**1. Introduction**

The phototrophic purple sulfur bacteria are *Gammaproteobacteria* that use sulfide and other reduced sulfur sources as photosynthetic electron donors and oxidize these to sulfate as the final oxidation product. The genus *Ectothiorhodospira* originally was included in the *Chromatiaceae* family [1]. Phenotypic differences and the distinction of *Ectothiorhodospira* from *Chromatiaceae* by oligonucleotide patterns of 16S rRNA molecules [2,3] led to their separation in a distinct family, the *Ectothiorhodospiraceae* [4].

The two families of purple sulfur bacteria can be distinguished by the most obvious differences in the oxidation of sulfide which are microscopically visible, the formation of elemental sulfur globules inside the cells in *Chromatiaceae* and outside the cells in *Ectothiorhodospiracae* [5–8]. Other characteristic properties that distinguish species of both families are different internal membrane systems, which are in the form of vesicles in *Chromatiaceae* and in the form of membrane stacks in *Ectothiorhodospiraceae* [5–7], and by a number of chemotaxonomic properties, including the quinone, lipid and fatty acid composition [9,10] and lipopolysaccharide structures [11–13]. Sequence analysis of the 16S rRNA gene from available type strains of *Ectothiorhodospiraceae* had demonstrated the clear divergence of two groups within this family which had been recognized as two different genera: slightly to moderately halophilic species of *Ectothiorhodospira* and extremely halophilic species, which were transferred to the new genus *Halorhodospira* [14].

**Citation:** Imhoff, J.F.; Kyndt, J.A.; Meyer, T.E. Genomic Comparison, Phylogeny and Taxonomic Reevaluation of the *Ectothiorhodospiraceae* and Description of *Halorhodospiraceae* fam. nov. and *Halochlorospira* gen. nov. *Microorganisms* **2022**, *10*, 295. https://doi.org/10.3390/ microorganisms10020295

Academic Editors: Matthew Sattley and Robert Blankenship

Received: 22 December 2021 Accepted: 23 January 2022 Published: 26 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Further comprehensive phylogenetic studies, including large numbers of phototrophic purple bacteria of *Alpha*-, *Beta*- and *Gammaproteobacteria*, comparing sequences of 16S rRNA genes and *bchXYZ* genes [15], *pufLM* genes [16] and the glycine/sarcosine methyltransferase essential for glycine betaine biosynthesis [17] all supported the recognition of three major groups of purple sulfur bacteria, the *Chromatiaceae*, the *Ectothiorhodospiraceae* (except *Halorhodospira* species) and the extremely halophilic *Halorhodospira* species.

In this communication we compare the genome information of a large number of available strains and species of the *Ectothiorhodospiraceae*, highlight some properties related to sulfur metabolism, electron transport and photosynthesis and discuss the phylogenetic and taxonomic status of these bacteria. The data are in support of a distinction of the *Halorhodospira* species from *Ectothiorhodospiraceae* and *Chromatiaceae* on a family level. The name *Halorhodospiraceae*, fam. nov. is proposed for this new family. *Ectothiorhodospiraceae* and *Halorhodospiraceae* families are defined. In addition, the separation of *Halo-rhodospira halochloris* and *Halorhodospira abdelmalekii* into the new genus *Halochlorospira* gen. nov. as *Halochlorospira halochloris* and *Halochlorospira abdelmalekii* is proposed.

According to genome information, including ANI, most of the chemotrophic genera, which have been assigned in recent years in the literature and in databases to the *Ectothiorhodospiraceae,* should be excluded from this family due to their large phylogenetic distance and significantly different phenotypic properties.

#### **2. Material and Methods**

#### *2.1. Genome Sequences Used in this Study*

Several genome sequences from phototrophic *Ectothiorhodospiraceae* and *Halo-rhodospiraceae* species have been published before [17–23] and a number of additional genome sequences have been established during this study (Table 1). The genome sequences from other strains of phototrophic bacteria used in this study are the following: *Allochromatium humboldtianum* DSM 21881 (JABZEO000000000), *Allochromatium vinosum* DSM 180 (CP001896), *Chromatium okenii* DSM 169 (NRRQ01000000), *Chromatium okenii* LaCa (PPGH01000000), *Halochromatium glycolicum* DSM 11080 (NRSJ01000000), *Halochromatium salexigens* DSM 4395 (NHSF01000000), *Marichromatium bheemlicum* DSM 18632 (JAAXKX000000000), and *Marichromatium gracile* DSM 203 (SMDC01000000). Those from chemotrophic species are: *Acidiferrobacter thiooxydans* m-1 (PSYR01000000), *Alkalilimnicola ehrlichii* MLHE-1 (CP000453), *Alkalispirillum mobile* DSM 12769 (RCDA01000000), *Aquisalimonas asiatica* CGMCC 1.6291 (FOEG01000000), *Arhodomonas aquaeolei* DSM 8974 (ARGF00000000), *Halofilum ochraceum* XJ16 (LVEG02000000), *Halopeptonella vilamensis* DSM 21056 (VMKO01000000), *Inmirania thermothiophila* DSM 100275 (RJVI01000000), *Nitrococcus mobilis* Nb-231 (AAOF00000000), *Oceanococcus atlanticus* 22II-S10r2 (AQQV00000000), *Spiribacter salinus* M19-40 (CP005963), *Thioalbus denitrificans* DSM 26407 (QPJY01000000), *Thioalkalivibrio sulfidophilus* HL-EbGr7 (CP001339), *Thioalkalivibrio versutus* AL2 (MVAR00000000), *Acidihalobacter prosperus* DSM 5130 (JQSG00000000), *Thiogranum longum* DSM 19610 (SMFX01000000), *Thiohalomonas denitrificans* HLD2 (FMWD01000000) and *Thiohalospira halophila* DSM 15071 (FOMJ01000000).

*Microorganisms* **2022**, *10*, 295


**Table 1.** Comparison of genome features of the *Ectothiorhodospiraceae* genomes in this study. The genomes are colored by groups as described in the text.

#### *2.2. Genomic DNA Extraction and Sequencing*

For the strains sequenced in this study, genomic DNA was prepared from frozen cells, using the GeneJET DNA purification kit (Thermo Fisher Scientific, Waltham, MA, USA). The quantity and purity of DNA were determined using Qubit and NanoDrop instruments and showed absorbance 260/280 ratios between 1.67 and 2.12. The DNA libraries were prepared with the Nextera DNA flex library prep kit (Illumina, Inc., San Diego, CA, USA). All genomes were sequenced using 500 µL of a 1.8pM library with an Illumina MiniSeq instrument, using paired-end sequencing (2 × 150 bp). Quality control of the reads was performed using FASTQC in BaseSpace (Illumina, version 1.0.0), using a kmer size of 5 and contamination filtering. The data for each was assembled de novo using Unicycler [24] in PATRIC [25]. The genome sequences were annotated using RAST (Rapid Annotations using Subsystem Technology; version 2.0; [26]).

#### *2.3. Whole Genome Comparison*

Average percentage nucleotide identity (ANIb) between the whole genomes was calculated using JSpecies [27]. JSpecies uses a pairwise genome comparison algorithm to measure the probability of genomes belonging to the same species, with an arbitrary species cutoff of 95%. The whole genome-based phylogenetic tree was generated within PATRIC [25], using the CodonTree pipeline which uses PGFams as homology groups. Among these selected genomes, 141 PGFams were found using the CodonTree analysis, and the aligned proteins and coding DNA from single-copy genes were used for RAxML analysis [28,29]. 100 rounds of bootstrapping were performed using the 'Rapid bootstrapping option. The resulting Newick file was used in iTOL for tree visualization [30]. Average amino acid identity (AAI) values were calculated from the proteome comparison in PATRIC. Only bi-directional hits were used for this analysis and *Hlr. halophila* SL1<sup>T</sup> *and Hlr. halochloris* DSM1059<sup>T</sup> (BN9850) as the reference strains. Pairwise 16S rRNA comparisons were performed using LALIGN (EMBL-EBI), using the genome-derived 16S rRNA sequences.

The alignment for the 16S rRNA comparisons was performed using Clustal Omega [31] which uses seeded guide trees and Hidden Markov Model (HMM) profiles to generate multiple sequence alignments. The phylogenetic tree was calculated by the neighbourjoining (NJ) method [32] in JALVIEW [33] and a Newick file was generated. The Jalview NJ method uses the BLOSUM62 substitution matrix to compute a sum of scores for the residue (base) pairs at each aligned position. iTOL was used to draw the phylogenetic trees expressed in the Newick phylogenetic tree format [30]. Due to the lack of complete 16S rRNA sequences from whole genomes, Genbank 16S rRNA sequences were used instead for *Ect. salini* (FM244738.1), *Ect. imhoffii*, *Arhodomonas aquaeoli* (M26631.2) and *Nitrococcus mobilis* (NR\_104912.1), for the 16S rRNA phylogenetic tree.

For the synteny analysis, comparative genome regions were generated using global PATRIC PGFam families to determine a set of genes that match a focus gene. The gene set is compared to the focus gene using BLAST and sorted by BLAST scores within PATRIC [25]. The Compare Region Viewer in PATRIC displays the focus gene along with the other genes in the same family and their flanking regions in their genomes. The *bch*B gene was used as a focus gene to analyze synteny of the photosynthetic and bacteriochlorophyll gene cluster.

#### **3. Results and Discussion**

With the availability of large numbers of genome sequences, the average nucleotide identity (ANI) of genomes became an established measure to compare the similarity/relatedness of bacteria and is an accepted alternative to the classical DNA–DNA hybridization method, which only provided reliable results if performed by experts familiar with this method. It has been suggested to recognize bacteria with ANI >95% as belonging to the same species [27], while bacteria with ANI <90% would be recognized in most cases as separate species. Those with values between 90 and 95% identity may be argued either way depending on other properties. As ANI is apparently more precise in the differentiation of closely related bacteria compared to the 16S rRNA gene sequences [27], we have analyzed

ANI from all phototrophic *Ectothiorhodospiraceae* species with available genome sequences with representatives of the genera *Ectothiorhodospira*, *Ectothiorhodosinus*, *Thio-rhodospira* and *Halorhodospira.* In addition, representative chemotrophic *Ectothiorhodospiraceae* with available genome sequences were included.

As the species of *Halorhodospira* are clearly distinct from other *Ectothiorhodospiraceae* genera and the separation of *Halorhodospira* species on a family level is proposed, in the following, the term *Ectothiorhodospiraceae* will be used in the strict sense with the exclusion of *Halorhodospira* species and species of *Halorhodospira* will be treated separately.

### *3.1. Genomes of Ectothiorhodospiraceae*

The *Ectothiorhodospiraceae* (excluding *Halorhodospira*) currently contain 10 named phototrophic species. The genome sequences of *Trs. sibirica* ATCC 700588<sup>T</sup> [18], *Ect.* strain PHS-1 [19], *Ect. haloalkaliphila* ATCC 51935<sup>T</sup> [20] and *Ect*. strain BSL-9 [21] have been previously published. In addition, genome sequence information for *Ect. magna* DSM 22250<sup>T</sup> , *Ect. marina* DSM 241<sup>T</sup> *, Ect. marismortui* DSM 4180<sup>T</sup> , *Ect. mobilis* DSM 237<sup>T</sup> *, Ect. shaposhnikovii* DSM 243<sup>T</sup> and *Ectothiorhodosinus mongolicus* DSM 15479<sup>T</sup> was presented in earlier publications [15–17]. Here we include additional genome sequences of type strains of *Ect. vacuolata* DSM 2111<sup>T</sup> and *Ect. variabilis* DSM 21381<sup>T</sup> and additional isolates (DG9, A-7R, A-7Y, B14B, WN21Y, WN21R, WN2R, BN9902 = C, BN9100 = YC6.1 and BN9905 = ScotB) and use this information for a comparison of species and strains (Table 1). Based on comparison of genome length and its G + C content (mol%), several groups of strains/species can be clearly distinguished (Table 1). The group containing *Ect. mobilis* DSM 237 and both *Ect. marismortui* strains (DSM 4180 and DG9) has a smaller genome size (2.62–2.80 Mbp) and higher percentage of 68.2–68.4 mol% G + C as compared to the majority of the genomes that have genome sizes from 3.20 Mbp to 3.79 Mbp (with the exception of *Ect*. spec. strain PHS-1 with 2.94 Mbp) and a G + C content of 62.3 to 63.7 mol%. *Ectothiorhodosinus mongolicus*, *Thiorhodospira sibirica* and *Ect. magna* apparently are more distinct and also have a lower G + C content (Table 1).

The ANI comparison of the phototrophic *Ectothiorhodospiraceae* (Table 2) clearly identifies three groups of species, with ANI values >80%: (i) including the type strains of *Ect. vacuolata*, *Ect. shaposhnikovii* and *Ect. magna* (colored orange in Table 2), with *Ect. magna* being clearly more distant to the others with ANI values between 83–87%; (ii) including the type strains of *Ect. haloalkaliphila*, *Ect. variabilis* and *Ect. marina* (colored yellow in Table 2); (iii) including the type strains of *Ect. mobilis* and *Ect. marismortui* (colored green in Table 2). More distant to these groups are *Ectothiorhodosinus mongolicus* and *Thiorhodospira sibirica*, with ANI values of <70% relative to the others.

The whole genome-based phylogenetic tree (Figure 1), constructed using unique PGFams as homology groups, supports the ANI analysis and confirms the three groups of strains and species as separate clades in the tree. It also includes representative genomes of chemotrophic species that had previously been considered as belonging to *Ectothiorhodospiraceae* but are phylogenetically at a distance, which suggests their grouping in a separate new family. The only chemotrophic bacteria that are especially closely related to the phototrophic *Ectothiorhodospiraceae* species, in particular to *Thiorhodospira sibirica*, are species of the two groups of *Thioalkalivibrio* (Figure 1). Of the two distinct groups, one includes, among others, the type species *Thioalkalivibrio versutus* and *Thioalkalivibrio halo-philus*, and the other, among others, *Thioalkalivibrio sulfidophilus* and *Thioalkalivibrio denitrificans* [34].


**Table 2.** Average percentage nucleotide identity (ANI) between pairs of genomes. ANI values >80% are colored and values above the species cutoff (>95%) are presented in bold.

In addition to the whole genome-based tree, we also constructed a 16S rRNA-based phylogenetic tree that includes a comparison of all sequenced phototrophic *Ectothiorhodospiraceae* (Figure 2), including *Ect. salini* and *Ect. imhoffii*, for which there is no genome sequence available. Included for comparison were 16S rRNAs of the chemotrophic species previously assigned under *Ectothiorhodospiraceae*. The results of the 16S rRNA tree are consistent with the whole genome-based tree and ANI comparisons.

These data show that the type strains of *Ect. variabilis* DSM 21381<sup>T</sup> (WN 22<sup>T</sup> ) and *Ect. haloalkaliphila* ATCC 51935<sup>T</sup> are highly similar. With an ANI of 98.3%, which is above the suggested 95% species cutoff, they may be regarded belonging to the same species. Both strains originate from soda lakes of the Wadi el-Natrun in Egypt [14,35,36]. Additional isolates from the Wadi el-Natrun (BN9902, WN21Y, WN21R, WN2R) are also highly similar to this species (Table 1 [36]). Despite some earlier results [14,36], 16S rRNA sequences support this high similarity, with 97.2% similarity between the two type strains, but may not necessarily demark species identity. This issue will have to be resolved in more detailed studies.

The results also support the recognition of the following closely related but distinct species. *Ect. marismortui* DSM 4180<sup>T</sup> is close to *Ect. mobilis* DSM 237<sup>T</sup> and *Ect. shaposhnikovii* DSM 243<sup>T</sup> close to *Ect. vacuolata* DSM 2111<sup>T</sup> . Previously, the taxonomic status of *Ect. marismortui* DSM 4180<sup>T</sup> and *Ect. vacuolata* DSM 2111<sup>T</sup> were disputed. They were recognized as distinct species by Imhoff [6] and Imhoff and Süling [14], though it was suggested by Ventura et al. [37,38] to combine *Ect. vacuolata* with *Ect. shaposhnikovii* and *Ect. marismortui* with *Ect. mobilis,* primarily based on data on DNA–DNA hybridization and ribotyping. The results of ribotyping, however, did not clearly support this conclusion. They revealed identity between two strains of *Ect. mobilis* (DSM 237<sup>T</sup> and DSM 240) but clear differences of these to *Ect. marismortui* DSM 4180<sup>T</sup> . They also revealed the identity of two strains of *Ect. vacuolata* (DSM 2111<sup>T</sup> and B3) with clear differences to *Ect. shaposhnikovii* DSM 243<sup>T</sup> [37]. In addition, the values of DNA–DNA hybridization were only 80–84% [37]. In contradiction to this suggestion and based on genetic studies of Rubisco and nitrogenase genes, Tourova et al. [39] recognized the four species as distinct species. The presented data on ANI and whole genome-based comparisons are in good agreement with considerations by Imhoff [6] and Tourova et al. [39], recognizing *Ect. vacuolata* DSM 2111<sup>T</sup> and *Ect. shaposhnikovii* DSM 243<sup>T</sup> as distinct species with only 90% ANI and also recognizing *Ect. marismortui* DSM 4180<sup>T</sup> and *Ect. mobilis* DSM 237<sup>T</sup> as distinct species with ANI values of 93%. Therefore, ANI supports the distinction of these four bacteria at the species level and also confirms the close relationships of the two couples of species.

**Figure 1.** Whole genome-based phylogenetic tree of the *Ectothiorhodospiraceae*. The support values for the phylogenetic tree are generated using 100 rounds of the 'Rapid bootstrapping' option of RaxML. The tree was rooted at midpoint and the branch length tree scale is defined as the mean number of substitutions per site, which is an average across both nucleotide and amino acid changes. *Oceanococcus* and *Acidiferrobacter* were found not to belong to any of the three families but were included in the tree as outgroups. **Figure 1.** Whole genome-based phylogenetic tree of the *Ectothiorhodospiraceae*. The support values for the phylogenetic tree are generated using 100 rounds of the 'Rapid bootstrapping' option of RaxML. The tree was rooted at midpoint and the branch length tree scale is defined as the mean number of substitutions per site, which is an average across both nucleotide and amino acid changes. *Oceanococcus* and *Acidiferrobacter* were found not to belong to any of the three families but were included in the tree as outgroups.

In addition to the whole genome-based tree, we also constructed a 16S rRNA-based phylogenetic tree that includes a comparison of all sequenced phototrophic *Ectothiorhodospiraceae* (Figure 2), including *Ect. salini* and *Ect. imhoffii*, for which there is no genome sequence available. Included for comparison were 16S rRNAs of the chemotrophic species previously assigned under *Ectothiorhodospiraceae*. The results of the 16S rRNA tree are con-

sistent with the whole genome-based tree and ANI comparisons.

**Figure 2.** 16S rRNA-derived phylogenetic tree for *Ectothiorhodospiraceae* species. The phylogenetic tree was calculated by the neighbour-joining (NJ) method [32] in Jalview [33]. The Jalview NJ method uses the BLOSUM62 substitution matrix to compute a sum of scores for the residue (base) pairs at each aligned position. The length of the branches is proportional to the number of nucleotide substitutions per site. iTOL was used to draw the phylogenetic trees expressed in the Newick phylogenetic tree format [30]. **Figure 2.** 16S rRNA-derived phylogenetic tree for *Ectothiorhodospiraceae* species. The phylogenetic tree was calculated by the neighbour-joining (NJ) method [32] in Jalview [33]. The Jalview NJ method uses the BLOSUM62 substitution matrix to compute a sum of scores for the residue (base) pairs at each aligned position. The length of the branches is proportional to the number of nucleotide substitutions per site. iTOL was used to draw the phylogenetic trees expressed in the Newick phylogenetic tree format [30].

*haloalkaliphila* ATCC 51935<sup>T</sup> are highly similar. With an ANI of 98.3%, which is above the suggested 95% species cutoff, they may be regarded belonging to the same species. Both The results also suggest the assignment of several new isolates to one or the other of the established species. Of two strains that were isolated as arsenite-oxidizing bacteria [40], strain BSL-9 originating from Big Soda Lake represents a distinct new species closely related

These data show that the type strains of *Ect. variabilis* DSM 21381<sup>T</sup> (WN 22T) and *Ect.* 

to *Ect. haloalkaliphila* and *Ect. marina*, while strain PHS-1 originating from Mono Lake is a strain of *Ect. vacuolata* (Figure 1 and Table 2). Strain DG9 is an additional isolate of *Ect. marismortui* from Berikei Sulfur Springs in Dagestan (Russia, [41]). Strains BN9100 (=YC6.1), an isolate from Solar Lake by H. Biebl (Braunschweig, Germany) and BN9905 (=ScotB), isolated from the seashore near Inverary (Scotland) by one of us (JFI), are new isolates of *Ect. marina.* Strains A-7R, A-7Y and B14B form a separate clade on the whole genome phylogenetic tree and have ANI values <90% with respect to any of the other strains, but high ANI values amongst themselves, and therefore represent a new species related to *Ect. shaposhnikovii* and *Ect. vacuolata* (Figure 1 and Table 2).

The present data support the distinction of *Ectothiorhodosinus mongolicus* [42] and also *Thiorhodospira sibirica* [43] in genera separate from *Ectothiorhodospira*. *Ect. magna* [44], though distantly related to other species of the genus, is placed inside a group of species together with *Ect. shaposhnikovii* and *Ect. vacuolata*.

#### *3.2. Genomes of Halorhodospira Species*

Four *Halorhodospira* species are currently recognized: *Hlr. halophila* [45,46], *Hlr. halochloris* [47], *Hlr. abdelmalekii* [48] and *Hlr. neutriphila* [49]. The genome sequence of *Hlr. halophila* SL1<sup>T</sup> was previously determined [22] and information on genome sequences of *Hlr. halophila* BN9626, BN9620, BN9630, *Hlr. neutriphila* DSM 15116<sup>T</sup> , *Hlr. halochloris* DSM 1059<sup>T</sup> and *Hlr. abdelmalekii* DSM 2110<sup>T</sup> were reported earlier [15–17].

We have now added genome sequences of additional strains of *Hlr. halochloris* and *Hlr. halophila*, which originate from the Wadi el-Natrun in Egypt [35,50] and from Mongolian soda lakes (Gorlenko, personal communication) (Table 1). The ANI, whole genome-based and also 16S rRNA-based phylogenetic tree comparison of the *Halorhodospira* species (Table 3, Figures 1 and 2) show a clear distinction of five different species.

**Table 3.** ANI, average percentage nucleotide identity between pairs of genomes. ANI values >90% are colored and values above the species cutoff (>95%) are presented in bold.


The first group consists of *Hlr. halophila* SL1<sup>T</sup> and *Hlr. halophila* BN9630. The close relationship of strain BN9630 to the *Hlr. halophila* type strain was demonstrated previously by 16S rRNA sequence analyses and fatty acid analyses [9,14] and is now supported by near identity in terms of the ANI (>99%) (Table 3). The 16S rRNA sequences delineated from the genome sequence actually are only two bases different between the two strains.

A second group is formed by strain BN9626 and a number of other isolates (BN9620, BN9621, BN9622, BN9624, BN9628, M38 and M39old), which differed from strain SL1<sup>T</sup> with respect to fatty acid composition [9] and revealed only 85–86% ANI identity to *Hlr. halophila* SL1<sup>T</sup> (Table 3). This indicates that these strains could be recognized as strains of a distinct species. The average amino acid identity (AAI) is 89.3% for 2414 orthologues of strains SL1<sup>T</sup> and BN9626, which is consistent with the ANI. The 16S rRNA identity (obtained from the genome sequences) between *Hlr. halophila* BN9626 and SL1<sup>T</sup> with 18 differences out of 1537 bp is 98.8%, which is within a range allowing species distinction according to [51] (see systematic conclusions, below). More detailed studies are required to clarify the situation.

Third, according to the whole genome phylogenetic tree (Figure 1) *Hlr. neutriphila* DSM 15166<sup>T</sup> [49] is clearly distinct from *Hlr. halophila*. The ANI to strains of *Hlr. halophila* is 77.4–77.6%, indicating considerably greater divergence. The 16S rRNA obtained from the genome sequence also averages 54 bases different from that of strains SL1<sup>T</sup> and BN9626 (96.5% identity), which places *Hlr. neutriphila* on a separate clade on the 16S rRNA tree (Figure 2). The AAI for *Hlr. neutriphila* and strains SL1<sup>T</sup> and BN9626 is 72.9% (1730 orthologs). The G + C content of the genome sequence from *Hlr. neutriphila* is 72 mol% and thus significantly higher than all other *Halorhodospira* species (Table 1).

Further, according to the genome-based phylogenetic tree, the green-colored, bacteriochlorophyll *b* containing species *Hlr. halochloris* and *Hlr. abdelmalekii* form a major branch distinct from the groups of red-colored species, which have bacteriochlorophyll *a* (Figure 1). The complete genome sequence of *Hlr. halochloris* DSM 1059<sup>T</sup> (=BN9850<sup>T</sup> ) (APO17372) was established by [23] and we have added a second genome sequence of this strain (NRRM00000000). The genome sequence supposedly determined from the *Hlr. halochloris* strain A<sup>T</sup> (=DSM 1059<sup>T</sup> ) by the authors of [52] turned out to be almost identical to the one from *Ect. haloalkaliphila* ATCC 51935<sup>T</sup> (AJUE00000000) and is not from *Hlr. halochloris*. It could originate from a culture contaminant or a mislabeled culture.

The genomes of the three *Hlr. halochloris* strains (BN9850<sup>T</sup> , BN9851 and BN9852) have a significantly lower G + C content (55.8–56.1 mol%) compared to *Hlr. abdelmalekii* (62.9 mol%) and other *Halorhodospira* species (>67.9 mol%; Table 1). The genome sequence of *Hlr. abdelmalekii* BN9840<sup>T</sup> (= DSM 2110<sup>T</sup> ) is significantly different from *Hlr. halochloris* (ANI of 72.5%), *Hlr. halophila* and *Hlr. neutriphila* (ANI of approximately 73–74%). There are 65 differences (95.8% identity out of 1551 bp) in the 16S rRNA between *Hlr. abdelmalekii* and *Hlr. halochloris,* as obtained from the genome sequences. The 16S rRNA of the green-colored species, *Hlr. halochloris* and *Hlr. abdelmalekii*, on average has 87 differences to the red-colored species. The AAI between the two green-colored species is 71.6% (1856 orthologs) and that for the two green-colored species and the three red-colored species is 68.6%.

#### *3.3. Genome-Delineated Properties*

#### 3.3.1. Sulfur and Thiosulfate Oxidation

In the *Chromatiaceae,* elemental sulfur is stored in the periplasm inside a proteinaceous membrane made up of one to five sulfur globule proteins rich in alternating Gly and Tyr residues [53,54]. None of the *Ectothiorhodospiraceae* and the *Halorhodospira* species, for that matter, store sulfur intracellularly and contain sulfur globule proteins; the sulfur is transported outside the cells and must be transported back inside for further oxidation, once all the sulfide and thiosulfate are exhausted. It is only then that it is further oxidized to sulfate in the cytoplasm. Detailed studies on processes of sulfur oxidation in purple sulfur bacteria have been made by Dahl and coworkers [55,56].

Most of the *Ectothiorhodospiraceae* species contain the thiosulfate-oxidizing enzymes SoxAXYZB, with the apparent exception of *Ect. magna* and *Trs. sibirica,* that only contain SoxY and Z. The genes are organized in two separate operons. As is apparent from the complete genome of *Ect*. BSL-9 containing a single chromosome, the genes *soxA* and *soxX* are in one location and *soxYZB* in another. In those *Halorhodospira* species that use thiosulfate, the diheme SoxA and SoxX are fused into the single triheme protein, SoxXA. *Hlr. abdelmalekii* and *Hlr. halochloris* do not use thiosulfate as a growth substrate and do not contain Sox enzymes. In *Chromatiaceae*, genes of the thiosulfate-oxidizing enzymes are also located in two operons; however, *soxB* is associated with *soxAX* and the two gene clusters are *soxAXB* and *soxYZ*.

#### 3.3.2. Glutathione

Glutathione is a common small molecule reductant in bacteria, which has several important functional roles [57]. It is present in most bacteria, though in some phototrophic purple bacteria it is replaced by the glutathione amide, in which the terminal glycine carboxylate is amidated, as was first shown for *Marichromatium gracile* [58]. In most bacteria, glutathione reductase has an Arg21 that, according to three-dimensional structural analysis, contributes to the binding of glutathione via a salt bridge to the carboxyl group of the

glutathione glycine residue [59]. As found for the glutathione reductase of *Marichromatium gracile* and other *Chromatiaceae*, the Arg21 is substituted by Glu21 and in consequence glutathione, but not its amide, is repelled from the binding site [57]. Thus, glutathione reductase Glu/Arg21 can be used as a proxy for the presence or absence of glutathione amide. We have found through genome sequencing that all *Chromatiaceae* and *Ectothiorhodospiraceae* (with the exceptions of *Ect. mobilis* DSM 237<sup>T</sup> and both strains of *Ect. marismortui*, DSM 4180<sup>T</sup> and DG9, where a glutathione reductase has not been found) have Glu21 and are thus likely to have glutathione amide. However, glutathione reductases of all *Halorhodospira* species have Arg21 like the majority of bacteria and are likely to have the normal glutathione.

#### 3.3.3. Carboxysome Genes

The carboxysome is an important structure in many autotrophic bacteria, such as the cyanobacteria [60] and many autotrophic proteobacteria [61]. The carboxysome forms a proteinaceous membrane that encloses Rubisco and carbonic anhydrase and its genes are associated with bicarbonate transporters. The carboxysome shields Rubisco from nonfunctional reactions with oxygen and provides CO<sup>2</sup> where it is needed. Though a definite proof for the physical structures in purple sulfur bacteria is lacking, we have found genes of putative carboxysome peptides A and B and carboxysome shell proteins *csoS1*, *csoS2* and *csoS3* in many *Chromatiaceae* and *Ectothiorhodospiraceae*. They are present in seven species of *Ectothiorhodospiraceae* but not in in *Ect. mobilis* DSM 237<sup>T</sup> and both strains of *Ect. marismortui* (DSM 4180<sup>T</sup> and DG9)*, Ect.* PHS-1 and *Trs. Sibirica*. Carboxysome genes appear to be absent from *Halorhodospiraceae*, though ribulose bisphosphate carboxylase, phosphoribulose kinase and enzymes involved ini the Calvin Cycle are present.

#### 3.3.4. Nitric Oxide Reduction

As mentioned above, *Ectothiorhodospira* strains A-7Y, A-7R and B14B are closely related but different from *Ect. shaposhnikovii* DSM 243<sup>T</sup> and *Ect. vacuolata* DSM 2111<sup>T</sup> . These three strains have 15 unique PGFams that are not present in any of the other *Ectothiorhodospiraceae* genomes and all of them contain a gene cluster for nitric oxide reductase (NOR) that seems to be missing from all the other species. The *norCBQD* cluster contains two NO reductase activation proteins and both the small and large NOR subunits (C and B), which show similarity to cytochrome oxidases. There have been few studies of nitric oxide reduction in the *Ectothiorhodospiraceae*, but these three strains appear to at least have the genetic capability for nitric oxide reduction. In addition to the overall genomic distinction, these genetic differences are another reason to consider these strains as a separate species.

#### 3.3.5. HIPIP

In *Chromatiaceae* the iron–sulfur protein HiPIP was shown to be the electron donor to the photosynthetic reaction center PufLM and the electron acceptor from the BC1 complex [62,63]. The majority of *Ectothiorhodospiraceae* contain either three or four HiPIP isozymes, with the exception of *Ets. mongolicus* and *Trs. sibirica*, which have only one. In addition, *Hlr. halophila* has genes for four HiPIP isoenzymes, two of which are abundant in cell extracts [64]. The gene for one of these isozymes is mixed in with the photosynthetic genes, suggesting that it is the electron donor in this species as it is in *Chromatiaceae* and presumably *Ectothiorhodospiraceae.* It is likely that it acts as a mediator of electrons between the *bc*<sup>1</sup> complex and photosynthetic reaction center in these bacteria. HiPIP II actually forms a strong complex with the reaction center and rapidly reacts with soluble HiPIP [65]. On the other hand, the green-colored *Halorhodospira* species do not have HiPIP and are likely to utilize cytochrome *c*5, which is present in all *Halorhodospira* species characterized to date, but not in the *Ectothiorhodospiraceae*. Cytochrome *c*<sup>5</sup> is known as a soluble mediator in the green sulfur bacteria in the family *Chlorobiaceae* and the membrane bound *c*<sup>5</sup> is a possible component of the *bc* complex [66].

In the green-colored *Halorhodospira* species, there are no HiPIP genes at all, which is an important difference to the red-colored *Halorhodospira* species. In this case, the electron mediator between the *bc*<sup>1</sup> complex and photosynthetic reaction center must be different. Perhaps due to the difficulty in growing *Hlr. halochloris* and *Hlr. abdelmalekii*, no experimental studies of electron transfer have been published. However, all of the *Halorhodospira* species produce a soluble cytochrome *c*<sup>5</sup> similar to those found in *Chlorobiaceae* species and which is thought to couple the *bc* complex to the photosynthetic reaction center. It is proposed that this cytochrome *c*<sup>5</sup> is the electron mediator in these two species.
