Genomic and phenotypic characterization of Burkholderia isolates from the potable water system of the International Space Station

The opportunistic pathogens Burkholderia cepacia and Burkholderia contaminans, both genomovars of the Burkholderia cepacia complex (BCC), are frequently cultured from the potable water system (PWS) of the International Space Station (ISS). Here, we sequenced the genomes and conducted phenotypic assays to characterize these Burkholderia isolates. All recovered isolates of the two species fall within monophyletic clades based on phylogenomic trees of conserved single-copy core genes. Within species, the ISS PWS strains all demonstrate greater than 99% average nucleotide identity (ANI), suggesting that they are of a highly similar genomic lineage and both individually may have stemmed from the two founding clonal strains before diverging into two unique sub strain populations. No evidence for horizontal gene transfer between the populations was observed. Differences between the recovered isolates can be observed at the pangenomic level, particularly within putative plasmids identified within the B. cepacia group. Phenotypically, the ISS-derived B. cepacia isolates generally exhibit a trend of lower rates and shorter duration of macrophage intracellularization compared to the selected terrestrial reference strain (though not significantly), and significantly lower rates of cellular lysis in 7 of the 19 isolates. ISS-derived B. contaminans isolates displayed no difference in rates of macrophage intracellularization compared to the selected reference, though generally increased rates lysis, with 2 of the 5 significantly increased at 12-hours post inoculation. We additionally find that ISS-isolated B. contaminans display hemolytic activity at 37°C not demonstrated by the terrestrial control, and greater antifungal capacity in the more recently collected isolates. Thankfully, the ISS-derived isolates generally exhibit 1-4 times greater sensitivity to common antibiotics used in their clinical treatments. Thus, despite their infection potential, therapeutic treatment should still have efficacy. Author Summary The International Space Station (ISS) is a unique built environment due to its isolation and recycling of air and water. Both microbes and astronauts inhabit the ISS, and the potential pathogenicity of the former is of great concern for the safety of the latter. The potable water dispenser (PWD) of the potable water system (PWS) on board the ISS was assembled in a cleanroom facility and then primed on Earth using an extensive process to ensure no gas bubbles existed within the lines that could lock the apparatus upon installation in orbit. The primed system sat dormant for 6 months before installation on the ISS. Microbial surveillance was conducted on the system after installation and the bacterial load was 85 CFU/mL, which exceeded the 50 CFU/mL limits set for ISS potable water. Over a microbial surveillance spanning 4.5 years, numerous strains of the potential pathogen Burkholderia have been isolated from the PWD. Here we sequenced and analyzed the genomes of these strains while also characterizing their potential pathogenicity. The genome analysis indicates it is likely that there were only two strains that were introduced on Earth that have subsequently undergone minimal diverging evolution. These strains retain pathogenicity, but remain susceptible to antibiotics, providing a potential therapeutic intervention in the event of infection.

pangenomic level, particularly within putative plasmids identified within the B. cepacia group. 23 Phenotypically, the ISS-derived B. cepacia isolates generally exhibit a trend of lower rates and 24 shorter duration of macrophage intracellularization compared to the selected terrestrial reference 25 strain (though not significantly), and significantly lower rates of cellular lysis in 7 of the 19 26 isolates. ISS-derived B. contaminans isolates displayed no difference in rates of macrophage 27 intracellularization compared to the selected reference, though generally increased rates lysis, with 28 2 of the 5 significantly increased at 12-hours post inoculation. We additionally find that ISS-29 isolated B. contaminans display hemolytic activity at 37°C not demonstrated by the terrestrial 30 control, and greater antifungal capacity in the more recently collected isolates. Thankfully, the 31 ISS-derived isolates generally exhibit 1-4 times greater sensitivity to common antibiotics used in 32 their clinical treatments. Thus, despite their infection potential, therapeutic treatment should still 33 have efficacy. 34 35

Author Summary
The International Space Station (ISS) is a unique built environment due to its isolation and 38 recycling of air and water. Both microbes and astronauts inhabit the ISS, and the potential 39 pathogenicity of the former is of great concern for the safety of the latter. The potable water 40 dispenser (PWD) of the potable water system (PWS) on board the ISS was assembled in a 41 cleanroom facility and then primed on Earth using an extensive process to ensure no gas bubbles 42 existed within the lines that could lock the apparatus upon installation in orbit. The primed system 43 sat dormant for 6 months before installation on the ISS. Microbial surveillance was conducted on 44 the system after installation and the bacterial load was 85 CFU/mL, which exceeded the 50 45 CFU/mL limits set for ISS potable water. Over a microbial surveillance spanning 4.5 years, 46 numerous strains of the potential pathogen Burkholderia have been isolated from the PWD. Here 47 we sequenced and analyzed the genomes of these strains while also characterizing their potential 48 pathogenicity. The genome analysis indicates it is likely that there were only two strains that were 49 introduced on Earth that have subsequently undergone minimal diverging evolution. These strains 50 retain pathogenicity, but remain susceptible to antibiotics, providing a potential therapeutic 51 intervention in the event of infection. 52 53 Introduction 54 55 Microbial surveillance of the surfaces, air, and potable water system (PWS) of the 56 International ISS has been implemented by National Aeronautics and Space Administration 57 (NASA) to ensure crew health within this unique closed environment. These efforts, which use 58 standard culturing techniques, have been conducted over twelve years and 22 missions and began 59 shortly after the potable water dispenser (PWD) was launched on STS-126 in November of 2008 60 . On-orbit operations using the PWD began in early 2009 and continues operations to this day. The 61 organisms Burkholderia cepacia and B. contaminans, both genomovars of the Burkholderia 62 cepacia complex (BCC), are frequently cultured from the PWD of the ISS. Our isolates were 63 collected between January 6, 2010 during mission 22 to August 6, 2014 during mission 40 ( Fig. 1, 64 S1 Table). The PWS in combination with the PWD is a water recycling system that utilizes containing genes linked to virulence and metabolism, quorum sensing, transcriptional regulation, 93 fatty acid biosynthesis, and transposition [13]. Here we characterize genomic and phenotypic 94 properties of 24 Burkholderia isolates derived from the ISS PWD. 95 96

98
The PWD unit of the ISS was assembled in a cleanroom facility and then primed on Earth 99 using an extensive process to ensure no gas bubbles existed within the lines that could lock the 100 apparatus upon installation in orbit. The primed system sat dormant for 6 months before 101 installation on the ISS [1,3]. Microbial surveillance was conducted on the system after installation 102 and the bacterial load was 85 CFU/mL, which exceeded the 50 CFU/mL limits set for ISS potable 103 water, leaving the sole source of water on the US module out-of-order [1]. In the meantime, the 104 Russian system was used as a back-up. The US system was flushed with the biocide iodine (I2), 105 first at what turned out to be a sub-inhibitory concentration of 4ppm, as subsequent measurements 106 revealed an increase in the microbial load [1]. Further testing revealed that 40ppm was the 107 necessary concentration of iodine flush to achieve the drinkable 50 CFU/mL bacterial load [1]. 108 Iodine flushes are still intermittently administered to the system after durations of PWS stagnation. 109 Our focal Burkholderia species are known to survive not only in distilled water, but also in iodine 110 solutions; therefore, these flushes may have reduced the overall microbial load of the PWS while 111 inadvertently selecting for the Burkholderia species within the system. Here we characterized the 112 Burkholderia species isolates obtained from the ISS PWS both genomically and phenotypically. 113 Twenty-four isolates collected over 4.5 years were chosen for sequencing. total GCs: 2,589 with genes contributed by all ("core"; 10.2%); 8,249 with genes contributed by 133 only single genomes ("singletons"; 32.5%); and 14,545 contributed by some mixture ("accessory") 134 (Fig 2A). A total of 25 B. contaminans genomes were incorporated (20 references from NCBI and 135 5 ISS-isolates), with an average gene count of 7,580 ± 539. From these, a total of 13,645 GCs were 136 generated: 3,534 core (25.9%); 3,436 singletons (25.2%); and 6,675 accessory GCs (49.0%; Fig  137  2B). It should be noted the disparity between the number of identified core, singleton, and 138 accessory genes in B. cepacia vs those identified in B. contaminans at this point is not meant to 139 convey anything about potential differences in gene-content diversity between the two species. 140 The number of genomes available/incorporated for each and the breadth of phylogenetic diversity 141 spanned by the incorporated genomes of the two groups both vary greatly. 142 We scanned for functional enrichment or depletion in the ISS-derived isolates as compared 143 to the reference genomes based on normalized frequencies of presence/absence across the two 144 groups (see Methods). It should be kept in mind that given the phylogenetic landscape of both ISS-145 derived groups -with each forming monophyletic clades within their respective species (Fig 1B)

146
-the functional differences observed may be due to evolutionary divergence as a whole, rather 147 than being due to their source of isolation (the ISS  which was the first B. contaminans isolate recovered from the ISS ( Fig 1C); these functional 174 annotations are presented in Table 3. All gene calls, sequences, and annotations are available in 175 S5  contain multiple copies of genes with annotations such as bacteriophage DNA transposition 201 protein, AAA+ family ATPase (s9, s36, s39, s57), Virulence-associated protein-VagC (virulence 202 associated gene C; s16, s28, s35, s36, s39), and additional elements of the T4SS, VirB8 (S6 Table). 203 We additionally see transposases-related genes in the B. cepacia isolates with a mean copy number 204 per putative plasmid of 17.57 ± 5.7 (mean ± 1 SD), and an integrase element with a mean copy 205 number of 2.47 ± 0.51, suggesting DNA rearrangements and duplications among the conjugative 206 plasmids of ISS B. cepacia may be a result of these mobile genetic elements. 207 As for B. contaminans, in addition to harboring elements of both the T4SS and Type II 208 secretion system (T2SS), the plasmids have a soluble lytic murein transglycosylase harboring a 209 putative invasion domain LysM at a mean copy number of 3 ± 2.8. The lytic transglycosylase, 210 LtgG, has recently reported for its role to control cell morphology and virulence in Burkholderia 211 pseudomallei [25]. Furthermore, we see each harboring a copy of the toxin component of the 212 MazEF toxin-antitoxin system suggesting these ISS B. contaminans isolates may be able to control 213 their transition to the dormant persister state [26]. Again, we see transposase-related genes with a 214 mean copy number of 5 ± 5.7. 215

B. cepacia putative-plasmid annotations (19) Annotation (COG)
Mean copy # ± 1 SD  LDH released in the cell culture media at six, eight, twelve and 24 hours post-inoculation (Fig 4). 233 Bacteria which were internalized by macrophage were quantified and reported in colony forming 234 units per milliliter (CFU/mL) measured at six and twelve hours post-inoculation ( Fig 5). 235 At six and eight hours post infection, little LDH-release was observed for both the ISS-236 derived B.cepacia isolates and terrestrial control strain ( Fig 4A). The B. contaminans ISS-derived 237 isolates appear to be triggering cell lysis at a slightly greater rate than the terrestrial control, though 238 with high variability across the experimental triplicates ( Fig 4B)  of triton-X, an example of full lysis (Fig 4). 246 At twelve hours post-infection it becomes apparent that the terrestrial strain of B. cepacia 247 ATCC25416 is able to survive intracellularly within macrophage to a greater degree than any of 248 the ISS isolates (Fig 5) Due to the identification of B. contaminans isolates harboring these two biosynthetic gene 261 clusters within our collection, we assayed for the ability to inhibit growth of the Aspergillus 262 fumigatus AF293 strain ( Fig 6A) and cause hemolysis (Fig 6B). None of the B. cepacia isolates 263 exhibited fungal inhibition or hemolysis, but each of the B. contaminans did to varying degrees. 264 The terrestrial control reference strain (B. contaminans J2956) displayed the least amount of fungal 265 inhibition ( Fig 6A) and little to no hemolysis (Fig 6B), while isolate s47 displayed a similar ability 266 to inhibit fungal growth but an added ability to lyse blood cells after 48 hours of growth. Though 267 time of isolate-recovery does not necessarily indicate total time that isolate spent aboard the ISS, 268 the isolates that display greater antifungal and hemolytic properties than s47 and the terrestrial 269 control were collected at a later date (Fig 1A and 1C an exception as it appears to be more sensitive to co-trimethoprim. All isolates appeared to be 282 moderately resistant to cefotaxime, yet susceptible to ceftazidime, also a 3 rd generation β-lactam 283 cephalosporin. 284 285 9 286 287 288 Biofilm Assay 289 290 The time of collection records show that the Burkholderia species are intermittently cultured 291 from the ISS PWD; there are numerous consecutive flights where Burkholderia could not be 292 cultivated from the PWD. One reason for this could be due to the proclivity for Burkholderia 293 species to form biofilms. This biofilm would be adapted over time to the pressure and flow 294 experienced during water removal from the system as needed for drinking or food hydration. We 295 observed the ability of the B. cepacia isolates to form biofilms to a slightly greater degree than 296 the B. cepacia terrestrial strains used for comparison (Fig 7A). We find that on a whole, B. 297 contaminans form biofilms more readily than B. cepacia, yet the ISS isolates of B. contaminans 298 have a diminishing ability to form biofilm in relation to the B. contaminans reference control 299 strain (Fig 7B). 300 301

303
Burkholderia genomic analyses 304 305 The ISS-isolated B. cepacia and B. contaminans both formed monophyletic clades when 306 placed in a phylogenomic tree with currently available references (Fig 1B). When comparing the 307 ISS isolates among one another, they display very few single nucleotide polymorphisms (SNPs) 308 suggesting low genetic diversity among the respective ISS PWD strains. In a pangenomics 309  and will likely reduce this microbial load. However, we find that the ISS B. cepacia isolates 330 analyzed in this study all harbored a Cu/Ag efflux pump (CusA) at a mean copy number of 3.05 ± 331 3.4 on a conjugative plasmid (Table 4). This suggest that they will able to share this ability to 332 pump silver out from within their membrane with the subpopulations of the B. cepacia bacterial 333 community that does not yet confer this resistance. 334 Similarly, we find a number of the identified enriched functions in our pangenomic analysis 335 are commonly associated with mobile genetic elements that can be found on conjugative elements. 336 A more in-depth plasmid analysis reveals that both species harbor elements of the Type IV 337 secretion system (T4SS) on putative plasmids as well as an enrichment of transposase-related 338 genes (S6 Table) In order to test our hypothesis that these strains have the capacity to be virulent due to their 397 plasmid gene content, we screened the ISS-derived B. cepacia and B. contaminans for the ability 398 to invade and colonize macrophage. The B. cepacia isolates were found to invade macrophage by 399 6 hours, with some (s33) being more effective colonizers than others. They appear to multiply, 400 then escape from the macrophage by 12 hours post-inoculation, possibly using the plasmid-401 encoded lysophospholipase mechanism to rupture the macrophage. Despite not contributing to a 402 longer-term infection, the cytotoxic byproducts generated by lysophospholipase degradation of 403 macrophage may play a physiological role in further stimulating the adhesion and differentiation 404 of lymphoid cells macrophages and activation and recruitment of additional macrophage and T-405 lymphocytes, among other immune response mechanisms [24]. This is in contrast to the B. cepacia 406 ATCC25416 terrestrial control strain, which has been noted for its ability to invade and carry out 407 a long-term colonization of macrophage which can lend to the formation of long-term chronic 408 infections [37]. We find that this strain displays the ability to remain intracellularized at 12-hours 409 post-inoculation (Fig 5) yet remains able to lyse macrophage (Fig 4). Accordingly, the B. cepacia 410 ATCC25416 reference strain contains additional virulence factors within its genome not found in 411 the ISS isolates such as elements of the Type 6 secretion system (T6SS off. This is followed by the addition of 1.5ul of USER Enzyme to the mixture, then incubation in 450 the thermocycler at 37C for 15 minutes with heated lit set to 47C. In order to obtain 700-900 bp 451 inserts, we used SPRIselect beads for a rightside clean-up of 0.25X, where the supernatant is 452 retained for a left side clean up using a 0.25X bead clean up then eluted in 7.5 ul of 0.1X TE buffer. 453 The fragmented, adapter ligated and size selected libraries were then amplified using 12.5 ul of 454 NEBNext Ultra II Q5 Master Mix with 2.5 ul i7 index primer and 2.5 ul i5 Universal PCR primer 455 for a total volume of 25 ul. The amplification was carried out at the following temperatures and 456 times: initial denaturation at 98°C for 30 seconds, followed by 4 cycles of denaturation at 98°C for 457 10 seconds and annealing/extension at 65°C for 75 seconds, and a final extension at 65°C for 5 458 minutes. A total of 58 libraries were generated, quality controlled to find library size using the 459 Agilent Bioanalyzer and high sensitivity DNA chip and double stranded DNA concentration was 460 quantified using Qubit Fluorometric Quantitation. Libraries were normalized to achieve a total of 461 700 pM of pooled library in 200 ul with an average library size of 750 bps.   Each concentric circle radiating out from the center represents an isolate, identified at the top of each next to the year they were isolated. Wrapping around the circles are the generated gene clusters (GCs), where a solid mark for an isolate at a given GC indicates that particular isolate contributed a gene to that gene cluster, and the absence of a solid color indicates that isolate did not contribute a gene to that gene cluster. The very outer layer of each is a blue and yellow line. The blue line highlights core gene clusters. The yellow line highlights accessory GCs. The putative-plasmid pangenome on the right does not include singletons in the visualization.   2  s1  5  s1  6  s1  8  s2  0  s2  2  s2  3  s2  4  s2  5  s2  8  s3  1  s3  3  s3  5  s3  6  s3  7  s3  9  s4  1 .cepacia and (B s9  s12  s15  s16  s18  s20  s22  s23  s24  s25  s28  s31  s33  s34  s35  s36  s37  s39  s41