We performed experiments with S. islandicus strain REY15A under three conditions defined by different pH and temperature combinations (Figure 1). These included the optimal growth condition control (pH 3.4, 76°C), an acid stress condition (pH 2.4, 76°C), and a cold stress condition (pH 3.4, 66°C). All cultures were grown as batch cultures in 1-L Erlenmeyer flasks containing 500-ml DY medium (Zhang et al., 2013) with D-arabinose (Sigma-Aldrich, USA) as the carbon and energy source in place of dextrin, hereafter AY medium. Cultures were incubated in Innova-42 shaking incubators (Eppendorf, Hauppauge, NY, USA) at 150 RPM and 66 or 76°C. To minimize culture evaporation at these high incubation temperatures, open trays of water were maintained in the incubator to ensure a saturated atmosphere. Culture flasks had loose-fitting plastic caps to allow for continuous atmospheric oxygen exchange with the media. We monitored culture growth by optical density measurements at 600 nm (OD₆₀₀) on a Genesys 10S UV-Vis spectrophotometer (Thermo Scientific, Waltham, MA, USA). OD₆₀₀ data were used to calculate growth rates using a rate calculation script from Cobban et al. (2020) and available on GitLab.¹ Calculations of growth rates for each condition only incorporated OD₆₀₀ data from the triplicate cultures designated for early stationary phase harvest given that monitoring of mid-log-designated cultures ceased after harvest.

Experiments for each growth condition consisted of one uninoculated control and six inoculated replicates. Cultures were inoculated with actively growing pre-cultures to starting optical densities (OD₆₀₀) of 0.02. Triplicate cultures were harvested at timepoints corresponding to mid-log and early stationary phase of each growth condition (Figure 2). Harvest timepoints were based on preliminary growth experiments characterizing S. islandicus growth under each growth condition (data not shown). For each harvest, biomass samples from the triplicate cultures were sacrificed, processed, and preserved for core lipid analysis, RNA sequencing, and proteomics (Figure 1). Given the anticipated biomass requirements for each downstream analysis and the variable OD₆₀₀ of each growth phase and condition, sampling volumes differed between experiments and are summarized in Supplementary Table 1. We also measured culture pH at each harvest to constrain average pH drifts over the course of growth under all conditions (Supplementary Table 1).

For each sampling time point, we prioritized biomass for RNA-seq as follows. The harvested culture was immediately mixed with an equal volume of chilled, sterile basal salts medium (AY medium without arabinose or tryptone) and placed on ice until centrifugation. Biomass pellets were collected by centrifuging the culture:basal salts solution for 18 min in an Eppendorf centrifuge with swinging bucket rotor set at 3,163 × g (max speed) and 4°C, and decanting the resulting supernatant. Pellets were immediately placed on dry ice and then transferred to a −80°C freezer until RNA extraction. In parallel, to preserve biomass for proteomics and lipid extractions, harvested culture was centrifuged (3,163 × g, 4°C, 18 min) and decanted, and resulting pellets were frozen in a −80°C freezer until further processing.

Frozen biomass pellets of S. islandicus harvested from growth experiments and designated for lipid extraction were freeze-dried prior to lipid extraction. We obtained core GDGT fractions from freeze-dried biomass using acid hydrolysis and solvent extraction as described previously, with modified 3N methanolic HCl incubation conditions of 65°C for 90 min (Zhou et al., 2020). Given that calditol-linked GDGTs found in Sulfolobales are resistant to hydrolysis by the methanolic HCl method used here, our results only include non-calditol core GDGTs. GDGT abundances were measured by ultra-high-performance liquid chromatography (UHPLC) coupled to an Agilent 6410 triple-quadrupole mass spectrometer (MS) as originally described by Becker et al. (2013) and previously implemented for related organisms (Zhou et al., 2020). To enable analysis of small sample aliquots, data was collected using a Selected Ion Method (SIM) for core GDGTs only. The resulting peak areas of GDGTs in each sample were used to calculate ring indices (RI) according to Eq. 1. Like previous studies of S. acidocaldarius (Cobban et al., 2020; Zhou et al., 2020), multiple isomers of GDGTs 3–5 were detected in some S. islandicus samples. Peak areas of minor isomers were summed with their respective major components before RI calculation.

Two sample t-tests were used to identify significant differences (p-value < 0.05) in RI and individual GDGT abundances between growth conditions.

Acid and cold stress impaired S. islandicus growth relative to optimal conditions (Figure 2). Optimal growth conditions yielded the fastest average doubling time observed: 6.8 ± 0.1 h. Doubling times under acidic pH and cold temperature stress conditions were 9.8 ± 0.1 and 14.0 ± 0.5 h, respectively. Maximum culture densities also differed among growth conditions (Figure 2). Cultures grown under optimal conditions reached an average maximum optical density of 1.48 ± 0.06 while acidic pH and cold temperature cultures reached average maximum optical densities of 0.75 ± 0.11 and 1.69 ± 0.04, respectively. An ANOVA of average maximum cell densities indicated that at least one mean significantly differed from the others (p < 0.001), and a post-hoc Tukey HSD test showed that the acidic pH cultures had a significantly lower maximum cell density than optimal or cold temperature cultures.

Saccharolobus islandicus produced GDGT-0 through -6 in all experimental conditions, while average GDGT cyclization differed among growth conditions and growth phases (Figure 3). Small amounts of GDGT-7 occurred in profiles from optimal and acid stress conditions, but its relative abundance never exceeded 1% (Supplementary Table 2). As previously observed in S. acidocaldarius, we also detected the late eluting isomers (denoted by ′ symbol) of GDGTs 3–5 in S. islandicus lipid profiles (Cobban et al., 2020; Zhou et al., 2020). Isomers were particularly abundant (up to 10% relative abundance) in GDGT profiles from acid stress conditions (Supplementary Table 2). The average cyclization of GDGTs, as interpreted using Ring Index (RI) values, ranged from 1.59 to 3.93 for each combination of growth condition and growth phase (Figure 3). Average RI values significantly increased from mid-log phase to early stationary phase in optimal and cold stress conditions, indicating greater GDGT cyclization during later growth (Figure 3). Average RI increased at later growth under acid stress as well, but not significantly (p = 0.09).

Both acid stress and cold stress induced significantly decreased average RI values relative to optimal conditions at each growth phase (Figure 3). At mid-log growth phase, S. islandicus grown under optimal conditions had an average RI value of 3.65 ± 0.05 while acid and cold stress resulted in average RI values of 1.59 ± 0.10 and 2.83 ± 0.05, respectively. Similarly at early stationary phase, optimal conditions yielded an average RI value of 3.93 ± 0.03 while acid and cold stress resulted in average RI values of 2.08 ± 0.29 and 3.67 ± 0.04, respectively. Though both stressors resulted in lower RI, comparison of GDGT profiles resulting from each stress show that acid and cold conditions induced distinct changes of different individual GDGT abundances (significance determined by t-test). For example, cold stress GDGT profiles showed decreased relative abundances of all detected GDGTs with ≥5 rings while GDGT-5 was the only higher-numbered ring showing decreased abundance in acid stress profiles (Figure 3). Unlike cold stress, the decrease in RI under acid stress conditions was driven largely by increased abundances of GDGTs 0–2 and a dramatic reduction of GDGT-4 (Figure 3). Differences in the underlying GDGT abundance changes induced by each stress condition highlights the value of interpreting complete lipid profiles in addition to RI values.

We obtained a total of 108.4 million raw reads from the 18 S. islandicus RNA samples derived from our growth experiments. Raw reads per sample averaged at 6.0 million (range: 3.7–10.1 million). After quality filtering and mapping fragments to the S. islandicus REY15A genome, an average of 91% (range: 65–95%) of fragments could be assigned to genes. Complete read and mapping data can be found in Supplementary Table 3.

To understand how S. islandicus transcriptomes responded to acid and cold stress, we compared the RNA-seq data from either stress condition versus optimal growth conditions as our control for baseline expression. Differential gene expression analysis showed that acid stress and cold stress each induced differentially expressed genes (DEGs) at both sampled growth phases though more DEGs occurred at early stationary phase in both stress conditions (Figure 4A).

We compared the DEGs from both stresses by growth phase and directionality (up or downregulated) to examine the number of DEG responses that were unique to either stress or common to both stresses in this study (Figure 4B). At both growth phases and for both expression directions, acid stress always yielded the highest proportion of observed DEGs (Figure 4B). At early stationary phase, a significant proportion of DEGs were commonly induced by both acid and cold stress, potentially representing more generalized stress responses by S. islandicus to environmental perturbations. Complete differential gene expression results are available in Supplementary Table 4. We present specific RNA-seq results with an initial focus on GDGT biosynthesis-related DEGs followed by an overview of broader DEG responses grouped by inducing stressor(s).

Quantitative proteomics revealed protein level changes associated with acid and cold stress responses of S. islandicus. We detected a total of 1,107 unique proteins, which account for 42% of the total predicted proteins (2,644) encoded by the S. islandicus REY15A genome. Differential abundance analysis revealed that 18 proteins significantly changed abundance levels under acid stress and 7 proteins significantly changed abundance under cold stress (Table 1). Of all the observed significant protein abundance changes, only six showed consistent transcriptomic changes at corresponding growth phase and stress condition including the heat shock protein (SiRe_2601) downregulated at early stationary phase under acid stress (Table 1). Another nine significant protein changes showed transcriptional changes at corresponding growth phase and stress condition, but with opposite directionality (Table 1 and Figure 7).

Some of the largest magnitude protein level changes in S. islandicus included the downregulation of formate dehydrogenase alpha subunit (SiRe_2464; log2FC = −3.31) and the upregulation of UDP-sulfoquinovose synthase Agl3 (SiRe_2623; log2FC = 2.98), both under acidic stress conditions during early stationary phase. SiRe_2464 is homologous to the non-canonical formate dehydrogenase (FDH) YjgC subunit from Bacillus subtilis (E-value 0.0, 53% identities), which complexes with a second subunit to form a FDH that couples formate oxidation with menaquinone reduction (Arias-Cartin et al., 2022). Agl3 forms UDP-sulfoquinovose, a component of S-layer and archaellum N-glycans that was important for S. acidocaldarius growth under high salt conditions (Meyer et al., 2013). Cold stress resulted in a strongly decreased abundance of the arginine biosynthesis protein argininosuccinate lyase ArgH (SiRe_1371; log2FC = −2.83) during mid-log phase. Despite these proteins showing some of the highest magnitude changes in our proteomic dataset, none had consistent transcriptional changes at corresponding growth phase and stress conditions, suggesting S. islandicus might engage significant post-transcriptional regulation in its acid and cold stress responses.

As expected, acid and cold stress both induced shifts in the average cyclization of S. islandicus GDGTs. While each stress caused changes to the relative abundances of different individual GDGTs, both sets of stress-induced changes resulted in lower RI values indicating lower average cyclization (Figure 3). The GDGT responses of various archaeal groups to a wide variety of stressors have been reviewed in the literature (e.g., Oger and Cario, 2013; Law et al., 2020; Rastädter et al., 2020), but we limit our discussion here to comparisons between our data and GDGT studies examining similar stress conditions. Lower RI at lower temperature is consistent with previous laboratory studies of thermoacidophiles as well as mesophilic neutrophiles (Figure 3; De Rosa et al., 1980a; Uda et al., 2001, 2004; Boyd et al., 2011; Elling et al., 2015; Jensen S. et al., 2015; Feyhl-Buska et al., 2016; Cobban et al., 2020).

Shifts in archaeal GDGT cyclization in response to pH changes are more complex. We observed lower RI at lower pH in S. islandicus (Figure 3), which is consistent with previous laboratory experiments with S. acidocaldarius (Cobban et al., 2020). However, a coherent empirical relationship between archaeal GDGT cyclization and pH has not yet been established (Siliakus et al., 2017). Moreover, comparison of our data to prior reports of GDGT shifts in response to pH is difficult given that several other studies only examined effects of higher pH’s rather than lower pH (e.g., Shimada et al., 2008; Boyd et al., 2011). Thermoacidophilic archaea such as S. islandicus maintain circumneutral intracellular pH while living in extremely acidic environments, resulting in steep proton gradients across their membranes. Shifting GDGT cyclization serves as a pH homeostasis mechanism for thermoacidophiles to maintain these proton gradients (Baker-Austin and Dopson, 2007; Slonczewski et al., 2009). Highly cyclized GDGTs form more tightly packed and less permeable membranes that may help limit passive proton diffusion and in turn protect against cytoplasmic acidification (Gabriel and Chong, 2000; Shinoda et al., 2005; Baker-Austin and Dopson, 2007). Thus, we initially predicted S. islandicus might increase RI as one means of counteracting the higher external concentration of protons at lower pH. In contrast, S. islandicus exhibited lower RI at more acidic pH, indicating that its membrane was more permeable. If increased membrane permeability allowed greater proton diffusion into the cell, the stunted maximum cell density of our cultures grown under acid stress (Figure 2) could represent cells succumbing to cytoplasmic acidification. Other factors can affect archaeal membrane permeability including GDGT headgroup composition and it follows that quantitative headgroup analysis and/or direct cytoplasmic pH measurements would help validate our proposed hypothesis (Oger and Cario, 2013).

Under both stress conditions, S. islandicus GDGT compositions differed by growth phase. RI increased at early stationary phase relative to mid-log phase within both the acid and cold stress conditions, as well as in optimal growth conditions (Figure 3). This is consistent with previous observations from other archaeal batch studies sampled at similar growth phases including S. islandicus and S. acidocaldarius as well as the with Picrophilus torridus and Nitrosopumilus maritimus (Elling et al., 2014; Jensen S. et al., 2015; Feyhl-Buska et al., 2016; Cobban et al., 2020). We suggest that our observations from S. islandicus are also consistent with the findings from continuous culture experiments of S. acidocaldarius and N. maritimus showing that RI increases as electron availability decreases (Hurley et al., 2016; Quehenberger et al., 2020; Zhou et al., 2020). As our batch cultures grew, S. islandicus actively consumed medium nutrients and thereby depleted its available energy supply. Here we observed increased RI values from mid-log to early stationary phase, which we infer as a transition to more energy limited conditions.

Notably, S. islandicus RI and doubling time did not linearly covary when comparing data from all tested growth conditions suggesting that RI changes are not a result of growth rate alone (Figure 8). pH and temperature are distinct physical parameters that likely affect cell physiology and thereby GDGT composition differently. If we consider the general effects of acid stress and cold stress independently, it is clear that both stressors induced slower growth and decreased RI relative to optimal conditions. These observations are consistent with the broad trend of environmental stress causing longer doubling times and decreased RI as observed in S. acidocaldarius (Cobban et al., 2020).

To provide context for the shifts in S. islandicus GDGT compositions induced by acid and cold stress, we examined expression changes of known GDGT biosynthesis genes under each condition. Most notable were the expression changes in the S. islandicus GDGT ring synthase genes, grsA and grsB. The grs genes encode the GrsA and GrsB enzymes that are required for cyclopentane ring formation in S. acidocaldarius and S. solfataricus (Zeng et al., 2019). GrsA produces GDGTs 1 through 4 and GrsB subsequently acts on GDGT-4 to produce GDGTs 5 through 8 (Zeng et al., 2019). A recent study using RT-qPCR and Western blot analysis examined GDGT ring synthase expression in S. acidocaldarius cultured at more acidic pH or colder temperature than its optimal conditions of pH 3.5 and 75°C (Yang et al., 2023). Acid stress (pH 2.5) induced increased grsB and GrsB expression in S. acidocaldarius while cold stress (65°C) caused decreased grsB expression and undetectable levels of GrsB (Yang et al., 2023). In contrast, S. acidocaldarius did not show differential expression of grsA nor GrsA in either stress condition at 48 or 72 h timepoints, which are the closest analogs to the mid-log and early stationary phase timepoints in our experiments (Yang et al., 2023). Results from our study indicated similar expression patterns of the grs genes in S. islandicus grown under acid stress or cold stress. S. islandicus did not differentially express grsA or GrsA in response to either stress, supporting constitutive transcription of grsA and insignificant GrsA protein level changes (Figure 5). However, S. islandicus showed differential expression of grsB as responses to both stress conditions in alignment with observations from S. acidocaldarius (Figure 5; Yang et al., 2023). Under cold stress at both mid-log and early stationary phases, S. islandicus downregulated grsB, and consistent with the gene’s expected function, cold stress lipid profiles had lower abundances of GDGTs with ≥5 rings (Figure 3). Under acid stress, S. islandicus upregulated grsB at mid-log phase, but changes in acid stress lipid profiles did not correspond as predicted by GrsB function. That is, no GDGTs with ≥5 rings significantly increased under acid stress (Figure 3). The isomer of GDGT-5 (GDGT-5′) increased under acid stress, but did not result in an overall increased abundance of GDGTs with 5 rings. The inconsistency between increased grsB transcription and the predicted changes in acid stress GDGT profiles may be related to the fact that like most of the DEGs in this study, the abundance of GrsB protein did not increase in tandem with higher transcription. Additionally, GrsB requires GDGT-4 as a substrate to produce more cyclized GDGTs, and GDGT-4 abundance was starkly lower in S. islandicus acid stress lipid profiles relative to optimal conditions (Figure 3; Zeng et al., 2019). Finally, GDGT profiles are likely not only affected by the GDGT ring synthase transcription and protein levels quantified here, but also by variable Grs enzyme activity under different temperature or pH conditions and by potential reworking of GDGT cyclization by as yet unidentified enzymes. Overall, acid and cold stress both affected the regulation of grsB in S. islandicus, but only cold stress conditions resulted in differential grsB expression and GDGT-level changes consistent with GrsB function. Our results suggest that transcriptional data may be helpful for predicting GDGT-level changes in cold stress conditions, but do not reflect Grs enzyme activity under all stress conditions.

Differential gene expression of other GDGT biosynthesis genes suggested that S. islandicus may have adapted its membrane to acid and cold stress by means other than GDGT cyclization. For example, the polar headgroups attached to membrane lipids can play a role in controlling membrane permeability (Shimada et al., 2008; Oger and Cario, 2013). Although we did not quantify headgroups in our experiments and despite only a few known headgroup synthesis genes, our RNA-seq data indicated that S. islandicus upregulated AIP phosphatase (SiRe_0209) under acid stress, suggesting that the proportion of archaetidylinositol headgroups may have increased in response to acidic conditions. We also observed differential expression of two sequential GDGT biosynthesis genes in response to acid and cold stress, though the directionality of their expression changes differed. S. islandicus upregulated GGPP synthase (SiRe_1931) and downregulated GGGP synthase (SiRe_1736) (Figure 5). The downregulation of GGGP synthase and the lack of differential expression changes of numerous other GDGT synthesis genes (Figure 5) suggest that S. islandicus did not increase overall core lipid synthesis in response to acid or cold stress. However, the upregulation of GGPP synthase suggests increased production of GGPP, which is an intermediate for the syntheses of multiple polyterpenes beyond just core lipids. A potentially increased supply of GGPP could have been funneled toward the synthesis of polyterpenes like quinones and polyprenols, which have been proposed as archaeal membrane regulators for environmental stress response (Salvador-Castell et al., 2019). Broader characterizations of archaeal membrane components in response to environmental stressors could help shed light on these other adaptative mechanisms.

RNA-seq enabled a comprehensive overview of S. islandicus’ acid and cold stress responses beyond just lipid biosynthesis. Overall, S. islandicus differentially expressed many more genes during early stationary phase than mid-log phase under both stresses (Figure 4), possibly related to prolonged exposure to a given stressor and/or onset of growth limitation. These observations show the value in analyzing stress-induced expression changes at multiple growth phases and at later growth phases in particular. We also found that both acid and cold stress induced many of the same transcriptional responses in S. islandicus, which we propose to represent candidate general stress response genes that can be examined in future studies testing different environmental stressors. One common transcriptional response to both acid and cold stress included downregulation of ETC terminal oxidases, suggesting decreased oxidative phosphorylation and thereby a reduction in ATP generation and proton pumping. The drop in ATP supply would stymie energy-requiring processes, which is consistent with the slower growth rates we observed under both stress conditions (Figure 2). In addition to a slowed growth rate, acid stress also caused a significant decrease in maximum cell density not seen in cold stress conditions (Figure 2). Decreased growth yield suggests that S. islandicus acid stress response was energetically costly and/or that S. islandicus could not successfully counteract the negative effects of acid stress.

Cell motility also decreased in response to both stressors. Expression changes of the archaellum and related regulator genes indicated that S. islandicus decreased archaellum activity in response to both acid and cold stresses, contrasting the increased motility response by S. acidocaldarius to nutrient limitation (Szabó et al., 2007; Lassak et al., 2012; Haurat et al., 2017; Hoffmann et al., 2017; Bischof et al., 2019). This difference could indicate that unlike nutrient limitation, cells facing pH or temperature stress may conserve energy rather than swimming to find more favorable conditions.

Saccharolobus islandicus upregulated other cellular processes as common responses to acid and cold stress. Transcription and translation activities increased, based on upregulation of genes encoding for RNA polymerase, ribosomal proteins, and translation initiation factors. Differential expression of VapBC toxin-antitoxin systems also indicated changes in translation across our experimental conditions. Upregulated HRR genes and purine synthesis genes suggest increased DNA repair activities, and increased expression of anaplerotic reaction genes suggests increased biosynthetic activity. Accordingly, DEGs supported increased biosynthesis of polyamines, histidine, cobalamin, and chorismate in response to both stressors.

The upregulation of polyamine synthesis under both acid and cold stress was notable given that polyamines are commonly referenced as cytoplasmic buffers and thereby an acid-specific adaptation (e.g., Slonczewski et al., 2009; Lewis et al., 2021). While upregulation of polyamine synthesis genes may have ameliorated cytoplasmic acidification in acid stress conditions, the advantage of increased cytoplasmic buffering is less clear under cold stress. As such, we suggest that polyamines may have contributed to the acid and cold stress responses of S. islandicus by some other mechanism(s). For example, polyamines are known to play an important role in translation as they are essential for ribosome activity (Londei et al., 1986). Given that observed DEGs indicated increased translation in response to both acid and cold stress, we propose that S. islandicus could have increased polyamine synthesis to support translation activity. Overall, upregulated polyamine synthesis as response to both stressors suggest that polyamines may contribute to the S. islandicus acid and cold stress responses by multiple mechanisms.

Low pH induced a strong physiological response in S. islandicus relative to cold temperature. The number of DEGs, magnitude of RI changes, and magnitude of cell density impairment were all greater in response to acidic stress. Consistent with the damage to protein and nucleic acids that can result from cytoplasmic acidification (Lund et al., 2014), acid-specific DEGs showed that S. islandicus upregulated genes related to degradation and repair processes for molecules including proteins, nucleic acids, and fatty acids. One of the most notable acid-specific transcriptional responses involved upregulation of an A-type ATPase that we suggest functions as a proton pump based on evidence from enzyme purification and gene sequence studies (Konishi et al., 1987; Lübben and Schäfer, 1987; Hochstein and Stan-Lotter, 1992; Grüber et al., 2014). Proton pumping expels protons from the cytoplasm and is a common mechanism to combat pH stress (Baker-Austin and Dopson, 2007; Slonczewski et al., 2009). Terminal oxidases of the ETC also function as proton pumps, but S. islandicus downregulated all three of its terminal oxidases under acid stress suggesting a need for an alternate proton pump like the A-type ATPase. Altogether, the acid-specific DEGs such as the A-type ATPase provide insight to the ways that S. islandicus is affected by and responds to decreased pH.

The insertion sequence (IS) DEGs observed in our experiments notably showed opposite changes in expression between acid and cold stress conditions. Under both stressors, S. islandicus differentially expressed IS elements belonging to the IS200/605, ISH3, or ISH6 families of transposable elements, which are restricted to, or ancestral to, archaea (Filée et al., 2007). Acid stress induced consistent downregulation of these transposable elements whereas cold stress induced almost entirely upregulation shifts. These observations suggest that transposition frequency in S. islandicus may increase under cold stress and decrease under acid stress.