It is now well established that the σE regulon of Escherichia coli is induced by misfolding of proteins in the periplasm and the outer membrane. htrA belongs to this regulon and encodes a periplasmic protease involved in the degradation of misfolded proteins. htrA transcription is also under the positive control of a two component signal transduction system CpxR CpxA. Closer examination of the putative signal transduction pathway modulating htrA transcription has led us to the identification of two new genes. Biochemical and genetic evidence shows that these two genes encode two phosphoprotein phosphatases, designated PrpA and PrpB. These are the first examples of typical serine/threonine and tyrosine phosphatases described in E.coli. PrpA and PrpB are involved in signaling protein misfolding via the CpxR CpxA transducing system. In addition, both PrpA and PrpB modulate the phosphorylated status of some other phosphoproteins in E.coli. Finally, we show that PrpA is a heat shock protein.
In Escherichia coli, the heat shock response is regulated overall by two alternative sigma factors, σ32 and σE/σ24, encoded by the rpoH and rpoE genes, respectively. Transcription of these two regulons is induced by various stresses and by protein misfolding in general. The stimuli arising from misfolded proteins are even more dramatic at high temperatures since proteins are more prone to aggregation. It is therefore not surprising that most of the protein chaperones and proteases are encoded by heat shock regulated genes. It is quite interesting that each of the two heat shock regulons have evolved to respond to protein misfolding in a defined cell compartment. Transcription of the rpoH regulon is induced by protein misfolding in the cytoplasm whereas transcription of rpoE, that was first believed to be induced exclusively by misfolded outer membrane proteins (Mecsas et al., 1993), appears to respond to the accumulation of any exported protein that is unstable or misfolded (Missiakas et al., 1995, 1996b; Raina et al., 1995). Many Eσ32‐transcribed genes have been shown to encode repair functions such as chaperones and proteases (reviewed by Missiakas et al., 1996a). The role of the σE regulon in preventing protein misfolding in the extracytoplasmic compartments is not yet fully deciphered. Interestingly, the elevated temperatures can induce both heat shock regulons. This is easily explained by the fact that the EσE polymerase transcribes both the rpoE and the rpoH genes (Erickson and Gross, 1989; Wang and Kaguni, 1989; Raina et al., 1995; Rouvière et al., 1995). htrA (degP) is the third known σE‐transcribed gene (Lipinska et al., 1988; Erickson and Gross, 1989). This gene encodes a periplasmic protease which appears to degrade specifically unstable proteins. How is the need for HtrA in the periplasm signaled to the σE regulon? We have addressed this question genetically by constructing a transcriptional fusion between the htrA promoter and the lacZ gene and searching for trans‐acting mutations affecting lacZ expression. Many mutations were found to map to the rpoE gene, but not all of them (Raina et al., 1995). It appeared that transcription of htrA is also regulated by a second sensing system, the CpxR CpxA couple (Danese et al., 1995; Raina et al., 1995). CpxA encodes a protein with homology to the classical histidine kinases of the two component systems (Weber and Silverman, 1988) such as EnvZ or CheA (see review by Parkinson, 1993). These kinases are embedded in the inner membrane and upon activation by an ‘outside’ stimulus they undergo autophosphorylation to activate, in turn, a cognate DNA‐binding regulatory protein by the transfer of the phosphate group. Based on sequence homologies with other regulatory proteins of two component systems, such as OmpR or CheA, it seems that CpxR (Dong et al., 1993) could be activated through phosphorylation by CpxA. In this study, we demonstrate that, in fact, it is a network of phosphorylation and dephosphorylation processes which fine‐tunes the transcriptional induction of htrA. This is quite similar to the signal transduction pathways observed in higher eukaryotes which occur through a cascade of events involving both kinases and phosphatases. Indeed, genetic analyses presented here reveal that E.coli encodes at least two type I phosphatase activities. In conjunction with the CpxA kinase, they modulate the transcription of some cellular components which may be important for protecting the cell upon accumulation of misfolded proteins in the periplasm. They do so by specifically signaling for induction of htrA transcription. PrpA phosphatase activity seems to be important for the general heat shock response of E.coli since overexpression of PrpA leads to the accumulation of the heat shock proteins.
We have previously described two complementary approaches which led to the identification of the rpoE gene (encoding σE) as well as other genes which could encode modulators of σE activity (Raina et al., 1995). One of these approaches was based on isolating trans‐acting mutations which down‐regulate the expression of reporter–promoter fusions of lacZ to rpoHP3 and htrA promoters. Another approach was based on identifying those genes which, when present on multicopy plasmids, were able to positively affect the transcription of such promoter fusions. Both led to the identification of the rpoE gene itself (Raina et al., 1995) as well as a few other additional genes. In this study, we specifically analyzed those loci which affect primarily htrA transcription in order to understand the different levels of sensing and signaling protein misfolding in the periplasm.
Isolation of mutations affecting htrA transcription
Among the original 64 Lac‐down mutants isolated earlier (Raina et al., 1995), at least three complementation groups were found to affect only the transcription of htrA–lacZ activity but not rpoHP3–lacZ (Figure 1A). Most of these mutants were simultaneously temperature sensitive for growth above 43°C. One group comprising six Lac‐down mutants was complemented by cosmids which hybridized to λ Kohara phages 334, 335 and 336 corresponding to the 41 min region of the E.coli chromosome. Another group of seven mutants was found to be complemented by cosmids hybridizing to λ 449 and 450 (61.5 min). The last group of four mutants was complemented by cosmids hybridizing to λ 539, 540 and 541 (88.5 min). These map positions were further verified by bacteriophage P1‐mediated transduction using known genetic markers. The location of mutations mapping at 41.5 min was confirmed by an observed 70% linkage with eda::Tn10 (CAG18486, Singer et al., 1989). Mapping of mutations located at 61.5 min was confirmed by linkage to mutS::Tn10 (90% linkage) and to rpoS::Tn10 (∼75% linkage). The assignment of the third complementation group (88.5 min) was achieved using clpQ::ΩCm (DM1674) as the linked marker (Missiakas et al., 1996c). This last group was complemented by clones carrying the two‐gene operon cpxR cpxA and therefore corresponded to a group reported earlier (Raina et al., 1995). The other two groups carried genes which were designated as prpA (41 min) and prpB (61.5 min) after further characterization which are described in the following sections.
Increased htrA transcription can be achieved by overexpression of the prp genes
In our previous studies (Missiakas et al., 1993; Raina et al., 1995), we had also observed that multicopy expression of genes not mapping to rpoE could induce transcription from the htrA gene and not rpoHP3 or rpoEP2 (the rpoE gene has two promoters, P2 is recognized by EσE; Raina et al., 1995). These clones, selected from a genomic library constructed in a p15A‐based vector (see Materials and methods), were analyzed more closely. It was found from their restriction pattern that 12 of them carried a common 2.8 kb Sau3A DNA fragment (pDM506, prpA+). These 12 clones were mapped on the E.coli chromosome and were shown to hybridize to bacteriophage λ 336 (19H3) of the Kohara library (Kohara et al., 1987). Interestingly, this corresponded to the 41 min region on the E.coli DNA chromosome to which one of the complementation groups comprising six trans‐acting mutations was also mapped. Further characterizations and subcloning experiments identified a 900 bp PvuII–AccI DNA fragment in pDM1695 (prpA+), which was found to induce the htrA–lacZ expression to the same extent as the original construct pDM506 (Figure 1B).
This multicopy cloning approach did not lead to the re‐isolation of the second locus mapping at 61.5 min, identified in the previous screening of trans‐acting mutations. To verify whether an induction of htrA could be observed with prpB‐carrying clones, we used the minimal DNA fragment subcloned from the cosmid complementing mutations at this locus and cloned it in the same vector used to construct the multicopy library. This new clone pDM1754 (prpB+) was assayed for β‐galactosidase activity of the htrA–lacZ fusion‐carrying strain. As shown on Figure 1B, a comparatively smaller induction of htrA transcription was detected using pDM1754 (prpB+). However, if cloned in a higher copy number vector (pBR322 based pSE420) pDM1757 prpB+, an ∼6‐fold induction of htrA transcription was observed (Table II).
PrpA and PrpB proteins have typical serine/threonine phosphatase signatures
Sequence analyses of the two clones revealed an ORF of 654 nucleotides (nt) long for both prpA and prpB genes. In the case of prpA (pDM506), the ORF is predicted to start with an ATG at nt position 467 and terminating with TAA at nt position 1123, and could encode a 24 257 Da polypeptide with a predicted pI of ∼7. These predictions are consistent with the observed pI and molecular weight as shown in Figures 6A and 11A (migration on 2D gel). Downstream of the prpA ORF are small stretches of DNA sequence which show significant homology to small stretches of DNA sequence found in the dicABCF region (Figure 2; Faubladier and Bouché, 1994). In addition, the prpA ORF is surrounded by large non‐coding regions ∼500 bp upstream and >1 kb at its 3′‐end. This is quite unusual in the E.coli genome where ORFs are generally tightly packed. In the case of prpB (pDM1755), the ORF also starts with an ATG at nt position 202 and terminates with a TAG at nt position 856. It is located 108 nt downstream of the stop codon of the mutS gene and is transcribed in the same orientation. The prpB gene seems to have its own promoter since clones not carrying mutS can still complement a prpB mutant allele. prpB ORF is predicted to encode a 25 082 Da polypeptide. This is again consistent with the observed size of the protein (see later protein purification data, Figure 6A). Sequence comparison with GenBank release 91 indicated no similarity with any known E.coli proteins but a significant homology to the serine/threonine family of type I eukaryotic phosphatases. An even higher degree of homology was shared with the phosphatase from bacteriophage λ (referred to in this study as the λ‐PP protein, λnin ORF221; Cohen et al., 1988) (Figure 3). Because of these observations, we designated the two genes as prp for protein phosphatase.
What appeared particularly noteworthy was the presence of most of the conserved residues shown to be part of the active site. Among those are the residues involved in metal binding. From the known phosphatase sequences and 3D structure (Goldberg et al., 1995), these residues would correspond to D24, H26, D53 and H80 for PrpA and D22, H24, D51 and H78 for PrpB (Figure 3). The only region with lower homology between PrpA, PrpB and λ‐PP sequences consists of a small stretch of amino acids located between amino acids 106 and 122. This region in PrpA is predicted to contain a coiled‐coil domain and might be important for interaction with the substrates. Consistent with its activity on htrA transcription as a phosphoprotein phosphatase, sequence analyses of four of the Lac‐down mutants isolated in the prpA gene showed changes in the highly conserved residues. One of the mutations was D24 to V (GAT to GTT) and two others corresponded to H26 to N (CAC to AAC) and H26 to L (CAC to CTC). The fourth sequenced mutation corresponded to a change L107 to Q (CTG to CAG). L107 lies in the predicted coiled‐coil domain of PrpA. Unlike wild‐type prpA, neither of these cloned prpA mutant alleles was able to induce htrA–lacZ activity when provided on a same‐copy‐number plasmid (Figure 1B). These results further confirm that it is the phosphatase activity of PrpA which influences the transcription of htrA. We also sequenced one of the mutant variants of prpB, i.e. prpB17. This variant was found to carry the mutation H78 to N (CAC to AAC). This histidine residue is again conserved in all type I phosphoprotein phosphatases (Figure 3).
Phenotypic analyses of the prpA and prpB null mutants
The multicopy effect observed with the two genes mapped at 41 min (prpA) and 61.5 min (prpB) led us to presume that the corresponding trans‐acting mutations isolated were the result of a loss of function of both PrpA and PrpB proteins. To confirm that this was indeed the case, the null alleles were constructed as described in Materials and methods and transduced into SR1458, the htrA–lacZ‐carrying strain. Their effects on htrA transcription, as judged from the results presented in Figure 1A, were similar to those obtained with the point mutations. For both genes, prpA and prpB, all the alleles isolated as either null or point mutations conferred a slightly temperature‐sensitive (Ts−) growth phenotype above 43°C. However, such effects were not bactericidal. Only a mild synergistic effect was observed when mutations in prpA and prpB were combined. In addition, null mutations in prpA or prpB genes conferred a slow growth phenotype, even at permissive temperatures. For example, the doubling time of prpA null mutant bacteria is ∼80 min at 30°C and that of prpB null mutant bacteria is ∼65 min as compared with a 45 min doubling time for the isogenic wild‐type bacteria at 30°C.
The prpA gene is heat shock regulated
A primer extension analysis was carried out with RNA isolated from wild‐type bacteria under a range of growth temperatures. The results of this analysis showed that prpA mRNA has at least one defined 5′‐end located 417 nt upstream of the putative ATG initiation codon (Figure 4A. The −10 (CACC) and −35 (GGCGAA) boxes of this start site present a good homology to the promoter consensus sequences of Eσ32‐transcribed genes (Figure 4B particularly in the −10 region. However, it may be noted that this promoter is quite weak compared with classical heat shock promoters. This may be due to a somewhat lower homology in the −35 region. In addition, prpA‐specific transcripts accumulated at 50°C (Figure 4A). It is known that, at such temperatures, heat shock genes or stress‐inducible psp genes (Weiner et al., 1991) are the only ones whose transcripts accumulate. To further substantiate that prpA is indeed transcribed by the Eσ32 holoenzyme, an in vitro run‐off assay was carried out using a linear 659 bp Sau3A–PstI DNA fragment as a template. As shown in Figure 5, purified Eσ32 holoenzyme initiates transcription at the same site as observed from in vivo RNA extracted from bacteria shifted to 50°C. Taken together, these results provide good evidence that prpA transcription is positively regulated by heat shock.
Purified PrpA and PrpB proteins exhibit phosphoprotein phosphatase activities in vitro
In order to compare the activities of the two putative E.coli phosphatases to some other known phosphatases, we decided to use the recently described bacteriophage λ phosphatase (λ‐PP) as a control (Cohen and Cohen, 1989; Barik et al., 1993; Zhuo et al., 1993). A profile of the three purified proteins is shown in Figure 6A. PrpA, PrpB and λ‐PP appeared to migrate to approximately the same distance as the molecular weight marker corresponding to 24–25 kDa. Surprisingly, λ‐PP migrates on SDS–PAGE as a slightly higher molecular weight species than the PrpA and PrpB proteins despite its similar predicted mass. During all purification steps, the phosphatase activity was measured using the classical phosphatase substrate p–nitrophenol phosphate (pNPP) and was correlated with fractionation of a 25 kDa species (this activity is referred as pNPPase activity). All biochemical assays presented have been performed in triplicate using each protein separately. Among the three phosphatases, PrpA exhibited a rather good pNPPase activity at high temperatures, retaining ∼90% of its activity at 65°C as compared with 20% for PrpB. This might be consistent with a role of the prpA gene product during the heat shock response and the finding that the gene is regulated by heat shock. The effect of various inhibitors was also tested on the pNPPase activity of the three enzymes (Table I). Despite their high sequence homologies, the pNPPase activities of the three enzymes are not inhibited to the same extent, depending on the salt or anion used. The pNPPase activity of PrpB was unaffected by most of the inhibitors except by zinc (Table I).
We also assayed the ability of the PrpA and PrpB phosphatases to dephosphorylate protein substrates at specific residues such as serine/threonine or tyrosine. For this purpose, two proteins, casein and myelin basic protein (MyBP), were phosphorylated using eukaryotic kinases which specifically phosphorylate at either serine/threonine or tyrosine. The assays were all performed under the same conditions of temperature, pH and phosphatase concentration (Figure 6B and C). In such conditions, it appeared that PrpA, like λ‐PP, is able to dephosphorylate casein and MyBP at both serine/threonine and tyrosine residues (Figure 6B and C). The phosphoprotein phosphatase activity of PrpB was best observed with the phospho‐MyBP substrates (Figure 6C). In this case again as with pNPPase activity, despite the high degree of homology between the three enzymes their catalytic properties appear to be slightly different. Hence like λ‐PP, E.coli phosphatases PrpA and PrpB have dual specificities in that they are both serine/threonine and tyrosine phosphatases.
The PrpA‐ and PrpB‐mediated increase in htrA transcription is dependent on the presence of functional CpxR and CpxA proteins
The genetic data presented here show that multiple unlinked mutations affecting htrA transcription could be isolated, including: rpoE, cpxR, cpxA, prpA and prpB. We tried to analyze the contribution of each locus to htrA transcription and whether the different mutations were epistatic, the aim being to understand the sequence of molecular events leading to htrA induction. To achieve this, we combined various mutations and examined their synergistic effect using the reporter LacZ activity from the htrA promoter. We first examined the combination of rpoE mutations with others. Double rpoE cpxR or rpoE prpA null mutations proved to affect htrA transcription most dramatically (Table II, reduction from 150 to ∼14 Miller units). This was not the case with the double null mutation prpA cpxR or prpB cpxR (Table II, most of them displayed ∼100 Miller units for their activity). Hence RpoE appears to act independently of PrpA as well as PrpB and CpxR/CpxA. This was further supported by the findings that the multicopy effect of prpA on the transcription of the htrA gene was still observed in an rpoE null background (induction from 38 to 94 Miller units). However, in strains lacking either the CpxR or CpxA protein no significant such induction was observed (Table II, 95 units in the cpxR mutant versus 109 for the isogenic cpxR mutant carrying the prpA gene onto the plasmid). Although the values for β‐galactosidase activities scale up quite differently (Table II), clearly transcription of htrA which is severely reduced in the rpoE mutant background, can still be stimulated 2‐ to 3‐fold upon overexpression of either prpA or prpB. A partial dependence on the presence of functional rpoE for the induction of htrA transcription by the Cpx pathway has also been observed by Danese et al. (1995). Similarly, the induction of htrA transcription by the overexpression of PrpB was also abolished in a cpxR null mutant background (Table II).
Protein misfolding in the extracytoplasm is sensed by a global Prp Cpx pathway
We then examined directly the involvement of a putative cpx prp pathway for sensing protein misfolding in the periplasm. Keeping in mind that htrA transcription in vivo is mainly induced by the accumulation of misfolded exported proteins, we took advantage of our knowledge of the Dsb proteins which are, so far, the best characterized folding catalysts. We have previously observed that mutations in the dsb genes, whose gene products catalyze the correct oxidation and folding of exported proteins (Missiakas et al., 1995), led to a 2‐ to 3‐fold induction of htrA transcription (Missiakas et al., 1995; Raina et al., 1995). Figure 7 compares the levels of inducibility of htrA transcription in a dsbD mutant background. Induction of htrA transcription is fully optimized only when functional CpxR/A and PrpA proteins are present (Figure 7). Clearly, part of the misfolding events is sensed by the whole σE regulon since transcription is induced from both htrA and rpoHP3 (Figure 7A). Interestingly, in an rpoE null mutant background, the accumulation of misfolded proteins associated with dsbD null mutation is still reflected by a minor induction of htrA transcription. No such increase is observed when using the rpoHP3–lacZ fusion in an rpoE null mutant background since rpoHP3 is exclusively transcribed by EσE (Figure 7B). Also, introducing either prpA or cpxR null mutation does not affect transcription from the rpoHP3 promoter (Figure 7B). It may be pointed out that, although the induction of htrA transcription is reduced from a factor of 3‐fold in the wild type to 2‐fold in cpx and prp mutants, it is reproducible and significant. It is particularly important since it is well established that a combination of htrA mutation is additive with any of the known mutations in dsb genes in terms of induction of the σE regulon (Missiakas et al., 1995; Raina et al., 1995).
P∼(CpxA CpxR) is a target of PrpA phosphatase action in vivo
From the genetic evidence presented, it seemed likely that CpxA and/or CpxR could be the targets of the phosphatase activity of PrpA. Bacteria carrying the cpxR cpxA operon on a plasmid under the control of an inducible ptac promoter were incubated in the presence of radioactive inorganic phosphate (32Pi). Incorporation of 32Pi by the CpxR CpxA system was estimated by immunoprecipitating CpxR from total E.coli extracts. This phosphorylation status of the CpxR protein was used to monitor the phosphorelay activity between CpxA and CpxR. Incorporation of 32Pi by the two component system was assayed in three isogenic backgrounds: wild‐type bacteria, bacteria co‐expressing the prpA gene on a plasmid with its own promoter, and bacteria lacking the functional PrpA protein (Figure 8). These results suggest that either the phosphate transfer between CpxA and CpxR or the stability of 32P∼CpxR are modulated by the presence of PrpA. Therefore, PrpA which has the signature of a typical type I phosphatase is able to dephosphorylate at serine and tyrosine residues but also at either histidine or aspartic acid residues. For some two component systems, an ‘aspartyl‐phosphatase’ activity such as that of CheZ has been found to specifically dephosphorylate the P∼CheY response regulator, providing a means to enhance the response time to signals issued by the chemotaxis transduction system (Parkinson, 1993). However, CheZ does not share any homology with typical phosphoprotein phosphatases. Hence we hereby provide the first evidence to date that type I phosphatases exist in E.coli and that they can modulate signal transduction pathways.
Overexpression of PrpA and PrpB relieves the envelope toxicity due to abnormal proteins
It has been previously reported that induction of the Cpx pathway can relieve the envelope toxicity otherwise observed in the presence of hybrid protein fusions like LamB–LacZ–PhoA (Danese et al., 1995). This has been shown to be dependent on the induction of htrA transcription mediated via the Cpx pathway (Danese et al., 1995). The expression of this hybrid fusion can be induced with maltose. Such an induction is toxic to the cell and makes bacteria sensitive to SDS (0.4%) unless HtrA is overproduced from a plasmid (Cosma et al., 1995). When such a fusion‐carrying strain was transformed with the plasmid pDM1695 containing the prpA+ gene, the toxicity was relieved to the same extent as when htrA was provided on a plasmid with a similar copy number (Table III). Overproduction of PrpB also helped to relieve this toxicity but only when the prpB gene was present on a plasmid with a higher copy number as observed by the decrease in the zone of growth inhibition (Table III). These results correlate overall with the relative induction of htrA transcription by the prpA+ or prpB+ gene on multicopy plasmids.
Influence of PrpA on other two component systems
The lack of two major periplasmic catalysts, DsbA and DsbB, leads to the induction of htrA transcription. In addition, dsbA and dsbB mutants are highly mucoid meaning that a third of the transcriptional machinery, the RcsB RcsC two component system, is also induced upon misfolding in the extracytoplasmic compartments. Mucoidy results from the production of colanic acid capsular polysaccharide. Capsule synthesis is promoted by proteins encoded by genes of the cps operon. Transcription of the cps genes is induced by two positive regulators, RcsA and phosphorylated RcsB. Synthesis of the capsule has been shown to be adaptive under certain external stresses such as desiccation, when strengthening of the cell envelope becomes an important mechanism of defense. Under normal conditions, capsular polysaccharide biogenesis is shut off since RcsA is rapidly degraded by the Lon protease. RscB, on the other hand, is part of the two component signaling system RcsB RcsC and is activated by phosphorylation, in a RcsC‐dependent manner (Gottesman and Stout, 1991). Overexpression of prpA in dsbA or dsbB backgrounds greatly reduces the mucoidy of the mutant cells. Quantification of such an RscB‐dependent transcriptional activity was also analyzed using cps–lacZ‐carrying strains. As can be seen in Table IV, a 2‐ to 3‐fold reduction of cps–lacZ activity was observed in dsbA or dsbB mutant strains carrying the multicopy prpA plasmid. It is likely that in this case, signaling for increased transcription of cps genes was inhibited by the phosphatase activity of PrpA acting either as a phospho‐histidine phosphatase on RcsC or as a phospho‐aspartic phosphatase on RcsB. It is quite interesting that an independent phosphatase enzyme such as PrpA can also fine tune the regulation of other two component systems in the cell.
Identification of in vivo targets of PrpA and PrpB
It is well known that some proteins get phosphorylated in E.coli (Freestone et al., 1995). The physiological importance of E.coli phosphatases in modulating the phosphorylation status of such proteins was addressed by examining the total protein profiles of wild‐type E.coli versus strains lacking prpA or prpB, on regular SDS–PAGE as well as on 2D gels (Figures 9A and B, and 10). Consistent with our model for the involvement of Prp proteins in dephosphorylation of phosphoproteins, an increased number of phosphoproteins was seen in extracts prepared from strains carrying mutations in prpA or prpB as compared with the isogenic wild‐type strain (Figure 9A). Approximately 20 different phosphoproteins were observed in both bacteria carrying null mutations of prpA or prpB and point mutants leading to a loss‐of‐function of the phosphatase activity (prpA26 and prpB17, see Table VI). As shown in Figure 9A and B, at least four additional phosphoproteins accumulate in either prpA or prpB mutant bacteria. Moreover, when wild‐type E.coli was transformed with a plasmid carrying either the prpA+ gene (pDM1695) or the prpB+ gene (pDM1755), most of the phosphorylated proteins seen in the wild‐type extract no longer accumulated (Figure 9A).
Figure 9B shows evidence for the accumulation of phosphorylated CpxA (P∼CpxA) in the prpA and prpB null mutant bacteria. The identification of the band corresponding to CpxA was achieved by running a sample extracted from cpxA null mutant strain. Bacteria deleted for both prpA and prpB show an increase in the accumulation of phosphorylated proteins, including P∼CpxA, as compared with the wild type (Figure 9B).
In an attempt to identify some of the phosphorylated proteins that are the targets of PrpA and PrpB phosphatases, some of these cell extracts were also analyzed by 2D equilibrium gel electrophoresis (Figure 10). Not all phosphoproteins could be resolved on such gels since the rank of ampholines used allows a good separation only between pH 7 and 4.5. In addition, there might be some membrane proteins which may not enter the gel. Hence mostly six different phosphoproteins (out of 20) accumulating in either prpA or prpB null mutant bacteria can be resolved on these gels (Figure 10). As a control, we also labeled bacteria carrying mutations in the cpxA gene, either a null mutation (cpxA::ΩCm) or cpxA* a chromosomal allele leading to a gain‐of‐function of CpxA. Then, to further substantiate that the Prp proteins act via the Cpx signal transduction pathway, extracts prepared from bacteria overexpressing the cpxR gene were also analyzed. Hence, the major spot corresponds to the P∼CpxR protein whose position on the 2D gel was assigned by running a 32P‐labeled extract from bacteria carrying the cpxR+ gene‐containing plasmid (pDM1787). Finally, the P∼CpxR protein was also found to accumulate in strains carrying the cpxA* allele (a change of T252 to R, see Table VI) and to disappear in strains lacking the cpxA gene (cpxA::ΩCm). This cpxA* allele corresponds to a mutation leading to a gain‐of‐function i.e. a hyper‐kinase activity of the CpxA protein and corresponds to a similar mutation envZ11 (T247R; Aiba et al., 1989). EnvZ is the histidine kinase most closely related to CpxA (Weber and Silverman, 1988).
Overexpression of PrpA induces the heat shock response
We then compared the global protein profiles as labeled with [35S]methionine of wild‐type and pDM1695‐(prpA+)‐carrying strain, by 2D equilibrium gel electrophoresis (Figure 11A). While an increase in the accumulation of Eσ32‐transcribed heat shock proteins was found, as judged from the overall protein profiles of the 2D gels, some other proteins were found to be present in diminished amounts (Figure 11A). These results were further supported by the observation of an increased transcription of Eσ32‐transcribed promoters as monitored from lon–lacZ, htpG–lacZ and groELS–lacZ promoter fusions (Table V). However, the increase in the activity of Eσ32‐dependent promoters is not as dramatic as with the htrA promoter (Table V). To confirm these results, we also examined in vivo the level of transcription of the major heat shock gene dnaK, by performing Northern blot analysis. Total RNA was extracted from isogenic bacteria carrying either the vector alone or the plasmid pDM1695 (prpA+) and probed for the accumulation of dnaK‐specific message. As shown in Figure 11B, at least a 2‐ to 3‐fold increase in the accumulation of dnaK‐specific message is observed even at normal temperatures (30°C) upon overexpression of prpA.
In addition to the heat shock proteins, the 2D gels revealed that accumulation of many other proteins was affected when prpA was present on a multicopy plasmid (Figure 11A, lower panel). As predicted from the calculated pI, the PrpA protein migrates on this 2D gel towards the basic end (Figure 11A, lower panel). Quite interesting is the separation of PrpA protein into two spots. This suggests that PrpA is modified in vivo and that this modification may possibly account for regulating its activity.
This study demonstrates that accumulation of misfolded proteins in the extracytoplasm is sensed by multiple transcriptional systems. The primary response depends on the presence of HtrA, a periplasmic protease. The levels of HtrA are controlled at the transcriptional level and the EσE polymerase is responsible for the synthesis of most, but not all, htrA transcripts. Mutations in the rpoE gene encoding the sigma factor σE lead to a severe, but partial, decrease in htrA transcription (Raina et al., 1995). In order to understand how signaling for increased htrA transcription is transduced between the two cell compartments, we performed extensive mutagenesis and isolated three additional loci which affected htrA transcription. These mutations were mapped to the cpxR cpxA operon, the others mapped into two new genes prpA and prpB. We found that these mutations affected specifically htrA transcription in a manner independent of the EσE transcription activity. Interestingly, such mutants had no effect on the transcriptional activity of the two other known genes of the rpoE regulon, namely rpoH or rpoE itself. The prpA gene was again isolated in a complementary genetic approach looking for gene products which in multicopy significantly induced htrA transcription.
PrpA and PrpB are prokaryotic type I‐like phosphatases
PrpA and PrpB are ∼50% identical at the amino acid level. They seem to be prototypes of classical eukaryotic type I serine/threonine phosphatases especially in their catalytic domains, based on significant sequence homology. Biochemical characterization proved that both PrpA and PrpB could hydrolyze pNPP, a typical substrate for phosphatase activity. In addition, PrpA and PrpB were both able to dephosphorylate protein substrates phosphorylated either at serine/threonine or tyrosine residues, with efficiencies comparable with that of λ‐PP, a recently characterized phosphatase with catalytic properties similar to type I phosphatases (Barik, 1993; Zhuo et al., 1993). Four point mutations leading to a loss of phosphatase activity were isolated in prpA. Of these, three mapped in residues D24 and H26. They correspond to residues D20 and H22 in λ‐PP which have been shown, by site‐directed mutagenesis, to be essential for the phosphatase activity of the λ protein (Zhuo et al., 1994). From the 3D structures of human type I phosphatase (Egloff et al., 1995; Goldberg et al., 1995) and calcineurin (type II phosphatase) (Griffith et al., 1995), these residues are also predicted to be part of the metal‐binding pocket in PrpA. Similarly, one prpB loss‐of‐function mutant was found to carry a change for the highly conserved residue H78 to N which corresponds to H76 in λ‐PP. The finding of phosphoprotein phosphatases in such a simple organism also raises the question about their in vivo role and function. If no requirement for a phosphatase activity in bacteriophage λ is yet known, E.coli phosphatases like PrpA and PrpB seem to play a direct role in the signal transduction pathways coupled with activation of gene transcription. Finally, it appears from this study that there are still some unknown features within the catalytic domains of phosphatases which make large differences in their properties. PrpA and PrpB are clear examples of this. Both in vivo and in vitro, they behave quite differently (towards the protein substrates casein and MyBP, for example) despite a high degree of homology (50% identity).
It is interesting that neither prpA nor prpB genes are essential for bacterial growth under normal conditions (30°C). This is probably consistent with the fact that, so far, none of the known components, periplasmic folding catalysts or periplasmic chaperones, are essential for bacterial growth under normal conditions. For example none of the dsb genes is essential, neither are any of the peptidyl prolyl isomerases (RotA, FkpA and SurA), nor OmpH/Skp; even though, in the absence of most of them, folding of exported proteins is retarded or impaired and the σE regulon is induced (Missiakas et al., 1996a; 1996b).
How do PrpA and PrpB influence htrA transcription?
Various lines of evidence are presented showing that the PrpA protein phosphatase activity, and to some extent that of PrpB, play an active role in the induction of htrA transcription. First, overexpression of PrpA and PrpB led to a 4‐ to 6‐fold increase in htrA transcription. Second, some chromosomal point mutations were isolated which led to a decrease in htrA transcription and were shown to map to the prpA and prpB genes. These specific mutations were identified as changes such as D24 to V, H26 to N and H26 to L in the case of prpA and H78 to N in the case of prpB. These mutants behaved quite like prpA or prpB null mutations. Third, overexpression of these mutant proteins in the same cloning system as the wild‐type prpA gene did not lead to the 4‐ to 6‐fold induction of htrA transcription. Since positions D24 and H26 correspond to highly conserved residues in the active site of type I phosphatases, this directly implicates the phosphatase activity of PrpA as the key element of its in vivo physiological role. Similarly, mutation H78 to N in prpB leads to a loss‐of‐function of the protein in vivo. The corresponding position has been shown to be essential for λ‐PP activity using site‐directed mutagenesis (Zhuo et al., 1994). Fourth, the phosphatase‐dependent activation of htrA transcription by either PrpA or PrpB was completely dependent on the presence of functional CpxR and CpxA proteins. The main pieces of evidence for this are the findings that the multicopy effect of prpA or prpB on htrA transcription was abolished in a cpxR cpxA null mutant and that no synergistic effect was observed when combined with cpxR cpxA mutations. Fifth, in vivo experiments showed that active PrpA and PrpB modulate the phosphorylated status of the Cpx system as reflected by the increased amount of P∼CpxR detectable in prpA or prpB null mutant backgrounds (Figures 8, 9 and 10) and a decrease in the accumulation of P∼CpxR when prpA+ is overexpressed (Figure 8). Since P∼CpxA also accumulates in prpA or prpB null mutant backgrounds (Figure 9B), it is likely that, like λ‐PP (Zhuo et al., 1993), both E.coli phosphatases have a histidine phosphatase activity. Such a phosphatase activity will affect the phosphorylated status of both CpxA and CpxR in vivo.
The CpxR and CpxA proteins (Weber and Silverman, 1988; Dong et al., 1993) are highly homologous to the OmpR EnvZ two component system. This implies that PrpA, which in vitro exhibits very good serine and tyrosine phosphatase activities, is behaving in vivo as a good histidine and/or aspartyl phosphatase. Interestingly, the homologue of PrpA in bacteriophage λ has been shown to be an efficient histidine phosphatase in vitro using P∼NRII or P∼CheA as the substrates (Zhuo et al., 1993). In Bacillus subtilis, the RapA and RapB proteins have been shown to be aspartyl‐phosphatases of the P∼Spo0F response regulator (Perego et al., 1996). Our observations show that decreasing the phosphorylated status of the Cpx system was important to obtain maximal transcription of htrA, in cases when the HtrA protease was highly required in the periplasm. Clearly, the phosphatase activity of PrpA and to some extent PrpB are important to fine‐tune the efficient transcription at the htrA promoter, possibly in two ways: (i) by allowing a faster turnover at the promoter since the main transcription, will yet come from the EσE transcriptional machinery; (ii) by preventing the binding of P∼CpxR to low affinity repression sites which may block the transcriptional process. It is very well documented in the case of OmpR, the regulatory protein most related to CpxR, that excess of P∼OmpR in the cell switches its function from activator to repressor (Rampersaud et al., 1994; Harlocker et al., 1995).
PrpA modulates the phosphorylated state of multiple two component systems as well as affecting the accumulation of major heat shock proteins
The E.coli protein phosphatases also affected other two component signal transduction systems. In the case of the RcsB RscC system, PrpA overexpression seemed to decrease the signal transduction process presumably by dephosphorylating P∼RcsC or P∼RcsB and thereby limiting DNA binding. It seems that like the λ‐PP, PrpA has a very general phosphatase activity with no strict substrate specificity. This might be of some advantage for the intruding bacteriophage. However, in E.coli, non‐specific dephosphorylation by PrpA may need to be curtailed. It is therefore interesting that transcription of prpA is quite weak, leading to very little accumulation of the protein in the cell. Such a situation changes upon heat shock since prpA transcription is induced.
Based on the observation that prpA overexpression leads to an overall altered protein profile as compared with the isogenic wild type (Figure 11A), and induces the accumulation of heat shock proteins at elevated levels (Figure 11A and Table V), it is tempting to speculate that the phosphorylated species of DnaK and GroEL (MacCarty and Walker, 1991; Sherman and Goldberg, 1992) might be acting directly as cellular thermometers by sensing temperature upshifts. Our findings that overexpression of PrpA triggers a heat shock response are consistent with a model that postulates that the unphosphorylated forms of chaperones do not bind the substrates very well (Sherman and Goldberg, 1992). Indeed, such a situation will lead to the accumulation of misfolded proteins and thereby trigger the σ32‐dependent heat shock response. It is equally interesting that in the case of PrpA overproduction, many proteins accumulate at reduced rates as compared with the isogenic wild type (Figure 11A).
It remains to be understood why protein misfolding occurring in the extracytoplasmic compartments signals a requirement for more HtrA protein using at least two different transcriptional regulation pathways. What makes HtrA so important? And what in the periplasm exactly triggers this requirement for HtrA? One possibility could be that the levels of HtrA are sensed directly by CpxA and by the σE regulon via the RseA protein (Missiakas et al., 1996a). RseA is located in the inner membrane and modulates σE activity by a direct protein–protein interaction (RseA–σE) which is presumably relieved upon extracytoplasmic stimuli (Missiakas et al., 1997). The only argument against sensing HtrA levels is that deleting the htrA gene alone induces neither the Cpx Prp nor the σE regulons (Raina et al., 1995), although this would mimic an extreme condition where HtrA is completely titrated out upon accumulation of too many misfolded proteins. Hence the ‘trigger(s)’ responsible for signaling extracytoplasmic stresses remain(s) to be identified. Interestingly, this type of intercompartmental signaling between periplasm and cytoplasm is also found in yeast. A membrane sensor protein, very similar to the bacterial histidine kinases such as CpxA, was found to play an important role in signaling protein misfolding occurring in the endoplasmic reticulum to the transcriptional machinery (Mori et al., 1993; Cox et al., 1993). The various levels of control which might modulate the activity of such a system are not known. It would be interesting to know if these other elements are similar to those found in E.coli.
Materials and methods
Bacterial strains and plasmids
The bacterial strains and plasmids used in this study are listed in Table VI.
Media and chemicals
Luria‐Bertani (LB) broth, MacConkey medium and M9 minimal medium were prepared as described by Miller (1992). Labeling experiments using [35S]methionine in the M9 high‐sulfur medium were performed as previously described (Missiakas et al., 1993). When necessary, the media were supplemented with ampicillin (100 μg/ml), tetracycline (15 μg/ml), kanamycin (50 μg/ml) or chloramphenicol (20 μg/ml). The indicator dye 5‐bromo‐4‐chloro‐3‐indolyl‐β‐d‐galactoside (X‐gal) was used at a final concentration of 40 μg/ml in the agar medium.
Isolation of trans‐acting mutations
Trans‐acting mutations which reduce β‐galactosidase expression from the htrA–lacZ transcriptional fusion exclusively, were isolated through two independent genetic approaches as described earlier (Raina et al., 1995). Briefly, P1 bacteriophage lysates of strain MC4100 (Lac−) carrying random insertions of either mini‐Tn10 (TetR) or mini‐Tn10 (KanR) transposons, were treated with the hydroxylamine mutagen (Miller, 1992). These mini‐Tn10 marked putative mutants were transduced into strains SR1458 and SR1710 carrying the single copy fusion htrA–lacZ and rpoHP–lacZ respectively. Alternatively, a mutD mutation was transduced into MC4100, and such a strain was used as a host to construct mutagenized mini‐Tn10 libraries. These mutagenized pools were transduced into SR1458 (htrA–lacZ) and SR1710 (rpoHP3–lacZ), and again screened on X‐gal‐containing plates for reduced β‐galactosidase activity. Bona fide candidates were also assayed in culture for β‐galactosidase activity, as described by Miller (1992). Two classes of mutations could be distinguished. The first ones affected transcription of both fusions htrA–lacZ and rpoHP3–lacZ and have already been described (Raina et al., 1995). The second class affected transcription of htrA–lacZ and not rpoHP3–lacZ. Three such complementation groups were obtained and a representative of each group is reported here with strain numbers: SR1572, SR1599 and SR2522. To map these new classes of mutations, complementing cosmid clones (selected from a cosmid library described by Raina et al., 1995) were identified because they rescued the Lac‐down phenotype and were able to recombine the linked TetR or KanR markers. DNAs from these cosmid clones were prepared for further subcloning experiments and for mapping to the E.coli DNA library in bacteriophage λ (Kohara et al., 1987) using 32P‐labeled random‐priming techniques (Sambrook et al., 1989).
Cloning of prp genes and ORF221 from λnin region
Chromosomal DNA isolated from the E.coli wild‐type strain MC4100 was used to construct a library in the p15A‐based vector pOK12 (Vieira and Messing, 1991), as described previously (Raina et al., 1995). This library was used to transform strain SR1458 carrying the single copy fusion htrA–lacZ. Transformants with an increased LacZ activity were selected on plates containing X‐gal. DNA was extracted from ∼50 such clones and transformed again in SR1458 (htrA–lacZ) and SR1710 (rpoHP3–lacZ). Clones breeding true were retained. Among them, 12 were found to induce the lacZ expression of htrA–lacZ but not of rpoHP3–lacZ. These clones were also found to revert the decreased β‐galactosidase activity from htrA–lacZ observed with one of the trans‐acting mutations isolated in strain SR1572. These clones were used for mapping experiments and 32P‐labeled nick‐translation (Sambrook et al., 1989). They hybridized to bacteriophage λ336 of the ordered E.coli genomic library (Kohara et al., 1987) and were shown to carry the wild‐type prpA gene.
The cosmid DNA complementing the mutation carried in SR1599 was shown to hybridize to bacteriophage λ449 and λ450 of the ordered E.coli genomic library (Kohara et al., 1987). Partial digestion with Sau3A and ligation in the BamHI site of the pOK12 vector (pDM1754) led to the identification of a 2.2 kb DNA fragment able to complement SR1599 mutant bacteria. A minimal clone (pDM1755) carrying a 1.3 kb DNA fragment was generated using exonuclease III which was sufficient to complement prpB mutant bacteria and used for sequencing. It was found to contain an ORF which corresponds to the gene designated as prpB.
Overexpression of PrpA was achieved by amplifying the minimal coding region of the prpA gene by PCR, using primers 5′‐AGGAAAATACATATGAAACAGGCT‐3′ and 5′‐GCGGTTGGATCCGCATTGAGG‐3′. The resulting amplified DNA product was cloned into the T7 promoter expression vector pAED‐4 (pDM1574 prpA+). The pAED‐4 vector was a kind gift of Dr S.Doering. The minimal prpB coding sequence was amplified by PCR, using the primers 5′‐gtaaaaccatggcatcta‐3′ and 5′‐taacaccggatccctcatgct‐3′. The PCR product was digested with NcoI and BamHI and cloned into pSE420 Invitrogen vector (pDM1757 prpB+) which contains the trc promoter and the lacIq repressor. The λ‐PP gene was amplified from DNA prepared from wild‐type bacteriophage λ, using 5′‐GTGAAACATATGCGCTAT‐3′ and 5′‐CGCTTTGGATCCTCATGCGCC‐3′ as primers. The resulting amplified DNA was cloned into vector pAED‐4 (pSR3040 λnin ORF221). All the PCR‐generated clones coding for PrpA, PrpB and λ‐PP were sequenced and shown not to carry any mutation. The cpxR cpxA operon was subcloned from the DNA of bacteriophage λ541 of the Kohara library (Kohara et al., 1986). A 2.4 kb NdeI–StuI DNA fragment carrying the whole operon was blunted and ligated into the EcoRV site of the pWSK30 vector (Wang and Kushner, 1991). The resulting plasmid pDM1787 was retained since it carried the operon in frame with the T7 RNA polymerase‐dependent promoter.
Construction of the chromosomal cpxA* allele
Site‐directed mutagenesis was used to replace Thr252 by an Arg using plasmid pDM1786. This change was based on the known allele of envZ11 which is known to confer a hyper‐kinase activity to the histidine kinase due to a loss of the autophosphatase activity (Aiba et al., 1989). Mutagenesis was performed using Quick change site‐directed mutagenesis kit from Stratagene and mutagenic primers 5′‐CACGAGCTGCGCCGCCCGCTGACGCGT‐3′ and 5′‐ACGCGTCAGCGGGCGGCGCAGCTCGTG‐3′. Replacement of the wild‐type cpxA gene for this mutant allele was performed by subcloning the mutant cpxA gene into the phagemid vector pBIP (Slater and Maurer, 1993) and as described earlier (Raina et al., 1995). Sucrose‐resistant and ampicillin‐sensitive colonies were retained. Loss of the wild‐type cpxA gene and replacement by the cpxA* allele was verified by scoring for resistance to 15 μg/ml amikacin (Weber and Silverman, 1988) and linkage to the marker clpQ::ΩCm (Missiakas et al., 1996c).
Disruption of the prpA and prpB genes
To construct a null allele of the prpA gene, an ΩCm cassette (Fellay et al., 1987) previously digested at BamHI and blunted, was introduced into the unique PstI site (also blunted using T4 DNA polymerase prior to ligation) of prpA coding region as shown in Figure 2. This site is located before the active site of the mature protein. An ΩTet cassette (Fellay et al., 1987) previously digested at SmaI, was introduced into the unique BstBI site of prpB coding region using plasmid pDM1754 (Figure 2). The BstBI site in prpB was blunted using T4 DNA polymerase prior to ligation. Transfer of these null alleles onto the chromosome was performed as described previously (Raina and Georgopoulos, 1990).
RNA isolation, Northern blot analysis and mapping of 5′ termini
Total cellular RNA was isolated by using the hot SDS–phenol extraction procedure (Sambrook et al., 1989). To define the transcriptional start site(s) of the prpA gene, ∼10 ng of an oligonucleotide probe 5′‐CCCGCAATTCTCTGATAAACG‐3′, which is complementary to nt positions 17–38 of the prpA sequence, was annealed to 10 μg of total cellular RNA. The annealed primer was extended by AMV reverse transcriptase (Promega), essentially as previously described (Raina and Georgopoulos, 1990). The primer extension products were electrophoresed on the same gel as the dideoxy sequencing reactions, using the same primer.
Purifications of RNAP core, σ70 and σ32 have been described previously. Run‐off transcription experiments were performed as described earlier (Raina et al., 1995). The template used was a linear DNA Sau3A–PstI DNA fragment of 659 bp from pDM1695 which contains the promoter region of prpA. For the S1 nuclease protection experiments, the same DNA fragment was used as a probe.
For Northern blot analysis of dnaK‐specific message, probes were made by the random priming technique (Sambrook et al., 1989), using a 508 bp EcoRI–NruI fragment prepared from plasmid pDM38 (dnaK+ dnaJ+; Missiakas et al., 1993), and by radiolabeling with [32P]dCTP (3000 Ci/mmol). Aliquots of 5 μg of RNA isolated from isogenic bacteria carrying either vector pOK12 alone or plasmid pDM1695 (prpA+) at 30°C were used for the Northern blot analyses.
Purification of proteins and immunoprecipitation
Escherichia coli bacteria carrying plasmid pDM1574 (prpA+) or pDM1757 (prpB+) or pSR3040 (λnin ORF221) were induced with 5 mM IPTG at an OD of 0.2 at 600 nm for 5 h. All purification steps were performed at 4°C. Cells were resuspended in buffer A [50 mM Tris–HCl, pH 7.8, 5 mM MnCl2, 5 mM DTT, 0.05 M NaCl, 20% (v/v) glycerol] and lysed by sonication.
PrpA protein pelleted with the membrane fraction
This pellet was resuspended in buffer A with 0.1% Triton X‐100 and spun at 15 000 g for 45 min, at 4°C. Proteins from the soluble fraction were discarded and the pellet containing aggregated PrpA (∼95% of the proteins) was dissolved in buffer A containing 3 M guanidium hydrochloride. Renaturation was done by dilution (20‐fold) in buffer A containing 50% glycerol. This solution was spun at 15 000 g for 45 min at 4°C to remove the insoluble particles and loaded onto a Q‐Sepharose column. Native PrpA protein was eluted with a linear NaCl gradient (0.1–0.6 M) at a concentration of ∼0.15 M NaCl.
PrpB and λ‐PP proteins were recovered from the soluble fraction of proteins in buffer A (after sonication of cells and centrifugation 15 000 g, 45 min, 4°C) and loaded onto Q‐Sepharose. Pools containing each of the proteins were further purified onto phenyl–Sepharose column equilibrated in buffer A containing 0.6 M NaCl (without MnCl2). The column was washed and proteins were eluted using a linear gradient of NaCl (0.6–0.01 M NaCl).
Fractions containing purified proteins, as judged by Coomassie Brilliant Blue‐stained SDS–PAGE, were pooled, dialyzed against buffer A and used directly for biochemical assays. 100% activity as depicted in Table II refers to specific pNPP phosphatase activities for λ‐PP, PrpA and PrpB after purification, i.e. 3500, 3900 and 3800 units/mg, respectively. Protein concentrations were estimated by using the Bradford assay (Bio‐Rad). In the case of PrpA, this measured value correlated well with the calculated extinction molar coefficient of 61 500/M/cm as determined from the absorbance spectrum of the fully unfolded protein.
For the immunoprecipitation experiment, after labeling cells were harvested (0.5 units at OD595 nm) and resuspended in 10 mM Tris–HCl pH 8, containing 1 mM EDTA and 2% Triton X‐100. After lysis, the supernatants were recovered by centrifugation (18 000 g, 30 min, 4°C) and incubated with 50 ml of Affi‐gel 10 beads (Bio‐Rad) coupled to IgGs purified from an anti‐CpxR serum using DEAE–TrysacrylM chromatography.
Standard pNPPase assays were performed in 50 mM Tris buffer pH 7.8 containing 2 mM MnCl2, 2 mM DTT and 10% glycerol, at 25°C for 10 min (or otherwise indicated). pNPP was used as a substrate at a concentration of 20 mM in a 1 ml reaction and the increase of p–nitrophenol upon addition of phosphatases was monitored at 405 nm on an Uvikon 940 spectrophotometer.
Dephosphorylation of protein substrates was assayed using either phosphorylated casein or myelin basic protein (MyBP)
Phosphorylated [32P]Ser/Thr‐casein and [32P]Tyr‐casein were prepared by phosphorylation of α‐casein (Sigma) using respectively the catalytic subunit of protein kinase A (Sigma) and pp60c‐src tyrosine kinase (Oncogene Science). In each case, a 1 ml reaction mixture was prepared using 3.2 mg/ml casein. Preparation of [32P]Ser/Thr‐casein was performed using 0.5 mM [γ‐32P]ATP with 20 mM DTT, 20 mM Mg‐acetate in Tris buffer 50 mM pH 7.4. Preparation of [32P]Tyr‐casein was performed using 1.5 mM [γ‐32P] ATP, 15 mM DTT, 20 mM MgCl2, 0.015% Brij 35, 0.1 mM EDTA in HEPES buffer 50 mM pH 7.5. Either 7 μg of the catalytic subunit of protein kinase A or 20 units of pp60c‐src were added and each independent reaction was incubated at 30°C for 4 h. Phosphorylated casein was recovered by TCA precipitation and extensively dialyzed against a 50 mM Tris buffer pH 7.8, at 4°C.
Dephosphorylation assays of [32P]casein were performed using 0.1–0.6 mg/ml of protein phosphatases in a 1 ml reaction. 100 μl of the mixture were withdrawn at different times and the reaction was quenched by adding TCA (20%). The pellets were washed thoroughly with 10% TCA, and all the soluble materials were pooled and counted for 32P released in the supernatant. The precipitated TCA material was resuspended in a neutralizing buffer (final pH 7) and counted separately.
Phosphorylated [32P]Ser/Thr‐MyBP and [32P]Tyr‐MyBP were prepared according to the manufacturer's instructions (New England Biolabs) using the catalytic subunit of protein kinase A and Abl protein tyrosine kinase, respectively. Dephosphorylation assays of [32P]MyBP were performed using 0.1–0.6 μg/ml of protein phosphatases in a 0.5 ml reaction. Aliquots of 60 μl of the mixture were withdrawn at different times and the reaction was quenched by adding TCA (20%). Soluble materials were counted for 32P released in the supernatant.
We thank John S.Parkinson for helpful comments on this manuscript. We thank D.Court, E.Lin, P.Silverman, H.Watanabe, S.Gottesman, T.Silhavy and P.Danese, for gift of the strains. We are grateful to C.Georgopoulos for initial support to this work (grant number: FN31‐31129‐91). This work was supported by grants from the Fond National Scientifique Suisse to S.R. and D.M. (FN31‐42429‐94), from the CNRS (3A9027) and the Fondation Recherche Médicale to D.M.
- Copyright © 1997 European Molecular Biology Organization