ARTICLE pubs.acs.org/jpr
Quantitative Proteomic Analysis of the Heat Stress Response in Clostridium difficile Strain 630 Shailesh Jain,† Ciaren Graham,†,‡ Robert L. J. Graham,†,§ Geoff McMullan,† and Nigel G. Ternan*,† †
School of Biomedical Sciences, University of Ulster, Cromore Road, Coleraine, Co. Londonderry, North Ireland, United Kingdom Jane Ann Nohl Division of Hematology and Centre for the Study of Blood Diseases, Keck School of Medicine, University of Southern California, Los Angeles, California, United States § The Proteome Exploration Laboratory, California Institute of Technology, Beckman Institute, Pasadena, California, 91125, United States ‡
bS Supporting Information ABSTRACT: Clostridium difficile is a serious nosocomial pathogen whose prevalence worldwide is increasing. Postgenomic technologies can now be deployed to develop understanding of the evolution and diversity of this important human pathogen, yet little is known about the adaptive ability of C. difficile. We used iTRAQ labeling and 2D-LC MS/MS driven proteomics to investigate the response of C. difficile 630 to a mild, but clinically relevant, heat stress. A statistically validated list of 447 proteins to which functional roles were assigned was generated, allowing reconstruction of central metabolic pathways including glycolysis, γ-aminobutyrate metabolism, and peptidoglycan biosynthesis. Some 49 proteins were significantly modulated under heat stress: classical heat shock proteins including GroEL, GroES, DnaK, Clp proteases, and HtpG were up-regulated in addition to several stress inducible rubrerythrins and proteins associated with protein modification, such as prolyl isomerases and proline racemase. The flagellar filament protein, FliC, was downregulated, possibly as an energy conservation measure, as was the SecA1 preprotein translocase. The up-regulation of hydrogenases and various oxidoreductases suggests that electron flux across these pools of enzymes changes under heat stress. This work represents the first comparative proteomic analysis of the heat stress response in C. difficile strain 630, complementing the existing proteomics data sets and the single microarray comparative analysis of stress response. Thus we have a benchmark proteome for this pathogen, leading to a deeper understanding of its physiology and metabolism informed by the unique functional and adaptive processes used during a temperature upshift mimicking host pyrexia. KEYWORDS: iTRAQ, proteomics, multidimensional, Clostridium difficile, heat stress, adaptation
’ INTRODUCTION Clostridia, it is proposed, appeared as a distinct and heterogeneous class of anaerobic bacteria some 2.7 billion years ago prior to the initial rise in oxygen.1 The genus Clostridium comprises a heterogeneous group of obligately anaerobic, Gram-positive, endospore forming bacilli that exhibit broad biocatalytic diversity with organisms such as C. acetobutylicum, C. beijerinckii, and C. cellulovorans routinely employed in the industrial production of biofuels.2 5 However, the genus is perhaps best known for a number of toxin-producing organisms, including C. difficile and C. botulinum, which have attracted considerable attention during the past few decades owing to the widespread socioeconomic impact attributable to them each year.6,7 C. difficile, a serious nosocomial pathogen, is said to be the most frequent cause of infectious bacterial diarrhea worldwide.8 C. difficile infection can also lead to potentially life-threatening pseudomembranous colitis (PMC) and toxic megacolon,9 r 2011 American Chemical Society
although these complications are rarer nowadays due to earlier diagnosis and better treatment. Numerous reports exist describing the pathogenesis of C. difficile, which is hypothesized to involve three principal events: an initial alteration of the indigenous colonic microflora by broad-spectrum antibiotics; germination of C. difficile spores and multiplication of vegetative cells unhindered by the indigenous microflora; and finally the release of the two main virulence factors, toxin A and toxin B, which eventually result in the classical symptoms associated with C. difficile infection (CDI).6,10 12 The incidence and severity of C. difficile infection has risen significantly among patients in hospitals worldwide,13 the fundamental reason being the alarming emergence of novel, hypervirulent strains Received: December 9, 2010 Published: July 25, 2011 3880
dx.doi.org/10.1021/pr200327t | J. Proteome Res. 2011, 10, 3880–3890
Journal of Proteome Research of C. difficile (e.g., ribotype 027),14,15 increased antimicrobial resistance,16 or possibly both. The publication by Sebaihia and co-workers in 2006 of the complete genome sequence of C. difficile strain 63016 has enabled researchers to carry out meaningful postgenomic analyses of aspects of C. difficile genetics and physiology. Recent publications have given insights into the evolution, lineage, and physiology of these organisms.8,17 C. difficile genomes, as might be expected for a notorious pathogen, contain a vast spectrum of genes primarily involved in resistance to antimicrobial agents, virulence, and host interaction, as well as the production of surface structures and various metabolic capabilities that allow survival within the challenging gut environment.16 However, it is notable that clones of the highly virulent and clinically significant ribotype 027, responsible for some 34% of reported UK CDI cases,18 contain an additional 234 genes, some of which are proposed to have been acquired over the past 20 years.17 Such genes, including those encoding binary toxin (cdtA and cdtB), various transcriptional regulators and two component systems and drug transporters, in addition to mutations conferring resistance to, for example, fluoroquinolones, have led to the evolution within a relatively short time scale of a highly effective and virulent pathogen. A core genome has been defined for C. difficile, representing some 16% of the genes within the C. difficile 630 genome;19 however, C. difficile as a whole is a genetically diverse species.17 Currently, some 29 C. difficile genome sequences are available within publicly accessible databases, and recent comparative genomic analysis has revealed the significance of horizontal gene transfer and large-scale recombination in shaping the C. difficile genome over both short and long time scales.17 This structural genomics data, along with recently developed tools for Clostridial functional genomics,20 makes it possible for researchers to now undertake systems biology investigations.21 Despite the fact that it is one of the leading causes of nosocomial infections worldwide14,22 and a significant burden upon patients and healthcare systems, little is known about the adaptive ability of C. difficile in response to hostile environments. Emerson et al.23 began to address this knowledge gap using microarray analysis of the transcriptional responses of C. difficile strain 630 to a variety of antimicrobial and environmental stresses. Transcriptional profiling identified a number of significantly regulated genes, operons, and pathways unique to, or common between, different stress conditions; thus, their study “...provides a starting point for detailed analysis of numerous genes whose expression is affected by stress and may therefore be involved in adaptation to the host environment.” To complement this transcriptomic approach, we now report the first comparative proteomic analysis of C. difficile strain 630 grown under a stress condition.
’ MATERIALS AND METHODS Reagents
All chemicals and reagents, of the highest purity available, were purchased from Sigma-Aldrich (Poole, U.K.), unless otherwise stated. Mass spectrometry grade water and acetonitrile (ACN) were purchased from Romil (Cambridge, U.K.) and sequencing grade trypsin was from Promega (Southampton, U.K.). All molecular biology reagents were purchased from Invitrogen (Renfrewshire, U.K.) save for qPCR reagents, which were obtained from Roche Diagnostics (Hertfordshire, U.K.).
ARTICLE
Cell Culture and Growth Conditions
Clostridium difficile strain 630 was a kind gift from Dr Peter Mullany of the Eastman Dental Institute, London and was routinely maintained on BHI agar (Oxoid) at 37 °C in a MACS MG500 Anaerobic workstation fitted with an airlock (Don Whitley Scientific, U.K.). The workstation was operated on a conventional anaerobic gas mixture containing 80% N2, 10% H2 and 10% CO2 and resazurin (1 mg L 1) was used in all growth media as a redox indicator. Routine growth of the organism involved the inoculation of autoclaved, prereduced BHI broth (100 mL) with a single actively growing colony from BHI agar. Cultures were grown overnight (∼16 h), and subsequently used as inocula at 5% (v/v) for growth in 1 L cultures, which were monitored by the increase in culture attenuance at 650 nm (D650) versus uninoculated BHI broth. Heat stress was induced in the early exponential phase (D650 = 0.3) of growth by transferring liquid cultures growing at 37 °C to a preheated 41 °C water-bath and incubation continued for a further 3 h to D650 = 1.1. Maintenance of anaerobiosis was confirmed by the observation that the resazurin remained colorless at all times. Attenuance was measured in the 41 °C culture bottle by briefly transferring it back to the anaerobic cabinet for removal of an aliquot, followed by a return to the 41 °C water bath. Cell Harvest and Lysis
C. difficile strain 630 cultures (1 L) were harvested under normal (37 °C) and heat-stress (41 °C) conditions at late-log phase (D650 = 1.1) of anaerobic growth by centrifugation in a sealed tube at 10 000 g for 15 min at 4 °C in the JA10 rotor of a Beckman J2-HS centrifuge (Beckman Instruments, Fullerton, CA). Inside the anaerobic cabinet, the spent broth supernatants were discarded and the cells were resuspended and washed using ice-cold 10 mM phosphate-buffered saline (pH 7.8). Cell pellets were again obtained by centrifugation at 10 000 g for 15 min at 4 °C, the PBS was decanted in the cabinet, and the pellets were then resuspended in 1.5 mL of 1 M triethylammonium bicarbonate (TEAB) buffer containing 0.05% SDS (w/v) on ice for 30 min with regular vortexing. To break the cells open, the samples were transferred to a 2 mL Lysing Matrix E tube (MP Biomedicals, Cambridge, U.K.), sealed, removed from the cabinet, and homogenized using the FastPrep FP120 Instrument. Cells were subjected to three 30 s disruptions at a speed setting of 5.5, with 2 min cooling on ice between treatments. The lysate was centrifuged in the Lysing tube at 10 000 g for 20 min at 4 °C (Beckman Allegra 64), and the protein concentration in the supernatant was determined using Bradford’s reagent24 (Bio-Rad Laboratories, Hertfordshire, U.K.). iTRAQ Reagent Labeling, Peptide Fractionation and Mass Spectrometry
The iTRAQ reagent labeling work was carried out essentially as previously described by Unwin et al.25,26 A total of 100 μg of protein from each sample was prepared for the labeling reaction according to the iTRAQ kit manufacturer’s instructions. The control samples (37 and 37 °C biological duplicate) and peptide samples were labeled with reagents 114 and 116 whereas the heat shock (41 and 41 °C biological duplicate) samples were labeled with reagents 115 and 117, respectively. Peptides were fractionated using strong cation-exchange chromatography (SCX) on an Agilent 1100 series instrument (Agilent, Wokingham, U.K.) as described previously.27,28 SCX fractions were resuspended in 130 μL of 2% ACN and 0.1% (v/v) formic acid and for mass 3881
dx.doi.org/10.1021/pr200327t |J. Proteome Res. 2011, 10, 3880–3890
Journal of Proteome Research spectrometry analysis, 60 μL of the peptide sample was loaded onto an UltiMate 3000 nanoflow liquid chromatography system (Dionex/LC Packings, Amsterdam). A μ-Precolumn Cartridge (300 μm 5 mm, 5 μm particle size) was placed prior to the C18 pepmap 100 (C18 silica-based) capillary column (75 μm 150 mm, 3 μm particle size, 100 Å) to enable desalting. Separation was over a 200 min solvent gradient from 2% ACN and 0.1% (v/v) formic acid to 80% ACN and 0.1% (v/v) formic acid online to a Q-TRAP 3200 Hybrid ESI Quadrupole linear ion trap mass spectrometer (ESI-Q-q-Qlinear ion trap-MS/MS) (Applied Biosystems/MDS SCIEX, Toronto, Canada). Data was acquired using a data dependent acquisition protocol in which, for each scan cycle, the two most abundant multiply charged peptides (2+ to 4+) in the MS scan with m/z between 400 and 1400 were selected for MS/MS. Each peptide was selected twice and then dynamically excluded for 1 min. Database Searching and Analysis
MS-data was processed by a “thorough” search against a combined C. difficile strain 630 genomic DNA and plasmid database (Refseq NC_0090989 and NC_008226, respectively) downloaded from NCBI (20/06/2007) and containing 3573 sequences in total,21 using the Paragon algorithm within ProteinPilot v1.1 software (Applied Biosystems). Search parameters included trypsin as the digest agent, with methylmethanethiosulfate-alkylation of cysteine, iTRAQ modification of lysine and N-termini, plus one missed cleavage. Variable modifications included iTRAQ labeling of tyrosine. Search parameters within the paragon algorithm were set with an MS tolerance of 1.2 Da, an MS/MS tolerance of 0.6 Da, and only peptides with a minimum confidence score of 70 were included in the final data set: this allowed only the highest quality MS/MS spectra to be used for protein identification and therefore relative quantification. A minimum of 3 spectra were required to identify and relatively quantify a protein. Quantitation was achieved by calculating a weighted average from all matching peptides along with a 95% confidence interval to assess accuracy of protein ratio. Only MS/MS spectra that were unique to a particular protein, and where the sum of the signalto-noise ratio for all of the peak pairs was >9, were used for quantification (default software settings). For each protein ratio a p-value was generated, thus allowing the results to be evaluated based on both the certainty of a change in expression, in addition to the magnitude of that change. P-value determination was calculated using the Student t-factor by dividing (Weighted Average of Log Ratios log Bias) by the Weighted Standard Deviation, thus allowing determination of the p-value, with n 1 degrees of freedom, where n is the number of peptides contributing to protein relative quantification. iTRAQ ratios greater than or equal to 1.5 or less than 0.7 and with a p-value less than 0.05 were flagged as being differentially expressed, similar to the ratios used by other researchers.27 32 Output files (MS/MS data) were analyzed using ProteinPilot software (Applied Biosystems) and FDR analysis was carried out using the embedded PSPEP tool. This tool reversed the protein database of 3573 sequences and reinterrogated all MS/MS data to generate a statistically validated protein list with all falsepositive proteins removed. Bioinformatics analysis was carried out as previously described by Graham et al.33 using PSORTb version 2.0.4,34 SignalP 3.0,35 and SecretomeP 2.0.36
ARTICLE
genes (16S rRNA, tpi) and GroEL were as described by Matsuda et al.37 and Lemee et al.38,39 Genomic DNA was extracted using the FastDNA SPIN kit for Soil (QBIOgene, Cambridge, U.K.) according to the manufacturer’s instructions. Primer sequences and optimized PCR conditions for individual genes are given in Supporting Information, Data File 1. RNA was extracted from late-log phase cells using QIAGEN’s RNeasy Mini kit. cDNA was synthesized using a reverse primer specific for each of the individual genes targeted. Total RNA (100 ng) was reverse transcribed using the SuperScript II Reverse Transcriptase kit (Invitrogen, Renfrewshire, UK). qPCR amplification was carried out in a LightCycler 2.0 Carousel-Based System using the LightCycler FastStart DNA MasterPLUS SYBR Green I kit. Absolute quantification of target gene expression was achieved using a standard curve constructed by amplifying known amounts of target gene PCR product. Each gene was assayed using RNA from biological duplicates, with technical qPCR triplicates. Standard deviations were an order of magnitude lower than the measured value for all samples. “No template” and “no reverse transcriptase” controls were included. Values were subsequently expressed as fold change ratios relative to the 37° control sample to allow comparison with the iTRAQ data.
’ RESULTS AND DISCUSSION While the epidemiology, toxin production ability and the infectious processes of numerous C. difficile strains is well characterized, there remains much to be discovered about the physiology and metabolism of this pathogen. Despite increasing availability of C. difficile genomic information,17 only a few groups have attempted to utilize postgenomic approaches, largely transcriptomic, to develop a systems-wide understanding of C. difficile.23,40,41 Most proteomic investigations of C. difficile are static snapshots21,42,43 that provide useful, benchmark, information about what cellular processes are functional. Comparative proteomics investigations, driven in the main by 2-dimensional gel electrophoresis (2DGE), have been applied to the less complex extracellular and surface proteomes of C. sordellii and C. perfringens.44 46 We believe therefore that there exists a relative dearth of robust comparative shotgun proteomics data on the response of C. difficile to changing conditions. In this investigation, we wished to quantify changes in the functional protein machinery of C. difficile strain 630 grown under a clinically realistic47 temperature stress (41 °C compared to 37 °C). We hypothesized that this would allow the identification of C. difficile proteins, and pathways, affected by and of importance in the heat stress response.48 A number of mechanisms are available to allow quantitative proteomic investigations to be undertaken. We utilized the 4 plex iTRAQ kit (Applied Biosystems) developed by Ross and coworkers.49 This is a well characterized method where peptides are tagged with amine-specific isobaric labels at their free aminotermini and lysine side chains. MS/MS fragmentation releases the labels producing ions with a m/z of 114, 115, 116, and 117 Da respectively. The peak area of the released reporter ions can then be used to assess the relative abundance of peptides and thus the proteins from which they are derived.50 52
Gene Expression Analysis
iTRAQ Methodology Generates a Robust, Validated, Proteomics Data Set
Primers for the coding region of target genes were designed using Invitrogen’s OligoPerfect software. Primers for reference
The aim of this investigation was to understand more about the cellular response of C. difficile strain 630 to heat stress (41 °C 3882
dx.doi.org/10.1021/pr200327t |J. Proteome Res. 2011, 10, 3880–3890
Journal of Proteome Research
ARTICLE
Figure 1. Functional categorization of 447 proteins identified within the proteome of Clostridium difficile 630.
relative to 37 °C). The MS/MS output files generated during the iTRAQ workflow were therefore analyzed using ProQuant and ProteinPilot software to generate a statistically validated list of 447 proteins (Supporting Information, Data File 2), representing some 12% of the predicted proteome of C. difficile strain 630. Our data set thus represents a benchmark 2D-LC MS/MS proteome, comparable, in terms of number of proteins identified and total proteome coverage achieved, to other recent iTRAQ driven investigations using Quadropole/ion trap or TOF mass spectrometers.29,53 57 Upon data curation, it was observed that the proteins identified varied extensively in terms of their physiochemical properties such as molecular mass (Mr) and pI (Supporting Information, Data File 3) and as in our previous work,58 60 we were able to assign functional roles for 411 (92%) of the identified C. difficile strain 630 proteins. These were known, or could be predicted in accordance to the SubtiList functional classification list61,62 allowing the reconstruction63,64 of a number of central metabolic pathways (Supporting Information, Data File 4). As previously reported,21 the majority of proteins identified under both growth conditions were involved in various aspects of protein synthesis (19.5%), the metabolism of amino acids and related molecules (15%), specific metabolic pathways (9%), and glycolysis (3.5%) (Figure 1). Pathways identified included glycolysis (10 proteins), acetylCoA fermentation to butyrate (10 proteins), the pentose phosphate pathway (5 proteins), isoleucine biosynthesis (5 proteins), lysine biosynthesis (7 proteins), fatty acid elongation (3 proteins), GABA metabolism (12 proteins) and peptidoglycan biosynthesis (11 proteins). As in some of our previous investigations, a large number of t-RNA synthetases (23 proteins), and ribosomal proteins (40 proteins) were detected.58 60 Relatively few significant changes in abundance were noted within these “core pathway” proteins, suggesting that C. difficile is relatively robust when encountering heat stress and that adaptation to higher
temperatures is not necessarily mediated by a global change in the abundance of the protein machinery of the cell alone. Changes in the Benchmark C. difficile Strain 630 Proteome under Heat Stress
To allow the identification of proteins affected by heat stress within the C. difficile strain 630 proteome, a comparative data analysis was carried out in which a g50% change in protein abundance was considered significant and sufficient to take into account systemic errors and biological variation within the data sets. Hence in the current investigation, fold changes of g1.5 or e0.7 were considered significant, with a fold change value of 1 representing no change in protein expression levels between the two C. difficile strain 630 growth conditions.27 32 Comparison of protein expression patterns from the biological replicates grown at 37 °C showed, with only a few exceptions, tight clustering of data points within the 0.7 to 1.5 fold change range (Figure 2a). This observation is indicative of the reliable and reproducible nature of the iTRAQ method, and gives confidence in our ability to carry out a comparative quantitative analysis. A markedly different picture was observed, however, when comparing the proteome of C. difficile strain 630 grown at 41 °C relative to that of the 37 °C culture, with a range of protein expression patterns being apparent (Figure 2b), many outside the 0.7 - 1.5 fold change range. From this comparative analysis, 50 proteins representing some 11% of the total number of proteins identified in this investigation, were found to be differentially expressed under heat stress conditions (Supporting Information Data File 5). This percent change, in protein expression, is consistent with the percent of significant protein expressional changes reported by other investigators in the field of bacterial proteomics.53,57,65 A total of 30 proteins were significantly upregulated under the applied heat stress while 19 proteins were significantly down-regulated. All expressionally changed proteins 3883
dx.doi.org/10.1021/pr200327t |J. Proteome Res. 2011, 10, 3880–3890
Journal of Proteome Research
ARTICLE
Figure 2. Ratiometric plots examining robustness and reproducibility of the iTRAQ quantitation procedure on the Clostridium difficile 630 proteome. (a) Plotted ratios for proteins using the 37 °C duplicate samples (116:114) respectively. Few ratios are greater than 1.5-fold, thus giving confidence in the cutoff values chosen. (b) Plotted ratios for proteins from one of the normal and heat shock data sets (117:114), clearly indicating the differences in protein expression pattern between the two growth conditions.
were functionally characterized with respect to their biochemical role within the organism. Proteomic Analysis of the Heat Stress Response in C. difficile Strain 630
As expected, a number of well-characterized heat shock proteins66 were identified (Supporting Information Data File 2) although surprisingly, not all of these were significantly changed (Supporting Information Data File 5) under the applied heat stress. Induction of the class I heat shock proteins GroEL (CD0194), GroES (CD0193) and DnaK (CD2461) was observed at 41 °C. These data are in accordance with the earlier work of Hennequin et al.67 on GroEL, and the more recent whole genome microarray data of Emerson et al.;23 in both investigations, GroEL
was heat, acid, and antibiotic stress inducible. The cochaperone DnaJ (CD2460) was also identified, but statistically we could not say that it was significantly modulated, having a fold change of only 1.38 and a p-value for the data of >0.05. This protein, along with any others for whom the p-value was in excess of 0.05, was not included in our final, statistically validated list of modulated proteins. Emerson et al.23 reported that although the DnaJK-GrpE operon was up-regulated under stress, the changes in mRNA levels were not consistent between genes, as might be expected for this apparently tricistronic operon. However, the up-regulation observed by Emerson et al.23 for DnaJ was, in all stresses, less than that observed for either DnaK or GrpE (CD2462, just below our cutoff at 1.45-fold up), and is therefore similar to what our proteomics data set reveals for the gene products. In Bacillus 3884
dx.doi.org/10.1021/pr200327t |J. Proteome Res. 2011, 10, 3880–3890
Journal of Proteome Research subtilis, it is known that to adjust cellular quantities of proteins derived from the dnaK operon, both transcriptional and posttranscriptional control methods are employed and a strategy of differential segmental mRNA stability is in place to fine-tune the expression of individual DnaK operon genes.68 A similar scenario may exist in C. difficile 630, resulting in different levels of gene products from the operon. Members of the class III group of heat shock proteins such as ClpP1 (CD3305) and ClpC (CD0026), encoding the subunits of ClpCP ATP-dependent protease, were up-regulated in response to heat stress. Previously, Derre et al.69 reported that these Class III proteins are required for stress survival, competence development, protein quality control and growth at high temperatures in B. subtilis and their induction in C. difficile suggests they may play a similar, chaperone-like role. ClpB (CD2020) was also upregulated during heat stress in C. difficile. Its modulation could be associated with the observed increased levels of DnaK, as ClpB is also a member of the protein-disaggregating chaperone machinery, rescuing aggregated proteins in association with the DnaK chaperone system.70,71 The former Class III heat-inducible protein, HtpG (CD0273, Hsp90), now reclassified as the single, first example of Class IV,72 is regulated by an unidentified transcriptional activator73 and was up-regulated in our investigation. HtpG has recently been reported to associate with the ribosomal 50S L2 protein during which its ATPase activity is stimulated74 and this interaction may contribute to ribosomal biogenesis and stability. Previously, heat-responsive genes that did not belong to one of the original three classes were collectively termed class IV genes, including, clpX, ftsH and lonB among others.69,75 However following analysis of HtpG by Versteeg et al.72 all genes of the former class IV heat inducible category (including the membrane bound serine proteases HtrA and HtrB, and their cognate sensor/regulator, CssS/CssR) are now reclassified as class V.72,73 Thus, we have adopted this nomenclature here. While certain of the (new) class VI heatinducible proteins such as Lon (CD3301), ClpX (CD3304), FtsH2 (CD3559), and FtsZ (CD2646) were detected, their abundance was not significantly altered by the applied heat stress (Supporting Information Data File 2). The mode of regulation for these proteins is still unknown, being neither CIRCE, CtsR, or σB dependent,66,73 and their lack of response to heat stress in C. difficile remains unexplained. The overlapping nature and membership of bacterial stress responses has been widely reported,73,76 and in the current investigation, we identified a number of differentially expressed proteins associated with the “general stress response” in bacteria. Rubrerythrins are stress responsive proteins initially described for their role in protecting strictly anaerobic bacteria from exposure to oxygen,77 although recently, nitric oxide reductase activities, playing a role in nitrosative stress response, have been demonstrated in certain of these proteins.78 We observed an increase in abundance of a number of rubrerythrin and rubredoxin-type proteins under heat stress. The rubrerythrin-like protein Hsp21 of C. acetobutylicum, which is encoded by the genes CA_C3597 and CA_C3598 in C. acetobutylicum ATCC824 is a general stress protein, inducible by heat.79 This reverse rubrerythin protein, in which the rubredoxin and rubrerythin domains are N and C terminal, respectively, unlike most other rubrerythrin proteins80,81 is not annotated as such in the C. difficile 630 genome. Upon blast searching the relevant amino acid sequences from C. acetobutylicum ATCC824 against the C. difficile 630 genome, good matches were obtained for two
ARTICLE
proteins that were detected in our investigation, namely CD1474 (up-regulated) and CD1524 (no change). Both of these genes are annotated in the C. difficile 630 genome as “putative rubrerythin” and are not located beside each other in the genome, unlike in C. acetobutylicum. Karlsson et al.40 also observed an increase in abundance of a 21 kDa “rubrerythrin” with the same N-terminal sequence as CD1524 and CD1474 upon cultivation of C. difficile VPI 10463 at 37 °C compared with 22 °C. However, they also reported a lower abundance of this protein at 42 °C than at either of the other two temperatures— an observation at odds with the current study. In contrast, Emerson et al.23 showed that transcript from both CD1474 and CD1524 was unaffected by heat but was up-regulated under acid and oxidative stress; therefore, the mode of regulation of these proteins is still unclear. We also observed an increase in abundance of the “classical” rubrerythrin (rbr, CD0825) in response to heat stress, but no significant change in the putative corresponding rubredoxin oxidoreductase (desulfoferrodoxin, CD0827) was observed. Emerson et al.23 also reported an increase in transcript from both CD0825 and CD0827 under heat stress. Interestingly, they saw no change in transcript of the other rbr gene, CD2845, under O2 stress, but recorded significant up-regulation under heat, acid, and alkali stress, suggesting that in clostridia these oxidative defense proteins are not regulated solely by exposure to oxygen. The increased expression of CD1157, annotated as a flavorubredoxin nitric oxide reductase, was observed. These proteins are involved in dissimilatory nitrate reduction, where they detoxify nitric oxide (NO), thus anaerobes associated with the GI tract can respond to the cytotoxic effects of nitrosative stress, which could be encountered during host inflammatory responses to infection. In Desulfovibrio gigas, transcripts of flavodiiron protein (FDP), rubredoxin/oxygen oxidoreductase (ROO) were increased by exposure to NO but not by oxygen.78 Emerson et al.23 showed that CD1157 was not significantly changed at the transcript level by heat shock or metronidazole but was upregulated under all other stress conditions. It appears from our data and that of Emerson et al.23 that there is significant overlap between the defense mechanisms induced by both heat and ROS/RNS stress in C. difficile 630. Molecular hydrogen is produced by a variety of clostridial fermentations, and thus, hydrogenases are of interest to those wishing to produce hydrogen as an energy source from renewable substrates.82 Clostridia possess Fe only hydrogenases, components of the anaerobic respiratory chain, that are 10-fold more active at hydrogen production than Ni Fe hydrogenases.83 In the current study, HymABC, encoded by the tricistronic operon CD3405 3407, increased in abundance during heat stress. Gao et al. saw a similar increase in expression of hydrogenases during growth of Shewanella oneidensis at elevated temperatures.84 As of yet, it is unclear why these Fe-only hydrogenases would be induced in response to heat stress, although Graentzdoerffer et al. provided evidence that these three Hym proteins were important in the shuttling of electrons to formate dehydrogenase enzymes in the Gram positive anaerobe Eubacterium acidaminophilum.85 This organism is thus able to use electrons generated from the oxidation of formate to reduce glycine, or a number of analogues, to the readily available energy source acetyl phosphate.86 This could implicate these enzymes in increased energy generation or amino acid fermentation reactions. The modulation of various oxidoreductases, as outlined above, 3885
dx.doi.org/10.1021/pr200327t |J. Proteome Res. 2011, 10, 3880–3890
Journal of Proteome Research
ARTICLE
Figure 3. Reverse transcription qPCR analysis of selected Clostridium difficile 630 genes under heat stress. Absolute quantitation experiments were performed in biological duplicates, with technical triplicates and for each individual gene, expressional fold-change values (hatched columns) are shown relative to the 37 °C control. Corresponding iTRAQ fold-changes (gray columns) are included for comparison and show good correlation between the data sets. 16S rRNA, tpi, and CD2849, whose expression did not change by more than 1.5-fold, were used as reference genes.
might suggest that electron flux across these pools of enzymes changes as a result the cellular response to heat stress.
following a change in growth temperature from 37 to 43 °C, similar to the temperature upshift used here.
Transcript Analysis of Heat Stress Responsive Genes
Changes in Protein Folding and Export Machinery in C. difficile Strain 630 during Heat Stress
It is evident from the existing literature that results obtained by “omics” technology should preferably be confirmed using another method.87 In the absence of suitable commercially available C. difficile specific antibodies for Western blotting analysis of proteins in whose expression we observed changes, we performed RT-q-PCR on RNA samples extracted from the cultures. We determined the expression of a number of genes encoding proteins whose abundance changed upon heat stress. Absolute quantitation of gene expression yielded expressional values for individual genes covering some 4 orders of magnitude, with standard deviations of less than 20%. These measurements of gene expression data were considered robust and therefore suitable for comparison with the proteomics data. We used 16S rRNA, tpi and CD2849 (putative bifunctional phn type protein) as reference genes; their expressional difference was
