The heat shock response in Escherichia coli is governed by the concentration of the highly unstable a factor or32. The essential protein HflB (FtsH), known to control proteolysis of the phage AcII protein, also governs o.32 degradation: an HflB-depleted strain accumulated o732 and induced the heat shock response, and the half-life of c732 increased by a factor up to 12 in mutants with reduced HflB function and decreased by a factor of 1.8 in a strain overexpressing HflB. The hflB gene is in the ftsJ-hflB operon, one promoter of which is positively regulated by heat shock and &-32. The AcIll protein, which stabilizes &-32 and Acll, appears to inhibit the HflB-governed protease. The E. coli HflB protein controls the stability of two master regulators, AclI and a32, responsible for the lysis-lysogeny decision of phage A and the heat shock response of the host.Cells protect themselves from thermal stress by induction of a set of proteins, the heat shock proteins, many of which are highly conserved from bacteria to man (1, 2). In Escherichia coli, this universal response is transcriptionally regulated by a special of factor, r32, which associates with RNA polymerase to initiate transcription from promoters containing oX32-specific recognition sequences (3, 4). The key determinant of heat shock regulation is the concentration of or32, a highly unstable protein (4). Its degradation involves the heat shock proteins DnaK, DnaJ, and GrpE, which are thought to chaperone cr32 to a proteolytic system (5-7). The protease responsible for ao32 degradation in E. coli-a central component of heat shock regulation-has not been identified.FtsH, a 70-kDa inner membrane protein (8, 9) essential for bacterial survival (10), was recently shown to be identical to HflB (11), which is involved in proteolysis of the AcMI protein (11, 12). The name HflB, which we now use, reflects the phenotype-"high frequency of lysogeny"-of hflB missense mutants (11,12). AcIl is the master regulator of the A phage lysis-lysogeny decision; high Acll concentrations shortly after infection favor lysogeny, whereas low concentrations favor lysis (12, 13). Since HflB is an essential protein, one can speculate that bacterial viability depends on the degradation (or processing) of one or more bacterial proteins by the HflBgoverned proteolytic pathway. However, until now, no bacterial substrates of this protease activity have been identified.The AcII protein stabilizes AcII (13). It also has been shown to stabilize &r32, thereby inducing the heat shock response (14).It is not known whether the protection afforded by AcIII is via direct interaction with AcIl and (r32 or via inhibition of one or more proteases involved in their degradation. Overproduction of AcIII is lethal in E. coli (15). HflB, an essential protein that governs an anti-AcII protease activity, seemed to be a good candidate for the target of AcIII action. If indeed AcIII inhibits the HflB-governed proteolytic pathway, then the simplest hypothesis to explain ao32 stabilization by AcIII is that...

Proteins with short nonpolar carboxyl termini are unstable in Escherichia coli. This proteolytic pathway is used to dispose of polypeptides synthesized from truncated mRNA molecules. Such proteins are tagged with an 11-amino-acid nonpolar destabilizing tail via a mechanism involving the 10Sa (SsrA) stable RNA and then degraded. We show here that the ATP-dependent zinc protease HflB (FtsH) is involved in the degradation of four unstable derivatives of the amino-terminal domain of the cI repressor: three with nonpolar pentapeptide tails (cI104, cI105, cI108) and one with the SsrA tag (cI-SsrA). cI105 and cI-SsrA are also degraded by the ClpP-dependent proteases. Loss of ClpP can be compensated for by overproducing HflB. In an in vitro system, cI108 and cI-SsrA are degraded by HflB in an energy-dependent reaction, indicating that HflB itself recognizes the carboxyl terminus. These results establish a tail-specific pathway for removing abnormal cytoplasmic proteins via the HflB and Clp proteases.[Key Words: 10Sa RNA; AAA ATPase family; intracellular proteolysis; tail-specific proteolysis; Clp protease; proteasome 26S] Received February 6, 1998; revised version accepted March 12, 1998. Living cells have elaborate mechanisms such as check points and coupling devices to evaluate their physiological state and environment to maintain a harmonious cell cycle. In addition, cell growth requires efficient systems to clean up erroneous metabolites and macromolecules that would otherwise accumulate in the cytoplasm and ultimately arrest growth. Abnormal proteins are a frequent and cumbersome type of cytoplasmic junk, and ingenious proteolytic systems have evolved to recognize and remove them. Abnormal proteins can arise in various ways: from denaturing treatments, from improper folding of newly synthesized polypeptides (often the case for foreign proteins), and from premature termination of transcription or translation (Gottesman 1996;Miller 1996).Cytoplasmic proteolysis serves several purposes (Gottesman 1996; Herman and D'Ari 1998). It permits rapid regulatory responses to specific signals, it creates protein turnover in times of starvation, and it removes abnormal proteins. The latter housekeeping role for cytoplasmic proteolysis is a major cell need, and inability to perform it quickly perturbs growth.Regulatory proteolysis is generally quite specific, each unstable regulator being the preferential substrate of a particular protease, whereas in housekeeping proteolysis, abnormal proteins are often substrates of several proteases (Maurizi et al. 1985). Such lack of specificity is perhaps not surprising, since virtually any protein can become ''abnormal'' by denaturation, in which case recognition presumably exploits general criteria, shared by most proteases. However, the specific features of proteins that target them for degradation remain largely unknown.In a systematic search for protease recognition signals in Escherichia coli, Parsell et al. (1990) showed that a stable protein can be destabilized by addition of nonp...

Aminoacyl-tRNA synthetase mutants of Escherichia coli are resistant to amdinocillin (mecillinam), a P-lactam antibiotic which specifically binds penicillin-binding protein 2 (PBP2) and prevents cell wall elongation with concomitant cell death. The leuS(Ts) strain, in which leucyl-tRNA synthetase is temperature sensitive, was resistant to amdinocillin at 37°C because of an increased guanosine 5'-diphosphate 3'-diphosphate (ppGpp) pool resulting from partial induction of the stringent response, but it was sensitive to amdinocillin at 25°C. We constructed a leuS(Ts) A(rod4-pbpA)::KMr strain, in which the PBP2 structural gene is deleted. This strain grew as spherical cells at 37°C but was not viable at 25°C. After a shift from 37 to 25°C, the ppGpp pool decreased and cell division was inhibited; the cells slowly carried out a single division, increased considerably in volume, and gradually lost viability. The cell division inhibition was reversible when the ppGpp pool increased at high temperature, but reversion required de novo protein synthesis, possibly of septation proteins. The multicopy plasmid pZAQ, overproducing the septation proteins FtsZ, FtsA, and FtsQ, conferred amdinocillin resistance on a wild-type strain and suppressed the cell division inhibition in the leuS(Ts) A(rodA-pbpA)::KMr strain at 25°C. The plasmid pAQ, in which theftsZ gene is inactivated, did not confer amdinocillin resistance. These results lead us to hypothesize that the nucleotide ppGpp activatesftsZ expression and thus couples cell division to protein synthesis.

Escherichia coli strains partially induced for the stringent response are resistant to mecillinam, a beta-lactam antibiotic which specifically inactivates penicillin-binding protein 2, the key enzyme determining cell shape. We present evidence that mecillinam resistance occurs whenever the intracellular concentration of the nucleotide ppGpp (guanosine 3'-diphosphate 5'-diphosphate), the effector of the stringent response, exceeds a threshold level. First, the ppGpp concentration was higher in a mecillinam-resistant mutant than in closely related sensitive strains. Second, the ppGpp pool was controlled by means of a plasmid carrying a ptac-relA' gene coding for a hyperactive (p)ppGpp synthetase, RelA'; increasing the ppGpp pool by varying the concentration of lac operon inducer IPTG resulted in a sharp threshold ppGpp concentration, above which cells were mecillinam resistant. Third, the ppGpp pool was increased by using poor media; again, at the lowest growth rate studied, the cells were mecillinam resistant. In all experiments, cells with a ppGpp concentration above 140 pmoles/A600 were mecillinam resistant whereas those with lower concentrations were sensitive. We discuss a possible role for ppGpp as transcriptional activator of cell division genes whose products seem to become limiting in the presence of mecillinam, when cells form large spheres. We confirmed the well-known inverse correlation between growth rate and ppGpp concentration but, surprisingly, for a given growth rate, the ppGpp concentration was lower in poor medium than in richer medium in which RelA' is induced. We conclude that, for E. coli growing in poor media, the concentration of the nucleotide ppGpp is not the major growth rate determinant.