The Larsemann Hills are an ice-free oasis situated in the Prydz Bay region in East Antarctica (69°30′S, 76°20′E; Stüwe et al., 1989). The study site is characterized by severe climatic conditions with low temperatures between -29°C and 0°C and little precipitation of about 250 mm a^-1 (Hodgson et al., 2001). Soil formation processes could not be observed and the studied material can be classified as dry and oligotrophic weathering debris. A detailed site description is given in an environmental study dealing with bacterial succession in two glacier forefields of the Larsemann Hills (Bajerski and Wagner, 2013). The studied bacterial strain was isolated from the weathering debris of the glacier forefield called “Black Valley Transect” in 446 m distance from the glacier (S 69°24.315; E 76°20.295; Bajerski et al., 2013). The harvested profile was characterized by a black organic mat mixed with coarse-grained sandy material at the surface and sandy permanently frozen ground in 1–6 cm depth. The soil pH of the study site ranged from acidic (pH 4.9) to alkaline (8.3), but the studied strain originated from a sampling site with almost neutral pH (6.8).

Chryseobacterium frigidisoli strain PB4^T was chosen to examine the effect of changing temperature and pH on the fatty acid composition of the bacterial cell membrane. This strain was described as a novel psychrotolerant bacterium (Bajerski et al., 2013) being able to grow between 0 and 25°C with an optimum growth at 20°C (Figure 1A). The gram-negative bacterium grows at pH values from pH 5.5–8.5 with optimum at pH 6.5 (Figure 1B). The turbidity of the cell culture was measured photometrical at 600 nm to determine the optical density (OD) during an incubation in R2A media (0.05% Proteose peptone, 0.05% Casamino acids, 0.05% yeast extract, 0.05% dextrose, 0.05% soluble starch, 0.03% dipotassium phosphate, 0.005% magnesium sulfate 7× H₂O, 0.03% sodium pyruvate (w/v); pH 7.2; Reasoner and Geldreich, 1985). The specific growth rate μ was determined in the exponential growth phase (linear range) using the measured ODs and was calculated as follows:

To study temperature adaption the strain was incubated at 0, 5, 10, 14, 20°C in R2A medium (pH 7.2). The impact of changing pH values on the membrane composition was analyzed using the R2A medium (Reasoner and Geldreich, 1985) modified with the following buffers: glycine (pH 4.0–5.0 and pH 10.0), MES [2-(N-morpholino)ethanesulfonic acid; pH 5.0–6.5], HEPES [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; pH 7.0-8.0], and BTP ([1,3-bis(tris(hydroxymethyl)methylamino)propane]; pH 8.5–9.5; Hoaki et al., 1994). Growth in dependence of pH was detected at pH 5.5, 6.0, 6.5, 7.0, 7.5, 8.0, and 8.5 after incubation at 18°C (in the range of the temperature optimum). For PL extraction cells from biological triplicates were harvested at the transition from late exponential to early stationary growth phase by centrifuging the combined cultures for 5 min at 10.000 g, washing with sterilized tab water and centrifuging again.

The cell pellets of C. frigidisoli cultivated under different temperature and pH conditions were extracted with a flow blending system using a solvent mixture of methanol (MeOH)/dichloromethane (DCM)/ammonium acetate buffer with a ratio of 2:1:0.8 (v/v) for 5 min (modified after Bligh and Dyer, 1959). Afterwards, the solvent ratio of the extract was changed to 1:1:0.9 by adding DCM and ammonium acetate buffer resulting into a phase separation of an organic and an aqueous phase. Subsequently, the aqueous phase was re-extracted three times with DCM. The combined organic phases were concentrated using a Turbo Vap system (Zymark). Afterwards, the extract was chromatographically separated into fractions of different polarity resulting, among other fractions, into a polar lipid (PL) fraction as described in Zink and Mangelsdorf (2004). This standard protocol was used mainly to clean up the PL fraction and to remove free fatty acids from the extract, thus they cannot mix up with the side chain fatty acids from the polar membrane lipids. Here we focused on the PL fraction since it contains the main cell membrane lipids of C. frigidisoli. An aliquot of the PL fraction was used for trans-esterification to obtain the methylated membrane fatty acids by following the method described by Müller et al. (1990). The aliquot of the PL fraction was dissolved in 50 μl of DCM/MeOH (9:1 v/v) in a 2 ml vial. Afterward 50 μl of trimethylsulfoniumhydroxid (TMSH) was added and the vial was sealed before placing in an oven for 2 h at 70°C. The trans-esterified (methylated) fatty acids were dried and subsequently dissolved in 50 μl DCM before analysis on a gas chromatographic system (Trace GC Ultra, Thermo Electron) coupled to a DSQ Thermo Electron Quadrupole mass spectrometer. The GC was equipped with a cold injection system operating in the splitless mode and a SGE BPX 5 fused silica capillary column (50 m length, 0.22 mm ID, 0.25 μm film thickness) using the following temperature program: initial temperature 50°C, 1 min isotherm, heating rate 3°C/min to 310°C, held isothermal for 30 min. Helium was used as a carrier gas with a constant flow rate of 1 mL min^-1. The injector temperature was programmed from 50 to 300°C at a rate of 10°C s^-1. Full scan mass spectra were recorded from m/z 50 to 600 at a scan rate of 2.5 scans s^-1.

Correlation and significance of the derived data was calculated with IBM SPSS Statistics 21 analyzing bivariate correlation using the Pearson correlation coefficient and testing two-tailed significance.

The detected polar membrane lipids of C. frigidisoli are phosphatidylethanolamines, ornithine lipids as well as flavolipins and flavocristamides. The major fatty acids (>10%, Figures 2, 3 and Tables 1, 2) determined in this study obtained after saponification of membrane lipids are iso- and anteiso-pentadecanoic acid (iso-, anteiso-C_15:0), iso-2-hydroxypentadecanoic acid (iso-2OH-C_15:0), iso-heptadeca-9-enoic acid (iso-C_17:1ω7) and a novel fatty acid, which, to our knowledge, has not been previously described. The structure of the newly discovered fatty acid was identified as anteiso-heptadeca-9,13-dienoic acid, a ω3,7-fatty acid with a branching position directly at the ω3 double bond. The experiments, which were performed to identify the detailed structure of the newly discovered fatty acid, are described elsewhere (Mangelsdorf et al., 2017). The identified major fatty acids are in accordance with the fatty acid composition of the strain description that identified iso-2OH-C_15:0 and iso-C_15:0 as the major fatty acids (Bajerski et al., 2013). Minor variations between this study and the presented results can be explained by different harvesting time points. In the strain description of Bajerski et al. (2013) cells were harvested at the end of the exponential growth phase, whereas in this study cell material was collected at the transition from late exponential to early stationary growth phase. The predominance of iso-2OH-C_15:0 and iso-C_15:0 at optimum growth seem to be shifted to other C₁₅ and longer unsaturated C₁₇ fatty acids under sub-optimal conditions (stress/stationary phase). While iso-C_15:0 is predominant in many Chryseobacterium species (Vandamme et al., 1994), the high amount of iso-2OH-C_15:0 appears to be distinctive in C. frigidisoli. It should be noted that the iso-C_17:1ω7 fatty acid was wrongly labeled as iso-C_17:1ω8 in Bajerski et al. (2013). Other fatty acids such as iso-tridecanoic acid (iso-C_13:0), iso-tetradecanoic acid (iso-C_14:0), pentadecanoic acid (C_15:0), iso-hexadecenoic acid (iso-C_16:1, presumably iso-C_16:1ω6), iso-hexadecanoic acid (iso-C_16:0), iso-2-methoxy-pentadecanoic acid (iso-2methoxy-C_15:0), anteiso-2-hydroxy-pentadecanoic acid (anteiso-2OH-C_15:0), hexadecanoic acid (C_16:0), anteiso-heptadec-9-enoic acid (anteiso-C_17:1ω7), octadec-9-enoic acid (C_18:1ω9) and octadecanoic acid (C_18:0) occur only in minor to trace amounts accounting in total for less than 10 % of the total cell membrane derived fatty acids pattern (Tables 1, 2). In some of the minor unsaturated components the exact double bond position could not be assigned to a standard or identified by derivatization experiments (Mangelsdorf et al., 2017), thus, they are simply labeled as unsaturated fatty acids (Figures 3, 4 and Tables 1, 2, respectively).

Overall, the fatty acid inventory of C. frigidisoli is characterized by branched (iso/anteiso) and unsaturated fatty acids in the shorter chain range between 15 and 17 carbon atoms. These features are used for instance by cells that are able to grow under harsh living conditions such as low temperatures or changing biogeochemical gradients to maintain membrane fluidity and functionality (Denich et al., 2003). By those means C. frigidisoli is well adapted to a psychrotolerant life style, because shorter chain fatty acids, branches and unsaturations increase the fluidity of the cell membrane at low temperatures (Russell, 1984). Furthermore, the whole genus Chryseobacterium is characterized by the presence of branched C₁₅ and C₁₇ fatty acids (Vandamme et al., 1994). The incorporation of anteiso-fatty acids, as it is shown in this study or for other psychrotolerant species such as C. frigidum (Kim et al., 2016) and C. haifense (Hantsis-Zacharov and Halpern, 2007) lowers the melting temperature of the cell membrane enhancing motion ability. Anteiso-C_15:0 (25.5°C) for instance has a critical lower main phase transition temperature than iso-C_15:0 (52.2°C; Suutari and Laakso, 1994). The incorporation of an unsaturation leads to an even stronger decrease in the melting temperature, while for instance a C_16:0 fatty acid has a solid-liquid phase transition temperature of 63°C, whereas C_16:1ω7 has a melting temperature of –1°C (Knothe and Dunn, 2009). In line with this, C. antarcticum incorporates iso-C_15:1 as the major fatty acid and C. frigidisoli and its closest relatives C. humi (Pires et al., 2010) and C. marinum (Lee et al., 2007) contain various unsaturated C₁₇ fatty acids in their cell membranes. In contrast, saturated and longer chain fatty acids (>C₁₇), representing adaptation to warmer conditions, play only a minor role in the fatty acid inventory of C. frigidisoli. In general, the genus Chryseobacterium comprises several psychrotolerant species isolated from cold-affected habitats indicating that Chryseobacterium have developed suitable genetic, biochemical, and physiological adaptations for a life at low temperature (Yi et al., 2005; Zhao et al., 2011; Bajerski et al., 2013).

As discussed in the previous paragraph the cell membrane of C. frigidisoli predominately contains branched shorter chained fatty acids partly with unsaturation as an adaptation to a psychrotolerant lifestyle. In the current study, we conducted temperature cultivation experiments at 0, 5, 10, 14, and 20°C. Between 20°C and 14°C the relative proportions of the fatty acids are not much different, however, below 14°C drastic changes are visible (Figure 4). This pattern is in accordance with the growth rate curve (Figure 1A) showing a significant change in the growth rate at 10°C. Generally, the fatty acid inventories at different cultivation temperatures are composed of the same fatty acids (Figure 2). However, there are differences in the relative distributions of the membrane fatty acids. While the anteiso-C_15:0 is more abundant than the iso-congener in the 0°C culture, the ratio of iso- to anteiso-C_15:0 changes drastically with increasing cultivation temperature with a clear dominance of the iso-C_15:0 in the 20°C culture (Figure 4). This change is clearly resembled in the iso/anteiso-C_15:0 ratio in Table 1. The relative proportion of iso-2OH-C_15:0 constantly increases from the 0°C to the 20°C culture (Figure 4), while the anteiso-2OH-C_15:0 shows an overall slight decrease between 0°C and 10°C and a more or less constant trend between 10°C and 20°C (Figure 4). The ratio of iso/anteiso-2OH-C_15:0 (Table 1) shows overall an increasing trend from 0°C to 20°C, however, the ratio value of the 10°C culture falls out of this trend due to the lower amount of anteiso-2OH-C_15:0. The change of the iso/anteiso-ratio allows the organism to adapt the cell membrane melting temperature in response to different ambient temperatures. The importance of anteiso-branched fatty acids at low temperatures was also reported in the temperature adaptation of Bacillus subtilis by Klein et al. (1999). They observed a relative dominance of anteiso-C_15:0 and anteiso-C_17:0 in the fatty acid inventory in cold shock experiments with B. subtilis.

Another adaptation to ambient temperature can be the chain length of the membrane fatty acids, which can be reduced with decreasing temperatures, since the melting temperature of the respective membrane fatty acids decreases with shorter chain length (Russell, 1989; Suutari and Laakso, 1994). However, in our study on C. frigidisoli the chain length does not seem to be a major regulation mechanism (Table 1) in response to changing temperature regimes. The overall chain length (short vs. long chain length) does not change significantly with temperature. Nevertheless, the fatty acid inventory of C. frigidisoli generally contains a high amount of shorter-chained fatty acids and the species uses other mechanisms (iso/ai ratios, unsaturation) to maintain the cell membrane fluidity at low temperature. In contrast to the chain length, temperature adaptation via the relative abundance of unsaturated fatty acids seems to play a significant role. The highly abundant monounsaturated iso-C_17:1ω7 fatty acid shows a strong increase from 20°C to 10°C, which can be interpreted as a temperature adaptation, since a higher relative proportion of unsaturated fatty acids lowers the membrane solid-liquid phase transition temperature (Russell, 1989), in strain PB4^T leading to an increased fluidity of the cell membrane (Figures 2, 4). At the same time the iso-C_15:0 strongly decreases. Thus, between 20 and 10°C there is a trend from iso-C_15:0 to iso-C_17:1ω7 (Figure 4). The iso-C_17:1ω7 has indeed a longer chain length (increase of melting temperature) but the unsaturation (strong decrease in melting temperature) more than compensates the chain length effect and seems to be an adaptation towards lower ambient temperatures. Unsaturation of fatty acids is a very effective temperature adaptation, because it has a higher effect on lipid fluidity and can be either realized by de novo synthesis or modification (desaturation) of already existing fatty acids (Russell, 1984). The longer chain length of iso-C_17:1ω7 might attenuate the adaptation effect introduced by the unsaturation. Below 10°C the relative amount of iso-C_17:1ω7 decreases again. The same is observed for other iso-fatty acids. In contrast, the anteiso-fatty acids generally increase between 10°C and 0°C (Figure 4). Therefore, below 10°C the data indicate that there is an interchange from iso-fatty acids to more anteiso-fatty acids, which represents a clear adaptation to cooler conditions (Kaneda, 1991). This interchange point in the fatty acid composition is also shown in the growth curve (Figure 1A) indicating that C. frigidisoli experiences stress at temperatures ≤10°C. Additionally, the anteiso-C_17:2ω3,7 continuously increases (Figures 2, 4) from 20°C to 0°C indicating a strong temperature adaptation of C. frigidisoli to lower ambient temperatures. The statistical analysis shows that this anteiso and bis-unsaturated fatty acid strongly correlates with temperature (p < 0.01, R² = 0.93). Its relative proportion increases with decreasing temperature and accounts for the highest amount of all fatty acids (25.73%) at 0°C (Table 1 and Figures 2, 4). Thus, the anteiso-C_17:3ω3,7 fatty acid seems to be an important and efficient regulation component for temperature adaption in C. frigidisoli. Detailed studies on changes in the PFLA profiles of microorganisms especially from extreme habitats are still rare and the discovery of the new fatty acid shows that the microorganisms develop various ways of stress response. There are several studies reporting the critical role of reducing the fatty acid chain length and altering the branching from iso to anteiso in reaction to low ambient temperatures (Paton et al., 1978; Annous et al., 1997). Russell (1984) stated that unsaturation is more effective than chain length reduction in low temperature adaptation. However, Mangelsdorf et al. (2009) showed that in microbial communities from Siberian permafrost the temperature regulation was mainly realized by chain length and the trend in the relative proportion of saturated and unsaturated fatty acids did not change significantly. Chain length and desaturation played also a minor role in the cold shock response of Bacillus subtilis and evidence for the dominance of anteiso branched fatty acids in the adaptation to low temperature was given (Klein et al., 1999). In our study, we observe a high importance of iso- and anteiso-branched fatty acids in cold adaption. Unsaturated fatty acids such as the iso-C_17:1ω7 show a concerted interplay between these structural species at different temperatures most likely induced by different temperature stress levels. Additionally, the newly discovered bis-unsaturated fatty acid anteiso-heptadeca-9,13-dienoic acid plays a major role in the cell membrane temperature adaption of C. frigidisoli.

For most of the major fatty acids (iso-C_17:1ω7, iso-C_15:0, anteiso-C_15:0, anteiso-2OH-C_15:0) of C. frigidisoli the relative abundance is rather constant at neutral pH with a plateau phase between pH 6.0 and 8 (Figure 5). An exception forms iso-2OH-C_15:0, which shows a maximum at pH 7.5. At the same time membrane fatty acids at pH minima and maxima (Figure 1B) show different compositions and variable adaptation mechanisms (Figure 5). At pH 8.5 the saturated iso- and anteiso-C_15:0 show a significant increase, while the unsaturated iso-C_17:1ω7, anteiso-C_17:3ω3,7 as well as the hydroxy fatty acids iso-2OH-C_15:0 and anteiso-2OH-C_15:0 are all decreasing. In contrast, at pH 5.5 iso-C_17:1ω7, anteiso-C_17:3ω3,7 and anteiso-2OH-C_15:0 increase, while iso- and anteiso-C_15:0 decrease. The iso-2OH-C_15:0 also decrease at lower pH. Thus, some opposite compositional trends can be observed at the different pH extremes outlined by the growth range of C. frigidisoli (Figure 1B).

Overall, the ratio of iso- to anteiso-C_15:0 is high at neutral pH (dominance of iso-congener) and regardless to the increasing or decreasing trend of the iso- and anteiso-congeners the ratio becomes lower towards the pH extremes indicating an increasing relative importance of the anteiso-branched fatty acids in response to high and low pH (Table 2). The introduction of more anteiso-branched fatty acids supports a higher flexibility of the cell membrane (Russell, 1989). The main challenge for the microorganism in an acidic or alkaline environment is the change of the ion strength that influences proton motive force of the cell membrane. The maintenance of the transmembrane electrical potential (Δψ) and the transmembrane ΔpH is mainly realized using transporters including active proton transport (Krulwich et al., 2011). A higher flexibility of the cell membrane could enable the incorporation of new membrane channels generating the ion transport through the cell membrane in reaction to the changing ion strength at different pH values.

Furthermore, different branching patterns in response to changing pH were discussed in literature before, which partly disagree with the results of the current study. Bååth and Anderson (2003) reported a decrease of iso-C_15:0 and iso-C_16:0 at increasing pH and no significant effect of the anteiso-branched fatty acids, while Männistö et al. (2007) observed an increase of anteiso-C_15:0 and no change of iso-C_15:0 at higher pH in bacterial communities in Arctic Fjelds of Finnish Lapland. Furthermore, an increase of branched (iso, anteiso) fatty acids was observed in response to low pH (Russell, 1984). In our study the relative proportions of the iso- and anteiso-C_15:0 increases at high pH and, interestingly, those of iso- and anteiso-2OH-C_15:0 decrease (Table 2 and Figures 3, 5). Thus, the hydroxy group as the only structural difference among these branched C_15:0 fatty acids seems to be less favorable for adaptation towards higher pH conditions. In contrast at pH 5.5 the iso-C_15:0 and anteiso-C_15:0 fatty acids significantly decrease, while the hydroxy congeners only slightly decrease (iso) or increase (anteiso). To our knowledge, this is the first study showing a pH dependent change in the relative amount of methyl and hydroxy-side chains of the fatty acids and we can only assume that an additional hydroxy group is somehow less favorable at alkaline pH in contrast to a lower pH. For acidophilic archaea it was shown that hydroxy groups on sugar moieties prevent the protons from penetrating the cell membrane at low pH (Wang et al., 2012). This shows that a hydroxylation might be a suitable adaptation mechanism in an acidic environment, which could explain why the relative proportion of the iso- and anteiso-2OH-C_15:0 is only less affected at lower pH.

Another interesting observation concerns the chain length of the respective fatty acids. C. frigidisoli incorporates significantly more short chain fatty acids, predominantly C₁₅, at higher pH (Figures 3, 5). The unsaturated iso-C_17:1ω7 and the anteiso-C_17:2ω3,7 significantly decrease at higher pH (Table 2 and Figure 2). At low pH the opposite trend can be observed, while iso-C_15:0, anteiso-C_15:0, iso-2OH-C_15:0 all decrease (anteiso-2OH-C_15:0 stays rather constant), iso-C_17:1ω7 and anteiso-C_17:2ω3,7 increase (Figure 5). A similar shift from short-chained saturated (pH 7) to long-chained mono-unsaturated fatty acids with decreasing pH was observed in Streptococcus mutans and other oral bacteria as a survival strategy in acidic environment, however, the function of this shift is currently not understood (Quivey et al., 2000; Fozo et al., 2004).

Similar to the temperature cultivation experiments, the anteiso-C_17:2ω3,7 fatty acid also seems to play a role in the cell membrane adaption at different pH conditions. With exception of the culture from pH 8, this fatty acid constantly decreases with increasing pH.

In contrast to the temperature adaptation, interpretation of the adaptation mechanisms concerning changing pH condition is rather difficult; in particular, since literature data are somehow contradicting and show no similar trends. Thus, the function of the observed fatty acid variations in the current study is not clear yet. At high pH the decrease in the long chain unsaturated fatty acids seem to support a compaction/stabilization of the cell membrane whereas the concomitantly shift to more short chain iso- and anteiso-C_15:0 with a relative increase of the anteiso-congener seem to attenuate this effect. At low pH the opposite trend may increase the membrane fluidity. Thus, the pH adaptation of the cell membrane derived fatty acid inventory appears to be a balanced interplay between stabilization and enhanced flexibility of the cell membrane presumably to either prevent or enable protons or other homeostatically active substances passing the cell membrane. Further studies are needed to discover the interrelation between the structural membrane adaptation and the effect on the membrane function at changing pH conditions.