Multiplex quantitative SILAC for analysis of archaeal proteomes: a case study of oxidative stress responses

Summary

Stable isotope labeling of amino acids in cell culture (SILAC) is a quantitative proteomic method that can illuminate new pathways used by cells to adapt to different lifestyles and niches. Archaea, while thriving in extreme environments and accounting for ~20–40% of the Earth’s biomass, have not been analyzed with the full potential of SILAC. Here we report SILAC for quantitative comparison of archaeal proteomes, using Haloferax volcanii as a model. A double auxotroph was generated that allowed for complete incorporation of ¹³C/¹⁵N-lysine and ¹³C-arginine such that each peptide digested with trypsin was labeled. This strain was found amenable to multiplex SILAC by case study of responses to oxidative stress by hypochlorite. 2,565 proteins were identified by LC-MS/MS analysis (q-value ≤ 0.01) that accounted for 64% of the theoretical proteome. Of these, 176 proteins were altered at least 1.5-fold (p-value < 0.05) in abundance during hypochlorite stress. Many of the differential proteins were of unknown function. Transcription factor homologs dominated those of known function including those related to oxidative stress by 3D-homology modeling and orthologous group comparisons. Thus, SILAC is found to be an ideal method for quantitative proteomics of archaea that holds promise to unravel gene function.

Introduction

Quantitative proteomics is a valuable tool for investigating protein abundances in cells exposed to different conditions. Two main types of quantitative proteomic methods are used in research: label and label-free (Megger et al., 2014). Labeling methods permit for the direct quantification of proteins and reduce bias between samples by allowing for multiplexing. Label-free methods avoid the expense of labeling but do not allow for the mixing of the different samples into a single pool for uniform processing (Ong et al., 2002; Megger et al., 2014). Proteomic studies that use stable isotope labeling in cell culture (SILAC) yield robust quantitative comparisons and allow for multiplexing (Ong et al., 2002). SILAC relies upon the incorporation of “light” (unlabeled) compared to “heavy” amino acids (labeled with isotopes such as ¹³C, ¹⁵N, ²H, and ¹⁸O) into the proteome during cell culture (Ong et al., 2002; Ong and Mann, 2006). SILAC is best performed when the heavy amino acids are essential for growth as seen for study of bacteria and eukaryotes (Ong and Mann, 2006). If essential, the externally added heavy amino acids are fully incorporated into the proteome after several cell doublings and can be used for comparison to cells grown with light amino acids.

Archaea are ideal for study of the mechanisms used by cells to withstand harsh environmental conditions, as many archaea thrive in extreme niches, such as hydrothermal vents, acidic hot springs, and hypersaline lakes (Pedone et al., 2004; Bidle et al., 2008; Kort et al., 2013). Archaeal proteomes have been analyzed quantitatively by use of isobaric tags for relative and absolute quantitation (iTRAQ), isotope-coded protein label (ICPL), and label-free methods [e.g., (Xia et al., 2006; Kirkland et al., 2008; Van et al., 2008; Humbard et al., 2009; Tebbe et al., 2009; Williams et al., 2011; Kort et al., 2013; Cerletti et al., 2015)]. However, archaeal proteomes have yet to be analyzed by the preferred method of multiplex SILAC using strains that require arginine and lysine for growth. The advantage of this latter approach is that the proteomes can be completely labeled with heavy (vs. light) arginine and lysine. These fully labeled proteomes are amenable to multiplexing and subsequent digestion with trypsin, a serine protease that cleaves carboxyl to lysine and arginine residues. This approach allows for the theoretical labelling of each tryptic peptide generated from the proteome and, thus, enhances the sensitivity of identification and quantification of proteins by a multiplexed SILAC approach.

Here we report the generation of a SILAC-compatible strain of Haloferax volcanii and use SILAC to investigate differential protein abundance during oxidative stress in this archaeon, originally isolated from the Dead Sea. To our knowledge, such a SILAC-based study that demonstrates complete labeling and quantitative comparison of archaeal proteomes using heavy (vs. light) arginine and lysine has not been previously reported. Our focus was on oxidative stress, often encountered by halophilic archaea in hypersaline environments that undergo cycles of desiccation and intense ultraviolet (UV) radiation resulting in the generation of reactive oxygen species (ROS) (Jones and Baxter, 2017). Our findings advance multiplex SILAC analysis of archaeal proteomes while providing an insight into the responses of archaea to oxidative stress at the proteome level.

Results and Discussion

Generation of an Hfx. volcanii SILAC compatible double auxotroph

Hfx. volcanii H26, a pHV2⁻ ΔpyrE2 derivative of DS2 (Allers et al., 2010), can biosynthesize all 20 standard amino acids when cultured in minimal medium and, thus, is not compatible for study by SILAC. To overcome this limitation, the pathways of lysine and arginine biosynthesis were targeted for deletion by homologous recombination. The rationale for generating this mutant strain was that the proteins analyzed by LC-MS/MS would first be enzymatically digested into peptides using trypsin, a serine protease which cuts carboxyl to lysine and arginine residues. Thus, growth of the Hfx. volcanii double auxotroph for lysine and arginine in medium supplemented with heavy lysine and arginine would theoretically label each tryptic peptide and allow for robust identification and quantification of proteins by a multiplexed SILAC approach.

To predict the best gene candidates for generating an Hfx. volcanii double auxotroph for lysine and arginine, we relied upon KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway predictions. The lysA (diaminopimelate decarboxylase EC 4.1.1.20) (HVO_1098) gene homolog was targeted for deletion based on its putative function in synthesis of L-lysine and CO₂ from meso-2,6-diaminoheptanedioate. Likewise, the gene homolog of argH (argininosuccinate lyase EC 4.3.2.1) (HVO_0048) was selected for deletion based on its predicted catalysis of the last step of arginine biosynthesis: the production of L-arginine and fumarate from L-argininosuccinate.

Using a markerless deletion strategy, the lysA and argH gene homologs were deleted from the Hfx. volcanii H26 genome. The mutations were found to confer amino acid auxotrophy. The H26 ΔlysA mutant (LM06) was unable to grow on glycerol minimal medium (GMM) (solid or liquid) unless supplemented with L-lysine (Figure 1A). The lysine auxotrophy of LM06 was relieved when the lysA homolog was ectopically expressed, compared to the empty vector control (Figure 1A). Similarly, the H26 ΔargH mutant (LM07) was unable to grow on solid and liquid GMM unless supplemented with L-arginine and was restored in growth to parental (‘wild-type’) levels when the argH gene homolog was ectopically expressed (Figure 1B). The double deletion strain LM08 (H26 ΔargH ΔlysA) was found auxotrophic for both lysine and arginine, with a concentration of 0.3 mM of each amino acid found sufficient to restore growth of this mutant to wild-type (Figure 2A).

Figure 1.

Hfx. volcanii ΔlysA and ΔargH mutations confer lysine and arginine auxotrophy, respectively. A. Top row: glycerol minimal medium (GMM) supplemented with 1 mM lysine. Bottom row: GMM alone. Left column, LM06 (H26 ΔlysA) was plated with H26 (parent). Center column, LM06 and H26 carry plasmid pJAM2918 for ectopic expression of lysA. Right column, LM06 and H26 contain the empty vector, pJAM202c. B. Top row: GMM supplemented with 1 mM arginine, bottom row: GMM alone. Left column, LM07 (H26 ΔargH) was plated with H26 (parent). Center column, LM07 and H26 carry plasmid pJAM2919 for ectopic expression of argH. Right column, LM07 and H26 contain the empty vector, pJAM202c. See methods for details.

Figure 2.

SILAC ready Hfx. volcanii strain and conditions for testing oxidative stress. A. Growth of the lysine and arginine double auxotroph Hfx. volcanii LM08 (H26 ΔlysA ΔargH) on glycerol minimal medium (GMM) supplemented with lysine and arginine at 0, 200, 300, 500 μM of each. Parent strain (H26) was grown in GMM without lysine or arginine supplementation. B. Cell survival after NaOCl treatment. H26 cells were grown in GMM to log phase (OD₆₀₀ 0.6–0.9) and treated with 0, 1.25, 2.5, 4, and 7.5 mM NaOCl for 20 min. Treated cells were plated at 10⁻⁷ dilution on ATCC974 solid medium and individual colony forming units (CFUs) were counted. C. H26 cells were grown to log phase (OD₆₀₀ 0.6–0.9) in GMM, treated with NaOCl at 0, 2, 5, 8, 11 mM and monitored for growth for a 15 h period after treatment. D. Ellman’s reagent assay for free sulfhydryl groups. LM08 cells were grown in GMM supplemented with 300 μM lysine and 300 μM arginine and treated with 0 mM (control) and 2.5 mM NaOCl for 20 min. Cells were lysed by sonication and total protein was assayed for sulfhydryl content by Ellman’s reagent. See methods for details.

Sodium hypochlorite causes oxidative stress in Hfx. volcanii

Hfx. volcanii was next investigated under conditions of oxidative stress. Reactive oxygen species (ROS) cause damage to biomolecules when their intracellular levels exceed the antioxidant and reducing capabilities of the cell (Cabiscol et al., 2000; Shang and Taylor, 2011). Sodium hypochlorite (NaOCl) in solution forms hypochlorous acid (HOCl) which is a potent oxidant that disrupts protein structure and causes irreversible protein aggregation, leading to a loss of function (Salo et al., 1990; Shang and Taylor, 2011). Damage by HOCl is not limited to proteins, lipids undergo peroxidation (Panasenko, 1997) and cell walls become more permeable (Sips and Hamers, 1981). Nucleotides also interact with HOCl causing disruption of purine and pyrimidine rings (Dennis et al., 1979) and damage to DNA (Takehara et al., 1994). Biological systems in environments with high concentrations of chloride are prone to the oxidation of Cl⁻ to hypochlorous acid (HOCl) or a similarly reactive chlorine electrophile (Ortiz-Bermudez et al., 2003; Wang, 2016). Thus, Hfx. volcanii was examined for its sensitivity to ROS damage by hypochlorite as this microbe thrives in hypersaline environments that have high concentrations of chloride.

The survival rate and protein oxidation state of Hfx. volcanii was found to be impaired when log-phase cells were exposed to NaOCl. Hfx. volcanii had a survival rate of 63% when treated with 2.5 mM NaOCl for 20 min and plated on GMM agar and did not recover when the NaOCl concentration was increased to 7.5 mM (Figure 2B). In GMM liquid cultures, Hfx. volcanii cells were sensitive, but recovered, from treatment with 2 mM NaOCl (Figure 2C). By contrast, the cells did not fully recover when treated with 5 mM NaOCl and appeared non-viable at 8–11 mM NaOCl (Figure 2C), consistent with the observations by plate assay (Figure 2B). When proteomes are oxidized, cysteine residues form promiscuous disulfide bonds (Pajares et al., 2015). Here a 20-min treatment of cells with 2.5 mM NaOCl was found to decrease (by 63 %) the level of free sulfhydryl groups detected in cell lysate by Ellman’s reagent, thus, revealing the formation of disulfide bonds in the proteome (Figure 2D). Based on these results, hypochlorite stress was found to reduce cell viability and stimulate the formation of disulfide bonds in Hfx. volcanii.

Two subcultures sufficient for full isotopic incorporation

To achieve full isotopic incorporation of the SILAC amino acids, Hfx. volcanii LM08 cells were subcultured twice and allowed to grow for 6.5 doublings in the labeled medium prior to analysis of the proteome by LC-MS/MS. Out of the 3,094 peptide groups identified, all had full incorporation of lysine (+8) and arginine (+6). In eukaryotes, [¹³C₆] arginine becomes [¹³C₅] proline through the arginase pathway (Ong et al., 2003; Van Hoof et al., 2007; Bendall et al., 2008). Hfx. volcanii has a predicted arginase homolog (EC:3.5.3.1) HVO_1575 (rocF), so proline conversion was monitored. Conversion of heavy proline (+5) from arginine occurred 34.3 ± 5.3% of the time from peptides in the top three most abundant proteins (Table 1). While this conversion could cause problems in studies where multiple isotopic forms of arginine are used (Van Hoof et al., 2007; Bendall et al., 2008; Bicho et al., 2010), conversion was not an issue in this study which relied upon the use of one form of heavy arginine coupled with light arginine.

Table 1.

Heavy proline conversion from heavy arginine^a.

Replicate	Heavy Prolines	Total Prolines	Proportion (heavy/ total)	Average (heavy / total)	Standard Deviation (heavy / total)
1	2640	6580	0.401	0.343	0.053
2	1169	3128	0.374
3	1211	4131	0.293
4	1005	3316	0.303

^aConversion calculated based on the top three most abundant proteins (UniProt: D4GZY6, D4GXX8, and D4GWB2), as determined by emPAI value. Data used to support these calculations are listed in Supplemental Table S1.

SILAC as an effective method to quantify the differential abundance of proteins in archaea

Using the SILAC method, Hfx. volcanii cells were treated with and without 2.5 mM NaOCl for 20 min (n=4). 2,565 proteins were identified across all four replicates representing 64% of the theoretical proteome (3,996 proteins, Proteome ID: UP000008243) at a FDR adjusted p-value (or q-value) of 0.01 or less implying that 1% of significant tests will result in false positives (Figure 3A, Supplementary Table S2). 1,806 proteins were identified (q-value ≤ 0.01) in all four replicates. To determine the efficiency of the tryptic digest, the peptides from the top 3 most abundant proteins by emPAI value were used to count missed cleavages out of total theoretical cleavages. The trypsin digest had a 48.6% mean efficiency and a standard deviation of 8.9%.

Figure 3.

Hfx. volcanii proteins identified by SILAC and found to be of differential abundance during NaOCl-mediated oxidative stress. A. Venn diagram of proteins identified in each replicate with at least two peptides with a FDR adjusted p-value (or q-value) of 0.01 or less. B. Volcano plot of protein abundance values displayed as log₂ of the fold change ratio (NaOCl treatment compared to control) and −log₁₀ of p-value. Black points are statistically significant with p-value < 0.05. See methods for details.

The 50 most and least abundant proteins in all four replicates, irrespective of condition were determined by emPAI value and further analyzed for putative function by TMpred and arCOG (Supplementary Table S3–S4). Of the most abundant proteins, about one-third (17 proteins) were found to have predicted TM domain(s), 28% (14 proteins) were associated with translation and ribosome structure, 18% (9 proteins) were predicted to function in amino acid transport and metabolism, and 16% (8 proteins) were involved in energy production and conversion. Only 6 (12%) of the 50 most abundant proteins were unannotated or of unknown function by arCOG. By contrast, when classified by arCOG, 64% (32 proteins) of the 50 least abundant proteins were of unknown function with 5 proteins associated with carbohydrate transport and metabolism, 3 proteins of signal transduction and 3 proteins of amino acid transport and metabolism. Nearly half (48%) of the least abundant proteins had predicted TM domains. When considering the 50 most and 50 least abundant proteins identified, 41% had predicted TM domains suggesting the coverage of TM domain proteins was well represented, as 24% of the theoretical proteome is predicted to be TM domain proteins (Kirkland et al., 2008). Note that proteins associated with translation are typically abundant (Beck et al., 2011), while proteins with TM domains are often underrepresented in proteomic studies (Schrimpf et al., 2009). Here we found proteins associated with translation, protein folding, transport, and central metabolism to be highly abundant, while proteins of unknown function to be least abundant in the Hfx. volcanii proteome.

When comparing protein abundance among haloarchaea, some interesting findings were noted. Of the 30 proteins determined to be of high abundance in Halobacterium salinarum (sp. NRC-1) by 2D-PAGE (Shukla, 2006), the translation elongation factor 1α (HVO_0359, Eef1A), type I glyceraldehyde-3-phosphate dehydrogenase (HVO_0481, GapB), peptidyl-prolyl cis-trans isomerase (HVO_2222, PpiA), a riboflavin/purine nucleoside binding protein (HVO_1401, YufN) (3D structural homolog of PDB:4IIL and 2FQX), and proteins of the ribosome were also found to be of high abundance in Hfx. volcanii. However, the hypothetical N-acyltransferase (NTRANS_0004) of chromatin remodeling proposed to be one of the most abundant proteins in Hfx. volcanii based on transcriptome data (Ammar et al., 2012) was not identified at the proteome level in this or previous study (Kirkland et al., 2008). NTRANS_0004 transcript maps to the same region of the Hfx. volcanii genome (NC_013967.1: 936059–936368) as the signal recognition particle sRNA ffs (HVO_RS19635; NC_013967.1:936051–936362). This overlap in transcript sequence combined with the inability to detect NTRANS_0004 at the proteome level suggests the transcript is the non-coding ffs.

A label-swap replication of SILAC experiments was used to test the effect of isotope labeling on differential protein abundance. In the control group, replicates 1 and 3 were labeled with heavy lysine (+8) and heavy arginine (+6), while replicates 2 and 4 were grown with light lysine and arginine. The replicates of the treatment group were inversely correlated. When comparing the control peptide abundance values for the 50 most abundant proteins by emPAI value, only one protein, a putative phosphoserine phosphatase (HVO_0880), was detected at a statistically significant fold change in abundance (0.93-fold at a p-value < 0.01) that was due to the light/heavy amino acids (Supplementary Table S5). The abundance of the unlabeled peptides of HVO_0880 were at 92.9% ± SD 5.9% of the same peptides with the lysine (+8) and arginine (+6) labeling. This limited difference observed due to the label swap revealed SILAC had little if any effect on protein abundance in Hfx. volcanii. Label swap experiments show SILAC labeling does not disturb peptide abundances between different label states in other organisms (Ong et al., 2002; Butter et al., 2013) but are not often performed in SILAC studies.

Proteomic changes caused by oxidative/hypochlorite stress

Significant changes, in the proteome, were observed when Hfx. volcanii cells were treated with sub-lethal doses of NaOCl (2.5 mM for 20 min) designed to detect the first protein responders of hypochlorite stress. 565 proteins were found to be of differential abundance due to the NaOCl treatment at a p-value of 0.05 or less (Supplementary Table S6, Figure 3B). Of the 565 proteins, 176 were found to be altered in abundance at least 1.5-fold: with 119 proteins up (Supplementary Table S7) and 57 proteins down (Supplementary Table S8). Nearly half (85 proteins) of the 176 proteins had no predicted function by arCOG analysis (Figure 4); however, some general trends in the differential proteome could be discerned by comparative genomics and protein homology modeling as outlined below.

Figure 4.

General classification of the differential proteome of Hfx. volcanii identified during oxidative stress. Proteins at least 1.5-fold up (blue) or down (red) in abundance after 20 min treatment with 2.5 mM NaOCl (at a p-value of 0.05 or less) were classified by arCOG. See methods for details.

ROS detoxification and sulfur mobilization

ROS detoxification and sulfur mobilization systems promote redox balance and facilitate repair of sulfur-containing biomolecules, such as Fe-S clusters. Here cysteine synthase (HVO_1439), cysteine desulfurase (HVO_A0635), and glutaredoxin/thioredoxin (Grx/Trx) (HVO_1081 and HVO_0337) homologs were found increased in abundance during hypochlorite stress. The sulfhydryl groups of cysteine and Grx/Trx are likely used to facilitate the reduction of disulfide bonds (Fig. 2D) and to mobilize sulfur (e.g., via persulfide groups) for repair of sulfur-containing biomolecules. Surprisingly, peroxiredoxin (Prx), catalase-peroxidase (KatG), methionine sulfoxide reductase (MsrA/B), superoxide dismutase (SOD1/2) and other ROS detoxification systems, while detected, were not of differential abundance suggesting these enzymes were at sufficient levels during the early stages of hypochlorite stress. In Hbs. salinarum, prx, sod1, msrA/B and katG transcripts are detected at increased abundance during severe oxidative stress (Kaur et al., 2010), while cysteine biosynthesis and grx/trx transcripts are found at elevated levels during both mild and severe ROS challenge (Whitehead et al., 2006; Boubriak et al., 2008; Kaur et al., 2010). Thus, an increase in levels of the sulfhydryl containing cysteine and Grx/Trx proteins appears to be a common defense mechanism for ROS challenge in haloarchaea, while ROS detoxification systems may be reserved for severe oxidative stress.

NAD⁺/kynurenine metabolism

De novo biosynthesis of NAD⁺ through kynurenine is redox sensitive and produces metabolic intermediates with the redox capacity to be protective, alter signal transduction, and other functions (Crowley et al., 2000; Massudi et al., 2012; Gonzalez Esquivel et al., 2017). Here we found homologs of NAD⁺ biosynthesis and kynurenine metabolism to be impacted at the protein level during hypochlorite stress including an increased abundance of nicotinamide-nucleotide adenylyltransferase (NadM, HVO_0782) and kynurenine formamidase (HVO_A0415) as well as a decreased abundance of nicotinate-nucleotide pyrophosphorylase [carboxylating] (NadC, HVO_2579) homologs. These results suggest ROS challenge may alter kynurenine/NAD⁺ metabolism to generate metabolites that are redox protective and/or promote signal transduction in Hfx. volcanii.

Ribosome biogenesis and translation

Proteins associated with ribosome biogenesis and translation were generally up during hypochlorite stress. Translation initiation factor IF-6 (HVO_0117), ribonuclease P (HVO_2556), an rRNA methylase (HVO_1669), a RimK-type ribosomal modification protein (HVO_0483), and 50S ribosomal proteins L44e (HVO_0701) and L13 (HVO_2778) were found highly up, while 10 other proteins of translation were found moderately up in abundance during the stress. Similarly, genes associated with translation are increased at the transcript level in Hbs. salinarum upon exposure to UV-C (McCready et al., 2005), UV-B (Boubriak et al., 2008), transition metals (Kaur et al., 2006), H₂O₂ and low doses of the superoxide (O₂⁻) generating paraquat (Kaur et al., 2006; Kaur et al., 2010). Whether these ROS-stimulated changes cause an increase in translation that is used for protein repair or alter translation to stall/protect the ribosomal machinery during oxidative stress remains to be determined. In eukaryotes, ROS-mediated stress typically results in a general decrease in translation, increase in tRNA halves that can inhibit translation, and preferential translation of mRNAs involved in the stress response (Holcik and Sonenberg, 2005; Thompson et al., 2008; Bakowska-Zywicka et al., 2016; Huang and Hopper, 2016).

Transcription and signal transduction

Transcription factor (TF) homologs dominated the proteins assigned general function that were altered at least 1.5-fold in abundance during hypochlorite stress. Of the 176 proteins altered at least 1.5-fold, over 15% (27 proteins) were predicted TFs with most found increased in abundance (21 TFs up vs. 6 TFs down) during the NaOCl treatment. Many of these TFs had conserved Cys/His residues that could theoretically sense redox status including homologs related by Phyre2-modeling to bacterial TFs that mediate survival against oxidative stress [e.g., HypR (HVO_0855), DosR (HVO_A0563), and IscR (HVO_A0588)] (Crack et al., 2012; Chim et al., 2014; Hillion and Antelmann, 2015). Proteins associated with signal transduction were also found at elevated levels during hypochlorite stress including a soluble methyl-accepting chemotaxis (MCP) sensory transducer homolog (HVO_2220) and a CheY-like response regulator (HVO_2012) which may function in a coordinated manner. Likewise, up in abundance was a DUF336 family member (HVO_A0272) related to bacterial HbpS, a heam-degrading redox sensor that regulates catalase and peroxidase expression through interaction with a two-component signal transduction system (Ortiz de Orue Lucana et al., 2009). HVO_0215 was also found highly up in abundance during stress, and while the function of this TM protein remains to be determined, its predicted location in the cell membrane, conserved Fe-S cluster, haem-binding motif, and lack of Fe-only hydrogenase active site residues suggests it may sense ROS challenge.

Metal homeostasis

High intracellular Mn/Fe ratios (Daly et al., 2004; Fredrickson et al., 2008; Kish et al., 2009) and metabolite complexes of Mn²⁺ (e.g., Mn-orthophosphate) (Daly et al., 2010; Culotta and Daly, 2013) are correlated with combating ROS stress and protecting proteins from oxidative damage. Proteomic responses of Hfx. volcanii that may promote a high Mn/Fe ratio were identified during hypochlorite stress. In particular, homologs (HVO_2397 and HVO_2398) of Mn²⁺/Zn²⁺ transporters (Desrosiers et al., 2007; Wichgers Schreur et al., 2011) and the ferritin-like DpsA (HVO_0536) were found up in abundance during hypochlorite stress. In Hbs. salinarum, DpsA binds Fe²⁺ (Zeth et al., 2004) and is upregulated at the transcript level by ROS challenge (Reindel et al., 2005; Kaur et al., 2010) suggesting Hfx. volcanii DpsA could sequester the free pools of Fe²⁺ that exacerbate ROS challenge. Homologs of siderophore biosynthesis (HVO_B0041 to _B0044) and heavy metal (Cu²⁺) transport (HVO_1751 of arCOG01576) were found significantly down in abundance, suggesting the uptake of Fe²⁺ and Cu²⁺ are reduced during hypochlorite stress. Down in abundance was also the DtxR-type HTH domain (IPR022687) protein HVO_0538 of arCOG02101 that includes Hbs. salinarum Idr2 (VNG0835G), a direct activator of putative siderophore biosynthesis gene expression and member of a transcriptional network that maintains Fe homeostasis (Schmid et al., 2011; Martinez-Pastor, 2017). Overall, the observed changes in the Hfx. volcanii proteome suggest hypochlorite stress signaled an increase in intracellular Mn/Fe ratios to protect the cells from ROS damage by altering the abundance of proteins associated with heavy metal uptake and Fe²⁺-binding.

Small proteins of unknown function

Hypochlorite stress impacted the abundance of numerous small proteins. The levels of six small CPxCG-related ‘Zn finger’ proteins were altered by the stress including increased abundance of HVO_1359 (68 aa), HVO_2057A (59 aa), HVO_0720 (100 aa), HVO_2982 (67 aa), and HVO_0758 (56 aa) and decreased abundance of HVO_1352 (62 aa). These Zn finger proteins are predicted to coordinate divalent metal ion(s) through conserved His/Cys residues and form DNA, RNA, and/or protein binding structures that regulate cell function. Of the Zn finger proteins, HVO_1352 and HVO_1359 were inversely correlated in abundance and encoded in genome synteny with HVO_1355, a small UPF0058 family protein (100 aa) that was up 6-fold suggesting this region of the genome is highly related by ROS challenge. HVO_1355 is predicted to bind divalent metal based on comparison to its homolog Hbs. salinarum Vng1086c (PDB: 2GF4). Other small proteins of unknown function altered by hypochlorite stress included HVO_1405 (147 aa), HVO_1563 (125 aa), HVO_2176 (66 aa), HVO_0805 (82 aa), and HVO_1516 (108 aa) that were increased in abundance and HVO_1512 (95 aa), HVO_1753 (65 aa), HVO_A0417 (102 aa), HVO_2897 (107 aa), and HVO_1188 (105 aa) that were decreased in abundance. By comparison, small CPxCG-related Zn finger proteins are increased at the transcript level by UV-mediated ROS challenge in Hbs. salinarum (McCready et al., 2005; Boubriak et al., 2008).

Lipid metabolism

Hypochlorite stress was found to increase the abundance of proteins associated with lipid metabolism by over 1.5-fold including homologs of sn-glycerol-1-phosphate dehydrogenase (HVO_0822, GldA), bifunctional short chain isoprenyl diphosphate synthase (HVO_2725, IsdA), farnesyl-diphosphate farnesyltransferase (HVO_1139, FdtF), glycerophosphodiester phosphodiesterase (HVO_B0291, GlpQ2), phosphatidylserine decarboxylase (HVO_0146, Psd), and a lipase/esterase homolog (HVO_0137). Of these, GldA forms the sn-glycerol-1-phosphate backbone of archaeal lipids (Caforio and Driessen, 2017) and is found increased at the transcript level by oxidative stress in Hbs. salinarum (Whitehead et al., 2006; Kaur et al., 2010). The increased abundance of proteins/transcripts associated with lipid metabolism may serve to replenish and/or modify the cell membrane and protect the haloarchaea from ROS damage.

DNA repair/recombination

DNA repair/recombination proteins were found increased in abundance during hypochlorite stress. Included in this group were the ATP-dependent helicase Hel308 (HVO_0971), Ski2-helical domain protein (HVO_0289), single-stranded-DNA-specific exonuclease RecJ1 (HVO_0073) and UvrC related protein (HVO_1745). Modest increases in abundance of the DNA repair/recombination associated DHH/RecJ family phosphoesterase RecJ1 (HVO_1018) and ssDNA binding replication protein A (RPA) (i.e., Rpa3 HVO_0292) of arCOG01510 were also observed. RPAs are widely associated with oxidative stress. In Hbs. salinarum, the RPAs are found overexpressed in extreme radiation-resistant mutants (DeVeaux et al., 2007) and increased in transcript level after ROS challenge (McCready et al., 2005; Whitehead et al., 2006; Boubriak et al., 2008; Kaur et al., 2010). Archaeal RPAs are thought to act as a platform for loading the Hel308 helicase onto aberrant ssDNA at blocked replication forks (Woodman et al., 2011). Thus, the enhanced levels of Hel308 and RPA observed here during hypochlorite stress may serve to coordinate DNA repair. Note the Hfx. volcanii UvrC- and Ski2-like proteins found increased in abundance are missing their respective nuclease domains. While the UvrC-like HVO_1745 has the two helix-hairpin-helix (HhH) motif used for non-specific DNA binding (Aravind et al., 1999), it is missing the RNase H domain and associated catalytic triad of UvrC (Karakas et al., 2007). Similarly, HVO_0289 is related to the C-terminal helical domain of Ski2 but lacks the central RecA1/2 domains needed for the helicase activity of Ski2 in eukaryotic exosome function (Halbach et al., 2012). The Hfx. volcanii Hel308 does have a RecA-like domain topology typical of Ski2 helicases (Woodman et al., 2011). Thus, the Hel308, RPA and Ski2-like proteins may be coordinately increased in abundance during hypochlorite stress to repair DNA.

Archaeal RecJ proteins are generally associated with DNA repair, recombination, and replication (Makarova et al., 2012; Oyama et al., 2016). Both Hfx. volcanii RecJ proteins found increased in abundance by hypochlorite stress had conserved nuclease active site residues, but were of distinct arCOG groups. HVO_0073 of arCOG00427 is a GINS-associated nuclease (GAN) homolog of the CMG (Cdc45/RecJ, MCM, GINS) complex of DNA replication and repair (Li et al., 2011; Oyama et al., 2016; Yi et al., 2017). By contrast, HVO_1018 is of arCOG00429 which includes: HAN (TK0155), a nuclease that binds Hef (helicase-associated endonuclease for fork-structured DNA)(Ishino et al., 2014) and VNG0779C, , which is increased at the transcript level by UV radiation (McCready et al., 2005; Boubriak et al., 2008). While Hfx. volcanii, Hef (HVO_3010) is involved in stalled replication fork repair (Lestini et al., 2010; Lestini et al., 2013) and was detected in this study, the abundance of Hef was found unaltered by the hypochlorite stress.

UspA domain proteins

Of the 40 Hfx. volcanii proteins with UspA universal stress domains (IPR006016), four were found increased in abundance during oxidative stress: HVO_0428 (149 aa), HVO_0612 (143 aa), HVO_2156 (301 aa), and HVO_2500 (Cat2, 748 aa). Cat2 had an added N-terminal cationic amino acid/polyamine transporter 1 (IPR002293) domain which along with UspAs are associated with oxidative stress. UspAs are important for oxidative stress resistance in bacteria (Nachin et al., 2005; Liu et al., 2007) and for modulating ROS generation in plants (Gutierrez-Beltran et al., 2017). Likewise, overexpression of cationic amino acid transporters can alleviate oxidative stress in mammals (Konstantinidis et al., 2014; Rajapakse et al., 2014). Increased transcript/protein levels of UspAs are not limited in archaea to hypochlorite stress. In the halophilic methanogen Methanohalophilus portucalensis, the MJ0557-like UspA is increased at the transcript level by hypo-salt stress (Shih and Lai, 2010), UspAs are increased in protein abundance by low temperature growth in Antarctic haloarchaea (Williams et al., 2017), and the UspA-like KdpQ is implicated in K⁺ homeostasis based on its genomic neighborhood in Hbs. salinarum (Strahl and Greie, 2008). Interestingly, little if any increase in abundance of UspA is noted at the transcript or protein level for Hbs. salinarum during recovery from ROS challenges that include gamma irradiation (Whitehead et al., 2006), H₂O₂/paraquat (Kaur et al., 2010), heavy metal (Kaur et al., 2006), and UV radiation (McCready et al., 2005; Boubriak et al., 2008). Thus, archaeal UspAs may be regulated by factors other than transcript level or are needed during specific stress conditions (e.g., hypochlorite, low temperature, and hypo-saline).

Ion pumps/channels

By contrast to the cationic amino acid transporter Cat2, which was enhanced during hypochlorite stress, a decrease in abundance was observed for two ion channels: the small conductance mechanosensitive ion channel homolog MscS (HVO_1165) and the ion channel pore/TrkA domain protein PchA1 (HVO_1137). Ion channels are associated with dissipating membrane potential and altering the redox state of biomolecules (e.g., glutathione) (Lee et al., 2016).

Proteases and chaperones

Components of post-translational modification and protein quality control (chaperones/proteases) were found impacted by hypochlorite stress. HVO_0102, a homolog of the stress responsive Zn²⁺ metalloprotease HtpX that contributes to protein quality control in bacteria (Shimohata et al., 2002; Sakoh et al., 2005; Akiyama, 2009) was found increased in abundance along with various putative chaperones: the proteasome assembly chaperone PAC2 (HVO_0697), a metal insertion GTPase (UreG, HVO_0150), and various AAA ATPases [i.e., PAN2 (HVO_1957), Cdc48a (HVO_2380), Cdc48c (HVO_1327) and MoxR-like protein (HVO_1941)]. A von Willebrand factor (vWF) A domain protein (HVO_1256) which was a predicted structural homolog of the 26S proteasome ubiquitin receptor (Rpn10, PDB: 2X5N) was also increased in abundance. Of these, the vWF-like HVO_1256 and MoxR-like HVO_1941 may function together as a chaperone-like system based on analogy to RavA-ViaA (Wong et al., 2017). Overall, we found the levels of HtpX and molecular chaperones to be elevated, while the abundance of 20S proteasomes and Lon-type proteases to be unaffected during the early stages of hypochlorite stress.

Post-translational modification systems

The ubiquitin-like sampylation and N^ε-lysine acetylation pathways of protein modification were impacted by NaOCl treatment. In particular, the level of the ubiquitin-like SAMP1 (HVO_2619) was found to be enhanced during hypochlorite stress, in-line with our previous finding that samp1ylation is stimulated by the mild oxidant DMSO (Dantuluri et al., 2016). To support a role of SAMP1 in the hypochlorite stress response, we examined a Δsamp1 mutant and found it to be hypersensitive to hypochlorite stress when compared to its parent (H26) (Figure 5). Other known components of sampylation (SAMP2/3, MsrA, E1 UbaA and JAMM1/2)(Humbard et al., 2010; Miranda et al., 2011; Miranda et al., 2014; Fu et al., 2016; Cao et al., 2017; Fu, 2017) were detected, but not impacted in protein abundance by the stress. Instead, the abundance of putative structural homologs of eukaryotic ubiquitin-binding proteins were found to be increased by the hypochlorite stress including the SWIM Zn²⁺ finger domain protein HVO_A0396 [related to the MEX E3 ubiquitin ligase (Nishito et al., 2006)] and HVO_1514 [related to the JAMM1 desampylase (Cao et al., 2017) but unable to coordinate the catalytic Zn²⁺ ion] based on homology modeling. Regarding acetylation, the histone acetyltransferase (HAT) homologs Pat1/2 were found inversely correlated in abundance during stress: Pat1 (HVO_1756) was found down 3.1-fold, while Pat2 (HVO_1821) was up 1.8-fold in abundance. Pat1/2 are proposed to mediate the N^ε-lysine acetylation of histones (Altman-Price and Mevarech, 2009). In eukaryotes, ubiquitination and acetylation of histones regulates chromatin structure (Caron et al., 2005).

Figure 5.

Hypersensitivity of an Hfx. volcanii small archaeal ubiquitin-like modifier protein mutant (Δsamp1) to oxidative stress as predicted by SILAC and observed by treatment with NaOCl. H26 (parent) and HM1041 (Δsamp1) strains were spot plated by serial dilution (panel A) or grown in liquid culture (panel B) on glycerol minimal medium (GMM) with 0, 0.8, or 3 mM NaOCl as indicated. Error bars indicate standard deviation of n = 9 culture replicates with the experiment found reproducible (n=4). See methods for details.

Conclusions

Here we generated a SILAC-ready strain of Hfx. volcanii and studied quantitative differences in its proteome due to hypochlorite stress conditions. We generated lysine and arginine single and double auxotrophs of this archaeon which allowed for full isotopic incorporation and sensitive quantitative coverage of the proteome after trypsin digestion. In Hfx. volcanii, lysA and argH are now found essential in lysine and arginine biosynthesis, respectively. We identified 2,565 proteins, covering 64% of the theoretical proteome (with a q-value of at least 0.01). While arginine to proline conversion is detected in the Hfx. volcanii LM08 strain at a rate of 34%, quantitative problems can be avoided by use of only one form of heavy arginine and may be overcome by deletion of rocF (HVO_1575) predicted to encode arginase (EC:3.5.3.1). We find the SILAC studies of Hfx. volcanii require only small amounts of cell material for whole-cell proteome studies, thus, minimizing costs. The SILAC method captured over half of the proteome, while maintaining coverage of membrane proteins, using 2 mg total protein (250 μg per replicate per experimental state). This method allows for conservative use of reagents, due to the small amount of protein needed and ability to reach 100 % of theoretical labeling efficiency. The development of this strain now allows for multiplexed quantitative proteomic analysis with accurate relative abundance values in archaea. Our case study of hypochlorite stress provides guidance for future studies to uncover mechanisms used by haloarchaea to overcome ROS challenge.

Experimental Procedures

Materials

Biochemicals were from Sigma Aldrich (St. Louis, MO, USA). Other inorganic and organic analytical grade chemicals were from Fisher-Scientific (Atlanta, GA, USA). Desalted oligonucleotide primers were from Integrated DNA Technologies (Coralville, IA, USA). Polymerases, ligase, and restriction endonucleases were from New England Biolabs (Ipswich, MA, USA). Amino acid isotopes were from Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA). Agarose for DNA analysis was purchased from Bio-Rad laboratories (Hercules, CA, USA).

Strains and media

Strains used in this study are listed in Table 2. E. coli strains Top10 and GM2163 were used for routine cloning and preparation of plasmid DNA for transformation into Hfx. volcanii, respectively (Dyall-Smith, 2009). E. coli strains were grown at 37 °C in Luria-Bertani (LB) medium supplemented with ampicillin (0.1 mg·ml⁻¹). Hfx. volcanii strains were grown at 42 °C at 200 rpm rotary shaking in ATCC974 (complex) medium or GMM, the latter medium with ammonium chloride as the nitrogen source (Dyall-Smith, 2009). Media was supplemented with ampicillin (0.1 mg·ml⁻¹), novobiocin (0.1 μg·ml⁻¹), L-lysine, or L-arginine as needed. L-Lysine and L-arginine concentrations were 0.3 mM unless specified otherwise. Culture plates were supplemented with 1.5% (w/v) agar for LB medium and 2.0% (w/v) agar for GMM and ATCC974 medium. Growth was measured by the optical density at 600 nm (OD₆₀₀) (where 1 OD₆₀₀ unit equals approximately 10⁹ CFU·ml⁻¹).

Table 2.

Strains, plasmids, and oligonucleotide primers used in this study^a.

Strain, plasmid, or primer	Description	Source or reference
E. coli strains
TOP10	F− recA1 endA1 hsdR17(rK − mK+) supE44 thi-1gyrA relA1	Invitrogen
GM2163	F– ara-14 leuB6 fhuA31 lacY1 tsx78 glnV44 galK2 galT22 mcrA dcm-6 hisG4 rfbD1 rpsL136 dam13::Tn9 xylA5 mtl-1 thi-1 mcrB1 hsdR2	New England Biolabs
H. volcanii strains
DS70	wild-type isolate DS2 cured of plasmid pHV2	(Wendoloski et al., 2001)
H26	DS70 ΔpyrE2	(Allers et al., 2004)
LM06	H26 ΔlysA (hvo_1098)	This study
LM07	H26 ΔargH (hvo_0048)	This study
LM08	H26 ΔlysA ΔargH	This study
HM1041	H26 Δsamp1 (hvo_2619)	(Miranda et al., 2011)
Plasmids
pTA131	Apr ; pBluescript II containing Pfdx-pyrE2	(Allers et al., 2004)
pJAM2912	Apr ; pTA131 based plasmid containing pre-deletion sequence for lysA	This study
pJAM2914	Apr ; pTA131 based plasmid for ΔlysA	This study
pJAM2916	Apr ; pTA131 based plasmid containing pre-deletion sequence for argH	This study
pJAM2917	Apr ; pTA131 based plasmid for ΔargH	This study
pJAM202c	Apr Nvr; Hfx. volcanii-E. coli shuttle P2rrn expression vector, empty vector	(Zhou et al., 2008)
pJAM2918	LysA complement	This study
pJAM2919	ArgH complement	This study
Primers
LysA 500 up SLIC	5′ aggaattcgatatcaCCGCTATCACGACGTGCTCC 3′a	This study
LysA 500 dwn SLIC	5′ cggtatcgataagctTCCGGGAACACCGACGCGT 3′	This study
LysA inverse up	5′ TCACTCGCGCAGCGCCTCCTCC 3′	This study
LysA inverse dwn	5′ CCAAGCACCCACACAGAATCATGAGCGTACTCAA 5′	This study
LysA 700 up	5′ AGAAGACCGGCTCCGACGTGACCT 3′	This study
LysA 700 dwn	5′ ACTGCGTCTCGCCCTCGACG 3′	This study
LysAcompSLICfwd	5′ ctttaagaaggagatatacaATGAGCGGCGGCGGGC 3′	This study
LysAcompSLICrev	5′ tatgctagttattgctcataTCAGTTGGGGATGTGCTCGG3′	This study
ArgH 500 up SLIC	5′ aggaattcgatatcaAACTCATCTCGTTCCTCAA 3′	This study
ArgH 500 dwn SLIC	5′ cggtatcgataagctAGTCCAACGTGCAATTTA 3′	This study
ArgH inverse up	5′ CTTACTCGTCGCTACCGC 3′	This study
ArgH inverse dwn	5′ GTCCGAACGCGAGACGCC 5′	This study
ArgH 700 up	5′ TGGTCCATCGACACGAACCTCTG 3′	This study
ArgH 700 dwn	5′ AAGCGCGTGATGTAGATGCTCCTGC 3′	This study
ArgHcompSLICfwd	5′ ctttaagaaggagatataca ATGGCAGGCGAGGACGGCGACT 3′	This study
ArgHcompSLICrev	5′ tatgctagttattgctcataTCAGACATAGCTCGAAACCTCCTCGTCGAG 3′	This study

^aSequence and ligase independent cloning (SLIC) overhangs to the plasmid vector are in lowercase.

Generation of mutant strains

Plasmids and primers used in this study are listed in Table 2. Plasmids were constructed using one step sequence and ligation independent cloning (SLIC) (Jeong et al., 2012). Gene deletions were created by the strategy shown previously by Allers et al. (Allers and Ngo, 2003). Gene homologs of lysA (HVO_1098; diaminopimelate decarboxylase EC 4.1.1.20) and argH (HVO_0048; argininosuccinate lyase EC 4.3.2.1) were targeted for markerless deletion on the Hfx. volcanii H26 genome. Colonies were screened for auxotrophy by growth on GMM plates with and without supplementation with L-arginine and/or L-lysine. The mutant strains were further analyzed by polymerase chain reaction (PCR) using genomic DNA as a template and primers that annealed outside of the plasmids used for homologous recombination. Mutants were further analyzed by Southern blotting in which genomic DNA was hybridized to digoxigenin-11-dUTP (Sigma Aldrich, St. Louis, MO, USA) labeled probes (Southern, 1975). The DNA hybrids were detected on the blots using alkaline phosphatase-conjugated antibody to digoxigenin and CDP-Star (Invitrogen, Thermo Fisher Scientific Waltham, MA, USA). The intensity of the DNA hybrid bands was visualized by X-ray film (Hyperfilm; Amersham Biosciences).

Growth assays for minimal amino acid

Hfx. volcanii ΔlysA and ΔargH mutants were tested for the minimum concentration of L-lysine and L-arginine required for optimal growth, respectively. Cells were grown in complex medium (ATCC974) to log phase (OD₆₀₀ of 0.8). Cells were harvested by centrifugation (4,000 x g), washed three times in GMM, resuspended in 5 ml GMM, and subcultured to an OD₆₀₀ of 0.01 in 25 ml of GMM supplemented with and without L-lysine or L-arginine (0 to 0.5 mM). Cultures were incubated with rotary shaking (200 rpm) at 42 °C in 125 ml Erlenmeyer flasks. Samples (0.1 to 1 ml) of culture were removed from 0–61 h for monitoring growth by OD₆₀₀.

Assays of cell survival after exposure to sodium hypochlorite (NaOCl)

Hfx. volcanii H26 cells were grown to log phase (OD₆₀₀ of 0.7) in 25 ml of GMM in 125 ml Erlenmeyer flasks in a rotary shaker (200 rpm) at 42 °C. To monitor survival after exposure to NaOCl, cells were treated with 0, 1.25, 2.5, 4, and 7.5 mM NaOCl (10 μl in dH₂O; diluted from Sigma Aldrich product 425044) at room temperature (RT) with 10 sec of rotary shaking every 5 min for 20 min. Treated-cells were serially diluted in ATCC974 liquid and plated at 10⁻⁶ to 10⁻⁷ on ATCC974 plates. The colony forming units (CFU) per ml were determined to assess cell survival. To test recovery after long-term exposure of NaOCl, cells were grown to log phase as described above. The log phase cells were treated with 0, 2, 5, 8, and 11 mM NaOCl. Growth was monitored by OD₆₀₀ for 15 h after treatment, and cells were plated for survival on ATCC974 plates. Experiments were performed in biological triplicate. To compare survival of Hfx. volcanii HM1041 (Δsamp1) to its parent H26, the cells were streaked onto GMM agar from −80 °C glycerol stocks and grown for 5 days in a closed plastic zippered bag at 42 °C. Isolated colonies were inoculated into 3 ml GMM in a capped 13 x 100 mm culture tube and incubated at 42 °C with rotary shaking (200 rpm). At an OD₆₀₀ 0.9, the cultures were normalized to 1 OD unit · ml ⁻¹ and serially diluted in GMM from 10⁻¹ to 10⁻⁶. Culture dilutions (20 μl) were spotted serially on GMM agar plates supplemented with 0.8 mM NaOCl and a mock control. Plates were placed in a closed plastic zippered bag in a 42 °C incubator and allowed to grow for 5 days. As an added assay, H26 (parent) and HM1041 (Δsamp1) strains were streaked from −80°C glycerol stocks onto solid complex medium (ATCC974) and incubated as described above. Colonies were inoculated into 25 ml of ATCC974 medium in 125 ml Erlenmeyer flasks. Cells were grown to log phase (OD₆₀₀ of 0.6–0.8) at 42 °C (200 rpm). Cells were washed twice with GMM by centrifugation (8,600 × g, 1 min at room temperature) and inoculated to a starting OD₆₀₀ of 0.1 unit in GMM (5 ml) supplemented with and without 3 mM NaOCl. Cells were incubated at 42°C (200 rpm) in capped 13 × 100 mm culture tubes. Growth was monitored at OD₆₀₀.

Ellman’s reagent assay for free sulfhydryl groups of NaOCl treated cells

Hfx. volcanii LM08 was grown to log phase (OD₆₀₀ of 0.8) in 50 ml GMM supplemented with 0.3 mM L-lysine and 0.3 mM L-arginine in 125 ml Erlenmeyer flasks in a rotary shaker (200 rpm) at 42 °C. The cultures were immediately treated with 0 and 2.5 mM NaOCl in biological triplicate as described for survival assays. After the 20-min treatment, cells were pelleted at 4,000 x g and resuspended in 1 ml reaction buffer (0.1 sodium phosphate pH 8.0 with 1 mM EDTA). Cells were lysed on ice by sonication (Sonic Dismembrator Model 500, Fisher Scientific) at 30% amplitude for 2 sec on, 2 sec off for 20 sec total, three times over. Cell lysate was clarified by centrifugation at 13,000 x g and passed through a 0.2 μm surfactant-free cellulose acetate (SFCA) filter (Nalgene, ThermoFisher Scientific, USA). The filtrate was quantified for total protein by the bicinchoninic acid (BCA) assay (Pierce, USA) and assayed for thiol content by Ellman’s reagent assay. In brief, the Ellman’s reagent was generated by dissolving 4 mg of 5,5′-dithio-bis-(2-nitrobenzoic acid) (DTNB) (ThermoFisher Scientific) in 1 ml reaction buffer (0.1 M sodium phosphate, pH 8.0, containing 1 mM EDTA). Reactions (2.8 ml total volume) were mixtures of 50 μl Ellman’s reagent, 0.25 ml protein (0.97 mg total), and 2.5 ml reaction buffer in borosilicate glass tubes (13 x 100 mm Fisher Scientific). The reactions were vortexed to mix and incubated at RT for 15 min. Sample (250 μl) was transferred in triplicate to a clear polystyrene flat bottom 96-well plate (Fisher Scientific) and absorbance was measured at 412 nm using a Biotek Synergy HTX Multi-Mode Reader. L-Cysteine hydrochloride monohydrate (Fisher BioReagents, Fisher Scientific) was used as a standard to quantify the sulfhydryl groups.

Isotopic incorporation

Hfx. volcanii LM08 was streaked with a toothpick from 20% v/v glycerol stocks stored at −80 °C onto a plate of GMM supplemented with 0.5 mM lysine and 0.5 mM arginine and grown at 42 °C for 5 days. Isolated colonies were transferred to 25 ml GMM supplemented with either heavy or light amino acids separately (0.3 mM each) and allowed to grow for 24 h (about 6.5 doublings) (in a 125 ml Erlenmeyer flask at 42 °C in a rotary shaker at 200 rpm). Cultures were subcultured into 25 ml GMM to a starting OD₆₀₀ of 0.01 and similarly grown in the same heavy or light medium for 24 h, twice sequentially for a total of three subcultures after 24 h each. The heavy amino acids were L-lysine+8 (L-lysine: 2HCl, ¹³C₆, 99%; ¹⁵N₂, 99%; Item No. CNLM-291-H-PK) and L-arginine+6 (L-arginine: HCl, ¹³C₆, 99%; Item No. CLM-2265-H-PK) (Cambridge Isotope Laboratories, Inc.). The light amino acids were L-lysine and L-arginine purchased from Sigma Aldrich (Item No. L8662 and A8094, respectively). The final cultures were harvested at log phase (OD₆₀₀ 0.6) by centrifugation at 4,000 x g. Proteins were extracted with TRIzol (ThermoFisher Scientific) as described previously (Kirkland et al., 2006). A preliminary study was done to determine the incorporation rate of heavy lysine (+8) and heavy arginine (+6) into the proteome. These preliminary untreated samples were analyzed by LC-MS/MS to monitor isotope incorporation, as described below, with the following exceptions: 1) sulfhydryl groups of the proteins were alkylated by reacting with 50 mM iodoacetamide in the dark at room temperature for 45 min, 2) tryptic peptides were separated by a 60-min linear gradient with the Easy-nLC 1200 system, 3) strong cation exchange (SCX) was not used for separation of peptides, and 4) peptide masses were detected by a Q Exactive Plus Hybrid Quadrupole-Orbitrap Mass Spectrometer.

Experimental design

Four biological replicates (n=4) of the Hfx. volcanii LM08 cells were grown in 25 ml of GMM at 42 °C in a rotary shaker (200 rpm) to late log phase (OD₆₀₀ 0.7–0.8). The cells were incubated for 20-min with 2.5 mM NaOCl (treatment groups) or dH₂O (control groups) as described for the survival assays. For two of the biological replicates, the control group cultures were labeled with (heavy) L-lysine (+8) and L-arginine (+6) and the treatment group was labeled with (light) L-lysine (+0) and L-arginine (+0), as described above. The other two replicates were grown with the labels switched: (heavy) L-lysine+8 and L-arginine+6 for the treatment group and (light) L-lysine+0 and L-arginine+0 for the control. This experimental designed allowed for comparison of the label swap as well as an assessment of the impact of oxidative stress on the Hfx. volcanii proteome.

Preparation of protein samples for analysis by mass spectroscopy

Culture volumes were normalized in each centrifuge tube to equivalent total OD units per sample, and cells were harvested by centrifugation (4,000 x g). All supernatant was carefully removed. The cell pellets were stored at −80 °C. Cell pellets of control and treatment groups were mixed at a 1:1 ratio (n=4 mixed samples), and the proteins were extracted with TRIzol, as described previously (Kirkland et al., 2006). The protein pellets were stored at −20 °C and processed as described previously (Mostafa et al., 2016) with a few modifications. Pellets were dissolved in a solubilization buffer [7 M urea, 2 M thiourea and 4% (w/v) CHAPS] and quantified by EZQ assay kit (Weist et al., 2008) according to manufacturer instructions (Invitrogen Inc., Eugene, OR, USA). Protein (500 μg per 100 μl buffer) was reduced by addition of 2.5 μl of 50 mM tris-(2-carboxyethyl phosphine (TCEP) and incubating the sample at 60 °C for 60 min. After cooling the sample, the protein was treated with 5 μl of 200 mM methyl methanethiosulfonate (MMTS) for 30 min in the dark and digested with 5 μg modified trypsin (Promega, Madison, WI, USA) at 37 °C for 16 h. Peptides were lyophilized and solubilized in 3% v/v acetonitrile in 0.1% v/v formic acid and desalted on a Macrospin C-18 reverse phase mini-column (The Nestgroup Inc., Southborough, MA, USA). Eluted peptides were lyophilized and fractionated by strong cation exchange chromatography (SCX) as described previously (Zhu et al., 2010) with gradient as in (Mostafa et al., 2016) at a flow rate of 200 μl/min. In brief, solvent A (25% v/v acetonitrile, 10 mM ammonium formate, and 0.1% v/v formic acid, pH 2.8) was applied for 10 min and a linear gradient of 0–20% solvent B (25% v/v acetonitrile and 500 mM ammonium formate, pH 6.8) was applied over 80 min before applying another 5-min gradient to 100% solvent B and holding for 10 min. Peptide elution was monitored by absorbance at 280 nm (A₂₈₀). SCX fractions (14 total) were pooled and lyophilized. In addition, replicate 1 had one post-run fraction and replicates 2–4 have 5 post-run fractions that were included in the LC-MS/MS analysis.

Reversed-phase liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis

Fractions were resuspended in LC solvent A (0.1% v/v formic acid in 3% v/v acetonitrile) and analyzed one at a time on an Easy-nLC 1200 system coupled to a Q-Exactive Orbitrap Plus MS (Thermo Fisher Scientific, Bremen, Germany). Peptides were concentrated on an Acclaim Pepmap 100 pre-column (20 mm × 75 μm; 3 μm-C18) and separated on a PepMap RSLC analytical column (250 mm × 75 μm; 2 μm-C18) at a flow rate of 300 nl/min using solvent A (0.1% v/v formic acid) and B (0.1% v/v formic acid and 99.9% v/v acetonitrile) at a gradient of 2–30% solvent B in 100 min, 30–98% solvent B in 10 min before an isocratic flow of 98% solvent B for 10 min. The separated peptides were analyzed on a Q-Exactive Plus MS (Thermo Fisher Scientific, Bremen, Germany) in positive ion mode and top 10 data dependent scanning with high collision dissociation (HCD) as previously described (Mostafa et al., 2016). Briefly, the spray voltage was 1800 V. The full MS resolution was 70,000 with a scan range of 350–1800 m/z, an AGC target of 3e6 and maximum IT of 250 ms. The MS/MS resolution was 17,500 with a scan range of 200–2000 m/z, an AGC target of 5e5, and maximum IT of 50 ms, the normalized collision energy of 27. The underfill ratio was 1%, intensity threshold was 2e5, charge exclusion was unassigned, 1, 7–8, and >8.

Data analysis and statistical rationale

Proteome Discoverer 2.1 (Thermo Scientific, Bremen, Germany) was used for protein identification and quantitation using the Hfx. volcanii DS2 database (downloaded from www.Uniprot.org April 2017) with 3,996 entries and NTRANS_0004 (Ammar et al., 2012). Proteome Discoverer was also used to generate the decoy database of the Hfx. volcanii proteome that was used to estimate the false discovery rate (FDR). Peptide and protein identification was set to an FDR adjusted p-value (or q-value) of 0.01 or less implying that 1% of significant tests would result in false positives. cRAP database was also searched for protein contaminants (none were detected at a q-value of 0.01 or less). Peptides were allowed two missed cleavages by trypsin and a static modification of methylthio on all cysteines. Dynamic modifications allowed were lysine +8, arginine +6, proline +5, methionine oxidation, N-terminal acetylation, and diglycine remnant on lysines. Mass tolerance for precursor ions was 10 ppm. Mass tolerance for fragment ions was 0.02 Da. Peptide abundances were mean normalized and scaled by Proteome Discoverer. Peptides identified for each protein were used to find the statistical significance of protein abundance using Welch’s t-test in R (version 3.3.3) to a p-value of < 0.05 (Cox et al., 2014). Ambiguous peptides and peptides with no or non-unique quantitative information were excluded in the analysis. The 50 most abundant proteins and 50 least abundant proteins were identified using the highest and lowest exponentially modified protein abundance index (emPAI) values, respectively (Ishihama et al., 2005). For the least abundant proteins, the proteins were identified in at least three replicates, and the most abundant proteins were found in all four replicates. The effect of label swapping was tested using Welch’s t-test at p-value < 0.01 using the peptide abundances from the control groups in the 50 most abundant proteins by emPAI value found in all four replicates. Proline conversion was calculated by taking the top three most abundant proteins by emPAI value and counting the number of heavy prolines (+5) out of total proline residues. Protein arCOG information (Makarova et al., 2007) was found in www.uniprot.org (The UniProt, 2017). Transmembrane domains were determined by Expasy TMpred server (Hofmann and Stoffel, 1993). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaino et al., 2016) partner repository with the dataset identifier PXD006877 (Username: reviewer36568@ebi.ac.uk and Password: IqT96pAU).

Supplementary Material

Supp TableS1. Table S1.

Peptide modification dataset used to calculate heavy proline conversion from heavy arginine.

Supp TableS2. Table S2.

Haloferax volcanii 2,565 proteins identified across all four replicates and 1,806 proteins identified in all four replicates at a q-value of 0.01 or less.

Supp TableS3. Table S3.

50 most abundant proteins of Haloferax volcanii identified by SILAC analysis.

Supp TableS4. Table S4.

50 least abundant proteins of Haloferax volcanii identified by SILAC analysis.

Supp TableS5. Table S5.

Statistical analysis of label-swap replication of SILAC experiments to examine the effect of labeling on protein abundance.

Supp TableS6. Table S6.

Haloferax volcanii proteins of differential abundance by NaOCl treatment at a p-value less than 0.05.

Supp TableS7. Table S7.

Haloferax volcanii proteins up at least 1.5-fold by NaOCl treatment at a p-value less than 0.05.

Supp TableS8. Table S8.

Haloferax volcanii proteins down at least 1.5-fold by NaOCl treatment at a p-value less than 0.05.

Originality-Significance Statement.

SILAC for robust, multiplexed quantitative comparison of proteomic responses of archaea is now demonstrated by case study of oxidative/hypochlorite stress responses of Haloferax volcanii.