Secretion of GOB Metallo-β-Lactamase in Escherichia coli Depends Strictly on the Cooperation between the Cytoplasmic DnaK Chaperone System and the Sec Machinery: Completion of Folding and Zn(II) Ion Acquisition Occur in the Bacterial Periplasm

Abstract

Metallo-β-lactamases (MβLs) are zinc-dependent enzymes produced by many clinically relevant gram-negative pathogens that can hydrolyze most β-lactam antibiotics. MβLs are synthesized in the bacterial cytoplasm as precursors and are secreted into the periplasm. Here, we report that the biogenesis process of the recently characterized MβL GOB-18 demands cooperation between a main chaperone system of the bacterial cytoplasm, DnaK, and the Sec secretion machinery. Using the expression of the complete gob-18 gene from the gram-negative opportunistic pathogen Elizabethkingia meningoseptica in Escherichia coli as a model system, we found that the precursor of this metalloenzyme is secreted by the Sec pathway and reduces cell susceptibility to different β-lactam antibiotics. Moreover, acting with different J proteins such as cytoplasmic DnaJ and membrane-associated DjlA as cochaperones, DnaK plays an essential role in the cytoplasmic transit of the GOB-18 precursor to the Sec translocon. Our studies also revealed a less relevant role, that of assisting in GOB-18 secretion, for trigger factor, while no significant functions were found for other main cytoplasmic chaperones such as SecB or GroEL/ES. The overall findings indicate that the biogenesis of GOB-18 involves cytoplasmic interaction of the precursor protein mainly with DnaK, secretion by the Sec system, and final folding and incorporation of Zn(II) ions into the bacterial periplasm.

The production of β-lactam-degrading enzymes (β-lactamases) represents the most-common mechanism of antibiotic resistance among gram-negative bacteria (13). β-lactamases are grouped into four classes (A through D) on the basis of sequence homology; classes A, C, and D are constituted of serine enzymes, and class B is formed exclusively by metallo-β-lactamases (MβLs), enzymes that employ one or two Zn(II) ions to cleave the β-lactam ring (11, 13, 14). MβLs are particularly worrisome in the clinical setting in that they can hydrolyze most β-lactam antibiotics and are resistant to all clinically employed inhibitors (11, 13). Comparisons of MβL structures belonging to the different subclasses highlight a similar αβ/βα fold and a conserved motif that forms the metal ion binding site (7, 11). Many studies aimed at uncovering a common mechanism of action, information that is fundamental for the design of a much-needed paninhibitor, have been done on different MβLs (11, 13, 31, 33). Less attention has been given to MβL biogenesis, a process that includes translocation of newly synthesized precursors across the inner membrane and requires cytoplasmic chaperones for precursor transit or folding, and the actual cell location(s) in which the metal ion is incorporated into the apoenzymes. This knowledge may uncover novel targets for inhibitors of different stages of MβL biogenesis, opening new possibilities for the treatment of infections due to recalcitrant MβL producers (19).

Most secretory proteins of gram-negative bacteria are translocated by evolutionarily conserved machinery such as Sec or Tat (5, 6, 34, 39, 45). The core of the Sec system is formed by the translocation pore SecYEG, a narrow conduit that allows passage of unfolded proteins only, and the cytoplasmic ATPase component SecA, which drives the polypeptide chain into and through the pore (hereafter referred to as the SecA-SecYEG complex) (34). In addition, in Escherichia coli and other members of the phylum Proteobacteria (48), the cytoplasmic protein SecB also functions as a Sec-dedicated chaperone, both preventing premature folding and aggregation of the precursors and facilitating their delivery to SecA (2, 26, 34, 46). Different genetic studies have identified two subsets of secretory proteins in E. coli, one specifically depending on SecB for secretion and another for which different functionally redundant cytoplasmic chaperones, including SecB, can assist in the secretion process (51). Among the latter, different studies have identified a number of heat shock proteins, such as the main chaperone systems DnaK (DnaK/DnaJ/GrpE) (51) and GroE (GroEL/GroES) (27), as well as other members of the E. coli σ³² regulon that have chaperone functions (1, 2). It is worth noting in this context the functional complementarity/cooperation among major bacterial chaperones such as DnaK, GroE, trigger factor (TF), and SecB in assisting the folding process of a similar subset of cytoplasmic polypeptides described in different studies (10, 12, 16, 18, 24, 46, 49).

The evolutionarily conserved Tat secretory system is specialized to assist in the secretion of folded proteins that demand prior assembly in the cytoplasm, such as those that harbor complex cofactors and even some proteins without cofactors, including a number of serine β-lactamases (5, 34, 39). Recent data also suggest that selection of either the Sec or Tat pathway is used by certain metalloproteins to overcome low intrinsic specificity in metal ion binding and/or scarce bioavailability of the desired metal ion in the bacterial periplasm (45).

We recently cloned the gene for the MβL GOB-18 from a clinical isolate of the opportunistic gram-negative pathogen Elizabethkingia meningoseptica and produced the recombinant enzyme in E. coli (33). Detailed characterization of this enzyme indicates a novel type of broad-spectrum MβL maximally active with only one Zn(II) ion in the active site (33). Here, by using genetic and biochemical approaches, we studied the secretion machinery and cytoplasmic chaperones assisting in the secretion process of GOB-18 in gram-negative bacteria. Our results show that secretion of GOB-18 into the E. coli periplasm occurs through the Sec machinery, indicating that final folding of the apoprotein and Zn(II) ion acquisition occur in the bacterial periplasm. Also, cytoplasmic transit of preGOB-18 requires the assistance of an extended DnaK chaperone system that involves cooperation between both cytoplasmic and membrane J proteins.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The E. coli laboratory strains and plasmids used in this work are listed in Table 1. All antibiotics (kanamycin, chloramphenicol, ampicillin, tetracycline, and cefotaxime) were purchased from Sigma-Aldrich (St. Louis, MO). Restriction endonucleases and other DNA-modifying enzymes were purchased from Promega (Madison, WI). Oligonucleotides were synthesized by Biosynthesis, Inc. (Lewisville, TX).

TABLE 1.

Laboratory strains and plasmids used or constructed in this study

Strain or plasmid	Genotype and/or relevant characteristic(s)a	Source or reference
Strains
E. coli
DH5α	supE44 ΔlacU169 φ80lacZΔM15 hsdR17 recA1 gyrA96 thi-1 relA1	41
MC4100	F−araD139 Δ(argF-lac) U169 rpsL150 relA1 deoC1 ptsF25 rpsR flbB301	10
KY1601	MC4100; ΔrpoH30::kan zhf-50::Tn10 (λpF12-PrpoDhs-lacZ) (Kmr)	44
KY1880	MC4100; ΔgroE68::tet/pKV1561; pBR322 derivative in which the E. coli groESL operon is under lac promoter control (Tcr, Ampr)	21
CWB213	MC4100; secA51(Ts) leuB::Tn10	6
RO104	MC4100; secB::Tn5 (Kmr)	6
MB104	RO104/pE63 (Kmr, Ampr)	This work
CU165	MC4100; zhd33::Tn10 secY40 (Tcr)	3
GP108	MC4100; dnaJ::Tn10-42 (Tcr)	17
GP109	MC4100; ΔcpbA::Kmr	17
WKG15	MC4100; ΔdjlA::ΩSpcr	17
GP110	MC4100; ΔdjlA::ΩSpcrdnaJ::Tn10-42 (Tcr)	17
GP111	MC4100; ΔdjlA::ΩSpcrΔcpbA::Kmr	17
GP112	MC4100; dnaJ::Tn10-42 (Tcr) ΔcpbA::Kmr	17
GP113	MC4100; ΔdjlA::ΩSpcrdnaJ::Tn10-42 (Tcr) ΔcpbA::Kmr	17
B1LK0	MC4100; ΔtatC	5
GE4101	MC4100; tig::Tn5 (Kmr)	Laboratory collection
CG799	C600; dnaK+thr::Tn10 (Tcr)	43
CG800	C600; dnak103 thr::Tn10 (Tcr)	43
MC1061	F−araD139 Δ(ara-leu) 7679 galK16 Δ(lac)X74 rpsL hsdR mcrA mcrB1	8
CAG13351	MC1061; dnaK(D201N) thr::Tn10 (Tcr)	50
BL21(DE3)pLysS	F−ompT (rB[minus] mB−) gal dcm λ(DE3) pLysS (Cmr)	33
Plasmids
pTrx6	pACYC184 derivative (Cmr); expression vector	36
p-preGOB	pTrx6 derivative (Cmr); expresses preGOB-18 gene under araBAD promoter control	This work
pKP-GOB-18	pET22b(+) derivative (Kmr); expresses pelB-gob-18 fusion protein (GOB-18 with the PelB signal sequence) under lac promoter control	33
pET-GOB-18	pETGEX derivative (Ampr); expresses gst-gob-18 (mature GOB-18 as a C-terminal fusion to GST)	33
p8760	pBAD24 derivative (Ampr); expresses torA-gfp fusion protein (GFP with the TorA signal sequence) under araBAD promoter control	3
pTorA-GOB	p8760 derivative (Ampr); expresses torA-gob-18 fusion (mature GOB-18 with the TorA signal sequence)	This work
pTF5	p8760 derivative (Ampr); expresses tig under araBAD promoter control	This work
pE63	pBR322 derivative (Ampr); expresses E. coli gpsA under araBAD promoter control	42
pNRK416	pBR322 derivative (Ampr); expresses E. coli dnaK under lacUV5 promoter control; lacIq	22

^aKm^r, kanamycin resistant; Amp^r, ampicillin resistant; Cm^r, chloramphenicol resistant; Tc^r, tetracycline resistant.

Bacterial growth conditions.

E. coli strains were grown aerobically at 30°C (unless otherwise indicated) in LB broth supplemented with an appropriate antibiotic, ampicillin (100 μg/ml), kanamycin (25 μg/ml), tetracycline (5 μg/ml), or chloramphenicol (15 μg/ml), when necessary.

Construction of expression vectors.

The complete gob-18 gene was amplified by PCR from chromosomal DNA obtained from an E. meningoseptica clinical strain (33). Forward (GOB4) and reverse (GOB5) primers (Table 2) were designed with NdeI and HindIII restriction sites, respectively. The amplification product was digested with NdeI and HindIII and ligated into the equivalent sites of pET24b(+) (Novagen), generating plasmid pET24-preGOB. Expression plasmid p-preGOB, which directs production of preGOB-18 in E. coli, including its original transit peptide, was constructed by digesting pET24-preGOB with XbaI and HindIII and subcloning the 923-bp fragment containing the preGOB-18 gene into the equivalent sites of pBluescript SKII(+) (Stratagene, La Jolla, CA). The resulting plasmid was digested with NdeI and KpnI, and the 921-bp fragment containing the preGOB-18 gene was subsequently cloned into the equivalent sites of plasmid pTrx6 (36). The resulting plasmid, p-preGOB, directs expression of preGOB-18 gene in E. coli under the control of the araBAD promoter.

TABLE 2.

Oligonucleotide primers designed for this study

Primer	Sequence (5′-3′)a
GOB4 (NdeI)	5′-GACATATGAGAAATTTTGCTACACTG-3′
GOB6 (HindIII)	5′-CTAAGCTTCATACTTATTTATCTTGGG-3′
GOBfus3 (NheI)	5′- TTGGCTAGCCAGGTAGTAAAAGAACC-3′
Tig1 (NdeI)	5′-TGGTCATATGCAAGTTTCAGTTGAAACCAC-3′
Tig2 (HindIII)	5′-CGGCAAGCTTTTACGCCTGCTGGTTCATC-3′

^aBoldface type indicates NdeI, HindIII, and NheI restriction sites.

To construct pTorA-GOB, which directs production of a TorA-GOB fusion protein in E. coli under the control of the araBAD promoter, the gob-18 gene corresponding to the mature protein was amplified by PCR from E. meningoseptica DNA, using GOBfus3 (Table 2), which was designed with a NheI restriction site, as the forward primer. The reverse primer used was the same as that described above (GOB6) for the amplification of gob-18. The amplicon thus obtained was digested with NheI and HindIII and ligated into the equivalent sites of p8760 (Table 1), resulting in pTorA-GOB.

The E. coli tig gene was amplified by PCR from chromosomal DNA obtained from E. coli MC4100. Forward (Tig1) and reverse (Tig2) primers (Table 2) were designed with NdeI and HindIII restriction sites, respectively. The amplification product was digested with NdeI and HindIII and ligated into the equivalent sites on p8760, generating pTF5. This plasmid directs production of TF in E. coli under the control of the araBAD promoter.

All DNA constructions described above were verified by sequencing at the DNA Sequencing Service of Maine University (Orono, ME). The complete nucleotide sequence of gob-18 was reported previously (33).

Evaluation of the susceptibility of E. coli to cefotaxime in the absence and presence of GOB-18 production.

The different parental strains and corresponding mutant derivatives transformed with either p-preGOB or plasmid vectors were grown aerobically at 30°C, unless otherwise specified, in LB liquid medium supplemented with 15 μg/ml chloramphenicol. When the absorbance at 600 nm (Abs₆₀₀) reached 0.3, l-arabinose was added to a final concentration of 0.01% (wt/vol), and incubation was continued for an additional 1-h period. At this stage, the MICs of cefotaxime were determined by agar dilution assay (35) in LB medium, using sequential twofold cefotaxime dilutions for each strain in cells expressing gob-18 directed by p-preGOB (C_GOB) or transformed with the plasmid vector only (intrinsic MIC, C_I). The ratio C_GOB/C_I then equaled 2ⁿ, in which n is the number of twofold dilutions of cefotaxime that separate the MIC for a given bacterial strain producing GOB-18 from that for the same strain bearing only the plasmid vector. Therefore, the n value provides a semiquantitative estimation of the reductions in antibiotic susceptibility (i.e., increments in resistance) conferred by GOB-18 production to a particular strain. The C_I values of cefotaxime for the strains used in this study varied between 0.025 and 0.050 μg/ml and were not significantly modified by mutations in the different chaperone genes described in Table 1 (data not shown).

MIC values for KY1601 bacteria bearing the ΔrpoH allele (Table 1) were determined at a permissive growth temperature for this strain, i.e., 20°C (44). MIC determinations for CU165 bacteria containing the cold-sensitive secY40 allele (Table 1) were conducted at 23°C, a temperature at which the conditionally defective phenotype was in evidence but which still allowed bacterial growth at rates close to that of the wild-type strain. Similar conditions were also used by other authors to study the impact of this defective secY allele in the secretion of serine β-lactamase fusions (39). MICs for CWB213 bacteria bearing the secA51(Ts) allele (Table 1) were determined at 30°C, a condition under which the mutant phenotype was in evidence but which still allowed cell growth.

Subcellular fractionation.

Bacterial cells harboring either p-preGOB or plasmid vector were grown and induced with l-arabinose as described above. The periplasmic fraction was prepared by lysozyme treatment under isosmotic conditions. Briefly, the cells were collected, rinsed once with a mixture of 20 mM Tris-HCl (pH 8.5) and 150 mM NaCl, and resuspended in 20 mM Tris-HCl (pH 8.5), 20% (wt/vol) sucrose, 0.5 mM phenylmethylsulfonyl fluoride, and 0.1 mM EDTA (180 μl per Abs₆₀₀ unit). The mixture was supplemented with lysozyme (1 mg/ml), and incubation was conducted for a further 30 min at 4°C. After centrifugation at 15,000 × g for 2 min at 4°C, the supernatant containing the periplasmic fraction was collected and stored at 4°C for further analysis. The cell pellet (i.e., the cytoplasmic fraction) was resuspended in the same volume as above of a mixture of 20 mM Tris-HCl (pH 8.5) and 0.5 mM phenylmethylsulfonyl fluoride, disrupted by sonication (Vibra-Cell VCX-600; Sonics & Materials), and clarified by centrifugation at 25,000 × g for 20 min at 4°C. The supernatant containing the soluble cytoplasmic fraction was collected and stored at 4°C for further analysis. The pellet was resuspended in the same volume of a mixture of 20 mM Tris-HCl (pH 8.5) and 1% (wt/vol) sodium dodecyl sulfate (SDS) and incubated in a boiling-water bath for 10 min prior to analysis by SDS-polyacrylamide gel electrophoresis (PAGE). The total cell extract was prepared by resuspending the cellular pellet obtained from centrifuging the whole bacterial culture in a mixture of 20 mM Tris-HCl (pH 8.5) and 1% (wt/vol) SDS (180 μl per Abs₆₀₀ unit), followed by treating the mixture in a boiling-water bath for 10 min prior to analysis by SDS-PAGE, as above.

Protein analysis.

Subcellular fractions of each strain (20 μl of each sample prepared as described above, equivalent to approximately 20 μg of total cell protein, and the equivalent cellular volume of cytoplasmic or periplasmic proteins) were resolved by SDS-PAGE, using 16% polyacrylamide gels (with an acrylamide:bis-acrylamide ratio of 30:0.8) (41). The polypeptide pattern was observed after staining the gels with Coomassie blue. For immunoblot analyses, the polypeptides were transferred to either nitrocellulose or polyvinylidene difluoride membranes, followed by detection of specific proteins by specific antibodies and alkaline phosphatase-conjugated goat anti-rabbit antibodies (Bio-Rad) (41). Antisera against GOB-18, DnaK, GroEL, or TF were generated in rabbits according to conventional procedures.

Detection of MβL activity in GOB-producing E. coli cells.

MβL activity was detected by a microbiological assay using cell extracts of the analyzed bacterial strains (32). Briefly, the cells were grown at 30°C and induced as indicated above, and extracts were prepared by disrupting the bacteria with silicon dioxide microbeads (32). Mueller Hinton agar plates were inoculated with a liquid culture of an indicator strain (E. coli ATCC 25922) previously adjusted to a 0.5 McFarland standard of turbidity, and antimicrobial commercial disks containing 30 μg cefotaxime (BBL, Cockeysville, MD) were placed on the centers of the agar plates. Three filter disks (C, C/E, and B) were placed at the periphery of the cefotaxime disk on each plate within the expected zone of inhibition (Fig. 1B). The disk designated C was loaded with 20 μl of cell-free bacterial extract prepared as described above, disk C/E (cell extract plus EDTA) was loaded with the same volume of extract previously supplemented with 50 mM EDTA, and disk B (buffer) was loaded with 20 μl of 10 mM Tris-HCl (pH 7.5) and used as a control to visualize the growth of the indicator E. coli strain. The plates were incubated overnight at 37°C, and the presence of a MβL enzyme in the extracts was judged by comparing the growth of the indicator E. coli cells around disks C and C/E (32).

FIG. 1.

GOB production reduces cefotaxime susceptibility in E. coli cells. (A) E. coli MC4100 cells harboring p-preGOB (lanes 2 and 4) or the plasmid vector (lanes 1 and 3) were grown to exponential phase in LB liquid medium and serially diluted (10-fold dilutions of each) as indicated. Five microliters of each dilution was plated onto LB solid agar supplemented with 0.001% (wt/vol) l-arabinose in the absence (−CTX, lanes 1 and 2) or presence (+CTX, lanes 3 and 4) of 0.1 μg/ml cefotaxime (CTX). Plasmid p-preGOB directs expression of the GOB precursor (preGOB) in the bacterial cells with its original signal peptide, resulting in increased cefotaxime resistance. (B) Microbiological assay using cell extracts of GOB-producing MC4100/p-preGOB cells to test MβL production. The dashed circle highlights the growth inhibition halo for the indicator E. coli strain in the vicinity of the CTX disk. Growth of the indicator strain around disk C (black arrowheads) marks the presence of a cefotaxime-hydrolyzing enzyme in the extracts (32). The absence of growth around disk C/E shows inhibition of this enzyme by EDTA, and therefore indicates a metalloenzyme. The different disks were loaded with cell extract (C), cell extract supplemented with 50 mM EDTA (C/E), or buffer (B; no extract). The addition of an equivalent amount of EDTA (disk E; see insert at the top right) did not inhibit growth of the indicator strain. (C) SDS-polyacrylamide electrophoretic gels of different subcellular fractions obtained from l-arabinose-induced MC4100/p-preGOB cells. (D) Immunoblots of the same fractions as shown in panel C tested with specific antibodies against GOB, DnaK, and GroEL. Lane 1, total extracts of noninduced MC4100/p-preGOB cells; lanes 2 to 5, l-arabinose-induced cells; lane 2, total extracts; lane 3, soluble cytoplasmic fractions; lane 4, insoluble cytoplasmic fractions; lane 5, periplasmic fractions. Lanes 1 to 3 were each loaded with an equivalent of 20 μg of total proteins and lanes 4 and 5 with an equivalent cell volume of insoluble and periplasmic fractions, respectively. Migrations of the different molecular mass markers are indicated at the right margins of panels C and D. For details, see Materials and Methods. pGOB, GOB precursor; mGOB, mature GOB.

RESULTS

Production of E. meningoseptica preGOB-18 in E. coli reduces the susceptibility of the cells to β-lactam antibiotics.

Plasmid p-preGOB was used to express the complete gob-18 gene (including its original transit peptide-coding region) in E. coli cells (Fig. 1). Production of preGOB-18 in the cells resulted in a marked increment in cell resistance to cefotaxime, as shown by the growth of MC4100 bacteria transformed with p-preGOB in the presence of this antibiotic compared to that of cells harboring the plasmid vector only (Fig. 1A, lanes 3 and 4). The inhibition of GOB-18-dependent β-lactamase activity by EDTA, which was determined by comparing the growth of the indicator E. coli around disks C and C/E (Fig. 1B), confirmed that a metalloenzyme was produced. GOB-18-mediated bacterial resistance to other β-lactam antibiotics, including penicillin G, imipenem, and meropenem (data not shown), was also observed. The same results were obtained for other E. coli strains bearing p-preGOB, including C600 and MC1061 (data not shown).

Cell fractionation followed by SDS-PAGE (Fig. 1C) and immunoblotting (Fig. 1D) indicated the presence of a polypeptide with a molecular mass corresponding to that of mature GOB-18 in the periplasmic fraction of MC4100/p-preGOB (Fig. 1D, lane 5). Moreover, polypeptides corresponding to the precursor and mature forms could also be detected in cytoplasmic fractions of all the above bacteria (Fig. 1D, lanes 2 to 4).

The overall results indicate that GOB can be secreted into the bacterial periplasm as an active MβL, therefore providing a model system for studying the mechanism(s) of biogenesis of these enzymes in gram-negative bacteria.

GOB is secreted by the Sec pathway.

In gram-negative bacteria, both the Sec pathway and the Tat machinery have been shown to participate in metalloprotein secretion (45). Analysis of the 18-amino-acid residues of the GOB signal peptide indicated features compatible with a Sec-dependent secretion mechanism (34). This inferred role of Sec in GOB secretion was therefore tested by genetic procedures, using strains containing conditionally deficient secY or secA mutant alleles (Fig. 2). The MICs of cefotaxime were determined for each strain, both in cells transformed with the plasmid vector (C_I) and in bacteria expressing preGOB-18 gene directed by plasmid p-preGOB C_GOB. The ratio C_GOB/C_I then equaled 2ⁿ, where n is a semiquantitative estimation of the reductions in antibiotic susceptibility (i.e., increments in cefotaxime resistance) attributable to GOB production in a particular genetic background.

FIG. 2.

GOB is secreted by the Sec pathway. (A) Cefotaxime resistance conferred to E. coli by GOB production. The wild-type (WT) and different mutant strains deficient in functions provided by either Sec or Tat components were transformed with p-preGOB, and semiquantitative measurements of cefotaxime susceptibility were estimated by determining the corresponding n value in each case. The MICs used for n calculations were obtained consistently in three or more independent experiments (for details, see Materials and Methods). Bars 1, 3, and 7, wild type (MC4100); bar 2, the secY40 mutant (CU165); bar 4, the secA mutant (CWB213); bar 5, wild type (MC4100/pE63); bar 6, the ΔsecB mutant (MB104); bar 8, the ΔtatC strain (B1LK0). (B and C) Immunoblot analyses of periplasmic and total extracts, respectively, from strains analyzed for panel A and tested with specific antibodies against GOB. The GOB form corresponding to the mature protein (mGOB) observed in total cell fractions most likely resulted from the proteolytic processing of preGOB by cytoplasmic proteases, as has been observed for other precursor proteins (12). pGOB, GOB precursor. (D) Same as for panel C but tested with antibodies against DnaK and GroEL. Each lane was loaded with ca 20 μg of total proteins (total extracts) or the equivalent cell volume of the corresponding periplasmic fractions. (E) Cefotaxime resistance conferred to E. coli by GOB chimeras in which the mature region was fused to the TorA signal sequence (MC4100/pTorA-GOB; bar 10) or to GST [BL21(DE3)/pLysS/pET-GOB-18; bar 12]. For comparison, the n values obtained for the corresponding strains transformed with p-preGOB are shown in bars 9 and 11. (F and G) Immunoblot analyses of periplasmic and total extracts, respectively, from MC4100/p-preGOB (lanes 9) and MC4100/p-TorA-GOB (lanes 10). (H) Detection of MβL activity by a microbiological assay using cell extracts derived from the following bacteria: MC4100/pBR322 (disk TEM-1), BL21(DE3)/pLysS/pET-GOB-18 (disk GST-GOB), BL21(DE3)/pLysS/pET-GOB-18 supplemented with 50 mM EDTA (disk GST-GOB/E), and MC4100/pTorA-GOB (disk TorA-GOB). The black arrowheads indicate β-lactamase activity present in extracts of bacteria producing TorA-GOB or GST-GOB. The TEM-1 β-lactamase, a resistance marker codified in plasmids such as pET-GOB-18 (Table 1), neither conferred cefotaxime resistance to the bacteria (bar 12 in panel E) nor hydrolyzed cefotaxime to an appreciable extent (disk TEM-1 in panel H). All strains were grown until an Abs₆₀₀ of 0.3 was reached, and they were induced for 1 h with 0.01% l-arabinose (or 0.5 mM IPTG [isopropyl-β-d-thiogalactopyranoside]) before the corresponding MICs were determined, or they were processed for the isolation of subcellular fractions or cell extracts as indicated in Materials and Methods. The growth temperature was 30°C in all cases, except for strain CU165 and the corresponding parental strain in this particular case, which were grown at 23°C.

As shown in Fig. 2A, the production of preGOB in the MC4100 wild-type strain resulted in large increments in cellular resistance to cefotaxime (n values of 5 and 6 for bacteria grown at 23°C and 30°C, respectively) (Fig. 2A, bars 1 and 3). In turn, GOB-producing secY or secA mutants were significantly more susceptible to cefotaxime activity, judging by the lower increments in resistance reflected by their reduced n values compared to those for the parental strain grown under the same temperature (Fig. 2A, bars 2 and 4). Also, concomitant reductions in periplasmic GOB contents were observed in the mutants (Fig. 2B, lanes 1 to 4). The overall observations shown above indicate a relevant participation of the Sec pathway in GOB secretion.

Whether the Tat pathway also contributes to GOB secretion was evaluated by analyzing cefotaxime susceptibility and periplasmic GOB contents in GOB-producing ΔtatC mutants (5) (Fig. 2). As shown in the figure, the same n value (n = 6) (Fig. 2A, bars 7 and 8) and similar periplasmic GOB contents were observed for the ΔtatC mutant cells and for the parental strain (Fig. 2B, lanes 7 and 8). These results tend to rule out a significant role for the Tat system in GOB secretion.

Production of a chimeric protein in E. coli in which GOB was fused to the C terminus of glutathione S-transferase (GST) resulted in the accumulation of high levels of cytoplasmic GST-GOB, which has MβL activity (33) (Fig. 2H). We therefore evaluated whether redirecting this enzyme to the Tat pathway by incorporating an N-terminal Tat-specific signal sequence (TorA-GOB) could result in a secreted MβL capable of increasing cellular resistance to cefotaxime. As shown in Fig. 2G, lane 10, the TorA-GOB fusion protein was detected in whole-bacteria extracts of MC4100/pTorA-GOB cells and found to generate active MβL species, judging by a microbiological assay using these extracts (Fig. 2H). However, cells producing TorA-GOB contained negligible amounts of periplasmic GOB (Fig. 2F, lane 10) and showed minimal increments in cellular resistance to cefotaxime (n = 1) compared to the same strain producing preGOB (n = 6) (Fig. 2E, bars 9 and 10). It is relevant in this context that, similarly to the case of TorA-GOB, the GST-GOB produced in the bacterial cytoplasm also resulted in null increments (n = 0) in cellular resistance to cefotaxime (Fig. 2E, bar 12).

The overall results shown above support a minor participation of the Tat pathway in GOB secretion and point to the Sec machinery as the major system responsible for this process.

Cytoplasmic chaperones play significant roles in GOB secretion in E. coli. (i) SecB.

Whether the SecB chaperone plays a significant role in GOB secretion was evaluated by using E. coli secB null mutants (ΔsecB). As shown in Fig. 2, the same increments in cellular resistance to cefotaxime and similar GOB contents in the periplasmic fractions were observed in secB null mutants and in the parental strain (Fig. 2A, bars 5 and 6, and B, lanes 5 and 6, respectively). These results indicate that SecB is not essential for GOB secretion, indicating either that other chaperones may play significant roles in the cytoplasmic transit of the GOB precursor or, alternatively, that newly synthesized preGOB chains can directly reach the Sec translocon without assistance. In the latter context, in vitro data indicate that chaperone assistance during cytoplasmic transit of some secretory proteins may not be mandatory and that SecA itself may select nascent chains for export by the SecYEG complex (23).

(ii) E. coli σ³² heat shock regulon members.

In E. coli, different members of the σ³² heat shock regulon, such as DnaK (DnaK/DnaJ/GrpE) (4, 51), GroE (GroEL/GroES) (27), and possibly others (1, 2), have been found to cooperate with, or even complement, SecB functions in the secretion of a subset of secretory proteins. The alternative σ³² factor that positively controls expression of the heat shock regulon is codified by the rpoH gene (53). By using E. coli rpoH null mutants, we next evaluated whether σ³² components could be involved in GOB biogenesis. As shown in Fig. 3, GOB-producing ΔrpoH mutants had minimal increments in cefotaxime resistance (n = 1) (Fig. 3A, bars 1 and 2) and diminished periplasmic GOB contents compared to the parental strain (Fig. 3B, lanes 1 and 2). Since whole-cell levels of GOB in both wild-type and ΔrpoH bacteria were similar (Fig. 3C, lanes 1 and 2), the overall results support the idea that GOB secretion is highly affected by the absence of σ³².

FIG. 3.

RpoH-dependent chaperones participate in GOB biogenesis. (A) Cefotaxime resistance conferred to the different strains analyzed by GOB production (Table 1). Bar 1, MC4100; bar 2, the ΔrpoH strain (KY1601); bar 3, CG799; bar 4, the ΔdnaK strain (CG800); bar 5, the ΔdnaK plus DnaK strain (CG800/pNRK416); bar 6, MC1601; bar 7, the dnaK(D201N) strain (CAG13351). (B and C) Immunoblot analyses of periplasmic and total extracts, respectively, of the strains analyzed for panel A tested with specific antibodies against GOB. pGOB, GOB precursor; mGOB, mature GOB. (D) Same as for panel C but tested with specific antibodies against DnaK and GroEL. For details, see the legend to Fig. 2.

The results shown above indicate an active requirement for at least one component of the σ³² regulon for GOB secretion. In an attempt to identify the nature of the component(s), we next evaluated a possible role for the main ATP-dependent cytoplasmic chaperone systems DnaK and GroE in this process.

(iii) DnaK.

In E. coli, the typical DnaK system is composed of three σ³² regulon members: the main chaperone, DnaK, and two ancillary cochaperones, DnaJ and GrpE (15, 18, 19). DnaK is the central component of the system and acts by reversibly binding to short hydrophobic stretches present in the extended protein chain of its substrate in ATP hydrolysis-driven cycles. DnaJ, a member of the J protein family, can also bind to similar sites and deliver the protein substrates to DnaK while simultaneously stimulating its ATPase activity and the docking of the substrate into DnaK. GrpE, in turn, is a nucleotide exchange factor that allows the exchange of ADP for ATP in DnaK and the release of the bound polypeptide for subsequent folding, recycling, or transferring to downstream systems (15, 18, 19).

Whether the DnaK system plays roles in GOB secretion was first evaluated by using two strains bearing different dnaK mutant alleles, an in-frame deletion mutant (strain CG800, designated ΔdnaK in Table 1) that totally lacks DnaK (Fig. 3D, lane 4), and another strain, CAG13351, that produces a mutant DnaK chaperone in which a single amino acid change (Asp for Asn at position 201; DnaKD201N in Table 1) partially blocks the efficient discharge of the bound polypeptide (50). As shown in Fig. 3A, GOB-mediated cefotaxime resistance increments were lower in both DnaK-deficient mutants (Fig. 3A, bars 4 and 7) compared to those for the corresponding parental strains (Fig. 3A, bars 3 and 6). The ΔdnaK mutants, in particular, showed null increments (n = 0) in cellular resistance to cefotaxime and an almost negligible accumulation of GOB in the periplasmic fraction (Fig. 3B, compare lanes 3 and 4). Expression of dnaK in trans in these mutants partially restored cefotaxime resistance (Fig. 3A, bar 5) and periplasmic GOB contents (Fig. 3B, lane 5).

The results shown above support the idea that DnaK plays a pivotal role in GOB biogenesis. It is noteworthy that GOB accumulation in the ΔdnaK mutants was almost abolished (Fig. 3C, lane 4). This result is contrary to that found for the ΔrpoH mutants (Fig. 3C, lane 2), which also lack DnaK (Fig. 3D, lane 2). A possible explanation for these differences is that the ΔdnaK mutant cells, contrary to the ΔrpoH mutants, show a deregulated heat shock response involving overproduction of several main proteases (53), which could be responsible for digesting preGOB in the absence of DnaK. In agreement with this explanation, the expression of dnaK in trans in the ΔdnaK mutants partially restored the accumulation of GOB in whole cells (Fig. 3C, lane 5).

Bacteria bearing the dnaK(D201N) allele also showed reduced increments in cefotaxime resistance (Fig. 3A, bars 6 and 7), reinforcing the above-proposed role of DnaK in GOB biogenesis. The still-significant resistance (n = 4) of these cells compared to the intrinsic levels found in the ΔdnaK mutants (n = 0) was probably due to the retention of the partial chaperone activity of DnaK(D201N) (10, 50). It is worth noting that cells bearing the dnaK(D201N) allele contained increased levels of the mutant DnaK chaperone (Fig. 3D, compare lanes 6 and 7) as the result of its defective functions in regulating the heat shock response (50). This DnaK increment was accompanied by an increased accumulation of GOB in these mutants (Fig. 3C, lanes 6 and 7), an observation that reinforced the above idea that reductions in GOB secretion were due to impaired DnaK functions and that binding of the GOB precursor to this chaperone stabilized this polypeptide against proteolytic degradation.

(iv) DnaK cochaperones, the J proteins.

Unexpectedly, the ΔdnaJ mutants lacking DnaJ showed increments in cefotaxime resistance (n = 5) (Fig. 4A, bar 2) and periplasmic GOB contents (Fig. 4B, lane 2) close to those found in the corresponding parental strain (n = 6) (Fig. 4A, bar 1). This finding suggests that DnaJ functions are less relevant than those of DnaK for GOB biogenesis or, alternatively, that other J-family proteins of E. coli could complement DnaJ as a DnaK cochaperone in this process. The latter explanation was in fact favored by the observation that E. coli mutants simultaneously lacking different J proteins, including the cytoplasmic-located DnaJ and CbpA and the inner membrane-located DjlA, showed only slight increments in cefotaxime resistance (n = 1) compared to those for the parental strain (n = 6) (Fig. 4A, bars 1 and 8). To further dissect the roles of DnaJ, CbpA, and DjlA, we next analyzed GOB-mediated cefotaxime susceptibility in mutants lacking either one or different combinations of these J proteins (Fig. 4A). As shown in the figure, cells lacking DjlA showed the lowest increments in GOB-mediated cefotaxime resistance (n = 4) (Fig. 4A, bar 4) among the three single mutants (Fig. 4A, bars 2 to 4). On the other hand, GOB-mediated cefotaxime resistance increments for the ΔcbpA mutants were the same as those for the parental strain (n = 6) (Fig. 4A, bars 1 and 3). In addition and in agreement with a more relevant role for both DnaJ and DjlA in GOB biogenesis, mutants lacking both of these two J proteins showed even lower increments in cefotaxime resistance (n = 2) (Fig. 4A, bar 6) than the corresponding individual mutants (Fig. 4A, bars 2 and 4). Conversely, double mutants lacking CbpA and any of the other J proteins showed cefotaxime resistance increments similar to those of the corresponding dnaJ or djlA single mutants (Fig. 4A, bars 2, 4, 5 and 7). In agreement with the above observations, analysis of the corresponding periplasmic fractions in all strains described above showed a strict correlation between GOB levels and the observed n values (Fig. 4B). Since GOB accumulation in whole cells was similar in all of these strains (Fig. 4C), and considering that DnaK is present in all cases (Fig. 4D), the overall results point to DnaJ and DjlA performing synergistic actions as DnaK cochaperones in GOB biogenesis.

FIG. 4.

J proteins acting as DnaK cochaperones in GOB biogenesis. (A) Cefotaxime resistance conferred by GOB production to different strains harboring p-preGOB. Bar 1, wild type (WT; MC4100); bar 2, the ΔdnaJ strain (GP108); bar 3, the ΔcbpA strain (GP109); bar 4, the ΔdjlA strain (WKG15); bar 5, the ΔdnaJΔcbpA strain (GP112); bar 6, the ΔdnaJΔdjlA strain (GP110); bar 7, the ΔcbpAΔdjlA strain (GP111); bar 8, the ΔdnaJΔcbpAΔdjlA strain (GP113). (B and C) Immunoblot analyses of periplasmic and total extracts, respectively, of the strains analyzed for panel A tested with specific antibodies against GOB. pGOB, GOB precursor; mGOB, mature GOB. (D) Same as for panel C but tested with specific antibodies against DnaK and GroEL. For details, see the legend to Fig. 2.

(v) GroE.

We next evaluated whether the GroE system (GroEL/GroES) plays any significant role in GOB biogenesis. The GroE system is essential for E. coli at all growth temperatures (24), a situation that precludes the use of deletion mutants. However, the use of cells bearing mutant alleles, such as groEL140 or groES30 (27), indicated no differences in GOB-mediated cefotaxime resistance compared to that for the wild-type strain (data not shown).

We further analyzed cefotaxime susceptibility and periplasmic GOB levels in cells overproducing GroEL/GroES (strain KY1880) (Table 1). In this strain, cellular contents of GroEL/GroES well above physiological levels have been found to result in increased interaction with GroEL substrates and reductions in their productive folding (18, 21). In the case studied here, however, KY1880 bacteria containing up to 10 times the normal levels of GroEL produced GOB in amounts similar to and showed GOB-mediated cefotaxime resistance identical to those of cells containing the physiological contents of these chaperones (n = 6) (Fig. 5). This result showed no effects of GroEL overproduction in the GOB secretion process and suggests no significant interactions between this chaperone and the GOB precursor in the cytoplasm.

FIG. 5.

The GroE system is not essential for GOB biogenesis. (A) Cefotaxime resistance conferred to the different E. coli cells harboring p-preGOB by GOB production. Bar 1, wild type (WT; MC4100); bar 2, KY1880 (MC4100; ΔgroE68::tet/pKV1561) (Table 1). (B) Immunoblot analyses of total extracts of the strains analyzed for panel A tested with specific antibodies against GOB (upper figure) or DnaK and GroEL (lower figure). For details, see the legend to Fig. 2. pGOB, GOB precursor; mGOB, mature GOB.

The above results tend to rule out any significant role of the GroE system in GOB biogenesis.

(vi) TF.

TF, the product of the tig gene, is a σ³²-independent chaperone in E. coli associated with the ribosome that interacts with most short nascent chains, assisting in their folding and/or delivery to downstream chaperone machineries (18, 38). Different studies have indicated that the chaperone activities of TF and DnaK in cytoplasmic protein biogenesis partially overlap (16, 38, 49), yet the role of TF in protein secretion is still obscure. It has been found to retard secretion of some proteins by sequestering nascent chains for long periods (29) and proposed to antagonize both SecB and DnaK/DnaJ functions (47).

In the case of GOB secretion (Fig. 6), cells lacking TF (the Δtig mutants) (Fig. 6E, lane 2) showed lower increments in cefotaxime resistance (n = 4) than the corresponding parental strain (n = 6) (Fig. 6A, bars 1 and 2). In addition, a concomitant reduction in GOB contents was observed in the periplasmic fraction of the mutants (Fig. 6B, lanes 1 and 2). In turn, the expression of tig Δtig and TF (Fig. 6E, lane 3) partially restored cefotaxime resistance (n = 5) (Fig. 6A, bar 3) and periplasmic GOB contents (Fig. 6B, lane 3). Given that no major differences were observed in the accumulation of GOB in these bacteria (Fig. 6C, lanes 1 and 2), the above results indicate that TF participates in GOB biogenesis. The much lower contribution of TF compared to that of DnaK (Fig. 3) may account for the minor increments in cefotaxime resistance found in the ΔrpoH mutants (n = 1) (Fig. 3), which lack DnaK (Fig. 3D, lane 2) as well as σ³² cytoplasmic proteases (44, 53).

FIG. 6.

TF plays roles in GOB biogenesis. (A) Cefotaxime resistance conferred to different E. coli cells harboring p-preGOB by GOB production. Bar 1, wild type (WT; MC4100); bar 2, the Δtig strain (GE4101); bar 3, the Δtig plus TF strain (GE4101/pTF5). (B and C) Immunoblot analyses of periplasmic and total extracts, respectively, of the strains analyzed for panel A tested with specific antibodies against GOB. pGOB, GOB precursor; mGOB, mature GOB. (D) Same as for panel C but tested with specific antibodies against DnaK. (E) Same as for panel C but tested with specific antibodies against TF. For details, see the legend to Fig. 2.

DISCUSSION

The results presented in this work indicate that secretion of the MβL GOB to the bacterial periplasm is essentially driven by the Sec machinery and that the main cytoplasmic chaperone system DnaK plays a major role in GOB biogenesis. The role of the SecA-SecYEG complex in GOB secretion was evidenced by the significantly reduced β-lactam resistance increments and periplasmic contents of this protein in cells deficient in either SecA or SecY (Fig. 2). Our results also indicate that SecB, which serves as a Sec-dedicated cytoplasmic chaperone for a subset of secretory proteins in E. coli (2, 4, 34, 51), is not required for GOB secretion (Fig. 2 and 3), opening up the possibility that other cytoplasmic chaperones could assist newly synthesized preGOB chains during transit to the Sec machinery. In this context, functional complementarities have been reported among the major E. coli cytoplasmic chaperones SecB, DnaK, TF, and GroEL/GroES, indicating that they may compete for the same pool of newly synthesized polypeptides (16, 18, 24, 46, 49). These observations show that at least some of these chaperones could functionally substitute for each other in assisting in protein secretion, and in fact, both the DnaK and GroE systems have been proposed to cooperate with SecB in assisting in the secretion of a subset of secretory proteins (2, 27, 51). In the study reported here, however, analyses of cytoplasmic chaperones assisting in GOB secretion indicated an essential involvement of DnaK in the process, while other major chaperone systems such as SecB and GroEL/GroES had no significant roles in GOB biogenesis (Fig. 2 to 6). The absence of DnaK abolished GOB-mediated cefotaxime resistance, and no other cytoplasmic chaperone could replace DnaK in restoring GOB secretion. The accumulation of preGOB was precluded in these mutants, suggesting that DnaK binds this precursor during cytoplasmic transit, protecting it from degradation by cytoplasmic proteases (Fig. 3). Consistent with this proposal, bacteria producing a DnaK mutant, such as DnaK(D201N), that were unable to efficiently discharge bound polypeptides (50) resulted in large increments in both the mutant chaperone and the GOB precursor in the cells (Fig. 3). Also, reductions in GOB secretion due to impairments in SecA functions resulted in higher cellular levels of DnaK and concomitantly increased accumulation of preGOB (Fig. 2).

Our results also indicate that in addition to the physiological DnaK cochaperone DnaJ (15), other J proteins such as DjlA played significant roles in GOB biogenesis, acting synergistically with DnaJ in this process (Fig. 4). DjlA, an inner membrane protein present in many gram-negative bacteria, is capable of performing DnaK cochaperone functions (17) and has recently been found to be involved in the pathogenesis of both Legionella dumoffii (37) and Vibrio tapetis (28). The redundancy of both cytoplasmic and membrane-located DnaK cochaperone functions found in this work on GOB secretion may thus serve to secure an efficient transit and discharge of subsets of presecretory proteins to the Sec translocon.

Whether TF plays a direct role in protein secretion remains uncertain (29, 47). Our results indicate that the absence of TF results in partial reductions in GOB secretion (Fig. 6). However, the large differences in secretion and/or stability of the GOB precursor observed in the ΔdnaK and Δtig mutants clearly point to DnaK as the most relevant system in assisting GOB biogenesis.

The overall conclusions of this work are summarized in the scheme shown in Fig. 7. The cytoplasmic transit of preGOB to the Sec machinery represents a complex process assisted mainly by an “extended” DnaK system in which cytoplasmic DnaJ and membrane-associated DjlA cochaperones can act synergistically and in which TF may collaborate by providing an alternative (but less efficient) pathway.

FIG. 7.

Proposed scenario of cytoplasmic chaperones and secretion systems assisting GOB biogenesis in bacteria. Newly synthesized preGOB chains interact with TF and are subsequently delivered to an “extended” DnaK system in which DnaK, aided by both cytoplasmic and membrane-located J proteins (step 1), protects the precursor from degradation by cytoplasmic proteases and assists in transit to the Sec complex (step 2). Lack of DnaK functions and augmented levels of σ³²-dependent proteases (as seen in the ΔdnaK mutants) increase proteolytic processing of the precursor (step 3). The GOB precursor may be transferred with less efficiency to the Sec translocon via TF (step 4). GOB folds and binds Zn(II) ions to generate an active MβL after secretion into the periplasm (step 5). For details, see the text. OM, outer membrane; IM, inner membrane.

Although the above results do not directly address the issue of GOB secretion in E. meningoseptica, they provide some hints about this process that are worth mentioning here. The lack of a requirement for SecB for GOB secretion observed in this work was not totally unexpected, given that SecB chaperones are restricted to the phylum Proteobacteria (to which E. coli is assigned) and seem to have evolved in their last common ancestor (48). E. meningoseptica, on the other hand, has been assigned to the class Flavobacteria (and the phylum Bacteroidetes) and is therefore well apart taxonomically from the Proteobacteria (25). In this context, and contrary to those for SecB, all DnaK system components (as well as TF) from all chaperone systems of the cytoplasm are absolutely conserved and ubiquitously distributed among all phyla that constitute the realm Bacteria (52). It is therefore reasonable to speculate that the DnaK chaperone system also plays a major role in the cytoplasmic transit of the GOB precursor in E. meningoseptica and that secretion of this MβL in this host may follow a pathway similar to that described in Fig. 7.

In the case of MβLs, the complete polypeptide chain must be present for the apoproteins to fold to a step that allows Zn(II) ion binding and the acquisition of activity (11, 33, 40). Secretion of these enzymes into the periplasmic space could then proceed in an extended conformation via the Sec machinery or, alternatively, via the Tat machinery in an already folded form containing the bound Zn(II) ion (34). The results presented here that show that GOB is secreted by the Sec machinery therefore unravel a fundamental step in MβLs biogenesis, indicating that the apoprotein folds and incorporates Zn(II) ions to gain the native state in the bacterial periplasm (Fig. 7). Secretion of MβLs by the Sec rather than by the Tat pathway may serve the important purpose of facilitating binding of the correct ion to gain the proper activity in the correct compartment, as proposed for other metalloproteins (45). In this context, we previously found that production of a recombinant GOB in a form restricted to the bacterial cytoplasm generated species capable of binding either Zn(II) or Fe(II) ions without apparent discrimination between them (33). Remarkably, the GOB form secreted to the E. coli periplasm contained only Zn(II) ions (33). Similar observations concerning the lack of selectivity between Zn(II) and Fe(II) ions were made recently by other authors using an L1 MβL form produced in the E. coli cytoplasm (20). Therefore, secretion of MβLs by the Sec rather than by the Tat pathway, as shown in this work, supports the proposal that a given secretory machinery may serve to overcome low intrinsic specificities of a given apoenzyme for the desired metal ion (45).

The emergence of novel enzymes with the ability to hydrolyze most of the currently available β-lactams and their dissemination as the result of horizontal gene transfer events among pathogenic bacteria constitute worrisome problems (13, 32). This situation has resulted in a need for new antimicrobials and for compounds capable of exerting synergistic effects with known antibiotics to treat infections due to multidrug-resistant strains. Characterization of the active sites of different MβLs advocates for a general mechanism of hydrolysis by these enzymes, opening the possibility of searching for a paninhibitor (31, 33). Yet no such inhibitor is available, a situation that necessitates alternative strategies for the treatment of broad-spectrum MβL-producing pathogenic bacteria. The results of the present work thus bear additional relevance when proposals that bacterial chaperones may represent therapeutic targets for new antimicrobials are considered (19). In this context, certain natural short peptides and fatty acyl benzamido derivatives have been described as highly selective inhibitors of bacterial DnaK functions (9, 19, 30). It is therefore tempting to speculate that combinations of different β-lactams and specific bacterial DnaK inhibitors may result in synergistic activities against bacteria that largely depend on the DnaK system for the biogenesis of a main resistance determinant, as is the case reported in this work.