Genomic and Physiological Characterization of Lignin-Derived Aromatic Catabolism Pathways in Roseobacters

University of Tennessee, Knoxville TRACE: Tennessee Research and Creative Exchange

Doctoral Dissertations Graduate School

12-2018

Michelle June Chua University of Tennessee, [email protected]

Follow this and additional works at: https://trace.tennessee.edu/utk_graddiss

Recommended Citation Chua, Michelle June, "Genomic and Physiological Characterization of Lignin-Derived Aromatic Catabolism Pathways in Roseobacters. " PhD diss., University of Tennessee, 2018. https://trace.tennessee.edu/utk_graddiss/5295

This Dissertation is brought to you for free and open access by the Graduate School at TRACE: Tennessee Research and Creative Exchange. It has been accepted for inclusion in Doctoral Dissertations by an authorized administrator of TRACE: Tennessee Research and Creative Exchange. For more information, please contact [email protected]. To the Graduate Council:

I am submitting herewith a dissertation written by Michelle June Chua entitled "Genomic and Physiological Characterization of Lignin-Derived Aromatic Catabolism Pathways in Roseobacters." I have examined the final electronic copy of this dissertation for form and content and recommend that it be accepted in partial fulfillment of the equirr ements for the degree of Doctor of Philosophy, with a major in Microbiology.

Alison Buchan, Major Professor

We have read this dissertation and recommend its acceptance:

Shawn R. Campagna, Elizabeth Fozo, Jill Mikucki, Erik Zinser

Accepted for the Council:

Dixie L. Thompson

Vice Provost and Dean of the Graduate School

(Original signatures are on file with official studentecor r ds.) Genomic and Physiological Characterization of Lignin-Derived Aromatic Catabolism Pathways in Roseobacters

A Dissertation Presented for the Doctor of Philosophy Degree The University of Tennessee, Knoxville

Michelle June Chua December 2018

ACKNOWLEDGEMENTS

I would like to first thank my parents and brother for supporting me and all my decisions. You have been the most reliable people throughout my life, and I am so lucky to have you as my family members.

Secondly, I would like to acknowledge everyone in Knoxville I have formed (what I believe are meaningful) relationships with: former and current members of the Buchan lab, Eric O., Alicia P., Kevin S., Yu-Kai W., and Jia W., the relationships with whom are too long to praise within these acknowledgements. Elizabeth McPherson is included in this group. She has listened to me yammer on for countless hours (something I am obviously not known for) and for some reason, still enjoys seeing me. I am also happy to have joined the Dungeons & Dragons group spear-headed by Alex Grossman. It is not often I feel like I fit in in a group, but the people there make me feel welcome, despite my awkwardness.

I also offer my gratitude to my committee members. Words cannot express how grateful I am that Drs. Alison Buchan, Shawn Campagna, Elizabeth Fozo, Jill Mikucki, and Erik Zinser joined and have stayed on my committee. I have said this before, I thank them all for their patience and helping me (what I hope) grow throughout my graduate school experience. A defining moment for me with Dr. Campagna was when he gave me a pep talk after I failed my first preliminary exam, a moment he may not remember, but is one I will remember for years to come. Thank you, Dr. Fozo, (during the irregular moments we have interacted) for being a kind and caring cheerleader for me. I thank Dr. Mikucki for supporting me (not only with emotional support, but food!) during my time as a graduate student with her and afterwards. I also thank Dr. Zinser for letting me rotate in his lab during my first semester at UT and for continuing to play a role in my dissertation. I very much enjoyed his Microbial Physiology class and non- standard Journal Clubs, which I will miss.

iii

Lastly, I want to acknowledge Dr. Buchan. I am extremely pleased that she let me join her lab. I have not only enjoyed the projects I have worked on, but also my interactions with Dr. Buchan. I appreciate her being patient with me when I am being difficult and also providing me with (emotional) support when I have needed it most. She is also hilarious, which makes my meetings with her and lab meetings quite enjoyable. Thank you, Alison.

ABSTRACT Lignin provides vascular plants with structural rigidity and is the most abundant aromatic polymer on Earth. Lignin is rich in phenolic structures and its structural complexity contributes to its biological recalcitrance, yet it is susceptible to degradation via microbial enzymes. Biogeochemically relevant marine bacteria in the Roseobacter clade are capable of transforming lignin and mineralizing lignin-derived aromatic monomers. Roseobacters are prevalent in temperate and polar oceans, marine sediments, sea ice, and hydrothermal vents, and representatives are readily cultivated in the lab. Roseobacter abundances are frequently highest in coastal oceans, including the salt marshes of the Southeastern United States, where the dissolved organic carbon pool is enriched in aromatic moeities. This dissertation provides new insight into the aromatic compound catabolism potential by members of the Roseobacter lineage through both genomic analyses of publicly available genome sequences and genetic and physiological characterization of the Roseobacter strain Sagitulla stellata. Approximately 85% of the 311 Roseobacter genomes analyzed in this study have genes that encode for enzymes for at least one ring-cleaving pathway. These findings highlight the prevalence of this trait amongst lineage members, but also revealed little correlation between phylogeny, based on 89 conserved protein sequences, and genomic potential. Laboratory-based studies focused on the ligninolytic Roseobacter, S. stellata, with an emphasis on pathways mediating the degradation of the hydroxycinnamates (HCAs) ferulate and p-coumarate, which are important for cross-linking in lignin. Two genes annotated as feruloyl-CoA- synthases, enzymes that catalyze the first step of HCA degradation, were found to have overlapping specificities for the two HCAs in S. stellata, highlighting a functional redundancy not yet reported for other aromatic compound degrading bacteria. Additional studies were performed to examine the physiological response of S. stellata to mixtures of aromatic monomers. A series of studies performed with binary mixtures of aromatic compounds revealed a strong substrate preference for HCAs and their catabolic intermediates relative to v benzoate. Collectively, these data indicate S. stellata is adapted to growing on a mixture of lignin-derived aromatics, which would be advantageous in the aromatic compound-rich salt marshes in which it and its relatives are found.

TABLE OF CONTENTS

CHAPTER I: INTRODUCTION ...... 1 I. Aromatic compounds in the environment ...... 2 Sources ...... 2 Lignin ...... 2 II. Aerobic catabolism of aromatic compounds in bacteria ...... 3 Aromatic catabolism: classic aerobic pathways ...... 3 Aromatic catabolism: aerobic “hybrid” pathways ...... 3 III. Significance of Roseobacter species ...... 4 Roseobacter species in the environment ...... 4 Aromatic catabolism by Roseobacter species ...... 4 Sagittula stellata E-37 ...... 5 IV. Objectives ...... 6 V. References ...... 8 VI. Appendix ...... 12 CHAPTER II: CHARACTERIZATION OF RING-CLEAVING PATHWAYS IN ROSEOBACTER GENOMES ...... 14 I. Abstract ...... 15 II. Introduction ...... 17 III. Methods ...... 20 IV. Roseobacter genomes...... 21 V. Aerobic peripheral pathways ...... 22 Protocatechuate, gentisate, and homogentisate pathways ...... 23 Benzoate and phenylacetate pathways ...... 25 VI. Evolutionary ecology of the marine Roseobacter clade ...... 26 VII. Conclusions ...... 28 VIII. References ...... 30 IX. Appendix ...... 36 CHAPTER III: FUNCTIONAL REDUNDANCY IN THE HYDROXYCINNAMATE CATABOLIC PATHWAYS OF THE SALT MARSH BACTERIUM SAGITTULA STELLATA E-37 ...... 44 I. Abstract ...... 45 II. Importance ...... 47 I. Introduction ...... 48 II. Materials and Methods ...... 51 Strains and growth conditions ...... 51 Generation of S. stellata mutants ...... 52 Biolog™ phenotypic microarrays of fcs mutants ...... 54 RNA isolation and reverse transcription ...... 55 Gene expression analyses ...... 55 Analysis of substrate concentrations ...... 56 Genome analyses ...... 56 III. Results ...... 57 vii

S. stellata possesses two disparately located hydroxycinnamoyl-CoA synthase homologs ...... 57 Genetic analysis reveals overlapping but distinct roles of fcs genes in HCA degradation ...... 59 Gene expression assays indicate fcs genes are growth substrate inducible61 Identification of a transporter for hydroxycinnamic acids in S. stellata ...... 61 IV. Discussion ...... 62 V. Acknowledgements ...... 67 VI. References ...... 68 VII. Appendix ...... 74 CHAPTER IV: HIERARCHICAL SUBSTRATE PREFERENCES BY THE MARINE BACTERIUM SAGITTULA STELLATA DURING GROWTH ON MIXTURES OF PLANT-DERIVED AROMATIC COMPOUNDS ...... 88 I. Abstract ...... 89 II. Introduction ...... 90 III. Materials and Methods ...... 93 Growth conditions and media ...... 93 Analysis of substrate concentrations ...... 94 Ash free dry mass measurements ...... 95 Transposon mutant library generation ...... 95 Genetic screen for mutants deficient in benzoate catabolism ...... 96 Location of Tn5 cassette and DNA sequencing of benzoate- mutants ...... 96 Statistical analyses ...... 97 IV. Results ...... 97 Peripheral pathway aromatic compound catabolism genes are disparately located on the S. stellata genome ...... 98 Substrate preference hierarchy in S. stellata ...... 99 S. stellata biomass yields vary amongst growth substrates ...... 100 box mutants show altered growth phenotypes on ferulate and vanillate .... 100 Gene expression data support the role of the box operon in benzoate degradation ...... 101 V. Discussion ...... 102 VI. Acknowledgements ...... 108 VII. References ...... 109 VIII. Appendix ...... 115 CHAPTER V: CONCLUSION ...... 123 I. References ...... 126 APPENDIX: GENOMIC AND PHYSIOLOGICAL CHARACTERIZATION AND DESCRIPTION OF MARINOBACTER GELIDIMURIAE SP. NOV., A PSYCHROPHILIC, MODERATE HALOPHILE FROM BLOOD FALLS, AN ANTARCTIC SUBGLACIAL BRINE ...... 128 I. Abstract ...... 129 II. Introduction ...... 131 III. Methods ...... 133

viii

Site description and organism isolation ...... 133 Scanning electron microscopy ...... 135 Genome sequence, assembly and annotation ...... 135 Phylogenetic analysis ...... 136 Growth rate experiments ...... 137 Fatty acid analysis ...... 138 Phenotypic analyses and carbon substrate utilization ...... 138 Denitrification and ferrous iron (Fe(II)) oxidation ...... 139 IV. Results and Discussion ...... 140 Morphology ...... 140 Phylogenetic relationships ...... 141 Genome summary ...... 143 Cold adaptation ...... 146 Salt adaptation ...... 150 Metabolic features ...... 152 Nitrate reduction ...... 153 Iron oxidation ...... 154 IV. Concluding remarks ...... 156 V. Acknowledgements ...... 157 VI. References ...... 158 I. Tables ...... 169 II. Figures ...... 172 VITA ...... 181

LIST OF TABLES

Table 2.1. Roseobacter prevalence in aromatic hydrocarbon-enriched environments...... 36 Table 2.2. Summary of criteria used to determine the presence or absence of ring-cleaving pathways in Roseobacter genomes...... 37 Table 2.3. Prevalence of ring-cleaving pathways in Roseobacter genomes...... 38 Table 2.4. Nucleotide and amino acid sequence identities of homogentisate pathway genes in Oceanicola sp. HL-35...... 39 Table 2.5. Nucleotide sequence identities of gentisate pathway genes in C. indicus P73 or M. alba DSM 26384...... 40 Table 2.6. Amino acid sequence identities of gentisate pathway genes in C. indicus P73 or M. alba DSM 26384...... 41 Table 3.1. Plasmids and strains used in this study...... 74 Table 3.2. Primers used in this study...... 75 Table 3.3. E-37 homologs to A. fabrum HCA catabolism proteins in NCBI...... 76 Table 4.1. Strains used in this study...... 115 Table 4.2. Growth rates of Sagittula stellata E-37 on single or binary substrates...... 116 Table 6.1. Comparison of phenotypic characteristics between strain BF04_CF4 and closely related cultured Marinobacter species...... 169 Table 6.2. Cellular fatty acid composition (%) of strain BF04_CF4 and related species of the Marinobacter genus...... 170 Table 6.3. Carbon sources utilized by strain BF04_CF4 and its closest characterized relatives...... 171

LIST OF FIGURES

Figure 1.1. Lignin backbone monomer units (H) p-hydroxyphenyl, (G) guaiacyl, and (S) syringyl...... 12 Figure 1.2. Comparison between aerobic (a) “classical” and (b) “hybrid” aromatic degradation pathways in bacteria...... 13 Figure 2.1. Evolutionary history of aromatic hydrocarbon degradation pathways in Roseobacter genomes...... 42 Figure 2.2. Evolutionary history of aromatic hydrocarbon degradation pathways among Nautella, Phaeobacter, or Leisingera spp...... 43 Figure 3.1. Select HCA degradation pathways initiated by CoA addition...... 77 Figure 3.2. Ferulate and p-coumarate standard curves...... 78 Figure 3.3. Wavelength scans (200-700 nm) of cell-free spent medium from wildtype, fcs1, and fcs2 mutant cultures grown on 2 mM ferulate (A-F) and p- coumarate (G-L)...... 79 Figure 3.4. Genetic landscape of the two fcs loci in S. stellata...... 80 Figure 3.5. Growth of S. stellata on 2 mM vanillin...... 81 Figure 3.6. Growth curves (A-D) and Biolog respiration assays (E-H) of wildtype S. stellata and fcs mutants...... 82 Figure 3.7. Catabolism of hydroxycinnamic acids by S. stellata wildtype and the fcs mutants...... 83 Figure 3.8. Comparison of growth rates of wildtype S. stellata and fcs mutants grown on various carbon sources...... 84 Figure 3.9. fcs gene expression for S. stellata grown on two hydroxycinnamates...... 85 Figure 3.10. Genetic landscape of box operon and upstream genes including fcs2 and the dctP operon...... 86 Figure 3.11. Growth curves of WT E-37 and dctP strains provided with either 2 mM ferulate or p-coumarate as the sole carbon source...... 87 Figure 4.1. Lignin-derived aromatic compounds that are catabolized through protocatechuate or benzoate in Sagittula stellata E-37...... 117 Figure 4.2. Organization of characterized and predicted aromatic compound catabolism genes in Sagittula stellata...... 118 Figure 4.3. Growth profiles and extracellular concentrations of (A) benzoate (2 mM), (B) vanillate (2 mM), (C) benzoate (1 mM) + vanillate (1 mM), (D) ferulate (2 mM), (E) benzoate (1 mM) + ferulate (1 mM), (F) p-coumarate (2 mM), and (G) benzoate (1 mM) + p-coumarate (1 mM) by Sagittula stellata...... 119 Figure 4.4. AFDM yields of Sagittula stellata (100 mL) grown on single or binary mixtures at 2 mM. p-Coum = p-coumarate, Fer = ferulate, Ben = benzoate, and Van = vanillate...... 120 Figure 4.5. Location of insertion sites of Tn5::KmR cassettes in the box operon of two Sagittula stellata mutants...... 121

Figure 4.6. Growth curves of S. stellata and both box mutants on acetate (7 mM) and various lignin-derived aromatic compounds at 2 mM: (A) acetate, (B) benzoate, (C) p-hydroxybenzoate, (D) p-coumarate, (E) vanillate, and (F) ferulate...... 122 Figure 6.1. Scanning electron microscope image of Marinobacter sp. strain BF04_CF4 showing rod-shaped cells and lack of flagella...... 172 Figure 6.2. Maximum likelihood phylogenetic tree showing the relationship between 16S rRNA genes from strain BF04_CF4, closely related Marinobacter clones and isolates, and select Marinobacter type strains. .. 173 Figure 6.3. Syntenic dotplot comparing Marinobacter sp. strains BF04_CF4 (query) and M. psychrophilus (reference) genomes...... 174 Figure 6.4. Comparison of Clusters of Orthologous Groups (COGs) between BF04_CF4 and M. psychrophilus...... 175 Figure 6.5. Effect of temperature (A), NaCl concentration (B), and pH (C) on the growth rates of strain BF04_CF4...... 176 Figure 6.6. Amino acid usage of predicted proteomes of Marinobacter sp. strain BF04_CF4 and Marinobacter lipolyticus SM19T...... 177 Figure 6.7. Amino acid substitution matrix from the alignments of helicase proteins in Marinobacter sp. strain BF04_CF4 and Marinobacter lipolyticus SM19T...... 178 Figure 6.8. Amino acid substitution matrix from the alignments of all homologous proteins in Marinobacter sp. strain BF04_CF4 and Marinobacter lipolyticus SM19T...... 179 Figure 6.9. Summary of the main features discussed in this text from the genomic analysis of BF04_CF4...... 180

xii

CHAPTER I: INTRODUCTION

I. Aromatic compounds in the environment

Sources Aromatic compounds are derived from a variety of biotic and abiotic sources. Major sources are of anthropogenic origin, such as coal or tar, but some microorganisms, including algal species, can also produce aromatic compounds as secondary metabolites (Abdel-Shafy & Mansour, 2016; Sasso et al., 2012; Seo et al., 2009). However, the most abundant source of aromatics in the environment are vascular plants which contain lignin, the most abundant aromatic polymer on Earth (Suhas et al., 2007).

Lignin Lignocellulose is found in all vascular plants and consists of cellulose, hemicellulose, and lignin. The lignin heterogenous biopolymer is the most abundant non-polysaccharide component of lignocellulose (Suhas et al., 2007), and lignin content varies between and even among plant species. Lignin is composed of three phenylpropanoids units: guaiacyl (G), syringyl (S), and p- hydroxyphenyl (H) (Figure 1.1), which are present in varying proportions in different plants. These phenylpropanoid units are derived from three hydroxycinnamyl alcohols (or monolignols): p-coumaryl, coniferyl, and sinapyl alcohols (Vanholme et al., 2010). Ferulate and p-coumarate are two hydroxycinnamates that are incorporated into lignin and plant cells walls. There are several mechanisms by which this occurs: 1) serving as precursors to monolignols, 2) acylated to the hemicellulose portion of lignocellulose, or 3) serving as an initiation site for lignification (a role which has only been demonstrated for ferulate) (Hatfield et al., 2017; Ralph, 2010).

Despite its recalcitrance due to its aromatic nature, lignin can be broken down by fungi and some bacterial species. Fungal degradation of lignin is generally faster and more widespread than bacterial degradation (Janusz et al., 2017). Arguably 2 more is understood about fungal-mediated degradation of lignin, during which extracellular lignin-modifying extracellular enzymes are produced. Canonically, low molecular weight (LMW) aromatic compounds are released from this initial stage of microbial degradation of lignin, either as byproducts or to aid in lignin degradation (Bergbauer & Newell, 1992; Bugg et al., 2011; Janusz et al., 2017; Raj et al., 2007). These LMW lignin-derived aromatics can then be used by other microorganisms in the environment, a process that plays a major role in carbon cycling due to the significant contribution of plants to aromatic abundance in the environment.

II. Aerobic catabolism of aromatic compounds in bacteria

Aromatic catabolism: classic aerobic pathways In aerobic microbial degradation of aromatic compounds, degradation typically proceeds via “catabolic funneling”, wherein diverse aromatic structures are transformed by various peripheral pathways into a limited number of central intermediates that are cleaved by dioxygenases and eventually enter the tricarboxylic acid cycle (Figure 1.2) (Díaz et al., 2013; Fuchs et al., 2011). A paradigm for aerobic aromatic compound catabolism in soil bacteria is the β- ketoadipate pathway. In this pathway, a diverse suite of aromatic compounds is converted to a central intermediate (catechol or protocatechuate), which then undergoes oxygen-dependent ring cleavage. Subsequent reactions give rise to tricarboxylic acid cycle intermediates (Harwood & Parales, 1996; Ornston & Stanier, 1966).

Aromatic catabolism: aerobic “hybrid” pathways An alternative pathway to aerobic aromatic catabolism also utilizes oxygenases, but begins with a CoA ligase to produce a non-aromatic epoxide. The intermediates for these alternative pathways are thioesters, and ring cleavage is performed by hydrolysis rather than using oxygen (Figure 1.2). Furthermore, β- 3 ketoadipyl-CoA is produced as an intermediate, similar to the β-ketoadipate pathway (Díaz et al., 2013; Fuchs et al., 2011). Benzoate and phenylacetate are known to be degraded by the latter pathway, with benzoate being fed through the benzoate oxidation (box) pathway (Gescher et al., 2002; Teufel et al., 2010).

III. Significance of Roseobacter species

Roseobacter species in the environment Bacteria in the Roseobacter clade within the Rhodobacteraceae family of Alphaproteobacteria are found nearly exclusively in marine environments. Widely found in coastal environments and oceans, where they can account for upwards of 20% of bacterioplankton (Wagner-Döbler & Biebl, 2006), this group of bacteria is metabolically diverse, with pathways for nitrate reduction, sulfur oxidation, phototrophy, and what this dissertation focuses on: aromatic catabolism (Luo & Moran, 2014; Newton et al., 2010). Given aromatic catabolism has been found in Roseobacter species, it is unsurprising that Roseobacter have been isolated and detected in environments that have been exposed to aromatic compounds, including oil-polluted sand, oilfield sediment, surface seawater and sediment enriched with aromatics, oil-contaminated microbial mats, Spartina dominated salt marshes, and Emiliania huxleyi blooms (Ankrah et al., 2014; Ansede et al., 2001; Buchan & González, 2010; Suarez-Suarez et al., 2012; Ying et al., 2007).

Aromatic catabolism by Roseobacter species Compared to soil bacteria, aromatic catabolism pathways in marine bacteria are relatively uncharacterized. Recent Roseobacter reviews have explored the presence of aromatic catabolism pathways in Roseobacter genomes, but only examined up to 32 genomes (Buchan & González, 2010; Newton et al., 2010), numbers that are now outdated given that genomes from over 400 bacterial species in the Rhodobacteraceae family can be found in online public databases (Petersen & Wagner-Döbler, 2017). 4

Six pathways for aromatic degradation have been identified in Roseobacter genomes and are referred to according to their central intermediate: protocatechuate, gentisate, benzoate, phenylacetate, homoprotocatechuate, and homogentisate. Benzoate and phenylacetate are the only aromatic compounds listed that can be degraded by aerobic “hybrid” pathways, while the other aromatics are catabolized by canonical aerobic pathways. From the most recent reviews, the most common aromatic pathway present in Roseobacter genomes was the protocatechuate (pca) pathway (found in ~80% of genomes analyzed) (Buchan & González, 2010; Newton et al., 2010). However, this representation may be different given the increase in available Roseobacter genomes in the last eight years.

Sagittula stellata E-37 One strain that has emerged as a model organism for the study of aromatic compounds in Roseobacter species is Sagittula stellata E-37. This bacterium was isolated from seawater off the coast of Georgia, where the vascular plant Spartina alterniflora is the primary producer (Gallagher et al., 1980; Haines, 1976; Schubauer & Hopkinson, 1984). S. stellata was isolated with the high molecular weight portion of pulp mill effluent as its sole carbon source and can partially transform synthetic lignin (Gonzalez et al., 1997). In addition, this strain possesses within its genome all six of the aromatic catabolic pathways previously identified in Roseobacter species. Furthermore, clones and isolates from salt marsh ecosystems have clustered with S. stellata genes (Ansede et al., 2001; Buchan et al., 2001; Dang & Lovell, 2002; Gifford et al., 2011; González et al., 1999). For example, of the Roseobacter isolates tested, the majority of the clones from salt marsh natural and enrichment communities were related to S. stellata pcaH genes (Buchan et al., 2001), genes that encode for the subunit of a protocatechuate ring-cleaving enzyme. Furthermore, the microcosms from which S. stellata was isolated consisted of up to 32% of the total rDNA (González et al., 1999).

There is also physiological evidence for S. stellata’s ability to degrade lignin- derived aromatic compounds, including the aromatic benzoate and other aromatics that feed through the pca pathway. Furthermore, the protocatechuate ring-cleaving enzyme protocatechuate 3,4-dioxygenase is induced by p- hydroxybenzoate (Buchan et al., 2000; Buchan et al., 2004). Further studies have elucidated that protocatechuate 3,4-dioxygenase is also involved in the degradation of ferulate, p-coumarate, quinate, and vanillate, suggesting that these lignin-derived aromatics are degraded through the pca pathway. Moreover, the boxA gene, a box operon gene in S. stellata, was induced on benzoate. S. stellata appears to have also developed an adaptive mechanism to utilize aromatics in the environment, where compounds are typically found in mixtures. S. stellata simultaneously catabolizes benzoate and p-hydroxybenzoate (Gulvik & Buchan, 2013), a metabolic process that appears to be uncommon in other bacteria that have been exposed to a mixture of both aromatics (Brzostowicz et al., 2003; Gaines et al., 1996; Nichols & Harwood, 1995; Pérez-Pantoja et al., 2015). Despite this recent accumulation of knowledge, the genes involved in the peripheral degradation pathways of lignin-derived aromatics by S. stellata are still uncharacterized. Moreover, it is unknown if simultaneous catabolism of lignin- derived aromatics by S. stellata is the standard mechanism for utilizing aromatics in its environment.

IV. Objectives

This dissertation focuses on aerobic catabolism of lignin-derived aromatic compounds in Roseobacter species, starting with a general focus on Roseobacter ring-cleaving pathways and then concentrating on the peripheral pathways involved in the degradation of one of the most abundant class of aromatics associated with lignin, hydroxycinnamates, by S stellata E-37. The individual objectives of each chapter are:

(i) to better characterize the phylogenetic distribution and gain insight into the evolutionary history of the aromatic catabolic pathways in Roseobacter genomes; (ii) to characterize genes involved in hydroxycinnamate catabolism and uptake in the Roseobacter strain Sagittula stellata E-37; (iii) and to continue research on how S. stellata responds to additional binary mixtures of aromatic substrates and the potential mechanism behind the substrate preference hierarchy displayed.

V. References

Abdel-Shafy, H. I., & Mansour, M. S. M. (2016). A review on polycyclic aromatic hydrocarbons: Source, environmental impact, effect on human health and remediation. Egyptian Journal of Petroleum, 25(1), 107-123. doi:https://doi.org/10.1016/j.ejpe.2015.03.011 Ankrah, N. Y. D., Lane, T., Budinoff, C. R., Hadden, M. K., & Buchan, A. (2014). Draft genome sequence of Sulfitobacter sp. CB2047, a member of the Roseobacter clade of marine bacteria, isolated from an Emiliania huxleyi bloom. Genome Announcements, 2(6). Ansede, J. H., Friedman, R., & Yoch, D. C. (2001). Phylogenetic analysis of culturable dimethyl sulfide-producing bacteria from a Spartina-dominated salt marsh and estuarine water. Applied and Environmental Microbiology, 67(3), 1210. Bergbauer, M., & Newell, S. Y. (1992). Contribution to lignocellulose degradation and DOC formation from a salt marsh macrophyte by the ascomycete Phaeosphaeria spartinicola. FEMS Microbiology Letters, 86(4), 341-347. doi:10.1111/j.1574-6968.1992.tb04826.x Brzostowicz, P. C., Reams, A. B., Clark, T. J., & Neidle, E. L. (2003). Transcriptional cross-regulation of the catechol and protocatechuate branches of the β-ketoadipate pathway contributes to carbon source- dependent expression of the Acinetobacter sp. Strain ADP1 pobA gene. Applied and Environmental Microbiology, 69(3), 1598. Buchan, A., Collier, L. S., Neidle, E. L., & Moran, M. A. (2000). Key aromatic- ring-cleaving enzyme, protocatechuate 3,4-dioxygenase, in the ecologically important marine Roseobacter lineage. Applied and Environmental Microbiology, 66(11), 4662. Buchan, A., & González, J. M. (2010). Roseobacter. In K. N. Timmis (Ed.), Handbook of Hydrocarbon and Lipid Microbiology (pp. 1335-1343). Berlin, Heidelberg: Springer Berlin Heidelberg. Buchan, A., Neidle, E. L., & Moran, M. A. (2001). Diversity of the ring-cleaving dioxygenase gene pcaH in a salt marsh bacterial community. Applied and Environmental Microbiology, 67(12), 5801-5809. Buchan, A., Neidle, E. L., & Moran, M. A. (2004). Diverse organization of genes of the β-ketoadipate pathway in members of the marine Roseobacter lineage. Applied and Environmental Microbiology, 70(3), 1658-1668. Bugg, T. D., Ahmad, M., Hardiman, E. M., & Rahmanpour, R. (2011). Pathways for degradation of lignin in bacteria and fungi. Natural product reports, 28(12), 1883-1896. Dang, H., & Lovell, C. R. (2002). Seasonal dynamics of particle‐associated and free‐living marine Proteobacteria in a salt marsh tidal creek as determined using fluorescence in situ hybridization. Environmental microbiology, 4(5), 287-295.

Díaz, E., Jiménez, J. I., & Nogales, J. (2013). Aerobic degradation of aromatic compounds. Current Opinion in Biotechnology, 24(3), 431-442. doi:https://doi.org/10.1016/j.copbio.2012.10.010 Fuchs, G., Boll, M., & Heider, J. (2011). Microbial degradation of aromatic compounds — from one strategy to four. Nature Reviews Microbiology, 9, 803. doi:10.1038/nrmicro2652 Gaines, G. L., Smith, L., & Neidle, E. L. (1996). Novel nuclear magnetic resonance spectroscopy methods demonstrate preferential carbon source utilization by Acinetobacter calcoaceticus. Journal of bacteriology, 178(23), 6833. Gallagher, J. L., Reimold, R. J., Linthurst, R. A., & Pfeiffer, W. J. (1980). Aerial production, mortality, and mineral accumulation-export dynamics in Spartina alterniflora and Juncus roemerianus plant stands in a Georgia salt marsh. Ecology, 61(2), 303-312. doi:10.2307/1935189 Gescher, J., Zaar, A., Mohamed, M., Schägger, H., & Fuchs, G. (2002). Genes coding for a new pathway of aerobic benzoate metabolism in Azoarcus evansii. Journal of bacteriology, 184(22), 6301-6315. Gifford, S. M., Sharma, S., Rinta-Kanto, J. M., & Moran, M. A. (2011). Quantitative analysis of a deeply sequenced marine microbial metatranscriptome. The Isme Journal, 5(3), 461-472. doi:10.1038/ismej.2010.141 González, J., Hodson, R., & Moran, M. (1999). Bacterial populations in replicate marine enrichment cultures: assessing variability in abundance using 16S rRNA-based probes. In Molecular Ecology of Aquatic Communities (pp. 69-75): Springer. Gonzalez, J., Mayer, F., Moran, M., Hodson, R., & Whitman, W. (1997). Sagittula stellata gen. nov., sp. nov., a lignin-transforming bacterium from a coastal environment. International Journal of Systematic and Evolutionary Microbiology, 47(3), 773-780. Gulvik, C. A., & Buchan, A. (2013). Simultaneous catabolism of plant-derived aromatic compounds results in enhanced growth for members of the Roseobacter lineage. Applied and Environmental Microbiology, AEM. 00405-00413. Haines, E. B. (1976). Stable carbon isotope ratios in the biota, soils and tidal water of a Georgia salt marsh. Estuarine and Coastal Marine Science, 4(6), 609-616. doi:https://doi.org/10.1016/0302-3524(76)90069-4 Harwood, C. S., & Parales, R. E. (1996). The β-ketoadipate pathway and the biology of self-identity. Annual Review of Microbiology, 50(1), 553-590. doi:10.1146/annurev.micro.50.1.553 Hatfield, R. D., Rancour, D. M., & Marita, J. M. (2017). Grass cell walls: a story of cross-linking. Frontiers in Plant Science, 7, 2056. Janusz, G., Pawlik, A., Sulej, J., Świderska-Burek, U., Jarosz-Wilkołazka, A., & Paszczyński, A. (2017). Lignin degradation: microorganisms, enzymes

involved, genomes analysis and evolution. FEMS Microbiology Reviews, 41(6), 941-962. doi:10.1093/femsre/fux049 Luo, H., & Moran, M. A. (2014). Evolutionary ecology of the marine Roseobacter clade. Microbiology and Molecular Biology Reviews, 78(4), 573. Newton, R. J., Griffin, L. E., Bowles, K. M., Meile, C., Gifford, S., Givens, C. E., et al. (2010). Genome characteristics of a generalist marine bacterial lineage. The Isme Journal, 4, 784. doi:10.1038/ismej.2009.150 https://www.nature.com/articles/ismej2009150#supplementary-information Nichols, N. N., & Harwood, C. S. (1995). Repression of 4-hydroxybenzoate transport and degradation by benzoate: a new layer of regulatory control in the Pseudomonas putida beta-ketoadipate pathway. Journal of bacteriology, 177(24), 7033. Ornston, L. N., & Stanier, R. Y. (1966). The conversion of catechol and protocatechuate to β-Ketoadipate by Pseudomonas putida : I. biochemistry. Journal of Biological Chemistry, 241(16), 3776-3786. Pérez-Pantoja, D., Leiva-Novoa, P., Donoso, R. A., Little, C., Godoy, M., Pieper, D. H., & González, B. (2015). Hierarchy of carbon source utilization in soil bacteria: Hegemonic preference for benzoate in complex aromatic compound mixtures degraded by Cupriavidus pinatubonensis JMP134. Applied and Environmental Microbiology. Petersen, J., & Wagner-Döbler, I. (2017). Plasmid transfer in the ocean – c Case study from the Roseobacter group. Frontiers in Microbiology, 8(1350). doi:10.3389/fmicb.2017.01350 Raj, A., Krishna Reddy, M. M., & Chandra, R. (2007). Identification of low molecular weight aromatic compounds by gas chromatography–mass spectrometry (GC–MS) from kraft lignin degradation by three Bacillus sp. International Biodeterioration & Biodegradation, 59(4), 292-296. doi:https://doi.org/10.1016/j.ibiod.2006.09.006 Ralph, J. (2010). Hydroxycinnamates in lignification. Phytochemistry Reviews, 9(1), 65-83. doi:10.1007/s11101-009-9141-9 Sasso, S., Pohnert, G., Lohr, M., Mittag, M., & Hertweck, C. (2012). Microalgae in the postgenomic era: a blooming reservoir for new natural products. FEMS Microbiology Reviews, 36(4), 761-785. doi:10.1111/j.1574- 6976.2011.00304.x Schubauer, J. P., & Hopkinson, C. S. (1984). Above- and belowground emergent macrophyte production and turnover in a coastal marsh ecosystem, Georgia. Limnology and Oceanography, 29(5), 1052-1065. doi:10.4319/lo.1984.29.5.1052 Sen, S., Patil, S., & Argyropoulos, D. S. (2015). Thermal properties of lignin in copolymers, blends, and composites: a review. Green Chemistry, 17(11), 4862-4887. Seo, J.-S., Keum, Y.-S., & Li, X. Q. (2009). Bacterial degradation of aromatic compounds. International Journal of Environmental Research and Public Health, 6(1). doi:10.3390/ijerph6010278

Suarez-Suarez, L. Y., Brunet-Galmes, I., Piña-Villalonga, J. M., Christie-Oleza, J. A., Peña, A., Bennasar, A., et al. (2012). Draft senome Sequence of Citreicella aestuarii Strain 357, a member of the Roseobacter clade isolated without xenobiotic pressure from a petroleum-polluted beach. Journal of bacteriology, 194(19), 5464. Suhas, Carrott, P. J. M., & Ribeiro Carrott, M. M. L. (2007). Lignin – from natural adsorbent to activated carbon: A review. Bioresource Technology, 98(12), 2301-2312. doi:https://doi.org/10.1016/j.biortech.2006.08.008 Teufel, R., Mascaraque, V., Ismail, W., Voss, M., Perera, J., Eisenreich, W., et al. (2010). Bacterial phenylalanine and phenylacetate catabolic pathway revealed. Proceedings of the National Academy of Sciences, 107(32), 14390. Vanholme, R., Demedts, B., Morreel, K., Ralph, J., & Boerjan, W. (2010). Lignin biosynthesis and structure. Plant Physiology, 153(3), 895. Wagner-Döbler, I., & Biebl, H. (2006). Environmental biology of the marine Roseobacter lineage. Annual Review of Microbiology, 60(1), 255-280. doi:10.1146/annurev.micro.60.080805.142115 Ying, J.-Y., Wang, B.-J., Dai, X., Yang, S.-S., Liu, S.-J., & Liu, Z.-P. (2007). Wenxinia marina gen. nov., sp. nov., a novel member of the Roseobacter clade isolated from oilfield sediments of the South China Sea. International Journal of Systematic and Evolutionary Microbiology, 57(8), 1711-1716. doi:doi:10.1099/ijs.0.64825-0

VI. Appendix

Figure 1.1. Lignin backbone monomer units (H) p-hydroxyphenyl, (G) guaiacyl, and (S) syringyl. From (Sen et al., 2015).

Figure 1.2. Comparison between aerobic (a) “classical” and (b) “hybrid” aromatic degradation pathways in bacteria. Blue arrows = aromatic ring activation steps; red arrows = aromatic ring cleavage steps; green arrows = central metabolism pathways. OX = ring-hydroxylating oxygenase; DH = dihydrodiol dehydrogenase; DOX = ring-cleavage dioxygenase; EX = aryl-CoA epoxidase. From (Díaz et al., 2013).

CHAPTER II: CHARACTERIZATION OF RING-CLEAVING PATHWAYS IN ROSEOBACTER GENOMES

Portions of the introductory text from this chapter were originally published by Alison Buchan, José M. González, and Michelle J. Chua:

Alison Buchan, José M. González, and Michelle J. Chua. “Roseobacter.” Handbook of Hydrocarbon and Lipid Microbiology 2 (2018).

The genomic analyses presented in this chapter will be published by Michelle J. Chua, Christopher Gulvik, José M. González, and Alison Buchan.

Author contribution statement: Michelle Chua performed the genome analyses, with support from Dr. Alison Buchan. Dr. José M. González generated Figure 2.1. Dr. Christopher Gulvik generated the box data from the Norwegian fjord samples. Michelle Chua led manuscript preparation, with support from Alison Buchan.

I. Abstract

With over 70 genera identified to date, the Roseobacter lineage falls within the Rhodobacteraceae family of the Alphaproteobacterial class. The overwhelming majority of cultivated representative strains have been isolated from marine environments and demonstrate metabolic versatility. Roseobacters are abundant in diverse marine environments enriched in aromatic compounds, including terrestrially derived organic matter-influenced coastal regions, algal blooms, and sites anthropogenically contaminated with aromatic hydrocarbons. Prior work has demonstrated that several Roseobacter strains are capable of utilizing a variety of aromatic monomers as primary growth substrates via ring-cleaving pathways well characterized in soil bacteria, but less studied in the marine Roseobacter lineage. Here, we provide a genomic and phylogenetic perspective into Roseobacter aromatic compound catabolic capabilities through the analysis of 311 genomes. The Roseobacter genomes ranged in size from 2.7 Mbp to 6.1 Mbp. Less than 6% of the genomes were closed, and the majority of the genomes are derived from cultured isolates. This analysis focused on five 15 complete ring-cleaving pathways. With approximately 86% of the genomes possessing diagnostic genetic markers, the homogentisate degradation pathway was the most prevalent in the dataset. In contrast, the benzoate oxidation and gentisate pathways were present in <20% of the genomes. Overall, a clear relationship between the presence of aromatic ring-cleaving genes and Roseobacter phylogeny, based on 89 conserved protein sequences, was not observed. The single exception was within a subgroup (Nautella, Phaeobacter, or Leisingera spp.) that contained ~45 genomes, most of which appear to possess complete pathways for the degradation of protocatechuate, homogentisate, and phenylacetate. This finding suggests that these aromatic catabolism genes have been gained and lost from Roseobacter genomes multiple times, perhaps in adapting to their ecological niches.

II. Introduction

The Roseobacter lineage consists of bacteria in the α-3 subclass of Proteobacteria that share >89% identity of the 16S rRNA gene sequence. Since the initial discovery and characterization of two Roseobacter strains in 1991 (Shiba, 1991), over 70 genera have been described within the Roseobacter clade in the Department of Energy’s Joint Genome Institute’s Integrated Microbial Genomes System. Predominantly found in marine environments, Roseobacter strains can comprise up to 20% of total bacteria in coastal environments and seawater (Buchan et al., 2005; Moran et al., 2007). This group of bacteria has also been isolated and detected in diverse environments, including associated with or as symbionts of aquatic macroorganisms, Antarctic lakes, and hypersaline soils (Brinkhoff et al., 2008; Buchan et al., 2005; Luo & Moran, 2014; Wagner-Dobler & Biebl, 2006). In addition to their abundance in natural systems, Roseobacters have emerged as excellent model systems due to the relative ease of culturing and genetic tractability (Christie-Oleza & Armengaud, 2015).

Members of the Roseobacter genus are metabolically diverse. Of broad interest are the sulfur and carbon transformations mediated by Roseobacters that impact biogeochemical processes (Luo & Moran, 2014). The particular focus of this chapter is the ability of Roseobacters to degrade naturally occurring aromatic compounds. The process of microbial degradation of aromatic compounds is typically performed via a process termed catabolic funneling. In microbial catabolic funneling, an array of aromatic compounds is transformed, via peripheral pathways, to one of a limited number of central intermediates that are then subject to ring cleavage, via lower central pathways. Subsequent reactions lead to tricarboxylic acid cycle intermediates. In oxic environments, microbes typically harness the power of molecular oxygen to cleave the aromatic ring. Most of the common ring-cleaving pathways examined here belonged to the canonical aerobic pathways that feed through β-ketoadipate, where a variety of

17 aromatic compounds is converted to a central intermediate (e.g. protocatechuate) that then undergoes oxygen-dependent ring cleavage. but two out of the five pathways use oxygen to produce an epoxide as the first intermediate, which is then processed as CoA thioesters that undergo hydrolysis- dependent ring cleavage.

Roseobacters are common in terrestrially derived organic matter-influenced coastal regions in the Arctic and southeastern region of the United States, which are rich in lignin-derived phenolics (Quigley, 2018; Sipler et al., 2017). For example, in southeastern coastal salt marshes the dissolved organic carbon pool is ~30% aromatic in nature (Moran & Hodson, 1990). Furthermore, Roseobacters have been found in close association with the microalgae Emiliania huxleyi, which produces p-coumaric acid as it ages, and other marine microalgae that can naturally produce the aromatic sinapic acid (Gonzalez et al., 2000; Seyedsayamdost et al., 2011; Simon et al., 2017; Stoica & Herndl, 2007; Wang & Seyedsayamdost, 2017). Roseobacters have been identified in deep sea hydrothermal vents where aromatic compounds can form hydrothermally (Crépeau et al., 2011; Ding et al., 2017; Wang et al., 2011). Lastly, Roseobacter strains can be exposed to aromatic compounds in oil-contaminated marine environments such as the Deepwater Horizon oil spill in the Gulf of Mexico (Buchan & Gonzalez, 2010; Yang et al., 2016).

Studies performed within the last two decades demonstrate that Roseobacter species are members of communities in aromatic hydrocarbon-contaminated environments. Compared to culture-dependent studies, culture-independent- based techniques have been more successful in identifying the presence of Roseobacter strains in environments contaminated with aromatic hydrocarbons (Table 2.1). This trend is particularly evident from Deepwater Horizon Spill studies (Kimes et al., 2013; Lamendella et al., 2014; Yang et al., 2016). For example, in sediment samples from the wellhead area, Roseobacters were not 18 detected in pre-spill sites, but contributed up to 20% of the clone library post-oil spill (Yang et al., 2016). The majority of the amendment-based studies demonstrated that the addition of aromatic hydrocarbons enriched for Roseobacter strains (Chronopoulou et al., 2015; Jimenez et al., 2011; Mishamandani et al., 2016; Sanni et al., 2015; Viggor et al., 2013; Zhou et al., 2009). While Roseobacter strains were detected in the original non-amended samples, these members persisted as the incubations continued, suggesting that these bacteria benefit from aromatic hydrocarbon addition (Jimenez et al., 2011; Viggor et al., 2013). In one amendment study, a bacterium closely related to Roseovarius spp. was found in association with a marine diatom, Skeletonema costatum, when crude oil was added to the culture (Mishamandani et al., 2016).

While clones related to Roseobacter were found in bacterial communities associated with decaying S. alterniflora, and several Roseobacter strains have been isolated from seawater using the high molecular weight fraction of pulp mill effluent, comparatively little is known of the distribution of Roseobacters in environments with other biotic sources of aromatics. In one recent study, flow- through systems were used to test the ability of marine or estuarine bacterial communities to decompose humic acids from peat bog freshwater rich in humic substances (Rocker et al., 2012). Approximately 38% (18/47) of the DGGE phylotypes belonged to the Alphaproteobacteria, and 61% of these phylotypes were affiliated with the Roseobacter clade. Of the isolates groups, 43% (6/14) were associated with Roseobacters. Furthermore, out of the 13 isolates that were tested for the degradation of aromatic compounds, three were Roseobacter strains and capable of degrading lignin-derived aromatics, which the authors suggested was an indicator for humic acid degradation. Further studies are required to determine a clearer role of Roseobacters in biotic aromatic degradation in the environment.

These aforementioned environmental studies provide evidence for the role of Roseobacters in aromatic compound degradation in largely marine niches. Additional research with cultivated strains has demonstrated that several Roseobacter strains are capable of utilizing a variety of aromatic monomers as primary growth substrates via ring-cleaving pathways well characterized in soil microbes (Buchan et al., 2004). Given the prevalence of Roseobacter genomes in public databases, a more comprehensive analysis of the genomic potential of lineage members to degrade aromatic compounds is now feasible. To that end, we conducted a genomic and phylogenetic characterization of 311 Roseobacter genomes. Efforts were focused on evaluating these genomes for the presence of 5 ring-cleaving pathways.

III. Methods

Three hundred and eleven Roseobacter genomes were obtained from the JGI Integrated Microbial Genomes System in June 2016 and included in this analysis. We analyzed the genomes for the presence of the protocatechuate (pca), gentisate (gdo), benzoate oxidation (box), homogentisate (hgd), and phenylacetate (paa) pathways by searching for the pcaH, nagI, boxB, hmgA, and paaA genes, respectively. Homologous genes were searched using the BLASTp program in the Joint Genome Institute Integrated Microbial Genomes system with the cutoff E-value of 10-5. Reference sequences for the genes involved in the pca, gdo, box, and paa pathways were obtained from the Roseobacter Sagittula stellata E-37 genome (GCA_000169415.1) and the hgd pathway from the Roseobacter Citreicella sp. SE45 genome (GCF_000161755.1). Organisms were considered positive for the pca pathway if (1) the pcaC (γ-carboxymuconolactone decarboxylase) gene was adjacent to the pcaHG (protocatechuate 3,4- dioxygenase) subunit, as this is a common feature of previously characterized strains (Buchan et al., 2004) and (2) the pcaB (β-carboxy-cis,cis-muconate cycloisomerase) and pcaD (β-ketoadipate enol-lactone hydrolase) genes were 20 located on the same operon, often at a distal site on the chromosome; for the gdo pathway, the presence of nagI (gentisate 1,2-dioxygenase), nagL (maleylpyruvate isomerase), and nagK (fumarylacetoacetate hydrolase) within a strain’s genome, which are found co-localized in other bacterial genomes (Jung et al., 2011; Lee et al., 2011; Shiba, 1991; Tomás-Gallardo et al., 2013); the box pathway: the presence of boxB (benzoyl-CoA oxygenase), boxA (benzoyl-CoA reductase), boxC (2,3-epoxybenzoyl-CoA dihydrolase), boxL (benzoate-CoA ligase), and boxR (XRE family transcriptional regulator), often found in one operon (Gescher et al., 2005; Schühle et al., 2003); the hgd pathway: the presence of the hmgAB (homogentisate dioxygenase and fumarylacetoacetate hydrolase, respectively) subunit and hmgC (maleylacetoacetate isomerase), found clustered together in other bacteria (Arias-Barrau et al., 2004; Mendez et al., 2011). The paa pathway: the presence of the paaABCDE (1,2-phenylacetyl- CoA multicomponent epoxidase) subunit and paaZ (ring-cleaving oxepin-CoA hydrolase), genes that are often clustered together (Martin & McInerney, 2009) (Table 2.2). Among the Roseobacter genomes with nearly complete aromatic ring-cleaving pathways, it is likely due to the (permanent) draft genome status given that accumulation of some aromatic ring intermediates can be toxic to bacterial cells (George & Hay, 2011). Some genes for the complete pathways were excluded as they are involved in the generation of tricarboxylic acid cycle intermediates, roles that are not specific to only one aromatic degradation pathway (Arias-Barrau et al., 2004; Buchan et al., 2004; Gescher, Johannes et al., 2005; Jung et al., 2011; Lee et al., 2011; Martin & McInerney, 2009; Mendez et al., 2011; Schuhle et al., 2003; Shiba, 1991; Tomás-Gallardo et al., 2013).

IV. Roseobacter genomes

Of the 311 Roseobacter genomes that were analyzed, the smallest genome size was 1.9 Mbp, the median was 4.1 Mbp, and the largest was 6.1 Mbp. The majority of the genomes were either draft or permanent draft status, with only 21

5.5% closed. The number of predicted genes ranged from 3,125 genes (Ketogulonicigenium vulgare WSH-001: 3.3 Mbp) to 5,149 genes (Octadecabacter arcticus 238, DSM 13978: 5.5 Mbp). Most of the genomes belonged to cultured isolates (95.18%; 296/311), while the 15 uncultured genomes were obtained from metagenomes or single cell sequencing. Isolation sites were largely marine, and 46% (143/311) were type strains from validly described strains. Roseobacter genomes have been previously shown to possess genes that encode for a variety of metabolic capabilities, including aerobic anoxygenic photoheterotrophy, nitrate reduction, and sulfur and/or carbon monoxide oxidation (Luo & Moran, 2014). Furthermore, horizontal gene transfer agents are common among Roseobacter genomes, with 46 out of 52 cultured Roseobacter genomes demonstrated to contain a (nearly) complete set of genes that encode for gene transfer agents (Luo & Moran, 2014), potential forms of gene acquisition in microorganisms. Many Roseobacter strains have been shown to possess plasmids, and some Roseobacters harbor genes encoding for aromatic compound catabolism, biofilm formation, secondary metabolite production, and phototrophy (Beyersmann et al., 2013; Dogs et al., 2013; Frank et al., 2014; Petersen et al., 2012; Wagner-Dobler et al., 2010). These plasmids can constitute up to 33% of a total genome, with 12 plasmids in several Marinovum algicola strains (Pradella et al., 2010). Many plasmids also contain genes for the Type IV secretion systems, which have been demonstrated to play a role in conjugation (Patzelt et al., 2016). Roseobacter genome content suggests that these bacteria have specific environmental adaptations, which could potentially be horizontally transferred and thus drive Roseobacter evolution.

V. Aerobic peripheral pathways

Aerobic degradation of aromatic compounds typically begins with enzymes of peripheral (or upper) pathways transforming aromatics to a few central 22 intermediates, which are then cleaved and converted to tricarboxylic acid cycle intermediates. The paradigm for the peripheral pathways involves oxygenases that mediate the transformation of a broad suite of aromatic compounds into one of a limited number of intermediates, generally dihydroxylated aromatic compounds, such as catechol, protocatechuate, gentisate, or homogentisate. An alternative mechanism for aerobic aromatic catabolism involves the introduction of oxygen through epoxidases to produce benzoyl-CoA or phenylacetyl-CoA, which are then converted to tricarboxylic acid cycle intermediates through the production of thioesters (Fuchs et al., 2011). Regardless of how oxygen is introduced, in all the aerobic catabolism pathways described here, β-ketoadipyl- CoA is produced as an intermediate to form acetyl-CoA and succinyl-CoA (Fuchs et al., 2011; Gescher et al., 2002; Grishin & Cygler, 2015).

Protocatechuate, gentisate, and homogentisate pathways In most of the aerobic central catabolism pathways discussed herein, dioxygenases are responsible for cleaving the aromatic ring intermediates protocatechuate, gentisate, or homogentisate (e.g. Figure 1.2). These pathways converge to form β-ketoadipate, a precursor to β-ketoadipyl-CoA and the intermediate which the β-ketoadipate pathway is named after. The most prevalent pathway found in Roseobacter genomes was the homogentisate pathway (73.95%) (Table 2.3). There are three major genes (hmgABC) involved in homogentisate metabolism (Arias-Barrau et al., 2004). Homogentisate dioxygenase (HmgA) is responsible for homogentisate ring cleavage to produce maleylacetoacetate, which is then isomerized by maleylacetoacetate isomerase (HmgC) to form fumarylacetoacetate. Fumarylacetoacetate hydrolase (HmgB) then converts fumarylacetoacetate to fumarate and acetoacetate, which are central carbohydrate intermediates. hmgA and hmgB were always found in the same operon, and often immediately adjacent in the Roseobacter genomes in which they are found, while hmgC was localized to a different region of the genome in some Roseobacter strains. Oceanicola sp. HL-35 (63.9% GC content)

23 is unique in it has two complete copies of this pathway, while the other strains had only one copy. The GC contents of the hmgA genes were 61.9% (K224DRAFT_1299) and 62.6% (K224DRAFT_2591), suggesting these genes were not horizontally transferred. There were no significant nucleotide sequence similarities found between the genes predicted to have the same function; however, the amino acid sequence identities were 62%, 53%, and 36% percent for the hmgABC genes, respectively (Table 2.4). While hmgR, a IclR-type transcriptional regulator that acts as a repressor, has been found in other bacterial strains (Arias-Barrau et al., 2004), no homologs of the hmgR gene were identified in the Roseobacter genomes.

The intermediates and products of gentisate degradation are similar to homogentisate catabolism, with the exception of pyruvate instead of acetate derivatives. After searching for the presence of three genes (nagILK), it was revealed that 17.68% of the Roseobacter genomes possessed the gentisate pathway (Table 2.3). While most of the Roseobacter genomes only contained one copy of each gene for this pathway, Celeribacter indicus P73 and Mameliella alba DSM 26384 possessed three complete copies of this pathway. The genes of one copy of a Celeribacter indicus P73 gdo pathway had a GC content that was ~7.8% lower than the Celeribacter indicus P73 genome (65.74%), suggesting these genes may be horizontally transferred (Rocha & Danchin, 2002). P73_4773- P73_4775 were also the genes that had the lowest nucleotide and amino acid sequence identities to their homologous genes (Tables 2.5-2.6).

Over half (63.67%) of the Roseobacter genomes contained genes involved in the protocatechuate pathway (Table 2.3). Protocatechuate 3,4-dioxygenase (encoded by pcaHG) is a well-characterized protein of this pathway that is responsible for catalyzing the ring cleavage of protocatechuate. pcaHG were found as adjacent pairs in all Roseobacter genomes examined. Furthermore, the gene encoding for p-hydroxybenzoate hydroxylase (PobA), the enzyme 24 responsible for the conversion of p-hydroxybenzoate to protocatechuate, was frequently found either up- or downstream (usually within 2 kb) of the pcaHG genes. The location of the pcaB, -C, and -D genes varied among the Roseobacter genomes (e.g. as pcaDCHGB, pcaDB, pcaDCB, or pcaCHG), which corroborates what has been studied previously (Alejandro-Marin et al., 2014; Buchan et al., 2004).

Benzoate and phenylacetate pathways In the alternative ring-cleaving pathways examined here, CoA thioesters serve as the intermediates, with the first committed step of each pathway being catalyzed by a ligase enzyme. In these pathways, the CoA intermediates are catabolized to form β-ketoadipyl-CoA, a shared intermediate with the β-ketoadipate pathway (e.g. Figure 1.2). The evolutionary advantage to these pathways has not yet been determined, but has been proposed to be useful under low or fluctuating oxygen concentrations (Fuchs et al., 2011).

Of the Roseobacter genomes examined, 16.61% contained aerobic benzoate oxidation genes, making it the least prevalent pathway among Roseobacter strains (Table 2.3). Generally, the genes encoding for benzoyl-CoA 2,3- epoxidase were separated by a conserved hypothetical protein. The genes that encode for 3,4-dehydroadipyl-CoA semialdehyde dehydrogenase (boxD; converts 3,4-dehydroadipyl-CoA semialdehyde to 3,4-dehydroadipyl-CoA) and β- ketoadipyl-CoA thiolase (boxE; converts β-ketoadipyl CoA to acetyl-CoA and succinyl-CoA) were not detected in the box operon, but the closest homologs were paaZ and paaJ, respectively, genes encoding for proteins involved in the phenylacetate catabolism pathway.

Experimental evidence of the box pathway has largely been described in cultured isolates including two Azoarcus spp. (Gescher et al., 2002; Valderrama et al., 2012), Bacillus stearothermophilus (Zaar et al., 2001), Burkholderia xenovorans

LB40 (Denef et al., 2006), Sagittula stellata E-37 (Gulvik & Buchan, 2013), and Thauera aromatica (Schuhle et al., 2003). Recently, the prevalence and genetic diversity of the box pathway in the natural environment was explored in samples from a Norwegian fjord. Using qPCR, it was determined that boxB abundance was 0.33 - 7.2% relative to bacterial-specific 16S rDNA copies in the fjord, while 37% and 50% of the boxB metagenome sequences analyzed were related to Roseobacters and Betaproteobacteria, suggesting that the box pathway is an environmentally relevant method for aromatic catabolism despite its low prevalence among Roseobacter genomes.

Genes for the phenylacetate pathway were found in a little over half (55.95%) of the Roseobacter genomes (Table 2.3). paaABCDE was found to be colocalized in Roseobacter genomes and have been demonstrated to encode the five- component 1,2-phenylacetyl-CoA epoxidase, the enzyme responsible for introducing oxygen into phenylacetyl-CoA in other bacteria (Grishin & Cygler, 2015). PaaZ (oxepin-CoA hydrolase) catalyzes the ring cleavage step of the paa pathway, and its gene was either found adjacent to paaABCDE or paaK and/or paaG, genes that encode for phenylacetate-coenzyme A ligase and the ring- opening phenylacetate-CoA oxygenase, respectively.

VI. Evolutionary ecology of the marine Roseobacter clade

In general, there did not appear to be a distinct relationship between the presence of aromatic degradation genes and phylogeny, suggesting different evolutionary histories of these pathways and Roseobacter species conservation. It is possible that most Roseobacter strains in this study have separately adapted to their individual isolation sites, perhaps acquiring genes from horizontal gene transfer, a form of gene acquisition that is abundant in Roseobacter strains (Beyersmann et al., 2013; Dogs et al., 2013; Frank et al., 2014; Luo & Moran, 2014; Petersen et al., 2012; Wagner-Dobler et al., 2010). However, a few distinct 26 trends were observed and highlighted in Figure 2.1. Thalassobacter and Octadecabacter spp. strains do not have genes for any of the aromatic catabolism pathways, while genome analysis revealed that Mameliella alba strains can degrade all the aromatics explored in this study. Maribius spp. and, with the exception of Ketogulonicigenium vulgare strain SPU B805, Ketogulonicigenium vulgare strains only have genes responsible for the pca pathway. In contrast, Maritimibacter spp. have genes encoding for all the pathways except for protocatechuate degradation. Lastly, Aliiroseovarius spp. appear to have the capability to degrade homogentisate and phenylacetate. Mameliella alba and Ketogulonicigenium vulgare strains are closely related within their respective clades, suggesting that gene acquisition might have occurred with speciation.

What is intriguing and the most striking observation, is the presence of one clade of the Roseobacter tree that contains genomes with genes for the complete pathways of protocatechuate, homogentisate, and phenylacetate degradation. Within this large clade are three smaller distinct clades, including clustering of Nautella, Phaeobacter, or Leisingera spp (Figure 2.2). The gene synteny among the Leisingera strains is unsurprising given that half of the isolates were cultured from the same or similar environment, an accessory nidamental gland of the Hawaiian bobtail squid, Euprymna scolopes (Collins et al., 2015). The relationship among the Nautella and Phaeobacter clades were less defined. While the Nautella spp. isolation sites were marine, they included biofilms or were associated with gilthead seabream larvae or an algal bleaching disease (Rodrigo-Torres et al., 2016b). The Phaebacter spp. were also isolated from variable sites, but 3 (Phaeobacter gallaeciensis BS107 and DSM 17395 and 26640) out of 10 were obtained from seawater containing scallop Pecten maximus larvae (Frank et al., 2014). Despite the relatively more recent discovery of the box pathway, there does not appear to be a clear relationship between

27 genes for individual aromatic degradation pathways and speciation timelines, perhaps indicating that genes for the box pathway were not recently acquired.

VII. Conclusions

With the increased number of sequenced Roseobacter genomes, we were provided a more in-depth perspective into trends and evolutionary relationships between Roseobacter genomes and their aromatic hydrocarbon degradation capabilities. Knowledge of the different aromatic degradation pathways does not necessarily correlate with prevalence, as the homogentisate and benzoate oxidation pathways are the most common and least abundant pathways in Roseobacter strains, respectively. For example, while there is evidence for homogentisate degradation in Roseobacter isolates in this review and other sources of literature (Buchan & Gonzalez, 2010; Moran, M. A. et al., 2007; Newton et al., 2010; Rodrigo-Torres et al., 2016a, 2016b; Zech et al., 2013), there does not appear to be knowledge of the specific steps and regulation of this pathway in Roseobacter strains, a trend that may need to be addressed given homogentisate degradation prevalence. Similar to the previous review, this updated literature also provides strong evidence for Roseobacter aromatic carbon degradation in the natural environment, and presence of aromatics appear to promote Roseobacter abundance. Furthermore, future studies should be performed to confirm that the Roseobacter strains in this study can degrade their respective aromatics and perhaps to determine the prevalence of these strains and their associated degradation pathways in the natural environment, particularly since generally, there was no clear relationship between phylogeny and aromatic hydrocarbon degradation capabilities. Exploration of these trends should also be increased in environments where lignin-derived aromatics are abundant, given that lignin is the most abundant phenolic polymer on Earth. It may also be useful to explore evolutionary relationships using the protein sequences of the aromatic compound degradation genes, an approach that is 28 beyond the scope of this study, but may confirm the relationships that were revealed in this review. As more resources and technological advances become available, a more updated pool of knowledge will emerge to fill these gaps and contribute to a more complete understanding of Roseobacter aromatic hydrocarbon degradation in natural ecosystems.

VIII. References

Alejandro-Marin, C. M., Bosch, R., & Nogales, B. (2014). Comparative genomics of the protocatechuate branch of the beta-ketoadipate pathway in the Roseobacter lineage. Mar Genomics, 17, 25-33. doi:10.1016/j.margen.2014.05.008 Arias-Barrau, E., Olivera, E. R., Luengo, J. M., Fernandez, C., Galan, B., Garcia, J. L., et al. (2004). The homogentisate pathway: a central catabolic pathway involved in the degradation of L-phenylalanine, L-tyrosine, and 3- hydroxyphenylacetate in Pseudomonas putida. J Bacteriol, 186(15), 5062- 5077. doi:10.1128/JB.186.15.5062-5077.2004 Beyersmann, P. G., Chertkov, O., Petersen, J., Fiebig, A., Chen, A., Pati, A., et al. (2013). Genome sequence of Phaeobacter caeruleus type strain (DSM 24564(T)), a surface-associated member of the marine Roseobacter clade. Stand Genomic Sci, 8(3), 403-419. doi:10.4056/sigs.3927623 Brinkhoff, T., Giebel, H. A., & Simon, M. (2008). Diversity, ecology, and genomics of the Roseobacter clade: a short overview. Archives of Microbiology, 189(6), 531-539. doi:10.1007/s00203-008-0353-y Buchan, A., & Gonzalez, J. M. (2010). Roseobacter. In K. N. Timmis (Ed.), Handbook of Hydrocarbon and Lipid Microbiology (pp. 1335–1343): Springer-Verlag Berlin Heidelberg. Buchan, A., Gonzalez, J. M., & Moran, M. A. (2005). Overview of the marine Roseobacter lineage. Applied and Environmental Microbiology, 71(10), 5665-5677. doi:10.1128/Aem.71.10.5665-5677.2005 Buchan, A., Neidle, E. L., & Moran, M. A. (2004). Diverse organization of genes of the beta-ketoadipate pathway in members of the marine Roseobacter lineage. Appl Environ Microbiol, 70(3), 1658-1668. Christie-Oleza, J. A., & Armengaud, J. (2015). Proteomics of the Roseobacter clade, a window to the marine microbiology landscape. Proteomics, 15(23- 24), 3928-3942. doi:10.1002/pmic.201500222 Chronopoulou, P. M., Sanni, G. O., Silas-Olu, D. I., van der Meer, J. R., Timmis, K. N., Brussaard, C. P., & McGenity, T. J. (2015). Generalist hydrocarbon- degrading bacterial communities in the oil-polluted water column of the North Sea. Microb Biotechnol, 8(3), 434-447. doi:10.1111/1751- 7915.12176 Collins, A. J., Fullmer, M. S., Gogarten, J. P., & Nyholm, S. V. (2015). Comparative genomics of Roseobacter clade bacteria isolated from the accessory nidamental gland of Euprymna scolopes. Frontiers in Microbiology, 6(123). doi:10.3389/fmicb.2015.00123 Crépeau, V., Cambon Bonavita, M.-A., Lesongeur, F., Randrianalivelo, H., Sarradin, P.-M., Sarrazin, J., & Godfroy, A. (2011). Diversity and function in microbial mats from the Lucky Strike hydrothermal vent field. FEMS Microbiology Ecology, 76(3), 524-540. doi:10.1111/j.1574- 6941.2011.01070.x 30

Denef, V. J., Klappenbach, J. A., Patrauchan, M. A., Florizone, C., Rodrigues, J. L. M., Tsoi, T. V., et al. (2006). Genetic and genomic insights into the role of benzoate-catabolic pathway redundancy in Burkholderia xenovorans LB400. Applied and Environmental Microbiology, 72(1), 585-595. Ding, J., Zhang, Y., Wang, H., Jian, H., Leng, H., & Xiao, X. (2017). Microbial community structure of deep-sea hydrothermal vents on the ultraslow spreading Southwest Indian Ridge. Frontiers in Microbiology, 8, 1012. doi:10.3389/fmicb.2017.01012 Dogs, M., Voget, S., Teshima, H., Petersen, J., Davenport, K., Dalingault, H., et al. (2013). Genome sequence of Phaeobacter inhibens type strain (T5(T)), a secondary metabolite producing representative of the marine Roseobacter clade, and emendation of the species description of Phaeobacter inhibens. Stand Genomic Sci, 9(2), 334-350. doi:10.4056/sigs.4448212 Frank, O., Pradella, S., Rohde, M., Scheuner, C., Klenk, H. P., Goker, M., & Petersen, J. (2014). Complete genome sequence of the Phaeobacter gallaeciensis type strain CIP 105210(T) (= DSM 26640(T) = BS107(T)). Stand Genomic Sci, 9(3), 914-932. doi:10.4056/sigs.5179110 Fuchs, G., Boll, M., & Heider, J. (2011). Microbial degradation of aromatic compounds - from one strategy to four. Nat Rev Microbiol, 9(11), 803-816. doi:10.1038/nrmicro2652 George, K. W., & Hay, A. G. (2011). Bacterial strategies for growth on aromatic compounds. In Advances in applied microbiology (Vol. 74, pp. 1-33): Elsevier. Gescher, J., Eisenreich, W., Wörth, J., Bacher, A., & Fuchs, G. (2005). Aerobic benzoyl‐CoA catabolic pathway in Azoarcus evansii: studies on the non‐ oxygenolytic ring cleavage enzyme. Molecular microbiology, 56(6), 1586- 1600. Gescher, J., Zaar, A., Mohamed, M., Schagger, H., & Fuchs, G. (2002). Genes coding for a new pathway of aerobic benzoate metabolism in Azoarcus evansii. Journal of Bacteriology, 184(22), 6301-6315. Gonzalez, J. M., Simo, R., Massana, R., Covert, J. S., Casamayor, E. O., Pedros-Alio, C., & Moran, M. A. (2000). Bacterial community structure associated with a dimethylsulfoniopropionate-producing North Atlantic algal bloom. Applied and Environmental Microbiology, 66(10), 4237-4246. doi:Doi 10.1128/Aem.66.10.4237-4246.2000 Grishin, A. M., & Cygler, M. (2015). Structural organization of enzymes of the phenylacetate catabolic hybrid pathway. Biology, 4(2), 424-442. doi:10.3390/biology4020424 Gulvik, C. A., & Buchan, A. (2013). Simultaneous catabolism of plant-derived aromatic compounds results in enhanced growth for members of the Roseobacter lineage. Applied and Environmental Microbiology, 79(12), 3716-3723.

Jimenez, N., Vinas, M., Guiu-Aragones, C., Bayona, J. M., Albaiges, J., & Solanas, A. M. (2011). Polyphasic approach for assessing changes in an autochthonous marine bacterial community in the presence of Prestige fuel oil and its biodegradation potential. Appl Microbiol Biotechnol, 91(3), 823-834. doi:10.1007/s00253-011-3321-4 Jung, J., Madsen, E., Jeon, C. O., & Park, W. (2011). Comparative genomic analysis of Acinetobacter oleivorans DR1: strain-specific genomic regions and gentisate biodegradation. Applied and Environmental Microbiology. Kimes, N. E., Callaghan, A. V., Aktas, D. F., Smith, W. L., Sunner, J., Golding, B., et al. (2013). Metagenomic analysis and metabolite profiling of deep- sea sediments from the Gulf of Mexico following the Deepwater Horizon oil spill. Front Microbiol, 4, 50. doi:10.3389/fmicb.2013.00050 Lamendella, R., Strutt, S., Borglin, S., Chakraborty, R., Tas, N., Mason, O. U., et al. (2014). Assessment of the Deepwater Horizon oil spill impact on Gulf coast microbial communities. Front Microbiol, 5, 130. doi:10.3389/fmicb.2014.00130 Lee, H. J., Kim, J. M., Lee, S. H., Park, M., Lee, K., Madsen, E. L., & Jeon, C. O. (2011). Gentisate 1,2-dioxygenase, in the third naphthalene catabolic gene cluster of Polaromonas naphthalenivorans CJ2, has a role in naphthalene degradation. Microbiology, 157(10), 2891-2903. doi:doi:10.1099/mic.0.049387-0 Luo, H., & Moran, M. A. (2014). Evolutionary ecology of the marine Roseobacter clade. Microbiol Mol Biol Rev, 78(4), 573-587. doi:10.1128/MMBR.00020- 14 Martin, F. J., & McInerney, J. O. (2009). Recurring cluster and operon assembly for Phenylacetate degradation genes. BMC Evolutionary Biology, 9(1), 36. doi:10.1186/1471-2148-9-36 Mendez, V., Agullo, L., Gonzalez, M., & Seeger, M. (2011). The homogentisate and homoprotocatechuate central pathways are involved in 3-and 4- hydroxyphenylacetate degradation by Burkholderia xenovorans LB400. PLoS One, 6(3), e17583. Mishamandani, S., Gutierrez, T., Berry, D., & Aitken, M. D. (2016). Response of the bacterial community associated with a cosmopolitan marine diatom to crude oil shows a preference for the biodegradation of aromatic hydrocarbons. Environ Microbiol, 18(6), 1817-1833. doi:10.1111/1462- 2920.12988 Moran, M. A., Belas, R., Schell, M. A., Gonzalez, J. M., Sun, F., Sun, S., et al. (2007). Ecological genomics of marine Roseobacters. Appl Environ Microbiol, 73(14), 4559-4569. doi:10.1128/AEM.02580-06 Moran, M. A., & Hodson, R. E. (1990). Contributions of degrading Spartina alterniflora lignocellulose to the dissolved organic carbon pool of a salt marsh. Marine ecology progress series. Oldendorf, 62(1), 161-168.

Newton, R. J., Griffin, L. E., Bowles, K. M., Meile, C., Gifford, S., Givens, C. E., et al. (2010). Genome characteristics of a generalist marine bacterial lineage. ISME J, 4(6), 784-798. doi:10.1038/ismej.2009.150 Patzelt, D., Michael, V., Päuker, O., Ebert, M., Tielen, P., Jahn, D., et al. (2016). Gene flow across genus barriers – conjugation of Dinoroseobacter shibae’s 191-kb killer plasmid into Phaeobacter inhibens and AHL- mediated expression of Type IV Secretion Systems. Frontiers in Microbiology, 7(742). doi:10.3389/fmicb.2016.00742 Petersen, J., Brinkmann, H., Bunk, B., Michael, V., Pauker, O., & Pradella, S. (2012). Think pink: photosynthesis, plasmids and the Roseobacter clade. Environ Microbiol, 14(10), 2661-2672. doi:10.1111/j.1462- 2920.2012.02806.x Pradella, S., Pauker, O., & Petersen, J. (2010). Genome organisation of the marine Roseobacter clade member Marinovum algicola. Archives of Microbiology, 192(2), 115-126. doi:10.1007/s00203-009-0535-2 Quigley, L. N. M. (2018). Characterizing the molecular mechanisms for the bacterial transformation of recalcitrant organic matter in coastal salt marshes. (Doctor of Philosophy), The University of Tennessee, Knoxville, Rocha, E. P., & Danchin, A. (2002). Base composition bias might result from competition for metabolic resources. TRENDS in Genetics, 18(6), 291- 294. Rocker, D., Brinkhoff, T., Gruner, N., Dogs, M., & Simon, M. (2012). Composition of humic acid-degrading estuarine and marine bacterial communities. FEMS Microbiol Ecol, 80(1), 45-63. doi:10.1111/j.1574-6941.2011.01269.x Rodrigo-Torres, L., Pujalte, M. J., & Arahal, D. R. (2016a). Draft genome sequence of Shimia marina CECT 7688(T). Mar Genomics, 28, 83-86. doi:10.1016/j.margen.2016.01.006 Rodrigo-Torres, L., Pujalte, M. J., & Arahal, D. R. (2016b). Draft genomes of Nautella italica strains CECT 7645(T) and CECT 7321: Two roseobacters with potential pathogenic and biotechnological traits. Mar Genomics, 26, 73-80. doi:10.1016/j.margen.2016.01.001 Sanni, G. O., Coulon, F., & McGenity, T. J. (2015). Dynamics and distribution of bacterial and archaeal communities in oil-contaminated temperate coastal mudflat mesocosms. Environ Sci Pollut Res Int, 22(20), 15230-15247. doi:10.1007/s11356-015-4313-1 Schuhle, K., Gescher, J., Feil, U., Paul, M., Jahn, M., Schagger, H., & Fuchs, G. (2003). Benzoate-coenzyme A ligase from Thauera aromatica: an enzyme acting in anaerobic and aerobic pathways. Journal of Bacteriology, 185(16), 4920-4929. Schühle, K., Gescher, J., Feil, U., Paul, M., Jahn, M., Schägger, H., & Fuchs, G. (2003). Benzoate-Coenzyme A Ligase from Thauera aromatica: an enzyme acting in anaerobic and aerobic pathways. Journal of Bacteriology, 185(16), 4920.

Seyedsayamdost, M. R., Case, R. J., Kolter, R., & Clardy, J. (2011). The Jekyll- and-Hyde chemistry of Phaeobacter gallaeciensis. Nature Chemistry, 3(4), 331-335. doi:10.1038/Nchem.1002 Shiba, T. (1991). Roseobacter litoralis gen. nov., sp. nov., and Roseobacter denitrificans sp. nov., aerobic pink-pigmented bacteria which contain bacteriochlorophyll a. Systematic and Applied Microbiology, 14(2), 140- 145. Simon, M., Scheuner, C., Meier-Kolthoff, J. P., Brinkhoff, T., Wagner-Dobler, I., Ulbrich, M., et al. (2017). Phylogenomics of Rhodobacteraceae reveals evolutionary adaptation to marine and non-marine habitats. ISME J, 11(6), 1483-1499. doi:10.1038/ismej.2016.198 Sipler, R. E., Kellogg, C. T. E., Connelly, T. L., Roberts, Q. N., Yager, P. L., & Bronk, D. A. (2017). Microbial community response to terrestrially derived dissolved organic matter in the coastal Arctic. Frontiers in Microbiology, 8. doi:ARTN 1018 10.3389/fmicb.2017.01018 Stoica, E., & Herndl, G. J. (2007). Bacterioplankton community composition in nearshore waters of the NW Black Sea during consecutive diatom and coccolithophorid blooms. Aquatic Sciences, 69(3), 413-418. doi:10.1007/s00027-007-0885-2 Tomás-Gallardo, L., Gómez-Álvarez, H., Santero, E., & Floriano, B. (2013). Combination of degradation pathways for naphthalene utilization in Rhodococcus sp. strain TFB. Microbial Biotechnology, 7(2), 100-113. doi:10.1111/1751-7915.12096 Valderrama, J. A., Durante-Rodriguez, G., Blazquez, B., Garcia, J. L., Carmona, M., & Diaz, E. (2012). Bacterial degradation of benzoate cross-regulation between aerobic and anaerobic pathways. Journal of Biological Chemistry, 287(13), 10494-10508. Viggor, S., Juhanson, J., Joesaar, M., Mitt, M., Truu, J., Vedler, E., & Heinaru, A. (2013). Dynamic changes in the structure of microbial communities in Baltic Sea coastal seawater microcosms modified by crude oil, shale oil or diesel fuel. Microbiol Res, 168(7), 415-427. doi:10.1016/j.micres.2013.02.006 Wagner-Dobler, I., Ballhausen, B., Berger, M., Brinkhoff, T., Buchholz, I., Bunk, B., et al. (2010). The complete genome sequence of the algal symbiont Dinoroseobacter shibae: a hitchhiker's guide to life in the sea. ISME J, 4(1), 61-77. doi:10.1038/ismej.2009.94 Wagner-Dobler, I., & Biebl, H. (2006). Environmental biology of the marine Roseobacter lineage. Annual Review of Microbiology, 60, 255-280. doi:10.1146/annurev.micro.60.080805.142115 Wang, R., & Seyedsayamdost, M. R. (2017). Roseochelin B, an algaecidal natural product synthesized by the Roseobacter Phaeobacter inhibens in response to algal sinapic acid. Organic Letters, 19(19), 5138-5141. doi:10.1021/acs.orglett.7b02424

Wang, Y., Yang, J., Lee, O. O., Dash, S., Lau, S. C. K., Al-Suwailem, A., et al. (2011). Hydrothermally generated aromatic compounds are consumed by bacteria colonizing in Atlantis II Deep of the Red Sea. The Isme Journal, 5, 1652. doi:10.1038/ismej.2011.42 https://www.nature.com/articles/ismej201142#supplementary-information Yang, T., Speare, K., McKay, L., MacGregor, B. J., Joye, S. B., & Teske, A. (2016). Distinct bacterial communities in surficial seafloor sediments following the 2010 Deepwater Horizon Blowout. Front Microbiol, 7, 1384. doi:10.3389/fmicb.2016.01384 Zaar, A., Eisenreich, W., Bacher, A., & Fuchs, G. (2001). A novel pathway of aerobic benzoate catabolism in the bacteria Azoarcus evansii and Bacillus stearothermophilus. Journal of Biological Chemistry, 276(27), 24997- 25004. Zech, H., Hensler, M., Kossmehl, S., Druppel, K., Wohlbrand, L., Trautwein, K., et al. (2013). Dynamics of amino acid utilization in Phaeobacter inhibens DSM 17395. Proteomics, 13(18-19), 2869-2885. doi:10.1002/pmic.201200560 Zhou, H. W., Wong, A. H., Yu, R. M., Park, Y. D., Wong, Y. S., & Tam, N. F. (2009). Polycyclic aromatic hydrocarbon-induced structural shift of bacterial communities in mangrove sediment. Microb Ecol, 58(1), 153-160. doi:10.1007/s00248-008-9456-x

IX. Appendix Table 2.1. Roseobacter prevalence in aromatic hydrocarbon-enriched environments. Sample Source Treatment Approach Roseobacter Reference prevalence Mangrove sediment Sediment supplemented with DGGE Not determined (Zhou, 2009) (Sai Keng mangrove phenanthrene, pyrene, and swamp in Hong Kong) benzo(a)pyrene Sediment/seawater in None DGGE Not determined (Mahjoubi, 2013) Tunisia near STIR, an oil refinery company Gulf of Mexico None Metagenomic Not determined (Kimes, 2013) (Deepwater Horizon sequencing oil spill) sediment Elmers’s Beach, None Targeted 16S rRNA Not determined (Lamendella, 2014) Grand Isle, LA sand gene sequencing and isolation Deepwater Horizon None Clone library ≤20% (Yang, 2016) wellhead area sediment Messina harbor, Italy Seawater supplemented with Clone library ≤5% (Cappello, 2012) surface seawater Arabian Light Crude Oil Ría de Ortigueira (A Artificial seawater medium DGGE Not determined (Jimenez, 2011) Coruña, Spain) fuel oil supplemented with fuel oil Surface seawater Seawater amended with oil, DGGE Not determined (Viggor, 2013) sample from the fuel, or hexadecane coastal area of Vilsandi Island Sediment and water None Clone library Not determined (Todorova, 2014) samples from Sozopol coastal region Netherlands Exclusive ONR7a medium amended 16S rRNA gene Not determined (Chronopoulou, 2015) Economic Zone water with seawater samples and amplification aromatic hydrocarbons Sediment and Sediment and seawater 454 Pyrosequencing ≤2.74% (Sanni, 2015) seawater from amended with weathered Pyefleet Channel North Sea crude oil Not applicable Non-axenic algal medium Barcoded 16S rRNA ≤~1% (Mishamandani, 2016) inoculated with the diatom gene Skeletonema costatum and pyrosequencing supplemented with Alaska North Slope (ANS) crude oil Sediment from Bohai Minimal salt medium Miseq sequencing ≤~12.5% (Wang, 2016) sea amended with sediment and n-alkanes

Table 2.2. Summary of criteria used to determine the presence or absence of ring-cleaving pathways in Roseobacter genomes. Pathway (diagnostic gene) Colocalization criteria Protocatechuate (pcaH) pcaHG subunit and adjacent pcaC gene pcaB and pcaD in same operon Gentisate (nagI) nagILK or nagIL,and/or nagIK subunits Homogentisate (hmgA) hmgA, hmgB, hmgC genes are found in one operon or hmgAB subunit and presence of hmgC gene Benzoate oxidation (boxA) boxAB subunit and boxC, boxL, and boxR genes are in one operon. Phenylacetate (paaA) paaABCDE subunit Presence of paaZ gene

Table 2.3. Prevalence of ring-cleaving pathways in Roseobacter genomes. Ring-cleavage pathway Roseobacter prevalence Protocatechuate Oxygenolytic 63.67% Gentisate Oxygenolytic 17.68% Homogentisate Oxygenolytic 73.95% Benzoate Non-oxygenolytic 16.61% Phenylacetate Non-oxygenolytic 55.95%

Table 2.4. Nucleotide and amino acid sequence identities of homogentisate pathway genes in Oceanicola sp. HL-35. Oceanicola sp. HL-35 (nucleotide) Oceanicola sp. HL-35 (amino acid) hmgA (K224DRAFT_1299)a (K224DRAFT_1299)a K224DRAFT_2591 No significant similarity found. 62% (E value 0.0) hmgB (K224DRAFT_1301)a (K224DRAFT_1301)a K224DRAFT_2592 No significant similarity found. 53% (E value E-151) hmgC (K224DRAFT_1300)a (K224DRAFT_1300)a K224DRAFT_2595 No significant similarity found. 36% (E value E-42) aReference sequences against which nucleotide or amino acid sequences were compared in BLASTn or BLASTp, respectively.

Table 2.5. Nucleotide sequence identities of gentisate pathway genes in C. indicus P73 or M. alba DSM 26384. Celeribacter indicus P73 Mameliella alba DSM 26384 nagI (P73_1454)a nagI (LX94DRAFT_03764)a P73_4775 79% (E value 0.0) LX94DRAFT_05172 100% (E value 0.0) P73_0175 100% (E value 0.0) LX94DRAFT_03761 100% (E value 0.0) nagL (P73_1455)a nagL (LX94DRAFT_03765)a P73_4774 No significant similarity found. LX94DRAFT_05171 100% (E value 0.0) P73_0176 99% (E value E-165) LX94DRAFT_03760 100% (E value 0.0) nagK (P73_1456)a nagK (LX94DRAFT_03766)a P73_4773 No significant similarity found. LX94DRAFT_05170 100% (E value 0.0) P73_0177 99% (E value 0) LX94DRAFT_03759 99% (E value E-170) aReference sequences against which nucleotide sequences were compared in BLASTn.

Table 2.6. Amino acid sequence identities of gentisate pathway genes in C. indicus P73 or M. alba DSM 26384. Celeribacter indicus P73 Mameliella alba DSM 26384 nagI (P73_1454)a (LX94DRAFT_03764)a P73_4775 85% (E value 0.0) LX94DRAFT_05172 100% (E value 0.0) P73_0175 100% (E value 0.0) LX94DRAFT_03761 100% (E value 0.0) nagL (P73_1455)a (LX94DRAFT_03765)a P73_4774 78% (E value E-128) LX94DRAFT_05171 100% (E value E-164) P73_0176 99% (E value E-165) LX94DRAFT_03760 100% (E value E-164) nagK (P73_1456)a (LX94DRAFT_03766)a P73_4773 81% (E value E-146) LX94DRAFT_05170 100% (E value E-161) P73_0177 99% (E value E-175) LX94DRAFT_03759 97% (E value E-170) aReference sequences against which amino acid sequences were compared in BLASTp.

Figure 2.1. Evolutionary history of aromatic hydrocarbon degradation pathways in Roseobacter genomes. Pink = pca, orange = gdo, green = box, light blue = hgd, and dark blue = paa. This maximum likelihood tree is based on an alignment of 89 peptides common to all genomes. The numbers at nodes represent bootstrap values. Agrobacterium fabrum C58 was the outgroup. (Further phylogenetic tree method generation details provided by co-author José Manuel González will be included in the published version of this manuscript.)

Figure 2.2. Evolutionary history of aromatic hydrocarbon degradation pathways among Nautella, Phaeobacter, or Leisingera spp. Pink = pca, orange = gdo, light blue = hgd, and dark blue = paa. Phylogenetic tree is a subclade of Figure 2.1 showing conservation of the pca, hgd, and paa pathways among Nautella, Phaeobacter, or Leisingera spp.

CHAPTER III: FUNCTIONAL REDUNDANCY IN THE HYDROXYCINNAMATE CATABOLIC PATHWAYS OF THE SALT MARSH BACTERIUM SAGITTULA STELLATA E-37

A version of this chapter was originally published by Ashley M. Frank, Michelle J. Chua, Christopher A. Gulvik, and Alison Buchan:

Ashley M. Frank, Michelle J. Chua, Christopher A. Gulvik, Alison Buchan. “Functional Redundancy in the Hydroxycinnamate Catabolic Pathways of the Salt Marsh Bacterium Sagittula stellata E-37.” Applied and Environmental Microbiology (accepted).

Author contribution statement: Drs. Ashley Frank, Christopher Gulvik and Alison Buchan developed initial project objectives and experimental design. Ashley Frank generated fcs mutants; Michelle Chua constructed the dctP mutant. Ashley Frank and Michelle Chua performed mutant growth characterizations. Michelle Chua performed qRT-PCR assays, with statistical support from Christopher Gulvik. Ashley Frank led fcs-data analysis. Ashley Frank, Michelle Chua and Alison Buchan wrote the manuscript.

I. Abstract

The hydroxycinnamates (HCAs) ferulate and p-coumarate are among the most abundant constituents of lignin and their degradation by bacteria is an essential step in the remineralization of vascular plant material. Here, we investigate the catabolism of these two HCAs by the marine bacterium, Sagittula stellata E-37, a member of the roseobacter lineage with lignolytic potential. Bacterial degradation of HCAs is often initiated by the activity of a hydroxycinnamoyl-CoA synthase. Genome analysis of S. stellata revealed the presence of two feruloyl-CoA (fcs) synthase homologs, an unusual occurrence amongst characterized HCA degraders. In order to elucidate the role of these homologs in HCA catabolism, fcs1 and fcs2 were disrupted using insertional mutagenesis, yielding both single and double fcs mutants. Growth on p-coumarate was abolished in the fcs double mutant, whereas maximum cell yield on ferulate was only 2% of wildtype. Interestingly, the single mutants demonstrated opposing phenotypes: the fcs1 45 mutant showed impaired growth (extended lag and ~60% of wildtype rate) on p- coumarate and the fcs2 mutant showed impaired growth (extended lag and ~20% of wildtype rate) on ferulate, pointing to distinct, but overlapping roles of the encoded fcs homologs with fcs1 primarily dedicated to p-coumarate utilization and fcs2 playing a dominant role in ferulate utilization. Finally, a TRAP family transporter was found to be required for growth on both HCAs. These findings provide evidence for functional redundancy in the degradation of HCAs in S. stellata E-37 and offer important insight into the genetic complexity of aromatic compound degradation in bacteria.

II. Importance

Hydroxycinnamates (HCAs) are essential components of lignin and are involved in various plant functions, including defense. In nature, microbial degradation of HCAs is influential to global carbon cycling. HCA degradation pathways are also of industrial relevance as microbial transformation of the HCA, ferulate, can generate vanillin, a valuable flavoring compound. Yet, surprisingly little is known of the genetics underlying bacterial HCA degradation. Here, we make comparisons to previously characterized bacterial HCA-degraders and use a genetic approach to characterize genes involved in catabolism and uptake of HCAs in the environmentally relevant marine bacterium, Sagitulla stellata. We provide evidence of overlapping substrate specificity between HCA degradation pathways and uptake proteins. We conclude S. stellata is uniquely poised to utilize HCAs found in the complex mixtures of plant-derived compounds in nature. This strategy may be common amongst marine bacteria residing in lignin- rich coastal waters and has potential relevance to biotechnology sectors.

I. Introduction

Constituting up to 30% of the cell wall of vascular plants, lignin represents the most abundant aromatic polymer on Earth and offers potential as a highly valuable source of renewable carbon for numerous applications, including hydrocarbon fuel (Laskar et al., 2013) and synthetic chemical production (Ragauskas et al., 2014). Lignin is generated from the radical polymerization of three hydroxycinnamyl alcohols (p-coumaryl [H], coniferyl [G], and sinapyl alcohol [S]), resulting in a structurally heterogeneous polymer harboring a variety of aromatic substituents linked through stable carbon-carbon and ether bonds (Ralph et al., 1998). Despite this chemical complexity, many fungi and a limited number of bacteria have been demonstrated to degrade lignin in nature (Bugg, Ahmad, Hardiman, & Rahmanpour, 2011). Primary depolymerization of lignin is typically accomplished by fungi through non-specific oxidation driven by extracellular oxidoreductases (Kirk & Farrell, 1987; Kuwahara et al., 1984; Tien et al., 1986). This initial enzymatic step liberates a pool of lower molecular weight aromatics that can serve as carbon and energy sources for microbes with appropriate ring fission pathways (Diaz et al., 2013; Fuchs, 2008; Harwood & Parales, 1996).

Among the most abundant and valuable aromatic compounds associated with lignin are the hydroxycinnamates (HCAs) p-coumarate (H unit derivative) and ferulate (G unit derivative) (Dixon et al., 2002; Hartley & Ford, 1989). Both HCAs are essential to plant integrity and offer additional potential for industrial and pharmaceutical applications (Kroon & Williamson, 1999). Ferulate and p- coumarate are common in crop plants such as fruits, vegetables, and coffee (Naczk & Shahidi, 2006; Torres-Mancera et al., 2011), but are most notably abundant in grasses (Sun et al., 2001). While p-coumarate is mostly esterified to other phenylpropanoid units of lignin (Ralph et al., 1994), ferulate crosslinks lignin to cell wall polysaccharides through ether and ester bonds (Ralph et al.,

1998). These ferulate crosslinks have been identified as a primary source of recalcitrance in grass lignocellulose (de Oliveira et al., 2015). Aside from providing a refractory architecture, HCAs belong to a family of phenylpropanoid defense molecules released in response to microbial invasion (Dixon et al., 2002; Macoy et al., 2015). Not surprisingly, microbes with the ability to degrade HCAs are able to deflect these phenylpropanoid-derived plant defenses, leading to enhanced virulence in plants (Lowe et al., 2015). A more detailed understanding of the genes involved in HCA degradation across different organisms may therefore inform strategies to mitigate plant disease. In addition to supporting plant health, HCAs also harbor value from human health and industrial perspectives as they can serve as dietary antioxidants (Fresco et al., 2006; Razzaghi-Asl, et al., 2013; Teixeira et al., 2013) and as precursors for a variety of commercial products. For instance, microbial strains have been engineered to convert p-coumarate into precursors for thermoplastics, flavorings, and cosmetics (Vargas-Tah & Gosset, 2015). Similarly, the bacterial degradation of ferulate, which often generates vanillin as an intermediate (Kumar & Pruthi, 2014), has been exploited for the biotechnological production of commercial vanillin flavoring (Muheim & Lerch, 1999; Plaggenborg et al., 2006; Priefert et al., 2001). Insight into the genes and pathways that modify these HCAs can thus guide technologies to generate commercially-relevant products.

To date, there have been two major pathways described for the aerobic bacterial degradation of HCAs, both of which are initiated by CoA activation of the HCA carboxylate side chain to generate a hydroxycinnamoyl-CoA (Figure 3.1). In both instances, the CoA addition is performed by a hydroxycinnamoyl-CoA synthase or ligase. The difference in the two described HCA pathways lies in the mechanism employed for removal of the acetate moiety, with one pathway proceeding through a β-oxidative route (Campillo et al., 2014; Otani et al., 2014) and the other through a non-β-oxidative route that generates an aldehyde intermediate (Masai et al., 2002; Parke & Ornston, 2003). While ferulate can be 49 degraded through either the β-oxidative (Campillo et al., 2014) or aldehyde pathway (Masai et al., 2002; Overhage et al., 1999; Parke & Ornston, 2003), ), p- coumarate is expected to be catabolized solely through the aldehyde pathway (Lowe et al., 2015; Parke & Ornston, 2003). The aldehyde pathways for ferulate and p-coumarate are homologous, differing only in the meta ring substituent

(ferulate = meta-OCH3, p-coumarate = meta-H) of the starting material and intermediates. Thus, after deacetylation in the aldehyde branch, the ferulate pathway generates vanillic aldehyde and then vanillate, whereas the p- coumarate pathway generates p-hydroxybenzaldehyde and then p- hydroxybenzoate. All known HCA pathways converge to form the central intermediate protocatechuate which undergoes ring cleavage, with products ultimately feeding into the TCA cycle (Harwood & Parales, 1996). In the few strains in which the genetics of HCA catabolism has been elucidated, the genes are typically co-localized and co-regulated (Lowe et al., 2015; Mattes et al., 2008). For example, in Agrobacterium fabrum, all ferulate degradation genes are localized to a single catabolon (Campillo et al., 2014).

For most bacteria with described HCA degradation pathways, the relevant CoA synthase is either reported specifically for ferulate, or more generically, for hydroxycinnamates as many characterized CoA synthases can act upon both ferulate and p-coumarate, as well as a variety of structurally related compounds. The hydroxycinnamoyl-CoA ligase from Acinetobacter baylyi ADP1 (HcaC) can transfer CoA to p-coumarate, ferulate, and caffeate (Parke & Ornston, 2003). Similarly, the annotated feruloyl-CoA ligase (FerA) from Sphingobium sp. SYK-6 has been demonstrated to convert p-coumarate, ferulate, caffeate, and sinapate (Masai et al., 2002). Although direct evidence is not yet available, the CoA ligase (Atu1416) from A. fabrum is also likely to accept both ferulate and p-coumarate based on transcriptomics data (Baude et al., 2016) and growth profiles from deletion constructs (Campillo et al., 2014).

The marine bacterium Sagitulla stellata E-37 has been developed as a model for the study of HCA degradation and the catabolism of other plant-derived aromatic compounds in coastal ocean environments (Buchan et al., 2005; Gonzalez et al., 1997; Gulvik & Buchan, 2013), where the dissolved organic carbon pool is highly aromatic in nature (Moran & Hodson, 1994). This isolate has been shown to selectively attach to and partially mineralize a synthetic form of lignin and demonstrates robust growth on a variety of aromatic monomers, including ferulate and p-coumarate (Buchan et al., 2000; Gonzalez et al., 1997; Gulvik & Buchan, 2013). However, the pathways used by this strain to degrade these HCAs are unknown. As described herein, the genome of S. stellata harbors two ORFs with significant homology to hydroxycinnamoyl-CoA synthases (both annotated as feruloyl-CoA synthases [Fcs]). Since Fcs proteins have been shown to have broad substrate specificities across para-substituted HCAs and are encoded by a single gene in other bacteria (Masai et al., 2002; Mitra et al., 1999), the presence of two disparately localized fcs homologs in S. stellata is an intriguing phenomenon, and suggests that dedicated pathways for p-coumarate and ferulate exist in this strain. This genetic arrangement in S. stellata presents insight into the complexity of HCA catabolism in lignin-rich coastal marine environments.

II. Materials and Methods

Strains and growth conditions Sagittula stellata E-37 was previously isolated from pulp mill effluent and enriched on Indulin AT, a commercially available Kraft lignin generated from pulp mills, as described by Gonzalez et al. (Gonzalez et al., 1997). Unless otherwise noted, wildtype S. stellata and any mutant derivatives were routinely maintained at 30°C in YTSS medium (per liter, 2.5 g yeast extract, 4 g tryptone, 15 g Sea Salts [Sigma-Aldrich, St. Louis, MO]). E. coli strains used for cloning and

51 conjugation experiments were maintained at 37°C in Lysogeny broth (per liter, 10 g tryptone, 5 g yeast extract, 10 g NaCl). Antibiotics were added to the E. coli growth medium to maintain selective pressure at 50 μg/mL kanamycin or 10 μg/mL tetracycline. S. stellata growth assays were routinely performed in Marine

Basal Medium (MBM) containing 1.5% (w/v) Sigma Sea Salts, 225 nM K2HPO4,

13.35 μM NH4Cl, 71 mM Tris-HCl (pH 7.5), 68 μM Fe-EDTA, trace metals and vitamins (38), and carbon (2 mM aromatic and 10 mM acetate, unless otherwise noted). Strains were incubated at 30°C, in the dark, with shaking. All glassware was combusted in an ashing furnace (Barnstead Thermolyne, Type F62700) for 4-24 hours prior to use to eliminate trace carbon. Seeding densities were ~1 x 106 CFU/mL and no-carbon added controls were run in parallel for comparison. Growth was monitored using a spectrophotometer at 540 nm. All strains used in this study are listed in Table 3.1.

Generation of S. stellata mutants A S. stellata fcs1 (SSE37_12324) mutant strain was produced using marker exchange mutagenesis following modifications of previously described procedures (Jones & Williams, 2003; Lidbury et al., 2014; Stibitz, 1994). Briefly, a 402 bp region upstream and a 585 bp region downstream of fcs1 was amplified to generate “A” and “B’ amplicons, respectively. During PCR amplification, a 21 bp scar sequence was added through the internal primers generating complementary appendages for overlap extension PCR. PCR mixtures for each A and B reaction included 1x GoTaq Buffer (Promega, Madison, WI), 200 μM deoxynucleotide triphosphates (Promega), 6 μM internal primer, 0.6 μM external primer, 0.025 U/μl GoTaq DNA Polymerase (Promega) and 1-2 ng/μl S. stellata genomic DNA. Thermocycling conditions consisted of a 3 min denaturation at 95°C followed by 31 cycles of 30 sec at 95°C, 30 sec at 56°C, 30 sec at 72°C and a final extension for 5 min at 72°C. The A and B amplicons were subsequently used as template DNA during overlap extension PCR to generate a fused “AB” product. This PCR was performed with 1x FailSafe Buffer F

(Epicentre, Madison, WI), 6 μM external primer A, 6 μM external primer B, 0.05 U/μl FailSafe enzyme mix (Epicentre), 1-2 ng/μl amplicon A, and 1-2 ng/μl amplicon B. Thermocycling conditions were the same as previous, except with a 10 min final extension. The resulting product was cloned into pCR™2.1-TOPO® (Invitrogen, Carlsbad, CA). Positive clones were selected on LB/kanamycin (50 μg/mL) and verified by sequence analysis. The AB insert was then subcloned into the mobilizable suicide plasmid, pARO180 (ATCC 77123), through shared HindIII and XbaI sites, generating plasmid pARO180_12324 and transformed into chemically competent E. coli JM109. Positive clones were selected on LB/ampicillin (50 μg/mL) agar and verified through amplification and sequencing. The ::sacBkanR cassette from pRMJ1 (Jones & Williams, 2003) was inserted into the middle of the AB insert at the SalI site, generating plasmid pARO180_12324_SK. The complete A-sacBkanR-B vector construct was then transformed into the mating strain, E. coli S17-1. The suicide vector was delivered to S. stellata through biparental mating with E. coli S17-1 (pARO180_12324_SK) on a YTSS agar plate. S. stellata transconjugants were enriched from the mating mixture by requiring growth on 5 mM p- hydroxybenzoate (a restrictive substrate for E. coli) and 50 μg/mL kanamycin. Allelic replacement was confirmed by PCR amplification across the chromosomal region of interest using primers 12324_AsacF/12324_BkmR (Table 3.2) and sequencing of the PCR product. PCR for this amplification was performed using the LongRange PCR kit (Qiagen, Valencia, CA), according to manufacturer instructions with the following thermocycling conditions: initial denaturation at 93°C for 3 min, followed by 30 cycles of 93°C denaturation for 15 sec, 57°C annealing for 30 sec, 68°C elongation for 12 min with no final extension.

The fcs2 (SSE37_24399) mutant was generated using insertional mutagenesis with the pKNOCK-Km plasmid (Alexeyev, 1999). A 231 bp internal region of SSE37_24399 (positions 983-1,213 bp) was cloned into pCR™2.1-TOPO®, subcloned into pKNOCK-Km to generate plasmid pKNOCK-Km_24399, and then 53 introduced into the E. coli mating strain BW20767 (Metcalf et al., 1996). Conjugation of BW20767 (pKNOCK-Km_24399) with wildtype S. stellata was performed as described above. Single-crossover homologous recombination resulted in integration of the entire plasmid into the fcs2 gene. The fcs1/fcs2 double mutant was generated by cloning the same 231 bp fragment of fcs2 into pKNOCK-Tc (Alexeyev, 1999) (making pKNOCK-Tc_24399), and bi-parental mating of the E. coli BW20767 harboring this plasmid with the fcs1 mutant strain of S. stellata. Different plasmids were used for chromosomal integration of fcs1 (pARO180) and fcs2 (pKNOCK) to eliminate the chance of crossover between homologous sections of the plasmids during generation of the double knockout.

A dctP (SSE37_24379) mutant was generated using insertional mutagenesis with the pKNOCK-Km vector harboring a 292 bp internal fragment (positions 423 to 714 bp) of dctP and following procedures outlined above. To confirm the role of SSE37_24379, complementation of the dctP with the broad-host-range plasmid pRK415 (Keen et al., 1988) harboring the wildtype dctP gene was attempted. Phenotypes on different substrates was determined by growth assays on HCAs and their intermediates.

All chromosomal disruptions were confirmed by PCR amplification and sequencing of junction sites. All plasmids and primers used in generation of mutants are listed in Tables 3.1 and 3.2, respectively.

Biolog™ phenotypic microarrays of fcs mutants Custom phenotypic microarrays were performed to assay for redox activity of strains when grown on defined carbon sources (10 mM acetate and 2 mM of each aromatic monomer [p-hydroxybenzoate, ferulate, p-coumarate]). All fcs mutants and wildtype S. stellata were pre-cultured in MBM with 10 mM acetate and supplemented with kanamycin and tetracycline, as necessary. Assays were performed in 96 well flat-bottom plates (Costar) containing 1X MBM, 1X Biolog™

Dye Mix G (Hayward, Ca), 107 cells/mL S. stellata, and a single carbon source. Plates were incubated at 30°C in a Omnilog reader (Biolog™), where digital colorimetric readings were captured every 30 minutes to monitor the intensity of dye reduction over time, recorded in Omnilog units/time.

RNA isolation and reverse transcription For total RNA extraction, strains were grown on 7 mM acetate or 2 mM HCA. Total RNA was extracted using the RNeasy Mini Kit (Qiagen) with slight modification; cells were lysed by vortexing at maximum speed for 10 min with 0.2 g of low-binding 200 µm zirconium beads (OPS Diagnostics, L.L.C., Lebanon, NJ). Genomic DNA was removed using the TURBO DNA-free™ Kit (Ambion, Austin, TX) per manufacturer’s instructions. cDNA synthesis was performed with M-MLV Reverse Transcriptase (Invitrogen) using 500 ng total RNA, as described in the product manual.

Gene expression analyses RT-qPCR was performed using 1X SYBR Premix Ex Taq (Perfect Real Time) (TaKaRa Bio Inc., Otsu, Japan) in 25 µL volumes consisting of 12.5 µL 1X SYBR Premix, 10 µL cDNA, and 1.25 µL forward and 1.25 µL reverse primers (10 µM). Genes of interest were normalized to three previously validated housekeeping genes rpoC, map, and alaS (Gulvik & Buchan, 2013). qPCR amplification included a 95°C denaturation for 3 min, followed by 40 cycles of 95°C for 20 sec, 57°C for 20 sec, and 72°C extension for 15 sec, and a final extension at 72°C for 5 min. Melt curves were obtained between 50-100°C at 1°C per second with readings every 1°C. cDNA was diluted with 10 mM Tris (pH 8.0), and end point

PCR was performed prior to RT-qPCR to ensure Ct values for each gene and replicate were between 15 and 30 cycles. Efficient genomic DNA removal (Ct values >5 cycles) was confirmed by including no-RT controls. Technical and biological replicates were included in triplicates for each gene. RT-qPCR

55 analyses were performed using qBASE (Hellemans et al., 2007) to normalize and relativize gene transcripts.

Analysis of substrate concentrations p-coumarate and ferulate concentrations were monitored in cell-free supernatants from S. stellata cultures spectrophotometrically. Wavelengths at which each substrate absorbed maximally in minimal medium were empirically determined by performing wavelength scans from 200-700 nm with 0.002% sample (1 mL total volume, diluted with sterile Milli-Q water). Molar extinction coefficients for each substrate in the minimal medium were calculated to be 0.21 L mol-1 cm-1 for ferulate and 0.40 L mol-1 cm-1 for p-coumarate. To quantify extracellular concentrations of substrates, culture aliquots were passaged through a 0.22 µm filter and absorbances at 283 nm (p-coumarate) and 308 nm (ferulate) were monitored with a Beckman DU 800 UV/Vis spectrophotometer. Concentrations were calculated from 10-point standard curves (0.3 – 3.0 mM) with r2 values of 0.97 for ferulate and 0.99 for p-coumarate (Figure 3.2). Full (200-700 nm) wavelength scans were performed for all cultures. The shape of the absorbance curves provides additional evidence of the specificity of the approach (Figure 3.3).

Genome analyses Protein sequences from organisms with experimental validation of function were used in homology searches against the S. stellata genome. These searches were performed using the BLASTp algorithm at NCBI.

III. Results

S. stellata possesses two disparately located hydroxycinnamoyl-CoA synthase homologs The S. stellata genome was queried for the presence of hydroxycinnamoyl-CoA synthase homologs through alignments to functionally validated protein sequences from Acinetobacter baylyi ADP1 (AAL54850), Agrobacterium fabrum C58 (NP_354423), and Sphingobium sp. SYK-6 (BAK67177). BLASTp alignments of the selected proteins against the S. stellata genome identified two ORF products, hereafter referred to as Fcs1 (WP_005859342) and Fcs2 (WP_005857142), which showed highest sequence similarity to A. baylyi and A. fabrum Fcs proteins. Fcs1 showed moderately higher sequence identity and expectation (E) values to both of these previously characterized Fcs proteins (identity 55 and 40%; E-value 0 and E-147, A. fabrum and A. baylyi, respectively) relative to Fcs2 (identity 40 and 36%; E-value E-140 and E-124, A. fabrum and A. baylyi, respectively). No significant homology was evident between the validated Fcs from Sphingobium and either of the S. stellata Fcs sequences. S. stellata Fcs1 and Fcs2 sequences share 41% identity and 57% similarity to one another and possess conserved sequence motifs diagnostic of hydroxycinnamoyl synthases (i.e., acyl-activating enzyme [AAE] consensus, putative AMP binding site, putative active site, and putative CoA binding site motifs).

S. stellata ORFs predicted to be involved in downstream conversions for the β- oxidative branch for HCA degradation were also identified (Figure 3.4, Table 3.3). BLASTp alignment of the β-oxidative acyl-CoA dehydrogenase (Atu1415) from A. fabrum (Campillo et al., 2014) against the S. stellata genome yielded homology to SSE37_24394 (identity 51%; E-value E-79), annotated as a 3-hydroxy-2- methylbutyryl-CoA dehydrogenase, which lies just upstream of fcs2. Interestingly, another acyl-CoA dehydrogenase (Atu1414) exists within the A. fabrum HCA catabolon, but was shown not to be essential to ferulate degradation in that 57 system (Baude et al., 2016). This protein demonstrates homology to the S. stellata acyl-CoA dehydrogenase SE37_12329 (identity 55%; E-value E-138) located next to fcs1. Lastly, the A. fabrum HMPKP β-ketothiolase, Atu1421, strongly aligns to S. stellata SSE37_12319 (identity 82%; E-value E-180), annotated as a 4-hydroxyphenyl-beta-ketoacyl-CoA hydrolase, located adjacent to fcs1.

The two fcs genes and several of the genes potentially involved in downstream conversions reside in two distinct genetic loci on the S. stellata chromosome (Figure 3.4). S. stellata harbors ORFs with high sequence similarity to many of the gene products within the A. fabrum catabolon (Table 3.3), nearly all of which are within 5 kb of fcs1. The single exception is an acyl-CoA dehydrogenase, which is adjacent to fcs2. Interestingly, fcs2 is immediately upstream of an operon that encodes for the recently characterized benzoyl-CoA oxidation (box) pathway, through which benzoate is degraded by this strain (Gulvik & Buchan, 2013). Additionally, fcs2 is just downstream from a three-gene cassette anticipated to encode a tripartite ATP-independent periplasmic transporter family (TRAP) which consists of a substrate-binding protein (DctP; SSE37_24379) and two membrane-bound proteins (DctQM; SSE37_24384, SSE37_24389) (Mulligan et al., 2011) (Figure 3.4). The proximity of the TRAP gene cassette to these catabolic genes suggests the encoded transporter may facilitate uptake of aromatic compounds.

Though S. stellata can use vanillin as a sole carbon source (Figure 3.5), genomic evidence for the aldehyde pathway, that generates vanillin as an intermediate, is not strong in S. stellata. Homology searches to characterized vanillin dehydrogenases (i.e., LigV [BAK65381] from Sphingobium sp. SKY-6 and HcaB [CAG68567] from A. baylyi) result in moderate-to-weak similarities (max identities: LigV = 37%; E-value ≤ E-76, HcaB = 34%; E-value ≤ E-82) to a number

58 of putative aldehyde dehydrogenases encoded in the S. stellata genome, none of which are adjacent to either fcs locus.

Genetic analysis reveals overlapping but distinct roles of fcs genes in HCA degradation To assess the contributions of the putative fcs homologs to hydroxycinnamate catabolism in S. stellata, mutant strains lacking either or both functional copies of fcs1 and fcs2 were generated and the phenotypes on a suite of substrates was determined. The three mutants (fcs1, fcs2, fcs1/fcs2) displayed wildtype growth on the control non-aromatic and aromatic carbon sources, acetate and p- hydroxybenzoate (POB), respectively (Figure 3.6A-B). In S. stellata, POB is converted to the central intermediate protocatechuate (PCA), which then undergoes ring fission (Gulvik & Buchan, 2013). Hence, wildtype growth on POB is a strong indicator that the ring cleavage pathway through which all other aromatic acids tested here are funneled (Buchan et al., 2005; Gonzalez et al., 1997; Gulvik & Buchan, 2013) is unaffected by the mutations.

Growth curve analyses on ferulate revealed that while the fcs1 mutant demonstrated wildtype growth, the fcs2 mutant showed an extended lag phase (Figure 3.7A), and significantly depressed growth rate (22% of wildtype; p ≤ 0.01, Figure 3.8). Yet, the culture ultimately reached wildtype cell densities in stationary phase (Figure 3.7A). The fcs1/fcs2 double mutant elicited a severely depressed growth phenotype on this substrate with a maximum absorbance reaching only ~30% of wildtype cultures. As ferulate imparts a maroon coloration to the medium that may inflate absorbance readings, viable counts were also performed as an additional measure of cell growth. Maximum viable cell abundances of the double mutant grown on ferulate were only 2% of wildtype cultures (7.27 x 108 [± 7.64 x 107] CFU/mL vs. 1.81 x 107 [± 1.56 x 106] CFU/mL). Consistent with this finding, only 30% of the provided ferulate was taken up by

59 the double mutant. In contrast to the double mutant, ferulate concentrations were drawn below the limit of detection for all other strains (Figure 3.7C).

The individual fcs mutants showed opposing phenotypes on p-coumarate. The fcs2 mutant displayed wildtype growth while the fcs1 mutant strain showed a lag phase double that of wildtype cultures (Figure 3.7B) and a significantly decreased exponential growth rate (61% of wildtype; p ≤ 0.01, Figure 3.8). Both single mutants ultimately reached wildtype cell densities on this substrate, whereas the double mutant failed to grow (Figure 3.7B). The abolished growth of the double mutant on p-coumarate was corroborated by the lack of p-coumarate depletion from the medium (Figure 3.7D). For all strains and substrates, the extracellular concentrations of substrates were consistent with the growth assays (Figure 3.7).

Given that many aromatic acids are toxic to bacterial cells (Sikkema, et al., 1994, 1995), a parallel set of growth experiments were performed using a Biolog™ array, which included a tetrazolium dye that is responsive to cellular respiration- associated redox changes (Borglin et al., 2012). Here, cells were incubated in 96-well plates supplemented with ferulate, p-coumarate, or control compounds (as previous) in the presence of a proprietary tetrazolium-based dye, and monitored for colorimetric changes indicative of dye reduction upon cellular respiration. The Biolog™ assays were conducted to cross-validate the OD-based growth studies, as measurement of respiration provides a supplemental analysis of cell growth. The general trends from these assays are consistent with absorbance-based measures of growth, with the exception of the ferulate Biolog™ measurements which may be due to interference of the maroon color imparted by ferulate (Figure 3.6). Regardless, the viable count and HCA depletion assays agree that the double mutant cannot effectively oxidize ferulate, while the fcs2 mutant is moderately delayed in its processing.

Gene expression assays indicate fcs genes are growth substrate inducible To complement the mutational analysis, RT-qPCR was performed to assess fcs gene expression when S. stellata was grown on ferulate or p-coumarate. Both fcs1 and fcs2 genes were significantly upregulated during growth on both HCAs relative to growth on the non-aromatic substrate acetate (p <0.005). However, relative gene expression of fcs1 far exceeded that of fcs2. fcs1 expression levels from p-coumarate and ferulate grown cells were 10-fold higher than those of fcs2 during growth on either HCA (Figure 3.9).

To expand our understanding of fcs gene regulation in S. stellata, additional sequence analysis was performed to identify putative transcriptional regulators. A putative MarR-type regulator (WP_005859349, SSE37_12344) is present within the fcs1 region (Figure 3.4). Alignment of the S. stellata MarR to this protein yields very weak identity (35%, E 3.2). Along those same lines, there is no notable similarity between the S. stellata predicted regulatory protein and the predicted regulator of the A. fabrum hca catabolon, GntR (NP_354412, Atu1405). Here, the S. stellata putative protein (WP_005861787) with highest sequence similarity to GntR demonstrates low identity (30%) and poor E value (E-6). Furthermore, it is not located near either fcs gene. A single putative regulator (SSE37_24449) is found within the fcs2 region. This gene is located within and co-transcribed with the box operon, but is not anticipated to be co-transcribed with fcs2 given the intergenic distance (>500 bp) (Figure 3.10).

Identification of a transporter for hydroxycinnamic acids in S. stellata Though Fcs1 appears to play a more dominant role in p-coumarate degradation and Fcs2 in ferulate degradation, our results suggested overlapping specificities, implying compensatory roles for each protein when provided a non-preferred HCA substrate. We next sought to determine whether this apparent relaxation of specificity was extended to proteins involved in uptake of these compounds.

We focused our efforts on the dctPQM cassette located adjacent to the fcs2 gene (Figure 3.4). Initial RT-qPCR studies of ferulate and p-coumarate grown cells showed dctP gene expression to be elevated 13-fold (±2) and 23-fold (±14), respectively, relative to acetate-grown cultures. To functionally determine the role of this putative aromatic transporter, dctP (SSE37_24379) was interrupted using insertional mutagenesis. The dctP mutant was unable to grow on either ferulate or p-coumarate (Fig S7), supporting its indispensable role in HCA transport. Partial restoration of activity was evident with a wildtype dctP complemented strain (Figure 3.11).

IV. Discussion

Despite the abundance of hydroxycinnamates (HCA) in lignin-rich environments and the demonstrated importance of bacteria in the recycling of this material, relatively little is known of the bacterial catabolic pathways used to remineralize these compounds. Using the model marine heterotrophic bacterium, Sagittula stellata E-37, we provide compelling evidence for the presence of multiple HCA degradation pathways with overlapping substrate specificities. We argue that this genetic arrangement allows the strain to be poised to utilize these compounds in nature, where they are found in complex mixtures of aromatics.

Our data support a primary role for fcs1 in p-coumarate catabolism and fcs2 in ferulate catabolism in S. stellata, with evidence of overlapping functionalities between the two enzymes encoded by these genes. The ability of each single mutant to ultimately reach wildtype cell densities on non-optimal substrates suggests a compensatory function for each of their gene products. The concept of overlapping substrate specificity for this class of enzymes is not novel, as many aromatic CoA ligases have been demonstrated to transform a variety of structurally related compounds. As previously mentioned, the hydroxycinnamoyl- CoA ligases from A. baylyi (HcaA) (Parke & Ornston, 2003), Sphingobium sp. 62

SYK-6 (FerA) (Masai et al., 2002), and A. fabrum (Atu1416) (Baude et al., 2016; Campillo et al., 2014) demonstrate broad substrate specificity. Although these well-documented CoA ligases convert multiple substrates, these strains do not possess multiple hydroxycinnamoyl-CoA ligase homologs nor is there evidence for compensatory function. Accordingly, single fcs gene knockouts were sufficient to prevent HCA utilization in both A. fabrum and Sphingobium (Campillo et al., 2014; Masai et al., 2002).

The expanded substrate range of the two Fcs proteins from S. stellata is further corroborated by transcriptional analyses of the wildtype strain, which show that both fcs1 and fcs2 are upregulated in cultures grown on either HCA. However, transcriptional responses to growth on HCAs were different, perhaps in part due to differences in their basal (non-induced) expression levels. S. stellata grown with acetate as the sole carbon source, had 5.2x (+/- 0.9) more fcs2 transcripts than fcs1 transcripts. This difference in basal gene expression suggests these genes are under different transcriptional regulation. Consequently S. stellata appears better poised to rapidly oxidize ferulate with Fcs2. If substantiated, these findings could indicate S. stellata is primed to respond to ferulate as a growth substrate.

The difference in expression of fcs1 and fcs2 in S. stellata suggests distinct regulation of the two genes. This is perhaps intuitive given the physical separation of the two fcs genes within the S. stellata chromosome. There are annotated regulators within both fcs genomic regions that could potentially serve this function, however, convincing genomic evidence is lacking. In A. baylyi, a MarR-type regulator (AAP78949) is responsible for regulating an hca operon (Parke & Ornston, 2003). While a putative MarR-type regulator (SSE37_12344) is located in the S. stellata fcs1 region, it shares little sequence similarity to the A. baylyi characterized protein. A single putative regulator (SSE37_24449) is found within the fcs2 region, but is not predicted to regulate HCA degradation based on 63 gene expression assays. This ORF (SSE37_24449) is predicted to encode a transcriptional regulator for benzoate catabolism and has been accordingly designated BoxR. SSE37_24449 shares 44% (E -68) sequence identity with two transcriptional regulators of benzoate catabolism in Azoarcus sp. CIB (BoxR [Ga0098266_114713] and BzdR [Ga0098266_111630]) (Barragán et al., 2005; Valderrama et al., 2012), the bacterium for which the novel aerobic box pathway was first described (Gescher, Eisenreich, Wörth, Bacher, & Fuchs, 2005). Finally, SSE37_24449 contains the Walker-A motif (consensus sequence: GLRGAGK(T/S)) which has been proposed to specifically interact with benzoyl- CoA (Barragán et al., 2005). Thus, there is currently no evidence to support the involvement of SSE37_24449 nor SSE37_12344 in direct genetic regulation of HCA catabolism in S. stellata.

The apparent relaxed substrate specificity for the HCA catabolic proteins appears to extend to proteins involved in transport. The TRAP transporter gene, dctP (SSE37_24379), is required for growth on both ferulate and p-coumarate. The upregulation of dctP in S. stellata cultures grown on p-coumarate or ferulate provides additional evidence that the products of the dctPQM cassette act on both p-coumarate and ferulate. The broad specificity of TRAP family members is well documented and includes aromatic acids (Mulligan et al., 2011). In Comamonas sp. strain DJ-12, transport of 4-chlorobenzoate and related compounds is mediated by a TRAP-family transporter (Chae & Zylstra, 2006). The Rhodopseudomonas palustris TRAP substrate-binding protein TarP has been shown to bind not only ferulate and p-coumarate, but also caffeate and cinnammate (Salmon et al., 2013). No other aromatic acid transporters have yet been identified in S. stellata and very few genes with sequence similarity to transporter proteins have been identified in aromatic compound catabolism gene clusters in this strain (Gulvik & Buchan, 2013).

The ability of the double fcs mutant to sustain limited growth on ferulate is intriguing. It is conceivable that S. stellata is able to access the methoxy side chain of ferulate, which is absent in p-coumarate, to support restricted growth. The spectrophotometric assay indicates the double mutant is capable of taking up ~30% of the provided substrate, though cellular growth is not consistent with complete utilization of this amount of ferulate. This could occur through removal of the entire methoxy group to generate p-coumarate, an unusable product for the double mutant, or through removal of the methyl group to yield caffeate, as observed in plants (Ni et al., 2015), a substrate S. stellata is unable to catabolize (Gulvik & Buchan, 2013). Although documentation of methoxy utilization from lignin-derived aromatics is typically observed with anaerobic methanogens or acetogens (Kato et al., 2015; Sembiring & Winter, 1990), S. stellata harbors a putative vanillate-O-demethylase (SSE37_02815) that could presumably remove the methyl group from ferulate. The energy and biomass derived from a C1 methyl group would be considerably less than that of C10 ferulate, which is consistent with the growth phenotype of the mutant. Furthermore, although results presented here indicate S. stellata predominantly degrades ferulate via a β-oxidative mechanism, it is feasible that additional parallel, not-yet-characterized pathway(s), for ferulate degradation exist in this strain and could explain this restricted growth phenotype.

While this study suggests a dominant role of fcs1 for p-coumarate utilization and fcs2 for ferulate utilization, an important ecological finding is the ability of both of these genes to be upregulated by its non-cognate HCA substrate. The cross- substrate utilization may reflect the diverse substrate pools to which the strain is adapted. The eastern US coastal salt marshes from which the strain was isolated (González et al., 1996) are rich in decaying plant matter, principally derived from the cordgrass Spartina alterniflora, which is rich in ferulate and p-coumarate (Bergbauer & Newell, 1992). Indeed, plant detritus is the principle source of dissolved organic carbon in these territories (Moran & Hodson, 1990). These 65 coastal marshes are also dominated by members of the Roseobacter lineage, to which S. stellata belongs (Gonzalez & Moran, 1997). The ability to degrade aromatic compounds is common amongst lineage members, as is the presence of multiple (≤6) ring-cleaving pathways (Newton et al., 2010). Roseobacters, in general, and S. stellata, in particular, appear to be uniquely adapted to accessing lignin-derived carbon. S. stellata is able to mineralize synthetic lignin (J. Gonzalez et al., 1997) and is adapted to growth on mixtures of aromatic compounds. The strain demonstrates a synergistic growth response when provided mixtures of two lignin degradation products, benzoate and p- hydroxybenzoate (Gulvik & Buchan, 2013). This phenomenon has been shown for other roseobacters (Gulvik & Buchan, 2013) and indicates a positive relationship between substrate complexity and aromatic compound utilization in this group of bacteria. In this context, the multi-substrate specificity of the two hydroxycinnamoyl-CoA synthases encoded by S. stellata seems conducive to its lifestyle, as it appears to have evolved an arsenal of tools to degrade the wide breadth of aromatics in its surroundings.

These insights into HCA catabolism of the marine bacterium S. stellata unveil the modularity and complexity of approaches for aromatic compound degradation within this organism. Through a continued understanding of the reactions used to transform lignin-derived compounds, there is potential for uncovering novel physiologies, pathways, and enzymes that could be of industrial use, as many valuable products can be generated from lignin (Ragauskas et al., 2014) and its HCA derivatives (Kumar & Pruthi, 2014; Priefert et al., 2001; Vargas-Tah & Gosset, 2015). The ease of handling and expanding genetic tractability of S. stellata and other roseobacters (Piekarski et al., 2009) offers an exciting opportunity for future engineering applications.

V. Acknowledgements

We are grateful to Elizabeth Fozo (UTK) for providing guidance on transcriptional analysis of the box operon. This research was supported by a grant from the National Science Foundation [OCE-1357242]. Additionally, AMF and CAG were supported as part of the Center for Direct Catalytic Conversion of Biomass to Biofuels (C3Bio), an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, award number DE-SC0000997.

Author contribution statement: AMF, CAG and AB developed initial project objectives and experimental design. AMF generated fcs mutants; MJC constructed dctP mutant. AMF and MJC performed mutant growth characterizations. MJC performed qRT-PCR assays, with statistical support from CAG. AMF led fcs-data analysis and manuscript preparation, with support from MJC and AB.

VI. References

Alexeyev, M. F. (1999). The pKNOCK series of broad-host-range mobilizable suicide vectors for gene knockout and targeted DNA insertion into the chromosome of gram-negative bacteria. BioTechniques, 26(5), 824-826, 828. Barragán, M. J. L., Blázquez, B., Zamarro, M. T., Mancheño, J. M., García, J. L., Díaz, E., & Carmona, M. (2005). BzdR, a repressor that controls the anaerobic catabolism of benzoate in Azoarcus sp. CIB, is the first Member of a new subfamily of transcriptional regulators. Journal of Biological Chemistry, 280(11), 10683-10694. Baude, J., Vial, L., Villard, C., Campillo, T., Lavire, C., Nesme, X., & Hommais, F. (2016). Coordinated regulation of species-specific hydroxycinnamic acid degradation and siderophore biosynthesis pathways in Agrobacterium fabrum. Applied and environmental microbiology, 82(12), 3515-3524. Bergbauer, M., & Newell, S. Y. (1992). Contribution to lignocellulose degradation and DOC formation from a salt marsh macrophyte by the ascomycete Phaeosphaeria spartinicola. FEMS Microbiology Letters, 86(4), 341-347. doi:10.1111/j.1574-6968.1992.tb04826.x Borglin, S., Joyner, D., DeAngelis, K. M., Khudyakov, J., D’haeseleer, P., Joachimiak, M. P., & Hazen, T. (2012). Application of phenotypic microarrays to environmental microbiology. Current Opinion in Biotechnology, 23(1), 41-48. doi:https://doi.org/10.1016/j.copbio.2011.12.006 Buchan, A., Collier, L. S., Neidle, E. L., & Moran, M. A. (2000). Key aromatic- ring-cleaving enzyme, protocatechuate 3, 4-dioxygenase, in the ecologically important marineRoseobacter lineage. Applied and environmental microbiology, 66(11), 4662-4672. Buchan, A., González, J. M., & Moran, M. A. (2005). Overview of the marine Roseobacter lineage. Applied and environmental microbiology, 71(10), 5665-5677. Bugg, T. D., Ahmad, M., Hardiman, E. M., & Rahmanpour, R. (2011). Pathways for degradation of lignin in bacteria and fungi. Nat Prod Rep, 28(12), 1883- 1896. doi:10.1039/c1np00042j Campillo, T., Renoud, S., Kerzaon, I., Vial, L., Baude, J., Gaillard, V., et al. (2014). Analysis of hydroxycinnamic acids degradation in Agrobacterium fabrum reveals a CoA-dependent, beta-oxidative deacetylation pathway. Applied and environmental microbiology, AEM. 00475-00414. Chae, J.-C., & Zylstra, G. J. (2006). 4-chlorobenzoate uptake in Comamonas sp. Strain DJ-12 is mediated by a Tripartite ATP-independent periplasmic transporter. Journal of bacteriology, 188(24), 8407. de Oliveira, D. M., Finger‐Teixeira, A., Rodrigues Mota, T., Salvador, V. H., Moreira‐Vilar, F. C., Correa Molinari, H. B., et al. (2015). Ferulic acid: a

key component in grass lignocellulose recalcitrance to hydrolysis. Plant biotechnology journal, 13(9), 1224-1232. Diaz, E., Jimenez, J. I., & Nogales, J. (2013). Aerobic degradation of aromatic compounds. Curr Opin Biotechnol, 24(3), 431-442. doi:10.1016/j.copbio.2012.10.010 Dixon, R. A., Achnine, L., Kota, P., Liu, C. J., Reddy, M. S., & Wang, L. (2002). The phenylpropanoid pathway and plant defence-a genomics perspective. Mol Plant Pathol, 3(5), 371-390. doi:10.1046/j.1364-3703.2002.00131.x Fresco, P., Borges, F., Diniz, C., & Marques, M. (2006). New insights on the anticancer properties of dietary polyphenols. Medicinal research reviews, 26(6), 747-766. Fuchs, G. (2008). Anaerobic metabolism of aromatic compounds. Annals of the New York Academy of Sciences, 1125(1), 82-99. doi:doi:10.1196/annals.1419.010 Gescher, J., Eisenreich, W., Wörth, J., Bacher, A., & Fuchs, G. (2005). Aerobic benzoyl-CoA catabolic pathway in Azoarcus evansii: studies on the non- oxygenolytic ring cleavage enzyme. Molecular Microbiology, 56(6), 1586- 1600. doi:10.1111/j.1365-2958.2005.04637.x Gonzalez, J., Mayer, F., Moran, M., Hodson, R., & Whitman, W. (1997). Sagittula stellata gen. nov., sp. nov., a lignin-transforming bacterium from a coastal environment. International Journal of Systematic and Evolutionary Microbiology, 47(3), 773-780. Gonzalez, J. M., & Moran, M. A. (1997). Numerical dominance of a group of marine bacteria in the alpha-subclass of the class Proteobacteria in coastal seawater. Appl Environ Microbiol, 63(11), 4237-4242. González, J. M., Whitman, W. B., Hodson, R. E., & Moran, M. A. (1996). Identifying numerically abundant culturable bacteria from complex communities: an example from a lignin enrichment culture. Applied and environmental microbiology, 62(12), 4433. Gulvik, C. A., & Buchan, A. (2013). Simultaneous catabolism of plant-derived aromatic compounds results in enhanced growth for members of the Roseobacter lineage. Applied and Environmental Microbiology, 79(12), 3716-3723. Hartley, R. D., & Ford, C. W. (1989). Phenolic constituents of plant cell walls and wall biodegradability. In Plant Cell Wall Polymers (Vol. 399, pp. 137-145): American Chemical Society. Harwood, C. S., & Parales, R. E. (1996). The beta-ketoadipate pathway and the biology of self-identity. Annu Rev Microbiol, 50, 553-590. doi:10.1146/annurev.micro.50.1.553 Hellemans, J., Mortier, G., De Paepe, A., Speleman, F., & Vandesompele, J. (2007). qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biology, 8(2), R19. doi:10.1186/gb-2007-8-2-r19

Jones, R. M., & Williams, P. A. (2003). Mutational analysis of the critical bases involved in activation of the AreR-regulated sigma54-dependent promoter in Acinetobacter sp. strain ADP1. Applied and environmental microbiology, 69(9), 5627. Kato, S., Chino, K., Kamimura, N., Masai, E., Yumoto, I., & Kamagata, Y. (2015). Methanogenic degradation of lignin-derived monoaromatic compounds by microbial enrichments from rice paddy field soil. Scientific Reports, 5, 14295. doi:10.1038/srep14295 https://www.nature.com/articles/srep14295#supplementary-information Keen, N. T., Tamaki, S., Kobayashi, D., & Trollinger, D. (1988). Improved broad- host-range plasmids for DNA cloning in Gram-negative bacteria. Gene, 70(1), 191-197. doi:https://doi.org/10.1016/0378-1119(88)90117-5 Kirk, T. K., & Farrell, R. L. (1987). Enzymatic "Combustion": the microbial degradation of lignin. Annual Review of Microbiology, 41(1), 465-501. doi:10.1146/annurev.mi.41.100187.002341 Kroon, P. A., & Williamson, G. (1999). Hydroxycinnamates in plants and food: current and future perspectives. Journal of the Science of Food and Agriculture, 79(3), 355-361. doi:10.1002/(SICI)1097- 0010(19990301)79:3<355::AID-JSFA255>3.0.CO;2-G Kumar, N., & Pruthi, V. (2014). Potential applications of ferulic acid from natural sources. Biotechnology Reports, 4, 86-93. Kuwahara, M., Glenn, J. K., Morgan, M. A., & Gold, M. H. (1984). Separation and characterization of two extracelluar H2O2-dependent oxidases from ligninolytic cultures of Phanerochaete chrysosporium. FEBS Letters, 169(2), 247-250. doi:doi:10.1016/0014-5793(84)80327-0 Laskar, D. D., Yang, B., Wang, H., & Lee, J. (2013). Pathways for biomass- derived lignin to hydrocarbon fuels. Biofuels, Bioproducts and Biorefining, 7(5), 602-626. doi:doi:10.1002/bbb.1422 Lidbury, I., Murrell, J. C., & Chen, Y. (2014). Trimethylamine N-oxide metabolism by abundant marine heterotrophic bacteria. Proceedings of the National Academy of Sciences, 111(7), 2710. Lowe, T. M., Ailloud, F., & Allen, C. (2015). Hydroxycinnamic acid degradation, a broadly conserved trait, protects Ralstonia solanacearum from chemical plant defenses and contributes to root colonization and virulence. Molecular Plant-Microbe Interactions, 28(3), 286-297. Macoy, D. M., Kim, W.-Y., Lee, S. Y., & Kim, M. G. (2015). Biotic stress related functions of hydroxycinnamic acid amide in plants. Journal of Plant Biology, 58(3), 156-163. Masai, E., Harada, K., Peng, X., Kitayama, H., Katayama, Y., & Fukuda, M. (2002). Cloning and characterization of the ferulic acid catabolic genes of Sphingomonas paucimobilis SYK-6. Applied and environmental microbiology, 68(9), 4416-4424. Mattes, T. E., Alexander, A. K., Richardson, P. M., Munk, A. C., Han, C. S., Stothard, P., & Coleman, N. V. (2008). The genome of Polaromonas sp.

strain JS666: insights into the evolution of a hydrocarbon-and xenobiotic- degrading bacterium, and features of relevance to biotechnology. Applied and environmental microbiology, 74(20), 6405-6416. Metcalf, W. W., Jiang, W., Daniels, L. L., Kim, S.-K., Haldimann, A., & Wanner, B. L. (1996). Conditionally replicative and conjugative plasmids carrying laczα for cloning, mutagenesis, and allele replacement in bacteria. Plasmid, 35(1), 1-13. doi:https://doi.org/10.1006/plas.1996.0001 Mitra, A., Kitamura, Y., Gasson, M. J., Narbad, A., Parr, A. J., Payne, J., et al. (1999). 4-hydroxycinnamoyl-CoA hydratase/lyase (HCHL)—an enzyme of phenylpropanoid chain cleavage from Pseudomonas. Archives of biochemistry and biophysics, 365(1), 10-16. Moran, M. A., & Hodson, R. E. (1990). Bacterial production on humic and nonhumic components of dissolved organic carbon. Limnology and Oceanography, 35(8), 1744-1756. doi:10.4319/lo.1990.35.8.1744 Moran, M. A., & Hodson, R. E. (1994). Dissolved humic substances of vascular plant origin in a coastal marine environment. Limnology and Oceanography, 39(4), 762-771. Muheim, A., & Lerch, K. (1999). Towards a high-yield bioconversion of ferulic acid to vanillin. Applied Microbiology and Biotechnology, 51(4), 456-461. Mulligan, C., Fischer, M., & Thomas, G. H. (2011). Tripartite ATP-independent periplasmic (TRAP) transporters in bacteria and archaea. FEMS Microbiology Reviews, 35(1), 68-86. doi:10.1111/j.1574- 6976.2010.00236.x Naczk, M., & Shahidi, F. (2006). Phenolics in cereals, fruits and vegetables: occurrence, extraction and analysis. J Pharm Biomed Anal, 41(5), 1523- 1542. doi:10.1016/j.jpba.2006.04.002 Newton, R. J., Griffin, L. E., Bowles, K. M., Meile, C., Gifford, S., Givens, C. E., et al. (2010). Genome characteristics of a generalist marine bacterial lineage. ISME J, 4(6), 784-798. doi:10.1038/ismej.2009.150 Ni, J., Tao, F., Du, H., & Xu, P. (2015). Mimicking a natural pathway for de novo biosynthesis: natural vanillin production from accessible carbon sources. Scientific Reports, 5, 13670. doi:10.1038/srep13670 https://www.nature.com/articles/srep13670#supplementary-information Otani, H., Lee, Y.-E., Casabon, I., & Eltis, L. D. (2014). Characterization of p- hydroxycinnamate catabolism in a soil Actinobacterium. Journal of bacteriology, JB. 02247-02214. Overhage, J., Priefert, H., & Steinbüchel, A. (1999). Biochemical and genetic analyses of ferulic acid catabolism in Pseudomonas sp. strain HR199. Applied and environmental microbiology, 65(11), 4837-4847. Parke, D., & Ornston, L. N. (2003). Hydroxycinnamate (hca) catabolic genes from Acinetobacter sp. strain ADP1 are repressed by HcaR and are induced by hydroxycinnamoyl-coenzyme A thioesters. Applied and environmental microbiology, 69(9), 5398-5409.

Piekarski, T., Buchholz, I., Drepper, T., Schobert, M., Wagner-Doebler, I., Tielen, P., & Jahn, D. (2009). Genetic tools for the investigation of Roseobacter clade bacteria. BMC Microbiol, 9, 265. doi:10.1186/1471-2180-9-265 Plaggenborg, R., Overhage, J., Loos, A., Archer, J. A., Lessard, P., Sinskey, A. J., et al. (2006). Potential of Rhodococcus strains for biotechnological vanillin production from ferulic acid and eugenol. Applied Microbiology and Biotechnology, 72(4), 745. Priefert, H., Rabenhorst, J., & Steinbüchel, A. (2001). Biotechnological production of vanillin. Applied Microbiology and Biotechnology, 56(3-4), 296-314. Ragauskas, A. J., Beckham, G. T., Biddy, M. J., Chandra, R., Chen, F., Davis, M. F., et al. (2014). Lignin valorization: improving lignin processing in the biorefinery. Science, 344(6185), 1246843. doi:10.1126/science.1246843 Ralph, J., Hatfield, R., Grabber, J., Jung, H.-J. G., Quideau, S., & Helm, R. (1998). Cell wall cross-linking in grasses by ferulates and diferulates. Lignin and lignan biosynthesis. Ralph, J., Hatfield, R. D., Quideau, S., Helm, R. F., Grabber, J. H., & Jung, H.-J. G. (1994). Pathway of p-coumaric acid incorporation into maize lignin as revealed by NMR. Journal of the American Chemical Society, 116(21), 9448-9456. doi:10.1021/ja00100a006 Razzaghi-Asl, N., Garrido, J., Khazraei, H., Borges, F., & Firuzi, O. (2013). Antioxidant properties of hydroxycinnamic acids: a review of structure- activity relationships. Current medicinal chemistry, 20(36), 4436-4450. Salmon, R. C., Cliff, M. J., Rafferty, J. B., & Kelly, D. J. (2013). The CouPSTU and TarPQM Transporters in Rhodopseudomonas palustris: redundant, promiscuous uptake systems for lignin-derived aromatic substrates. PLOS ONE, 8(3), e59844. doi:10.1371/journal.pone.0059844 Sembiring, T., & Winter, J. (1990). Demethylation of aromatic compounds by strain B10 and complete degradation of 3-methoxybenzoate in co-culture with Desulfosarcina strains. Applied Microbiology and Biotechnology, 33(2), 233-238. doi:10.1007/BF00176531 Sikkema, J., de Bont, J. A., & Poolman, B. (1994). Interactions of cyclic hydrocarbons with biological membranes. Journal of Biological Chemistry, 269(11), 8022-8028. Sikkema, J., de Bont, J. A., & Poolman, B. (1995). Mechanisms of membrane toxicity of hydrocarbons. Microbiological Reviews, 59(2), 201. Stibitz, S. (1994). [35] Use of conditionally counterselectable suicide vectors for allelic exchange. In Methods in Enzymology (Vol. 235, pp. 458-465): Academic Press. Sun, R.-C., Sun, X.-F., & Zhang, S.-H. (2001). Quantitative determination of hydroxycinnamic acids in wheat, rice, rye, and barley straws, maize stems, oil palm frond fiber, and fast-growing poplar wood. Journal of Agricultural and Food Chemistry, 49(11), 5122-5129.

Teixeira, J., Gaspar, A., Garrido, E. M., Garrido, J., & Borges, F. (2013). Hydroxycinnamic acid antioxidants: an electrochemical overview. BioMed research international, 2013. Tien, M., Kirk, T. K., Bull, C., & Fee, J. A. (1986). Steady-state and transient- state kinetic studies on the oxidation of 3,4-dimethoxybenzyl alcohol catalyzed by the ligninase of Phanerocheate chrysosporium Burds. J Biol Chem, 261(4), 1687-1693. Torres-Mancera, M. T., Cordova-López, J., Rodríguez-Serrano, G., Roussos, S., Ramírez-Coronel, M. A., Favela-Torres, E., & Saucedo-Castañeda, G. (2011). Enzymatic extraction of hydroxycinnamic acids from coffee pulp. Food Technology and Biotechnology, 49(3), 369. Valderrama, J. A., Durante-Rodriguez, G., Blazquez, B., Garcia, J. L., Carmona, M., & Diaz, E. (2012). Bacterial degradation of benzoate: Cross-regulation between aerobic and anaerobic pathways. Journal of Biological Chemistry. Vargas-Tah, A., & Gosset, G. (2015). Production of cinnamic and p- hydroxycinnamic acids in engineered microbes. Frontiers in bioengineering and biotechnology, 3, 116.

VII. Appendix

Table 3.1. Plasmids and strains used in this study.

Plasmids Description Source

pCR™2.1-TOPO® TA plasmid for cloning Invitrogen

pRMJ1 Plasmid harboring sacB-kanR cassette for marker exchange Jones & Williams, mutagenesis 2003 pARO180 D. Park Mobilizable suicide plasmid for marker exchange mutagenesis (ATCC 77123) pARO180_12343 pARO180 containing fused flanking regions of fcs1 This study (SSE37_12324) cloned into the XbaI and HindIII sites of the MCS pARO180_12324_SK pARO180_12324 with sacB-kanR from pRMJ1 inserted into the This study middle of the fcs1 insert through the designed SalI site pKNOCK-Km Mobilizable suicide plasmid harboring kanamycin resistance gene. Alexeyev, 1999 Used for site-directed mutagenesis via chromosomal integration (Addgene 46262) pKNOCK-Km_24399 pKNOCK-Km with small internal region (231 bp) of fcs2 cloned into This study the BamHI and XhoI sites of the MCS pKNOCK-Tc Mobilizable suicide plasmid harboring tetracycline resistance gene. Alexeyev, 1999 Used for site-directed mutagenesis via chromosomal integration (Addgene 46259) pKNOCK-Tc_24399 pKNOCK-Tc with small internal region (231 bp) of fcs2 cloned into This study the BamHI and XhoI sites of the MCS pKNOCK-Km_24379 pKNOCK-Km with 292 bp internal region of dctP cloned into the This study HindIII and BamHI sites of the MCS pRK415 Broad host range plasmid for DNA cloning in Gram-negative Keen et al., 1988 bacteria used for dctP complementation pRK415_24379 pRK415 containing a 1.7 kb HindIII-BamHI DNA fragment that includes the region 579 bp upstream of the dctP start codon and This study 83 bp downstream of the dctP stop codon Strains

Gonzalez & Moran, E-37 Wildtype Sagitulla stellata E-37 1997 E-37 fcs1::sacBkanR E-37 fcs1 mutant where fcs1 (SE37_12324) is replaced with sacB- This study kanR from pARO180_12324_SK E-37 fcs2::pKNOCK- E-37 fcs2 mutant where fcs2 (SSE37_24399) is interrupted by This study Km pKNOCK-Km_24399 E-37 fcs1::sacBkanR E-37 fcs1/fcs2 double mutant generated by interrupting fcs2 fcs2::pKNOCK-Tc (SSE37_24399) with pKNOCK-Tc_24399 in strain E-37 This study fcs1::sacBkanR E-37 dctP::pKNOCK- E-37 dctP mutant wherein dctP (SSE37_24379) was interrupted This study Km by pKNOCK-Km_24379 E-37 dctP-pKNOCK- Km_dctP::pRK415 E-37 dctP mutant complemented with pRK415_24379 This study E-37 dctP-pKNOCK- Km_pRK415 E-37 dctP mutant transformed with empty pRK415 vector This study E. coli TOP10 Cloning strain for pCR™2.1-TOPO® plasmid Invitrogen E. coli JM109 Subcloning and transformation strain Invitrogen E. coli S17-1 Mating strain for marker exchange mutagenesis ATCC 47055

E. coli BW20767 Metcalf et al., 1996 Mating strain for pKNOCK mutagenesis, derived from S17-1 (ATCC 47084) 74

Table 3.2. Primers used in this study.

Name Sequence Function

12324_A_ex GATCACTGCGCCGATCTT Primer pair amplifies A region (402 bp upstream 12324_A_in gattcgaggagcgatagagctGTCGAC of 12324) with 21 bp scar SalI site appended to TTCTCCTCCTGGCGTTTTAG* the 3’ end. A_ex primer is also use with B_ex for cross-over pCR to generate the A-scar-B product

12324_B_ex GGAGCCCAACAGCATCAT Primer pair amplifies B region (585 bp 12324_A_in agctctatcgctcctcgaatcGACCGGC downstream of 12324) with 21 bp scar appended GCCCGGTT* to the 5’ end. B_ex primer is also use with A_ex for cross-over PCR to generate the A-scar-B product

12324_Asac_F GGATCAGGATGTCGTCGTTC Primer pair binds E-37 DNA upstream and 12324_Bkm_R CAGCTTCTCCTCGATTCGCT downstream of fcs1. Used to verify replacement of gene with sacBkanR cassette

24399_int_F TGGTGTTTTATGCAGGCGCG Primer pair amplifies 231 bp internal region of 24399_int_R TTTCGCAGCGCATGTCTTCG fcs2 (SSE37_24399) used for cloning into pKNOCK plasmids pKNOCKKm_752_F ACGGCTGACATGGGAATTCC Primer pair binds around MCS of pKNOCK-Km to pKNOCKKm_901_R GCGGAATTAATTCGACGCGTC amplify inserts off of plasmid pKNOCKTc_425_F GAATTCCCCTCCACCGCGG Primer pair binds around MCS of pKNOCK-Tc to pKNOCKTc_558_R TGATCAAGCTGACGCGTCCT amplify inserts off of plasmid

24399_884_F GAACGCTGGCCTTCAACGTG Primer pair binds E-37 DNA upstream and 24399_1383_R GAAATCCTCCGAAATCCGCCC downstream of the 231 bp internal region of fcs2. Used to verify insertion of pKNOCK plasmids in fcs2 of E-37

24379_423_F TCTGGAAGGCATGAAGATCC Primer pair amplifies 292 bp region of dctP 24379_714_R ATCGATCACCTCCTGCAGAT (SSE37_24379) used for cloning into pKNOCK- Km

24379_367966_F TCTAGAAGCTTTGTTTGATGGA Primer pair amplifies 1.7 kb HindIII-BamHI DNA 24379_369669_R TCAACGCACT fragment which encompasses the region 579 bp TGCGTAGGATCCATCAGGATC upstream of the dctP start codon and 83 bp AGGAACGTCAGC downstream of the dctP stop codon

24379_690_F AGAGGATCTGCAGGAGGTGA Primer pair amplifies 188 bp region of dctP 24379_877_R ACGTCTCGTAGATCGGGTTG (SSE37_24379) used for quantifying dctP expression in WT E-37 on hydroxycinnamic acids

12324_562_F CTGACAGCGACACACCTGAC Amplifies 181 bp region of (SSE37_12324) for 12324_742_R CCTTCAGGAAGGTGAAGCAC quantifying fcs1 expression in WT E-37 and fcs2 mutant

24399_1249_F CTCAACGACCCGAAGAAGAC Amplifies 187 bp region of fcs2 (SSE37_24399) 24399_1435_R ACAATTCCAGACGCAGGTTC for quantifying fcs2 expression in WT E-37 and fcs1 mutant

15096_340_F GCCCATATCTGGTTCCTCAA Primer pair amplifies 159 bp region of rpoC 15096_498_R TTCCTCTTCGGTCAGCATCT (SSE37_15096) for use as housekeeping gene 1 of 3 in RT-qPCR map_13553_620_F GCATGTTCTTCACCATCGAG Primer pair amplifies 166 bp region of map map_13553_785_R GCGGGAGAGAGGGTAAAGAT (SSE37_13553) for use as housekeeping gene 2 of 3 in RT-qPCR

* 21 bp scar sequence indicated in lower case. SalI site underlined.

Table 3.3. E-37 homologs to A. fabrum HCA catabolism proteins in NCBI.

A. fabrum S. stellata locusa homolog locusa Putative function Coverage IDb E-value

Atu1414 SSE37_12329 acyl-CoA dehydrogenase 83% 54% 3e-138 (NP_354421) (WP_005859343)

Atu1415 SSE37_24394 acyl-CoA dehydrogenase 100% 51% 1e-79 (NP_354422) (WP_005857140)

Atu1416 SSE37_12324 feruloyl-CoA synthase 98% 55% 0.0 (NP_354423) (WP_005859342)

Atu1416 SSE37_24399 feruloyl-CoA synthase 97% 40% 2e-140 (NP_354423) (WP_005857142)

Atu1417 SSE37_12349 enoyl-CoA hydratase 96% 67% 6e-126 (NP_354425) (WP_005859350)

Atu1421 SSE37_12319 β-ketothiolase 100% 82% 8e-180 (NP_354429) (WP_005859340) a NCBI protein accession numbers are provided in parenthesis beneath loci. bID= % amino acid identity

O R = OMe; ferulate ferulate product O- R = H; p-coumarate p-coumarate product HO R Hydroxycinnamate ß-oxidative SSE37_12324 (Fcs1) Hcs aldehyde SSE37_24399 (Fcs2) O CoA shared S

HO R Hydroxycinnamoyl-CoA Ech SSE37_12349 OH O OH O CoA CoA S S

HO HO H SSE37_12329 OMe Ech Ech Acd HPHP-CoA SSE37_24394 HMPHP-CoA SSE37_12349 SSE37_12349

O O O O A y CoA

l a

S H H d w

HO HO HO h

h t

OMe OMe H y a HMPKP-CoA Vanillin d

p p-hydroxybenzaldehyde

Bkt Adh Adh p

v i

a t SSE37_12319 O O

t a

h - - d O O

i x

HO HO a o

- OMe H

ß Vanillate p-hydroxybenzoate

VanAB PobA SSE37_02815 (VanA) SSE37_18837 O

O-

HO OH

Protocatechuate

Figure 3.1. Select HCA degradation pathways initiated by CoA addition. The left branch exhibits the β-oxidative pathway through which ferulate is degraded. The right branch shows the aldehyde pathway through which both ferulate and p-coumarate can be processed. E-37 locus tags are provided at reactions for which there are strong homologs to functionally validated proteins and/or genome annotations. Accession numbers for the designated proteins are as follows: SSE37_12324 (WP_005859342), SSE37_24399 (WP_005857142), SSE37_12349 (WP_005859350), SSE37_24394, (WP_005857140), SSE37_12329 (WP_005859343), SSE37_12319 (WP_005859340), SSE37_18837 (WP_040603982), SSE37_02815 (WP_005862022). Hcs; hydroxycinnamoyl-CoA synthase, Ech; enoyl-CoA hydratase, Acd; acyl-CoA dehydrogenase, Bkt; β-ketothiolase, Adh; aldehyde dehydrogenase, PobA, p-hydroxybenzoate oxygenase, VanA, vanillate-O-demethylase. HMPHP-CoA, 4-hydroxy- 3-methoxyphenyl-β-hydroxypropionyl-CoA; HPHP-CoA, 4-hydroxyphenyl-β-hydroxypropionyl- CoA; HMPKP-CoA, 4-hydroxy-3-methoxyphenyl-β-ketopropionyl-CoA. Dashed arrows indicate reactions unique to the β-oxidative pathway, light solid lines represent reactions unique to the aldehyde pathways, and bold solid lines indicate reactions shared between the two branches. This figure is adapted from Overhage et al., 1999 and Lowe et al., 2015. 77

Figure 3.2. Ferulate and p-coumarate standard curves. Assayed concentration range was 0 to 3 mM.

Figure 3.3. Wavelength scans (200-700 nm) of cell-free spent medium from wildtype, fcs1, and fcs2 mutant cultures grown on 2 mM ferulate (A-F) and p-coumarate (G-L). Sampling time points are indicated for each panel set. Concentrations of ferulate and p-coumarate in Figures 3C and D are derived from similar scans.

SSE37_12324 SSE37_12329 SSE37_12344 Atu1416 (55%) Atu1414 (54%) SSE37_12349 cobP cobT cobS 1 fcs1 acd Atu1417 (67%)

bkt 2 3 4 ech SSE37_12319 Atu1421 (82%)

SSE37_24394 SSE37_24399 Atu1415 (51%) Atu1416 (40%) dctP dctQ dctM acd fcs2 boxL boxB boxC

box pathway 1kb

Figure 3.4. Genetic landscape of the two fcs loci in S. stellata. Locus tags for predicated HCA-related degradation genes in E-37 and associated A. fabrum homologs (denoted with Atu prefix) are provided with percent protein identity above corresponding gene arrows. Dark pink arrows represent genes involved in the β-oxidative and/or the aldehyde pathway described in Figure 1. Light pink arrows indicate genes involved only in the β-oxidative pathway. Gene abbreviations are as follows: bkt = β-ketothiolase, fcs = feruloyl-CoA synthase, acd = acyl-CoA dehydrogenase, ech = enoyl-CoA hydratase. Other functionally related genes are coordinately color coded and include a predicted operon for cobalamin biosynthesis (SSE37_12299, SSE37_12304, SSE37_12309); the dct operon (SSE37_24379, SSE37_24384, SE37_24839) predicted to encode a C4-dicarboxylate TRAP transporter; and a portion of the box pathway (SSE37_24404, SSE37_24409, SSE37_24414, SSE37_24419, SSE37_24424) encoding enzymes for benzoate degradation. Dark grey represents genes for which a defined role in either HCA degradation or another pathway is not predicted: (1) SSE37_12314 is a protein of unknown function (DUF 3237); (2) SSE37_12334 is annotated as an aldo/ketoreductase; (3) SSE37_12339 is annotated as an FAD-dependent oxidoreductase; (4) SSE37_12344 is annotated as a MarR-like transcriptional regulator as discussed in the text. The complete box operon is provided in Figure 3.10.

Figure 3.5. Growth of S. stellata on 2 mM vanillin.

Figure 3.6. Growth curves (A-D) and Biolog respiration assays (E-H) of wildtype S. stellata and fcs mutants. Panels A and E exhibit growth and respiration on the non-aromatic carbon source, acetate, to assess any defects in central metabolism. Panels B and F exhibit growth and respiration on p- hydroxybenzoate (POB) to assess defects in the protocatechuate pathway, through which ferulate and its metabolites are known to proceed. Panels C and G demonstrate growth and respiration on ferulate. Panels D and H demonstrate growth and respiration on p-coumarate. Profiles for growth and respiration curves are consistent except for the fcs1 and fcs1/fcs2 mutants on ferulate. This may be due to interference of a maroon coloration that appears during growth on this substrate. Supplementary analyses with viable counts confirm minimal growth of the fcs1/fcs2 double mutant on ferulate (see main text). Panels C and D are identical to Figure 3.6A and 3.6B and are provided here to facilitate direct comparison between growth dynamics monitored via optical density versus redox-dye based assays. 82

A. 2 mM Ferulate C. 2.5 2

1.5 WT E-37 fcs1 fcs2 fcs1/2

a l

) 1.5

u m

1 r

4 1

(

D 0.5

O 0.5

0 0 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Time (hours) Time (hours)

2 mM p-Coumarate 1.8 B. D. 1.6

1 1.4

t a

0.8 r 1.2

) m

m 1

0.6 o

0 0.8

5 ( M 0.6

0.4 D m 0.4 O 0.2 0.2 0 0 0 10 20 30 40 50 0 10 20 30 40 Time (hours) Time (hours)

Figure 3.7. Catabolism of hydroxycinnamic acids by S. stellata wildtype and the fcs mutants. Growth of strains on ferulate (A) or p-coumarate (B), with corresponding concentrations of ferulate (C) or p-coumarate (D) in cell-free spent medium. Wildtype is shown with black closed circles, the fcs1 mutant with blue open circles, the fcs2 mutant with red open circles, and the fcs1/fcs2 double mutant with yellow open circles. Averages from a minimum of three biological replicates are shown along with one standard deviation from the mean.

140 fcs1

)

fcs2

r o

u 120

o fcs1/fcs2

i s

t 100

a 80 r

e *"

e a

n *"

e 60

(

e o

p 40

d *"

l i % 20 w *" 0 AC POB FA CA

Figure 3.8. Comparison of growth rates of wildtype S. stellata and fcs mutants grown on various carbon sources. Bars represent the growth rate percent relative to wildtype (set to 100%). AC, acetate; POB, p- hydroxybenzoate, FA, ferulate; CA, p-coumarate. Error bars represent the standard deviation from three normalized replicates. Grey asterisks indicate a significant difference (p ≤ 0.01) from the wildype growth rate for the respective substrate.

Figure 3.9. fcs gene expression for S. stellata grown on two hydroxycinnamates. Gene expression was relativized to growth on 7 mM acetate and normalized to three housekeeping genes. Blue bars represent fcs1 expression and red bars represent fcs2 expression. Identical letters above bars indicate non-significant values, whereas non-identical letters represent a significant difference (*= p-value<0.005). Averages from a minimum of three biological and two technical replicates are shown along with one standard deviation from the mean.

Figure 3.10. Genetic landscape of box operon and upstream genes including fcs2 and the dctP operon. Locus tags for S. stellata are listed above the genes. Gene 2: Lactonase. Grey genes: hypothetical protein. Gene 9: thioesterase. Black gene: transcriptional regulator. For co- transcription assays, RNA from benzoate-grown S. stellata was reverse transcribed with random hexamers and used as template in end-point PCRs. Arrows indicate location of primers used for co-transcription studies. Gel image above each set of arrows show results of cDNA amplification. Genomic DNA from E-37 was used as a positive control. No RT controls were performed for all reactions to ensure no DNA contamination. A, D: Lane 1: positive control, Lane 2: negative control, Lane 3: E-37 cDNA. B, E: Lane 2: negative control, Lane 3: positive control, Lane 4: E-37 cDNA. C: Lane 1: negative control, Lane 2: positive control, Lane 3: E-37 cDNA. F, H: Lane 1: negative control, Lane 2: positive control, Lane 3: E-37 cDNA, Lane 4: E-37 cDNA. G: Lane 2: positive control, Lane 3: negative control, Lane 4: E-37 cDNA.

Figure 3.11. Growth curves of WT E-37 and dctP strains provided with either 2 mM ferulate or p-coumarate as the sole carbon source. Complementation of the dctP mutant was attempted by placing a wildtype copy of dctP in pRK415. Empty vector controls are shown in yellow.

CHAPTER IV: HIERARCHICAL SUBSTRATE PREFERENCES BY THE MARINE BACTERIUM SAGITTULA STELLATA DURING GROWTH ON MIXTURES OF PLANT-DERIVED AROMATIC COMPOUNDS

Author contribution statement: Alison Buchan and Michelle Chua developed project objectives and experimental design. Michelle Chua performed all experimental work and data analysis. Michelle Chua led manuscript writing with support from Alison Buchan.

I. Abstract

Mineralization of lignin requires cleavage of the stable aromatic ring and is mediated by microbial decomposers. In nature, plant-derived aromatics are most often found in complex mixtures, yet little is known of how microbes process individual aromatic compounds found in such mixtures. Sagittula stellata is a model heterotrophic marine bacterium capable of utilizing a wide variety of lignin- derived aromatic compounds as primary growth substrates. Prior work revealed that when S. stellata is provided binary mixtures of two lignin-derived aromatics, p-hydroxybenzoate and benzoate, the bacterium simultaneously catabolizes these substrates, displaying a synergistic growth response. In order to test whether this physiological response is broadly characteristic of this strain, growth on a variety of binary aromatic substrate mixtures was tested. In contrast to previous studies, these assays revealed that S. stellata preferentially utilizes the hydroxycinnamates (HCAs) ferulate and p-coumarate, or vanillate, over benzoate. Growth rates on benzoate with either ferulate or vanillate were significantly decreased, at least 47% and 72.5%, respectively, compared to single substrate treatments. Insertional transposon mutagenesis was used to obtain two mutants deficient in benzoate degradation. Characterization of these strains reveal cross-regulation between the benzoate oxidation and HCA or vanillate degradation pathways, perhaps at the level of substrate uptake. These results indicate S. stellata has several mechanisms for mixed substrate utilization and provides insight into how salt marsh microbes transform lignin-derived aromatics in natural environments.

II. Introduction

Plants are the primary contributors of homocyclic aromatic compounds to natural systems. Lignocellulose comprises the bulk of vascular plant biomass and consists principally of the carbohydrate polymers cellulose and hemicellulose and the aromatic polymer lignin (Hodson et al., 1984). Lignin is the most abundant aromatic polymer on earth and the second most abundant organic polymer (Leisola et al., 2012). Lignin-derived aromatic compounds can only be degraded by organisms with suitable enzymes, which are produced by bacteria and fungi (Cragg et al., 2015). The microbial degradation of aromatics is often termed “catabolic funneling” whereby a diverse array of compounds is converted into one of a limited number of intermediates that then undergo ring-cleavage. Subsequent reactions lead to the production of tri-carboxylic acid cycle intermediates (Fuchs et al., 2011) (Figure 4.1). The β-ketoadipate pathway, a paradigm for microbial aerobic aromatic compound catabolism, has two parallel branches. In this pathway, a diverse suite of aromatic compounds are converted to one of two central intermediates (catechol or protocatechuate), which then undergo oxygen-dependent ring cleavage (Harwood & Parales, 1996; Ornston & Stanier, 1966). The majority of the ring-cleaving enzymes employed by aerobic organisms use molecular oxygen to open the stable ring (Fuchs et al., 2011). However, in a more recently discovered pathway for benzoate degradation, designated the box pathway, the ring is prepared for cleavage by CoA activation (Gescher et al., 2002).

When bacteria are provided mixtures of aromatic substrates, substrate preferences are frequently observed. In the bacteria in which this phenomenon has been studied, several underlying mechanisms were found to regulate these preferences, including substrate transporter inhibition, activator protein inhibition, or the catabolite repression control (Crc) protein (Morales et al., 2004; Nichols & Harwood, 1995; Zhan et al., 2009). Historically, mixed substrate studies include

90 benzoate as one of the substrates of mixtures, and it is often found to be the preferred substrate over other aromatics, including p-hydroxybenzoate, phthalate, and phenol (Ampe & Lindley, 1995; Brzostowicz et al., 2003; Choi et al., 2007; Donoso et al., 2011; Gaines et al., 1996; Mazzoli et al., 2007; Nichols & Harwood, 1995; D. Perez-Pantoja et al., 2015; Zhan et al., 2009).

The dissolved organic carbon pool in the coastal oceans are highly enriched in aromatic moeities (Hedges et al., 1997). This pool is derived from a mixture of allochthonous and autochthonous sources, including riverine inputs of terrestrially-derived organic matter and local inputs by marine vascular plants, principally grasses, such as Spartina species (Moran & Hodson, 1994). The mineralization of this material is an important component of the marine carbon cycle and principally mediated by estuarine microbes (Gallagher et al., 1976). Members of the Roseobacter clade of marine bacteria have emerged as models for the degradation of lignin-derived aromatic compounds in the coastal ocean due to their abundance and activities in these systems (reviewed in (Brinkhoff et al., 2008; Buchan et al., 2005; Luo & Moran, 2014). Several Roseobacter strains have been isolated from salt marsh ecosystems, where the lignin-rich cordgrass Spartina alterniflora is the dominant primary producer (Hladik et al., 2013); S. alterniflora lignocellulose contributes 44% of the dissolved organic carbon pool (Moran & Hodson, 1990).

In contrast to the hierarchical substrate preference described for many soil bacteria, Roseobacter strains have been shown to simultaneously catabolize the lignin-derived aromatics benzoate and p-hydroxybenzoate (Gulvik & Buchan, 2013). Furthermore, evidence of synergetic growth responses was revealed in one representative Roseobacter, Sagitulla stellata. The growth rate of S. stellata on the mixed substrates was significantly increased relative to growth on equimolar concentrations of each individual substrate, suggesting an evolutionary advantage to simultaneously catabolizing two aromatics in salt 91 marsh ecosystems. S. stellata possesses genes for six ring-cleaving pathways and is capable of degrading a variety of lignin-derived aromatic compounds (Buchan et al., 2004; Gulvik & Buchan, 2013). Moreover, S. stellata has been demonstrated to transform and partially mineralize synthetic lignin (Gonzalez et al., 1997). When catabolizing lignin-derived aromatics, S. stellata transforms the compounds via peripheral pathways to a few key central intermediates, such as protocatechuate or benzoyl-CoA. In S. stellata, lignin-derived aromatics are principally degraded through the protocatechuate (pca) pathway, with benzoate as the only known aromatic degraded through the benzoate oxidation (box) pathway (Figure 4.1) (Gulvik & Buchan, 2013).

Hydroxycinnamates or HCAs (primarily ferulate and p-coumarate) are the most abundant phenolic acids in S. alterniflora cell walls (Opsahl & Benner, 1995). Thus, HCAs are influential to microbial decomposition of lignin in salt-marshes. Within plant cell walls, ferulate and p-coumarate are bound to lignin polymers via ester linkages (Hatfield et al., 2016). Vanillate is an intermediate of ferulate degradation in S. stellata (Figure 4.1). While simultaneous catabolism of benzoate and p-hydroxybenzoate has been studied in S. stellata, it is not clear if this is a common metabolic response to other aromatic mixtures.

Genetic regulation often underlies catabolite repression. Co-localization of functionally related genes into operons for coordinated gene expression is common for many bacteria in which aromatic compound degradation has been studied (e.g., (Choi et al., 2005; Collier et al., 1998; Patrauchan et al., 2005; Danilo; Perez-Pantoja et al., 2008; Xu et al., 2003). However, there is limited colocalization of functionally related genes mediating aromatic compound catabolism in S. stellata (Gulvik & Buchan, 2013). Furthermore, while the ring- cleaving pathways are well-defined for S. stellata (Buchan et al., 2001; Buchan et al., 2004), the peripheral pathways for conversion of aromatic compounds to these central intermediates are less defined. For example, the genes mediating 92 vanillate degradation in S. stellata are unknown (Frank & Chua et al., accepted). In addition, it has been recently determined that the enzymes mediating HCA degradation in this strain show functional redundancy and are disparately located on the chromosome (Frank & Chua et al., accepted). As a consequence, substrate preference hierarchies for this strain are difficult to predict from genome analysis alone. To that end, this study seeks to characterize the physiological response of S. stellata to a variety of binary mixtures of aromatic compounds. The results provide insight into the fate of phenolic monomers in natural systems.

III. Materials and Methods

Growth conditions and media Sagittula stellata was isolated from Georgia coastal waters by enrichment with indulin, the high molecular weight fraction of pulp mill effluent (Gonzalez et al., 1997). For the acetate and aromatic substrate growth studies, S. stellata was grown on a defined medium: Aromatic Basal Medium [ABM; per liter: 1.5%

(wt/vol) sea salts, 225 nM K2HPO4, 13.35 µM NH4Cl, 71 mM Tris-HCl (pH 7.5), 68 µM Fe-EDTA, trace metals, and vitamins] at 30°C. S. stellata cells were pre- conditioned on ABM supplemented with 10 mM acetate. For growth analyses, 10 mM acetate pre-conditioned cells were transferred for an initial inoculum of

OD540≈0.005 to either 7 mM acetate or 2 mM aromatic substrate, unless otherwise noted. Cells grown in liquid media were incubated at 30°C at 200 rpm, in the dark. Negative controls (without inoculum) were also incubated for each growth experiment. Glassware was combusted for at least 4 hours at 450°C to eliminate residual carbon. Escherichia coli strains used for cloning and conjugation were grown in Lysogeny Broth [LB, per liter: 10 g tryptone, 5 g yeast extract, 10 g NaCl] at 37°C, 200 rpm. Antibiotics were added at 50 μg/mL kanamycin for selection purposes. Growth was measured using a GENESYS™

20 Visible Spectrophotometer at 540 nm for S. stellata and 600 nm for E. coli strains (Table 4.1). Higher optical densities were observed in the ferulate cultures due to the production of an intermediate that displays a maroon color (Frank, Chua et al., accepted). Acetate and aromatic acids were purchased from Fisher Scientific (Waltham, MA) and Sigma-Aldrich, respectively. For routine maintenance, S. stellata was grown on YTSS medium [0.25% (wt/vol) yeast extract, 0.4% (wt/vol) tryptone, 1.5% (wt/vol) sea salts (Sigma-Aldrich, St. Louis, MO)].

Analysis of substrate concentrations Two methodologies were used to quantify aromatic compound concentrations in cultures. With cultures employing vanillate as a growth substrate, high performance liquid chromatography (HPLC) was used. For other cultures, UV spectrophotometry was used. For each HPLC sample, 1 mL of culture was centrifuged at 4°C at ~18,000 x g for 10 minutes. The supernatant was drawn off and filtered through a 13 mm 0.2 µm pore size PVDF filter (Whatman, Maidstone, United Kingdom) prior to injection into the HPLC instrument. Benzoate and vanillate were separated using a Waters 2695 HPLC instrument connected to a reverse phase 3.9- by 150-mm guarded Nova-Pak C18 column coupled to a Waters 2996 photodiode array (PDA) detector (Waters Corp., Milford, MA). The samples were eluted isocratically with a mobile phase consisting of 30% acetonitrile and 70% of 0.07% phosphoric acid at 0.8 mL/minute. Benzoate and vanillate elute at approximately 3.8 (273 nm) and 2.2 (259 nm) minutes, respectively. The λmax values for benzoate and vanillate were 273 and 259 nm, respectively. Benzoate and vanillate standard curves were developed from 0 to 3.0 mM benzoate or vanillate. The standards were prepared in increments of 0.3 mM, resulting in 10 concentrations for each substrate. Benzoate and vanillate stock solutions were dissolved in Milli-Q water and dimethyl sulfoxide (DMSO), respectively, and dilutions for the standard curves were prepared in sterile ABM.

The peak areas of each analyte were calculated using ApexTrack’s integration tool of the Empower 2 Pro software package (Waters).

For cultures employing ferulate and p-coumarate, aromatic substrate concentrations were monitored in S. stellata cultures spectrophotometrically as described previously (Frank, Chua, accepted). Briefly, culture aliquots were passaged through a 0.22 µm filter and absorbances at 222 nm (benzoate), 283 nm (p-coumarate), and 308 nm (ferulate) were monitored with a Beckman DU 800 UV/Vis spectrophotometer (Indianapolis, IN)). Concentrations were calculated from 10-point standard curves (0.0 to 3.0 mM) with r2 values of 0.97 for ferulate, 0.99 for p-coumarate, and 0.99 for benzoate.

Ash free dry mass measurements S. stellata cells were grown to early stationary phase in 250 mL baffled flasks in 100 mL volumes [2 mM total aromatic substrate] and centrifuged at 6,000 x g for 10 minutes at 4°C. The supernatant was filtered onto pre-combusted (450°C for 1 hour) and pre-weighed fiberglass GF/F filters (0.7-µm nominal pore size, Whatman Ltd., Maidstone, ME) to collect cells that did not pellet. Pellets from centrifugation were resuspended in ~0.5 mL spent medium, and the resuspensions were filtered onto their respective filters. Filters were dried at 60°C until the weight was stable. Filters were then combusted at 450°C for 4 hours and weighed to determine Ash Free Dry Mass (AFDM).

Transposon mutant library generation A Tn5 random transposon mutant library for S. stellata was generated using the Tn5::KmR cassette detailed in Larsen et al. (Larsen, Wilson, Guss, & Metcalf, 2002) with modifications as described in Cude et al (Cude et al., 2012). Briefly, 0.5 mL of overnight S. stellata and E. coli EA145 pRL27, a diaminopimelate (DAP) auxotroph, were transferred to 10 mL YTSS or LB plus 1 mM DAP, respectively. Cultures were grown to OD540 ~0.4, and a conjugative mating

95 culture containing 0.2 mL of E. coli donor and 0.2 mL of S. stellata recipient was set up. Mating mixtures were immediately centrifuged at 11,200 x g for 2 min, and the pellet was resuspended in 1 mL fresh YTSS. The mixture was centrifuged at 11,200 x g for 2 min again and the pellet resuspended in 15 μL fresh YTSS. The entire volume of the mating mixture was spotted onto YTSS agar containing 1 mM DAP and the plate incubated at 30°C overnight. The mating mixture was transferred to 1 mL YTSS, diluted, and plated onto YTSS agar containing 50 mg/L kanamycin. To inhibit fungal contamination, 25 µg/mL cycloheximide was added to the YTSS selection plates.

Genetic screen for mutants deficient in benzoate catabolism The S. stellata Tn5 mutant library was spotted onto large YTSS agar plates (9.5 in x 9.5 in) using a flame-sterilized 96 pin replicator. Mutants positive for growth on YTSS were transferred to ABM agar plates supplemented with 10 mM sodium acetate as the sole carbon source. Acetate-grown Tn5 mutants were subsequently plated onto ABM agar with 2 mM sodium benzoate as the carbon source. The plates were incubated for 2 days at room temperature. To confirm growth deficiencies on benzoate, Tn5 mutants were grown in ABM liquid medium supplemented with 2 mM benzoate. Growth was monitored for at least 55 hours. Tn5 mutants positive for growth on YTSS and acetate but negative for growth on benzoate were further characterized.

Location of Tn5 cassette and DNA sequencing of benzoate- mutants Arbitrary PCR was used to determine the location of the Tn5::KmR cassette in the mutants of interest following the methodology of O’Toole (O'Toole et al., 1999). Briefly, genomic DNA was extracted from mutants at mid-log phase using the DNeasy Blood and Tissue Kit, and 100 ng genomic DNA was used for PCR with 0.2 mM dNTPs (final concentration) and GoTaq® (Promega, USA) Reaction Buffer and DNA Polymerase. The random sequence primer ARB6 (5’ GGC CAC GCG TCG ACT AGT ACN NNN NNN NNN ACG CC 3’) and transposon specific

96 primers TNPR17OUT (5’AAC AAG CCA GGG ATG TAA CG 3’) or TNPR13OUT (5’ CAG CAA CAC CTT CTT CAC GA 3’) were used to amplify DNA upstream or downstream of the insertion, respectively. The initial PCR amplification included denaturation at 95°C for 5 min, 5 cycles of 30 sec denaturation at 94°C, 30 sec annealing at 30°C, and 1 min extension at 72°C, and 30 cycles of 30 sec denaturation at 94°C, 30 sec annealing at 45°C, and 1 min extension at 72°C, and a final extension at 72°C for 5 min. The PCR products from the first round of amplifications were used for a nested PCR with the primers ARB2 (5’ GGC CAC GCG TCG ACT AGT AC 3’) and TNPR17Nest or TNPR13Nest (5’ CTG ACA TGG GGG GGT ACC 3’) or TNPR13Nest (5’ CTA GAG TCG ACC TGC AGG CAT 3’). Nested PCR consisted of a 5 min denaturation step at 95°C, 30 cycles of 30 second denaturation at 94°C, 30 second annealing at 45°C, and 1 min extension at 72°C, and a final extension at 72°C for 10 min. Nested PCR products were sequenced at the University of Tennessee Genomics Core Facility. The transposons were located by manually mapping the homologous sequences to the S. stellata draft genome (Genbank accession: AAYA00000000.1).

Statistical analyses Statistical analyses of growth rates were performed using one-way analysis of variance in Excel 2016 (Microsoft® Office 2016). If p-value ≤0.05, Student’s t- tests were performed. For AFDM measurements, one-way analysis of variance was performed with Tukey’s post hoc tests in SigmaPlot 11.0.

IV. Results

Despite the prevalence of complex aromatic mixtures in nature, little is known of how individual microbes process components of these mixtures. In this study, we used S. stellata to explore how this environmentally relevant bacterium responds to binary mixtures of lignin-derived aromatics that are processed

97 through separate ring-cleaving pathways. In this strain, benzoate is the only known substrate to feed through the box pathway, but several substrates (ferulate, vanillate, p-coumarate, and p-hydroxybenzoate) feed through the pca pathway (Figure 4.1). Previous research indicates that S. stellata demonstrates a synergistic growth response to equimolar binary mixtures of benzoate and p- hydroxybenzoate, compared to growth on each substrate alone (Gulvik & Buchan, 2013). Our study sought to determine if that phenomenon is broadly observable with other aromatic compounds (i.e. ferulate, vanillate, or p- coumarate) destined for the pca pathway and benzoate.

Peripheral pathway aromatic compound catabolism genes are disparately located on the S. stellata genome Genes encoding for ferulate, p-coumarate, or vanillate catabolism are dispersed across the S. stellata genome (Figure 4.2). S. stellata possesses two homologs encoding feruloyl-CoA synthases, the enzymes responsible for transforming ferulate and p-coumarate (Frank and Chua et al., accepted). These are termed fcs1 and fcs2 and located in two different operons. fcs2 is adjacent to the box operon. In contrast, fcs1 is located on a different operon and the genes immediately surrounding it have not been characterized, though are predicted to be involved in HCA degradation in E-37. Vanillate monooxygenase is a heterodimer that converts vanillate to protocatechuate and is encoded by genes designated vanAB, which are traditionally co-localized in bacteria in which this enzyme has been characterized (Morawski et al., 2000; Segura et al.. S. stellata possesses homologs to vanAB, however, they are not co-localized. In fact, vanA is upstream of genes encoding tripartite ATP-independent periplasmic transporters, while vanB is downstream of ATP-binding cassette transporter subunits (Figure 4.2).

In contrast to the peripheral pathway genes described above, all of the box pathway genes are co-localized in a single operon on the S. stellata genome and

98 contain a putative transcriptional regulator, designated boxR (Figure 4.2). pobA encodes a p-hydroxybenzoate hydroxylase that converts p-hydroxybenzoate to protocatechuate. pobA is co-localized and co-transcribed with pcaHG, which include the protocatechuate 3,4-dioxygenase that is responsible for ring cleavage of protocatechuate. Regulation of this operon is mediated by a LysR-type transcriptional regulator termed pcaQ (Buchan et al., 2004) (Figure 4.2).

Substrate preference hierarchy in S. stellata Growth dynamics of S. stellata provided with equimolar binary mixtures of benzoate with vanillate, ferulate, or p-coumarate revealed substrate preferences (Figure 4.3). The growth rate and final cell yields on the binary mixture of benzoate and vanillate were at least 72.5% and 14.8% lower, respectively, than in single substrate cultures (Table 4.2 and Figure 4.4). Furthermore, the substrate concentration profiles revealed that benzoate was not removed from the medium until vanillate concentrations were nearly depleted (~0.001 mM) (Figure 4.3C). Obvious diauxic growth phases were not evident, but temporal resolution may not have been sufficient to capture this.

As vanillate has been demonstrated to be an intermediate of ferulate catabolism in other bacteria (Narbad & Gasson, 1998), S. stellata growth on binary mixtures of ferulate and benzoate was also tested. As expected, growth dynamics were similar to S. stellata growth on vanillate with benzoate. Benzoate was not removed from the medium until ferulate concentrations dropped to ~0.007 mM (~40 hours post inoculation) (Figure 4.3E), suggesting that ferulate is preferentially consumed over benzoate. Furthermore, the growth rate was at least 47% decreased in the binary mixture compared to single substrate ferulate or benzoate cultures (Table 4.2).

Given that S. stellata simultaneously catabolizes p-hydroxybenzoate and benzoate, and p-coumarate is predicted to generate p-hydroxybenzoate as an

99 intermediate, we hypothesized that p-coumarate would also be simultaneously catabolized with benzoate. However, the peak for benzoate was still detected when p-coumarate was depleted from the medium (20+ hours post inoculation) (Figure 4.3G). Therefore, along with ferulate and vanillate, p-coumarate is degraded before benzoate. The growth rate and final optical densities were unaffected in the cultures receiving binary mixtures relative to growth on 2 mM p- coumarate (Table 4.2, Figure 4.3).

S. stellata biomass yields vary amongst growth substrates Biomass yields for S. stellata cells were estimated in the binary mixtures and single substrate cultures to determine if there was a correlation between growth rates and total organic carbon yields. Biomass yields of S. stellata were highest in the p-coumarate culture (24.2 ± 0.8 mg), and lowest in the benzoate and vanillate cultures (17.5 ± 1.3 mg). S. stellata yields were statistically similar amongst the benzoate, vanillate, ferulate, and ferulate + benzoate cultures (Figure 4.4), demonstrating there is no clear relationship between growth rates and biomass yields. Carbon molar mass calculations indicate the following sequence: Benzoate (= 68.8% or 1.38 mM) > p-coumarate (= 66.2% or 1.32 mM) > ferulate (= 61.9% or 1.24 mM) > vanillate (= 57.1% or 1.14 mM). box mutants show altered growth phenotypes on ferulate and vanillate As the binary substrate experiments suggest cross-regulation between the box and vanillate and box and HCA pathways, we generated box operon mutants to probe this hypothesis further. A genetic screen of a 6,048-member Tn5 mutant library of S. stellata uncovered two mutants with a null phenotype when provided with benzoate as the sole carbon source. DNA sequencing localized the Tn5::KmR cassette to two regions within the box operon of S. stellata: the boxC gene and an intergenic region (Figure 4.5). BoxC is responsible for the third step of benzoate degradation in the box pathway and produces a semialdehyde and formic acid (Gescher et al., 2005). The second mutant has a Tn5 insertion

100 between two genes encoding for conserved hypothetical proteins which, in turn, are situated between the boxA and boxB genes. boxAB genes are cotranscribed and encode for two subunits of one enzyme that is responsible for the conversion of benzoyl-CoA to 2,3-epoxybenzoyl-CoA, the reactant for BoxC (Zaar et al., 2004).

Growth on benzoate was abolished in both of the box mutants (Figure 4.6B), which is unsurprising given that BoxABC have been demonstrated to be involved in benzoate degradation in Azoarcus evansii (Gescher et al., 2005; Zaar et al., 2004), and S. stellata shares 54% (E-value = 6e-146), 59% (E-value = 0), 54% (E-value = 0) identities with their respective protein sequences in A. evansii. Growth was unaffected in complex medium and in defined media with p- hydroxybenzoate or p-coumarate as the sole carbon. However, growth of the box mutants was 14.8% (box_6,531::Tn5) and 16.7% (boxC::Tn5) and 13.7% (box_6,531::Tn5) and 15.8% (boxC::Tn5) lower on vanillate or ferulate, respectively (Figure 4.6).

Gene expression data support the role of the box operon in benzoate degradation The expression of genes encoding for key enzymes in the box pathway was differentially affected in the two box mutants, despite the box genes sharing the same promoter (Dr. Elizabeth Fozo, personal communication) and being cotranscribed (Figure 3.10). Relative to wildtype, when grown on 1 mM p- hydroxybenzoate and 1 mM benzoate, boxA expression in box_6,531::Tn5 was diminished by 99.5%, but boxB (44% decrease) and boxL (27% decrease) expression were less affected by the Tn5 insertion, suggesting the conserved hypothetical protein is important for boxA function on p-hydroxybenzoate and benzoate. However, the Tn5 insertion in the intergenic region of the two conserved hypothetcal genes appears to also affect box gene expression on acetate, as the expression of each gene was slightly downregulated (boxA: 32%

101 decrease, boxB: 30% decrease, boxL: 49% decrease). This may explain why the growth rate of box_6,531::Tn5 was impaired on 7 mM acetate (Figure 4.6A). boxA (99% decrease) and boxB (99.1% decrease) expression were significantly downregulated in boxC::Tn5, but boxL expression (1% decrease) was unaffected in p-hydroxybenzoate and benzoate cultures. This may be unsurprising as BoxC accelerates BoxAB rates in Azoarcus evansii (Gescher et al., 2005). boxC and box_6,531::Tn5 box gene expression profiles may also explain why both mutants displayed similar growth dynamics on ferulate or vanillate.

V. Discussion

Plant-derived aromatic compounds are typically found in heterogenous mixtures in nature (Seo et al., 2009). A previous study demonstrated S. stellata simultaneously catabolizes benzoate and p-hydroxybenzoate, but this study revealed a strong substrate preference for HCAs and vanillate relative to benzoate. Collectively, these results suggest S. stellata has developed separate and distinct mechanisms for catabolic regulation of structurally related aromatic compounds. Contrary to the results of this study, when other largely soil bacteria were given a mixture of carbon substrates, benzoate (with catechol as the intermediate) was usually the preferred substrate (Ampe & Lindley, 1995; Choi et al., 2007; Mazzoli et al., 2007; Nichols & Harwood, 1995; D. Perez-Pantoja et al., 2015; Zhan et al., 2009). These studies demonstrated repression by benzoate by either inhibiting the transcription of catabolic enzymes, transporters, or transcriptional regulators (Ampe & Lindley, 1995; Choi et al., 2007; Nichols & Harwood, 1995; D. Perez-Pantoja et al., 2015; Zhan et al., 2009). Ours is the first study where substrate preference hierarchy has been demonstrated in a bacterium with the box pathway. It is unlikely that HCAs and vanillate are used before benzoate solely due to presence of the box pathway as demonstrated by Gulvik and Buchan (2013), wherein the box pathway-containing Variovorax paradoxus EPS was shown to preferentially degrade p-hydroxybenzoate over 102 benzoate (with catechol as an intermediate). Another unexpected result was the lower biomass yields on 2 mM benzoate relative to growth on other substrate concentrations. Carbon molar mass calculations suggest growth on 2 mM benzoate would produce the most biomass, but the lower than expected biomass could be explained by the inherent toxicity of benzoate (Bosund, 1963). Furthermore, previous studies indicate that the addition of benzoate in bacterial cultures resulted in either unaffected biomass yields because benzoate needed to be quickly degraded for detoxification, or even reduced biomass yields because soluble, unused byproducts were produced (D. Perez-Pantoja et al., 2015; Rudolph & Grady, 2001), suggesting that the reason for benzoate repression is unrelated to biomass yields. Given that of all the mixed substrate cultures, only the optical density of the vanillate and benzoate culture was significantly lower than their respective single substrate cultures, it is unsurprising that the lowest biomass yields were observed in the vanillate and benzoate culture. Additionally, of the individual substrates, vanillate provides the lowest carbon/molar mass value.

In terrestrial systems, fungi are the primary mediators of lignocellulose degradation to water-soluble phenolic compounds, which can be used as energy and carbon sources by fungi and bacteria (Boer et al.. While there is evidence for initial decomposition of S. alterniflora by fungi (Newell et al., 1996) in salt marsh ecosystems, studies have shown that bacteria are largely responsible for degrading the lignin and polysaccharide components of lignocellulose (Benner et al., 1986; Benner et al., 1984; Gallagher et al., 1976). Trends from S. alterniflora composition studies suggest that there is an evolutionary advantage to consuming HCAs such as ferulate and p-coumarate over benzoate. For example, cinnamyl phenols are removed at faster rates than vanillyl, p-hydroxyl, and syringyl phenols from S. alterniflora decomposition (Benner et al., 1991). This pattern was also demonstrated in two different degradation studies involving S. alterniflora, where ferulic acid accounted for 57% and 82% of the total lignin lost 103

(Haddad et al., 1992). In fact, p-coumaric and ferulic acids are bound to lignin by ester linkages (Higuchi et al., 1967; Smith, 1955), which are reactive and susceptible to cleavage.

Vanillate, ferulate, and p-coumarate are abundant in lignocellulose (Benner et al., 1984; Bergbauer & Newell, 1992; Goni & Thomas, 2000; Haddad et al., 1992; Hedges & Parker, 1976). For example, in a study involving lignocellulose incubation with Phaeosphaeria spartinicola, a fungus that is associated with decaying S. alterniflora leaves (Buchan et al., 2003; Y. Newell et al., 1989), there were still higher contents of cinnamyl phenols and vanillic acid compared to p- hydroxyl phenols in the lignocellulose after incubation, a factor that could contribute to the benefit of S. stellata simultaneously catabolizing p- hydroxybenzoate and benzoate. Among the decomposition studies where benzoic acid concentrations were measured, benzoic acid contributed only up to 1.3% of lignocellulose, while the amounts of vanillic, ferulic, and p-coumaric acids were up to 30% (Goni & Thomas, 2000; Hedges & Parker, 1976), hinting at an adaptive advantage to microbes using these substrates over benzoate.

Substrate preference hierarchy between benzoate and vanillate or HCAs has not extensively been studied in other bacteria. As far as we know, the only study involving cross-regulation between these pathways was with Acinetobacter baylyi. The hca (p-coumarate) and van (vanillate) operons were shown to be repressed in the presence of benzoate (Bleichrodt et al., 2010). Our results, however, suggest that benzoate degradation is repressed in the presence of vanillate or the HCAs.

Genes for enzymes involved in aromatic degradation in Rhodococcus sp. DK17, Acinetobacter baylyi ADP1, and Acinetobacter pittii PHEA-2 are in close proximity to each other and a genetic regulator (Choi et al., 2005; Collier et al., 1998; Patrauchan et al., 2005; Danilo; Perez-Pantoja et al., 2008; Xu et al., 104

2003). Conversely, many of the genes for lignin-derived aromatic catabolism in S. stellata are disparately located in the genome (Figure 4.2). While the box genes are cotranscribed (Frank and Chua et al., accepted), pobA is cotranscribed with pcaCHG genes and a highly conserved ORF (Buchan et al., 2004), but pcaBD (SSE37_21460, SSE37_21470) are located in another operon. Similar to S. stellata, the genes for p-hydroxybenzoate degradation in Cupriavidus pinatubonensis JMP134 are not located in the same operon (Danilo; Perez-Pantoja et al., 2008), which is in contrast to the other bacteria that preferentially utilize benzoate over other aromatic compounds and suggests that gene organization may not readily reveal substrate preference hierarchies. In S. stellata, Fcs1 and Fcs2 both act on ferulate and p-coumarate, with genes encoding for ferulate and p-coumarate degradation enzymes found adjacent to fcs1 or fcs2. Genes in the vanillate degradation pathway are currently undetermined in S. stellata. In other bacteria, vanAB encode for vanillate monooxygenase, which demethylates vanillate to form protocatechuate, and are colocalized in an operon (Merkens et al., 2005; Segura et al., 1999; Venturi et al., 1998). However, probable S. stellata vanAB genes are found on different operons (Figure 4.2), which is a common feature in other Roseobacter genomes and belies the role of these genes in Roseobacter vanillate degradation. Given the gene synteny among Roseobacter strains, it is possible that other Roseobacters also utilize vanillate or HCAs before benzoate in a mixed substrate culture.

A boxR homolog (SSE37_24449) is present in the box operon, but it is also unlikely that this regulator is inhibited by vanillate or HCAs. SSE37_24449 shares 44% (E value = 3e-68) and 43% (E value = 2e-71) identities with the BoxR (Ga0098266_114713) and BzdR (Ga0098266_111630) proteins of Azoarcus sp. CIB, respectively. The C-terminal domain of the Azoarcus sp. BzdR protein has been proposed to interact with benzoyl-CoA through a consensus nucleotide- binding site containing a Walker A motif (Barragan et al., 2005). Furthermore, 105 the benzoyl-CoA analogue phenylacetyl-CoA was unable to alleviate the repression of BoxR or BzdR in Azoarcus sp. CIB (Barragan et al., 2005; Valderrama et al., 2012), suggesting that only intermediates of the bzd (anaerobic benzoate degradation) or box pathways can act as inducers. It is unlikely that the HCA pathway intermediates would act as inducers for BoxR in S. stellata as the intermediates are not structurally similar to benzoyl-CoA. The Walker A motif (GLRGAGK(T/S)) was found from residue 137-144 of S. stellata BoxR, suggesting that this transcriptional regulator has a similar mechanism to that described for BoxR and BzdR in Azoarcus.

The catabolite repression control (Crc) protein is an alternative mechanism for catabolite repression, whererin Crc binds an A-rich motif surrounding the ribosome binding site, thus inhibiting translation (Moreno et al., 2009). There are two Crc protein encoding genes in S. stellata, SSE37_04605 or SSE37_11639, with 26% (E-value = 8e-16) and 26% (E-value = 7e-14) identities to the Crc protein encoding gene (Ga0097782_182211) of Pseudomonas aeruginosa PAO1, a model organism for Crc catabolite repression (Sonnleitner et al.. Both S. stellata crc genes contain the conserved catalytic residues identified by (Ruiz- Manzano, Yuste, & Rojo, 2005) within SSE37_04605 (Asp221 and His251) and SSE37_11639 (Asp223 and His253). However, there does not appear to be a characteristic A-rich motif in the box operon, suggesting that the box operon is not repressed post-transcriptionally in the presence of vanillate or either HCA.

A gene encoding for a transporter was not found in the S. stellata box operon. However, a cluster of genes encoding for a tripartite ATP-independent periplasmic (TRAP) transporter is located upstream of the box operon and encoded by dctPQM. dctP (SSE37_24379) encodes the periplasmic-binding subunit, and a prior mutagenesis study indicates DctP plays a role in ferulate and p-coumarate uptake, but does not act on benzoate (Frank and Chua et al., accepted). Despite the lack of physiological evidence for benzoate transport, 106 dctP is upregulated when S. stellata is grown on benzoate (93-fold (±13)), ferulate (13-fold (±2)), or p-coumarate (23-fold (±14)), suggesting transport could play a role in substrate preference hierarchy in S. stellata. When S. stellata is grown on benzoate and either HCA, the dctPQM genes would be upregulated, but only ferulate or p-coumarate would be transported, so S. stellata would consume ferulate or p-coumarate before benzoate. While dctP expression was not measured in benzoate and HCA cultures, dctP upregulation was observed in benzoate and vanillate cultures (1.8x increase when vanillate was being used, and 6x increase during benzoate utilization).

While further studies are required to understand the underlying regulatory network(s) dictating the substrate preference hierarchies identified in this study, we now know that S. stellata appears to have developed several mechanisms for adapting to the heterogenous mixture of aromatic compounds representative of salt march ecosystems. Results from this study demonstrate that S. stellata preferentially utilizes aromatics that represent the most abundant and easily accessible fractions within lignocellulose. This study provides further evidence for the metabolic versatility of S. stellata and may provide a broader basis for understanding how salt marsh microbes transform lignin-derived aromatics in mixtures representative of natural environments.

107

VI. Acknowledgements

This research was supported by grants from the Department of Energy and National Science Foundation [OCE-1357242].

We would also like to thank Ernesto Zuleta from Dr. Joseph Bozell’s for assisting in HPLC training and Dr. Robbie Martin and Mary Hadden for generating the S. stellata Tn5 mutant library.

108

VII. References

Ampe, F., & Lindley, N. D. (1995). Acetate utilization is inhibited by benzoate in Alcaligenes eutrophus: evidence for transcriptional control of the expression of acoE coding for acetyl coenzyme A synthetase.. Journal of bacteriology, 177(20), 5826-5833. Barragan, M. J. L., Blazquez, B., Zamarro, M. T., Mancheno, J. M., Garcia, J. L., Diaz, E., & Carmona, M. (2005). BzdR, a repressor that controls the anaerobic catabolism of benzoate in Azoarcus sp CIB, is the first member of a new subfamily of transcriptional regulators. Journal of Biological Chemistry, 280(11), 10683-10694. doi:10.1074/jbc.M412259200 Benner, R., Fogel, M. L., & Sprague, E. K. (1991). Diagenesis of belowground biomass of Spartina alterniflora in salt‐marsh sediments. Limnology and Oceanography, 36(7), 1358-1374. Benner, R., Moran, M. A., & Hodson, R. E. (1986). Biogeochemical cycling of lignocellulosic carbon in marine and freshwater ecosystems: Relative contributions of procaryotes and eucaryotes. Limnology and Oceanography, 31(1), 89-100. doi:DOI 10.4319/lo.1986.31.1.0089 Benner, R., Newell, S. Y., Maccubbin, A. E., & Hodson, R. E. (1984). Relative contributions of bacteria and fungi to rates of degradation of lignocellulosic detritus in salt-marsh sediments. Appl Environ Microbiol, 48(1), 36-40. Bergbauer, M., & Newell, S. Y. (1992). Contribution to lignocellulose degradation and DOC formation from a salt marsh macrophyte by the ascomycete Phaeosphaeria spartinicola. Fems Microbiology Ecology, 86(4), 341-347. Bleichrodt, F. S., Fischer, R., & Gerischer, U. C. (2010). The beta-ketoadipate pathway of Acinetobacter baylyi undergoes carbon catabolite repression, cross-regulation and vertical regulation, and is affected by Crc. Microbiology, 156(Pt 5), 1313-1322. doi:10.1099/mic.0.037424-0 Boer, W., Folman, L. B., Summerbell, R. C., & Boddy, L. (2005). Living in a fungal world: impact of fungi on soil bacterial niche development. FEMS Microbiol Rev, 29(4), 795-811. doi:10.1016/j.femsre.2004.11.005 Bosund, I. (1963). The Action of Benzoic and Salicylic Acids on the Metabolism of Microorganisms (Vol. 11). Brinkhoff, T., Giebel, H. A., & Simon, M. (2008). Diversity, ecology, and genomics of the Roseobacter clade: a short overview. Archives of Microbiology, 189(6), 531-539. doi:10.1007/s00203-008-0353-y Brzostowicz, P. C., Reams, A. B., Clark, T. J., & Neidle, E. L. (2003). Transcriptional cross-regulation of the catechol and protocatechuate branches of the beta-ketoadipate pathway contributes to carbon source- dependent expression of the Acinetobacter sp. strain ADP1 pobA gene. Appl Environ Microbiol, 69(3), 1598-1606. Buchan, A., Gonzalez, J. M., & Moran, M. A. (2005). Overview of the marine roseobacter lineage. Appl Environ Microbiol, 71(10), 5665-5677. doi:10.1128/AEM.71.10.5665-5677.2005 109

Buchan, A., Neidle, E. L., & Moran, M. A. (2001). Diversity of the ring-cleaving dioxygenase gene pcaH in a salt marsh bacterial community. Appl Environ Microbiol, 67(12), 5801-5809. doi:10.1128/AEM.67.12.5801-5809.2001 Buchan, A., Neidle, E. L., & Moran, M. A. (2004). Diverse organization of genes of the beta-ketoadipate pathway in members of the marine Roseobacter lineage. Appl Environ Microbiol, 70(3), 1658-1668. Buchan, A., Newell, S. Y., Butler, M., Biers, E. J., Hollibaugh, J. T., & Moran, M. A. (2003). Dynamics of bacterial and fungal communities on decaying salt marsh grass. Appl Environ Microbiol, 69(11), 6676-6687. Choi, K. Y., Kim, D., Sul, W. J., Chae, J.-C., Zylstra, G. J., Kim, Y. M., & Kim, E. (2005). Molecular and biochemical analysis of phthalate and terephthalate degradation by Rhodococcus sp. strain DK17. FEMS microbiology letters, 252(2), 207–213. Choi, K. Y., Zylstra, G. J., & Kim, E. (2007). Benzoate catabolite repression of the phthalate degradation pathway in Rhodococcus sp strain DK17. Applied and Environmental Microbiology, 73(4), 1370-1374. Collier, L. S., Gaines, G. L., & Neidle, E. L. (1998). Regulation of benzoate degradation in Acinetobacter sp. strain ADP1 by BenM, a LysR-type transcriptional activator. Journal of bacteriology, 180(9), 2493-2501. Cragg, S. M., Beckham, G. T., Bruce, N. C., Bugg, T. D., Distel, D. L., Dupree, P., et al. (2015). Lignocellulose degradation mechanisms across the Tree of Life. Curr Opin Chem Biol, 29, 108-119. doi:10.1016/j.cbpa.2015.10.018 Cude, W. N., Mooney, J., Tavanaei, A. A., Hadden, M. K., Frank, A. M., Gulvik, C. A., et al. (2012). The production of the antimicrobial secondary metabolite indigoidine contributes to competitive surface colonization in the marine roseobacter Phaeobacter sp. strain Y4I. Applied and Environmental Microbiology, 78(14), 4771-4780. doi:10.1128/AEM.00297- 12 Donoso, R. A., Perez-Pantoja, D., & Gonzalez, B. (2011). Strict and direct transcriptional repression of the pobA gene by benzoate avoids 4- hydroxybenzoate degradation in the pollutant degrader bacterium Cupriavidus necator JMP134. Environmental Microbiology, 13(6), 1590- 1600. doi:10.1111/j.1462-2920.2011.02470.x Fuchs, G., Boll, M., & Heider, J. (2011). Microbial degradation of aromatic compounds - from one strategy to four. Nat Rev Microbiol, 9(11), 803-816. doi:10.1038/nrmicro2652 Gaines, G. L., 3rd, Smith, L., & Neidle, E. L. (1996). Novel nuclear magnetic resonance spectroscopy methods demonstrate preferential carbon source utilization by Acinetobacter calcoaceticus. J Bacteriol, 178(23), 6833- 6841. Gallagher, J. L., Pfeiffer, W. J., & Pomeroy, L. R. (1976). Leaching and microbial utilization of dissolved organic carbon from leaves of Spartina alterniflora. Estuarine and Coastal Marine Science, 4(4), 467-471. doi:Doi 10.1016/0302-3524(76)90021-9

110

Gescher, J., Eisenreich, W., Worth, J., Bacher, A., & Fuchs, G. (2005). Aerobic benzoyl-CoA catabolic pathway in Azoarcus evansii: studies on the non- oxygenolytic ring cleavage enzyme. Mol Microbiol, 56(6), 1586-1600. doi:10.1111/j.1365-2958.2005.04637.x Gescher, J., Zaar, A., Mohamed, M., Schagger, H., & Fuchs, G. (2002). Genes coding for a new pathway of aerobic benzoate metabolism in Azoarcus evansii. J Bacteriol, 184(22), 6301-6315. Goni, M. A., & Thomas, K. A. (2000). Sources and transformations of organic matter in surface soils and sediments from a tidal estuary (north inlet, South Carolina, USA). Estuaries, 23(4), 548-564. Gonzalez, J. M., Mayer, F., Moran, M. A., Hodson, R. E., & Whitman, W. B. (1997). Sagittula stellata gen. nov., sp. nov., a lignin-transforming bacterium from a coastal environment. Int J Syst Bacteriol, 47(3), 773-780. doi:10.1099/00207713-47-3-773 Gorke, B., & Stulke, J. (2008). Carbon catabolite repression in bacteria: many ways to make the most out of nutrients. Nat Rev Microbiol, 6(8), 613-624. doi:10.1038/nrmicro1932 Gulvik, C. A., & Buchan, A. (2013). Simultaneous catabolism of plant-derived aromatic compounds results in enhanced growth for members of the Roseobacter lineage. Appl Environ Microbiol, 79(12), 3716-3723. doi:10.1128/AEM.00405-13 Haddad, R. I., Newell, S. Y., Martens, C. S., & Fallon, R. D. (1992). Early diagenesis of lignin-associated phenolics in the salt marsh grass Spartina alterniflora. Geochimica Et Cosmochimica Acta, 56(10), 3751-3764. Harwood, C. S., & Parales, R. E. (1996). The beta-ketoadipate pathway and the biology of self-identity. Annu Rev Microbiol, 50, 553-590. doi:10.1146/annurev.micro.50.1.553 Hatfield, R. D., Rancour, D. M., & Marita, J. M. (2016). Grass Cell Walls: A Story of Cross-Linking. Front Plant Sci, 7, 2056. doi:10.3389/fpls.2016.02056 Hedges, J. I., Keil, R. G., & Benner, R. (1997). What happens to terrestrial organic matter in the ocean? Organic Geochemistry, 27(5), 195-212. doi:https://doi.org/10.1016/S0146-6380(97)00066-1 Hedges, J. I., & Parker, P. L. (1976). Land-Derived Organic-Matter in Surface Sediments from Gulf of Mexico. Geochimica Et Cosmochimica Acta, 40(9), 1019-1029. Higuchi, T., Ito, Y., Shimada, M., & Kawamura, I. (1967). Chemical properties of milled wood lignin of grasses. Phytochemistry, 6(11), 1551-&. Hladik, C., Schalles, J., & Alber, M. (2013). Salt marsh elevation and habitat mapping using hyperspectral and LIDAR data. Remote Sensing of Environment, 139, 318-330. doi:https://doi.org/10.1016/j.rse.2013.08.003 Hodson, R. E., Christian, R. R., & Maccubbin, A. E. (1984). Lignocellulose and lignin in the salt marsh grass Spartina alterniflora: initial concentrations and short-term, post-depositional changes in detrital matter. Marine Biology, 81(1), 1-7.

111

Larsen, R. A., Wilson, M. M., Guss, A. M., & Metcalf, W. W. (2002). Genetic analysis of pigment biosynthesis in Xanthobacter autotrophicus Py2 using a new, highly efficient transposon mutagenesis system that is functional in a wide variety of bacteria. Archives of Microbiology, 178(3), 193-201. doi:10.1007/s00203-002-0442-2 Leisola, M., Pastinen, O., & Axe, D. D. (2012). Lignin--designed randomness. Bio-complexity, 2012. Luo, H., & Moran, M. A. (2014). Evolutionary ecology of the marine Roseobacter clade. Microbiol Mol Biol Rev, 78(4), 573-587. doi:10.1128/MMBR.00020- 14 Mazzoli, R., Pessione, E., Giuffrida, M. G., Fattori, P., Barello, C., Giunta, C., & Lindley, N. D. (2007). Degradation of aromatic compounds by Acinetobacter radioresistens S13: growth characteristics on single substrates and mixtures. Archives of Microbiology, 188(1), 55-68. Merkens, H., Beckers, G., Wirtz, A., & Burkovski, A. (2005). Vanillate metabolism in Corynebacterium glutamicum. Current Microbiology, 51(1), 59-65. doi:10.1007/s00284-005-4531-8 Morales, G., Linares, J. F., Beloso, A., Albar, J. P., Martínez, J. L., & Rojo, F. (2004). The Pseudomonas putida Crc global regulator controls the expression of genes from several chromosomal catabolic pathways for aromatic compounds. Journal of bacteriology, 186(5), 1337. Moran, M. A., & Hodson, R. E. (1990). Contributions of degrading Spartina alterniflora lignocellulose to the dissolved organic carbon pool of a salt marsh. Marine Ecology Progress Series, 62(1-2), 161-168. Moran, M. A., & Hodson, R. E. (1994). Dissolved humic substances of vascular plant origin in a coastal marine environment. Limnology and Oceanography, 39(4), 762-771. Morawski, B., Segura, A., & Ornston, L. N. (2000). Repression of Acinetobacter vanillate demethylase synthesis by VanR, a member of the GntR family of transcriptional regulators. FEMS microbiology letters, 187(1), 65-68. doi:10.1111/j.1574-6968.2000.tb09138.x Moreno, R., Marzi, S., Romby, P., & Rojo, F. (2009). The Crc global regulator binds to an unpaired A-rich motif at the Pseudomonas putida alkS mRNA coding sequence and inhibits translation initiation. Nucleic Acids Res, 37(22), 7678-7690. doi:10.1093/nar/gkp825 Narbad, A., & Gasson, M. J. (1998). Metabolism of ferulic acid via vanillin using a novel CoA-dependent pathway in a newly-isolated strain of Pseudomonas fluorescens. Microbiology-Uk, 144, 1397-1405. doi:Doi 10.1099/00221287-144-5-1397 Newell, S. Y., Porter, D., & Lingle, W. L. (1996). Lignocellulolysis by ascomycetes (fungi) of a saltmarsh grass (smooth cordgrass). Microsc Res Tech, 33(1), 32-46. doi:10.1002/(SICI)1097- 0029(199601)33:1<32::AID-JEMT5>3.0.CO;2-2

112

Nichols, N. N., & Harwood, C. S. (1995). Repression of 4-hydroxybenzoate transport and degradation by benzoate: a new layer of regulatory control in the Pseudomonas putida beta-ketoadipate pathway. Journal of bacteriology, 177(24), 7033-7040. O'Toole, G. A., Pratt, L. A., Watnick, P. I., Newman, D. K., Weaver, V. B., & Kolter, R. (1999). [6] Genetic approaches to study of biofilms. In Methods in Enzymology (Vol. 310, pp. 91-109): Academic Press. Opsahl, S., & Benner, R. (1995). Early diagenesis of vascular plant tissues: Lignin and cutin decomposition and biogeochemical implications. Geochimica Et Cosmochimica Acta, 59(23), 4889-4904. doi:https://doi.org/10.1016/0016-7037(95)00348-7 Ornston, L. N., & Stanier, R. Y. (1966). The conversion of catechol and protocatechuate to beta-ketoadipate by Pseudomonas putida. J Biol Chem, 241(16), 3776-3786. Patrauchan, M. A., Florizone, C., Dosanjh, M., Mohn, W. W., Davies, J., & Eltis, L. D. (2005). Catabolism of benzoate and phthalate in Rhodococcus sp. strain RHA1: redundancies and convergence. Journal of bacteriology, 187(12), 4050-4063. Perez-Pantoja, D., De la Iglesia, R., Pieper, D. H., & González, B. (2008). Metabolic reconstruction of aromatic compounds degradation from the genome of the amazing pollutant-degrading bacterium Cupriavidus necator JMP134. FEMS microbiology reviews, 32(5), 736-794. Perez-Pantoja, D., Leiva-Novoa, P., Donoso, R. A., Little, C., Godoy, M., Pieper, D. H., & Gonzalez, B. (2015). Hierarchy of carbon source utilization in soil bacteria: Hegemonic preference for benzoate in complex aromatic compound mixtures degraded by Cupriavidus pinatubonensis JMP134. Applied and Environmental Microbiology, 81(12), 3914-3924. Rudolph, J. M., & Grady, C. P. L. (2001). Effect of media composition on yield values of bacteria growing on binary and ternary substrate mixtures in continuous culture. Biotechnology and Bioengineering, 74(5), 396-405. doi:DOI 10.1002/bit.1130 Ruiz-Manzano, A., Yuste, L., & Rojo, F. (2005). Levels and activity of the Pseudomonas putida global regulatory protein Crc vary according to growth conditions. J Bacteriol, 187(11), 3678-3686. doi:10.1128/JB.187.11.3678-3686.2005 Segura, A., Bunz, P. V., D'Argenio, D. A., & Ornston, L. N. (1999). Genetic analysis of a chromosomal region containing vanA and vanB, genes required for conversion of either ferulate or vanillate to protocatechuate in Acinetobacter. J Bacteriol, 181(11), 3494-3504. Segura, A., Bünz, P. V., D’Argenio, D. A., & Ornston, L. N. (1999). Genetic Genetic analysis of a chromosomal region containing vanA and vanB, genes required for conversion of either ferulate or vanillate to protocatechuate in Acinetobacter Journal of bacteriology, 181(11), 3494- 3504.

113

Seo, J. S., Keum, Y. S., & Li, Q. X. (2009). Bacterial degradation of aromatic compounds. Int J Environ Res Public Health, 6(1), 278-309. doi:10.3390/ijerph6010278 Smith, D. C. C. (1955). Ester groups in lignin. Nature, 176(4475), 267-268. Sonnleitner, E., Abdou, L., & Haas, D. (2009). Small RNA as global regulator of carbon catabolite repression in Pseudomonas aeruginosa. Proceedings of the National Academy of Sciences, 106(51), 21866-21871. doi:10.1073/pnas.0910308106 Sudtachat, N., Ito, N., Itakura, M., Masuda, S., Eda, S., Mitsui, H., et al. (2009). Aerobic vanillate degradation and compound metabolism in Bradyrhizobium japonicum. Applied and Environmental Microbiology, 75(15), 5012. Valderrama, J. A., Durante-Rodriguez, G., Blazquez, B., Garcia, J. L., Carmona, M., & Diaz, E. (2012). Bacterial degradation of benzoate: cross-regulation between aerobic and anaerobic pathways. J Biol Chem, 287(13), 10494- 10508. doi:10.1074/jbc.M111.309005 Venturi, V., Zennaro, F., Degrassi, G., Okeke, B. C., & Bruschi, C. V. (1998). Genetics of ferulic acid bioconversion to protocatechuic acid in plant- growth-promoting Pseudomonas putida WCS358. Microbiology, 144 ( Pt 4), 965-973. doi:10.1099/00221287-144-4-965 Xu, Y., Chen, M., Zhang, W., & Lin, M. (2003). Genetic organization of genes encoding phenol hydroxylase, benzoate 1, 2-dioxygenase alpha subunit and its regulatory proteins in Acinetobacter calcoaceticus PHEA-2. Current Microbiology, 46(4), 0235-0240. Y. Newell, S., Fallon, R., & Miller, J. (1989). Decomposition and microbial dynamics for standing, naturally positioned leaves of the salt-marsh grass Spartina alterniflora (Vol. 101). Zaar, A., Gescher, J., Eisenreich, W., Bacher, A., & Fuchs, G. (2004). New enzymes involved in aerobic benzoate metabolism in Azoarcus evansii. Mol Microbiol, 54(1), 223-238. doi:10.1111/j.1365-2958.2004.04263.x Zhan, Y. H., Yu, H. Y., Yan, Y. L., Ping, S. Z., Lu, W., Zhang, W., et al. (2009). Benzoate catabolite repression of the phenol degradation in Acinetobacter calcoaceticus PHEA-2. Current Microbiology, 59(4), 368-373.

114

VIII. Appendix

Table 4.1. Strains used in this study. Strains Description Source E-37 Wildtype Sagitulla stellata E-37 (Gonzalez, J. M. et al., 1997) E-37 boxC::Tn5 E-37 boxC mutant wherein a single This study Tn5 cassette was inserted into boxC (SSE37_24419) E-37 box_6,531::Tn5 E-37 box_6,531 mutant where a This study single Tn5 cassette was inserted in the intergenic region between chp (SSE37_24429) and hp (SSE37_24434) Escherichia coli EA145 Mating strain used for generation of (Larsen et al., 2002) E-37 Tn5 random transposon mutant library

115

Table 4.2. Growth rates of Sagittula stellata E-37 on single or binary substrates. Ben: 2 mM benzoate, Van: 2 mM vanillate, Van + Ben: 1 mM vanillate + 1 mM benzoate, Fer: 2 mM ferulate, Fer + Ben: 1 mM ferulate + 1 mM benzoate, Coum: 1 mM p-coumarate, Coum + Ben: 1 mM p-coumarate + 1 mM benzoate. Averages and standard deviations are from three biological replicates. Different letters indicate statistical significance from each substrate culture (p<0.05).

Ben Van Van + Ben Fer Fer + Ben p-Coum p-Coum + Ben Growth 0.120 ± 0.178 ± 0.033 ± 0.083 ± 0.044 0.170 ± 0.191 rates 0.004a 0.005b 0.003c 0.001d ± 0.0005e 0.008b ± 0.016b

116

Figure 4.1. Lignin-derived aromatic compounds that are catabolized through protocatechuate or benzoate in Sagittula stellata E-37. This figure is adapted from (Gulvik & Buchan, 2013).

117

Figure 4.2. Organization of characterized and predicted aromatic compound catabolism genes in Sagittula stellata. Locus tags for S. stellata are located above or below genes predicted or determined to be involved in aromatic compound degradation. Gene clusters are organized by substrate as listed to the left of clusters. dctPQM genes: TRAP transporter subunits, sdr: short-chain dehydrogenase, fcs: feruloyl-CoA synthase, gene 2: lactonase (SSE37_24409), grey genes: hypothetical protein, gene 9: thioesterase (SSE37_24444), black genes: transcriptional regulator, vanA: vanillate O-demethylase monooxygenase subunit, tPort1234: ABC transporter subunits, vanB: vanillate O-demethylase ferredoxin subunit, ech: enoyl-CoA hydratase, pobA: 4- hydroxybenzoate 3-monooxygenase, pcaQCHG: genes involved in protocatechuate degradation. Percentage homology to Pseudomonas aeruginosa PAO1 gene homologs and associated locus tags are listed above or below the proposed vanillate genes.

118

Figure 4.3. Growth profiles and extracellular concentrations of (A) benzoate (2 mM), (B) vanillate (2 mM), (C) benzoate (1 mM) + vanillate (1 mM), (D) ferulate (2 mM), (E) benzoate (1 mM) + ferulate (1 mM), (F) p-coumarate (2 mM), and (G) benzoate (1 mM) + p-coumarate (1 mM) by Sagittula stellata.

Solid lines indicate growth of S. stellata as measured by OD540. Dashed lines represent the concentrations of aromatic compounds in cell-free aliquots. For panels C, E, G: blue lines = benzoate. Ben = benzoate, Van = vanillate, Fer = ferulate, and Coum = p-coumarate. Open circles = single substrate, closed circles = binary mixtures. Averages and standard deviations are from three biological replicates.

119

Figure 4.4. AFDM yields of Sagittula stellata (100 mL) grown on single or binary mixtures at 2 mM. p-Coum = p-coumarate, Fer = ferulate, Ben = benzoate, and Van = vanillate. Averages and standard deviations are from three biological replicates. Different letters indicate statistical significance, and same letters indicate results are statistically indistinguishable (p<0.05).

120

Figure 4.5. Location of insertion sites of Tn5::KmR cassettes in the box operon of two Sagittula stellata mutants. boxL: benzoate-CoA ligase (WP_005857144.1). 2: lactonase (WP_005857146.1). 3: conserved hypothetical protein (WP_005857148.1). boxC: benzoyl-CoA dihydrodiol lyase (WP_005857150.1). boxB: benzoyl-CoA oxygenase (WP_005857152.1). 4: conserved hypothetical protein (WP_005857154.1). 5: hypothetical protein (EBA09447.1). boxA: benzoyl- CoA reductase (EBA09448.1). 6: thioesterase (WP_005857160.1). boxR: transcriptional regulator (WP_005857163.1). Annotations were determined from BLASTp alignments against the Azoarcus evansii genome. Arrows indicate open reading frames. Numbers beneath each gene indicate the base pair length of their respective genes. Triangles represent transposon Tn5 in two independent mutants.

121

Figure 4.6. Growth curves of S. stellata and both box mutants on acetate (7 mM) and various lignin-derived aromatic compounds at 2 mM: (A) acetate, (B) benzoate, (C) p- hydroxybenzoate, (D) p-coumarate, (E) vanillate, and (F) ferulate. Growth curve and letter colors represent S. stellata strain type. Black: Sagittula stellata. Red: boxC::Tn5. Orange: box_6,531::Tn5. Letters indicate significant differences (p<0.05) between the growth rates of the exponential phase of each growth curve. F: Higher optical densities were observed due to maroon color production by S. stellata on ferulate. Optical densities for both box mutants were indistinguishable in panels (E) and (F) and therefore overlapped. Averages and standard deviations are from a minimum of three biological replicates.

122

CHAPTER V: CONCLUSION Bacteria in the marine Roseobacter clade are biogeochemically important and largely heterotrophic (Buchan et al., 2005; Luo & Moran, 2014; Wagner-Dobler & Biebl, 2006). This dissertation explored the heterotrophic pathways Roseobacter use to aerobically degrade aromatic compounds derived from the most abundant aromatic polymer on Earth, lignin. The first focus of the dissertation was to determine the evolutionary history of ring-cleaving aromatic degradation pathways in over 300 Roseobacter genomes. The remaining chapters of the dissertation focused on the specific Roseobacter strain S. stellata E-37. Roseobacter species are common in environments rich in aromatic compounds, including salt marshes (Ansede et al., 2001; Buchan et al., 2001; Buchan et al., 2003), where S. alterniflora is the dominant primary producer. Therefore, it is unsurprising that of the Roseobacter genomes examined, 85.5% (266/311) contained genes for at least one complete ring-cleaving aromatic degradation pathway. However, an obvious relationship between gene presence and phylogeny did not emerge (Chapter 2), suggesting these genes were acquired at different times by the different Roseobacter strains. This overall lack of synteny between aromatic catabolic genes and Roseobacter phylogeny could be attributed to horizontal gene transfer, a mechanism that is common among Roseobacter species (Beyersmann et al., 2013; Dogs et al., 2013; Frank et al., 2014; Luo & Moran, 2014; Petersen et al., 2012; Wagner-Dobler et al., 2010). However, what is clear is the propensity for Roseobacter strains to inhabit aromatics-enriched environments. Through genome analyses, we also discovered the presence of several complete copies of gentisate or homogentisate degradation pathways in three Roseobacter strains. It is possible that the enzymes for which these genes encode have uncharacterized functions and/or alternative substrates, a concept that Chapter 3 delves into with the genes encoding for feruloyl-CoA synthases in S. stellata.

123

Hydroxycinnamates including ferulate and p-coumarate are major components of lignin in smooth cordgrass and are useful for biotechnology due to the production of other aromatic chemicals, such as vanillin (Opsahl & Benner, 1995; Rosazza et al., 1995). Enzymatic and genetic pathways for hydroxycinnamate degradation have been elucidated in soil bacteria (Calisti et al., 2008; Campillo et al., 2014; Overhage et al., 1999). However, relatively little is known regarding the enzymes involved in hydroxycinnamate degradation in marine bacteria, including S. stellata E-37. Therefore, the second aim of this dissertation was to locate and characterize these hydroxycinnamate genes in the S. stellata genome. Through genome mining, we found two genes annotated as feruloyl-CoA synthases, a protein that is responsible for the first step in hydroxycinnamate degradation. The presence of two copies of this gene is novel among bacteria that can utilize ferulate; therefore, we sought to ascertain why S. stellata retained two copies of this gene. This was accomplished through the generation of insertional mutagenesis mutants and physiological characterization of these mutants. Results indicated there is promiscuity and overlapping function between the two feruloyl-CoA synthase proteins, with each enzyme playing a more dominant role in the degradation of either ferulate (fcs2) or p-coumarate (fcs1). Moreover, after insertional mutagenesis of a gene encoding for a TRAP transporter periplasmic subunit adjacent to fcs2, we determined this transporter system is required for ferulate or p-coumarate uptake, providing another example for hydroxycinnamate nonspecificity in S. stellata. Given S. stellata was isolated from an environment where aromatic moieties are abundant, we believe these trends in enzyme promiscuity prime S. stellata for the presence of mixed substrates in salt marshes.

Along the same vein, Chapter 4 identified aromatic substrate preference hierarchy in S. stellata when the bacterium is given more than one aromatic substrate. The results from this study varied from previous work (Gulvik & Buchan, 2013), but also hints at S. stellata’s adaptation to salt marshes. The 124 preference for hydroxycinnamates suggests S. stellata would utilize ferulate or p- coumarate prior to other lignin-derived aromatics (e.g. benzoate) that are less abundant in salt marshes. Furthermore, the observation of simultaneous catabolism of benzoate and p-hydroxybenzoate by S. stellata corroborates other studies, where simultaneous catabolism coincides with substrate concentrations that are growth-limiting (Harder & Dijkhuizen, 1982). Gulvik and Buchan (2013) suggested that the presence of multiple transporters was the underlying mechanism for simultaneous catabolism of benzoate and p-hydroxybenzoate by S. stellata. Results from Chapter 4 also provide evidence for transporter involvement in how S. stellata interacts with lignin-derived aromatic mixtures. The TRAP family transporter system found upstream to the fcs2 and box operon genes are responsible for hydroxycinnamate, but not benzoate uptake. Despite this, dctP is still upregulated when S. stellata is grown on benzoate; therefore, we hypothesize that this potential cross-regulation plays a role in hydroxycinnamate preference over benzoate in S. stellata.

We now know that S. stellata selectively utilizes either simultaneous catabolism or substrate preference hierarchy, depending on the mixtures of lignin-derived aromatics it is given. What is still unknown is if these same patterns will emerge if S. stellata were provided a mixture of lignin-derived aromatics at environmentally relevant substrate concentrations. Furthermore, these experiments were only performed in binary mixtures, but in salt marsh ecosystems, S. stellata would likely be exposed to several lignin-derived aromatics; therefore, it may be prudent to re-perform these experiments with more aromatics. However, this research provides the foundations to potential future studies and gives insight into how Roseobacters contribute to carbon cycling and have adapted to specific niches, particularly salt marshes.

125

I. References

Ansede, J. H., Friedman, R., & Yoch, D. C. (2001). Phylogenetic analysis of culturable dimethyl sulfide-producing bacteria from a Spartina-dominated salt marsh and estuarine water. Appl Environ Microbiol, 67(3), 1210-1217. doi:10.1128/AEM.67.3.1210-1217.2001 Beyersmann, P. G., Chertkov, O., Petersen, J., Fiebig, A., Chen, A., Pati, A., et al. (2013). Genome sequence of Phaeobacter caeruleus type strain (DSM 24564(T)), a surface-associated member of the marine Roseobacter clade. Stand Genomic Sci, 8(3), 403-419. doi:10.4056/sigs.3927623 Buchan, A., Gonzalez, J. M., & Moran, M. A. (2005). Overview of the marine roseobacter lineage. Appl Environ Microbiol, 71(10), 5665-5677. doi:10.1128/AEM.71.10.5665-5677.2005 Buchan, A., Neidle, E. L., & Moran, M. A. (2001). Diversity of the ring-cleaving dioxygenase gene pcaH in a salt marsh bacterial community. Appl Environ Microbiol, 67(12), 5801-5809. doi:10.1128/AEM.67.12.5801-5809.2001 Buchan, A., Newell, S. Y., Butler, M., Biers, E. J., Hollibaugh, J. T., & Moran, M. A. (2003). Dynamics of bacterial and fungal communities on decaying salt marsh grass. Appl Environ Microbiol, 69(11), 6676-6687. Calisti, C., Ficca, A. G., Barghini, P., & Ruzzi, M. (2008). Regulation of ferulic catabolic genes in Pseudomonas fluorescens BF13: involvement of a MarR family regulator. Appl Microbiol Biotechnol, 80(3), 475-483. doi:10.1007/s00253-008-1557-4 Campillo, T., Renoud, S., Kerzaon, I., Vial, L., Baude, J., Gaillard, V., et al. (2014). Analysis of hydroxycinnamic acid degradation in Agrobacterium fabrum reveals a coenzyme A-dependent, beta-oxidative deacetylation pathway. Appl Environ Microbiol, 80(11), 3341-3349. doi:10.1128/AEM.00475-14 Dogs, M., Voget, S., Teshima, H., Petersen, J., Davenport, K., Dalingault, H., et al. (2013). Genome sequence of Phaeobacter inhibens type strain (T5(T)), a secondary metabolite producing representative of the marine Roseobacter clade, and emendation of the species description of Phaeobacter inhibens. Stand Genomic Sci, 9(2), 334-350. doi:10.4056/sigs.4448212 Frank, O., Pradella, S., Rohde, M., Scheuner, C., Klenk, H. P., Goker, M., & Petersen, J. (2014). Complete genome sequence of the Phaeobacter gallaeciensis type strain CIP 105210(T) (= DSM 26640(T) = BS107(T)). Stand Genomic Sci, 9(3), 914-932. doi:10.4056/sigs.5179110 Gulvik, C. A., & Buchan, A. (2013). Simultaneous catabolism of plant-derived aromatic compounds results in enhanced growth for members of the Roseobacter lineage. Appl Environ Microbiol, 79(12), 3716-3723. doi:10.1128/AEM.00405-13 Harder, W., & Dijkhuizen, L. (1982). Strategies of mixed substrate utilization in microorganisms. Philosophical Transactions of the Royal Society of 126

London. B, Biological Sciences, 297(1088), 459-480. doi:10.1098/rstb.1982.0055 Luo, H., & Moran, M. A. (2014). Evolutionary ecology of the marine Roseobacter clade. Microbiol Mol Biol Rev, 78(4), 573-587. doi:10.1128/MMBR.00020- 14 Opsahl, S., & Benner, R. (1995). Early diagenesis of vascular plant tissues: Lignin and cutin decomposition and biogeochemical implications (Vol. 59). Overhage, J., Priefert, H., & Steinbuchel, A. (1999). Biochemical and genetic analyses of ferulic acid catabolism in Pseudomonas sp. Strain HR199. Appl Environ Microbiol, 65(11), 4837-4847. Petersen, J., Brinkmann, H., Bunk, B., Michael, V., Pauker, O., & Pradella, S. (2012). Think pink: photosynthesis, plasmids and the Roseobacter clade. Environ Microbiol, 14(10), 2661-2672. doi:10.1111/j.1462- 2920.2012.02806.x Rosazza, J. P. N., Huang, Z., Dostal, L., Volm, T., & Rousseau, B. (1995). Review: Biocatalytic transformations of ferulic acid: An abundant aromatic natural product. Journal of Industrial Microbiology, 15(6), 457-471. doi:10.1007/bf01570016 Wagner-Dobler, I., Ballhausen, B., Berger, M., Brinkhoff, T., Buchholz, I., Bunk, B., et al. (2010). The complete genome sequence of the algal symbiont Dinoroseobacter shibae: a hitchhiker's guide to life in the sea. ISME J, 4(1), 61-77. doi:10.1038/ismej.2009.94 Wagner-Dobler, I., & Biebl, H. (2006). Environmental biology of the marine Roseobacter lineage. Annu Rev Microbiol, 60, 255-280. doi:10.1146/annurev.micro.60.080805.142115

127

APPENDIX: GENOMIC AND PHYSIOLOGICAL CHARACTERIZATION AND DESCRIPTION OF MARINOBACTER GELIDIMURIAE SP. NOV., A PSYCHROPHILIC, MODERATE HALOPHILE FROM BLOOD FALLS, AN ANTARCTIC SUBGLACIAL BRINE

128

A version of this chapter was originally published by Michelle J Chua, Richard L Campen, Lindsay Wahl, Joseph J Grzymsk, and Jill A Mikucki:

Michelle J Chua, Richard L Campen, Lindsay Wahl, Joseph J Grzymski, Jill A Mikucki. “Genomic and physiological characterization and description of Marinobacter gelidimuriae sp. nov., a psychrophilic, moderate halophile from Blood Falls, an antarctic subglacial brine.” FEMS Microbiology Ecology 94 (2018).

Author contribution statement: Drs. Joseph Grzymski and Jill Mikucki developed the initial project objectives and prepared the genome for sequencing. Jill Mikucki isolated Marinobacter gelidimuriae. Lindsay Wahl and Dr. Richard Campen generated the images for Figure 6.1 and 6.2, respectively. Michelle Chua performed the experiments for Figure 6.5, collected the data for Tables 6.1-6.3, and performed the dentrification and ferrous iron oxidation experiments. Michelle Chua and Jill Mikucki performed genome analyses. Joseph Grzymski created the images for Figures 6.3, 6.6-6.8. All authors helped prepare the manuscript.

I. Abstract

Antarctic subice environments are diverse, underexplored microbial habitats. Here, we describe the ecophysiology and annotated genome of a Marinobacter strain isolated from a cold, saline, iron-rich subglacial outflow of the Taylor Glacier, Antarctica. This strain (BF04_CF4) grows fastest at neutral pH (range 6– 10), is psychrophilic (range: 0°C–20°C), moderately halophilic (range: 0.8%–15% NaCl) and hosts genes encoding potential low temperature and high salt adaptations. The predicted proteome suggests it utilizes fewer charged amino acids than a mesophilic Marinobacter strain. BF04_CF4 has increased concentrations of membrane unsaturated fatty acids including palmitoleic (33%) and oleic (27.5%) acids that may help maintain cell membrane fluidity at low temperatures. The genome encodes proteins for compatible solute biosynthesis 129 and transport, which are known to be important for growth in saline environments. Physiological verification of predicted metabolic functions demonstrate BF04_CF4 is capable of denitrification and may facilitate iron oxidation. Our data indicate that strain BF04_CF4 represents a new Marinobacter species, Marinobacter gelidimuriae sp. nov., that appears well suited for the subglacial environment it was isolated from. Marinobacter species have been isolated from other cold, saline environments in the McMurdo Dry Valleys and permanently cold environments globally suggesting that this lineage is cosmopolitan and ecologically relevant in icy brines.

130

II. Introduction

Greater than 80% of the Earth's biosphere is below 5°C and includes permafrost, sea ice, glaciers and much of the Earth's oceans. Despite this fact, a limited number of psychrophilic microorganisms have been characterized. Ice covers 10% of the Earth's terrestrial surface, with the majority (∼90%) being in Antarctica (Priscu & Christner, 2004). Antarctic subglacial environments are permanently cold and dark due to ice covers up to 4.8 km thick (Fretwell et al., 2013). A diversity of subglacial ecosystems exist in Antarctica, ranging from shallow freshwater wetland systems (Fricker et al., 2013), deep closed basin lakes (Siegert et al., 2001) and saline brines (Mikucki et al., 2015). The microbial molecular diversity from the first direct sampling of an Antarctic subglacial lake, freshwater Subglacial Lake Whillans, was recently reported (Christner et al., 2014). Blood Falls, which is sourced from a highly saline (>8% salinity) subsurface aquifer (Mikucki et al., 2015), is, to date, the only other Antarctic subglacial environment to be sampled and characterized (Mikucki et al., 2004; Mikucki et al., 2009; Mikucki & Priscu, 2007). While, little information is available on how microorganisms adapt to the cold, dark constraints imparted by subglacial environments, these ecosystems are now known to harbor diverse and metabolically active microbial communities (Christner et al., 2014; Christner et al., 2008; Mikucki & Priscu, 2007). Because these features are difficult to access, microbial isolates from these systems offer important material for studying microbial ecophysiology of subglacial environments.

The challenges of living under icy conditions are numerous. Cells require a liquid milieu for metabolic processes, and at subzero temperatures, this can require living under high solute concentrations. Temperature and salinity extremes can alter both microbial biomolecules and normal cellular functions. For instance, low temperatures decrease membrane fluidity, inhibit transcription and translation, and inactivate proteins and ribosomes (D'Amico et al., 2006; Margesin & Miteva,

131

2011; Phadtare, S., 2004; Piette et al., 2011; Siddiqui et al., 2013). Turgor pressure is essential for bacterial cell growth and division, but hypersalinity induces a decrease in the water activity and turgor pressure (Csonka, 1989; Oren 2006; Overmann, 2006). Thus, microorganisms living in cold, saline environments must evolve physiological and genomic adaptations to grow under such conditions.

Currently, forty-one species in the genus Marinobacter have been described. These isolates have been cultivated from diverse environments with saline properties including hydrothermal sediments, alkaline serpentine muds and an oil-producing well (Handley et al., 2009; Huu et al., 1999; Papke et al., 2013; Takai et al., 2005). However, to our knowledge, no Marinobacter species from the McMurdo Dry Valleys, Antarctica have been fully characterized. The Marinobacter genus is within the class Gammaproteobacteria and includes Gram-negative, aerobic, motile, halotolerant or halophilic, rod-shaped bacteria (Grimaud, 2010). The majority of isolated Marinobacter strains are mesophilic. To date, only one psychrophilic strain, Marinobacter psychrophilus 20041T, originally isolated from sea ice in the Canadian Basin, has been characterized and is recognized as a formal microbial species (Zhang et al., 2008).

Here, we describe genomic and physiological characteristics and ecological context of the psychrophilic, moderately halophilic Marinobacter sp. strain BF04_CF4 that was isolated from Blood Falls, an iron-rich saline subglacial Antarctic outflow. Comparison of the annotated high quality draft genome with other closely related Marinobacter genomes, along with physiological characteristics indicated that BF04_CF4 represents a novel species within the Marinobacter genus that we name Marinobacter gelidimuriae sp. nov. We discuss how the physiological traits and genomic potential of strain BF04_CF4 provide insight into survival and growth in the conditions that prevail below the Taylor Glacier. Many of the liquid oases for life in McMurdo Dry Valleys of 132

Antarctica are briny and occupied by Marinobacter spp., thus they appear to be cosmopolitan in the cold, dry valley deserts.

III. Methods

Site description and organism isolation Taylor Valley is located in the McMurdo Dry Valleys, the largest ice-free region in Antarctica (Levy et al., 2012). Taylor Glacier is an outlet glacier of the East Antarctic Ice Sheet, which terminates into Taylor Valley. There is evidence for widespread brine saturated sediments below Taylor Glacier spanning from the glacier snout to at least 5.75 km up glacier (Foley et al., 2016; Hubbard et al., 2004; Mikucki et al., 2015). Blood Falls, a feature at the glacier terminus, is the only known surface release point of this large subsurface aquifer (Mikucki et al., 2015). Subglacial outflow at Blood Falls has previously been characterized as cold, saline and ferrous (Lyons et al., 2005; Mikucki et al., 2009). The microbial diversity and ecology of Blood Falls has been described previously and was shown to contain numerous heterotrophic species, including representatives of the Marinobacter genus, and chemosynthetic organisms that utilize iron and sulfur compounds for growth (Mikucki et al., 2004; Mikucki et al., 2009; Mikucki & Priscu, 2007). The brine from Blood Falls eventually flows into its proglacial lake, Lake Bonney. Lake Bonney has two lobes and a permanent ice cover that is ∼3.5 m thick (Fountain et al., 2013). The lake water is cold, ranging from approximately −5°C at ∼40 m depth to approximately 6°C in the upper depths. Lake Bonney is also meromictic with a halocline between 13 and 15 m in the west lobe and 17 and 20 m in the east lobe (Spigel & Priscu, 1996; Ward & Priscu, 1997). Salinity in Lake Bonney ranges from fresh surface waters to hypersaline (up to 15%) bottom waters (Spigel & Priscu, 1996). The deeper waters (∼22–25 m) of the west lobe of Lake Bonney are geochemically similar to brine collected at Blood Falls (Lyons et al., 2005).

133

Samples of Blood Falls brine were collected during an active subglacial discharge event on 1 November 2004. Outflow was collected directly from the source for enrichment work and was saline (∼8% salinity), contained no 14 detectable oxygen and had a CDIC value of −999.3‰ (essentially radiocarbon dead), indicating that the brine sampled for this study had limited contact with the atmosphere (Mikucki et al., 2009). The oxidation/reduction potential of the brine measured during sampling was 100 mV, which supported the oxygen measurements indicating the brine was suboxic. At the time of collection, the brine was cold (−5.2°C), circumneutral (pH = 6.2) and contained iron and sulfate at concentrations of 3.45 mM total Fe [>97% as Fe(II)] and 50 mM SO4, respectively; no sulfides were detected (Mikucki et al., 2009). Measured + dissolved species included inorganic nitrogen (94 µM; 100% as NH4 ; i.e. no nitrate or nitrite was detected) and dissolved inorganic carbon (55 mM) and dissolved organic carbon (420 µM) (Mikucki et al., 2009). Brine was collected directly into precombusted 74 ml serum vials by holding vials under the outflow until overflowing. The serum vials were then crimp-sealed with an autoclaved butyl rubber stopper without headspace. Serum vials were kept in the dark at 4°C until returned to McMurdo Station (∼1 week). Brine (100 μl) was removed from the serum vial with a sterile syringe, inoculated onto Marine Agar (Bacto) plates in a UV-sterilized laminar flow hood, and spread with a sterile cotton swab. Plates were then incubated at 2–4°C for ∼1 month until colonies appeared. Colonies were picked with a sterile pipette tip and restreaked through three passages to ensure purity. A single colony was then picked and gently vortexed in 1× TBE buffer. An aliquot was stained with SYBR gold nucleic acid stain per the manufacturer's protocol (Invitrogen, Thermo Scientific, Waltham, MA, USA), and a single morphotype was confirmed using epifluorescence microscopy; another aliquot was used for DNA sequencing of PCR-amplified 16S rRNA genes to further confirm purity. The same procedure was repeated for strain BF14_3D that was collected from Blood Falls on 30 November 2014. While this isolate was

134 not described in detail for this study, we have included it in our phylogenetic analyses and ecological discussion.

Scanning electron microscopy A 1 ml aliquot of cells grown in liquid culture was allowed to settle and attach to poly-l-lysine (Sigma, St. Louis, MO, USA; 300,000MW) coated cover slips for 30 min. A saturated solution of HgCl2 (in double distilled H2O) was added to a stock solution of 2% OsO4 (in double distilled H2O) in the ratio of six parts 2% OsO4 to one part HgCl2. Cells were fixed in this solution for 30 min at 4°C and rinsed three times with double distilled H2O. Cells were then dehydrated with ethanol for 5 min each in 30%, 50%, 70% and 95% ethanol and washed three times with 100% ethanol for 10 min each. Cells were dehydrated by critical point drying with liquid CO2. Cover slips were mounted onto Al stubs with silver paint and sputter coated with AuPd (http://www.dartmouth.edu/∼emlab/manuals/sempreps/pseudomonas.html). Images were obtained with an XL-30 ESEM-FEG (Environmental Scanning Electron Microscopy-Field Emission Gun) microscope (FEI Company, Hillsboro, OR, USA).

Genome sequence, assembly and annotation A 50 ml volume of strain BF04_CF4 culture was obtained after growth to mid-log phase at 4°C for 2 weeks. Cells were pelleted at 12 000 rpm for 8 min and flash frozen. High-molecular weight DNA (∼500 ng) was extracted using a CTAB buffer/organic solvent extraction protocol. DNA was RNase treated, quality was confirmed on a 1% agarose gel and shipped to the Joint Genome Institute (JGI) for draft sequencing using Illumina sequencing technology. The draft genome sequences were assembled according to JGI SOP protocols (Mavromatis et al., 2009) and made available as a part of the Integrated Microbial Genomes and Microbiomes (IMG/M) data warehouse (Markowitz et al., 2014). An improved high-quality permanent draft genome was produced using the Velvet v. 1.1.04,

135

ALLPATHS v. R41043 assembly method and Prodigal 2.5 gene calling. Hypothetical proteins were further analyzed and annotated with HHpred (probability cut-off: 85%) (Soding et al., 2005). Amino acid analysis of shared proteins between strain BF04_CF4 and the mesophilic Marinobacter lipolyticus SM19T (95% sequence identity) (Martin, S. et al., 2003) was performed after reverse best hit Basic Local Alignment Search Tool (BLAST) analysis (Grzymski et al., 2008) to determine the core genome. Proteins were considered homologous between the two strains if they were top BLAST hit in an all against all BLAST at an E-value cut-off of E−50. Substitutions were calculated from each aligned protein pair. Alignments were performed using Clustal (Larkin et al., 2007).

Phylogenetic analysis The 16S rRNA gene sequence for BF04_CF4 generated previously by Mikucki and Priscu (2007) showed 100% identity to that obtained from the sequenced genome, and was used for the phylogenetic analysis. The 16S rRNA gene for strain BF14_3D was amplified by PCR from a genomic DNA sample prepared from a single isolated colony, using the MoBio PowerSoil® DNA Isolation Kit (Carlsbad, CA, USA). PCR reactions contained 1× AmpliTaq Gold 360 MasterMix (Applied Biosystems), 1 ng of template genomic DNA and 0.2 µM of the bacterial primers 27F (AGAGTTTGATCMTGGCTCAG) and 1492R (CGGTTACCTTGTTACGACTT) under the following conditions: denaturation at 95°C for 10 min; 32 cycles at 95°C for 45 s, 50°C for 1 min and 72°C for 90 s; and a final extension at 72°C for 10 min. Sanger sequencing was performed by Eurofins Genomics (Huntsville, AL, USA) in both directions using primers 27F and 1492R. The raw sequences were processed using Geneious 5.5.6 (Kearse et al., 2012), with the ends of the reads containing low quality bases trimmed, and the resulting sequences aligned to generate a consensus sequence with a length of 1411 bp.

136

A nucleotide BLAST search was performed against the nonredundant nucleotide collection using the online version of BLAST (http://blast.ncbi.nlm.nih.gov.proxy.lib.utk.edu:90). The top hits for BF04_CF4 and BF3D_14, along with selected sequences from representative isolates were aligned using the SILVA incremental aligner (SINA) v1.2.11 at http://www.arb- silva.de/aligner/ (Pruesse et al., 2012). The final alignment of 1375 bp, trimmed to the length of the shortest sequence, was imported into MEGA version 6.06 (Tamura et al., 2013) for phylogenetic analysis. A maximum-likelihood phylogenetic tree was constructed with the Tamura-Nei model with discrete Gamma distribution (TN93 + G), using the highest subtree-pruning-regrafting heuristic inference, and 1000 bootstrap replicates.

Genome comparisons between strain BF04_CF4 and other sequenced Marinobacter species were conducted using online software for pairwise average nucleotide identities (ANI) and DNA–DNA hybridization (DDH) estimates. Pairwise average nucleotide identities were computed using the genome comparison tools in the IMG/M microbial genome annotation and analyses software available through JGI. An additional parameter for species consideration is the DDH percentage between two closely related organisms. We used the genome to genome distance calculator (GGDC; version 2.1) for a genome-based species delineation (Auch et al., 2010a; Auch et al., 2010b; Meier-Kolthoff et al., 2013). GGDC calculates an intergenomic distance under three different distance formulas.

Growth rate experiments The effects of temperature, NaCl concentration and pH on the growth of strain BF04_CF4 were measured using a modified marine broth. To prepare the marine broth basal medium (MBBM), the following chemicals were added to 1 l of Milli-Q water: ferric citrate (0.1 g), MgCl2 (5.9 g), MgSO4 (3.24 g), CaCl2 (1.8 g), KCl

(0.55 g), NaHCO3 (0.16 g), KBr (0.08 g), SrCl2 (34 mg), H3BO3 (22 mg), sodium

137 silicate (4 mg), NaF (2.4 mg), NH4NO3 (1.6 mg) and Na2HPO4 (8 mg). MBBM (supplemented with 5 g peptone, 1 g yeast extract and 19.45 g NaCl) was used for the temperature experiments. The final pH of supplemented MBBM was approximately 7.6. Cells were grown to mid-log phase and transferred to supplemented MBBM (final cell concentration: 2% of total volume). Strain BF04_CF4 was then incubated at 0°C, 4°C, 10°C, 15°C, 20°C and 25°C. To test the effect of salinity, MBBM was supplemented with 5 g peptone, 1 g yeast extract and NaCl to final concentrations of 0%, 0.8%, 3%, 6%, 8%, 15% and 20%. For the pH experiments, the pH of MBBM (supplemented with 5 g peptone, 1 g yeast extract and 19.45 g NaCl) was adjusted to 5, 6, 7, 8, 9, 10 or 11 with 1 and 10 M HCl or NaOH. Strain BF04_CF4 was grown at 15°C for NaCl and pH experiments, the temperature at which fastest growth was observed. Growth was measured by absorbance at 600 nm with a GENESYS 20 visible spectrophotometer (Thermo Scientific, Waltham, MA, USA).

Fatty acid analysis Marinobacter lipolyticus SM19T (DSM 15157T) was obtained from DSMZ (German collection of microorganisms and cell cultures; Braunschweig, Germany) for fatty acid methyl ester (FAME) comparison. Strain BF04_CF4 and M. lipolyticus were grown in duplicate at 15°C in supplemented MBBM until mid-log phase. Cells were then pelleted by centrifugation at 5000 g and 15°C for 10 min. Pellets were sent to Microbial ID Inc. (Newark, DE, USA; http://www.microbialid.com/) for analysis of fatty acid methyl ester profiles.

Phenotypic analyses and carbon substrate utilization The Gram stain was conducted on isolated colonies (Bartholomew, 1962) and results were confirmed by the KOH test as detailed in (Buck, 1982). Motility was determined by growth on BBL motility test medium (5% NaCl) according to the manufacturer's instructions.

138

Organic carbon utilization profiles for BF04_CF4 were generated with Biolog PM1 and PM2A Microplates (BIOLOG Inc., Hayward, CA, USA). Briefly, BF04_CF4 cells were pelleted as above, washed and transferred to a minimal medium (MM) for inoculating Biolog Microplates according to Shivaji et al.

(Shivaji et al., 2005). MM was prepared with (per liter of MilliQ water): 1 g NH4Cl,

0.075 g K2HPO4, 1.45 g CaCl2, 60 g NaCl, 6.15 g MgCl2, 0.75 g KCl and 0.028 g

FeSO4 without added carbon. Cell pellets were resuspended in MM, and Biolog Redox Dye mix D was added (1× final concentration) to the cell suspension. Cell suspensions (150 µL) were pipetted into each Biolog well (8 × 105 cells final concentration). Absorbance was measured at 530 nm with a Synergy HT Multi- Mode Microplate Reader (BioTek, Winooski, VT, USA) every 12 h for 2 weeks. Carbon substrate wells corresponding to D- and L-alanine, L-arginine, glucose, L-glutamine, glycerol, L-isoleucine, leucine, L-methionine, L-ornithine, D-xylose, acetate and D-fructose showed color change in Biolog microplates after 2 weeks. BF04_CF4 was then grown in triplicate in MM with 10 mM of each carbon source that was positive via Biolog to confirm growth. BF04_CF4 was incubated at 15°C in 10 ml volumes with 5% inoculum. Growth was measured following two transfers. MM (with no added carbon source) was inoculated with strain BF04_CF4 as a negative control for substrate incubations.

Denitrification and ferrous iron (Fe(II)) oxidation

Cells were cultivated under anaerobic conditions [N2/CO2 80:20 (v/v)] at 15°C in a synthetic medium: (per liter of MilliQ water) 3.7 g NH4Cl, 6.2 g MgSO4 x 7H2O,

1.5 g CaCl2 and 0.75 g KCl, supplemented with 60 g NaCl, 5 g peptone, 1 g acetate and 4 mM nitrate (KNO3) (Gauthier et al., 1992). pH was adjusted to 7.0 with 1 M NaOH. Nitrate reduction was determined by a colorimetric assay as described by Lanyi (Lanyi, 1987). Briefly, 0.1 ml of reagent A (sulphanilic acid, 8 g; 5 N acetic acid) and 0.1 ml of reagent B (N,N-dimethyl-l-naphthylamine, 6 ml; 5 N acetic acid, 1000 ml) were added to uninoculated controls and cells at mid- log phase. Change in media color to red was observed directly within 1 min. The

139 reduction of nitrate forms nitrous acid in the aqueous medium. The nitrous oxide then reacts with sulphanilic acid and naphthylamine to form a visible red, compound. An absence of color change indicates either nitrate reduction did not occur or nitrite was further reduced to N2. Following the 1 min observation, 25 mg of zinc dust was added to the culture to test for the absence of nitrate and therefore, complete reduction to N2. Pseudomonas stutzeri strain DCP-Ps1 was used as a positive control for the nitrate reduction phenotype (Sanford et al., 2012). To test for enzymatic Fe(II) oxidation, BF04_CF4 cells were grown anaerobically [N2/CO2 80:20 (v/v)] in synthetic medium with the addition of 5 mM acetate and 10 mM Fe(II) (FeSO4 · 7H2O) and nitrate. Cultures and abiotic controls were incubated at 15°C for approximately 2 weeks. Evidence for Fe(II) oxidation is a change in the media from clear to the formation of orange–red precipitate (ferric iron).

IV. Results and Discussion

Morphology Strain BF04_CF4 cells are Gram-negative and a single, rod-shaped morphology (average length = ∼1.7 μm; average width = ∼0.5 μm) was observed using scanning electron microscopy (Figure 6.1). There was variation in cell length, with the shortest cell viewed as ∼850 nm and the longest at ∼2.3 µm. These observations are consistent with previous descriptions of the Marinobacter genus (Bowman & McMeekin, 2005). Cells of BF04_CF4 did not have evidence of flagella and motility was not observed. The cell membrane appeared to contain pores under high magnification. Colonies were round, white to cream-colored and 1–1.5 mm in diameter when grown on Marine Agar (Difco, Thermo Scientific, Waltham, MA, USA).

140

Phylogenetic relationships 16S rRNA gene sequence analysis placed strain BF04_CF4 in the Marinobacter genus and comparative analysis in the NCBI nucleotide database using BLAST revealed highest sequence identity to both uncultured clones and cultured isolates collected from other cold and saline environments including other brines within the McMurdo Dry Valleys (Figure 6.2). The top BLAST hits for BF04_CF4 were predominantly of Antarctic origin. BF04_CF4 shared the highest nucleotide identity (99.9%) to the uncultured clone 5III.9 isolated from sediments in the west lobe of Lake Bonney (Tang et al., 2013) and strain ELB17 (99.2% identity to BF04_CF4), which was originally isolated from 17 m depth in the east lobe of Lake Bonney. Blood Falls, or the subglacial source of brine to Blood Falls, supplies a significant amount of solutes to the west lobe of Lake Bonney (Lyons et al., 2005), so it is possible that members of the population represented by strain BF04_CF4 are transported from Blood Falls or below the Taylor Glacier into the west lobe and eventually into the east lobe. Water from the west lobe has been shown to migrate to the east lobe via a process termed ‘chemocline leakage’ (Doran et al., 2014).

Strain BF04_CF4 was collected from Blood Falls during the austral summer of 2004, yet showed high-sequence identity to environmental clones obtained from a sample collected during an active outflow event in 1999 (Mikucki & Priscu, 2007). Clones BF_C65 and BF_C90 shared 98.4% and 98.2% sequence identity with BF04_CF4 respectively, and these clones comprised 3.7% of the clone library (Mikucki & Priscu, 2007). BF04_CF4 showed somewhat lower sequence identity (96.4%) to the strain BF14_3D, which was isolated during an active outflow event in 2014. Another Antarctic strain (R-36953), sharing 99.1% sequence identity, was isolated from Forlidas Pond, located in the Ellsworth Land Region of Antarctica, which is a considerable distance from the dry valleys (Peeters et al., 2011). However, not all BLAST hits with high-sequence identity (>97%) were geographically constrained to Antarctica, and included isolates and 141 clones of arctic and subarctic origin as well. Therefore, although the high identity in sequences between clones and isolates from Blood Falls and Lake Bonney may partially be explained by hydrologic connectivity, it does not fully explain the high identity to strains from more distant locations.

Our phylogenetic analysis also revealed that the two Marinobacter isolates from Blood Falls (BF04_CF4 and BF14_3D) belong to one of two apparent clades (Figure 6.2). These clades contained clones and isolates derived from both Antarctica and elsewhere, including the Arctic and sub-Arctic. Such diverse Marinobacter species appear to be cosmopolitan in cold brines (Niederberger et al., 2010; Perreault et al., 2007; Perreault et al., 2008). Collectively this suggests that, the diversity of Marinobacter in Blood Falls and other briny dry valley lakes may not be the result of local speciation, but rather a result of diverse Marinobacter species present in the ancient marine waters that originally formed the brines.

The closest characterized species for which BF04_CF4 shared high 16S rRNA gene sequence similarity were the psychrophilic M. psychrophilus 20041T (97.8%) and psychrotolerant Marinobacter antarcticus ZS2–30T (96.3%), originally isolated from coastal sediments near Larsemann Hills, Antarctica (Liu et al., 2012). These rRNA gene sequence identities indicate that strain BF04_CF4 is a distinct species from M. antarcticus and is close to the species cut-off (97% or greater) with M. psychrophilus. Although >97% 16S rRNA gene sequence similarity is widely used to classify different species (Stackebrandt & Goebel, 1994), recent studies have shown that strains with less than 99% 16S rRNA gene sequence similarity benefit from additional comparisons such as DDH or determination of an ANI (Meier-Kolthoff et al., 2013). In silico-based estimates of DDH between BF04_CF4 and M. psychrophilus was 44–48.4% (Table S1), which is far below the >70% DDH species delineation cut-off. The genome of strain BF04_CF4 had 142 an ANI of 92.95% to bidirectional best hits of the strain M. psychrophilus (Table S1).

For decades, the ‘gold standard’ for delineating microbial species has been >97% 16S rRNA gene sequence similarity (Stackebrandt & Goebel, 1994). Even before the broad accessibility of microbial genome sequencing, scientists have debated this threshold as potentially deficient (Stackebrandt et al., 2002). With the growing database of sequenced microbial genomes, ANI has emerged as a highly robust metric, with a threshold of 95%–96% delineating species (Chan et al., 2012; Konstantinidis et al., 2006; Konstantinidis et al., 2005). A recent meta- analysis of available genomes corroborates this finding and determined that 95– 96% ANI corresponds to 98.65% 16S rRNA gene sequence similarity (Kim et al., 2014). The authors conclude that >98% is a better standard for species designation when examining high quality, full-length 16S rRNA gene sequences. We contend that the differences in 16S rRNA sequence, ANI and DDH values indicate that BF04_CF4 is a distinct species from M. psychrophilus.

Genome summary The BF04_CF4 genome is composed of 3,633,797 base pairs with a G + C content of 54.63%, which is comparable to other closely related Marinobacter species (Table 6.1); for example, M. psychrophilus contains 53.84% G + C content. The permanent high-quality draft genome contained 136 scaffolds and 139 contigs. The genes encode for 3753 proteins, 2784 of which have been associated with a predicted function. The RNA genes were re- annotated because they were broken between contigs and incomplete in length. There are likely a total of 92 RNA genes with 6 total rRNA in 2 rRNA operons, 45 tRNA genes, 11 other miscellaneous RNAs and 30 group II catalytic introns with high identities to integrases from other bacteria, fragments and un-annotated parts of the deep sea, cold-adapted Photobacterium profundum SS9 genome (Vezzi et al., 2005). The BF04_CF4 genome contains at least 153 transposases

143 or insertion sequence elements, 6 resolvases and 59 phage-related genes or integrases. The presence of insertion sequence elements has been implicated in genetic mobility and versatility, which may allow for rapid adaptations to new environmental conditions (Touchon & Rocha, 2007). Several transposase flanking-genes encoded for other transposases, phage-related proteins, nucleases, nucleic acid-binding domains, chemotaxis proteins, and proteins of unknown function. Phage-related proteins include site-specific recombination integrases, another factor in genome flexibility and modification (Argos et al., 1986; Hochhut et al., 2006; Ogier et al., 2010). Other putative phage-related genes in strain BF04_CF4 include a major coat protein Gp8, phage tail, capsid, and maturation and terminase proteins (Table S1). The majority of the nucleotide sequences from the phage-related proteins did not produce sequence identities except a phage integrase (IMG gene ID: 2515470055). The nucleotide sequence identity of the phage integrase was 93% identical to the sequence of plasmid pMAQU02 in M. aquaeolei VT8. Presence of both insertion sequence elements and phage genes in the genome may indicate horizontal gene transfer events in strain BF04_CF4.

Our survey of the BF04_CF4 genome and comparisons made to its close relative, M. psychrophilus (Figure 6.3), further support the differences between these two strains and reveal distinct attributes in BF04_CF4 that may reflect its unique ecological history. A comparison of cluster of orthologous genes (COG) gene abundances between the two strains showed that, overall, BF04_CF4 has a higher percentage of genes that are not in COG databases (42.8% vs. 26.4%).

The genome BF04_CF4 has a greater number of defense mechanisms (3.71%) and mobilome (1.88%) related genes (Figure 6.4) compared to M. psychrophilus (1.93% and 0.48% respectively), including genes for restriction modification systems not found in M. psychrophilus. BF04_CF4 also appears to have a complete Type I-C/Dvulg CRISPR/cas system (Makarova et al., 2015; 144

Nam et al., 2012). M. psychrophilus lacks any CRISPR/cas genes, as do most other sequenced Marinobacter species. The CRISPR/cas gene set found in BF04_CF4 was also found in strain ELB17, where these genes share between 90–99% protein identity. CRISPR/cas systems act as an immune system for bacteria, providing resistance against phages (Hille & Charpentier, 2016). Often CRISPR/cas systems are acquired by horizontal gene transfer events, which could explain the lack of a homologous CRISPR/cas system in other Marinobacter strains. The presence of a CRISPR/cas system in BF04_CF4, as well as numerous mobile elements, may be indicative of higher levels of phage predation for this strain.

Marinobacter psychrophilus has a higher abundance of cell motility genes (3.4%) compared to BF04_CF4 (1.7%), which likely correlates with their phenotypes; M. pscyhrophilus was motile (Zhang et al., 2008), whereas we did not observe motility in BF04_CF4. Of the currently 52 sequenced Marinobacter genomes, only strains BF04_CF4 and two strains originally isolated from Lake Vida (strains LV10MA510–1, and LV10R520–4) contain genes for gas vesicles (Table S1). Gas vesicles are typically used by microbes for buoyancy in stratified water columns (Walsby, 1994) and provide an advantage to aerobic halophilic microorganisms because oxygen is less soluble in salt-concentrated water (Oren 2002; Pfeifer, 2006). Strain BF04_CF4 may use gas vesicles to float to the surface of the Blood Falls brine conduit for access to oxygen; it has also been speculated that in slow-growing organisms, utilization of gas vesicles to maintain buoyancy actually decreases energy costs in comparison to flagella utilization (Jung et al., 2004; Walsby, 1994). However, little is known about the subglacial structure of the brine or how gas vesicles might be an adaptive advantage in brine-saturated sediments, which is significantly deeper in the subsurface (>450 m deep). Chivian et al. (Chivian et al., 2008) reported the presence of genes for the formation of gas vesicles in ‘Candidatus Desulforudis audoxyiator,’ whose genome was constructed from an extremely low diversity sample of fracture 145 water of a South African gold mine at 2.8 km depth. Perhaps there are yet unknown functions of gas vesicles that convey a growth advantage in the deep subsurface.

Cold adaptation Strain BF04_CF4 grew at 0°C (0.15 cell division per day) with fastest growth observed at 15°C (0.8 cell divisions per day); no growth was observed at 25°C, indicating the bacterium can be described as psychrophilic (Table 6.1). Psychrophiles grow optimally at temperatures of 15°C or lower, tolerate a maximum temperature of approximately 20°C, and have a minimum growth temperature at or below 0°C (Morita, 1975). The laboratory-observed growth temperature range for BF04_CF4 was 0–20°C (Table 6.1) that was comparable to the range reported for M. psychrophilus (range = 0°C–22°C) (Zhang et al., 2008). However, Blood Falls outflow was approximately −5°C when samples were collected which was considerably lower than the strain's optimal (fastest) growth temperature (Topt). Using the Weibull distribution (Figure 6.5A), the growth rate of BF04_CF4 was predicted to be approximately 0.07 divisions per day at −4.5°C suggesting BF04_CF4 could be active and growing slowly in Blood Falls or the subglacial source to Blood Falls, rather than merely surviving in a dormant state. Laboratory-determined optimal growth temperatures are generally higher than in situ temperatures, and this trend is commonly found in microorganisms isolated from low-temperature environments (Bakermans, C. & Nealson, 2004; Ward, B. B. & Priscu, 1997). Deviations can be explained by psychrophiles maximizing growth yield as growth rates decrease at low temperatures by lowering rates of RNA and protein synthesis (Bakermans & Nealson, 2004; Feller & Gerday, 2003; Ward & Priscu, 1997), another possible mechanism for cold adaptation. In fact, increased growth rates at higher temperatures, including Topt, could be an indication of heat stress (Siddiqui et al., 2013).

146

Several key cold adaptation traits are common among psychrophilic genomes including that of BF04_CF4. These traits include genes encoding for cold shock proteins (Csps), chaperones, and RNA helicases (Lauro et al., 2011), all of which have possible roles in maintaining translation capacity. Translation is affected by low temperature-induced changes in mRNA secondary structures because secondary structures prevent ribosome movement (Phadtare & Inouye, 2008). There are five copies of RNA chaperone cspA cold shock protein genes in strain BF04_CF4. This is similar to other Marinobacter species including M. psychrophilus, which also contains 5 copies of this gene. Cold shock protein homologues have also been found in the genomes of other well-characterized psychrophiles however at different abundances, including Colwellia psychrerythraea 34H (4 csps) (Methé et al., 2005), Desulfotalea psychrophila Lsv54 (8 csps) (Rabus et al., 2004), Psychrobacter arcticus273–4 (3 csps) (Ayala-del-Rio et al., 2010), and Psychromonas ingrahamii 37 (12 csps) (Riley et al., 2008). The csp sequences of BF04_CF4 ranged from 48 to 81% identity to the aforementioned psychrophilic csps. However, within the Marinobacter genus, all five BF04_CF4 csp sequences shared highest identity (>90%) to the csps of M. psychrophilus, ELB17, BSs20148 and two isolates from Lake Vida (LV10R520–4 and LV10R510–11A) (Table S1). Strain BF04_CF4 has three cspA thermoregulators which are 5΄ untranslated (UTR) RNAs that regulate the expression of cspA and likely other protein-encoding genes during stress response. Again, the BF04_CF4 cspA thermoregulator sequences were most similar (>95%) to sequences of M. psychrophilus and strain BSs20148. These thermoregulators and the low in situ temperature of Blood Falls could induce an increase in the number of stable mRNA secondary structures. BF04_CF4 also contains four DEAD-box containing genes homologous to csdA, dbpA, rhlB and rhlE that encode proteins responsible for ATP-dependent RNA helicase activity (Annie et al., 2004; Kuhn, 2012). The DEAD-box containing genes are all most closely related to cold-adapted M. psychrophilus and strains ELB17 and BSs20148 (Table S1). In addition, strain 147

BF04_CF4 possesses genes for protein chaperones, which are associated with assisting in three-dimensional protein structure formation (Hoffmann et al., 2010). Protein chaperone genes in strain BF04_CF4 encode for the trigger factor protein, DnaK, DnaJ, GrpE, GroEL, and GroES, most of which are also most closely related to M. psychrophilus and strains ELB17 and BSs20148 (Table S1). The trigger factor is a molecular chaperone that has peptidyl-prolyl cis-trans isomerase activity (Kandror & Goldberg, 1997) that accelerates folding of growing polypeptide chains (Nagradova, 2008). The protein chaperones DnaK, DnaJ, GrpE, GroEL and GroES also assist in protein folding, and studies have shown upregulation of these genes in psychrotolerant and psychrophilic microorganisms (Gao et al., 2006; Hartl & Hayer-Hartl, 2009; Qiu et al., 2006; Suzuki et al., 2004; Ting et al., 2010).

FAME analysis of strain BF04_CF4 incubated at 15°C revealed the presence of mostly unsaturated fatty acids (Table 6.2). The percentage of unsaturated fatty acids in BF04_CF4 was higher than the mesophile M. lipolyticus (which has an optimal growth temperature = 37°C). However, when grown at 15°C palmitoleic, oleic and vaccenic acids were the most abundant unsaturated fatty acids produced by both strains (Table 6.2; (Russell, 2008)). This is not unexpected since most organisms increase the concentration of unsaturated fatty acids in their cell membranes at lower temperatures (Gounot, 1991; Hebraud & Potier, 1999; Marr & Ingraham, 1962). All the genes necessary for palmitoleic acid biosynthesis via the type II fatty acid synthetic system, including fabB and fabF, which encode for β-ketoacyl-ACP synthase I and II respectively, and are enzymes involved in fatty acid unsaturation, are present in the BF04_CF4 genome (Table S1). M. psychrophilus and ELB17 were among the few Marinobacter strains that do not contain a copy of fabF (6 out of the 52 Marinobacter genomes in IMG lacked fabF). The fabF gene in BF04_CF4, showed highest identity to beta-ketoacyl-ACP synthase II from Shewanella species including a gene (WP_014609653) from S. 148 putrefaciens (70% identity) and a gene (WP_011635827) from S. frigidimarina (68%). A strain closely related to S frigidimarina (98% 16S rRNA gene sequence similarity) has been previously isolated from Blood Falls (Mikucki, J. A. & Priscu, 2007). In contrast, all other Marinobacters from cold locales, showed highest sequence identity (92–93%) to fabF from other Marinobacter species. While the exact role of fabF in the BF04_CF4 genome is unknown, it appears to have been acquired via horizontal gene transfer from the Shewanella and may provide a unique advantage in Blood Falls. All the genes required for oleic acid production are present in BF04_CF4 except stearoyl-CoA desaturase (Table S1), which is responsible for the first step of oleic acid production. It is possible that strain BF04_CF4 does contains stearoyl-CoA desaturase, but that the sequence data is absent in our analysis due to the draft genome status; alternatively oleic acid production could proceed via a different pathway in BF04_CF4. Regardless, the FAME results (Table 6.2) demonstrate BF04_CF4 is capable of synthesizing palmitoleic and oleic acids, which may help maintain cell membrane homeoviscosity to optimize transfer of nutrients and waste at colder temperatures in environments such as Blood Falls.

Our analysis shows only subtle amino acid usage differences between strains BF04_CF4 and its mesophilic relative, M. lipolyticus. The mesophilic strain uses more negatively charged amino acids than the psychrophile in the shared genome and in its overall predicted proteome (Figure 6.6). Changes in amino acid usage between mesophiles and psychrophiles are common, but not always indicative of cold adaptation (Feller & Gerday, 1997; Grzymski et al., 2008), as there are multiple factors involved in tolerating low temperatures. Charged amino acids form salt bridges that provides rigidity to protein folds. Minimizing these folds could be an important mechanism in a cold, saline subglacial environment. Given the role of helicases as cold shock proteins in cold adaptation and abiotic stress response, the amino acid composition and substitutions in this class of proteins were specifically examined. Between strains BF04_CF4 and M. 149 lipolyticus helicases, the substitutions D → N, D → A/S, E → Q, S → A, and E → A were more commonly found in the psychrophile than the mesophile (Figure 6.7). These patterns are also well defined when the entire shared predicted proteome is calculated by reciprocal BLAST and examined (Figure 6.8). For example, there are ∼1% fewer D and E amino acids in the shared proteome of BF04_CF4 compared M. lipolyticus. This reduction in negatively charged, salt- bridge forming amino acids is likely an adaptation to cold (Feller & Gerday, 1997; Grzymski et al., 2008).

Overall, BF04_CF4 possesses genomic traits found among other psychrophiles; including numerous genetic mobile elements, increased membrane lipid unsaturation, a reduction of charged amino acids, and the presence of genes encoding for cold shock proteins, chaperones, and helicases. Our analysis did not bring to light significant differences in gene abundances for proteins known to be involved in psychrophilic adaptations. Psychrophilic genomes have also been demonstrated to contain redundant genes involved in cold adaptation (De Maayer et al., 2014). A high number of tRNA genes have been associated with psychrophilic genomes (De Maayer et al., 2014) because diffusion rates are slower at low temperatures, and translation rates are dependent on tRNA diffusion rates (Bulmer, 1991). However, strain BF04_CF4 has at least 45 tRNA genes, which is comparable to all other Marinobacter species with sequenced genomes, including mesophiles, suggesting that the number of tRNA genes may not necessarily indicate a psychrophilic lifestyle.

Salt adaptation Strain BF04_CF4 grew optimally (fastest) in 6% (60 g L−1) NaCl and did not reproduce at 0% NaCl, and therefore can be characterized as a moderate halophile (Figure 6.5B). Optimal growth rates for moderate halophiles fall between ∼2.9 and ∼14.6% NaCl concentration (Kushner, 1978). Growth was observed in cultures up to 15% NaCl concentrations, thus strain BF04_CF4 could

150 grow under in situ brine salinity (∼8%) (Mikucki & Priscu, 2007). The genome of BF04_CF4 contains genes important for adaptation to saline environments. Cells maintain osmotic balance by two main strategies, which include concentrating non-ionic solutes or actively pumping protons (Oren 2008). These approaches prevent water loss and provide a neutral solvent environment within the cell. Because salts can act to prevent or depress the freezing of water, many psychrophiles reside in halophilic habitats (Kelly, 2011) and thus have strategies that help the cell contend with both conditions.

The initial response to high osmolarity in a bacterium is to increase intracellular concentrations of K+ and pump out protons (White, 2007). Genes for Trk transport systems, one of the most conserved K+ uptake systems in bacteria (Heermann & Jung, 2012), are present in BF04_CF4 (Table S1). The BF04_CF4 genome contains genes for glutamate and trehalose synthesis and glycine betaine, proline, and carnitine transport (Table S1). The accumulation of these molecules in the cytoplasm can serve as both osmo- and cryoprotectants for microbial cells. Genes for ATP-binding and permease components of an ectoine/hydroxyectoine ABC transporter for exogenous ectoine import and an ectoine synthase gene involved in ectoine biosynthesis were identified in BF04_CF4 (Table S1). BF04_CF4 also contained a gene for choline and glycine betaine aldehyde dehydrogenase (IMG gene ID: 2515470750), an enzyme required for glycine betaine biosynthesis from choline, which is common in psychrophilic genomes (Bakermans et al., 2012). A gene from BF04_CF4 coding for glutamate synthase (IMG Gene ID: 2515472203) was in the top ten BLAST hits and had 64% sequence identity for a glutamate synthase gene (Accession: ABM04053) in Psychromonas ingrahamii 37, a well-characterized psychrophile from sea ice (Riley et al., 2008). The symport of proline and Na + are commonly found in the genomes of psychrophiles (Bowman, 2008; Methé et al., 2005) and we detected all the genes of the ABC-type proline/glycine betaine transport system (opuA) and genes for the high-affinity proline-specific uptake 151 sodium:solute symporter family (opuE) were also detected. These genes shared highest identity with M. psychrophilus (97–98%) but also showed low identity to Paenibacillus darwinianus genes (41–51%). P. darwinianus was originally isolated from irradiated Antarctic soils (Dsouza et al., 2014a) and shows similar genomic features (Dsouza et al., 2014b) suggesting the accumulation of compatible solutes for osmoprotection.

The BF04_CF4 genome also contains capD (capsular polysaccharide biosynthesis protein) genes which are involved in the production of exopolymeric substances (EPS), extracellular organic molecules that can act as cryoprotectants at subzero temperatures (Table S1). This is similar to Colwellia psychrerythraea 34H, Psychrobacter arcticus 273–4, and P. ingrahamii, isolates capable of growth at subzero temperatures. In experiments, C. psychrerythraea showed increased EPS production in response to lower temperatures and increased salinity (Marx et al., 2009) suggesting EPS production can provide an advantage to cells in both cold and saline environments. Strain BF04_CF4 can grow at in situ Na+ concentrations (∼8%) and contains genes involved in a variety of strategies for tolerating high salt concentrations, including proteins for maintaining cell osmotic homeostasis and EPS production. These factors indicate BF04_CF4 has adaptations that would support growth in the salinity of Blood Falls.

Metabolic features BF04_CF4 grew on a limited number of carbon substrates including D- and L- alanine, a small amino acid, acetate and L-leucine (Table 6.3). Analysis of the BF04_CF4 genome revealed incomplete versions of pathways typically found in heterotrophs. Consistent with BF04_CF4's inability to utilize glucose, fructose bisphosphatase and the oxaloacetate-decarboxylating malate dehydrogenase genes of the glycolysis and gluconeogenesis pathways were notably absent. The non-oxidative branch of the pentose phosphate pathway is complete in strain

152

BF04_CF4, but only genes encoding for glucose-6-phosphate 1-dehydrogenase for the oxidative branch were found in the genome (IMG gene ID: 2515470888). The complete TCA cycle, common in most microbes, is also present in strain BF04_CF4. This is consistent with the ability of strain BF04_CF4 to utilize acetate as a sole carbon source. M. psychrophilis and M. antarcticus have different carbon utilization profiles compared to strain BF04_CF4 (Table 6.3), although M. antarcticus also utilizes L-alanine as an energy source (Liu et al., 2012; Zhang et al., 2008). M. psychrophilus and M. antarcticus metabolize D- serine and L-serine (Liu et al., 2012; Zhang et al., 2008), respectively, while strain BF04_CF4 lacks genes for D-serine ammonia-lyase and L-serine deaminase, which are required to convert serine to pyruvate and ammonium. Acetate has been proposed as a preferable carbon source for microorganisms in cold environments because transporters are not required for acetate uptake, and acetyl coenzyme A is produced in at most, a two-step reaction (Ayala-del-Rio et al., 2010). Alanine conversion to pyruvate only requires alanine racemase and/or D-alanine:quinoneoxidoreductase, enzymes present in BF04_CF4 (Table S1). These traits suggest energetic efficiency in BF04_CF4 and could help with growth in the subglacial environment below Taylor Glacier.

Nitrate reduction Strain BF04_CF4 coupled acetate metabolism to nitrate reduction under anaerobic conditions. Replicates of BF04_CF4 and the positive control showed no color change after the addition of denitrification reagents and zinc, indicating nitrate had been reduced beyond nitrite to N2. The BF04_CF4 genome contained genes encoding for nitrate, nitrite, nitric oxide, and nitrous oxide reductases (Table S1), all proteins necessary for denitrification. Nitrate and nitrite are low to below detection in Blood Falls brine (Mikucki et al., 2009) and may be a consequence of an active nitrate reducing population.

153

Close relatives of BF04_CF4 that are capable of denitrification have been isolated from other brines in the dry valleys. For example, strain ELB17, was capable of denitrification in laboratory experiments (Ward & Priscu, 1997). Using indirect immunofluorescence staining, the maximum concentration of ELB17 cells was found between 17 and 35 m depth in the east lobe and 13 and 22 m depth in the west lobe, which corresponds with the highest concentrations of nitrous oxide production (Lee et al., 2004; Ward & Priscu, 1997), an intermediate of microbial denitrification under low oxygen concentrations. Similar to Lake Bonney, there are high levels of inorganic nitrogen including nitrous oxide in Lake Vida brine

(58.8–86.6 µmol/L N2O) (Murray et al., 2012). These authors concluded that the nitrate and ammonium detected were of atmospheric origin; however, isotopic measurements of the nitrous oxide were inconclusive. Marinobacter sp. strains from Lake Vida ice (15.9 m) were capable of nitrous oxide production (Trubl, 2013); therefore, Marinobacter species may use nitrate as a terminal electron acceptor and contribute to nitrous oxide concentrations under in situ, suboxic conditions in Antarctic brines. While our genomic evidence from strain BF04_CF4 and that of strain ELB17 supports the notion that the nitrous oxide profiles in Lake Bonney are of biogenic origin, both biotic and abiotic reactions could be responsible for the concentrations observed. Alternatively, the dissolved iron (up − to 55 µM Fe [total]) in Lake Bonney could react with nitrite (up to ∼50 µM NO2 ) in the water column (Ward et al., 2003) to form nitrous oxide (Picardal, 2012), an abiotic process similar to what has been demonstrated between nitrite-rich (16.1– 34.5 µM) Don Juan Pond brine and Fe(II)-bearing minerals (Samarkin et al., 2010).

Iron oxidation BF04_CF4 has genes encoding for enzymes involved in iron acquisition and dissimilatory iron oxidation, but not iron reduction (Table S1). Genes in BF04_CF4 encode for ABC-type cobalamin/Fe3+-siderophore transport systems used in iron acquisition for assimilatory cellular needs such as enzyme

154 production. During energy acquisition from iron oxidation, electrons from Fe(II) are transported through one of two pathways. One pathway involves electron transfer to cytochrome c oxidase, which then transfers the electrons to oxygen.

Alternatively, electrons are shuttled from the bc1 complex, ubiquinone, and then to the NADH-ubiquinone reductase to regenerate NADH for biosynthesis and carbon fixation (Valdes et al., 2008). BF04_CF4 has genes encoding for both pathways including two operons with cytochrome c oxidase genes (coxA, coxB and coxC); there is one operon in the genome that includes genes for the bc1 complex (petA, petB and petC). These genes are also found in the representative iron-oxidizing bacterium Acidithiobacillus ferrooxidans (Accession: AF220499). The concentration of iron in the Blood Falls outflow was high (0.4- 3.45 mM) and primarily in the reduced form (>97% as Fe(II)); theoretically, iron oxidation coupled to nitrate reduction could be thermodynamically favorable under the brine Eh (100 mV) and pH (6.2) conditions if nitrate is available (Straub et al., 2001; Straub et al., 1996). However, the iron is thought to be oxidized abiotically when the brine is exposed to the atmosphere at the glacier surface

(Mikucki et al., 2009); both pH and Eh of the brine increase rapidly when exposed to air, making iron oxides the more favorable form. BF04_CF4 grew best at neutral pH with a range of 6–10 (Figure 6.5C), which indicates it is capable of growth at the in situ brine pH as well as the higher values measured when brine is exposed to the surface (Mikucki et al., 2004). It is currently unknown if the Fe(II) is microbially oxidized at the surface as well, but it is possible BF04_CF4 is involved in this process. In fact, several Marinobacter species are capable of dissimilatory iron oxidation (Handley & Lloyd, 2013); Marinobacter santoriniensis NKSG1, koreensis DSM 17924T, and aquaeolei DSM 11845T were shown to chemoheterotrophically oxidize iron coupled to nitrate reduction (Handley et al., 2009). The ability of BF04_CF4 to oxidize Fe(II) enzymatically was also tested, and when compared to the abiotic control, a red-orange precipitate formed in the BF04_CF4 cultures after approximately two weeks. Thus, it is possible that BF04_CF4 is biotically oxidizing Fe(II) to Fe(III); but the biotic process can be 155 difficult to discern due to the rapid abiotic Fe(II) oxidation by nitrite and to a lesser extent nitrate (Picardal, 2012). It is more likely that BF04_CF4 is reducing nitrate to nitrite, thus allowing for abiotic Fe(II) oxidation, a reaction that might also occur in Blood Falls. Perhaps a similar process is responsible for the lack of measureable nitrite in Blood Falls, however further experimentation would be required to confirm this.

IV. Concluding remarks

Strain BF04_CF4 was isolated from a cold, saline subglacial brine and coinciding with its isolation source, exhibits genomic and physiological adaptations to low temperatures and high salt concentrations (Figure 6.9). BF04_CF4 shares similar genomic characteristics to psychrotrophs and psychrophiles, including essential genes in the fatty acid synthase II system. The isolate also has the capability to produce compatible solutes usually involved in osmotic stress response. The isolate grows optimally (fastest) under cold and salty conditions and contains genes for energy efficiency and metabolic pathways potentially useful for growth in the geochemical conditions that exist below the Taylor Glacier and in McMurdo Dry Valley lakes. The metabolic activities of BF04_CF4 demonstrate the strain could contribute to nitrogen and iron cycling in Lake Bonney and Blood Falls. Close relatives of BF04_CF4 have been detected in subsequent brine collections from the conduit feeding Blood Falls, other brine lakes in the McMurdo Dry Valleys and elsewhere in Antarctica and globally that have similar ecological characteristics suggesting the Marinobacter lineage may be cosmopolitan to cold salty brines. These data support the notion that BF04_CF4 is not a transient interloper at Blood Falls, rather that the Marinobacter lineage, represented by the strain BF04_CF4 we characterized here, is a versatile, ecologically relevant organism in the Blood Falls microbial community and many of the briney lakes in the McMurdo Dry Valleys.

156

V. Acknowledgements

We thank Christine Foreman (Montana State University) for field assistance, Chuck Daghlian, (Dartmouth College) for electron microscopy expertise, the Antarctic Support Contractor for the United States Antarctic Program, Rob Sanford (University of Illinois) for providing Pseudomonas stutzeri strain DCP- Ps1, Elizabeth McPherson (University of Tennessee-Knoxville) for denitrification test reagents, Susan Pfiffner (University of Tennessee-Knoxville) for FAME analysis advice, Aharon Oren (The Hebrew University of Jerusalem) for assistance with nomenclature, the JGI sequencing and annotation team, and the editor and anonymous reviewers whose comments greatly improved the manuscript.

157

VI. References

Annie, P. h.-G., K., B. R., Isabelle, I., Shane, R. C., A., M. G., & W., S. R. (2004). Physical and functional interactions among RNase E, polynucleotide phosphorylase and the cold-shock protein, CsdA: evidence for a ‘cold shock degradosome’. Molecular Microbiology, 54(5), 1409-1421. doi:doi:10.1111/j.1365-2958.2004.04360.x Argos, P., Landy, A., Abremski, K., Egan, J. B., Haggard‐Ljungquist, E., Hoess, R. H., et al. (1986). The integrase family of site‐specific recombinases: regional similarities and global diversity. The EMBO Journal, 5(2), 433- 440. doi:doi:10.1002/j.1460-2075.1986.tb04229.x Auch, A. F., Klenk, H.-P., & Göker, M. (2010). Standard operating procedure for calculating genome-to-genome distances based on high-scoring segment pairs. Standards in Genomic Sciences, 2(1), 142-148. doi:10.4056/sigs.541628 Auch, A. F., von Jan, M., Klenk, H.-P., & Göker, M. (2010). Digital DNA-DNA hybridization for microbial species delineation by means of genome-to- genome sequence comparison. Standards in Genomic Sciences, 2(1), 117-134. doi:10.4056/sigs.531120 Ayala-del-Rio, H. L., Chain, P. S., Grzymski, J. J., Ponder, M. A., Ivanova, N., Bergholz, P. W., et al. (2010). The genome sequence of Psychrobacter arcticus 273-4, a psychroactive Siberian permafrost bacterium, reveals mechanisms for adaptation to low-temperature growth. Appl Environ Microbiol, 76(7), 2304-2312. doi:10.1128/AEM.02101-09 Bakermans, C., Bergholz, P. W., Rodrigues, D. F., Vishnivetskaya, T., Ayala-del- Río, H. L., & Tiedje, J. M. (2012). Genomic and expression analyses of cold-adapted microorganisms. In M. RV & W. LG (Eds.), Polar Microbiology: Life in a Deep Freeze (pp. 126–155). Washington, D. C: ASM Press. Bakermans, C., & Nealson, K. H. (2004). Relationship of critical temperature to macromolecular synthesis and growth yield in Psychrobacter cryopegella. J Bacteriol, 186(8), 2340-2345. Bartholomew, J. W. (1962). Variables influencing results, and the precise definition of steps in Gram staining as a means of standardizing the results obtained. Stain Technology, 37(3), 139-155. doi:10.3109/10520296209117723 Bowman, J. (2008). Genomic analysis of psychrophilic prokaryotes. In M. R, S. F, M. J-C, & G. C (Eds.), Psychrophiles: From Biodiversity to Biotechnology (pp. 265–284). Berlin: Springer Bowman, J., & McMeekin, T. (2005). Marinobacter. In B. DJ, K. NR, S. JT, & G. GM (Eds.), Bergey's Manual of Systematics of Archaea and Bacteria (Vol. 2, pp. 444–447). New York: Springer. Buck, J. D. (1982). Nonstaining (KOH) method for determination of gram reactions of marine bacteria. Appl Environ Microbiol, 44(4), 992-993. 158

Bulmer, M. (1991). The selection-mutation-drift theory of synonymous codon usage. Genetics, 129(3), 897-907. Chan, J. Z., Halachev, M. R., Loman, N. J., Constantinidou, C., & Pallen, M. J. (2012). Defining bacterial species in the genomic era: insights from the genus Acinetobacter. BMC Microbiol, 12, 302. doi:10.1186/1471-2180-12- 302 Chivian, D., Brodie, E. L., Alm, E. J., Culley, D. E., Dehal, P. S., DeSantis, T. Z., et al. (2008). Environmental genomics reveals a single-species ecosystem deep within earth. Science, 322(5899), 275. Christner, B. C., Priscu, J. C., Achberger, A. M., Barbante, C., Carter, S. P., Christianson, K., et al. (2014). A microbial ecosystem beneath the West Antarctic ice sheet. Nature, 512(7514), 310-313. doi:10.1038/nature13667 Christner, B. C., Skidmore, M. L., Priscu, J. C., Tranter, M., & Foreman, C. M. (2008). Bacteria in subglacial environments. In M. R, S. F, M. JC, & G. C (Eds.), Psychrophiles: From Biodiversity to Biotechology (pp. 51–71). Berlin: Springer. Csonka, L. N. (1989). Physiological and genetic responses of bacteria to osmotic stress. Microbiol Rev, 53(1), 121-147. D'Amico, S., Collins, T., Marx, J. C., Feller, G., & Gerday, C. (2006). Psychrophilic microorganisms: challenges for life. EMBO Rep, 7(4), 385- 389. doi:10.1038/sj.embor.7400662 De Maayer, P., Anderson, D., Cary, C., & Cowan, D. A. (2014). Some like it cold: understanding the survival strategies of psychrophiles. EMBO Rep, 15(5), 508-517. doi:10.1002/embr.201338170 Doran, P. T., Kenig, F., Knoepfle, J. L., Mikucki, J. A., & Lyons, W. B. (2014). Radiocarbon distribution and the effect of legacy in lakes of the McMurdo Dry Valleys, Antarctica. Limnology and Oceanography, 59(3), 811-826. doi:doi:10.4319/lo.2014.59.3.0811 Dsouza, M., Taylor, M. W., Ryan, J., MacKenzie, A., Lagutin, K., Anderson, R. F., et al. (2014). Paenibacillus darwinianus sp. nov., isolated from gamma- irradiated Antarctic soil. Int J Syst Evol Microbiol, 64(Pt 4), 1406-1411. doi:10.1099/ijs.0.056697-0 Dsouza, M., Taylor, M. W., Turner, S. J., & Aislabie, J. (2014). Genome-based comparative analyses of Antarctic and temperate species of Paenibacillus. PLoS One, 9(10), e108009. doi:10.1371/journal.pone.0108009 Feller, G., & Gerday, C. (1997). Psychrophilic enzymes: molecular basis of cold adaptation. Cell Mol Life Sci, 53(10), 830-841. Feller, G., & Gerday, C. (2003). Psychrophilic enzymes: hot topics in cold adaptation. Nat Rev Microbiol, 1(3), 200-208. doi:10.1038/nrmicro773 Foley, N., Tulaczyk, S., Auken, E., Schamper, C., Dugan, H., Mikucki, J., et al. (2016). Helicopter-borne transient electromagnetics in high-latitude environments: An application in the McMurdo Dry Valleys, Antarctica. GEOPHYSICS, 81(1), WA87-WA99. doi:10.1190/geo2015-0186.1

159

Fountain, A. G., Dana, G. L., Lewis, K. J., Vaughn, B. H., & Mcknight, D. H. (2013). Glaciers of the Mcmurdo Dry Valleys, Southern Victoria Land, Antarctica. In J. C. Priscu (Ed.), Ecosystem Dynamics in a Polar Desert: the Mcmurdo Dry Valleys, Antarctica (Vol. 72`, pp. 65–75). Washington, DC: American Geophysical Union. Fretwell, P., Hamish D. Pritchard, David G. Vaughan, J. L. Bamber, N. E. Barrand, R. Bell, & al, C. B. e. (2013). Bedmap2: improved ice bed, surface and thickness datasets for Antarctica. Cryosphere(7), 375–393. Fricker, H. A., Powell, R., Priscu, J. C., Tulaczyk, S., Anandakrishnan, S., Christner, B. C., et al. (2013). Siple Coast subglacial aquatic environments: the Whillans Ice Stream Subglacial Access Research Drilling Project. In M. J. Siegert & M. C. I. Kennicutt (Eds.), Antarctic Subglacial Aquatic Environments (Vol. 192, pp. 199–219). Washington, DC: American Geophysical Union. Gao, H., Yang, Z. K., Wu, L., Thompson, D. K., & Zhou, J. (2006). Global transcriptome analysis of the cold shock response of Shewanella oneidensis MR-1 and mutational analysis of its classical cold shock proteins. J Bacteriol, 188(12), 4560-4569. doi:10.1128/JB.01908-05 Gauthier, M. J., Lafay, B., Christen, R., Fernandez, L., Acquaviva, M., Bonin, P., & Bertrand, J. C. (1992). Marinobacter hydrocarbonoclasticus gen. nov., sp. nov., a new, extremely halotolerant, hydrocarbon-degrading marine bacterium. Int J Syst Bacteriol, 42(4), 568-576. doi:10.1099/00207713-42- 4-568 Gounot, A. M. (1991). Bacterial life at low temperature: physiological aspects and biotechnological implications. Journal of Applied Bacteriology, 71(5), 386- 397. doi:doi:10.1111/j.1365-2672.1991.tb03806.x Grimaud, R. (2010). Marinobacter. In K. N. Timmis (Ed.), Handbook of Hydrocarbon and Lipid Microbiology (pp. 1289–1296). Berlin: Springer. Grzymski, J. J., Murray, A. E., Campbell, B. J., Kaplarevic, M., Gao, G. R., Lee, C., et al. (2008). Metagenome analysis of an extreme microbial symbiosis reveals eurythermal adaptation and metabolic flexibility. Proc Natl Acad Sci U S A, 105(45), 17516-17521. doi:10.1073/pnas.0802782105 Handley, K. M., Hery, M., & Lloyd, J. R. (2009). Marinobacter santoriniensis sp. nov., an arsenate-respiring and arsenite-oxidizing bacterium isolated from hydrothermal sediment. Int J Syst Evol Microbiol, 59(Pt 4), 886-892. doi:10.1099/ijs.0.003145-0 Handley, K. M., & Lloyd, J. R. (2013). Biogeochemical implications of the ubiquitous colonization of marine habitats and redox gradients by Marinobacter species. Front Microbiol, 4, 136. doi:10.3389/fmicb.2013.00136 Hartl, F. U., & Hayer-Hartl, M. (2009). Converging concepts of protein folding in vitro and in vivo. Nat Struct Mol Biol, 16(6), 574-581. doi:10.1038/nsmb.1591

160

Hebraud, M., & Potier, P. (1999). Cold shock response and low temperature adaptation in psychrotrophic bacteria. J Mol Microbiol Biotechnol, 1(2), 211-219. Heermann, R., & Jung, K. (2012). K+ supply, osmotic stress and the KdpD/KdpE two–component system. In G. R & B. D (Eds.), Two–Component Systems in Bacteria (pp. 181–198). Norfolk: Caister Academic Press. Hille, F., & Charpentier, E. (2016). CRISPR-Cas: biology, mechanisms and relevance. Philos Trans R Soc Lond B Biol Sci, 371(1707). doi:10.1098/rstb.2015.0496 Hochhut, B., Wilde, C., Balling, G., Middendorf, B., Dobrindt, U., Brzuszkiewicz, E., et al. (2006). Role of pathogenicity island-associated integrases in the genome plasticity of uropathogenic Escherichia coli strain 536. Mol Microbiol, 61(3), 584-595. doi:10.1111/j.1365-2958.2006.05255.x Hoffmann, A., Bukau, B., & Kramer, G. (2010). Structure and function of the molecular chaperone Trigger Factor. Biochim Biophys Acta, 1803(6), 650- 661. doi:10.1016/j.bbamcr.2010.01.017 Hubbard, A., Lawson, W., Anderson, B., Hubbard, B., & Blatter, H. (2004). Evidence for subglacial ponding across Taylor Glacier, Dry Valleys, Antarctica. Annals of Glaciology, 39, 79-84. doi:10.3189/172756404781813970 Huu, N. B., Denner, E. B., Ha, D. T., Wanner, G., & Stan-Lotter, H. (1999). Marinobacter aquaeolei sp. nov., a halophilic bacterium isolated from a Vietnamese oil-producing well. Int J Syst Bacteriol, 49 Pt 2, 367-375. doi:10.1099/00207713-49-2-367 Jung, D. O., Achenbach, L. A., Karr, E. A., Takaichi, S., & Madigan, M. T. (2004). A gas vesiculate planktonic strain of the purple non-sulfur bacterium Rhodoferax antarcticus isolated from Lake Fryxell, Dry Valleys, Antarctica. Arch Microbiol, 182(2-3), 236-243. doi:10.1007/s00203-004-0719-8 Kandror, O., & Goldberg, A. L. (1997). Trigger factor is induced upon cold shock and enhances viability of Escherichia coli at low temperatures. Proceedings of the National Academy of Sciences of the United States of America, 94(10), 4978-4981. Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., et al. (2012). Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics, 28(12), 1647-1649. doi:10.1093/bioinformatics/bts199 Kelly, L. (2011). Psychrophile. In G. M, A. R, Q. JC, C. HJ, I. WM, P. DL, & V. M (Eds.), Encyclopedia of Astrobiology (pp. 1385–1386). Berlin: Springer. Kim, M., Oh, H.-S., Park, S.-C., & Chun, J. (2014). Towards a taxonomic coherence between average nucleotide identity and 16S rRNA gene sequence similarity for species demarcation of prokaryotes. International Journal of Systematic and Evolutionary Microbiology, 64(2), 346-351. doi:doi:10.1099/ijs.0.059774-0

161

Konstantinidis, K. T., Ramette, A., & Tiedje, J. M. (2006). The bacterial species definition in the genomic era. Philos Trans R Soc Lond B Biol Sci, 361(1475), 1929-1940. doi:10.1098/rstb.2006.1920 Konstantinidis, K. T., & Tiedje, J. M. (2005). Genomic insights that advance the species definition for prokaryotes. Proceedings of the National Academy of Sciences of the United States of America, 102(7), 2567-2572. doi:10.1073/pnas.0409727102 Kuhn, E. (2012). Toward understanding life under subzero conditions: the significance of exploring psychrophilic "cold-shock" proteins. Astrobiology, 12(11), 1078-1086. doi:10.1089/ast.2012.0858 Kushner, D. (1978). Life in high salt and solute concentrations: halophilic bacteria. In K. DJ (Ed.), Microbial Life in Extreme Environments. (pp. 317– 368). London: Academic Press. Lanyi, B. (1987). Classical and rapid identification methods for medically important bacteria. In C. RR & G. R (Eds.), Methods in Microbiology (Vol. 19, pp. 1–68). Orlando: Academic Press Inc. Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P. A., McWilliam, H., et al. (2007). Clustal W and Clustal X version 2.0. Bioinformatics, 23(21), 2947-2948. doi:10.1093/bioinformatics/btm404 Lauro, F. M., Allen, M. A., Wilkins, D., Williams, T. J., & Cavicchioli, R. (2011). Psychrophiles: Genetics, Genomics, Evolution. In K. Horikoshi (Ed.), Extremophiles Handbook (pp. 865-890). Tokyo: Springer Japan. Lee, P. A., Mikucki, J. A., Foreman, C. M., Priscu, J. C., DiTullio, G. R., Riseman, S. F., et al. (2004). Thermodynamic constraints on microbially mediated processes in lakes of the McMurdo Dry Valleys, Antarctica. Geomicrobiology Journal, 21(3), 221-237. doi:10.1080/01490450490275884 Levy, J. S., Fountain, A. G., Welch, K. A., & Lyons, W. B. (2012). Hypersaline “wet patches” in Taylor Valley, Antarctica. Geophysical Research Letters, 39(5). doi:doi:10.1029/2012GL050898 Liu, C., Chen, C. X., Zhang, X. Y., Yu, Y., Liu, A., Li, G. W., et al. (2012). Marinobacter antarcticus sp. nov., a halotolerant bacterium isolated from Antarctic intertidal sandy sediment. Int J Syst Evol Microbiol, 62(Pt 8), 1838-1844. doi:10.1099/ijs.0.035774-0 Lyons, W. B., Welch, K. A., Snyder, G., Olesik, J., Graham, E. Y., Marion, G. M., & Poreda, R. J. (2005). Halogen geochemistry of the McMurdo dry valleys lakes, Antarctica: Clues to the origin of solutes and lake evolution. Geochimica et Cosmochimica Acta, 69(2), 305-323. doi:https://doi.org/10.1016/j.gca.2004.06.040 Makarova, K. S., Wolf, Y. I., Alkhnbashi, O. S., Costa, F., Shah, S. A., Saunders, S. J., et al. (2015). An updated evolutionary classification of CRISPR–Cas systems. Nature Reviews Microbiology, 13, 722. doi:10.1038/nrmicro3569 https://www.nature.com/articles/nrmicro3569#supplementary-information

162

Margesin, R., & Miteva, V. (2011). Diversity and ecology of psychrophilic microorganisms. Res Microbiol, 162(3), 346-361. doi:10.1016/j.resmic.2010.12.004 Markowitz, V. M., Chen, I. M., Palaniappan, K., Chu, K., Szeto, E., Pillay, M., et al. (2014). IMG 4 version of the integrated microbial genomes comparative analysis system. Nucleic Acids Res, 42(Database issue), D560-567. doi:10.1093/nar/gkt963 Marr, A. G., & Ingraham, J. L. (1962). Effect of temperature on the composition of fatty acids in Escherichia coli. J Bacteriol, 84(6), 1260-1267. Martin, S., Marquez, M. C., Sanchez-Porro, C., Mellado, E., Arahal, D. R., & Ventosa, A. (2003). Marinobacter lipolyticus sp. nov., a novel moderate halophile with lipolytic activity. Int J Syst Evol Microbiol, 53(Pt 5), 1383- 1387. doi:10.1099/ijs.0.02528-0 Marx, J. G., Carpenter, S. D., & Deming, J. W. (2009). Production of cryoprotectant extracellular polysaccharide substances (EPS) by the marine psychrophilic bacterium Colwellia psychrerythraea strain 34H under extreme conditions. Can J Microbiol, 55(1), 63-72. doi:10.1139/W08-130 Mavromatis, K., Ivanova, N. N., Chen, I. M., Szeto, E., Markowitz, V. M., & Kyrpides, N. C. (2009). The DOE-JGI standard operating procedure for the annotations of microbial genomes. Stand Genomic Sci, 1(1), 63-67. doi:10.4056/sigs.632 Meier-Kolthoff, J. P., Göker, M., Spröer, C., & Klenk, H.-P. (2013). When should a DDH experiment be mandatory in microbial taxonomy? Archives of Microbiology, 195(6), 413-418. doi:10.1007/s00203-013-0888-4 Methé, B. A., Nelson, K. E., Deming, J. W., Momen, B., Melamud, E., Zhang, X., et al. (2005). The psychrophilic lifestyle as revealed by the genome sequence of Colwellia psychrerythraea 34H through genomic and proteomic analyses. Proceedings of the National Academy of Sciences of the United States of America, 102(31), 10913-10918. doi:10.1073/pnas.0504766102 Mikucki, J. A., Auken, E., Tulaczyk, S., Virginia, R. A., Schamper, C., Sørensen, K. I., et al. (2015). Deep groundwater and potential subsurface habitats beneath an Antarctic dry valley. Nature Communications, 6, 6831. doi:10.1038/ncomms7831 https://www.nature.com/articles/ncomms7831#supplementary-information Mikucki, J. A., Foreman, C. M., Sattler, B., Berry Lyons, W., & Priscu, J. C. (2004). Geomicrobiology of Blood Falls: an iron-rich saline discharge at the terminus of the Taylor Glacier, Antarctica. Aquatic Geochemistry, 10(3), 199-220. doi:10.1007/s10498-004-2259-x Mikucki, J. A., Pearson, A., Johnston, D. T., Turchyn, A. V., Farquhar, J., Schrag, D. P., et al. (2009). A contemporary microbially maintained subglacial ferrous "ocean". Science, 324(5925), 397-400. doi:10.1126/science.1167350

163

Mikucki, J. A., & Priscu, J. C. (2007). Bacterial diversity associated with Blood Falls, a subglacial outflow from the Taylor Glacier, Antarctica. Appl Environ Microbiol, 73(12), 4029-4039. doi:10.1128/AEM.01396-06 Morita, R. Y. (1975). Psychrophilic bacteria. Bacteriol Rev, 39(2), 144-167. Murray, A. E., Kenig, F., Fritsen, C. H., McKay, C. P., Cawley, K. M., Edwards, R., et al. (2012). Microbial life at −13 °C in the brine of an ice-sealed Antarctic lake. Proceedings of the National Academy of Sciences, 109(50), 20626. Nagradova, N. K. (2008). Peptidyl–prolyl cis/trans isomerases as protein folding catalysts endowed with additional functions. In N. K. Nagradova (Ed.), Foldases: Enzymes Catalyzing Protein Folding (pp. 51–103). New York: Nova Science Publishers, Inc. Nam, K. H., Haitjema, C., Liu, X., Ding, F., Wang, H., DeLisa, M. P., & Ke, A. (2012). Cas5d protein processes pre-crRNA and assembles into a cascade-like interference complex in subtype I-C/Dvulg CRISPR-Cas system. Structure, 20(9), 1574-1584. doi:10.1016/j.str.2012.06.016 Niederberger, T. D., Perreault, N. N., Tille, S., Lollar, B. S., Lacrampe-Couloume, G., Andersen, D., et al. (2010). Microbial characterization of a subzero, hypersaline methane seep in the Canadian High Arctic. ISME J, 4(10), 1326-1339. doi:10.1038/ismej.2010.57 Ogier, J. C., Calteau, A., Forst, S., Goodrich-Blair, H., Roche, D., Rouy, Z., et al. (2010). Units of plasticity in bacterial genomes: new insight from the comparative genomics of two bacteria interacting with invertebrates, Photorhabdus and Xenorhabdus. BMC Genomics, 11, 568. doi:10.1186/1471-2164-11-568 Oren, A. (2002). The cellular structure of halophilic microorganisms. In O. A (Ed.), Halophilic Microorganisms and Their Environments (pp. 69–123). Dordrecht: Kluwer Academic Publishers. Oren, A. (2006). Life at high salt concentrations. In D. M, F. S, R. E, S. KH, & S. E (Eds.), The Prokaryotes: Volume 2: Ecophysiology and Biochemistry (3rd ed., Vol. 2, pp. 263–282). Singapore: Springer Science and Business Media, LLC. Oren, A. (2008). Microbial life at high salt concentrations: phylogenetic and metabolic diversity. Saline Systems, 4, 2-2. doi:10.1186/1746-1448-4-2 Overmann, J. (2006). Principles of enrichment, isolation, cultivation, and preservation of prokaryotes. In D. M & F. S (Eds.), The Prokaryotes: Vol. 1: Symbiotic Associations, Biotechnology, Applied Microbiology (3rd ed., Vol. 1, pp. 20–136). Singapore: Springer Science and Business Media, LLC. Papke, R. T., de la Haba, R. R., Infante-Dominguez, C., Perez, D., Sanchez- Porro, C., Lapierre, P., & Ventosa, A. (2013). Draft genome sequence of the moderately halophilic bacterium Marinobacter lipolyticus strain SM19. Genome Announc, 1(4). doi:10.1128/genomeA.00379-13

164

Peeters, K., Hodgson, D. A., Convey, P., & Willems, A. (2011). Culturable diversity of heterotrophic bacteria in Forlidas Pond (Pensacola Mountains) and Lundstrom Lake (Shackleton Range), Antarctica. Microb Ecol, 62(2), 399-413. doi:10.1007/s00248-011-9842-7 Perreault, N. N., Andersen, D. T., Pollard, W. H., Greer, C. W., & Whyte, L. G. (2007). Characterization of the prokaryotic diversity in cold saline perennial springs of the Canadian high Arctic. Appl Environ Microbiol, 73(5), 1532-1543. doi:10.1128/AEM.01729-06 Perreault, N. N., Greer, C. W., Andersen, D. T., Tille, S., Lacrampe-Couloume, G., Lollar, B. S., & Whyte, L. G. (2008). Heterotrophic and autotrophic microbial populations in cold perennial springs of the high arctic. Appl Environ Microbiol, 74(22), 6898-6907. doi:10.1128/AEM.00359-08 Pfeifer, F. (2006). Gas vesicles of archaea and bacteria. In S. JM (Ed.), Complex Intracellular Structures in Prokaryotes (Vol. 2, pp. 115–140). Berlin: Springer. Phadtare, S. (2004). Recent developments in bacterial cold-shock response. Curr Issues Mol Biol, 6(2), 125-136. Phadtare, S., & Inouye, M. (2008). Cold–shock proteins. In R. Margesin, F. Schinner, J. C. Marx, & C. Gerday (Eds.), Psychrophiles: from Biodiversity to Biotechnology (pp. 191–209). Berlin, Heidelberg: Springer. Picardal, F. (2012). Abiotic and microbial interactions during anaerobic transformations of Fe(II) and [Formula: see text]. Front Microbiol, 3, 112. doi:10.3389/fmicb.2012.00112 Piette, F., Struvay, C., & Feller, G. (2011). The protein folding challenge in psychrophiles: facts and current issues. Environ Microbiol, 13(8), 1924- 1933. doi:10.1111/j.1462-2920.2011.02436.x Priscu, J., & Christner, B. (2004). Earth’s icy biosphere. In B. AT (Ed.), Microbial Diversity and Bioprospecting (pp. 130–145). Washington, DC: ASM Press. Pruesse, E., Peplies, J., & Glockner, F. O. (2012). SINA: accurate high- throughput multiple sequence alignment of ribosomal RNA genes. Bioinformatics, 28(14), 1823-1829. doi:10.1093/bioinformatics/bts252 Qiu, Y., Kathariou, S., & Lubman, D. M. (2006). Proteomic analysis of cold adaptation in a Siberian permafrost bacterium--Exiguobacterium sibiricum 255-15 by two-dimensional liquid separation coupled with mass spectrometry. Proteomics, 6(19), 5221-5233. doi:10.1002/pmic.200600071 Rabus, R., Ruepp, A., Frickey, T., Rattei, T., Fartmann, B., Stark, M., et al. (2004). The genome of Desulfotalea psychrophila, a sulfate-reducing bacterium from permanently cold Arctic sediments. Environ Microbiol, 6(9), 887-902. doi:10.1111/j.1462-2920.2004.00665.x Riley, M., Staley, J. T., Danchin, A., Wang, T. Z., Brettin, T. S., Hauser, L. J., et al. (2008). Genomics of an extreme psychrophile, Psychromonas ingrahamii. BMC Genomics, 9, 210. doi:10.1186/1471-2164-9-210

165

Russell, N. (2008). Membrane components and cold sensing. In R. Margesin, F. Schinner, J. C. Marx, & C. Gerday (Eds.), Psychrophiles: From Biodiversity to Biotechology (pp. 177–188). Berlin: Springer. Samarkin, V. A., Madigan, M. T., Bowles, M. W., Casciotti, K. L., Priscu, J. C., McKay, C. P., & Joye, S. B. (2010). Abiotic nitrous oxide emission from the hypersaline Don Juan Pond in Antarctica. Nature Geoscience, 3, 341. doi:10.1038/ngeo847 https://www.nature.com/articles/ngeo847#supplementary-information Sanford, R. A., Wagner, D. D., Wu, Q., Chee-Sanford, J. C., Thomas, S. H., Cruz-Garcia, C., et al. (2012). Unexpected nondenitrifier nitrous oxide reductase gene diversity and abundance in soils. Proc Natl Acad Sci U S A, 109(48), 19709-19714. doi:10.1073/pnas.1211238109 Shivaji, S., Gupta, P., Chaturvedi, P., Suresh, K., & Delille, D. (2005). Marinobacter maritimus sp. nov., a psychrotolerant strain isolated from sea water off the subantarctic Kerguelen islands. Int J Syst Evol Microbiol, 55(Pt 4), 1453-1456. doi:10.1099/ijs.0.63478-0 Siddiqui, K. S., Williams, T. J., Wilkins, D., Yau, S. A., Michelle A.; Brown, V.;, M., et al. (2013). Psychrophiles. Annual Review of Earth and Planetary Sciences, 41, 87-115. Siegert, M. J., Ellis-Evans, J. C., Tranter, M., Mayer, C., Petit, J. R., Salamatin, A., & Priscu, J. C. (2001). Physical, chemical and biological processes in Lake Vostok and other Antarctic subglacial lakes. Nature, 414(6864), 603- 609. doi:10.1038/414603a Soding, J., Biegert, A., & Lupas, A. N. (2005). The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Res, 33(Web Server issue), W244-248. doi:10.1093/nar/gki408 Spigel, R. H., & Priscu, J. C. (1996). Evolution of temperature and salt structure of Lake Bonney, a chemically stratified Antarctic lake. Hydrobiologia, 321(3), 177-190. doi:10.1007/bf00143749 Stackebrandt, E., Frederiksen, W., Garrity, G. M., Grimont, P. A., Kampfer, P., Maiden, M. C., et al. (2002). Report of the ad hoc committee for the re- evaluation of the species definition in bacteriology. Int J Syst Evol Microbiol, 52(Pt 3), 1043-1047. doi:10.1099/00207713-52-3-1043 Stackebrandt, E., & Goebel, B. M. (1994). Taxonomic note: a place for DNA-DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. International Journal of Systematic and Evolutionary Microbiology, 44(4), 846-849. doi:doi:10.1099/00207713-44- 4-846 Straub, K. L., Benz, M., & Schink, B. (2001). Iron metabolism in anoxic environments at near neutral pH. FEMS Microbiology Ecology, 34(3), 181- 186. doi:doi:10.1111/j.1574-6941.2001.tb00768.x Straub, K. L., Benz, M., Schink, B., & Widdel, F. (1996). Anaerobic, nitrate- dependent microbial oxidation of ferrous iron. Appl Environ Microbiol, 62(4), 1458-1460.

166

Suzuki, Y., Haruki, M., Takano, K., Morikawa, M., & Kanaya, S. (2004). Possible involvement of an FKBP family member protein from a psychrotrophic bacterium Shewanella sp. SIB1 in cold-adaptation. Eur J Biochem, 271(7), 1372-1381. doi:10.1111/j.1432-1033.2004.04049.x Takai, K., Moyer, C. L., Miyazaki, M., Nogi, Y., Hirayama, H., Nealson, K. H., & Horikoshi, K. (2005). Marinobacter alkaliphilus sp. nov., a novel alkaliphilic bacterium isolated from subseafloor alkaline serpentine mud from Ocean Drilling Program Site 1200 at South Chamorro Seamount, Mariana Forearc. Extremophiles, 9(1), 17-27. doi:10.1007/s00792-004-0416-1 Tamura, K., Stecher, G., Peterson, D., Filipski, A., & Kumar, S. (2013). MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0. Molecular Biology and Evolution, 30(12), 2725-2729. doi:10.1093/molbev/mst197 Tang, C., Madigan, M. T., & Lanoil, B. (2013). Bacterial and archaeal diversity in sediments of west Lake Bonney, McMurdo Dry Valleys, Antarctica. Appl Environ Microbiol, 79(3), 1034-1038. doi:10.1128/AEM.02336-12 Ting, L., Williams, T. J., Cowley, M. J., Lauro, F. M., Guilhaus, M., Raftery, M. J., & Cavicchioli, R. (2010). Cold adaptation in the marine bacterium, Sphingopyxis alaskensis, assessed using quantitative proteomics. Environmental Microbiology, 12(10), 2658-2676. doi:doi:10.1111/j.1462- 2920.2010.02235.x Touchon, M., & Rocha, E. P. (2007). Causes of insertion sequences abundance in prokaryotic genomes. Mol Biol Evol, 24(4), 969-981. doi:10.1093/molbev/msm014 Trubl, G. (2013). Insights into the origin of N2O in Lake Vida brine. (Master’s thesis), University of Nevada, Reno, Valdes, J., Pedroso, I., Quatrini, R., Dodson, R. J., Tettelin, H., Blake, R., 2nd, et al. (2008). Acidithiobacillus ferrooxidans metabolism: from genome sequence to industrial applications. BMC Genomics, 9, 597. doi:10.1186/1471-2164-9-597 Vezzi, A., Campanaro, S., D'Angelo, M., Simonato, F., Vitulo, N., Lauro, F. M., et al. (2005). Life at depth: Photobacterium profundum genome sequence and expression analysis. Science, 307(5714), 1459-1461. doi:10.1126/science.1103341 Walsby, A. E. (1994). Gas vesicles. Microbiological Reviews, 58(1), 94-144. Ward, B., Granger, J., Maldonado, M., & Wells, M. (2003). What Limits Bacterial Production in the Suboxic Region of Permanently Ice-Covered Lake Bonney, Antarctica? (Vol. 31). Ward, B. B., & Priscu, J. C. (1997). Detection and characterization of denitrifying bacteria from a permanently ice-covered Antarctic Lake. Hydrobiologia, 347(1), 57-68. doi:10.1023/a:1003087532137 White, D. (2007). Homeostasis. In W. D. (Ed.), The Physiology and Biochemistry of Prokaryotes (3rd ed., pp. 405–416). New York: Oxford University Press. Zhang, D. C., Li, H. R., Xin, Y. H., Chi, Z. M., Zhou, P. J., & Yu, Y. (2008). Marinobacter psychrophilus sp. nov., a psychrophilic bacterium isolated

167 from the Arctic. Int J Syst Evol Microbiol, 58(Pt 6), 1463-1466. doi:10.1099/ijs.0.65690-0

168

I. Tables

Table 6.1. Comparison of phenotypic characteristics between strain BF04_CF4 and closely related cultured Marinobacter species. Isolation Geographic Cell size Nitrate Temperature %NaCl pH GC source location (µm) utilization optimum and optimum and optimum content range (°C) range (w/v) and range (%) Marinobacter sp. Subglacial Blood Falls, Average: Yes 15 and 0–20 6.0 and 0.8– 7.0 and 54.63 strain BF04_CF4 outflow Antarctica 0.5 × 1.7 15.0 6.0–10.0 Marinobacter sp. Perennially Lake Bonney, ND Yes 12–15* and 1.8–3.5 and ND 54.34 strain ELB17 ice-covered Antarctica ND ND [99.2%] (Ward lake and Priscu 1997) Marinobacter Sea-ice Canadian Basin in 0.3–0.4 × Yes 16–18 and 0– ND and 2.0– 6.0–9.0 and 53.84 psychrophilus Arctic Ocean 1.0–2.5 22 8.0 5.0–10.0 20041T [97.8%] (Zhang et al. 2008) Marinobacter Antarctic Chinese Antarctic 0.5–0.8 × Yes 25 and 4–35 3.0–4.0 and 7.0 and 54.86 antarcticus ZS2– intertidal Zhongshan 1.5–2.3 0.0–25.0 5.0–10.5 30T [96.3%] (Liu sandy Station on the et al. 2012) sediment Larsemann Hills Marinobacter Saline soil Cádiz, Spain 0.3–0.5 × No 37 and 15–40 7.5 and 1.0– 7.5 and 56.76 lipolyticus SM19T 2.5–3.5 15.0 5.0–10.0 [94.5%] (Martin et al. 2003) Percent 16S rRNA gene sequence identity to strain BF04_CF4 are in brackets. *Resolution (variation) of temperatures did not allow for optimum temperature identification. G + C number of bases determined from genome statistics in the Integrated Microbial Genomes & Microbiomes (IMG/M) system. ND: no data available.

169

Table 6.2. Cellular fatty acid composition (%) of strain BF04_CF4 and related species of the Marinobacter genus.

Marinobacter Marinobacter Marinobacter antarcticus Marinobacter sp. strain psychrophilus ZS2–30T (96.3%) lipolyticus strain BF04_CF4 20041T (97.8%) SM19 (95.1%) 16:1 ω7c 33.1 22.2 - 22.0 18:1 ω9c 27.5 13.4 14.8 18.3 16:0 13.0 14.0 15.9 15.7 12:0 3OH 6.5 10.2 7.5 7.5 18:1 ω7c 5.9 1.6 - 11.4 18:0 4.4 1.2 1.2 3.0 12:0 3.0 4.4 5.2 5.9 17:1 ω8c 1.1 8.6 2.5 0.8 17:0 0.6 7.8 0.9 0.3 16:1 ω9c 0.0 8.8 10.1 6.3 Percent 16S rRNA gene sequence identities to strain BF0F_CF4 are in parentheses. The mean of two replicate fatty acid profiles from BF04_CF4 and SM19 grown at 15°C grown to mid log phase is presented. M. psychrophilus 20041T was grown at 15°C for 5 days (Zhang et al. 2008) and CM. antarcticus ZS2–30T was grown at 16°C for 5 days (Liu et al. 2012). -: No data reported.

170

Table 6.3. Carbon sources utilized by strain BF04_CF4 and its closest characterized relatives.

Marinobacter sp. Marinobacter Marinobacter antarcticus ZS2– strain BF04_CF4 psychrophilus 20041T 30T (96.3%) (Zhang et al. 2010) (97.8%) (Zhang et al. 2008) Acetate + ND ND L-alanine + - + D-alanine + - - D-arabinose - + - Glycerol - + - L-lactate - - + L-leucine + - - D-serine - + - L-serine - - + Sucrose - + - Marinobacter psychrophilus 20041T and antarcticus ZS2–30T were isolated from the Arctic and Antarctic, respectively. Percent 16S rRNA gene sequence identities to strain BF04_CF4 are in parentheses. +: positive for growth; -: negative for growth. ND: No data available.

171

II. Figures

Figure 6.1. Scanning electron microscope image of Marinobacter sp. strain BF04_CF4 showing rod-shaped cells and lack of flagella. Cells have an average length and width of ~1.7 and 0.5 μm, respectively. The scale bar is 1 μm.

172

Figure 6.2. Maximum likelihood phylogenetic tree showing the relationship between 16S rRNA genes from strain BF04_CF4, closely related Marinobacter clones and isolates, and select Marinobacter type strains. Strain BF04_CF4 is indicated by the black circle; strain BF14_3D is indicated by the black diamond; Marinobacter clones isolated from Blood Falls are indicated by the black squares; Marinobacter type strains are indicated by the white diamonds. Bootstrap values greater than 50 (1000 replicates) are indicated at nodes. The scale bar represents 0.01× nucleotide substitutions per sequence position. The outgroup was Shewanella sp. strain BF02_Schw from Blood Falls, Antarctica (DQ677870).

173

Figure 6.3. Syntenic dotplot comparing Marinobacter sp. strains BF04_CF4 (query) and M. psychrophilus (reference) genomes. Colored lines and dots represent regions of similarity between the genomes, and line breaks represent sites with genome variation. Blue points for regions of similarity found on parallel strands and red points for regions of similarity found on antiparallel strands. Gaps, inversions and deletions as well as small areas of repeated sequence are present. IMG uses MUMmer to generate dotplot comparisons.

174

Figure 6.4. Comparison of Clusters of Orthologous Groups (COGs) between BF04_CF4 and M. psychrophilus. Abundance of genes assigned to COG functional categories in IMG for BF04_CF4 (2453 genes) and M. psychrophilus (3108 genes). BF04_CF4 had 1646 genes that were not found in COG; M. psychrophilus had 982. Genome annotations and the identification of common genes across organisms in IMG based on three gene clustering methods: COG, Pfam and IMG orthologue clusters (Mavormatis et al. 2009).

175

Figure 6.5. Effect of temperature (A), NaCl concentration (B), and pH (C) on the growth rates of strain BF04_CF4. Fitted curves were created from averages of specific growth rates under each condition, and open circles represent three replicates of the growth rates. Subzero temperature growth rates were predicted using the Weibull distribution (4 parameter) in SigmaPlot. Modeled temperature value is indicated by an X.

176

Figure 6.6. Amino acid usage of predicted proteomes of Marinobacter sp. strain BF04_CF4 and Marinobacter lipolyticus SM19T. Data for each organism are shown for all predicted proteins and for only those predicted proteins found in each genome by reverse BLAST. Ali: aliphatic, Aro: aromatic, Neg: negative, Pos: positive.

177

Figure 6.7. Amino acid substitution matrix from the alignments of helicase proteins in Marinobacter sp. strain BF04_CF4 and Marinobacter lipolyticus SM19T.

178

Figure 6.8. Amino acid substitution matrix from the alignments of all homologous proteins in Marinobacter sp. strain BF04_CF4 and Marinobacter lipolyticus SM19T.

179

Figure 6.9. Summary of the main features discussed in this text from the genomic analysis of BF04_CF4. Text color indicates the potential involvement with the following processes. Blue: cold adaptation; red: salt adaptation; purple: cold and salt adaptations; green: metabolic capabilities.

180

VITA

Michelle J. Chua was born in Decatur, Georgia and grew up in Stone Mountain, Georgia. She graduated from Chamblee High School and subsequently attended the University of Georgia in Athens, Georgia, where she received degrees in Microbiology and German, a language she has not had practice with for several years and thus has forgotten much of. Following a semester off, Michelle started graduate school at Georgia State University, where she worked with Dr. Kuk- Jeong Chin and attempted to isolate and characterize anaerobic, sulfate- reducing, aromatic compound-degrading bacteria from petroleum-contaminated sites in Brunswick, Georgia and the Gulf of Mexico. After receiving her Master’s degree, she spent a semester working as a waitress at Olive Bistro, the food from which she often still misses. In August 2012, she started her PhD at the University of Tennessee in Knoxville. Until August 2015, she worked with Dr. Jill Mikucki, during which she generated and analyzed enough data (with the help of her coauthors) to publish a paper on the characterization of the physiology and genome of a bacterial isolate from Antarctic subglacial outflow. Starting in August 2015, she joined and continued working in the lab of Alison Buchan, where she performed experiments necessary to write the chapters of this dissertation. Her defense is scheduled for October 12, 2018.

181