Volcanic Soils as Sources of Novel CO-Oxidizing Paraburkholderia and Burkholderia: Paraburkholderia hiiakae sp. nov., Paraburkholderia metrosideri sp. nov., Paraburkholderia paradisi sp. nov., Paraburkholderia peleae sp. nov., and Burkholderia alpina sp. nov. a Member of the Burkholderia cepacia Complex

Previous studies showed that members of the Burkholderiales were important in the succession of aerobic, molybdenum-dependent CO oxidizing-bacteria on volcanic soils. During these studies, four isolates were obtained from Kilauea Volcano (Hawai‘i, USA); one strain was isolated from Pico de Orizaba (Mexico) during a separate study. Based on 16S rRNA gene sequence similarities, the Pico de Orizaba isolate and the isolates from Kilauea Volcano were provisionally assigned to the genera Burkholderia and Paraburkholderia, respectively. Each of the isolates possessed a form I coxL gene that encoded the catalytic subunit of carbon monoxide dehydrogenase (CODH); none of the most closely related type strains possessed coxL or oxidized CO. Genome sequences for Paraburkholderia type strains facilitated an analysis of 16S rRNA gene sequence similarities and average nucleotide identities (ANI). ANI did not exceed 95% (the recommended cutoff for species differentiation) for any of the pairwise comparisons among 27 reference strains related to the new isolates. However, since the highest 16S rRNA gene sequence similarity among this set of reference strains was 98.93%, DNA-DNA hybridizations (DDH) were performed for two isolates whose 16S rRNA gene sequence similarities with their nearest phylogenetic neighbors were 98.96 and 99.11%. In both cases DDH values were <16%. Based on multiple variables, four of the isolates represent novel species within the Paraburkholderia: Paraburkholderia hiiakae sp. nov. (type strain I2T = DSM 28029T = LMG 27952T); Paraburkholderia paradisi sp. nov. (type strain WAT = DSM 28027T = LMG 27949T); Paraburkholderia peleae sp. nov. (type strain PP52-1T = DSM 28028T = LMG 27950T); and Paraburkholderia metrosideri sp. nov. (type strain DNBP6-1T = DSM 28030T = LMG 28140T). The remaining isolate represents the first CO-oxidizing member of the Burkholderia cepacia complex: Burkholderia alpina sp. nov. (type strain PO-04-17-38T = DSM 28031T = LMG 28138T).

Previous studies showed that members of the Burkholderiales were important in the succession of aerobic, molybdenum-dependent CO oxidizing-bacteria on volcanic soils. During these studies, four isolates were obtained from Kilauea Volcano (Hawai'i, USA); one strain was isolated from Pico de Orizaba (Mexico) during a separate study. Based on 16S rRNA gene sequence similarities, the Pico de Orizaba isolate and the isolates from Kilauea Volcano were provisionally assigned to the genera Burkholderia and Paraburkholderia, respectively. Each of the isolates possessed a form I coxL gene that encoded the catalytic subunit of carbon monoxide dehydrogenase (CODH); none of the most closely related type strains possessed coxL or oxidized CO. Genome sequences for Paraburkholderia type strains facilitated an analysis of 16S rRNA gene sequence similarities and average nucleotide identities (ANI). ANI did not exceed 95% (the recommended cutoff for species differentiation) for any of the pairwise comparisons among 27 reference strains related to the new isolates. However, since the highest 16S rRNA gene sequence similarity among this set of reference strains was 98.93%, DNA-DNA hybridizations (DDH) were performed for two isolates whose 16S rRNA gene sequence similarities with their nearest phylogenetic neighbors were 98.96 and 99.11%. In both cases DDH values were <16%. Based on multiple variables, four of the isolates represent novel species within the Paraburkholderia: Paraburkholderia hiiakae sp. nov. (type

INTRODUCTION
Soils have long been recognized as important biological sinks for carbon monoxide (CO), a critical reactant in the troposphere, yet the microbiology of soil CO oxidizers remains largely unstudied (e.g., Bartholomew and Alexander, 1982;Bender and Conrad, 1994;Conrad, 1996;King, 1999). Early work by Conrad (1988) established rates of atmospheric CO uptake by different soil types along with some of the controls of uptake, but those studies did not identify populations that were active in situ or lead to the isolation and characterization of novel CO oxidizers. More recently, molecular ecological approaches have revealed an unexpectedly large diversity of soil aerobic CO oxidizers, including members of the phylum Actinobacteria and class Ktedonobacteria (phylum Chloroflexi), as well as diverse α-, β-, and δ-Proteobacteria and Euryuarchaeota (e.g., King, 2003a;Dunfield and King, 2004;King and Weber, 2007;Weber and King, 2010a,b;Quiza et al., 2014;King and King, 2014a,b;McDuff et al., 2016).
Recent work with north temperate deciduous forest soils has been particularly interesting, since it has revealed that a group of δ-Proteobacteria related to the myxobacterium, Haliangeum ochraceum, plays significant roles in high-affinity atmospheric CO uptake (Quiza et al., 2014). Haliangeum ochraceum is intriguing itself, since it harbors the smallest known cox gene operon comprised only of the three structural genes for carbon monoxide dehydrogenase (CODH) and one accessory gene. Whether this or aspects of its CODH structure are related to its capacity of atmospheric CO uptake is unknown. Likewise, whether this group accounts for atmospheric CO uptake in other soils is unknown.
Previous work with a rapidly developing forest colonizing volcanic cinders identified members of the β-Proteobacteria, and Burkholderiales in particular, as important contributors to the CO-oxidizing community, a community that as a whole was involved with rapid atmospheric CO consumption (King, 2003b;King, 2010b, 2012). During these studies several novel CO oxidizers were isolated and their ability to consume atmospheric CO was established. Based on a battery of molecular, biochemical, and physiological analyses, we describe here these and a related isolate as new species within the genera Burkholderia and Paraburkholderia (Yabuuchi et al., 1992;Sawana et al., 2014). We propose the new isolates as potential models for understanding atmospheric CO oxidation by a widely distributed group of terrestrial β-Proteobacteria.

Isolation
Paraburkholderia isolates DNBP6-1 T , I2 T , PP52-1 T , and WA T were isolated by Weber and King (2012) from enrichments initiated with forest soil from a 1959 tephra deposit (Pu'u Puai) located on Kilauea Volcano (19 • 24 ′ 22.5 ′′ N × 155 • 15 ′ 18.2 ′′ W); this site has been described previously (King, 2003b;Gomez-Alvarez et al., 2007;. Burkholderia isolate PO-04-17-38 T was isolated by F.A. Rainey (University of Alaska, Anchorage) and colleagues from enrichments initiated with soil obtained above the tree line at an altitude of about 4357 m on Pico de Orizaba (Mexico) a dormant stratovolcano (see Callegan et al., 2008 for additional details on the site).
Three strains (WA T , I2 T , and PP52-1 T ) were enriched in basal salts media (King, 2003a) with various carbon sources (WA T , xylose; I2 T pyruvate; PP52-1 T ; mannose). All enrichments were assayed for CO oxidation after adding CO (100 ppm final concentration) to the headspaces of sealed 160-ml serum bottles, and monitoring headspace concentrations at intervals using gas chromatography (King, 1999). Cultures that oxidized CO were used to inoculate solidified versions of the basal salts media used for the original enrichments. Individual colonies were selected from plates, and used to inoculate small volumes of liquid media that were monitored for CO oxidation using gas chromatography as before. Cultures that oxidized CO were purified further by plating, colony selection, and transfer to liquid media; additional CO uptake assays were conducted as necessary. Strain DNBP6-1 T was isolated similarly, except that nutrient broth (0.8 g L −1 ) containing penicillin (500 µg ml −1 ) was used for enrichment. Strain PO-04-17-38 T was isolated using a medium comprised of 10% R2A (DSMZ medium 830) at 15 • C without regard to its capacity for CO oxidation. It was subsequently identified as a CO oxidizer after screening as above. CO uptake capacity for the isolates was assayed following King and King (2014a) with stationary phase liquid cultures (10 ml in 160-ml serum bottles) amended with approximately 200-ppm headspace CO concentrations; cultures were incubated at 30 • C with shaking at 200 rpm; cell protein concentrations were determined at the end of the uptake assays using a kit based on bicinchoninic acid (Pierce Protein Research Products; Thermo Scientific).
The SINA alignment tool (Pruesse et al., 2007) was used to align isolate 16S rRNA gene sequences with sequences derived from their closest phylogenetic neighbors and related taxa [determined from the EZtaxon application (Kim et al., 2012)]. Alignments were adjusted manually as necessary using MEGA7 (Tamura et al., 2013). Maximum likelihood analyses were also performed using MEGA7 with a general time reversible model and 100 and 1,000 bootstrap replicates, respectively.
EZTaxon (Kim et al., 2012) was used to obtain 16S rRNA gene sequences similarities for a set of type strains representing the phylogenetic neighborhood of the CO-oxidizing isolates. Genomes for strains with 16S rRNA gene sequence similarities > 98.0% were then used to generate parallel pairwise comparisons of average nucleotide identity (ANI). ANI was calculated using the comparative genome toolkit from the Integrated Microbial Genomes/Microbiome Samples website (https://img.jgi.doe.gov/ cgi-bin/m/main.cgi).
RecA genes were amplified using primers Bur3 (forward) and Bur4 (reverse) to further clarify the phylogenetic positions of isolates DNBP6-1 T , I2 T , PP52-1 T , and WA T ; the PCR protocol followed the methods of Payne et al. (2005). Purified amplicons of the correct size (385 bp) were sequenced bi-directionally as above. MUSCLE was used in the MEGA7 platform (Tamura et al., 2013) to align partial recA gene sequences for the isolates' close phylogenetic neighbors. Phylogenetic analyses were also performed using MEGA7. Sequences were deposited in Genbank with the following accession numbers: DNBP6-1 T , KY305132; I2 T , KY305131; PP52-1 T , KY305133; WA T , KY3051130.

Morphological and Physiological Characterization
Routine microscopy and staining methods were used for basic isolate characterization (Gerhardt et al., 1994). pH ranges suitable for growth were determined by cultivating the isolates in R2A media with pH adjusted between values of 5.5 and 9.5. A phosphate buffer (0.1 M) was used to prepare media with pH values from 5.5 to 6.5; a CO 2 /sodium bicarbonate/sodium carbonate buffer was used to prepare media with pH values from 7.5 to 9.5. Temperature optima were assessed similarly using cultures grown from 5 to 50 • C (R2A medium at pH 6.5).
Sole carbon source metabolism patterns were assessed with Biolog GN2 plates (Biolog, Inc.; Hayward CA, USA) following the manufacturer's recommendations. Sole carbon source assimilation, enzymatic reactions (including oxidase and catalase), nitrate reduction, and other biochemical traits were also assayed with API 20NE strips following the manufacturer's recommendations (bioMérieux SA; Marcy l'Etoile, France). In addition, the ability of isolates to grow with selected sole carbon and energy sources was assessed in liquid culture with the basal salts medium above containing 25 mM of individual carbon sources.

Phospholipid Fatty Acid Characterization
Phospholipid fatty acid analyses for the isolates were carried out by the DSMZ Identification Service using standard extraction and analytical methods (Miller, 1982;Kuykendall et al., 1988). After methylation and gas chromatographic quantitation, individual fatty acids were identified using the standard protocol of the Sherlock Microbial Identification System (MIDI Inc.).

DNA G+C Content and DNA-DNA Hybridization
DNA base composition (mol% G+C) for all strains was also determined by the Identification Service of the DSMZ (Braunschweig, Germany) using the method of Mesbah et al. (1989). The Identification Service of DSMZ performed DNA-DNA hybridizations for two isolates, Paraburkholderia sp. DNBP6-1 T and Paraburkholderia sp. PP52-1 T , with their closest phylogenetic neighbors (P. bryophila LMG 23644 T and P. mimosarum DSM 21841 T , respectively) using the protocols of Cashion et al. (1977), De Ley et al. (1970 and Huss et al. (1983).

RESULTS AND DISCUSSION
Phylogenetic analyses of 16S rRNA gene sequences (Figure 1) showed that the Pico de Orizaba isolate (PO-04-17-38 T ) clustered with the genus Burkholderia. Results from analyses conducted with EZtaxon (Kim et al., 2012) further showed that PO-04-17-38 T was most closely related to B. stabilis LMG 14294 T with a 16S rRNA gene sequence similarity of 97.49%. This level of similarity is considered consistent with species novelty and is less than similarity values (i.e., > 98.7-99.0%) for which DNA-DNA hybridization assays have been proposed for establishing species distinctions (Stackebrandt and Ebers, 2006).
Phylogenetic analyses also showed that the Kilauea Volcano isolates clustered with the genus Paraburkholderia. Pairwise comparisons revealed that each of the isolates shared 16S rRNA gene sequence similarities no greater than 98.1% with other isolates, a difference consistent with species demarcation (Stackebrandt and Ebers, 2006). However, 16S rRNA gene sequence similarities for the isolates and their closest type species phylogenetic neighbors were somewhat higher: 98.53% for Paraburkholderia sp. I2 T vs. P. oxyphila NBRC 105797 T ; 98.80% for Paraburkholderia sp. WA T vs. P. oxyphila NBRC 105797 T ; 98.96% for Paraburkholderia sp. PP52-1 T vs. P. mimosarum DSM FIGURE 1 | Maximum likelihood analysis of partial 16S rRNA gene sequences from CO-oxidizing Paraburkholderia and Burkholderia isolates, and related taxa. Bootstrap support ≥ 50% is indicated below the branches. A discrete Gamma distribution was used to model evolutionary rate differences among sites. All gapped positions and positions with missing or ambiguous data were removed, leaving 1263 positions in the final dataset. Asterisks (*) indicate confirmed CO oxidizers; ‡ indicates isolates with form I CO dehydrogenase genes identified in their genome sequences. 21841 T ; and 99.11% for Paraburkholderia sp. DNBP6-1 T vs. P. bryophila LMG 223644 T .
To determine whether these similarities were consistent with the delineation of novel Paraburkholderia species, 16S rRNA gene sequence similarities were generated using EZTaxon (Kim et al., 2012) for pairs of type species in the phylogenetic neighborhood of the isolates. The maximum sequence similarity in this set of comparisons was 98.93%. For all pairs with similarities ≥ 98.0-98.93%, genome sequences were used to generate a set of average nucleotide identities (ANI; Supplementary Table 1). ANI did not exceed 95% (a recommended lower cutoff for species differentiation; Richter and Rosselló-Móra, 2009;Kim et al., 2014;Yarza et al., 2014) in any comparison (Supplementary Table 1, Supplementary Figure 1). Moreover, a more extensive analysis of ANI and 16S similarities involving multiple genera from multiple phyla yielded comparable resuts (Kim et al., 2014). This observation supports designation of isolates I2 T and WA T as novel species based on similarities with their closest phylogenetic neighbor, P. oxyphila NBRC 105797 T (98.53 and 98.80%, respectively).
Phylogenetic analyses of partial recA gene sequences were also consistent with species level differentiation (Figure 2). Primers Bur3 and Bur4 have been previously shown to discriminate successfully among numerous Burkholderia (Paraburkholderia) species (Payne et al., 2005;Hall et al., 2015). Although the topologies of phylogenetic trees in this study and that of Payne et al. (2005) did not have strong bootstrap support, the overall topologies were consistent with the topologies of phylogenetic trees based on 16S rRNA gene sequences, and the isolates in this study were clearly distinct from related taxa (Figure 2).
BLAST analyses of Burkholderia and Paraburkholderia genomes and previously published culture-based assays of CO oxidation potential (Weber and King, 2012) showed that none of the closest type strain phylogenetic neighbors of the isolates in this study harbored form I carbon monoxide dehydrogenase (cox) genes or oxidized CO. However, genome analyses revealed form I cox genes in P. heleia SA42 T , P. nodosa LMG 23741 T , Paraburkholderia sp. CH1-1, and Paraburkholderia sp. H160. Previous studies have confirmed CO oxidation by P. nodosa LMG 23741 T , P. xenovorans LB400 T , and Paraburkholderia sp. LUP (King, 2003a;Weber and King, 2012). All of these isolates are phylogenetically distinct from those in this study (Figure 1).
All isolates were Gram-negative, non-spore forming, nonmotile, CO-oxidizing rods [CO oxidation capacity was reported previously by Weber and King (2012)]. All were catalase positive, and all but PP52-1 T were oxidase positive ( Table 1). Colonies formed on solid pyruvate-yeast extract media [PYE, (Weber and King, 2012)] were circular with entire margins, with white coloration for I2 T , DNBP6-1 T , and PP52-1 T ; colonies for WA T and PO-04-17-38 T were off-white.
Three isolates (PO-04-17-38 T , PP52-1 T , and WA T ) grew with pH optima at 6.5, while the optimum for the remaining two was 7.5; all grew at pH 5.7 and showed an upper limit of pH 8-pH 8.5 (Table 1). Isolate PO-04-17-38 T grew with a notably lower temperature optimum (25 • C) and minimum (5 • C) than for other isolates in this study, but its response to temperature was similar to that of its closest phylogenetic neighbor, Burkholderia stabilis LMG 14294 T ( Table 1). The remaining isolates grew optimally at 30 • C with a minimum at 15 • C and an upper limit of 40 • -45 • C; similar values were observed for their phylogenetic neighbors ( Table 1).
Positive reactions on Biolog GN-2 plates varied from a low of 20 substrates for PO-04-17-38 T (mostly sugars and a few simple organic acids) to 72 of 95 substrates for DNBP6-1 T (Supplementary Table 2). With the exception of PO-04-17-38 T , the isolates used a variety of sugars, sugar derivatives, organic acids, and amino acids that reflected the broad substrate utilization patterns reported for Paraburkholderia (e.g., Chen et al., 2007;Vandamme et al., 2007b;Compant et al., 2008;Vanlaere et al., 2008b;Aizawa et al., 2011;Otsuka et al., 2011). Nonetheless, substrate use differed for each of the isolates, and for three of the isolates when compared with their phylogenetic neighbors. Previously published BIOLOG GN-2 substrate utilization reactions facilitated comparisons between strains PP52-1 T and P. mimosarum DSM 21841 T , I2 T and P. nodosa LMG 23741 T , and DNBP6-1 T and P. bryophila LMG 23644 T (Supplementary Table 2). At least 10% of the 95 substrate reactions differed in each of these paired comparisons. It must be noted however that strain variability and variability among assays could reduce or increase these differences.
Results from API 20NE strips also revealed differences in substrate assimilation among the CO-oxidizing isolates and some  of their close phylogenetic neighbors. Notably, isolate PO-04-17-38 T did not assimilate any of the substrates in the panel, while its nearest phylogenetic neighbor, B. stabilis LMG 14294 T , assimilated all but mannitol, N-acetylglucosamine and gluconate ( Table 1). The lack of substrate assimilation by PO-04-17-38 T contrasts with its ability to oxidize substrates in Biolog GN2 plates (Supplementary Table 2), and to grow in liquid culture with arabinose, glucose, mannitol, and mannose (Supplementary Table 3). The lack of substrate assimilation by PO-04-17-38 T in the API 20NE panel was repeatable, however, which suggests that assay conditions for the API tests do not reliably reflect the capacity of PO-04-17-38 T to use substrates. Strain I2 T differed from its closest phylogenetic neighbor, P. oxyphila OX-01 T , in its ability to assimilate arabinose, but not adipate, citrate or phenylacetate ( Table 1). Strain PP52-1 T differed from its closest phylogenetic neighbor, P. mimosarum DSM 21841 T , in its ability to assimilate mannose, caprate, adipate, and phenylacetate, but not malate ( Table 1). Strain DNBP6-1 T differed from its closest phylogenetic neighbor, P. bryophila LMG 23644 T , in its inability to assimilate glucose, arabinose, mannose, malate, gluconate, caprate, or citrate ( Table 1). Strain WA T differed from its closest phylogenetic neighbor, P. oxyphila OX-01 T , in its inability to assimilate malate ( Table 1).
The ability of the isolates to grow on various substrates in liquid culture (Supplementary Table 3) largely paralleled observations from Biolog GN2 plates (Supplementary Table 2). Strain PO-041783 T grew with the fewest substrates (11 of 39, mostly sugars and a few organic acids) and was inhibited by several, while the Paraburkholderia isolates grew with 19-22 of 39 substrates. None of the strains were able to grow with glycine, phthalate, or solvents and alcohols, but two strains (I2 T and WA T ) were able to use dimethylamine and trimethylamine.
The isolates obtained in this study share multiple characteristics with members of the genera Burkholderia and Paraburkholderia into which they were placed on the basis of 16S rRNA gene analyses (Figure 1). Although several CO-oxidizing members of Paraburkholderia have been identified previously (King, 2003a), PO-04-17-38 T represents the first CO-oxidizing member of the Burkholderia, and in particular the Burkholderia cepacia complex, a group that harbors a number of important pathogens (Peeters et al., 2016). This observation is notable, since genomic sequencing of a large number of Burkholderia has yet to reveal any putative CO oxidizers, while at least 5 putative CO oxidizers have been identified among the Paraburkholderia. In addition, a phylogenetic analysis has shown that the PO-04-17-38 T coxL gene clusters most closely with coxL from isolate DBNP6-1 T , which suggests that a horizontal gene transfer event from Paraburkholderia to PO-04-17-38 T might account for its apparently unusual capacity to oxidize CO.

Description of Burkholderia alpina sp. nov.
Burkholderia alpina (al.pi'na. L. fem. adj. alpina, pertaining to the Alps and generally from or inhabiting mountainous regions, especially above the tree line, alpina referring to an isolate from an alpine altitude).
Description of Paraburkholderia paradisi sp. nov.