Caldalkalibacillus thermarum TA2.A1 was grown in 5-L flat-bottomed shake flasks at 65°C with a working volume of 1 L rich medium, the composition of which was described previously (McMillan et al., 2009). A smaller pre-inoculum was grown beforehand in a 500-ml round-bottomed shake flask with a 100-ml working volume. The larger shake flasks were inoculated at 1% from the pre-inoculum. The incubations lasted for ~18 h. Cells were harvested by centrifugation at 8,000 × g and were subsequently washed. The fraction appropriated to whole cell proteomics (±100 mg cell wet weight) was washed in ice-cold phosphate-buffered saline (PBS) and thereafter flash frozen with liquid nitrogen and stored at −80°C. The fraction used for membrane proteomics was washed in 50 mM Tris–HCl buffer, pH 8.0, with 10% (w/v) glycerol and 2 mM MgCl₂, and thereafter also flash frozen and stored at −80°C.

Membrane preparation was performed as described earlier (McMillan et al., 2007), with modifications. Briefly, samples were thawed gradually by placing them on top of a layer of ice. Extraction started with a lysis step in 50 mM Tris-HCl buffer, pH 8.0, with 10% (w/v) glycerol, 2 mM MgCl₂, 1 mM phenylmethylsulphonyl fluoride, and 1 mM dithiothreitol. Lysis was initiated by the addition of 1.2 mg ml⁻¹ lysozyme and 0.2 mg DNaseI. Lysed cells were subsequently passed through a cell disruptor twice at 20,000 psi. As a trial, the number of passages through the cell disruptor was decreased to a single passage, but that yielded no discernible difference (Supplementary Figure 1). The resulting liquid fraction, discarding the denatured fraction, was centrifuged at 8,000 × g for 10 min. The resulting supernatant was ultracentrifuged at 180,000 × g for 45 min. The supernatant was discarded, and the pellet was resuspended in ~2 ml of 50 mM Tris-HCl buffer, pH 8.0, with 10% (w/v) glycerol and 2 mM MgCl₂. When required, extracted membranes were flash frozen and stored at −80°C. C. thermarum TA2.A1 membranes were solubilised in 200 μM ammonium-bicarbonate (ABC) buffer with either 2% (w/v) n-tetradecylphosphocholine (FOS-choline-12, Anatrace) or 0.01% (w/v) sodium dodecyl sulfate (SDS) for 30 min at 50°C, 65°C, or 80°C, while gently shaking at 300 rpm. Extracted membrane samples were diluted to a final concentration of 4 mg ml⁻¹ protein. After solubilisation, samples were frozen at −20°C until further preparation.

The following protocols were based on extensive work done before (den Ridder et al., 2022, 2023) to optimize the protocols. For whole cell proteomics, the pellet stored after washing with PBS was thawed, and 29 mg of the sample was used and transferred to a LoBind Eppendorf tube. The pellet was dissolved in 0.175 ml of 1 M triethylammonium bicarbonate buffer, and 0.15 g of glass beads (105–212 μm, Sigma Aldrich) were added. The sample was subjected to a vortexing regime that was repeated thrice; the sample was vortexed at maximum speed for 90 s and then placed on ice for 30 s. Afterward, the sample was incubated at 80°C for 3 min at 1,000 rpm and subsequently ultrasonicated for 10 min. The sonicated sample was centrifuged, and the resulting supernatant (250 μl) was carefully pipetted into a new LoBind Eppendorf tube. The proteins were then precipitated by adding 62.5 μl of 6.1 N tricholoacetic acid by vortexing and subsequent incubation at 4°C for 30 min. After centrifugation, the resulting pellet was washed with 200 μl of ice-cold acetone and then centrifuged again. The washed pellet was redissolved in 50 μl of 6 M urea. The sample was then subjected to the proteolytic digestion protocol as described below.

After thawing, in the case of the membrane extract, samples were transferred to a LoBind Eppendorf tube, vortexed, and an additional 50 μl of 200 mM ABC buffer was added. A total of 100 μl (≈100 μg protein) of that mix was used for reduction and alkylation. For the whole cell sample, proteolytic digestion occurred straight after lysate preparation. In this case, the preceding protocol yielded 30 μg of protein. For the following part, two volumes are given for every step; the lower volume corresponds to the whole cell sample and the higher volume to the membrane extracts. The reduction was performed by incubation with 8.5 or 30 μl of 10 mM dithiothreitol at 37°C for 60 min at 300 rpm. Alkylation was subsequently performed by adding 8.5 or 30 μl of 20 mM iodoacetamide and incubating at 20°C for 30 min in the dark. Samples were then further diluted with 285 or 400 μl ABC, and 5 or 20 μl of 0.1 μg μl⁻¹ trypsin was added for digestion. Digestion occurred for ~18 h at 37°C and 300 rpm. After digestion, all samples were treated equally again. After digestion, peptides were extracted by solid-phase extraction in an Oasis HLB 96-well μElution Plate (Waters) according to manufacturer specifications. Peptides were eluted in two phases: (1) with 200 μl of 2% formic acid in 80% methanol and (2) with 200 μl of 1 mM ABC in 80% methanol. The peptides were then collected in an Integrated SpeedVac^TM System (Thermo Scientific) at 50°C for 2 h. Samples were stored at −20°C until analysis.

Shotgun proteomic analysis was performed as described earlier (Kleikamp et al., 2021; Pabst et al., 2021). Speed vac dried samples were dissolved in 20 μl of 3% acetonitrile plus 0.1% trifluoroacetic acid. Protein concentrations were estimated by NanoDrop ND-1000 spectrophotometry (Thermo Scientific). An aliquot corresponding to ~0.5 μg of protein was analyzed using a one-dimensional shotgun proteomics approach using a nano-liquid chromatography system consisting of an EASY nano-LC 1200 (Acclaim PepMap RSLC RP C18 separation column, 50 × 150 mm, 2 μm, Cat. No. 164568) and a QE plus Orbitrap mass spectrometer (Thermo Fisher Scientific, Germany). The flow rate was maintained at 350 nl/min over a linear gradient from 5 to 25% solvent B over 88 min, then from 25 to 55% over 60 min, followed by back equilibration to starting conditions. Data were acquired from 5 to 240 min. Solvent A was H₂O containing 0.1% formic acid, and solvent B consisted of 80% ACN in H₂O and 0.1% FA. The Orbitrap was operated in data-dependent acquisition (DDA) mode, acquiring peptide signals from 385 to 1,250 m/z at 70 K resolution in full MS mode with a maximum ion injection time (IT) of 75 ms and an automatic gain control (AGC) target of 3E6. The top 10 precursors were selected for MS/MS analysis and subjected to fragmentation using higher-energy collisional dissociation (HCD). MS/MS scans were acquired at 17.5 K resolution with an AGC target of 2E5 and an IT of 75 ms, a 2 m/z isolation width, and normalized collision energy (NCE) of 28. Data were analyzed against the proteome database from C. thermarum [GenBank: CP082237.1, strain Caldalkalibacillus thermarum TA2.A1 complete genome, Tax ID: 559292 (de Jong et al., 2020)] using PEAKS Studio X (Bioinformatics Solutions Inc., Waterloo, Canada), allowing for 20 ppm parent ion and 0.02 m/z fragment ion mass error, three missed cleavages, carbamidomethylation and methionine oxidation, and N/Q deamidation as variable modifications. Peptide spectrum matches were filtered against 1% false discovery rates (FDR), and identifications with ≥2 unique peptides were considered significant. Results from PEAKS were exported to “proteins.csv,” listing the identified proteins.

Proteins from database searching files (Supplementary Table 1) were further processed using Python (version 3.9). Files were converted to “Pandas” dataframes and counted based on the following two criteria: “Area” > 0 and “Unique Peptides” ≥ 2. Venn diagrams were then visualized using the “Matplotlib” functions “venn2_unweighted” or “venn3_unweighted.” For the Venn diagrams concerning only membrane proteins, an additional criterion was added (TMH's ≥ 1). For the heat map, the raw absolute area was used directly, after filtering using Supplementary Table 2. The heat maps were visualized using the function ‘vizus.gene_exp_hmap' from the package ‘Bioinfokit'.

We first performed a conventional whole-cell lysate shotgun proteomics experiment to detect the soluble protein fraction. The full genome of C. thermarum TA2.A1 (de Jong et al., 2020) contains 3,085 genes (NCBI; accession: CP082237), of which 1,244 (Figure 1A) were detected in this experiment (~40% of the genome). DTU's TMHMM tool was used to distinguish which of these proteins were likely integral membrane proteins, using a single TMH as the cut-off (Krogh et al., 2001). The C. thermarum TA2.A1 genome contains 754 proteins out of the total 3,085 that contain a single predicted TMH. It should be noted that proteins associated with the membrane (peripheral membrane proteins) but lacking a TMH are not “soluble proteins” and are also included in our analysis. Peripheral membrane proteins are membrane-associated via hydrophobic patches, lipid anchors, or lateral helices (Boes et al., 2021). However, these proteins cannot be identified from genetic information or be unequivocally identified or in a whole-cell proteome analysis without supporting biochemical data. Our analysis confirmed that only 124 of the 1,244 proteins detected contain a TMH. Considering that roughly a quarter of all proteins in the complete genome will be either integral or peripheral membrane proteins (Wallin and Von Heijne, 1998; Sawada et al., 2007), this reinforces the notion that the membrane proteome is generally underrepresented in conventional whole cell lysate proteomics experiments. Nevertheless, the whole-cell proteomics approach yields a membrane proteome with a considerable proportion of membrane proteins. For instance, the majority of the aerobic ETC proteins were detected. The integral membrane protein NADH dehydrogenase type I (Ndh-1) was detectable, as was the peripheral membrane protein NADH dehydrogenase type II (Ndh-2) (Godoy-Hernandez et al., 2019; Godoy-Hernandez and McMillan, 2021; Juergens et al., 2021). Other integral membrane proteins, such as succinate dehydrogenase (Sdh) and the cytochrome b₆c₁ complex (a hybrid complex III), were also detectable, as well as the cytochrome oxidase aa₃ complex. The cytochrome oxidase aa₃ complex is expected to be the primary terminal oxidase under fully aerobic conditions. However, we also detected a single subunit pertaining to the microaerophilic cytochrome oxidase ba₃ complex. Surprisingly, the subunit detected contains 13 TMHs, while the soluble subunits were undetectable. However, caution should be taken with such a result because to confidently state that C. thermarum TA2.A1 expressed two terminal oxidases under the growth conditions we imposed, detecting the additional subunits provides greater clarity toward the notion of functionality. Finally, the ATP synthase was also detected, revealing that we can get very close to detecting a fully functional C. thermarum TA2.A1 pathway for oxidative phosphorylation from whole cell proteomics.

To better evaluate the possible expression of the cytochrome oxidase ba₃, as well as many more functional proteins, increased detection is paramount. Unfortunately, membrane proteins are lesser in number in a cell, so a whole-cell proteome analysis including soluble proteins dwarfs the lower signals of membrane proteins (Tan et al., 2008). Isolating the membrane provides an initial step toward enrichment. Second, detergent solubilisation can singularly isolate proteins and is an easily accessible method to purify proteins for this reason (McMillan et al., 2007; Pullara et al., 2013; Mathieu et al., 2019). Proteins are in a crowded space in the biological membrane, and this crowding can prevent access by the proteases necessary for proteomics. Finally, as previously mentioned, for peripheral membrane proteins lacking biochemical description, we cannot rely on genetic information and cannot be unequivocally identified in whole-cell proteome analysis (Boes et al., 2021).

To obtain better coverage of the membrane proteome, a modified membrane extraction protocol was followed. While extracting the membrane fraction from the rest of the cell constituents added a very modest increase in protein hits, adding a membrane solubilisation step significantly increased the detection of membrane proteins (Figures 1A, B). The breadth of solubilisation methods available in the literature (Schuck et al., 2003; Eshaghi et al., 2005; Williams et al., 2007; Wang et al., 2009; Zaparty et al., 2009; Burton and Martin, 2012; Altinisik Kaya et al., 2018) leaves a plethora of options available, but it is well noted that there is no common optimal method. Given our knowledge of membrane proteins, we opted for the use of detergents. Most detergents are not compatible with mass spectrometry, thus limiting options. Fortunately, two extremely effective detergents for membrane protein solubilisation were feasible [FOS-choline-12 and sodium dodecyl sulfate (SDS)]. Additionally, the effect of the temperature (50°C, 65°C, and 80°C) used for solubilisation was assessed. For solubilisation with FOS-choline-12, using 65°C or 80°C yielded optimal results, while for SDS, 50°C was clearly optimal (Figures 1C, D). Additionally, cross-comparison between the two detergents showed that each detergent appears to solubilise the membrane protein fraction selectively (Figure 1B). Finally, we tested whether changing the digestion enzyme made a difference in protein detection. An in silico digestion (Supplementary Figure 2) showed potential improvement in the optimal peptide length range using both LysC and chymotrypsin as digestion enzymes. Contrary to expectations, neither increased the detection of membrane proteins significantly, having only a modest effect (Figures 2, 3). Combining all proteomics experiments allowed for the construction of an overview of all detected proteins containing a TMH, as well as zooming in on the ETC and known transporters.

The main goal of our study was to identify as many membrane proteins as possible through the optimisation of membrane proteomics. In total, 158 membrane proteins were detected using the standard whole cell proteome method; although a full aerobic ETC was detected, not all subunits of each complex were found. Solubilisation enabled the detection of the other two remaining subunits pertaining to the microaerophilic cytochrome oxidase ba₃ (Figure 2A). The aerobic and most efficient ETC C. thermarum TA2.A1 consists of Ndh-1, Sdh, cytochrome complex b₆c₁, cytochrome c, and cytochrome oxidase aa₃. The expression of the F₁F_o ATP synthase completes an aerobic oxidative phosphorylation pathway. All these complexes were detected (Figure 3).

Besides the expected aerobic ETC, somewhat surprisingly, ETC components that have been proposed to be more tuned toward a microaerophilic lifestyle, Ndh-2 and cytochrome oxidase ba₃, were expressed under aerobic conditions. The ba₃ complex proteins were among the few extra proteins detected after digestion with chymotrypsin or LysC instead of the standard trypsin digestion. The detection of Ndh-2 is in line with earlier research on this enzyme in C. thermarum TA2.A1, where it has been studied as a potential analog for drug targets (Heikal et al., 2014; Godoy-Hernandez et al., 2019; Godoy-Hernandez and McMillan, 2021). Finding the presence of cytochrome oxidase ba₃ was more surprising, as it is only half as efficient as aa₃ (Von Ballmoos et al., 2012). The expression could be due to the culturing conditions for this trial. A simple shake flask batch was used, which could have led to some oxygen limitation at the end of the batch phase, forcing C. thermarum TA2.A1 to adopt a more austere lifestyle. Such questions could be solved by culturing under more controlled conditions, such as a chemostat. We would expect that culturing at lower dissolved oxygen would also trigger the expression of the other two terminal oxidases, namely cytochromes bb₃ and bd.

In terms of carbon import machinery, a range of different transporters was detected, pertaining to the SugABC, RbsABC, MalEFGK, DctPQM, and UgpABC complexes (Figures 2A, 3). Considering the use of sucrose as a carbon source, the detection of multiple subunits of SugABC was expected (Figure 2B). MalEFGK is annotated as a maltose importer but might transport sucrose as well; this dual specificity is described in Streptococcus mutans (Kilic et al., 2007). The detection of possibly two sucrose importers means this research has failed to highlight the genes encoding for the sucrose:sodium symporter that originally sparked the interest in C. thermarum TA2.A1 (Peddie et al., 1999, 2000). The detection of the TRAP-type C4-dicarboxylate importers DctPQM, RbsABC complex, and UgpABC complex (Figures 2A, 3) may be due to their scavenging functions rather than their favored presence under our culturing conditions. In line with other alkaliphiles, no defined medium for C. thermarum TA2.A1 exists, and it still requires the presence of tryptic peptone extract to grow, which could contain a substrate for all of these transporters. In agreement with this proposition, several amino acid importers, oligopeptide importers, and importers of related compounds, like glutathione, were also detected. Finally, the proteins translocase SecDF and the flagellum proteins were also detected.

Homeostasis of monovalent cations is crucial for any alkaliphile, and generally, this is coupled with proton transport in an antiport protein (Krulwich and Cheng, 1994). C. thermarum TA2.A1 has two types of sodium:proton antiporters; a few subunits of the multisubunit Mrp were detected under the conditions used. The active ATP powered NatABC was also expressed. C. thermarum TA2.A1 relies on the TrkAH system for potassium uptake. The low-affinity TrkAH system found in most bacteria has been highlighted as a strategy to combat saline stress (Kraegeloh et al., 2005; Guo et al., 2009). C. thermarum TA2.A1 is capable of growing up to 6% NaCl (Xue et al., 2006), perhaps in part due to its reliance on this potassium uptake system. Two proteins pertaining to the A subunit of the potassium:proton antiporter TrkAH were detected (Figures 2B, 4); the trkA genes encoding for the expression of these proteins are not part of a single operon or are they close to any of the other trkA or trkH genes. As an importer of inorganic phosphate, the subunit S of the Pts complex was detected only in the experiments using alternative digestion enzymes (Figures 2B, 4). For the import of trace metals, transporters for zinc, molybdate, magnesium, and manganese were detected (Figure 4).

Other metal transporters, such as cadmium, copper and iron, were also detected. Notably, these were all described as exporters (Yoon and Silver, 1991; Rensing et al., 2000). The ferrous exporter FetAB was previously described as an exporter and implied as an important protein in avoiding problems with radical formation (Nicolaou et al., 2013). Several drug exporters were also detected, further shedding light on the possible mechanisms C. thermarum TA2.A1 might use to remain dominant in a given environment. Finally, several importers related to osmolytes were detected. The incorporation of transporters of glycine betaine and ectoine could be another method used by C. thermarum TA2.A1 to combat the challenges posed by the high pH at which it lives. Especially glycine betaine was abundantly available in the rich medium used; the strategy of utilizing compatible solutes should be perfectly viable.

This study is the first report on the proteome of C. thermarum TA2.A1, in which 1,398 proteins (45.3% of all potential proteins) were detected in at least one of the experiments. Of this, 158 contain at least a single TMH (11.3% of all identified proteins) and are therefore considered membrane proteins. The total coverage of the membrane proteome is 20.9%. C. thermarum TA2.A1 is an alkaliphile, and therefore this study focuses primarily on proteins identified in the ETC and on transporters. For an alkaliphile, homeostasis of protons, sodium, and potassium is crucial, and this study detected transporters for all of these compounds. Importers of various compounds related to central carbon metabolism and anabolism were also detected, as well as a complete ETC. Combined, this study gives a comprehensive overview of the membrane proteins required by an alkaliphile to survive and provides a valuable benchmark for further research on this organism. The authors specifically note its adaptability to a broad pH range as an area for further research. The conditions under which C. thermarum TA2.A1 employs its alternative terminal oxidases are also yet to be revealed. In that regard, performing solubilisation of membrane extracts as an addition to whole cell lysate proteomics will increase the chances of detecting these alternative terminal oxidases.