The RipA and RipB Peptidoglycan Endopeptidases Are Individually Nonessential to Mycobacterium smegmatis

ABSTRACT

Mycobacteria possess a series of Rip peptidoglycan endopeptidases that have been characterized in various levels of detail. The RipA and RipB proteins have been extensively studied and are dl-endopeptidases, and RipA has been considered essential to Mycobacterium smegmatis and Mycobacterium tuberculosis. We show here that the ripA and ripB genes are individually dispensable in M. smegmatis and that at least one of the genes must be expressed for viability. We characterized strains carrying in-frame deletion mutations of ripA and ripB and found that both mutant strains exhibited increased susceptibility to a limited number of antibiotics and to detergent but that only the ΔripA mutant displayed hypersusceptibility to lysozyme. We also constructed and characterized ΔripD and ΔripA ΔripD mutants and found that the single mutant had only an intermediate lysozyme hypersusceptibility phenotype compared to that of wild-type cells while loss of ripD in the ΔripA background partially rescued the antibiotic and lysozyme phenotypes of the ΔripA mutant.

IMPORTANCE We show that the RipA endopeptidase, which has been considered essential for cell division in certain mycobacteria, is not essential but that at least it or a similar protein, RipB, must be expressed by the bacteria for viability. This work is the first description of strains carrying single deletion mutations of RipA, RipB, and a novel endopeptidase-like protein, RipD.

INTRODUCTION

One of the most important characteristics of mycobacteria is a complex cell envelope, the biosynthesis of which is the target of several antimycobacterial drugs (1–3). Our research is focused upon the assembly and maintenance of the peptidoglycan (PG) layer, which surrounds the entirety of the cell and covalently anchors the other components of the cell envelope. The biosynthesis of the PG layer is inhibited by β-lactam antibiotics, which have recently been reevaluated as potential antitubercular drugs (4, 5).

The mycobacterial PG is composed of glycan strands of N-acetylglucosamine and N-acetyl/N-glycolylmuramic acid with a high degree of interpeptide cross-links (6–9). Assembly is orchestrated by a variety of transglycosylases and transpeptidases, including the penicillin binding proteins (PBPs) PonA1, PonA2, PbpA, and PbpB, as well as several ld-transpeptidases (reviewed in reference 10). In order to incorporate new precursor material into the preexisting cell wall, the old cell wall material must first be metabolized by lytic transglycosylases, endopeptidases, and amidases (11). Cell wall turnover has been extensively studied in Gram-negative bacteria, but comparatively little is known about the balance of degradation/synthesis of the PG of mycobacteria. Most of the research has focused on the Rpf resuscitation-promoting factors, which are PG glycan hydrolases, and on several PG endopeptidases (11, 12).

The endopeptidases belong to the NlpC/P60 superfamily (13), with the first NlpC/P60 mycobacterial proteins identified in Mycobacterium marinum and given the names iipA and iipB (invasion and intracellular persistence) because they were identified in a screen for mutants defective for intracellular survival (14). Subsequently, these genes were renamed ripA and ripB (Rpf interaction protein) after it was found that RipA interacts with the resuscitation-promoting factors RpfB and RpfE (15). RipA is one of four homologs in Mycobacterium tuberculosis, encoded by Rv1477 (ripA), Rv1478 (ripB), Rv2190c (ripC), Rv1566c (ripD) (16). There is another related protein, encoded by Rv0024, that may be another endopeptidase, but this has not been studied (17). Orthologous proteins exist in Mycobacterium smegmatis: MSMEG_3145 (ripA), MSMEG_3146 (ripB), MSMEG_4256 (ripC), MSMEG_3477 (ripD), and MSMEG_1686 (Rv0024). The Rip proteins have NlpC/P60 domains in their C termini, with RipA, RipB, and Rv0024 possessing the typical NlpC/P60 Cys-His-Glu catalytic triad (17–19). RipA has an N-terminal domain that is missing from RipB, but both enzymes are dl-endopeptidases that cleave the bond between the d-glutaminyl and meso-diaminopimelyl residues in the PG peptides (17, 18). RipC has a variant Cys-His-His catalytic triad, is activated by FtsX, and has a role in cell wall restructuring during cell division, and a ripC mutant of M. tuberculosis is attenuated in the mouse model (20, 21). RipD differs from the other Rip enzymes by having Ala-Ser residues substituted for the canonical Cys-His catalytic residues in the NlpC/P60 domain. RipD can bind PG but lacks endopeptidase activity (22).

The RipA and RipB proteins have received the most scrutiny, but the contributions of these two proteins to cell wall physiology are unclear because no single null mutant of either gene has been studied. The ripA and ripB genes comprise a bicistronic operon with ripA proximal to the promoter, which is regulated by the MtrAB two-component system (15, 23). A previous study with M. smegmatis reported that RipA is essential and RipB is not, but that work used a depletion strain in which the ripAB promoter was replaced with a tetracycline-regulated promoter, which resulted in loss of ripAB expression under noninducing conditions (15, 24). Another study, using a genome-wide screening of a saturated transposon library of M. tuberculosis, failed to identify ripA insertional mutants while it did identify ripB mutants (25). In contrast, another study reported that both of the RipA and RipB homologs in M. marinum (IipA and IipB) are not essential and have some overlapping functions (14). The M. marinum studies were performed using an engineered iipA insertional mutation (iipA::kan) that had a polar effect on iipB expression. Thus, it would appear that loss of the ripAB operon may be lethal for M. smegmatis and M. tuberculosis but not for M. marinum. Alternatively, perhaps the essentiality of the operon in the first two of these species is due to a requirement for only ripA expression. However, we hypothesized that M. smegmatis and M. tuberculosis require that at least one of the genes in the ripAB operon is expressed. We tested this idea genetically using deletion mutants of M. smegmatis. Here, we provide data consistent with the view that neither ripA nor ripB is individually essential to M. smegmatis but that the organism requires at least one of the genes to maintain viability. In addition to characterizing novel ripA and ripB single deletion mutants, we also characterized novel ripD deletion mutants and found that loss of ripD could partially rescue the phenotypes of an ΔripA mutant.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Strains are listed in Table 1. Wild-type (WT) M. smegmatis strains are PM145 and PM1482. For technical reasons, the ripA essentiality and reverse transcription-PCR (RT-PCR) assays were done using strains with a PM145 lineage while all phenotypic analyses were done using mutants made in the PM1482 background. Escherichia coli strain DH10B was used for all routine plasmid constructions, and strain HB101 was used for construction of plasmids bearing the res1-hyg-res1 cassette to avoid resolution of the hyg gene. E. coli cultures were grown at 37°C in Luria-Bertani (LB) broth (Difco/BD Bioscience, San Jose, CA) or on LB agar. Mycobacterial cultures were grown at 37°C in Middlebrook 7H9 broth and were plated on Middlebrook 7H10 agar (Difco/BD Biosciences). The Middlebrook medium was supplemented with 0.05% Tween 80, 0.2% glycerol, and ADS (0.5% bovine serum albumin fraction V, 0.2% dextrose, and 0.85% NaCl). When needed, l-lysine was added to all Middlebrook media at 40 μg/ml. Antibiotics were used at the following concentrations: hygromycin, 100 μg/ml for M. smegmatis and 200 μg/ml for E. coli; kanamycin, 50 μg/ml for E. coli and 25 μg/ml for M. smegmatis; apramycin, 50 μg/ml for E. coli and M. smegmatis; and streptomycin, 1,000 μg/ml for M. smegmatis. Hygromycin and bovine serum albumin fraction V were obtained from Roche Applied Science (Indianapolis, IN); all other antibiotics and additives were obtained from Sigma-Aldrich Chemical (St. Louis, MO).

TABLE 1.

Bacterial strains used in this study

Strain	Description	Reference or source
E. coli
DH10B	F− mcrA Δ(mcrBC-hsdRMS-mrr) [ϕ80dΔlacZΔM15] ΔlacX74 deoR recA1 endA1 araD139 Δ(ara-leu)7697 galU galK λ− rpsL nupG	32
HB101	F− hsdS20 proA2 leuB1 glnV44 ara-14 galK2 lacY1 rpsL20 xyl-5 mtl-1 recA13	33
M. smegmatis
PM145	mc21212 (lysA ept-4 rpsL5)	27
PM3167	PM145/pMP1188 (ΔripA::res1-hyg-res1 primary recombinant)	This work
PM3183	PM3167 ΔripA::res1-hyg-res1/pMP1213 (secondary recombinant)	This work
PM3206	PM3183 ΔripA::res1/pMP1213 (resolved secondary recombinant)	This work
mc2155	ept-1	34
PM1482	ept-1 ΔlysA4 rpsL6 ΔblaS ΔblaE	35
PM3207	PM1482/pMP1187 (ΔripA primary recombinant)	This work
PM3221	PM3207 ΔripA (secondary recombinant)	This work
PM3293	PM1482/pMP1243 (ΔripB::res1-hyg-res1 primary recombinant)	This work
PM3297	PM3293 ΔripB::res1-hyg-res1 (secondary recombinant)	This work
PM3301	PM3297 ΔripB::res1 (resolved secondary recombinant)	This work
PM3414	PM1482/pMP1244 (ΔripD::res1-hyg-res1 primary recombinant)	This work
PM3415	PM3414 ΔripD::res1-hyg-res1 (secondary recombinant)	This work
PM3416	PM3415 ΔripD::res1 (resolved secondary recombinant)	This work
PM3419	PM3221/pMP1244 (ΔripD::res1-hyg-res1 primary recombinant)	This work
PM3420	PM3419 ΔripD::res1-hyg-res1 (secondary recombinant)	This work
PM3421	PM3420 ΔripD::res1 (resolved secondary recombinant)	This work

Plasmids and DNA manipulation.

Plasmids used in this study are shown in Table 2. Basic molecular biology techniques were performed as previously described (26). Plasmids used for recombination were prepared with Qiagen columns (Qiagen, Valencia, CA). DNA fragments were isolated using agarose electrophoresis and were purified with a QIAquick gel purification kit (Qiagen). Restriction and DNA modification enzymes were obtained from Fermentas (Hanover, MD) or New England BioLabs (Beverly, MA). Oligonucleotides were synthesized by Eurofins (Huntsville, AL). Electroporation of M. smegmatis was done as previously described (27). DNA sequencing was performed by Genewiz (South Plainfield, NJ).

TABLE 2.

Plasmids used in this study

Plasmid	Description	Reference or source
pMV261	Kmr E. coli-Mycobacterium shuttle vector, contains the groEL promoter; ColE1/pAL500 oriM	36
pMP854	pMV261 γδ resolvase	37
pMP1189	pMV261 ripABa	This work
pMP1267	pMV261 ripABb	This work
pMP1190	pMV261 ripA	This work
pMP1191	pMV261 ΔripA ripB	This work
pMP1247	pMV261 ripD	This work
pMP349	pMV261 with aacC41 replacing aphA-1; Apr	38
pMP1212	pMP349 ripAB	This work
pMP1213	pMP349 ripA	This work
pMP1214	pMP349 ΔripA ripB	This work
pMP918	rpsL+ Kmr suicide allelic exchange vector	This work
pMP1186	pMP918 ripA region	This work
pMP1225	pMP918 ripB region	This work
pMP1227	pMP918 ripD region	This work
pMP1187	pMP1186 ΔripA	This work
pMP1226	pMP1225 ΔripB	This work
pMP1228	pMP1227 ΔripD	This work
pMP1064	pBluescript KS res1-hyg-res1 cassette flanked by SmaI sites	This work
pMP1188	pMP1187 ΔripA::res1-hyg-res1	This work
pMP1243	pMP1226 ΔripB::res1-hyg-res1	This work
pMP1244	pMP1228 ΔripD::res1-hyg-res1	This work
pMN467	pMS2 (Kmr E. coli-Mycobacterium shuttle vector) with mycobacterial codon-optimized gfp	39

^aripAB from M. smegmatis.

^bripAB from M. tuberculosis.

Construction of the M. smegmatis ΔripA, ΔripB, and ΔripD suicide plasmids for allelic exchange.

The ripA, ripB, and ripD regions were amplified from mc²155 genomic DNA using iProof High-Fidelity DNA Polymerase from Bio-Rad (Hercules, CA). Both amplicons were cloned into the M. smegmatis suicide vector pMP918 (Km^r and rpsL⁺) via NheI and XbaI, resulting in pMP1186 (ripA), pMP1225 (ripB), and pMP1227 (ripD). These plasmids were used for inverse PCRs to create in-frame deletions to produce pMP1187 (ΔripA), pMP1226 (ΔripB), and pMP1228 (ΔripD). The deletion alleles were then marked with an SmaI-digested res1-hyg-res1 cassette cloned into a unique DraI site at each deletion site. This produced plasmids pMP1188 (ΔripA::res1-hyg-res1), pMP1243 (ΔripB::res1-hyg-res1), and pMP1244 (ΔripD::res1-hyg-res1), which were used for allelic exchange. Plasmid pMP1187 (ΔripA) was used for allelic exchange in the WT strain PM1482, and pMP1188 (ΔripA::res1-hyg-res1) was used for allelic exchange in WT strain PM145. Plasmids pMP1243 (ΔripB::res1-hyg-res1) and pMP1244 (ΔripD::res1-hyg-res1) were used for allelic exchange in WT PM1482. Additionally, pMP1244 was also used with the ΔripA strain PM3321 to create the ΔripA ΔripD::res1 double deletion strain). Deletions marked with the res1-hyg-res1 cassette were resolved in frame using the γδ resolvase plasmid pMP854, and all alleles were confirmed by PCR and sequence analysis.

RT-PCR amplification of ripB mRNA.

Reverse transcription-PCR (RT-PCR) was performed to amplify ripB message from PM145 (WT), PM3183 (ΔripA::res1-hyg-res1), and PM3206 (ΔripA::res1). Nucleic acids were isolated from saturated cultures of these strains using a FastRNA Pro Blue kit (MP Biomedicals, Solon, OH). Genomic DNA was degraded using a Turbo DNA-Free kit from Ambion (Lithuania). cDNA was synthesized using Superscript II reverse transcriptase from Invitrogen (Carlsbad, CA) and used in PCR amplification reactions with iProof High-Fidelity DNA Polymerase from Bio-Rad (Hercules, CA). The PCR products were separated on a 0.8% agarose gel, stained with ethidium bromide, and then imaged using a Fluorochem 5500 from Alpha Innotech (Santa Clara, CA). The entire image was brightened by 60% using PowerPoint software during the preparation of the figure for publication in order to reveal any bands in the dark background.

Growth curves.

Log-phase broth cultures (optical density at 600 nm [OD₆₀₀] of 0.5 to 0.8) in 7H9 medium were used to inoculate 10 ml of 7H9 medium, and the OD was adjusted to an OD₆₀₀ of ∼0.05. Cultures were incubated at 37°C with shaking, and optical density was measured at time points 0, 6, 12, 24, and 48 h using a Beckman Du530 Life Science UV/Vis Spectrophotometer. Viable plating was also performed at each time point, with plates incubated at 37°C for 3 days.

Antibiotic susceptibility.

Log-phase broth cultures (OD₆₀₀ of 0.5 to 0.8) in 7H9 medium were adjusted to an OD₆₀₀ of ∼0.5. Two hundred microliters of the suspension was mixed with 3 ml of molten 0.7% noble agar from Sigma-Aldrich Chemical (St. Louis, MO) and then immediately poured onto 7H10 plates. Once the top agar was solidified, either prepared Sensi-Disc disks (BD Biosciences) or compound-soaked sterile disks (BD 3 Biosciences) using antibiotics from Sigma-Aldrich Chemical (St. Louis, MO) were placed on the plates. The antibiotics tested were ampicillin (50 μg), carbenicillin (100 μg), piperacillin (5 μg), ceftriaxone (100 μg), imipenem (10 μg), rifampin (5 μg), erythromycin (15 μg), vancomycin (30 μg), ciprofloxacin (50 μg), and isoniazid (10 μg). Sodium dodecyl sulfate (SDS) from EM Science (Gibbstown, NJ) was used at a concentration of 20% SDS in 10-μl volumes on sterile disks. After incubation at 37°C for 48 h, the zones of inhibition were measured to the nearest millimeter. Six repetitions were done for each strain per antibiotic or chemical.

Lysozyme susceptibility.

Log-phase broth cultures (OD₆₀₀ of 0.5 to 0.8) in 7H9 medium were adjusted to an OD₆₀₀ of ∼0.5. Cultures were serially diluted and plated onto 7H10 plates containing 0, 200, or 1,000 μg/ml lysozyme (catalog number L-7651, lot number 77H7032; Sigma-Aldrich Chemical). Plates were then incubated at 37°C for 3 days. Changes in viability were measured by comparing the counts of cultures plated onto medium lacking lysozyme (N₀) with that of cultures plated onto medium with lysozyme (N_*).

Light microscopy.

Log-phase broth cultures (OD₆₀₀ of 0.5 to 0.8) in 7H9 medium were used for microscopy. pMN437, a mycobacterial codon-optimized green fluorescent protein (GFP) expression plasmid, was used to transform PM1482 (WT) and PM3221 (ΔripA). One milliliter of culture for each was centrifuged, and the cell pellet was resuspended in 100 μl of residual medium. Six microliters of Mowiol mounting medium (6 g of glycerol, 2.4 g of Mowiol 4-88, 6 ml of distilled H₂O [dH₂O], and 12 ml of 0.2 M Tris buffer, pH 7.2) was used to fix 4 μl of the bacterial suspension onto glass slides (method courtesy of K. Derbyshire, Wadsworth Center, New York State Department of Health, Albany, NY). Cells were visualized using an Olympus BX41 microscope at a magnification of ×100, and images were taken using a QImaging Exi Aqua camera.

Statistical analysis.

GraphPad software (GraphPad Software, Inc., La Jolla, CA) was used to analyze data using an unpaired Student's t test to compare results with the mutant to those with the WT or to compare results with two mutants to each other, as indicated in the figure legends.

RESULTS

Deletion of the ripA and ripB genes in M. smegmatis.

The ripA (MSMEG_3145) and ripB (MSMEG_3146) genes exist as a bicistronic operon in mycobacteria. We set out to create in-frame deletions of both ripA and ripB individually, using suicide plasmids bearing the genomic regions flanking ripAB with in-frame deletions of ripA (pMP1188) or ripB (pMP1243). The deletions were marked with a novel resolvable hygromycin cassette (res1-hyg-res1). The wild-type res-hyg-res cassette is resolved by expression of the γδ resolvase, leaving a single 136-bp res site at the insertion point; however, the res site has translational stops in all three reading frames (28). The new res1-hyg-res1 cassette has point mutations in the flanking res sites (res1) that remove a stop codon in one of the open reading frames such that resolution of the cassette can result in the res1 site being in frame within the deleted gene of interest. We used an rpsL⁺-based kanamycin resistance suicide vector for allelic exchange of the marked ripB gene and successfully generated an ΔripB::res1-hyg-res1 mutant, strain PM3297, in which the hygromycin marker was resolved to yield the final mutant PM3301 (ΔripB::res1). However, our initial attempts to create an ΔripA::res1-hyg-res1 mutant in a similar fashion failed, suggesting that disruption of ripA was lethal. We decided to test this formally using a series of genetic assays.

ripA essentiality tests.

We performed essentiality tests using the Km^r Hyg^r Sm^s ripA rpsL primary recombinant strain (PM3167), containing both the WT and the ΔripA::res1-hyg-res1 alleles in its genome (Fig. 1). This recombinant was transformed with replicating apramycin resistance plasmids bearing either the entire ripAB operon (pMP1212), ripA (pMP1213), ripB (pMP1214), or the vector control (pMP349). We then measured the number of WT (Hyg^s) or ΔripA::res1-hyg-res1 (Hyg^r) secondary recombinants in the Sm^r Km^s population after outgrowth in apramycin-containing medium followed by streptomycin counterselection (Fig. 1). We found that the primary recombinant bearing the empty vector resulted in only WT secondary recombinants (Table 3). However, primary recombinants bearing ripAB, ripA, or ripB plasmids generated ΔripA::res1-hyg-res1 secondary recombinants at frequencies of 12%, 9%, and 4%, respectively (Table 3). The ripAB operon plasmid transformant yielded mutants in all three trials, but the results with the ripA or ripB plasmid transformants were not uniformly successful.

FIG 1.

Allelic exchange at the ripAB locus. The integrated rpsL⁺ counterselectable suicide plasmid for the allelic exchange of wild-type (WT) ripA with the ΔripA::res1-hyg-res1 mutation is shown, along with the two potential outcomes of secondary recombination (mutant or WT). Km, kanamycin; Hyg, hygromycin; Sm, streptomycin.

TABLE 3.

ripA essentiality tests in M. smegmatis

Complementing plasmid	No. of secondary recombinantsa				Frequency of ΔripA (%)
	Trial 1	Trial 2	Trial 3	Total
pMP349 (vector control)	0 (23)	0 (22)	0 (24)	0 (69)	0
pMP1212 (ripAB)	2 (24)	6 (25)	1 (25)	9 (74)	12
pMP1213 (ripA)	0 (23)	6 (21)	0 (22)	6 (66)	9
pMP1214 (ripB)	0 (26)	4 (22)	0 (26)	4 (74)	5

^aNumber of ΔripA::res1-hyg-res1 clones (total number of secondary recombinants screened).

Plasmid stability tests in the ΔripA::res1-hyg-res1 mutant.

We hypothesized that the inability to generate ΔripA::res1-hyg-res1 mutants in the absence of complementing plasmids is because M. smegmatis requires at least RipA or RipB for viability and that a polar effect from ΔripA::res1-hyg-res1 could prevent downstream expression of ripB (Fig. 2A). Alternatively, the inconsistent ability to obtain mutants in two of the three complemented strains described in Table 3 might suggest that spontaneous suppressor mutations might play a role in the ability to disrupt ripA. In order to test these ideas, we performed plasmid stability tests with some of the complemented mutants. First, we resolved out the hygromycin cassette in the ΔripA::res1-hyg-res1 allele in one of the strains in the hope that resolution of this cassette in frame would alleviate polarity. We did this by transforming an ΔripA::res1-hyg-res1 secondary recombinant (PM3183), bearing the ripA complementing plasmid pMP1213, with pMP854, which encodes the γδ resolvase. Antibiotic selection was maintained for the ripA plasmid (apramycin) and the γδ resolvase plasmid (kanamycin). We screened transformants for the loss of hygromycin resistance and confirmed that the hyg gene in the original ΔripA::res1-hyg-res1 allele was resolved in frame to generate ΔripA::res1 by sequencing of a PCR product obtained from genomic DNA prepared from a selected clone (strain PM3206) (Fig. 2A). After curing the strain of the resolvase plasmid, we then conducted plasmid stability tests. The WT/pMP1213 (ripA), ΔripA::res1-hyg-res1/pMP1213, and resolved ΔripA::res1/pMP1213 strains were grown in the presence of apramycin, subcultured into liquid medium lacking antibiotics, grown for 48 h (approximately 16 generations), plated for viable counts, and then screened for apramycin resistance as an indicator of plasmid curing. We found that the ripA-complementing plasmid was lost from both the wild-type strain and the resolved ΔripA::res1 strain but not from the ΔripA::res1-hyg-res1 mutant (Table 4). The ability to cure the complementing plasmid from the ΔripA::res1 strain suggested that the barrier to ripA disruption was likely a polar effect and that our ability to disrupt the gene in the complemented strains was due to complementation. If the ability to disrupt the gene was due to extragenic suppressor mutations, then we should have been able to cure the pMP1213 plasmid from the strain bearing the ΔripA::res1-hyg-res1 mutation.

FIG 2.

Genetic organization of the ripAB locus and polarity. (A) Expression of the WT ripAB operon, polarity effect of the ΔripA::res1-hyg-res1 allele, and the polarity relief of the in-frame ΔripA::res1 allele after resolution by the γδ resolvase. The open box below the ripB gene indicates the region targeted for PCR analysis of ripB expression. The X represents loss of ripAB mRNA due to polarity. (B) Ethidium bromide-stained DNA gel electrophoresis of the 265-bp RT-PCR ripB product from total RNA prepared from WT strain PM145 (lanes 2 and 3), the ΔripA::res1-hyg-res1/pMP1213 (ripA) complemented mutant strain PM3183 (lanes 4 and 5), and the ΔripA::res1/pMP1213 (ripA) strain PM3206 (lanes 6 and 7). Lanes 2, 4, and 6 represent reaction mixtures that included reverse transcriptase (RT), while lanes 3, 5, and 7 represent control reaction mixtures lacking RT. The WT genomic DNA PCR amplification control is shown in lane 1.

TABLE 4.

Plasmid stability tests in M. smegmatis ΔripA

Strain	Plasmid	No. of apramycin-sensitive clones (total no. of clones screened)			Frequency of plasmid loss (%)
		Expt 1	Expt 2	Total
PM1482 (wild type)	pMP1213 (ripA)	26 (100)	19 (100)	45 (200)	22.5
PM3183 (ΔripA::res1-hyg-res1)	pMP1213 (ripA)	0 (100)	0 (100)	0 (200)	0
PM3206 (ΔripA::res1)	pMP1213 (ripA)	86 (100)	98 (100)	184 (200)	92

The res1-hyg-res1 allele has a polar effect on ripB expression.

We tested the supposition that the res1-hyg-res1 insertion in ripA has a polar effect on ripB expression by analyzing ripB expression in the WT strain and the Hyg^r nonpermissive ripA deletion strain PM3183 complemented with the ripA plasmid (ΔripA::res1-hyg-res1/pMP1213 strain) and the resolved (Hyg^s) derivative which is permissive for ripA deletion, strain PM3206 (ΔripA::res1/pMP1213) (Fig. 2A). Total RNA was prepared from cultures of these strains and used for ripB-specific RT-PCR. As shown in Fig. 2B, we were able to amplify ripB message from the WT and ΔripA::res1/pMP1213 strains but not from the ΔripA::res1-hyg-res1/pMP1213 strain, indicating a polar effect from the ΔripA::res1-hyg-res1 allele.

ΔripA and ΔripB strains do not display a growth or morphological defect.

The results above prompted us to engineer a new mutant using a suicide plasmid bearing an unmarked, in-frame deletion of ripA, as described in Materials and Methods. Unlike the attempts with the marked mutation, we were able to easily construct the ΔripA mutant PM3221. We then used this mutant and the ΔripB::res1 mutant, PM3301, for phenotypic analyses. It was previously reported that an M. smegmatis ripAB depletion strain has a severe growth defect (24). Our ripA and ripB deletion strains do not have a substantial change in log-phase growth, as determined by either optical density or viable plate counts, and although the ΔripA mutant does exhibit a small decease in viable counts at the 48-h time point (Fig. 3A and B), we do not see this consistently in other growth curves (data not shown). Both the M. smegmatis ripAB depletion strain and the M. marinum iipA::kan mutant have septation defects and a filamentous morphology (14, 24). To investigate this with our M. smegmatis ΔripA mutant, we transformed the deletion strain with a plasmid (pMN467) that expresses a codon-optimized GFP to visualize the cells microscopically and found that the mutant did not show any morphological differences compared to the WT morphology (Fig. 3C and D).

FIG 3.

Growth, viability, and morphology of the ΔripA and ΔripB mutants. (A) Optical density semilog plot (inset, linear plot). (B) Viable CFU counts in WT (PM1482), ΔripA (PM3221), and ΔripB (PM3301) strains. *P < 0.0001. (C and D) Fluorescence microscopy of the WT (PM1482/pMN467 GFP⁺) and ΔripA (PM3221/pMN467 GFP⁺) strains, respectively.

ΔripA and ΔripB strains have increased susceptibility to antibiotics and detergent, but only ripA is important for lysozyme resistance in M. smegmatis.

The ΔripA and ΔripB strains were tested by disk diffusion to determine if they have any changes in antibiotic susceptibilities. We found that both strains showed increased susceptibilities to rifampin, erythromycin, and vancomycin (Fig. 4A). We also tested both mutants with ampicillin, carbenicillin, piperacillin, ceftriaxone, imipenem, ciprofloxacin, and the ΔripA mutant with isoniazid, but saw no differences in susceptibilities compared to those of the WT (data not shown). Because of the increased sensitivity to large antibiotics, we decided to test envelope permeability using SDS (29) and found that both the ΔripA and ΔripB mutants have an increase in sensitivity to SDS compared to that of the WT (Fig. 4A). We found that ripAB and ripA could complement the antibiotic and SDS phenotypes of the ΔripA mutant, while ripB only partially complemented the antibiotic phenotypes and did not complement the SDS phenotype of the ΔripA strain (Fig. 4B). In contrast, all three plasmids could fully complement the susceptibility phenotype of the ΔripB strain (Fig. 4C).

FIG 4.

Antibiotic and detergent susceptibility of ΔripA and ΔripB mutants by disk diffusion assays. (A) Susceptibility of the WT (PM1482), ΔripA (PM3221), and ΔripB (PM3301) strains. (B) Susceptibility of the complemented WT and ΔripA strains: PM1482/pMV261 (WT/VC), PM3221/pMV261 (ΔripA/VC), PM3221/pMP1189 (ΔripA/ripAB), PM3221/pMP1190 (ΔripA/ripA), and PM3221/pMP1191 (ΔripA/ripB). (C) Susceptibility of the complemented WT and ΔripB strains: PM1482/pMV261 (WT/VC), PM3301/pMV261 (ΔripB/VC), PM3301/pMV261 (ΔripB/ripAB), PM3301/pMP1190 (ΔripB/ripA), and PM3301/pMP1191 (ΔripB/ripB). *P < 0.001; **, P < 0.0001. VC, vector control.

Since loss of RipA or RipB could have an effect on peptidoglycan turnover, we wanted to see if the mutants had altered susceptibilities to lysozyme. We found that the ΔripA mutant was more susceptible to lysozyme but that the ΔripB mutant showed no such increase in susceptibility compared to that of the WT (Fig. 5A). The lysozyme phenotype of the ΔripA mutant could be complemented back to that of the WT by either ripA or ripB but not by the entire ripAB operon (Fig. 5B).

FIG 5.

Lysozyme susceptibility of ΔripA and ΔripB mutants. Data represent viable counts on medium with lysozyme (N_*) divided by viable counts on medium lacking lysozyme (N₀). (A) Susceptibility of WT (PM1482), ΔripA (PM3221), and ΔripB (PM3301) strains. (B) Lysozyme susceptibility of complemented WT and ΔripA strains: PM1482/pMV261 (WT/VC), PM3221/pMV261 (ΔripA/VC), PM3221/pMP1189 (ΔripA/ripAB), PM3221/pMP1190 (ΔripA/ripA), and PM3221/pMP1191 (ΔripA/ripB). *, P < 0.05; **, P < 0.001; ***, P < 0.0001.

The M. tuberculosis ripAB operon complements the M. smegmatis ΔripB mutant, but not the ΔripA mutant, phenotype.

Previous studies with the iipAB operon in M. marinum had shown that the polar ΔiipA::kan mutation could be complemented by the M. tuberculosis ripAB genes in terms of antibiotic and lysozyme susceptibilities (14). We demonstrated here that complementation of the ΔripB M. smegmatis mutant with the M. tuberculosis ripAB operon produced a WT phenotype in susceptibility testing with antibiotics and SDS (Fig. 6A). However, we did not observe any complementation of the ΔripA mutant with the M. tuberculosis ripAB operon in susceptibility testing with antibiotics and SDS (Fig. 6B) or lysozyme (Fig. 6C).

FIG 6.

Susceptibility phenotypes of ΔripA and ΔripB mutants bearing the M. tuberculosis or M. smegmatis ripAB operon. (A) Disk diffusion assays of complemented WT and ΔripB strains as follows: PM1482/pMV261 (WT/VC), PM3301/pMV261 (ΔripB/VC), PM3301/pMP1189 (ΔripB/Msm ripAB), and PM3301/pMP1267 (ΔripB/Mtb ripAB). (B) Disk diffusion assays of complemented WT and ΔripA strains as follows: PM1482/pMV261 (WT/VC), PM3221/pMV261 (ΔripA/VC), PM3221/pMP1189 (ΔripA/Msm ripAB), and PM3221/pMP1267 (ΔripA/Mtb ripAB). (C) Lysozyme susceptibility in the following strains: PM1482/pMV261 (WT/VC), PM3221/pMV261 (ΔripA/VC), PM3221/pMP1189 [ΔripA/ripAB(Msm)], and PM3221/pMP1267 [ΔripA/ripAB(Mtb)]. *, P < 0.0001. Mtb, complementation with ripAB of M. tuberculosis; Msm, complementation with ripAB of M. smegmatis.

Deletion of the ripD gene in M. smegmatis.

Of the four rip genes in mycobacteria, only ripD has not been characterized genetically. Previous studies of RipD have been focused upon biochemical analyses of the purified M. tuberculosis RipD protein (22). We decided to analyze ripD deletion mutants made in either a WT background or a ripA deletion mutant. We chose the latter mutant because loss of ripA had a more profound effect on phenotype than the loss of ripB. As with the ripA and ripB deletion constructs, we created a suicide plasmid bearing a resolvable hygromycin cassette (res1-hyg-res1) with a ripD in-frame deletion and used it to create ΔripD::res1 (PM3416) and ΔripA ΔripD::res1 (PM3421) mutant strains. We encountered no obstacles to the creation of the ΔripD mutant in either background. Similar to results with the ΔripA and ΔripB mutants, we did not see any growth or viability defects with these mutants (data not shown).

Loss of ripD partially rescues the phenotypes of the ΔripA mutant.

Since we observed increased susceptibilities to antibiotics (rifampin, erythromycin, and vancomycin) and SDS in our ΔripA and ΔripB mutants, we wanted to determine if a ΔripD mutant would also exhibit these same phenotypes. However, we saw no change in the susceptibility of the ΔripD mutant to these antibiotics and SDS compared to the susceptibility of the WT (Fig. 7A). We also tested to see if there were changes in β-lactam susceptibility in the mutant, but we saw no changes in susceptibilities to ampicillin, imipenem, and ceftriaxone compared to those of the wild type (data not shown). Introduction of the ΔripD mutation into the ΔripA mutant partially rescued the antibiotic, but not the SDS, susceptibility of that mutant (Fig. 7B). Complementation of the ΔripA ΔripD double mutant with ripD on a multicopy plasmid resulted in an augmentation of the original ΔripA phenotype (Fig. 7C).

FIG 7.

Susceptibility phenotypes of the ΔripD and ΔripA ΔripD mutants by disk diffusion assay. (A) WT/VC (PM1482/pMV261), ΔripD/VC (PM3416/pMV261), and ΔripD/ripD (PM3416/pMP1247) complemented strains. (B) WT (PM1482), ΔripA (PM3221), ΔripD (PM3416), and ΔripA ΔripD (PM3421) strains. (C) WT/VC (PM1482/pMV261), ΔripA/VC (PM3221/pMV261), ΔripA ΔripD/VC (PM3421/pMV261), and ΔripA ΔripD/ripD (PM3421/pMP1247) complemented strains. *, P < 0.05; ***, P < 0.0001 (results compared to those of the WT); ⁺, P < 0.05; ⁺⁺, P < 0.01; ⁺⁺⁺, P < 0.0001 (results compared to those of the ΔripA mutant).

We next examined the lysozyme phenotype and found that the ΔripD mutant was more susceptible to lysozyme than the wild type but was intermediate to the phenotype of the ΔripA mutant (Fig. 8A). The ΔripD mutant could be fully complemented by ripD in trans (Fig. 8A). In a similar fashion, the ΔripA ΔripD double mutant exhibited increased susceptibility to lysozyme compared to that of the WT, but it also was more resistant to lysozyme than the ΔripA and ΔripD single mutants (Fig. 8B). Furthermore, complementation of the ΔripA ΔripD double mutant with ripD essentially restored the susceptibility phenotype to that of the ripA single mutant (Fig. 8C). It appears that loss of RipD has a minimal effect on wild-type cells but is able to partially rescue most of the phenotypes of the ΔripA mutant.

FIG 8.

Lysozyme susceptibility of the ΔripD and ΔripA ΔripD mutants. Data represent viable counts on medium with lysozyme (N_*) divided by viable counts on medium lacking lysozyme (N₀). (A) WT/VC (PM1482/pMV261), ΔripA/VC (PM3221/pMV261), ΔripD/VC (PM3416/pMV261), and ΔripD/ripD (PM3416/pMP1247) strains. *, P < 0.01. (B) WT (PM1482), ΔripA (PM3221), ΔripD (PM3416), and ΔripA ΔripD (PM3421) strains. (C) WT/VC (PM1482/pMV261), ΔripA ΔripD/VC (PM3421/pMV261), ΔripA/VC (PM3221/pMV261), and ΔripA ΔripD/ripD (PM3421/pMP1247) strains. *, P < 0.05; **, P < 0.01 (results compared to those of the WT); ⁺, P < 0.01; ⁺⁺, P < 0.0001 (results compared to those of the ΔripA mutant).

DISCUSSION

We have demonstrated that the two closely related mycobacterial dl-endopeptidases, RipA and RipB, are individually dispensable in M. smegmatis. This is in contrast to previous studies that concluded that RipA is essential to M. smegmatis and likely M. tuberculosis as well and also plays a critical role in cell division. The earlier studies on the ripAB operon in M. smegmatis used a strain in which expression of the entire operon was shut down, with subsequent growth problems and severe septation defects, which could be rescued by ripA complementation (24). Similar morphological defects were seen in a polar M. marinum iipA::kan mutant (14). We surmised that the defect in the M. smegmatis ripAB depletion mutant was really the result of the loss of expression of both genes and that in-frame deletion mutants would be viable. This idea was also suggested by the observation that a comprehensive M. tuberculosis transposon mutant library lacks ripA mutants but has ripB mutants (25). We think that the lack of ripA insertional mutants in that collection is due to a polar effect on ripB expression such that both proteins were not expressed. In this work we tested these ideas by constructing a marked ripA deletion allele and showing that it could be used only to make an M. smegmatis mutant in a strain bearing an extra copy of ripA, ripB, or the entire operon. We then demonstrated that the insertion had a polar effect on ripB expression, which, once removed, allowed ripB expression and rescue of the ΔripA mutant. We conclude from these genetic analyses that RipA and RipB are individually nonessential and that loss of both gene products is lethal. We saw no growth or morphological changes at any level in any of the mutants in this study, in contrast to previous studies in M. smegmatis and M. marinum. Again, we think that the difference here is that the previous studies examined mutants in which both genes were not expressed. The fact that the loss of ripAB is lethal to an M. smegmatis depletion strain while a polar M. marinum iipA::kan mutant does not have a viability defect might be explained by species-specific differences or by differences between the levels of expression of ripB (iipB) in the strains.

While the two proteins have overlapping functionality from an essentiality standpoint, they do not have entirely overlapping functions in general, as the individual mutants have somewhat different phenotypes. We found that both mutants had increased susceptibility to rifampin, erythromycin, and vancomycin. Increases in susceptibility to rifampin, erythromycin, and ciprofloxacin were also seen with the iipA::kan mutant of M. marinum, but we failed to detect any increase in susceptibility to ciprofloxacin (14). Increased susceptibility to rifampin was also seen in an M. smegmatis ripC mutant, suggesting a common susceptibility phenotype for endopeptidase mutants (20). Our mutants had no changes in susceptibility to PG-specific β-lactam antibiotics or isoniazid, which inhibits mycolic acid biosynthesis, suggesting that the PG and the cell envelope may not have large defects in architecture. However, the increase in susceptibilities to antibiotics that are large suggests a particular alteration to the envelope that may somehow increase permeability to those compounds. The increased susceptibilities of both mutants to the detergent SDS also suggests a permeability change in the envelope. Detergent susceptibility was also seen in an ripC mutant of M. tuberculosis, suggesting another phenotype that may be common among strains lacking endopeptidases (21).

The antibiotic and SDS susceptibility phenotypes of the ΔripB mutant could be fully complemented back to that of the wild type with either gene or with the entire ripAB operon carried on multicopy plasmids; however, only the ripA gene or the ripAB operon could fully complement the ΔripA mutant, with the ripB gene partially complementing the mutant. A similar phenomenon was noted with the inability of ripB to complement the antibiotic susceptibility of the M. marinum iipA::kan mutant (14). We think that this suggests that there is a subtle difference in the roles of these two proteins. RipA has been shown to localize to the poles and the septum of growing cells (30); perhaps RipB does not localize in the same way or does not interact with the same partners as RipA, and so it cannot efficiently substitute for RipA under these particular stresses. Alternatively, a study has shown that purified RipA functions on high-molecular-weight PG preparations while RipB seems to be inactive on such substrates, suggesting that the two proteins recognize slightly different types of PG peptides (17). However, that study used PG purified from Bacillus subtilis, which lacks modifications found in mycobacterial PG (7), and, thus, such data on substrate specificities should be interpreted with caution. Alternatively, the differences in cross-complementation might be due to multicopy effects since the complementing genes are expressed from the ripAB promoter on a multicopy plasmid.

A difference in RipA and RipB functionalities is also evident by the observation that only the ΔripA mutant is hypersusceptible to lysozyme, but in contrast to the antibiotic and SDS phenotypes, this phenotype can be fully complemented by either ripA or ripB but not the entire operon. We do not know why ripB complements this defect and not the others, but this may suggest that the different phenotypes may result from different mechanisms. The inability of the ripAB operon to complement the lysozyme defect may be the result of excessive endopeptidase activity in the presence of lysozyme. We have noted that wild-type cells bearing the ripAB operon on a multicopy plasmid are very sick (data not shown), suggesting that higher than normal levels of endopeptidases are detrimental, which has been noted before with RipA (31). The ripAB plasmid is not detrimental to the ΔripA mutant in any other assay but the lysozyme assay, suggesting that the damage inflicted by the lysozyme is important in this phenomenon. Given that RipA interacts with RfpB and RpfE, both of which are PG hydrolases, we speculate that loss of RipA somehow disturbs the regulation of the Rpf proteins. This might explain why the ΔripA mutant but not the ΔripB mutant is hypersusceptible to lysozyme. However, the ability of ripB to complement this defect suggests an alternative mechanism, but we do not know if RipB interacts with any of the Rpf proteins. It was noted that the M. marinum iipA::kan mutant was also hypersusceptible to lysozyme but was only fully complemented by the entire operon (14). It is difficult to compare results between the two species since the M. marinum mutant was essentially a double mutant. Still, there may be some species-specific differences in how these proteins function.

The species differences are most apparent when we examined the ability of the genes from M. tuberculosis to complement the M. smegmatis mutants. We demonstrated that while the M. smegmatis ripAB operon could complement the antibiotic and SDS phenotypes of both mutants, the ripAB genes of M. tuberculosis complemented only the ΔripB mutant. We think that the lack of complementation of the ΔripA mutant is due to the domain structure of RipA.

RipA has an N-terminal domain that occludes the active site of the enzyme, which may have to be removed in order for the enzyme to function on high-molecular-weight PG substrates (17, 18, 31). It was previously shown that this is likely a species-specific cleavage event (31); thus, the failure of the M. tuberculosis ripAB operon to complement the ΔripA mutant is likely because the protease of M. smegmatis does not recognize the M. tuberculosis RipA protein. We noticed a similar loss of complementation in the lysozyme phenotype of the ΔripA mutant with the M. tuberculosis ripAB operon. In this case, we assume that the ripB of M. tuberculosis is expressed and partially functional since this plasmid can still complement the ΔripB antibiotic and SDS phenotypes. Since the M. smegmatis ripB gene can complement the lysozyme phenotype of the ΔripA mutant, the inability of the M. tuberculosis ripAB operon to complement this phenotype may be due to the inability to interact with other proteins because of species-specific differences in the interacting partner(s).

Because there is some functional overlap between RipA and RipB, we wanted to explore the role of the lesser-studied RipD protein, which lacks the canonical endopeptidase active site but is still able to bind PG (22), in order to determine if perhaps loss of RipD would affect the same phenotypes as loss of RipA or RipB. Our rationale was that if some phenotypes are mirrored in a ΔripD mutant, then the mechanism behind the phenotype is not directly related to PG cleavage since RipD is incapable of this activity. We found that there was essentially no antibiotic or detergent phenotype in the ΔripD mutant, but there was an intermediate lysozyme susceptibility phenotype, which could be complemented with the wild-type ripD gene. This would suggest that the lysozyme phenotype may be due to the physical loss of the protein and not necessarily the loss of a protein with endopeptidase activity. On the other hand, the lysozyme phenotypes of the ripA and ripD deletion mutants might result from two completely different mechanisms. Remarkably, when we introduced the ΔripD mutation into the ripA deletion background, we found that the loss of ripD partially rescued both the antibiotic and lysozyme phenotypes but not the SDS phenotype of the ripA mutant. Importantly, complementation of the double ΔripA ΔripD mutant with ripD restored the susceptibility seen in the single ΔripA mutant. This would suggest that in the ripA deletion mutant, the antibiotic and lysozyme phenotypes are partially due to the presence of RipD. Perhaps RipD is involved with recruiting other PG-degrading enzymes to the PG, and this may become dysregulated in the ripA mutant. Future experiments aimed at determining if the essential nature of the ripAB operon could be abrogated in a ripD deletion mutant or if loss of ripD could partially rescue the phenotype of the ΔripB mutant, along with further investigating the interactions between the endopeptidases and PG biosynthetic proteins, will help sort out the complexities in this system and provide a better understanding of mycobacterial cell wall metabolism.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.