STR), and chloramphenicol (CHL). These antimicrobials are categorized as bactericidal inhibitors of cell-wall synthesis (EMB and AMP), bactericidal aminoglycoside antibiotics causing mRNA misreading and inhibit protein biosynthesis (KAN, STR), and a bacteriostatic protein synthesis inhibitor (CHL) 290.

222 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion Heterogeneous fluorescence distribution attributable to concentration gradients was not observed (Fig. 44). The cells continued asymmetric division for more than 9 hours with 25 mg/mL EMB, whereas division halted after the second filial generation with 25 µg/mL AMP and after the first cell division with 25 µg/mL KAN and 50 µg/mL STR. Cell division ceased after initial antibiotic contact with 50 µg/mL KAN and 50 µg/mL CHL (Video 29, Appendix 8.4 and Supplemented CD). The mean PI fluorescence was found to be lower following the addition of cell wall synthesis inhibitors and 25 µg/mL KAN than with the addition of antibiotics that impair protein biosynthesis at a concentration at 50 µg/mL (Fig. 51). The mean CALv fluorescence was increased by the presence of an antibiotic (Fig. 51) relative to untreated colonies (Fig. 48 B). The striking variation in the standard deviation is primarily attributable to the microbial phenotypic heterogeneity of bacteria that die, are lysed, or temporally tolerate the presence of potentially fatal antibiotic concentrations. Furthermore, variation in the mean single-cell fluorescence is likely attributable to differences in fluorochrome containment in antibiotic-affected cells. The cells were enlarged by cell wall synthesis inhibitors, whereas EMB resulted in the asymmetric division of bulky cells with sharp

223 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 51: Continuous monitoring of cell wall integrity and metabolic activity of wild-type C. glutamicum ATCC 13032 colonies treated with different antibiotics using PI (red) and CvAM (blue). The total cell numbers of at least 6 colonies (black solid line) and the mean PI and CALv fluorescence values are indicated along with the standard deviation of all cells over time. (A) Continuous treatment with the cell wall synthesis inhibitor EMB (25 µg/mL). (B) Continuous treatment with the cell wall synthesis inhibitor AMP (25 µg/mL). (C) Continuous treatment with 25 µg/mL KAN, which triggers mRNA misreading. (D) Continuous treatment with 50 µg/mL KAN, which triggers mRNA misreading. (E) (figure legend continues on next page▼)

224 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion (▲continued figure legend of Fig. 51) Continuous treatment with 50 µg/mL STR, which triggers mRNA misreading. (F) Continuous treatment with 50 µg/mL CHL, which inhibits protein synthesis.

Figure 52: Fractions of different cell states following continuous antibiotic treatment of C. glutami-cum ATCC 13032 cells at 25 µg/mL for 12 h and at 50 µg/mL for 16.6 h. Dead cells are identified by a significant increase in PI fluorescence (PI+/CALv-), antibiotic-tolerant cells are identified by the conversion of CvAM to CALv and its retention (PI-/CALv+), lysed cells are non-fluorescent and pale in phase contrast images (PI-/CALv-), segmented cells are bipolar with a dead pole and a tolerant pole (PI+/CALv-/PI-/CALv+). Growing cells were defined as non-inhibited with respect to cell elongation for C. glutamicum, whose cells typically undergo snapping cell division. The bactericidal antibiotics EMB and AMP are inhibitors of cell wall synthesis. The bacteriostatic mRNA inhibitor KAN was tested at 25 µg/mL and 50 µg/mL. STR and the bacteriostatic CHL inhibit protein biosynthesis.

poles, and AMP led to elongated cells with apical enlargement (Fig. 53 A-B). KAN and STR induced cell rounding and accounted for a fraction of the lysed cells (Fig. 53 C-E). KAN also initiated total cell 225 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion disintegration. CHL halted cell division, resulting in segmented cells independently exhibiting metabolic activity or viability (Fig. 53 F). The bactericidal antibiotics EMB and AMP inhibit cell wall synthesis and its repair mechanisms. Both antibiotics exerted a remarkable impact on C. glutamicum cells (Fig. 52 A, Fig. 53 A-B, and Fig. 51 A- B). Furthermore, both antibiotics resulted in reduced intracellular fluorescence compared to the reference (Fig. 48 B) and to other fluorescent cells treated with KAN, STR, or CHL (Fig. 53 C-F) due to the compromised intracellular retention of molecules by leaky cell walls. The bacteriostatic antibiotics KAN, STR, and CHL inhibit bacterial protein biosynthesis. In addition to cell death and cell lysis (Fig. 52 B- C), these bacteriostatic antibiotics resulted in substantial increases in CALv fluorescence over time (Fig. 51 C-F) compared to non-treated reference cells (Fig. 49 B). Fractions of the total numbers of cells treated with KAN, STR, or CHL switched to an antibiotic-tolerant state accompanied by residual enzymatic activity (conversion of CvAM to CALv) and increased CALv retention (Fig. 52 and 53). Cell growth was impaired via different mechanisms, resulting in phenotypic variation, as shown in Fig. 52-53. The status of antibiotic- treated cells was altered, as indicated by intracellular fluorescence. In contrast to untreated cells (Fig. 52, no antibiotics), the addition of antibiotics resulted in formation of a subpopulation of highly CALv-

226 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

227 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion Figure 53: Antibiotic-induced cell death and antibiotic tolerance of wild-type C. glutamicum ATCC 13032. Mean single-cell fluorescence traces are shown for single cells from one representative growth chamber stained continuously with the cell death indicator PI (right) and the metabolic activity indicator CALv (middle). The micrographs (right) show representative cells from the final time-lapse images. Dead (PI+/CALv-), lysed (PI-/CALv-), antibiotic-tolerant (PI-/CALv+), or segmented cells with a dead cell pole and a surviving cell pole (PI+/CALv-/PI-/CALv+) can be distinguished. Cells were continuously treated with (A) 25 µg/mL EMB (a deformed cell that retained cell wall integrity is marked with *), (B) 25 µg/mL AMP (a deformed cell that partially retained CALv fluorescence is marked with *), (C) 25 µg/mL KAN (a segmented cell with a cell pole that is PI+ and a PI- cell pole with CALv fluorescence is marked with *), (D) 50 µg/mL KAN (heterogeneous PI+ cells with bright PI fluorescence (*) and pale PI fluorescence (**) are marked), (E) 50 µg/mL STR (dead, lysed cells with rapid PI fluorescence loss (*) and dead cells with constant PI fluorescence (**) are compared), (F) and 50 µg/mL CHL (segmented cells with halted growth and independently stained cell poles; two cells are marked, one with a PI+ pole and a blue fluorescent PI- pole (*) and another with two cell poles exhibiting CALv fluorescence (**)). fluorescent non-growing cells (PI-/CALv++) that were considered antibiotic tolerant as well as subpopulations of dead PI+ cells (PI+/CALv-) and non-fluorescent cells (PI-/CALv-) that lost their intracellular content following lysis. Fractions of dead and lysed cells differed according to the antibiotic applied and its impact mechanism on cell wall synthesis or translational processes (Fig. 52). Mean colony fluorescence with a high standard deviation confirmed changes in individual fluorescence profiles based on antibiotic treatment (see Fig. 53, Fig. 51). Bacteria with injured cell walls lost intracellular CALv fluorescence while PI intruded and underwent DNA intercalation (Fig. 54). The mean single-cell PI fluorescence equilibrium differed according to the antibiotic. Cell wall-impairing antibiotics (EMB and AMP) resulted, by far, in the lowest mean

228 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion single-cell PI fluorescence values (Fig. 51 and Fig. 53). Heterogeneous PI+ cells were observed with 50 µg/mL KAN (Fig. 53 D, cells marked with *). Bacterial growth arrest was not a specific indicator of cell death, as several cells remained unstained by PI (PI-) even after growth halted. Residual CALv fluorescence revealed bacterial survival among cells subjected to treatment with all six antibiotics, even after 16 hours (apparent antibiotic tolerance). Thus, PI fluorescence indicates bacteria that are permeabilised by an antibiotic. Cell membrane disintegration and a concurrent increase in PI fluorescence differed among cells by antibiotic (Fig. 53). PI+ cells exhibited increased maximum mean single-cell fluorescence values of 1000-1500 AU. PI+ cells continuously cultivated at 25 µg/mL AMP (Fig. 53 B, *), 50 µg/mL STR (Fig. 53 E, *) or 50 µg/mL CHL (Fig. 53 F, *) showed subtle decreases in mean single-cell fluorescence over time. Although the fluorescence profiles of these cells resembled a Bateman function, that describes concentration dependent uptake and decay, fluorescence loss was more rapid than that induced by photo bleaching, which accounted for 1 % to 3 % of the total mean single-cell PI fluorescence across all imaging frames (Fig. 49). Lysed cells that retained their cell shape as visible ghost cells demonstrated rapid reductions in intracellular PI and CALv

229 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion fluorescence due to DNA loss and cell membrane destruction (Fig. 53 E, cell marked with *, Fig. 54). In contrast, the PI fluorescence values of dead bacteria rapidly reached high and stable plateaus (Fig. 53 E, cell marked with **). Reduced PI fluorescence in these cells correlated with decreased DAPI signals as determined by additional endpoint staining of total DNA (Fig. 54), indicating possible DNA decay or fractional DNA loss. Furthermore, the apparent fractional tolerance of segmented cells exhibited two different viable states. In addition to fully living cells, segmented cells with dead cell poles that were PI+ and cell poles that retained cell wall integrity were observed in the presence of 50 µg/mL CHL (Fig. 53 C and 53 F, cells marked with * and **, respectively). However, a fraction of the bacteria remained PI- while demonstrating remarkably increased CALv fluorescence (CALv++). These cells did not stain red or lyse and tolerated continuous antibiotic treatment in a non-growing but metabolically active state during the observed time frame (Fig. 53 A-B, cells and single-cell fluorescence traces marked with *). In addition, an endpoint staining experiment was performed. Therefore, the cell state of apparent antibiotic tolerant cell survival and cell division arrested segmented C. glutamicum cells with the possibility of bipolar viability appearance have been tested by staining

230 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion with DAPI (cell permeant DNA indicator) and nile red (biomembrane indicator), respectively. C. glutamicum cells have been treated with 50 µg/mL CHL as cells presented in Fig. 51, Fig. 52, and Fig. 53. In spite of dual staining, cells were continuously stained by CvAM or PI addition to the perfusion medium. After comparable long continuous treatment of the cells with CHL as in previous experiments, cell fixation with paraformaldehyde and endpoint counterstaining was performed with nile red (additionally to CALv fluorescence) or DAPI (additionally to PI fluorescence). Representative cells are given for CALv/nile red stained cells (Fig. 54 A) and PI/DAPI stained cells (Fig. 54 B), respectively. Intact cell membranes, indicated by nile red, correlated with enzymatic converted CvAM with reduced active CALv efflux. Cells with destroyed lipid bilayer were considered as lysed cells, which are non-fluorescent cells (Fig. 54 A). Furthermore, lysed cells showed to lack DNA as indicated by non-fluorescent cells in Fig. 54 B. Whereas, cells with compromised cell wall integrity were stained with PI and DAPI.

231 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 54: Validation of the dynamic dual counterstaining with PI and CvAM by endpoint staining of C. glutamicum cells treated 48 h with 50 µg/ml CHL with the lipid membrane indicator nile red or total DNA staining with DAPI. (A) Representative CHL treated C. glutamicum cells after nile red endpoint staining and PFA fixation. CvAM was supplied continuously to the cells during CHL treatment. Nile red indicates intact lipid biolayers (pink). CALv fluorescence (blue) is produced when CvAM is hydrolysed and the fluorescent product is retained intracellularly. (B) Representative CHL treated C. glutamicum cells after DAPI endpoint staining and PFA fixation. PI (red) was supplied continuously to the cells during CHL treatment. DAPI (light blue) is a cell permeant total DNA stain.

232 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion 5.5.8 Programmed Cell Death (PCD) of E. coli

Toxin-antitoxin modules are present in a wide range of bacteria. Their functions and triggers are currently under intensive investigation to determine their potential antimicrobial use. In the present study, we performed dynamic PI staining and undertook the time-resolved observation of a pneumococcal zeta toxin (PezT) described by Mutschler et al. (2011) 75. E. coli BL21CodonPlus(DE3)-RIL bearing pET28b(pezTΔC242) (Fig. 55 A), pET28b(pezA/pezT) (Fig. 55 B), or pET28b(pezTΔC242(D66T)) (Fig. 55 C) expressed an inactivated toxin, the C-terminal truncated toxin or the antitoxin-toxin complex, respectively, after induction with 100 µM IPTG after 2.6 h of cultivation (Video 30, Appendix 8.4 and Supplemented CD). Cells underwent lysis after expressing the truncated toxin, which impairs cell wall synthesis. Extracellular DNA was stained immediately and exhibited red fluorescence. Non-lysed PI+ cells were observed when toxin-producing mutants were cultivated but were rarely observed (at a frequency of 1.4 %) during cultivation of the strain expressing the antitoxin-toxin complex (Fig. 55 E, Fig. 55 F, yellow arrow, and (Video 30, Appendix 8.4 and Supplemented CD). Thus, cell death occurred independent of induced toxin production. Rare cells containing the truncated toxin-bearing plasmid remained in a non-replicating PI- state at a frequency of 1.5 % (Fig. 55 F). These

233 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion resistant cells did not die and were not lysed during 5.6 h of IPTG treatment (Fig. 55 F, white arrow). Recovery was not achieved by reverting to a LB medium without the inducer more than 4 h after IPTG induction. Resuscitation was attempted until all resistant cells became PI+ (data not shown). Continuous PI fluorescent time-lapse imaging enabled us to distinguish between toxin-induced lytic cell death, incidental non-lytic cell death and toxin resistance. Prior to the induction of toxin expression, single cells exhibited PI fluorescence and were overgrown by non-fluorescent bacteria until toxin production began (Fig. 55 C). Therefore, the percentage of live cells increased to almost 100 % and then decreased subsequent to toxin production.

234 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 55: PI to determine the bacterial survival rate following toxin-antitoxin module expression in E. coli BL21CodonPlus(DE3)-RIL. Strains harbouring the plasmids pET28b(pezTΔC242(D66T)), pET28b(pezA/pezT), or pET28b(pezTΔC242) produced (A) an inactivated toxin, (B) the antitoxin-toxin complex or (C) the C- terminal truncated toxin. Micrographs showing an E. coli colony with (D) inactivated toxin expression, (E) a colony of cells before and after expression of the antitoxin- toxin module, and (F) a colony expressing the truncated toxin, which caused cell lysis, are presented. Cells stained prior to IPTG induction (2.6 h) or prior to lysis are indicated with a yellow arrow. A white arrow indicates a cell that resisted toxin expression but did not resuscitate following a backshift to growth medium.

235 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion 5.5.5 Temporally Resolved Programmed Cell Death (PCD) in Yeast

Baker’s yeast is a simple model organism used to study apoptotic phenotypes and lethal cell differentiation. Given that starvation induces cell death and PCD in S. cerevisiae 291, we analysed yeast cells during microfluidic cultivation with a fresh supply of YPD medium combined with dynamic dual staining using PI/CgAM or PI/PO-PRO- 1 after pre-cultivation in a shaking flask under famine conditions. Famine conditions were initiated by (i.) medium replacement with 0.9 % NaCl (w/v) (starvation conditions) and (ii.) prolonged pre- cultivation in YPD (nutrient limitation conditions). In contrast to the rapid necrosis, that occurred among yeast cells in- between the 30-min imaging period, which was also observed in reference experiments, apoptotic phenotypes exhibited death rates that were relatively prolonged, as described below. Yeast PCD involves a complex functional network 292, and interactions between PCD and cell ageing, mating, and autophagy pathways, as well as the epigenetics of PCD, have been recently reviewed 191,293–296. In contrast to apoptotic PCD, which is characterized by genetic regulation and energy dependence, necrosis occurs in an uncontrolled manner after the swelling of cells or organelles, sudden loss of plasma membrane integrity, or the occurrence of cellular dysfunction 297.

236 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion Nutrient stress during pre-cultivation promotes chronological ageing, triggering PCD 294. We observed a variety of single-cell death phenotypes in yeast that were temporally resolved by fluorescent time- lapse imaging with dual staining using either PI/PO-PRO-1 or PI/CgAM (Fig. 56-58 and Videos 31-34, Appendix 8.4 and Supplemented CD). Given that PO-PRO-1 stains dsDNA via intercalation comparable to PI, time-resolved, single-cell, dual-fluorescence imaging permitted us to distinguish necrosis (a sudden change from PO-PRO-1-/PI- to PO- PRO-1+/PI+) and apoptosis (PO-PRO-1+/PI- to PO-PRO-1+/PI+) over time, as shown in the schematic diagram in Fig. 57. The competing adsorption of both dsDNA dyes was not observed, although PO-PRO- 1 diffusion is assumed to be higher due to its smaller molecular size compared to PI 264,298. PO-PRO-1 is not considered problematic for use in single-cell studies, as shown by Wlodkovic et al. (2009), or for use at higher concentrations with mammalian cells 86 and has been tested in B. subtilis (Fig. 48). The other non-invasive counterstain method presented here for single- cell-death studies employed non-toxic fluorogenic esterase substrates (Fig. 48 and Fig. 50). CgAM is taken up as an esterase substrate into

237 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 56: Budding and cell death. (A) A budding yeast cell was injured at the budding neck (marked with red arrow) and subsequently died. (B) Schematic drawing of an apoptotic mother cell that lost its (figure legend continues on next page▼)

238 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion (▲continued figure legend of Fig. 56) membrane potential at the same time as its bud cell, which was injured near the budding neck. Both cells were PO-PRO-1+/PI+. (C) The pictures indicate that spatiotemporally resolved PI fluorescence diverges from the merged images in (A). (D) Schematic drawing of cell recovery due to cell budding. The cell became PO-PRO-1+ after membrane potential loss. The cell eventually initiated PCD while undergoing replication and proceeded with budding and actin- assisted DNA distribution (AT-rich regions appear fluorescent blue). (E) A cell exhibiting apoptotic-like behaviour followed by recovery due to budding is marked with white arrows. PO-PRO-1 fluorescence increased and was maintained for one hour before budding was initiated. DNA passage into the daughter cell was observable due to the presence of the DNA indicator PO-PRO-1. Mother and daughter yeast cells proceeded to budding followed by the dilution of fluorescence in the next filial generation. the cytosol and secreted by active efflux pumps or sequestered in either vacuoles or in the cytosol if ATP is depleted 299,300. Aged cells sequestering calcein green (CALg) in their vacuoles were assumed to have lost their V-ATPase activity prior to achieving a PI+ state due to the loss of organelle function. As with PO-PRO-1, apoptotic-like phenotypes appeared as CALg+ before presenting as CALg+/PI+ (Fig. 58). It could be distinguished between necrotic-like phenotypes by employing non-toxic dual staining (Fig. 57 and 58) to observe cells exhibiting the hallmarks of ageing or undergoing lethal autophagy, the apoptosis of budding mother or zygote cells, and shmoo mating with aged cells. PCD induction was primarily observed in the budding descendants of progenitor yeast cells derived from stationary, nutrient- deprived pre-cultures. However, dynamic live-cell analysis with PI and PO-PRO-1 permits hours of time-resolved monitoring of yeast fission and the subsequent loss of membrane potential indicated by 239 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion PO-PRO-1 loading prior to PI uptake, as shown in Fig. 56 A-B. The spatial resolution of PI fluorescence revealed injury close to the budding neck (Fig. 56 A and Fig. 56 C, marked with red arrows), where the replicated DNA from the mother is passed to the daughter cell. The mother and daughter cells were still connected and shared the same fate: the initiation of death. In contrast, the absence of PI fluorescence in a PO-PRO-1+ cell and the loss of PO-PRO-1 fluorescence are indicative of pre-apoptotic cells able to undergo growth recovery (Fig. 56 D-E, Video 35 (see Appendix 8.4 and Supplemented CD) among rapidly growing cells. PO-PRO-1 is a very selective indicator of double-stranded DNA that is not sequestered in a manner consistent with other cell components 298. PO-PRO-1 fluorescence decreased between two imaging time points, which cannot be explained by bleaching, and this property was passed on to emerging daughter cells during meiosis (Fig. 56 E). Cells seeded after famine conditioning in a pre-culture shaking flask were partly growth-inhibited but were not PI+. Based on PO-PRO-1 staining, these growth-inhibited yeast cells formed small circular blue fluorescent patches localized close to the cell membrane. Extrachromosomal ribosomal DNA circles (ERC) are indicative of the replicative ageing of yeast cells 301. Interestingly, not all cells were stained, similar to early apoptotic cells undergoing DNA fragmentation.

240 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 57: PI combined with PO-PRO-1 to indicate apoptosis in yeast cells. (A) Necrotic cells and late apoptotic necrotic-like cells both lose their membrane potential (PO-PRO-1+) and cell integrity (PI+) concurrently (white arrow). During apoptosis, cells induce DNA fragmentation, which reduces PI fluorescence (right image, white arrow), whereas early phase apoptotic cells lose their membrane potential (green arrow). (B) Although vacuoles are not stained by PI or PO-PRO-1, vacuole enlargement can be observed via phase-contrast imaging (white arrow). (C) A daughter cell that underwent macroautophagy produced an autophagosome containing DNA (white arrow) labelled by PI immediately prior to cell death (PI+). The vesicles remained stable for hours prior to rupture. (D) Budding shmoos that mated with aged cells underwent apoptosis. The mating cells exhibited enlarged vacuoles and dense engulfment (white arrows). PI fluorescence transitioned from the mating cell to the associated shmoo. (E) The progression of PO-PRO-1/PI labelling is shown. Loss of membrane potential allows PO-PRO-1 to enter the cell and stain the entire cell intensely with blue fluorescence via DNA intercalation. Apoptotic cells are initially PO-PRO-1-positive (PO-PRO-1+) before exhibiting PI fluorescence (PI+).

241 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 58: PI combined with CALg to indicate apoptosis in yeast cells. (A) A yeast cell with a necrotic-like phenotype exhibited increased calcein green fluorescence (CALg+) during the induction of apoptosis until the cell membrane ruptured, and calcein was not retained intracellularly (indicated by a light blue circle). PI fluorescence increased immediately after cell disintegration (PI+). (B) Vacuole enlargement and an increase in vacuole pH characterized ageing in yeast, which was indicated by a local increase in calcein fluorescence. (C) A cell that formed two autophagy vesicles (indicated with a white arrow) labelled by calcein immediately prior to cell death (PI+) is shown. The vesicles remained stable for hours before disappearing after cell contact with budding yeast (marked with *). (D) A budding zygote that generated two normally growing daughter cells (indicated with 1 and 2) is shown. After the third bud (3) formed, the yeast cell died, along with the bud. (E) The progression of PI/CgAM labelling is shown. The entire cell will be stained green if calcein efflux is inhibited. Partial calcein accumulation in vacuoles occurs in aged cells. Ageing cells accumulate calcein green in their enlarged vacuoles, presumably due to the loss of V-ATPase. Cells are initially calcein-positive (CALg+) before acquiring PI fluorescence (PI+).

242 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion Necrotic-like phenotypes immediately exhibited CALg+/PI+ and PO-PRO-1+/PI+, respectively (Fig. 57 A and Fig. 58 A). Spontaneous cloudy leakage of CALg fluorescence was an indicator of cell membrane rupture, as shown in Fig. 58 A, after 5.2 h of cultivation. CALg fluorescence increased for one hour prior to latent cell swelling and radical calcein release into the surrounding environment immediately before the cell became PI+. A fraction of cells exhibited remarkable vacuole enlargement, which has been shown to appear in aged yeast cells 302, followed by PI staining during the late phase of cultivation. Aged cells accumulated CALg in their vacuoles, which remained intact (Fig. 58 B and Video 32). Aged cells that produced enlarged vacuoles (Fig. 57 B, white arrow) were partially PO-PRO-1+. During the late cultivation phase, aged cells became PI+/PO-PRO-1+ (Fig. 57 B). In terms of autophagy in yeast, there are several cell recovery mechanisms, each with a known potential for failure 303. Macroautophagy was rarely observed to end in fatality (Fig. 57 C, Fig. 58 C, and Video 31). Cells that immediately became PI+ formed two separate vesicles (Fig 58 C, white arrow) that were distinguishable based on interior CALg fluorescence and slight PI fluorescence surrounding the vesicle exterior. Both vesicles remained stable for more than 3 hours. After 4 hours, the vesicles disappeared following contact with the bud of a growing sibling.

243 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion Furthermore, shmoo-mediated mating between aged cells with enlarged vacuoles was observed. Shmooing and mating were not imaged during microfluidic cultivation. However, cells underwent shmooing or mating during pre-cultivation and were stochastically seeded in the microcultivation chamber for observation. Although zygote budding prior to cell death was observed (Fig. 58 D and Video 33), shmoos lost their membrane potential after mating with cells that had previously lost their membrane potential. The cells formed a condensed, small vesicle (Fig. 57 D, white arrows, Video 34) inside the enlarged vacuole concurrent with slight PI fluorescence surrounding the vacuole. The vacuole was likely under pressure because bursting was observed. PI fluorescence did not appear inside the vacuoles, resulting in the heterogeneous distribution of fluorescence inside the shmoo-cell aggregate. Additionally, the autophagy of a budding mother cell was observed (Fig. 57 C, white arrow). The mother and daughter cells both lost their membrane potential simultaneously. The cells did not divide terminally because first the daughter and then the mother cell became PI+. The lysosome exhibited dual fluorescent signals (PO-PRO- 1+/PI+), indicating the presence of DNA content. The vesicle remained stable for more than ten hours with minimal fading of fluorescence.

244 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion

Figure 59: Stress-triggered cell rescue via membrane permeability transmission pore formation. (A) The fluorescent traces of PI and PO-PRO-1 in cells with initially local PO-PRO-1 staining at the cell membrane due to mitochondrial permeability transition pore (MPTP) formation are shown. The spatiotemporal resolution of the intracellular fluorescence revealed a necrotic-like phenotype, an apoptotic-like phenotype, and cell resuscitation as explained in (B) and depicted in (C) – (E). (B) After exposure to nutrient starvation (figure legend continues on next page▼)

245 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion (▲continued figure legend of Fig. 59) conditions, single cells were observed to be permeable, to some extent, to dye loading with PO-PRO-1 due to its small molecular size. Blue fluorescent patches near the cell membrane were assumed to indicate extrachromosomal rDNA circles (ERC), which are thought to influence the life span and chronological ageing of yeast. This can lead to apoptosis (PI-/PO-PRO-1+) and necrotic-like apoptosis (PI+/PO-PRO-1+). Apoptotic-like cells were shown to be capable of recovery and division. (C) An aged cell with an enlarged vacuole (white arrow) showed initial partial membrane permeability to PO-PRO-1 due to MPTP formation. The loss of membrane potential followed. This necrotic-like phenotype showed decreased size, stained PI+, and demonstrated induced death over several hours. The size reduction and period of death are characteristic of necrotic-like apoptosis. (D) Cells that temporally were partially permeable to PO-PRO-1 were thought to have formed MPTPs and to have lost their membrane potential. The fluorescence of this apoptotic-like phenotype increased over time. (E) Cells that perform MPTP formation are capable of recovery, as shown here. The cell internalized PO-PRO-1 and remained fluorescent for more than 8 h before the fluorescence disappeared from most intracellular areas. The reduction in fluorescence was followed by budding and cell division.

PO-PRO-1 is a much smaller molecule than PI and enters the cells due to perturbations in mitochondrial permeability. Mitochondrial permeability transition plays a role in MPTP formation, leading to necrosis (Fig. 59 C) or apoptosis (Fig. 59 D) 304. Mitochondrial permeability transition is the reversible differentiation of mitochondria resulting in increased permeability to solutes smaller than 1500 Da, depolarization, swelling, and ATP production 304. Cells are able to recover via the microautophagy of dysfunctional mitochondria (Fig. 59 E) 303. We observed three phenotypes distinguishable based on their fluorescent PO-PRO-1 traces: a necrotic-like phenotype, an apoptotic-like phenotype, and a resuscitative phenotype (Fig. 59).

246 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion 5.5.11 Discussion

We demonstrated a one-step, non-invasive, dynamic PI staining method inside a microfluidic cultivation device that supported the real-time observation of cell death events among prokaryotic and eukaryotic cells. Non-toxic counterstains (CALv, CALg, and PO- PRO-1) were selected to facilitate real-time testing for survivor cells or to obtain additional information regarding cell status. The continuous supply of fluorochromes in media ensured optimal distribution in dense cell cultures over the experimental period. Thus, optimal fluorochrome concentrations could be realized in the µM or smaller range, avoiding non-specific cell staining that occurs due to fluorochrome uptake at high concentrations. Furthermore, no experimental disruptions due to sampling were necessary. Cellular- triggered differentiation and subsequent death were observed in real- time for selected cells and colonies under specific cultivation conditions of interest. This permitted observation of the development of rapid cell death phenotypes (e.g., lysis) or more complex PCD occurring over time under low ATP consumption conditions. PI staining is among the most widely used methods to detect cellular death. However, this staining method is discussed very inconsistently in the literature 259–264. Some researchers have described contradictory staining results, and the conventional staining protocol is prone to

247 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion error during certain steps and using parameters, such as the dye incubation time, wash buffers, and dye concentration 261. The often- mentioned occurrence of false PI+ cells may be attributable to the use of dye concentrations that are too high. Although considered impermeable, viable cells may be stained by diffusion-driven uptake if the concentration gradient at the outer cell wall boundary is sufficiently large. Although a longer staining duration increases PI+ cell numbers 259, low-level, optimized PI perfusion is non-toxic for use with microfluidic cultivation. The effects of longer staining duration may be explained by the on-going death of moribund cells during endpoint sample staining due to the bactericidal impacts of storage, nutrient deprivation, osmotic shock, counterstain toxicity, or high PI concentrations, given that the toxicity of assay conditions is generally not taken into account. However, viable cell numbers have also been overestimated due to the stronger binding of SYTO 9 to DNA binding sites in comparison to PI 260,264. We determined that a constant PI concentration of 0.1 µM was sufficient for bacteria and yeast. A concentration of 10 µM PI, which corresponds to 6.7 µg/mL, results in a concentration gradient between the inside and outside of viable gram-positive cells that facilitates partial PI intrusion. However, yeast staining using 6 µg/mL PI have been reported to provide inconsistent staining results 259. Previous

248 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion studies have employed batch staining approaches with PI concentrations ranging from 4 µM to 500 µM for gram-negative E. coli 268,278, 3 µM for gram-positive Mycobacterium smegmatis 19, and 30 µM to 149.6 µM for S. cerevisiae 284,305. Continuous PI contact with cancer cells has been achieved with concentrations ranging from 0.37 µM to 3 µM PI in microfluidic devices 86,285. Furthermore, we demonstrated the presence of segmented cells with a PI+ dead cell pole and an antibiotic-tolerant, metabolically active cell pole attributable to the incomplete cell division of stressed cells caused by the addition of antibiotics. Extrinsic growth perturbation (as shown in this study for C. glutamicum utilizing bacteriostatic antibiotics) as well as intrinsic stress 30 lead to growth arrest and the inhibition of cell division. In contrast to microscope-based analytical systems, FACS analysis possesses the disadvantage that aggregated PI- and PI+ cells are considered falsely as PI-stained cells 259. Thus, if a surviving PI- cell is attached to a PI+ cell or the cell pole is recovered, misinterpretations are possible. The continuous addition of counterstains together with PI is challenging if toxicity and growth impairment must be avoided. We found CAM derivatives and the membrane potential-indicating stain PO-PRO-1 to be suitable for non-toxic counterstaining in combination with PI. These stains are appropriate for live-cell imaging applications 16,85,298 and permit the indication of residual cell functionality in

249 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion survivor cells or cells in the early stages of PCD. In particular, the temporal resolution of single-cell dye uptake is crucial for cell death analyses. The use of a microfluidic device allows for an integrated approach involving non-toxic dynamic PI staining during cell cultivation. Detailed examination of cellular PI uptake as well as time-resolved observation of phenotypic differentiation in individual dying cells were possible in stressed cell populations, in contrast to the reference cells that proceeded growth. Here, we present the novel use of a microfluidic cultivation device for the time-resolved analysis of cellular survival and studies of PCD employing fluorescent dyes to detect intracellular changes. The dynamic analysis of single-cell viability comprises more than the differentiation of dead and alive cells. Temporal differentiation permitted intermediate states and intermediate changes in cell status to be distinguished (alive to dead or lysed, moribund to resuscitated, alive to autolysed, dead to lytic decay). Our single-cell fluorescence analysis is particularly relevant for studies examining phenotypically heterogeneous and spontaneously occurring cell survival or lethal cell differentiation. Thus, our method has the potential to contribute to studies of autolysis, autophagy, antibiotic tolerance, spontaneous resistance, epigenetic triggered cell differentiation, membrane integrity, drug testing, medical care and many other areas of research.

250 Time-Resolved, Single-Cell Analysis of Induced and Programmed Cell Death via Non-Invasive Propidium Iodide and Counterstain Perfusion Its broad applicability is not only limited to typal specifications and may also inspire applications involving biofilms, mammalian cells or thin tissue layers. Our dynamic live/dead staining method can be adapted to other existing media-perfused microfluidic cultivation devices for single-cell fluorescence imaging if the duration of cellular death or survival is of interest. Such analyses will allow lingering questions to be answered during molecular biological studies. Additionally, our approach can be integrated into fluorophore expression studies through the use of multiplexed imaging.

251 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

5.6 Real-Time Antibiotic Susceptibility Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Experiments have been performed by A. Singh and preliminary results are published in his Master thesis. I have adapted the preliminary results for calculation of normalized growth, chi tests, and categorized fluorescence results. I have concepted the manuscript and figures. A. Singh has concepted the videos and the supporting information of the manuscript. Therefore, the manuscript first authorship has been shared by mutual agreement. Spore samples of B. athrophaeus have been kindly provided by J. Arreola (RG of Prof. Dr.-Ing. M. J. Schöning of the University of Applied Sciences Aachen).

Chapter Abstract: Antibiotic sensitivity and spontaneous antibiotic tolerance of single cells in isogenic bacterial colonies is known to be heterogeneous with a tendency of subfraction formation of survivors. Especially, antibiotic tolerance is not genetic based or inherited to daughter cells, and can reappear in vitro and in vivo. The susceptibility and tolerance to antibiotics is both measured by time-consuming killing curves. Whereas antibiotic tolerance is undetectable, if conventional agar plate based antibiotic susceptibility testing is performed, and non-growing tolerant bacteria in low cell number are

252 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device determined erroneous as antibiotic sensitive. The potential of antibiotic tolerance formation and insensitivity to antibiotics harbour the risk of reinfection, regrowth and chronic infection diseases and even worse, it is the cradle of antibiotic resistance. Thus, there is a tremendous gap in antibiotic research, that has to be solved for improvement of testing, diagnostics and patience treatment. We present a novel microfluidic approach for an antibiotic sensitivity testing that is able to indicate instantaneous appearance of antibiotic tolerant bacteria using a microfabricated cultivation device for parallelized analysis under controlled conditions. The microfluidic fluorescence time-lapse imaging single-cell analysis showed antibiotic treatment specific phenotype development using non-invasive fluorescence in situ staining of genetically unmodified gram-positive wild type bacteria. A spore forming and a non-sporulating model bacterium were used for proof of principle, since both bacteria are close relatives of severe human pathogens, e.g., M. tuberculosis, C. diphteriae, B. anthracis, and MRSA, that are known for persistence.

5.6.1 Introduction to Current Antibiotic Sensitivity Research

Survival of single antibiotic tolerant bacteria harbours the potential of regrowth in the medical treated host 240. Antibiotic tolerance is characterized by a reduction of bacterial metabolic activity, 253 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device remarkably impaired cell growth, and it is not based on resistance genes, although a genetic disposition due to virulence genes has been stated recently 68,306,307. While antibiotic resistant bacteria precede growth in presence of an antibiotic, single antibiotic tolerant microorganisms are non-replicating or growth reduced cells 68. Therefore, growth impaired antibiotic tolerant cells are easily overseen in conventional cultivation based count plate methods 306. The occurrence of single antibiotic tolerant phenotypes in an isogenic colony has been reported for many severe diseases causing an antibiotic failure that getting worse, further tolerant bacteria can evolve antibiotic resistant bacteria 308–310. While antibiotic resistant infection can be treated by higher medication concentration, antibiotic tolerant bacteria can be treated by pro-longed antibiotic treatments to avoid persister subpopulations formation 306,309. Opportunistic persistence or evolution of antibiotic resistance and regrowth were observed with gram-negative bacteria such as E. coli and Comamonas denitrificans cells in the remaining cytoplasm surrounding of lysed cells in antibiotic gradient evolving microfluidic devices 311,312. Furthermore, phases of lethal antibiotic addition were demonstrated to provoke dynamic persistence of M. smegmatis with regrowth after days 313 and persistence as phenotypic switch in E. coli with regrowth in resupplied medium after hours 314 in a microfluidic device. Thus, the bacterial survival in antibiotic treatment is a complex multiparameter deterministic process as revised in 68,306,308,315.

254 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Bacterial survival after antibiotic treatment increase the demand for single cell resolved point-of-care diagnostic for phenotypical heterogeneity in antimicrobial susceptibility of relevant bacteria. Conventional cultivation based antibiotic susceptibility testing is limited in precision, resolution, potential for automation and acceleration of analysis. Conventional agar plate based assays are prone to a bias by dilution, plating and accuracy of visual appraisal 316. However, conventional microbiological diagnostic does not determine antibiotic tolerance of these non-growing or growth reduced bacteria. Antibiotic susceptibility testing based on agar plating as well as microfluidic antibiotic susceptibility testing establish bacterial growth analysis at different antibiotic concentrations 311,316–318. At the moment there are only few approaches of reliable tolerance assays either in conventional microbiological cultivation assays or in microfluidics. In addition to killing curve performance to our knowledge based on resuscitation and secondary growth on agar plates 319 or on GFP expression under control of virulence genes combined with single-cell growth analysis 307. Microfluidic antibiotic susceptibility test methods intend not only better precision of minimal inhibitory concentration determination of antimicrobials, but also claim to fasten up obtainment time of results 317,318,320,321. Furthermore, microfluidic implementation of antibiotic concentration gradients have been demonstrated by mostly complicated skilled approaches using an agarose channel system 318, a

255 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device hexagonal microwell array connected by nanoslit channels 312, a sink and source channel system in multi-layered chips 311, an automated digital microdroplet system 321, a microchamber array with integrated mixing and filling microvalves to test synergistic or antagonistic interactions of antibiotics on E. coli 322, prior freeze dried antibiotics in culture chamber array combined with confocal reflection microscopy 323, and modulated multidimensional antibiotic containing flow microsegments 324. Fluorescence-aided antibiotic susceptibility testing access intracellular changes upon antibiotic stress 16,17,257. Mostly, fluorescence is given by fluorophore expression after genetic modification. In microfluidic devices, fluorescent fusion protein expressing bacteria are implemented for antibiotic susceptibility testing of gram-positive actinobacteria 19,313 and firmicutes 325, as well as gram-negative bacteria 312,322. Also the use of a fluorogenic substrate has been shown for efflux pump inhibitor antibiotic testing in a femtoliter droplet array 326 and in microfluidic perfusion devices 16. Further, enzymatic reaction on resazurin to resorufin was used for viability testing of antibiotic treated E. coli 321. Also, implementation of the cell death indicators was shown with Sytox green to visualise antibiotic sensitive dead S. aureus cells 327 and with PI in combination with GFP or YFP expression 240,257 and in combination with noninvasive counterstaining of a fluorogenic substrate or a membrane potential stain 17.

256 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

We developed a microfluidic approach that elucidates temporal resolved formation of dormant cell forms and cell death during antibiotic treatment. Bacillus subtilis is a model organism to study spore formation in bacteria. These spores of the class bacilli are known to endure decades and under severe environmental conditions until regrowth (B. antracis) 328. Non-sporulating microorganism Corynebacterium glutamicum is closely related to pathogenic Corynebacteriaceae and Mycobacteriaceae with reported antibiotic persistent phenotypes 329. For proof of principle gram positive wild type microorganisms were treated with ampicillin (AMP), chloramphenicol (CHL), kanamycin (KAN), streptomycin (STR) and mytomycin C (MMC). Cell death and bacterial survival under antibiotic stress were indicated by non-invasive fluorescence imaging under temporal resolution. Phenotypic differentiation and cellular heterogeneity can be studied with the antibiotic of choice.

5.6.2 Microfluidic Antibiotic Susceptibility and Survivor Assay

The microfluidic antibiotic susceptibility test approach was integrated in a microfluidic device approved for bacterial cultivation 34,36,38 and temporal resolved viability studies 16,17. The microfluidic device has four channels with two microarrays of growth chambers for each main channel (Fig. 60 A). The microarrays are substructured in smaller 257 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device perfusion channels embedding the microchambers for bacterial cultivation in between (Fig. 60 B). Bacteria are seeded initially in the growth chambers (Fig. 60 C) and perfused with BHI containing nutrients, the antibiotic of interest in adjusted concentration and fluorescence indicators at a non-toxic homeopathic concentration. Modelling and CFD simulation of nutrient distribution have been published in 38. We developed appropriate fluorochrome combinations for the bacterial families of the here used test microorganism for non-invasive dynamic fluorescence staining as shown in detail by Krämer et al. (2015) and Krämer et al. (2016) 16,17. The mechanisms of the fluorescence indicators for fluorescence time-lpase imaging of firmicutes (B. subtilis) and actinobacteria (C. glutamicum) are shown in Fig. 60 D. Membrane potential loss of B. subtilis and spores of B. athrophaeus is indicated by the negative charged PO-PRO-1 (Fig. 60 D i.)) that intrude cells (Fig. 61 A and C) or dye repellent Bacillus spores (Fig. 61 D), if their membranes are damaged. The charged dye molecules pass the cell and membrane, attach to the DNA and fluoresce brightly blue (PO-PRO-1+).

258 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Figure 60: Microfluidic test principle of antibiotic susceptibility and tolerance testing. (A) The four-channelled microfluidic device is fabricated of PDMS and glass. Every channel harbours a microcultivation chamber array for continuous perfusion of trapped bacteria with adjusted concentrations of nutrients, antibiotics and fluorescence indicators. (B) The microstructure array contains parallelised growth chambers to seed bacteria and image cell growth and cell death under defined conditions. (C) The growth chambers are supplied with a combination of dyes for real- time observation of cell survival and antibiotic tolerance, and with nutrients to analyse antibiotic concentration sensitivity by cell growth. The cells are contacted with nutrients, sensing molecules, and antibiotics by diffusion (D) Non-invasive fluorescence imaging is used for determination of (i) membrane potential loss (PO-PRO-1), (ii) cell disintegration (PI) and (iii) metabolic activity (CvAM hydrolysis to CALv). (figure legend continues on next page▼)

259 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

(▲continued figure legend of Fig. 60) The negative net-charging of living cells shield gram-positive bacteria from intrusion of negative charged molecules such as PO- PRO-1 or PI. PO-PRO-1 fluorescence indicate dying firmicute cells and mother cells after sporulation. Dormant endospores are PO-PRO-1 repellent what lead to non- fluorescent cell poles in case of antibiotic tolerance and bear the risk of regrowth. PI is a cell death indicator to detect antibiotic sensitivity. The esterified CALv (CvAM) is non-charged, taken up by actinobacteria, and hydrolysed to the charged CALv that is intracellularly trapped due to its negative charging. Metabolic active cells actively efflux CALv, whereas dormant cells with decreased metabolic activity accumulate CALv what result in increased fluorescence. As shown in Krämer et al. 2016, propidium iodide (PI) is a universally usable death indicator for cell death (Fig. 60 D ii.)) and was used for B. subtilis (Fig. 61 A and C) and C. glutamicum (Fig. 61 B and E) to visualise antibiotic sensitivity 17. PI intrudes strictly cell wall injured cells, intercale with their DNA. If cells are disintegrated they are red fluorescent and considered as dead (PI+). Unstained and non-lysed B. subtilis cells are considered as vegetative cells in presence of antibiotics (Fig. 61 F). In addition, mature, unimpaired Bacillus spores remained unstained as shown with B. athropaeus spores (Fig. 61 D and F). Although, rescusitation experiments have been deomonstrated in the microfluidic device recently for C. glutamicum cells treated for 1 h with AMP or CHL 16 and for B. athrophaeus spores in Fig. 61 F, these experiments are not in focus here.

260 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Figure 61: Cell categorisation of the fluorescence in situ staining of antibiotic impaired (A) Bacillus subtilis 168 and (B) Corynebacterium glutamicum ATCC 13032 cells. (C) Bacillus subtilis cells are able of spore formation. Vegetative cells remain unstained (PI-/PO-PRO-1-), whereas dying cells are blue fluorescent (PO- PRO-1-), dead cells blue and red fluorescent (PI+/PO-PRO-1+), and lysed cells are fragmented as shown in the phase contrast. (D) In addition, mature dye repellent Bacillus spores can be distinguished in viable and impaired by the membrane potential indicator as shown with B. athropaeus spores. (E) C. glutamicum cells are non- sporulating gram-positive cells. (figure legend continues on next page▼) 261 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

(▲continued figure legend of Fig. 61) Lysed cells are fragmented as also unstained putative intact cells with a sudden loss of previous fluorescence. Dead cells also stain red fluorescent (PI+) and loose their CALv fluorescence, whereas antibiotic tolerant remain brighter CALv fluorescence (CALv++) than moderate fluorescent growth unimpaired cells (CALv+). Due to antibiotic presence cell division halted cells, can show a bipolar phenotype with two independent cell poles that are lysed, dead or antibiotic tolerant. (F) Spore resuscitation is indefinite in time for antibiotic impaired spores. Here unimpaired regrowth of an unimpaired B. athrophaeus spore is shown within 2 hours.

In difference to B. subtilis, which remains length grows or lyses, non- PI stained C. glutamicum remain its cell shape probably due to its cell wall that contains heigh molecular weight lipids in the outer cell wall area and a wax-like mycolic acid layer covalently linked to polysaccharides in the interior of the cell wall barrier 147. Therefore, remaining metabolic activity in non-replicating antibiotic tolerant C. glutamicum cells was determined as described in detail by Krämer et al. 2015 and shown in application by Krämer et al. 2016 16,17. The esterified violet fluorogenic calcein acetoxymethyl ester (CvAM) is a neutral molecule with a supposed structure as shown in (Fig. 60 D iii.)). CvAM has been demonstrated previously as non-toxic fluorogenic substrate of the actinobacteria C. glutamicum and M. tuberculosis for microbial diagnostic 16,87,142. The fluorogenic substrate is taken up by actinobacteria, and hydrolysed by intracellular esterases to ethanol and the violet fluorescent calcein (CALv) that is intracellularly trapped due to its negative charge. Metabolic active cells actively efflux CALv and show moderate CALv fluorescence, whereas dormant cells and antibiotic tolerant cells with 262 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device decreased metabolic activity accumulate CALv what result in increased bright CALv fluorescence (Fig. 61 E), whereas lysed cells show sudden fluorescence loss (Fig. 61 E) 16.

5.6.3 Microfluidic Antibiotic Susceptibility Testing Compared to Conventional Plate Count Assay

Antibiotic susceptibility of B. subtilis 168 and C. glutamicum ATCC 13403 were determined by the bacterial growth in microfluidic cultivation chambers or on BHI media plates normalized by the mean growth parameter of all performed experiments without antibiotic addition. Therefore, the growth rate was determined by cell number over time of five cultivated colonies and the colony forming units of three media plates were determined at the same cultivation conditions, respectively. Normalized growth data of both microorganisms is compared for AMP, CHL, KAN and STR concentrations at 0 to 10 µg/mL liquid BHI broth and on solid BHI medium (Fig. 62). The normalized growth results of reference condition without any antibiotic and a low antibiotic concentration (0.1 µg/mL) were comparable for both wild type bacteria and the above-mentioned antibiotic substances using both cultivation methods. For 1 µg/mL CHL and B. subtilis and for 1 µg/mL AMP and C. glutamicum, respectively, the observed normalised growth was higher than resulted by microfluidic cultivation. 263 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Figure 62: Antibiotic susceptibility of (A) Bacillus subtilis 168 and (B) Corynebacterium glutamicum ATCC 13032 for AMP, CHL, KAN and STR. All antibiotics were tested in a low (0.1 µg/mL), middle (1 µg/mL) and high (10 µg/mL) concentration with both microorganisms. The novel microfluidic approach (n = 5 colonies) is compared with results of the conventional viable count plating method (n = 3 colonies). Bacterial growth was measured by cell number determination in microfluidic cultivation chambers or colony forming units counting on plates and normalized by mean of all reference values. 264 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

5.6.4 Determination of Time-Resolved Antibiotic Sensitivity and Antibiotic Tolerance

For successful antibiotic stewardship, the key points are the selection of an antibiotic, antibiotic concentration and the duration of antibiotic uptake. Conventional MIC testing helps to decide the first two key points and leave the answer to the best duration of antibiotic treatment open. Temporal resolved antibiotic sensitivity (cell death), antibiotic tolerance (cell survival), and antibiotic insensitivity (remaining cell growth) at low (0.1 µg/mL), middle (1 µg/mL), and high (10 µg/mL) concentration of bactericidal (AMP, KAN, STR), and bacteriostatic (CHL) antibiotics. The non-invasive dual fluorescence labelling indicated time resolved antibiotic induced single-cell death and antibiotic tolerant phenotypes. Although, the count plate tests suggested 10 µg/mL of every antibiotic as sufficient to induce cell death of B. subtilis and C. glutamicum cells, incomplete antibiotic susceptibility was observed comparing the intracellular PI fluorescence with the particular counterstain fluorescence (Fig. 63 and 64). An eligible antibiotic susceptibility of B. subtilis was given if the previously increased PO-PRO-1 fluorescence in a cell is followed by an increase of PI fluorescence. If solely the membrane potential increased, the antibiotic has neiter completed killing nor resulted cell lysis, that result total loss of fluorescence. Increased PO-PRO-1

265 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device indicated a potential of endospore formation in absence of PI fluorescence. Hence, B. subtilis showed the highes incipient antibiotic sensitivity to AMP at 1 µg/mL and CHL at 10 µg/mL. KAN resulted merely loss of membrane charge of B. subtilis, whereas cell growth proceeded with STR. Non-sporulating C. glutamicum reduce its metabolism that is indicated by accumulation of CALv as prior described 16,17. The uptake and hydolysis of CAMv to fluorescent CALv pursued, whereas CALv efflux arrested. In case of cell death or lysis, CALv fluorescence is lost and the intracellular fluorescence switch to red fluorescence of PI or the cell is non-fluorescent. C. glutamicum showed the same highest sensitivities as B. subtilis. However, the highest concentration of AMP or CHL resulted antibiotic tolerant cells accumulating CALv in absence of PI fluorescence. An incease of AMP or CHL concentration did not prevent antibiotic tolerance as shon previously by Krämer et al. 2016 17. At 1 µg/mL CHL, the bacteriostatic properties of the antibiotic were not given. In difference to the bacteriostatic protein inhibitor CHL, the bacteriocidal antibiotics KAN and STR, which also target at the protein biosynthesis by induction of mRNA misreading, impaired C. glutamicum at a low concentration of 0.1 µg/mL. Growth impaired cells died partly or switched to the quiescent antibiotic tolerant phenotype.

266 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Figure 63: Antibiotic sensitivity and tolerance of Bacillus subtilis 168 for AMP, CHL, KAN and STR. All antibiotics were tested in a low (0.1 µg/mL), middle (1 µg/mL) and high (10 µg/mL) concentration together with a reference channel without antibiotic addition. The intracellular fluorescence of the cell death indicator PI and the membrane potential indicator PO-PRO-1 were normalized to the single-cell fluorescence at initiation of antibiotic (figure legend continues on next page▼)

267 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

(▲continued figure legend of Fig. 63) contact to compare the killing efficiency (high PI fluoresccence) of the antibiotic with the occurence of antibiotic tolerance (high CALv fluoresccence) of low, middle, and high concentrations. Exemplarily, micrographs of B. subtilis colonies in presence of a middle and high antibiotic concentration are given.

268 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

Figure 64: Antibiotic sensitivity and tolerance of Corynebacterium glutamicum ATCC 13403 for AMP, CHL, KAN and STR. All antibiotics were tested in a low (0.1 µg/mL), middle (1 µg/mL) and high (10 µg/mL) concentration together with a reference channel without antibiotic addition. The intracellular fluorescence of the cell death indicator PI and the hydrolysis product CALv of fluorogenic CvAM were normalized to the single-cell fluorescence (figure legend continues on next page▼) 269 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device

(▲continued figure legend of Fig. 61) at initiation of antibiotic contact to compare the killing efficiency (high PI fluorescence) of the antibiotic with the occurrence of antibiotic tolerance (high CALv fluorescence) of low, middle, and high concentrations. Exemplarily, micrographs of C. glutamicum colonies in presence of a middle and high antibiotic concentration are given.

5.6.5 Discussion and Conclusion

A microfluidic approach has been developed to test antibiotics in different concentrations upon their aqueous solubility limit to achieve comparable results on bacterial growth at half time as it is done conventionally with media plating. Additionally, non-growing survivor phenotypes of gram positive sporulating (B. subtilis) and non- sporulating (C. glutamicum) bacteria were unmasked by non-invasive fluorescence imaging. These dormant cell forms are known to survive antibiotic treatments due to growth arrest and a reduced metabolism. Therefore, they remain in a temporal antibiotic tolerant state whereby these cells bear the risk of recovery and recolonization when antibiotics are absent again 68,330,331. Our microfluidic antibiotic susceptibility assay allows elaborating appropriate conditions of antibiotic treatment according choice of antimicrobial substance, its concentration and residence time for the particular target microorganism. The evolution of antibiotic tolerance is considered to depend on carbon flux and bacterial growth impairing microenvironment conditions in vivo and in vitro in M. tuberculosis 331 and in bacteria in

270 Antibiotic Susceptability Testing and Antibiotic Tolerance Sensing in a Microfluidic Fluorescence Indicator Perfusion Device general 308. Furtermore, the bacterial cell wall influences antibiotic uptake and efficiency 147. Antibiotic addition has been shown to induce changes of the mycobacterial energy metabolism 257 that reduces the metabolic activity and the accessibility of the cell for the antimicrobial action of the antibiotic. These led to the hypothesis that antibiotics have self-propagating impact on their own non-gene-based tolerance. GFP expression strategies of antibtiotic sensitivity sensing have the disadvantage that a temporal gap between growth arrest or loss of GFP fluorescence and PI fluorescence appearance is given as shown for the actinobacterium M. smegmatis in 19. This stress the importance of novel antibiotic sensitivity testing that helps to address in addition antibiotic tolerance potential of antibiotic treatment at defined antibiotic concentration and time of action.

271 Discussion and Conclusion

6. Discussion and Conclusion

Intracellular implementation of fluorescent probe molecules for undisturbed, temporal resolved observation of living cells is challenging due to the requirement of physiological sensing conditions to preserve cell growth and function. Thus, phototoxicity, chemotoxicity, metabolic burden or cytostatic interaction with the DNA inside the cell have to be avoided. The challenge is even enhanced if the cells are localized in closed microcultivation systems, where space is defined small and cell sampling is technically impaired. In addition, unwanted interaction of the molecular fluorogenic probe with the microfluidic device surfaces should not been given due to ionic interfacial adhesion or lipophilic material intrusion 17. However, the molecular probe must be hydrophilic enough to be soluble in media, neutral to cationic charged to traverse the cell wall and cell membrane, or it have to be integrated in a prior step into the microbial DNA for self-assembling of fluorescent fusion proteins 28,80. The latter well-established multiplexable fluorescence approach is a prior introduction of heterologous fluorescent fusion proteins, which are associated with promotor activities of a gene of interest 22,67. The genetic integration of heterologous genes to generate intracellular fluorescence or bioluminescence upon gene expression is a frequently used fluorescence-based imaging method 22,25,67,332. It requires

272 Discussion and Conclusion

functionable cells, which are ready to express genes upon promotor activation and to translate them into proteins 260. Unfortunately, the expression can be an additional metabolic burden and cells have to be genetically modified in advance. To avoid overshadowed temporal response by different ripening times of multiple parallel integrated fluorophores or timing bias due to kinetics of promotor induction, different extend of intracellular protein expression, the expression dynamics have to be evaluated 41,82,333–335. Thus, biosensor strains can be constructed to qualify intracellular metabolism 23,60. Furthermore, bimolecular fluorescence complementation is used as FRET biosensors , e.g. with CFP and YFP or Venus and mCherry, for analyses of intracellular molecular mechanisms, transmembrane transport, or quantitative fluorescence measurements in living cells 257,336–338. However, also unintended FRET can occur with two fluorophores or fluorochromes in molecular proximity. This inner filter effect has to be determined and corrected if observable 339. Although, sophisticated FISH methods has been reported early for high sensitivity of cellular metabolism analyses of bacteria and multiplexed fluorescence imaging of heterogeneous microbial communities, however the approach sacrifice the lives of bacterial consortia and is not compatible with live-cell imaging 340,341. Therefore, temporal resolution of developing lineage trees and single- cell tracking over time is impaired using FISH. Nevertheless stable

273 Discussion and Conclusion long-time labelling with relatively high cell fluorescence in comparison to fluorescent protein based approaches can be realized 341. Another approach for intracellular fluorescence sensing is addition of fluorogenic molecules in the cellular environment to indicate distinctive cell features. Hence, the uptake of and interaction with these fluorogenic tracer molecules can be followed over time. This is a novel approach to develop living cell analyses methods to characterize cell functions, metabolism or rare spontaneous phenotypes such as tolerant or resistant cells. However, this approach is limited to nontoxic fluorochromes at prior optimized concentration 16,17. Thus, relatively high intracellular fluorescence signals can be achieved for successful fluorochrome applications, if fluorescence background is additionally ideal low. The challenges were manifold that had to be tackled for the development of this multiplexed fluorescence methods: i.) The used dye concentrations had to be harmless for replicating microorganisms, ii.) it may not trigger cell stress reactions or unwanted phenotypic differentiation, iii.) the probe molecules should have favourable cell retention,

insensitive to bleaching and bleading, iv.) with low background fluorescence or adsorption to the microfluidic device material, 274 Discussion and Conclusion

v.) without emission or excitation spectra overlap, vi.) and they should not block cell functionalities as molecular transport or replication or transcription, vii.) further excitation should be non-phototoxic while bright intracellular fluorescence is given with filter sets, which were already established in the microscope.

Thus, strategies with conventional biosensor application and novel developed fluorescence in situ staining (FISS) have been developed and evaluated for implementation of intracellular fluorescence sensing in microfluidic cultivation devices for environmentally controlled cultivations with the biotechnologically important gram-positive C. glutamicum ATCC 13032 in particular and with other relevant prokaryotes and eukaryotes. The orchestration of microfluidic cultivation environment control of evolving microcolonies and intracellular fluorescent probe molecule implementation enabled advanced visualisation of microbial stress studies under different conditions. The reaction repertoire of microorganism upon stress conditions has a plethora of complex processes and is involved in every area were microorganism can be found. Stress triggers can be defined as abiotic or biotic and as intrinsic or extrinsic. All these stressors of living cells have in common, that they are determined by their resulted stress response, effect on cell functionality, phenotypic changes, changes in productivity, viability, metabolic activity and changes in growth or 275 Discussion and Conclusion metabolism. Microorganisms have gone through a myriad of evolutionary steps to cope with challenging environmental changes that endanger their live. If it comes to stress of cellular systems, the researcher is confronted with the variation of outcomes as individuals react to stress differently due to differences in their physiological state, that influence their survival and their stress response. The heterogeneous response of cells to environmental stress could be explained by different hypothesis. The origin of phenotypic heterogeneities in isogenic cell populations is supposed to be either by toggling genetic switches (intrinsic stress), by microenvironmental inhomogeneities, or by stochastic maldistribution of biomolecules during cell division (extrinsic stress) 52. Furthermore, genetic switches have noisiness, what impacts phenotypic heterogeneity in isogenic cell populations. Indeed, the triggers of phenotypic heterogeneity are not fully cleared, yet. It is noteworthy, that toggling mechanisms of genetic switches, imbalance of molecule distributions after division, and environmental bias have to be considered as the trigger for subpopulation formation. However, a complex interaction cannot be denied or stated at the moment of intrinsic and extrinsic triggers. It is noteworthy, that phenotypic heterogeneity in isogenic colonies is not exclusively explained by genetic mechanisms. Some differentiations of phenotypic appearance are induced by environmental influences simplified as stress. Stress can be caused by

276 Discussion and Conclusion

absence of required nutritive components (famine stress, oxygen limitation), chemical exposure with antimicrobials or salt (chemostess, ionic stress, osmostress), physical parameters such as heat/chill (thermostress), draught, pressure, light exposure (photostress), and by physicochemical triggers that appear by an interaction of chemical and physical stress mechanism, e.g., light exposure induced oxygen radicals (oxidative stress). Phenotypic heterogeneity in isogenic populations is supposed as survival mechanism of life endangering stress conditions. To visualise evolving phenotypic heterogeneity, the ease of reliable microenvironmental control and unbiased contactless sensing in living cells has to be given in combination. Especially, if life endangering stress is applied to analyse bacterial adaption, impact of the observation technique must not introduce a bias in outcome. Therefore, real-time fluorescence sensing strategies were established for time resolved intracellular sensing in living ancestor cells and their descendants in a microfluidic cultivation device. To understand the processes that are going on in stressed cells, several examples of multiplexed fluorescence imaging methods were implemented as measurement tools in a microfluidic cultivation device platform. The fluorescence sensing methods were performed on the one hand with conventional biosensor strains expressing fluorescent fusion proteins, on the other hand a novel approach with in situ fluorescence staining (FISS) was developed.

277 Discussion and Conclusion

The involved promotor induction of genes of the prophage CGP3 and of the SOS gene cascade gene recA have been analysed by state-of- the-art prior introduction of heterologous yellow and red fluorescent fusion proteins in the model organism Corynebacterium glutamicum ATCC 13032 at full nutrient supply and under nutrient starvation conditions without carbon, nitrogen, phosphate, or iron 22,80. The integration (plasmid-borne or chromosome-integrated) influenced the excitation parameters in difference to FISS due to differences in resulted protein to gene ratios 22,30,72. The novel, developed FISS methods were preliminary designed and tested with the model organism C. glutamicum ATCC 13032 that is of relevance for microbiological research as well as biotechnological production processes. If applicable the FISS methods had been transferred or adapted to other bacterial species and yeast S. cerevisiae. A combinatory approach of FISS and fluorescent fusion proteins expression has been shown exemplarily. The fluorescence analyses have been proofed to be non-invasive on microbial growth and physiology. The advantageous environmental control of microfluidic devices had been employed for experimental set ups with a selection of environmental stress scenarios for microorganisms to apply the molecular sensing methods that I have developed and evaluated.

278 Discussion and Conclusion

6.1 Time-Resolved Visualisation of Heterogeneous Promotor Induction Events

Fluorescence time-lapse imaging has been shown to be a precise tool in combination with microfluidic devices for single-cell analyses 53,62,69. In difference to FACS, that help to analyse population fractions of SOS+ cells and phage+ cells at different cultivation snap shots 30,96, single-cell time-lapse imaging in microfluidic devices reveals the individual cell fates in an isogenic population 22. Thus, different fluorophores inserted by heterologous gene integration approaches had been tested and evaluated for biosensor strain development for spontaneous SOS response and sporadic phage induction with C. glutamicum ATCC 13032. A dual reporter strain was achieved with green and red fluorescence of eYFP (SOS response, phage induction), Venus (SOS response), and e2-Crimson (SOS response, phage induction) with a defined difference in expression and ripening and determined photobleaching dynamic. The photobleaching for eYFP, Venus and E2-Crimson has been determined to correct fluorescence loss by time in future. Thus, quantitative determination of single-cell expression dynamics could be performed, if stress triggers and prophage inducing conditions are directly in focus of subsequent research. But quantitative expression dynamic studies and promotor activity determination depend on an effective promotor inducer as shown in 53,62,69.

279 Discussion and Conclusion

In addition, the fluorescence time-lapse microscopy procedure for measuring gene expression dynamics in reporter strains has been established with custom-made PMMA fluorescence beads from PolyAN, Berlin. These beads had tailor-made fluorescence intensities adjusted to the eYFP expressing reporter strains to correct emission fluctuation due to biases of the exciting light source over time. The PMMA beads were self-immobilized to the oxygen plasma treated main channel walls and. In difference to the approach of Young et al. used on medium pads, the PMMA fluorescence beads could be integrated in the microfluidic cultivation, and needed no prior adjustment of the exposure time to compensate lamp intensity fluctuations from experiment to experiment 62. Furthermore, the lamp intensity bias could be corrected continuously to correct the fluorescence course of spontaneous expressed fluorescent fusion proteins in C. glutamicum ATCC 13032 cells. Time-lapse imaging of C. glutamicum ATCC 13032 cultivated under constant conditions in microfluidic devices revealed that spontaneous SOS response and prophage induction were switched on independently in absence of extracellular stimuli in a population minority 22,36. Although, a significant fraction of SOS+ cells were additionally phage+, this state was followed by growth arrest, halted cell division, and absent recovery of the cell. SOS response in C. glutamicum ATCC 13032 was reported to end in a non-growing state or in cell recovery as it was previously stated as SOS recovery

280 Discussion and Conclusion

by DNA repair or pulsing SOS response in E. coli 96, in B. subtilis 53 and in C. glutamicum ATCC 13032 22,30. Several cells remained in SOS+/non-dividing state with branched cell elongation as reported by SOS-induced gene activation of divS 30,100. Non-growing cells that spontaneously induced SOS response were tested previously to exclude the cell death indicator PI and defined as senescence-like SOS+ E. coli cells 96,104. Similar studies of the senescence state of C. glutamicum ATCC 13032 SOS biosensor strains could be performed in microfluidic cultivation in future. The role of conserved prophage elements CGP1-3 in the bacterial genome of C. glutamicum ATCC 13032 is not completely elucidated, yet. Nevertheless, the major prophage CGP3 has underwent intensive studies by Frunzke and coworkers 22,30,72,73,91. Under environmental control CGP3 prophage genes were shown to be induced even under absence of an obvious inducing trigger in exponentially growing isogenic cell populations using prior introduction of fluorescent fusion proteins by plasmid-based approach as well as into the bacterial genome 22. The promotors of the SOS cascade and the prophage induction of CGP3 could be determined to be both active (SOS+/phage+) in one third of stressed cells under non-induced conditions. An initially by Nanda et al. (2014) stated correlation of SOS response (SOS+) and prophage gene induction (phage+) could be found only partly in stressed cells with lethal SOS+/phage+ state 22. Although,

281 Discussion and Conclusion spontaneous SOS response is followed remarkably often by phage induction and cell division arrest, a distinct trigger of the prophage element could not be defined so far that allows also SOS-independent prophage induction. Temporal resolved spontaneous phenotype development under full nutrient supply and with nutrient deprivation state showed, that phage induction is a lethal mechanism that survived in the genome of its host. A remarkable part of phage+ cells developed previously the SOS+ state. The SOS+ state has been shown to be involved in cell division impairment, resuscitation and re-initiation of cell growth is in the reach of possibility for cells in this state. Also, the impact of CGP3 on its host fitness is still not cleared in detail and how its genes survive evolutionary selection. Knockout mutants without CGP3 are reported to show no special phenotype 42. The role of remaining prophage elements such as CGP3 in C. glutamicum ATCC 13032 and their influence on host fitness are of interest since it’s reported that bacteriophages are capable to control pathogenesis and metabolism in corynebacteria 105. The SOS-independent prophage induction is under further investigation in the future. Recently, the prophage genes of CGP3 are under intensive examination, the controlling inducer of prophage induction is not reported 22,30,42,72,73,91. However, Donovan et al. (2015) have reported nucleotide hydrolysis by the CGP3 protein AlpC in vitro with a preference to ATP in comparison to GTP and proofed the protein to be involved in egoistic

282 Discussion and Conclusion

phage transport to the cell membrane 91. This may lead to the hypothesis that C. glutamicum ATCC 13032 missuses the prophage induction as an apoptotic-like altruistic cell death pathway to remove ATP poor, damaged, and stressed cells in the population with a lytic strategy. If a direct trigger of prophage gene induction can be addressed, the promotor activities could be determined of a significant cell number for quantitative expression dynamic analysis in the future. Intermittent phases of nutritive stress were applied for C. glutamicum ATCC 13032, for which the metabolism routes of glucose, iron, nitrogen and phosphate have been well characterised in literature 115,123,342,343. However, extended limitation of carbon and phosphate limitation increased phage+ and SOS+ cells in a minority of the isogenic population (1 – 2 %) in comparison to cultures without nitrogen or iron (<< 1). Regrowth experiments of stationary C. glutamicum ATCC 13032 cells and cells of the dtxR (major regulation gene for iron uptake) knockout mutant revealed remarkable induction of the prophage CGP3. The prophage induction has been finally increased to ~20 % in the WT and in the ΔdtxR mutant, if ancestor cells were precultivated in iron- containing CGXII + 4 % GLC for three days. Under iron free precultivation, resuscitation of the WT failed completely, whereas the prophage induction in the ΔdtxR mutant even increased 4-fold in comparison to the iron containing precultivation. Therefore,

283 Discussion and Conclusion intracellular absence of iron and a depleted intracellular iron pool is supposed to benefit prophage induction. Hence, an indirect phage induction mechanism or a more complex role of the genetic regulation node dtxR might be given in C. glutamicum, which has not been revealed, yet. Iron limitation is known to be an essential transition molecule in oxygen transfer and catalytic processes 92,114,115. Iron homeostasis is controlled by the regulator DtxR to control a non-toxic intracellular iron concentration 122,123. C. glutamicum ATCC 13032 ΔdtxR mutant has shown increased relative phage+ induction as upregulation of CGP3 genes has been reported previously of this mutant strain 73. Further, dtxR is known to be involved in corynephage associated toxin production in the relative strain C. diphteriae 114.

However, absence of phototoxicity as promotor trigger of pRecA, pLys, and pInt2 could be stated with the help of fluorogenic ester substrates and the metabolic activity of nutrient deprived SOS+ and phage+ cells could be determined. Metabolic activity measurement using CvAM perfusion of nutritive depleted C. glutamicum ATCC 13032 cell cultures indicated under carbon limitation conditions that phage+ cells belonged mainly to three categories: i.) lysed CALv- cells, ii.) ATP depleted cells with reduced CALv secretion, and iii.) SOS+ cells depleted in ATP, that switch to the phage+ state. Thus, the energy transfer molecule ATP or induction of dormancy genes might play a fuelling role of the induction of the genome innate prophage CGP3.

284 Discussion and Conclusion

This additional cell heterogeneity of the phage+ population promoter was independent of gene-based promotor activity and resulted from the developed multiplexed fluorescence imaging approach using biosensor strains expressing fluorescent proteins and nontoxic molecular probe integration with an AM-coupled fluorogenic substrate perfusion.

6.2 Acetoxymethylester (AM)-Enabled Intracellular Fluorescence Sensing

Integration of intracellular fluorescence sensing was developed with novel single-cell imaging methods using self-embedding molecular probes in the cytoplasm without the requirement of genetic modification. These approaches were, once evaluated and established, immediately useable for the production strain C. glutamicum ATCC 13032 and its mutants. Thus, non-DNA coded intracellular parameters could be addressed. The acetoxymethyl ester (AM) moiety has been developed to shuttle probe molecules for molecular sensor integration into mammalian cells 154. This shuttle function of AM-bound fluorochromes has been shown for an actinobacterium 16. Therefore, similarities of the cell wall were speculated to be the reason for AM-coupled fluorochrome uptake and subsequent intracellular fluorescence signalling, that failed to occur in living cells of E. coli or B. subtilis and for other aerobic

285 Discussion and Conclusion firmicutes 233. A hypothetic phylogenetic reason for uptake of AM- bound sensing molecules might be given and has to be explored in future. The transport mechanism could be visualized by TIRF microscopy or other high-resolution fluorescence microscopy. The use of fluorescence assays with AM-coupled dyes were shown for C. glutamicum 16,217 and other actinobacteria 87,142,158,216 in literature to determine intracellular pH or viability. Additionally CgAM was used to stain proteobacteria 175,176, a cyanobacterium 179 and the anaerobic firmicute C. acetobutylicum 176. Studies with C. glutamicum ATCC 13032 in presence of AM-coupled molecular probes demonstrated successful use of those for an actinobacterium for several commercially available fluorogenic sensing molecules. This has been tested for direct measurement of several intracellular parameters to present novel sensing strategies for lineage tree and single-cell analyses. Novel intracellular sensing strategies based on AM-coupled molecular probes has been developed for C. glutamicum ATCC 13032 to measure metabolic activity, apparent dormant cell states, intracellular oxygen radical formation, phototoxicity, and internal pH in living single cells. Thus, temporal resolved single-cell studies with AM- bound indicators showed to be a novel methodical approach for studying the metabolism of production strains such as C. glutamicum ATCC 13032. These approaches could be transferred for microbiological studies of anaerobic bacteria or M. tuberculosis 87,142.

286 Discussion and Conclusion

Phototoxicity is a major issue for (multiplexed) time-lapse fluorescence microscopy. However, phototoxicity and its influence on living cells are only rarely reported to be checked for single-cell fluorescence studies. If phototoxicity is determined comparative growth studies with and without illumination are used 62. A direct measurement approach of phototoxicity is to determine ROS that occur due to photonic excitation. The AM-coupled DHCAM indicated

1 O2 appearance in the cytoplasm after uptake and hydrolysis. Thus, absence of phototoxicity for all multiplexed fluorescence time- lapse imaging methods was tested and could be stated to be absent. In

1 addition, internal formation O2 of respiratory chain gene devoid mutants could be detected with DHCAM. This molecular probe molecule showed fast response, excellent cell retention, more specificity than CM-H2DCFDA, and revealed heterogenous and

1 sporadic O2 presence in the mutant ΔctaD in absence of thiamine in the perfusion medium.

1 O2 appearance immediately stopped cell growth in the mutant ΔctaD, whereby the mode of action of the scavenger thiamine could be revealed to be a prevention of growth arrest by ROS formation instead of improvement of growth as suggested by population originated

207 growth determination by OD600 measurement . Whereas the mutant

1 Δqcr showed absence of O2, which confirmed the previous conclusion of the function and importance for ROS formation of the iron-sulphur cluster of the cytochrome reductase of the cytochrome

287 Discussion and Conclusion

+ bc1-aa3 supercomplex if the subsequent cytochrome oxidase as H

207 drainage is deleted . The general ROS indicator CM-H2DCFDA revealed continuous increase of unspecific oxidative molecules built

- - by intracellular metabolism such as H2O2, •O2 , HO•, ROO , •NO, ONOO- over time 190. The non-AM-bound fluorogenic substrate showed less intracellular retention and specificity than DHCAM. The shuttle function of the enzymatic labile moiety AM carries the key of uptake of complex esterified molecules as the studied probe molecules in this PhD thesis. Once AM was cleaved off the probe molecule, this was captured in the cytosol. Exclusively, promiscuous ABC transporters are supposed to actively transport out the fluorescent probe molecules as found for CALv. C. glutamicum ATCC 13032 harbours ABC-type multidrug transport systems 256 and bears ABC-type multidrug transporter genes involved in homeostasis 211. Nevertheless, revealing the molecular mechanism of CALv efflux has importance to further develop the metabolic activity sensing with CvAM to a method of bacterial multidrug resistance screening as already established for tumor cells 85,244. Thus, drug related bacterial ABC transport mechanism can be considered as a further important application of our metabolic activity sensing method in future. Since effects induced by antibiotics are already demonstrated in relation to actinobacterial ATP metabolism of non- growing and growing phenotypes to elucidate evolving antimicrobial resistance or tolerance 17,87,142,257.

288 Discussion and Conclusion

The active efflux of CALv has been demonstrated in dependency of glucose availability for C. glutamicum. The transfer mechanism is considered by an ATPase, what could be elucidated by knockout mutants and follow up studies. Although, efflux of CALv is not fully understood, yet, the conversion from CvAM to CALv in combination with a sampling free microfluidic approach is a powerful tool to gain new insights in the metabolic activity of growing and non-growing bacteria. Non-growing, dormant or resistant cells exhibit a large potential for bacterial survival of antimicrobial substances such as antibiotics. The mechanism of AM-coupled calceins was thoroughly analysed with violet fluorogenic CvAM. The fluorogenic parent molecule of CALv, AM-bound CvAM, is taken up by C. glutamicum ATCC 13032 16 and its relatives M. luteus and M. tuberculosis 87,142, but excluded by bacteria as E. coli or B. subtilis. Intracellular CALv fluorescence is a universal indicator that the cell has active esterases and the energetic capability to perform transport mechanisms to secrete CALv. Therefore, comparative metabolic activity sensing relying on co-factor independent hydrolysis and fluorochrome trafficking could be demonstrated with intermittent deprivation of carbon, iron and the iron chelator PCA. The detection of cell to cell differences of metabolic activity with CALv and derivatives was established successfully for actinobacteria as valuable analytical tool for multiple applications in single-cell

289 Discussion and Conclusion studies such as growth perturbation tests and single-cell analyses of toxicity resistance in continuous microcultivation. The superior environmental control in microfluidic devices was used for simulation of alternating cultivation by starvation phases and stress conditions under additive antibiotic feed for C. glutamicum ATCC 13032. The rising importance of microfluidic approaches was recently reviewed for temporal resolved live cell analyses of bacterial metabolism and human or animal microbiome at presence and in future 344. Fluorochrome incorporation, ester conversion, and light exposure were found to be non-toxic for the bacterial growth under fluorogenic calcein substrate perfusion. The fluorescence signal was not found to be prone to photobleaching or photobleading, which did not explain the tremendous and fast reduction of the mean single-cell fluorescence following carbon re-supply after a starvation phase of 10 – 12 h. A tremendous increase in the mean fluorescence under carbon limitation supported the conclusion that CALv is secreted energy dependent, because in the absence of glucose intracellular ATP was reduced. Another hypothesis for the drastic change in mean fluorescence can be induction of esterase activity and enhanced calcein production. However, increasing extracellular CvAM concentrations did not result in a constant increase of CALv fluorescence. Therefore, a constitutive intracellular esterase activity is applicable. Nevertheless, the CvAM esterase activity of

290 Discussion and Conclusion

C. glutamicum ATCC 13032 has not been characterized yet and require more experimental insights e.g. by gene expression analysis Non-growing cells that arose after intermittent limitation of carbon or iron proved to be metabolically active. As long as iron was omitted in the perfusion medium, the absence of the important nutrient inhibited growth and cell division. However, lack of iron did not result in the same bright CALv fluorescence as under the lack of carbon. Under lack of carbon, non-growing but metabolically active cells appeared in every analysed microcolony and differed phenotypically according to their mean single-cell fluorescence after carbon re-supply. The resuscitation promoting factor Rpf2 is reported to trigger the regrowth of non-growing starved C. glutamicum ATCC 13032 cells after a switch from famine to feast condition 93,253. The complex regulation of Rpf2 may explain the heterogeneous CALv fluorescence of dormant cells observed after carbon limitation. Comparative studies

of wild type expressing Prpf2 promoted fluorescent fusion proteins with GFP, YFP or RFP and the Rpf2 devoid mutant would be of interest in alternating presence and absence of different carbon sources to unravel the mechanism of heterogeneous regrowth of dormant C. glutamicum ATCC 13032 cells. Whereas, regrowth after antibiotic caused growth perturbation showed heterogeneous cell shape and fluorescence phenotypes as discussed and concluded in the following chapter. The use of metabolic activity sensing with CvAM for antimicrobial testing was demonstrated with

291 Discussion and Conclusion an exemplary use of antibiotics. The antibiotics differ in mode of action and they generated different mean single-cell fluorescence traces of descendants after antibiotic exposure. These studies could be continued with C. glutamicum ATCC 13032 biosensor strains with introduces antibiotic resistance genes. Krämer et al. (2015) showed the recovery and regrowth of stressed bacteria and the impact on their descendants could be shown after one hour of addition of AMP and CHL 16. The enhanced heterogeneity after AMP contact was supposed to be induced by disturbance of CvAM uptake and CALv transport mechanism or CALv containment due to the antibiotic impact on the cell wall synthesis. CHL impaired mean single-cell fluorescence especially during exposure and shortly after. This can be explained by rather reduced esterase activity than influence of CALv leakage. AMP addition is partly explained by the impaired cell wall integrity due to bacterial cell wall synthesis inhibition. Furthermore, tolerance of C. glutamicum ATCC 13032 to AMP is influenced in contrast to CHL by the expression level of the potential multidrug resistance gene cepA encoding an efflux pump like protein 258. Also, the heterogeneous resistance of C. gltuamicum ATCC 13032 cells to NIG and VAL showed to challenge the calibration of the internal pH measurement calibration using AM-coupled pH probes.

The pHint sensing molecules pHrodo Red and pHrodo Green were taken up by C. gltuamicum ATCC 13032 cells prior permeabilization

292 Discussion and Conclusion

with antibiotics. However, the antibiotic treatment influenced the calibration of the pHint due to the duration to equalize pHint with pHex. In addition, antibiotic treatment showed to be more homogenous with VAL in presence of BHI medium than with NIG and VAL added to calibration buffers. To reveal the perturbation of pHint of

C. gltuamicum ATCC 13032 by changes of pHex, metabolism and growth has to be elucidated more in detail in future with further improvement of the method calibration 212. Moreover, there are other AM-coupled molecular probes commercially available that could be tested to establish further molecular sensing methods for actinobacteria and their intracellular cationic micronutrient metabolism 154,157,219,220 or cytoplasmic pH 157,166,218. Furthermore, the possibilities of molecular probe development are not fully exploited for bacterial single-cell analyses using microfluidic devices, yet. A broad innovative use of fluorogenic substrates would advance rapid future bacterial point of care diagnostics or rapid enzyme activity screening of bacterial libraries in hours instead of days of conventional cultivation if integrated in microfluidic cultivation devices 141.

293 Discussion and Conclusion

6.3 Time-Resolved Single-Cell Survival Sensing Using Fluorescent In Situ Staining

CALv fluorescence were established as counterstain method of real- time viability studies of C. glutamicum using the red fluorescent PI. Hence, non-dividing cells without red fluorescence could be defined as viable but non-dividing with bright enhanced CALv fluorescence or lysed without CALv fluorescence. This phenotypic discrimination allowed to define antibiotic treated cells as lysed, apparent antibiotic tolerant or bipolar with a dead pole and a surviving cell pole 17. Furthermore, the hypothesis that CvAM/CALv can be used for studies of the pathogens C. diphteriae, M. leprae, and M. tuberculosis as affirmed for latter recently 16,87,142. A growth-independent antibiotic susceptibility screening of non- culturable surviving M. tuberculosis, incubated in chemostats, was performed by Hendon-Dunn et al. (2016) to analyse once a day the viability using FACS combined with chemostat cultivation and dual labelling with CvAM and Sytox green 87,142. Thus, a plating and in vivo independent reliable screening of antibiotics against M. tuberculosis could be concepted. However, the cultivation and analysis of cell survival was consecutive and non-integrated in difference to the advanced microfluidic approach installed by sensing integration in a micro cultivation device. Novel sensing tools that are based on molecular probe integration in the microfluidic cultivation, have been developed to visualize the 294 Discussion and Conclusion

evolving heterogeneity of single cells in isogenic populations with the help of fluorescence time-lapse imaging. The sensing methods have been proved to be without impact on growth and cell physiology. The sensing probes are either uncharged fluorogenic substrate, which have to be converted by cytoplasmic enzymes after uptake, or the molecules are charged fluorogenic molecules, which enter cell wall injured or cell membrane impaired cells. The PI staining in combination with cultivation in microfluidic devices allowed to demonstrate the broad applicability for scientific as well as industrial concerns that involve survival and viability of prokaryotic or eukaryotic cells. The advanced novel real-time viability sensing allowed insights in bacterial antibiotic tolerance, spontaneous toxin resistance, visualization of filial phenotype development of previously stressed yeast ancestor cells to give raise to studies of evolving population minorities of antibiotic tolerance, genetic induced resistance, programmed cell death, apoptotic mitochondrial dysfunction in model cell systems. The method is transferable in other fluid supplied optic sensing systems established in a lab and is not based on sampling systems or additional integration of additional measurement equipment 305. Indeed, cell viability determination has broad importance for productivity, environmental response, clinical diagnostic and disease control, host and microbe relation, necrosis and apoptosis studies, cell differentiation, bacterial survival, chemotoxicity and drug

295 Discussion and Conclusion development, fine chemical production, bacterial resistance and tolerance, drug test screening. In near futute, chemical synthesis of organic molecules, appropriate for bacterial uptake and relevant enzymatic activity specificity, has a great future perspective in flow biochemistry. The conversion of fluorogenic compounds during cultivation answers, if bacteria are viable under the condition of interest 16,126,132. The role of enzymes in virulence and bacterial survival is of high interest for basic research as well as future oriented therapeutic strategies 129,131,133.

The design and use of fluorogenic compounds with fluorescent scaffolds emitting appropriate fluorescence 16,130 allow multiplexed time-resolved imaging in combined use with expression of GFP derivatives after genetic modification of bacteria 80,145. These would give the possibility to determine protein interaction, promotor activities, enzyme function and biochemical reactions in living cells by multiplexed fluorescence measurements. Therefore, a variation of novel nontoxic fluorescent scaffolds with enzyme-labile moieties will be of rising interest in future for bacterial diagnostics. Therefore, nontoxic fluorescent scaffolds with emission variety of minor Stokes shift will be of importance in future. Microfluidic approaches will facilitate screening and diagnostic use of new potential fluorogenic substrates to replace conventional incubation until colony growth is given 146.

296 Discussion and Conclusion

The efforts resulted easy transferrable fluorescence in situ staining (FISS) methods that can be flexibly integrated in small volume cell cultivations, whereas the invasive cell perforation and sample preparation of fluorescence in situ hybridisation is absent. The modular integration of the FISS methods has the benefit that it is readily available for integration in studies of biosensor stains expressing differently fluorescent fluorophores or knockout strains without fluorophore expression. FISS does not require additional genetic modification. Thus, the implementation of metabolic activity sensing of dual reporter strains for SOS response/prophage induction has been shown. This PhD thesis demonstrated a one-step, non-invasive, dynamic PI staining method inside a microfluidic cultivation device that supported the real-time observation of cell death events among prokaryotic and eukaryotic cells. Different non-toxic counterstains (CALv, CALg, and PO-PRO-1) were selected to facilitate real-time testing for survivor cells or to obtain additional information regarding cell status. The continuous supply of fluorochromes in media ensured optimal distribution in dense cell cultures over the experimental period. Thus, optimal fluorochrome concentrations could be realized in the µM or smaller range, avoiding non-specific cell staining that occurs due to fluorochrome uptake at high concentrations. Furthermore, no experimental disruptions due to sampling were necessary. Cellular- triggered differentiation and subsequent death were observed in real-

297 Discussion and Conclusion time for selected cells and colonies under specific cultivation conditions of interest. This permitted observation of the development of rapid cell death phenotypes (e.g., lysis) or more complex PCD occurring over time under low ATP consumption conditions. PI staining is among the most widely used methods to detect cellular death. However, this staining method is discussed very inconsistently in the literature 259–264. Some researchers have described contradictory staining results, and the conventional staining protocol is prone to error during certain steps and using parameters, such as the dye incubation time, wash buffers, and dye concentration 261. The often- mentioned occurrence of false PI+ cells may be attributable to the use of dye concentrations that are too high. Although considered impermeable, viable cells may be stained by diffusion-driven uptake if the concentration gradient at the outer cell wall boundary is sufficiently large. Although a longer staining duration increases PI+ cell numbers 259, low-level, optimized PI perfusion is non-toxic for use with microfluidic cultivation. The effects of longer staining duration may be explained by the on-going death of moribund cells during endpoint sample staining due to the bactericidal impacts of storage, nutrient deprivation, osmotic shock, counterstain toxicity, or high PI concentrations, given that the toxicity of assay conditions is generally not taken into account. However, viable cell numbers have also been

298 Discussion and Conclusion

overestimated due to the stronger binding of SYTO 9 to DNA binding sites in comparison to PI 260,264. We determined that a constant PI concentration of 0.1 µM was sufficient for bacteria and yeast. A concentration of 10 µM PI, which corresponds to 6.7 µg/mL, results in a concentration gradient between the inside and outside of viable gram-positive cells that facilitates partial PI intrusion. However, yeast staining using 6 µg/mL PI have been reported to provide inconsistent staining results 259. Previous studies have employed batch staining approaches with PI concentrations ranging from 4 µM to 500 µM for gram-negative E. coli 268,278, 3 µM for gram-positive Mycobacterium smegmatis 19, and 30 µM to 149.6 µM for S. cerevisiae 284,305. Continuous PI contact with cancer cells has been achieved with concentrations ranging from 0.37 µM to 3 µM PI in microfluidic devices 86,285. Furthermore, we demonstrated the presence of segmented cells with a PI+ dead cell pole and an antibiotic-tolerant, metabolically active cell pole attributable to the incomplete cell division of stressed cells caused by the addition of antibiotics. Extrinsic growth perturbation (as shown in this study for C. glutamicum utilizing bacteriostatic antibiotics) as well as intrinsic stress 30 lead to growth arrest and the inhibition of cell division. In contrast to microscope-based analytical systems, FACS analysis possesses the disadvantage that aggregated PI- and PI+ cells are considered falsely as PI-stained cells 259. Thus, if a surviving PI-

299 Discussion and Conclusion cell is attached to a PI+ cell or the cell pole is recovered, misinterpretations are possible. The continuous addition of counterstains together with PI is challenging if toxicity and growth impairment must be avoided. We found CAM derivatives and the membrane potential-indicating stain PO-PRO-1 to be suitable for non-toxic counterstaining in combination with PI. These stains are appropriate for live-cell imaging applications 16,85,298 and permit the indication of residual cell functionality in survivor cells or cells in the early stages of PCD. In particular, the temporal resolution of single-cell dye uptake is crucial for cell death analyses. The use of a microfluidic device allows for an integrated approach involving non-toxic dynamic PI staining during cell cultivation. Detailed examination of cellular PI uptake as well as time-resolved observation of phenotypic differentiation in individual dying cells were possible in stressed cell populations, in contrast to the reference cells that proceeded growth. Here, we present the novel use of a microfluidic cultivation device for the time-resolved analysis of cellular survival and studies of PCD employing fluorescent dyes to detect intracellular changes. The dynamic analysis of single-cell viability comprises more than the differentiation of dead and alive cells. Temporal differentiation permitted intermediate states and intermediate changes in cell status to be distinguished (alive to dead or lysed, moribund to resuscitated,

300 Discussion and Conclusion

alive to autolysed, dead to lytic decay). Our single-cell fluorescence analysis is particularly relevant for studies examining phenotypically heterogeneous and spontaneously occurring cell survival or lethal cell differentiation. Thus, our method has the potential to contribute to studies of autolysis, autophagy, antibiotic tolerance, spontaneous resistance, epigenetic triggered cell differentiation, membrane integrity, drug testing, medical care and many other areas of research. Its broad applicability is not only limited to typal specifications and may also inspire applications involving biofilms, mammalian cells or thin tissue layers. Our dynamic live/dead staining method can be adapted to other existing media-perfused microfluidic cultivation devices for single-cell fluorescence imaging if the duration of cellular death or survival is of interest. Such analyses will allow lingering questions to be answered during molecular biological studies. Additionally, our approach can be integrated into fluorophore expression studies through the use of multiplexed imaging. We have developed a microfluidic test approach of antibiotics in different concentrations upon their aqueous solubility limit to achieve comparable results on bacterial growth at have time as it is done conventionally with media plates. Additionally, non-growing survivor phenotypes of gram positive sporulating and non-sporulating bacteria were unmasked by non-invasive fluorescence imaging. These dormant cell forms are known to survive antibiotic treatments due to growth arrest and a reduced metabolism. Therefore, they remain in a temporal

301 Discussion and Conclusion antibiotic tolerant state whereby these cells bear the risk of recovery and recolonization when antibiotics are absent again 68,330,331. Our microfluidic antibiotic susceptibility assay allows elaborating appropriate conditions of antibiotic treatment according choice of antimicrobial substance, its concentration and residence time for the particular target microorganism. The evolution of antibiotic tolerance is considered to depend on carbon flux and bacterial growth impairing microenvironment conditions in vivo and in vitro in M. tuberculosis 331 and in bacteria in general 308. Furthermore, the bacterial cell wall influences antibiotic uptake and efficiency 147. Antibiotic addition has been shown to induce changes of the mycobacterial energy metabolism 257 that reduces the metabolic activity and the accessibility of the cell for the antimicrobial action of the antibiotic. These led to the hypothesis that antibiotics have self-propagating impact on their own non-gene-based tolerance. GFP expression strategies of antibiotic sensitivity sensing have the disadvantage that a temporal gap between growth arrest or loss of GFP fluorescence and PI fluorescence appearance is given as shown for the actinobacterium M. smegmatis in 19. This stress the importance of novel antibiotic sensitivity testing that helps to address in addition antibiotic tolerance potential of antibiotic treatment at defined antibiotic concentration and time of action.

302 Outlook

7. Outlook

There are five possible main strategies of potential interest to pursue in a nut shell for future investigations by single-cell fluorescence analyses using and advancing the microfluidic cultivation device, which was described in this PhD thesis. Hence, novel unconventional multiplexed fluorescence sensing could be implemented in the microfluidic channels using self-integrating sensing molecules.

i.) Further, the hydrophobic PDMS surface could be altered chemically more hydrophilic with the challenge to keep microstructural resolution. ii.) In addition, (controlled) channel wall tethered microenvironment sensors and improved microscopic resolution would enhance the insight in single-cell survival and stress triggered adaption in live endangering microenvironments. iii.) The microfluidic environmental control could be advanced with microvalves for pulses of shifting stress conditions for time resolved single-cell cell differentiation under stress conditions and cell recovery.

303 Outlook iv.) A remaining challenge in microcultivation devices is the single-cell isolation of cell minorities of interest comparable to FACS. i.) PDMS surface modification:

Microfluidic devices in combination with fluorescence time-lapse microscopy analyses reveal real-time insight in single-cell phenomena as phenotype development, cellular differentiation and intracellular processes. For dynamic in vivo visualisation, gene modification and fluorophore expression or bioluminescence gene introduction are mainly used as the gold standard for fluorescence time-lapse imaging of bacteria 59,345. Genetic modification with subsequent fluorescent fusion protein expression is the conventionally used method to introduce fluorescence in bacteria to visualize gene-based phenotypes of microorganisms. Expression of fluorescent proteins is the state-of-the-art techniques to perform studies of gene functionality and genetic regulatory systems. At present, fluorescent proteins are available in all colours of the rainbow for fluorescent protein expression studies and multiplexed single-cell fluorescence analysis 22,67,81. Unfortunately, it has been found that heterogeneous protein expression has shortcomings due to noise in regulatory genes 53, bias in inductor-promotor efficiency 46 and influence of minor codon exchange on host fitness 82. Further, it is dependent to oxygen, 304 Outlook therefore, oxidative maturation of GFP and GFP-derivatives is limited in anoxic environments if not alternative fluorescent fusion proteins are used 333. Thus, the use of fluorescent protein expression can be influenced in anaerobic bacteria, what drives the demand for alternative strategies as presented recently by Geva-Zatorsky et al. (2015) with designed oligosaccharides and click chemistry for metabolizable building blocks of the bacterial cell wall of anaerobic bacteria 227. For this purpose, biomolecules were metabolically engineered and coupled to derivatives of fluorophores for integration into the bacterial cell wall of growing bacteria 227,346. Also, fluorescent D-amino acids were built by metabolic cross-linkage of derivatives of coumarin, fluorescein to D-amino acids for temporal resolved tracking of nascent cell wall formation of E. coli, B. subtilis, Staphylococcus aureus, and other species 346. Different metabolic labelling strategies of bacteria to visualize growth, division or secretion have been summarized by Siegrist et al. 347. These labelling strategies have the potential to raise to a competitive strategy to fluorophore expression in microfluidic device coupled fluorescence imaging, if a simple “one shot” addition procedure to cultivation medium can be performed for fluorescence single-cell analyses. However, a predominant challenge is to avoid phototoxicity or chemotoxicity of fluorescence imaging using cell dyes for live cell staining under microenvironmental control 3.

305 Outlook

In addition to fusion proteins, conjugation of biomolecules to organic fluorochromes such as ATTO dyes or ALEXA labelling or surface tagging with fluorogen-activating proteins could be performed 12,14. Additionally, the alternative dynamic staining, that avoid growth impact, has been demonstrated for several gene-based and non- genetically determined single-cell analyses. Thus, also prophage- triggered bulging could be visualised by an fluorescence time-lapse imaging approach using an styryl membrane indicator and a cell impermeant cell death indicator recently 348. A similar method could be used to clarify if phage vesicles are formed with prophage CGP3. Fluorescence-based sensing approaches used for eukaryotic cells as reviewed by Fritsche and Mandenius (2010) are of high interest for antimicrobial drug testing and bacterial stress studies for basic research 349. In this thesis, time-resolved single-cell fluorescence analysis have been presented to perform studies of metabolic activity, viability, dormancy, tolerance and resistance of bacterial subpopulation raising over time 16,17. The ability to take up AM-coupled compounds is a distinguishing feature of bacteria to differentiate bacterial consortia and bacterial metabolism. In addition to intracellular sensing, the esterification of compound could be advantageous to overcome antibiotic therapy shortcomings due to resistance or tolerance of severe pathogens especially their persisters. In almost the same manner, cyclodextrins

306 Outlook and siderophores are described as trojan horses to surpass the bacterial cell wall to deliver antimicrobials 350. Hence, esterification of e.g. fluoroquinolones could revolutionize tuberculosis treatment and help to treat e.g. C. dipheriae or M. tuberculosis persister cells or antibiotic resistance spreading in future. However, fluorogenic substrates can help to visualize molecular mechanisms in living cells and transport mechanism. ii.) Channel wall tethered microenvironmental sensors:

The development of non-invasive fluorescence staining methods using lipophilic fluorogenic sensor molecules for analyses of the bacterial membrane, biotechnological lipid production for pharmaceutical precursors or fuel additives could be a next step to broaden the tool box of sensor molecules. Hence, styryl dyes or Nile red could be considered for promising sensing strategies, if the PDMS of the microfluidic device could be tailored more hydrophilic 8,83. Membrane staining using a styryl dye combined with GFP expression had been demonstrated for time-lapse imaging of bacterial antibiotic testing on agar pads 8. Passivation of the microfluidic channel walls with bovine serum albumin (BSA) enabled background suppression and PDMS penetration of Nile red 351. The surface coating or integration of polyethylene oxide (PEO)- moieties into the PDMS matrix is a promising strategy to shift the

307 Outlook

Figure 65: Optical sensor integration in microfluidic devices for measurement of microenvironmental parameters. (A) Sensing films on surfaces of microfluidic channels or chambers by (B) direct facing of the microfluidic surfaces or (C) with an additional protecting cover layer. (D) Locally applied sensing spots that consist of (E) locally printed or attached sensor films or (F) anchored sensing microbeads. (G) Dynamic sensing with (H) dissolved flown indicator molecules or (I) dispended nanoparticles or microparticles.

308 Outlook surface hydrophobicity more hydrophilic 352,353. To decrease surface hydrophobicity, UV-grafting of PEO and other monomeric polymer species was used to generate stable surface modification for electrophoresis application 354. Else, surface treatment with polyvinyl alcohol (PVA) has been shown for microfluidic oil-in-water droplet approaches 355. This surface treatments could be considered for containment of lipophilic medium components in the liquid phase in future, if oxygen supply is not critically diminished.

Furthermore, absorbance of H2O2, alcohols or lipophilic organic solvents by PDMS could be prevented by chemical surface modification of PDMS microchannels for visualisation of relevant bioproduction processes, bioconversion, and extracellular enzyme biosensing in microfluidic incubation devices 356,357. For cell wall tethered biosensor molecules (e.g., enzymes, antibodies), local surface functionalisation of PDMS microstructures enable controlled biomolecule immobilisation for sensing microenvironments 357. iii.) Microfluidic environmental control:

Strategies for extracellular sensing implementation in microfluidic channels are summarized by Gruber et al. 2017 and could be advantageous of use for future comparative single-cell studies under alternating stress conditions. These main principles are static sensing layers that are anchored locally cover surfaces as sensor films (Fig. A – C) or as sensing spots 309 Outlook

(Fig. D – F) in microchannels of the microfluidic devices, and dynamic sensors that are feed continuously to the flow as dissolved hydrophilic component (Fig. G – H) or as microbeads or nanobeads in suspension (Fig. I) 358. However, it will be a Pareto driven approach to integrate robust, sensitive luminescent sensing of microenvironments in microfluidic devices in single-cell studies using fluorescence imaging. Important environmental parameters, that are measured in real time, are substrate concentration, extracellular metabolites, pH, dissolved oxygen, carbon dioxide or temperature 358. The extracellular pH could be measured by pulses or continuous feed of dissolved sensor molecules or as dissolved oxygen with integrated sensor spots or sensor films in the microfluidic device 218,359. Another important feature is the use of extracellular microbeads for calibration to compare series of fluorescent single-cell studies or even quantify intracellular biomolecules, the dynamic building process (e.g. transcription), and intracellular localisation 62,360. Another approach of sensing integration are strategies for implementation of mass spectrometry and liquid chromatography in microfluidic devices. These approaches are highly sophisticated for measurement of cell colonies 361. This could help to quantify intracellular metabolite concentrations, although, the sampling is challenging. In addition, evaluation of intracellular enzyme kinetics could be performed in future 143.

310 Outlook

Furthermore, coupling of the microfluidic cultivation device and FISS or fluorophore expression with high-resolution fluorescence microscopy techniques could improve to unravel fluorescent molecule transport or biomolecule building processes in single bacterial cells. Molecular intracellular mechanisms such as fluorescent molecule trafficking or diffusion dynamics could be visualized better by high- resolution microscopy in comparison to epifluorescence microscopy (e.g. STED, STORM) 11,12. Light sheet microscopy reduces phototoxic impact potential and improves a volumetric insight in fluorescence distribution in biological matrices 362. Programmable microenvironmental manipulation with the help of microfluidic devices with integrated microvalve technology would advance the possibilities of single-cell analyses using fluorescence single-cell imaging 363. Microfluidic devices enable advanced microenvironmental control and perturbation of environmental conditions 364. Precise liquid handling enables to create via dilution concentration gradient profiles for rapid high throughput matrix screenings for single-cell PCR based taxonomy studies 363, optical tweezer assisted single-cell whole genome amplification of non-culturable cells 37,365. iv.) Single cell selection:

A remaining challenge in microcultivation devices is the single-cell isolation of phenotypic differentiated cell minorities to recultivate these cells. Probst et al. (2013) have successfully demonstrated 311 Outlook regrowth of single E. coli cells in neighbouring empty cultivation chambers after cell selection by optical tweezers 37. Else, integration of a fluorescence-activated droplet sorter (FADS) into the microfluidic device combines the advantages of FACS and microfluidic device based cultivation in combination with fluorescence imaging for isolation of enzymatic active and fluorescent cell minorities 366. This successful cell isolation approaches could be advanced by a valve controlled high-throughput microfluidic PCR application 363,367. This would help to elucidate if certain phenotypic differentiations are gene-based due to spontaneous mutations or happen despite of an isogenic DNA-panel. Spontaneous mutations or mechanisms of genetic exchange are speculated to introduce phenotypic heterogeneity such as antibiotic tolerance 68. Beyond genetic based decision making in the cell, phenotypic heterogeneity of isogenic cells occur and is not totally unravelled, yet 65. Microfluidic devices are valuable tools for advanced diagnostics. Improved automized image analyses would allow future approaches for studying gene-independent cell differentiation as cell death, bet- hedging strategies, intracellular product accumulation, intracellular radical formation, antibiotic tolerance and resistance. Therefore, microfluidic testing of metabolism and cell viability could advance future drug testing.

312 Outlook

Figure 66: Tremendous demand of data storage and big data analysis of multiplexed fluorescence imaging data. Own data set (green) with up to four imaging channels used (phase contrast + three different fluorescence channels) of a three-year working term in comparison with mixed data of researchers working between one to five years using non-multiplexed imaging. Figure kindly provided by C. Sachs (Research group of Dr. Katharina Nöh).

Therefore, automisation of multi-channel fluorescence image analysis is crucial. Multiplexed fluorescence images rapidly expand data scoring in comparison to imaging studies using one or two imaging channels (phase contast plus one fluorescence signal) as adumbrated by Fig. 66. Thus, semi-automated image analysis, as performed here, which need proofreading and manual corrections of every frame hamper especially multiplexed single-cell studies with high variation 313 Outlook in cellular phenotypes. The automisation of imaging and microfluidic cultivation is highly advanced by artefact-free, full autimisation of image analyses33,368 . Hence, algorithsm for classifier definition combined with iterative machine learning could provide online and on demand image analysis while the running longtime experiment in future 369.

314 Appendix

8. Appendix

Videos are supplemented on CD. Additional contributions are described at the end of the appendix.

8.1 Supplementary Material and Videos of SOS Response and Phage Induction Studies

The dual reporter strain C. glutamicum ATCC 13032::PrecA- venus/pJC1-Plys-e2-crimson showed different phenotypes of spontaneous stress response and sporadic prophage induction under standard cultivation condition. The stress response (SOS+, yellow fluorescent cells), the prophage induced cells (phage+, red fluorescent), and SOS+/phage+ cells (composite). Videos were published in 22. Video 1: A dual reporter strain colony with SOS+/phage- cell that proceeded cell division Video 2: A dual reporter strain colony with SOS+/phage- minority with flickering and constant Venus fluorescence Video 3: A dual reporter strain colony with SOS-/phage+, SOS+/phage+, and elongated putative divS inhibited SOS+ cell minority

315 Appendix

The reporter strain C. glutamicum ATCC 13032::PrecA-eyfp showed comparative phenotypes of spontaneous stress response as the dual reporter strain under standard cultivation condition. Videos were published in 30. Video 4: Cells of the SOS reporter strain exhibited SOS+ phenotype with proceeded cell division Video 5: Cells of the SOS reporter strain exhibited SOS+ phenotype with reduced cell division rate Video 6: Putative divS inhibited Cell of the SOS reporter strain exhibited SOS+ phenotype with elongated cell shape The dual reporter strain C. glutamicum ATCC 13032::PrecA- venus/pJC1-Plys-e2-crimson exhibited stagnation of SOS response and prophage induction if a carbon source was absent for 24 h after and before standard cultivation was applied. Prophage induction was not significant under starvation conditions, wheras rare SOS response was present as shown in Video 7. The video was published in 22.

Video 7: A dual reporter strain colony with appearance of a SOS+/phage- under starvation condition under intermittent carbon supply

316 Appendix

Figure S1: SOS response and prophage induction distribution of C. glutamicum ATCC 13032::PrecA-venus/pJC1-Plys-e2-crimson under intermittent hunger condition. (n = 25 colonies with 10 h intermittent hunger phase)

8.2 Supplementary Videos and Supplementary Text of ROS Visualization in Living Bacteria

The knockout mutant ΔctaD showed without thiamine addition different spontaneous ROS development, wheras wether the whole colony initiated simultaneous CALox fluorescence as shown in Video 8 (after ~ 14.5 h) and Video 9 (after ~ 12.7 h), or no ROS formation appeared as shown in Video 10, or flickering CALox fluorescence in non-growing cells as observed in Video 11. The addition of the radical scavenger thiamine reduced CALox fluorescence to rare fluorescence foci in the polar region of single cells.

317 Appendix

Video 8: Simultaneous ROS formation in a ΔctaD colony after 14.5 h of cultivation Video 9: Simultaneous ROS formation in a ΔctaD colony after 12.7 h of cultivation Video 10: Absence of ROS formation in a growing ΔctaD population Video 11: Unsteady ROS formation visualised by flickering CALox fluorescence in ΔctaD cells Video 12 Prevented ROS formation in ΔctaD cells in presence of 0.2 µg/L thiamine Video 13 Prevented ROS formation in ΔctaD cells in presence of 200 µg/L thiamine The knockout mutant Δqcr showed no CALox fluorescence without thiamine addition and sporadic initial ROS prescence that disappeared before cell division was initiated with addition of thiamine.

Video 14: Absence of ROS formation in a growing Δqcr population Video 15: Disappearence of initial ROS in a single Δqcr cell and proceeded cell division in presence of 0.2 µg/L thiamine Video 16: Disappearence of initial ROS in a single Δqcr cell and proceeded cell division in presence of 200 µg/L thiamine

8.3 Supplementary Videos of Metabolic Activity Sensing

The metabolic activity sensing over time is presented by exemplary videos under different conditions described more in detail in the results part of the publication. The single cell time lapse movies give 318 Appendix an impression of the real-time monitoring of the dynamic growth and metabolic activity changes. The colony growth and calcein fluorescence is shown in comparison using the complex medium brain heart infusion (BHI, BD, Germany) and the minimal medium CGXII as described by Keilhauer et al. (1993) 70, respectively. All videos of 8.3 were previously published in 16. Video 17: Metabolic activity sensing under reference conditions. Example colonies growing in BHI medium with the pH 6.6, pH 7.0 and pH 7.4 are given in comparison, respectively. Video 18: Metabolic activity sensing at different media pH. The cell growth and calcein fluorescence of different temporary nutrient deprived colonies grown in minimal medium CGXII is compared to the undisturbed standard cultivation condition with CGXII + 4 % glucose (CGXII + 4 % GLC. After 3.7 hours, the perfusion medium was switched to procatechuate (PCA) free conditions (CGXII + 4 % GLC – PCA), CGXII medium without iron (CGXII + 4 % GLC – iron), or without carbon (CGXII – PCA). The limitation of iron and carbon arrested the growth of C. glutamicum cells until full media supply was returned after 15.2 hours. Under limitation of iron chelator and after resupply of full nutrient supply following iron limitation a phenotype of elongated cells frequently could be observed. Bursting cells only appeared after returning iron containing perfusion medium to iron depleted cells. Non-growing

319 Appendix bacteria with clearly remained metabolic activity could be found under all three conditions of nutrient limitation. Video 19: Intermittent iron supply Video 20: Elongated cells after iron depletion Video 21: Intermittent carbon supply Video 22: Non-growing cells after carbon depletion Video 23: Intermittent iron chelator supply Video 24: Intermittent iron supply with single cell events Video 25: Bursting iron depleted cells Video 26: Spontaneous non-growing cell after iron depletion The impact on single cell growth and calcein fluorescence by short term exposure of antibiotics at a concentration of 10 µg/mL are shown for bactericidal ampicillin and bacteriostatic chloramphenicol. The growth arrest and cell deforming effect of AMP was stronger compared to CHL at the same concentration. Non-viable cells were shrinking and fading.

Video 27: Short term growth impairment by AMP Video 28: Short term growth impairment by CHL

8.4 Supplementary Videos of Dynamic Viability Staining

All videos of 8.3 were previously published in 17.

Video 29: Cell death and antibiotic tolerance of Corynebacterium glutamicum cells after the addition of antibiotics.

320 Appendix

Phenotypic analysis of C. glutamicum colonies during the continuous addition of antibiotics revealed a heterogeneous cell response. Cell death and extracellular DNA derived from lysed cells are indicated by red PI fluorescence. Lysed cells exhibited no fluorescence. Cells possessing residual metabolic activity indicated by an increase in blue CALv fluorescence remain in an antibiotic-tolerant state. A reference colony without the addition of antibiotics is shown for comparison. Video 30: Programmed cell death of Escherichia coli. The initiation of single-cell death in E. coli BL21CodonPlus(DE3)- RIL cells producing a truncated lytic zeta toxin (PezT), an inactivated PezT or the antitoxin-toxin complex are indicated by red PI fluorescence. After the induction of toxin expression with IPTG, cell survival (resistance) and the prior induction of non-lytic cell death were observed in a subpopulation that did not undergo complete cell lysis. Non-lytic single-cell death was also observed in E. coli cells expressing the inactivated toxin. Video 31: Macroautophagy in Saccharomyces cerevisiae. Cell death (indicated by red PI fluorescence) after the induction of autophagy. PO-PRO-1 staining (blue fluorescence) indicates the loss of membrane potential and the presence of DNA fractions in the autophagosome. CALg (green fluorescence) accumulates in the autophagosome. CALg was generated from CgAM by intracellular esterases and was retained if no ATPase-mediated active transport was present.

321 Appendix

Video 32: Ageing in Saccharomyces cerevisiae. Cell ageing of yeast is indicated by the enlargement and loss of function of vacuoles, which induces cell death. CALg (green fluorescence) accumulated in vacuoles that lost their vATPase activity. Cell death is indicated by PI (red fluorescence). Video 33: Dying zygote. The sudden cell death (red PI fluorescence) of a bud can initiate a loss of function (green CALg fluorescence) and subsequent cell death. Video 34: Shmoo-mediated mating of aged cells. Vacuole enlargement following nutrient stress likely appeared due to the ageing of yeast cells, resulting in cell death (red PI fluorescence). If shmoos mate with aged cells during programmed cell death, which initiates upon membrane potential loss (blue PO-PRO-1 fluorescence) followed by cell disintegration (red PI fluorescence), the mating shmoo dies following the cell death of the mating partner. Video 35: Fluctuation in membrane potential followed by cell recovery. Saccharomyces cerevisiae cell recovery was observed following PO- PRO-1 (blue fluorescence) uptake. Cell budding was initiated, and the PO-PRO-1 that intercalated with the DNA was distributed between the mother and daughter cells.

322 Appendix

8.5 Summary of previous findings of A. Koch-Koerfges

Knockout strains C. glutamicum ATCC 13032 ΔctaD and C. glutamicum ATCC 13032 Δqcr were provided by A. Koch- Koerfges (Research group of Prof. Dr. Michael Bott). Results, figures, and tables kindly provided from A. Koch-Koerfges 207.

It was shown that both supercomplex mutants showed a by 50% impaired growth compared to the wild type on complex medium 201,205. However, on CGXII minimal medium with 4 % (w/v) GLC as sole carbon and energy source, growth of ΔctaD was nearly absent in static shaking culutes, while Δqcr showed 50% of wild type again. Growth of ΔctaD could be reserved by addition of thiamine, which has been describes as an inhibitor of oxidative stress and lipid peroxidation 207. Koch-Koerfges (2011) could show increased ROS production in both mutants with TBARs assay 207.

323 Appendix

Table S1. Growth parameters of C. glutamicum ATCC 13032 wild type, ΔctaD, and Δqcr during cultivation in shaking flasks in CGXII + 222 mM GLC. In case of ΔctaD the values for thiamine supplemented medium are also shown. Each value represents the mean of at least 3 independent experiments and the corresponding standard deviation (σ). Kindly provided by A. Koch-Koerfges 207. Parameter wt σ ΔctaD σ ∆ctaD + σ ∆qcr σ thiamine

OD600 67 ± 6 3.54 ± 0.61 44 ± 3 39 ± 4

µ (h-1) 0.4 ±0.01 0.12 ± 0.01 0.33 ± 0.01 0.22 ± 0.04

324 Appendix

Figure S2: Schematic overview of the respiratory chain and oxidative 204 phosphorylation of C. glutamicum based on . The F1FO-ATP synthase presumably requires 3-4 H+ for the synthesis of one molecule ATP. Glucose oxidation in the central metabolism is indicated and the dehydrogenases of the respiratory chain that are also part of the TCA cycle are highlighted in orange. DH, dehydrogenase; G6P, glucose 6-phosphate; MK, menaquinone; OR, oxidoreductase; PDHc, Pyruvate dehydrogenase complex; PTSGlc, glucose specific permease of the phosphotransferase system; TCA, tricarobxylic acid. Kindly provided by A. Koch-Koerfges 207.

325 Appendix

Table S2. TBARs formation in C. glutamicum ATCC13032 wild type, and its ΔctaD, Δqcr, and ΔqcrA mutants. Mean values and standard deviation (σ) of three independent experiments are given. Kindly provided by A. Koch-Koerfges 207.

TBARs (µM)

8h σ 14 h σ 24 h σ 48 h σ

WT 0.7 ±0.1 0.6 ±0.4 1.9 ±1.0 1.2 ±1.3

Δqcr 1.5 ±0.1 4.0 ±0.8 2.6 ±1.4 2.4 ±0.9

ΔqcrA 1.2 ±0.2 4.0 ±0.7 2.6 ±0.1 2.8 ±0.3

ΔctaD 2.3 ±0.3 4.2 ±0.7 7.2 ±1.4 5.2 ±0.2

ΔctaD + thiamine 1.3 ±0.1 2.5 ±0.1 1.9 ±0.7 1.3 ±0.5

326 Appendix

Figure S3: Scheme of radical formation in C. glutamicum ΔctaD due to the presence and activity of the two cytochrome bc1 complex subunits QcrA and QcrB (bL = low potential heme, bH = high potential heme). Arrows designate the movement of electrons. The reduction of the iron-sulfur cluster of the Rieske protein QcrA can be catalyzed via electron transfer from MKH2 or by O2- in the Haber-Weiss reaction. The iron-sulfur cluster could react with hydrogen peroxide in the Fenton reaction causing the formation of OH- and OH• (second step in Haber-Weiss reaction). The hydroxyl OH• attacks the unsaturated fatty acids of the cytoplasmic membrane (CM) and the unsaturated mycolic acids of the outer membrane (OM), causing lipid peroxidation (LPO). Due to the extra cytoplasmic localization of the iron-sulfur cluster of the Rieske protein, it is probably not accessible for repair by the Suf system. AG, arabinogalactan complex, Cat, catalase; CM, cytoplasmic membrane; LPO, lipid peroxidation; MK, menaquinone; OM, outer membrane; PG, peptidoglycan; ROS, reactive oxygen species; Sod, superoxide dismutase; Suf system, required for assembly and repair of cytoplasmic iron-sulfur clusters. Kindly provided by A. Koch-Koerfges 207.

327 Appendix

Figure S4: Comparison of C. glutamicum wild type and its ΔctaD, Δqcr and ΔqcrA mutants with respect to growth (A, D, G), glucose consumption (B), lactate consumption (E), dissolved oxygen (C), pyruvate formation (F), and pH of the supernatant (H). The strains were cultivated in glucose minimal medium (A- C), lactate minimal medium (D-F) or BHI complex medium (G, H) in 500 ml shake flasks with a 50 ml culture volume. The wild type is indicated by black squares (-■-), ΔctaD by grey triangles (-▲-), ΔctaD supplemented with 0.2 mg/l thiamine by red circles (-●-), Δqcr by blue circles (-●-) and ΔqcrA by green squares (-■-). The values and standard deviation from of at least three independent cultivations of each strain are shown except for the DO measurements (panel C), where one representative experiment of three independent ones is shown. Kindly provided by A. Koch- Koerfges 207.

328 References

9. References

(1) Lane, N. The unseen world: reflections on Leeuwenhoek (1677) “Concerning little animals.” Philos. Trans. R. Soc. London B 2015, 370 (1677), 20140344 DOI: 10.1098/rstb.2014.0344.

(2) Wollman, A. J. M.; Nudd, R.; Hedlund, E. G.; Leake, M. C. From Animaculum to single molecules: 300 years of the light microscope. Open Biol. 2015, 5, 150019 DOI: 10.1098/rstl.1739.0085.

(3) Coutu, D. L.; Schroeder, T. Probing cellular processes by long- term live imaging-historic problems and current solutions. J. Cell Sci. 2013, 126 (Pt 17), 3805–3815 DOI: 10.1242/jcs.118349.

(4) Stramer, B. M.; Dunn, G. A. Cells on film - the past and future of cinemicroscopy. J. Cell Sci. 2015, 128 (1), 9–13 DOI: 10.1242/jcs.165019.

(5) Malone, R. H. A simple apparatus for isolating single organisms. J. Pathol. Bacteriol. 1919, 22 (2), 222–223 DOI: 10.1002/path.1700220207.

329 References

(6) Schlimpert, S.; Flärdh, K.; Buttner, M. Fluorescence Time- lapse Imaging of the Complete S. venezuelae Life Cycle Using a Microfluidic Device. J. Vis. Exp. 2016, No. 108, e53863 DOI: 10.3791/53863.

(7) Skinner, S. O.; Sepúlveda, L. A.; Xu, H.; Golding, I. Measuring mRNA copy number in individual Escherichia coli cells using single-molecule fluorescent in situ hybridization. Nat. Protoc. 2013, 8 (6), 1100–1113 DOI: 10.1038/nprot.2013.066.

(8) Sass, P.; Josten, M.; Famulla, K.; Schiffer, G.; Sahl, H.-G.; Hamoen, L.; Brötz-Oesterhelt, H. Antibiotic acyldepsipeptides activate ClpP peptidase to degrade the cell division protein FtsZ. Proc. Natl. Acad. Sci. U. S. A. 2011, 108 (42), 17474– 17479 DOI: 10.1073/pnas.1110385108.

(9) Grünberger, A.; Probst, C.; Heyer, A.; Wiechert, W.; Frunzke, J.; Kohlheyer, D. Microfluidic Picoliter Bioreactor for Microbial Single Cell Analysis: Fabrication, System Setup and Operation. J. Vis. Exp. 2013, 82, e50560.

(10) Dusny, C.; Grünberger, A.; Probst, C.; Wiechert, W.; Kohlheyer, D.; Schmid, A. Technical bias of microcultivation environments on single-cell physiology. Lab Chip 2015, 15 (8), 1822–1834 DOI: 10.1039/c4lc01270d.

(11) Haagensen, J. A. J.; Regenberg, B.; Sternberg, C. Advanced Microscopy of Microbial Cells. In Adv. Biochem. 330 References

Engin./Biotechnol.; 2010; Vol. 124, pp 21–54.

(12) Turkowyd, B.; Virant, D.; Endesfelder, U. From single molecules to life: microscopy at the nanoscale. Anal. Bioanal. Chem. 2016, 408 (25), 6885–6911 DOI: 10.1007/s00216-016- 9781-8.

(13) Tsien, R. Y. The Green Fluorescent Protein. Annu. Rev. Biochem. 1998, 67, 509–544 DOI: 10.1146/annurev.biochem.67.1.509.

(14) Schneider, J. P.; Basler, M. Shedding light on biology of bacterial cells. Philos. Trans. R. Soc. B Biol. Sci. 2016, 371 (1707), 0–10 DOI: 10.1098/rstb.2015.0499.

(15) Wouters, K.; Maes, E.; Spitz, J. A.; Roeffaers, M. B. J.; Wattiau, P.; Hofkens, J.; Springael, D. A non-invasive fluorescent staining procedure allows Confocal Laser Scanning Microscopy based imaging of Mycobacterium in multispecies biofilms colonizing and degrading polycyclic aromatic hydrocarbons. J. Microbiol. Methods 2010, 83 (3), 317–325 DOI: 10.1016/j.mimet.2010.09.014.

(16) Krämer, C.; Singh, A.; Helfrich, S.; Grünberger, A.; Wiechert, W.; Nöh, K.; Kohlheyer, D. Non-invasive microbial metabolic activity sensing at single cell level by perfusion of calcein acetoxymethyl ester. PLoS One 2015, 10 (10), e0141768.

331 References

(17) Krämer, C. E. M.; Wiechert, W.; Kohlheyer, D. Time-resolved, single-cell analysis of induced and programmed cell death via non-invasive propidium iodide and counterstain perfusion. Sci. Rep. 2016, 6 (August), 32104 DOI: 10.1038/srep32104.

(18) Daddi Oubekka, S.; Briandet, R.; Fontaine-Aupart, M. P.; Steenkeste, K. Correlative time-resolved fluorescence microscopy to assess antibiotic diffusion-reaction in biofilms. Antimicrob. Agents Chemother. 2012, 56 (6), 3349–3358 DOI: 10.1128/AAC.00216-12.

(19) Golchin, S. A.; Stratford, J.; Curry, R. J.; McFadden, J. A microfluidic system for long-term time-lapse microscopy studies of mycobacteria. Tuberculosis (Edinb). 2012, 92 (6), 489–496 DOI: 10.1016/j.tube.2012.06.006.

(20) Tomokuni, Y.; Goryo, K.; Katsura, A.; Torii, S.; Yasumoto, K.; Kemnitz, K.; Takada, M.; Fukumura, H.; Sogawa, K. Loose interaction between glyceraldehyde-3-phosphate dehydrogenase and phosphoglycerate kinase revealed by fluorescence resonance energy transfer-fluorescence lifetime imaging microscopy in living cells. FEBS J. 2010, 277 (5), 1310–1318 DOI: 10.1111/j.1742-4658.2010.07561.x.

(21) Patteson, A. E.; Gopinath, A.; Goulian, M.; Arratia, P. E. Running and tumbling with E. coli in polymeric solutions. Sci. Rep. 2015, 5, 15761 DOI: 10.1038/srep15761.

332 References

(22) Helfrich, S.; Pfeifer, E.; Krämer, C.; Sachs, C. C.; Wiechert, W.; Kohlheyer, D.; Nöh, K.; Frunzke, J. Live cell imaging of SOS and prophage dynamics in isogenic bacterial populations. Mol. Microbiol. 2015, 98 (4), 636–650 DOI: 10.1111/mmi.13147.

(23) Mustafi, N.; Grünberger, A.; Mahr, R.; Helfrich, S.; Nöh, K.; Blombach, B.; Kohlheyer, D.; Frunzke, J. Application of a Genetically Encoded Biosensor for Live Cell Imaging of L- Valine Production in Pyruvate Dehydrogenase Complex- Deficient Corynebacterium glutamicum Strains. PLoS One 2014, 9 (1), e85731 DOI: 10.1371/journal.pone.0085731.

(24) Contag, C. H.; Contag, P. R.; Mullins, J. I.; Spilman, S. D.; Stevenson, D. K.; Benaron, D. A. Photonic detection of bacterial pathogens in living hosts. Mol. Microbiol. 1995, 18 (4), 593–603 DOI: 10.1111/j.1365- 2958.1995.mmi_18040593.x.

(25) Waidmann, M. S.; Bleichrodt, F. S.; Laslo, T.; Riedel, C. U. Bacterial luciferase reporters: the Swiss army knife of molecular biology. Bioeng. Bugs 2011, 2 (1), 8–16 DOI: 10.4161/bbug.2.1.13566.

(26) Amann, R.; Fuchs, B. M. Single-Cell Identification in Microbial Communities by Improved Fluorescence In Situ Hybridization Techniques. Nat. Rev. 2008, 6 (5), 339–348.

(27) Wendeberg, A. Fluorescence In Situ Hybridization for the 333 References

Identification of Environmental Microbes. Cold Spring Harb. Protoc. 2010, 2010 (1), 8.

(28) Krämer, C. E. M.; Wiechert, W.; Kohlheyer, D. Artificial fluorogenic substrates in microfluidic devices for bacterial diagnostics in biotechnology. J. Flow Chem. 2016, 6 (1), 3–7 DOI: 10.1556/1846.2015.00035.

(29) Errington, J. L-form bacteria, cell walls and the origins of life. Open Biol. 2013, 3 (1), 120143 DOI: 10.1098/rsob.120143.

(30) Nanda, A. M.; Heyer, A.; Krämer, C.; Grünberger, A.; Kohlheyer, D.; Frunzke, J. Analysis of SOS-induced spontaneous prophage induction in Corynebacterium glutamicum at the single-cell level. J. Bacteriol. 2014, 196 (1), 180–188 DOI: 10.1128/JB.01018-13.

(31) Grünberger, A. M. Single-Cell Analysis of Microbial Production Strains in Microfluidic Bioreactors Single-Cell Analysis of Microbial Production Strains in Microfluidic Bioreactors, RWTH Aachen, 2015, Vol. 114.

(32) Probst, C.; Grünberger, A.; Wiechert, W.; Kohlheyer, D. Polydimethylsiloxane (PDMS) sub-micron traps for single-cell analysis of bacteria. Micromachines 2013, 4 (4), 357–369 DOI: 10.3390/mi4040357.

(33) Sachs, C. C.; Gr??nberger, A.; Helfrich, S.; Probst, C.;

334 References

Wiechert, W.; Kohlheyer, D.; N??h, K. Image-based single cell profiling: High-throughput processing of mother machine experiments. PLoS One 2016, 11 (9), 1–15 DOI: 10.1371/journal.pone.0163453.

(34) Grünberger, A.; Paczia, N.; Probst, Christopher Schendzielorz, Georg Eggeling, L.; Noack, S.; Wiechert, Wolfgang Kohlheyer, D. A disposable picolitre bioreactor for cultivation and investigation of industrially relevant bacteria on the single cell level. Lab Chip 2012, 12 (11), 9 DOI: 10.1039/c2lc40156h.

(35) Gruenberger, A.; Probst, C.; Heyer, A.; Wiechert, W.; Frunzke, J.; Kohlheyer, D. Microfluidic picoliter bioreactor for microbial single-cell analysis: fabrication, system setup, and operation. J. Vis. Exp. 2013, 82, 50560 DOI: 10.3791/50560.

(36) Grünberger, A.; Probst, C.; Helfrich, S.; Nanda, A.; Stute, B.; Wiechert, W.; Lieres, E. von; Nöh, K.; Frunzke, J.; Kohlheyer, D. Spatiotemporal microbial single-cell analysis using a high- throughput microfluidics cultivation platform. J. Cytom. A 2015, 87 (12), 1101–1115.

(37) Probst, C.; Grünberger, A.; Wiechert, W.; Kohlheyer, D. Microfluidic growth chambers with optical tweezers for full spatial single-cell control and analysis of evolving microbes. J. Microbiol. Methods 2013, 95 (3), 470–476 DOI: 10.1016/j.mimet.2013.09.002.

335 References

(38) Westerwalbesloh, C.; Grünberger, A.; Stute, B.; Weber, S.; Wiechert, W.; Kohlheyer, D.; Lieres, E. Von. Lab on a Chip Modeling and CFD simulation of nutrient distribution in picoliter bioreactors for bacterial growth studies on single-cell level. Lab Chip 2015, 15 (21), 4177–4186 DOI: 10.1039/C5LC00646E.

(39) Grünberger, A.; Ooyen, J. van; Paczia, N.; Rohe, P.; Schendzielorz, G.; Eggeling, L.; Wiechert, W.; Kohlheyer, D.; Noack, S. Beyond Growth Rate 0.6: Corynebacterium glutamicum Cultivated in Highly Diluted Environments. Biotechnol. Bioeng. 2013, 110 (1), 18.

(40) Unthan, S.; Grünberger, A.; van Ooyen, J.; Gätgens, J.; Heinrich, J.; Paczia, N.; Wiechert, W.; Kohlheyer, D.; Noack, S. Beyond growth rate 0.6: What drives Corynebacterium glutamicum to higher growth rates in defined medium. Biotechnol. Bioeng. 2014, 111 (2), 359–371 DOI: 10.1002/bit.25103.

(41) Binder, D.; Probst, C.; Grünberger, A.; Hilgers, F.; Loeschcke, A.; Jaeger, K. E.; Kohlheyer, D.; Drepper, T. Comparative single-cell analysis of different E. Coli expression systems during microfluidic cultivation. PLoS One 2016, 11 (8), 1–19 DOI: 10.1371/journal.pone.0160711.

(42) Baumgart, M.; Unthan, S.; Sivalingam, J.; Bott, M.; Noack, S.;

336 References

Frunzke, J.; Rückert, C.; Sivalingam, J.; Grünberger, A.; Kalinowski, J.; et al. Construction of a prophage-free variant of Corynebacterium glutamicum ATCC 13032 for use as a platform strain for basic research and industrial biotechnology. Appl. Environ. Microbiol. 2013, 79 (19), 6006–6015 DOI: 10.1128/AEM.01634-13.

(43) Scholz, K. E.; Okrob, D.; Kopka, B.; Grünberger, A.; Pohl, M.; Jaeger, K.-E.; Krauss, U. Synthesis of Chiral Cyanohydrins by Recombinant Escherichia coli Cells in a Micro-Aqueous Reaction System. Appl. Environ. Microb. 2012, 78 (14), 3.

(44) Mustafi, N.; Grünberger, A.; Kohlheyer, D.; Bott, M.; Frunzke, J. The Development and Application of a Single-Cell Biosensor for the Detection of L-Methionine and Branched-Chain Amino Acids. Metab. Eng. 2012, 14 (4), 449–457.

(45) Schendzielorz, G.; Dippong, M.; Grünberger, A.; Kohlheyer, D.; Yoshida, A.; Binder, S.; Nishiyama, C.; Nishiyama, M.; Bott, M.; Eggeling, L. Taking Control over Control: Use of Product Sensing in Single Cells to Remove Flux Control at Key Enzymes in Biosynthesis Pathways. ACS Synth. Biol. 2014, 3 (1), 21–29 DOI: 10.1021/sb400059y.

(46) Binder, D.; Grünberger, A.; Loeschcke, A.; Probst, C.; Bier, C.; Pietruszka, J.; Wiechert, W.; Kohlheyer, D.; Jaeger, K.-E.; Drepper, T. Light-responsive control of bacterial gene

337 References

expression: precise triggering of the lac promoter activity using photocaged IPTG. Integr. Biol. (Camb). 2014, 6 (8), 755–765 DOI: 10.1039/c4ib00027g.

(47) Schneider, C. A.; Rasband, W. S.; Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 2012, 9 (7), 671–675 DOI: 10.1038/nmeth.2089.

(48) Helfrich, S. High-Throughput Live-Cell Imaging for Investigations of Cellular Heterogeneity in Corynebacterium glutamicum, RWTH Aachen, 2016, Vol. 130.

(49) Schindelin, J.; Arganda-Carreras, I.; Frise, E.; Kaynig, V.; Longair, M.; Pietzsch, T.; Preibisch, S.; Rueden, C.; Saalfeld, S.; Schmid, B.; et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 2012, 9 (7), 676–682 DOI: 10.1038/nmeth.2019.

(50) Helfrich, S.; Azzouzi, C. E.; Probst, C.; Seiffarth, J.; Grünberger, A.; Wiechert, W.; Kohlheyer, D.; Nöh, K. Vizardous : Interactive Analysis of Microbial Populations with Single Cell Resolution. Bioinformatics 2015, 31 (23), 3875– 3877.

(51) Munsky Neuert, G., Oudenaarden, A. van, B. Using Gene Expression Noise to Understand Gene Regulation. Science (80- . ). 2012, 336, 5.

338 References

(52) Elowitz, M. B.; Levine, A. J.; Siggia, E. D.; S, S. P. Stochastic Gene Expression in a Single Cell. Science (80-. ). 2002, 297 (5584), 1183–1187 DOI: 10.1126/science.1070919.

(53) Locke Young, J. W., Fontes, M., Hernández Jiménez, M. J., Elowitz, M. B., J. C. W. Stochastic Pulse Regulation in Bacterial Stress Response. Science (80-. ). 2011, 334 (6054), 366–369.

(54) Kiviet, D. J.; Nghe, P.; Walker, N.; Boulineau, S.; Sunderlikova, V.; Tans, S. J. Stochasticity of metabolism and growth at the single-cell level. Nature 2014, 514 (7522), 376– 379 DOI: 10.1038/nature13582.

(55) Bryant, J. A.; Sellars, L. E.; Busby, S. J. W.; Lee, D. J. Chromosome position effects on gene expression in Escherichia coli K-12. Nucleic Acids Res. 2014, 42 (18), 11383–11392 DOI: 10.1093/nar/gku828.

(56) Uhlendorf, J.; Miermont, A.; Delaveau, T.; Charvin, G.; Fages, F.; Bottani, S.; Batt, G.; Hersen, P. Long-term model predictive control of gene expression at the population and single-cell levels. Proc. Natl. Acad. Sci. U. S. A. 2012, 109 (35), 14271– 14276 DOI: 10.1073/pnas.1206810109.

(57) Sauer, C.; Syvertsson, S.; Bohorquez, L. C.; Cruz, R.; Harwood, C. R.; van Rij, T.; Hamoen, L. W. Effect of Genome Position on Heterologous Gene Expression in Bacillus subtilis : An 339 References

Unbiased Analysis. ACS Synth. Biol. 2016, 5 (9), 942–947 DOI: 10.1021/acssynbio.6b00065.

(58) Kou, S.; Cheng, D.; Sun, F.; Hsing, I. Microfluidics and Microbial Engineering. Lab Chip 2015, 16 (3), 432–446 DOI: 10.1039/C5LC01039J.

(59) Barbier, M.; Damron, F. H. Rainbow vectors for broad-range bacterial fluorescence labeling. PLoS One 2016, 11 (3), 1–16 DOI: 10.1371/journal.pone.0146827.

(60) Liu, D.; Evans, T.; Zhang, F. Applications and advances of metabolite biosensors for metabolic engineering. Metab. Eng. 2015, 31, 35–43 DOI: 10.1016/j.ymben.2015.06.008.

(61) Contag, C. H.; Bachmann, M. H. Advances in In Vivo Bioluminescence Imaging of Gene Expression. Annu. Rev. Biomed. Eng. 2002, 4 (1), 235–260 DOI: 10.1146/annurev.bioeng.4.111901.093336.

(62) Young, J. W.; Locke, J. C. W.; Altinok, A.; Rosenfeld, N.; Bacarian, T.; Swain, P. S.; Mjolsness, E.; Elowitz, M. B. Measuring single -cell gene expression dynamics in bacteria using fluorescence time-lapse microscopy. Nat. Protoc. 2012, 7 (1), 80–88 DOI: 10.1038/nprot.2011.432.

(63) Hashimoto, M.; Nozoe, T.; Nakaoka, H.; Okura, R.; Akiyoshi, S.; Kaneko, K.; Kussell, E.; Wakamoto, Y. Noise-driven growth

340 References

rate gain in clonal cellular populations. Proc. Natl. Acad. Sci. U. S. A. 2016, 113 (12), 3251–3256 DOI: 10.1073/pnas.1519412113.

(64) Long, Z.; Olliver, A.; Brambilla, E.; Sclavi, B.; Lagomarsino, M. C.; Dorfman, K. D. Measuring bacterial adaptation dynamics at the single-cell level using a microfluidic chemostat and time-lapse fluorescence microscopy. Analyst 2014, 139 (20), 5254–5262 DOI: 10.1039/c4an00877d.

(65) Geiler-Samerotte, K. A.; Bauer, C. R.; Li, S.; Ziv, N.; Gresham, D.; Siegal, M. L. The details in the distributions: Why and how to study phenotypic variability. Current Opinion in Biotechnology. 2013, pp 752–759.

(66) Young, J. W.; Elowitz, M. B. Mixed Messages: How Bacteria Resolve Conflicting Signals. Mol. Cell 2011, 42, 2 DOI: 10.1016/j.molcel.2011.05.005.

(67) Schlüter, J. P.; Czuppon, P.; Schauer, O.; Pfaffelhuber, P.; McIntosh, M.; Becker, A. Classification of phenotypic subpopulations in isogenic bacterial cultures by triple promoter probing at single cell level. J. Biotechnol. 2015, 198, 3–14 DOI: 10.1016/j.jbiotec.2015.01.021.

(68) Dhar, N.; McKinney, J. D. Microbial Phenotypic Heterogeneity and Antibiotic Tolerance. Curr. Opin. Microbiol. 2007, 10 (1), 9 DOI: 10.1016/j.mib.2006.12.007. 341 References

(69) Young, J. W.; Locke, J. C. W.; Elowitz, M. B. Rate of Environmental Change Determines Stress Response Specifity. PNAS 2013, 110 (10), 4140–4145.

(70) Keilhauer, C.; Eggeling, L.; Sahm, H. Isoleucine Synthesis in Corynebacterium glutamicum: Molecular Analysis of the ilvB- ilvN-ilvC Operon. J. Bacteriol. 1993, 175 (17), 5595–5603.

(71) Nanda, A. M. Characterization of the spontaneous SOS response in Corynebacterium glutamicum and its effect on the lysogenic prophage CGP3, Heinrich Heine Universität Düsseldorf, 2015.

(72) Pfeifer, E.; Hünnefeld, M.; Popa, O.; Polen, T.; Kohlheyer, D.; Baumgart, M.; Frunzke, J. Silencing of cryptic prophages in Corynebacterium glutamicum. Nucleic Acids Res. 2016, 44 (21), 10117–10131 DOI: 10.1093/nar/gkw692.

(73) Frunzke, J.; Bramkamp, M.; Schweitzer, J.-E.; Bott, M. Population Heterogeneity in Corynebacterium glutamicum ATCC 13032 Caused by Prophage CGP3. J. Bacteriol. 2008, 190 (14), 9.

(74) Studier, F. W.; Daegelen, P.; Lenski, R. E.; Maslov, S.; Kim, J. F. Understanding the Differences between Genome Sequences of Escherichia coli B Strains REL606 and BL21 ( DE3 ) and Comparison of the E . coli B and K-12 Genomes. J. Mol. Biol. 2009, 394 (4), 653–680 DOI: 10.1016/j.jmb.2009.09.021. 342 References

(75) Mutschler, H.; Gebhardt, M.; Shoeman, R. L.; Meinhart, A. A Novel Mechanism of Programmed Cell Death in Bacteria by Toxin-Antitoxin Systems Corrupts Peptidoglycan Synthesis. PLOS Biol. 2011, 9 (3), e1001033.

(76) Ziegelhoffer, E. C.; Donohue, T. J. Bacterial responses to photo-oxidative stress. Nat. Rev. Microbiol. 2009, 7 (12), 856– 863 DOI: 10.1038/nrmicro2237.

(77) Breeuwer, P.; Drocourt, J.; Rombouts, F. M.; Breeuwer Drocourt, J.-L., Rombouts, F. M., Abee, T., P.; Breeuwer, P.; Drocourt, J.; Rombouts, F. M. A Novel Method for Continuous Determination of the Intracellular pH in Bacteria with the Internally Conjugated Fluorescent Probe 5 (and 6-)- Carboxyfluorescein Succinimidyl Ester. Appl. Environ. Microb. 1996, 62 (1), 6.

(78) Balaban, N. Q.; Gerdes, K.; Lewis, K.; McKinney, J. D. A problem of persistence: still more questions than answers? Nat. Rev. Microbiol. 2013, 11 (8), 587–591 DOI: 10.1038/nrmicro3076.

(79) Beaumont, H. J. E.; Gallie, J.; Kost, C.; Ferguson, G. C.; Rainey, P. B. Experimental evolution of bet hedging. Nature 2009, 462 (7269), 90–93 DOI: 10.1038/nature08504.

(80) Krämer, C.; Kohlheyer, D. Mikrobiologische Einzelzell- Phänotypencharakterisierung im Mikrochip. BioSpektrum 343 References

2016, 22 (1), 48–50 DOI: 10.1007/s12268-016-0658-3.

(81) Zimmer, M. GFP: from jellyfish to the Nobel prize and beyond. Chem. Soc. Rev. 2009, 38 (10), 2823 DOI: 10.1039/b904023d.

(82) Rahmen, N.; Schlupp, C. D.; Mitsunaga, H.; Fulton, A.; Aryani, T.; Esch, L.; Schaffrath, U.; Fukuzaki, E.; Jaeger, K.-E.; Büchs, J. A particular silent codon exchange in a recombinant gene greatly influences host cell metabolic activity. Microb. Cell Fact. 2015, 14 (1), 156 DOI: 10.1186/s12934-015-0348-8.

(83) Tracy, B. P.; Gaida, S. M.; Papoutsakis, E. T. Flow cytometry for bacteria: enabling metabolic engineering, synthetic biology and the elucidation of complex phenotypes. Curr. Opin. Biotechnol. 2010, 21 (1), 85–99 DOI: 10.1016/j.copbio.2010.02.006.

(84) Díaz, M.; Herrero, M.; García, L. A.; Quirós, C. Application of flow cytometry to industrial microbial bioprocesses. Biochem. Eng. J. 2010, 48 (3), 385–407 DOI: 10.1016/j.bej.2009.07.013.

(85) Byrd, T. F.; Hoang, L. T.; Kim, E. G.; Pfister, M. E.; Werner, E. M.; Arndt, S. E.; Chamberlain, J. W.; Hughey, J. J.; Nguyen, B. a; Schneibel, E. J.; et al. The microfluidic multitrap nanophysiometer for hematologic cancer cell characterization reveals temporal sensitivity of the calcein-AM efflux assay. Sci. Rep. 2014, 4, doi:10.1038/srep05117 DOI: 10.1038/srep05117.

344 References

(86) Wlodkowic, D.; Skommer, J.; McGuinness, D.; Faley, S.; Kolch, W.; Darzynkiewicz, Z.; Cooper, J. M. Chip-Based Dynamic Real-Time Quantification of Drug-Induced Cytotoxicity in Human Tumor Cells. Anal. Chem. 2009, 81 (16), 6952–6959.

(87) Hendon-Dunn, C. L.; Doris, K. S.; Thomas, S. R.; Allnutt, J. C.; Marriott, A. A. N.; Hatch, K. A.; Watson, R. J.; Bottley, G.; Marsh, P. D.; Taylor, S. C.; et al. A Flow Cytometry Method for Rapidly Assessing Mycobacterium tuberculosis Responses to Antibiotics with Different Modes of Action. Antimicrob. Agents Chemother. 2016, 60 (7), 3869–3883 DOI: 10.1128/AAC.02712-15.

(88) Galhardo, R. S.; Hastings, P. J.; Rosenberg, S. M. Mutation as a Stress Response and the Regulation of Evolvability. Crit. Rev. Biochem. Mol. 2007, 42, 37.

(89) Michel, B. After 30 years of study, the bacterial SOS response still surprises us. PLoS Biol. 2005, 3 (7), e255 DOI: 10.1371/journal.pbio.0030255.

(90) Witkin, E. M. RecA Protein in the SOS Response: Milestones and Mysteries. Biochimie 1991, 73, 9.

(91) Donovan, C.; Heyer, A.; Pfeifer, E.; Polen, T.; Wittmann, A.; Kr??mer, R.; Frunzke, J.; Bramkamp, M. A prophage-encoded actin-like protein required for efficient viral DNA replication in 345 References

bacteria. Nucleic Acids Res. 2015, 43 (10), 5002–5016 DOI: 10.1093/nar/gkv374.

(92) Wennerhold, J.; Krug, A.; Bott, M. The AraC-type Regulator RipA Represses Aconitase and Other Iron Proteins from Corynebacterium under Iron Limitation and Is Itself Repressed by DtxR. J. Biol. Chem. 2005, 280 (49), 40500–40508 DOI: 10.1074/jbc.M508693200.

(93) Jungwirth, B.; Emer, D.; Brune, I.; Hansmeier, N.; Pühler, A.; Eikmanns, B. J.; Tauch, A. Triple transcriptional control of the resuscitation promoting factor 2 (rpf2) gene of Corynebacterium glutamicum by the regulators of acetate metabolism RamA and RamB and the cAMP-dependent regulator GlxR. FEMS Microbiol. Lett. 2008, 281 (2), 190–197 DOI: 10.1111/j.1574-6968.2008.01098.x.

(94) Berenstein. Prophage Induction by Ultraviolet Light in Acinetobucter calcoaceticus. J. Gen. Microbiol. 1986, 132, 4.

(95) Amir, A.; Kobiler, O.; Rokney, A.; Oppenheim, A. B.; Stavans, J. Noise in Timing and Precision of Gene Activities in a Genetic Cascade. Mol. Syst. Biol. 2007, 3 (71), 10.

(96) Pennington, J. M.; Rosenberg, S. M. Spontaneous DNA breakage in single living Escherichia coli cells. Nat. Genet. 2007, 39 (6), 797–802 DOI: 10.1038/ng2051.

346 References

(97) Shimoni, Y.; Altuvia, S.; Margalit, H.; Biham, O. Stochastic analysis of the SOS response in Escherichia coli. PLoS One 2009, 4 (5) DOI: 10.1371/journal.pone.0005363.

(98) Nagai, T.; Ibata, K.; Park, E. S.; Kubota, M.; Mikoshiba, K.; Mijawaki, A. A Variant of Yellow Fluorescent Protein with Fast and Efficient Maturation for Cell-Biological Applications. Nat. Biotechnol. 2002, 20, 4.

(99) Donovan, C.; Bramkamp, M. Cell division in Corynebacterineae. Front. Microbiol. 2014, 5 (April), 132 DOI: 10.3389/fmicb.2014.00132.

(100) Ogino, H.; Teramoto, H.; Inui, M.; Yukawa, H. DivS, a novel SOS-inducible cell-division suppressor in Corynebacterium glutamicum. Mol. Microbiol. 2008, 67 (3), 597–608 DOI: 10.1111/j.1365-2958.2007.06069.x.

(101) Yamamotoya, T.; Dose, H.; Tian, Z.; Fauré, A.; Toya, Y.; Honma, M.; Igarashi, K.; Nakahigashi, K.; Soga, T.; Mori, H.; et al. Glycogen Is the Primary Source of Glucose during the Lag Phase of E. coli proliferation. Biochim. Biophys. Acta 2012, 1824, 7.

(102) Seibold, G.; Dempf, S.; Schreiner, J.; Eikmanns, B. J. Glycogen formation in Corynebacterium glutamicum and role of ADP- glucose pyrophosphorylase. Microbiology 2007, 153 (4), 1275– 1285 DOI: 10.1099/mic.0.2006/003368-0. 347 References

(103) McCool, J. D.; Long, E.; Petrosino, J. F.; Sandler, H. A.; Rosenberg, S. M.; Sandler, S. J. Measurement of SOS expression in individual Escherichia coli K-12 cells using fluorescence microscopy. Mol. Microbiol. 2004, 53 (5), 1343– 1357 DOI: 10.1111/j.1365-2958.2004.04225.x.

(104) Singletary, L. A.; Gibson, J. L.; Tanner, E. J.; McKenzie, G. J.; Lee, P. L.; Gonzalez, C.; Rosenberg, S. M. An SOS-regulated type 2 toxin-antitoxin system. J. Bacteriol. 2009, 191 (24), 7456–7465 DOI: 10.1128/JB.00963-09.

(105) Groman, N. B. Evidence for the active role of bacteriophage in the conversion of nontoxigenic Corynebacterium diphtheriae to toxin production. J. Bacteriol. 1955, 69 (1), 9–15.

(106) Casjens, S. Prophages and bacterial genomics: What have we learned so far? Mol. Microbiol. 2003, 49 (2), 277–300 DOI: 10.1046/j.1365-2958.2003.03580.x.

(107) Aertsen, A.; Houdt, R. Van. An SOS response induced by high pressure in Escherichia coli. J. … 2004, 186 (18), 6133–6141 DOI: 10.1128/JB.186.18.6133.

(108) Lemire, S.; Figueroa-bossi, N.; Bossi, L. Bacteriophage crosstalk: Coordination of prophage induction by Trans-Acting antirepressors. PLoS Genet. 2011, 7 (6) DOI: 10.1371/journal.pgen.1002149.

348 References

(109) Engelberg-Kulka, H.; Glaser, G. Addiction Modules and Programmed Cell Death and Antideath in Bacterial Cultures. Annu. Rev. Microbiol. 1999, 53 (29), 43–70.

(110) Kawano, M.; Aravind, L.; Storz, G. An antisense RNA controls synthesis of an SOS-induced toxin evolved from an antitoxin. Mol. Microbiol. 2007, 64 (3), 738–754 DOI: 10.1111/j.1365- 2958.2007.05688.x.

(111) Foster, P. L. Stress-Induced Mutagenesis in Bacteria. Crit. Rev. Biochem. Mol. Biol. 2007, 42 (5), 38.

(112) Shee, C.; Ponder, R.; Gibson, J. L.; Rosenberg, S. M. What limits the efficiency of double-strand break-dependent stress- induced mutation in Escherichia coli? J. Mol. Microbiol. Biotechnol. 2012, 21 (1–2), 8–19 DOI: 10.1159/000335354.

(113) Gibson, J. L.; Lombardo, M.-J.; Thornton, P. C.; Kenneth, H. H.; Galhardo, R. S.; Beadle, B.; Habib, A.; Magner, D. B.; Frost, L. S.; Herman, C.; et al. The σE Stress Response Is Required for Stress-Induced Mutation and Amplification in Escherichia coli. Mol. Microbiol. 2010, 77 (2), 16.

(114) Merchant, S. S.; Helmann, J. D. Elemental Economy: Microbial Strategies for Optimizing Growth in the Face of Nutrient Limitation. Adv. Microb. Physiol. 2012, 60 (91), 120.

(115) Bott, M. Offering surprises: TCA cycle regulation in

349 References

Corynebacterium glutamicum. Trends Microbiol. 2007, 15 (9), 417–425 DOI: 10.1016/j.tim.2007.08.004.

(116) Seibold, G. M.; Hagmann, C. T.; Schietzel, M.; Emer, D.; Auchter, M.; Schreiner, J.; Eikmanns, B. J. The transcriptional regulators RamA and RamB are involved in the regulation of glycogen synthesis in Corynebacterium glutamicum. Microbiology 2010, 156 (Pt 4), 1256–1263 DOI: 10.1099/mic.0.036756-0.

(117) Seibold, G. M.; Breitinger, K. J.; Kempkes, R.; Both, L.; Kramer, M.; Dempf, S.; Eikmanns, B. J. The glgB-encoded glycogen branching enzyme is essential for glycogen accumulation in Corynebacterium glutamicum. Microbiology 2011, 157 (11), 3243–3251 DOI: 10.1099/mic.0.051565-0.

(118) Rittmann, D.; Sorger-Herrmann, U.; Wendisch, V. F. Phosphate starvation-inducible gene ushA encodes a 5’ nucleotidase required for growth of Corynebacterium glutamicum on media with nucleotides as the phosphorus source. Appl. Environ. Microbiol. 2005, 71 (8), 4339–4344 DOI: 10.1128/AEM.71.8.4339-4344.2005.

(119) Woo, H. M.; Noack, S.; Seibold, G. M.; Willbold, S.; Eikmanns, B. J.; Bott, M. Link between Phosphate Starvation and Glycogen Metabolism in Corynebacterium glutamicum, Revealed by Metabolomics. Appl. Environ. Microbiol. 2010, 76

350 References

(20), 6910–6919 DOI: 10.1128/AEM.01375-10.

(120) Li, L.; Rao, N. N.; Kornberg, A. Inorganic polyphosphate essential for lytic growth of phages P1 and fd. Proc. Natl. Acad. Sci. U. S. A. 2007, 104 (6), 1794–1799 DOI: 10.1073/pnas.0610763104.

(121) Manganelli, R. Polyphosphate and stress response in mycobacteria. Mol. Microbiol. 2007, 65 (2), 258–260 DOI: 10.1111/j.1365-2958.2007.05819.x.

(122) Wennerhold, J.; Bott, M. The DtxR Regulon of Corynebacterium glutamicum The DtxR Regulon of Corynebacterium glutamicum †. J. Bacteriol. 2006, 188 (8), 2907–2918 DOI: 10.1128/JB.188.8.2907.

(123) Brune, I.; Werner, H.; Hüser, A. T.; Kalinowski, J.; Pühler, A.; Tauch, A. The DtxR protein acting as dual transcriptional regulator directs a global regulatory network involved in iron metabolism of Corynebacterium glutamicum. BMC Genomics 2006, 7, 21 DOI: 10.1186/1471-2164-7-21.

(124) Zhao, N.; Darby, C. M.; Small, J.; Bachovchin, D. A.; Jiang, X.; Burns-Huang, K. E.; Botella, H.; Ehrt, S.; Boger, D. L.; Anderson, E. D.; et al. Target-based screen against a periplasmic serine protease that regulates intrabacterial ph homeostasis in mycobacterium tuberculosis. ACS Chem. Biol. 2015, 10 (2), 364–371 DOI: 10.1021/cb500746z. 351 References

(125) Shim, J. U.; Olguin, L. F.; Whyte, G.; Scott, D.; Babtie, A.; Abell, C.; Huck, W. T. S.; Hollfelder, F. Simultaneous determination of gene expression and enzymatic activity in individual bacterial cells in microdroplet compartments. J. Am. Chem. Soc. 2009, 131 (42), 15251–15256 DOI: 10.1021/ja904823z.

(126) Ishiguro, K.; Washio, J.; Sasaki, K.; Takahashi, N. Real-time monitoring of the metabolic activity of periodontopathic bacteria. J. Microbiol. Methods 2015, 115, 22–26 DOI: 10.1016/j.mimet.2015.05.015.

(127) Hoefel, D.; Grooby, W. L.; Monis, P. T.; Andrews, S.; Saint, C. P. A comparative study of carboxyfluorescein diacetate and carboxyfluorescein diacetate succinimidyl ester as indicators of bacterial activity. J. Microbiol. Methods 2003, 52 (3), 379–388 DOI: 10.1016/S0167-7012(02)00207-5.

(128) Lyu, F.; Xu, M.; Cheng, Y.; Xie, J.; Rao, J.; Tang, S. K. Y. Quantitative detection of cells expressing BlaC using droplet- based microfluidics for use in the diagnosis of tuberculosis. Biomicrofluidics 2015, 9 (4), 44120 DOI: 10.1063/1.4928879.

(129) Lun, S.; Bishai, W. R. Characterization of a novel cell wall- anchored protein with carboxylesterase activity required for virulence in Mycobacterium tuberculosis. J. Biol. Chem. 2007, 282 (25), 18348–18356 DOI: 10.1074/jbc.M700035200.

352 References

(130) Tallman, K. R.; Beatty, K. E. Far-red fluorogenic probes for esterase and lipase detection. Chembiochem 2015, 16 (1), 70– 75 DOI: 10.1002/cbic.201402548.

(131) Akopian, T.; Kandror, O.; Tsu, C.; Lai, J. H.; Wu, W.; Liu, Y.; Zhao, P.; Park, A.; Wolf, L.; Dick, L. R.; et al. Cleavage specificity of Mycobacterium tuberculosis ClpP1P2 protease and identification of novel peptide substrates and boronate inhibitors with anti-bacterial activity. J. Biol. Chem. 2015, 290 (17), 11008–11020 DOI: 10.1074/jbc.M114.625640.

(132) Vorobyeva, A. G.; Stanton, M.; Godinat, A.; Lund, K. B.; Karateev, G. G.; Francis, K. P.; Allen, E.; Gelovani, J. G.; McCormack, E.; Tangney, M.; et al. Development of a bioluminescent nitroreductase probe for preclinical imaging. PLoS One 2015, 10 (6), e0131037 DOI: 10.1371/journal.pone.0131037.

(133) Shao, Q.; Zheng, Y.; Dong, X.; Tang, K.; Yan, X.; Xing, B. A covalent reporter of ??-lactamase activity for fluorescent imaging and rapid screening of antibiotic-resistant bacteria. Chem. - A Eur. J. 2013, 19 (33), 10903–10910 DOI: 10.1002/chem.201301654.

(134) Burke, H. M.; Gunnlaugsson, T.; Scanlan, E. M. Recent advances in the development of synthetic chemical probes for glycosidase enzymes. Chem. Commun. 2015, 51 (53), 10576–

353 References

10588 DOI: 10.1039/C5CC02793D.

(135) Manafi, M.; Kneifel, W.; Bascomb, S. Fluorogenic and Chromogenic Substrates Used in Bacterial Diagnostics. Microbiol. Rev. 1991, 55 (3), 335–348.

(136) Orenga, S.; James, A. L.; Manafi, M.; Perry, J. D.; Pincus, D. H. Enzymatic substrates in microbiology. J. Microbiol. Methods 2009, 79 (2), 139–155 DOI: 10.1016/j.mimet.2009.08.001.

(137) Veal, D. A.; Deere, D.; Ferrari, B.; Piper, J.; Attfield, P. V. Fluorescence staining and flow cytometry for monitoring microbial cells. Journal of Immunological Methods. 2000, pp 191–210.

(138) Rowland, B. Fluorescence-based detection of lacZ reporter gene expression in intact and viable bacteria including Mycobacterium species. FEMS Microbiol. Lett. 1999, 179 (2), 317–325.

(139) Diaper, J. P.; Edwards, C. The use of fluorogenic esters to detect viable bacteria by flow cytometry. J. Appl. Bacteriol. 1994, 77 (2), 221–228 DOI: 10.1111/j.1365-2672.1994.tb03067.x.

(140) Najah, M.; Calbrix, R.; Mahendra-Wijaya, I.; Beneyton, T.; Griffiths, A.; Drevelle, A. Droplet-based microfluidics platform for ultra-high-throughput bioprospecting of cellulolytic

354 References

microorganisms. Chem. Biol. 2014, 21 (12), 1722–1732.

(141) Hosokawa, M.; Hoshino, Y.; Nishikawa, Y.; Hirose, T.; Yoon, D. H.; Mori, T.; Sekiguchi, T.; Shoji, S.; Takeyama, H. Droplet- based microfluidics for high-throughput screening of a metagenomic library for isolation of microbial enzymes. Biosens. Bioelectron. 2015, 67, 379–385 DOI: 10.1016/j.bios.2014.08.059.

(142) Hendon-Dunn, C. L.; Thomas, S. R.; Taylor, S. C.; Bacon, J. A Flow Cytometry Method for Assessing M. tuberculosis Responses to Antibiotics. Methods Mol. Biol. 2018, 1736, 51– 57 DOI: 10.1007/978-1-4939-7638-6_5.

(143) Di Carlo, D.; Aghdam, N.; Lee, L. P. Single-Cell Enzyme Concentrations, Kinetics, and Inhibition Analysis Using High- Density Hydrodynamic Cell Isolation Arrays Single-Cell Enzyme Concentrations , Kinetics , and Inhibition Analysis Using High-Density Hydrodynamic Cell Isolation Arrays. Anal. Chem. 2006, 78 (14), 4925–4930 DOI: 10.1021/ac060541s.

(144) Gengenbacher, M.; Kaufmann, S. H. E. Mycobacterium tuberculosis: success through dormancy. FEMS Microbiol. Rev. 2012, 36 (3), 514–532 DOI: 10.1111/j.1574- 6976.2012.00331.x.

(145) Shim, J.; Patil, S. N.; Hodgkinson, J. T.; Bowden, S. D.; Spring, D. R.; Welch, M.; Huck, W. T. S.; Hollfelder, F.; Abell, C., J. 355 References

Controlling the Contents of Microdroplets by Exploiting the Permeability. Lab Chip 2011, 11 (6), 1132–1137.

(146) Cellier, M.; Fabrega, O. J.; Fazackerley, E.; James, A. L.; Orenga, S.; Perry, J. D.; Salwatura, V. L.; Stanforth, S. P. 2- Arylbenzothiazole, benzoxazole and benzimidazole derivatives as fluorogenic substrates for the detection of nitroreductase and aminopeptidase activity in clinically important bacteria. Bioorganic Med. Chem. Lett. 2015, doi:10.1016/j.bmcl.2015.10.099 DOI: 10.1016/j.bmc.2011.03.043.

(147) Lambert, P. A. Cellular impermeability and uptake of biocides and antibiotics in Gram-positive bacteria and mycobacteria. J. Appl. Microbiol. Symp. Suppl. 2002, 92, 46S–54S.

(148) Nagano, K.; Nikaido, H. Kinetic behavior of the major multidrug efflux pump AcrB of Escherichia coli. Proc. Natl. Acad. Sci. U. S. A. 2009, 106 (14), 5854–5858 DOI: 10.1073/pnas.0901695106.

(149) Castañeda-García, A.; Do, T. T.; Blázquez, J. The k+ uptake regulator trka controls membrane potential, pH homeostasis and multidrug susceptibility in mycobacterium smegmatis. J. Antimicrob. Chemother. 2011, 66 (7), 1489–1498 DOI: 10.1093/jac/dkr165.

(150) Kaidoh, K.; Kimura, M.; Miyauchi, S.; Nara, T.; Kamo, N. 356 References

Drug extrusion in Corynebacterium glutamicum. Microb. Drug Resist. 1997, 3 (4), 345–350 DOI: 10.1089/mdr.1997.3.345.

(151) Delcour, A. H. Outer Membrane Permeability and Antibiotic Resistance. Biochim Biophys Acta 2010, 1794 (5), 808–816 DOI: 10.1016/j.bbapap.2008.11.005.Outer.

(152) Turner, R. D.; Vollmer, W.; Foster, S. J. Different walls for rods and balls: The diversity of peptidoglycan. Mol. Microbiol. 2014, 91 (5), 862–874 DOI: 10.1111/mmi.12513.

(153) Grynkiewicz, G.; Poenie, M.; Tsien, R. Y. A New Generation of Ca2 + Indicators with Greatly Improved Fluorescence Properties. J. Biol. Chem. 1985, 260 (6), 3440–3450.

(154) Tsien, R. Y.; Pozzan, T.; Rink, T. J. Calcium homeostasis in intact lymphocytes: Cytoplasmic free calcium monitored with a new, intracellularly trapped fluorescent indicator. J. Cell Biol. 1982, 94 (2), 325–334 DOI: 10.1083/jcb.94.2.325.

(155) Liminga, G.; Nygen, P.; Larsson, R. Microfluorometric Evaluation of Calcein Acetoxymethyl Ester as a Probe for P- Glycoprotein-Mediated Resistance: Effects of Cyclosporin A and Its Nonimmunosuppressice Analogue SDZ PSZ 833. Experimental Cell Research. 1994, pp 291–296.

(156) Ogawa, M.; Kosaka, N.; Regino, C. a S.; Mitsunaga, M.; Choyke, P. L.; Kobayashi, H. High sensitivity detection of

357 References

cancer in vivo using a dual-controlled activation fluorescent imaging probe based on H-dimer formation and pH activation. Mol. Biosyst. 2010, 6 (5), 888–893 DOI: 10.1039/b917876g.

(157) Clementi, E. A.; Marks, L. R.; Roche-Håkansson, H.; Håkansson, A. P. Monitoring changes in membrane polarity, membrane integrity, and intracellular ion concentrations in Streptococcus pneumoniae using fluorescent dyes. J. Vis. Exp. 2014, No. 84, e51008 DOI: 10.3791/51008.

(158) Chitarra, L. G. L.; Breeuwer, P.; Abee, T.; Bulk, R. W. R. The use of fluorescent probes to assess viability of the plant pathogenic bacterium Clavibacter michiganensis subsp. michiganensis by flow cytometry. Fitopatol. Bras. 2006, 31 (4), 349–356.

(159) Yamaguchi, N.; Nasu, M. Flow cytometric analysis of bacterial respiratory and enzymatic activity in the natural aquatic environment. J. Appl. Microbiol. 1997, 83, 43–52.

(160) Pan, Y.; Kaatz, L. Use of Image-Based Flow Cytometry in Bacterial Viability Analysis Using Fluorescent Probes. In Current Protocols in Microbiology; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2012; Vol. 27, p 11.

(161) Adam, G.; Duncan, H. Development of a sensitive and rapid method for the measurement of total microbial activity using fluorescein diacetate ( FDA ) in a range of soils. Soil Biol. 358 References

Biochem. 2001, 33, 943–951.

(162) Tsuji, T.; Kawasaki, Y.; Takeshima, S.; Sekiya, T.; Tanaka, S. A new fluorescence staining assay for visualizing living microorganisms in soil. Appl. Environ. Microbiol. 1995, 61 (9), 3415–3421.

(163) Porter, J.; Diaper, J.; Edwards, C. Direct measurements of natural planktonic bacterial community viability by flow cytometry . These include : Direct Measurements of Natural Planktonic Bacterial Community Viability by Flow Cytometry. Appl. Environ. Microbiol. 1995, 61 (7), 2783–2786.

(164) Schumann, R.; Schiewer, U.; Karsten, U.; Rieling, T. Viability of bacteria from different aquatic habitats. II. Cellular fluorescent markers for membrane integrity and metabolic activity. Aquat. Microb. Ecol. 2003, 32 (2), 137–150 DOI: 10.3354/ame032137.

(165) Tawakoli, P. N.; Al-Ahmad, A.; Hoth-Hannig, W.; Hannig, M.; Hannig, C. Comparison of different live/dead stainings for detection and quantification of adherent microorganisms in the initial oral biofilm. Clin. Oral Investig. 2013, 17 (3), 841–850 DOI: 10.1007/s00784-012-0792-3.

(166) Dive, C.; Cox, H.; Watson, J. V; Workman, P. Polar fluorescein derivatives as improved substrate probes for flow cytoenzymological assay of cellular esterases. Mol. Cell. 359 References

Probes 1988, 2 (2), 131–145 DOI: 10.1016/0890- 8508(88)90035-7.

(167) Jepras, R. I.; Carter, J.; Pearson, S. C.; Paul, F. E.; Wilkinson, M. J. Development of a robust flow cytometric assay for determining numbers of viable bacteria. Appl. Environ. Microbiol. 1995, 61 (7), 2696–2701.

(168) Shen Zhou, N.-Y., Liu, S.-J., X.-H. Degradation and Assimilation of Aromatic Compounds by Corynebacterium glutamicum: Another Potential for Applications for This Bacterium? Appl. Microbiol. Biotechnol. 2012, 95, 13.

(169) Xu, Y.; Wang, S. H.; Chao, H. J.; Liu, S. J.; Zhou, N. Y. Biochemical and molecular characterization of the gentisate transporter genk in Corynebacterium glutamicum. PLoS One 2012, 7 (7), 1–8 DOI: 10.1371/journal.pone.0038701.

(170) Brinkrolf, K.; Brune, I.; Tauch, A. Transcriptional regulation of catabolic pathways for aromatic compounds in Corynebacterium glutamicum. Genet. Mol. Res. 2006, 5 (4), 773–789.

(171) Chaudhry, M. T.; Huang, Y.; Shen, X. H.; Poetsch, A.; Jiang, C. Y.; Liu, S. J. Genome-wide investigation of aromatic acid transporters in Corynebacterium glutamicum. Microbiology 2007, 153 (3), 857–865 DOI: 10.1099/mic.0.2006/002501-0.

360 References

(172) S. Weston; Parish, C. Calcein: a novel marker for lymphocytes. Cytometry 1992, 13 (7), 739–749.

(173) Lichtenfels, R.; Biddison, W. E.; Schulz, H.; Vogt, A. B.; Martin, R. CARE-LASS (calcein-release-assay), an improved fluorescence- based test system to measure cytotoxic T lymphocyte activity. J.Immunol.Methods 1994, 172 (2), 227– 239.

(174) Sayers, E. W.; Barrett, T.; Benson, D. A.; Bryant, S. H.; Canese, K.; Chetvernin., V.; Church, D. M.; Dicuccio, M.; Edgar, R.; Federhen, S.; et al. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 2009, 37 (SUPPL. 1), 5–15 DOI: 10.1093/nar/gkn741.

(175) Hiraoka, Y.; Kimbara, K. Rapid Assessment of the Physiological Status of the Polychlorinated Biphenyl Degrader Comamonas testosteroni TK102 by Flow Cytometry Rapid Assessment of the Physiological Status of the Polychlorinated Biphenyl Degrader Comamonas testosteroni TK102 by Flow. Appl. Environ. Microbiol. 2002, 68 (4), 2031–2035 DOI: 10.1128/AEM.68.4.2031.

(176) Benomar, S.; Ranava, D.; Cárdenas, M. L.; Trably, E.; Rafrafi, Y.; Ducret, A.; Hamelin, J.; Lojou, E.; Steyer, J.-P.; Giudici- Orticoni, M.-T. Nutritional stress induces exchange of cell material and energetic coupling between bacterial species. Nat.

361 References

Commun. 2015, 6, 6283 DOI: 10.1038/ncomms7283.

(177) Mullineaux, C. W.; Mariscal, V.; Nenninger, A.; Khanum, H.; Herrero, A.; Flores, E.; Adams, D. G. Mechanism of intercellular molecular exchange in heterocyst-forming cyanobacteria. EMBO J. 2008, 27 (9), 1299–1308 DOI: 10.1038/emboj.2008.66.

(178) Müller, J.; Overmann, J. Close interspecies interactions between prokaryotes from sulfureous environments. Front. Microbiol. 2011, 2 (JULY), 1–11 DOI: 10.3389/fmicb.2011.00146.

(179) Mariscal, V.; Herrero, A.; Nenninger, A.; Mullineaux, C. W.; Flores, E. Functional dissection of the three-domain SepJ protein joining the cells in cyanobacterial trichomes. Mol. Microbiol. 2011, 79 (4), 1077–1088 DOI: 10.1111/j.1365- 2958.2010.07508.x.

(180) Nieves-Morión, M.; Mullineaux, C. W.; Flores, E. Molecular Diffusion through Cyanobacterial Septal Junctions. MBio 2017, 8 (1), e01756-16 DOI: 10.1128/mBio.01756-16.

(181) Sayali, K.; Sadichha, P.; Surekha, S. Microbial Esterases : An overview. J . Curr . Microbiol . App . Sci 2013, 2 (7), 135–146.

(182) Du, D.; van Veen, H. W.; Murakami, S.; Pos, K. M.; Luisi, B. F. Structure, mechanism and cooperation of bacterial multidrug

362 References

transporters. Curr. Opin. Struct. Biol. 2015, 33 (Figure 1), 76– 91 DOI: 10.1016/j.sbi.2015.07.015.

(183) Nestorovich, E. M.; Danelon, C.; Winterhalter, M.; Bezrukov, S. M. Designed to penetrate: time-resolved interaction of single antibiotic molecules with bacterial pores. Proc. Natl. Acad. Sci. U. S. A. 2002, 99 (15), 9789–9794 DOI: 10.1073/pnas.152206799.

(184) Sojo, V.; Pomiankowski, A.; Lane, N. A Bioenergetic Basis for Membrane Divergence in Archaea and Bacteria. PLoS Biol. 2014, 12 (8) DOI: 10.1371/journal.pbio.1001926.

(185) Rohnstock, A.; Lehmann, L. Evaluation of the probe dihydrocalcein acetoxymethylester as an indicator of reactive oxygen species formation and comparison with oxidative DNA base modification determined by modified alkaline elution technique. Toxicol. Vitr. 2007, 21 (8), 1552–1562 DOI: 10.1016/j.tiv.2007.05.001.

(186) Bacellar, I. O. L.; Tsubone, T. M.; Pavani, C.; Baptista, M. S. Photodynamic efficiency: From molecular photochemistry to cell death. Int. J. Mol. Sci. 2015, 16 (9), 20523–20559 DOI: 10.3390/ijms160920523.

(187) Lee Seo, J., Kim, E.-S., Lee, H.-S., Kim, P., J.-Y. Adaptive Evolution of Corynebacterium glutamicum Resistant to Oxidative Stress and Its Global Gene Expression Profiling. 363 References

Biotechnol. Lett. 2013, 35, 9.

(188) Lushchak, V. I. Adaptive Response to Oxidative Stress: Bacteria, Fungi, Plants and Animals. Comp. Biochem. Phys. C 2011, 153, 16.

(189) Henle, E. S.; Linn, S. Repair of DNA Damage by Iron / Hydrogen Peroxide *. No. 20.

(190) Dwyer, D. J.; Belenky, P. A.; Yang, J. H.; MacDonald, I. C.; Martell, J. D.; Takahashi, N.; Chan, C. T. Y.; Lobritz, M. A.; Braff, D.; Schwarz, E. G.; et al. Antibiotics induce redox-related physiological alterations as part of their lethality. Proc. Natl. Acad. Sci. U. S. A. 2014, 111 (20), E2100-9 DOI: 10.1073/pnas.1401876111.

(191) Carmona-Gutierrez, D.; Eisenberg, T.; Büttner, S.; Meisinger, C.; Kroemer, G.; Madeo, F. Apoptosis in yeast: triggers, pathways, subroutines. Cell Death Differ. 2010, 17 (5), 763– 773 DOI: 10.1038/cdd.2009.219.

(192) Cabiscol Tamarit, J., Ros, J., E. Oxidative Stress in Bacteria and Protein Damage by Reactive Oxygen Species. Internatl. Microbiol. 2000, 3, 6.

(193) Mols, M.; Abee, T. Primary and secondary oxidative stress in Bacillus. Environ. Microbiol. 2011, 13 (6), 1387–1394 DOI: 10.1111/j.1462-2920.2011.02433.x.

364 References

(194) Krömer, J. O.; Bolten, C. J.; Heinzle, E.; Wittmann, C. Physiological response of Corynebacterium glutamicum to oxidative stress induced by deletion of the transcriptional repressor McbR. Microbiology 2008, 154, 3917–3930 DOI: 10.1099/mic.0.2008/021204-0.

(195) Cumming, B. M.; Lamprecht, D. A.; Wells, R. M.; Saini, V.; Mazorodze, J. H.; Steyn, A. J. C. The Physiology and Genetics of Oxidative Stress in Mycobacteria. Microbiol. Spectr. 2014, 2 (3), MGM2-0019-2013 DOI: 10.1128/microbiolspec.MGM2- 0019-2013.Correspondence.

(196) Si, M.; Zhao, C.; Zhang, B.; Wei, D.; Chen, K.; Yang, X.; Xiao, H. Overexpression of Mycothiol Disulfide Reductase Enhances Corynebacterium glutamicum Robustness by Modulating Cellular Redox Homeostasis and Antioxidant Proteins under Oxidative Stress. Sci. Rep. 2016, 6 (29491), 1–14 DOI: 10.1038/srep29491.

(197) Shafey, H. M. El; Ghanem, S. Regulation of expression of sodA and msrA genes of Corynebacterium glutamicum in response to oxidative and radiative stress. Genet. Mol. Res. 2015, 14 (1), 2104–2117.

(198) Imlay, J. A. Diagnosing oxidative stress in bacteria: not as easy as you might think. Curr. Opin. Microbiol. 2015, 24, 124–131 DOI: 10.1016/j.mib.2015.01.004.

365 References

(199) Chen, M.; Yin, M. Design and development of fluorescent nanostructures for bioimaging. Prog. Polym. Sci. 2014, 39 (2), 365–395 DOI: 10.1016/j.progpolymsci.2013.11.001.

(200) Anthony, J. R.; Warczak, K. L.; Donohue, T. J. A transcriptional response to singlet oxygen, a toxic byproduct of photosynthesis. Proc Natl Acad Sci U S A 2005, 102 (18), 6502– 6507 DOI: 10.1073/pnas.0502225102.

(201) Niebisch, A.; Bott, M. Molecular analysis of the cytochrome bc 1 - aa 3 branch of the Corynebacterium glutamicum respiratory chain containing an unusual diheme cytochrome c 1. Arch. Microbiol. 2001, 175 (4), 282–294 DOI: 10.1007/s002030100262.

(202) Koch-Koerfges, A.; Kabus, A.; Ochrombel, I.; Marin, K.; Bott, M. Physiology and global gene expression of a Corynebacterium glutamicum ΔF1FO-ATP synthase mutant devoid of oxidative phosphorylation. Biochim. Biophys. Acta - Bioenerg. 2012, 1817 (2), 370–380 DOI: 10.1016/j.bbabio.2011.10.006.

(203) Dwyer, D. J.; Camacho, D. M.; Kohanski, M. a; Callura, J. M.; Collins, J. J. Antibiotic-induced bacterial cell death exhibits physiological and biochemical hallmarks of apoptosis. Mol. Cell 2012, 46 (5), 561–572 DOI: 10.1016/j.molcel.2012.04.027.

366 References

(204) Bott, M.; Niebisch, A. The respiratory chain of Corynebacterium glutamicum. J. Biotechnol. 2003, 104 (1–3), 129–153 DOI: 10.1016/S0168-1656(03)00144-5.

(205) Niebisch, A.; Bott, M. Purification of a cytochrome bc1-aa3 supercomplex with quinol oxidase activity from Corynebacterium glutamicum: Identification of a fourth subunit of cytochrome aa3 oxidase and mutational analysis of diheme cytochrome c1. J. Biol. Chem. 2003, 278 (6), 4339–4346 DOI: 10.1074/jbc.M210499200.

(206) Koch-Koerfges, A.; Pfelzer, N.; Platzen, L.; Oldiges, M.; Bott, M. Conversion of Corynebacterium glutamicum from an aerobic respiring to an aerobic fermenting bacterium by inactivation of the respiratory chain. Biochim. Biophys. Acta 2013, 1827 (6), 699–708 DOI: 10.1016/j.bbabio.2013.02.004.

(207) Koch-Koerfges, A. Novel insights into the energy metabolism of Corynebacterium glutamicum by comprehensive analysis of mutants defective in respiration or oxidative phosphorylation, Heinrich Heine Universität Düsseldorf, 2011.

(208) Ning, C.; Wang, X.; Li, L.; Zhu, Y.; Li, M.; Yu, P.; Zhou, L.; Zhou, Z.; Chen, J.; Tan, G.; et al. Concentration Ranges of Antibacterial Cations for Showing the Highest Antibacterial Efficacy but the Least Cytotoxicity against Mammalian Cells: Implications for a New Antibacterial Mechanism. Chem. Res.

367 References

Toxicol. 2015, 28 (9), 1815–1822 DOI: 10.1021/acs.chemrestox.5b00258.

(209) Eggeling, L.; Sahm, H. The Cell Wall Barrier of Corynebacterium glutamicum and Amino Acid Efflux. J. Biosci. Bioeng. 2001, 92 (3), 201–213.

(210) Abrahams, K. A.; Besra, G. S. Mycobacterial cell wall biosynthesis: a multifaceted antibiotic target. Parasitology 2016, 1–18 DOI: 10.1017/S0031182016002377.

(211) Follmann, M.; Ochrombel, I.; Krämer, R.; Trötschel, C.; Poetsch, A.; Rückert, C.; Hüser, A.; Persicke, M.; Seiferling, D.; Kalinowski, J.; et al. Functional genomics of pH homeostasis in Corynebacterium glutamicum revealed novel links between pH response, oxidative stress, iron homeostasis and methionine synthesis. BMC Genomics 2009, 10, doi: 10.1186/1471-2164-10-621 DOI: 10.1186/1471-2164-10-621.

(212) Booth, I. R. Regulation of cytoplasmic pH in bacteria. Microbiol. Rev. 1985, 49 (4), 359–378 DOI: 10.1016/S1569- 2582(97)80142-0.

(213) van Beilen, J. W. A.; Brul, S. Compartment-specific pH monitoring in Bacillus subtilis using fluorescent sensor proteins: A tool to analyze the antibacterial effect of weak organic acids. Front. Microbiol. 2013, 4 (JUN), 1–11 DOI: 10.3389/fmicb.2013.00157. 368 References

(214) Martinez, K. A.; Kitko, R. D.; Mershon, J. P.; Adcox, H. E.; Malek, K. A.; Berkmen, M. B.; Slonczewski, J. L. Cytoplasmic pH response to acid stress in individual cells of Escherichia coli and Bacillus subtilis observed by fluorescence ratio imaging microscopy. Appl. Environ. Microbiol. 2012, 78 (10), 3706– 3714 DOI: 10.1128/AEM.00354-12.

(215) Siegumfeldt, H.; Björn Rechinger, K.; Jakobsen, M. Use of fluorescence ratio imaging for intracellular pH determination of individual bacterial cells in mixed cultures. Microbiology 1999, 145 (7), 1703–1709 DOI: 10.1099/13500872-145-7-1703.

(216) Futsaether, C. M.; Kjeldstad, B.; Johnsson, A. Measurement of the intracellular pH of Propionibacterium acnes: comparison between the fluorescent probe BCECF and 31P-NMR spectroscopy. Can. J. Microbiol. 1993, 39 (2), 180–186.

(217) Leyval, D.; Debay, F.; Engasser, J. M.; Goergen, J. L. Flow cytometry for the intracellular pH measurement of glutamate producing Corynebacterium glutamicum. J. Microbiol. Methods 1997, 29 (2), 121–127 DOI: 10.1016/S0167- 7012(97)00031-6.

(218) Hunter, R. C.; Beveridge, T. J. Application of a pH-Sensitive Fluoroprobe (C-SNARF-4) for pH Microenvironment Analysis in Pseudomonas aeruginosa Biofilms. Appl. Environ. Microb. 2005, 71 (5), 10.

369 References

(219) Gangola, P.; Rosens, B. P. Maintenance of Intracellular Calcium in Escherichia coli *. J. Biol. Chem. 1987, 262 (26), 12570–12574.

(220) Sabnis, R. W. (Ram W. Handbook of biological dyes and stains : synthesis and industrial applications; 2010.

(221) Votyakova, T.; Kaprelyants, A. Influence of viable cells on the resuscitation of dormant cells in Micrococcus luteus cultures held in an extended stationary phase: the population effect. Appl. … 1994, 60 (9), 3284–3291.

(222) Unge, A.; Tombolini, R.; Mølbak, L.; Jansson, J. K. Simultaneous Monitoring of Cell Number and Metabolic Activity of Specific Bacterial Populations with a Dual gfp- luxAB Marker System. Appl. Environ. Microbiol. 1999, 65 (2), 813–821.

(223) Seletzky, J. M.; Noack, U.; Fricke, J.; Hahn, S.; Büchs, J. Metabolic activity of Corynebacterium glutamicum grown on L -lactic acid under stress. Appl. Microbiol. Biotechnol. 2006, 72 (6), 1297–1307 DOI: 10.1007/s00253-006-0436-0.

(224) Tuson, H. H.; Auer, G. K.; Renner, L. D.; Hasebe, M.; Tropini, C.; Salick, M.; Crone, W. C.; Gopinathan, A.; Huang, K. C.; Weibel, D. B. Measuring the Stiffness of Bacterial Cells from Growth Rates in Hydrogels of Tunable Elasticity. Mol. Microbiol. 2012, 84 (5), 18. 370 References

(225) Besant, J. D.; Sargent, H.; Kelley, S. O. Rapid electrochemical phenotypic profiling of antibiotic-resistant bacteria. Lab Chip 2015, 15 (13), 2799–2807 DOI: 10.1039/C5LC00375J.

(226) Gómez-Sjöberg, R.; Leyrat, A. A.; Pirone, D. M.; Chen, C. S.; Quake, S. R. Versatile, Fully Automated, Microfluidic Cell Culture System. Anal. Chem. 2007, 79 (22), 8557–8563.

(227) Geva-Zatorsky, N.; Alvarez, D.; Hudak, J. E.; Reading, N. C.; Erturk-Hasdemir, D.; Dasgupta, S.; von Andrian, U. H.; Kasper, D. L. In vivo imaging and tracking of host–microbiota interactions via metabolic labeling of gut anaerobic bacteria. Nat. Med. 2015, 21 (9), 1091–1100 DOI: 10.1038/nm.3929.

(228) Stephens, P. J.; Hewitt, C. J.; Powell, J. R.; Badley, R. A.; Nebe- von-Caron, G. Analysis of Bacterial Function by Multi-Colour Fluorescence Flow Cytometry and Single Cell Sorting. J. Microbiol. Meth. 2000, 42 (1), 97–114.

(229) Rotmant, B.; Papermaster, B. W. MEMBRANE PROPERTIES OF LIVING MAMMALIAN CELLS AS STUDIED BY ENZYMATIC HYDROLYSIS OF FLUOROGENIC ESTERS*. Med. Res. Counc. Spec. Rept. Ser. Proc. Proc. Am. J. Med 1956, 46 (31), 368–1461 DOI: 10.1073/pnas.55.1.134.

(230) Haugland, R. P.; MacCoubrey, I. C.; Moore, P. L. Dual- Fluorescence Cell Viability Assay Using Ethidium Homodimer and Calcein Am. 5,314,805, 1994. 371 References

(231) Poulsen, C. R.; Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Static and dynamic acute cytotoxicity assays on microfluidic devices. Anal. Chem. 2005, 77 (2), 667–672 DOI: 10.1021/ac049279i.

(232) Joux, F.; Lebaron, P. Use of fluorescent probes to assess physiological functions of bacteria at single-cell level. Microbes Infect. 2000, 2 (12), 1523–1535.

(233) Comas, J.; Vives-Rego, J. Enumeration, viability and heterogeneity in Staphylococcus aureus cultures by flow cytometry. J. Microbiol. Methods 1998, 32 (1), 45–53 DOI: 10.1016/S0167-7012(98)00003-7.

(234) Hammes, F.; Egli, T. Cytometric methods for measuring bacteria in water: advantages, pitfalls and applications. Anal. Bioanal. Chem. 2010, 397 (3), 1083–1095 DOI: 10.1007/s00216-010-3646-3.

(235) Shapiro, H. M. Microbial analysis at the single-cell level: Tasks and techniques. J. Microbiol. Methods 2000, 42 (1), 3–16 DOI: 10.1016/S0167-7012(00)00167-6.

(236) Halldorsson, S.; Lucumi, E.; Gómez-Sjöberg, R.; Fleming, R. M. T. Advantages and challenges of microfluidic cell culture in polydimethylsiloxane devices. Biosens. Bioelectron. 2015, 63, 218–231 DOI: 10.1016/j.bios.2014.07.029.

372 References

(237) Hartmann, M.; Barsch, A.; Niehaus, K.; Pühler, A.; Tauch, A.; Kalinowski, J. The glycosylated cell surface protein Rpf2, containing a resuscitation-promoting factor motif, is involved in intercellular communication of Corynebacterium glutamicum. Arch. Microbiol. 2004, 182 (4), 299–312 DOI: 10.1007/s00203-004-0713-1.

(238) Evangelopoulos, D.; da Fonseca, J. D.; Waddell, S. J. Understanding anti-tuberculosis drug efficacy: rethinking bacterial populations and how we model them. Int. J. Infect. Dis. 2015, 32, 76–80 DOI: 10.1016/j.ijid.2014.11.028.

(239) Marienfeld, S.; Uhlemann, E.; Schmid, R.; Krämer, R.; Burkovski, A. Ultrastructure of the Corynebacterium glutamicum cell wall. Antonie Van Leeuwenhoek 1997, 72 (4), 291–297.

(240) Manina, G.; Dhar, N.; McKinney, J. D. Stress and Host Immunity Amplify Mycobacterium tuberculosis Phenotypic Heterogeneity and Induce Nongrowing Metabolically Active Forms. Cell Host Microbe 2015, 17 (1), 32–46 DOI: 10.1016/j.chom.2014.11.016.

(241) Bhargava, P.; Collins, J. Boosting Bacterial Metabolism to Combat Antibiotic Resistance. Cell Metab. 2015, 21 (2), 154– 155.

(242) Peng, B.; Su, Y.; Li, H.; Han, Y.; Guo, C.; Tian, Y.; Peng, X. 373 References

Exogenous Alanine and/or Glucose plus Kanamycin Kills Antibiotic-Resistant Bacteria. Cell Metab. 2015, 21 (2), 249– 261.

(243) Dhar, N.; Dubée, V.; Ballell, L.; Cuinet, G.; Hugonnet, J.-E.; Signorino-Gelo, F.; Barros, D.; Arthur, M.; McKinney, J. D. Rapid cytolysis of Mycobacterium tuberculosis by faropenem, an orally bioavailable β-lactam antibiotic. Antimicrob. Agents Chemother. 2015, 59 (2), 1308–1319 DOI: 10.1128/AAC.03461-14.

(244) Feller, N.; Broxterman, H. J.; Währer, D. C. R.; Pinedo, H. M. ATP-dependent efflux of calcein by the multidrug resistance protein (MRP): no inhibition by intracellular glutathione depletion. FEBS Lett. 1995, 368 (2), 385–388 DOI: 10.1016/0014-5793(95)00677-2.

(245) Wang, C.; Cai, H.; Zhou, Z.; Wan, H. Metabolic Analysis of a Corynebacterium glutamicum IdhA Mutant During an Efficient Succinate Production Using pH-Control Under Oxygen Deprivation. In Advances in Applied Biotechnology; Zhang, T.- C., Nakajima, M., Eds.; Lecture Notes in Electrical Engineering; Springer: Berlin, Heidelberg, 2015; Vol. 332, pp 375–387.

(246) Braissant, O.; Bachmann, a.; Bonkat, G. Microcalorimetric assays for measuring cell growth and metabolic activity:

374 References

Methodology and applications. Methods 2015, 76, 27–34 DOI: 10.1016/j.ymeth.2014.10.009.

(247) Foudeh, A. M.; Trigui, H.; Mendis, N.; Faucher, S. P.; Veres, T.; Tabrizian, M. Rapid and specific SPRi detection of L. pneumophila in complex environmental water samples. Anal. Bioanal. Chem. 2015, 407 (18), 5541–5545 DOI: 10.1007/s00216-015-8726-y.

(248) Timmins, G. S. Detecting virulence and drug-resistance mycobacterial phenotypes in vivo. Trends Microbiol. 2015, 23 (6), 321–323 DOI: 10.1016/j.tim.2015.02.013.

(249) Hammes, F.; Berney, M.; Egli, T. Cultivation-independent Assessment of Bacterial Viability. Adv. Biochem. Engin. Biotechnol. 2011, 124, 123–150.

(250) Ullrich, S.; Karrasch, B.; Hoppe, H.; Jeskulke, K.; Mehrens, M. Toxic effects on bacterial metabolism of the redox dye 5-cyano- 2,3-ditolyl tetrazolium chloride. Appl. Environ. Microbiol. 1996, 62 (12), 4587–4593.

(251) Liebl, W.; Klamer, R.; Schleifer, K.-H. Requirement of Chelating Compounds for the Growth of Corynebacterium glutamicum in Synthetic Media. Appl. Microbiol. Biotechnol. 1989, 32 (2), 205–210.

(252) Dertz, E. A.; Stintzi, A.; Raymond, K. N. Siderophore-

375 References

Mediated Iron Transport in Bacillus subtilis and Corynebacterium glutamicum. J. Biol. Inorg. Chem. 2006, 11 (8), 1087–1097.

(253) Schröder, J.; Tauch, A. Transcriptional regulation of gene expression in Corynebacterium glutamicum: the role of global, master and local regulators in the modularand hierarchical gene regulatory network. FEMS Microbiol. Rev. 2010, 34 (5), 685– 737 DOI: 10.1111/j.1574-6976.2010.00228.x.

(254) Keep, N. H.; Ward, J. M.; Cohen-Gonsaud, M.; Henderson, B. Wake up ! Peptidoglycan lysis and bacterial non-growth states. Trends Microbiol. 2006, 14 (6), 271–276 DOI: 10.1016/j.tim.2006.04.003.

(255) Essodaı, M.; Broxterman, H. J.; Garnier-Suillerot, A. Kinetic Analysis of Calcein and Calcein - Acetoxymethylester Efflux Mediated by the Multidrug Resistance Protein and P- Glycoprotein. Biochemistry 1998, 37 (8), 2243–2250.

(256) Schluesener, D.; Fischer, F.; Kruip, J.; Rögner, M.; Poetsch, A. Mapping the Membrane Proteome of Corynebacterium glutamicum. Proteomics 2005, 5 (5), 1317–1330.

(257) Maglica, Z.; Özdemir, E.; McKinney, J. D. Single-Cell Tracking Reveals Antibiotic-Induced Changes in Mycobacterial Energy Metabolism. MBio 2015, 6 (1), 1–11 DOI: 10.1128/mBio.02236-14.Editor. 376 References

(258) Sim, S. Y.; Hong, E. J.; Kim, Y.; Lee, H. S. Analysis of cepA encoding an efflux pump-like protein in Corynebacterium glutamicum. J. Microbiol. 2014, 52 (4), 278–283 DOI: 10.1007/s12275-014-3461-1.

(259) Davey, H. M.; Hexley, P. Red but not dead? Membranes of stressed Saccharomyces cerevisiae are permeable to propidium iodide. Environ. Microbiol. 2011, 13 (1), 163–171 DOI: 10.1111/j.1462-2920.2010.02317.x.

(260) Lehtinen, J.; Nuutila, J.; Lilius, E.-M. Green fluorescent protein-propidium iodide (GFP-PI) based assay for flow cytometric measurement of bacterial viability. Cytom. Part A 2004, 60 (2), 165–172 DOI: 10.1002/cyto.a.20026.

(261) Williams, S. C.; Hong, Y.; Danavall, D. C. A.; Howard-Jones, M. H.; Gibson, D.; Frischer, M. E.; Verity, P. G. Distinguishing between living and nonliving bacteria: Evaluation of the vital stain propidium iodide and its combined use with molecular probes in aquatic samples. J. Microbiol. Methods 1998, 32 (3), 225–236.

(262) Shi, L.; Günther, S.; Hübschmann, T.; Wick, L. Y.; Harms, H.; Müller, S. Limits of Propidium Iodide as a Cell Viability Indicator for Environmental Bacteria. Cytom. Part A 2007, 71 (8), 592–598.

(263) Gião, M. S.; Wilks, S. A.; Azevedo, N. F.; Vieira, M. J.; Keevil, 377 References

C. W. Validation of SYTO 9/Propidium Iodide Uptake for Rapid Detection of Viable but Noncultivable Legionella pneumophila. Microb. Ecol. 2009, 58 (1), 56–62.

(264) Stiefel, P.; Schmidt-Emrich, S.; Maniura-Weber, K.; Ren, Q. Critical aspects of using bacterial cell viability assays with the fluorophores SYTO9 and propidium iodide. BMC Microbiol. 2015, 15 (36), 1–9 DOI: 10.1186/s12866-015-0376-x.

(265) Mathey, R.; Dupoy, M.; Espagnon, I.; Leroux, D.; Mallard, F.; Novelli-Rousseau, A. Viability of 3h grown bacterial microcolonies after direct Raman identification. J. Microbiol. Methods 2015, 109, 67–73 DOI: 10.1016/j.mimet.2014.12.002.

(266) Benoit, M. R.; Conant, C. G.; Ionescu-Zanetti, C.; Schwartz, M.; Matin, A. New device for high-throughput viability screening of flow biofilms. Appl. Environ. Microbiol. 2010, 76 (13), 4136–4142 DOI: 10.1128/AEM.03065-09.

(267) Booth, R.; Noh, S.; Kim, H. A multiple-channel, multiple-assay platform for characterization of full-range shear stress effects on vascular endothelial cells. Lab Chip 2014, 14 (11), 1880– 1890 DOI: 10.1039/c3lc51304a.

(268) Cho, S.; Kang, D.; Sim, S.; Geier, F.; Kim, J.; Chang, S.; Edel, J.; Wootton, R. a Droplet-Based Microfluidic System for High- Throughput Screening of Photosensitisers Against Microbial Organisms. MicroTAS 2012 2012, 85 (17), 332–334. 378 References

(269) Han, J.-H.; Heinze, B. C.; Yoon, J.-Y. Single cell level detection of Escherichia coli in microfluidic device. Biosens. Bioelectron. 2008, 23 (8), 1303–1306 DOI: 10.1016/j.bios.2007.11.013.

(270) Zordan, M. D.; Grafton, M. M. G.; Acharya, G.; Reece, L. M.; Cooper, C. L.; Aronson, A. I.; Park, K.; Leary, J. F. Detection of pathogenic E. coli O157:H7 by a hybrid microfluidic SPR and molecular imaging cytometry device. Cytom. Part A 2009, 75 (2), 155–162 DOI: 10.1002/cyto.a.20692.

(271) Cho, B. S.; Schuster, T. G.; Zhu, X.; Chang, D.; Smith, G. D.; Takayama, S. Passively driven integrated microfluidic system for separation of motile sperm. Anal. Chem. 2003, 75 (7), 1671– 1675 DOI: 10.1021/ac020579e.

(272) Zhang, T.; Fang, H. H. P. Quantification of Saccharomyces cerevisiae viability using BacLight. Biotechnol. Lett. 2004, 26 (12), 989–992 DOI: 10.1023/B:BILE.0000030045.16713.19.

(273) Packard, M.; Wheeler, E.; Alocilja, E.; Shusteff, M. Performance Evaluation of Fast Microfluidic Thermal Lysis of Bacteria for Diagnostic Sample Preparation. Diagnostics 2013, 3 (1), 105–116 DOI: 10.3390/diagnostics3010105.

(274) Komen, J.; Wolbers, F.; Franke, H. R.; Andersson, H.; Vermes, I.; van den Berg, A. Viability analysis and apoptosis induction of breast cancer cells in a microfluidic device: effect of cytostatic drugs. Biomed. Microdevices 2008, 10 (5), 727–737 379 References

DOI: 10.1007/s10544-008-9184-5.

(275) Liu, W.; Li, L.; Wang, J.; Tu, Q.; Ren, L.; Wang, Y.; Wang, J. Dynamic trapping and high-throughput patterning of cells using pneumatic microstructures in an integrated microfluidic device. Lab Chip 2012, 12 (9), 1702–1709 DOI: 10.1039/c2lc00034b.

(276) Toh, Y.-C.; Zhang, C.; Zhang, J.; Khong, Y. M.; Chang, S.; Samper, V. D.; van Noort, D.; Hutmacher, D. W.; Yu, H. A novel 3D mammalian cell perfusion-culture system in microfluidic channels. Lab Chip 2007, 7 (3), 302–309 DOI: 10.1039/b614872g.

(277) Desjardins, R.; Boulos, L.; Barbeau, B. Methods LIVE / DEAD ® Bac Light E : application of a new rapid staining method for direct enumeration of viable and total bacteria in drinking water. J. Microbiol. Methods 1999, 37, 77–86.

(278) Mcclain, M. A.; Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Flow Cytometry of Escherichia coli on Microfluidic Devices. Anal. Chem. 2001, 73 (21), 5334–5338.

(279) Lu, H.; Schmidt, M. a; Jensen, K. F. A microfluidic electroporation device for cell lysis. Lab Chip 2005, 5 (1), 23– 29 DOI: 10.1039/b406205a.

(280) Toh, Y.-C.; Lim, T. C.; Tai, D.; Xiao, G.; van Noort, D.; Yu, H. A microfluidic 3D hepatocyte chip for drug toxicity testing. Lab

380 References

Chip 2009, 9 (14), 2026–2035 DOI: 10.1039/b900912d.

(281) Ziółkowska, K.; Stelmachowska, A.; Kwapiszewski, R.; Chudy, M.; Dybko, A.; Brzózka, Z. Long-term three- dimensional cell culture and anticancer drug activity evaluation in a microfluidic chip. Biosens. Bioelectron. 2013, 40 (1), 68– 74 DOI: 10.1016/j.bios.2012.06.017.

(282) Valero, A.; Merino, F.; Wolbers, F.; Luttge, R.; Vermes, I.; Andersson, H.; van den Berg, A. Apoptotic cell death dynamics of HL60 cells studied using a microfluidic cell trap device. Lab Chip 2005, 5 (1), 49–55 DOI: 10.1039/b415813j.

(283) Chung, B. G.; Manbachi, A.; Khademhosseini, A. A microfluidic device with groove patterns for studying cellular behavior. J. Vis. Exp. 2007, 7 (270), doi:10.3791/270 DOI: 10.3791/270.

(284) Palková, Z.; Váchová, L.; Valer, M.; Preckel, T. Single-cell analysis of yeast, mammalian cells, and fungal spores with a microfluidic pressure-driven chip-based system. Cytometry. A 2004, 59 (2), 246–253 DOI: 10.1002/cyto.a.20049.

(285) Kobel, S.; Valero, A.; Latt, J.; Renaud, P.; Lutolf, M. Optimization of microfluidic single cell trapping for long-term on-chip culture. Lab Chip 2010, 10 (7), 857–863 DOI: 10.1039/b918055a.

381 References

(286) Kim, S.; Thiessen, P. A.; Bolton, E. E.; Chen, J.; Fu, G.; Gindulyte, A.; Han, L.; He, J.; He, S.; Shoemaker, B. A.; et al. PubChem substance and compound databases. Nucleic Acids Res. 2016, 44 (D1), D1202–D1213 DOI: 10.1093/nar/gkv951.

(287) Hettne, K. M.; Williams, A. J.; van Mulligen, E. M.; Kleinjans, J.; Tkachenko, V.; Kors, J. a. ChemSpider: An Online Chemical Information Resource. J. Cheminform. 2010, 2 (1), 3 DOI: 10.1186/1758-2946-2-3.

(288) Duineveld, P. C.; Lilja, M.; Johansson, T.; Inganäs, O. Diffusion of Solvent in PDMS Elastomer for Micromolding in Capillaries. Langmuir 2002, 18, 9554–9559.

(289) Warnes, S. L.; Keevil, C. W. Mechanism of copper surface toxicity in vancomycin-resistant enterococci following wet or dry surface contact. Appl. Environ. Microbiol. 2011, 77 (17), 6049–6059 DOI: 10.1128/AEM.00597-11.

(290) Law, V.; Knox, C.; Djoumbou, Y.; Jewison, T.; Guo, A. C.; Liu, Y.; Maciejewski, A.; Arndt, D.; Wilson, M.; Neveu, V.; et al. DrugBank 4.0: shedding new light on drug metabolism. Nucleic Acids Res. 2014, 42 (D1091-7), doi:10.1093/nar/gkt1068 DOI: 10.1093/nar/gkt1068.

(291) Eisler, H.; Fröhlich, K.-U.; Heidenreich, E. Starvation for an essential amino acid induces apoptosis and oxidative stress in yeast. Exp. Cell Res. 2004, 300 (2), 345–353 DOI: 382 References

10.1016/j.yexcr.2004.07.025.

(292) Wanichthanarak, K.; Cvijovic, M.; Molt, A.; Petranovic, D. yApoptosis: yeast apoptosis database. Database 2013, 2013, Article ID bat068 DOI: 10.1093/database/bat068.

(293) Tower, J. Programmed cell death in aging. Ageing Res. Rev. 2015, 23 (Pt A), 90–100 DOI: 10.1016/j.arr.2015.04.002.

(294) Carmona-Gutierrez, D.; Büttner, S. The many ways to age for a single yeast cell. Yeast 2014, 31 (8), 289–298 DOI: 10.1002/yea.

(295) Müller, M.; Lu, K.; Reichert, A. S. Mitophagy and mitochondrial dynamics in Saccharomyces cerevisiae. Biochim. Biophys. Acta 2015, 1853 (10), 2766–2774 DOI: 10.1016/j.bbamcr.2015.02.024.

(296) Allis, D.; Cell, M.; Carmona-Gutierrez, D.; Madeo, F. Yeast Unravels Epigenetic Apoptosis Control : Deadly Chat within a Histone Tail. Mol. Cell 2006, 24 (2), 167–169 DOI: 10.1016/j.molcel.2006.10.004.

(297) Eisenberg, T.; Carmona-Gutierrez, D.; Büttner, S.; Tavernarakis, N.; Madeo, F. Necrosis in yeast. Apoptosis 2010, 15 (3), 257–268 DOI: 10.1007/s10495-009-0453-4.

(298) Hoshi, H.; O’Brien, J.; Mills, S. L. A novel fluorescent tracer for visualizing coupled cells in neural circuits of living tissue.

383 References

J. Histochem. Cytochem. 2006, 54 (10), 1169–1176 DOI: 10.1369/jhc.6A6935.2006.

(299) Lemasters, J. J.; Nieminen, a L.; Qian, T.; Trost, L. C.; Elmore, S. P.; Nishimura, Y.; Crowe, R. a; Cascio, W. E.; Bradham, C. a; Brenner, D. a; et al. The mitochondrial permeability transition in cell death: a common mechanism in necrosis, apoptosis and autophagy. Biochim. Biophys. Acta 1998, 1366 (1–2), 177–196 DOI: 10.1016/S0005-2728(98)00112-1.

(300) Prudêncio, C.; Sansonetty, F. Rapid detection of efflux pumps and their relation with drug resistance in yeast cells. Cytometry 2000, 35 (1), 26–35.

(301) Lewinska, A.; Miedziak, B.; Kulak, K.; Molon, M.; Wnuk, M. Links between nucleolar activity, rDNA stability, aneuploidy and chronological aging in the yeast Saccharomyces cerevisiae. Biogerontology 2014, 15 (3), 289–316 DOI: 10.1007/s10522- 014-9499-y.

(302) Lippuner, A. D.; Julou, T.; Barral, Y. Budding yeast as a model organism to study the effects of age. FEMS Microbiol. Rev. 2014, 38 (2), 300–325 DOI: 10.1111/1574-6976.12060.

(303) Reggiori, F.; Klionsky, D. J. Autophagic processes in yeast: Mechanism, machinery and regulation. Genetics 2013, 194 (2), 341–361 DOI: 10.1534/genetics.112.149013.

384 References

(304) Rodriguez-Enriquez, S.; He, L.; Lemasters, J. J. Role of mitochondrial permeability transition pores in mitochondrial autophagy. Int. J. Biochem. Cell Biol. 2004, 36 (12), 2463–2472 DOI: 10.1016/j.biocel.2004.04.009.

(305) Nunes, P. S.; Kjaerulff, S.; Dufva, M.; Mogensen, K. B. Real- time direct cell concentration and viability determination using a fully automated microfluidic platform for standalone process monitoring. Analyst 2015, 140 (12), 4007–4020 DOI: 10.1039/C5AN00478K.

(306) Brauner, A.; Fridman, O.; Gefen, O.; Balaban, N. Q. Distinguishing between resistance, tolerance and persistence to antibiotic treatment. Nat. Rev. Microbiol. 2016, 14 (5), 320–330 DOI: 10.1038/nrmicro.2016.34.

(307) Arnoldini, M.; Vizcarra, I. A.; Peña-Miller, R.; Stocker, N.; Diard, M.; Vogel, V.; Beardmore, R. E.; Hardt, W. D.; Ackermann, M. Bistable expression of virulence genes in salmonella leads to the formation of an antibiotic-tolerant subpopulation. PLoS Biol. 2014, 12 (8) DOI: 10.1371/journal.pbio.1001928.

(308) Ayrapetyan, M.; Williams, T. C.; Oliver, J. D. Bridging the gap between viable but non-culturable and antibiotic persistent bacteria. Trends Microbiol. 2014, 23 (1), 7–13 DOI: 10.1016/j.tim.2014.09.004.

385 References

(309) Fauvart, M.; de Groote, V. N.; Michiels, J. Role of persister cells in chronic infections: Clinical relevance and perspectives on anti-persister therapies. J. Med. Microbiol. 2011, 60 (6), 699–709 DOI: 10.1099/jmm.0.030932-0.

(310) Levin-Reisman, I.; Ronin, I.; Gefen, O.; Braniss, I.; Shoresh, N.; Balaban, N. Q. Antibiotic tolerance facilitates the evolution of resistance. Science (80-. ). 2017, 355 (6327), 826–830 DOI: 10.1126/science.aaj2191.

(311) Li, B.; Qiu, Y.; Glidle, A.; Cooper, J.; Shi, H.; Yin, H. Single cell growth rate and morphological dynamics revealing an “opportunistic” persistence. Analyst 2014, 139 (13), 3305–3313 DOI: 10.1039/c4an00170b.

(312) Zhang, Q.; Lambert, G.; Liao, D.; Kim, H.; Robin, K.; Tung, C.-K.; Pourmand, N.; Austin, R. H. Acceleration of Emergence of Bacterial Antibiotic Resistence in Connected Microenvironments. Science (80-. ). 2011, 333, 4.

(313) Wakamoto, Y.; Dhar, N.; Chait, R.; Schneider, K.; Signorino- Gelo, F.; Leibler, S.; McKinney, J. D. Dynamic persistence of antibiotic-stressed mycobacteria. Science 2013, 339 (6115), 91– 95 DOI: 10.1126/science.1229858.

(314) Balaban, N. Q.; Merrin, J.; Chait, R.; Kowalik, L.; Leibler, S. Bacterial Persistence as a Phenotypic Switch. Science (80-. ). 2004, 305 (10), 1622–1625. 386 References

(315) Maisonneuve, E.; Gerdes, K. Molecular mechanisms underlying bacterial persisters. Cell 2014, 157 (3), 539–548 DOI: 10.1016/j.cell.2014.02.050.

(316) Hombach, M.; Maurer, F. P.; Pfiffner, T.; Furrer, R. Standardization of Operator-Dependent Variables Affecting Precision Susceptibility Testing. J. Clin. Microbiol. 2015, 53 (12), 3864–3869 DOI: 10.1128/JCM.02351-15.Editor.

(317) Elf, J.; Biology, M.; Uppsala, S.-; Baltekin, Ö.; Boucharin, A.; Andersson, D. I. Fast Antibiotic Susceptibility Testing ( FASTest ) based on single cell growth rate measurements. bioRxiv 2016, 71407 DOI: 10.1101/071407.

(318) Choi, J.; Jung, Y.-G.; Kim, J.; Kim, S.; Jung, Y.; Na, H.; Kwon, S. Rapid antibiotic susceptibility testing by tracking single cell growth in a microfluidic agarose channel system. Lab Chip 2013, 13 (2), 280–287 DOI: 10.1039/C2LC41055A.

(319) Gefen, O.; Chekol, B.; Strahilevitz, J.; Balaban, N. Q. TDtest: easy detection of bacterial tolerance and persistence in clinical isolates by a modified disk-diffusion assay. Sci. Rep. 2017, 7 (July 2016), 41284 DOI: 10.1038/srep41284.

(320) Altobelli, E.; Mohan, R.; Mach, K. E.; Lai, M.; Sin, Y.; Anikst, V.; Buscarini, M.; Kin, P.; Gau, V.; Banaei, N.; et al. Integrated Biosensor Assay for Rapid Uropathogen Identification and Phenotypic Antimicrobial Susceptibility Testing. Eur. Urol. 387 References

Focus 2016, 1–7 DOI: 10.1016/j.euf.2015.12.010.

(321) Churski, K.; Kaminski, T. S.; Jakiela, S.; Kamysz, W.; Baranska-Rybak, W.; Weibel, D. B.; Garstecki, P. Rapid screening of antibiotic toxicity in an automated microdroplet system. Lab Chip 2012, 12 (9), 1629 DOI: 10.1039/c2lc21284f.

(322) Mohan, R.; Mukherjee, A.; Sevgen, S. E.; Sanpitakseree, C.; Lee, J.; Schroeder, C. M.; Kenis, P. J. a. A multiplexed microfluidic platform for rapid antibiotic susceptibility testing. Biosens. Bioelectron. 2013, 49, 118–125 DOI: 10.1016/j.bios.2013.04.046.

(323) Takagi, R.; Fukuda, J.; Nagata, K.; Yawata, Y.; Nomura, N.; Suzuki, H. A microfluidic microbial culture device for rapid determination of the minimum inhibitory concentration of antibiotics. Analyst 2013, 138 (4), 1000 DOI: 10.1039/c2an36323b.

(324) Cao, J.; Kürsten, D.; Schneider, S.; Knauer, A.; Günther, P. M.; Köhler, J. M. Uncovering Toxicological Complexity by Multi- Dimensional Screenings in Microsegmented Flow: Modulation of Antibiotic Interference by Nanoparticles. Lab Chip 2012, 12 (11), 474–484.

(325) de Jong, I. G.; Beilharz, K.; Kuipers, O. P.; Veening, J.-W. Live Cell Imaging of Bacillus subtilis and Streptococcus pneumoniae using Automated Time-lapse Microscopy. J. Vis. Exp. 2011, 388 References

No. 53, 1–6 DOI: 10.3791/3145.

(326) Iino, R.; Hayama, K.; Amezawa, H.; Sakakihara, S.; Kim, S. H.; Matsumono, Y.; Nishino, K.; Yamaguchi, A.; Noji, H. A single- cell drug efflux assay in bacteria by using a directly accessible femtoliter droplet array. Lab Chip 2012, 12 (20), 3923 DOI: 10.1039/c2lc40394c.

(327) Kalashnikov Lee, J. C., Campbell, J., Sharon, A., Sauer-Budge, A. F., M. A Microfluidic Platform for Rapid, Stress-Induced Antibiotic Susceptibility Testing of Staphylococcus aureus. Lab Chip 2012, 12, 10.

(328) Laflamme, C.; Lavigne, S.; Ho, J.; Duchaine, C. Assessment of bacterial endospore viability with fluorescent dyes. J. Appl. Microbiol. 2004, 96 (4), 684–692 DOI: 10.1111/j.1365- 2672.2004.02184.x.

(329) Hett, E. C.; Rubin, E. J. Bacterial growth and cell division: a mycobacterial perspective. Microbiol. Mol. Biol. Rev. 2008, 72 (1), 126–56, table of contents DOI: 10.1128/MMBR.00028-07.

(330) Gefen, O.; Balaban, N. Q. The Importance of Being Persistent: Heterogeneity of Bacterial Populations under Antibiotic Stress. FEMS Microbiol. Rev. 2009, 33, 14.

(331) Baek, S.-H.; Li, A. H.; Sassetti, C. M. Metabolic regulation of mycobacterial growth and antibiotic sensitivity. PLoS Biol.

389 References

2011, 9 (5), e1001065 DOI: 10.1371/journal.pbio.1001065.

(332) Anetzberger, C.; Schell, U.; Jung, K. Single cell analysis of Vibrio harveyi uncovers functional heterogeneity in response to quorum sensing signals. BMC Microbiol. 2012, 12 (1), 209 DOI: 10.1186/1471-2180-12-209.

(333) Drepper, T.; Eggert, T.; Circolone, F.; Heck, A.; Krauß, U.; Guterl, J.-K.; Wendorff, M.; Losi, A.; Gärtner, W.; Jaeger, K.- E. Reporter proteins for in vivo fluorescence without oxygen. Nat. Biotechnol. 2007, 25 (4), 443–445 DOI: 10.1038/nbt1293.

(334) Buckley, A. M.; Petersen, J.; Roe, A. J.; Douce, G. R.; Christie, J. M. LOV-based reporters for fluorescence imaging. Curr. Opin. Chem. Biol. 2015, 27, 39–45 DOI: 10.1016/j.cbpa.2015.05.011.

(335) Teramoto, J.; Yamanishi, Y.; Magdy, E.-S. H.; Hasegawa, A.; Kori, A.; Nakajima, M.; Arai, F.; Fukuda, T.; Ishihama, A.; Teramoto Yamanishi, Y., Magdy, E.-S. H., Hasegawa, A., Kori, A., Nakajima, M., Arai, F., Fukuda, T., Ishihama, A., J. Single Live-Bacterial Cell Assay of Promoter Activity and Regulation. Genes Cells 2010, 15 (11), 12 DOI: 10.1111/j.1365-2443.2010.01449.x.

(336) Zhang, X. E.; Cui, Z.; Wang, D. Sensing of biomolecular interactions using fluorescence complementing systems in living cells. Biosens. Bioelectron. 2016, 76, 243–250 DOI: 390 References

10.1016/j.bios.2015.07.069.

(337) Poelarends, G. J.; Mazurkiewicz, P.; Konings, W. N. Multidrug transporters and antibiotic resistance in Lactococcus lactis. Biochim. Biophys. Acta - Bioenerg. 2002, 1555 (1–3), 1–7 DOI: 10.1016/S0005-2728(02)00245-1.

(338) Potzkei, J.; Kunze, M.; Drepper, T.; Gensch, T.; Jaeger, K.-E.; Buechs, J. Real-time determination of intracellular oxygen in bacteria using a genetically encoded FRET-based biosensor. BMC Biol. 2012, 10 (1), 28 DOI: 10.1186/1741-7007-10-28.

(339) Li, J.-K.; Humphrey, A. E. Factors Affecting Culture Fluorescence When Monitoring Bioreactors. J. Ferment. Bioeng. 1992, 74 (2), 8.

(340) Valm, A. M.; Welch, J. L. M.; Rieken, C. W.; Hasegawa, Y.; Sogin, M. L.; Oldenbourg, R.; Dewhirst, F. E.; Borisy, G. G. Systems-level analysis of microbial community organization through combinatorial labeling and spectral imaging. Proc. Natl. Acad. Sci. 2011, 108 (10), 4152–4157 DOI: 10.1073/pnas.1101134108.

(341) Musat, N.; Foster, R.; Vagner, T.; Adam, B.; Kuypers, M. M. M. Detecting metabolic activities in single cells, with emphasis on nanoSIMS. FEMS Microbiol. Rev. 2012, 36 (2), 486–511 DOI: 10.1111/j.1574-6976.2011.00303.x.

391 References

(342) Panhorst, M.; Sorger-Herrmann, U.; Wendisch, V. F. The pstSCAB operon for phosphate uptake is regulated by the global regulator GlxR in Corynebacterium glutamicum. J. Biotechnol. 2011, 154 (2–3), 149–155 DOI: 10.1016/j.jbiotec.2010.07.015.

(343) Silberbach Schäfer, M., Hüser, A. T., Kalinowski, J., Pühler, A., Krämer, R., Burkovski, A., M. Adaptation of Corynebacterium glutamicum to Ammonium Limitation: A Global Analysis Using Transcriptome and Proteome Techniques. Appl. Environ. Microb. 2005, 71 (5), 12.

(344) Liu, Y.; Walther-Antonio, M. Microfluidics: A new tool for microbial single cell analyses in human microbiome studies. Biomicrofluidics 2017, 11 (6), 61501 DOI: 10.1063/1.5002681.

(345) Hutchens, M.; Luker, G. D. Applications of bioluminescence imaging to the study of infectious diseases. Cellular Microbiology. 2007, pp 2315–2322.

(346) Kuru, E.; Hughes, H. V.; Brown, P. J.; Hall, E.; Tekkam, S.; Cava, F.; de Pedro, M. a; Brun, Y. V; VanNieuwenhze, M. S. In Situ probing of newly synthesized peptidoglycan in live bacteria with fluorescent D-amino acids. Angew. Chemie 2012, 51 (50), 12519–12523 DOI: 10.1002/anie.201206749.

(347) Siegrist, M. S.; Swarts, B. M.; Fox, D. M.; Lim, S. A.; Bertozzi, C. R. Illumination of growth, division and secretion by metabolic labeling of the bacterial cell surface. FEMS 392 References

Microbiol. Rev. 2015, 39 (2), 184–202 DOI: 10.1093/femsre/fuu012.

(348) Toyofuku, M.; Cárcamo-Oyarce, G.; Yamamoto, T.; Eisenstein, F.; Hsiao, C.-C.; Kurosawa, M.; Gademann, K.; Pilhofer, M.; Nomura, N.; Eberl, L. Prophage-triggered membrane vesicle formation through peptidoglycan damage in Bacillus subtilis. Nat. Commun. 2017, 8 (1), 481 DOI: 10.1038/s41467-017- 00492-w.

(349) Fritzsche, M.; Mandenius, C.-F. Fluorescent cell-based sensing approaches for toxicity testing. Anal. Bioanal. Chem. 2010, 398 (1), 181–191 DOI: 10.1007/s00216-010-3651-6.

(350) Santos, R. S.; Figueiredo, C.; Azevedo, N. F.; Braeckmans, K.; De Smedt, S. C. Nanomaterials and molecular transporters to overcome the bacterial envelope barrier: Towards advanced delivery of antibiotics. Adv. Drug Deliv. Rev. 2017 DOI: 10.1016/j.addr.2017.12.010.

(351) Kim, H. S.; Weiss, T. L.; Thapa, H. R.; Devarenne, T. P.; Han, A. A microfluidic photobioreactor array demonstrating high- throughput screening for microalgal oil production. Lab Chip 2014, 14 (8), 1415 DOI: 10.1039/c3lc51396c.

(352) Boxshall Wu, M.-H., Cui, Z., Cui, Z., Watts, J. F., Baker, M. A., K. Simple Surface Treatments to Modify Protein Adsorption and Cell Attachment Properties within a 393 References

Poly(dimethylsiloxane) Micro-Bioreactor. Surf. Interface Anal. 2006, 38, 4.

(353) Doherty, E. A. S.; Meagher, R. J.; Albarghouthi, M. N.; Barron, A. E. Microchannel Wall Coatings for Protein Separations by Capillary and Chip Electrophoresis. Electrophoresis 2003, 24, 21.

(354) Hu Ren, X., Bachman, M., Sims, C. E., Li, G. P., Allbritton, N., S. Surface Modification of Poly(dimethylsiloxane) Microfluidic Devices by Ultraviolet Polymer Grafting. Anal. Chem. 2002, 74, 7.

(355) Trantidou, T.; Elani, Y.; Parsons, E.; Ces, O. Hydrophilic surface modification of PDMS for droplet microfluidics using a simple, quick, and robust method via PVA deposition. Microsystems Nanoeng. 2017, 3, 16091 DOI: 10.1038/micronano.2016.91.

(356) Kim, B.-Y.; Hong, L.-Y.; Chung, Y.-M.; Kim, D.-P.; Lee, C.- S. Solvent-Resistant PDMS Microfluidic Devices with Hybrid Inorganic/Organic Polymer Coatings. Adv. Funct. Mater. 2009, 19 (23), 3796–3803 DOI: 10.1002/adfm.200901024.

(357) Ibarlucea, B.; Fernández-Sánchez, C., Demming, S.; Büttgenbach, S.; Llobera, A. Selective Functionalisation of PDMS-Based Photonic Lab on a Chip for Biosensing. Analyst 2011, 136, 7. 394 References

(358) Gruber, P.; Marques, M. P. C.; Szita, N.; Mayr, T. Integration and application of optical chemical sensors in microbioreactors. Lab Chip 2017, 17 (16), 2693–2712 DOI: 10.1039/c7lc00538e.

(359) Lee, K. S.; Boccazzi, P.; Sinskey, A. J.; Ram, R. J. Microfluidic Chemostat and Turbidostat with Flow Rate, Oxygen, and Temperature Control for Dynamic Continuous Culture. Lab Chip 2011, 11, 11.

(360) Valencia-Burton, M.; McCullough, R. M.; Cantor, C. R.; Broude, N. E. RNA visualization in live bacterial cells using fluorescent protein complementation. Nat. Methods 2007, 4 (5), 421–427 DOI: 10.1038/nmeth1023.

(361) Oedit, A.; Vulto, P.; Ramautar, R.; Lindenburg, P. W.; Hankemeier, T. Lab-on-a-Chip hyphenation with mass spectrometry: strategies for bioanalytical applications. Curr. Opin. Biotechnol. 2015, 31, 79–85 DOI: 10.1016/j.copbio.2014.08.009.

(362) Daetwyler, S.; Huisken, J. Fast Fluorescence Microscopy with Light Sheets. Biol. Bull. 2016, 231 (1), 14–25 DOI: 10.1086/689588.

(363) Leung, K.; Zahn, H.; Leaver, T.; Konwar, K. M.; Hanson, N. W.; Page, a. P.; Lo, C.-C.; Chain, P. S.; Hallam, S. J.; Hansen, C. L. A programmable droplet-based microfluidic device applied to multiparameter analysis of single microbes and 395 References

microbial communities. Proc. Natl. Acad. Sci. 2012, 109 (20), 7665–7670 DOI: 10.1073/pnas.1106752109.

(364) Dusny, C.; Schmid, A. Microfluidic single-cell analysis links boundary environments and individual microbial phenotypes. Environ. Microbiol. 2015, 17 (6), 1839–1856 DOI: 10.1111/1462-2920.12667.

(365) Landry, Z. C.; Giovanonni, S. J.; Quake, S. R.; Blainey, P. C. Optofluidic Cell Selection from Complex Microbial Communities for Single-Genome Analysis. In Biophysical Chemistry; 2013; Vol. 257, pp 61–90.

(366) Baret, J.-C.; Miller, O. J.; Taly, V.; Ryckelynck, M.; El-Harrak, A.; Frenz, L.; Rick, C.; Samuels, M. L.; Hutchison, J. B.; Agresti, J. J.; et al. Fluorescence-activated droplet sorting (FADS): efficient microfluidic cell sorting based on enzymatic activity. Lab Chip 2009, 9 (13), 1850–1858 DOI: 10.1039/b902504a.

(367) Zhang, R.; Gong, H.-Q.; Zeng, X. D.; Sze, C. C. A High- Throughput Microfluidic Biochip to Quantify Bacterial Adhesion to Single Host Cells by Real-Time PCR Assay. Anal. Bioanal. Chem. 2013, 405, 6.

(368) Lamprecht, M.; Sabatini, D.; Carpenter, A. CellProfilerTM: free, versatile software for automated biological image analysis. Biotechniques 2007, 42 (1), 71–75 DOI: 10.2144/000112257. 396 References

(369) Jones, T. R.; Carpenter, A. E.; Lamprecht, M. R.; Moffat, J.; Silver, S. J.; Grenier, J. K.; Castoreno, A. B.; Eggert, U. S.; Root, D. E.; Golland, P.; et al. Scoring diverse cellular morphologies in image-based screens with iterative feedback and machine learning. Proc. Natl. Acad. Sci. U. S. A. 2009, 106 (6), 1826–1831 DOI: 10.1073/pnas.0808843106.

397