Fatty acid profiles and desaturase-encoding genes are different in thermo- and psychrotolerant strains of the Bacillus cereus Group

The Bacillus cereus Group consists of closely-related bacteria, including pathogenic or harmless strains, and whose species can be positioned along the seven phylogenetic groups of Guinebretière et al. (I–VII). They exhibit different growth-temperature ranges, through thermotolerant to psychrotolerant thermotypes. Among these, B. cytotoxicus is an atypical thermotolerant and food-poisoning agent affiliated to group VII whose thermotolerance contrasts with the mesophilic and psychrotolerant thermotypes associated to the remaining groups I–VI. To understand the role of fatty acid (FA) composition in these variable thermotypes (i.e. growth behavior vs temperatures), we report specific features differentiating the FA pattern of B. cytotoxicus (group VII) from its counterparts (groups I–VI). The FA pattern of thermotolerant group VII (B. cytotoxicus) displayed several specific features. Most notably, we identified a high ratio of the branched-chain FAs iso-C15/iso-C13 (i15/i13) and the absence of the unsaturated FA (UFA) C16:1(5) consistent with the absence of ∆5 desaturase DesA. Conversely, phylogenetic groups II–VI were characterized by lower i15/i13 ratios and variable proportions of C16:1(5) depending on thermotype, and presence of the DesA desaturase. In mesophilic group I, thermotype seemed to be related to an atypically high amount of C16:1(10) that may involve ∆10 desaturase DesB. The levels of i15/i13 ratio, C16:1(5) and C16:1(10) UFAs were related to growth temperature variations recorded between thermotypes and/or phylogenetic groups. These FA are likely to play a role in membrane fluidity and may account for the differences in temperature tolerance observed in B. cereus Group strains.


Background
In bacteria, the fatty acid (FA) composition of the cell membrane varies according to environmental conditions, as it is involved in bacterial adaptation to environmental changes such as temperature, pressure, and O 2 availability [1][2][3][4]. Among these changes, the effect of temperature on bacterial FA composition is a prominent focus of research. Desaturases responsible for producing unsaturated FAs (UFAs) have been shown to play a role in low-temperature adaptation. The effect of incorporation of low-melting-point UFAs is to maintain membrane fluidity under the physical stress due to cold [5][6][7][8][9]. The number of desaturases varies depending on species, e.g. two desaturases have been identified in Bacillus cereus stricto or ss [10] whereas B. subtilis counts only one known desaturase [11]. The FA composition of bacterial cells is also known to vary according to species, especially in Bacillus and related genera [12,13], and has Open Access *Correspondence: marie-helene.guinebretiere@avignon.inra.fr 5 INRA, UMR408 SQPOV, Site Agroparcs, 228 route de l' Aérodrome, CS40509, 84914 Avignon Cedex 9, France Full list of author information is available at the end of the article been included among important features for describing new taxa of the aerobic endospore-forming bacteria [14].
The Bacillus cereus Group (B. cereus sensu lato or sl) includes bacterial strains with a wide range of growth temperatures. These strains can be classified by growthtemperature range, from psychrotrophic to thermotolerant strains [15]. These ranges of growth temperatures fit with the seven major phylogenetic groups (I-VII) established by Guinebretière et al. [15,16] in the B. cereus Group (see Table 1). This 7-macrogroup classification is the most complete phylogenetic description of the B. cereus Group and is coherent with all MLST, AFLP, MLEE and genomic data produced since 2004 in the literature [17]. It offers a unique setting to investigate the relation between temperature adaptation and hypothetical factors and can be used to resolve problems tied to effective species delimitation in the Group. Indeed, B. cereus sl contains seven closely-related species that, although not all genomospecies, can be clearly positioned by reference to each other in the classification of Guinebretière et al.
(Additional file 1). Some of them have been described on the basis of singular phenotypic or pathologic traits such as rhizoidal colonies (B. mycoides), psychrotolerance (B. weihenstephanensis), insecticidal properties (B. thuringiensis) [18], enterotoxins (B. cereus sensu ss) [19] and anthrax (B. anthracis) [20]. Only two are true genomospecies: the harmless species B. pseudomycoides (mesophilic group I) and the pathogenic species B. cytotoxicus (thermotolerant group VII) [21], both of which thus share a homogenous growth-temperature range. B. mycoides/B. weihenstephanensis (psychrotolerant group VI), and B. anthacis (a clonal lineage within the highly mesophilic group III) also have a homogenous growth-temperature range as they belong each to a unique phylogenetic group, whereas B. cereus ss and B. thuringiensis represent highly heterogeneous thermotypes across phylogenetic groups II-VI (Additional file 1) [15,16].
Bacillus cytotoxicus, though relatively rare, is known as one of the most virulent pathogenic species of B. cereus sl [16,21]. Its pathogenicity is mainly attributed to the greater expression and cytotoxic activity of the poreforming cytotoxin K1 (CytK-1) [22][23][24][25], a variant of the cytotoxin K found in many strains of B. cereus sl [16]. Bacillus cytotoxicus comprises solely thermotolerant strains [15,21]. In addition to all these particularities, B. cytotoxicus has been described as displaying a specific FA composition in B. cereus sl [21]. However, this specific FA composition of B. cytotoxicus has actually only been validated with reference to 4 of the 7 phylogenetic groups, and the relation between FA composition and the various thermotypes found in B. cereus sl has not yet been studied.
The aim of this study was to determine to what extent the FA composition of B. cytotoxicus (phylogenetic group VII) differs from that of all the other phylogenetic groups of the B. cereus Group, and to what extent these differences relate to the growth-temperature ranges of the groups. In addition, as desaturases are known to play a role in low-temperature adaptation through FA composition and membrane fluidity, we also investigated the presence of desaturase-encoding genes among the available B. cereus sl genomes and the existence of a putative relation between their presence and the FA composition of strains exhibiting different thermotypes.

Strains
The studied strains are listed in Additional file 2 and were representative of all seven phylogenetic groups. Minimal growth temperature (T min ) and maximal growth temperature (T max ) were used to determine growth-temperature range (T min -T max ). Their group affiliation and growthtemperature range were determined in previous studies in 2008 and 2013 [15,21]. As a rule, each phylogenetic group has its own growth-temperature range [15], as presented in Table 1. In the previous work of 2008, T min and T max were determined using a standard test described in the Bergey's Manual [26], with temperature fluctuation of the incubators being not greater than ±0.2°C for all tested temperatures.
All these strains are referenced in the previous works of 2008 and 2013 [15,21].These strains have since been conserved in our Laboratory (UMR408 Collection) and the original sources are presented in Additional file 2.

Thermotypes
The thermotypes were determined from the growth-temperature ranges (T min -T max ) as the resulting phenotype, and are presented in Table 1 by reference to growth-temperature range.

Table 1 Distribution of key fatty acids (FA), desA and desB according to phylogenetic group in B. cereus sl
Phylogenetic groups and growth temperature ranges are described in previous works [15,21].
FA proportions obtained from the FAMEs extractions and GC-MS analysis for a total of 21 representative strains are presented as mean value ± standard deviation (sd) (n = 2-6 representative strains, with 1-3 replicates).
FA nomenclature is: iX: iso-FA with X carbons; aX: anteiso-FA with X carbons; nX: saturated FA with X carbons. For unsaturated FA, symbol ":" is prefixed to the number of unsaturations in FA chain; position of unsaturation in FA chain is indicated between parentheses. The symbol "*" indicates that the FA is unsaturated by DesA and "#" indicates that the FA is unsaturated by DesB.

In silico analysis
A total of 210 B. cereus sl genomes available in databases at the time of the search were used for this study. Genome affiliations to the phylogenetic groups were established previously [28], as described in [16], using panC sequence similarity.
Two desaturases have been described in B. cereus, i.e. DesA and DesB [10], which are responsible for two different types of unsaturation. The DesA enzyme adds a double bond in the ∆5-position of a saturated FA (SFA) while DesB creates an unsaturation in the ∆10-position of an SFA. The presence of each desaturase-encoding gene was thus investigated among all the available B. cereus sl genomes. The search for desA and desB orthologs with reference to locus_tag BC_2983 and BC_0400, respectively, in the ATCC 14579T genome (i.e. sequence loci described for desaturases in the ATCC 14579 genome) was performed via the Integrated Microbial Genomes (IMG) interface [29]. No other ∆5 or ∆10 desaturaseencoding genes have been reported in the B. cereus Group. First, candidate homologs were identified based on BLASTp similarities with a 1e-2 E-value cutoff and with low-complexity soft masking (-F'm S') turned on. Second, orthologous relationships between BC_2983 or BC_0400 genes and their respective homologous genes in all other genomes were established through bidirectional best blast hits.

FA composition in B. cereus sensu lato: general features
The B. cereus Group displays a specific overall FA pattern setting it apart from the other species of the Bacillus genus [21,30], with short-chain branched FAs (12C and 13C) and a characteristic predominance of iso-C13:0 (Additional file 2). Whatever the phylogenetic group analyzed, three major FAs were identified: iso-C15:0 (i15), iso-C13:0 (i13) and C16:0 (n16). For better visibility, only these three major SFAs and the 7 UFAs (previously linked to cold adaptation [3,27]) are listed in Table 1. While the n16 SFA did not range widely according to thermotype, the proportion of the other two SFAs (i13 and i15) varied for the most sharply-contrasting thermotypes, i.e. the thermotolerant type (group VII, B. cytotoxicus), the highly mesophilic type (group III), and the highly psychrotolerant type (group VI). The SFA i15 accounted for more than 1/13.3 of total FAs in group VII (B. cytotoxicus) and thus constituted a marker for this group.

i15/i13 ratio as a rough indicator of thermotype
We calculated the i15/i13 ratio, defined as proportion of iso-C15:0 divided by proportion of iso-C13:0, for strains belonging to each phylogenetic group (Fig. 1a).

Presence of Δ5/Δ10-desaturases and proportion of C16:1(5) as an accurate indicator of thermotype in the most recent branch of B. cereus sl
UFAs unsaturated at the ∆5 position were present in the most recent phylogenetic groups (II-VI, see Additional file 1) yet absent from thermotolerant group VII (B. cytotoxicus) and mesophilic group I (Table 1), particularly C16:1(5). We therefore searched for the presence of the genes encoding for the desaturases responsible for the synthesis of UFAs unsaturated at the ∆5 or ∆10 location (desA and desB respectively) in this group of bacteria. The results indicated that, contrary to desB, orthologs of desA were not found in all B. cereus sl genomes ( Table 1): none of the genomes in groups I and VII contained an ortholog of desA gene. The 2.5-4% of negative genomes in groups III and IV may be due to information missing from draft genomes. However, in groups I (B. pseudomycoides strains) and VII (B. cytotoxicus strains), the ortholog of desA was truly absent, consistent with the near-zero concentration of ∆5 UFAs observed in groups I and VII.
Considering only groups II to VI, C16:1(5) proportion increased with thermotype, from the highly mesophilic to the highly psychrotolerant groups (with no significant difference between the two psychrotolerant groups II and VI) (Fig. 1b). Despite being produced in low amounts, C16:1(5) proportion appeared a good parameter to discriminate psychrotolerant thermotypes (groups II, VI) from other thermotypes and even to discriminate between mesophilic thermotypes (groups III, IV, V), including those that were not discriminated by i15/i13 ratio (groups II, V, IV).
Atypically in group-I strains, UFAs unsaturated at the ∆10 location were produced in higher proportions than in the other groups (Table 1), particularly C16:1(10), offsetting the lack of UFAs unsaturated at the ∆5 location. This difference presumably contributes to membrane fluidity and allows group I (B. pseudomycoides) to exhibit a mesophilic thermotype. In contrast to mesophilic group I (B. pseudomycoides), the thermotolerant group VII (B. cytotoxicus) seems unable to offset this deficiency by producing larger amounts of UFAs unsaturated at the ∆10 location.

FA composition and putative link with cold adaptation in the B. cereus Group
Our study highlighted the relation between i15/i13 ratio and B. cereus Group thermotypes. Another study reported that proportion of i13 was strongly reduced in a B. cereus ATCC 14579 mutant displaying growth impairment at low temperature compared with its parental strain during growth at low temperature [27]. The i15/i13 ratio recalculated from these data [27] was much higher in the mutant than in the parental strain, emphasizing the putative role of a low i15/i13 ratio for psychrotolerance ability.
The C16:1(5) UFA presumably appeared with ∆5 desaturase DesA in the most recent branch of the phylogeny in B. cereus sl containing groups II-VI (see Additional file 1). Its absence in phylogenetic groups I and VII indicates a link with the whole evolution of the B. cereus Group. Indeed, groups I and VII belong to two other independent branches at the base of the phylogenetic tree (see Additional file 1). Taken together, these results converge towards a differential process of evolution involving ∆5 UFAs for groups II-VI and ∆10 UFAs for group I. As group VII is basal to the evolutionary tree, followed by group I and then the remaining groups, we can posit that the ancestor of the B. cereus Group was devoid of ∆5 UFAs and went through adaptation in a few steps. The first step would involve an increase of ∆10 UFAs through group I. The second step would involve ∆5 UFAs through groups II-VI, with a more efficient adaptation from mesophily to psychrotolerance. This is also consistent with the absence of ∆5 desaturase (DesA) in groups I and VII. Interestingly, the same kind of configuration (absence in groups I and VII) was also observed for the two-component system CasK/R, which was recently described as playing a role in B. cereus-Group cold adaptation [28]. Through this observation, we can also posit that the lack of key genes such as casK/R and desA might be related to the inability of B. cytotoxicus strains to grow at temperatures below 20°C [21], and that these genes probably took part in a more complex mechanism of adaptation in the most recent branch of the phylogeny.

Conclusion
A link was established between the FA pattern of B. cereus sl strains and ability or inability to grow at low temperature. The FA profile of B. cytotoxicus (group VII) is highly specific compared to that of the phylogenetic groups I-VI and is relatable to its atypical thermotolerance: high i15/i13 ratio, absence of UFAs unsaturated at the ∆5 location (particularly C16:1(5)), absence of a ∆5 desaturase (DesA). In contrast, the presence of ∆5 desaturase DesA and subtle amounts of C16:1(5) seem to be associated with an advanced mechanism of adaptation, resulting in a large panel of thermotypes through groups II-VI (from mesophily to psychrotolerance). Mesophilic strains of Group I seem to exhibit a specific intermediate state of evolution involving a fairly atypical amount of the ∆10-desaturated C16:1(10).

Availability of supporting data
The datasets supporting the results of this article are included in Additional file 2. Authors' contributions SD organized and drafted the manuscript, analyzed the acquired data, performed the statistical analysis, interpreted the results and contributed to the write-up of the results. MHG conceived the study design, performed FAMEs extraction, contributed to data analysis and interpretation of results, and organized and edited the manuscript. BDS performed GC-MS analysis and extracted the data. CNT and VB were involved in editing and writing up the final version of the manuscript. JB supervised the analysis and interpretation of the outcomes and was involved in writing up the different versions of the manuscript. All the authors read and approved the final manuscript. 1