Chromosomal organization of biosynthetic gene clusters, including those of nine novel species, suggests plasticity of myxobacterial specialized metabolism

Introduction Natural products discovered from bacteria provide critically needed therapeutic leads for drug discovery, and myxobacteria are an established source for metabolites with unique chemical scaffolds and biological activities. Myxobacterial genomes accommodate an exceptional number and variety of biosynthetic gene clusters (BGCs) which encode for features involved in specialized metabolism. Methods In this study, we describe the collection, sequencing, and genome mining of 20 myxobacteria isolated from rhizospheric soil samples collected in North America. Results Nine isolates were determined to be novel species of myxobacteria including representatives from the genera Archangium, Myxococcus, Nannocystis, Polyangium, Pyxidicoccus, Sorangium, and Stigmatella. Growth profiles, biochemical assays, and descriptions were provided for all proposed novel species. We assess the BGC content of all isolates and observe differences between Myxococcia and Polyangiia clusters. Discussion Continued discovery and sequencing of novel myxobacteria from the environment provide BGCs for the genome mining pipeline. Utilizing complete or near-complete genome sequences, we compare the chromosomal organization of BGCs of related myxobacteria from various genera and suggest that the spatial proximity of hybrid, modular clusters contributes to the metabolic adaptability of myxobacteria.

Rhizospheric soil samples collected from shrubs and trees were screened for bacterial swarms using standard prey-baiting and filter paper degradation methods (Mohr et al., 2016Mohr, 2018) to isolate environmental myxobacteria (Supplementary Figure 1) (Adaikpoh et al., 2020). Morphology screening of visible swarms facilitated the isolation of myxobacteria from multiple genera with a specific focus on lesser-studied myxobacteria. A total of 20 environmental isolates of putative myxobacteria including 8 agarolytic isolates were obtained as monocultures (Figure 1). Lesser-studied myxobacteria include genera with agarolytic phenotypes such as Nannocystis, Polyangium, and Sorangium, hence numerous agarolytic isolates with similar morphologies were advanced for genome sequencing (Mohr, 2018). We have previously discussed 4 of the 20 environmental isolates (SCHIC03, SCPEA02, NCCRE02, and NCSPR01) (Ahearne et al., 2021). Genome sequencing of all isolates provided five complete genomes, seven draft genomes with ≤3 contigs, three draft genomes with 5-8 contigs, and five lower-quality genome assemblies with ≤44 contigs (Table 1). Genome sizes ranged from 9,459,689 to 13,831,693 Mb, and GC content varied from 68.1 to 71.5%. High-quality assemblies enabled subsequent whole-genome comparison approaches for phylogenetic analysis and assessment of biosynthetic gene cluster content and organization.

Phylogenetic relationships of isolated myxobacteria
Initial phylogenetic analysis using 16S rRNA sequences of type strain myxobacteria [obtained from the List of Prokaryotic names with Standing in Nomenclature (LPSN)] suggested the environmental isolates included 1 Archangium, 5 Corallococcus, 3 Myxococcus, 6 Nannocystis, 1 Polyangium, 2 Pyxidicoccus, 1 Sorangium, and 1 Stigmatella (Supplementary Figure 2). Utilizing genome data from isolates and the type strain of myxobacteria, sequence similarities were determined using average nucleotide identity (ANI) and digital DNA-DNA hybridization values (dDDH) according to the established methods for the taxonomic assignment of myxobacteria (Chambers et al., 2020;Livingstone et al., 2020). Resulting ANI and dDDH values indicated 10 of the 20 environmental isolates to be novel species with values below the respective cutoffs of 95% and 70% when compared to most similar type strains (Figure 2 and Supplementary Table 1). Isolate MIWBW is most phylogenetically similar to Archangium gephyra DSM2261 T and A. gephyra Cbvi76 [previously referred to as Cystobacter violaceus Cbvi76 (Stevens et al., 2014)] ( Figure 2A). Of the four published type strain of Archangium, A. gephyra DSM2261 T is currently the only representative sufficiently sequenced for comparative genomic analysis. More rigorous analysis comparing sequenced representatives of closely related Cystobacter and Melittangium reinforced MIWBW as a novel species. This analysis also revealed Cystobacter gracilis DSM 14753 T to be an outlier within the three genera with ANI values below 77.5 for all included representatives. Isolate SCHIC03 is most phylogenetically similar to Myxococcus stipitatus DSM 14675 T when compared to eight Myxococcus-type strains ( Figure 2B). As observed by Chambers et al. (2020) the ANI value between the established type strain species Myxococcus xanthus DSM 16526 T and Myxococcus virescens DSM 2260 T is above the threshold for novel species. Initial 16S rRNA analysis suggested that environmental isolates RBIL2, FL3, BB15-2, and NCELM were all novel Nannocystis species. However, of the three Nannocystis-type strains, there was no genome data for Nannocystis pusilla DSM 14622 T (also referred to as N. pusilla Na p29 T ). Subsequent sequencing of N. pusilla DSM 14622 T and comparison including our genome data for N. pusilla DSM 14622 T revealed RBIL2 to be a subspecies of N. pusilla that is slightly above the novel species threshold ( Figure 2E). Isolates BB15-2, FL3, and NCELM are most phylogenetically similar to Nannocystis exedens DSM 71 T and are significantly distinct from each other. Our proposed addition of three Nannocystis doubles the current member total. Isolate RJM3 is most phylogenetically similar to Polyangium fumosum DSM 14688 T ( Figure 2C). However, only 3 of 10 Polyangium-type strains have sufficient 16S rRNA and genome sequence data suitable for thorough analysis (Lang and Reichenbach, 2013;Wang et al., 2021). Isolate SCPEA02 is most phylogenetically similar to Pyxidicoccus caerfyrddinensis CA032A T ( Figure 2G) (Chambers et al., 2020). The proposed addition to Pyxidicoccus will make SCPEA02 only the fourth type of Pyxidicoccus strain. Isolate WIWO2 is most phylogenetically similar to Sorangium cellulosum Soce56, but no type strain of Sorangium has been sufficiently sequenced for comparative genomics ( Figure 2D). Alternatively, 16S RNA . /fmicb. . analysis suggests that WIWO2 is most phylogenetically similar to Sorangium kenyense Soce 375 T (Supplementary Figure 2). Isolate NCWAL01 is most phylogenetically similar to Stigmatella aurantiaca DSM17044 T and St. aurantiaca DW4_3-1. Interestingly, our analysis indicates ANI values above the threshold for novel species for all three Stigmatella-type strains ( Figure 2F). Environmental isolates NCSPR01 and NCRR are highly similar subspecies of Corallococcus coralloides DSM 2259 T (Supplementary Table 1 and Supplementary Figure 3A). As previously suggested by Ahearne et al. (2021), isolate NCCRE02 is a subspecies of Corallococcus exiguus DSM 14696 T (Supplementary Table 1 and Supplementary Figure 3A). Isolate BB12-1 is likely a subspecies of Corallococcus terminator CA054A T , and isolate BB11-1 is potentially a novel species of Corallococcus (Supplementary Table 1 and Supplementary Figure 3B). However, fragmented genome assemblies for BB11-1 and BB12-1 limited our confidence in precise taxonomic placement. Isolate MISCRS is a subspecies of Myxococcus fulvus DSM 16525 T , and isolate NMCA is a subspecies of M. xanthus DSM 16526 T (Supplementary Table 1 and Supplementary Figure 3C). Isolate MSG2 is a subspecies of Py. caerfyrddinensis CA032A T (Supplementary Table 1 and Supplementary Figure 3D). Isolate SCPEA04 is a Nannocystis highly similar to NCELM, and isolates RBIL2 and ILAH1 are both subspecies of N. pusilla DSM 14622 T (Supplementary Table 1 and Supplementary Figure 3E).

Physiological and biochemical analyses of nine novel genomospecies
All isolated strains swarmed on VY/2 media, and growth characteristics at various pH values and temperatures were analyzed for all nine novel species (Table 2). All nine strains grew at 25-30 • C, and SCPEA02 grew at temperatures up to 40 • C. Growth at pH 7 was observed in all strains, and SCHIC03, NCELM, and SCPEA02 all grew at pH 6-9. Agarolytic strains include BB15-2, WIWO2, FL3, NCELM, and RJM3. Metabolic activity was assessed for all strains (Table 3), and none were able to reduce nitrate or metabolize arginine, glucose, or urea. All strains were able to hydrolyze esculin, and all except FL3 and WIWO2 hydrolyzed .
/fmicb. . gelatin. SCPEA02 and SCHIC03 were the only strains that did not exhibit alkaline phosphatase activity. MIWBW was the only strain to demonstrate both trypsin and a-chymotrypsin activity and possessed overlapping characteristics with A. gephyra (Lang et al., 2015). The growth and activity of SCHIC03 were most similar to M. stipitatus and M. fulvus (Chambers et al., 2020). The growth profiles and biochemical activities of BB15-2, FL3, and NCELM were similar to those of other Nannocystis; however, all three demonstrated comparatively limited pH-dependent growth (Mohr et al., 2018a). Unlike Nannocystis konarekensis, N. exedens, and N. pusilla, none of the Nannocystis strains grew at pH 10. Temperature and pH-dependent growth ranges for RJM3 were notably different from those of other Polyangium, which all grow at temperatures above 30 • C and pH ranges of 6-8.5 (Wang et al., 2021). The growth profile and biochemical activity of SCPEA02 were closely aligned with the reported activities of Py. caerfyrddinensis (Chambers et al., 2020). The growth profile and biochemical activity of WIWO2 overlapped somewhat with those of the recently described Sorangium species (Mohr et al., 2018b). The characterization and description of all Stigmatella-type strains pre-date the present description methodology; however, St. aurantiaca and NCELM have similar growth and biochemical profiles (Kleinig and Reichenbach, 1969;Reichenbach et al., 1969).

Biosynthetic potential of myxobacterial isolates
AntiSMASH analysis of BGCs in all sequenced isolates provided notable differences in BGC contents, sizes, and similarities with previously characterized clusters (Medema et al., 2011;Blin et al., 2021b). A total of 735 BGCs were predicted from 20 genome assemblies, and only 36 were identified by antiSMASH to be fragmented clusters (∼5%) with the vast majority of fragmented BGCs included in sequenced Corallococcus strains (20 fragmented BGCs). Genome data from sequenced Archangium, Myxococcus, Polyangium, Pyxidicoccus, Sorangium, and Stigmatella provided three or fewer fragmented BGCs total from each genus. Of our isolated strains, MIWBW had the most predicted BGCs with 52 total (zero fragmented), and NMCA1 had the least with 23 total (zero fragmented). The average length of BGCs from all 20 sequenced isolates was ∼56 kb. Clusters from Nannocystis strains were significantly shorter (average size ∼30 kb) than Corallococcus, Myxococcus, and Pyxidicoccus clusters ( Figure 3A). Clusters from Pyxidicoccus strains were significantly longer than those from Archangium, Nannocystis, Polyangium, and Sorangium clusters. Interestingly, Myxococcia clusters were significantly longer than Polyangiia BGCs ( Figure 3B).  All identified BGCs were compared with the 1,225,071 BGCs and 29,955 gene cluster families (GCFs) included in the BiG-FAM database (Kautsar et al., 2021a,b). Utilizing a previously established clustering threshold (T = 900) to determine distance from database GCFs, we evaluated our 735 BGCs for similarity with BiG-FAM BGCs. Clusters below the arbitrary threshold have similarities with BiG-FAM GCFs and are likely less novel than clusters above the threshold Waschulin et al., 2022). Clusters from Pyxidicoccus strains (SCPEA02 and MSG2) had the highest average distance (1165), and somewhat predictably Myxococcus strains had the lowest (824) with an average distance below the threshold ( Figure 3C). Average distance of Pyxidicoccus BGCs was significantly higher than the average distances of Corallococcus, Myxococcus, and Nannocystis clusters. Although Nannocystis and Polyangium are lesser-studied myxobacteria, the average distances of clusters from members of each genus were just above the threshold for novelty (923 and 924, respectively). No significant difference was observed between the average distances of Myxococcia and Polyangiia clusters ( Figure 3D).
Of the 735 predicted BGCs, 384 clusters (∼50%) had distances above the threshold. The removal of clusters below the threshold revealed differences in the remaining cluster types across genera (Figure 4). Subsequent comparison of these 384 BGCs with cluster similarities identified during antiSMASH analysis revealed that 52 clusters were either highly homologous to characterized clusters deposited in the MiBIG database  or included embedded clusters with high similarity to known clusters. For example, the myxochelin BGC was found to be embedded in 10 clusters that scored above the BiG-FAM threshold (Gaitatzis et al., 2001;Li et al., 2008). Although co-clustering likely impedes the analysis of novelty and similarity to BiG-FAM GCFs, we suggest that .
or a cluster with high similarity to it is present in all strains excluding members of the class Polyangiia and NCWAL01.

Genomic organization of BGCs
AntiSMASH analysis of complete or near-complete genome data from FL3, MIWBW, MSG2, NCRR, NCSPR03, NMCA1, SCHIC03, and SCPEA02 provided contiguous sequence data sufficient to observe the genome organization of BGCs. Cluster data from related myxobacteria and complete genome data from the antiSMASH database were used to compare BGC organization between related strains (Figures 6, 7). Similarities between BGC content and genome organization were observed between subspecies ( Figures 6A, B) and related strains within the same genus ( Figures 6C, 7). Biosynthetic gene clusters were dispersed throughout all genomes, and cluster-dense genomic segments were observed during the comparison of BGC organization. Notably, cluster-dense segments include hybrid-type clusters such as PKS-NRPS clusters or clusters including more than one cluster type. The myxochelin and myxoprincomide BGCs are located within a cluster-dense region in all sequenced environmental Myxococcia (Supplementary Figure 8). Clusters highly similar to carotenoid, geosmin, and VEPE/AEPE/TG-1 BGCs are often located in less-dense segments of analyzed chromosomes. Chromosomal segments with increased adjacency of hybrid and modular clusters were observed for all analyzed myxobacteria albeit less apparent in FL3 and N. exedens DSM 71 T ( Figure 7C). Other than small differences in cluster content between analyzed subspecies, such as the absence of the myxovirescin BGC from .
/fmicb. . Ordinary one-way ANOVA with multiple comparisons was used to determine the significance of genus-level analysis, and Welch's t-test was used to determine the significance of class-level analysis (for genus-level analysis Archangium n = , Corallococcus n = , Myxococcus n = , Nannocystis n = , Polyangium n = , Pyxidicoccus n = , Sorangium n = , Stigmatella n = ; p < . and class-level analysis Myxococcia n = , Polyangiia n = ; p < . ). Asterisks indicate the associated p values included in the figure descriptions. Figure 6A), numerous inversions of clusters resulting in changes in cluster organization were observed. Apparent BGC inversions were predominantly located within or near cluster-dense regions, and inversions were often relegated to core biosynthetic genes of single clusters with proximal genes unchanged between strains (Supplementary Figure 9). Additional synteny analysis of strains with observed BGC inversions using a set of 10 homologous housekeeping genes revealed highly similar genome organization with no observed inversions (Supplementary Figures 10, 11) (Veltri et al., 2016).

Proposal of nine novel species from the seven genera of Myxococcota
We propose nine candidate strains to represent novel species in the genera Archangium, Myxococcus, Nannocystis, Polyangium, Pyxidicoccus, Sorangium, and Stigmatella. Comparative genomics including differences in genome content, phylogeny, and biosynthetic capacities, as well as physiological and biochemical analyses support the following distinctions: Archangium lansinium

Myxococcota taxonomy
Myxobacteria are excellent resources for the discovery of therapeutics and are suggested to be keystone taxa influencing polymicrobial community structure in soil (Herrmann et al., 2017;Baltz, 2019;Bader et al., 2020;Perez et al., 2020;Petters et al., 2021). Recent discoveries of novel Corallococcus, Myxococcus, and Pyxidicoccus species as well as species from lesser-studied genera indicate an abundance of uncharacterized myxobacteria (Mohr et al., 2012(Mohr et al., , 2018aIizuka et al., 2013;Garcia et al., 2014Garcia et al., , 2016Yamamoto et al., 2014;Sood et al., 2015;Awal et al., 2016Awal et al., , 2017Moradi et al., 2017;Garcia and Muller, 2018;Chambers et al., 2020;Livingstone et al., 2020;Wang et al., 2021Wang et al., , 2022Zhou et al., 2021;Babadi et al., 2022). Our investigation of rhizospheric soil samples provided 20 environmental myxobacteria including 9 proposed novel species. As an initial attempt to isolate myxobacteria from the soil, we were surprised by the effectiveness of morphology screening to enable the discovery of myxobacteria from a variety of genera. We suspect that improved genome data will clarify the observed discrepancies in type strain differentiation and recommend that high-quality genome data be provided for all newly described type strain myxobacteria. We demonstrate that established comparative genome analysis thresholds for the designation of novel species  (Chambers et al., 2020;Wang et al., 2022). However, we note the significant differences in BGC content between Myxococcus and Pyxidicoccus strains included in this analysis. Overall, our phylogenetic analysis provides further support for comparative genomic approaches to identify and classify myxobacteria. The primary limitation is the absence of quality genome data for established type strains within genera such as Archangium, Polyangium, and Sorangium.

Expansion of the genus Nannocystis
The current type strain Nannocystis includes N. exedens DSM 71 T , N. konarekensis DSM 104509 T , and N. pusilla DSM 14622 T . Our investigation resulted in the discovery of an additional three proposed type strains doubling the current total of Nannocystis (Mohr et al., 2018a). We also report genome data for all three proposed type strain species as well as three Nannocystis subspecies and N. pusilla DSM 14622 T . When compared to other myxobacteria, the analysis of all sequenced Nannocystis and our isolates provided notable differences in BGC content, such as the absence of a cluster similar to the myxochelin BGC, the presence of multiple siderophore and phosphonate cluster types, and smaller cluster sizes. The resulting genome data for an additional eight Nannocystis will improve future efforts to characterize and describe the members of this underexplored genus of myxobacteria.

FIGURE
Spatial organization of the myxochelin BGC across all isolates with cluster homology. Gene names assigned by homology to the mxyochelin BGC deposited in the MiBIG database (BGC ), and ribbons indicate shared gene identity between clusters. Image generated using CAGECAT (version . ) using an identity threshold of . (van den Belt et al., ).

Genomic organization of BGCs and adaptability of specialized metabolism
Afforded complete or near-complete genome sequence data, we report the first comparative analysis of spatial organization of myxobacterial BGCs. Dissimilar from Streptomyces localization of BGCs in the extremities of linear genomes (Karoonuthaisiri et al., 2005;Lioy et al., 2021), clusters were distributed throughout the circular genomes of myxobacteria. Observed cluster-rich regions replete with modular BGCs, and noted inversions of biosynthetic genes could, however, contribute to metabolic differentiation similar to how terminal compartments of Streptomyces chromosomes enable spatial reorganization and conditional expression of BGCs during metabolic development and sporulation (Lioy et al., 2021). We suggest that BGC-enriched regions may benefit BGC evolution and contribute to the metabolic adaptability of myxobacteria. Compartmentalization of modulartype clusters with highly homologous domains may benefit module duplication and deletion events associated with the evolution of BGCs (Fischbach et al., 2008;Medema et al., 2014;Chevrette et al., 2020). Vertical inheritance of the myxochelin cluster is apparent in all sequenced Myxococcia. Our data also reveal vertical transfer and the likely concerted evolution of myxoprincomide-type clusters across sequenced Myxococcia (Medema et al., 2014). Alternatively, the presence of the myxovirescin trans-AT PKS cluster within a cluster-rich region of the M. xanthus DK1622 genome and the . /fmicb. . absence of a homologous cluster in NMCA1 indicate horizontal acquisition. Although the absence of the myxovirescin cluster in NMCA1 provides an alternative explanation, the absence of the myxovirescin cluster in all other sequenced Myxococcus and Myxococcota members currently deposited in the antiSMASH database supports horizontal acquisition by DK1622. Regardless, the presence of clusters in DK1622 and their absence in other myxobacteria demonstrate metabolic adaptability among myxobacterial genomes with BGC-enriched segments. Further investigation of the chromosomal organization of BGCs in myxobacteria is required to determine functional impacts on metabolic adaptability and cluster evolution.
The type strain (MIWBW T = TSD-326 T = NCCB 100916 T ) was isolated from soil collected in Summer 2021 from the roots of a white basswood tree near the city of Lansing, Michigan, USA (42.73 • N, 84.48 • W).
Vegetative cells glide on solid media. Cells grow as slightly orange swarms and develop rounded, stalking fruiting bodies .
The type strain (BB15-2 T = NCCB 100934 T ) was isolated from soil collected in Summer 2020 from the roots of a blueberry bush near the city of Bainbridge Island, Washington, USA (47.65 • N, -122.55 • W).
The type strain (FL3 T = TSD-332 T = NCCB 100918 T ) was isolated from soil collected in Fall 2020 from the roots of a southern live oak near the city of Palm Coast, Florida, USA (29.59 • N, −81.21 • W).
The type strain (NCELM T = NCCB 100919 T ) was isolated from soil collected in Spring 2020 from the roots of an elm tree near the city of Asheville, North Carolina, USA (35.63 • N, −82.55 • W).
The type strain (RJM3 T = NCCB 100920 T ) was isolated from soil collected in Fall 2020 from the roots of a red Japanese maple near the village of Mundelein, Illinois, USA (42.26 • N, −88.0 • W).
The type strain (SCPEA02 T = TSD-328 T = NRRL B-65670 T = NCCB 100921 T ) was isolated from soil collected in Spring 2020 from the roots of a peach tree near Parkway Farm in Landrum, South Carolina, USA (35.14 • N, −82.12 • W).
The type strain (NCWAL01 T = TSD-329 T = NCCB 100923 T ) was isolated from soil collected in Spring 2020 from the roots of a walnut tree near the city of Asheville, North Carolina, USA (35.63 • N, −82.55 • W).

Isolation of myxobacteria
Bacteriolytic myxobacteria were isolated by the Escherichia coli baiting method (Mohr, 2018). Briefly, an E. coli lawn was grown overnight at 37 • C and resuspended in 1 mL of an antifungal solution (250 µg/mL of cycloheximide and nystatin). A 300 µl of the solution was spread across a WAT agar (1.5% agar, 0.1% CaCl 2 ) plate and air-dried. Previously air-dried soil was wetted with the antifungal solution to a mud-like consistency, and a pea-sized amount was placed on the dried E. coli WAT plate. The plate was incubated at 25 • C for up to 4 weeks. After 3 days of incubation, the plates were checked daily for the appearance of lytic zones or fruiting bodies in the E. coli lawn. Using a syringe needle, the lytic zones were moved to a plate of VY/4 (Baker's yeast 2.5 g/L, CaCl 2 × 2H 2 O 1.36 g/L, vitamin B 12 0.5 mg/L, and agar 15 g/L). The swarm edge was repeatedly used to inoculate a fresh VY/4 plate until pure cultures were obtained. Isolates were cultivated continuously at 25-30 • C on VY/4. Isolation of cellulolytic myxobacteria was accomplished using the filter paper method (Mohr, 2018). A small square of autoclaved filter paper was placed in the center of a ST21 agar plate (1 g/L of K 2 HPO 4 , 20 mg/L of yeast extract, 14 g/L of agar, 1 g/L of KNO 3 , 1 g/L of MgSO 4× 7H 2 O, 1 g/L of CaCl 2 × 2H 2 O, 0.1 g/L of MnSO 4× 7H 2 O, and 0.2 g/L of FeCl 3 ). A peasized amount of soil, wet with the antifungal solution, was placed at the edge of the filter paper. Plates were incubated at 25 • C for up to 2 months. After 2 weeks, the plates were checked every 2 days for cellulose degradation and fruiting body formation. Fruiting bodies were moved to a fresh ST21 plate with filter paper repeatedly until pure monocultures were observed. Isolates were cultivated continuously at 25-30 • C on VY/4.

Cultivation of isolates
All isolates were maintained on VY/4 media. Growth in liquid cultures was achieved using CYH/2 media (0.75 g/L of casitone, 0.75 g/L of yeast extract, 2g/L of starch, 0.5 g/L of soy flour, 0.5 g/L of glucose, 0.5 g/L of MgSO 4× 7H 2 O, 1 g/L of CaCl 2 × 2H 2 O, 6 g/L of HEPES, 8 mg/L of EDTA-Fe, and 0.5 mg/L of vitamin B 12 ).

Microscopy
A Zeiss stereo discovery.V12 microscope using Axiocam 105 and a Plano Apo S 1.0X objective was used to observe fruiting bodies and swarming patterns.

Enzymatic assays
Enzymatic activity was assessed for myxobacteria utilizing commercial API ZYM (bioMérieux, France) and API NE (bioMérieux, France) kits. Each isolate strain was suspended in 0.85% NaCl to an OD 600 of 0.7 and 0.1 for API NE. API ZYM strips were incubated for 4.5 h at 37 • C, and API NE strips were incubated for 24 h at 37 • C. After incubation, specific reagents were added to the cupule and evaluated according to the manufacturer's instructions.

Growth conditions
For most of the myxobacteria tested, strains were grown on VY/4 (pH 7.2) for 5 to 7 days and resuspended in deionized water to an OD 600 of 0.5. For the genera Nannocystis and Polyangium, strains were grown for 5 to 7 days in CYH/2 media, centrifuged, washed, and resuspended in sterile distilled water. The optimal growth temperature was tested by inoculating VY/4 plates with 25 µl of the 0.5 OD 600 suspension for the given myxobacteria. Plates were incubated at 20, 25, 30, 35, and 40 • C for up to 14 days. Optimal pH was assessed by plating the 0.5 OD 600 solution on VY/4 plates buffered to pH 5, 6, 7, 8, 9, or 10. The pH conditions were buffered at a pH of 5 to 6 with 25 mM MES buffer, 7 to 8 with 25 mM HEPES buffer, and 9 to 10 with 25 mM TRIS buffer in VY/4 plates and incubated at 25 • C for 2 weeks. Comparisons of swarm diameters were used to determine optimal growth conditions.

BGC analysis
FASTA files for all sequenced isolates were uploaded for analysis using antiSMASH (version 6.1.1) using relaxed detection strictness with all extra features toggled on (Blin et al., 2021b). Resulting antiSMASH job IDs from analyzed isolates were submitted as queries using the BiG-FAM database v1.0.0 (1,225,071 BGCs and 29,955 GCFs) to assess BGC similarity to database clusters . All BGCs with >900 distance from model GCFs were subsequently dereplicated manually to remove characterized BGCs not clustered with GCFs within the BiG-FAM database. The antiSMASH database v3.0 (147,571 BGCs) was used to analyze BGCs from myxobacteria . BiG-SCAPE v1.1.0 was used to analyze all BGCs.gbk files from sequenced isolates as well as all .gbk files from myxobacteria with sequenced genomes deposited in the antiSMASH database with the "hybrids-off " and "MiBIG" parameters Navarro-Munoz et al., 2020).

Author contributions
AA, KP, and TK: isolation of environmental myxobacteria. KP, TK, and MH: growth profiles, biochemical assays, and imaging for isolates. AA, KP, and SD: genome sequencing. AA, KP, and DS: BGC analyses, manuscript preparation, and editing. DS: supervision and administration. All authors have read and approved the final manuscript.

Funding
This research was supported by the National Institute of Allergy and Infectious Diseases (1 R15 AI137996) and the National Institute of General Medical Sciences (1 P20 GM130460).
The reviewer DW is currently organizing a Research Topic with the author DS.

Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.