Chemical modification of the Ca(2+)-ATPase of rabbit skeletal muscle sarcoplasmic reticulum: identification of sites labeled with aryl isothiocyanates and thiol-directed conformational probes.

The Ca(2+)-ATPase protein of rabbit skeletal muscle sarcoplasmic reticulum is a single polypeptide chain of 1001 amino-acid residues. Among these residues are 24 Cys, 9 of which have previously been shown to be accessible to one or more thiol-specific reagents. Many studies on the structure and function of this Ca(2+)-ATPase have made use of sulfhydryl-directed, conformationally-sensitive probes, but the labeling sites for these probes have been directly identified in only a few cases, causing uncertainty in the interpretation of results. In the present work, we have investigated the Ca(2+)-ATPase labeling sites for three thiol-directed spectroscopic probes: fluorescein 5'-maleimide (Fmal), 4-dimethylaminophenyl-azo phenyl-4'-maleimide (DABmal), and 4-dimethylaminophenylazophenyl-4'-iodoacetamide (DABIA). Labeled Ca(2+)-ATPase was digested exhaustively with trypsin, and labeled peptides were purified and sequenced in order to identify the labeled Cys residues. Our results do not support the widely held assumptions that Cys-344 and Cys-364 are the most reactive residues with maleimide-based reagents, while Cys-670 and Cys-674 react most rapidly with iodoacetamide derivatives. We found instead that Fmal reacted most rapidly with Cys-471, followed by Cys-364, and more slowly with Cys-498, -525, -614 and -636. DABmal reacted most rapidly with Cys-364, followed by Cys-614, and more slowly with Cys-471, -498, -636 and -670. Cys-344 was not labeled by either Fmal or DABmal. DABIA reacted with the same six Cys residues, including Cys-670, as were labeled with DABmal, but in much lower yield. There was no evidence for labeling of Cys-674 with DABIA. The high reactivity of Fmal, but not the more hydrophobic DABmal, with Cys-471 is of interest because of previous studies suggesting that the accessibility of Cys-471 is influenced by ATP and that fluorescein derivatives bind to a hydrophobic pocket in the ATP binding site. Another derivative, fluorescein-5'-isothiocyanate (FITC), is thought to label the catalytic site of the Ca(2+)-ATPase and has been widely used as a conformational probe in structure-function studies on this and related proteins. We reinvestigated the chemical modification of the Ca(2+)-ATPase by FITC and 4-dimethyl-aminophenyl-4'-isothiocyanate (DABITC). Incorporation of stoichiometric amounts of FITC resulted in a nearly complete loss of ATPase activity. Labeling and inactivation of the Ca(2+)-ATPase by FITC did not occur in the presence of ATP. DABITC was less reactive than FITC, and did not inactivate the Ca(2+)-ATPase to any significant extent.(ABSTRACT TRUNCATED AT 400 WORDS)

The Rz1 gene product of bacteriophage lambda is a lipoprotein localized in the outer membrane of Escherichia coli.

The Rz1 gene of bacteriophage lambda is located within the Rz1 lysis gene. It codes for the 6.5-kDa prolipoprotein (Rz1) which undergoes N-terminal signal sequence cleavage and post-translational lipid modification of the N-terminal Cys of the mature protein. Globomycin, the antibiotic which inhibits bacterial signal peptidase II, specific for prolipoproteins containing diacylglyceryl cysteine [Hayashi and Wu, J. Bioenerg. Biomembr. 22 (1990) 451-471] inhibits the N-terminal sequence cleavage of the Rz1 precursor. The mature protein is rich in Pro, which constitutes 25% of its amino acids (aa). Using a computer-predicted, synthetic, 15-aa antigenic determinant of Rz1 polyclonal anti-Rz[46-60] antibodies, were obtained, and employed to localize Rz1 in bacterial fractions. In induced Escherichia coli lambda lysogens Rz1 was found almost exclusively in the outer membrane (OM). In a strain overproducing Rz1 from the pSB54 plasmid, it was distributed in all the fractions, OM, fraction A and inner membrane (IM). Expression of Rz1 from the pSB54 caused enlargement of fraction A, corresponding to the adhesion sites of OM and IM. Such an enlargement was previously observed in induced lambda lysogens, shortly before the onset of lysis.

Site-directed mutagenesis of the HtrA (DegP) serine protease, whose proteolytic activity is indispensable for Escherichia coli survival at elevated temperatures.

The HtrA(DegP) 48-kDa serine protease of Escherichia coli is indispensable for bacterial survival at elevated temperatures. It contains the amino-acid sequence Gly208AnsSerGlyGlyAlaLeu, which is similar to the consensus sequence GlyAspSerGlyGlyProLys surrounding the active Ser residue of trypsin-like proteases. Mutational alteration of Ser210 eliminated proteolytic activity of HtrA. An identical effect was observed when His105 was mutated. The mutated HtrA were unable to suppress thermosensitivity of the htrA bacteria. These results suggest that Ser210 and His105 may be important elements of the catalytic domain and indicate that the proteolytic activity of HtrA is essential for the survival of cells at elevated temperatures.

Identification of the site phosphorylated by casein kinase II in smooth muscle caldesmon.

Phosphorylation of avian gizzard caldesmon by casein kinase II was investigated. The enzyme incorporates about 1 mol of phosphate per mol of caldesmon. All sites of phosphorylation are located in short chymotryptic peptides with Mr 25-27 kDa or in the short N-terminal peptide formed after cleavage of chicken gizzard caldesmon at Cys153. The primary structure of the tryptic peptide containing the main site of duck gizzard caldesmon phosphorylation is S-E-V-N-A-Q-N-X-V-A-E-D-E-T-K, where X is an unidentified residue, presumed to be phosphoserine. Thus, Ser73 is the main site phosphorylated by casein kinase II in avian gizzard caldesmon.

IbpA and IbpB, the new heat-shock proteins, bind to endogenous Escherichia coli proteins aggregated intracellularly by heat shock.

IbpA/B, 16 kDa heat-shock proteins were recently described as recognizing heterologous protein inclusion bodies in Escherichia coli cells; the corresponding genes formed an operon regulated by the rpoH gene product, sigma 32 protein (Burland et al (1993) Genomics 16, 551; Allen et al (1992) J Bacteriol 174, 6938; Chuang et al (1993) Gene 134, 1; Chuang and Blattner (1993) J Bacteriol 175, 5242). We have found that IbpA/Bs also recognize endogenous bacterial proteins aggregated intracellularly by heat shock. IbpA/B proteins were isolated and purified from the aggregates (the S fraction), identified by amino acid microsequencing and used as immunogen for anti-IbpA/B serum preparation. Western blotting with the serum showed that in cells growing at 30 degrees C IbpA/B were located in the bacterial outer membrane and appeared in the S fraction after heat shock. Then the cellular level of the IbpA/B proteins increased about 20-fold as estimated by densitometry of the Western blots. In the E coli rpoH strain the level of IbpA/B was higher than in wild type before the heat shock and rose to still higher levels after it. This result pointed to a regulation of ibpA/B operon by another factor, besides that of sigma 32.

Bacteriophage lambda lysis gene product modified and inserted into Escherichia coli outer membrane: Rz1 lipoprotein.

Lysis proteins of bacteriophage lambda were localized in different parts of the host envelope: S in the inner membrane,36 Rz in the membrane adhesion sites,14 and Rz1 in the outer membrane. The R gene product, the transglycosylase destroying bacterial murein, is a soluble protein. Computer-assisted analysis of the Rz1 protein amino acids sequence revealed that its N-terminal part contained the site 15VVVG [symbol: see text] C20, which could be recognizable for the SPase II and cleaved leaving lipid modified C20 as the N-terminal amino acid of the mature protein. Microsequencing of the Rz1 protein isolated from the expression products of E. coli [pSB54] carrying the Rz1 gene showed that the N-terminal part of the protein was cleaved as predicted. Lipid labeling with [3H]palmitate confirmed the expectation that Rz1 was a lipoprotein. E. coli [pSB54] treated with globomycin accumulated prolipoprotein, the Rz1 precursor, which was detectable by the anti-Rz1 serum on electropherograms as the 6.5-kDa protein, larger than mature protein. Physiological function of the Rz1 protein remains to be discovered, but as a first hint we noticed that it evokes increase of the fraction of adhesion sites of outer and inner membranes when overproduced from pSB54. The same effect was observed in induced E. coli (lambda) just before the lysis onset, however, one should be cautious in interpreting the results obtained in conditions of the overproduction of the Rz1 lipoprotein.

Erythrocyte Ca2(+)-ATPase: activation by enzyme oligomerization versus by calmodulin.

The subject of our studies is the mechanism of activation of the erythrocyte Ca2(+)-ATPase. Using purified, detergent solubilized enzyme it was found that equivalent maximal Ca2(+)-ATPase activity is obtained either upon addition of calmodulin or upon increase of enzyme concentration. Three independent methods, including Ca2(+)-ATPase activity, polarization of the enzyme modified with an external fluorescent probe, and efficiency of fluorescence resonance energy transfer between enzyme molecules have established that the concentration dependent activation is due to enzyme oligomerization. The oligomers bind calmodulin with a lower stoichiometry (0.5 mol calmodulin/mol Ca2(+)-ATPase), higher Ca2+ affinity (KCa = pCa 7.4), and higher cooperativity for Ca2+ (nH = 2.6) than the monomeric form (stoichiometry = 1 mol calmodulin/mol Ca2(+)-ATPase, KCa = pCa 7.0, nH = 1.1). The Ca2+ dependence of calmodulin binding and activation of monomers indicates that calmodulin binds before the Ca2(+)-ATPase activity is exhibited, demonstrating that the activation of this enzyme form is totally dependent on calmodulin. In contrast, oligomers reveal very similar Ca2+ dependence for calmodulin binding and for Ca2(+)-ATPase activity as well as for Ca2+ binding (assessed by tryptophan fluorescence), and for the oligomerization process (assessed by fluorescence energy transfer). The calmodulin antagonist drug 48/80 inhibits the calmodulin dependent activity of the monomers (I50 = 1.4 micrograms/ml) but has no effect on the activity of oligomers, confirming that calmodulin plays no role in the activation of the oligomeric enzyme. Our studies indicate that the erythrocyte Ca2(+)-ATPase can be activated by its high affinity, Ca2+ dependent binding of calmodulin or by a Ca2+ dependent oligomerization process which may involve calmodulin binding site.(ABSTRACT TRUNCATED AT 250 WORDS)

Divergent effects of ATP on the binding of the DnaK and DnaJ chaperones to each other, or to their various native and denatured protein substrates.

Using the native proteins lambda P, lambda O, delta 32, and RepA, as well as permanently unfolded alpha-carboxymethylated lactalbumin, we show that DnaK and DnaJ molecular chaperones possess differential affinity toward these protein substrates. In this paper we present evidence that the DnaK protein binds not only to short hydrophobic peptides, which are in an extended conformation, but also efficiently recognizes large native proteins (RepA, lambda P). The best substrate for either the DnaK or DnaJ chaperone is the native P1 coded replication RepA protein. The native delta 32 transcription factor binds more efficiently to DnaJ than to DnaK, whereas unfolded alpha-carboxymethylated lactalbumin or native lambda P binds more efficiently to DnaK than to the DnaJ molecular chaperone. The presence of nucleotides does not change the DnaJ affinity to any of the tested protein substrates. In the case of DnaK, the presence of ATP inhibits, while a nonhydrolyzable ATP analogues markedly stimulates the binding of DnaK to all of these various protein substrates. ADP has no effect on these reactions. In contrast to substrate protein binding, DnaK binds to the DnaJ chaperone protein in a radically different manner, namely ATP stimulates whereas a nonhydrolyzable ATP analogue inhibits the DnaK-DnaJ complex formation. Moreover, the DnaKc94 mutant protein lacking 94 amino acids from its C-terminal domain, which still possesses at ATPase activity and forms a transient complex with protein substrates, does not interact with DnaJ protein. We conclude that the DnaK-ADP form, derived from ATP hydrolysis, possesses low affinity to the protein substrates but can efficiently interact with DnaJ molecular chaperone.

The Clp ATPases define a novel class of molecular chaperones.

The Clp ATPases were originally identified as a regulatory component of the bacterial ATP-dependent Clp serine proteases. Proteins homologous to the Escherichia coli Clp ATPases (ClpA, B, X or Y) have been identified in every organism examined so far. Recent data suggest that the Clp ATPases are not only specificity factors which help to 'present' various protein substrates to the ClpP or other catalytic proteases, but are also molecular chaperones which can function independently of ClpP. This review discusses the recent evidence that the Clp ATPases are indeed molecular chaperones capable of either repairing proteins damaged during stress conditions or activating the initiation proteins for Mu, lambda or P1 DNA replication. A mechanism is suggested to explain how the Clp ATPases 'decide' whether to repair or destroy their protein substrates.

An iodoacetamide spin-label selectively labels a cysteine side chain in an occluded site on the sarcoplasmic reticulum Ca(2+)-ATPase.

Sarcoplasmic reticulum vesicles were labeled with [14C]iodoacetamide spin-label (ISL) under conditions where time courses of the reaction predicted that one amino acid residue would be preferentially labeled. Solubilized tryptic peptides were separated by high-performance liquid chromatography following extensive digestion, and amino acid sequences were determined for major and minor radio-labeled peptides. Only one radio-labeled residue, Cys-674 on the Ca(2+)-ATPase, could be identified. Extensive incubation with excess label increased nonspecific labeling, but did not produce detectable amounts of any other reactive side chain residue. Time courses of the iodoacetamide spin-label reaction were compared to those of 6-(iodoacetamido)fluorescein (IAF), and the ISL reaction was found to be more selective, in accordance with previous studies showing that IAF labeled both Cys-674 and Cys-670 [Bishop, J. E., Squire, T. C., Bigelow, D. J., & Inesi, G. (1988) Biochemistry 27, 5233-5240]. Titrations with spin-broadening reagents NiCl2 and Ni-EDTA showed Cys-674 to be in a region with very low solvent accessibility. These titrations also showed the ATPase to be distributed between two alternating conformations based on the accessibility of the label to NiCl2.

Stabilizing and destabilizing effects on plasma membrane Ca(2+)-ATPase activity.

We have examined the temperature-dependent effects of several organic compounds on the activity of the purified Ca(2+)-ATPase of erythrocytes. The monomeric enzyme was activated either by interaction with calmodulin or by oligomerization in the absence of calmodulin. Of the four homologous solute series studied including polyols, alkanols, aprotic solvents, and N-methyl derivatives of formamide and acetamide only polyols stabilized the enzyme over a broad range of concentration and temperature. Similarity of Ca(2+)-ATPase activity patterns at 25 and 37 degrees C and in the presence of glycerol is in agreement with indirect, stabilizing interactions. Glycerol also protected the Ca(2+)-ATPase from thermal denaturation at 45 degrees C. Within each homologous series, inhibitory effects increased with increasing solute concentration and with increasing structural similarity to detergents, indicating that direct destabilizing interactions are responsible for the observed inhibition. These were comparable to the destabilizing effect of urea. Oligomers were more resistant to all inhibitory solutes as compared to calmodulin-activated monomers suggesting that the nonpolar patches of the oligomerized enzyme are less accessible to solutes.

Formation of the preprimosome protects lambda O from RNA transcription-dependent proteolysis by ClpP/ClpX.

Using the bacteriophage lambda DNA replication system, composed entirely of purified proteins, we have tested the accessibility of the short-lived lambda O protein to the ClpP/ClpX protease during the various stages of lambda DNA replication. We find that binding of lambda O protein to its orilambda DNA sequence, leading to the so-called "O-some" formation, largely inhibits its degradation. On the contrary, under conditions permissive for transcription, the lambda O protein bound to the orilambda sequence becomes largely accessible to ClpP/ClpX-mediated proteolysis. However, when the lambda O protein is part of the larger orilambda:O.P.DnaB preprimosomal complex, transcription does not significantly increase ClpP/ClpX-dependent lambda O degradation. These results show that transcription can stimulate proteolysis of a protein that is required for the initiation of DNA replication.

Structure-function analysis of the Escherichia coli GrpE heat shock protein.

We have isolated various missense mutations in the essential grpE gene of Escherichia coli based on the inability to propagate bacteriophage lambda. To better understand the biochemical mechanisms of GrpE action in various biological processes, six mutant proteins were overexpressed and purified. All of them, GrpE103, GrpE66, GrpE2/280, GrpE17, GrpE13a and GrpE25, have single amino acid substitutions located in highly conserved regions throughout the GrpE sequence. The biochemical defects of each mutant GrpE protein were identified by examining their abilities to: (i) support in vitro lambda DNA replication; (ii) stimulate the weak ATPase activity of DnaK; (iii) dimerize and oligomerize, as judged by glutaraldehyde crosslinking and HPLC size chromatography; (iv) interact with wild-type DnaK protein using either an ELISA assay, glutaraldehyde crosslinking or HPLC size chromatography. Our results suggest that GrpE can exist in a dimeric or oligomeric form, depending on its relative concentration, and that it dimerizes/oligomerizes through its N-terminal region, most likely through a computer predicted coiled-coil region. Analysis of several mutant GrpE proteins indicates that an oligomer of GrpE is the most active form that interacts stably with DnaK and that the interaction is vital for GrpE biological function. Our results also demonstrate that both the N-terminal and C-terminal regions are important for GrpE function in lambda DNA replication and its co-chaperone activity with DnaK.

Co-chaperones Bag-1, Hop and Hsp40 regulate Hsc70 and Hsp90 interactions with wild-type or mutant p53.

Using highly purified proteins, we have identified intermediate reactions that lead to the assembly of molecular chaperone complexes with wild-type or mutant p53R175H protein. Hsp90 possesses higher affinity for wild-type p53 than for the conformational mutant p53R175H. The presence of Hsp90 in a complex with wild-type p53 inhibits the binding of Hsp40 and Hsc70 to p53, consequently preventing the formation of wild-type p53-multiple chaperone complexes. The conformational mutant p53R175H can form a stable heterocomplex with Hsp90 only in the presence of Hsc70, Hsp40, Hop and ATP. The anti-apoptotic factor Bag-1 can dissociate Hsp90 from a pre- assembled complex wild-type p53 protein, but it cannot dissociate a pre-assembled p53R175H-Hsp40- Hsc70-Hop-Hsp90 heterocomplex. The results presented here provide possible molecular mechanisms that can help to explain the observed in vivo role of molecular chaperones in the stabilization and cellular localization of wild-type and mutant p53 protein.

Fluorescence energy transfer studies of purified erythrocyte Ca2+-ATPase. Ca2+-regulated activation by oligomerization.

Fluorescence resonance energy transfer has been used to study oligomerization of the purified erythrocyte Ca2+-ATPase. The energy transfer efficiency has been measured at different enzyme concentrations, from fluorescein 5'-isothiocyanate attached on one enzyme molecule to eosin 5-maleimide or tetramethylrhodamine 5-isothiocyanate attached on another enzyme molecule. The energy transfer efficiency showed a sigmoid dependence on enzyme concentration and was half-maximal at 10-12 nM enzyme; this dependence on enzyme concentration closely resembled previously demonstrated dependence of Ca2+-ATPase activity and polarization of the fluorescein 5'-isothiocyanate enzyme (Kosk-Kosicka, D., and Bzdega, T. (1988) J. Biol. Chem. 263, 18184-18189). Thus, the three independent methods establish that enzyme concentration-dependent oligomerization is a mechanism of activation of the erythrocyte Ca2+-ATPase. Further energy transfer studies demonstrated that enzyme oligomerization required calcium. This calcium dependence was characterized by high affinity (half-maximal energy transfer at pCa 7.15) and cooperativity (Hill coefficient of 2.36), being very similar in both respects to the Ca2+ dependence of the Ca2+-ATPase activity. The data indicated that the oligomerization process produced a highly cooperative, Ca2+-regulated activation of the enzyme at physiologically relevant Ca2+ concentrations. These studies show that the Ca2+-ATPase can be fully activated by a Ca2+-dependent oligomerization mechanism, which is independent of the previously described activation by calmodulin. We propose two pathways for the activation of the Ca2+-ATPase, taking into account the interdependencies between the Ca2+, calmodulin, and enzyme concentrations.

ATP hydrolysis is required for the DnaJ-dependent activation of DnaK chaperone for binding to both native and denatured protein substrates.

Using two independent experimental approaches to monitor protein-protein interactions (enzyme-linked immunosorbent assay and size exclusion high performance liquid chromatography) we describe a general mechanism by which DnaJ modulates the binding of the DnaK chaperone to various native protein substrates, e.g. lambda P, lambda O, delta 32, P1, RepA, as well as permanently denatured alpha-carboxymethylated lactalbumin. The presence of DnaJ promotes the DnaK for efficient DnaK-substrate complex formation. ATP hydrolysis is absolutely required for such DnaJ-dependent activation of DnaK for binding to both native and denatured protein substrates. Although ADP can stabilize such as an activated DnaK-protein complex, it cannot substitute for ATP in the activation reaction. In the presence of DnaJ and ATP, DnaK possesses the affinity to different substrates which correlates well with the affinity of DnaJ alone for these protein substrates. Only when the affinity of the DnaJ chaperone for its protein substrate is relatively high (e.g. delta 32, RepA) can a tertiary complex DnaK-substrate-DnaJ be detected. In the case that DnaJ binds weakly to its substrate (lambda P, alpha-carboxymethylated lactalbumin), DnaJ is only transiently associated with the DnaK-substrate complex, but the DnaK activation reaction still occurs, albeit less efficiently.

Structure-function analysis of the zinc-binding region of the Clpx molecular chaperone.

The ClpX heat shock protein of Escherichia coli is a member of the universally conserved Hsp100 family of proteins, and possesses a putative zinc finger motif of the C(4) type. The ClpX is an ATPase which functions both as a substrate specificity component of the ClpXP protease and as a molecular chaperone. Using an improved purification procedure we show that the ClpX protein is a metalloprotein complexed with Zn(II) cations. Contrary to other Hsp100 family members, ClpXZn(II) exists in an oligomeric form even in the absence of ATP. We show that the single ATP-binding site of ClpX is required for a variety of tasks, namely, the stabilization of the ClpXZn(II) oligomeric structure, binding to ClpP, and the ClpXP-dependent proteolysis of the lambdaO replication protein. Release of Zn(II) from ClpX protein affects the ability of ClpX to bind ATP. ClpX, free of Zn(II), cannot oligomerize, bind to ClpP, or participate in ClpXP-dependent proteolysis. We also show that ClpXDeltaCys, a mutant protein whose four cysteine residues at the putative zinc finger motif have been replaced by serine, behaves in similar fashion as wild type ClpX protein whose Zn(II) has been released either by denaturation and renaturation, or chemically by p-hydroxymercuriphenylsulfonic acid.

Sarcolipin, the "proteolipid" of skeletal muscle sarcoplasmic reticulum, is a unique, amphipathic, 31-residue peptide.

The sarcoplasmic reticulum of rabbit skeletal muscle contains a small "proteolipid," i.e., a protein which is soluble in acidic CHCl3/CH3OH. We propose the name sarcolipin for this small protein, to signify its lipid-like solubility and association with the sarcoplasmic reticulum. We have determined the following amino acid sequence for sarcolipin, using protein chemistry methods: M E R S T R E L C L N F T V V L I T V I L I W L L V R S Y Q Y. This 31-residue sequence includes a 19-residue hydrophobic segment which probably spans the sarcoplasmic reticulum membrane. The molecular weight calculated from the sequence, 3733, agrees with that measured by fast atom bombardment mass spectrometry, showing that sarcolipin contains no attached fatty acyl or other prosthetic groups.

DjlA is a third DnaK co-chaperone of Escherichia coli, and DjlA-mediated induction of colanic acid capsule requires DjlA-DnaK interaction.

DjlA is a 30-kDa type III membrane protein of Escherichia coli with the majority, including an extreme C-terminal putative J-domain, oriented toward the cytoplasm. No other regions of sequence similarity aside from the J-domain exist between DjlA and the known DnaK (Hsp70) co-chaperones DnaJ (Hsp40) and CbpA. In this study, we explored whether and to what extent DjlA possesses DnaK co-chaperone activity and under what conditions a DjlA-DnaK interaction could be important to the cell. We found that the DjlA J-domain can substitute fully for the J-domain of DnaJ using various in vivo functional complementation assays. In addition, the purified cytoplasmic fragment of DjlA was shown to be capable of stimulating DnaK ATPase in a manner indistinguishable from DnaJ, and, furthermore, DjlA could act as a DnaK co-chaperone in the reactivation of chemically denatured luciferase in vitro. DjlA expression in the cell is tightly controlled, and even its mild overexpression leads to induction of mucoid capsule. Previous analysis showed that DjlA-mediated induction of the wca capsule operon required the RcsC/RcsB two-component signaling system and that wca induction by DjlA was lost when cells contained mutations in either the dnaK or grpE gene. We now show using allele-specific genetic suppression analysis that DjlA must interact with DnaK for DjlA-mediated stimulation of capsule synthesis. Collectively, these results demonstrate that DjlA is a co-chaperone for DnaK and that this chaperone-co-chaperone pair is implicated directly, or indirectly, in the regulation of colanic acid capsule.