Delft University of Technology Molecular and physiological adaptations to low temperature in thioalkalivibrio strains isolated from soda lakes with different temperature regimes

The genus Thioalkalivibrio comprises sulfur-oxidizing bacteria thriving in soda lakes at high pH and salinity. Depending on the geographical location and the season, these lakes can strongly vary in temperature. To obtain a comprehensive understanding of the molecular and physiological adaptations to low temperature, we compared the responses of two Thioalkalivibrio strains to low (10°C) and high (30°C) temperatures. For this, the strains were grown under controlled conditions in chemostats and analyzed for their gene expression (RNA sequencing [RNA-Seq]), membrane lipid composition, and glycine betaine content. The strain Thioalkalivibrio versutus AL2T originated from a soda lake in southeast Siberia that is exposed to strong seasonal temperature differences, including freezing winters, whereas Thioalkalivibrio nitratis ALJ2 was isolated from an East African Rift Valley soda lake with a constant warm temperature the year round. The strain AL2T grew faster than ALJ2 at 10°C, likely due to its 3-fold-higher concentration of the osmolyte glycine betaine. Moreover, significant changes in the membrane lipid composition were observed for both strains, leading to an increase in their unsaturated fatty acid content via the Fab pathway to avoid membrane stiffness. Genes for the transcriptional and translational machinery, as well as for counteracting cold-induced hampering of nucleotides and proteins, were upregulated. Oxidative stress was reduced by induction of vitamin B12 biosynthesis genes, and growth at 10°C provoked downregulation of genes involved in the second half of the sulfur oxidation pathway. Genes for intracellular signal transduction were differentially expressed, and interestingly, AL2T upregulated flagellin expression, whereas ALJ2 downregulated it. IMPORTANCE In addition to their haloalkaline conditions, soda lakes can also harbor a variety of other extreme parameters, to which their microbial communities need to adapt. However, for most of these supplementary stressors, it is not well known yet how haloalkaliphiles adapt and resist. Here, we studied the strategy for adaptation to low temperature in the haloalkaliphilic genus Thioalkalivibrio by using two strains isolated from soda lakes with different temperature regimes. Even though the strains showed a strong difference in growth rate at 10°C, they exhibited similar molecular and physiological adaptation responses. We hypothesize that they take advantage of resistance mechanisms against other stressors commonly found in soda lakes, which are therefore maintained in the bacteria living in the absence of low-temperature Citation Ahn A-C, Jongepier E, Schuurmans JM, Rijpstra WIC, Sinninghe Damsté JS, Galinski EA, Roman P, Sorokin D, Muyzer G. 2021. Molecular and physiological adaptations to low temperature in Thioalkalivibrio strains isolated from soda lakes with different temperature regimes. mSystems 6:e01202-20. https://doi.org/10.1128/mSystems.01202-20. Editor Rachel Mackelprang, California State University, Northridge Copyright © 2021 Ahn et al. This is an openaccess article distributed under the terms of the Creative Commons Attribution 4.0 International license. Address correspondence to Gerard Muyzer, g.muijzer@uva.nl. Low-temperature adaptation in Thioalkalivibrio Received 16 November 2020 Accepted 30 March 2021 Published 27 April 2021 March/April 2021 Volume 6 Issue 2 e01202-20 msystems.asm.org 1 RESEARCH ARTICLE D ow nl oa de d fr om h ttp s: //j ou rn al s. as m .o rg /jo ur na l/m sy st em s on 0 3 Ju ne 2 02 1 by 1 54 .5 9. 12 4. 11 3. pressure. A major difference, however, was detected for their glycine betaine content at 10°C, highlighting the power of this osmolyte to also act as a key compound in cryoprotection.


Molecular and Physiological Adaptations to Low Temperature in
Thioalkalivibrio Strains Isolated from Soda Lakes with Different Temperature Regimes ABSTRACT The genus Thioalkalivibrio comprises sulfur-oxidizing bacteria thriving in soda lakes at high pH and salinity. Depending on the geographical location and the season, these lakes can strongly vary in temperature. To obtain a comprehensive understanding of the molecular and physiological adaptations to low temperature, we compared the responses of two Thioalkalivibrio strains to low (10°C) and high (30°C) temperatures. For this, the strains were grown under controlled conditions in chemostats and analyzed for their gene expression (RNA sequencing [RNA-Seq]), membrane lipid composition, and glycine betaine content. The strain Thioalkalivibrio versutus AL2 T originated from a soda lake in southeast Siberia that is exposed to strong seasonal temperature differences, including freezing winters, whereas Thioalkalivibrio nitratis ALJ2 was isolated from an East African Rift Valley soda lake with a constant warm temperature the year round. The strain AL2 T grew faster than ALJ2 at 10°C, likely due to its 3-fold-higher concentration of the osmolyte glycine betaine. Moreover, significant changes in the membrane lipid composition were observed for both strains, leading to an increase in their unsaturated fatty acid content via the Fab pathway to avoid membrane stiffness. Genes for the transcriptional and translational machinery, as well as for counteracting cold-induced hampering of nucleotides and proteins, were upregulated. Oxidative stress was reduced by induction of vitamin B 12 biosynthesis genes, and growth at 10°C provoked downregulation of genes involved in the second half of the sulfur oxidation pathway. Genes for intracellular signal transduction were differentially expressed, and interestingly, AL2 T upregulated flagellin expression, whereas ALJ2 downregulated it.
IMPORTANCE In addition to their haloalkaline conditions, soda lakes can also harbor a variety of other extreme parameters, to which their microbial communities need to adapt. However, for most of these supplementary stressors, it is not well known yet how haloalkaliphiles adapt and resist. Here, we studied the strategy for adaptation to low temperature in the haloalkaliphilic genus Thioalkalivibrio by using two strains isolated from soda lakes with different temperature regimes. Even though the strains showed a strong difference in growth rate at 10°C, they exhibited similar molecular and physiological adaptation responses. We hypothesize that they take advantage of resistance mechanisms against other stressors commonly found in soda lakes, which are therefore maintained in the bacteria living in the absence of low-temperature to low temperature, we studied gene expression by transcriptomics, including strain-specific responses to temperature, and analyzed their adaptation of membrane lipid composition and the production of the cryoprotectant glycine betaine at both 10°C and 30°C.

RESULTS AND DISCUSSION
For our experiment, we selected two Thioalkalivibrio strains from geographical locations with different temperature regimes. Based on their different origins, we expected that these strains would exhibit different responses to low temperature.
Batch cultivation showed that T. versutus AL2 T grew much faster than T. nitratis ALJ2 (m max of 0.064/h versus 0.024/h, respectively) at 10°C, while this was the opposite at 30°C (m max of 0.28/h versus 0.43/h) ( Table 1). To gain further insights into the mechanisms behind these differences in growth rate, the bacteria were subsequently grown at 10°C and 30°C under controlled conditions in chemostats until steady state. As the growth rate of the bacterial culture is set by the dilution rate and the samples were taken at the steady state of the chemostat cultures, all changes in the gene expression and the membrane lipids originate from the difference in temperature and strain between the reactors and are not influenced by the growth state of the culture, as is the case in batch cultivation. The strength of repeatability of chemostat cultivation is illustrated by the close clustering of the samples from the same condition in the principal-component analysis (PCA) (Fig. S1C). The PCA also shows that the gene expression profiles of both strains were markedly different between the two temperature regimes and strains.
Temperature adaptation in both Thioalkalivibrio strains consisted of molecular and physiological shifts in several biological processes and components, which are depicted in a conceptual model (Fig. 1). A summary of the expression data for the different categories of genes mentioned in Fig. 1 and in the sections below is given in Table S9. Statistical differences in temperature responses between the two strains (Tables S6 and S7) are highlighted in the text by giving the Gene Ontology (GO) ID of the orthologs possessing a strain-temperature interaction.
Category 1: membrane lipid composition. Changes in temperature may have substantial effects on the fluidity of the membrane and consequently also on the functioning of membrane-associated metabolic pathways, such as respiratory electron transfer. Therefore, the adaptation of the lipid composition is crucial to counteract cold-induced membrane stiffness and to maintain the fluidity of the membrane (20,35). The head group composition of intact polar membrane lipids did not change significantly with temperature and, for both Thioalkalivibrio strains, was mainly composed of phosphatidylcholine with smaller amounts of lysophosphatidylcholine and phosphatidylglycerol (Table 2). Furthermore, phosphatidylethanolamine was present at low abundance in both strains and increased in concentration at 10°C for T. nitratis ALJ2. Diphosphatidylglycerol was found in T. nitratis ALJ2, whereas it was detected only in low quantities for T. versutus AL2 T grown at 10°C (Table 2). To our knowledge, the head group composition of the intact polar membrane lipids was measured for the first time for Thioalkalivibrio by this study. In strong contrast to the head group composition, major changes in the fatty acid (FA) composition of the cell membrane were detected as a result of the difference in cultivation temperature (Tables 3 and 4). At 10°C, a decrease of lactobacillic acid (C 19 cyclopropyl FA) in both strains and of 10-methylhexadecanoic acid (10-Me-C 16:0 ) in T. versutus AL2 T , as well as an increase of octadec-11-enoic acid (C 18:1D11 ) in both strains and of hexadec-9-enoic acid (C 16:1D9 ) in T. versutus AL2 T , was observed in comparison to the FA composition at 30°C ( Fig. 1; Table 3). The saturated FA hexadecanoic acid (C 16:0 ) remained abundant (20 to 24%) at both temperatures. An increase in the proportion of unsaturated FA at low temperatures is a well-known mechanism to keep the membrane fluidity optimal by lowering its phase-transition temperature (35,36). Apart from increasing the proportion of unsaturated FAs, a reduction in chain length and increase in branching of FA are also used by bacteria as an adaptation to lower temperature (20). Therefore, it is somewhat surprising that T. versutus AL2 T reduced the amount of branched FA and both species almost maintained the ratio of shorter-to longer-chain FA (Table 4). This suggests that the high abundance of unsaturated FA provides sufficient membrane fluidity for growth at 10°C. With respect to the shift in proportion of the C 19 cyclopropane FA, i.e., from ca. 40% at 30°C to 3 to 12% at 10°C (Table 3), its increased presence has been associated with heat resistance (37-39) but has also been detected in low-temperature adaptation (20,38).
The changes in FA composition correlated with the differential expression of genes responsible for the membrane polar lipid biosynthesis of both Thioalkalivibrio strains ( Fig. 1; Tables 3 and 4; Table S9). Desaturation of FAs is performed either by the fatty acid desaturases (Des) or by the FabA dehydratase/isomerase, which replaces single bonds with doubled bonds in the carbon chain (40,41). The genes desA and desC, coding for D12 and D9 desaturases, respectively, were detected only in the genome of T. versutus AL2 T but were not differentially expressed at 10°C. However, at 10°C, both strains upregulated fabA (Table S5), and T. nitratis ALJ2 also upregulated a malonyl coenzyme A (malonyl-CoA)-acyl carrier protein transacylase gene, fabD. Thus, the Fab   Table S9.

Low-Temperature Adaptation in Thioalkalivibrio
March/April 2021 Volume 6 Issue 2 e01202-20 msystems.asm.org 5 pathway appears to be a major desaturation mechanism at low temperatures in Thioalkalivibrio. This is in good agreement with the fact that the abundant unsaturated FAs (C 16:1D9 and C 18:1D11 ) are omega-7 FAs, which are produced by the Fab pathway. Lactobacillic acid is produced from C 18:1D11 FA by cyclopropanation by the cyclopropane-fatty-acyl-phospholipid synthase. The methylene donor is a methyl group on Sadenosylmethionine (42). Multiple genes encoding this enzyme were detected in the genomes of both strains. Two of them were significantly downregulated at 10°C in T. nitratis ALJ2 but not in T. versutus AL2 T (strain-temp interaction for OG0001444) (Table S7).
Other differentially expressed genes involved in the membrane lipid biosynthesis included in T. nitratis ALJ2 the upregulation at 10°C of the squalene-associated FAD-dependent desaturase hpnE and the squalene synthase hpnD, which are both involved in the production of the neutral lipid squalene ( Fig. 1; Table S9). Interestingly, squalene was only detected in a minor fraction (,1%) of the total lipids, even though it was found in high concentrations in the Thioalkalivibrio strain ALJ15 (43), and in T. paradoxus ARh1 T its derivative lanosterol constituted up to 50% of the total lipids (44). Furthermore, multiple genes coding for proteins involved in peptidoglycan, lipopro-   tein, and lipopolysaccharide synthesis were found to be differentially expressed in both strains as well as a decrease in expression at 10°C for a range of genes containing a PEP-CTERM domain (Table S9). PEP-CTERM domain-containing proteins are believed to constitute a protein export sorting system, which is linked to exopolysaccharide protein expression (45). Category 2: compatible solutes and antifreeze proteins. A key adaptation to reduced temperature is the production of compatible solutes and antifreeze proteins. Among these compatible solutes, glycine betaine is, apart from its feature as an osmolyte (46), also known to enhance cryotolerance in bacteria (21,(47)(48)(49). Accordingly, the glycine betaine concentration in both strains was higher at 10°C than at 30°C (Fig. 2). Moreover, the glycine betaine content was 3-fold higher for T. versutus AL2 T than for T. nitratis ALJ2 at 10°C ( Fig. 1 and 2). Hence, T. versutus AL2 T appears to have a clear advantage for growing at low temperatures compared to T. nitratis ALJ2. Osmolytes provide cryotolerance by protecting the cytoplasmic proteins from denaturation during freezing (50) and by reducing the cytoplasmic freezing point, which prevents the formation of ice crystals inside the cell (51,52).
Glycine betaine can be taken up by the cell via the betaine-carnitine-choline transporter (BCCT) (53) or by the ATP-driven ABC glycine betaine transporter (54). Both Thioalkalivibrio strains contain four genes for BCCT, and T. versutus AL2 T also possesses the ATP-dependent betaine transporter. Interestingly, one of the annotated BCCT genes is downregulated in both strains at 10°C, whereas the others are not differentially expressed ( Fig. 1; Table S9).
Another option for intracellular accumulation of glycine betaine is de novo synthesis from the precursor choline (55) or from glycine in a three-step methylation process (56). Neither Thioalkalivibrio strain encodes BetABI, which is responsible for the synthesis of glycine betaine from choline (55). However, both strains do possess the de novo synthesis pathway from glycine, which is catalyzed by two methyltransferases, glycine sarcosine methyltransferase and sarcosine dimethylglycine methyltransferase (56). Some bacteria also shield themselves from intracellular ice formation by the production of antifreeze or ice-binding proteins, which bind to ice crystals and prevent them from growing (57). A gene encoding such an antifreeze-like protein was overexpressed in T. versutus AL2 T at the low temperature ( Fig. 1; Table S9) but was not detected in the genome of T. nitratis ALJ2. Another molecule protecting cells against cold stress is the polyamine spermidine (58). This might be explained by its properties of protecting DNA from oxidative damage, intervening in transcriptional and translational regulation, and playing a role in the structure and the stability of nucleic acids (59). At 10°C, T. nitratis ALJ2 upregulates the expression of genes for two enzymes facilitating spermidine production, polyamine aminopropyltransferase and adenosylmethionine decarboxylase ( Fig. 1; Table S9), while their expression is consistently high in T. versutus AL2 T (strain-temp interactions for OG0001863 and OG0000401) (Table S6). Category 3: stability of nucleic acids and proteins. A decrease in temperature causes an increase in the negative supercoiling of DNA (23) and also an increase in the stability of the secondary structures of mRNA (24,25), which negatively affects the efficiency of replication, recombination, transcription, and translation (28). Furthermore, a decrease in temperature can lead to protein misfolding (29). To counteract these effects, genes coding for helicases, chaperones, and proteins of the transcription and translational machinery were generally upregulated in both Thioalkalivibrio strains growing at 10°C ( Fig. 1; Table S9).
Multiple genes annotated for nucleotide synthesis and repair were upregulated with growth at the low temperature in both strains (Table S9). Furthermore, several transcription regulators were differentially expressed in Thioalkalivibrio. Cold shock proteins (CSP) counteract the formation of secondary structures in mRNA and thereby improve the transcription and translation efficiency by acting as chaperones of nucleic acids (28). Interestingly, only one CSP was slightly upregulated in T. nitratis ALJ2, whereas the other was downregulated (b values of 0.68 [P adj = 6.63 Â 10 25 ] and 21.30 [P adj = 3.53 Â 10 22 ], respectively). Other low-temperature-induced genes involved in transcription included the genes for DEAD box-containing RNA helicase (rhlE) and the RNase PH in both strains (Table S5), the DNA-binding protein of the HU family in T. versutus AL2 T , and the transcription termination/antitermination factor NusA in T. nitratis ALJ2 ( Fig. 1; Table S9).
Both strains induced the gene expression of multiple tRNAs, genes involved in the synthesis of tRNAs and amino acids, the 5S rRNA and several ribosomal protein-encoding genes ( Fig. 1; Table S9). Multiple noncoding RNAs (ncRNA) implicated in translation were upregulated as well and include RtT sRNA (small RNA processed from the tyrT transcript) and signal recognition particle (SRP) RNA, whereas the RNase P RNA component and the tmRNA were downregulated ( Fig. 1; Table S9). Furthermore, the gene encoding the translation initiation factor IF-1 was upregulated at 10°C in T. nitratis ALJ2, while it was consistently highly expressed in T. versutus AL2 T (strain-temp interaction for OG0000216) ( Fig. 1; Table S6).
The genes of the chaperones GroEL and GroES of the Hsp60 system were highly expressed in both Thioalkalivibrio strains at 10°C ( Fig. 1; Table S5). These chaperones enable the proper folding of proteins, which is essential under denaturing conditions (30), such as cold stress (29). In contrast, Hsp20 heat shock proteins were downregulated at 10°C in the two tested Thioalkalivibrio strains (Table S4). Other protein chaperones induced under cold stress are the peptidyl-prolyl isomerases (PPIase), including the FKBP (FK506 binding protein) and the cyclophilin types (60)(61)(62). In T. nitratis ALJ2, a FKBP-and a cyclophilin-type PPIase were strongly upregulated at 10°C, but another FKBP-type PPIase was highly downregulated ( Fig. 1; Table S9). Contrary to the effect in T. nitratis ALJ2, the cyclophilin-type PPIase was consistently highly expressed in T. versutus AL2 T (strain-temp interaction for OG0001177) (Table S6).
Finally, the secD and yajC gene components of the general secretory pathway (Sec) were upregulated in both Thioalkalivibrio at 10°C ( Fig. 1; Table S5). The complex formed by SecDF-YajC might play an important role in cold adaptation, as inactivation of this complex induces cold sensitivity (63).
Category 4: protection against oxidative stress. With the reduction of temperatures, oxygen solubility in the medium increases (32), which can lead to the formation of reactive oxygen species (ROS), oxidative stress, and subsequently damage to DNA, proteins, and lipids (31). Neither a superoxide dismutase nor a catalase was differentially expressed at 10°C in the two Thioalkalivibrio strains. However, multiple peroxiredoxins were upregulated in T. nitratis ALJ2 (strain-temp interaction for OG0001493) (Table S6), and a glutathione S-transferase was upregulated in both strains at 10°C ( Fig. 1; Table S9). In contrast, genes involved in vitamin B 12 biosynthesis were strongly upregulated at 10°C for both strains ( Fig. 1; Tables S5 and S9). Interestingly, genes for vitamin B 12 production were also highly upregulated in Thioalkalivibrio under arsenite stress, where vitamin B 12 was previously proposed as an antioxidant in this bacterium (64). Its antioxidant capacity has already been demonstrated in both prokaryotic and eukaryotic cells under oxidative stress (65)(66)(67). Furthermore, for cold stress, genes encoding vitamin B 12 biosynthesis have been shown to be expressed and linked to oxidative stress protection (60,68).
Category 5: sulfur, carbon, and energy metabolism. Thioalkalivibrio strains are sulfur-oxidizing bacteria (17) that under standard thiosulfate-limited steady state conditions completely oxidize thiosulfate to sulfate. However, accumulation of elemental sulfur was observed in the chemostats of both Thioalkalivibrio strains growing at 10°C, whereas elemental sulfur was not observed in the chemostats at 30°C. This might indicate a decrease in the metabolic activity of the second oxidation step from elemental sulfur to sulfate compared to the first oxidation step of thiosulfate to elemental sulfur at low temperatures. Indeed, the multienzyme sox system, which is responsible for the oxidation of thiosulfate and the release of sulfate and SoxY-bound sulfane sulfur (69,70), did not change significantly in expression, except for the downregulation of soxB in T. versutus AL2 T . Interestingly, the heterodisulfide reductase (hdr) genes, which are thought to be responsible for the further processing of the sulfane sulfur (71), also did not change their expression.
In contrast, genes involved in the transformation of sulfite to sulfate were downregulated at the low temperature in both Thioalkalivibrio strains. These include genes for the indirect sulfite oxidation pathway with sat and aprAB (72,73), which are present only in T. nitratis ALJ2, and genes for the two direct pathways, including the quinonedependent sulfite oxidoreductase SoeABC of cluster 1 (as defined in reference 64) (74) and the cytochrome c-dependent oxidoreductase SorAB (Table S4) (72). While both strains downregulated sorA at 10°C, the effect was more pronounced in T. nitratis ALJ2 (strain-temp interaction for OG0000759) ( Table S7). The genes encoding SoeABC of cluster 2 (as defined in reference 64), however, were upregulated in T. nitratis ALJ2. Furthermore, the sulfide dehydrogenase gene fccAB was also downregulated at 10°C in both Thioalkalivibrio strains ( Fig. 1; Table S9). A decrease in the expression of genes responsible for the oxidation of elemental sulfur to sulfate at the low temperature was also observed for the psychrotolerant bacterium Acidithiobacillus ferrivorans SS3 at 8°C, where hdr, sat, and apr were downregulated and sox genes were significantly upregulated (68).
T. versutus AL2 T appears to increase its capacity to fix carbon in response to growth at low temperatures, as it induces the transcription of the small and the large subunit of RubisCO (ribulose-biphosphate carboxylase), as well as a subunit of the carboxysome. Moreover, T. nitratis ALJ2 upregulates a carbon storage regulator csrA ( Fig. 1; Table S9). These results suggest an increased carbon demand at lower temperatures.
Looking at genes encoding electron transport chain proteins, it can be seen that the NuoL subunit of NADH-quinone oxidoreductase was significantly induced in T. versutus AL2 T and the NuoM subunit was downregulated at 30°C in T. versutus AL2 T but not in T. nitratis ALJ2 (strain-temp interaction for OG0000591) ( Fig. 1; Table S7). The induction of NADH-quinone oxidoreductase subunits at low temperatures was also found in other studies (68,75). Furthermore, at the low temperature, multiple subunits of the ATP synthase were also significantly induced in both Thioalkalivibrio strains and a gene encoding a Na 1 /H 1 antiporter was upregulated in T. versutus AL2 T ( Fig. 1; Tables S5 and S9). The upregulation of genes involved in the energy generation reflects an increased need for ATP to perform adaptations in response to low temperature, for example, the de novo production of glycine betaine (76). changes in DNA, RNA, and proteins (25,77). These signals are transmitted and can induce physiological adaptations to cope with a change in environmental temperature. For example, two-component systems with sensor histidine kinases and response regulator proteins are known to be involved in the sensing and transmitting of environmental signals, including temperature (78). Methyl-accepting chemotaxis proteins were found to be involved in sensing temperature in Escherichia coli and to alternate accordingly its swimming behavior (25). Indeed, gene expression results for Thioalkalivibrio grown at 10°C revealed several genes encoding methyl-accepting chemotaxis proteins and a gene encoding a two-component system protein with differential expression ( Fig. 1; Table S9). Moreover, multiple genes encoding GGDEF-domain proteins, which are involved in signal transduction by producing the secondary messenger bis-(39-59)-cyclic dimeric GMP (c-di-GMP) (79), were found to be differentially expressed as well ( Fig. 1; Tables S4, S5, and S9). A GGDEF domain-containing protein (OG0000928) is upregulated at 10°C in T. versutus AL2 T but downregulated in T. nitratis ALJ2, while a methyl-accepting chemotaxis protein (OG0000306) is strongly downregulated at 10°C in T. nitratis ALJ2. Furthermore, a methyl-accepting chemotaxis protein (OG0001236) is strongly upregulated at 10°C for both strains (strain-temp interactions for OG0000928, OG0000306, and OG0001236) (Table S7). Flagella are important for the motility of bacteria in their aqueous environment and to respond to chemotaxis (80), as are type IV pili enabling twitching motility on surfaces (81). Furthermore, gene expression of flagellum components and motility is primarily influenced by various environmental conditions in bacteria (82,83). In particular, movement rates of the flagellum and, with this, swimming speed are physically directly influenced by temperature (84). Thioalkalivibrio strains are motile with a single polar flagellum (17). At the low temperature, all fli, flg, and flh flagellumbuilding blocks as well as genes assisting in the assembly of the flagellum (flgANM and fliST) were upregulated for T. versutus AL2 T . In contrast, for T. nitratis ALJ2, genes for structural flagellum proteins were downregulated, with the exception of the genes encoding the filament and the filament cap of the flagellum (fliCD) and their respective chaperones (fliST) (85) (Fig. 1; Tables S4, S5, and S9). However, comparing expression levels of the flagellum genes of both strains to each other revealed that both strains shift to a similar level at 10°C (strain-temp interactions for OG0000450-52, OG0000486-9, OG0001024-28, OG0001070-71, OG0001073-74, OG0001257-58, OG0001286-87, OG0001559-64, OG0001613-15, OG0001644-50, OG0002008, OG0002052, and OG0002056) (see the box plots in Table S7). Type IV pilus synthesis genes did not change their expression with temperature in T. versutus AL2 T , but certain of these genes showed a downregulation at 10°C in T. nitratis ALJ2 (strain-temp interactions for OG0001191, OG0001834, OG0001850, OG0002020, and OG0002096) ( Fig. 1; Tables S7 and S9). When the flagellum is damaged, the cell is able to repair it by incorporating new flagellum protein units (86,87). The upregulation of the flagellum units in T. versutus AL2 T could be explained by the increased need for repair at low temperatures. However, T. nitratis ALJ2 might reduce its mobility under these shearing circumstances and thus downregulate its flagellum genes.
Conclusion. In this study, we investigated the low-temperature responses of two Thioalkalivibrio strains isolated from soda lakes with different temperature regimes. T. versutus AL2 T , which originates from a soda lake with freezing periods during winter, had a clear growth advantage at 10°C compared to T. nitratis ALJ2. Remarkably, T. nitratis ALJ2 exhibited an adaptation response to the low temperature similar to that of T. versutus AL2 T , even though T. nitratis ALJ2 was isolated from a Kenyan soda lake with a stable warm temperature the year round. The exception was the 3-fold-higher concentration of the osmolyte glycine betaine in T. versutus AL2 T , identifying this compatible solute as an essential actor in cryoprotection of the cold-adapted strain. We hypothesize that the adaptation to low-temperature stress invokes multiple "basic" stress responses that trigger inherent protection mechanisms against, e.g., oxidative stress or protein denaturation. Moreover, glycine betaine is also one of the key factors in protection against osmotic stress. Therefore, even though T. nitratis ALJ2 does not face cold stress in its environment, it must cope with other common soda lake stressors, such as high UV radiation or high salinity. This study enlightens differences and commonalities in adaptation by strains of the same genus with different temperature regime backgrounds and contributes to the general understanding of low-temperature adaptation in bacteria.

MATERIALS AND METHODS
Strains and growth conditions. Growth parameters (lag phase [hours], maximum growth rate [m max ] [per hour], and stationary phase [cells per milliliter]) of T. versutus AL2 T and T. nitratis ALJ2 were determined from axenic cultures grown in duplicate 500-ml batch-mode reactors at 10°C and 30°C by cell count measurements using flow cytometry. The batch reactors were magnetically stirred at 700 rpm and aerated with 0.5 liter/min of compressed air. The medium contained 17.5 g/liter Na 2 CO 3 , 13.9 g/liter NaHCO 3 , 6.1 g/liter NaCl, 1 g/liter K 2 HPO 4 , 0.2 g/liter MgCl 2 , 40 mM Na 2 S 2 O 3 , 5 mM KNO 3 and 1:1,000 trace metals (88). As the pH is influenced by temperature, it was adjusted to 9.8 at 10°C and 30°C. To obtain the growth parameters, the cell count measurements of the batch cultures were fit into a logistic fit model using the R package GroFit (89) (lag phase) and Jupyter Notebook (https://jupyter.org) (maximal growth rate and stationary phase).
To study the response to low temperature, T. versutus AL2 T and T. nitratis ALJ2 were grown at 10°C and 30°C in 500-ml chemostat reactors under continuous cultivation with a dilution rate of 0.010 6 0.00027/h. The dilution rate was regularly monitored by measuring the pump rate over time with a burette placed between the medium vessel and the peristaltic pump. The chemostats were continuously aerated with compressed air at 0.5 liter/min and magnetically stirred at 600 rpm. Metallic baffles inside the reactor were used to achieve full aeration. All reactors were inoculated with 20 ml of bacterial cultures and were kept in batch mode for 2 days at 30°C to obtain a dense culture. The 30°C reactors were then switched to a continuous culture mode, whereas the 10°C reactors were transferred to 10°C and kept there for 4 days to acclimatize before being switched to a continuous culture mode. The composition of the growth medium for the continuous cultivation was kept the same as described for the batch cultivation except for using 20 mM Na 2 S 2 O 3 and an increase of the pH to 10. Per condition, two reactors were run in parallel and each reactor was run twice using approximately 100 ml from the previous steady state culture as starting material, providing in the end four replicates for each condition. Figure S1A gives a schematic illustration of the experimental chemostat setup. Cell material was harvested at steady state, i.e., after at least five volume changes of the reactor. To assess steady state, cell count measurements by flow cytometry were performed (Fig. S1B) during the entire chemostat run and thiosulfate and sulfate concentrations were measured during the 3 days before the culture entered steady state growth (Table S1).
Cell counting by flow cytometry. Cell counting by flow cytometry was first used to determine the growth curves of the batch cultivation and subsequently the growth parameters, which were used to set up the continuous cultivation experiment. Second, flow cytometry was also used to count cells during the continuous cultivation experiment in order to assess the presence of the steady state at the sampling time.
Cell material was harvested from the reactors throughout the experiment and fixed with formaldehyde at a final concentration of 1% (wt/vol). The samples were incubated for 1 h at 4°C and thereafter centrifuged at 19,000 Â g for 10 min at 4°C. The supernatant was discarded, and the cell pellet was dissolved in 0.5 ml TE buffer (10 mM Tris-HCl, 1 mM disodium EDTA [pH 8.0]), directly flash-frozen in liquid nitrogen, and stored at 280°C until analysis.
For the flow cytometry analysis, the samples were diluted in TE buffer to reach cell counts below 2,500 events per ml. Samples were preheated for 5 min to 35°C, and SYBR green I (Invitrogen, Carlsbad, CA, USA) was added. The stained samples were incubated for 10 min at 35°C and then kept in the dark until measurement. Flow cytometry was performed on a BD Accuri C6 flow cytometer (BD Bioscience, Franklin Lakes, NJ, USA) with 50 ml of cell sample with a medium flow rate (35 ml/min). To distinguish the events of stained bacterial cells from the background noise, a dot plot of the FL1-A (excitation at 488 nm, fluorescence at 533/30 nm) and FL3-A (excitation at 488 nm, fluorescence at 670 nm) was used that was gated from the counts of FL1-A between fluorescence values of 10,000 and 1,000,000. A threshold at 10,000 on FL-1 H was set.
Analysis of thiosulfate and sulfate. Thiosulfate and sulfate were analyzed using a protocol as published before (90,91). Briefly, samples were filtered over 0.2-mm polyethersulfone (PES) membrane filters (VWR International, Radnor, PA, USA) and stored at 220°C until further processing. Thiosulfate and sulfate concentrations were measured by ion chromatography (Compact IC 761) (Metrohm, Herisau, Switzerland) with an anion column (Metrohm Metrosep A Supp 5; 150/4.0 mm) equipped with a precolumn (Metrohm Metrosep A Supp 4/5 Guard). The ion chromatography system included a chemical suppressor (Metrohm), a CO 2 suppressor (853; Metrohm), and a conductivity detector (Metrohm). In addition, suppressors for eluent conductivity and carbon dioxide were used (Metrohm). The mobile phase consisted of 3.2 mM sodium carbonate, 1 mM sodium bicarbonate solution, and 1% (vol/vol) acetone and had a flow rate of 0.7 ml/min. RNA sequencing and data analysis. Bacterial cells were harvested from the steady state cultures, and the RNA was extracted and sequenced (RNA sequencing [RNA-Seq]) as described previously (64). The reference sequences of T. versutus AL2 T (NZ_MVAR00000000.1) and T. nitratis ALJ2 (NZ_ARKB00000000.1) were

Low-Temperature Adaptation in Thioalkalivibrio
March/April 2021 Volume 6 Issue 2 e01202-20 msystems.asm.org 11 downloaded from the NCBI RefSeq FTP server. For the general transcriptomic analysis, RNA-Seq reads were analyzed using the software programs kallisto v0. 44.0 (92) and sleuth v0. 30.0 (93) as described previously (64) ( Table S2). A gene was considered differentially expressed if jbj was .1 and its P adj value was ,0.05. The b value is defined as a biased estimator of the log(fold change) on a natural-log scale (93). General information on the RNA-Seq reads is presented in Table S3.
To assess strain-specific gene expression responses to temperature, one-to-one orthologs were identified with OrthoFinder v2.3.11 (94) and analyzed with kallisto v0.46.2 and sleuth v0.30.0. Main effects of strain and temperature were evaluated with the Wald test, and their interaction effect with a likelihood ratio test (LRT). Functional annotations were obtained with InterProScan v5.40-77.0 (95) including Gene Ontology (GO) terms, PFAM A domain content (database version 32), and InterProScan terms (Tables S4  to S8).
Membrane lipid analysis. Bacterial cells were harvested as for RNA-Seq analysis, stored at 280°C, and then lyophilized. The lyophilized cells were hydrolyzed with 1 N KOH-methanol (96%) by refluxing for 1 h. The hydrolysate was adjusted to pH 4 with 2 N HCl-methanol, and after addition of H 2 O, it was extracted with dichloromethane (DCM). The fatty acids in this DCM extract were converted to fatty acid methyl esters (FAMEs) by methylation with diazomethane. FAMEs were analyzed by gas chromatography (GC) and GC-mass spectrometry (GC-MS) as described previously (96). Double-bond positions of the monounsaturated FAMEs were determined using the mass spectra of their dimethyl disulfide derivatives as described previously (97). The measurement was done on three samples per condition.
The intact polar lipids (IPLs) were extracted from the lyophilized cells using a modified Bligh-Dyer technique and analyzed by high-performance liquid chromatography (HPLC)-electrospray ionization (ESI)-MS as described previously (96). The measurement was done on two samples per condition.
Determination of the concentration of glycine betaine. T. versutus AL2 T and T. nitratis ALJ2 were grown in triplicate at 10°C and 30°C in 1-liter batch cultures in the culture medium that was used for the chemostat experiment. These cultures had to be produced separately, as the continuous cultivation experiments did not generate enough biomass to perform all the analyses. Cultures were centrifuged at 10,000 Â g for 5 min and the supernatant was removed. The pellet was resuspended in a small volume, transferred to a 50-ml Greiner tube, and centrifuged at 7,000 Â g. The pellets were stored at 220°C and then freeze-dried. From this, 30 mg of dry bacterial biomass was used for extraction with 500 ml methanol-chloroform-water (10:5:4) according to a modified protocol from reference 98 as described in reference 99. Phase separation was achieved by adding 130 ml chloroform and 130 ml water, followed by a 5min centrifugation step at 10,000 Â g. The upper, polar phase was removed and evaporated at reduced pressure in a SpeedVac vacuum concentrator at 50°C and 1,500 Pa. The residue was dissolved in 100 ml of an 80% (vol/vol) acetonitrile-water mixture and subsequently analyzed by isocratic HPLC on a LiChroCART aminopropyl column (Merck, Darmstadt, Germany) with 80% (vol/vol) acetonitrile-water as the mobile phase and at a flow rate of 1 ml/min.
The remaining lower chloroform phase (including the layer of insoluble cell material) was evaporated overnight. The dry residue was resuspended in 1 ml of 0.1 M NaOH and boiled for 5 min. The solubilized material was diluted and subjected to total protein determination using the Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the recommendations of the manufacturer. The total protein content was measured to standardize the glycine betaine concentration.
Data availability. The raw RNA-Seq data sets have been deposited in the NCBI Sequence Read Archive under SRA accession numbers SRX7551034 to SRX7551049.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, EPS file, 2.1 MB.

ACKNOWLEDGMENTS
We acknowledge Danielle Davidson, Gertjan Bon, Nicole Galalla, Mariska Beekman, Pieter Slot, Bas van Beusekom, and Denis Rijnsburger for their technical support, and we are grateful to Gjorgji Strezoski and Nikoletta Tsiarta for their help with the coding of the growth model. We thank Gijs Kuenen for the helpful discussions and Natasha Barbolini for the English proofreading. We declare no competing financial interests.