|Home | About | Journals | Submit | Contact Us | Français|
The distinctive feature of the GroES-GroEL chaperonin system in mediating protein folding lies in its ability to exist in a tetradecameric state, form a central cavity, and encapsulate the substrate via the GroES lid. However, recombinant GroELs of Mycobacterium tuberculosis are unable to act as effective molecular chaperones when expressed in Escherichia coli. We demonstrate here that the inability of M. tuberculosis GroEL1 to act as a functional chaperone in E. coli can be alleviated by facilitated oligomerization. The results of directed evolution involving random DNA shuffling of the genes encoding M. tuberculosis GroEL homologues followed by selection for functional entities suggested that the loss of chaperoning ability of the recombinant mycobacterial GroEL1 and GroEL2 in E. coli might be due to their inability to form canonical tetradecamers. This was confirmed by the results of domain-swapping experiments that generated M. tuberculosis-E. coli chimeras bearing mutually exchanged equatorial domains, which revealed that E. coli GroEL loses its chaperonin activity due to alteration of its oligomerization capabilities and vice versa for M. tuberculosis GroEL1. Furthermore, studying the oligomerization status of native GroEL1 from cell lysates of M. tuberculosis revealed that it exists in multiple oligomeric forms, including single-ring and double-ring variants. Immunochemical and mass spectrometric studies of the native M. tuberculosis GroEL1 revealed that the tetradecameric form is phosphorylated on serine-393, while the heptameric form is not, indicating that the switch between the single- and double-ring variants is mediated by phosphorylation.
GroEL, an essential chaperonin, is known to form a ring-shaped structure for sequestering substrate proteins from the crowded cellular milieu and is responsible for the occurrence of various cellular processes, such as de novo folding, transport, and macromolecular assembly, within a biologically relevant time scale (7, 26, 48, 53). In Escherichia coli, GroEL, along with its cofactor GroES, assists the folding of about 10 to 30% of cytosolic proteins, among which some are known to be essential for cell viability (15, 26, 27, 31). GroEL was originally identified as the host factor responsible for phage λ and T4 capsid protein assembly and was subsequently shown to be essential for cell viability (17, 20). E. coli groEL is found in an operonic arrangement with groES (groESL), and its expression is regulated by multiple promoter elements.
GroEL function has been shown to be a complex interplay between its interaction with and encapsulation of substrate proteins, with concomitant conformational changes induced by ATP binding, hydrolysis, and GroES binding (24, 56, 62). E. coli GroEL exists as a homotetradecamer forming two isologous rings of seven identical subunits each. Crystallographic analyses have delineated the three-domain architecture of GroEL monomers and the GroES-GroEL interactions (4, 63). The central region of the GroEL polypeptide, spanning amino acid residues 191 to 376, constitutes the GroES and substrate polypeptide-binding apical domain. The equatorial ATPase domain spanning two extremities of the GroEL polypeptide, that is, residues 6 to 133 and 409 to 523, is responsible for the ATPase activity and the bulk of intersubunit interactions. The hinge-forming intermediate domain, spanning two regions on the polypeptide, namely, residues 134 to 190 and 377 to 408, connects the said two domains in the tertiary structure. The conformational changes resulting from ATP binding and hydrolysis at the equatorial domain are coupled to those occurring at the apical domain via this hinge region (4, 63).
The usual size limit for the substrate proteins, as shown by both in vitro and in vivo studies, is around 57 kDa, although the cis cavity is reported to theoretically accommodate larger proteins, on the order of 104 kDa (10, 27, 35, 46). Productive in vivo folding of the proteins larger than the usual size limit, such as the 86-kDa maltose binding protein fusion and 82-kDa mitochondrial aconitase, has also been reported (9, 29). Since such large substrates are difficult to accommodate in the central cavity, it has been suggested that their productive folding might occur outside the cis cavity. These studies therefore indicate that the substrate recognition patterns of GroEL may be more diverse than initially thought.
Recent genome annotation studies of various bacteria have revealed that a few bacterial genomes possess multiple copies of groEL genes (2, 18, 30). The Mycobacterium tuberculosis genome bears two copies of groEL genes (groELs). One of these, groEL1, is arranged in an operon, with the cognate cochaperonin groES being the first gene, while the second copy, groEL2, exists separately on the genome (13). Recombinant mycobacterial GroELs were shown to possess biochemical features that deviated significantly from the trademark properties of E. coli GroEL. The most striking feature of M. tuberculosis GroELs, however, was their oligomeric state, where contrary to expectations, in vitro they did not form the canonical tetradecameric assembly when purified from E. coli. The proteins instead existed as lower oligomers (dimers) irrespective of the presence or absence of cofactors, such as the cognate GroES or ATP (40, 41). Furthermore, they displayed weak ATPase activities and GroES independence in preventing aggregation of the denatured polypeptides.
Evolutionary studies of M. tuberculosis groEL sequences have suggested rapid evolution of the groEL1 gene, yet without turning these into pseudogenes (21). The other hypothesis suggests that M. tuberculosis, being an organism that grows slowly, might require GroEL function that does not utilize ATP rapidly but, rather, with a slow turnover rate. Alternately, additional mechanisms might exist in M. tuberculosis which could mediate regulated oligomerization of M. tuberculosis chaperonins. Such regulation might help in the controlled utilization of ATP in nutrient-deprived M. tuberculosis, as observed for other chaperones, such as small heat shock proteins (23).
In the present study, we have exploited the unusual oligomeric status of the recombinant M. tuberculosis GroELs to study the significance of oligomer formation for GroEL's function as a molecular chaperone. Furthermore, we have explored the possibility of the existence of regulated oligomerization for native M. tuberculosis GroELs in their natural setting. We first show that M. tuberculosis groEL genes are not capable of complementing a conditional allele of E. coli groEL, namely, groEL44. The results of phenotypic and biochemical analyses of GroEL variants obtained by gene shuffling and domain swapping suggest that the impaired chaperoning ability of recombinant M. tuberculosis GroELs is a consequence of their inability to form higher-order oligomers in E. coli and that oligomerization is the prelude to the formation of an active GroEL chaperonin. Further, by immunochemical and mass spectrometric (MS) analysis of native mycobacterial GroELs, we show that M. tuberculosis GroEL1 exists in multiple oligomeric forms, viz., monomeric, dimeric, heptameric (single ring), and tetradecameric (double ring) forms, and that the switch between single-ring and double-ring variants is operated by phosphorylation on a serine residue. These observations suggest that the determinants of oligomerization for M. tuberculosis GroEL1 are distinct from those of its E. coli counterpart and that it does oligomerize in M. tuberculosis (its native environment), whereas it loses its oligomerization capability when expressed in E. coli. It could thus be possible that M. tuberculosis GroEL1 requires a certain native M. tuberculosis protein, probably a eukaryotic-like Ser-Thr protein kinase, to oligomerize properly, though the precise reason cannot be discerned by these observations.
The molecular biology procedures employed in this study were performed according to the standard protocols (47). All chemicals, enzymes, and antibiotics were purchased from Sigma, Inc. Antibodies IT3 and IT56 were procured via an NIH-NIAID TB Vaccine Testing and Research Materials contract awarded to Colorado State University, CO; anti-Cpn60.1Mtb (Cpn60.1 of M. tuberculosis) was a kind gift from A. R. M. Coates; and the phosphoaminoacyl-specific antibodies antiphosphoserine polyclonal antibody and phosphothreonine monoclonal antibody were purchased from Assay Designs and Rockland Immunochemicals, respectively. E. coli was cultured in standard LB broth supplemented as appropriate. E. coli SV2 is a derivative of the E. coli K-12 strain B178 (galE groESL+) which bears a temperature-sensitive allele of groEL, namely, groEL44. E. coli LG6 and E. coli MGM100 are derivatives of MG1655 wherein the chromosomal groESL operon is placed downstream from the lactose- or isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible Plac and l-arabinose-inducible PBAD promoters, respectively (26, 36, 54). These strains were kind gifts from Alan Fersht, Arthur Horwich, and Millicent Masters, respectively. Coliphages λcIB2 and T4GT7 were sourced from laboratory stocks. M. tuberculosis H37Rv was propagated on Middlebrook 7H11 medium and cultured in Middlebrook 7H9 liquid medium supplemented with oleic acid-albumin-dextrose-catalase and Tween 80, as appropriate (BD Biosciences). For the studies of the effects of temperature shift, a fraction of the M. tuberculosis culture at mid-log phase was subjected to heat shock for 3 to 6 h at 42°C. Lysates from the resulting cultures were prepared according to the standard protocols (44). The plasmids and oligonucleotide primers used in this study are listed in Tables S1 and S2, respectively, in the supplemental material. The open reading frames (ORFs) encoding M. tuberculosis GroES, GroEL1, and GroEL2 and E. coli GroEL/S were cloned under the arabinose-inducible PBAD promoter (22). M. tuberculosis groEL1 was amplified from an M. tuberculosis genomic DNA library, a kind gift from Stewart Cole (13), using primers SCM7F and SCM5R and cloned into the NdeI and SmaI sites of pBAD24N to obtain plasmid pSCM1604. M. tuberculosis groEL2 was sourced from the plasmid pSCM1000 upon digestion with restriction endonucleases XbaI and HindIII, and the resulting fragment was cloned into pBAD18 that was digested with the same enzymes to generate plasmid pSCM1605 in which the groEL2 ORF is preceded by an optimally placed vector-borne ribosome binding site. E. coli groEL was amplified from the genomic DNA of strain MG1655 using primers SCM4F and SCM3R and cloned into the NcoI and HindIII sites of pBAD24 to generate plasmid pSCM1608. Furthermore, two additional plasmids, pSCM1602 and pSCM1603, were constructed in which groEL1 and groEL2, respectively, were placed into pBAD24 downstream from and in an operonic arrangement with M. tuberculosis groES.
Stationary-phase cultures of E. coli strains SV2 and LG6 containing plasmids pSCM1601, pSCM1602, pSCM1603, and pBAD24 were serially diluted. Five microliters of each serially diluted culture of E. coli SV2 were spotted onto LB agar plates supplemented with either 0.2% d-glucose (to repress the PBAD promoter) or 0.2% l-arabinose (to induce the expression of cloned groEL genes). The cultures of E. coli LG6 were spotted onto LB agar supplemented with 0.2% d-glucose (to repress the PBAD promoter), 0.2% l-arabinose (to induce the expression of cloned groEL genes), or 1 mM IPTG (to induce the chromosomal copy of the E. coli groESL operon). The plates bearing SV2 cultures were incubated at permissive (30°C) and restrictive (42°C) temperatures, and those bearing LG6 cultures were incubated at 30°C.
Since an ORF encoding an active version of GroEL is expected to complement the temperature-sensitive phenotype associated with the groEL44 allele, it allows facile genetic selection for the isolation of M. tuberculosis GroEL variants with increased chaperone activity. Toward this end, amplified M. tuberculosis groEL1 and groEL2 ORFs were used for multigene DNA shuffling, following the reported method (51, 58). Briefly, PCR products were subjected to limited DNase I digestion and fragments of 50 to 150 base pairs were recovered from agarose gels, which were further used as a template for a primerless assembly PCR. The assembled PCR product was used for the next round of PCR, with E. coli groEL-specific primers, according to the method of Zhao and Arnold (66). For the final round of PCR, E. coli groEL-specific primers were used, taking into consideration the presence of the 13-residue repeat (GGM)4M at the carboxyl terminus, alteration in the size and chemical nature of which is thought to affect GroEL's function in folding a few of the substrate polypeptides (16, 52). The product obtained after the final PCR was digested and cloned under the PBAD promoter, and the library of plasmids was recovered in E. coli strain DH5α. A pooled plasmid preparation from this library was used to transform E. coli strain SV2, and direct selection for recovery of plasmids encoding active versions of groEL was employed by plating the transformation mixture onto LB-ampicillin-l-arabinose plates at the restrictive temperature of 42°C. Nine plasmids (pSCM1622 to pSCM1637) were chosen for further study (see Table S1 in the supplemental material). A detailed protocol followed for gene shuffling is presented in the supplemental material.
Two chimeric M. tuberculosis-E. coli hybrid ORFs were generated, namely, groELMEF and groELMER, by employing overlap extension PCR (60) wherein the equatorial domains of M. tuberculosis GroEL1 and E. coli GroEL were mutually exchanged (see Fig. Fig.4A).4A). The resulting plasmids, bearing groELMEF and groELMER under the PBAD promoter, were designated pSCM1609 and pSCM1611, respectively. Phenotypic analysis of the cloned genes was performed in E. coli strains SV2 and MGM100. Detailed methods are presented in the supplemental material.
Cultures of E. coli SV2 bearing plasmids pSCM1604, pSCM1605, pSCM1608, pSCM1609, pSCM1611, and pBAD24 obtained after culturing in 0.4% d-maltose with the appropriate antibiotic were overlaid in soft agar on the surface of LB plates containing either 0.2% d-glucose or 0.2% l-arabinose. The plaque-forming abilities of bacteriophages λc1B2 and T4GT7 were assessed by spotting 5 μl each of 100-fold diluted suspensions of bacteriophage stocks. The plates were incubated at 30°C.
The ATPase activity of the purified GroELs was quantified with a colorimetric assay performed as described previously (25, 41, 57). Briefly, 50 μl of the reaction buffer containing 50 mM Tris-HCl (pH 8.0), 10 mM KCl, 10 mM MgCl2, and 2.5 μM of each GroEL was incubated with 1 mM ATP at 37°C for 20 min. Enzymatic reactions were terminated by the addition of 200 μl of an acidic solution of malachite green. The amount of inorganic phosphate liberated was measured at 655 nm using a Nanodrop ND-1000. Control reactions without ATP and GroEL were performed. A standard curve with monobasic potassium phosphate was generated concurrently with each experiment.
Pig heart citrate synthase aggregation was performed as reported previously (5). Briefly, 0.15 μg/ml citrate synthase was incubated at 43°C in the presence or absence of equimolar oligomer ratios of different GroEL variants in 40 mM HEPES-KOH buffer (pH 7.5). The ability of the said chaperones to prevent the aggregation of citrate synthase was monitored for 20 min on a Hitachi F-4000 spectrofluorimeter with emission and excitation wavelengths set at 465 nm and corresponding band passes set at 3.0 nm. The temperature of the sample was maintained with a Julabo circulating water bath and was monitored by using a Physitemp type T microcouple.
The protocol for denaturing, refolding, and assaying the activity of pig heart mitochondrial citrate synthase was followed essentially as reported previously (41, 50). Briefly, citrate synthase at 15 μM was denatured in denaturation buffer containing 100 mM Tris-HCl (pH 8.0), 6 M guanidine-HCl, and 1 mM dithiothreitol for 30 min at 25°C. The denatured citrate synthase was diluted 100-fold into the refolding buffer containing 20 mM potassium phosphate buffer (pH 7.5), 10 mM MgCl2, 2 mM ATP, 1 mM oxaloacetic acid, and 1 μM GroEL variant in the presence or absence of 2 μM E. coli GroES, and the activity of the recovered citrate synthase was assayed after 60 min. The enzyme activity was assayed as the consumption of acetyl-coenzyme A (CoA) revealed by the decrease in absorption at 233 nm for 90 s. Refolding reactions in the absence of the chaperonins and with the 1.5 nM native citrate synthase were set as controls.
Cell lysates were prepared from normally grown (at 37°C) and heat-shocked (at 42°C) M. tuberculosis cultures that were normalized for optical density. Two to 3 mg total protein of the cell lysates from both the cultures were resolved by gel filtration chromatography on a Superdex S200 16/60 (GE Biosciences, Inc.), and the resulting fractions were loaded on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels and probed for the presence of M. tuberculosis GroEL1 by immunoblotting, using anti-Cpn60.1Mtb (rabbit polyclonal) antibody at a 1:10,000 dilution. Fractions bearing individual oligomeric forms of M. tuberculosis GroEL1, as determined by comparison to the molecular weight standards that were also resolved by gel filtration on Superdex S200 16/60, were pooled, and an equal proportion of each pool was resolved on 10% SDS-PAGE gels and probed by immunoblotting using anti-Cpn60.1Mtb for quantitative determination of the level of each oligomeric form.
Using antibodies specific to M. tuberculosis GroEL1 and GroEL2, these proteins were immunoprecipitated from M. tuberculosis cell lysates using standard protocols (3). Briefly, 50 to 100 μg total protein of cell lysates from the normally grown and heat-shocked cultures were mixed separately with antibodies anti-Cpn60.1Mtb and IT56 at equal titers and were incubated at 4°C overnight with constant mixing. Forty microliters of protein A-coupled Sepharose beads (GE Biosciences) equilibrated in buffer P (50 mM Tris [pH 8.0], 100 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 0.5% NP-40) was added to each tube and was incubated for 3 h with constant mixing. The protein A-coupled Sepharose beads, which were bound by the antibody-GroEL complex, were collected by centrifugation, and the pellet was washed three times in buffer P to remove the unbound cell proteins. Equal proportions of the protein-bound beads were resolved on 10% SDS-PAGE gels to estimate the levels of GroELs in the M. tuberculosis cell lysates. The immunoprecipitated samples were probed further by immunoblotting for the presence of phosphorylation using antibodies specific to phosphoseryl and phosphothreonyl residues at 1:1,000 dilutions. Once it was confirmed that the phosphorylation was on the seryl residue, equal amounts of individual oligomeric forms of GroEL1 were probed for the presence of phosphorylation using antiphosphoserine antibody as described above. To ensure equal loading, the transferred proteins were visualized by staining with Ponsue red, followed by immunoblotting with the antiphosphoseryl antibody.
Different oligomeric forms of recombinant M. tuberculosis GroEL1 were resolved on 10% SDS-PAGE gels, and the Coomassie-stained bands corresponding to GroEL1 were sliced out as polyacrylamide gel plugs. These were subjected to in-gel digestion using trypsin gold (Promega), followed by a matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) study for peptide mass fingerprinting and protein sequencing using an Ultraflex MALDI-tandem TOF (TOF-TOF) instrument (Bruker Daltonics). The peptides which hosted a seryl residue were subjected to tandem MS (MS-MS) using an ion trap mass spectrometer (Agilent) to determine the presence of phosphorylation. The MALDI-TOF and MS-MS data were analyzed by using the MASCOT search engine or Spectrum Mill software. In these experiments, purified recombinant GroEL1 was included as a control for peptide identification. The MS studies were carried out in collaboration with The Centre for Genomic Applications (TCGA), New Delhi, India.
Owing to their unusual behavior in vitro, we tested whether the two M. tuberculosis groEL ORFs could complement the loss of GroEL function in E. coli. This was tested in two E. coli strains, SV2, which harbors a temperature-sensitive groEL44 allele, and LG6, in which the expression of the chromosomal groESL operon is under the control of the Plac promoter. The expression of M. tuberculosis groEL1 and groEL2 from the PBAD promoter in SV2 did not lead to rescue of its temperature-sensitive phenotype even when the M. tuberculosis groELs were coexpressed with their cognate groES (Fig. (Fig.1A).1A). For these studies, a plasmid bearing E. coli groESL under the control of the PBAD promoter, which complemented the temperature-sensitive phenotype of the groEL44 allele, was also employed. Immunoblots with GroEL1- and GroEL2-specific antibodies established that the proteins were expressed and, hence, that the lack of complementation was not due to the lack of expression (Fig. (Fig.1B1B).
Likewise, M. tuberculosis GroELs could not support the growth of E. coli LG6 (Fig. (Fig.1C).1C). The inability of M. tuberculosis groELs to complement the loss of GroEL function in E. coli is consistent with the weakened in vitro chaperonin activity shown by our earlier studies of M. tuberculosis GroELs (41).
In order to investigate the molecular features which lead to the differences in E. coli and M. tuberculosis GroELs, we generated a pool of groEL variants via DNA shuffling, starting with ORFs encoding M. tuberculosis GroEL1 and GroEL2 as template DNA. ORFs capable of encoding active versions of GroEL were selected in E. coli SV2. The mutants thus generated were chimeras of M. tuberculosis groEL1 and groEL2 and E. coli groEL. Sequence analysis of the clones and the comparison of G+C contents show that the gene-shuffled groEL variants are in fact derived from M. tuberculosis groELs (Table (Table1).1). A quantitative phenotypic analysis of nine of the gene-shuffled groEL variants is shown in Fig. Fig.2.2. Multiple sequence alignment of the polypeptides encoded by the said groEL alleles with the E. coli and M. tuberculosis GroEL sequences interestingly showed that their putative apical domains are subject to considerable sequence variation, with some variants bearing fairly large deletions and insertions, whereas their putative equatorial domains are conserved among the variants and are analogous to E. coli GroEL (Fig. (Fig.3;3; also see the supplemental material). This feature is most noticeable in the polypeptide encoded by the variant groELSp24, which bears a deletion of about 90 amino acids in its putative apical domain but has an equatorial domain homologous to that of E. coli GroEL and is able to complement the temperature-sensitive phenotype of the groEL44 allele (see the supplemental material). These studies lend additional support to the supposition that the ability to oligomerize, due to the presence of an “E. coli GroEL-like” equatorial domain, correlates with biologically relevant GroEL activity and suggest that the apical domain can tolerate substantial variation. However, variants GroELSp22 and GroELSp25 displayed weak complementation, despite possessing homologous equatorial domains. A future comprehensive study including site-directed mutagenesis might be able to examine such behavior.
Having obtained evidence that the selection pressure on the equatorial domain is relatively more stringent than that on the apical domain but that, nonetheless, the equatorial domains of M. tuberculosis GroEL orthologs have diverged to a large extent from that of E. coli GroEL, we studied the effect of mutual exchange of equatorial domains between M. tuberculosis GroEL1 and E. coli GroEL. For these studies, two ORFs, groELMEF and groELMER, were generated and placed under the expression control of the PBAD promoter (see Materials and Methods and Fig. Fig.4A).4A). Complementation studies with groELMEF and groELMER showed that the exchange of the equatorial domain in M. tuberculosis GroEL1 with that of E. coli GroEL, specified by the groELMEF ORF, turns it into a functional GroEL in vivo. groELMEF was able to complement the groEL44 allele at the restrictive temperature of 42°C to an extent similar to that exhibited by E. coli groEL, whereas at 45°C, groELMEF displayed weak complementation. In contrast, groELMER, in which the DNA encoding the equatorial domain of E. coli GroEL is replaced with that encoding the corresponding domain from M. tuberculosis GroEL1, was able to exhibit only very weak complementation at 42°C (Fig. (Fig.4B).4B). Curing the plasmids from these transformants led to the loss of complementation, confirming that the phenotype was indeed vector borne (data not shown).
In E. coli, the development of bacteriophages like lambda (λ) and T4 requires a functional GroEL/S system (20). To gauge the extent of restoration of GroEL function present in GroELMEF, its ability to support phage morphogenesis was studied. The expression of groELMEF, like that of E. coli groEL, was able to support the growth of both phage λ and T4 in the groEL44 mutant strain SV2, whereas the expression of M. tuberculosis groEL1, groEL2, and groELMER did not promote phage development (Fig. (Fig.4C).4C). These studies thus lend credence to the notion that the reason for the observed functional difference between the E. coli and M. tuberculosis GroELs must be a consequence of differences between the equatorial domains of the two molecules.
To avoid potential interference of the GroEL polypeptide encoded by the resident groEL44 allele in the complementation exhibited by groELMEF, we studied the effect of groELMEF expression under conditions of depletion of endogenous GroEL. Toward this end, plasmids bearing tac promoter (Ptac)-driven expression of groELMEF and one in which groELMEF was cloned in an operonic arrangement with E. coli groES were generated. Similar plasmids capable of expressing E. coli groES and E. coli groEL individually and the E. coli groESL operon were included as controls. The said plasmids were transformed into strain MGM100, which does not form colonies in the absence of arabinose unless a functional copy of groES-groEL is expressed (See Materials and Methods). Then, the ability of groELMEF to support the growth of MGM100 in the absence of exogenous l-arabinose supplementation was assessed. Of the various plasmids tested, only plasmids coexpressing E. coli groES with either groELMEF or E. coli groEL allowed colony formation in the absence of l-arabinose (Fig. (Fig.5).5). These results clearly show that GroELMEF can substitute for E. coli GroEL in vivo and requires E. coli GroES for its activity, which is not influenced by the resident GroEL of strain SV2.
Having shown that GroELMEF is functional in vivo, we wished to test whether the ability of GroELMEF to complement the defect in the groEL44 allele is a consequence of it existing in higher-order oligomeric form and thus being able to form a cavity for encapsulation, features that are lacking in the parental M. tuberculosis GroEL1 (40). Since the polypeptides encoded by the gene-shuffled variants of groEL, groELSp24, and groELSp32 bear considerable variation (Fig. (Fig.3;3; also see Fig. S1 in the supplemental material), similar attributes of the encoded polypeptides were studied. Gel filtration studies with Sephacryl S300 showed that GroELMEF was capable of existing in a higher oligomeric state, similar to that seen for E. coli GroEL, whereas GroELMER displayed a lower oligomeric character (Fig. (Fig.6),6), a property reminiscent of that seen for M. tuberculosis GroEL1 (59). One noticeable aspect of the oligomeric properties of GroELMEF is that it existed in equilibrium between higher and lower oligomeric states, with the higher oligomeric state being the predominant species, which presumably explains its somewhat weakened ability to substitute for E. coli GroEL in vivo at 45°C (Fig. (Fig.4B).4B). Since the higher oligomeric state of GroELMEF displays gel filtration characteristics similar to those of E. coli GroEL, it is reasonable to presume that the said state corresponds to a tetradecameric assembly. We have also examined the oligomeric states of GroELSp24 and GroELSp32, variants that bear deletions in the putative intermediate apical domain boundary. Both the variants displayed tendencies to exist in a higher oligomeric state, and their gel filtration profiles were consistent with their predicted molecular weights.
GroEL-assisted folding of substrate proteins typically involves (i) binding of the polypeptides by virtue of exposed hydrophobic interactions; (ii) sequestration of polypeptides into the cavity, thereby preventing irreversible aggregation; (iii) ATP hydrolysis; and (iv) GroES-dependent refolding of substrate polypeptides. Since the gel filtration results revealed that GroELMEF, GroELSp24, and GroELSp32 are oligomeric, we studied their biochemical features in hydrolyzing ATP and refolding the model substrate, citrate synthase. The ability of GroELMEF to exist in a higher oligomeric state correlated with its ability, in comparison to that displayed by GroELMER, to refold chemically denatured substrate protein (Fig. 7A and B). Similarly, GroELSp24 and GroELSp32, with their ability to oligomerize, exhibited canonical chaperonin properties (Fig. 7A and B).
The well-known chaperonin properties of GroEL, as mentioned above in (i) and (ii), require two important properties, i.e., binding to exposed hydrophobic surfaces of substrate polypeptides and encapsulation in the cavity to promote folding. The former is typically studied by measuring the prevention of aggregation of substrate polypeptides, while the latter can be studied by monitoring the recovery of folded substrate proteins. Remarkably, GroELMEF was poor at preventing aggregation of the substrate protein, citrate synthase, at elevated temperatures for more than 5 min. On the other hand, GroELMER prevented citrate synthase aggregation for at least 20 min (Fig. (Fig.7C).7C). Thus, GroELMEF, despite not being able to efficiently prevent substrate aggregation, is able to fold substrates due to its competence in encapsulation. On the other hand, GroELMER, despite being able to prevent aggregation, is not able to promote refolding of substrates due to its inability to encapsulate. Similarly, GroELSp32 was able to prevent the aggregation of citrate synthase, whereas GroELSp24 was not (Fig. (Fig.7C).7C). The results for GroELMER suggest that lower oligomeric GroELs possess substrate binding activity but are inefficient in promoting refolding.
Having established that the impaired oligomerization is the basis for the inactivity of recombinant mycobacterial GroELs, we tested the nature of the oligomerization of M. tuberculosis GroELs in their natural settings. For these studies, we considered three aspects concerning the biology of M. tuberculosis and GroEL. First, multiple sequence alignment of bacterial GroEL homologues showed that several positions in the M. tuberculosis GroELs' equatorial domains were altered from different residues to serine/threonine, indicating potential phosphorylation sites (40, 41). Second, the M. tuberculosis genome encodes 11 eukaryotic-like serine/threonine kinases (ELKs), and emerging evidence that ELKs play a role in different cellular processes, including stress adaptation, across bacterial genera tempted us to speculate that there is a role for ELKs in GroEL oligomerization (38, 61, 65). Third, some heat shock proteins, including mitochondrial and chloroplast Hsp60 homologues, exhibit heat-induced or concentration-dependent oligomerization. The mitochondrial Hsp60 homologue, although predominantly found as a single ring entity, is known to display a concentration-dependent tetradecamer formation (14, 34). Hence, we attempted to explore whether M. tuberculosis GroEL1 exhibits regulated oligomerization and, if so, to determine the source of its regulation. The resolution of cell lysates on native PAGE gels, followed by probing with M. tuberculosis GroEL1-specific antibody, revealed the existence of M. tuberculosis GroEL1 in four different oligomeric forms, i.e., monomeric, dimeric, heptameric (single ring), and tetradecameric (double ring) forms (data not shown).
In order to separate these forms, M. tuberculosis cell lysates were resolved by molecular exclusion chromatography and an equal proportion of each fraction was probed with GroEL1-specific antibody to quantitatively determine the presence of each oligomeric form at 37°C and under heat shock conditions. It was observed that the tetradecameric form existed at both 37°C and 42°C. The presence of the tetradecameric form in cultures grown at 37°C indicates that the switch between heptameric and tetradecameric forms is not solely temperature mediated. The dimeric form was also detected under both conditions, and the monomeric form, although present, is at extremely low levels compared to the levels of the other, higher-order forms.
Our earlier comparative analysis with E. coli GroEL had indicated that at different positions, there are systematic mutations of glutamates to serines or threonines in the sequence of M. tuberculosis GroEL1 (41). This suggested a possibility that the serine or threonine residues are the potential sites of phosphorylation. Therefore, investigating the presence and nature of phosphorylation was attempted by immunoprecipitation of GroEL1 and GroEL2 from M. tuberculosis cell lysates, followed by immunoblotting with antiphosphoseryl and antiphosphothreonyl antibodies. These immunochemical analyses revealed that GroEL1 is phosphorylated on serine residues but not on any threonine residue (Fig. (Fig.8B).8B). Having established that GroEL undergoes phosphorylation, we further investigated the phosphorylation status of the various oligomeric forms of GroEL1. Strikingly, we observed that only the tetradecameric form was phosphorylated; the heptameric and dimeric forms were not phosphorylated (Fig. (Fig.8C).8C). The band in the lane corresponding to the heptameric form, as shown in Fig. Fig.8C,8C, was seen to be migrating faster than M. tuberculosis GroEL1 when compared with the Ponsue red-stained bands on nitrocellulose membranes. Furthermore, MS analyses of this band revealed it to be the 53.5-kDa glutamine synthetase A1 of M. tuberculosis. These observations lead us to believe that a phosphorylation switch might be mediating the conversion from the heptameric single-ring form to the tetradecameric double-ring form.
MS analysis of M. tuberculosis GroEL1 by MALDI-TOF followed by MS-MS fragmentation of the resulting peptides confirmed that serine-393 is phosphorylated in the tetradecameric form (Fig. (Fig.9A).9A). However, we do not rule out the possibility that a few more serines could be phosphorylated, and if so, these might be located in the critical positions governing the inter-ring contacts. Our results thus reveal that there exist multiple oligomeric forms of GroEL1 molecules in M. tuberculosis and that the conversion from the heptameric to the tetradecameric form is mediated by phosphorylation of Ser residues.
The critical features of the chaperonin function reside in the ATP-driven cycles of binding, encapsulation, and controlled release of substrate polypeptides that lead to productive folding (43, 45, 48). Detailed studies have revealed mechanistic and physiological characteristics of the isologous ring form of the GroEL/S machine (11, 42, 45, 53, 54). According to the current understanding, the following properties of GroEL are significant in controlling its activity: (i) oligomerization mediated by the equatorial domain, resulting in the formation of the folding chamber and the encapsulation of substrate polypeptides (48, 59); (ii) recognition of substrate polypeptides mediated by the apical domain (6, 49); (iii) conformational changes between the said two domains driven by ATP and GroES binding/release (32, 63, 64); and (iv) ATP hydrolysis (19, 55). Impairment of any of these properties significantly alters the functioning of GroEL (37, 40).
Some prokaryotic species bear multiple genes encoding GroEL, one usually cotranscribed with the cognate groES orthologue (2, 18, 30). Since sequence conservation among GroELs from different species is an indication that the mechanism of GroEL is universally conserved, paralogous copies of GroEL in these organisms might provide redundancy of chaperonin function. However, biochemical and biophysical characterization of both of the recombinant M. tuberculosis GroELs showed that they failed to oligomerize in vitro despite possessing a high sequence homology with E. coli GroEL (40), suggesting a stringent requirement of the host environment for oligomerization. Recent studies have shown that in Mycobacterium smegmatis, GroEL1, a paralog of M. tuberculosis GroEL1, is involved in biofilm formation, whereas GroEL2 is thought to provide housekeeping chaperonin function (39). The observation that GroEL1 can physically associate with KasA, a component of the mycolic acid synthesis pathway, is thought to support the notion that GroEL1 may play the role of chaperonin in the process of biofilm formation. Furthermore, M. tuberculosis GroEL1 is also known to induce host inflammatory responses (28).
Since oligomerization is compromised in recombinant M. tuberculosis GroELs, studying the mutations accumulated in positions responsible for intersubunit interactions seems reasonable. Sequence analysis of the shuffled-gene products suggested that the apical domain is capable of absorbing considerable variation in its amino acid sequence. The large deletions observed in the shuffled-gene products occur mostly in the apical domain (Fig. (Fig.3).3). The tolerance of variations by the apical domain can be reconciled with the observation that GroEL can recognize a wide repertoire of substrates with almost no sequence specificity. Thus, the said domain may bear a latent plasticity. The domain-swapping experiments, by generating E. coli and M. tuberculosis GroEL hybrids, showed that the lack of in vitro activity of M. tuberculosis GroELs was indeed due to the loss of oligomerization (41) (Fig. (Fig.6)6) and that the loss of oligomerization can be exclusively attributed to the equatorial domain of M. tuberculosis GroEL, which when replaced with that from E. coli GroEL, might circumvent the requirement of its native environment for oligomerization.
Interestingly, the results of the ATPase activity assays and refolding studies with chemically denatured citrate synthase are in concurrence with those of the in vivo studies showing that GroELMEF is able to exhibit characteristics of a chaperonin (Fig. 7A and B). However, GroELMEF is poor at protecting the substrate from aggregation beyond 5 min, whereas GroELMER acts similarly to E. coli GroEL in protecting the substrate from aggregation for at least 15 min (Fig. (Fig.7C).7C). Furthermore, GroELSp24 displays characteristics similar to those of E. coli GroEL. These observations therefore suggest that the variations in apical domain can, to some extent, be absorbed without impairing chaperonin function as long as the molecule retains its ability to encapsulate the substrate proteins. It is not surprising to find that GroELMER possesses substrate binding capacity, since it bears the apical domain of E. coli, yet is unable to refold substrate polypeptides due to its inability to form tetradecamers. It might be possible that GroELMER, despite its inability to oligomerize, exhibits very weak complementation in vivo due to this property (Fig. (Fig.4B).4B). Furthermore, the behavior of GroELSp32 can be rationalized on the basis that the overt variation in GroELSp32, a segmental deletion in its putative apical domain, could be neutral, leading to substantial polypeptide binding, but could contribute to its weakened chaperonin activity. These results therefore confirm that mere substrate recognition is not sufficient for the chaperonin function. On the contrary, oligomerization is an important attribute of GroEL.
With the establishment of the fact that oligomerization is the principal attribute of GroEL function and based on our earlier bioinformatic analysis of M. tuberculosis groEL genes, which showed that these genes have not accumulated mutations so as to become pseudogenes during the process of evolution, a need for GroEL's function in vivo in M. tuberculosis is indicated (21). We therefore attempted to explore the oligomeric status of the M. tuberculosis GroELs under native conditions. Surprisingly, M. tuberculosis GroEL1 exhibited multiple oligomeric forms, which is the first-ever observation of this for a bacterial chaperone. GroEL homologues from mammalian mitochondria and from chloroplasts have been shown to exhibit multiple oligomeric forms where the conversion from single-ring form to double-ring form is concentration and GroES dependent (61, 65). Attempts to make a single-ring version of E. coli GroEL by site-directed mutagenesis of four critical residues at the inter-ring surface had yielded a mutant, SR1, that was compromised in GroES release (62). However, the conversion of a temperature-sensitive GroEL mutant, GroEL44, into single-ring form, SR44, resulted in an active chaperonin (8). Studies of the assembly of bacterial chaperonin were attempted with purified E. coli GroEL involving cofactors and denaturing agents, such as Mg2+, GroES, urea, and guanidium chloride, which showed the existence of bacterial chaperonin in multiple oligomeric forms under various denaturing conditions tested (33). In addition, several small heat shock proteins and Hsp90 were shown to exhibit temperature-regulated oligomerization. Hence, we set out to explore the possibility of regulated oligomerization of M. tuberculosis GroEL1 and the source of regulation, if any. Immunochemical studies, followed by MS analysis of M. tuberculosis GroEL1 and GroEL2 from the M. tuberculosis cell lysates, confirmed that GroEL1 is phosphorylated on serine-393.
In summary, this work shows the requirement of oligomerization for the function of GroEL. Although M. tuberculosis GroEL1 is capable of oligomerization in its native environment, it cannot do the same in E. coli, but the exchange of their equatorial domains renders M. tuberculosis GroEL active, suggesting specific oligomerization capability of M. tuberculosis GroEL's equatorial domain only in the M. tuberculosis cellular milieu. Furthermore, E. coli GroEL can be rendered nonfunctional and indistinguishable from M. tuberculosis GroEL1 by replacing its equatorial domain with the one present in GroEL1. Since GroELMEF can substitute for the lack of GroEL function in E. coli, it suggests that despite the heterologous apical domain of M. tuberculosis GroEL1 borne on GroELMEF, the chimeric protein (and therefore M. tuberculosis GroEL1) may recognize the same cellular substrates in vivo as does its E. coli counterpart. Despite the large functional difference between M. tuberculosis GroEL1 and E. coli GroEL, a noteworthy aspect is that GroEL1 retains the highly conserved ATP binding pocket that bears amino acid residues identical to those in E. coli GroEL. This tempts us to speculate that oligomerization may be a prerequisite for GroEL ATPase activity and, therefore, the ability to assist the refolding of substrate polypeptides. These studies therefore reveal that the basis of the reduced activity of recombinant M. tuberculosis GroEL1 in E. coli was its impaired oligomerization. We also show, for the first time, that mycobacterial GroEL1 exists in different oligomeric forms and that the switch between the single-ring (heptamer) and double-ring (tetradecameric) GroEL forms is mediated by phosphorylation. Thus, we propose that the naturally synthesized GroEL exists in equilibrium between a dimer and a heptamer and that heptamer-to-tetradecamer conversion is mediated by phosphorylation (Fig. (Fig.9B).9B). Since a similar strategy of chaperonin oligomerization operates in mitochondrial GroEL, this study, in addition to probing M. tuberculosis GroEL biology, might also shed light on the evolution of mitochondria and chloroplasts, which are thought to be the bacterial symbionts of the eukaryotes (1).
We thank Arthur Horwich, Jon Beckwith, Anthony R. M. Coates, Stewart Cole, Alan Fersht, Costa Georgopoulos, and Millicent Masters, as well as NBRP, E. coli strain (NIG, Japan), for providing strains, plasmids and antibodies. We are grateful for materials supplied under NIH-NIAID contract no. HHSN266200400091C, Tuberculosis Vaccine Testing and Research Materials, awarded to Colorado State University, Fort Collins, CO. We thank Arthur Horwich, Peter Lund, Ranjan Sen, Nasreen Z. Ehtesham, T. Ramakrishna Murti, and Vijaya Gopal for helpful discussions and support in spectroscopic studies and The Centre for Genomic Applications, New Delhi, for assistance with the MS studies. We thank Kshama Goyal for help in sequence analysis and the generation of Fig. Fig.33.
This work was supported by grants from the Wellcome Trust, United Kingdom, and the Department of Biotechnology, India. C.M.S.K. and G.K. were supported by fellowships from the Council for Scientific and Industrial Research, India, and A.A.S. was supported by a fellowship from the Department of Science and Technology, India. S.C.M. is a Wellcome Trust International Senior Research Fellow. The authors declare no financial conflict of interest.
Author contributions. C.M.S.K. conducted all the molecular biological, biochemical and genetic experiments. C.M.S.K. and A.A.S. designed the genetic experiments with E. coli, and C.M.S.K. and C.V.S. performed the gene-shuffling studies. C.M.S.K., G.K., A.K.T., and S.C.M. designed the experiments with M. tuberculosis, and C.M.S.K. and G.K. conducted these. Overall supervision and design was by S.C.M., A.K.T., and A.A.S.
Published ahead of print on 28 August 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.