CRISPR Inhibition of Essential Peptidoglycan Biosynthesis Genes in Mycobacterium abscessus and Its Impact on β-Lactam Susceptibility

ABSTRACT

We utilized a CRISPR interference (CRISPRi) assay to control the gene expressions of two predicted essential peptidoglycan biosynthesis genes, pbpB and cwIM, in Mycobacterium abscessus and to evaluate their contribution to β-lactam susceptibility. Our results showed that CRISPR inhibition of each gene led to a significant 3-log₁₀ reduction in CFU in the presence of imipenem but not for cefoxitin. These results demonstrate that CRISPRi provides an experimental approach to study drug/target interactions in M. abscessus.

TEXT

The rise in incidence of Mycobacterium abscessus infections in patients with cystic fibrosis and in transplant recipients highlights an emerging environmental pathogen that is recalcitrant to treatment with most antibiotics and associated with a poor clinical outcome (1–3). Clarithromycin resistance, which is commonly detected in clinical M. abscessus subsp. abscessus isolates, is associated with treatment success rates of only 25 to 40% (4–7). Combination drug regimens, including the use of two different β-lactams, such as imipenem (a carbapenem) and cefoxitin (a cephamycin), have been tried but the clinical efficacy remains inconclusive (8). Interestingly, we and others have shown that other β-lactams, used together, are synergistic in vitro and in vivo (9–11). For example, the combination of ceftaroline or imipenem with ceftazidime showed a 4- to 32-fold reduction in the MIC₅₀ to either drug alone (9). These findings support the hypothesis that optimal β-lactam treatment will require targeting multiple proteins that are part of the complex and highly redundant battery of enzymes involved in peptidoglycan (PG) biosynthesis (8, 10, 12).

We recently created a saturated Himar1 transposon insertion mutant library in M. abscessus ATCC 19977, comparatively assessed its genomic content to similar libraries in Mycobacterium tuberculosis and Mycobacterium avium, and catalogued genetic elements essential for in vitro growth (13). Included in this analysis were 28 genes predicted to be involved in peptidoglycan biosynthesis and remodeling, three of which were deemed essential to M. abscessus as follows: MAB_2000 (penicillin-binding membrane protein PbpB), MAB_3167c (penicillin-binding lipoprotein), and MAB_4942 (N-acetylmuramoyl-l-alanine amidase CwlM). Among them, pbpB and cwIM are essential in both M. tuberculosis and M. abscessus (13, 14).

The interaction of proteins with essential functions and selected antibiotics (e.g., β-lactams, which target PG biosynthesis) in M. abscessus has only recently been explored due to the availability of new tools to genetically control the expression of essential genes. CRISPR interference (CRISPRi), which uses a nuclease-deactivated Cas9 (dCas9) paired with a single-guide RNA (sgRNA) to sterically hinder transcription at the sgRNA base-pairing genomic locus (15) has been developed as a specific and efficient approach for gene knockdown in both essential and nonessential genes. In this study, we utilized the CRISPRi assay (16, 17) to control essential gene expression in M. abscessus, enabling us to experimentally confirm the essentiality of PbpB and CwIM and to evaluate their contribution to β-lactam susceptibility.

The plasmid pLJR962 (Addgene no. 115162) (16), containing an anhydrotetracycline (ATc)-inducible dCas9_Sth1 (from Streptococcus thermophilus) and an ATc-inducible sgRNA along with a kanamycin selection marker was used as the vector for CRISPRi in M. abscessus. sgRNA was designed using the method described recently (17) in Geneious 9.1.5 software by identifying Sth1 dCas9 protospacer adjacent motif (PAM) sequences against the reference ATCC 19977 genome. We then extracted ∼26 nucleotide sgRNA targeting sequences upstream of each PAM, and only sgRNA targeting sequences in which the transcription-initiating nucleotide was an “A” or “G” were kept for further processing (17) (Table 1). The sgRNA forward and reverse oligonucleotides were ordered from IDT (Integrated DNA Technologies, Inc., Coralville, Iowa), annealed, and ligated into the pLJR962 backbone (pLJR962-sgRNA) (16, 17). pLJR962-sgRNA constructs were initially created in Escherichia coli DH10B and then isolated and electroporated into M. abscessus ATCC 19977 with selection for kanamycin.

TABLE 1.

Primers used in this study

Oligonucleotide name	Sequences 5′–3′	Purpose
sth-cwlM-gRNA-F2	GGGAGGAAGTTCCCGTCGACCAGACCGGT	cwlM sgRNA
sth-cwlM-gRNA-R2	AAACACCGGTCTGGTCGACGGGAACTTCC
sth-pbpb-gRNA-F2	GGGAGGCTGAATGGCCGCGATGGCCAAGGT	pbpB sgRNA
sth-pbpb-gRNA-R2	AAACACCTTGGCCATCGCGGCCATTCAGCC
cwlM_RT-F(n)	CTACCACTTCGGCAACCTAC	cwlM RT PCR
cwlM_RT-R(n)	GCCACTTCTCGCTGAATGA
pbpb_RT-F(n)	GATTCATGGACCTGGTGGAC	pbpB RT PCR
pbpb_RT-R(n)	TCGATCGGTGGCACAATAC
MABrpo-F1(RT)	ATCTCGGTGGTCAGGAAGTA	control RT PCR
MABrpo-R1(RT)	TGCTGTCCTCGAACAACATC

We firstly examined the essentiality of pbpB and cwIM under different degrees of ATc-induced gene inhibition. In brief, overnight cultures of pLJR962 ATCC 19977 transformants with or without target sgRNA were diluted to an optical density at 600 nm (OD₆₀₀) of 0.2 and then serially diluted and plated (10 μl) on 7H10 agar plates supplemented with ATc at concentrations ranging from 0 to 500 ng/mL. As shown in Fig. 1, we identified sgRNAs targeted to each gene that successfully suppressed the growth of M. abscessus at ATc concentrations of ≥100 ng/mL. Plating of the dilutions provided a direct method to quantify the CFU and the effects of CRISPRi. For either construct targeting cwIM or pbpB, ATc concentrations of 100 and ≥150 ng/mL reduced the CFU counts by ≥1 and ≥4 log₁₀, respectively, compared to the empty vector control, confirming the essentiality of both genes in M. abscessus under these in vitro conditions.

FIG 1.

The growth of M. abscessus ATCC 19977 CRISPRi constructs exposed to different ATc concentrations.

The same CRISPR plasmids were electroporated into a Mycobacterium bolletii clinical isolate and evaluated for growth in the presence of differing concentrations of ATc. The results mimicked the findings that we observed with M. abscessus ATCC 19977, supporting the finding that cwlM and pbpB are also essential in M. bolletii (data not shown).

With the knowledge that cwlM and pbpB are predicted to be involved in peptidoglycan synthesis in M. abscessus (13, 18), we investigated whether reducing their gene expression would impact their susceptibility to β-lactam antibiotics. We determined the imipenem and cefoxitin MICs against M. abscessus ATCC 19977 to be 2 μg/mL and 8 μg/mL, respectively, using the microdilution method in 7H9 broth (19, 20). In the absence of ATc, MIC values remained constant with and without the CRISPRi plasmids. To test whether inhibiting the gene expression of cwIM and pbpB affects susceptibility to β-lactams, we plated each strain on agar containing imipenem or cefoxitin, with or without ATc at 50 ng/mL, a concentration that does not reduce the growth of cwlM and pbpB CRISPRi constructs in the absence of other drugs (Fig. 1). Imipenem and cefoxitin were tested at one-half and 1× MIC. This plating approach provided a direct method to evaluate any synergistic drug/protein interactions. Synergy was defined as ≥2-log₁₀ reduction in CFU at 72 h with the combination of antibiotic and ATc compared with the antibiotic alone. As shown in Fig. 2, one-half of the MIC of imipenem (1 μg/mL) reduced CFU of both cwlM and pbpB CRISPRi constructs by approximately 1 log₁₀ compared to the no-drug control in the absence of ATc induction, while imipenem 2 μg/mL caused a >2-log₁₀ reduction of CFU, consistent with its MIC. The addition of ATc at 50 ng/mL caused a 3-log₁₀ reduction in CFU of the cwlM and pbpB CRISPRi constructs compared to the empty vector control in the presence of imipenem. Addition of ATc caused a smaller, ≤1-log₁₀ reduction in CFU of the cwlM and pbpB CRISPRi constructs compared to the empty vector control in the presence of cefoxitin, which did not meet our definition of synergy. We also tested the synergistic activity between imipenem and ATc against the M. bolletii strain with cwlM and pbpB CRISPRi constructs, and a similar synergy was observed in the presence of imipenem 2 μg/mL (one-half MIC) and 50 ng/mL ATc (see Fig. S1 in the supplemental material).

FIG 2.

The interaction of ATc and either imipenem or cefoxitin and their effects on the growth of M. abscessus ATCC 19977 CRISPRi constructs.

In addition, we tested the cwIM and pbpB transcriptional response in the presence of CRISPRi constructs with 50 ng/mL ATc treatment. In brief, overnight cultures of cwIM and pbpB CRISPRi ATCC 19977 strains were inoculated onto 7H10 agar plates with and without 50 ng/mL ATc and incubated for 72 h at 37°C. Total RNAs were extracted with FastPrep (MP Biomedicals) and RNeasy minikit (Qiagen), followed by on-column DNase digestion using RNase-Free DNase set (Qiagen) with an additional TURBO DNase treatment (ThermoFisher). Reverse transcription-quantitative PCR (qRT-PCR) was run in an Mx3005 real-time PCR system using qRT-PCR Brilliant II SYBR kit (Agilent). In the presence of ATc, transcription of cwIM and pbpB was reduced by 28% ± 14% (mean ± standard deviation [SD]) and 36% ± 10%, respectively.

Essential bacterial proteins can be important therapeutic targets. Our recent sequencing of a saturated transposon library created in the M. abscessus ATCC 19977 clinical isolate (13) provided the first comprehensive prediction of essential genes in the genome of this emerging pathogen. However, such essentiality predictions require experimental confirmation. Genetic constructs enabling conditional knockdown of gene expression are an important means of confirming essentiality predictions and investigating whether inhibiting essential proteins can provide therapeutic benefit. With a focus on peptidoglycan biosynthesis enzymes and their importance as β-lactam drug targets, we identified two genes predicted to be essential in both M. abscessus and M. tuberculosis and targeted them for CRISPRi to confirm their essentiality and to study the consequences of suppressing their expression in combination with each of the first-line β-lactams used in the treatment of M. abscessus infections.

Although both cwIM and pbpB have been annotated, there is currently no experimental data to directly dissect their biology in M. abscessus. However, their homologues in M. tuberculosis, both shown to be essential, have been studied to elucidate their roles in peptidoglycan biosynthesis and remodeling. The M. tuberculosis CwlM protein (Rv3915) is a unique cytoplasmic regulatory protein that, when phosphorylated, interacts with and dramatically increases the activity of MurA, the first enzyme in peptidoglycan precursor synthesis (21–23). Under nutrient starvation, CwlM is dephosphorylated, and peptidoglycan synthesis is arrested (23). We hypothesized that the M. abscessus CwlM protein has a similar function in peptidoglycan synthesis. The MAB_2000 (pbpB) gene homologue in M. tuberculosis (Rv2163c) encodes PbpB, also known as PBP-3 or FtsI, an essential d,d-transpeptidase required for cell division and targeted in drug discovery programs (24, 25).

The CRISPRi approach provided the opportunity to control gene expression with ATc titration, repress essential gene expression, and evaluate the consequences. Given the likelihood that both cwIM and pbpB encode for proteins vital for peptidoglycan synthesis in M. abscessus, we investigated their interaction with β-lactams to determine whether suppressing their gene expression would alter susceptibility to these important antibiotics. Although we did not observe reductions in CFU when cwIM or pbpB expression was inhibited approximately 25 to 35% by 50 ng/mL ATc, we did observe a significant 3-log₁₀ reduction in CFU in the presence of imipenem for both genes and a modest effect in the presence of cefoxitin. One hypothesis to explain the potentiation of imipenem is that it targets one or both essential M. abscessus proteins, and the result of CRISPRi inhibiting gene expression is to create a “hypomorph” that is inhibited by lower drug concentrations. Indeed, meropenem and faropenem inactivate PbpB of M. tuberculosis (26). A related possibility is that imipenem inhibits one or more proteins in the same peptidoglycan synthesis and remodeling pathway(s) as PbpB or CwlM, and it is potentiated by reduced activity of the pathway resulting from cwIM or pbpB silencing. For example, reduced expression of cwIM may globally reduce PG synthesis by inhibiting precursor synthesis, thus causing additive effects with subinhibitory imipenem concentrations. One hypothesis for the greater potentiation of imipenem compared to cefoxitin is that imipenem is a more effective inhibitor of other enzymes or pathways that are able to compensate for the reduced activity of PbpB or CwlM. This is supported by the observation that carbapenems inhibit l,d-transpeptidases more than cephalosporins (8) and that imipenem is known to target multiple PG synthesis enzymes in comparison to cefoxitin, including l,d-transpeptidases (Ldt_Mab1, Ldt_Mab2, Ldt_Mab4, Ldt_Mab5) and d,d-carboxypeptidase (12).

Although further studies are needed to investigate these and other hypotheses, the results presented here demonstrate that CRISPRi provides an experimental approach to catalogue drug/target interactions, and this strategy can be extended to target these and other peptidoglycan remodeling proteins and investigate the consequences of inhibiting multiple genes versus different drug combinations in the same assay.