Engineering of Hyaluronic Acid Synthases from Streptococcus Equi Subsp

Engineering of Hyaluronic Acid Synthases from Streptococcus equi subsp. zooepidemicus and Pasteurella multocida Towards Improved HA Chain Length and Titer

Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation

vorgelegt von

John Cyrus Baltazar Mandawe, MSc

aus Paraῆaque, Philippinen

Berichter: Universitätsprofessor Dr. rer. nat. Ulrich Schwaneberg

Universitätsprofessor Dr.-Ing. Lars M. Blank

Tag der mündlichen Prüfung: 31.Oktober 2018

Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek verfügbar.

"It is not the critic who counts; not the man who points out how the strong man stumbles, or where the doer of deeds could have done them better. The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood; who strives valiantly; who errs, who comes short again and again, because there is no effort without error and shortcoming; but who does actually strive to do the deeds; who knows great enthusiasms, the great devotions; who spends himself in a worthy cause; who at the best knows in the end the triumph of high achievement, and who at the worst, if he fails, at least fails while daring greatly, so that his place shall never be with those cold and timid souls who neither know victory nor defeat." -Theodore Roosevelt

Acknowledgements

“I've learned that people will forget what you said, people will forget what you did, but people will never forget how you made them feel” - Maya Angelou

My deepest gratitude to Prof. Dr. Ulrich Schwaneberg for allowing me the opportunity to pursue my PhD studies. I have learned and grown so much because of you.

My sincere appreciation to Prof. Dr. Lars Blank and Prof. Dr. Lothar Elling as supportive project collaborators and as members of my dissertation committee. I also extend my gratitude to Prof. Dr. Uwe Conrath as the chairperson of my dissertation committee.

I acknowledge the Deutsche Bundesstiftung Umwelt (DBU) for the financial support from 2013-2016. Very special thanks to SeSaM-Biotech for giving me the opportunity to sustain myself for the last several months of my doctoral studies so that I could bring all these efforts to fruition.

I acknowledge those who contributed to the pmHAS KnowVolution publication: Dr. Belen Infanzon, Anna Eisele, Henning Zaun, Dr. Jürgen Kuballa, Dr. Mehdi D. Davari, Dr. Felix Jakob, Prof. Dr. Lothar Elling and Prof. Dr Ulrich Schwaneberg. Also, special thanks to Shohana Islam for the PLICing primers and JARA-HPC from RWTH Aachen University under projects JARA0169 for granting computer simulations resources. Thank you as well to Dr. Gaurao Dhoke for providing assistance with the homology model manipulations for the ChemBioChem cover page.

I thank Dr. Felix Jakob for the supervision and assistance and the members of the Biohybrid subgroup for the scientific discussions. I would like to thank every single member of the Schwaneberg group, not only for their technical and administrative support, but also for being amazing colleagues. Thank you to Dr. Kristin Rübsam for the help with confocal microscopy. I also extend my special appreciation to Shohana Islam, Patrizia Pazdzior and Dr. Juliana Kurniadi for their support and kindness during the stormy days. I ingrain this indebtedness to my memory.

My sincerest thanks to my brothers Paul, George and Ringo (“The Beatles”), and my families and friends in Canada and Germany for their support, inspiration and encouragement throughout the years. Tim, Swanny, Mom Astrid and Anne, thank you for welcoming me into your respective families and for your kindness, care and thoughtfulness over the years. I highly appreciate you.

Finally, I dedicate this PhD to my parents, Maria Lucila and Alvin, who gave up their education to raise their “Beatles”. Thank you for all the sacrifices you have made for us and for your endless love and support. You have instilled in me the values of hard work, diligence, perseverance, independence, responsibility, integrity and respect. I offer you this doctoral degree. I keep the life lessons. Todo para la familia!

III

Publications

Parts of this thesis have been published:

Mandawe J, Infanzon B, Eisele A, Zaun H, Kuballa J, Davari MD, et al. Directed Evolution of Hyaluronic Acid Synthase from Pasteurella multocida Towards High Molecular Weight Hyaluronic Acid. ChemBioChem. 2018;19:1414-23.

Parts of this thesis will be published:

Mandawe J*, Anand D*, Jakob F, Zaun H, Kuballa H, Schwaneberg U. A facile and inexpensive toolbox for differential hyaluronic acid synthesis: From kilodalton to megadalton scale (manuscript in preparation).

*shared first authorship

Other publications:

Vargas WA, Mandawe J and Kenerley CM. Plant-derived sucrose is a key element in the symbiotic association between Trichoderma virens and maize plants. Plant Physiol. 2009;151:792-808.

Jakob F, Martinez R, Mandawe J, Hellmuth H, Siegert P, Maurer KH, Schwaneberg U. Surface charge engineering of a Bacillus gibsonii subtilisin protease. Appl Microbiol Biotechnol. 2012;97:6793-6802.

Abstract Hyaluronan (hyaluronic acid; HA) is the non-sulfated glycosaminoglycan product of HA synthases that can vary in length from 103-107 Daltons. Depending on polymer length, concentration and localization, HA possesses variable physicochemical properties that serve useful in pharmaceutical, biomedical and cosmetic applications worth billions ($) in commercial valuation worldwide. A deeper molecular understanding of the control of HA polymerization by the synthase machinery can, therefore, facilitate the tuned production of HA, according to the intended application. The general objective of this doctoral investigation, therefore, is to implement protein engineering principles to Class I and Class II HA synthases, with supplication of computational modeling, to improve HA production by means of polymer length and quantity.

The Class I HAS from Streptococcus equi subsp. zooepidemicus (seHAS) was first subjected to protein engineering. seHAS was recombinantly expressed in three microbial hosts (E. coli, B. subtilis and S. cerevisiae) and could direct HA synthesis in all three hosts. E. coli was selected for enzyme engineering due to easy handling and quick doubling time. Fluorescent epitope tagging confirmed the presence of seHAS and its localization in the outer membrane of the Gram-negative host, contrary to the inner membrane reported in literature. Of the many screening systems attempted to be established, the agarose gel electrophoresis screening platform was the most reliable and could discriminate between empty vector controls, wild type and seHAS variants. Screening of the site-saturation mutagenesis libraries (of the conserved Cys226, Cys262, and Cys281 and polar membrane residues Lys48 and Glu327) and one random mutagenesis library (1392 error-prone PCR variants) failed to identify one seHAS variant with improved chain length specificity. However, alternative positive results were discovered. Site-saturation mutagenesis variants (K48L and K48E) produced consistently monodispersed low molecular weight (LMW; < 0.5 MDa) HA products, while epPCR variants H2 (N345S/F403L) and A6 (R347S/F362S) produced high molecular weight (HMW; >1 MDa) HA with polydispersity lower than that of seHAS- WT. Homology model analysis hinted at the potential role of HA-HAS interaction in the control of HA polymerization. The discovery of these new positions bifurcates into another dimension of HA, which is chain polydispersity. A better understanding of these product-enzyme interactions can provide clues for production of monodispersed HA.

The second protein engineering campaign involved the Class II HAS from Pasteurella multocida (pmHAS). The knowledge-gaining directed evolution (KnowVolution) approach successfully improved the enzymatic activity of the membrane-associated pmHAS. Two screening systems were simultaneously employed to detect improvements in enzymatic output: agarose gel electrophoresis for chain length and the CTAB turbidimetric assay for HA titer. With CTAB, absorbance values of HA synthesized by pmHAS-expressing E. coli BL21 GOLD (DE3) cells were at least 5-fold higher than that

V of the baseline (empty vector control). Through KnowVolution, seven improved epPCR variants out of 1392 were identified, eight prospective beneficial positions from these variants were saturated and the most beneficial amino acid substitutions (T40L, V59M and T104A) were recombined to generate the final variant (pmHAS-VF). Production of HA up to 4.7 MDa and with a two-fold improvement in mass-based total turnover number over wild type was achieved. This is the first case of a Class II HA synthase directed evolution and an example of a simultaneous dual property improvement resulting from protein engineering. The most complete and validated model to date of pmHAS32-703 was also generated to gain molecular insight into the improved properties. The substitutions in pmHAS-VF are located at the N-terminal domain, away from either glycosyltransferase active sites of pmHAS, suggesting their non-catalytic role. Molecular dynamics simulations reveal the improved flexibility of the N-terminal region allowing it to swing from the GlcNAc-transferase domain to the GlcA-transferase domain. This suggests a newly found importance of the N-terminal domain in HA synthesis. Overall, the ability to synthesize longer HA polymers at higher output brings promise to improved HA production.

In the last chapter entitled, “Methods for Differential Hyaluronic Acid Synthesis by pmHAS”, methods for production of either HMW HA or LMW HA were investigated. HA synthesis using purified pmHAS was slow (days) despite the addition of the tetrasaccharide synthesis initiator (HA4) and produced only LMW HA. HA synthesis with lysate from sonicated pmHAS-expressing E. coli cells was faster, was HA4- independent and produced LMW HA. When pmHAS-expressing E. coli cells were subjected to freeze-thaw and lysozyme treatment, polydispersed HMW HA were generated. More importantly, the strain E. coli BL21 GOLD (DE3) expressing pmHAS was demonstrated to autonomously produce HA, suggesting that this strain naturally possesses metabolic pathways to synthesize the precursors: UDP-GlcA and UDP- GlcNAc. Moreover, the addition of nucleotide sugar precursors prompts the production of fewer but longer HA polymers. When the in vitro synthesized HA products were boiled for 5 min, up to 97 % of protein contaminants (detected by BCA assay and Image J analysis) could be precipitated out of solution, albeit at the expense of HA chain depolymerization. This work not only illustrates that cell disruption influences HA production but also offers quick and inexpensive methods to generate semi-purified LMW or HMW HA.

In these investigations, protein engineering and computational modeling were employed complementarily to dissect both Class I seHAS and Class II pmHAS, not only having industrial relevance in mind, but to also contribute to the existing knowledge of HAS biology. Ultimately, protein engineering is a powerful tool for tuning HA production with respect to polymer length, quantity and size distribution.

Abbreviations aa amino acid amp ampicillin bp base pair B. subtilis Bacillus subtilis Da Dalton ddH2O double distilled water DNA deoxyribonucleic acid dNTP deoxynucleoside triphosphate E. coli Escherichia coli e.g. exempli gratia (for example) epPCR error prone polymerase chain reaction EV empty vector HA(S) hyaluronic acid or hyaluronan (synthase) HMW high molecular weight i.e. id est (that is) IPTG isopropyl-β-D-thiogalactopyranoside IVS in vitro synthesis kan kanamycin kb kilobase(s) kDa kiloDalton(s) LB Luria-Bertani medium LMW low molecular weight MTP microtiter plate PBS phosphate buffered saline PCR polymerase chain reaction PDB Protein Data Bank PLICing phosphorothioate-based ligase independent gene cloning pmHAS HA synthase from Pasteurella multocida S. cerevisiae Saccharomyces cerevisiae SDM site-directed mutagenesis SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis seHAS HA synthase from Streptococcus equi subsp. zooepidemicus SeSaM sequence saturation mutagenesis SOC super optimal broth with catabolic repressor SSM site saturation mutagenesis Taq Thermus aquaticus TB terrific broth UDP-GlcA uridine diphosphate-glucuronic acid UDP-GlcNAc uridine diphosphate-N-acetylglucosamine VF final variant WT wild type YASARA Yet Another Scientific Artificial Reality Application YPD yeast extract-peptone-dextrose (medium)

VII

Table of Contents

Contents Acknowledgements ...... III Publications ...... IV Abstract...... V Abbreviations ...... VII Table of Contents...... VIII 1. Chapter 1: Review of the Literature ...... 11 1.1. Introduction to Hyaluronan ...... 11 1.1.1. Structure of Hyaluronan ...... 12 1.1.2. Rheological Properties and Biological Functions of Hyaluronan ...... 13 1.1.3. Applications of Hyaluronan ...... 14 1.2. Synthesis of Hyaluronan ...... 15 1.2.1. Class I HA Synthases ...... 17 1.2.1. Class II HA Synthase ...... 23 1.3. Catabolism of Hyaluronan ...... 25 1.4. Production of Hyaluronan - State of the Art ...... 27 1.5. Protein Engineering ...... 29 1.5.1. Rational Design vs Directed Evolution ...... 29 1.5.2. Diversity Generation Methods ...... 30 1.5.3. Screening and Selection ...... 32 1.6. Purpose of the Study ...... 33 2. Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer ...... 34 2.1. Project Objective ...... 34 2.2. Materials and Methods ...... 35 2.2.1. Generation of seHAS Constructs ...... 35 2.2.2. Cell cultivation and Protein Production ...... 42 2.2.3. In vitro HA Biosynthesis ...... 44

VIII

Table of Contents

2.2.4. Development of Protocols for Screening Mutant Libraries in 96-Well Microtiter Plate Format ...... 46 2.2.5. Computational Modeling of seHAS ...... 49 2.2.6. Cellular Localization of seHAS using Fluorescence Microscopy ...... 49 2.2.7. Diversity Generation ...... 51 2.3. Results ...... 54 2.3.1. Selection of Most Suitable Microbial HA Production System for the seHAS Engineering Campaign ...... 54 2.3.2. Establishment of the Screening System in 96-Well Microtiter Plate Format for the seHAS Engineering Campaign ...... 64 2.3.3. Rational Evolution of seHAS ...... 76 2.3.4. Random Mutagenesis of seHAS ...... 83 2.3.5. Investigating the Validity of the Directed Evolution Hits ...... 87 2.4. Discussion ...... 93 2.5. Conclusion ...... 102 3. Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer ...... 103 3.1. Declaration ...... 103 3.2. Project Objective ...... 104 3.3. Materials and Methods ...... 106 3.3.1. Generation of pmHAS Construct, Cell Cultivation and Protein Production ……………………………………………………………………………………………………….106 3.3.2. Diversity Generation ...... 107 3.3.3. In vitro HA Biosynthesis ...... 110 3.3.4. MTP-Based Screening of Variants ...... 110 3.3.5. Characterization of pmHAS-WT and pmHAS-VF ...... 111 3.3.6. Molecular Modeling of pmHAS ...... 112 3.4. Results ...... 113 3.4.1. Investigating the Possibility of the CTAB Turbidimetric Assay and Agarose Gel Electrophoresis as Screening Systems ...... 114 3.4.2. Directed evolution of pmHAS for Improved Polymerizing Activity ...... 115 3.4.3. Characterization of pmHAS-VF and pmHAS-WT with Respect to HA Titer and Polymer Length ...... 124

Table of Contents

3.4.4. Structural Insight into the Improved Polymerizing Activity of pmHAS ... 126 3.5. Discussion ...... 129 3.6. Conclusion ...... 133 4. Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS ...... 134 4.1. Declaration ...... 134 4.2. Project Objective ...... 134 4.3. Materials and Methods ...... 135 4.3.1. Generation of pmHAS Constructs ...... 135 4.3.2. Cell Cultivation ...... 135 4.3.3. Purification of pmHAS-WT and pmHAS-VF by Nickel-NTA Affinity Chromatography ...... 136 4.3.4. Qualitative Analytical Methods ...... 137 4.3.5. Quantitative Analytical Methods ...... 139 4.3.6. Hyaluronic Acid Synthesis ...... 140 4.3.7. Hyaluronan Digestion by Hyaluronidase ...... 141 4.4. Results ...... 141 4.4.1. HA Biosynthesis with Purified pmHAS-WT and pmHAS-VF ...... 142 4.4.2. HA Biosynthesis with Crude Lysate of pmHAS-Expressing E. coli ...... 144 4.4.3. Investigating the Influence of Cell Disruption on HA Biosynthesis ...... 149 4.4.4. Proof-of-Principle for Differential HA Production ...... 153 4.5. Discussion ...... 158 4.6. Conclusion ...... 162 5. Chapter 5: Final Summary and Remarks ...... 163 6. Appendix ...... 166 6.1. Figures ...... 166 6.2. Tables ...... 171 7. References ...... 174 Declaration ...... 193 Curriculum vitae ...... 194

Chapter 1: Review of the Literature

1. Chapter 1: Review of the Literature "If you can't explain it simply, you don't understand it well enough." - Albert Einstein

Some contents of this chapter have been published [1] and were reproduced with permission from the publisher. Credit is given to the original source:

John Mandawe, Belen Infanzon, Anna Eisele, Henning Zaun, Jürgen Kuballa, Mehdi D. Davari, Felix Jakob, Lothar Elling and Ulrich Schwaneberg: Directed Evolution of Hyaluronic Acid Synthase from Pasteurella multocida Towards High Molecular Weight Hyaluronic Acid. ChemBioChem. 2018. 19. 1414-1423. Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.

1.1. Introduction to Hyaluronan Since its initial isolation from vitreous humour of bovine eyes in 1934 [2], hyaluronic acid (hyaluronan, HA) has proven to be an important biopolymer owing to its intrinsic biological and chemical properties. In 1937, Kendall isolated HA from the capsules of Groups A and C Streptococci [3]. Hyaluronic acid is a linear glycosaminoglycan comprised of (-4)-D-GlcA-(β1-3)-D-GlcNAc(β1->) disaccharide repeats [4] and is ubiquitous in nature, being readily produced by viruses [5], prokaryotes [6, 7] and vertebrates [8-10]. In relation to humans, HA is found in tissues including synovial fluid, eyes, skin and umbilical cord [11, 12] as well as throughout the central nervous system [13].

In 1949, Ragan and Mayer performed a comparative study of normal and rheumatic synovial fluids by assessing the concentration and viscosity in both sample types [14]. In 1951, Ogston and Stanier published the structure of HA in aqueous solution and found the association between viscosity and concentration of HA [15]. Four years later, a more developed light scattering technique shed light into the helical configuration of HA in solution [16]. To date, scientific knowledge and technology have enabled the chemistry and biology of HA to be elucidated, as well as its wide array of human applications.

Chapter 1: Review of the Literature

1.1.1. Structure of Hyaluronan Hyaluronan is a linear, anionic polysaccharide that is always chemically identical irrespective of the source [17]. The main difference between HA products is their molecular weight. To illustrate, high molecular weight (HMW) HA can contain 5 000 disaccharide units corresponding to 2 MDa in size and is approximately 5 µm in length [18].

Figure 1-1: Basic subunit of HA. Alternating GlcA and GlcNAc monomers bonded by ß-1,3 and ß-1,4 glucorinidic linkages.

The glucoronidic linkage has been reported by Weissmann et al in 1952, confirming that alternating glucosamine and glucuronic acid residues in the HA chain and the presence of β-1,3 linkage through a series of chemical derivation and elemental analysis [4]. Naturally, HA is unbranched, without chemical modifications and can vary in length. It is now known that polymer size influences HA activity, thereby making HA an information-rich system [19]. The distinct alternating nature of the β-1,3 and β-1,4 glycosidic linkages within the HA molecule contribute to its properties (Figure 1-1). The hydroxyl groups, carboxylate moieties of GlcA and acetamide groups are situated equatorially which minimize stearic constraints for chemical modifications and provide hydrophilicity to the HA polymer [20]. The simple hydrogen atoms (-CH), on the other hand, occupy the axial positions. This creates hydrophobic faces in the secondary structure of HA that enable the formation of an energetically favourable meshwork-like β-sheet tertiary structure. This tertiary structure in return is stabilized by intermolecular hydrogen bonding between the oxygen of the carboxyl group and hydrogen of the acetylamine group [21, 22] and in aqueous solution, with the participation of water

Chapter 1: Review of the Literature molecules [17]. Combined hydrophobic forces, hydrogen bonding and electrostatic interaction are ultimately responsible for the intrinsic propensity of HA molecules to aggregate and form HA matrices [21]. On the supramolecular level, structures of HA polymers vary from elongated chains, ‘relaxed’ (non-condensed) helices, condensed rod- shaped structures and ‘clips’).

HA behaves as an independent molecule in diluted solutions. HA molecules typically take on a coiled conformation [17] and x-ray data revealed that HA as salt of Na+ or K+ exists as a double helix [23] and can even adapt a supra-helical conformation forming a dense microgel [24]. By adapting a dynamic coil conformation, HA can bind huge quantities of water and at concentration exceeding 1 mg/mL, individual coils “entangle” forming a three-dimensional network or web that give rise to the rheological properties of HA [17, 22].

1.1.2. Rheological Properties and Biological Functions of Hyaluronan The rheological (viscoelastic) properties of HA provide insight into how and why HA is involved in a wide array of biological processes. It has been stated above that HA can adapt various macromolecular confirmations, with helix being the most energetically favourable [25]. These structures can therefore be changed by external factors including pressure, salts, polymer size and concentration and temperature and define the viscoelastic properties of HA [17]. The governing forces acting on the HA molecule, especially electrostatic repulsion between the negative charges, contribute to the strained conformation. With increased concentration, HA chains overlap and shielding of the charges occurs resulting in a network structure. The transition from helix to coil (intramolecular melting) occurs depending on the external stimulus and transition from ordered coil to random coil is made possible by weakening intermolecular interactions, while the transition from random coil to rod conformation is only possible for HA samples with at least a molar mass of 17 kDa [26]. Viscosity depends on the concentration of HA in solution [27] and temperature [28], while elasticity has been shown to change proportionally with HA concentration and molecular weight [29, 30].

Chapter 1: Review of the Literature

It is therefore these characteristics and viscoelasticity determinants of HA that define the properties suitable for biological functions.

HA is highly hygroscopic. One gram of HA can bind up to six liters of water [31]. This, on the molecular level, is the result of the hydrogen bonding interaction with the carboxyl and N-acetyl functionalities within the molecule. With this capacity for large hydrodynamic volume and the ability to form solutions with high viscosity and elasticity, HA has functions in space filing, lubrication and shock absorption [12, 32, 33].

1.1.3. Applications of Hyaluronan One of the earliest applications of HA as a medicinal agent occurred during the second World War, when bandages used to treat soldiers were infused with extracts from the umbilical cord to accelerate the healing process [17]. In addition, arthritis from race horses were relieved by intrajoint injection while intraocular lens implantations gave rise to the first use of HA in ophthalmology [34]. The involvement of HA in biological, pathological, and therapeutic processes in humans have been explored in several other review articles [35-38].

HA-related diseases, especially during old age, are tightly coupled to HA homeostasis, which ultimately affects HA chain length and concentration in the body. Not only are the rheological properties important for the biological function of HA but also are the localization [13] and the polymer size [19, 39]. The polymer size of HA defines its application. For example, low molecular weight (<5 kDa) HA have been shown to induce angiogenesis, prevent tumor progression [40, 41] and prevent differentiation during nerve cell repair [13]. The expanding applications of HMW HA include ophthalmology [42], joint lubrication [20, 43], osteoarthritis treatment [44], wound healing [45], tissue engineering to induce cell proliferation and migration [46] and drug delivery [47]. Moreover, the non-immunogenic [3, 48], hygroscopic nature [2] and its ability to scavenge free-radicals [49] make HA (especially low molecular weight) an excellent active agent in cosmetic and anti-aging applications cosmetics [50, 51]. In 2016, the global market value of HA was appraised at $ 8.3 billion by Grand View Research and this trend will continue to rise [52].

Chapter 1: Review of the Literature

To date, HA classification according to molecular size has not been clearly defined. Various reports [39, 53, 54] have attempted to assign arbitrary molecular weight cut-offs (LMW is interchangeably termed as “oligosaccharides” or having a size < 0.5-0.7 MDa; HMW > 1 MDa; MMW is sometimes not considered in literature or defined as the middle between LMW and HMW according to the author’s arbitrary range), unfortunately, this convention has not been standardized by the hyaluronan community. For purpose of this thesis, LMW HA is considered < 1 MDa, while HMW HA is considered > 1 MDa.

Practical applications of HA require (near) absolute purity. Moreover, preparation of uniformly sized HA is also mandatory since the product distribution of HA is naturally polydispersed. Polydispersity is measured by the ratio Mw/Mn (Mw: mass weighted molecular weight, Mn: number weighted molecular weight) and is used to characterize the molecular weight distribution of HA [55]. This polydispersity index ratio is a crucial HA characteristic that must be considered due to size-dependent effects.

1.2. Synthesis of Hyaluronan Naturally-occurring HA is a linear undecorated glycosaminoglycan [13] synthesized by a family of glycosyltransferases (GT) called hyaluronan synthases (HAS). Separated into two classes by amino acid sequence, membrane topology and mode of action [56], HAS are theorized to be a result of convergent evolution rather than horizontal gene transfer [57]. Class I HAS (mammalian, viral, amphibian, and bacterial) are membrane-bound synthases that catalyze the alternating transfer of GlcNAc and GlcA to the nascent HA chain in the plasma membrane [58-61]. Moreover, this systematic and processive polymerization occurs either at the reducing [62-64] or non-reducing end of the growing HA chain depending on the subclass [65]. To date, regulatory mechanisms governing chain length specificity and titer are still under investigation. Information is still insufficient to formulate a detailed mechanism of hyaluronan polymerization, however, it is for certain that regulation of HA synthesis occurs on multiple levels.

Chapter 1: Review of the Literature

The biosynthetic pathway for HA is depicted in Figure 1-2. Glucose is first phosphorylated by a hexokinase to form glucose-6-phosphate, a critical building block for the two precursor pathways. To form UDP-GlcA, the pathway involves the enzymes phosphoglucomutase, UDP-glucose pyrophosphorylase and dehydrogenase, while synthesizing UDP-GlcNAc requires more enzymes (hasE, glmS, glmM and hasD). The two parallel metabolic branches converge to produce the precursor nucleotide sugars, which are consequently polymerized by the hyaluronic acid synthase to form hyaluronic acid. To complete the HA biosynthetic pathway in E. coli, the kfiD and hasA genes along with glucosamine and glucose supplementation are required [66]. It is apparent that the convergence of both UDP-GlcNAc and UDP-GlcA pathways is important for enzymatic conversion of HA molecules by the HA synthase.

Figure 1-2: General overview of the hyaluronic acid biosynthetic pathway [67-70]. The red font represents the E. coli homologues, the green represents S. zooepidemicus, blue represents B. subtilis and the black represents standard nomenclature.

HAS belong to the glycosyltransferase (GT) family that catalyzes the transfer of activated sugar unit to an acceptor to assemble homo- or heteropolysaccharide molecules [71]. Typically, GTs are membrane-integrated but have cytosolic domains that are actively involved in polymer synthesis whereby sugar monomers from nucleotide-activated sugars are transferred to the polymer chain [72]. HA synthases belong to the GT family

Chapter 1: Review of the Literature that are classified into two classes according to their amino acid sequence, membrane association and mode of HA polymerization [56, 73]. In family-2 of GTs, it is proposed that reactions occur via SN2-like nucleophilic displacement reaction by the acceptor resulting in inversion of the configuration of the donor’s anomeric C1 carbon from α to β [71, 72]. Moreover, the nascent polysaccharide chain is translocated across the plasma membrane through the transmembrane domain(s) of the glycosyltransferase as exemplified with the cellulose synthase [74] and hyaluronan synthase [63, 75].

Aspartate residues from three important sequence motifs are reported to be crucial for interacting with the donor and acceptor molecules and overall catalytic activity [71]. The “DDG” motif is involved in nucleotide binding; “DXD” is responsible for coordinating divalent cations like Mg2+ and Mn2+; and “TED” in cellulose or “GDD” in HAS synthase is responsible for deprotonating the acceptor during glycosyltransfer [71, 74].

HAS are multifunctional GTs because of their ability to recognize two substrates (UDP- GlcA and UDP-GlcNAc) to synthesize heterosaccharides with alternating β-1,3 and β-1,4 linkages and is proven thus far to be solely responsible for the translocation of its polymer product through the membrane [71, 73, 75]. The crystal structure of any HA synthase has yet to be solved, however, topology predictions suggest that bacterial HAS have 4-5 transmembrane regions while vertebrate HAS contain 6-7 transmembrane domains. As previously stated, the cytosolic domains are catalytically important [76]. Functionally, Class I viral/vertebrate HAS [64], streptococcal HAS [77] and Class II pmHAS [78] have been determined to be a monomer, however, recent evidences suggest that streptococcal HAS form a homo-dimer to be catalytically active [64]. This reflects an unsolved facet in the biology of HAS that requires further work.

1.2.1. Class I HA Synthases A majority of HA synthases belong to Class I (Figure 1-3) and have been identified from green algae virus, to bacteria to humans. The Class I HAS from prokaryotes and vertebrates share approximately 30 % sequence identity and a common membrane topology [59, 60]. Collectively, Class I HAS share the following features [69]:

Chapter 1: Review of the Literature

1. They share amino acid similarities in the central region of the protein sequence and possess a single glycosyltransferase (GT2) family module; 2. They are integral membrane enzymes with four to six predicted transmembrane domains; 3. They are lipid-dependent for processive HA polymerization and extrusion; 4. The synthetic process is processive and can occur either at the reducing end or non-reducing end of the HA polymer; 5. The cytoplasmic domain of the enzyme is involved in substrate binding and HA polymerization; 6. The heterosaccharide product is extruded through the plasma membrane using a pore provided by the enzyme itself [33].

Figure 1-3: HA polymerization by seHAS. Class I seHAS is membrane-embedded and requires molecules of cardiolipin (orange circles) for stability and activity. The cytosolic domain binds activated nucleotide sugar monomers and facilitates glycosyltransferase reactions by adding GlcA or GlcNAc to the growing chain in an alternating fashion. The growing HA chain is elongated and translocated through the pore of the synthase itself across the cell membrane until its eventual release. The reducing end elongation reaction is specified in the figure [65].

Chapter 1: Review of the Literature

More detailed comparisons between Class I HAS and Class II HAS have been outlined in other excellent review articles [37, 56, 69].

In mammals, three HAS isoforms (HAS1, HAS2 and HAS3), arguably as a result of ancestral gene duplication, are responsible for HA production on the cell membrane and polymer extrusion [38, 79]. Each mammalian HAS isoform produces polymers of different lengths and quantity. For instance, recombinant expression experiments revealed that HAS1 exhibited 4- and 10-fold lower hyaluronan synthesizing capacity than the HAS2 and HAS3, respectively, with HAS2 producing HA (>3.6 MDa) longer than HAS1 and HAS3 (0.12-1.00 MDa) [80]. It is then postulated that differential regulation of HAS may be necessary to modulate cellular processes [33]. Literature suggests that these differences are loosely coupled with the regulation of the synthases’ catalytic activities, like phosphorylation and ubiquitination [81]. Moreover, it has been shown that deletion of HAS2 leads to severe cardiovascular abnormalities during early mammalian embryonic development and ultimate lethality, highlighting its necessity [82].

In HA-producing bacteria like Streptococcus equisimilis, three genes, hasC (encoding UDP-Glucose pyrophosphorylase; [83]), hasB (encoding UDP-Glucose dehydrogenase; [84]) and hasA (encoding HA synthase; [6]) are required [65, 85]. Being a major component in capsules, HA not only serves adhering and protective purposes, but helps evade phagocytosis by the host’s immune system and promote virulence [86]. Other natural HA-producing pathogenic bacteria include Streptococcus pyogenes, Streptococcus uberis [6, 87], Pasteurella multocida [88] and Cryptococcus neoformans [89].

Many works have been published to understand Class I HAS. Arguably, the most studied Class I member is the streptococcal HAS. Early investigations extensively studied and capitalized on the natural ability Streptococcus spp. to synthesize HA. Biochemical and radioactive experiments were performed by Dorfman and colleagues to elucidate the components required in synthesis of HA by Group A Streptococcus as well as the biochemical pathways involved in the process [90-92]. Moreover, work on HAS from Streptococcus pyogenes (spHAS) reported the membrane localization of the enzyme

Chapter 1: Review of the Literature and that the sugar nucleotide precursors and Mg2+ were sufficient for HA synthesis [92]. The streptococcal enzyme system has also been characterized [61]: pH 7.1 for optimal enzymatic activity; 10 mM Mg2+ for divalent cation with preference over 1 mM Mn2+; Km of 5 x 10-5 M for UDP-GlcA and Km of 5 x 10-4 M for UDP-GlcNAc. In 1979, Sugahara et al demonstrated that HA was released from the enzyme after elongation or completion and declared that no initiator or lipid intermediates were found that could assist in the polymerization process [93]. Mechanistic understanding of HA chain growth was contributed by Prehm who proposed that synthesis occurs at the reducing end of the HA chain in an alternating fashion [94].

The emergence of molecular cloning has sparked more interest in Class I HAS and paved the way for more structural and mechanistic characterization studies. In 1993, the gene locus responsible for HA biosynthesis was isolated and identified. This was elegantly proven by co-expressing the hasA and hasB genes in acapsular Streptococcus strains and Enteroccus faecalis and directing HA production by the naturally non-HA producing bacteria [6, 95]. Furthermore, the 42-kDa recombinant HAS from Streptococcus pyogenes was also heterologously expressed in the Gram-negative E. coli. Kinetics studies reveal the rapid simultaneous polymerization of radiolabeled precursors (10-30 monosaccharides/s) in relatively discrete manner, suggesting that the enzyme is processive [96]. This means the enzyme remains associated with the growing HA chain until released. The glycosyltransferase reactions conform to an inversion mechanism to form β-glycoside bonds perhaps coupled to stepwise movements of the extending polymer [97].

Sharing a 72% protein sequence identity with spHAS, seHAS from Group C Streptococcus equisimilis was shortly cloned, expressed and characterized [7]. The synthase, comprised of 417 amino acids (48 kDa) was expressed in E. coli SURE cells and was supplied with the UDP-sugar precursors to produce HMW HA with a polymerization rate of 160 monosaccharides/s [7]. Further characterization work using radiation inactivation and mass spectrometry revealed that the functional enzyme is a monomer and requires 16 molecules of cardiolipin for full activity [77]. Purification of the recombinant seHAS, accounted for less than 10% of the total membrane protein and

Chapter 1: Review of the Literature corroborated that indeed cardiolipin is necessary for the HA synthetic activity. This was further demonstrated when the addition of cardiolipin molecules restored the activity of the freshly purified seHAS or retained 60% of the initial activity of seHAS purified two months prior [98]. Kinetic characterization using isolated streptococcal membranes resulted in Km values of 51±5 µM and 60±7 µM for UDP-GlcA and UDP-GlcNAc, respectively, and a specific activity of 12 000 nmol/h/mg [99]. When purified, the broad pH profile ranged from pH 6.5 to 10.5, retaining the catalytic constant of the enzyme. The “naked” synthase itself requires exogenous cardiolipin to confer HA polymerizing capability and restore enzyme stability [77, 98].

Comparative sequence analysis of seHAS against other HAS reveals that seHAS contains four conserved cysteines. To further understand this significance, functional studies involving sulfhydryl reagents demonstrated inhibition of seHAS activity in time- and dose-dependent manner. Moreover, mutating the cysteine residues to alanine/serine negatively influenced the enzyme specific activity and reduced the HA product size distribution. C226 and C262 were the most affected suggesting their involvement in other seHAS functions [100]. To further dissect the importance of cysteine residues, inhibition and substrate protection studies revealed that three of the four conserved cysteines (C262, C281 and C367) were clustered very close together at the membrane- enzyme interface, while C226 is located at the inner surface of the cell membrane. Moreover, both C226 and C262 are located in proximity to a UDP binding site, which further substantiates their possible importance in seHAS functions [101]. Further mutational studies on the conserved cysteines showed uncoordinated negative effects on HA synthesis and/or product size. This essentially clarified that indeed the polymerizing activity is independent of product size control [102].

Another seHAS residues that have been examined are the conserved polar residues Lys48 and Glu327 located in membrane domains 2 and 4 [103]. In this work, the effects of substituting Lys48 to Arg or Glu and Glu327 to Lys, Asp or Gln were monitored with respect to specific enzyme activities and HA polymer product size. All variants displayed inferior HA synthase capability compared to the wild type suggesting that disrupting the interaction between the two membrane domains through the two polar residues might

Chapter 1: Review of the Literature have negative influences on enzyme activity and/or chain elongation and/or translocation [103].

During the time frame of this thesis, directed evolution campaign on seHAS heterologously expressed in Bacillus subtilis produced slight improvement on HA production and molecular mass values. It was only when tuaD (encoding UDP-glucose dehydrogenase) and glmU (encoding acetyl-transferase and UDP-GlcNAc pyrophosphorylase) were overexpressed in combination with the best variant that considerable improvement was achieved (2.8 g/L; 2.6 MDa)[104]. Analysis on the amino acid level suggested that the introduction of positively charges residues facilitated improved HAS-HA interaction leading to increase chain length, while the improvement in production was attributed to perturbations in the β-1,3 and β-1,4 transferase sites [104].

Site-specific scanning mutations involving the C-terminal seHAS portion containing the conserved HA binding motif in tandem (B-X7-B) reiterated once more that HA polymerizing activity is decoupled from HA size control [105]. Lys398, Arg406 and C- terminal deletion variants produced uncoordinated HA titer and chain length response, where the first part of the tandem motif (underlined) 398-KLYSLFTIRNADWGTRKKLL- 417 is involved with HA size control and the second with the rate of polymer synthesis [105]. This demonstrates how HA-HAS interaction can help regulate HA production rate and product molecular size. A follow up molecular dynamics experiment on the C- terminal revealed that (1) R413 (second B in B-X7-B) is crucial for HA synthesis as deletion abolished seHAS function (i.e. catalysis); (2) R406-R413 form an α–helical structure that interact loosely with HA and contribute to the stability of HA-HAS complex (i.e. catalysis); and (3) a three-fold increase in HA size (up to 2.29 MDa) by rational design due to enhanced HA binding affinity suggested that residues 414-417 are involved in HA chain retention making way for longer HA chain production (i.e. translocation)[76].

Chapter 1: Review of the Literature

1.2.1. Class II HA Synthase HAS from the Gram-negative Pasteurella multocida (pmHAS) [EC 2.4.1.212] is a membrane-associated glycosyltransferase (Figure 1-4) comprised of 972 amino acid residues [88]. Since pmHAS has a sequence, topology and mechanistic traits different from Class I HAS, it has been suggested to be a class of its own [88]. A number of key properties distinguish pmHAS from other Class I HAS [69]: 1. pmHAS possesses two glycosyltransferase (GT2) modules that independently transfer GlcA and GlcNAc to the nascent HA polymer; 2. pmHAS is a peripheral membrane protein and lipid-independent; 3. pmHAS adds UDP-sugars to the non-reducing end of the growing HA polymer in a non-processive way.

Figure 1-4: HA polymerization by pmHAS. Class II pmHAS is membrane-associated at the C-terminal end (represented by black rectangle). The enzyme is comprised of two glycosyltransferase sites – one for UDP-GlcA and the other for UDP-GlcNAc. The activated nucleotide sugar monomers are added to the non-reducing end of the growing HA chain in an alternating fashion. The growing HA chain is elongated and translocated by the synthase itself across the cell membrane until its eventual release through a mechanism that is still not fully elucidated. The nonreducing end elongation reaction is specified in the figure [65].

Chapter 1: Review of the Literature

Over time, research on pmHAS has focussed on functional and structural characterization to gain insight into its catalytic activity [106-109]. Deletion of the last 269 C-terminal amino acids still retained the glycosyltransferase activity and transformed pmHAS1-703 into a recombinant soluble cytoplasmic protein. This suggested that the C-terminal domain is involved in membrane association [106]. N- terminal truncation studies revealed that residues 1-117 are not required for functional transferase and polymerizing activity [107]. Mutational analyses showed that pmHAS possesses two independent active sites [106, 108] for GlcNAc-transferase activity (residues 161-267, “Domain A1”) and for GlcA-transferase activity (residues 443-547, “Domain A2”) [107]. Furthermore, it was shown that mutating Asp196, Asp247 or Asp249 to Asn/Lys/Glue impairs GlcNAc-transferase activity, while substitutions in Asp477, Asp527 or Asp529 abolished the GlcA-transferase activity. It was concluded that (1) the DXD and DGS consensus motifs shared among Class I HAS, essential for HA synthase activity, are also applicable to the Class II pmHAS and (2) the first aspartate residue is essential for HA synthesis. The latter is supported by the crystal structure of SpsA protein (a Bacillus glycosyltransferase) complexed with UDP, showing that the DXD motif co-ordinates metal ion binding with the beta phosphate and the ribose moiety of the UDP-sugar [110].

In addition, another sequence motif (WGGED) was shown to be essential for GlcNAc- transferase activity since E369H/D/Q or D370/N/K/E substitutions abolished function [107]. Sugar transferase domain swapping of the N-terminal pmHAS and C-terminal chondroitin synthase and vice versa showed that Domain A1 dictates hexosamine transfer, with residues 225-265 discriminating between UDP-GlcNAc and UDP-GalNAc. This corroborates that Domain A1 confers GlcNAc-transferase activity [107]. Lastly, pmHAS catalyzes the sequential elongation of HA at the non-reducing end of the nascent HA polymer [111] with its β-1,3-glucuronyltransferase and β-1,4-N- acetylglucosaminyltransferase domains working synergistically as one polypeptide [106, 109]. Due to the non-processive mechanism of HA polymerization, it is concluded that beside the two UDP-sugar binding sites, two other HA binding sites exist in pmHAS [65, 106].

Chapter 1: Review of the Literature

Having similarities to cellulose synthase (e.g. addition to non-reducing end, functional monomer), it can be assumed (with reservation) that the mode of action of pmHAS (substrate binding, glycosyl transfer, UDP release, polymer extension and extrusion) can be modeled after cellulose synthase, the mechanism of which has been deciphered by McNamara and colleagues [112]. Despite such efforts and evidences from other GT family members, molecular understanding of the control of HA synthesis is far from completion. Furthermore, the crystal structure of pmHAS has yet to be resolved, which hampers any structural and mechanistic insights into any pmHAS structure-based engineering study.

1.3. Catabolism of Hyaluronan Early works to understand the physicochemical properties of HA have begun in the 1950’s with reports of decreasing viscosity of HA solution due to degradation by UV light [113] and increasing intensity of x-ray irradiation [114]. In 1954, Linker et al. demonstrated the decomposition HA by bacterial hyaluronidase and reported the structural formula of the decomposition products (disaccharides) of HA by analytical values for reducing sugar, uronic acid-hexosamine ration and paper chromatography [115]. Moreover, HA is also chemically degradable. In a degradation kinetics experiment with hydrochloric acid (up to 2.0 M) at 40 °C, 60 °C and 80 °C, it was shown that acid hydrolysis was random and follows a first order kinetics with linear dependence on temperature. Moreover, the authors report that the glycosidic β-1,4 linkage between GlcNAc and GlcA is primarily cleaved in acid hydrolysis [116].

In humans, it has been estimated that approximately 15 grams of HA are present in a 70-kg human body [117], with highest amount of HA are found in connective tissues like umbilical cord, synovial fluid, vitreous body, and skin (approximately 5 grams) [118] and the lowest having been found in blood serum [119, 120]. The turnover of HA has previously been reviewed [120]. Briefly, this process utilizes both the lymphatic pathways and the local metabolism. In skin and joints, local metabolism accounts for

Chapter 1: Review of the Literature

20-30 % of HA turnover with the remainder being removed by the lymphatic pathways. In the bloodstream, about 85-90% are directed to the liver for elimination. The kidneys extract about 10% but only 1-2% are lost in the urine. The half-life of HA varies considerably depending on the tissue type [120]. For example, HA has a half-life of 1 day, 20 days and 70 days in the skin, cartilage and vitreous body, respectively [33]. Receptor proteins like cytoplasmic domain peptide 44 (CD44) [121] and HA receptor for endocytosis (HARE) [122] mediate the endocytosis and degradation of HA, while the lymph vessel-specific receptor (LYVE-1) has been shown to be involved in HA uptake and transport from tissue to the lymph [123].

Hyaluronidases (HAse) are the endo-β-acetyl-hexoaminidases or endo-β- glucuronidases that degrade HA into disaccharides, tetrasaccharides or hexasaccharides [117]. A putative scheme for HA catabolism (Figure1-5) has been proposed by [117] and the degradation of HA by hyaluronidase has been outlined by [124].

Figure 1-5: Catabolism of HA by hyaluronidases, adapted from [117]. HMW HA are first bound extracellularly by HA-specific receptors like CD44 and then hydrolyzed by the hyaluronidase 2 into intermediate size HA (~50 disaccharides). The resulting fragments are then internalized by endocytosis, delivered to endosomes and transported to lysosomes. Further degradation occurs via hyaluronidase 1. HYAL1 degrades the 20 kDa HA fragments at the β-1,4 linkages between N-acetylglucosamine and glucuronic acid into smaller oligosaccharides such that they become substrates for the sequential action of the β-glucuronidase and β-N- acetylglucosaminidase [117]. The β-glucuronidase hydrolyzes the terminal non-reducing glucuronic acid, leaving a non-reducing β-N-acetylglucosamine that can then be hydrolyzed by β-N-acetylglucosaminidase [124, 125]. The sugar monomers are then processed for other metabolic purposes.

Chapter 1: Review of the Literature

1.4. Production of Hyaluronan - State of the Art HA production and extraction methods have evolved over the decades. In 1937, Kendall, Heidelberger and Dawson extracted a polysaccharide from the cultural liquid of the haemolytic streptococcus by precipitation with acetic acid and ethanol [3]. This was further improved by precipitation of HA from cultural liquid of Group A Streptococcus yielding 50-75 mg per liter of culture [126]. In 1965, hyaluronic acid from human umbilical cord was isolated by a phenol extraction and further purified by digesting the isolate with ribonuclease, deoxyribonuclease and trypsin. Resultantly, the white fibrous material was verified to be hyaluronic acid by elemental analysis, chromatography and hyaluronidase digestion [127].

The growing demand for HA over the years demanded improved production technologies. This entails development in extraction technology from animal sources. HA is extracted from rooster combs by tissue homogenization and use of large volumes of organic solvents and multi-step chromatography for purification [128, 129]. Various attempts have been made to optimize HA extraction from rooster combs using aqueous- alcohol mixture and even in water at high temperatures (80-100 °C) followed by subsequent removal of nucleic acid and protein contaminants. An exhaustive list of HA extraction and purification experiments have been beautifully compiled elsewhere [17]. Most of the animal-derived HA have high molecular weight (>1 MDa) and can yield 7.5 mg per gram of tissue [120].

It is of utmost importance to purify HA obtained from animal sources because of concerns regarding the risk of cross-species viral infection and allergic reactions to animal-derived HA (e.g. avian allergy. One successful case study for purified HA was reported by Healon [130]. Briefly, homogenized animal tissues were extracted by ethanol:chloroform and treated several times, followed by three rounds water:chloroform extraction and treating the aqueous fractions with hydrochloric acid to pH 4-5 before adding equal volume of chloroform. Considering the yield of 0.8 g HA per 1 kg of chicken combs, production of pure, non-pyrogenic HA with this method proves to be cumbersome and time- and cost-inefficient. While HA extraction from

Chapter 1: Review of the Literature animal tissues and purification methods have evolved since then, newer microbial production methods have been established.

Currently, production of HA by microbial fermentation are considerably higher compared to animal-derived counterparts. The cumulative research on HA biosynthesis by bacterial and mammalian synthases not only gained some understanding of reaction mechanisms but also provided ways by which HA synthesis can further be tailored for improved production. To start, it has been reported that Streptococcus [67] and Pasteurella [131] species can create a capsule layer composed of hyaluronan. The drawback however is that they are pathogenic to humans and animals. With these, HA is produced during the optimized cultivation process and harvested at the stationary phase when the HA accumulates in the culture liquid. This is followed by subsequent purification methods. Notably, fermentation by the Streptococcus equi ATCC 6580 mutant strain can yield 6-7 g/L of 3.2 MDa HA [132], however, the source is a zoonotic pathogen. To date, genetically modified and generally regarded as safe (GRAS) strains [133] including B. subtilis [68], S. thermophilus [134], C glutamicum [135], L. lactis [136, 137], and P. pastoris [138] are employed as microbial factories for industrial scale production of HA. Many HA productivity and/or HA size improvement studies have been reported.

Metabolic engineering of biosynthetic pathways in HA synthesis have been explored [66, 134, 139, 140] and process engineering of microbial fermentative parameters like oxygen supplementation [66, 141], temperature [142], pH [143], and feeding conditions [66, 144, 145] have also been examined. The standard amongst the GRAS HA producers is the metabolically engineered B. subtilis strain TPG223 co-expressing pmHAS/tuaD– gtaB that produces 6.8 g/L HA with a molecular weight of 4.5 MDa [136]. For such reasons, microbial production of HA can now achieve high product yield and quality. It is noteworthy that this is made possible primarily with metabolic and/or process engineering but the contributing potential of protein engineering in HA production has barely been unraveled.

Chapter 1: Review of the Literature

1.5. Protein Engineering Since the release of “On the origin of species” in 1859, it has been generally accepted within the scientific community that evolution is the process whereby species of organisms arise through a selection of traits conducive to the fitness, survival and reproduction of such species. Occurring over a long period of time, the critical interplay between the evolving trait of such species and external selective influences dictate which will be favoured by natural selection. This is the idea of “survival of the fittest”. In a controlled laboratory setting, the power of evolution is harnessed by the introduction of mutation(s) and recombination on the gene level and selecting for the desired trait [146]. The Darwinian evolution of replicase in a test tube was first reported by Mills et al in the latter part of the 1960’s [147]. They demonstrated that 83% aberration of the original genome lead to a 15-fold improvement in replication efficiency of the enzyme. Since then, laboratory-scale evolution through protein engineering has gradually become a standard procedure to redesign proteins for improved enzymatic properties or to discover new functionalities.

1.5.1. Rational Design vs Directed Evolution Generally, two approaches in protein engineering are applied: rational design and directed evolution. Rational design is a branch of protein engineering whereby molecular structure is paramount for predicting which amino acid residues can be modified to elicit improved desired properties [148, 149]. Molecular knowledge is facilitated by x-ray crystallography or computational modelling. Although structure- function relationship can be established, the structures do not account for potentially influential variables including post-translational modifications, cell localization, intermolecular interactions (solvents, salts, etc.) and intramolecular interactions governed by protein folding and stability [148]. Despite this drawback, rational design eliminates the need to generate vast mutant libraries and consequently minimizes screening efforts. For instance, x-ray structure was used to rationally redesign the active site of the E. coli alkaline phosphatase resulting in a 40-fold enhancement in catalytic activity [150].

Contrary to rational design, directed evolution can be applied to enzymes with little or no prior structural or functional knowledge. Moreover, beneficial substitutions obtained

Chapter 1: Review of the Literature with this approach may not necessarily arise from the active site of the enzyme [151]. With directed evolution, mutations are introduced to the gene of interest, selection pressure is established according to the final desired trait, improved variants are selected according to the customized specifications and the mutations of the improved variant(s) are analyzed on the gene and amino acid level. This can then be used as the parent gene for the next cycle of directed evolution until the desired trait or improvement has been achieved. In this matter, “you get what you screen for”, which is the generally accepted First Law of directed evolution [152]. This emphasizes the importance of establishing the selection criteria towards the tailored enzyme functionality.

1.5.2. Diversity Generation Methods Since mutations are introduced on the genetic level, one of the important aspects of a directed evolution campaign is the generation of molecular diversity [153]. For this reason, various methods have been and continue to be established. DNA breeding or gene shuffling is one of the classical ways to generate genetic diversity. Parental DNA (multigene) of related and identical set of sequences are fragmented, shuffled and reassembled by self-priming polymerase reaction to produce chimeric sequence [154]. For example, homologous β-lactamase genes from four types of bacteria were shuffled to produce a recombinant enzyme with 270-fold catalytic improvement compared to the parental wild type gene [155]. This technology was applied on the genomic level as genome shuffling pertaining to the natural evolution of prokaryotic organisms and as exon shuffling for eukaryotic protein evolution. Resultantly, biomedical advances came to fruition with these technologies [148].

Arguably, the most familiar method to introduce random mutation in a gene is the error prone PCR method [156]. This method capitalizes on the high error rate in nucleotide incorporation by the Thermus aquaticus (Taq) polymerase [157, 158]. The Taq p0lymerase is especially useful due to its lack for 3’-5’ proofreading activity and high incorporation error rate of 2 x 10-4 mutations per base pair per duplication [159]. Introduction of mutations is still insufficient for gene randomization and is therefore further increased (1-20 nucleotides per kb) by the introduction of MnCl2, varying dNTPs and increasing the amplification cycles [158, 160]. The method casting epPCR (cePCR)

Chapter 1: Review of the Literature was recently developed using high mutation load PCR to improve the number of identified beneficial positions in a given library [161]. Due to poor mutational load and the inherent mutational bias of the Taq polymerase, the standard epPCR can only achieve 34% of natural diversity on the protein level [162, 163]. These limitations can be circumvented by performing the Sequence Saturation Method (SeSaM) instead to generate random mutations at universal sites in the gene caused by the promiscuous base-pairing of universal bases [164]. Another recombination-based method is Staggered Extension Process (StEP). Briefly, the template is primed followed by cycles of denaturation and abbreviated extensions creating recombination cassettes until full- length genes are formed [165]. Similarly, the phosphorothioate-based DNA recombination (PTRec) method is an enzyme-free method based on phosphorothioate- based ligase independent cloning (PLICing) [166]. In this method, DNA fragments are amplified using phosphorothioated primers and subsequently the complementary overhangs are hybridized to generate chimeric DNA constructs. Moreover, this technique can be used in combination with gene- and exon- shuffling [167].

With regards to focussed mutagenesis, Site-Directed Mutagenesis (SDM) [168-170] and Site-Saturation Mutagenesis (SSM) [171] are the most popular method of choice for introduction of point mutations or for randomization of amino acids at one position made possible by the mutagenic primers, respectively. To take this further, the Iterative Saturation Mutagenesis (ISM) focuses on rationally chosen sites in an enzyme and performing iterative saturation cycles to maximize the probability of finding cooperative effects of newly introduced mutations [172]. Similar to ISM is the semi-rational combinatorial active site-saturation test (CAST) method. Based on the 3-dimensional structure of an enzyme, amino acids in close proximity to the main site of interest are randomized simultaneously allowing for possibility of cooperative effects [173]. Finally, the OmniChange multisite-saturation mutagenesis method was developed to allow for the simultaneous saturation of five independent codons using phosphorothioate-based chemistry [174]. This method not only addresses issues with “localizable” properties of the enzyme, but also facilitates the discovery of possible cooperative effects between amino acid substitutions in one run.

Chapter 1: Review of the Literature

Most recently, the four-phase protein engineering strategy Knowledge-gaining directed evolution (KnowVolution) was proposed to address minimal screening efforts while gaining molecular understanding of the improved enzymatic properties [175]. In Phase I (Identification), a directed evolution experiment using standard diversity generation methods is performed to identify variants with potentially beneficial positions that enhance enzymatic properties. In Phase II (Determination), the prospective positions are subjected to site-saturation mutagenesis and screening to gain insight into factors (charge, size, etc.) that give rise to property improvement. In Phase III (Selection), structural inspection (if possible) is performed to determine the proximity and interaction capability of the amino acid substitutions that may be considered for recombination. In Phase IV (Recombination), the chosen amino acid substitutions are recombined individually or simultaneously to generate an improved enzyme variant with the desired property.

1.5.3. Screening and Selection With an array of available genetic manipulation strategies, the main limitation of directed evolution campaigns is the sensitivity towards the desired properties of interest and the selection capacity of the screening system [146, 151]. It is also desirable that the introduction of mutations only affects enzymatic function and does not interfere with other cellular activities and cell viability. For this reason, directed evolution campaigns using emulsions for fluorescence-activated cell sorting have been developed. In an elegant experiment, ultrahigh throughput cell-free compartmentalization platform was used for directed cellulase evolution which resulted in a variant with 13.3-fold improvement in specific activity over wild type [176]. These ultrahigh throughput screening methodologies not only offers acceptable coverage of the generated protein sequence space (up to 107 events per hour) but also reduces cost, time and screening efforts. However, screening with standard microtiter plates still continues to be implemented in most directed evolution campaigns. The screening library size also presents practical limitations that perhaps will never address the theoretical sequence space. For example, a protein with only 50 amino acid residues has a theoretical sequence space of (20)50 or 1.124 * 1065 combinations. The HA synthase from Streptococcus equi subsp. zooepidemicus, with 417 amino acid residues, has an even

Chapter 1: Review of the Literature more astronomical sequence space of (20)417. Overall, a good screening system must not only cater to the desired enzyme property but must offer considerable coverage of the generated sequence space.

1.6. Purpose of the Study The general purpose of this investigation is to apply protein engineering principles towards improved production of hyaluronic acid by Class I seHAS and Class II pmHAS. Sufficient foundational knowledge about glycosyltransferases, more specifically HA synthases, exist to propel the research forward. However, major questions remain unanswered. For instance, the exact mechanism for HA chain elongation, termination and extrusion remain unsolved. This work not only aims to shed light into the complexity of HA synthases but also to unravel unreported structural and mechanistic insights about HA synthases. HA is a truly remarkable biopolymer with boundless applications. Gaining an understanding of the control of HA polymerization by these biological machineries on the molecular level will enable scientists to tune the production of HA according to their desired properties and application. In this work, protein engineering and computational modeling were employed complementarily to dissect both Class I seHAS and Class II pmHAS, not only having industrial relevance in mind, but also to contribute to the existing knowledge of HAS biology.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

2. Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer “Fall seven times and stand up eight” – Japanese Proverb

2.1. Project Objective HAS from Streptococcus equi subsp. zooepidemicus was used in this project to improve the production of HA. This work was part of the collaboration funded by the Deutsche Bundesstiftung Umwelt in an attempt to synthesize HA in a more economical and sustainable way. This work was focused on the “Engineering of Hyaluronic Acid Synthase Towards the Production of Defined Hyaluronic Acid Polymers for Biomedical Applications”. To this end, seHAS was subjected to a series of rational design and directed evolution campaigns to generate improved variants that produce HMW HA with improved titer. The microbial host most suitable for HA production and screening was determined. A high-throughput screening platform was established, conducive to the selection pressure or the final improved property. In this work, seHAS-WT was used as reference for improvement in enzymatic properties, while variants were screened to some degree with regards to HA titer, but primarily for chain length. The generation of a seHAS variant with improved polymerizing capability (HA chain length >2.5 MDa) was the ultimate goal. Concurrently, in the absence of crystal structure, a homology model of seHAS was generated to gain insight into the enzyme structure-function relationship on the molecular level. The experimental overview is presented in Figure 2-1.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-1: Overview of the strategy for engineering of seHAS. The most suitable HA production host was selected between E. coli, B. subtilis and S. cerevisiae. The screening system for improved seHAS polymerizing activity (more and longer HA polymers compared to wild type) had to be establish. Combined efforts of rational design (based on literature) and random mutagenesis (epPCR) were implemented to generate libraries of seHAS variants. Finally, a homology model was generated to understand the improved properties on the molecular level.

2.2. Materials and Methods

2.2.1. Generation of seHAS Constructs The first phase of the seHAS engineering was the determination of the appropriate microbial host for HAS expression and HA production. In this work, three host systems were initially employed including Escherichia coli, Bacillus subtilis and Saccharomyces cerevisiae. The synthetic non-codon optimized seHAS gene (EC 2.4.1.212; NCBI Reference Sequence: NC_012470.1; Invitrogen, Life Technologies Karlsruhe, Germany) was subcloned into the respective vectors.

2.2.1.1. Preparation of pET-22b(+)-seHAS for Expression in Escherichia coli The preparation of the construct pET-22b(+)-seHAS using classical ligation- dependent cloning was a three-part process - the introduction of the NdeI restriction

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer site upstream of the synthetic seHAS gene in the pMK-RQ vector, the double NdeI- EcoRI double digest of the gene insert and pET-22b(+) vector backbone and ligation with T4 ligase.

The NdeI restriction site was introduced in the plasmid pMK-RQ-seHAS for ligation- dependent cloning. The recipient pET-22b(+)-vector already has both NdeI and EcoRI sites. The NdeI restriction site was introduced by encoding the sequence into the forward primer (Appendix Table A-1). The PCR was comprised of 0.2 mM dNTPs, 40 ng pMK-RQ-seHAS template, 0.4 µM each of the forward and reverse primer, 1 U PfuS polymerase in 1 X PfuS buffer. The PCR was performed as follows: initial denaturation at 94 °C for 2 min, then 30 cycles of denaturation at 94 °C for 15 s, annealing at 55 °C for 30 s and extension at 72 °C for 90 s. A final extension at 72 °C for 4 min completed the DNA amplification and the products were cooled to 8 °C. The amplicons were purified as prescribed (Qiagen PCR purification kit, Hilden, Germany) and subjected to double restriction digest (Components in 50 µL reaction: 5 µL 10 X CutSmart

Buffer, 2 µg pMK-RQ-seHAS, 2 µL EcoRI, 2 µL NdeI and ddH2O up to 50 µL, Conditions: 37 °C, 2 h).

The pET-22b(+) empty vector was extracted from overnight E. coli culture by standard plasmid preparation (NucleoSpin® Plasmid Kit, Macherey-Nagel, Düren, Germany). The DNA was subjected to double digestion (Components in 50 µL reaction: 5 µL 10 X

CutSmart Buffer, 5 µg pMK-RQ-seHAS, 5 µL EcoRI, 5 µL NdeI and ddH2O up to 50 µL, Conditions: 37 °C, 5 h, no shaking in the Eppendorf Thermomixer).

The digested products were resolved in 0.8 % agarose gel at 100 V for 30 min. The DNA fragments (⁓1.2 kb for the seHAS gene and ⁓5.5 kb for the pET-22b(+) vector) were extracted from the agarose gel using the gel extraction kit (Roche Life Science, Indianapolis, United States), eluted with 15 µL sterile Milli-Q water (Millipore, Billerica, USA) and quantified using the NanoDrop photometer 1000 (NanoDrop Technologies, Wilmington, DE, USA).

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer The purified DNA molecules were ligated (16 °C, 16 h) at a 5:1 insert-vector mole ratio consisting of 1 µL pET-22b(+) plasmid, 2.5 µL seHAS insert, 1 µL T4 DNA ligase, 1 µL

10 X T4 DNA ligase buffer and ddH2O up to 10 µL.

Five microliters of the recombinant pET-22b(+)-seHAS plasmid were transformed by heat-shock into chemically competent E. coli DH5α (Stratagene, La Jolla, CA, USA).

One hundred microliters were plated on LBAMP (100 µg/mL) agar plates and cells were grown overnight at 37 °C. Colonies were selected for colony PCR to verify the ligation of the seHAS gene. The colony PCR consisted of ½ colony, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 1 U Taq polymerase (in house) in 1 X Taq buffer. PCR was performed as follows: initial denaturation at 94 °C for 3 min, followed by 25 cycles of denaturation at 94 °C for 30 s, annealing at 56 °C for 30 s and extension at 72 °C for 90 s. A final extension at 72 °C for 10 min completed the DNA amplification and the products were cooled to 8 °C. The insert gene PCR was verified by resolving in 0.8 % agarose gel at 100 V for 30 min and confirmed by sequencing.

The recombinant pET-22b(+)-seHAS plasmid was transformed into chemically competent E. coli BL21 GOLD (DE3) cells (Agilent technologies, Santa Clara, USA): Approximately 20 ng DNA was added to 50 µL competent cells and incubated on ice for 30 min. DNA uptake by E. coli cells was facilitated by heat-shock treatment (42 °C, 45 s), followed by a 2-min incubation on ice. Cell recovery was achieved by addition of 900 µL pre-warmed SOC medium followed by incubation (37 °C, 1 h, 200 rpm). The regenerated cells were plated on LBAMP agar plates and incubated overnight at 37 °C.

Colonies were grown in LBAMP medium and glycerol stocks were prepared by adding glycerol to 30 % final concentration.

2.2.1.2. Preparation of pHY300PLK-seHAS for Expression in Bacillus subtilis The preparation of the construct pHY300PLK-seHAS using classical ligation- dependent cloning was a three-part process. The NotI restriction site was introduced downstream of the promoter region of the pHY300PLK (vector) and upstream of the seHAS gene in the pMK-RQ plasmid. Two separate PCRs were performed to produce the vector and insert with the NotI site. The other cloning site is XmaI, which is

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer inherent in the pHY300PLK vector but not in the pMK-RQ. The two DNA fragments were ligated with T4 ligase and transformed into E. coli DH5α prior to sequence verification. The primers used in this process are listed in Appendix Table A-1.

The seHAS gene was amplified using a forward primer with the NotI sequence and a reverse primer with the XmaI sequence. The PCR consisted of 40 ng of pMK-RQ- seHAS, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 1 U PfuS polymerase in 1 X PfuS buffer. PCR was performed as follows: initial denaturation at 94 °C for 2 min, then 30 cycles of denaturation at 94 °C for 15 s, annealing at 55 °C for 30 s and extension at 72 °C for 90 s. A final extension at 72 °C for 4 min completed the DNA amplification and the products were cooled to 8 °C. The insert gene PCR was verified by electrophoresis at 100 V for 30 min in 0.8 % agarose gel.

The vector backbone was amplified using forward and reverse primers with the integrated NotI sequence. The PCR consisted of 40 ng of pHY300PLK plasmid, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 5 U Phusion HF polymerase (New England Biolabs, Ipswich, MA, United States) in 1 X PfuS buffer. PCR was performed as follows: initial denaturation at 98 °C for 30 s, then 25 cycles of denaturation at 98 °C for 15 s, annealing at 52 °C for 15 s and extension at 72 °C for 3 min. A final extension at 72 °C for 6 min completed the DNA amplification and the products were cooled to 8 °C. The vector DNA was verified by electrophoresis at 100 V for 30 min in 0.8 % agarose gel.

The insert and vector amplicons were extracted from agarose gel using a kit (Roche Life Science, Indianapolis, United States), eluted with 30 µL of sterile ddH2O and quantified using the NanoDrop photometer 1000 (NanoDrop Technologies, Wilmington, DE, USA).

The gel extracted insert and vector DNA were separately subjected to XmaI and NotI double digestion consisting of 5 µL 10 X CutSmart Buffer, 3 µg respective DNA, 3 µL XmaI, 3 µL NotI and ddH2O up to 50 µL. The DNA was digested (37 °C, 3 h, 0 rpm in the Eppendorf Thermomixer), purified using a PCR clean-up kit (Macherey-Nagel,

Düren, Germany) and eluted with 15 µL ddH2O.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer The purified DNA were ligated (16 °C, 16 h) at a 5:1 insert-vector mole ratio consisting of 1 µL vector, 0.91 µL insert, 1 µL T4 DNA ligase, 1 µL 10 X T4 DNA ligase buffer and ddH2O up to 10 µL.

Five microliters of the recombinant pHY300PLK-seHAS plasmid were transformed by heat-shock into E. coli DH5α. Transformants (100 µL) were plated on LBAMP agar plates and cells were grown overnight at 37 °C.

Six colonies from the ligation plate plus two colonies from the relegation negative control plate were selected for colony PCR. The colony PCR consisted of ½ colony, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 1 U Taq polymerase (in house) in 1 X Taq buffer. PCR was performed as follows: initial denaturation at 94 °C for 3 min, followed by 25 cycles of denaturation at 94 °C for 30 s, annealing at 56 °C for 30 s and extension at 72 °C for 90 s. A final extension at 72 °C for 10 min completed the DNA amplification and the products were cooled to 8 °C. The insert gene PCR was verified by 0.8 % agarose gel electrophoresis at 100 V for 30 min. The positive clones were determined by agarose gel electrophoresis and confirmed by sequencing.

A two-step site-directed mutagenesis was performed to remove the NotI restriction site from the recombinant pHY300PLK-seHAS plasmid. The PCR comprised of 0.2 mM dNTPs, 20 ng pHY300PLK-seHAS plasmid, 1 U PfuS polymerase in 1 X PfuS buffer. Half of the mixture was supplied with 0.4 µM of the forward primer, while the other half with 0.4 µM of the reverse primer. The first step of the PCR was performed as follows: initial denaturation at 98 °C for 30 s, then three cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 125 s. The PCR mixtures were combined and then partitioned in half for the second step of the PCR as follows: initial denaturation at 98 °C for 30 s, then 20 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 125 s. A final extension at 72 °C for 3 min completed the DNA amplification and the products were cooled to 8 °C. The parental DNA was digested by DpnI overnight at 37 °C with 1 µL of DpnI per 50 µL reaction. The mutant plasmids were purified using the PCR Clean-up kit (Macherey-Nagel, Düren, Germany) and eluted with Milli-Q water (Millipore,

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer Billerica, USA). Only 20 ng of the mutant plasmid were transformed into chemically competent E. coli DH5α cells and grown on LBKAN (50 µg/mL) agar plates (37 °C, 16 h). DNA sequencing of selected transformants verified the removal of the NotI site.

The construct pHY300PLK-seHAS (without NotI site) was transformed into naturally competent B. subtilis DB104 [177]. A B. subtilis DB104 colony (or glycerol stock) was grown (37 °C, 16 h, 250 rpm) in 10 mL SpC minimal culture medium (in 100 mL: 10 mL 10 X T-Base, 1.0 mL 50 % glucose, 1.5 mL 1.2 % MgSO4, 2.0 mL 10 % yeast extract,

2.5 mL 1 % casamino acids). The overnight culture was diluted to starting OD600 of 0.5 in SpC medium and allowed to grow (37 °C, 3 h 10 min, 250 rpm). The culture was diluted at 1:1 ratio with the starvation medium, SpII, (in 100 mL: 10 mL 10 X T-Base,

1.0 mL 50 % glucose, 7.0 mL 1.2 % MgSO4, 1.0 mL 10 % yeast extract, 1.0 mL 1 % casamino acids and 0.5 mL 100 mM CaCl2) and supplemented with histidine to 25 µg/mL. Competence was achieved with further incubation (37 °C, 3 h, 250 rpm) and the cells were immediately used for transformation.

Four microliters of the pHY300PLK-seHAS construct (⁓250 ng/µL) were mixed with 500 µL of competent cells in 1.5 mL Eppendorf tube and incubated (37 °C, 30 min, 250 rpm) to facilitate transformation. LB medium (300 µL) was added for the recovery phase (37 °C, 30 min, 200 rpm). The cells were plated on LBTET (15 µg/mL) agar plates and grown overnight at 37 °C. Cells were cultivated in TBTET (37 °C, 16 h, 220 rpm).

2.2.1.3. Preparation of pYES2-seHAS for Expression in Saccharomyces cerevisiae The preparation of the construct pYES2-seHAS via homologous recombination by the Saccharomyces cerevisiae was a two-part process: the amplification of the seHAS gene and pYES2 vector (Invitrogen, Karlsruhe, Germany) using primers suitable for homologous recombination and transformation into Saccharomyces cerevisiae. The primers used in this process (Appendix Table-A1) were designed to have at least 30 bp overhang complementary to the vector backbone.

The seHAS gene was amplified by standard PCR consisting 20 ng of pMK-RQ -seHAS template plasmid, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 1 U PfuS polymerase in 1 X PfuS buffer. PCR was performed as follows: initial

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer denaturation at 94 °C for 2 min, followed by 28 cycles of denaturation at 94 °C for 15 s, annealing at 55 °C for 30 s and extension at 72 °C for 90 s. A final extension at 72 °C for 4 min completed the DNA amplification and the products were cooled to 8 °C. The seHAS amplicon was verified by electrophoresis (100 V, 30 min) in 0.8 % agarose gel.

The vector backbone PCR consisted of 40 ng of pYES2 plasmid, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 5 U Phusion HF polymerase (New England Biolabs, Ipswich, MA, United States) in 1 X Phusion buffer. PCR was performed as follows: initial denaturation at 98 °C for 30 s, followed by 25 cycles of denaturation at 98 °C for 15 s, annealing at 56 °C for 15 s and extension at 72 °C for 3 min. A final extension at 72 °C for 6 min completed the DNA amplification and the products were cooled to 8 °C. The vector DNA was visualized as previously described.

The parental DNA from the insert and vector amplification were subjected to DpnI digestion (37 °C, 16 h) with 1 µL of DpnI per 50 µL reaction. PCR clean-up was performed (Macherey-Nagel, Düren, Germany) and DNA were eluted with 40 µL Milli- Q water (Millipore, Billerica, USA). The insert and vector were quantified in NanoDrop photometer 1000 (NanoDrop Technologies, Wilmington, DE, USA) prior to transformation.

2.2.1.3.1. Preparation of Competent Saccharomyces cerevisiae Cells Saccharomyces cerevisiae were streaked on YPD-agar (in 500 mL: 10 g peptone, 5 g yeast extract, 10 g glucose and 7.5 g agar) and incubated for 48 h at 30 °C. Pre-culture was prepared by inoculating 4 mL YPD-medium (in 500 mL: 10 g peptone, 5 g yeast extract, 10 g glucose) with a single colony and grown (30 °C, 16 h, 250 rpm). The main culture of 100mL (pre-warmed 30 °C in 500 mL shake flask) was inoculated with the starter culture to a starting OD600 of 0.2. The cells were grown until OD600 of 1.5 was reached. The cells were harvested (room temperature, 5 min, 3000 g in Eppendorf 5810R centrifuge; Eppendorf AG, Hamburg, Germany) and resuspended in 50 mL sterile ddH2O. The cells were once again centrifuged and washed with 25 mL sterile ddH2O. After centrifugation, the cells were resuspended with 1 mL frozen competent cell buffer (FCC: 42.5 g ddH2O, 2.5 g glycerol and 5 g DMSO, filter sterilized using 0.45

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer µm Whatman filter). The cells were partitioned into 50 µL aliquots, slowly frozen and stored in the -80 °C freezer until required.

2.2.1.3.2. Transformation into Saccharomyces cerevisiae Single-strand carrier DNA (2 mg/mL salmon sperm DNA; Thermo Fisher Scientific, Schwerte, Germany) was activated by boiling at 99 °C for 10 min and subsequently cooling in ice. The competent Saccharomyces cerevisiae cells were thawed at 37 °C for 30 s and then centrifuged (room temperature, 2 min, 13000 g in Centrifuge 5424; Eppendorf, Hamburg, Germany). The supernatant was carefully aspirated out and to the cell pellet were added in the following order: 260 µL PEG3350 (50 % w/w), 36 µL 1 M lithium acetate, 50 µL activated carrier DNA and 14 µL of seHAS insert and pYES2 plasmid DNA at 5:1 mole ratio. The cell pellet was resuspended by vortexing and the transformation mixture was incubated (42 °C, 2 h). The mixture was spun down (room temperature, 1 min, 3000 g in Centrifuge 5424) and the pellet was resuspended with 1 mL YPD medium. Cells continued to incubate at 30 °C for 1 h. The cells were centrifuged once more (room temperature, 1 min, 3000 g in Centrifuge 5424) and the pellet was resuspended this time with 400 µL ddH2O. Cells were plated onto selective synthetic complete (uracil dropout) agar plates containing 2% glucose (SC-U) using a spreader and incubated 30 °C for 3-4 days with the plates sealed with parafilm. Successful homologous recombination was confirmed by colony PCR, double restriction enzyme digest and sequencing.

2.2.2. Cell cultivation and Protein Production

2.2.2.1. Cultivation of Escherichia coli Cells Harbouring pET-22b(+)- seHAS (wild type or variant) in Shake Flask E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-seHAS-WT (or variant) were grown as a starter culture in 15 mL TBAMP (100 µg/mL; 37 °C, 24 h, 200 rpm in Multitron II Infors Shaker; Einsbach, Germany). The inoculum was added at an initial

OD600 of 0.05 to 200 mL Terrific Broth (TB; 1.2 % peptone (w/v), 2.4 % yeast extract

(w/v), 0.4 % glycerol (v/v), 0.23 % KH2PO4 (w/v), 1.25 % K2HPO4 (w/v)) medium supplemented with ampicillin (100 µg/mL). Once cell density has reached OD600 0.6 -

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 0.8, protein expression induced with IPTG (1 mM) and lowering the temperature to 30 °C. Cells were grown up to 24 h and harvested by centrifugation (4 °C, 20 min, 11 279 g in Sorvall RC6 Plus; Thermo Fisher Scientific, Schwerte, Germany). The cell pellet was stored at -80 °C until further use.

2.2.2.2. Cultivation of Bacillus subtilis Cells Harbouring pHY300PLK- seHAS in Shake Flask Pre-culture B. subtilis DB104 cells harbouring pHY300PLK-seHAS were grown in

TBTET medium (15 µg/mL; 37 °C, 24 h, 220 rpm in Multitron II Infors Shaker; Einsbach, Germany). A portion was used to prepare glycerol stocks to a final 30 % glycerol. To prepare the main culture, overnight culture was diluted to the initial

OD600 of 0.1 in TBTET medium and the cells were continued to be cultivated (37 °C, 24 h, 220 rpm) to allow for protein production. Cells were harvested by centrifugation (4 °C, 20 min, 11 279 g in Sorvall RC6 Plus; Thermo Fisher Scientific, Schwerte, Germany) and stored at -80 °C prior to in vitro synthesis experiments.

2.2.2.3. Cultivation of Saccharomyces cerevisiae Cells Harbouring pYES2-seHAS in Shake Flask Saccharomyces cerevisiae cells harbouring pYES2-seHAS were grown as starter culture by inoculating 10 mL YPD-medium (in 500 mL: 10 g peptone, 5 g yeast extract, 10 g glucose) with a single colony and grown (30 °C, 16 h, 220 rpm). The main culture of 100mL YPD (pre-warmed 30 °C in 500 mL shake flask) was inoculated with the starter culture to a starting OD600 of 0.2. The main culture was grown (30 °C, 48 h, 220 rpm), harvested by centrifugation (4 °C, 20 min, 11 279 g in Sorvall RC6 Plus; Thermo Fisher Scientific, Schwerte, Germany) and frozen prior to in vitro synthesis experiments.

2.2.2.4. Cultivation of Escherichia coli Cells Harbouring pET-22b(+)- seHAS (Wild Type or Variants) in 96-Deep Well Microtiter Plate E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-seHAS-WT (or variants from respective libraries) were inoculated into the starter 96-well MTP (PS-F-bottom,

Greiner Bio-One, Frickenhausen, Germany) containing 150 µL LBAMP medium (100 µg/mL) and cultivated to saturation (37 °C, 24 h, 900 rpm, 70 % humidity in Multitron II Infors Shaker; Einsbach, Germany). Thirty-five microliters of the pre-culture were

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer used to inoculate the 500 µL TBAMP (100 µg/mL) main culture in 96-deep-well MTP. Cells were cultivated (37 °C, ⁓3 h, 900 rpm, 70 % humidity) until the cell density has reached OD600 0.6 - 0.8. After which, IPTG was added to 1 mM to induce protein expression. Cultivation resumed at a lower temperature (30 °C, 24 h, 900 rpm, 70 % humidity). Cells were pelleted (4 °C, 15 min, 3220 g in Eppendorf 5810R centrifuge; Eppendorf AG, Hamburg, Germany), washed with 500 mL PBS pH 7.4 (137 mM NaCl,

2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4) under these conditions: 24 °C, 10 min, 900 rpm on TiMix Shaker, Edmund Bühler (Hechingen, Germany). Preparations were centrifuged once more (4 °C, 15 min, 3220 g in Eppendorf 5810R centrifuge). The pellets were stored at -80 °C until further use. The remainder of the pre-culture was preserved as master plate by adding glycerol to 30 % and stored at -80 °C.

2.2.3. In vitro HA Biosynthesis

2.2.3.1. Flask Method (E. coli BL21 GOLD (DE3) Harbouring pET- 22b(+)-seHAS Wild Type or Variants) E. coli BL21 GOLD (DE3) expressing seHAS (wild type or variants) were harvested by taking 1.5 mL aliquots. The samples were pelleted (room temperature, 1 min, 11000 g in Centrifuge 5424; Eppendorf, Hamburg, Germany) and the supernatant was discarded. The pellet was resuspended and washed with 750 µL PBS pH 7.4 (137 mM

NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4), centrifuged (room temperature, 1 min, 11000 g in Centrifuge 5424) and resuspended with 500 mL PBS pH 7.4. Cells were disrupted in ice by sonication (30 % amplitude, 10 s ON, 10 s OFF, 2 min in Vibra-Cell Ultrasonicator; Sonics Sonics & Materials, Inc, Newtown, Connecticut, USA). The disrupted cells were supplemented with 500 µL in vitro synthesis cocktail (Final concentrations: 0.4 mM UDP-GlcA, 0.4 mM UDP-GlcNAc, 4 mM MgCl2, 0.4 mM DTT) to start HA synthesis (37 °C, x h, 0 rpm) depending on the nature of the experiment. HA products were resolved in 0.5 % agarose gel at 100 V for 30 min and visualized with Stains-All.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.2.3.2. Flask Method (B. subtilis DB104 Harbouring pHY300PLK- seHAS) B. subtilis cells were prepared similarly to E. coli (Section 2.2.3.1) until the lysis procedure. Washed pellets were resuspended with 750 µL PBS pH 7.4 and supplemented with lysozyme to 0.4 mg/mL to initiate peptidoglycan layer degradation (37 °C, 20 min), followed by sonication in ice (30 % amplitude, 10 s ON, 10 s OFF, 2 min in Vibra-Cell Ultrasonicator). The disrupted cells were supplemented with 500 µL in vitro synthesis cocktail (Final concentrations: 0.4 mM UDP-GlcA, 0.4 mM UDP- GlcNAc, 4 mM MgCl2, 0.4 mM DTT) to start HA synthesis (37 °C, x h, 0 rpm) depending on the nature of the experiment. HA products were resolved in 0.5 % agarose gel at 100 V for 30 min and visualized with Stains-All.

2.2.3.3. Flask Method (S. cerevisiae cells Harbouring pYES2-seHAS) S. cerevisiae cells were prepared for in vitro HA synthesis similar to E. coli. Please refer to section 2.2.3.1.

2.2.3.4. 96-Deep Well Microtiter Plate Method E. coli BL21 GOLD (DE3) cells expressing seHAS-WT (or variants from respective libraries) in 96-deep well MTPs were initially washed once with 500 µL of 0.1 M Tris pH 7.6 (room temperature, 10 min, 900 rpm on TiMix Shaker, Edmund Bühler; Hechingen, Germany) and pelleted (24 °C, 15 min, 3220 g in Eppendorf 5810R). Cells were resuspended again with the 200 µL of 0.1 M Tris pH 7.6 and subjected to three rounds of freeze (-80 °C, 15 min)-thaw (37 °C, 30 min, 200 rpm) cycles and then permeabilized with Polymyxin B sulfate salt (Merck KGaA, Darmstadt, Germany; final 20 µg/mL; 37 °C, 1 h, 200 rpm). The permeabilized cells were centrifuged (4 °C, 20 min, 3220 g in Eppendorf 5810R) and the supernatant was discarded. The remaining cell pellet containing the membrane-bound seHAS was resuspended with 250 µL PBS pH 7.4 (room temperature, 10 min, 1000 rpm on TiMix Shaker) and transferred to a clean V-bottom microtiter plate (Greiner Bio-One, Frickenhausen, Germany). Once again, the cell suspension was centrifuged (24 °C, 15 min, 3220 g in Eppendorf 5810R) and the supernatant was discarded. To the pellet, 150 µL of in vitro synthesis cocktail was added (final concentrations: 0.4 mM UDP-GlcA, 0.4 mM UDP-GlcNAc, 4 mM

MgCl2, 0.4 mM DTT) to allow for HA polymerization (37 °C, 20 h, 0 rpm). The final

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer product was clarified (24 °C, 20 min, 3200 g in Eppendorf 5810R) before HA screening, analysis and/or characterization.

2.2.4. Development of Protocols for Screening Mutant Libraries in 96-Well Microtiter Plate Format

2.2.4.1. CTAB Turbidimetric Assay The CTAB turbidimetric assay in a 96-well format was slightly modified [178]. This method depends on the formation of an insoluble complex between HA and cetyltrimethylammonium bromide, where the turbidity is linearly proportional to the quantity of HA in the system [179]. Briefly, 60 µL of HA (in vitro synthesis products or commercial standard (Hyaluronic Acid Na-salt > 2.O MDa, GfN & Selco, Wald- Michelbach, Germany) were added to 20 µL of 0.1 M phosphate buffer, pH 7.0 and incubated (37 °C, 15 min, 0 rpm). The addition of 160 µL of pre-warmed CTAB reagent (2.5 g in 100 mL 2 % NaOH) to the sample initiated the precipitation reaction. The end-point absorbance was measured within five minutes of the reaction (400 nm, 2- second shake, 5-second settle time) with the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland).

2.2.4.2. Alcian Blue Colorimetric Assay The Alcian Blue colorimetric assay relies on the complexation of the polyanionic hyaluronic acid with the basic dye Alcian Blue 8GX [180]. This is important for determination of molecular binding ratios of Alcian Blue to HA [181]. Unlike CTAB, an inverse relationship between the amount of HA and the absorbance of solution is established. In a microtiter plate, 80 µL of the HA sample (in vitro synthesized or commercial) in 110 µL 3 % acetic acid and 10 µL of the Alcian Blue stock (1.0 g Alcian Blue 8GX (Merck KGaA, Darmstadt, Germany) dissolved in 100 mL 3 % glacial acetic acid, pH 2.5) were thoroughly mixed (room temperature, 30 s, 900 rpm on TiMix Shaker, Edmund Bühler; Hechingen, Germany). Instead of the classical 30-second microwave step, the precipitation was facilitated by heating up the samples to 65 °C for 10 min in the oven. The samples were centrifuged (24 °C, 5 min, 2900 g in Eppendorf 5810R) and equilibrated to room temperature for 2.5 h. Another round of

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer centrifugation was performed and the clarified supernatants were transferred to a clean flat-bottom microtiter plate (Greiner Bio-One, Frickenhausen, Germany). The endpoint absorbance was measured at the wavelength of 540 nm with the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland). A dilution series of standard HA (0-400 µg/mL in PBS buffer pH 7.4; > 2.O MDa, GfN & Selco, Wald- Michelbach, Germany) was performed for HA quantification.

2.2.4.3. 2-AA Fluorescent Labeling of Hyaluronan The fluorescent labeling of HA using 2-anthranilic acid was performed using the GlycoProfileTM 2-AA Labeling Kit (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany). Prior to labeling, the HA samples were completely dehydrated. The labeling solution was prepared just before modifying HA: 150 µL of glacial acetic acid was added to 350 µL DMSO and thoroughly mixed. Only 100 µL of the acetic acid-DMSO solution was added to dissolve 6 mg 2-AA. All were added to dissolve 6 mg of sodium cyanoborohydride.

To label the HA products, 5 µL of the labeling solution was added to the dehydrated HA samples in 96-well MTP (black PS-F-bottom, Greiner Bio-One), mixed (room temperature, 1 min, 1000 rpm on TiMix shaker) and incubated at 65 °C for 3 h in the drying oven protected from light exposure. The mixture was diluted with 50 µL of PBS pH 7.4 (room temperature, 1 min, 1000 rpm on TiMix shaker). The samples were cooled at ambient temperature prior to proceeding with post-labeling analysis. Fluorescence was determined using the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland) using the following parameters: excitation wavelength: 315 nm; emission wavelength: 400 nm; 10 s shaking at 2 mm amplitude; gain: 70; number of flashes: 50). Dilutions were performed if the initial fluorescence measurements exceeded 50 000 RFUs.

2.2.4.4. Anion Exchange Chromatography Features of the manual anion exchange chromatography were adapted from the protocol by [182] using disposable 10 mL polypropylene column (Thermo Fisher Scientific, Schwerte, Germany). The column was filled with 100 µL of 50 % slurry (Source 30Q 30 µm anion beads, GE Healthcare Europe GmbH; Freiburg, Germany).

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer The column was equilibrated with 1 mL of Mobile Phase A (Buffer A1: 10 mM sodium phosphate, 20 mM sodium sulfate pH 7.0). Twenty-five microliters of the HA sample (0.5 mg/mL commercial HA or in vitro synthesis product) were loaded onto the column. Gradient elution was performed by increasing the quantity of Mobile Phase B (10 mM sodium phosphate, 1 M sodium sulfate pH 7.0) to determine the concentration of sodium sulfate that starts eluting the HA products. Eluates were collected at each step and resolved in 0.5 % agarose gel at 100 V for 50 min and stained with Stains-All.

To adapt to the 96-deep well (high throughput) format, the following steps were followed. Each column of the 96-deep well chromatography plate was filled with 300 µL of 50 % the anion resin (Source 30Q 30 µm anion beads, GE Healthcare Europe GmbH; Freiburg, Germany). The columns were equilibrated with 2 mL of Mobile Phase A (Buffer A1:10 mM sodium phosphate, 20 mM sodium sulfate pH 7.0). Three hundred fifty microliters of the in vitro synthesis HA products were loaded onto the columns. The columns were washed with 2 mL of Buffer A1 and HA were eluted with Buffer A4 (10 mM sodium phosphate, 160 mM sodium sulfate pH 7.0). Eluates were collected at each step and resolved in 0.5 % agarose gel at 100 V for 50 min and stained with Stains-All.

2.2.4.5. Agarose Gel Electrophoresis For visual analysis of the HA products, 15 µL HA products were mixed with 3 µL of 6 X DNA gel loading dye (Merck KGaA, Darmstadt, Germany) and applied to 0.5 % agarose gel for electrophoresis at 100 V for 50-60 min. The agarose gel was soaked in 30 % ethanol for 1 h and stained in darkness for at least 8 h with Stains-All (Merck KGaA, Darmstadt, Germany; 0.1 mg/mL in 30 % EtOH). The stain was discarded and replaced with distilled water. The gel was exposed to white light for destaining and scanned with Canon Scan 5600F as TIF. For HA molecular weight determination, either the GeneRuler™ 1 kb DNA Ladder (Fermentas, St. Leon-Rot, Germany) or Select-HA HI and Mega Ladder (Hyalose, amsbio, Frankfurt, Germany) or both were used.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.2.5. Computational Modeling of seHAS Homology modeling of the structure of seHAS was performed using YASARA Structure Version 13.3.26 [183] with the default settings (PSI-BLAST iterations: 3, E value cut-off: 0.5, templates: 5, OligoState: 4). A position-specific scoring matrix (PSSM) was used to score the 19 obtained template structures. The crystal structure of cellulose synthase from Rhodobacter sphaeroides (PDB ID: 4HG6; [74]) was selected as a template for construction of the seHAS homology model. The obtained model is a monomer with 345 of 419 target residues (82.3 %) aligned to template residues. Among these aligned residues, the sequence identity is 20.0 % and the sequence similarity is 40.3 % (BLOSUM62 score is > 0). The constructed homology model has a Z score of -2.253.

2.2.6. Cellular Localization of seHAS using Fluorescence Microscopy The goal of this experiment was to verify seHAS expression and identify cellular localization. The EstA cell surface display detection system [184] was employed. The E-tag epitope sequence (13 aa: G-A-P-V-P-Y-P-D-P-L-E-P-R) was incorporated into the flexible outer membrane loop of seHAS determined from the seHAS computational model. Phe28-Gly29 were chosen as the insertion site for the E-tag epitope. The primers were designed accordingly (see Appendix Table A-1).

2.2.6.1. Introduction of the E-tag Sequence by PCR To introduce the E-tag sequence overlap PCR was performed. The PCR consisted of 50 ng of pET-22b(+)-seHAS plasmid, 0.4 µM of the forward and reverse primer, 0.2 mM dNTPs and 1 U PfuS polymerase in 1 X PfuS buffer. PCR was performed as follows: initial denaturation at 94 °C for 2 min, followed by 20 cycles of denaturation at 94 °C for 15 s, annealing at 50-58 °C for 30 s and extension at 72 °C for 3 min 30 s. A final extension at 72 °C for 10 min completed the DNA amplification and the products were cooled to 8 °C. The amplicons were visualized by 0.8 % agarose gel electrophoresis at 100 V for 30 min. The PCR products were pooled, resolved in 0.8 % agarose gel and extracted using the gel extraction kit (Roche Life Science, Indianapolis, United States).

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.2.6.2. Heat-shock Transformation into Chemically Competent E. coli BL21 GOLD (DE3) cells Approximately 10 µL (1.5 mg) DNA was added to 50 µL competent E. coli BL21 GOLD (DE3) cells (Agilent technologies, Santa Clara, USA) and incubated on ice for 30 min. DNA uptake by E. coli cells was facilitated by heat-shock treatment (42 °C, 45 s), followed by 2-min incubation on ice. Cell recovery was achieved by addition of 900 µL pre-warmed SOC medium followed by incubation (37 °C, 1 h, 200 rpm). Before plating, the cells were gently centrifuged (room temperature, 1 min, 2000 g in Centrifuge 5424; Eppendorf, Hamburg, Germany) and 600 µL supernatant was discarded. The remaining cells were resuspended and transferred onto LBAMP (100

µg/mL) agar plates and incubated overnight at 37 °C. Cells were cultivated in LBAMP medium, DNA were extracted (NucleoSpin® Plasmid Kit; Macherey-Nagel, Düren, Germany) and verified by sequencing.

2.2.6.3. Antibody Staining Protocol Notes: This protocol was provided by Dr. Kristin Rübsam. PBS pH 7.4 with 0.2 % BSA was used and all steps were performed on ice. As soon as the antibody was added, the samples were kept away from light.

Cell suspension of 300 µL (pET22b±푠푒퐻퐴푆±퐸푇퐴퐺) post 1mM IPTG induction (37 °C,

20 h) in TBAMP were pelleted (4 °C, 2 min, 15871 g in Centrifuge 5424; Eppendorf, Hamburg, Germany). The pellet was resuspended with 300 µL PBS and centrifuged (4 °C, 2 min, 15871 g in Centrifuge 5424) to discard the supernatant. The pellet was resuspended with 10 µL PBS containing 1 % anti E-epitope antibody labelled with FITC (fluorescein isothiocyanate; Novus Biologicals, Wiesbaden, Germany). After 10 min incubation on ice and in darkness, 500 µL PBS buffer was added to wash off unbound antibodies. The resuspension was centrifuged for the last time (4 °C, 2 min, 15871 g in Centrifuge 5424) and the supernatant was discarded. Finally, the pellet was resuspended with 50 µL PBS and 2 µL was mounted on glass slides. Dilutions were performed as necessary to keep approximately the same cell count.

The samples were analyzed using the Leica Application Suite confocal microscope (courtesy of Dr. Kristin Rübsam). Images were taken with an excitation wavelength of

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 488 nm, an emission wavelength of 519 nm, 10 % intensity, 63 x optical magnification, 2.1 x digital zoom and a gain of 981. An overlay of transmitted light and fluorescent images was used to visualize the fluorescent cells.

2.2.7. Diversity Generation

2.2.7.1. Site-Saturation Mutagenesis of Conserved Cysteines (C226, C262 and C281) Non-exhaustive substitutions on the conserved cysteines of the Class I streptococcal HAS first showed the effects of sequence perturbations on HA titer and chain length specificity [101]. It was therefore of interest to saturate the respective positions of the conserved cysteines to generate improved seHAS variants.

NNK primers were designed (see Appendix Table A-1) and ordered from Eurofins Genomics. A two-step site-saturation mutagenesis was performed to introduce mutations at positions 226, 262 and 281. The PCR comprised of 0.2 mM dNTPs, 20 ng pET-22b(+)-seHAS, 1 U PfuS polymerase in 1 X PfuS buffer. Half of the mixture was supplied with 0.4 µM of the forward primer, while the other half with 0.4 µM of the reverse primer. The first step of the PCR was performed as follows: initial denaturation at 98 °C for 30 s, then five cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 2 min 20 s. The PCR mixtures were combined and then partitioned in half for the second step of the PCR as follows: initial denaturation at 98 °C for 30 s, then 18 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 2 min 20 s. A final extension at 72 °C for 10 min completed the DNA amplification and the products were cooled to 8 °C. The parental DNA was digested by DpnI (37 °C, 16 h) with 1 µL of DpnI per 50 µL reaction. The mutant plasmids were purified using the PCR clean-up kit (Macherey-

Nagel, Düren, Germany) and eluted with sterile ddH2O. One hundred nanogram each of the mutant plasmids were then transformed separately into chemically competent

E. coli BL21 GOLD (DE3) cells and plated on LBAMP (100 µg/mL) agar plates. Cells were grown overnight and colonies were picked for deep well MTP-based experiments.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.2.7.2. Site-Saturation Mutagenesis of Polar Intramembrane Residues Lys48 and Glu327 Mutations of two conserved intramembrane polar residues of seHAS could affect HA product size [103]. It was therefore of interest to saturate Lys48 from membrane domain 2 and Glu327 from membrane domain 4 of seHAS to discover possible beneficial substitutions that improve HA chain length.

NNK primers were designed (see Appendix Table A-1) and ordered from Eurofins Genomics. A two-step site-saturation mutagenesis was performed to randomize positions 48 and 327, respectively. The PCR consisted of 0.2 mM dNTPs, 20 ng pET- 22b(+)-seHAS, 1 U PfuS polymerase in 1 X PfuS buffer. Half of the mixture was supplied with 0.4 µM of the forward primer, while the other half with 0.4 µM of the reverse primer. The first step of the PCR was performed as follows: initial denaturation at 98 °C for 30 s, then seven cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 2 min 30 s. The PCR mixtures were combined and then partitioned in half for the second step of the PCR as follows: initial denaturation at 98 °C for 30 s, then 18 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 2 min 30 s. A final extension at 72 °C for 10 min completed the DNA amplification and the products were cooled to 8 °C. The parental DNA was digested by DpnI (37 °C, 16 h) with 1 µL of DpnI per 50 µL reaction. The mutant plasmids were purified using the PCR clean-up kit (Macherey-

Nagel, Düren, Germany) and eluted with sterile ddH2O. One hundred nanogram each of the mutant plasmids were then transformed separately into chemically competent

E. coli BL21 GOLD (DE3) cells and plated on LBAMP (100 µg/mL) agar plates. Cells were grown overnight and colonies were picked for deep well MTP-based experiments.

2.2.7.3. epPCR seHAS Library Generation A random mutagenesis library of pmHAS was generated using the error-prone PCR approach [185] in combination with megaprimer PCR of whole plasmids (MEGAWHOP) cloning method [186]. The optimal manganese concentration was determined by adding increasing amounts of MnCl2 (0.05, 0.10, 0.20, 0.25 and 0.50 mM) to each of the five separate insert PCRs (0.2 mM dNTPs, 20 ng pET-22b(+)- seHAS plasmid, 0.4 µM forward primer, 0.4 µM reverse primer, 1 U Taq polymerase in

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 1 x Taq polymerase buffer, in-house). PCR was performed using the following cycle conditions: initial denaturation 94 °C for 2 min, 25 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s, elongation at 68 °C for 2 min, final elongation was performed at 68 °C for 10 min and terminated at 8 °C. The amplicons were purified (NucleoSpin® Gel and PCR Clean-up kit; Macherey-Nagel, Düren, Germany) and eluted with sterile ddH2O.

To introduce mutations in the seHAS gene via MEGAWHOP, the amplicons were used as megaprimers for the subsequent PCR. The individual reactions consisted of 0.3 mM dNTPs, 60 ng pET-22b(+)-seHAS plasmid, 500 ng purified amplicons and 1 U PhuS polymerase in 1 x PhuS buffer. PCR was initiated by the proofreading of megaprimers at 72 °C for 5 min and initial denaturation 98 °C for 1 min 30 s. Subsequently, 24 cycles were performed: denaturation at 98 °C for 45 s, annealing at 55 °C for 45 s, elongation at 72 °C for 4 min, final elongation was performed at 72 °C for 10 min and terminated at 8 °C. The parental DNA was digested with 20 U DpnI overnight at 37 °C. The amplicons were purified (NucleoSpin® Gel and PCR Clean-up kit, Macherey-

Nagel, Düren, Germany) and eluted with sterile ddH2O.

The DpnI-digested products (10 µL) were transformed into chemically competent E. coli BL21 GOLD (DE3) according to standard protocol and plated on LBAMP (100 µg/mL) agar plates. The cells were cultivated overnight at 37 °C. Colonies were picked (87 epPCR variants, 6 pET-22b(+)-seHAS-WT and 3 pET-22b(+) empty vector) to generate one 96-well MTP library per manganese concentration. Cells were cultivated, proteins were produced and in vitro synthesis of HA was performed. HA products were resolved in 0.5 % agarose gel and stained with Stains-All. The percentage of active variants was calculated based on the number of HA bands (active seHAS variant) observed divided by 87. With the optimal concentration of 0.10 mM, a random library of 1392 epPCR variants (16 microtiter plates) was generated.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

2.3. Results

This section is divided into four major parts. First, the selection of the most suitable microbial HA production system for the directed evolution campaign is outlined. Second, the establishment of the screening system in 96-microtiter plate format is provided. Third, the results of the rational evolution experiments (SSM) are provided. Lastly, the outcome of the directed evolution of seHAS is presented.

2.3.1. Selection of Most Suitable Microbial HA Production System for the seHAS Engineering Campaign

2.3.1.1. Escherichia coli The successful ligation-dependent subcloning of the seHAS gene into the pET-22b(+) vector is evident in Figure 2-2A,B and was verified by sequencing. Heterologous seHAS expression by E. coli BL21 GOLD (DE3) cells had little effect on the cells according to the congruent growth curves in Figure 2-2C. seHAS overexpression, however, could not be ascertained from the SDS gel (Figure 2-2D). Regarding HA synthesis, the empty vector control did not produce HA as expected, while E. coli cells expressing seHAS displayed time-dependent HA polymerization (Figure 2-3). Polydispersed HA products (blue smear) at maximum 1.5 MDa could already be synthesized within 15 minutes of in vitro synthesis and chain elongation continued proportionally within the allotted synthesis duration. HA synthesis was repeated for up to 4 hours (Figure 2-4). HA production was time-dependent and plateaud starting at t = 3 hours since the longest polymers could not be further extended thereafter. Moreover, 30 minutes after initiation of HA synthesis, seHAS was capable of producing HA up to 2.4 MDa in molecular weight. To verify that HA was the polymer product, digestion by bovine testis hyaluronidase was performed overnight (Figure 2- 5). The fingerprint blue smears of the HA products and the commercial HA (- lanes) are abrogated by the presence of hyaluronidase (+ lanes). Ultimately, E. coli cells expressing seHAS were demonstrated to synthesize HA in a time-dependent fashion, however, protein expression could not be shown.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

A B

C D

Figure 2-2: Construction of pET-22b(+)-seHAS for expression in Escherichia coli BL21 GOLD (DE3) cells. (A) NdeI-EcoRI double digest of the seHAS insert gene (right image, ~1.3 kb) from parental plasmid pMK-RQ seHAS and pET-22b(+) vector backbone (left image, ~5.5kb). (B) Confirmation of successful T4 ligation of the seHAS gene into the pET- 22b(+) vector by colony PCR. M is the GeneRuler™ 1 kb DNA Ladder (Fermentas, St. Leon- Rot, Germany), V is the linearized pET-22b(+)-seHAS. (C) Exemplar cell cultivation time course of E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-EV and pET-22b(+)-seHAS in TBAMP starting from 37 °C and reduced to 30 °C after IPTG induction to 1 mM at OD600 of 0.8. (D) Exemplar seHAS expression profile from 0, 3, 24 and 48 h after IPTG induction, under aforementioned cultivation conditions. The PageRuler Prestained Protein ladder 10-180 kDa (Thermo Fisher Scientific, Schwerte, Germany) was used. The expected size for seHAS is 48 kDa.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-3: HA biosynthesis by pET-22b(+)-seHAS in Escherichia coli BL21 GOLD (DE3) cells. E. coli cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)- seHAS-WT were grown in TBAMP (37 °C, 200 rpm) until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 200 rpm, 24 h). Cells were pelleted from 1.5 mL aliquots, sonicated and supplemented with in vitro synthesis cocktail. In vitro synthesis products were collected during the two-hour synthesis duration. Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Nucleic acids stain purple, while HA stains blue. Both GeneRuler™ 1 kb DNA ladder (purple) and Select-HA Hi Ladder (blue) were used as standards.

Figure 2-4: Verifying time-dependent HA biosynthesis by pET-22b(+)-seHAS in Escherichia coli BL21 GOLD (DE3) cells. E. coli cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)-seHAS-WT were grown in TBAMP (37 °C, 200 rpm) until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 200 rpm, 24 h). Cells were pelleted from 1.5 mL aliquots, sonicated and supplemented with in vitro synthesis cocktail. In vitro synthesis products were collected during the four-hour synthesis duration. Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Nucleic acids stain purple, while HA stains blue. The GeneRuler™ 1 kb DNA ladder, Select-HA Hi Ladder and 2.4 MDa HA standard were used as standards.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-5: Proving HA biosynthesis by seHAS by hyaluronidase digestion. E. coli cells harbouring pET-22b(+)-seHAS- WT were processed to synthesize HA as previously described. In vitro synthesis HA products were digested with hyaluronidase (1 U/µL, bovine testis HYAL; Sigma- Aldrich, MO, USA) at 1/10 the reaction volume (37 °C, 18 h). Digestion of 2 MDa commercial HA (GfN & Selco, Wald- Michelbach, Germany) was also used for control. Digested and undigested HA samples were resolved in 0.5 % agarose gel and stained with Stains-All.

2.3.1.2. Bacillus subtilis

The successful ligation-dependent subcloning of the seHAS gene into the pHY300PLK vector is evident in Figure 2-6A,B and was verified by sequencing. Heterologous seHAS expression in B. subtilis DB104 cells could not be verified by the SDS gel (Figure 2-6D), however, cell growth reached the lag phase within 21 hours of cultivation with an optical density of 5 (Figure 2-6C). With respect to HA synthesis, three conditions were tested: synthesis with complete in vitro synthesis cocktail, synthesis with only UDP-GlcNAc and synthesis with only UDP-GlcA. As seen in Figure 2-7, HA polymerization was not time-dependent. The HA products were consistently LMW throughout the time course and never exceeded 0.5 MDa in size. Moreover, HA synthesis was possible despite lacking either of the two nucleotide sugar precursors. This observation was further verified by another experiment with varying durations of protein expression and HA synthesis (Figure 2-8). HA production was limited to only low molecular weight (LMW) HA (<0.5 MDa), however, time-dependence with respect to quantity was observed. As the synthesis time increased, the amount of HA (thickness and darkness of the smear) increased accordingly, peaking at the 3-hour timepoint. No sign of HA production was observed at the very beginning of synthesis (t = 0 h), but occurs thereafter with or without supplying the precursors. Taken together,

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer overexpression of seHAS in B. subtilis DB104 could not be confirmed, however, seHAS can facilitate the synthesis of only LMW HA in a time-dependent manner, irrespective of exogenous addition of nucleotide sugar precursors.

A B

C D

Figure 2-6: Construction of pHY300PLK-seHAS for expression in Bacillus subtilis DB104 cells. (A) Amplification of the seHAS insert gene (left image, ~1.3 kb) and pHY300PLK vector backbone (middle image, ~5.4kb) harbouring the XmaI-NotI restriction sites for T4 ligation. (B) Confirmation of successful T4 ligation of the seHAS gene into the pHY300PLK vector by colony PCR. M is the GeneRuler™ 1 kb DNA Ladder (Fermentas, St. Leon-Rot, Germany), EV is the parental pHY300PLK plasmid. (C) Exemplar cell cultivation time course of B. subtilis DB104 cells harbouring pHY300PLK-seHAS in TBTET (37 °C, 220 rpm, 28 h) to allow for protein production. (D) Exemplar seHAS expression profile for up to 28 hours of expression, under aforementioned cultivation conditions in 12 % SDS gel. The PageRuler Prestained Protein ladder 10-180 kDa (Thermo Fisher Scientific, Schwerte, Germany) was used as reference protein ladder. The expected size for seHAS is 48 kDa.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-7: HA biosynthesis by pHY300PLK-seHAS in Bacillus subtilis DB104 cells. B. subtilis DB104 cells harbouring pHY300PLK-seHAS were grown in TBTET (37 °C, 200 rpm, 24 h). Cells were pelleted from 1.5 mL aliquots, sonicated and supplemented with full in vitro synthesis cocktail, solution lacking UDP-GlcNAc and solution lacking UDP-GlcA. In vitro synthesis products were collected hourly during the three-hour synthesis duration (37 °C). Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Nucleic acids stain purple, while HA stains blue. Both GeneRuler™ 1 kb DNA ladder and Select-HA Hi Ladder were used as standards.

Figure 2-8: Verifying time-dependent HA biosynthesis by pHY300PLK-seHAS in Bacillus subtilis DB104 cells. B. subtilis DB104 cells harbouring pHY300PLK-seHAS were grown in TBTET (37 °C, 220 rpm, 28 h). At 2, 22 and 28 h after inoculation, 1.5 mL aliquots were taken. Cells were pelleted, washed, resuspended in PBS and sonicated (40 % amplitude, 10 s ON, 10 s OFF, 2 min) and supplemented with either complete in vitro synthesis cocktail or only PBS buffer. In vitro synthesis products were collected hourly during the three-hour synthesis duration (37 °C). Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Nucleic acids stain purple, while HA stains blue. Both GeneRuler™ 1 kb DNA ladder and Select-HA Hi Ladder were used as standards.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.3.1.3. Saccharomyces cerevisiae

A B

C D

Figure 2-9: Construction of pYES2-seHAS for expression in Saccharomyces cerevisiae. (A) Amplification of the seHAS insert gene (left image, ~1.3 kb) and pYES2 vector backbone (middle image, ~5.4 kb) with complementary overhangs suitable for homologous recombination. (B) Confirmation of successful integration of the seHAS gene into the pYES2 vector by colony PCR. M is the GeneRuler™ 1 kb DNA Ladder (Fermentas, St. Leon-Rot, Germany). Prospective clones 3 and 5 were further verified by sequencing. (C) Exemplar cell cultivation time course of Saccharomyces cerevisiae cells harbouring either pYES2-(EV) or pYES2-seHAS in YPD medium to allow for protein expression. (D) Exemplar seHAS expression profile in 12 % SDS gel for up to 48 hours of expression. The PageRuler Prestained Protein ladder 10-180 kDa (Thermo Fisher Scientific, Schwerte, Germany) was used. The expected size for seHAS is 48 kDa.

seHAS was successfully incorporated into the pYES2 vector (Figure 2-9A,B; Clones 3 and 5) and was verified by sequencing. Cultivation of S. cerevisiae cells harbouring pYES2-seHAS or the empty vector in YPD medium display similar growth patterns with the lag phase starting from 30 hours of cultivation with an optical density of 30 (Figure 2-9C). Heterologous seHAS expression by S. cerevisiae cells was not discernable in the SDS gel (Figure 2-9D). To determine the ability to synthesize HA, two conditions were tested: synthesis with the complete in vitro synthesis cocktail and

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer synthesis without. As expected, no HA could be synthesized by the non-expressing cells (Figure 2-10A). S. cerevisiae cells, whether expressing seHAS for 28 h or 48 h, polymerized HA chains with time. High molecular weight (HMW) HA (> 2.5 MDa) were produced within 1 hour of synthesis and continued to elongate until 3 hours, at which point the polymerization stagnated. HA polymerization, however, was only possible when seHAS was supplied with the complete in vitro synthesis cocktail. Taken together, seHAS expressed in S. cerevisiae could also direct HMW HA with the polymer length increasing with time, up to 3 hours.

Figure 2-10: Time-dependent HA biosynthesis by seHAS in Saccharomyces cerevisiae. S. cerevisiae cells harbouring pYES2-(EV) (A) or pYES2-seHAS (B) were grown in YPD medium (30 °C, 220 rpm, 48 h). At 28 and 48 hours after inoculation, 1.5 mL aliquots were taken. Cells were pelleted, washed, resuspended in PBS and sonicated (40 % amplitude, 10 s ON, 10 s OFF, 2 min) and supplemented with either complete in vitro synthesis cocktail or only PBS buffer. HA products were collected hourly during the allotted four-hour synthesis (37 °C). Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Nucleic acids stain purple, while HA stains blue. Both GeneRuler™ 1 kb DNA ladder and Select-HA Hi Ladder were used as standards.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.3.1.4. Selection of E coli as the seHAS Expression System It has been demonstrated that HA can be synthesized by seHAS recombinantly- expressed in E. coli, B. subtilis and S. cerevisiae. B. subtilis can produce HA autonomously but is limited to only LMW. Due to this inability to produce long chain HA, the possibility of using B. subtilis as microbial host was discounted. Both E. coli and S. cerevisiae displayed similar HA polymerizing capability as hosts, and for the easier processing and shorter cultivation periods, E. coli was selected for microbial HA production.

2.3.1.4.1. seHAS Membrane Localization Detection One challenge in working with membrane-bound enzyme is protein expression. SDS- PAGE could not conclusively show the expression of seHAS in all three hosts. To address this, a method for cell-surface display technology was adapted (Figure 2-11) [184]. Briefly, a 13-amino acid epitope sequence was inserted in the extracellular flexible loop of seHAS (Phe28-Gly29). Detection of this epitope was facilitated by FITC-conjugated anti-ETAG antibody. Binding of the antibody to the epitope was observed via fluorescence microscopy (Figure 2-12). As expected, the empty vector control showed no fluorescence. E. coli cells expressing only seHAS-WT also showed no fluorescence, while E. coli cells expressing seHAS-WT-ETAG emitted fluorescence. Qualitatively, this suffices to confirm the presence and localization of seHAS to the outer membrane. To assess the effect of the sequence insertion in seHAS, a comparative HA synthesis study was performed (Figure 2-13). Considering the three- hour HA synthesis period, cells containing only pET-22b(+) did not produce HA, while cells harbouring seHAS-WT produced the expected HA smear, while those seHAS-WT- ETAG showed diminished HA production. Nevertheless, this approach verifies the expression of seHAS in E. coli.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

A B

Figure 2-11: Determining the presence of seHAS by membrane localization. (A) The EstA cell surface display detection system [184] was employed by incorporating the E-tag epitope sequence (13 aa: G-A-P-V-P-Y-P-D-P-L-E-P-R) into a portion of seHAS. Visual inspection of the YASARA-generated homology model of seHAS [183] shows a prospective flexible loop between the first and second intramembrane α-helices suitable for the E-tag detection system. Phe28-Gly29 were chosen as the insertion site for the E-tag epitope. (B) Successful incorporation of the E-tag into seHAS is confirmed by the subsequent fluorescence detection. The membrane topology of seHAS was reconstructed from recent data [60, 64, 104] consisting of 6 membrane domains (purple), 2 extracellular flexible loops (dark blue) and 5 cytosolic domains.

Figure 2-12: Confirming the presence of seHAS by confocal microscopy. The E-tag epitope sequence (13 aa: G-A-P-V-P-Y-P-D-P-L-E-P-R) was introduced between Phe28-Gly29 of flexible loop between the first and second intramembrane α-helices. E. coli BL21 cells harbouring pET-22b(+)±−seHAS±ETAG were grown to express seHAS accordingly (30 °C, 20 h), pelleted and processed for FITC-conjugated antibody detection. The samples were analyzed using the Leica Application Suite confocal microscope (courtesy of Dr. Kristin Rübsam). Images were taken with an excitation wavelength of 488 nm, an emission wavelength of 519 nm, 10 % intensity, 63 x optical magnification, 2.1 x digital zoom and a gain of 981. An overlay of transmitted light and fluorescent images was used to determine fluorescent cells.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-13: Determining the effect of ETAG sequence incorporation on HA polymerizing activity. E. coli cells harbouring empty pET-22b(+) (negative control), pET- 22b(+)-seHAS, pET-22b(+)-seHAS-ETAG were grown in TBAMP (37 °C, 200 rpm) until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 200 rpm, 24 h). Cells were pelleted from 1.5 mL aliquots, sonicated and supplemented with in vitro synthesis cocktail. In vitro synthesis products were collected at 0 h, 3 h and 20 h thereafter. Samples were resolved in 0.5 % agarose gel and stained with Stains- All. Nucleic acids stain purple, while HA stains blue. The GeneRuler™ 1 kb DNA ladder and Select-HA Hi Ladder were used as standards.

2.3.2. Establishment of the Screening System in 96-Well Microtiter Plate Format for the seHAS Engineering Campaign

2.3.2.1. CTAB Turbidimetric Assay Optimization This method depends on the formation of an insoluble complex between HA and cetyltrimethylammonium bromide, where the turbidity is linearly proportional to the quantity of HA in the system [179]. Figure 2-14 depicts the linear association with a co- efficient of determination of 0.9994, up to 300 µg/mL HA. To determine its applicability as a screening system, a 96-well MTP was cultivated with E. coli BL21 GOLD (DE3) cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)- seHAS-WT (Figure 2-15). After expression and in vitro synthesis, HA products quantified by the CTAB assay have a mean absorbance value of 0.1219, standard

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer deviation of 0.0256 and a 21% percentage error for the wild type. These values, however, were statistically insignificant with respect to the absorbances from the empty vector controls. To corroborate this, the standard deviation test was repeated twice more (Figure 2-15). Three independent runs showed average absorbances (λ = 400 nm) fluctuating from 0.14-0-17 and 0.10-0.14 for the WT samples and empty vector samples, respectively. Considering the standard deviation of the samples, it can be concluded that the CTAB assay cannot discriminate between signals from WT and empty vector samples. Therefore, CTAB is unsuitable for the quantitative screening of seHAS libraries.

A B

Figure 2-14. Establishing the CTAB assay as a turbidimetric screening platform for seHAS engineering. (A) The linear concentration-absorbance regression (y = 0.0027x + 0.0375; R2 = 0.9994) from 0-312.5 g/mL was established to quantify HA. (B) Determination of the co-efficient of variation of the CTAB turbidimetric screening system. The apparent co- efficient of variation is obtained from the absorbance values from a full plate of BL21 GOLD (DE3)-pET-22b(+)-seHAS-WT. A mean absorbance (λ = 400 nm) of 0.13 was obtained for the seHAS-WT.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-15: Screening by CTAB. E. coli BL21 GOLD (DE3) cells harbouring empty pET- 22b(+) (negative control) and pET-22b(+)-seHAS-WT were grown in TBAMP (37 °C, 900 rpm) in 96-deep well MTP until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 900 rpm, 24 h), in vitro synthesis was facilitated and HA products were quantified by CTAB as previously described. The experiment was performed three times.

2.3.2.2. Alcian Blue Colorimetric Assay The Alcian Blue colorimetric assay relies on the complexation of the polyanionic hyaluronic acid with the basic dye Alcian Blue 8GX [180]. This is important for determination of molecular binding ratios of Alcian Blue to HA [181]. Unlike CTAB, an inverse relationship between the amount of HA and the absorbance of solution is established. An adapted protocol for 96-well HA quantification using Alcian Blue was performed [187]. In keeping with literature, Figure 2-16 shows the inverse relationship between the HA amount and photometric absorbance. The response curve ranging 0- 400 µg/mL HA produced a second-order polynomial with a correlation of R2 = 0.9804. To test its applicability as a screening system, a 96-well MTP was cultivated with E. coli BL21 GOLD (DE3) cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)-seHAS-WT. After expression and in vitro synthesis, HA products were quantified using Alcian Blue. As evidenced in Figure 2-17, absorbances (λ=540

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer nm) ranged between 0.12-0.14 for the buffer, empty vector control and wild type samples. The same experiment was repeated and photometric measurements continued to show insignificant absorbance differences between the three sample types. These demonstrate that the Alcian Blue colorimetric assay is also unsuitable as a screening system because it cannot discriminate signals between the empty vector and wild type samples.

Figure 2-16: Alcian Blue colorimetric assay in 96-well MTP format. HA samples were prepared as described in Materials and Methods. The endpoint absorbances (λ=540 nm) from each HA standard dilution generated a second-degree polynomial for HA quantification. Insert: An exemplar of HA dilutions with Alcian Blue in MTP.

Figure 2-17: 96-well MTP screening by Alcian Blue assay. E. coli BL21 GOLD (DE3) cells harbouring empty pET-22b(+) (negative control) and pET- 22b(+)-seHAS-WT were grown in TBAMP (37 °C, 900 rpm) in 96-deep well MTP until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 900 rpm, 24 h) and in vitro synthesis was facilitated. HA products were assayed using Alcian Blue colorimetric assay as previously described. The experiment was performed twice.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.3.2.3. 2-AA Fluorescent Labeling of Hyaluronan It was suspected that basal HA production by seHAS-WT presented detection sensitivity challenges in both CTAB and Alcian Blue assays. To address this, HA products were modified to facilitate quantitative fluorescence detection. The fluorophore, 2-anthranilic acid (2-AA), was used to derivatize HA at the reducing ends by reductive amination [188, 189]. With this strategy, the glycosaminoglycan is labeled in a two-step reductive amination process. First, a Schiff's base is formed with the nucleophilic attack on the carbonyl carbon of the acyclic reducing sugar by the dye. The resulting imine group is then subjected to reduction to yield a labeled HA. The principle and methodology are explained in Figure 2-18.

Figure 2-18: Principle of fluorescent labelling of HA with 2-anthranilic acid. The fluorescent labeling of HA was performed according to instructions in the GlycoProfileTM 2-AA Labeling Kit (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany). The two-step reductive amination process is depicted above and can also be found in the Materials and Methods. Insert: The chemical reaction involved with labeling glycosaminoglycans. First, a Schiff's base is formed with the nucleophilic attack on the carbonyl carbon of the acyclic reducing sugar by the dye. The resulting imine group is then subjected to reduction to yield a labeled HA

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer A labeling experiment requiring dilutions with 10-13 to 10-11 mol HA per reaction (n=3) was prepared. One set were derivatized with 2-AA, while the other remained unlabeled as control (Figure 2-19). On average, 2-AA labeled HA emitted fluorescence of 4 x 105 RFUs, while the unlabeled counterparts showed an average of 60 RFUs. It is also noteworthy that the fluorescence did not positively correlate with the amount of derivatized HA (R2=0.2115). Nevertheless, the disparity in fluorescence between the two treatments illustrated an improvement in HA detection sensitivity. Further separation of the conjugated fluorophores from non-conjugated should be made possible by anion exchange chromatography. In tandem, both methods can be used for high throughput screening of seHAS variants for improved HA chain length.

Figure 2-19: Fluorescent labelling of commercial HA. Stock solution of 1.0 mg/mL HA (Hyaluronic Acid Na-salt > 2.O MDa, GfN & Selco, Wald-Michelbach, Germany) was prepared in PBS pH 7.4. The samples were diluted and various volumes (n=3) were prepared to have 10- 13 to 10-11 mol HA per reaction (based on MW=2.12 MDa). Another set was prepared as “non- labeled” samples. The samples were prepared as per instruction and fluorescence was determined using the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland) under the following conditions: excitation wavelength=315 nm; emission wavelength=400 nm; 10 s shaking at 2 mm amplitude; gain=70; number of flashes=50. Insert: fluorescence values for the buffer control.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.3.2.4. Anion Exchange Chromatography Anion exchange chromatography has been previously used to determine the peak molecular mass and the molecular mass distribution of HA [182, 190]. Briefly, the population of HA of various lengths are separated according to overall charge. Longer HA chains possess more electronegative charges from the carboxyl groups of glucuronic acid and bind stronger to the immobilized cation surface compared to shorter chains. Therefore, the longer HA species elute with increasing salt concentration (Figure 2-20).

A B

Figure 2-20: (A) Principle and (B) application of anion exchange chromatography. A) The population of HA of various lengths are separated according to overall charge. Longer chains possess more electronegative charges and bind stronger to the immobilized cation surface compared to shorter chains. Separation of polydispersed HA is then made possible by eluting with increasing salt concentration. A protocol suitable for individual or 96-well format was established using the mobile phase (sodium phosphate and sodium sulfate) reported by [182]. B) This is an exemplary vacuum-assisted anion exchange in a 96-well format. A 96-column plate (“column”) is filled with the equilibrated anion resin and then the HA sample. This plate is placed above the “collector” plate for eluate collection. Upon vacuum activation, elution of the mobile phase with increasing salt concentration containing HA is made possible. The eluates can be used in tandem with 2-AA labeling for highly sensitive HA detection or visualized on agarose gel.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer An experiment was performed to test the lower limit of detection and optimal salt concentration to elute HA. Figure 2-21 shows that for preparations containing 0.10 mg HA or lower, separation could not be visualized on agarose gel. For preparations above 0.25 mg HA, HA are retained by the quaternary ammonium anion exchange groups of the resin and only eluted starting with 140 mM sodium sulfate (Buffer A3). To proceed, 0.25 mg commercial HA were bound to 100 µL anion resin and eluted step- wise with either 100 µL or 200 µL of each buffer. The agarose gel (Figure 2-22) shows that with 100 µL elution volume, the minimal elution concentration is 225 mM sodium sulfate, whereas when eluted with 200 µL, HA fractions are already observed with 140 mM sodium sulfate. This suggested that 140 mM sodium sulfate was sufficient to elute HMW HA from the anion resin.

Figure 2-21: Lower limit detection and optimal elution salt concentration determination. Commercial HA of various amount (0.05-0.50 mg; > 2.0 MDa) were loaded into the vacuum-assisted 96-microcolumn set up. Respective wells of the 96-microcolumn plate was added 200 µL of 50 % slurry of the anion resin (GE Source 30Q 30 µm anion beads). The column was equilibrated with 2 mL of Mobile Phase A (Buffer A1:10 mM sodium phosphate, 20 mM sodium sulfate pH 7.0). The commercial HA were applied onto the respective columns. The columns were washed with 2 mL of Buffer A1 and HA were eluted with 400 µL Buffer A2-D in a step-wise manner (10 mM sodium phosphate, X mM sodium sulfate pH 7.0). Eluates were collected at each step and resolved in 0.5 % agarose gel at 100 V for 50 min and stained with Stains-All. The concentrations (mM) of Na2SO4 pH 7.0 vary as follows: A2: 120; A3: 140; A4: 160; A5: 180; B: 225; C: 300; D: 500. C: crude sample; F: flow through; W: wash.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-22: Elution volume optimization. Commercial HA (0.25 mg; > 2.0 MDa) was loaded into the vacuum-assisted 96-microcolumn set up. Respective wells of the 96- microcolumn plate were added 200 µL of 50 % slurry of the anion resin (GE Source 30Q 30 µm anion beads). The column was equilibrated with 2 mL of Mobile Phase A (Buffer A1:10 mM sodium phosphate, 20 mM sodium sulfate pH 7.0). The commercial HA were applied onto the respective columns. The columns were washed with 2 mL of Buffer A1 and HA were eluted with either 100 µL or 200 µL Buffer A3-D in a step-wise manner (10 mM sodium phosphate, various mM sodium sulfate pH 7.0). Eluates were collected at each step and resolved in 0.5 % agarose gel at 100 V for 50 min and stained with Stains-All as previously described. The GeneRuler™ 1 kb DNA ladder and Select-HA Hi Ladder were used as standards. The concentrations (mM) of Na2SO4 pH 7.0 vary as follows: A3: 140; B: 225; C: 300; D: 500. C: crude sample; F: flow through; W: wash.

This condition was applied to a 96-well format. A complete 96-microcolumn set up was prepared with 150 µL anion resin per well and applied with in vitro synthesized HA products (350 µL) from E. coli cells expressing seHAS-WT. HA products from the complete 96-well MTP were resolved in agarose gel for reference. HA that were bound to the microcolumn and eluted with 160 mM sodium sulfate were resolved in another agarose gel slab. Figure 2-23A reveals that the intensity of the HA smears from the reference gel were quite comparable, except those from column 8 that appear to be stronger than the rest. Those passed through the anion exchange column and eluted with 160 mM sodium sulfate (Figure 2-23B) generated HA signals that highly vary in intensities and distributions. More alarmingly, the bottom rows (Column 9-12) appear to have fewer HA smears, signifying a loss of samples or poor anion exchange separation. This illustrates the unreliability of anion exchange chromatography as a screening system. Because of this, the tandem approach with 2-AA fluorescence labeling was not pursued.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-23: Anion exchange chromatography of in vitro synthesized HA from a full 96-deep well MTP. HA synthesized by seHAS-WT in 96-deep well MTP (A) were loaded into the vacuum-assisted 96-microcolumn set up. Respective wells of the 96-microcolumn plate was added 300 µL of 50 % slurry of the anion resin (GE Source 30Q 30 µm anion beads). The column was equilibrated with 2 mL of Mobile Phase A (Buffer A1:10 mM sodium phosphate, 20 mM sodium sulfate pH 7.0). HA products (350 µL) were applied onto the respective columns. The columns were washed with first with 2 mL of Buffer A1 and then eluted with 500 µL elution Buffer A4 (10 mM sodium phosphate, 160 mM sodium sulfate pH 7.0). Eluates were collected and resolved in 0.5 % agarose gel at 100 V for 50 min and stained with Stains-All (B). GeneRuler™ 1 kb DNA ladder was used as the standard.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.3.2.5. Screening for Increased Chain Length Using Agarose Gel Electrophoresis The final option to screen for improved HA chain length was agarose gel electrophoresis. Previous reports have used gel electrophoresis to visualize HA, either as fluorescently conjugated [191, 192] or by staining with a dye [193-195]. However, this method has not been utilized as a screening system for HA production.

Figure 2-24: HA screening with agarose gel electrophoresis. E. coli BL21 GOLD (DE3) cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)-seHAS-WT were grown in TBAMP (37 °C, 900 rpm) in 96-deep well MTP until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 900 rpm, 24 h), in vitro synthesis was performed as previously described. HA synthesis products were resolved in 0.5 % agarose gel and stained with Stains-All. With this method, up to 6 MTPs (540 variants) can be screened in one day, primarily to detect improvement in polymer length.

To test this prospect, a full 96-well MTP was prepared for cell cultivation, protein expression and HA synthesis. The synthesis products were resolved in 0.5 % agarose gel and stained with Stains-All. Figure 2-24 shows in vitro synthesized HA visualized on agarose gel. Expectedly, the wild type produced the characteristic blue smear and

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer the negative controls (empty vector and blank) did not. Furthermore, uniformity with respect to HA chain distribution and signal intensity are evident throughout the gel. The agarose gel detection proved to be the most consistent and reliable out of all the screening systems attempted. This method was therefore selected as the screening system for the seHAS engineering campaigns.

Dilution Conc µg (µg/mL) (in 15 µL) A 625 9.38 B 500 7.50 C 313 4.69 D 250 3.75 E 156 2.34 F 125 1.88 G 78 1.17 H 31 0.47 I 16 0.24 J 8 0.12 K 4 0.06 L 0 0.00

Figure 2-25: Detection sensitivity of agarose gel electrophoresis. Serial dilution (0- 625 µg/mL) of commercial HA in phosphate buffer pH 7.0 (> 2.O MDa) were prepared and resolved in 0.5 % agarose gel (100 V, 50 min) and stained with Stains-All. The table below shows the concentration and amount of HA loaded in each well. GeneRuler™ 1 kb DNA ladder was used as the standard.

To determine the sensitivity of the agarose gel system with Stains-All, dilutions of commercial HA were prepared from 0-625 µg/mL and visualized on the agarose gel. Figure 2-25 shows the lowest HA concentration still visible on the agarose gel was 8 µg/mL, which equates to 0.12 µg of HA. Conversely, 500 µg/mL (7.50 µg HA in total) appears to be the upper limit of detection. Taken together, the agarose gel/Stains-All system has a detection range of 8-500 µg/mL and should be sufficient to distinguish signals variants from the wild type. With this screening system, a workflow for in vitro HA synthesis and library screening has been established to discriminate the polymerizing activities of variants generated for the seHAS engineering campaigns (Figure 2-26).

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-26: Established 96-well MTP HA synthesis and screening protocol. E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-seHAS (WT or variant; empty vector) are cultivated in TBAMP (37 °C, 900 rpm) in 96-deep well MTP until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 900 rpm, 18-24 h). Cells are harvested and permeabilized with 20 µg/mL Polymyxin B prior to in vitro synthesis as previously described. HA synthesis products are resolved in 0.5 % agarose gel and visualized with Stains-All.

2.3.3. Rational Evolution of seHAS

2.3.3.1. Site-Saturation Mutagenesis of the Conserved Cysteines (C226, C262 and C281) Site-directed mutagenesis of the conserved cysteines of streptococcal HAS first showed the effects of amino acid substitution on enzymatic activity [101, 196]. These cysteine residues are clustered at the inner surface of the cell membrane, with C226 and C262 located in proximity to a UDP-binding site [101], and can influence the rate of sugar assembly and effectively the product size [102]. For such reasons, a homology model of seHAS was generated (Figure 2-27) and revealed that three of the four conserved cysteines are in close proximity to each other. To further substantiate this, cys-to- ala/ser substitutions at positions 226, 262 and 281 decreased seHAS activity, suggesting to some degree their contribution to seHAS catalytic activity [100]. The three residues (C226, C262 and C281) were, therefore, selected for site-saturation mutagenesis and screened for improved chain length specificity.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

A B

Figure 2-27: Homology modeling of seHAS by YASARA. (A) The primary sequence of seHAS was aligned with cellulose synthase (PDB 4HG6). 82.3 % were aligned to template residues, with sequence similarity and identity of 40.3 % and 20.0 %, respectively. In the enzyme are β-1,4 linked chain of 18 glucose molecules. (B) Respective conserved cysteine residues from the modeled seHAS structure are shown in green. Due to proximity, C226, C262 and C281 were chosen for site-saturation mutagenesis and screened for production of improved HA polymer length.

Three site-saturation mutagenesis libraries were generated resulting in 522 seHAS variants. The first screen resulted in 56 variants (10.7 % of total) retaining their ability to synthesize HA when the substitutions are introduced (Figure 2-28). The 56 active variants were rescreened in triplicates and HA production was compared against seHAS-WT (Figure 2-29). Three C226 variants (P1.A4, P1.B10 and P1.C3), five C262 variants (P1.H3, P2.A7, P2.B2, P2.C3 and P2.C7) and one C281 variant (P2.G2) performed comparably against seHAS-WT and were selected for sequencing analysis. All prospective beneficial variants (“hits”) were effectively wild type on the amino acid level due to silent mutations (TGC to TGT). This explains the comparable HA signals observed on the agarose gels. To further gain insight into the amino acid substitutions that impaired HA synthesis, the amino acid sequence of six variants (two per position)

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer were investigated. Threonine and proline substitutions were found in position 226, arginine and glycine for position 262 and arginine and tyrosine for position 281. With caution due to the small sample size, substitutions from cysteine to a polar or positively charged amino acid may impose negative consequences on HA synthesis. Taken together, no variant or potential amino acid substitutions were identified in this SSM study that could improve HA synthesis.

Figure 2-28: Site-saturation mutagenesis of the neighbouring conserved cysteines. The conserved residues (C226, C262 and C281) were saturated using NNK primers. E. coli BL21 GOLD (DE3) cells expressing seHAS SSM mutants were cultivated at two microtiter plates per position (174 variants, 12 seHAS-WT and 6 empty vector controls). In vitro synthesis was performed as described in Materials and Methods and HA products were resolved in 0.5 % agarose gel to screen for HA chain length improvement.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-29: Rescreening of site-saturation mutagenesis hits. Out of 522 variants generated (174 per position), 56 variants were rescreened in triplicates. (A) Scheme depicts the placement of the SSM variants in the 96-well format. Each microtiter plate has a built-in positive control (seHAS-WT) and negative controls (empty vector and TB medium). (B) Cells were cultivated in 96-deep well MTP and protein expression was induced with 1 mM IPTG. In vitro synthesis was performed as usual and HA products were resolved in 0.5 % agarose gel. HA production and chain length were compared against those of seHAS-WT.

2.3.3.2. Site-Saturation Mutagenesis of the Polar Intramembrane Residues, Lys48 and Glu327 Another conserved residues of interest are the polar intramembrane residues, Lys48 and Glu327 from seHAS membrane domains 2 and 4, respectively. Kumari and others have investigated the effects of changing Lys48 to Arg or Glu and Glu327 to Lys, Asp or Gln and showed that in general, all variants synthesized relatively smaller HA than seHAS-WT [103]. Moreover, the authors claimed that while Glu327 may be involved in seHAS stability, both Glu327 and Lys48 “are involved in the ability of HAS to synthesize very large HA”. It was therefore of interest to saturate positions 48 and 327 and discover the effects of remaining amino acid substitutions on HA polymerization (Figure 2-30).

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-30: Site-saturation mutagenesis of the polar intramembrane residues Lys48 and Glu327 - Motivation. The membrane topology of seHAS was reconstructed from recent data [60, 64, 104] consisting of 6 membrane domains (purple), 2 extracellular flexible loops (dark blue) and 5 cytosolic domains. The polar membrane residues, Glu327 and Lys48, believed to be involved in seHAS stability and HMW HA production [103] are marked in red stars. Glu327 and Lys48 were saturated using NNK primers for site-saturation mutagenesis and screened by agarose gel electrophoresis for improved HA chain length.

Two SSM libraries were generated (174 variants per position) and were screened by agarose gel electrophoresis. The first round of screening identified 21 variants that produced HA (Figure 2-31). Rescreening in triplicates (Figure 2-32) verified that 19 could still synthesize HA and 8 putative hits produced HMW HA (variants E1, D3, E3, C5 and A3 for position 327 and variants C1, B8 and H11 for position 48). Sequencing revealed that all prospective K48 and E327 variants retained their wild type amino acid sequences. Another interesting observation was that repeated experiments with the K48 plates consistently produced LMW HA (variants F2, G5 and H7 from Plate K48.2; Figure 2-33). These variants were selected for sequencing. Both variants G5 and H7 harboured K48L substitutions, while F2 producing more LMW HA possessed a K48E substitution. This is in agreement with Kamari and colleagues, who showed that a K48E substitution results in production of short HA chains (<0.5 MDa). Overall, the site-saturation mutagenesis at positions 48 and 327 failed to generate variants that produce HMW HA. Alternatively, seHAS variants that synthesize low polydispersed LMW HA (<500 kDa) were discovered.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Lys48

Glu32 7

Figure 2-31: Site-saturation mutagenesis of the polar intramembrane residues Lys48 and Glu327. Glu327 and Lys48 were saturated using NNK primers. E. coli BL21 GOLD (DE3) cells expressing seHAS SSM variants were cultivated at two microtiter plates per position (174 variants, 12 seHAS-WT and 6 empty vector controls). In vitro synthesis was performed as described in Materials and Methods and HA products were resolved in 0.5 % agarose gel to screen for HA chain length improvement. The upper gels represent HA production by Lys48 variants and the lower gels represent HA production by Glu327 variants.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-32: Rescreening of Lys48 and Glu327 site-saturation mutagenesis hits. Out of 348 variants generated (174 per position), 21 notable variants were rescreened in triplicates. E. coli BL21 GOLD (DE3) cells expressing seHAS SSM variants were cultivated in 96-deep well MTP and protein expression was induced with 1 mM IPTG. In vitro synthesis was performed as usual and HA products were resolved in 0.5 % agarose gel. Each microtiter plate has a built-in positive control (seHAS-WT) and negative controls (empty vector). HA product length were compared against those of seHAS-WT.

K48E

K48L K48L Figure 2-33: Identification of reported LMW HA producing variant. Amino acid substitutions from the three selected Lys48 SSM variants were determined. Two harboured K48L substitution, while one had a K48E amino acid exchange. The latter has been previously reported to produce LMW HA products [103].

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.3.4. Random Mutagenesis of seHAS

Figure 2-34: Determination of optimal manganese concentration for the epPCR library. (A) The optimal manganese concentration was determined by adding increasing amount of MnCl2 to each of the five separate insert PCRs. The amplicons were purified and were used as megaprimers for the subsequent PCR. Parental DNA were DpnI-digested and the remainder were transformed into chemically competent E. coli BL21 GOLD (DE3) cells. (B) Colonies were picked to generate one 96-well MTP per manganese concentration. Cells were cultivated, proteins were produced and in vitro synthesis of HA was performed. HA products were resolved in 0.5 % agarose gel and stained with Stains-All. The percentage of active variants was calculated based on the number of HA bands observed (active seHAS variant) divided by the total number. The concentration of 0.1 mM Mn2+ yielded active/inactive clones closest to 50 % and was deemed as suitable for generation of the epPCR library.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer The last approach to engineer seHAS was through error-prone PCR. epPCR is a standard procedure to introduce mutations at the gene level by increasing the error rate of the polymerase typically by adding manganese to the reaction [160]. Initially, the optimal manganese concentration was determined by preparing five separate PCRs with increasing amounts of manganese. After using the mutant amplicons as megaprimers for PCR (Figure 2-34A), the mutant plasmids were transformed into E. coli cells. Cell cultivation and in vitro synthesis were performed as usual. HA products were resolved in 0.5 % agarose gel and stained with Stains-All (Figure 2-34B). The percentage of active variants was calculated based on the number of HA bands observed (active seHAS variant) divided by the total. The concentration of 0.1 mM Mn2+ yielded active/inactive clones closest to 50 % and was deemed as suitable concentration for the epPCR library. A total of 1392 seHAS epPCR variants were generated and screened. As a screening criterion, a “hit” is identified when the HA product is higher on the agarose gel compared to that of seHAS-WT. Figure 2-35 shows the stained agarose gels in which 21 promising epPCR variants produced longer HA relative to seHAS-WT. Rescreening on MTP showed 12 true promising variants producing HMW HA comparably with that of the wild type (Figure 2-36). The corresponding amino acid substitutions of these variants were determined by sequencing.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-35: Screening of epPCR library by agarose gel electrophoresis. An epPCR mutant library of 1392 epPCR variants was generated. E. coli BL21 GOLD (DE3) cells harbouring empty pET-22b(+) (negative control), pET-22b(+)-seHAS-WT and epPCR variants were cultivated and processed for in vitro synthesis as previously described. HA synthesis products were resolved in 0.5 % agarose gel to screen for improved chain length. Each row has internal controls (2 positive seHAS-WT; 1 negative pET-22b(+)-EV) for polymer length comparison. GeneRuler™ 1 kb DNA ladder was used as the standard. Twenty-one promising variants were identified.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

Figure 2-36: Rescreening of potentially improved epPCR variants (96-well MTP). Out of 1392 epPCR variants was produced, 21 were identified to produce HA longer than seHAS-WT. Cells were cultivated (n=3) and prepared for in vitro HA synthesis as previously described. HA products were resolved by agarose gel electrophoresis and visualized using Stains-All. Each row has internal controls (2 positive seHAS-WT; one negative pET-22b(+)- (EV) for polymer length comparison. GeneRuler™ 1 kb DNA ladder was used as the standard. The transferability of the results from the MTP format to shake flask was then investigated (Figure 2-37). Cell cultivation, protein expression and in vitro HA synthesis were performed as described. Due to the inability to purify and quantify the seHAS variants, the amount of enzymes were normalized to 40 mg cells per mL of buffer prior to in vitro synthesis. The agarose gel shows that 11 of the 12 hits produced HA of equal length as the wild type. Interestingly, E11, F5, H5 and H7 are the same clone and produced HMW HA with lower polydispersity than that of seHAS-WT. The same applies for variants A11 and G10 however HA production by G10 was considerably weaker. Variant H2, harbouring N345S/F403L substitutions, is capable of producing HA of similar quantity and polydispersity as the wild type. Overall, the

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer epPCR variants can only polymerize HA that rivals that of seHAS-WT, however, the polydispersity of the HA polymers has been lowered, suggesting an improvement in polymerizing control of seHAS.

A6 B7 E11 E12 F5 G10 G12 G1 H2 H5 H7 A11 (P11.G5) (P11.H6) (P14.F5) (P9.G3) (P11.F5) (P2.C2) (P9.G3) (P16.A8) (P7.H3) (P11.F5) (P11.D4) (P14.C7)

R347S D287N R61W V335A R61W I8L V335A --- N345S R61W R61W I111V F362S D116G D116G Y22F F403L D116G D116G D151G T175A T175A S80P T175A T175A D287E E97G

Figure 2-37: Rescreening of potentially improved epPCR variants (shake flask). Twelve promising epPCR variants were cultivated for HA synthesis. Cells were harvested and normalized to 40 mg/mL in PBS and sonicated (40 % amplitude, 10 s ON, 10 s OFF, total 90 s ON). In vitro synthesis was facilitated and HA products were resolved by agarose gel electrophoresis as previously described. Controls (2 positive seHAS-WT; one negative pET- 22b(+)-EV) were prepared concurrently for HA chain comparison. GeneRuler™ 1 kb DNA ladder was used as the standard. The respective amino acid substitutions are tabulated below the figure.

2.3.5. Investigating the Validity of the Directed Evolution Hits To explain for the false identification of improved seHAS variants, variables within the in vitro synthesis procedure were re-examined. First, the correlation between the volume of culture, optical density and cell mass was established (Figure 2-38A). The regression curve shows the positive linear relationship between the volume of cell culture and the optical density (λ= 600 nm, R2=0.9957) and cell mass (R2=0.9986),

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer respectively. Alternatively, the optical (cell) density is linearly proportional to the cell concentration (Figure 2-38B). This suggests that the quantity of cells (effectively the amount of membrane-bound seHAS) used to synthesize HA is a variable that must be stringently controlled.

A B

Figure 2-38: Establishing the relationship between cell mass, cell volume and optical density. E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-seHAS were cultivated in TBAMP (37 °C, 200 rpm) in a shake flask until OD600 of 0.8, at which point protein expression was induced with IPTG to 1 mM. Cells were grown overnight (30 °C, 200 rpm, 20 h) and harvested. Cells were aliquoted with different volumes, wet cell masses were calculated and the corresponding optical density OD600 were measured. (A) A double y-axes scatterplot showing the co-dependence of cell mass and optical density on the volume of aliquots. (B) A linear relationship between the optical density and cell concentration.

To test this further, HA synthesis for 2 hours with equal amounts of UDP-sugars was performed with increasing amount of E. coli cells expressing seHAS-WT. With the normalized cell concentration of 40 mg/mL, cells were sonicated. Various volumes of the crude cell lysate were centrifuged and the pellets were used for HA synthesis. Figure 2-39A demonstrates the inverse effect of cell mass (effectively amount of seHAS) on HA chain length. Fewer cells produced fewer, longer and less polydispersed HA chains, while increasing the cell amounts produces more but relatively shorter HA.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

A B

Figure 2-39: Investigating parameters that influence HA polymer length. E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-seHAS or just empty vector were cultivated in a shake flask using the established parameters. (A) The effect of cell mass (seHAS) on HA polymer length. Cells were harvested, normalized to 40 mg/mL in PBS and sonicated. In separate tubes, various cell lysate volumes (see above) were pelleted and supplied with the standard in vitro synthesis cocktail to allow for HA polymerization (37 °C, 0 rpm, 2 hrs). PBS pH 7.4 was added to each tube to ensure consistent reaction volumes. (B) The effect of quantity of nucleotide sugar precursors on HA polymer length. Identical preparations were followed, however, a fixed volume of 800 µL cell lysate was supplemented with varying concentrations of UDP-GlcA and UDP-GlcNAc. (C) The effect of synthesis duration on HA polymer length. Identical preparations with standard conditions were followed, aside from HA polymerization which was allotted for 4 h with sampling at specified time points. For all three experiments, HA syntheses were halted by the addition to EDTA to 40 mM and immediately boiling the samples (99 °C, 2 min). HA products were resolved in 0.5 % agarose gel and stained with Stains-All. The GeneRuler™ 1 kb DNA ladder was used as the standard, with the corresponding sizes in kb (purple) and kDa (blue).

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer The effect of the amount of supplied precursors was investigated by allowing HA synthesis with normalized cell amount for 2 hours. Figure 2-39B shows a gradual stepwise positive correlation between precursor amount and HA production. Using up to 0.20 mM of precursors produces HA up to 1.5 MDa, while supplementation with at least 0.40 mM of precursors improved HA length and quantity. From 1.0 mM onwards, the intensity of HA smears (darker blue hue) begins to strengthen, suggesting higher HA concentration.

Finally, to gain insight into the effect of the duration of HA synthesis, the amounts of cells used and the supplied precursors were kept constant. Figure 2-39C shows an upward trend in HA polymer length with time. It is noteworthy that at t = 0 h, up to 1.5 MDa HA polymers are already observed, with a steady increase in chain length up to 3 h. There is little or no difference in the HA production profile produced after 3 h or 4 h. Polydispersity of HA also increases with time. Overall, this shows that the duration of the HA synthesis is critical, especially when screening for improved chain length.

Taken together, quantities of E. coli cells (effectively enzymes) and precursors as well as the HA synthesis duration must be tightly controlled for effective comparison of all HA products. A final flask-based comparison of HA production by seHAS-WT and variants was done (24 mg normalized cells, 0.4 mM precursors, 30 min synthesis duration). Figure 2-40 shows the comparative HA production by the seHAS variants. Against the two seHAS-WT, no variant could produce longer HA chains under the specified conditions. Variants with single (V335A), double (N345S/F403L and R347S/F362S) and triple substitutions (R61W/D116G/T175A and I111V/D151G/D287E) produced HA polymers that nearly rivaled those of seHAS-WT, however, quantity was relatively weaker. The two most notable variants are H2 (N345S/F403L) and A6 (R347S/F362S) having produced the highest amount of HA (most intense smear). Equally important is that they generated more uniform HA with lower polydispersity than those of seHAS-WT.

Ultimately, one round of directed evolution failed to improve the polymerizing property of seHAS-WT. The graphical summary of the seHAS engineering experiments

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer is depicted in Figure 2-41 and Figure 2-42, both of which will be referred to in detail in the Discussion section.

A6 A11 B7 E11 E12 F5 G10 G12 H2 H5 H7 (P11.G5) (P14.C7) (P11.H6) (P14.F5) (P9.G3) (P11.F5) (P2.C2) (P9.G3) (P7.H3) (P11.F5) (P11.D4)

R347S I111V D287N R61W V335A R61W I8L V335A N345S R61W R61W F362S D151G D116G D116G Y22F F403L D116G D116G D287E T175A T175A S80P T175A T175A E97G

Figure 2-40: Rescreening of epPCR hits using new HA synthesis conditions. The final epPCR variants and seHAS-WT were cultivated separately in shake flasks, harvested, normalized to 40 mg/mL in PBS and sonicated as previously described. In separate tubes, 600 µL cell lysate (24 mg wet cell weight) were used, supplemented with 200 µL PBS and 200 µL in vitro synthesis cocktail (final concentrations: 0.04 mM UDP-GlcA, 0.04 mM UDP-GlcNAc, 4 mM MgCl2, 0.4 mM DTT) to allow for HA polymerization (37 °C, 0 rpm, 2 h). HA synthesis was halted by the addition to EDTA to 40 mM and immediately boiling (99 °C, 2 min). HA products were resolved in 0.5 % agarose gel and stained with Stains-All. Controls (seHAS-WT as positive; pET-22b(+)-EV as negative) were also prepared for comparison. The GeneRuler™ 1 kb DNA ladder was used as the standard, with the corresponding sizes in kb (purple) and kDa (blue). The corresponding amino acid substitutions are tabulated below the figure.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

B Figure 2-41: Graphical summary of the protein engineering of seHAS. (A) The membrane topology of seHAS was reconstructed from recent data [60, 64, 104] consisting of 6 membrane domains (purple), 2 extracellular flexible loops (dark blue) and 5 cytosolic domains. The conserved cysteines [101] and the polar membrane residues [103] are marked in red stars. Known glycosyltransferase sequence motifs like DXD [197] for UDP-sugar binding, SGPL and QXXRW [74] for formation of glycosidic linkages and B-X7-B [105, 198] motif for HA binding were also included in the topology. Final epPCR variants and their respective substitutions appear in orange circles. (B) Homology model of seHAS in cartoon view with all the substitutions from the final epPCR variants generated by YASARA [183].

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer

A B

Figure 2-42: Dissecting HA-HAS interaction by analyzing the homology model of seHAS-WT. (A) epPCR variant A6 (substitutions at Arg347 and Phe362) participating in interaction with HA (depicted by red molecule). (B) epPCR variant H2 (substitutions at Asn345 and Phe403). Residue 345, located at the “lid” of the synthase pore, directly interacts with HA while Phe403 is clustered as part of the B-X7-B motif (depicted as mesh) for HA binding.

2.4. Discussion

The goal of this investigation was to engineer HAS from Streptococcus zooepidemicus to improve the production of HA with respect to chain length (>2.5 MDa). To this end, variants generated from either focussed or random mutagenesis approaches were recombinantly expressed in E. coli and the resulting HA products were screened by agarose gel electrophoresis for improved polymer length.

This work has demonstrated that B. subtilis does not require supplementation with nucleotide sugar precursors to synthesize HA. This suggests that it naturally encodes the genes necessary to synthesize UDP-GlcA and UDP-GlcNAc. Analysis of the biosynthetic pathway for HA production revealed that indeed B. subtilis has homologues to the genes for UDP-glucose dehydrogenase (hasB/tuaD), UDP-glucose pyrophosphorylase (hasC/gtaB) and UDP-GlcNAc pyrophosphorylase (hasD/gcaD) to synthesize UDP-GlcA and UDP-GlcNAc, respectively. Heterologous expression of seHAS suffices to direct HA synthesis [68]. Despite this, only LMW HA were generated by B. subtilis which does not meet the goal of the investigation. Moreover, for a

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer controlled library screening it is important that every variant is supplied with standardized amounts of UDP-GlcA and UDP-GlcNAc for HA synthesis, devoid of endogenous levels of the precursor sugar monomers that may affect screening.

E. coli was chosen as the microbial host for the easy handling and short turnover time. E. coli is a Gram-negative bacterium encapsulated by two membrane layers [199]. In this work, overexpression of seHAS was not evidenced by SDS-PAGE, however, expression and cellular localization was confirmed using the EstA cell surface display detection system [184]. Our selected transmembrane loop of seHAS for antibody detection has been corroborated to be an extracellular flexible loop via UV-dependent crosslinking studies [64]. This means that seHAS must be embedded in the outer membrane of E. coli, and not in the inner membrane of E. coli. Literature states otherwise [75]. Membrane topologies of other Class I HAS have been predicted to have 4-6 transmembrane domains [60] most likely creating a pore for HA translocation and extrusion [63, 75, 200], with cytosolic domains most likely involved in UDP-sugar binding, glycosyltransferase activity and HA chain elongation [76, 101, 104, 197]. There has not been a report of any HAS spanning both the inner and outer membrane of the microbial host. With the inner membrane and periplasmic space posing as a barrier between the cytosol (containing the UDP-sugars) and the outer membrane (hosting the synthase), the glycosyltransferase activity of the enzyme becomes more difficult to explain. HAS from S. pyogenes expressed in E.coli Top 10 cells was reported to produce HA (<1.9 MDa) that encapsulates E. coli as evidenced by Alcian Blue staining and HA accumulation in the growth medium [70]. However, mechanism by which this occurred was not explained. The prospect of nucleotide sugar transporters bringing the substrate monomers to the synthase does not mediate the dilemma as these anti- transporters import nucleotide sugars from the cytosol into the lumen of organelles and not extracellularly [201]. seHAS is membrane protein and presents purification challenges. Although it has been accomplished, the purified enzyme loses activity and requires reconstitution with molecules of cardiolipin to restore activity [64, 99]. This lipid-dependence necessitates an undisrupted quaternary structure for full seHAS functionality. To address this,

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer seHAS was not purified and HA synthesis was facilitated by only disrupting the membrane of E. coli with Polymyxin B. This was also advantageous since cell permeabilization with Polymyxin B was conducive to the high-throughput screening platform. Mechanistically, Polymyxin B first inserts its hydrophobic N-terminal to the outer membrane, thus weakening and destabilizing the lipid packing and then penetrates the cell further to disrupt the inner membrane [202]. This is sufficient to allow the influx of the UDP-sugars, ready to be polymerized by the membrane- embedded seHAS. This helps to also explain how the synthase machinery gains access to the nucleotide sugars and negates the issue with cellular localization.

With the flask system, a time-dependent HA synthesis was observed. Within 15 minutes of adding the precursor sugar monomers, approximately 1.0-1.5 MDa HA are produced. Within 30 minutes, up to 2.5 MDa are synthesized. Achieving these polymer chain lengths is possible considering that streptococcal HAS have a polymerization rate of 160 monosaccharides/s [7]. Exact comparison with seHAS in this work was limited by the inability to quantify seHAS in the outer membrane of E. coli. The time course experiment of HA production by E. coli also revealed that 3 hours were sufficient to synthesize the longest possible HA polymer. This could be due to the resolving limitation of the agarose gel or the limitation of the supplied precursors. In this work, 0.5 % agarose gel was used for electrophoretic separation of the polydispersed HA products. Preliminary experiments (results not shown) using 0.4 % and 0.3 % agarose to create larger pores in the gel matrix for the larger HA chains, showed no difference in HA migration pattern. The largest HA polymers did not migrate any farther from the loading well with higher agarose concentrations. By deduction, the plateau observed in chain length after 3 hours must be attributed to the limited supply of nucleotide sugars. Moreover, smears and not bands were observed in the agarose gel reflecting the natural propensity of the enzyme to produce HA chains of various lengths.

Directed evolution campaigns require screening of thousands, if not millions, of variants for the improved property. It is therefore imperative to establish a high- throughput screening system. There have been numerous reports on qualitative

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer detection (e.g. radioactive labeling-paper chromatography [7, 93], Alcian Blue staining-microscopy [180] and gel electrophoresis [194, 195]) and quantitative measurements (e.g. carbazole assay [203], Alcian Blue [181, 187], CTAB [178], viscosity measurements [204], ELISA-like assay with HA-binding proteins (amsbio), gel permeation chromatography-dynamic light scattering [205, 206] of HA with respect to viscosity, chain length and quantity. However, a suitable screening system must be established based on the desired improved property, in this case, chain length.

The CTAB turbidimetric assay [104] and Alcian Blue Colorimetric assay [187] have been used as screening systems for HA production. Preliminary validation efforts in this work resulted in both methods failing to discriminate between the empty vector control and seHAS-WT, a critical feature necessary to screen mutant libraries. Works by Zhang and colleagues and Yu and colleagues failed to show validation of the CTAB and Alcian Blue assays as screening systems, respectively, before screening their libraries of HAS variants. No comment or discussion of preliminary work using empty vector and wild type seHAS could be found in their articles. One plausible explanation perhaps was the sensitivity of the screening systems to low quantities of HA products.

To address this, a tandem 2-AA fluorescence tagging-anion exchange system was also planned. Fluorescently tagging HA with 2-AA addresses the sensitivity problem while anion exchange chromatography separates HMW HA from LMW HA. The optimal elution concentration was determined but when tested with HA produced by seHAS- WT in MTP, inconsistent migration patterns (i.e. varying amounts and size range of HA) were observed on the agarose gel. This suggested that anion exchange chromatography was not a reliable high-throughput screening system for improve polymer length. For this reason, the tandem 2-AA tagging and anion exchange approach was aborted. Alternatively, the agarose gel electrophoresis was ultimately selected as the screening platform. Based on the principle of charge (size) separation followed by staining with a cationic carbocyanine dye (Stains-All), it was possible to visualize HA production. This system was validated by visualizing HA produced by seHAS-WT in a full 96-well MTP. Uniform and consistent HA migration patterns were observed on the gels.

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer Previous mutagenesis experiments on the conserved cysteines of seHAS only substituted cysteine to alanine or serine [101, 196]. The goal of the site-saturation mutagenesis was to determine the effect of the remaining amino acid substitutions on the catalytic activity of seHAS. Screening of 522 variants resulted in 56 retaining the HA synthesis capability, suggesting that the HA synthesis capability of approximately 90% of the mutant population has been impaired. Sequencing of a small samples size revealed that substitution from cysteine to a polar/negatively charged amino acid residue can be detrimental to synthase activity. This perhaps provides another clue as to why these cysteine residues are evolutionarily conserved. Inhibition and substrate protection studies revealed that three of the four conserved cysteines (C262, C281 and C367) are clustered very close together at the membrane-enzyme interface, while C226 is located at the inner surface of the cell membrane. Moreover, both C226 and C262 are located in proximity to a UDP binding site [101]. It is possible that substitution from cysteine to threonine (C226) or arginine (C262) may inhibit the binding of nucleotide sugars to seHAS or that the bulky side groups of the substituted amino acids pose stearic hindrance that prevent HA chain elongation/translocation.

Selective substitutions on the polar intramembrane residues, Glu327 and Lys48, of seHAS suggested that Glu327 may be involved with seHAS stability and Lys48 may interact with Glu327 to synthesize HMW HA [103]. Similar to the conserved cysteines, positions 48 and 327 were independently saturated to determine whether the remaining substitutions can improve HAS activity or HA chain length. Resultantly, no seHAS variant with substantial property improvement was generated. Two variants (K48L and K48E) were proven to produce monodispersed LMW (< 500 kDa) HA products, with the latter having been reported in literature [103]. As previously suggested, the interaction of K48 with Glu327 in membrane domain 4 may have been destabilized, resulting in LMW HA chains. Production of non-dispersed LMW HA is uncharacteristic of seHAS, which naturally produces polydispersed HMW HA. Another candidate for the control of HA polymerization might have been discovered. Lys48 is critical for deciding the length of HA being polymerized. The uniform length of the HA product suggests, to some degree, a rhythmic discontinuance of the HA elongation and/or a premature release of the HA chain extracellularly. Nevertheless,

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer this new finding provides practical importance, especially with respect to the production of monodispersed LMW HA for respective purposes.

One round of directed evolution of seHAS failed to generate variants that produce HA longer than that of seHAS-WT, however, the polydispersity of the HA polymers has been decreased. Single (V335A), double (N345S/F403L and R347S/F362S) and triple substitutions (R61W/D116G/T175A and I111V/D151G/D287E) resulted in decent HA production but still did not fare as well as the wild type. Analysis of the amino acid substitutions using the most updated seHAS topology and the generated homology model combined with the knowledge of previous seHAS mutational experiments revealed the following (see Figure 2-41):

a) Variant G12 (V335A) – The non-polar amino acid substitution is in close proximity to Glu327 argued to be have the ability to synthesize HMW HA [103]. Val335 is located in the intramembrane domain 4 so it makes sense that an aliphatic side chain is retained in the hydrophobic portion of the enzyme. However, it is not clear how a substitution to a smaller aliphatic amino acid (alanine) can be advantageous. Homology model inspection reveals that Val335 belongs to the “pore” of seHAS for HA translocation, however, the side chain faces away from the growing heteropolysaccharide chain. Therefore, contribution to improved HA polymerization control cannot be accounted for and further functional studies involving Val335 are required.

b) Variant A11 (I111V/D151G/D287E) – All three amino acid residues are cytosolic. I111 situated in the periphery of the enzyme distal from the main catalytic core. D151G substitution would suggest a disruption in UDP-sugar binding motif of DXD known to be important for β-glycosyltransferases [197]. It has been shown that the first aspartate is essential and deletion or substitution leads to eventual impairment of glycosyltransferase activity [107]. In this experiment, a D151G substitution retained synthase activity, suggesting that 151-DAD-153 is not a UDP- sugar binding site. The last amino acid exchange D287E is situated close to but facing away from the catalytic centre of seHAS. This substitution conserved the

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer negative charge, in exchange for a slightly larger side group. The effect of the substitutions could not be explained. Further investigations are required. c) Variant F5 (R61W/D116G/T175A) – The three residues belong to the cytosolic domain that bridges membrane domains 2 and 3. This domain possesses the conserved UDP-sugar binding motif, DXD, (151-DAD-153 and 159-DSD-161) as well as the regions 227-SGPL-YRR-235 and 293-LKQQNRW-299 associated with the glycosyltransferase activity (β-1,3 and β-1,4 linkage formation) [101, 104]. Structural inspection reveals that the three substitutions are situated in the periphery of the enzyme distal from the main catalytic core. This suggests that little or no interaction is possible with the HA/precursors or the catalytic site and that the effect on HA synthesis may have been caused by another unexplained factor. d) Variant H2 (N345S/F403L; see also Figure 2-42) – Both the predicted seHAS topology and homology models are in agreement that Asn345 is located in the external loop connecting membrane domains 4 and 5. Positionally, Asn345 directly interacts with the HA as it is being translocated out of the membrane. There has not been any documentation of Asn345 participating in HA chain elongation, while the mechanism of HA release remains unclear. Since the amino

acid substitution was conserved from one polar uncharged (-CH2CONH2) to

another (-CHCH3OH), the substitution perhaps plays a role in HAS-HA interaction. On the same note, the second substitution in Phe403 belongs to the

(B-X7-B) HA-binding motif that has already been shown to be associated with HA size control [105]. The authors reported that full saturation of the basic amino

acids in the HA binding motif (effectively K-X7-R) results in shorter HA products

compared to the original. Furthermore, one amino acid deletion of the K-X7-R residues starting from the C-terminal resulted in shorter HA products or lack

thereof in some instances. Amino acid substitutions of the X7 residues, however, were not explored. Nevertheless, the study showed that perturbation of the HA binding motif has influence on HA chain length. Substitution from a bulky

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer aliphatic amino acid (Phe) to a small aliphatic amino acid (Leu) in this HA binding motif could only produce HA that are shorter than those of seHAS-WT. These clues lead to the possibility that N345S/F403L participate in HA-HAS interaction to control HA synthesis.

e) Variant A6 (R347S/F362S; see also Figure 2-42) – Both the predicted seHAS topology and homology model concur that Arg347 is situated in the outer flexible loop connecting membrane domains 4 and 5, while Phe362 is part of membrane domain 5. Both amino acids directly interact with the growing HA chain with Phe362 facing the lumen of the enzyme pore and Arg347 on the “lid” of the pore. It is notable that both amino acid substitutions resulted in serine. Firstly, the bulky aliphatic aromatic group of Phe362 in the lumen was replaced by a polar residue, thereby decreasing hydrophobic interactions with the hydrophobic regions of HA inside the pore. This weakens the HA-HAS interaction and can facilitate faster translocation of the polymer chain. Secondly, the branched positively charged side group of Arg347 at the tip of the pore is replaced by a polar side chain. This positively charged residue has the capacity to participate in ionic interaction with HA, owing to its alternating negative charges. Abolishing this positive charge weakens the affinity of the HA to HAS (or vice versa), thereby allowing the HA chain to be released from the pore more easily. Conversely, the new serine residues can participate in hydrogen bonding with HA but nevertheless, interaction has been weakened. These double substitutions also demonstrate the importance of HAS-HA interaction in the control of HA polymer synthesis.

The objective of producing HMW HA longer than that of seHAS-WT in these engineering campaigns was not achieved. False hits from agarose gel screening revealed that HA synthesis conditions must be strictly controlled for proper product comparison. Further investigations revealed that the synthesis time, amount of UDP- sugars and amount of enzyme all (essentially enzyme-to-substrate ratio) influence the synthesis of HA and all three are strongly connected. The factor of time was not an

100

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer issue in these investigations as in vitro HA syntheses were performed simultaneously on the same MTP. The amount of precursor monomers supplied to the in vitro synthesis system was related to polymer size but surprisingly in a step-wise manner. Less than 0.2 mM of the UDP-sugars produced HA within the same size range, while supplying 0.4 mM or higher facilitated for longer HA chain production of comparable size range. The effect of concentration of UDP-sugars on HA size has been investigated [207]. It was argued that the substrate-to-HAS ratio directly controls HA size because the total amount of nucleotide sugar molecules available in the substrate pool limits the extent by which HA molecules are synthesized. On the same note, it was shown that increasing the amount of HAS, with a fixed amount of precursor monomers, inversely affected HA chain length. This is true from both the economic and evolutionary perspective. Competition arises in an increasing population (HAS) with limited available resources (monomers). In literature, HA synthesis using increasing amounts of membrane-bound HAS (2 µg to 14 µg to 26 µg) generated an HA size distribution shift from 2.3-4.2 MDa to 0.6-2.5 MDa to 0.5-1.6 MDa, respectively [105].

One directed evolution campaign for Class I seHAS [104] was published before the completion of this seHAS engineering investigation, using B. subtilis as the production host and CTAB turbidimetric assay for screening. Unlike this directed evolution campaign, their work focussed on engineering the cytosolic domain of seHAS for improved HA production coupled with metabolic engineering by upregulating the expression of precursor genes for UDP-GlcA and UDP-GlcNAc, boosting HA production titer to 2.8 g/L and the molecular weight to 2.6 MDa. Another difference noted was that HA was obtained from the variants by SDS capsule release and ethanol precipitation prior to the CTAB measurements. No cell disruption suggested that HA was secreted out of the cells. In our work, sensitivity issues were encountered as HA production by seHAS-WT was basal (estimated 0.2 g/L). HA production with B. subtilis was at least 6-fold higher (>1.22 g/L) thereby making the CTAB assay more suitable for screening. These differences perhaps made the engineering of seHAS more feasible.

101

Chapter 2: Engineering of Class I HAS from Streptococcus equi subsp. zooepidemicus Towards Improved HA Chain Length and Titer 2.5. Conclusion

seHAS was subjected to a series of rational design and directed evolution campaigns but failed to generate improved variants that produce HMW HA with improved titer. While factors pertaining to HA synthesis are multilayered, it is conceivable that positions associated with the control of HA polymerization by seHAS may have been discovered. Site-saturation mutagenesis generated two variants (K48L and K48E) that consistently produce monodispersed LMW HA products, while epPCR variants H2 (N345S/F403L) A6 (R347S/F362S) produce HMW HA with polydispersity lower than that of seHAS-WT. Molecular insights provide clues toward HA-HAS interaction as one regulatory mechanism to influence HA polymerization. The discovery of these new positions provides the starting point in the future to dissect the control of HA chain polydispersity by HA synthases.

102

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

3. Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer “If the wind will not serve, take to the oars” – Latin Proverb

3.1. Declaration Contents of this chapter have been published [1] and were reproduced with permission from the publisher. Credit is given to the original source:

103

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.2. Project Objective

In the latter part of this doctoral research project, the rapid evolution of seHAS towards improved production of HA by titer and length was reported [104]. This devalued the novelty of the seHAS engineering work. Moreover, executing a directed evolution campaign on the membrane-bound seHAS using the techniques presented in Chapter 2 presented numerous challenges and did not meet the goal of the project, despite discovering positions potentially involved in HA polymerizing control. This chapter aims to determine the plausibility of improving the enzymatic activity of the membrane- associated pmHAS (Class II) by the process of KnowVolution. With such an approach, two screening systems were simultaneously employed to detect improvement in enzymatic properties, namely CTAB turbidimetric assay and agarose gel electrophoresis. Resultantly, this effort culminates in the evolution of pmHAS towards improved HA titer (productivity determined by CTAB) and HA chain length (evidenced by agarose gel). This, therefore, is the first case of a directed evolution of a Class II HA synthase and an example of a simultaneous dual property improvement through protein engineering. The overview of the pmHAS KnowVolution campaign is depicted in Figure 3-1.

104

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

Figure 3-1. Overview of the pmHAS KnowVolution campaign. The goal was to generate a pmHAS variant producing longer HA chains at a higher output compared to the wild type. Phase I resulted in the identification of seven improved epPCR variants. In Phase II, eight beneficial positions from the most improved epPCR variants were determined and saturated by site-saturation mutagenesis. Phase III entailed the selection of four beneficial substitutions based on the screening results and analysis of the pmHAS model. In Phase IV, the most beneficial substitutions were recombined to generate the final variant. In parallel, epPCR variant, pmHAS-P6.H8 (V59M/T104A), was used as another parent for recombination. One cycle of directed evolution was performed to identify the best performing variant, pmHAS-VF (T40L/V59M/T104A).

Site-directed mutagenesis on the conserved cysteines [101] and intramembrane polar residues (K48 and E327) [103] of the Class I streptococcal HAS first showed the effects of amino acid substitutions on HA titer and chain length specificity. The same synthase was evolved for improved production and molecular weight [104], demonstrating the applicability of protein engineering to a membrane-bound synthase (Class I). Until our recent publication, Class II pmHAS has never been subjected to a directed evolution campaign for improved HA production and chain length specificity.

With this intent, the four-phase protein engineering strategy Knowledge-gaining directed evolution (KnowVolution; Figure 3-1) [175]) was implemented to boost the

105

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer synthase activity of pmHAS while gaining molecular understanding of the improved properties. In Phase I, a standard random mutagenesis approach by epPCR was performed to generate and identify pmHAS variants producing longer polymers at higher output. In Phase II, the potentially beneficial positions of the identified variants were subjected to site-saturation mutagenesis. In Phase III, a pmHAS homology model was generated and a structural inspection was performed to see whether amino acid substitutions are in close proximity or might even interact with each other. In Phase IV, the best and most reasonable amino acid substitutions were recombined by site-directed mutagenesis to engineer the most-improved pmHAS variant. Throughout the process, a pmHAS variant capable of producing high molecular weight HA was generated and the possible involvement of the N-terminus of pmHAS in HA synthesis was unraveled.

3.3. Materials and Methods

Chemicals were obtained from AppliChem (Darmstadt, Germany), Sigma Aldrich Chemie (Taufkirchen, Germany) or VWR International (Darmstadt, Germany) unless otherwise stated. All oligonucleotides were purchased from Eurofins MWG Operon (Ebersberg, Germany) while the DNA polymerases (PhuS and Taq) were produced in- house. Sequencing was done by Eurofins MWG Operon (Ebersberg, Germany). The list of primers used in this study can be found in the Appendix Table A2.

3.3.1. Generation of pmHAS Construct, Cell Cultivation and Protein Production

3.3.1.1. Subcloning of the pmHAS Gene The gene encoding the truncated, soluble pmHAS1-703 [106] [GenBank accession number: AF036004.2, EC 2.4.1.212] was ordered from Thermo Fisher Scientific as a codon-optimized version for E. coli expression. pmHAS1-703 was subcloned by ligation- dependent cloning into the pET-22b(+) vector at NdeI and NotI restriction sites. The recombinant plasmid also contains a C-terminal hexahistidine tag (pmHAS1-703-His6)

106

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer and is now termed “pmHAS-WT”. The vector pET-22b(+)-pmHAS1-703-His6 was transformed by heat-shock into E. coli BL21 GOLD (DE3) cells (Agilent Technologies, Santa Clara, USA) and the successful subcloning was confirmed by sequencing.

3.3.1.2. Cell Cultivation and Protein Expression in 96-Well Microtiter Plate Colonies from generated libraries were inoculated into the starter 96-well MTP (PS-F- bottom, Greiner Bio-One, Frickenhausen, Germany) containing 100 µL Luria-Bertani broth supplemented with ampicillin to 100 µg/mL and cultivated to saturation (37 °C, 24 h, 900 rpm, 70 % humidity) in Multitron II Infors Shaker (Einsbach, Germany). Ten microliters of the pre-culture were used to inoculate the main 96-well MTP containing Terrific Broth (150 µL TB; 1.2 % peptone (w/v), 2.4 % yeast extract (w/v), 0.4 % glycerol

(v/v), 0.23 % KH2PO4 (w/v), 1.25 % K2HPO4 (w/v)) supplemented with ampicillin (100

µg/mL). Cells were grown (37 °C, 900 rpm, 70% humidity) until OD600 0.6-0.8. After which, IPTG was added to 1 mM to induce protein production (25 °C, 24 h, 900 rpm, 70 % humidity). Expression cultures were harvested (4 °C, 15 min, 3220 g) in Eppendorf 5810R centrifuge (Eppendorf AG, Hamburg, Germany) and stored at -80 °C until further use. The remainder of the pre-culture was preserved as master plate by adding glycerol to 30 % (w/v) and frozen at -80 °C until further use.

3.3.2. Diversity Generation

3.3.2.1. Generation of pmHAS Random Mutagenesis Library by Error Prone (ep)PCR A random error-prone PCR mutagenesis library was generated [185] in tandem with phosphorothioate-based ligase-independent gene cloning (PLICing)[166]. The optimal manganese concentration was determined by adding increasing amounts of MnCl2 (0.0, 0.2, 0.4, 0.6 and 0.8 mM) to each of the five separate insert PCRs (0.2 mM dNTPs, 20 ng pET-22b(+)-pmHAS-WT plasmid, 0.4 µM phosphorothioated forward primer, 0.4 µM phosphorothioated reverse primer, 1 U Taq polymerase in 1 x Taq polymerase buffer, in-house). PCR was performed using the following cycle conditions: initial denaturation 94 °C for 2 min, 25 cycles of denaturation at 95 °C for 15 s, annealing at 65 °C for 30 s, elongation at 72 °C for 2.5 min, final elongation was performed at 72 °C for 5

107

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer min and terminated at 8 °C. The amplicons were purified (NucleoSpin® Gel and PCR Clean-up kit, Macherey-Nagel, Düren, Germany) and eluted with Milli-Q water (Millipore, Billerica, USA).

To amplify the vector backbone, the PCR consisted of 0.2 mM dNTPs, 20 ng pET-22b(+) plasmid, 0.4 µM phosphorothioated forward primer, 0.4 µM phosphorothioated reverse primer and 1 U PhuS polymerase in 1 x PhuS buffer. PCR was performed using the following cycle conditions: initial denaturation 94 °C for 2 min, 25 cycles of denaturation at 95 °C for 30 s, annealing at 65 °C for 15 s, elongation at 72 °C for 3.5 min, final elongation was performed at 72 °C for 10 min and terminated at 8 °C. The template DNA was digested with 20 U DpnI to digest the parental DNA. The PCR products were purified (NucleoSpin® Gel and PCR Clean-up kit, Macherey-Nagel, Düren, Germany) and eluted with Milli-Q water (Millipore, Billerica, USA).

A ratio of purified 0.04 pmol insert: 0.10 pmol pET-22b(+) vector were hybridized to generate the pET-22b(+)-pmHAS_His6 variants. The PLICing products from each concentration were then transformed separately into chemically competent E. coli BL21

GOLD (DE3) and plated on LBAMP (100 µg/mL) agar plates. Colonies (87 epPCR variants, 6 pET-22b(+)-pmHAS-WT and 3 pET-22b(+) empty vector) were picked to generate one 96-well MTP library per manganese concentration. Cells were cultivated, proteins were produced and in vitro synthesis of HA was performed. HA products were resolved in 0.5 % agarose gel and stained with Stains-All. The percentage of active variants was calculated based on the number of HA bands (active pmHAS variant) observed divided by the total number. The manganese concentration producing 50 % active clones was determined to be 0.07 mM. The 50 % active/inactive ratio was chosen since directed evolution studies have reported 1-3 amino acid substitutions in protein variants in a library of 1000-2000 clones with >50 % active population [208, 209]. A random mutant library of 1392 epPCR variants was generated using 0.07 mM Mn2+ and screened using the CTAB turbidimetric assay.

108

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.3.2.2. Determination of Beneficial Substitutions by Site-Saturation Mutagenesis Eight positions (40, 59, 90, 104, 175, 186, 313 and 480) from the best HA producing epPCR variants were saturated using NNK primers for site-saturation mutagenesis (Appendix Table A2). A two-step PCR was used comprising of 0.2 mM dNTPs, 20 ng pET-22b(+) -pmHAS_His6, 1 U PfuS polymerase in 1 X PfuS buffer. Half of the mixture was added 0.2 µM of the forward primer, while the other half was supplemented with 0.2 µM of the reverse primer. The first step of the PCR was performed as follows: initial denaturation at 98 °C for 30 s, then seven cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 3.75 min. The PCR mixtures were combined and then partitioned in half for the second step of the PCR as follows: initial denaturation at 98 °C for 30 s, then 18 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 30 s and extension at 72 °C for 3.75 min. A final extension at 72 °C for 10 min completed the DNA amplification and the products were cooled to 8 °C. The parental DNA was digested overnight by DpnI at 37 °C with 1.5 µL of BSA and 1.5 µL of DpnI per 50 µL reaction. The mutant plasmids were purified using the Macherey-Nagel kit (Macherey-Nagel, Düren, Germany) and eluted with Milli-Q water (Millipore, Billerica, USA). One hundred nanogram each of the mutant plasmids were then transformed separately into chemically competent E. coli BL21 GOLD (DE3) cells and plated on LBAMP (100 µg/mL) agar plates. Libraries (174 variants, 12 pET-22b(+)- pmHAS-WT and 6 pET-22b(+)-EV) were generated and screened using the CTAB assay.

3.3.2.3. Recombination of Beneficial Positions by Site-Directed Mutagenesis pmHAS-WT was used to introduce the substitutions (T40L, V59L, T104L and I175N) to generate single, double and triple mutants. In parallel, a T40L or I175N substitution (single or combined) was added to the epPCR variant, pmHAS-P6.H8 (V59M and T104A), which can synthesize more and longer HA chains than pmHAS-WT. The same two-step PCR conditions described above, using the respective primers, were used for site-directed mutagenesis. For primer details, see Appendix Table A2. Site-directed mutagenesis was verified by sequencing and analyzed using Clone Manager 9 Professional Edition Software (Scientific & Educational Software, Cary, NC, USA). The final variant (T40L/V59M/T104A) is referred throughout as “pmHAS-VF”.

109

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.3.3. In vitro HA Biosynthesis The expressed libraries in 96-well MTPs were initially washed with 150 µL of PBS (room temperature, 10 min, 900 rpm on TiMix Shaker, Edmund Bühler, Hechingen, Germany) and pelleted (24 °C, 15 min, 3220 g in Eppendorf 5810R centrifuge, Hamburg, Germany). Cells were resuspended with the same volume of PBS. Cell lysis was initiated with two rounds of freeze-thaw cycles (-80 °C for 15 min then 37 °C for 30 min) and then finalized with lysozyme digestion (2 mg/mL; 37 °C, 1 h, 900 rpm). The lysed cells were centrifuged (4 °C, 20 min, 2900 g in Eppendorf 5810R centrifuge) and 150 µL of the supernatant were transferred to a clean V-bottom MTP (PS-V-bottom, Greiner Bio- One, Frickenhausen, Germany). Seventeen microliters of the in vitro synthesis cocktail were added (final concentrations: 0.4 mM UDP-GlcA, 0.4 mM UDP-GlcNAc, 4 mM

MgCl2, 0.4 mM DTT) to allow for HA polymerization (37 °C, 20 h, 0 rpm). The final product was clarified (24 °C, 20 min, 2900 g in Eppendorf 5810R centrifuge) before HA screening, analysis and/or characterization.

3.3.4. MTP-Based Screening of Variants

3.3.4.1. 96-Well Microtiter Plate Screening of HA by CTAB Turbidimetric Assay The 96-well format of the CTAB turbidimetric assay was slightly modified [178]. Turbidity is based on the formation of an insoluble complex between HA and cetyltrimethylammonium bromide and is linearly proportional to the quantity of HA in the system [179]. Briefly, 60 µL of HA (in vitro synthesis products or HA standard (Hyaluronic Acid Na-salt > 2.O MDa, GfN & Selco, Wald-Michelbach, Germany) were added to 20 µL of 0.1 M phosphate buffer, pH 7.0 and incubated (37 °C, 0 rpm, 15 min). Precipitation occurred upon addition of 160 µL of pre-warmed CTAB reagent (2.5 g in 100 mL 2 % NaOH). The end-point absorbance was measured within five minutes of the reaction (400 nm, 2-second shake, 5-second settle time) with the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland).

3.3.4.2. Visualization of HA Production by Agarose Gel Electrophoresis For visual analysis of the HA products, 15 µL HA products were mixed with 3 µL of 6 X DNA gel loading dye (Thermo Fisher Scientific, Schwerte, Germany) and applied to 0.5

110

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer % agarose gel for electrophoresis at 100 V for 50-60 min. The agarose gel was soaked in 30 % ethanol for 1 h and stained in darkness for at least 8 h with Stains-All (0.1 mg/mL in 30 % EtOH). The stain was discarded and replaced with distilled water. The gel was exposed to white light for destaining and scanned with the Canon Scan 5600F scanner. The image was saved as a TIF file.

3.3.5. Characterization of pmHAS-WT and pmHAS-VF

3.3.5.1. Quantification of pmHAS Expression in MTP by ImageJ To compare the obtained improvements under screening conditions, enzymatic conversion by pmHAS-WT and pmHAS-VF were done in 96-well microtiter plates. Separately, pmHAS-WT and pmHAS-VF were expressed and HA syntheses were performed as previously described. The synthesis products were centrifuged (4 °C, 20min, 2900 g; Eppendorf 5810R centrifuge) and the supernatant containing the cell lysate and synthesis products were pooled together. Samples were prepared for SDS- PAGE [210]. In parallel, a dilution series of lysozyme (0-3 mg/mL) was prepared for protein quantification. The HA and lysozyme samples were boiled at 98 °C for 5 min and resolved in 12% SDS-PAGE gel, initially at 90 V for 15 min, then 120 V for 1.5 h. The gel was stained with Coomassie Brilliant Blue (R-250). The gel was scanned with Canon Scan 5600F and the image was saved as a TIF file.

The quantify the amount of proteins in the gel, ImageJ analysis was employed [211]. Briefly, the image type was converted to a 32-bit grey scale and inverted to have a black background and white bands. Using the “rectangle” function, the protein bands were highlighted consistently to generate the peak areas. The integrated densities under each curve corresponding to the pmHAS (or lysozyme) bands were measured. A standard curve of the lysozyme dilution series (0-3 mg/mL) and the corresponding band intensities was plotted. The integrated densities of the lysozyme bands were interpolated within the quadratic curve to calculate the corresponding lysozyme concentration in solution (http://www.math.com/students/calculators/source/quadratic.htm). The concentration of pmHAS involved in HA synthesis was calculated by multiplying the known lysozyme concentration by the pmHAS/lysozyme ratio.

111

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.3.5.2. Molecular weight determination of HA products by ImageJ. To determine the molecular weight range of the HA products, HA products were resolved in 0.5 % agarose gel for electrophoresis at 100 V for 150 min. ImageJ [211] analysis was used to determine the migration distances of the HA bands and calculate the corresponding molecular weights. Briefly, the distance of migration of each of the HA bands from the Select-HA HI and Mega Ladder (Hyalose, amsbio, Frankfurt, Germany) were measured using the "straight line" option. The distances were measured in arbitrary units (au). A standard curve of log molecular weight of each band and the corresponding migration distance were plotted. The migration distance of the HA products (highest point and lowest point of the smear) were interpolated within the linear equation to calculate the corresponding molecular weight range.

3.3.6. Molecular Modeling of pmHAS A homology model of pmHAS was generated using the YASARA Structure version 17.4.17 [183]. pmHAS-WT shares 63 % sequence identity with the chondroitin polymerase from Escherichia coli strain K4 (K4CP). Two different crystallographic structures were selected manually as template (PDB ID: 2Z86 and 2Z87) [212]. A total of 5 models were generated, analyzed, sorted by Z-score and a monomer model (amino acid residues 32-703) based on 2Z86 was selected.

The changes in the Gibbs free energy (ΔG) induced by the substitutions at positions 40, 59 and 104 in the pmHAS-WT structure were calculated with increased ionic strength by FoldX method (version 3.0 beta3) [213] in YASARA Structure version 17.4.17 [183]. The structure of the homology model of pmHAS was rotamerized and energy minimized using the “RepairObject” command to correct the residues that have non-standard torsion angles. Then, structures of singly/doubly substituted variants were constructed using the “mutate (multiple) residue” command and the ΔG values (=ΔGvariant-ΔGwt) were calculated using the YASARA FoldX-plugin [213] for FoldX method [214].

The model was then further evaluated for protein geometry by Structural Analysis and Verification Server tools (SAVES) PROCHECK (Ramachandran plot, ERRAT) [215] and ProSA [216, 217].

112

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer The structural models of pmHAS-WT and pmHAS-VF were neutralized and solvated in a periodic box containing TIP3P [218] water and 0.9 % (low ionic strength) NaCl. All MD simulations were performed in triplicate using AMBER14 force field [219], and YASARA software package (Ver. 17.4.17) [183]. The simulation parameters were kept at the default values defined by the macro. Electrostatics were calculated using a cut-off of 7.86 Å; long-range interactions were calculated by using the particle-mesh Ewald integration. Bond length to hydrogen atoms and bond angles in water were constrained to speed up the simulation [218]. After initial minimization by steepest descent and simulated annealing until convergence (<0.02 kJ mol−1 per atom during 200 steps) were reached. MD simulation was performed for 95 ns at 298 K by rescaling the time- averaged atom velocities using a Berendsen thermostat [220] and a solvent density of 0.997 g L−1. Snapshots were taken every 25 ps and the recorded trajectories were statistically analyzed using YASARA and VMD1.8 [221].

3.4. Results

The results are organized into four subsections: (1) the establishment of the screening system is presented; (2) the generation and selection of the best pmHAS epPCR variants and the identification and recombination of the most beneficial substitutions are highlighted; (3) the characterization of pmHAS-WT and pmHAS-VF with respect to product output (quantity and polymer length) is presented; and (4) molecular insights into the improved pmHAS synthase activity are provided.

113

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.4.1. Investigating the Possibility of the CTAB Turbidimetric Assay and Agarose Gel Electrophoresis as Screening Systems The recombinant expression of pmHAS-WT by E. coli was confirmed by the SDS gel (Figure 3-2A). The HA synthesizing capability of the soluble pmHAS-WT was also demonstrated by the thick blue band below the 509 kDa on the agarose gel, which is not visible with the empty vector control (Figure 3-2B). This suggested that HA production by pmHAS-WT can be differentiated qualitatively on the agarose gel. To determine whether this was plausible quantitatively, HA products from E. coli cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)-pmHAS-WT were quantified using the CTAB assay. Figure 3-3 shows the linear regression of the assay and the co- efficient of variation by the empty vector and wild type. A mean absorbance of 0.36 (six times the empty vector) was obtained for the pmHAS-WT, reiterating the suitability of CTAB as a screening platform for the directed evolution of pmHAS.

A B

Figure 3-2. Investigating the expression of pmHAS and its capability to synthesize HA. E. coli cells harbouring empty pET-22b(+) (negative control) and pET-22b(+)-pmHAS-WT were cultivated in TBAMP (37 °C, 900 rpm) until OD600 of 0.8, at which point protein expression was induced by adding IPTG to 1 mM. Cells were grown overnight (25 °C, 900 rpm, 16 h) and harvested. (A) Whole cell SDS-PAGE (12 %) was performed by running at 120 V for 90 min and the gel was stained with Coomassie Brilliant Blue (R-250). pmHAS is expected to run at 82 kDa. The PageRuler Prestained Protein ladder 10-180 kDa (Thermo Fisher Scientific) was used as reference protein ladder. (B) Cells were normalized to 50 mg/mL, sonicated and centrifuged. To 200 µL supernatant, 250 µL IVS cocktail stock was added and HA synthesis was facilitated (37 °C, 20 h, 0 rpm). HA products were resolved in 0.5 % agarose gel at (100 V, 50 min) and stained using Stains-All. GeneRuler™ 1 kb DNA Ladder (Fermentas) was used as a marker and the corresponding sizes in kDa are specified in blue.

114

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

A B

Figure 3-3. Establishing the CTAB assay as a turbidimetric screening platform for pmHAS engineering. (A) The linear concentration-absorbance regression (y = 0.003x + 0.0414; R2 = 0.9975) from 0-312.5 g/mL was established to quantify HA. (B) Determination of the co-efficient of variation of the CTAB turbidimetric screening system. The apparent co- efficient of variation (--) is obtained from the absorbance values from a full plate of BL21 GOLD (DE3)-pET-22b(+)-pmHAS-WT. The co-efficient for the empty vector (--) is 15 %, while the true co-efficient of variation (--) is 16 %. These values represent the true absorbance after subtracting the background of the empty vector. A mean absorbance of 0.36 obtained for the pmHAS-WT was six times the empty vector. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

3.4.2. Directed evolution of pmHAS for Improved Polymerizing Activity The sequential scheme for the knowledge-gaining directed evolution of pmHAS has been outlined in Chapter 3.2 Project Objective.

3.4.2.1. Phase I: Identification of Improved Variants by Diversity Generation A suitable screening system is essential for the success of a directed evolution campaign. To this end, the CTAB turbidimetric assay in a 96-well MTP format was employed as a high-throughput screening platform. To determine the reliability of the screening system, the co-efficient of variation was determined (Figure 3-3B). A complete 96-well microtiter plate of E. coli BL21 GOLD (DE3) harbouring pET-22b(+)–pmHAS-WT or empty vector (pET-22b(+); negative control) were grown respectively for HA synthesis. A basal mean absorbance at λ=400 nm of 0.06 was obtained for the empty vector with a co-efficient of variation of 15 %, while a mean absorbance of 0.36 (six times the empty

115

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer vector) was obtained for the pmHAS-WT with a co-efficient of variation of 14 %. The true co-efficient of variation was 16 % and was calculated from the difference in absorbance values between pmHAS-WT and empty vector.

The error-prone PCR random library was generated using the optimal concentration 0.07 mM manganese chloride (Figure 3-4). The 50 % active/inactive ratio was chosen since directed evolution studies have reported 1-3 amino acid substitutions in protein variants in a library of 1000-2000 clones with >50 % active population [208, 209]. A library of 1392 epPCR variants (16 MTPs) was generated. The quality of the epPCR library was determined by sequencing 10 random clones resulting in a mutational frequency of 1.8 mutations per kilobase, 74 % transition (Ts) mutations, 21 % transversion (Tv) mutations and a Ts/Tv ratio of 3.5.

Figure 3-4: Phase I – Determining the optimal manganese concentration for the epPCR library. Mutant pmHAS amplicons were cloned into the pET-22b(+) backbone, transformed separately into E. coli BL21 GOLD (DE3) cells and plated on LBAMP agar plates (100 µg/mL). Colonies (87 epPCR variants, 6 pET-22b(+)-pmHAS-WT and 3 pET-22b(+) empty vector) were picked to generate one plate per manganese concentration in a V-bottom microtiter plate. In vitro HA products from each plate were resolved in 0.5 % agarose gel and visualized with Stains-All. The percentage of active clones was calculated for each manganese concentration and 0.07 mM (at 50 %) was determined optimal for the epPCR library. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

Preliminary CTAB screening of the epPCR library identified sixteen potentially beneficial variants with improved HA titer over pmHAS-WT. The minimal selection criterion of 1.5-fold increase in absorbance (i.e. improvement in HA titer) over pmHAS- WT was deliberately set to establish the selection pressure. Moreover, this selection cut-

116

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer off value also serves to compensate for the 16 % co-efficient of variation in the CTAB assay. Rescreening of the sixteen variants (n=5) and statistical analysis were performed. Table 3-1 shows the variants producing more HA than pmHAS-WT (from 1.18- to 2.35- fold improvement), seven of which exceed the 1.5-fold threshold and the best four (P6.H8, P14.C6, P1.H11 and P16.C10) showing statistically significant improvements. The latter were identified and considered for Phase II of KnowVolution.

Table 3-1: Phase I - Identification of Improved epPCR Variants.

CTAB screening of 1392 epPCR variants (0.07 mM Mn2+) identified sixteen potentially improved variants. Rescreening (n=5) verified the seven best epPCR hits, four of which were statistically significant. Statistical analysis was performed using the following equation, where V is for the absorbance (A400) by the variant and WT is for the absorbance by wild type, Z is the ratio of the variant and wild type and Dx is for the corresponding standard deviations: Z±Dz = 2 2 1/2 (V±Dv)/(WT±Dwt), where Z = V/WT and Dz = Z * [(Dv/V) +(Dwt/WT) ] . Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

Re- Percentage Standard Percentage epPCR screening Standard Ratio Error Deviation Error Variant Average Deviation (Z) (Average) (Dz) (Ratio) (A400) EV 0.0642 0.0025 3.88 0.18 0.015 8.3 WT 0.3588 0.0263 7.32 1.00 0.104 10.4 P14.A10 0.4242 0.0431 10.16 1.18 0.148 12.5 P7.G4 0.4270 0.0416 9.74 1.19 0.145 12.2 P7.D4 0.4393 0.0674 15.34 1.22 0.208 17.0 P11.F10 0.4450 0.0410 9.20 1.24 0.146 11.8 P11.G3 0.4490 0.0207 4.61 1.25 0.108 8.7 P3.G12 0.4523 0.0595 13.16 1.26 0.190 15.1 P11.F2 0.4535 0.0639 14.09 1.26 0.201 15.9 P11.H10 0.4626 0.0534 11.54 1.29 0.176 13.7 P6.B12 0.4842 0.0755 15.60 1.35 0.233 17.2 P3.H2 0.5584 0.0425 7.62 1.56 0.164 10.6 P11.G4 0.5653 0.0863 15.27 1.58 0.267 16.9 P4.A2 0.6142 0.0694 11.30 1.71 0.231 13.5 P6.H8 0.6246 0.0574 9.19 1.74 0.205 11.8 P14.C6 0.6434 0.0208 3.23 1.79 0.144 8.0 P1.H11 0.6683 0.1094 16.37 1.86 0.334 17.9 P16.C10 0.8416 0.0240 2.85 2.35 0.184 7.9

117

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.4.2.2. Phase II: Determination of Beneficial Positions that Lead to Improvement in pmHAS Properties The amino acid substitutions from the seven best epPCR variants were identified (Table 3-2). Eight potentially beneficial positions (40, 59, 90, 104, 175, 186, 313, and 480) from variants P3.H2, P6.H8, P1.H11 and P16.C10 were selected for site-saturation mutagenesis (SSM) primarily on the basis of improvement in HA productivity. The top variants, P16.C10 and P1.H11, demonstrated the best improvement in HA titer over pmHAS-WT, therefore, it was of interest to saturate their respective positions 40, 175, 186, 313 and 480. Moreover, variant P6.H8 had a significantly improved HA titer while producing longer HA chains than pmHAS-WT (Figure 3-5). Therefore, positions 59 and 104 were also chosen for site-saturation mutagenesis to gain insight into their potential role in the control of HA polymerization. Lastly, the lone position 90 from variant P3.H2 was chosen for site-saturation mutagenesis to determine whether substitutions at this position can have additive beneficial effect towards HA production. In summary, eight potentially beneficial positions were determined for site-saturation mutagenesis in Phase II.

Table 3-2: Phase II - Determination of Beneficial Positions for Site- Saturation Mutagenesis.

The seven most improved hits identified in Phase I were selected and sequenced to determine the introduced amino acid substitutions. Eight positions (bold) from P3.H2, P6.H8, P1.H11 and P16.C10 were selected for site-saturation mutagenesis. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

epPCR Ratio/WT Amino Acid Substitutions Variant P16.C10 2.35 T40P/E313V/I175T P1.H11 1.86 T40P/H186L/T480A P14.C6 1.79 F384L/K400R/K542E/N569D P6.H8 1.74 V59M/T104A P4.A2 1.71 N/A P11.G4 1.58 N14D P3.H2 1.56 D90G

118

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

Figure 3-5: Phase I – Qualitative screening of HA production by epPCR Variants. The sixteen best epPCR variants were rescreened in quintuplicates. The HA products were resolved in 0.5 % agarose gel at 100 V for 50-60 min and stained with Stains-All. Although variant P16.C10 performed the best in the CTAB screening, the agarose gel shows that variant P6.H8 produced HA chains longer than pmHAS-WT. Each row has a built-in negative control (empty vector) and two positive controls (pmHAS-WT). A broken line aligning the upper end of the HA produced by pmHAS-WT is used for comparison of HA products. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

119

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.4.2.3. Phase III: Selection of Beneficial Positions and Amino Acid Substitutions for Recombination To select for the most promising positions and amino acid substitutions, we relied on the re-screening results from the site-saturation mutagenesis of the determined positions in Phase II. Screening of two microtiter plates (174 variants, 12 wild type and 6 empty vector control) for each of the eight selected positions resulted in 42 SSM variants with least 1.5-fold higher HA titer than pmHAS-WT. The re-screening of the preliminary hits in multiples (n=5) identified the 14 most beneficial amino acid substitutions primarily arising from positions 40, 59 and 175 (Table 3-3). The best substitutions were T40L with 2.1-fold, V59L with 1.9-fold and I175L/N with 1.9-fold improvement in HA titer over pmHAS-WT. Statistical analysis corroborated the significance of the amino acid substitutions and showed that the HA titer improvement exceeded the 1.5-fold selection threshold (Table 3-3). Therefore, these substitutions are suitable candidates for the recombination phase (Phase IV) of KnowVolution. With the I175L and I175N substitutions producing identical results, the I175N substitution was selected to determine whether introduction of a positively charged asparagine from the non-polar isoleucine at position 175 will have an effect on HA synthesis.

A homology model of pmHAS was also generated (Figure 3-9A) for structural inspection of the identified beneficial positions and amino acid substitutions. The purpose was to discover if the substitutions are clustered and are worth further consideration. The eight selected positions (40, 59, 90, 104, 175, 186, 313 and 480) are dispersed throughout pmHAS. Positions 175, 186 and 313 are located in the GlcNAc-transferase domain, however, they are distant from the binding and active sites (247-DCD-249, 527-DSD-529, 196- DGS-198, and 477-DGS-479). Position 480 is located in the GlcA-transferase site and is proximal to the active/binding domain. Lastly, positions 40, 59, 90 and 104 are located in the flexible N-terminal domain of pmHAS, a region of the synthase previously reported to be not required for HA synthase activity [107].

120

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer Table 3-3: Phase II - Determination of beneficial amino acid substitutions for recombination.

Eight positions (40, 59, 90, 104, 175, 186, 313 and 480) were randomized by site-saturation mutagenesis. Fourteen pmHAS variants exhibited at least a 1.5-fold improvement in HA production over wild type. The best substitutions (in bold) were T40L with 2.1-fold, V59L with 1.9-fold and I175N/L with 1.9-fold improvement. Statistical analysis was performed as previously described. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

First Second Amino Standard SSM Screening Screening Acid Plate Well Deviation %Error Variant Ratio Ratio Substitu- (D ) (V/WT) (Z) z tion 1 40-1 C2 1.5 2.1 0.40 19.4 T40L 2 40-2 A6 1.8 1.8 0.37 20.6 T40R 3 40-2 A7 1.7 1.8 0.36 20.0 T40N 4 59-1 H3 1.7 1.9 0.34 18.2 V59L 5 59-1 H5 1.5 1.6 0.31 19.3 V59T 6 59-1 A6 1.6 1.7 0.34 19.8 V59G 7 59-2 H9 1.5 1.5 0.28 19.5 V59R 8 104-1 G9 1.5 1.6 0.32 19.9 T104L 9 104-2 A8 1.1 1.6 0.30 18.6 T104N 10 175-1 A9 1.5 1.6 0.38 23.5 I175V 11 175-1 G9 1.4 1.9 0.38 19.4 I175L 12 175-1 H9 1.4 1.9 0.37 20.2 I175N 13 175-2 F2 1.6 1.7 0.37 21.9 I175R 14 175-2 G2 1.6 1.8 0.34 18.7 I175G

Furthermore, there was also an interest in the Phase I epPCR variant P6.H8 (V59M/T104A) for producing longer HA chains at higher quantity compared to the wild type (Table 3-2; Figure 3-5). It was hypothesized that additional substitution(s) to variant P6.H8 can further improve its polymerizing activity. We introduced the substitutions T40L and I175N to variant P6.H8 as one parent and the substitutions V59L and T104L alongside T40L and I175N to pmHAS-WT as the other parent. Ultimately, in Phase III of KnowVolution, we identified a cluster of beneficial amino acids at the N-terminal of pmHAS prompting an investigation on the possible role of the N-terminus in HA synthesis.

121

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.4.2.4. Phase IV: Recombination of the beneficial substitutions The amino acid substitutions from Phase III were introduced via site-directed mutagenesis to pmHAS-WT and variant P6.H8 as previously stated. The recombination variants from pmHAS-WT and pmHAS-P6.H8 were independently screened twice (n=6; Figure 3-6). To ensure comparability of HA production, the enzyme-to-substrate ratio was maintained. On average, between 1.2-1.7-fold improvement in HA titer compared to the wild type was observed for the single, double and triple substitution variants. No correlation could be established between the type or number of substitutions and the improvement in HA synthesis by the variants. The pmHAS-P6.H8-T40L (pmHAS-VF as white bars), however, showed a 1.9-fold improvement. This was corroborated by the second sample set that showed a 2.1-fold improvement in HA titer, the highest improvement amongst all the recombination variants.

Single Mutants Double Mutants Triple Mutants

Figure 3-6: Phase IV - Recombination of beneficial substitutions in pmHAS-WT and pmHAS-P6.H8. The substitutions (T40L, V59L, T104L and I175N) selected in Phase III of KnowVolution were introduced to pmHAS-WT to generate four single, six double and four triple mutants. In parallel, a T40L and I175N mutation (alone or combined) was added to the epPCR variant pmHAS-P6.H8 (V59M/T104A) previously discovered to produce more and longer HA chains than pmHAS-WT. After repeated CTAB screening, the variant pmHAS-P6.H8- T40L (pmHAS-VF) showed the best improvement (~2.0-fold over WT) as demonstrated by the two white bars. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

122

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer Due to the variations in standard deviation, pmHAS-VF could not yet be concluded as the best variant. A confirmatory experiment (n=8) comparing pmHAS-WT, epPCR variant P6.H8, pmHAS-VF and the respective individual substitutions was performed (Figure 3-7). Visual confirmation by agarose gel showed that the single substitution variants and pmHAS-WT produced HA approximately up to 3.5 MDa, epPCR variant P6.H8 with up to 4.0 MDa and pmHAS-VF producing up to 4.5 MDa HA. This verifies that pmHAS-VF produces longer HA chains than the wild type or any of the single substitution variants. Furthermore, regarding HA titer, the single variants V59M and T104A showed 1.5-fold and 1.6-fold improvement, respectively, while the T40L performed relatively comparable to pmHAS-WT. Variant P6.H8 and pmHAS-VF showed a 1.9-fold and 2.1-fold improvement, respectively. This confirms that pmHAS- VF yields twice more HA than the wild type synthase. Taken together, it was demonstrated that pmHAS-VF possesses two improved polymerizing properties i.e. product titer and chain length specificity.

Figure 3-7: Comparison of HA chain A length and titer produced by pmHAS- WT, pmHAS-VF and respective single substitution variants. E. coli BL21 GOLD (DE3) cells harbouring pET-22b(+)-pmHAS- WT and variants were cultivated in microtiter plates (n=8) and lysed followed by in vitro HA synthesis as previously described. (A) The HA products were resolved in 0.5 % agarose gel to visualize product chain length and polydispersity. Select-HA HI and Mega Ladder (Hyalose) were used as standards. (B) The same HA products were quantified through the CTAB turbidimetric assay with HA > 2.O MDa as the standard (GfN & Selco). The B improvement in absorbance over pmHAS- WT (grey bars) and the respective standard deviations were calculated. The HA concentrations were calculated using the the standard curve from Figure 3-3A. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

123

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.4.3. Characterization of pmHAS-VF and pmHAS-WT with Respect to HA Titer and Polymer Length In order to compare the obtained improvements under screening conditions, pmHAS- VF (T40L/V59M/T104A) was ultimately selected for comparison against pmHAS-WT. HA production by pmHAS-WT and pmHAS-VF were performed in a 96-well MTP format. Densitometric analysis using ImageJ (Figure 3-8; Appendix Figure A2), reveal a protein concentration of 0.40 mg/mL (5.73 mg enzyme in 14.40 mL reaction) for pmHAS-WT, compared to 0.27 mg/mL (3.86 mg enzyme in 14.40 mL reaction) for pmHAS-VF. On the other hand, CTAB quantification (Table 3-4) of the in vitro synthesis HA products suggest an HA concentration of 0.143 mg/mL (2.06 mg in 14.40 mL) produced by pmHAS-WT and 0.216 mg/mL (3.11 mg in 14.40 mL) by pmHAS-VF.

The mass-based turnover number, TTNmass, or total turnover number (TTN) [222] was determined by dividing the amount (mg) of HA catalyzed per mg of pmHAS in the synthesis reaction. Taken together, pmHAS-WT has a TTNmass of 0.36 mg HA per mg enzyme, while pmHAS-VF has TTNmass of 0.80 mg HA per mg enzyme under applied conditions, overall a 2.2-fold improvement over wild type.

The HA products were resolved on 0.5 % agarose gel for visual inspection (Figure 3-8B). The smears on the agarose gel reflect the polydispersity of the HA products. Staining with the carbocyanine-based dye Stains-All show that pmHAS-VF produces longer HA chains than pmHAS-WT. ImageJ calculations reveal a molecular weight range of 0.4-3.5 MDa and 0.4-4.7 MDa for pmHAS-WT and pmHAS-VF, respectively, representing an improvement in polymer elongation of up to 1.2 MDa, albeit at the expense of polydispersity (Appendix Figure A-2).

124

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

Figure 3-8. Determination of pmHAS expression and HA production levels. (A) Comparing the expression of pmHAS-WT and pmHAS-VF “normalized” to lysozyme concentration. Dilutions of pure lysozyme (from 3.00 mg/mL to 0.06 mg/mL) were used for the calibration curve (ImageJ, Appendix Figure A-3) to quantify pmHAS. The expected molecular weight of pmHAS is 82 kDa, while lysozyme is 14 kDa. (B) Comparing HA production by pmHAS-WT and pmHAS-VF. The in vitro HA synthesis products were normalized to 0.1 mg/mL and resolved in 0.5 % agarose gel at 90 V for 2.5 h. HA was visualized using Stains-All. Select-HA HI and Mega Ladder (Hyalose, amsbio, Frankfurt, Germany) were used as standards. A standard curve of the migration distance of each band and their corresponding molecular weight (ImageJ, Appendix Figure A-2) was plotted to calculate the molecular weight range of the in vitro synthesized HA. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

Table 3-4: Quantification of HA produced by pmHAS-WT and pmHAS-VF in microtiter plate. Enzymatic conversion by pmHAS-WT and pmHAS-VF were performed in 96-well MTP. HA products were pooled together to a total of approximately 15 mL. The HA products were quantified using the CTAB assay. For CTAB controls, PBS pH 7.4, in vitro synthesis cocktail and the lysate from empty vector control culture were used. The linear regression of y = 0.003x + 0.0414 (Figure 3-3A) was used to calculate the corresponding concentration of the HA products. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

Average Standard Percentage Avg- Concentration Sample (A400) Deviation Error IVS (µg/mL) PBS 0.043 0.0012 2.66 -0.015 -19 IVS 0.059 0.0015 2.60 0.000 -14 (Ref) EV 0.055 0.0015 2.76 -0.003 -15 WT 0.532 0.0113 2.12 0.473 143 VF 0.753 0.0326 4.33 0.694 216

125

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

3.4.4. Structural Insight into the Improved Polymerizing Activity of pmHAS Figure 3-9A shows the homology model of pmHAS-WT. The homology model generated was further evaluated against the protein geometry by using different tools of SAVES [223]. PROCHECK identifies only 3 amino acids having stereo chemical clashes with respect to Ramachandran plot. But not all of them were neither part of the active site nor the binding domain. The Ramachandran plot highlighted that there are 89.4 % residues in the most favored region, 10.1 % in the allowed region, 0.3 % in the generously allowed region, and 0.2 % in the disallowed region, while ERRAT plot also exhibited a 93.15 overall quality factor. The ß factor value calculated with YASARA was 2.6. Consequently, ProSA displayed quite a similar model quality (Z-score) for homology model (−10.01) and template (−9.16).

A model of the core pmHAS-WT72- 688 has already been reported by Kooy and coworkers [109]. In our model, we completed the C-terminal portion of the core synthase (up to residue 703) and improved the N-terminal domain by further extending by 40 amino acid residues. The core part of our structural model of pmHAS-WT32-703 is similar to the one previously reported by Kooy and coworkers and retains the consensus topology of the template and other GT-A proteins. The topology of all GT-A is highly conserved [224]. As described by Romero-García et al,[224] the consensus topology of secondary structure elements is formed by seven β-strands that adopt an extended and twisted β- sheet flanked by three α-helices at each side of the β-sheet platform. The conserved DXD motif, necessary for co-ordinating divalent cations and/or a ribose ring [72], is located in the center between β-strand 4 and α- helix 4. Interestingly, β-strands 5 and 7 cross each other in the structure; this allows the formation of a parallel platform of β- strands that extends up to the DXD motif [107, 212, 224].

The pmHAS-VF model (T40L/V59M/T104A) was also generated using FoldX method based on the homology model of pmHAS-WT. Changes in the Gibbs free energy (ΔΔG) values for amino acid substitution in units of kcal/mol were calculated with FoldX 3.0.

126

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer All the substitutions were stabilizing (T40L: -0.19, V59M: -0.38, T104A: 1.84. Values less than 2 kcal/mol are considered as stabilizing).

In order to generate hypotheses regarding the structural changes leading to the substitutions in pmHAS variants, MD simulation studies of pmHAS-WT and pmHAS- VF were performed in 0.9 % (0.15 M) and 7.5 % (1.3 M) NaCl. The resulting 95 ns trajectories were analyzed, focusing on the environment of substituted amino acids (T40, V59, and T104). Experimental results show that the pmHAS-VF produces longer HA than pmHAS-WT. The amino acids T40 and V59 are located in the variable region suggesting a more flexible loop (amino acid residues 32-68; Figure 3-9B).

Figure 3-9: Molecular modeling of pmHAS. (A) Structural homology model of pmHAS- WT. The notable residues improved in the directed evolution campaign (T40, V59 and T104) are shown in green. The other five positions are shown in blue, while the active (247-DCD-249 and 527- DSD-529) and binding sites (196-DGS-198 and 477-DGS-479) of both domains are highlighted in red and orange, respectively. (B) Root Mean Square Fluctuation (RMSF) of amino acid residues in pmHAS-WT and pmHAS-VF during MD simulations. The blue line represents pmHAS-WT, while the grey line represents pmHAS-VF. Reproduced from [1] with permission from © Wiley- VCH Verlag GmbH & Co. KGaA.

127

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer

Figure 3-10: Molecular dynamics simulation snapshots of pmHAS-WT (blue) and pmHAS-VF (grey). Secondary structural elements of the N-terminal region are indicated by ribbons for α-helices. The residues improved by SDM (T40L, V59M and T104A) are highlighted in green. The active and binding sites of both transferase domains are highlighted in red and orange, respectively. After 95 ns of MD simulations of pmHAS-WT and pmHAS-VF, the structures maintain the global fold. No important conformational changes have been detected in the conserved region. Only with pmHAS-VF, the variable region is observed to flip from one domain to the other. Furthermore, the root mean square fluctuations (RMSF) of the substitutions show an improved flexibility of the N-terminus of pmHAS-VF but not of pmHAS- WT. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

The structural stability of both pmHAS-WT and pmHAS-VF models in water were estimated by calculation of the root mean square deviation (RMSD) from the initial structure in MD simulations. The RMSD values of pmHAS-WT and pmHAS-VF rapidly reached about 0.43 nm and 0.91 respectively in the first 40 ns and reached equilibrium at 0.64 nm and 0.85 after 95 ns equilibration. After 95 ns, the simulations were stable, no more changes in secondary structure were observed; the structures kept the global fold and essentially, all the secondary structure elements of the conserved region obtained by the previous homology modeling. No important conformational changes were detected in the conserved region.

128

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer Taken together, the MD simulations show that the substitutions (T40L/V59M/T104A) in the N-terminal part of the pmHAS-VF lead to increased flexibility (Figure 3-9B). With the improved flexibility (Figure 3-10), the N-terminal domain appears to interact with the GlcNAc-transferase domain within the first 10 ns of simulation and then immediately swings away towards the GlcA-transferase domain.

3.5. Discussion

With the fast-evolving pace of protein engineering, various techniques have been developed to improve properties of proteins. KnowVolution was used in this study to improve the TTNmass and chain length specificity of pmHAS, albeit at the expense of higher polydispersity. With KnowVolution, screening efforts were minimized, while gaining molecular insight into the improved enzymatic properties [175]. KnowVolution has been successfully applied to enzymes like glucose oxidase for decreased oxygen dependency [225], phytase for improved thermal and pH stability [226] and cellulose for increased specific activity in eutectic solvents [227]. Other successful applications of KnowVolution have already been catalogued [175]. In this study, pmHAS was improved through a KnowVolution campaign to yield a variant with a 2.2-fold improvement in

TTNmass (improved production from 0.143 g/L to 0.216 g/L HA) and chain length specificity (from 0.4-3.5 MDa to 0.4-4.7 MDa).

The CTAB turbidimetric assay was used to screen for improved titer, in combination with agarose gel electrophoresis for improved polymer length. A linear sensitivity of up to 0.3 mg/mL HA with CTAB was established with a true co-efficient of variation of 16 %, thereby satisfying the acceptance criterion of 20 % or less for a high-throughput screening assay [228]. Furthermore, a hit was stringently identified for having at least a 1.5-fold improvement in titer over wild type. Other groups have employed the CTAB assay in a 96-well microtiter plate format [104, 178], reiterating its suitability as a screening system platform for the pmHAS KnowVolution campaign.

129

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer In this investigation, pmHAS was heterologously expressed by E. coli producing up to 0.216 g/L HA. Due to the wide discrepancies in previous reports pertaining to HAS isoform, host/expression systems, HA synthesis conditions, the use of metabolic and/or process engineering, and HA detection/quantification method, it is difficult to compare our results with previous reported data. HA production by recombinant E. coli have been documented to range between 0.0325 g/L [229] to 0.19 g/L with up to 2 MDa in molecular weight (gel filtration chromatography) [70], which are in close alignment with our results. The GRAS Agrobacterium sp. ATCC 31749 expressing pmHAS and kfiD has been reported to produce 0.30 g/L of HA (carbazole assay) with a molecular weight range of 0.7-2.0 MDa (aqueous size exclusion chromatography) [230]. Fermentative HA production by the food-grade S. thermophilus expressing both hasA and hasB produced 1.2 g/L HA (HA binding protein assay) with a molecular weight of 1 MDa (size exclusion chromatography) [134]. In comparison with other GRAS strains, metabolically engineered E coli JM109 cells expressing pmHAS reported an HA production of 3.8 g/L (carbazole assay) with a molecular weight of 1.5 MDa (aqueous size exclusion chromatography) by fed-batch fermentation [66], while metabolically engineered B. subtilis strain TPG223 co-expressing pmHAS/tuaD–gtaB yielded 6.8 g/L HA (modified carbazole assay) with a molecular weight of 4.5 MDa (high performance size exclusion chromatography) [136]. This nearly rivals that of streptococcal fermentation which can yield up to 7 g/L HA (capillary viscometry) [132].

With respect to size, pmHAS-VF produced HA polymers up to 4.7 MDa, representing an improvement in chain length specificity by up to 1.2 MDa. To the best of our knowledge, this is the longest HA polymer generated by enzymatic conversion by an engineered pmHAS variant. In general, HA produced by eukaryotic cells are longer compared to microbial production. For instance, the mammalian HAS2 expressed in rat 3Y1 fibroblast generated HA that exceeds 2 MDa (gel filtration chromatography) [79]. DG42 from Xenopus laevis was reported to synthesize HA in vitro with an average size range of 6-12 MDa (gel filtration chromatography) [9]. More recently, HA produced by Heterocephalus glaber were reported to range from 6-12 MDa in molecular weight (pulse field gel electrophoresis) [231]. While mammalian cells have the capability to produce high molecular weight HA, HA titer is compromised.

130

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer Thus far, only one directed evolution campaign for Class I szHAS expressed in B. subtilis has been reported [104]. In this work, the engineered intracellular domain of szHAS improved HA production from 1.22 g/L to 2.24 g/L (CTAB assay) and chain length modestly from 1.20 to 1.36 MDa (high-performance size-exclusion chromatography coupled with multi-angle laser-light scattering). In tandem with metabolic engineering, the upregulation of tuaD (encoding UDP-glucose dehydrogenase) and glmU (encoding acetyl-transferase and UDP-GlcNAc pyrophosphorylase) further improved the titer to 2.8 g/L and the molecular weight to 2.6 MDa. The improvement in production was attributed to perturbations in the β-1,3 and β-1,4 transferase sites, while the improvement in chain length was associated with the introduction of positively charged residues (R197, H203, H297 and K300) to the HA binding site facilitating HA-HAS interaction.

In order to gain molecular understanding on the improvements in the polymerizing activity of pmHAS, structural models of pmHAS-WT32-703 and pmHAS-VF32-703 were generated. Visual inspection of the models revealed that the final substitutions (T40L, V59M and T104A) belong to the N-terminal variable region of pmHAS formed by three alpha-helices and two loops (Figure 3-9A), considerably distant from either of the two glycosyltransferase active sites. This eliminates the possibility that these substitutions are directly involved in enzyme catalysis. Furthermore, the screening data (Figure 3-7) demonstrated that pmHAS variants harbouring the individual substitutions T40L, V59M and T104A produced lesser and relatively shorter HA products compared to pmHAS-VF. This, therefore, implicates the additive effect of the substitutions to the improvement in pmHAS activity.

Various reports have shown that dynamic loop flexibility can modulate carbohydrate binding and catalysis in GT-A enzymes regardless of their mechanism (inverting or retaining) as reviewed by Urresti and colleagues [232]. Similarly, it has been shown that the catalytic activity of β-1,4-galactosyltransferase is modulated by the conformational change brought forth by its two flexible loops [233]. Interestingly, T40L is a part of the HAS tandem-motif region described to be involved in both the control of HA synthesis rate and the control of HA size for seHAS [105]. Furthermore, molecular dynamics

131

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer simulations revealed that the increase in flexibility caused by the substitutions result in the N-terminal region to first interact with the GlcNAc-transferase and then turn towards the GlcA-transferase domain of pmHAS (Figure 3-9B; Figure 3-10). This is only observed with the variant pmHAS-VF but not with the wild type. This may also imply that HA is released faster from the enzyme binding sites. This would have an impact on the processivity of HA chains by producing more HA of higher chain length, as evidenced by the increase in polydispersity of the HA products. Thus, the results of this protein engineering campaign suggest that the N-terminal domain may contribute to HA synthesis. The latter is in misalignment with the pmHAS N-terminal truncation experiment by Jing and DeAngelis who showed that residues 1-117 of pmHAS are not required for functional transferase and polymerizing activity [107].

One also cannot rule out the effect of HAS-HA interaction on product size determination [65]. In a site-directed mutagenesis experiment involving HAS from Xenopus laevis [234], perturbation of one amino acid residue produced HA chains of varying sizes. In particular, increasing salt concentrations produced higher molecular weight HA. This suggests that the high salt concentration forces interaction between the hydrophobic amino acids of pmHAS and the hydrophobic faces of the HA chain, leading to HA retention at the enzyme. Applied to pmHAS, two polar threonine residues were substituted to hydrophobic leucine and alanine (T40L, V59M and T104A), thereby increasing the hydrophobicity of pmHAS. It lends support to the theory that alteration in the strength of the catalyst-polymer interaction may influence the control of HA polymerization.

In summary, this is the first successful directed Class II pmHAS evolution campaign in which the HA chain length specificity was increased (up to 1.2 MDa), TTNmass was doubled, and a potential role of the N-terminus in chain length specificity was discovered.

132

Chapter 3: Directed Evolution of Class II HAS from Pasteurella multocida Towards Improved HA Chain Length and Titer 3.6. Conclusion

Directed evolution can improve pmHAS properties i.e. improved titer and chain length specificity. Interestingly, the molecular dynamics simulations show that the substitutions (T40L/V59M/T104A) in the N-terminal part of the pmHAS-VF lead to increased flexibility. With the improved flexibility, the N-terminal domain appears to interact with the GlcNAc-transferase domain and then immediately swings away towards the GlcA-transferase domain. The N-terminal flexible region of pmHAS was discovered to play a role in improving the TTNmass and chain length of HA. The KnowVolution protein engineering strategy can be applied to improve properties of other HA synthases. Protein engineering has the potential to complement process and metabolic engineering efforts by offering possibilities to redesign properties of pmHAS.

133

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

4. Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS “Millions saw the apple fall, but Newton asked why” - Bernard Baruch

4.1. Declaration

Portions of this chapter are intended for publication in the journal, Microbial Cell Factories (or similar) entitled, “A facile and inexpensive toolbox for differential hyaluronic acid synthesis: From kilodalton to megadalton scale”.

4.2. Project Objective

With the success of the pmHAS KnowVolution campaign, a final variant capable of synthesizing up to 4.7 MDa of HA with a 2.2-fold improvement in mass-based turnover number was successfully achieved. This protein engineering feature proves to be an addition to the HA production toolbox predominantly involving metabolic and/or process engineering.

HA synthesis is achieved either by natural HA producers (Group A [3] and Group C Streptococci [235] and Pasteurella multocida [236]) or by heterologously-expressed HAS in non-native HA producing hosts [68, 133, 134]. The evolution of HA production has shifted over the last decades from extraction from animal sources to microbial production. The emergence of metabolic engineering and process engineering have fuelled further advancement in HA production. More than ever, methodologies have been established to produce HMW HA [136], LMW HA [237, 238] and with higher yield [132, 239] to address the growing needs and global demands for HA, owing to its diverse properties.

To further gain understanding into the control of HA polymerization, this chapter dissects deeper into various methods of HA production. Synthesis using purified

134

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS pmHAS, production with crude cell lysates from physical or enzymatic cell disruption and a deeper investigation into the influence of cell disruption on HA biosynthesis were explored. This study offers various methods by which HMW and LMW HA can be differentially synthesized.

4.3. Materials and Methods

Chemicals were obtained from AppliChem (Darmstadt, Germany), Sigma Aldrich Chemie (Taufkirchen, Germany) or VWR International (Darmstadt, Germany) unless otherwise stated.

4.3.1. Generation of pmHAS Constructs Both pmHAS constructs were generated from the work by Mandawe et al. [1]. Briefly, the gene encoding the truncated, soluble pmHAS1-703 [106] (GenBank accession number: AF036004.2, EC 2.4.1.212) was subcloned into pET-22b(+) vector using NdeI and NotI restriction sites. The resulting construct contains a C-terminal hexahistidine tag

(pmHAS1-703-His6) and is referred to as “pmHAS-WT”. The recombinant plasmid pET-

22b(+)-pmHAS1-703-His6 was transformed into chemically competent E. coli BL21 GOLD (DE3) cells (Agilent technologies, Santa Clara, USA) and was verified by sequencing. pmHAS-WT was subjected to one round of KnowVolution resulting in the final variant harbouring triple mutations (T40L/V59M/T104A). This improved variant has been termed, “pmHAS-VF” [1].

4.3.2. Cell Cultivation

4.3.2.1. Cultivation of Escherichia coli Cells Harbouring pET22b(+)±pmHAS variants in Shake Flask E. coli BL21 GOLD (DE3) cells harbouring the pET-22b(+)-pmHAS construct (or variant) were grown as a starter culture in 15 mL TBAMP (100 µg/mL; 37 °C, 24 h, 200 rpm). The inoculum was added at an initial OD600 of 0.05 to 200 mL Terrific Broth (TB;

1.2 % peptone (w/v), 2.4 % yeast extract (w/v), 0.4 % glycerol (v/v), 0.23 % KH2PO4

135

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

(w/v), 1.25 % K2HPO4 (w/v)) medium supplemented with ampicillin (100 µg/mL). Once

OD600 has reached between 0.6-0.8, protein expression was induced with IPTG to 1 mM and lowering the temperature to 25 °C. Cells were grown up to 24 h and harvested by centrifugation (4 °C, 20 min, 11 279 g) using Sorvall RC6 Plus (Thermo Fisher Scientific, Schwerte, Germany). The cell pellet was stored at -80 °C until further use.

4.3.2.2. Cell Cultivation and Protein Expression in 96-Well Microtiter Plate Colonies from generated libraries were inoculated into the starter 96-well MTP (PS-F- bottom, Greiner Bio-One, Frickenhausen, Germany) containing 100 µL Luria-Bertani broth supplemented with ampicillin to 100 µg/mL and cultivated to saturation (37 °C, 24 h, 900 rpm, 70 % humidity) in Multitron II Infors Shaker (Einsbach, Germany). Ten microliters of the pre-culture were used to inoculate the main 96-well MTP containing 150 µL TB in each well supplemented with ampicillin (100 µg/mL). Cells were grown (37

°C, 900 rpm, 70 % humidity) until OD600 0.6-0.8. After which, IPTG was added to 1 mM to induce protein production (25 °C, up to 24 h, 900 rpm, 70 % humidity). Expression cultures were harvested (4 °C, 15 min, 3220 g) in Eppendorf 5810R centrifuge (Eppendorf AG, Hamburg, Germany) and frozen at -80 °C until further use.

4.3.3. Purification of pmHAS-WT and pmHAS-VF by Nickel-NTA Affinity Chromatography E. coli cells expressing pmHAS (freshly harvested or previously pelleted) were resuspended in cold PBS pH 7.4 by vortexing and aspiration to a concentration of 50 mg/mL. PMSF protease inhibitor was added to 0.1 mM before sonication for 8 min at 60 % amplitude, 15 s ON, 30 s OFF using Vibra-Cell Ultrasonicator (Sonics Sonics & Materials, Inc, Newtown, Connecticut, USA). The sonication product was pelleted by centrifugation (4 °C, 20 min, 11 279 g) using Sorvall RC6 Plus (Thermo Fisher Scientific, Schwerte, Germany) and the supernatant was filtered through a 0.80 µm Whatman filter before passing the nickel column.

In a disposable 10 mL polypropylene column (Thermo Fisher Scientific, Schwerte, Germany), approximately 6 mL of resin (Ni sepharose 6 Fast Flow; GE Healthcare

136

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Europe GmbH, Freiburg, Germany) were pre-equilibrated using 5 column volumes of the binding buffer (50 mM Na2HPO4/NaH2PO4, 500 mM NaCl, 5 mM imidazole, pH 7.4). The filtered lysate was manually passed into the column, while collecting the flow through in 50 mL falcon tube. Two extra column volumes of the binding buffer were passed onto the column. This makes the “Flow Through”. Once completed, the contaminants were washed off the column with at least 10 column volumes of the wash buffer (50 mM Na2HPO4/NaH2PO4, 500 mM NaCl, 30 mM imidazole, pH 7.4). Finally, the his-tagged pmHAS was eluted slowly using the elution buffer (50 mM

Na2HPO4/NaH2PO4, 500 mM NaCl, 250 mM imidazole, pH 7.4). Fifteen microliter aliquots were taken at each stage of the purification process for SDS-PAGE analysis. The final eluate was dialyzed overnight (Spectra/Por 4 Dialysis Membrane 12-14 kDa MWCO; Thermo Fisher Scientific, Schwerte, Germany) in PBS buffer pH 7.4 (or in 50 mM Tris-Cl pH 7.2) and the buffer exchange was finalized with successive rounds (4 °C, 3220 g, in Eppendorf 5810R centrifuge) of Amicon filtration (Amicon Ultra-15 Centrifugal Filter Units, Merck KGaA Darmstadt, Germany) until the imidazole concentration was reduced to < 50 µM. The purification fractions and the purified enzymes were visualized in a 12 % SDS gel and quantified using the BCA Protein Assay (Novagen/Merck KGaA, Darmstadt, Germany) and immediately used for HA biosynthesis. The remainder were stored in +4 °C until further use.

4.3.4. Qualitative Analytical Methods

4.3.4.1. Visualization of HA Production by Agarose Gel Electrophoresis and Staining with Stains-All Fifteen microliters of HA products were mixed with 3 µL of 6 X DNA gel loading dye (Thermo Fisher Scientific, Schwerte, Germany) and applied to 0.5 % agarose gel for electrophoresis in 1 X TAE-buffer (50 X TAE-buffer pH 7.8: 242 g Tris base, 18.61 g

EDTA disodium salt, 57.1 mL glacial acetic acid in 1L ddH2O) at 100 V for 50-60 min. The agarose gel slab was soaked in 30 % ethanol for 1 h and stained in darkness for at least 8 h with Stains-All (0.1 mg/mL in 30 % EtOH). The stain was discarded and replaced with distilled water. The gel was exposed to white light for destaining and scanned with Canon Scan 5600F as TIF. GeneRuler™ 1 kb DNA ladder and Select-HA

137

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

HI and Mega Ladder (Hyalose, amsbio, Frankfurt, Germany) were used as the molecular weight standards.

4.3.4.2. Visualization of HA Production by Agarose Gel Electrophoresis and Staining with Roti-GelStain Five microliters of HA products were mixed with 1 µL of 6 X DNA gel loading dye (Thermo Fisher Scientific, Schwerte, Germany) and applied to 0.8 % agarose gel prestained with Roti-GelStain at 5 µL per 100 mL agarose (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) for electrophoresis in 1 X TAE-buffer at 100 V for 30 min. DNA was visualized under UV using the U: Genius3 Gel Imaging System (Syngene, Fisher Scientific, Schwerte, Germany)

4.3.4.3. SDS Polyacrylamide Gel Electrophoresis For SDS-PAGE, a 5 % (w/v) acrylamide stacking gel (720 µL ddH2O, 130 µL acrylamide mix (40 %), 130 µL 1.0 M Tris pH 6.8, 10 µL 10 % SDS, 10 µL 10 % APS and 1 µL

TEMED) and 12 % (w/v) acrylamide resolving gel (2.1 mL ddH2O, 1.5 mL acrylamide mix (40 %), 1.3 mL 1.5 M Tris pH 8.8, 50 µL 10 % SDS, 50 µL 10 % APS and 2 µL TEMED) were prepared as previously described [240].

Purification fractions or in vitro HA synthesis products were prepared by combining with 4x SDS-PAGE loading dye to 1 X and boiling at 100 °C for 5 min. Electrophoresis was initially executed at 90 V for 15 min, then 120 V for 1.5 h, until the dye front has reached the bottom edge of the gel. The gels were stained in Coomassie Brilliant Blue (R-250) solution (0.4 % (w/v) Coomassie, 50 % v/v methanol, 10 % v/v acetic acid) for 20 min followed by de-staining in fixing solution (30 % (v/v) methanol and 10 % (v/v) acetic acid) until minimal or no background stain is observed. The gel was visualized with the Canon Scan 5600F and saved as TIF. The PageRuler Prestained Protein ladder 10-180 kDa (Thermo Fisher Scientific, Schwerte, Germany) was used as reference protein ladder.

138

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

4.3.5. Quantitative Analytical Methods

4.3.5.1. Bicinchoninic Acid (BCA) Assay for Protein Quantification The BCA Protein Assay (Novagen/Merck KGaA, Darmstadt, Germany) was adapted to the microtiter plate format. Briefly, 25 µL of protein sample was added to 200 µL of the BCA working reagent (50 parts BCA solution: 1 part 4 % cupric sulfate). Proper mixing (room temperature, 30 s, 600 rpm) was ensured on the TiMix Shaker (Edmund Bühler, Hechingen, Germany), followed by the colorimetric reaction (37 °C, 30 min, no shaking). The samples were cooled to room temperature for 5 min and the absorbance (562 nm, 2-second shake, 5-second settle time) was measured with the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland). Protein concentrations were calculated using the prepared BSA standard curve.

4.3.5.2. ImageJ Analysis for Protein Quantification ImageJ was employed for quantification soluble proteins in the cell lysates [211]. To this end, the SDS gel containing the proteins was initially saved as TIF. The image type was converted to a 32-bit grey scale and inverted to have a black background and white bands. Using the “rectangle” selection tool, each column of protein bands was highlighted individually and consistently using the rectangular selection tool. The corresponding integrated densities for each column were measured. The net integrated density of each column was calculated by subtracting the background signal (blank column). The percentage of proteins lost was calculated using the following equation:

[integrated density(before) -integrated density(after)]/integrated density(before) * 100 %

4.3.5.3. CTAB Turbidimetric Assay for Hyaluronan Quantification The 96-well microtiter plate format CTAB turbidimetric assay was slightly modified [178]. This method depends on the formation of an insoluble complex between HA and cetyltrimethylammonium bromide, where the turbidity is linearly proportional to the quantity of HA in the system [179]. Briefly, 60 µL of HA (in vitro synthesis products or HA standard (Hyaluronic Acid Na-salt > 2.O MDa, GfN & Selco, Wald-Michelbach, Germany) were added to 20 µL of 0.1 M phosphate buffer, pH 7.0 and incubated (37 °C, 0 rpm, 15 min). Precipitation occurred upon addition of 160 µL of pre-warmed CTAB

139

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS reagent (2.5 g in 100 mL 2% NaOH). The end-point absorbance was measured within 5 min of the reaction (400 nm, 2-second shake, 5-second settle time) with the Tecan Sunrise microplate reader (Tecan Trading AG, Switzerland).

4.3.6. Hyaluronic Acid Synthesis

4.3.6.1. HA Biosynthesis with Purified pmHAS-WT and pmHAS-VF The standard HA biosynthesis conditions (37 °C, 24 h) using 200 µg/mL purified pmHAS-WT and/or pmHAS-VF utilizes the in vitro synthesis cocktail (final concentrations: 0.4 mM UDP-GlcA, 0.4 mM UDP-GlcNAc, 4 mM MgCl2, 0.4 mM DTT in PBS pH 7.4), unless otherwise stated. For the magnesium-manganese comparison experiment, the Tris-Cl pH 7.2 was used for MnCl2 due to incompatibility (precipitation) with the PBS buffer. For experiments requiring the synthesis initiator, HA4, (HA oligosaccharide 4mer: GlcA β 1→3(GlcNAc β 1→4 GlcA)1 β 1→3GlcNAc; Cosmo Bio Co. LTD, Tokyo, Japan) concentrations were taken into account, while keeping the reaction volumes constant.

4.3.6.2. HA Biosynthesis with Crude Lysate of Sonicated E. coli Cells Expressing pmHAS-WT or pmHAS-VF E. coli cells expressing pmHAS (freshly harvested or previously pelleted) were resuspended to 50 mg/mL in PBS buffer and sonicated in ice (60 % amplitude, 10 s ON, 20 s OFF, 5 min) using Vibra-Cell Ultrasonicator (Sonics Sonics & Materials, Inc, Newtown, Connecticut, USA), unless otherwise stated. The crude pmHAS lysate was clarified by centrifugation (4 °C, 20 min, 3220 g in Eppendorf 5810R centrifuge). In another clean tube, the clarified lysate was supplemented with the standard in vitro synthesis cocktail to allow for synthesis (37 °C, 24 h), unless otherwise stated. Whenever required, aliquots were taken at various time points, boiled for 2 min at 99 °C and immediately stored in the -80 °C freezer for future use. In vitro HA synthesis products were further processed depending on the experiment, otherwise, were resolved in 0.5 % agarose gel to visualize HA with Stains-All, in 0.8 % agarose gel for DNA staining Roti- GelStain or in 12 % SDS gel for protein visualization.

140

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

4.3.6.3. HA Biosynthesis with Crude Lysate (Freeze-Thaw and Lysozyme) of E. coli cells Expressing pmHAS-WT or pmHAS-VF E. coli cells expressing pmHAS (freshly harvested or previously pelleted) in microtiter plate were disrupted by two cycles of freeze (-80 °C, 10 min)-thaw (37 °C, 15 min) and lysozyme digestion (2 mg/mL in PBS pH 7.4; 37 °C, 1.5 h, 900 rpm). The microtiter plate was centrifuged (24 °C, 20 min, 3220 g in Eppendorf 5810R centrifuge). For synthesis in a new microtiter plate, the clarified lysates were supplemented with the standard in vitro synthesis cocktail. For synthesis in falcon tubes, the clarified lysates were pooled together and supplemented with the standard in vitro synthesis cocktail. HA synthesis was facilitated at 37 °C for 18 h, unless otherwise stated. For experiments requiring the synthesis initiator, HA4, (Cosmo Bio Co., LTD, Tokyo, Japan) concentrations were taken into account while keeping the reaction volumes constant.

E. coli cells expressing pmHAS (freshly harvested or previously pelleted) from shake flask were processed similarly as above, however, cells were first normalized to 50 mg/mL (or specific OD600).

4.3.7. Hyaluronan Digestion by Hyaluronidase The production of hyaluronic acid was confirmed by hyaluronidase digestion. In vitro synthesis HA products were digested by hyaluronidase (2 mg/mL in PBS or 1 U/µL, bovine testis HYAL; Sigma-Aldrich, MO, USA) at 1/10 the reaction volume. Quick digest was facilitated at 37 °C for 2 h, otherwise performed overnight.

4.4. Results

The results section is partitioned into four different parts: (1) enzymatic conversion of HA using purified pmHAS; (2) production of HA with crude cell lysate from sonicated or lysozyme-disrupted pmHAS-expressing E. coli cells; (3) investigation of the influence of cell disruption on HA product length; (4) provision of proof-of-principle for a cheap production of semi-purified HMW or LMW HA.

141

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

4.4.1. HA Biosynthesis with Purified pmHAS-WT and pmHAS-VF pmHAS-WT and pmHAS-VF were successfully purified via nickel affinity chromatography. Figure 4-1 shows the purification scheme from the crude extract to elution. A considerable protein band (fraction “E”) just under the 100 kDa marker is observed with the wild type, however, lower quantity but high purity fractions are obtained after dialysis. The purified enzymes were quantified using the BCA assay (Appendix Figure A-4).

Figure 4-1: Immobilized metal affinity chromatography purification of pmHAS- WT and pmHAS-VF. Clarified cell lysates were loaded onto the pre-equilibrated nickel- NTA column, washed and eluted with increasing imidazole concentration (5 mM, 30 mM and 250 mM, respectively). The purified fractions were collected and dialyzed overnight through 50 MWCO membrane in cold 50 mM phosphate buffer pH 7.4 or in 50 mM Tris pH 7.2. Aliquots were prepared and resolved in 12 % SDS-PAGE gel and stained with Coomassie Brilliant Blue (R-250). pmHAS is expected around 82 kDa. PageRuler Prestained Protein ladder (Thermo Fisher Scientific) was used C: crude lysate; F: flow through; W: wash fraction; E: pmHAS after dialysis in PBS pH 7.4; E‡: is pmHAS after dialysis in Tris pH 7.2.

To test the polymerizing capability of the synthases, separate reactions consisting of the precursors and enzymes were performed. It has been reported in literature that manganese is the preferred divalent metal ion for HA polymerization by pmHAS, more efficient than magnesium at lower concentrations [131, 241]. However, a conflicting report claims that magnesium is a more preferred metal ion by pmHAS [109]. The enzyme preparations were dialyzed respectively against Tris and PBS buffers to investigate the dependence on divalent cations. The effect of the metal ions was examined using 4 mM Mg2+ in PBS as buffer and 2 mM Mn2+ in Tris buffer. Under specified conditions, it is apparent that HA is produced only by pmHAS-VF and slightly larger and more LMW HA are produced in the PBS reaction system (Figure 4-2).

142

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Figure 4-2: Comparative enzymatic conversion by purified pmHAS-WT and pmHAS-VF in PBS and Tris buffer. HA synthesis with 200 µg/mL pmHAS in PBS pH 7.4 (containing 4 mM MgCl2) and Tris pH 7.2 (containing 2 mM MnCl2) with 4 mM each of UDP-GlcA and UDP-GlcNAc at 37 °C for 24 h. Products were resolved in 0.5 % agarose gel and stained with Stains-All. HA stains blue, while DNA stains purple. GeneRuler™ 1 kb DNA Ladder (Fermentas, St. Leon-Rot, Germany) was used as a marker and the corresponding sizes in kDa are specified in blue.

To understand the weaker polymerizing performance of the pmHAS-WT, a synthesis initiator (an HA tetrasaccharide: GlcA β 1→3(GlcNAc β 1→4 GlcA)1 β 1→3GlcNAc) was used. Williams and colleagues reported the improvement in HA production when using the HA4 acceptor [242]. Reactions using increasing amounts of the purified enzyme and supplemented with 1.6 µM HA4 acceptor were performed. According to Figure 4-3, 18 h of synthesis reaction was not sufficient to produce traceable amounts of HA, however, after three days, it is observed that LMW HA are produced when at least 75 µg/mL of the purified enzyme are used.

Taken together, these sets of experiments reveal that synthesis with magnesium as the divalent cation is sufficient to produce LMW HA. Moreover, HA polymerization is assisted by the addition of the HA4 acceptor and occurs at a very slow rate.

143

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Figure 4-3: HA4 acceptor-assisted HA biosynthesis by purified pmHAS-WT. HA synthesis with varying concentrations of pmHAS-WT in standard in vitro synthesis cocktail supplemented with 1.6 µM HA4 synthesis initiator (Cosmo Bio Co., LTD, Tokyo, Japan) was performed over three days at 37 °C. Products after 18 h and 72 h of synthesis were resolved in 0.5 % agarose gel and stained with Stains-All. HA stains blue, while DNA stains purple. GeneRuler™ 1 kb DNA ladder was used as a marker and the corresponding sizes in kDa are specified in blue.

4.4.2. HA Biosynthesis with Crude Lysate of pmHAS-Expressing E. coli A time course of HA production (Figure 4-4) using the lysates of sonicated, normalized E. coli cells expressing pmHAS shows production of LMW HA starting at t=0 h up to 18 h. The amounts of generated polymers appear to remain constant throughout the time course. To delineate the HA bands (blue) from nucleic acid smears (purple), samples were resolved in agarose gel and stained with Roti-GelStain. A time-dependent DNA degradation was also observed.

To determine how the pmHAS quantity influences HA synthesis, polymerization using decreasing amounts lysate volume of sonicated, normalized cells was performed. Figure 4-5 shows that production of LMW HA was already observed at t=0 h. Moreover, HA is produced starting with the lowest lysate concentration of 5 mg/mL and the smear steadily enlarges with increasing pmHAS content. After 18 hours of synthesis, identical pattern in HA production is observed, however, DNA was completely degraded. These results were corroborated by the repeated experiment (Figure 4-5).

144

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Figure 4-4: Time-dependent HA biosynthesis with crude lysate of E. coli cells expressing pmHAS-WT. Cells expressing pmHAS-WT were resuspended to 50 mg/mL in PBS buffer and sonicated (60 % amplitude, 15 s ON, 15 s OFF, 5 min). The crude pmHAS lysate was supplemented with the in vitro synthesis cocktail to prompt HA production. Aliquots were taken at various time points and boiled for 2 min at 99 °C and immediately stored in the freezer. Samples were resolved in 0.5 % agarose gel to visualize HA with Stains-All, and in 0.8 % agarose gel for Roti-GelStain (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) staining to visualize nuclei acids. GeneRuler™ 1 kb DNA ladder was used as a marker and the corresponding sizes in kDa are specified in blue.

Figure 4-5: Concentration-dependent HA biosynthesis with crude lysate (sonication) of E. coli cells expressing pmHAS-WT. Cells expressing pmHAS-WT were resuspended to 50 mg/mL in PBS buffer and sonicated (60 % amplitude, 15 s ON, 15 s OFF, 10 min). The crude pmHAS lysate clarified by centrifugation and various dilutions (0-45 mg/mL) were supplied with the in vitro synthesis cocktail separately to facilitate synthesis. Aliquots were taken immediately and after 18 h. For t = 0 h, samples were immediately boiled at 99 °C for 10 min and stored at -80 °C. Products were resolved in 0.5 % agarose gel and stained with Stains- All to visualize HA. GeneRuler™ 1 kb DNA ladder was used as a marker and the corresponding sizes in kDa are specified in blue. In the second attempt, Select-HA Mega Ladder (Hyalose, amsbio, Frankfurt, Germany) was used as the standard. Insert in Attempt #2: pmHAS-WT dilution expression profile in 12 % SDS gel.

145

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

A similar concentration-dependent HA biosynthesis experiment was performed but cell lysates were obtained from tandem freeze-thaw and lysozyme treatment of pmHAS- expressing E. coli cells. Contrary to sonication, highly polydispersed HMW HA (from 0.5 to > 6.1 MDa) were synthesized (Figure 4-6). Moreover, HA quantity and chain length decrease proportionally with the cell lysate volume. The SDS gel shows the proportionate amounts of lysozyme and pmHAS in each reaction.

Figure 4-6: Concentration-dependent HA biosynthesis with crude lysate (freeze- thaw and lysozyme) of E. coli cells expressing pmHAS-WT. Cells expressing pmHAS- WT in microtiter plate were disrupted by two rounds of freeze-thaw and 1.5 h lysozyme digest. The microtiter plate was centrifuged and the lysates were pooled together and diluted accordingly before adding the in vitro synthesis cocktail. HA synthesis was facilitated at 37 °C for 18 h. Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Both GeneRuler™ 1 kb DNA ladder and Select-HA Mega Ladder were used as the standard. Concomitantly, samples were resolved in 12 % SDS to demonstrate the corresponding protein dilution. pmHAS and lysozyme are expected around 82 kDa and 14 kDa, respectively.

146

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

To determine whether the HA4 is beneficial for HA synthesis with cell lysate from the freeze-thaw lysozyme disruption, syntheses with varying amounts of HA4 (0-40 µM) were performed. Figure 4-7 shows the identical, highly polydispersed HMW HA polymers irrespective of the amount of the initiator supplied to the reaction. Moreover, the SDS gel proves the consistent amounts of lysozyme and pmHAS present in each reaction. Taken together, these experiments showed that HA synthesis using crude cell lysate is superior to HA synthesis with purified pmHAS.

Figure 4-7: HA4-independent HA biosynthesis with crude lysate (freeze-thaw and lysozyme) of E. coli cells expressing pmHAS-WT. Cells expressing pmHAS-WT in microtiter plate were disrupted by two rounds of freeze-thaw cycles and 1.5 h lysozyme digest. The microtiter plate was centrifuged and the lysates were pooled together. In separate synthesis reactions (37 °C, 18 h), increasing amounts of HA4 (0-40 µM) were added alongside the in vitro synthesis cocktail. Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Both GeneRuler™ 1 kb DNA ladder and Select-HA Mega Ladder were used as the standard. Concomitantly, samples were resolved in 12 % SDS to demonstrate the protein content for each reaction. pmHAS and lysozyme are expected around 82 kDa and 14 kDa, respectively. PageRuler Prestained Protein ladder (Thermo Fisher Scientific, Schwerte, Germany) was used.

147

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Due to the stark contrast in production outcome by the crude and purified pmHAS, it was hypothesized whether other components in the cell lysate contribute to HA synthesis. Various synthesis combinations using crude cell lysate from sonicated pmHAS-WT (and empty vector control), purified pmHAS-VF and supplementation with in vitro synthesis cocktail were prepared as outlined in Figure 4-8. The agarose gel shows that the IVS solution alone, cell lysate from pmHAS-deficient E. coli cells alone and purified pmHAS-VF alone did not produce HA. Surprisingly, the lysate from pmHAS-expressing E. coli cells alone was able to produce LMW HA, identical to when E. coli cells are subjected to sonication as previously mentioned. When the purified pmHAS-VF was supplemented with IVS solution, no HA production was observed after 18 h. When the pmHAS-WT lysate was supplemented with IVS solution, HA was also produced at a comparable size and quantity as the control. HA production was also observed when the lysate from pmHAS-WT was combined with the purified pmHAS-VF and/or the lysate from the empty vector control, however, at lower amounts. Finally, when the pmHAS-VF was supplemented with the lysate from the empty vector control and IVS solution, HA could still not be synthesized. Ultimately, this experiment reveals three important pieces of information: (1) production of HA with the cell lysate is more superior than with purified synthase; (2) components in the cell lysate do not participate in HA synthesis; and (3) pmHAS expression can direct E. coli cells to synthesize HA, in the absence of nucleotide sugar precursors.

148

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Figure 4-8: Investigating cellular components that may participate in HA synthesis. Separately, cells harbouring the pET-22b(+) empty vector and expressing pmHAS- WT were resuspended to 50 mg/mL, sonicated and clarified. Purified pmHAS-VF (200 µg/mL) was also used for HA synthesis (37 °C, 18 h). Various combinations synthesis components to the final volume of 500 µL were used as shown in the scheme above. Samples were resolved in 0.5 % agarose gel and stained with Stains-All. The GeneRuler™ 1 kb DNA ladder was used as a marker and the corresponding sizes in kDa are specified in blue.

4.4.3. Investigating the Influence of Cell Disruption on HA Biosynthesis To further understand the disparity in HA production by cells treated by sonication or freeze-thaw and lysozyme digestion, a criss-cross experiment was executed. pmHAS- expressing E. coli cells were grown either in a shake flask or microtiter plate, disrupted either by sonication or freeze-thaw/lysozyme and HA syntheses were performed either in a microtube or microtiter plate. Figure 4-9 shows that the mode of cultivation and vessel for HA production have no influence on HA synthesis. The cell disruption method, however, is critical. pmHAS from cells that have been sonicated could only produce LMW HA (<0.5 MDa), while pmHAS from cells that have been lysed by freeze- thaw and lysozyme had the capacity to produce HMW HA (>1.5 MDa). This experiment

149

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS was repeated and corroborated the results (Figure 4-9B). In the second experiment, the SDS gel also shows the presence of pmHAS and lysozyme in the synthesis reaction.

A B

Figure 4-9: Investigating the effect of cell cultivation, cell disruption and reaction vessel on HA biosynthesis. E. coli BL21 GOLD (DE3) cells expressing pmHAS-WT were cultivated in a shake flask or microtiter plate format as described in Materials and Methods. Cells were harvested, normalized to an OD600 of 10, pelleted and washed. To the first half, cells were disrupted by freeze-thaw and lysozyme digestion and the other half were sonicated. The lysates were centrifuged prior to addition of the in vitro synthesis cocktail respectively. (A) Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Both GeneRuler™ 1 kb DNA ladder and Select-HA Mega Ladder were used as the standard. (B) Repeated experiment. In parallel, in vitro synthesis products were also resolved in 12 % SDS to demonstrate the protein content for each reaction. PageRuler Prestained Protein ladder (Thermo Fisher Scientific, Schwerte, Germany) was used as protein ladder. pmHAS and lysozyme are expected around 82 kDa and 14 kDa, respectively. FT: freeze-thaw; L: lysozyme digestion; S: sonication; T: synthesis in microtube; M: synthesis in MTP.

To further investigate the influence of cell disruption, E. coli cells expressing pmHAS- WT were first treated by freeze-thaw and lysozyme followed by sonication of up to 60 % power (Figure 4-10). As expected, no HA production was observed with the empty

150

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS vector samples. pmHAS from the lysozyme preparation produced highly polydispersed HMW HA, while the sonication generated LMW HA. A finer observation of the HA bands shows that HA arising from those treated with 20% sonication power are slightly longer than those from 40 % and 60 %. Moreover, application of hyaluronidase to the HA products abolished the typical blue HA signals on the gel or fragmented the HMW commercial HA control to smaller polymers, confirming the production of HA.

Figure 4-10: Verifying the effect of cell disruption on HA biosynthesis. Separately, cells harbouring the pET-22b(+) empty vector and cells expressing pmHAS-WT were resuspended to 50 mg/mL. Cells were then subjected to two rounds of freeze-thaw cycle followed by 1.5-h lysozyme digestion at 37 °C. Cells were partitioned into four different tubes for subsequent sonication (X %, 10 s ON, 20 s OFF, 5 min). Crude lysates were clarified by centrifugation prior to addition of the in vitro synthesis cocktail (37 °C, 18 h). To half of each in vitro synthesis product, 50 µL of hyaluronidase (50 U bovine testis HYAL; Sigma-Aldrich, MO, USA) was added and quick digest was facilitated (37 °C, 2 h). Samples were resolved in 0.5 % agarose gel and stained with Stains-All. Both GeneRuler™ 1 kb DNA ladder and Select-HA Mega Ladder were used as standards.

151

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Figure 4-11: Exploring HA biosynthesis without supplementation of HA precursors and verifying the effect of cell disruption on HA biosynthesis. Separately, cells harbouring the pET-22b(+) empty vector, pET-22b(+)-pmHAS-WT and pET-22b(+)-pmHAS- VF were cultivated in TBAMP and induced with IPTG to 1 mM. At 0 h, 3 h and 20 h after IPTG induction, 2 mL aliquots were collected, pelleted and quickly frozen. Cell pellets were resuspended to 50 mg/mL with PBS. Half were subjected to two rounds of freeze-thaw cycle and 1.5-h lysozyme digestion and the other half were subjected to sonication. Commercial HA standard (Hyaluronic Acid Na-salt > 2.O MDa, GfN & Selco, Wald-Michelbach, Germany) was also treated accordingly for reference. Crude lysates were clarified by centrifugation. Only lysates from t = 20 h were subjected to hyaluronidase treatment (20 U bovine testis HYAL in 200 µL reaction; Sigma-Aldrich, MO, USA) for 2 h at 37 °C. Samples were resolved in 0.5 % agarose gel and stained with Stains-All. The GeneRuler™ 1 kb DNA ladder was used as a marker and the corresponding sizes in kDa are specified in blue. M: marker; HA*: untreated HA standard; HA: treated HA standard.

To address the possibility of the pmHAS-expressing E. coli cells autonomously producing HA, a cell expression time course and HA synthesis (without precursor supplementation) were performed. Figure 4-11 clearly reinforces the disparity in HA polymer size distribution previously observed with either sonication or lysozyme treatment. At t = 0 h of cell expression, no HA is produced from either treatment. However, after only 3 hours of expression, visible traces of HMW HA are observed with the lysozyme-treated cells and LMW HA with the sonication treatment. After 20 hours of expression, strong distinct HA signals are observed with both pmHAS-WT and pmHAS-VF with either treatment, despite the lack of UDP-GlcA and UDP-GlcNAc.

152

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Importantly, the commercial HA control (2.12 MDa) was subjected to the freeze-thaw and lysozyme digestion and no remarkable difference is observed after treatment. In contrast, the HMW HA control was converted to smaller HA fragments (<0.5 MDa) with the sonication treatment. HA production was again verified by hyaluronidase digestion. Taken together, results show once again that HA was synthesized by the E. coli expressing pmHAS, without supplying the HA precursors.

4.4.4. Proof-of-Principle for Differential HA Production

The performance of E. coli cells expressing pmHAS-WT and pmHAS-VF were compared and the effect of supplementing UDP-sugars to the synthesis pool was investigated. Figure 4-12A, reveals the following with regards to pmHAS-WT: There is no substantial difference in HA products when the cells are sonicated. LMW HA are produced but thicker smears are observed when supplemented with HA precursors. A clearer distinction is observed in the lysozyme-treated pmHAS-WT expressing cells. As previously observed, polydispersed HMW HA are produced. The addition of the precursors prompted production of slightly longer HA chains. When the samples are boiled, signs of HA degradation appear and causing the HA samples to be identical in HA size range and quantity. With respect to the lysozyme-treated pmHAS-VF expressing cells, HMW HA are produced. With the addition of the UDP-sugars, elongation in HA chain length is observed. Boiling of the samples resulted in slight degradation causing the HA products to have similar size range and quantity. When sonicated, a majority of the HA pool are of the LMW range. However, unlike the wild type counterpart, HA polymerization by pmHAS-VF continue to occur at the HMW range as observed with the faint blue smears.

153

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Figure 4-12: Differential HA biosynthesis. Separately, cells harbouring pET-22b(+)- pmHAS-WT and pET-22b(+)-pmHAS-VF were cultivated as described in Materials and Methods and resuspended to 50 mg/mL with PBS. Half were subjected to two rounds of freeze- thaw cycle and 1.5-h lysozyme digestion and the other half were sonicated. Crude lysates were clarified by centrifugation. Half of the samples were supplemented with the standard in vitro synthesis cocktail and half were not, prior to HA synthesis (37 °C, 20 h). The synthesis products were further partitioned in halves, where one set were submerged in boiling water bath for 5 min and the other was not. (A) Samples were resolved in 0.5 % and 0.8 % agarose gel for HA and nucleic acid visualization, respectively. The GeneRuler™ 1 kb DNA ladder was used as a marker and the corresponding sizes in kDa are specified in blue. M: marker; -: unboiled synthesis products; +: boiled HA products. (B) In parallel, samples were loaded onto 12% SDS to demonstrate the protein content after each treatment. PageRuler Prestained Protein ladder (Thermo Fisher Scientific, Schwerte, Germany) was used for protein ladder. pmHAS and lysozyme are expected around 82 kDa and 14 kDa, respectively.

154

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

Table 4-1: Differential hyaluronan biosynthesis

ImageJ§ IVS HA † BCA % (Net % Sample Treatment solution Boil (µg/mL) (mg/mL) Lost Integ. Lost added Density) - - 184±6 1.19 466319 N/A N/A Freeze- - + 182±5 N/A N/A Thaw

+ Lysozyme + - 114±1 1.22 424453 WT - -96 -91 + + 119±3 0.05 40158

- - 201±3 0.48 615179 22b(+) - -76 -79 - + 210±7 0.12 127882

pET Sonication + - 165±4 0.85 739364 -80 -79 + + 172±4 0.17 157332

- - 240±7 1.40 283355 -83 -81 Freeze- - + 237±6 0.24 53023 Thaw

VF + Lysozyme + - 183±7 1.32 300709 - -97 -93 + + 197±6 0.04 21200

22b(+) - - 245±7 0.18 382014 - -67 -67 - + 246±9 0.06 88172 pET Sonication + - 190±9 0.52 463739 -74 -74 + + 196±6 0.13 118541 † HA concentrations were determined by CTAB turbidimetric assay § Protein quantification was performed using ImageJ based on the integrated density of the protein bands

The DNA profile of the in vitro synthesis reactions (Roti-GelStain) reveal that no DNA is observed when the cells are treated with freeze-thaw and lysozyme, however, varying amounts are observed when sonicated. This suggests that the lysozyme treatment is the “cleaner” method for HA synthesis and does not promote DNA shearing. The SDS gels in Figure 4-12B show the effects of boiling the reaction samples. The untreated (-) lanes

155

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS represent the crude lysates, while the (+) lanes show how much proteins have been precipitated out of the solution.

To quantify the amount protein lost from boiling, BCA assay and ImageJ analysis were performed (Table 4-1). Depending on the treatment, from 67-97 % of the proteins were removed simply by boiling the in vitro synthesized HA products for 5 minutes. Moreover, protein quantitation by BCA and ImageJ are in close agreement with each other and therefore provides confidence in the protein quantification method.

To quantify the amount of HA produced, the CTAB assay was performed (Table 4-1). When supplemented with UDP-sugars, the concentration of HA generally decreases by approximately 25 % depending on the treatment, however, the chain length is improved as seen in the agarose gels. Boiling has little effect on HA quantity when considering the standard deviations. It is also important to highlight that pmHAS-VF generally produces more HA polymers compared to pmHAS-WT (183-246 mg/mL vs. 114-210 mg/mL).

To apply these findings, HA were synthesized from lysates of E. coli cells expressing pmHAS-WT or pmHAS-VF. HA synthesis was supplemented with the standard in vitro synthesis cocktail to promote synthesis of longer glycosaminoglycan polymers. In Figure 4-13A, the agarose gel shows production of < 6. 1 MDa HA by pmHAS-VF compared to < 4.6 MDa by pmHAS-WT. CTAB quantification showed a concentration of 149±3.9 mg/mL and 102±0.8 mg/mL for pmHAS-VF and pmHAS-WT, respectively (Table 4-2). As expected with the freeze-thaw and lysozyme treatment, the presence of nucleic acid could not be detected (Figure 4-13B). Protein quantification by ImageJ analysis revealed a residual protein of 4.8% and 4.0% for pmHAS-VF and pmHAS-WT, respectively, predominantly arising from the residual lysozyme in solution (Table 4-2). Taken together, we have shown production of semi-purified HMW HA by minimal supplementation of HA precursors and boiling the samples for 5 min.

156

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

A B

Figure 4-13: HA biosynthesis and purification. pmHAS-WT and pmHAS-VF were expressed in E. coli cells in MTP (1 mM IPTG, 25 °C, 20 h, 900 rpm). Cells were harvested and subjected to two rounds of freeze-thaw cycle and 1.5-h lysozyme digestion. Crude lysates were clarified by centrifugation and supplemented with the in vitro synthesis cocktail to allow for HA synthesis (37 °C, 20 h). The synthesis products were pooled together in 50 mL falcon tube and exposed to boiling water bath for 5 min. The mixtures were cooled to room temperature prior to centrifugation. The clarified supernatants were analyzed. In vitro synthesis products were resolved in 0.5 % and 0.8 % agarose gel for staining with Stains-All (A) and Roti-GelStain (B), respectively. Both GeneRuler™ 1 kb DNA ladder and Select-HA Mega Ladder were used as standards. In parallel, samples were loaded onto 12 % SDS for protein quantification. PageRuler Prestained Protein ladder (Thermo Fisher Scientific, Schwerte, Germany) was used. Expected sizes: pmHAS: 82 kDa; lysozyme: 14 kDa.

Table 4-2: HA biosynthesis and purification

Residual DNA Residual Protein§ HA† Synthase (µg/mL) (%) (µg/mL) pmHAS-WT Not traceable 4.0 102 ± 0.8 pmHAS-VF Not traceable 4.8 149 ± 3.9 † HA concentrations were determined by CTAB turbidimetric assay § Protein quantification was performed using ImageJ based on the integrated density of the protein bands

157

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

4.5. Discussion

pmHAS was first discovered by cloning and recombinantly expressing the putative has gene in E. coli K5 cells to direct HA synthesis [88]. Over the years, conflict in literature has arisen as some argue that Gram-negative bacteria like E. coli do not produce HA because either they have an incomplete biosynthetic pathway or they express components of the pathway on a basal level [69]. In fact, Mao and coworkers had to co- express UDP-glucose dehydrogenase (kfiD) from E. coli K5 and pmHAS in recombinant E. coli JM109 cells and supplemented the culture media with glucosamine to generate HA up to 3.8 g/L HA [66].

In this work, the strain E. coli BL21 GOLD (DE3) was able to autonomously synthesize HA when expressing pmHAS. This suggests that this strain is equipped with the metabolic machinery to supply the precursor nucleotide sugars for pmHAS to direct HA synthesis. The presence of divalent cations is also important for polymerization [61, 109, 131]. Tracing the components of the cultivation medium (TB) used in these investigations, only the yeast extract could supply the divalent cations. Quantitative spectrochemical analysis of yeast extract revealed a content of 1270 µg/g dry yeast extract weight and 2.3 µg/g dry yeast extract weight for Mg2+ and Mn2+, respectively [243]. In a 50 mL TB medium, 1.25 mM Mg2+ are present, a value considerably lower than the optimum (10 mM). Nevertheless, this suboptimal condition sufficed to prompt HA polymerization during cell cultivation and pmHAS expression.

Production yield of HA from pmHAS-WT is approximately 184 µg/mL, while yield from the improved variant pmHAS-VF reached 240 µg/mL. This underscores the low production yield of HA owing to the basal production of the UDP-sugar precursors and the suboptimal quantities of magnesium ions important for synthase activity. Moreover, this also reinforces that the engineered pmHAS variant performs better than the wild type [1].

E. coli can produce HA based on its natural metabolic machinery and increasing the pool of nucleotide sugar precursors by supplementation promotes longer heterosaccharide chain production. The duration of pmHAS expression can also

158

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS influence HA production. For example, HA produced by pmHAS expressed for 3 hours are inferior with respect to quantity and chain length compared to pmHAS after 20 hours of expression.

Previous reports of HA production by non-native HA producing microbial hosts (e.g. B. subtilis [68, 136], L. lactis [137], P. pastoris [138] and E. coli [70, 88] assume that HA is produced and secreted into culture medium and/or form another layer that encapsulates the microbe [88, 103]. While the faculty and mechanism for HA translocation in Class I HAS remain inconclusive, two different explanations have been proposed: (1) a nearby membrane-bound ATP-binding cassette transport system for secreting HA out of the cell [244] similar to other bacterial polysaccharides being exported after assembly [245] and (2) the extended HA is translocated out of the cell through a pendulum motion of pmHAS [63]. Briefly, this model describes the one- monomer-at-a-time extrusion of the growing HA chain to the cell exterior by the swinging motion of the two functional domains ("arms") of HAS alternating between the binding state (off) and active transferase state (on). There is a growing consensus, however, that HA is translocated by the enzyme autonomously [75, 246]. Knowledge of the mechanism of HA translocation by the Class II pmHAS is far less progressive. However, knowledge of HA export mechanism by pmHAS in this investigation was made unnecessary by cell disruption.

We showed that pmHAS synthesizes HA, which accumulates intracellularly, in the absence of exogenous UDP-GlcA and UDP-GlcNAc. Furthermore, cell disruption can influence the chain length of the accumulated HA products. Cells that were subjected to sonication produced LMW HA, while those subjected to gentle freeze-thaw and lysozyme digestion could generate HMW HA. Sonication results from the sound (mechanical) waves emitted by the sonotrode to elicit cavitation causing cell disruption and content release. Concurrently, intracellular HA are depolymerized into shorter, monodispersed chains. The same phenomenon was observed when HA derived from rooster combs [247] and human umbilical cord [248] were subjected to ultrasonication, owing to the weak glycosidic HA linkages. The second cell disruption method initially uses the freeze-thaw process, which forms ice-crystals that disrupt the outer membrane

159

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS of E. coli. This provides access for lysozyme to cleave the β-1,4 glycosidic linkage between the N-acetyl muramic acid and N-acetyl glutamic acid in the peptidoglycan layer of E. coli [249]. In this work, the tandem cell disruption methods allowed for production of HMW polydispersed HA, not prone to lysozyme cleavage. It has been proven that lysozyme does not cleave the glycosidic linkages of HA [250]. Ultimately, we propose methods for differential microbial HA production.

The prospect of an assisting agent in HA polymerization was also investigated from the observation that the HA production output of the crude pmHAS was better than the purified synthase. First, the purified pmHAS required the initiator (HA4) while the crude enzyme did not. Second, HA polymerization with the purified pmHAS was slower compared to the crude enzyme. Third, the purified pmHAS could only produce LMW HA (< 0.5 MDa), while the crude pmHAS was more versatile by producing either HMW or LMW HA, depending on the cell disruption method. To mimic the synthesis conditions with crude pmHAS, purified pmHAS was supplemented with nucleotide sugar precursors and lysate from pmHAS-deficient E. coli cells. Resultantly, no HA production was observed. This failure to mimic the activity of the crude synthase suggested that no soluble cellular components could assist and salvage the synthase activity of crude pmHAS. The notion of the cell membrane (or its components) participating in HA production is ruled out because the in vitro synthesis process only utilized clarified cell lysates.

One report of preparative one-pot multienzyme HA polymerization from cheap monosaccharides required 20 mM MgCl2, 10 mM GlcNAc, 10 mM GlcA, 24 mM UTP, 24 mM ATP, and varied amounts of four enzymes to synthesize the UDP-sugar precursors. Addition of pmHAS catalyzed HA polymerization for 30 hours, resulting in 70 % conversion to produce LMW HA (0.015-0.55 MDa) [251]. The concentration of magnesium ions and the available UDP-sugars in our in vitro synthesis cocktail are 5- and 17.5-fold lower than the one-pot system, respectively (assuming 7 mM UDP-sugars were generated from 70% conversion rate). This may explain the low production yield of our HA synthesis. Two aspects in agreement with our results are the production of LMW HA and the long duration of HA synthesis. This is also supported in literature [252].

160

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

More recently, Eisele and colleagues optimized a nucleotide sugar regeneration enzyme module for one-pot HA synthesis by pmHAS. Using sucrose and 10 mM GlcNAc and 0.02 mM UDP as starting materials, HMW HA (2.3 MDa) were synthesized in under 10 hours [241]. Encompassing this technology with the improved pmHAS-VF can lead to another alley for an efficient production of clean HWM HA.

Thermal stability is one aspect that must be considered when producing hyaluronic acid. Boiling the in vitro synthesis HA products at 99 °C for 5 min resulted in chain depolymerization. The thermal degradation of HA increases with temperature and time, suggesting the relativity of thermostability to HA chain length [253]. In a more related case, HA (1.6-1.8 MDa) were subjected to heat treatment (90 °C) that resulted with a molar mass loss of 12 % and 17.2 % after one hour and two hours, respectively, and increase in polydispersity as determined by SEC-MALLS and electrophoretic light scattering. In the same work, subjecting BSA to 90 °C heat caused significant aggregation within a few minutes [253]. This was in accord with our observation of white filamentous aggregates forming during the boiling process, presumably from the soluble proteins in solution. A compromise between protein decontamination and HA depolymerization must be considered. Nevertheless, boiling the HA products allows for up to 97 % of proteins eliminated from solution.

It is known that E. coli is a gram-negative bacterium that possesses an outer membrane, periplasmic space and inner membrane. Anchored to the outer membrane are lipopolysaccharides that serve to maintain its integrity. Lipopolysaccharides are a form of immunostimulatory endotoxin [199, 254]. E. coli-based HA production therefore requires either a downstream post-production endotoxin removal technology or an upstream endotoxin-free microbial host. To address the latter, an engineered endotoxin- free E. coli BL21 GOLD (DE3) strain, metabolically impaired to disrupt the synthesis of critical LPS component, can be used as the microbial host of pmHAS [255]. Moreover, downstream HA purification using ultrafiltration and ions exchange steps must be implemented, while at the same time physically preserving the HA products [256]. HA purification, however, is beyond the scope of this study.

161

Chapter 4: Methods for Differential Hyaluronic Acid Synthesis by pmHAS

4.6. Conclusion

This work entails the provision of an inexpensive HA production methodology by simply hijacking the metabolic machinery of E. coli. HA production with the purified pmHAS is slow and basal despite providing the initiator for HA synthesis. HA production with the enzyme in clarified cell lysate is faster, does not require an initiator and is more versatile. The method of cell disruption influences the length of the HA products. Cell disruption by sonication produces monodispersed LMW HA, while cell disruption by freeze-thaw and lysozyme generates polydispersed HMW HA. Supplementation with the nucleotide sugar precursors promotes the synthesis of longer HA polymers. Boiling of the samples for 5 minutes removes up to 97 % of protein contaminants, however, causes slight HA depolymerization. Ultimately, foundations have been laid to differentially produce inexpensive and semi-pure HA.

162

Chapter 5: Final Summary and Remarks

5. Chapter 5: Final Summary and Remarks “The more you know, the more you know you don't know” – Aristotle

Efforts on this doctoral thesis were directed towards the production of hyaluronic acid by Class I seHAS and Class II pmHAS. Knowledge about glycosyltransferases continue to accumulate and this work not only provided testament on the complexity of HA synthases but gave rise to a series of unasked questions that have yet to be considered.

In Chapter II (seHAS engineering), seHAS was successfully cloned and expressed in three recombinant microbial hosts: E. coli, B. subtilis and S. cerevisiae. Production of HA was possible in all three. E. coli was chosen for downstream experimentation because it was the most favourable with respect to handling and time constraint. Fluorescent cellular localization studies demonstrated the presence of seHAS in the outer membrane of the Gram-negative host, contrary to the inner membrane reported in literature. Screening systems attempted in this protein engineering campaign failed to discriminate signals from empty vector controls and seHAS-WT, most likely due to basal HA production. Agarose gel electrophoresis was resultantly employed to identify improved variants generated from site-saturation mutagenesis libraries (conserved cysteines (C226, C262 and C281) and polar residues (Lys48 and Glu327)) and random mutagenesis (epPCR) library for improve chain length. Both protein engineering campaigns failed to generate one improved seHAS variant, however, positive and unintended results were obtained. Site-saturation mutagenesis variants (K48L and K48E) consistently produced monodispersed LMW HA products, while epPCR variants, A6 (R347S/F362S) and H2 (N345S/F403L), could direct synthesis of HMW HA with lower polydispersity compared to that of the wild type. Inspection of the homology model points towards the involvement of HA-HAS interaction in the control of HA polymerization. The discovery of these new positions bifurcates into another alley of polymerizing control of HAS, which is HA chain polydispersity. While methods and influencers of HA production are multifaceted, this work may have generated another sub-branch that leads to a better understanding of the control of HA polymerization by Class I HAS. Ultimately, production of industrially-relevant monodispersed HA may be possible in the future.

163

Chapter 5: Final Summary and Remarks

In Chapter III (pmHAS engineering), the KnowVolution approach successfully improved the enzymatic activity of the membrane-associated pmHAS. With such an approach, two screening systems were simultaneously employed to detect improvements in enzymatic properties: agarose gel electrophoresis and the CTAB turbidimetric assay. With CTAB, HA produced by pmHAS-expressing E. coli BL21 GOLD (DE3) were easily distinguished from the empty vector control. The four-phase KnowVolution strategy generated the final variant, pmHAS-VF, capable of producing HA up to 4.7 MDa and with a two-fold improvement in mass-based total turnover number over pmHAS-WT. This is the first case of a directed evolution of a Class II HA synthase and an example of a simultaneous dual property improvement through protein engineering. Computational work generated the "most complete", validated model of pmHAS32-703. The substitutions in pmHAS-VF (T40L, V59M and T104A) are located at the N-terminal domain, away from either glycosyltransferase active sites of pmHAS, suggesting their non-catalytic role. Molecular dynamics simulations reveal the improved flexibility of pmHAS-VF allowing it to interact with either glycosyltransferase domains of the synthase. As the N-terminus of pmHAS was previously thought to be delineated from HA polymerizing activity, these recent findings suggest a newly found involvement of the flexible N-terminal domain in HA synthesis.

This work not only improved the polymerizing activity of pmHAS, but also contributed to the understanding of the synthase on the molecular level. Deeper investigations are required to further elucidate the contributory role of the N-terminus of pmHAS in HA synthesis and raises further questions: Is it involved in the strict alternating GlcA- GlcNAc sequence of polymerization? Does it directly interact with the growing HA chain? Does it also contribute to the stability of the HA-HAS interaction? What factors dictate the flexible N-terminal region to swing between glycosyltransferase domains of pmHAS? Nevertheless, the groundwork has been laid with the improved pmHAS variant. The ability to synthesize longer HA polymers at higher output brings promise to improved enzymatic conversion.

In Chapter IV (Differential HA Synthesis), methods for production of either HMW HA or LMW HA were investigated. HA synthesis using purified pmHAS required the

164

Chapter 5: Final Summary and Remarks

synthesis initiator (HA4), was slow and produced only LMW HA. HA synthesis using lysate from sonicated pmHAS-expressing E. coli cells did not require HA4, was faster compared to synthesis with purified pmHAS and produced LMW HA. When instead subjected to freeze-thaw and lysozyme treatment, HMW HA were generated. More importantly, E. coli BL21 GOLD (DE3) cells expressing pmHAS were demonstrated to autonomously produce HA, suggesting that the strain naturally synthesizes the precursors, UDP-GlcA and UDP-GlcNAc. Moreover, addition of the in vitro synthesis cocktail elicited production of fewer but longer HA polymers. When the in vitro synthesized HA products were boiled for 5 min, up to 97 % of protein contaminants (detected by BCA and Image J) were removed at the expense of slight HA depolymerization. Ultimately, methods for inexpensive LMW or HMW HA production and semi-purification have been presented.

In closing, this doctoral work entails scientific contributions in the fields of protein engineering and HA synthase biology (glycobiology), including the following:

• Identification of several new positions in seHAS that are potentially involved in the control of HA polymerization; • Development of a dual-screening platform for the directed evolution of pmHAS – CTAB assay for HA quantification and agarose gel for chain length detection; • Generation of a pmHAS variant capable of producing HMW HA at twice the output, which can be utilized for industrial applications; • Molecular understanding of the improved pmHAS activity and unraveling the potential role of the flexible N-terminal region of pmHAS in HA synthesis; • Development of inexpensive and facile methods for production of semi-purified HMW or LMW HA.

165

Appendix

6. Appendix

6.1. Figures

166

Appendix

Figure A-1: Plasmid constructs generated for this thesis work. The pET-22b(+), pHY- 300PLK and pYES2 constructs were transformed into E. coli BL21 GOLD (DE3), B. subtilis DB104 and S. cerevisiae, respectively, for seHAS or pmHAS expression and HA production.

167

Appendix

A B

C Migration Distance (au) Molecular Weight (kDa) log MW HA ladder 1 1.040 6100 3.7853 HA ladder 2 1.316 4570 3.6599 HA ladder 3 1.822 3050 HA ladder 4 2.169 1510 3.1790 HA ladder 5 2.427 1138 3.0561 HA ladder 6 2.640 940 2.9731 HA ladder 7 2.854 667 2.8241 HA ladder 8 3.067 509 2.7067

HAmax -VF 1.271 4701 3.672

HA min - VF 3.253 414 2.617

HAmax - WT 1.511 3503 3.544 HAmin - WT 3.253 414 2.617 Figure A-2: Molecular weight determination of HA produced by pmHAS-WT and pmHAS-VF by ImageJ. In vitro HA syntheses by pmHAS-WT and pmHAS-VF were performed in 96-well MTP as previously described. HA products were pooled together to a total of approximately 15 mL. ImageJ was employed for molecular weight determination of the HA products [211]. (A, C) Briefly, the distance of migration of each of the standard HA bands were measured using the "straight line" option. The distances were measured in arbitrary units (au). (B) A standard curve of log MW and the corresponding migration distances were plotted. A co- efficient of determination (R2) of 0.9985 was obtained from linear equation of y = -1.8786 x + 8.1696, when ladder band 3 is discounted. The upper (HAmax) and lower values (HAmin) of the HA smear from pmHAS-WT and pmHAS-VF were interpolated to obtain the molecular weight range of the HA products. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

168

Appendix

Lysozyme (mg/mL) Area Raw Integrated Density B 3.000 0.058 666804 2.000 0.058 598676 1.000 0.058 379349 0.500 0.058 247132 0.250 0.058 140850 0.125 0.058 64921 0.063 0.058 31368

Sam ple Raw Integrated Density pmHAS/Lysozyme Ratio WT-pmHAS 1354050 0.521 Lysozyme 2596920 VF-pmHAS 876435 0.302 Lysozyme 2904195

Sam ple Raw Integrated Density Corresponding conc (mg/mL) pmHAS/Lysozyme Ratio Estimated pmHAS (mg/mL) D pmHAS-WT 0.521 0.398 Lysozyme 304952.666 0.764 pmHAS-VF 0.302 0.268 Lysozyme 344830.667 0.887

169

Appendix

Figure A-3: Quantification of pmHAS expression by ImageJ. In vitro HA syntheses by pmHAS-WT and pmHAS-VF were performed in 96-well MTP as previously described. HA products were pooled together to a total of approximately 15 mL. ImageJ was employed to quantify pmHAS expression [211]. Briefly, the image type was converted to a 32-bit grey scale and was inverted to have a black background and white bands. (A) The bands were highlighted to generate the peak areas. (B) The integrated densities under each curve corresponding to the pmHAS (or lysozyme) bands were measured. (C) A standard curve of lysozyme (0–3 mg/mL) and the corresponding integrated intensities were plotted. A co-efficient of determination (R2) of 0.9999 was obtained from the second-degree polynomial equation. (D) The integrated densities of the lysozyme bands were interpolated to the quadratic curve to calculate the corresponding lysozyme concentration in solution (http://www.math.com/students/calculators/source/quadratic.htm). The concentration of pmHAS involved in HA synthesis was calculated by multiplying the known lysozyme concentration by the pmHAS/lysozyme ratio. Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA.

Figure A-4: Bicinchoninic acid (BCA) assay for protein quantification. A dilution series of BSA standard up to 1.00 mg/mL was performed. To facilitate the reaction, a triplicate of 25 µL standard were incubated with 200 µL BCA working reagent in microtiter plate and incubated 37 °C for 30 min. The reaction was cooled to room temperature (10 min) and the absorbance was measured at 562 nm. The regression curve (y=1.3022x + 0.1013; R2= 0.9952) was used to interpolate the corresponding protein concentration.

170

Appendix

6.2. Tables

Table A1: PCR Primers used for seHAS engineering

Purpose Primer Name 5’ – 3’ Tm (°C)

Introduce NotI Fwd_not_in GCTCCTTCTACTATGGGAAAGGTC 58.6 into CTCATCATGCGGCCGCCTCCTCCTC Rev_not_in2 54 pHY300PLK TTTATTCAG CTGAATAAAGAGGAGGAGAGTGA Remove NotI fwd_not_Out 59.7 GTAATGAGAACATTAAAAAACCTC out of GAGGTTTTTTAATGTTCTCATTACT pHY300PLK Rev_not_out 59.7 CACTCTCCTCCTCTTTATTCAG pHY300PLK Seq Cloning CAGATTTCGTGATGCTTGTCAGG 56.2 Primer_F verification and pHY300PLK Seq Sequencing CGTTAAGGGATCAACTTTGGGAG 55.6 Primer_R GAATATTAAGCTTGGTACCAGG Insert seHAS_F_rek_2 ATG AGA ACA TTA AAA AAC CTC 59.4 amplification ATA ACT G for homologous CGTTACTAGTGGATCCGAGCTCTC rec by S. seHAS_R_rek_2 ACAATAATTTTTTACGTGTTCCCCA 54.3 cerevisiae G Vector pYES_F_rek2 GAG CTC GGA TCC ACT AGT AAC G 56.6 amplification pYES_R_rek2 for homologous (changed, start CTCATCCTGGTACCAAGCTTAATAT 55.2 rec by S. immediately upstream TC cerevisiae of MFα sequence) Cloning fw_CalB_Sequenzing GCCATTTTCCAACAGCAC 52.5 verification and 51.6 Sequencing rev_CalB_Sequenzing GCGTGAATGTAAGCGTG Introduce NdeI TAG AGC ATA TGA GAA CAT TAA seHAS_F_SDM2 59.4 restriction site AAA ACC TCA TAA CTG TTG TGG C for Ligation- Dependent seHAS_R_SDM2 GAA AGG AAG GCC CAT GAG GC 58.6 Cloning (E. coli) GTATCCAGATCCGCTGGAACCGCG G29 down_F 53.7 GGT GCT AAA GGA AGC TTG TC E-tag CGCGGTTCCAGCGGATCTGGA CAA detection F28 up_R AGA GAT AAA CAT TGA CGT AAA 53.5 TCA AC F01_epseHAS GCG GAT AAC AAT TCC CCT CTA G 55.0 epPCR R01_epseHAS CGA CGG AGC TCG AAT TCC TAC 57.3 CGT TAC AGG TAA TAT CCT TGT F01_C226_NNK SSM on TNN KTC AGG TCC GCT TAG CG 63.7-66.9 Position C226 CGC TAA GCG GAC CTG AMN NAA R01_C226_NNK CAA GGA TAT TAC CTG TAA CG

171

Appendix

GTA AGT ATT GGT GAT GAC AGG F01_C262_NNK NNK TTG ACC AAC TAT GCA ACT SSM on G 63.0-65.9 Position C262 CAG TTG CAT AGT TGG TCA AMN R01_C262_NNK NCC TGT CAT CAC CAA TAC TTA C CTG TTT ATC AAT CCA CTG CTA F01_C281_NNK AAN NKA TTA CAG ATG TTC CTG SSM on ACA AGA TG 62.1-64.7 Position C281 CAT CTT GTC AGG AAC ATC TGT R01_C281_NNK AAT MNN TTT AGC AGT GGA TTG ATA AAC AG GCT GAT AGC TTA CCT ATT AGT F01_K48_NNK SSM on CNN KAT GTC CTT ATC 57.1-61.5 Position K48 GAT AAG GAC ATM NNG ACT AAT R01_K48_NNK AGG TAA GCT ATC AGC GCC CTA TGG ACC ATA CTT NNK F01_E327_NNK SSM on GTG TCT ATG TTT ATG ATG 60.6-64.1 Position E327 CAT CAT AAA CAT AGA CAC MNN R01_E327_NNK AAG TAT GGT CCA TAG GGC

172

Appendix

Table A2: PCR primers used for the directed pmHAS evolution campaign [a].

Purpose Primer Name 5’ – 3’ Tm (°C)

PLICing Primer Backbone gatgatgatgatgGCTGCTACCCATGGTA 61.9 Amplification (R) TATCTC PLICing Primer Backbone accaccaccactGAGATCCGGCTGCTAAC 66.5 epPCR [b] Amplification (F) PLICing Primer Insert catcatcatcatcATCACAGCAGCGGCCT 65.3 Amplification (F) GG PLICing Primer Insert agtggtggtggtGGTGGTGCTCGAGTG 67.5 Amplification (R) GTC GAA TTC CAG ATC NNK AAA pmH40_SSM_Fwd 59.2-63.2 TGC AAA GAA AAA CTG AG CTC AGT TTT TCT TTG CAT TTM pmH40_SSM_Rev 59.2-63.2 NNG ATC TGG AAT TCG AC Site- GCC CAT CTG AGC NNK AAT AAA pmH59_SSM_Fwd 58.9-63.1 Saturated GAA GAA AAA GTT AAC G CGT TAA CTT TTT CTT CTT TAT pmH59_SSM_Rev 58.9-63.1 Mutagenesis TMN NGC TCA GAT GGG C TGG AAA CTG CTG NNK GAG AAA pmH104_SSM_Fwd 60.2-64.1 AAA AGC GAA AAT G CAT TTT CGC TTT TTT TCT CMN pmH104_SSM_Rev 60.2-64.1 NCA GCA GTT TCC A CGT CGA ATT CCA GAT CCT TAA T40L_Fwd 58.1 ATG CAA AG CTT TGC ATT TAA GGA TCT GGA T40L_Rev 58.1 ATT CGA CG GTG CCC ATC TGA GCC TTA ATA V59L_Fwd 59.3 AAG AAG AAA AAG CTT TTT CTT CTT TAT TAA GGC V59L_Rev 59.3 Site- TCA GAT GGG CAC GGA AAC TGC TGT TGG AGA AAA Directed T104L_Fwd 59.2 AAA GCG Mutagenesis CGC TTT TTT TCT CCA ACA GCA T104L_Rev 59.2 GTT TCC [c] GTG CCC ATC TGA GCA TGA ATA SDM V59M_Fwd 59.6 AAG AAG AAA AAG CTT TTT CTT CTT TAT TCA TGC TCA SDM V59M_Rev 59.6 GAT GGG CAC GAA ACT GCT GGC CGA GAA AAA SDM T104A_Fwd 59.4 AAG C GCT TTT TTT CTC GGC CAG CAG SDM T104A_Rev 59.4 TTT C [a] Reproduced from [1] with permission from © Wiley-VCH Verlag GmbH & Co. KGaA. [b] small case denotes the phosphorothioated bases [c] underlined bases indicate the site of mutation

173

References

7. References

[1] Mandawe J, Infanzon B, Eisele A, Zaun H, Kuballa J, Davari MD, et al. Directed Evolution of Hyaluronic Acid Synthase from Pasteurella multocida Towards High Molecular Weight Hyaluronic Acid. ChemBioChem. 2018;19:1414-23.

[2] Meyer K, Palmer JW. The Polysaccharide of the Vitreous Humor. J Biol Chem. 1934;107:629-34.

[3] Kendall FE, Heidelberger M, Dawson MH. A Serologically Inactive Polysaccharide Elaborated by Mucoid Strains of Group A Hemolytic Streptococcus. J Biol Chem. 1937;118:61-9.

[4] Weissmann B, Meyer K. Structure of Hyaluronic Acid. The Glucuronodic Linkage. J Am Chem Soc. 1952;74:4729.

[5] DeAngelis PL, Jing W, Graves MV, Burbank DE, Van Etten JL. Hyaluronan Synthase of Chlorella Virus PBCV-1. Science. 1997;278:1800-3.

[6] DeAngelis PL, Papaconstantinou J, Weigel PH. Isolation of a Streptococcus pyogenes Gene Locus that Directs Hyaluronan Biosynthesis in Acapsular Mutants and in Heterologous Bacteria. J Biol Chem. 1993;268:14568-71.

[7] Kumari K, Weigel PH. Molecular Cloning, Expression, and Characterization of the Authentic Hyaluronan Synthase from Group C Streptococcus equisimilis. J Biol Chem. 1997;272:32539-46.

[8] Shyjan AM, Heldin P, Butcher EC, Yoshino T, Briskin MJ. Functional Cloning of the cDNA for a Human Hyaluronan Synthase. J Biol Chem. 1996;271:23395-9.

[9] Pummill PE, Achyuthan AM, DeAngelis PL. Enzymological Characterization of Recombinant XenopusDG42, A Vertebrate Hyaluronan Synthase. J Biol Chem. 1998;273:4976-81.

[10] Usui T, Nakajima F, Ideta R, Kaji Y, Suzuki Y, Araie M, et al. Hyaluronan Synthase in Trabecular Meshwork Cells. Br J Opthalmol 2003;87:357-60.

[11] Necas J, Bartosikova L, Brauner P, Kolar J. Hyaluronic Acid (Hyaluronan): A Review. Vet Med (Prague, Czech Repub). 2008;53:397-411.

[12] Kogan G, Šoltés L, Stern R, Mendichi R. Hyaluronic Acid: A Biopolymer with Versatile Physico-Chemical and Biological Properties. New York: Nova Science Publishers; 2007.

[13] Sherman LS, Matsumoto S, Su W, Srivastava T, Back SA. Hyaluronan Synthesis, Catabolism, and Signaling in Neurodegenerative Diseases. Int J Cell Biol. 2015;2015:10.

174

References

[14] Ragan C, Meyer K. The Hyaluronic Acid of Synovial Fluid in Rheumatoid Arthritis. J Clin Inv. 1949;28:56-9.

[15] Ogston AG, Stanier JE. The Dimensions of the Particle of Hyaluronic Acid Complex in Synovial Fluid. Biochem J. 1951;49:585-90.

[16] Laurent TC, Gergely J. Light Scattering Studies on Hyaluronic Acid. J Biol Chem. 1955;212:325-33.

[17] Selyanin MA, Boykov PY, Khabarov VN. Hyaluronic Acid: Preparation, Properties, Application in Biology and Medicine.2014.

[18] Cowman MK, Matsuoka S. Experimental Approaches to Hyaluronan Structure. Carbohydr Res. 2005;340:791-809.

[19] Stern R, Asari AA, Sugahara KN. Hyaluronan Fragments: An Information-Rich System. Eur J Cell Biol. 2006;85:699-715.

[20] Tamer TM. Hyaluronan and Synovial Joint: Function, Distribution and Healing. Interdiscip Toxicol. 2013;6:111-25.

[21] Dahiya P, Kamal R. Hyaluronic Acid: A Boon in Periodontal Therapy. N Am J Med Sci. 2013;5:309-15.

[22] Morris ER, Rees DA, Welsh EJ. Conformation and Dynamic Interactions in Hyaluronate Solutions. J Mol Biol. 1980;138:383-400.

[23] Atkins ED, Phelps CF, Sheehan JK. The Conformation of the Mucopolysaccharides. Hyaluronates. Biochem J. 1972;128:1255-63.

[24] Ribitsch G, Schurz J, Ribitsch V. Investigation of the Solution Structure of Hyaluronic Acid by Light Scattering, SAXS, and Viscosity Measurements. Colloid Polym Sci. 1980;258:1322-34.

[25] Haxaire K, Braccini I, Milas M, Rinaudo M, Perez S. Conformational Behavior of Hyaluronan in Relation to its Physical Properties as Probed by Molecular Modeling. Glycobiology. 2000;10:587-94.

[26] Simulescu V, Mondek J, Kalina M, Pekař M. Kinetics of Long-Term Degradation of Different Molar Mass Hyaluronan Solutions Studied by SEC-MALLS. Polym Degrad Stab. 2015;111:257-62.

[27] Saito S, Hashimoto T, Morfin I, Lindner P, Boué F. Structures in a Semidilute Polymer Solution Induced under Steady Shear Flow as Studied by Small-Angle Light and Neutron Scattering. Macromolecules. 2002;35:445-59.

[28] Cleland RL. Effect of Temperature on the Limiting Viscosity Number of Hyaluronic Acid and Chondroitin 4‐Sulfate. Biopolymers. 1979;18:1821-8.

175

References

[29] Jensen CE, Koefoed J. Flow Elasticity of Hyaluronate Solutions. J Colloid Sci. 1954;9:460-5.

[30] Cowman MK, Schmidt TA, Raghavan P, Stecco A. Viscoelastic Properties of Hyaluronan in Physiological Conditions. F1000Research. 2015;4:622.

[31] Laurent TC, Fraser JRE. Hyaluronan. FASEB J. 1992;6:2397-404.

[32] Sutherland IW. Novel and Established Applications of Microbial Polysaccharides. Trends Biotechnol. 1998;16:41-6.

[33] Tammi MI, Day AJ, Turley EA. Hyaluronan and Homeostasis: A Balancing Act. J Biol Chem. 2002;277:4581-4.

[34] Balazs E, Miller D, Stegmann R. Viscosurgery and the Use of Na-Hyaluronate in Intraocular Lens Implantation. Lecture, Cannes, France. 1979;5:56-9.

[35] Liang J, Jiang D, Noble PW. Hyaluronan as a Therapeutic Target in Human Diseases. Adv Drug Deliv Rev. 2016;97:186-203.

[36] Dicker KT, Gurski LA, Pradhan-Bhatt S, Witt RL, Farach-Carson MC, Jia X. Hyaluronan: A Simple Polysaccharide with Diverse Biological Functions. Acta Biomater. 2014;10:1558-70.

[37] Schiraldi C, La Gatta A, De Rosa M. Biotechnological Production and Application of Hyaluronan. In: Elnashar M, editor. Biopolymers. London, UK: InTech; 2010. p. 387- 412.

[38] Lee JY, Spicer AP. Hyaluronan: A Multifunctional, MegaDalton, Stealth Molecule. Curr Opin Cell Biol. 2000;12:581-6.

[39] Albano GD, Bonanno A, Cavalieri L, Ingrassia E, Di Sano C, Siena L, et al. Effect of High, Medium, and Low Molecular Weight Hyaluronan on Inflammation and Oxidative Stress in an In Vitro Model of Human Nasal Epithelial Cells. Mediators Inflamm. 2016;2016:1-13.

[40] Spinelli FM, Vitale DL, Demarchi G, Cristina C, Alaniz L. The Immunological Effect of Hyaluronan in Tumor Angiogenesis. Clin Trans Immunol. 2015;4:e52.

[41] Boregowda RK, Appaiah HN, Siddaiah M, Kumarswamy SB, Sunila S, Thimmaiah KN, et al. Expression of Hyaluronan in Human Tumor Progression. J Carcinog. 2006;5:2.

[42] Rah MJ. A Review of Hyaluronan and its Ophthalmic Applications. Optometry (St Louis, Mo). 2011;82:38-43.

[43] Ogston AG, Stanier JE. The Physiological Function of Hyaluronic Acid in Synovial Fluid; Viscous, Elastic and Lubricant Properties. J Physiol. 1953;119:244-52.

176

References

[44] Barbucci R, Lamponi S, Borzacchiello A, Ambrosio L, Fini M, Torricelli P, et al. Hyaluronic Acid Hydrogel in the Treatment of Osteoarthritis. Biomaterials. 2003;23:4503-13.

[45] Neuman MG, Nanau RM, Oruña-Sanchez L, Coto G. Hyaluronic Acid and Wound Healing. J Pharm Pharm Sci. 2015;18:53-60.

[46] Allison DD, Grande-Allen KJ. Review. Hyaluronan: A Powerful Tissue Engineering Tool. Tissue Eng. 2006;12:2131-40.

[47] Pedrosa SS, Goncalves C, David L, Gama M. A Novel Crosslinked Hyaluronic Acid Nanogel for Drug Delivery. Macromol Biosci. 2014;14:1556-68.

[48] McBride WH, Bard JB. Hyaluronidase-Sensitive Halos Around Adherent Cells. Their Role in Blocking Lymphocyte-Mediated Cytolysis. J Exp Med. 1979;149:507-15.

[49] Myint P, Deeble DJ, Beaumont PC, Blake SM, Phillips GO. The Reactivity of Various Free Radicals with Hyaluronic Acid: Steady-State and Pulse Radiolysis Studies. Biochim Biophys Acta. 1987;925:194-202.

[50] Nobile V, Buonocore D, Michelotti A, Marzatico F. Anti-Aging and Filling Efficacy of Six Types Hyaluronic Acid Based Dermo-Cosmetic Treatment: Double Blind, Randomized Clinical Trial of Efficacy and Safety. J Cosmet Dorm. 2014;13:277-87.

[51] Papakonstantinou E, Roth M, Karakiulakis G. Hyaluronic Acid: A Key Molecule in Skin Aging. Dermatoendocrinol. 2012;4:253-8.

[52] Research GV. Hyaluronic Acid Market Size To Reach $13.4 Billion By 2024. San Francisco, California: Grand View Research; 2016.

[53] Yang C, Cao M, Liu H, He Y, Xu J, Du Y, et al. The High and Low Molecular Weight Forms of Hyaluronan Have Distinct Effects on CD44 Clustering. J Biol Chem. 2012;287:43094-107.

[54] Cyphert JM, Trempus CS, Garantziotis S. Size Matters: Molecular Weight Specificity of Hyaluronan Effects in Cell Biology. Int J Cell Biol. 2015;2015:563818.

[55] Rogošić M, Mencer HJ, Gomzi Z. Polydispersity Index and Molecular Weight Distributions of Polymers. Eur Polym J. 1996;32:1337-44.

[56] DeAngelis PL. Hyaluronan Synthases: Fascinating Glycosyltransferases from Vertebrates, Bacterial Pathogens, and Algal Viruses. Cell Mol Life Sci. 1999;56:670-82.

[57] DeAngelis PL. Evolution of Glycosaminoglycans and their Glycosyltransferases: Implications for the Extracellular Matrices of Animals and the Capsules of Pathogenic Bacteria. Anat Rec. 2002;268:317-26.

177

References

[58] Prehm P. Hyaluronate is Synthesized at Plasma Membranes. Biochem J. 1984;220:597-600.

[59] Weigel PH, Hascall VC, Tammii M. Hyaluronan Synthases. J Biol Chem. 1997;272:13997-4000.

[60] Heldermon C, DeAngelis PL, Weigel PH. Topological Organization of the Hyaluronan Synthase from Streptococcus pyogenes. J Biol Chem. 2001;276:2037-46.

[61] Stoolmiller AC, Dorfman A. The Biosynthesis of Hyaluronic Acid by Streptococcus. J Biol Chem. 1969;244:236-46.

[62] Tlapak-Simmons VL, Baron CA, Gotschall R, Haque D, Canfield WM, Weigel PH. Hyaluronan Biosynthesis by Class I Streptococcal Hyaluronan Synthases Occurs at the Reducing End. J Biol Chem. 2005;280:13012-8.

[63] Weigel PH. Hyaluronan Synthase: The Mechanism of Initiation at the Reducing End and a Pendulum Model for Polysaccharide Translocation to the Cell Exterior. Int J Cell Biol. 2015;2015:1-15.

[64] Blackburn MR, Hubbard C, Kiessling V, Bi YC, Kloss B, Tamm LK, et al. Distinct Reaction Mechanisms for Hyaluronan Biosynthesis in Different Kingdoms of Life. Glycobiology. 2018;28:108-21.

[65] Weigel PH, DeAngelis PL. Hyaluronan Synthases: A Decade-Plus of Novel Glycosyltransferases. J Biol Chem. 2007;282:36777-81.

[66] Mao Z, Shin HD, Chen R. A Recombinant E. coli Bioprocess for Hyaluronan Synthesis. Appl Microbiol Biotechnol. 2009;84:63-9.

[67] Chong BF, Blank LM, McLaughlin R, Nielsen LK. Microbial Hyaluronic Acid Production. Appl Microbiol Biotechnol. 2005;66:341-51.

[68] Widner B, Behr R, Von Dollen S, Tang M, Heu T, Sloma A, et al. Hyaluronic Acid Production in Bacillus subtilis. Appl Environ Microbiol. 2005;71:3747-52.

[69] Sze JH, Brownlie JC, Love CA. Biotechnological Production of Hyaluronic Acid: A Mini Review. 3 Biotech. 2016;6:67.

[70] Yu H, Stephanopoulos G. Metabolic Engineering of Escherichia coli for Biosynthesis of Hyaluronic Acid. Metab Eng. 2008;10:24-32.

[71] Bi Y, Hubbard C, Purushotham P, Zimmer J. Insights Into the Structure and Function of Membrane-Integrated Processive Glycosyltransferases. Curr Opin Struct Biol. 2015;34:78-86.

[72] Lairson LL, Henrissat B, Davies GJ, Withers SG. Glycosyltransferases: Structures, Functions, and Mechanisms. Annu Rev Biochem. 2008;77:521-55.

178

References

[73] Weigel PH. Functional Characteristics and Catalytic Mechanisms of the Bacterial Hyaluronan Synthases. IUBMB life. 2002;54:201-11.

[74] Morgan JL, Strumillo J, Zimmer J. Crystallographic Snapshot of Cellulose Synthesis and Membrane Translocation. Nature. 2013;493:181-6.

[75] Hubbard C, McNamara JT, Azumaya C, Patel MS, Zimmer J. The Hyaluronan Synthase Catalyzes the Synthesis and Membrane Translocation of Hyaluronan. J Mol Biol. 2012;418:21-31.

[76] Yang J, Cheng F, Yu H, Wang J, Guo Z, Stephanopoulos G. Key Role of the Carboxyl Terminus of Hyaluronan Synthase in Processive Synthesis and Size Control of Hyaluronic Acid Polymers. Biomacromolecules. 2017;18:1064-73.

[77] Tlapak-Simmons VL, Kempner ES, Baggenstoss BA, Weigel PH. The Active Streptococcal Hyaluronan Synthases (HASs) Contain a Single HAS Monomer and Multiple Cardiolipin Molecules. J Biol Chem. 1998;273:26100-9.

[78] Pummill PE, Kane TA, Kempner ES, DeAngelis PL. The Functional Molecular Mass of the Pasteurella hyaluronan Synthase is a Monomer. Biochim Biophys Acta. 2007;1770:286-90.

[79] Itano N, Sawai T, Yoshida M, Lenas P, Yamada Y, Imagawa M, et al. Three Isoforms of Mammalian Hyaluronan Synthases Have Distinct Enzymatic Properties. J Biol Chem. 1999;274:25085-92.

[80] Brinck J, Heldin P. Expression of Recombinant Hyaluronan Synthase (HAS) Isoforms in CHO Cells Reduces Cell Migration and Cell Surface CD44. Exp Cell Res. 1999;252:342-51.

[81] Viola M, Vigetti D, Karousou E, D'Angelo ML, Caon I, Moretto P, et al. Biology and Biotechnology of Hyaluronan. Glycoconj J. 2015;32:93-103.

[82] Camenisch TD, Spicer AP, Brehm-Gibson T, Biesterfeldt J, Augustine ML, Calabro A, et al. Disruption of Hyaluronan Synthase-2 Abrogates Normal Cardiac Morphogenesis and Hyaluronan-Mediated Transformation of Epithelium to Mesenchyme. J Clin Inv. 2000;106:349-60.

[83] Crater DL, Dougherty BA, van de Rijn I. Molecular Characterization of hasC from an Operon Required for Hyaluronic Acid Synthesis in Group A Streptococci. Demonstration of UDP-Glucose Pyrophosphorylase Activity. J Biol Chem. 1995;270:28676-80.

[84] Dougherty BA, van de Rijn I. Molecular Characterization of hasB from an Operon Required for Hyaluronic Acid Synthesis in Group A Streptococci. Demonstration of UDP-Glucose Dehydrogenase Activity. J Biol Chem. 1993;268:7118-24.

179

References

[85] Csoka AB, Stern R. Hypotheses on the Evolution of Hyaluronan: A Highly Ironic Acid. Glycobiology. 2013;23:398-411.

[86] Wessels MR, Moses AE, Goldberg JB, DiCesare TJ. Hyaluronic Acid Capsule is a Virulence Factor for Mucoid Group A Streptococci. Proc Natl Acad Sci USA. 1991;88:8317-21.

[87] Blank LM, Hugenholtz P, Nielsen LK. Evolution of the Hyaluronic Acid Synthesis (has) Operon in Streptococcus zooepidemicus and Other Pathogenic Streptococci. J Mol Evol. 2008;67:13-22.

[88] DeAngelis PL, Jing W, Drake RR, Achyuthan M. Identification and Molecular Cloning of a Unique Hyaluronan Synthase from Pasteurella multocida. J Biol Chem. 1998;273:8454-8.

[89] Jong A, Wu CH, Chen HM, Luo F, Kwon-Chung KJ, Chang YC, et al. Identification and Characterization of CPS1 as a Hyaluronic Acid Synthase Contributing to the Pathogenesis of Cryptococcus neoformans Infection. Eukaryot Cell. 2007;6:1486-96.

[90] Roseman S, Ludowieg J, Moses FE, Dorfman A. The Biosynthesis of Hyaluronic Acid by Group A Streptococcus: II. Origin of the Glucuronic Acid. J Biol Chem. 1954;206:665-9.

[91] Dorfman A, Roseman S, Moses FE, Ludowieg J, Mayeda M. The Biosynthesis of Hyaluronic Acid by Group A Streptococcus: III. Origin of the n-Acetylglucosamine Moiety. J Biol Chem. 1955;212:583-92.

[92] Markovitz A, Cifonelli JA, Dorfman A. The Biosynthesis of Hyaluronic Acid by Group A Streptococcus. VI. Biosynthesis from Uridine Nucleotides in Cell-Free Extracts. J Biol Chem. 1959;234:2343-50.

[93] Sugahara K, Schwartz NB, Dorfman A. Biosynthesis of Hyaluronic Acid by Streptococcus. J Biol Chem. 1979;254:6252-61.

[94] Prehm P. Synthesis of Hyaluronate in Differentiated Teratocarcinoma Cells. Mechanism of Chain Growth. Biochem J. 1983;211:191-8.

[95] DeAngelis PL, Papaconstantinou J, Weigel PH. Molecular Cloning, Identification, and Sequence of the Hyaluronan Synthase Gene from Group A Streptococcus pyogenes. J Biol Chem. 1993;268:19181-4.

[96] DeAngelis PL, Weigel PH. Immunochemical Confirmation of the Primary Structure of Streptococcal Hyaluronan Synthase and Synthesis of High Molecular Weight Product by the Recombinant Enzyme. Biochemistry. 1994;33:9033-9.

[97] Tlapak-Simmons VL, Baron CA, Weigel PH. Characterization of the Purified Hyaluronan Synthase from Streptococcus equisimilis. Biochemistry. 2004;43:9234-42.

180

References

[98] Tlapak-Simmons VL, Baggenstoss BA, Clyne T, Weigel PH. Purification and Lipid Dependence of the Recombinant Hyaluronan Synthases from Streptococcus pyogenes and Streptococcus equisimilis. J Biol Chem. 1999;274:4239-45.

[99] Tlapak-Simmons VL, Baggenstoss BA, Kumari K, Heldermon C, Weigel PH. Kinetic Characterization of the Recombinant Hyaluronan Synthases from Streptococcus pyogenes and Streptococcus equisimilis. J Biol Chem. 1999;274:4246-53.

[100] Kumari K, Tlapak-Simmons VL, Baggenstoss BA, Weigel PH. The Streptococcal Hyaluronan Synthases are Inhibited by Sulfhydryl-Modifying Reagents, but Conserved Cysteine Residues are not Essential for Enzyme Function. J Biol Chem. 2002;277:13943-51.

[101] Kumari K, Weigel PH. Identification of a Membrane-Localized Cysteine Cluster Near the Substrate-Binding Sites of the Streptococcus equisimilis Hyaluronan Synthase. Glycobiology. 2005;15:529-39.

[102] Weigel PH, Baggenstoss BA. Hyaluronan Synthase Polymerizing Activity and Control of Product Size are Discrete Enzyme Functions that can be Uncoupled by Mutagenesis of Conserved Cysteines. Glycobiology. 2012;22:1302-10.

[103] Kumari K, Baggenstoss BA, Parker AL, Weigel PH. Mutation of Two Intramembrane Polar Residues Conserved Within the Hyaluronan Synthase Family Alters Hyaluronan Product Size. J Biol Chem. 2006;281:11755-60.

[104] Zhang L, Huang H, Wang H, Chen J, Du G, Kang Z. Rapid Evolution of Hyaluronan Synthase to Improve Hyaluronan Production and Molecular Mass in Bacillus subtilis. Biotechnol Lett. 2016;38:2103-8.

[105] Baggenstoss BA, Harris EN, Washburn JL, Medina AP, Nguyen L, Weigel PH. Hyaluronan Synthase Control of Synthesis Rate and Hyaluronan Product Size are Independent Functions Differentially Affected by Mutations in a Conserved Tandem B- X7-B Motif. Glycobiology. 2017;27:154-64.

[106] Jing W, DeAngelis PL. Dissection of the Two Transferase Activities of the Pasteurella multocida Hyaluronan Synthase: Two Active Sites Exist in One Polypeptide. Glycobiology. 2000;10:883-9.

[107] Jing W, DeAngelis PL. Analysis of the Two Active Sites of the Hyaluronan Synthase and the Chondroitin Synthase of Pasteurella multocida. Glycobiology. 2003;13:661-71.

[108] Kooy FK, Beeftink HH, Eppink MHM, Tramper J, Eggink G, Boeriu CG. Kinetic and Structural Analysis of Two Transferase Domains in Pasteurella multocida Hyaluronan Synthase. J Mol Catal B: Enzym. 2014;102:138-45.

181

References

[109] Kooy FK, Beeftink HH, Eppink MHM, Tramper J, Eggink G, Boeriu CG. Structural and Functional Evidence for Two Separate Oligosaccharide Binding Sites of Pasteurella multocida Hyaluronan Synthase. Adv Enzyme Res. 2013;01:97-111.

[110] Charnock SJ, Davies GJ. Structure of the Nucleotide-Diphospho-Sugar Transferase, SpsA from Bacillus subtilis, in Native and Nucleotide-Complexed Forms. Biochemistry. 1999;38:6380-5.

[111] DeAngelis PL. Molecular Directionality of Polysaccharide Polymerization by the Pasteurella multocida Hyaluronan Synthase. J Biol Chem. 1999;274:26557-62.

[112] McNamara JT, Morgan JL, Zimmer J. A Molecular Description of Cellulose Biosynthesis. Annu Rev Biochem. 2015;84:895-921.

[113] Balazs EA, Laurent TC. Viscosity Function of Hyaluronic Acid as a Polyelectrolyte. J Polym Sci. 1951;6:665-7.

[114] Caputo A. Depolymerization of Hyaluronic Acid by X-rays. Nature. 1957;179:1133- 4.

[115] Linker A, Meyer K, Hoffman P. The Production of Unsaturated Uronides by Bacterial Hyaluronidases. J Biol Chem. 1956;219:13-25.

[116] Tommeraas K, Melander C. Kinetics of Hyaluronan Hydrolysis in Acidic Solution at Various pH Values. Biomacromol. 2008;9:1535-40.

[117] Stern R. Hyaluronan Catabolism: A New Metabolic Pathway. Eur J Cell Biol. 2004;83:317-25.

[118] Banks J, Kreider JW, Bhavanandan VP, Davidson EA. Anionic Polysaccharide Production and Tyrosinase Activation in Cultured Human Melanoma Cells. Cancer Res. 1976;36:424-31.

[119] Laurent TC, Fraser JR. The Properties and Turnover of Hyaluronan. Ciba Found Symp. 1986;124:9-29.

[120] Fraser JR, Laurent TC, Laurent UB. Hyaluronan: Its Nature, Distribution, Functions and Turnover. J Intern Med. 1997;242:27-33.

[121] Hua Q, Knudson CB, Knudson W. Internalization of Hyaluronan by Chondrocytes Occurs via Receptor-Mediated Endocytosis. J Cell Sci. 1993;106:365-75.

[122] Zhou B, Weigel JA, Fauss L, Weigel PH. Identification of the Hyaluronan Receptor for Endocytosis (HARE). J Biol Chem. 2000;275:37733-41.

[123] Prevo R, Banerji S, Ferguson DJP, Clasper S, Jackson DG. Mouse LYVE-1 is an Endocytic Receptor for Hyaluronan in Lymphatic Endothelium. J Biol Chem. 2001;276:19420-30.

182

References

[124] Triggs-Raine B, Natowicz MR. Biology of Hyaluronan: Insights from Genetic Disorders of Hyaluronan Metabolism. World J Biol Chem. 2015;6:110-20.

[125] Linker A, Meyer K, Weissmann B. Enzymatic Formation of Monosaccharides from Hyaluronate. J Biol Chem. 1955;213:237-48.

[126] Roseman S, Moses FE, Ludowieg J, Dorfman A. The Biosynthesis of Hyaluronic Acid by Group A Streptococcus. I. Utilization of 1-C14-Glucose. J Biol Chem. 1953;203:213-25.

[127] Taussig A. A Simple Method for the Preparation of Hyaluronic Acid. Can J Biochem. 1965;43:132-4.

[128] Boas NF. Isolation of Hyaluronic Acid from the Cock's Comb. J Biol Chem. 1949;181:573-5.

[129] Kang DY, Kim WS, Heo IS, Park YH, Lee S. Extraction of Hyaluronic Acid (HA) from Rooster Comb and Characterization Using Flow Field-Flow Fractionation (FlFFF) Coupled with Multiangle Light Scattering (MALS). J Sep Sci. 2010;33:3530-6.

[130] Balazs EA. Ultrapure Hyaluronic Acid and the Use Thereof. In: States U, editor. United States1979.

[131] DeAngelis PL. Enzymological Characterization of the Pasteurella multocida Hyaluronic Acid Synthase. Biochemistry. 1996;35:9768-71.

[132] Kim J-H, Yoo S-J, Oh D-K, Kweon Y-G, Park D-W, Lee C-H, et al. Selection of a Streptococcus equi Mutant and Optimization of Culture Conditions for the Production of High Molecular Weight Hyaluronic Acid. Enzyme Microb Technol. 1996;19:440-5.

[133] Fang X, Duan RS, Yang HY, Liu JF. Hyaluronic Acid Production by Genetic Modified GRAS Strains. Adv Mater Res (Durnten-Zurich, Switz). 2014;950:13-7.

[134] Izawa N, Serata M, Sone T, Omasa T, Ohtake H. Hyaluronic Acid Production by Recombinant Streptococcus thermophilus. J Biosci Bioeng. 2011;111:665-70.

[135] Cheng F, Luozhong S, Guo Z, Yu H, Stephanopoulos G. Enhanced Biosynthesis of Hyaluronic Acid Using Engineered Corynebacterium glutamicum via Metabolic Pathway Regulation. Biotechnol J. 2017;12.

[136] Jia Y, Zhu J, Chen X, Tang D, Su D, Yao W, et al. Metabolic Engineering of Bacillus subtilis for the Efficient Biosynthesis of Uniform Hyaluronic Acid with Controlled Molecular Weights. Bioresour Technol. 2013;132:427-31.

[137] Kaur M, Jayaraman G. Hyaluronan Production and Molecular Weight is Enhanced in Pathway-Engineered Strains of Lactate Dehydrogenase-Deficient Lactococcus lactis. Metab Eng Commun. 2016;3:15-23.

183

References

[138] Jeong E, Shim WY, Kim JH. Metabolic Engineering of Pichia pastoris for Production of Hyaluronic Acid with High Molecular Weight. J Biotechnol. 2014;185:28- 36.

[139] Chien LJ, Lee CK. Enhanced Hyaluronic Acid Production in Bacillus subtilis by Coexpressing Bacterial Hemoglobin. Biotechnol Prog. 2007;23:1017-22.

[140] Chen WY, Marcellin E, Steen JA, Nielsen LK. The Role of Hyaluronic Acid Precursor Concentrations in Molecular Weight Control in Streptococcus zooepidemicus. Mol Biotechnol. 2014;56:147-56.

[141] Ogrodowski CS, Hokka CO, Santana MHA. Production of Hyaluronic Acid by Streptococcus. Appl Biochem Biotechnol. 2005;121-124:753-62.

[142] Armstrong DC, Johns MR. Culture Conditions Affect the Molecular Weight Properties of Hyaluronic Acid Produced by Streptococcus zooepidemicus. Appl Environ Microbiol. 1997;63:2759-64.

[143] Johns R, Goh L-T, Oeggerli A. Effect of pH, Agitation and Aeration on Hyaluronic Acid Production by Streptococcus zooepidemicus. Biotechnol Lett. 1994;16:507-12.

[144] Khue NTH, Vo PTM. Study of Complex Nutrients, Temperature and Salts on Hyaluronic Acid Production In Streptococcus zooepidemicus ATCC 43079. J Appl Pharm Sci. 2013;3:12-5.

[145] Blank LM, McLaughlin RL, Nielsen LK. Stable Production of Hyaluronic Acid in Streptococcus zooepidemicus Chemostats Operated at High Dilution Rate. Biotechnol Bioeng. 2005;90:685-93.

[146] Arnold FH, Volkov AA. Directed Evolution of Biocatalysts. Curr Opin Chem Biol. 1999;3:54-9.

[147] Mills DR, Peterson RL, Spiegelman S. An Extracellular Darwinian Experiment with a Self-Duplicating Nucleic Acid Molecule. Proc Natl Acad Sci USA. 1967;58:217-24.

[148] Stemmer W, Holland B. Survival of the Fittest Molecule. Am Sci. 2003;91:526-33.

[149] Shin H, Cho B-K. Rational Protein Engineering Guided by Deep Mutational Scanning. Int J Mol Sci. 2015;16:23094-110.

[150] Du MHL, Lamoure C, Muller BH, Bulgakov OV, Lajeunesse E, Ménez A, et al. Artificial Evolution of an Enzyme Active Site: Structural Studies of Three Highly Active Mutants of Escherichia coli Alkaline Phosphatase. J Mol Biol 2002;316:941-53.

[151] Dalby PA. Optimising Enzyme Function by Directed Evolution. Curr Opin Struct Biol. 2003;13:500-5.

184

References

[152] Schmidt-Dannert C, Arnold FH. Directed Evolution of Industrial Enzymes. Trends Biotechnol. 1999;17:135-6.

[153] Wong TS, Zhurina D, Schwaneberg U. The Diversity Challenge in Directed Protein Evolution. Comb Chem High Throughput Screen. 2006;9:271-88.

[154] Stemmer WP. Rapid Evolution of a Protein in vitro by DNA Shuffling. Nature. 1994;370:389-91.

[155] Crameri A, Raillard S-A, Bermudez E, Stemmer WPC. DNA Shuffling of a Family of Genes from Diverse Species Accelerates Directed Evolution. Nature. 1998;391:288.

[156] Cirino PC, Mayer K, Umeno D. Generating Mutant Libraries Using Error-Prone PCR. Totowa, NJ: Humana Press; 2003.

[157] Leung DW. A Method for Random Mutagenesis of a Defined DNA Segment Using a Modified Polymerase Chain Reaction. Technique. 1989;1:11-5.

[158] Cadwell RC, Joyce GF. Randomization of Genes by PCR Mutagenesis. PCR Methods Appl. 1992;2:28-33.

[159] Keohavong P, Thilly WG. Fidelity of DNA Polymerases in DNA Amplification. Proc Natl Acad Sci USA. 1989;86:9253-7.

[160] Lin-Goerke JL, Robbins DJ, Burczak JD. PCR-Based Random Mutagenesis Using Manganese and Reduced dNTP Concentration. Biotechniques. 1997;23:409-12.

[161] Yang J, Ruff AJ, Arlt M, Schwaneberg U. Casting epPCR (cepPCR): A Simple Random Mutagenesis Method to Generate High Quality Mutant Libraries. Biotechnol Bioeng. 2017;114:1921-7.

[162] Wong TS, Roccatano D, Zacharias M, Schwaneberg U. A Statistical Analysis of Random Mutagenesis Methods Used for Directed Protein Evolution. J Mol Biol. 2006;355:858-71.

[163] Wong TS, Wong TS, Roccatano D, Schwaneberg U. Challenges of the Genetic Code for Exploring Sequence Space in Directed Protein Evolution. Biocatal Biotransform. 2007;25:229-41.

[164] Wong TS, Tee KL, Hauer B, Schwaneberg U. Sequence Saturation Mutagenesis (SeSaM): A Novel Method for Directed Evolution. Nucleic Acids Res. 2004;32:e26.

[165] Zhao H, Giver L, Shao Z, Affholter JA, Arnold FH. Molecular Evolution by Staggered Extension Process (StEP) in vitro Recombination. Nat Biotechnol. 1998;16:258-61.

185

References

[166] Blanusa M, Schenk A, Sadeghi H, Marienhagen J, Schwaneberg U. Phosphorothioate-Based Ligase-Independent Gene Cloning (PLICing): An Enzyme-Free and Sequence-Independent Cloning Method. Anal Biochem. 2010;406:141-6.

[167] Marienhagen J, Dennig A, Schwaneberg U. Phosphorothioate-Based DNA Recombination: An Enzyme-Free Method for the Combinatorial Assembly of Multiple DNA Fragments. Biotechniques. 2012;52:1-6.

[168] Carter P. Site-Directed Mutagenesis. Biochem J. 1986;237:1-7.

[169] Zheng L, Baumann U, Reymond JL. An Efficient One-Step Site-Directed and Site- Saturation Mutagenesis Protocol. Nucleic Acids Res. 2004;32:e115.

[170] Hemsley A, Arnheim N, Toney MD, Cortopassi G, Galas DJ. A Simple Method for Site-Directed Mutagenesis Using the Polymerase Chain Reaction. Nucleic Acids Res. 1989;17:6545-51.

[171] Steffens DL, Williams JGK. Efficient Site-Directed Saturation Mutagenesis Using Degenerate Oligonucleotides. J Biolmol Tech. 2007;18:147-9.

[172] Reetz MT, Carballeira JD. Iterative Saturation Mutagenesis (ISM) for Rapid Directed Evolution of Functional Enzymes. Nat Protoc. 2007;2:891-903.

[173] Reetz MT, Wang LW, Bocola M. Directed Evolution of Enantioselective Enzymes: Iterative Cycles of CASTing for Probing Protein-Sequence Space. Angew Chem Int Ed Engl. 2006;45:1236-41.

[174] Dennig A, Shivange AV, Marienhagen J, Schwaneberg U. OmniChange: The Sequence Independent Method for Simultaneous Site-Saturation of Five Codons. PLoS One. 2011;6:e26222.

[175] Cheng F, Zhu L, Schwaneberg U. Directed Evolution 2.0: Improving and Deciphering Enzyme Properties. Chem Commun (Cambridge, U K). 2015;51:9760-72.

[176] Körfer G, Pitzler C, Vojcic L, Martinez R, Schwaneberg U. In Vitro Flow Cytometry-Based Screening Platform for Cellulase Engineering. Sci Rep. 2016;6:26128.

[177] Vojcic L, Despotovic D, Martinez R, Maurer K-H, Schwaneberg U. An Efficient Transformation Method for Bacillus subtilis DB104. Appl Microbiol Biotechnol. 2012;94:487-93.

[178] Oueslati N, Leblanc P, Harscoat-Schiavo C, Rondags E, Meunier S, Kapel R, et al. CTAB Turbidimetric Method for Assaying Hyaluronic Acid in Complex Environments and Under Cross-Linked Form. Carbohydr Polym. 2014;112:102-8.

[179] Di Ferrante N. Turbidimetric Measurement of Acid Mucopolysaccharides and Hyaluronidase Activity. J Biol Chem. 1956;220:303-6.

186

References

[180] Sugiyama T. The Mechanism of Specific Staining of Acid Mucopolysaccharide by Alcian Blue: On the Influence of Urea Treatment on the Stainability. Acta Histochem Cytochem. 1968;1:16-30.

[181] Whiteman P. The Quantitative Measurement of Alcian Blue–Glycosaminoglycan Complexes. Biochem J. 1973;131:343-50.

[182] Karlsson G, Bergman R. Determination of the Distribution of Molecular Masses of Sodium Hyaluronate by High-Performance Anion-Exchange Chromatography. J Chromatogr A. 2003;986:67-72.

[183] Krieger E, Koraimann G, Vriend G. Increasing the Precision of Comparative Models with YASARA NOVA—A Self‐Parameterizing Force Field. Proteins Struct Funct Bioinf. 2002;47:393-402.

[184] Becker S, Theile S, Heppeler N, Michalczyk A, Wentzel A, Wilhelm S, et al. A Generic System for the Escherichia coli Cell-Surface Display of Lipolytic Enzymes. FEBS Lett. 2005;579:1177-82.

[185] Wilson DS, Keefe AD. Random Mutagenesis by PCR. Curr Protoc Mol Biol. 2008/02/12 ed2001.

[186] Miyazaki K, Takenouchi M. Creating Random Mutagenesis Libraries Using Megaprimer PCR of Whole Plasmid. Biotechniques. 2002;33:1033-4, 6-8.

[187] Yu H, Tyo K, Alper H, Klein-Marcuschamer D, Stephanopoulos G. A High- Throughput Screen for Hyaluronic Acid Accumulation in Recombinant Escherichia coli Transformed by Libraries of Engineered Sigma Factors. Biotechnol Bioeng. 2008;101:788-96.

[188] Seyfried NT, Blundell CD, Day AJ, Almond A. Preparation and Application of Biologically Active Fluorescent Hyaluronan Oligosaccharides. Glycobiology. 2005;15:303-12.

[189] Bigge JC, Patel TP, Bruce JA, Goulding PN, Charles SM, Parekh RB. Nonselective and Efficient Fluorescent Labeling of Glycans Using 2-Amino Benzamide and Anthranilic Acid. Anal Biochem. 1995;230:229-38.

[190] Holmbeck S, Lerner L. Separation of Hyaluronan Oligosaccharides by the Use of Anion-Exchange HPLC. Carbohydr Res. 1993;239:239-44.

[191] Kooy FK, Ma M, Beeftink HH, Eggink G, Tramper J, Boeriu CG. Quantification and Characterization of Enzymatically Produced Hyaluronan with Fluorophore-Assisted Carbohydrate Electrophoresis. Anal Biochem. 2009;384:329-36.

[192] Jackson P. The Analysis of Fluorophore-Labeled Glycans by High-Resolution Polyacrylamide-Gel Electrophoresis. Anal Biochem. 1994;216:243-52.

187

References

[193] Lee HG, Cowman MK. An Agarose Gel Electrophoretic Method for Analysis of Hyaluronan Molecular Weight Distribution. Anal Biochem. 1994;219:278-87.

[194] Cowman MK, Chen CC, Pandya M, Yuan H, Ramkishun D, LoBello J, et al. Improved Agarose Gel Electrophoresis Method and Molecular Mass Calculation for High Molecular Mass Hyaluronan. Anal Biochem. 2011;417:50-6.

[195] Volpi N, Maccari F. Detection of Submicrogram Quantities of Glycosaminoglycans on Agarose Gels by Sequential Staining with Toluidine Blue and Stains-All. Electrophoresis. 2002;23:4060-6.

[196] Heldermon CD, Tlapak-Simmons VL, Baggenstoss BA, Weigel PH. Site-Directed Mutation of Conserved Cysteine Residues Does Not Inactivate the Streptococcus pyogenes Hyaluronan Synthase. Glycobiology. 2001;11:1017-24.

[197] Breton C, Imberty A. Structure/Function Studies of Glycosyltransferases. Curr Opin Struct Biol. 1999;9:563-71.

[198] Wang C, Entwistle J, Hou G, Li Q, Turley EA. The Characterization of a Human RHAMM cDNA: Conservation of the Hyaluronan-Binding Domains. Gene. 1996;174:299-306.

[199] Silhavy TJ, Kahne D, Walker S. The Bacterial Cell Envelope. Cold Spring Harbor Perspectives in Biology. 2010;2:a000414.

[200] Morgan JL, McNamara JT, Fischer M, Rich J, Chen HM, Withers SG, et al. Observing Cellulose Biosynthesis and Membrane Translocation in Crystallo. Nature. 2016;531:329-34.

[201] Hadley B, Maggioni A, Ashikov A, Day CJ, Haselhorst T, Tiralongo J. Structure and Function of Nucleotide Sugar Transporters: Current Progress. Comput Struct Biotechnol J. 2014;10:23-32.

[202] Yu Z, Qin W, Lin J, Fang S, Qiu J. Antibacterial Mechanisms of Polymyxin and Bacterial Resistance. BioMed Res Int. 2015;2015:11.

[203] Bitter T, Muir HM. A Modified Uronic Acid Carbazole Reaction. Anal Biochem. 1962;4:330-4.

[204] Matsuoka S, Cowman MK. Equation of State for Polymer Solution. Polymer. 2002;43:3447-53.

[205] Liu Y, Bo S, Zhu Y, Zhang W. Determination of Molecular Weight and Molecular Sizes of Polymers by High Temperature Gel Permeation Chromatography with a Static and Dynamic Laser Light Scattering Detector. Polymer. 2003;44:7209-20.

[206] Jing W, Haller FM, Almond A, DeAngelis PL. Defined Megadalton Hyaluronan Polymer Standards. Anal Biochem. 2006;355:183-8.

188

References

[207] Baggenstoss BA, Weigel PH. Size Exclusion Chromatography-Multiangle Laser Light Scattering Analysis of Hyaluronan Size Distributions Made by Membrane-Bound Hyaluronan Synthase. Anal Biochem. 2006;352:243-51.

[208] Zhao J, Kardashliev T, Joëlle Ruff A, Bocola M, Schwaneberg U. Lessons From Diversity of Directed Evolution Experiments by an Analysis of 3,000 Mutations. Biotechnol Bioeng. 2014;111:2380-9.

[209] Tee KL, Schwaneberg U. Directed Evolution of Oxygenases: Screening Systems, Success Stories and Challenges. Comb Chem High Throughput Screening. 2007;10:197- 217.

[210] Laemmli UK. Cleavage of Structural Proteins During the Assembly of the Head of Bacteriophage T4. Nature. 1970;227:680-5.

[211] Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of Image Analysis. Nat Methods. 2012;9:671-5.

[212] Osawa T, Sugiura N, Shimada H, Hirooka R, Tsuji A, Shirakawa T, et al. Crystal Structure of Chondroitin Polymerase from Escherichia coli K4. Biochem Biophys Res Commun. 2009;378:10-4.

[213] Van Durme J, Delgado J, Stricher F, Serrano L, Schymkowitz J, Rousseau F. A Graphical Interface for the FoldX Forcefield. Bioinf. 2011;27:1711-2.

[214] Guerois R, Nielsen JE, Serrano L. Predicting Changes in the Stability of Proteins and Protein Complexes: A Study of More Than 1000 Mutations. J Mol Biol. 2002;320:369-87.

[215] Laskowski RA, MacArthur MW, Moss DS, Thornton JM. PROCHECK: A Program to Check the Stereochemical Quality of Protein Structures. J Appl Crystallogr. 1993;26:283-91.

[216] Sippl MJ. Recognition of Errors in Three-Dimensional Structures of Proteins. Proteins. 1993;17:355-62.

[217] Wiederstein M, Sippl MJ. ProSA-Web: Interactive Web Service for the Recognition of Errors in Three-Dimensional Structures of Proteins. Nucleic Acids Res. 2007;35:W407-10.

[218] Miyamoto S, Kollman PA. SETTLE: An Analytical Version of the SHAKE and RATTLE Algorithm for Rigid Water Models. J Comput Chem. 1992;13:952-62.

[219] Case D, Babin V, Berryman J, Betz R, Cai Q, Cerutti D, et al. The FF14SB Force Field. Amber. 2014;14:29-31.

189

References

[220] Krieger E, Darden T, Nabuurs SB, Finkelstein A, Vriend G. Making Optimal Use of Empirical Energy Functions: Force‐Field Parameterization in Crystal Space. Proteins Struct Funct Bioinf. 2004;57:678-83.

[221] Humphrey W, Dalke A, Schulten K. VMD: Visual Molecular Dynamics. J Mol Graphics. 1996;14:33-8.

[222] Rogers TA, Bommarius AS. Utilizing Simple Biochemical Measurements to Predict Lifetime Output of Biocatalysts in Continuous Isothermal Processes. Chem Eng Sci. 2010;65:2118-24.

[223] Lab U-D. The Structure Analysis and Verification Server Version 4. In: Holton T, editor. UCLA-DOE Lab2017.

[224] Romero-García J, Francisco C, Biarnés X, Planas A. Structure-Function Features of a Mycoplasma Glycolipid Synthase Derived from Structural Data Integration, Molecular Simulations, and Mutational Analysis. PLoS One. 2013;8.

[225] Arango Gutierrez E, Mundhada H, Meier T, Duefel H, Bocola M, Schwaneberg U. Reengineered Glucose Oxidase for Amperometric Glucose Determination in Diabetes Analytics. Biosens Bioelectron. 2013;50:84-90.

[226] Shivange AV, Dennig A, Schwaneberg U. Multi-Site Saturation by OmniChange Yields a pH- and Thermally Improved Phytase. J Biotechnol. 2014;170:68-72.

[227] Lehmann C, Sibilla F, Maugeri Z, Streit WR, Dominguez de Maria P, Martinez R, et al. Reengineering CelA2 Cellulase for Hydrolysis in Aqueous Solutions of Deep Eutectic Solvents and Concentrated Seawater. Green Chem. 2012;14:2719-26.

[228] Iversen PW, Beck B, Chen YF, et al. HTS Assay Validation. In: Sittampalam GS, Coussens NP, Brimacombe Kea, editors. Assay Guidance Manual [Internet]. Bethesda (MD): Eli Lilly & Company and the National Center for Advancing Translational Sciences; 2012.

[229] Jongsareejit B, Bhumiratana A, Morikawa M, Kanaya S. Cloning of Hyaluronan Synthase (sz-has) Gene from Streptococcus zooepidemicus in Escherichia coli. SciAsia. 2007;33:389.

[230] Mao Z, Chen RR. Recombinant Synthesis of Hyaluronan by Agrobacterium sp. Biotechnol Prog. 2007;23:1038-42.

[231] Tian X, Azpurua J, Hine C, Vaidya A, Myakishev-Rempel M, Ablaeva J, et al. High- Molecular-Mass Hyaluronan Mediates the Cancer Resistance of the Naked Mole Rat. Nature. 2013;499:346-9.

[232] Urresti S, Albesa-Jove D, Schaeffer F, Pham HT, Kaur D, Gest P, et al. Mechanistic Insights into the Retaining Glucosyl-3-Phosphoglycerate Synthase from Mycobacteria. J Biol Chem. 2012;287:24649-61.

190

References

[233] Ramakrishnan B, Boeggeman E, Ramasamy V, Qasba PK. Structure and Catalytic Cycle of β-1,4-Galactosyltransferase. Curr Opin Struct Biol. 2004;14:593-600.

[234] Pummill PE, DeAngelis PL. Alteration of Polysaccharide Size Distribution of a Vertebrate Hyaluronan Synthase by Mutation. J Biol Chem. 2003;278:19808-14.

[235] Maclennan AP. The Production of Capsules, Hyaluronic Acid and Hyaluronidase by 25 Strains of Group C Streptococci. J Gen Microbiol. 1956;15:485-91.

[236] Carter GR, Annau E. Isolation of Capsular Polysaccharides from Colonial Variants of Pasteurella multocida. Am J Vet Res. 1953;14:475-8.

[237] Jin P, Kang Z, Yuan P, Du G, Chen J. Production of Specific-Molecular-Weight Hyaluronan by Metabolically Engineered Bacillus subtilis 168. Metab Eng. 2016;35:21- 30.

[238] Liu L, Du G, Chen J, Zhu Y, Wang M, Sun J. Microbial Production of Low Molecular Weight Hyaluronic Acid by Adding Hydrogen Peroxide and Ascorbate in Batch Culture of Streptococcus zooepidemicus. Bioresour Technol. 2009;100:362-7.

[239] Vazquez JA, Montemayor MI, Fraguas J, Murado MA. Hyaluronic Acid Production by Streptococcus zooepidemicus in Marine By-Products Media from Mussel Processing Wastewaters and Tuna Peptone Viscera. Microb Cell Fact. 2010;9:46.

[240] Sambrook J, Russell DW. Molecular Cloning: A Laboratory Manual (3rd ed.). New York: Cold Spring Harbor; 2001.

[241] Eisele A, Zaun H, Kuballa J, Elling L. In Vitro One-Pot Enzymatic Synthesis of Hyaluronic Acid from Sucrose and N-Acetylglucosamine: Optimization of the Enzyme Module System and Nucleotide Sugar Regeneration. ChemCatChem. 2018;0.

[242] Williams KJ, Halkes KM, Kamerling JP, DeAngelis PL. Critical Elements of Oligosaccharide Acceptor Substrates for the Pasteurella multocida Hyaluronan Synthase. J Biol Chem. 2006;281:5391-7.

[243] Grant CL, Pramer D. Minor Element Composition of Yeast Extract J Bacteriol. 1962;84:869-70.

[244] Ouskova G, Spellerberg B, Prehm P. Hyaluronan Release from Streptococcus pyogenes: Export by an ABC Transporter. Glycobiology. 2004;14:931-8.

[245] Fath MJ, Kolter R. ABC Transporters: Bacterial Exporters. Microbiol Rev. 1993;57:995-1017.

[246] Medina AP, Lin J, Weigel PH. Hyaluronan Synthase Mediates Dye Translocation Across Liposomal Membranes. BMC Biochem. 2012;13:1-9.

191

References

[247] Hafsa J, Chaouch MA, Charfeddine B, Rihouey C, Limem K, Le Cerf D, et al. Effect of Ultrasonic Degradation of Hyaluronic Acid Extracted from Rooster Comb on Antioxidant and Antiglycation Activities. Pharm Biol. 2017;55:156-63.

[248] Kubo K, Nakamura T, Takagaki K, Yoshida Y, Endo M. Depolymerization of Hyaluronan by Sonication. Glycoconjugate J. 1993;10:435-9.

[249] Madden M, Song C, Tse K, Wong A. The Inhibitory Effect of EDTA and Mg2+ on the Activity of NADH Dehydrogenase in Lysozyme Lysis. J Exp Microbiol Immunol. 2004;5:8-15.

[250] Berger LR, Weiser RS. The Beta-Glucosaminidase Activity of Egg-White Lysozyme. Biochim Biophys Acta. 1957;26:517-21.

[251] Li S, Wang SS, Fu X, Liu XW, Wang PG, Fang JQ. Sequential One-Pot Multienzyme Synthesis of Hyaluronan and its Derivative. Carbohydr Polym. 2017;178:221-7.

[252] De Luca C, Lansing M, Martini I, Crescenzi F, Shen G-J, O'Regan M, et al. Enzymic Synthesis of Hyaluronic Acid with Regeneration of Sugar Nucleotides. J Am Chem Soc. 1995;117:5869-70.

[253] Mondek J, Kalina M, Simulescu V, Pekař M. Thermal Degradation of High Molar Mass Hyaluronan in Solution and in Powder; Comparison with BSA. Polym Degrad Stab. 2015;120:107-13.

[254] Raetz CR, Dowhan W. Biosynthesis and Function of Phospholipids in Escherichia coli. J Biol Chem. 1990;265:1235-8.

[255] Mamat U, Wilke K, Bramhill D, Schromm AB, Lindner B, Kohl TA, et al. Detoxifying Escherichia coli for Endotoxin-Free Production of Recombinant Proteins. Microb Cell Fact. 2015;14:57.

[256] Schwach-Abdellaoui K, Fuhlendorff BL, Longin F, Lichtenberg J. Development and Applications of a Novel, First-in-Class Hyaluronic Acid from Bacillus. Green Polymer Chemistry: Biocatalysis and Biomaterials: American Chemical Society; 2010. p. 305-11.

192

Declaration

Hiermit versichere ich, dass ich die vorliegende Arbeit selbstständig verfasst und keine anderen als die angegebenen Quellen und Hilfsmittel benutzt habe, dass alle Stellen der Arbeit, die wörtlich oder sinngemäß aus anderen Quellen übernommen wurden, als solchekenntlich gemacht sind und dass die Arbeit in gleicher oder ähnlicher Form noch keiner Prüfungsbehörde vorgelegt wurde.

John Cyrus Baltazar Mandawe Aachen, November 2018

193

Curriculum vitae

Personal Data

Name: John Cyrus Baltazar Mandawe

Date of birth: 17.07.1983

Place of birth: Paraῆaque, Philippines

Nationality: Canadian

Education

2013 –2018 RWTH Aachen University, Aachen, Germany PhD fellow Doctoral Thesis: “Engineering of Hyaluronic Acid Synthases from Streptococcus equi subsp. zooepidemicus and Pasteurella multocida Towards Improved HA Chain Length and Titer”

2007 –2010 York University, Toronto, Ontario, Canada MSc - Biology Master’s Thesis: “Interaction Studies of the Yeast Ubiquitin Specific Processing Protease 15 (Ubp15) N-Terminal Domain”

2002 –2007 York University, Toronto, Ontario, Canada BSc – Biology and Chemistry Bachelor’s Thesis: “Rhythmicity of Diacylglycerol in Neurospora crassa”

1997–2002 Saint Thomas Aquinas Catholic Secondary School, Brampton, Ontario, Canada Ontario Secondary School Diploma

John Cyrus Baltazar Mandawe, Aachen

194