HtrA is the unique surface housekeeping protease in Lactococcus lactis and is required for natural protein processing



We identified an exported protease in Lactococcus lactis ssp. lactis strain IL1403 belonging to the HtrA/DegP family. Inactivation of the chromosomal gene (htrALl) encoding this protease (HtrALl) results in growth thermo-sensitivity at very high temperatures (above 37°C for L. lactis). The role of htrALl in extracellular proteolysis under normal growth conditions was examined by testing the stability of different exported proteins (i.e. fusions, a heterologous pre-pro-protein or a native protein containing repeats), having different locations. In the wild-type (wt) strain, degradation products, including the C-terminal protein ends, were present in the medium, indicating that proteolysis occurs during or after export to the cell surface; in one case, degradation was nearly total. In contrast, proteolysis was totally abolished in the htrA strain for all five proteins tested, and the yield of full-length products was significantly increased. These results suggest that HtrALl is the sole extracellular protease that degrades abnormal exported proteins. In addition, our results reveal that HtrALl is needed for the pro-peptide processing of a natural pro-protein and for maturation of a native protein. We propose that in lactococci, and possibly in other Gram-positive organisms with small sized-genomes, a single surface protease, HtrA, is totally responsible for the housekeeping of exported proteins.


Proteolysis in bacteria is a widely studied phenomenon that ensures correct functioning of the cell. It is involved in essential functions such as native protein turn-over and recycling, protein activation and abnormal protein degradation, as well as nutrient recognition and degradation into peptides and amino acids that can be assimilated. Proteolysis is thereby involved in diverse processes, including nutrition and growth, protein export and secretion, stress resistance and survival, and regulation (Visick and Clarke, 1995; Gottesman, 1996; Gottesman et al., 1997). Proteolysis is ensured in the different cell compartments by the presence of differentially localized proteases. Escherichia coli and Bacillus subtilis are known to possess numerous exported proteases (Pero and Sloma, 1993; Wülfing and Plückthun, 1994; Gottesman 1996; Margot and Karamata, 1996).

The first and best studied envelope protease that has a confirmed role in housekeeping is the E. coli HtrA/DegP/Do protease. This periplasmic protein is a trypsin-like serine protease that at very high temperatures (above 42°C) is induced and essential for growth (Lipinska et al., 1988, 1989, 1990; Strauch and Beckwith, 1988; Strauch et al., 1989; Pallen and Wren, 1997). The various studies on E. coli HtrA/DegP have shown that it is involved in the degradation of exported proteins that are misfolded or aggregated (Strauch and Beckwith, 1988; Strauch et al., 1989; Lipinska et al., 1990; Pallen and Wren, 1997). It has recently been shown that E. coli HtrA can display a dual function of both chaperone and protease. Using a misfolded form of the periplasmic α-amylase MalS as a model, Spiess et al. (1999) showed that, at low temperatures, HtrA chaperone function dominates, leading to stimulation of MalS folding, whereas at normal and high temperatures, HtrA degrades misfolded MalS. Finally, HtrA function is probably important for prokaryotic life, as several copies of the broadly conserved htrA genes are often present in bacteria (Pallen and Wren, 1997; our blast analyses of completely sequenced genomes). For example, E. coli possesses three exported serine proteases of the HtrA family, HtrA/DegP, HhoA/DegQ and HhoB/DegS (Waller and Sauer, 1996), and B. subtilis contains two putative genes homologous to E. coli htrA, yyxA and ykdA (Pallen and Wren, 1997).

For Lactococcus lactis, a model lactic acid bacterium that is widely used in the dairy industry, proteolysis is essential for bacterial growth and plays a key role in determining dairy product quality. Several studies have shown that L. lactis contains a complex proteolytic system, involving several intracellular peptidases and a single cell surface-anchored protease, PrtP. PrtP, a plasmid-encoded serine protease, provides a nitrogen source for cell growth by degrading milk caseins (Pritchard and Coolbear, 1993; Kunji et al., 1996).

PrtP was routinely claimed to be the only general cell surface protease in L. lactis, probably because of the focus of research studies on nitrogenous nutrition and the failure to purify any other surface protease (note that some strains of L. lactis possess another surface protease, NisP which is transposon-encoded; NisP is specifically needed for processing the nisin precursor after its secretion, to release the fully matured and active nisin bacteriocin; van der Meer et al., 1993). Other surface proteases have recently been identified in distantly related lactic acid bacteria; a metallo-protease has been purified from cell wall fractions of Lactobacillus delbrueckeii ssp. bulgaricus (Stefanitsi and Garel, 1997) and, recently, an htrA homologue was found in Lactobacillus helveticus and shown to be stress-inducible (Smeds et al., 1998).

Recent observations in our laboratory suggested that an L. lactis strain (MG1363), which is devoid of the PrtP surface protease, still has extracellular proteolytic activity. We previously generated a collection of exported lactococcal fusion proteins having an export-specific reporter at the C-terminus. The reporter, ΔSPNuc, was developed from the Staphylococcus aureus secreted nuclease (Nuc) for applications in Gram-positive bacteria (Poquet et al., 1998). Interestingly, many of the fusions produced in MG1363 were at least partially degraded during or after export (Poquet et al., 1998; unpublished data). As the E. coli DegP/HtrA protease was previously shown to degrade fusions between E. coli envelope proteins and the export-specific reporter PhoA (Strauch and Beckwith, 1988), we suspected that the unknown lactococcal protease responsible for degradation of the ΔSPNuc fusion proteins could belong to the HtrA family.

In this study, we made use of the recently completed diagnostic sequence of L. lactis ssp. lactis strain IL1403 (Bolotin et al., 1999) to find a homologue of the E. coli htrA gene. Gene inactivation showed that this lactococcal homologue, htrALl, is essential for growth at very high temperatures and is involved in surface proteolysis. Our results show that, under normal growth conditions, HtrALl is involved in abnormal protein degradation. They also reveal a previously unreported role of HtrA in pro-peptide processing as well as in native protein maturation. Disruption of the lactococcal htrALl gene results in full stabilization of all tested exported proteins. We have thus identified a new and unique surface housekeeping protease in L. lactis that appears to be responsible for both degradation and maturation of exported proteins in L. lactis.


Identification and inactivation of L. lactis ssp. lactis htrALl gene

We hypothesized that the observed degradation of different exported fusion proteins in L. lactis strain MG1363 (Poquet et al., 1998; and not shown) could be mediated by an HtrA homologue. A conserved signature domain of the HtrA protease family, corresponding to the consensus motif around the catalytic serine residue (Pallen and Wren, 1997), was used to screen the possible open reading frames (ORFs) of the genome diagnostic sequence of L. lactis ssp. lactis strain IL1403 (Bolotin et al., 1999), a close relative of L. lactis ssp. cremoris strain MG1363. A single putative htrALl gene was identified (GenBank accession number AF155705) and found to be the only L. lactisIL1403 homologue of E. coli htrA/degP, hhoA/degQ and hhoB/degS genes.

The putative htrALl gene encodes a 408-amino-acid (aa) protein HtrALl. A putative ρ independent terminator of transcription is predicted by the GCG terminator program (AAAAGTCTTCTGTAAATAGAAGGCTTTT; complementary nucleotides are in bold). HtrALl is homologous to E. coli HtrA/DegP/Do, HhoA/DegQ and HhoB/DegS proteases (34% identity in each case), B. subtilis ORFs, YyxA (45% identity) and YkdA (46% identity), Lb. helveticus HtrA (47% identity) and Streptococcus pneumoniae ORF, spHtrA (56% identity). HtrALl appears to be a typical HtrA trypsin-like serine protease as it contains the catalytic triad Ser (S239), His (H127) and Asp (D157), embedded within three consensus motifs, DAYVVTNYH127VI, D157LAVLKIS, and GNS239GGALINIEGQVIGIT (conserved residues are in bold; Pallen and Wren, 1997). The N-terminal HtrALl sequence is hydrophobic and is predicted by the psort program (http://psort. to form a transmembrane segment (L9LLTGVVGGAIALGGSAI26). The positive-inside rule for integral membrane protein topology (von Heijne, 1989) predicts that HtrALl is an N-in, C-out membrane protein, with the bulk of it, including the catalytic site, exposed to the cell surface. This disposition suggests that HtrALl degrades exported proteins. HtrALl, like other HtrA proteins from Gram-positive bacteria, contains a single C-terminal PDZ domain (aa positions 285–383). In E. coli HtrA, PDZ domains were recently shown to be responsible for protein oligomerization (Sassoon et al., 1999), although a role in substrate binding cannot be excluded (Pallen and Wren, 1997; Sassoon et al., 1999).

The L. lactis ssp. lactis strain IL1403 htrALl gene was inactivated by single cross-over homologous recombination into the chromosome, using a chloramphenicol-resistant (CmR) non-replicative plasmid (pJIM2481) containing an internal htrALl fragment. The resultant htrA strain produces a truncated HtrALl protein that is missing the catalytic serine. As a positive control for integration, single cross-over integration was also performed, using a fragment corresponding to the 5′ end of the gene, including expression signals. In the resulting strain (referred to as htrA+/htrA), both wild-type (wt) and truncated gene copies are present. Strains were characterized phenotypically for growth and proteolysis.

Thermo-sensitivity of L. lactis ssp. lactis htrA strain

Growth of the htrA strain was compared with that of the htrA+/htrA strain and of two control strains, the wt IL1403 strain and a CmR derivative, IL6179 (Schouler et al., 1998). At 30°C, the usual growth temperature for lactococci, growth rates in liquid for all the strains were comparable. On plates, the htrA strain colonies were slightly smaller and, in the absence of selection, a few large chloramphenicol-sensitive (CmS) colonies, probably wt revertants, appeared. More pronounced differences in growth and viability of htrA compared with the other strains were observed in cultures maintained at or above 37°C. The htrA colonies on plates incubated at 37°C were tiny and large CmS-revertant colonies appeared in the absence of selection. Upon a shift-up to 39°C, growth in liquid cultures of the htrA strain was rapidly arrested, whereas growth of the other strains continued exponentially, regardless of whether chloramphenicol (Cm) selection was imposed (not shown) or not (Fig. 1). Viability was markedly affected in the case of htrA; 102- and 104-fold drops in viable CmR cells were observed after incubation at 39°C for 5 h and 22 h respectively, compared with only 101- and 102-fold respective drops for the other strains. After prolonged incubation (22 h) at 39°C in the absence of selection, CmS revertants outgrew the htrA cultures (not shown; note that even if the reversion frequency remains the same whatever the temperature, revertants outgrowth at 39°C increases the apparent reversion frequency). In all tests, properties of the htrA+/htrA strain were identical to those of the wt strain.

Figure 1.

Thermosensitivity of the htrA mutant strain. Overnight cultures of wt L. lactis ssp. lactis IL1403 (CmS, squares), and the mutant htrA+/htrA (CmR, circles) and htrA (CmR, triangles) strains were grown at 30°C without (for IL1403) or with Cm (for mutants). Cultures were diluted 200-, 200-, and 100-fold respectively, in medium without Cm, and grown at 30°C. At an OD600 of 0.1, cultures were split; one part was maintained at 30°C (closed symbols), and the other was grown at 39°C (open symbols). Viability of strains was evaluated at different time points by plating dilutions of the strains on plates without (for IL1403) or with selection (for mutants). Similar results were obtained when cultures were maintained in Cm (not shown).

The above results demonstrate that inactivation of htrALl results in growth thermo-sensitivity. They are consistent with a role for htrALl in heat shock response and in survival at high temperature.

Exported fusion proteins are stabilized in the L. lactis ssp. lactis htrA strain

Cell surface proteolysis of abnormal proteins was examined in the htrA strain. A collection of fusions between L. lactis ssp. cremoris MG1363 exported proteins and the ΔSPNuc reporter had been generated previously (Poquet et al., 1998). Several of the fusions were found to be degraded in the wt strain, MG1363 (Poquet et al., 1998; and not shown). Three fusions (see Experimental procedures for descriptions), having different cellular locations and exhibiting instability in MG1363, were selected to test the effects of htrALl inactivation: (i) the secreted protein Usp–ΔSPNuc (ii) the lipoprotein Nlp4–ΔSPNuc, and (iii) a large tripartite fusion, Exp5–ΔSPNuc, which comprises the N-terminal end of a penicillin-binding protein homologue and an internal segment of a cytoplasmic protein homologue fused to the reporter. This fusion, which allows export of the cytoplasmic segment, was found to be highly degraded in MG1363 (Poquet et al., 1998; and not shown).

Plasmids encoding the above fusions (respectively pVE8009, pVE8024 and pVE8021) were established in the htrA strain and in the isogenic wt IL1403 and htrA+/htrA strains, and cultures were grown to late exponential phase. Protein extracts from total cultures or from cell and medium fractions were analysed by Western blotting using polyclonal NucA antibodies (the NucA moiety is present at the C-terminal reporter end of protein fusions; Poquet et al., 1998). Note that, owing to the nature of these antibodies, only degradation products including the C-terminal protein ends can be detected.

We noted that for all fusion proteins, Western blot profiles were identical in the wt IL1403 and htrA+/htrA strains (Fig. 2), suggesting that the presence of a truncated HtrALl protein copy did not interfere with wt HtrALl activity. For wt IL1403 and htrA+/htrA strains producing the secreted Usp–ΔSPNuc (Fig. 2A) and the lipoprotein Nlp4–ΔSPNuc (Fig. 2B), three bands appeared on the Western blot, corresponding in size to the precursor, the mature form and a NucA-sized degradation product. The tripartite fusion, Exp5–ΔSPNuc (Fig. 2C), was barely detectable in the wt IL1403 and htrA+/htrA strains. Two very faint bands could be distinguished, one of high molecular weight compatible with the size of the full length protein, and the other corresponding to a NucA degradation product. It appears that Exp5–ΔSPNuc undergoes nearly total degradation in IL1403.

Figure 2.

Western blotting of ΔSPNuc fusion proteins in WT and mutant htrA contexts. Protein extracts of late exponential phase cultures (OD600 of 0.6–0.8) of IL1403, htrA and htrA+/htrA strains, expressing the exported fusion proteins, were submitted to SDS–PAGE and Western blotting, using polyclonal antibodies against commercial S. aureus NucA. For each panel, commercial NucA control is in the right-hand lane.

A. Usp–ΔSPNuc. Cultures of IL1403, htrA and htrA+/htrA strains producing Usp–ΔSPNuc were fractionated. T, total fraction; C, cells fraction; and M, medium fraction. Subcellular fractionation showed the expected locations for the three detected bands: (i) pre. is the putative precursor found in the cells fraction; (ii) mat. is the putative mature form secreted into the medium after signal peptide cleavage; and (iii) NucA corresponds to the NucA-sized degradation products that are both secreted and cell associated, as previously observed (Poquet et al., 1998). We note that samples corresponding to the htrA strain are underloaded compared with the other samples presented, so that protein stabilization in the htrA strain is less obvious for Usp–ΔSPNuc than for Nlp4–ΔSPNuc (Fig. 2B) and for Exp5–ΔSPNuc (Fig. 2C).

B. Nlp4–ΔSPNuc. Total protein extracts (cells plus medium) of cultures of IL1403, htrA and htrA+/htrA strains producing Nlp4–ΔSPNuc were prepared. The three detected bands correspond to: (i) pre., the putative precursor in the cells fraction; (ii) mat., the putative mature form secreted into the medium after signal peptide cleavage; and (iii) NucA, the NucA-sized degradation products.

C. Exp5–ΔSPNuc. Total protein extracts (cells plus medium) of cultures of IL1403, htrA and htrA+/htrA strains producing Exp5–ΔSPNuc were prepared. The two detected bands correspond to: (i) int., the putative intact form; and (ii) NucA, the NucA-sized degradation products.

In sharp contrast, for each of the three fusions, the profile observed in the htrA strain was markedly different than that observed in the wt IL1403 or htrA+/htrA strains (Fig. 2). In the htrA strains producing either Usp–ΔSPNuc or Nlp4–ΔSPNuc, only the precursor and the mature form were observed; no NucA degradation products were detected (Fig. 2A and B). In the htrA strain producing Exp5–ΔSPNuc, only the full-sized protein is detected. For all protein fusions, and particularly for Exp5–ΔSPNuc, the amounts of high molecular weight forms were greater in the htrA strain than in the other two strains (Fig. 2).

The above results demonstrate that HtrALl is necessary for the degradation of all three fusions, Usp–ΔSPNuc, Nlp4–ΔSPNuc and Exp5–ΔSPNuc. We therefore propose that HtrALl is the protease that degrades the ΔSPNuc fusions. An alternative possibility is that HtrALl only acts as a chaperone (Spiess et al., 1999) that gives access to an as yet unknown protease. We consider this hypothesis unlikely for a variety of reasons. First, HtrALl has all the characteristics of documented serine proteases of the HtrA family. Second, E. coli HtrA, which has been shown to have a dual function, reportedly acts as a chaperone only at low temperatures (28°C), but as a protease at low, normal and high temperatures (28°C, 37°C and 42°C), and it is inactive in vitro only below 22°C; (Spiess et al., 1999). The experiments in L. lactis were performed at its optimal growth temperature (30°C), conditions under which HtrALl protease is expected to be active, even if it might act as a chaperone at lower temperatures. Third, examination of the full lactococcal genome (Bolotin et al., 1999) did not reveal any potentially exported protease other than HtrALl.

Interestingly, Usp–ΔSPNuc produced in the wt IL1403 and htrA+/htrA strains gave rise to NucA-sized degradation products in the medium fraction (Fig. 2A). This location of NucA-sized products indicates that proteolysis occurred after translocation of the protein across the cytoplasmic membrane and is in agreement with the prediction that HtrALl is located at the cell surface. The absence of NucA-sized degradation products in the cell fraction of the htrA strain is consistent with their location at the cell surface in the wt IL1403 and MG1363 strains, as previously proposed (Liebl et al., 1992; Poquet et al., 1998; see Fig. 2A).

Tests were also performed on two exported heterologous proteins, both of which were stabilized in the htrA mutant. In particular, the Staphylococcus hyicus lipase gave rise to multiple degradation products in the wt IL1403 strain and was stabilized in the htrA mutant (data not shown). Taken together, these results demonstrate that the htrALl mutation results in a spectacular stabilization of exported (cell surface-associated or secreted) proteins of different types (heterologous or fusion proteins), with an increase in the amounts of intact protein. As HtrALl is required to degrade exported abnormal proteins, and as degradation is totally abolished in the htrA mutant, we suggest that HtrALl is the unique housekeeping surface protease in L. lactis.

Exp5–ΔSPNuc is misfolded in the htrA strain

As shown above, Exp5–ΔSPNuc is particularly sensitive to proteolysis compared with the other two ΔSPNuc fusions studied. In the MG1363 strain, previous studies have revealed a degradation pattern of Exp5–ΔSPNuc with numerous intermediate-sized bands, probably a result of cleavage at different sites in the Exp5 part of the fusion (Poquet et al., 1998; and not shown). In IL1403, the faint NucA signal (Fig. 2C) suggested that HtrALl efficiently recognizes at least one site in the NucA part of Exp5–ΔSPNuc, leading to its near complete degradation (cleavage differences in the two subspecies may be due to different activities and/or production levels and/or folding activities of the HtrALI proteases). The near complete degradation of the NucA moiety of Exp5–ΔSPNuc in IL1403 was surprising in view of the well-documented NucA stability (Davis et al., 1977; Miller et al., 1987; Liebl et al., 1992; Suciu and Inouye, 1996; Le Loir et al., 1998; Poquet et al., 1998) and may suggest that the Exp5–ΔSPNuc fusion imposes constraints on the structure of the NucA moiety. In agreement with this hypothesis, we observed that Exp5–ΔSPNuc is unique among the tested fusions as it displays a Nuc phenotype in the htrA strain, but is Nuc+ in both the IL1403 strain and in wt CmS revertants of the htrA strain (not shown). We propose that the Exp5 moiety of Exp5–ΔSPNuc impedes proper folding of NucA and thus causes a Nuc phenotype in the htrA strain, and that in the wt IL1403 strain, the few NucA molecules (Fig. 2C) that remain stable can assume an active conformation.

Taken together, the above results demonstrate that HtrALl is able to cleave a misfolded exported protein at different sites. This is in agreement with its essential role in cell survival at high temperatures, a condition known to favour the accumulation of misfolded/aggregated proteins (Strauch et al., 1989; Lipinska et al., 1990; Pallen and Wren, 1997). Finally, all our results strongly suggest that HtrALl performs a general housekeeping function.

Pro-peptide processing of S. aureus nuclease Nuc is mediated by HtrALl

The S. aureus nuclease is a pre-pro-protein that is secreted into the medium by signal peptide cleavage of the precursor as the active pro-Nuc form (or NucB), which undergoes secondary processing to the final active NucA form (Davis et al., 1977; Shortle, 1983). Pro-peptide processing has been demonstrated in numerous heterologous hosts (Miller et al., 1987; Liebl et al., 1992), including E. coli (Suciu and Inouye, 1996). In L. lactis, Nuc is also secreted as pro-Nuc and secondarily processed into NucA (Le Loir et al., 1998). The protease involved in wt Nuc pro-peptide processing in L. lactis is unknown and we asked whether it could be HtrALl.

The native S. aureus nuc gene (on plasmid pNuc3; Le Loir et al., 1998) was established in wt IL1403, htrA+/htrA and htrA strains. Protein extracts from late exponential phase cultures were analysed using Western blotting (Fig. 3). Four bands were detected in IL1403 and htrA+/htrA strains corresponding to (i) the cellular precursor, (ii) and (iii) a doublet for secreted mature pro-Nuc forms, and (iv) the secondary processed NucA form, which is present in both medium and cell fractions. In contrast, no NucA was detected in the htrA strain, although both precursor and pro-Nuc were present. These results show that HtrALl is involved in wt Nuc pro-peptide processing after its translocation to the cell surface, presumably through its function as a protease.

Figure 3.

Western blotting of Nuc protein in WT and mutant htrA contexts. Late exponential phase (OD600 0.6–0.8) cultures of IL1403, htrA and htrA+/htrA strains producing wt S. aureus Nuc were fractionated and treated as in Fig. 2A. The four detected bands correspond to: (i) pre. (pre-pro-prot.), the putative precursor (pre-pro-protein) in the cells fraction, (ii) and (iii) pro. (pro-prot.), a doublet of the putative mature pro-protein secreted into the medium after signal peptide cleavage (the doublet might be due to alternative processing (Y. Le Loir, personal communication) and is more visible in the htrA strain owing to protein stabilization), and (iv) NucA, the secondary-processed NucA form obtained after pro-peptide processing. Commercial NucA control is in the right-hand lane.

HtrALl is needed for maturation of the native lactococcal autolysin AcmA

The autolysin AcmA, an N-acetyl muramidase, is a native L. lactis exported protein. It undergoes maturation via an unknown protease, giving rise to two active smaller forms (Buist et al., 1995; Chapot-Chartier, 1996). AcmA is exported via a classical signal peptide and is both cell surface associated and secreted into the medium. Surface association is thought to rely on its affinity for peptidoglycan via repeat sequences present at its C-terminal end. The two smaller active forms are presumably devoid of some of the repeats (Buist et al., 1995). We asked whether HtrALl could be involved in this natural processing event.

Protein extracts of late exponential phase cultures of htrA and control strains IL1403 and htrA+/htrA were analysed using PAGE and zymogram (Fig. 4). As previously reported (Buist et al., 1995; Chapot-Chartier, 1996), four bands exhibiting muramidase activity were detected in control strains. Note that all these bands are attributed to AcmA activity, as an MG1363 acmA strain exhibits no lytic activity by zymograms (Buist et al., 1995). In IL1403 and htrA+/htrA strains, the active precursor is exclusively cell associated, whereas the mature form and two active processed products are in both cell and medium fractions. In contrast, the htrA strain contains only the precursor and the full mature forms. These results show that HtrALl is involved in AcmA processing after its translocation to the cell surface, presumably through its function as a protease.

Figure 4.

Zymogram of AcmA protein in WT and mutant htrA contexts. Cultures of IL1403, htrA and htrA+/htrA were fractionated as in Fig. 2 A and submitted to SDS–PAGE on a gel containing autoclaved Micrococcus lysodeikticus cells. AcmA activity was then revealed as described previously (Buist et al., 1995). The four detected bands correspond to: (i) pre., the putative precursor in the cells fraction (ii) mat., the putative mature form obtained after signal peptide cleavage, and (iii) deg., two active degradation products.


This study establishes the existence of a chromosomally encoded housekeeping surface protease, HtrALl, in L. lactis. It thus disproves the accepted supposition that L. lactis possesses only one general surface protease, PrtP, that is plasmid encoded and needed for nitrogen utilization. HtrALl is predicted to be a membrane-bound serine protease, with its catalytic site exposed to the cell surface, thus exposing it to exported substrates. Studies of an htrA mutant strain showed that: (i) htrALl is essential for growth at high temperatures (i.e. above 37°C in L. lactis); and (ii) under normal growth conditions, htrALl is involved in surface proteolysis of abnormal proteins, as shown for three fusion proteins and one heterologous protein having different final extracytoplasmic locations. Thus HtrALl exhibits the functions previously described for the E. coli HtrA (Strauch and Beckwith, 1988; Lipinska et al., 1989; 1990; Strauch et al., 1989; Pallen and Wren, 1997). It is probable that, as seen for the E. coli (Lipinska et al., 1988) and Lb. helveticus (Smeds et al., 1998) htrA genes, htrALl expression will be induced under stress conditions.

Two novel roles for HtrA were discovered in L. lactis. In addition to its role in the elimination of abnormal proteins, we observed that HtrALl proteolytic activity is involved in both pro-peptide processing and in maturation of a native host protein. HtrALl was shown to process the wt S. aureus nuclease pro-Nuc form to the final processed NucA form. It is also required to mature the L. lactis chromosomally encoded muramidase AcmA to two smaller active forms. In E. coli, only two natural HtrA substrates have been described, corresponding to the plasmid-encoded colicin A lysis lipoprotein (Cal; Cavard et al., 1989; Pallen and Wren, 1997) and the periplasmic α-amylase MalS (Spiess et al., 1999). However, in the first case, non-physiological conditions, including drug addition, were needed to reveal Cal degradation by HtrA (Cavard et al., 1989), and in the second case, MalS misfolding was only observed in a dsbA context (Spiess et al., 1999). Our results in L. lactis lead us to suggest that the normal processing/maturation of numerous proteins in different hosts will involve HtrA family members.

How can we reconcile two seemingly different functions for HtrA, the degradation of abnormal proteins, and the maturation or processing of natural proteins? Degradation by HtrA of misfolded or slowly folding proteins, such as fusions or heterologous proteins, or degradation of aggregates under stress conditions (Strauch and Beckwith, 1988; Lipinska et al., 1990; Pallen and Wren, 1997) is probably due to the exposure of regions that are usually buried in the proteins. We suggest that the folding properties of both pro-Nuc and AcmA make them substrates for HtrALl.

(i) Although S. aureus NucA forms a very compact domain, it has been demonstrated that the pro-peptide region, which is not needed to prevent nuclease activity, forms a flexible, and thus accessible, arm (Davis et al., 1977, 1979; Suciu and Inouye, 1996). Previous in vitro studies have shown that the secreted staphylococcal V8 protease (StsP) can process Nuc pro-peptide. However, it is not known whether this protease has a role in this phenomenon in vivo (Davis et al., 1977). Although not part of the HtrA family, the V8 protease does share its homologous catalytic domain with HtrA (Koonin et al., 1997). This similarity, together with our present results, leads us to suggest that a S. aureus HtrA homologue could cleave the Nuc pro-peptide in vivo. Nuc pro-peptide processing has also been observed in other diverse hosts, including E. coli (Suciu and Inouye, 1996), Corynebacterium glutamicum (Liebl et al., 1992) and B. subtilis, in which it was shown to depend on a serine protease activity (Miller et al., 1987). In B. subtilis and L. lactis, the pro-peptide cleavage site was confirmed to be one of the two sites found in the native host producing Nuc, S. aureus (Marrakechi and Cozzone, 1981; Miller et al., 1987; Le Loir et al., 1998). We suggest that the HtrA of the respective hosts may be responsible for Nuc pro-peptide processing.

ii) The processing of AcmA presumably removes C-terminal repeats that are distinct from the catalytic site and that are believed and that are believed to mediate peptidoglycan binding (Buist et al., 1995). It is possible that the repeat region also forms a less compact structure and is thereby more accessible to HtrALl degradation. Note that numerous Gram-positive proteins, in particular pathogenicity determinants and cell wall-binding proteins, contain repeat sequences (Navarre and Schneewind, 1999). Taking into account results with AcmA, it is possible that repeat sequences that are known to be processed may also be substrates for HtrA. Their maturation could possibly affect their pathogenic or binding properties and/or activity. It is tempting to speculate that the reduced virulence of htrA mutants from different species could be due not only to reduced viability under conditions of infection (Johnson et al., 1991; Pallen and Wren, 1997), but also to an inability to process host determinants involved in pathogenicity. For cell wall binding proteins, it is interesting to note that, in L. lactis, AcmA processing results in an altered specificity. The AcmA matured products degrade the cell walls of Micrococcus lysodeikticus but not of L. lactis, whereas full-length AcmA degrades both (Buist et al., 1995; Chapot-Chartier, 1996). This suggests an ecological role for AcmA processing by HtrALl, in which degradation products lacking their C-terminal repeats would be preferentially released into the medium (Buist et al., 1995; Chapot-Chartier, 1996) and active against competing, heterologous cells.

It is remarkable that inactivation of a single gene in L. lactis, htrALl, resulted in total and cumulative stabilization of all the exported proteins tested. In contrast, pulse–chase studies in E. coli showed that htrA inactivation only slows down the kinetics of proteolysis (Strauch and Beckwith, 1988). The partial effects of the htrA mutation in E. coli is attributed to the presence of several other envelope proteases (Strauch and Beckwith, 1988; Wülfing and Plückthun, 1994; Gottesman, 1996). Complex arrays of proteases are also found in numerous other bacteria, using both genetic and sequence analyses (Pero and Sloma, 1993; Wülfing and Plückthun, 1994; Gottesman, 1996; Margot and Karamata, 1996; Koonin et al., 1997; Pallen and Wren, 1997).

The existence in L. lactis of a unique surface protease, as shown by the htrA mutant strain study, was also confirmed by analysis of the L. lactis genome sequence (2.4 Mb; Bolotin et al., 1999). We also failed to find homologues of other known exported proteases (including ApeA = protease I, OmpP, OmpT, Prc = Tsp, Prt = protease III, SohB and SppIV of E. coli and Mpr, NprB, NprE, SubE, SubF, SubT, SubV and WprA of B. subtilis). Examination of the S. pneumoniae genome (2.2 Mb) and the Streptococcus pyogenes genome (2.2 Mb; both sequences are more than 95% complete) also suggests the presence of a single HtrA protease in each species. We postulate that Gram-positive bacteria with small genomes may minimize their genetic information and, in consequence, encode a single housekeeping surface protease. In the case of pathogens, the uniqueness of HtrA may make this protease a potentially interesting target for antibacterial drugs, particularly if HtrA indeed is involved in the maturation of pathogenic surface proteins.

The L. lactis htrA mutant strain may be a valuable tool for improving the yield, stability and activity of heterologous exported proteins. In E. coli and B. subtilis, for example, the stability of exported proteins was improved by inactivating several genes coding for exported proteases (four proteases in E. coli, Meerman and Georgiou, 1994; and six in B. subtilis, Wu et al., 1991). Nevertheless, considerable residual degradation of exported heterologous proteins was observed (Wu et al., 1991; Meerman and Georgiou, 1994). Furthermore, strain viability was greatly affected, even at the normal growth temperature (Wu et al., 1991; Meerman and Georgiou, 1994). Recently, a B. subtilis strain deficient in seven secreted proteases was also constructed, but it still retains about 0.14% of the extracellular protease activity of the wt strain (Ye et al., 1999). In L. lactis, inactivation of the unique htrALl gene allows complete stability of exported proteins as well as good growth when cells are cultivated at 30°C. In the future, this strain may have useful applications in the dairy, health or biotechnological fields by providing improved export efficiency of enzymes, antigens or therapeutic drugs.

Experimental procedures

Bacterial strains, growth conditions, plasmids and DNA preparations

E. coli strains used were: (i) TG1 [supE hsdΔ5 thiΔ(lac-proAB) F′ (traD36 proAB+lacIqlacZΔM15)], (ii) JIM4646, a TG1 repA+ derivative (E. Guedon, P. Renault, and C. Delorme, personal communication), and (iii) JIM5589 [JIM4646(pJIM2481); see below, E. Guedon, P. Renault, and C. Delorme, personal communication]. They were grown on Luria–Bertani medium at 37°C with shaking. L. lactis ssp. lactis strains used were IL1403 (wt, plasmid free, Chopin et al., 1984) and three isogenic derivatives: (i) IL6179 (hsdR, CmR, Schouler et al., 1998), which contains the same Cm resistance gene as that used to inactivate htrALl, and present in pJIM2481 (E. Guedon, P. Renault, and C. Delorme, personal communication); (ii) the htrA mutant (CmR, see below); and (iii) htrA+/htrA (CmR, see below).

Lactococci were grown on M17 medium (Terzaghi and Sandine, 1975) supplemented with 0.5 or 1% glucose at 30°C, without shaking. For thermo-sensitivity studies, growth (OD600 measure) and viability at 30°C, 37°C or 39°C were tested either with or without Cm selection using, respectively, the strain IL6179 or IL1403 as a positive control.

Plasmids used in this study were: pGEMT (ampicillinR, Promega); pJIM2481, a RepA-dependant replication plasmid (CmR, E. Guedon, P. Renault, and C. Delorme, personal communication); pVE8009, pVE8024 and pVE8021, pFUN derivative plasmids encoding Usp–ΔSPNuc, Nlp4–ΔSPNuc and Exp5–ΔSPNuc respectively (erythromycinR, Poquet et al., 1998); and pNuc3, a pIL253 derivative plasmid encoding wt S. aureus Nuc (erythromycinR, Le Loir et al., 1998). Plasmids were maintained by adding ampicillin (100 µg ml−1) or Cm (20 µg ml−1) to the medium for E. coli, or erythromycin (5 µg ml−1) for L. lactis.

General procedures for DNA preparation and handling were performed as has been described previously (Sambrook et al., 1989). Enzymes for cloning and sequencing were used, and PCR was carried out, as recommended by the suppliers. Electroporation of L. lactis was performed as has been previously described (Poquet et al., 1998).

Inactivation of L. lactis ssp. lactis IL1403 htrALl gene

For PCR amplification of htrALl gene fragments, the following oligonucleotides were used: (i) the ‘G’ primer, 5′-GTT TCC ACT TTT CTG TGG-3′, located upstream of the htrALl putative promoter; (ii) the ‘F’ primer 5′-GGA GCC A(G/T)(A/C/T) GC(A/G/C/T) (C/T)T(A/G/T) GG-3′, located downstream from the initiator methionine codon, and (iii) the ‘A’ primer, 5′-TT(A/T) CC(A/T) GG(A/G) TT(A/G/T) AT(A/G/C/T) GC-3′, located upstream of the catalytic site serine codon (primer degeneracy of ‘F’ and ‘A’ was needed for experiments not reported here). PCR amplification from L. lactis ssp. lactis IL1403 chromosomal DNA was performed using primer couples F+A or G+A. Two htrALl fragments, respectively, the 665 bp internal fragment, FA, and the 902 bp promoter-containing fragment, GA, were obtained.

FA and GA PCR fragments of htrALl were cloned into linear vector pGEMT and the resulting plasmids, pVE8035 and pVE8036, were verified using digestion and sequencing. The pJIM2481 vector was then inserted into each of them at the NotI site. The resulting co-integrates were introduced into E. coli TG1, using Cm selection. The pGEMT moiety of both co-integrates (pVE8037 and pVE8038) was subsequently deleted using an AatII and SacI double digestion and DNA polymerase I Klenow fragment treatment. The resulting constructs were introduced into E. coli JIM4646, using Cm selection. The final plasmids, pVE8039 and pVE8040, correspond, respectively, to FA (internal fragment) or GA (promoter-containing fragment) of htrALl subcloned into the pJIM2481 vector.

L. lactis non-replicative plasmids pVE8039 and pVE8040 were transformed into L. lactis ssp. lactis IL1403 competent cells. Selection for single cross-over integrants was performed with Cm selection (2.5 µg ml−1) under low-stress conditions, i.e. at room temperature under anaerobiosis. For each transformation, CmR clones were analysed using Southern blotting employing FA or GA PCR fragments as probes, and were confirmed to have the expected structure. The final mutant strains are referred to as htrA (corresponding to a disruption of the IL1403 chromosomal htrALl gene, htrALl::pVE8039, CmR) and htrA+/htrA (containing both wt and truncated copies of htrALl gene, htrALl::pVE8040, CmR). Note that an htrA derivative from MG1363 strain has not been constructed.

Description of proteins

Three protein fusions to the ΔSPNuc reporter (i.e. the 155 C-terminal aa of S. aureus Nuc protein, including the pro-peptide cleavage region and the NucA moiety), Usp–ΔSPNuc, Nlp4–ΔSPNuc and Exp5–ΔSPNuc, were studied (Poquet et al., 1998). In each case, native expression signals upstream of the ORF fused to the reporter were present; (i) the Usp–ΔSPNuc (198 aa) contains the 43 N-terminal aa (including the 27-aa classical signal peptide) of the precursor of L. lactis secreted protein Usp45 (Poquet et al., 1998); (ii) the Nlp4–ΔSPNuc (249 aa) contains the 94 N-terminal aa (including the 24-aa lipoprotein-type signal peptide) of the precursor of L. lactis lipoprotein Nlp4, a homologue of L. lactis PrtM and B. subtilis PrsA (Poquet et al., 1998); (iii) the Exp5–ΔSPNuc fusion (610 aa) is a tripartite fusion, comprising the N-terminal end (175 aa) of a penicillin-binding protein homologue (including a predicted 48-aa classical signal peptide), an internal segment (280 aa) of a cytoplasmic protein homologue, and ΔSPNuc. The full-length Exp5–ΔSPNuc protein detected in the htrA strain (Fig. 2C) is entirely cell associated (not shown). This may be owing to the presence of an uncleavable export signal (a transmembrane domain instead of a signal peptide) or to properties of the penicillin-binding protein moiety of Exp5–ΔSPNuc.

The S. aureus pre-pro-Nuc protein is 228 aa long and contains a predicted 60-aa signal peptide. Its pro-peptide is 19 or 21 aa long, depending on the bacterial host [processing occurs at both sites in S. aureus (Davis et al., 1977; Marrakechi and Cozzone, 1981), and at +21 in B. subtilis (Miller et al., 1987) and L. lactis (Le Loir et al., 1998)].

The AcmA precursor contains 437 aa, including a 57-aa signal peptide. The size of the smallest active degradation product is about 30 kDa (Buist et al., 1995; Chapot-Chartier, 1996). Note that cell association of some of the active degradation products was previously observed in the IL1403 strain (Chapot-Chartier, 1996), in contrast to the MG1363 strain (Buist et al., 1995).

Protein detection

Protein extracts and cell fractionation, SDS–PAGE and Western blotting of ΔSPNuc fusions and wt Nuc protein were performed as has been previously described (Poquet et al., 1998). For AcmA protein detection, zymograms were performed as has been previously described using autoclaved Micrococcus lysodeikticus cells as a substrate (Buist et al., 1995).


We thank A. Sorokin and S. D. Ehrlich for communicating the L. lactis ssp. lactis strain IL1403 sequence prior to publication. We are indebted to C. Delorme, E. Guedon and P. Renault for providing strains JIM4646 and JIM5589, to M.-C. Chopin for providing strain IL6179, to Y. Le Loir and P. Langella for giving plasmid pNuc3 and competent cells of JIM4646, to F. Gotz, S. Drouault and P. Renault for providing S. hyicus lipase and to M.-P. Chapot-Chartier for her help in AcmA activity detection. We are very grateful to B. Nicolas for photography. We thank our colleagues P. Langella, Y. Le Loir, and J.-C. Piard for frequent discussion during the course of this work. We are thankful to E. Maguin and M. van de Guchte for helpful suggestions and critical reading of the manuscript.