Identifying Vulnerable Pathways in Mycobacterium tuberculosis by Using a Knockdown Approach

Abstract

We constructed recombinant strains of Mycobacterium tuberculosis in which expression of specific genes was downregulated to identify vulnerable drug targets. Growth phenotypes in macrophages and culture were used to rank targets: the dprE1, clpP1, and fadD32 operons were the best targets and glnA1, glnE, pknL, regX3, and senX3 were poor targets.

TEXT

Novel drug targets are required for tuberculosis; essentiality is a common first step in validating a target, but there is increasing interest in identifying vulnerable targets for which incomplete inhibition effects a lethal phenotype (1, 7, 10). We utilized the tetracycline-inducible system to generate knockdown strains of a number of genes/operons in Mycobacterium tuberculosis to determine essentiality and vulnerability simultaneously. We selected genes involved in protein turnover(clpP1 and clpP2), cell wall biosynthesis (fadD32 and dprE1 operons), nitrogen assimilation (glnA1 and glnE), and gene regulation (pknL, regX3, and senX3) (Fig. 1 A).

Fig. 1.

Construction of recombinant strains by promoter replacement. (A) Chromosomal arrangement of the targeted genes. Arrows indicate the genes expressed; lines indicate the region of the target gene amplified. (B) Schematic of a recombination delivery plasmid carrying the 5′ end of the gene cloned under the control of P_tet. The TetO operator sites located within the promoters are indicated (Op). Recombination between the plasmid and the chromosomal gene results in replacement of the native promoter with the P_tet promoter. A nonfunctional fragment of the 5′ end of the gene is also incorporated.

Knockdown strains for each gene/operon were constructed as follows: P_tet was cloned from pTACT2 (3) into p4NIL (3) as a NotI fragment; for each gene, a promoterless 5′ region was cloned as a PacI fragment downstream (Table 1). Plasmids were transformed into M. tuberculosis (5), and single-crossover recombinants were selected (11) (Fig. 1B). No recombinants were obtained when the medium lacked tetracycline. When tetracycline (100 to 200 ng/ml) was added, 3 to 18 transformants were obtained. Genotypes were confirmed by Southern hybridization (not shown).

Table 1.

Primers^a

Primer	Sequence
clpP1 F	TTAATTAAGTGAGCCAAGTGAGGAC
clpP1 R	TTAATTAAGCTTGCCCTTGGTACCTG
glnA1 F	TTAATTAAGTAAAGGAGCATTCTGTGACGGAA
glnA1 R	TTAATTAACCTGCCATCAGCATGGCCGAGAAC
glnE F	TTAATTAACTTTGCGCTCCGTCGTG
glnE R	TTAATTAAGGCCATTAGCACGTAGT
fadD32 F	TTAATTAAATGTTTGTGACAGGAGAGAGTG
fadD32 R	TTAATTAACTATGTAGAGGTGGTCCTTGAA
pknL F	TTAATTAACCGTGGTCGAAGCTGGCACGAG
pknL R	TTAATTAACCCAGGATGACGCCGGTAGACG
regX3 F	TTAATTAAATGACCAGTGTGTTGATT
regX3 R	TTAATTAAATCGCTCATCTCCGAGTCGT
senX3 F	TTAATTAACCTTGTGACTGTGTTCTCGG
senX3 R	TTAATTAAATGAGCACCTTCTCGGCGAA
16S rRNA F	GATCCTGGCTCAGGACGA
16S rRNA R	CACCTTCCGGTACGGCTAC

^aThe primers used for cloning 5′ gene fragments are listed; each primer contained a PacI restriction sit (TTAATTAA) at the 5′ end. Amplification primers for the 16S rRNA are also given.

We analyzed growth of the recombinant strains and obtained a range of phenotypes which reflect cellular requirements for viability; in particular, the cell wall biosynthesis genes gave rise to severe phenotypes. Growth of the dprE1 knockdown strain was severely reduced without tetracycline (Fig. 2A); the fadD32 recombinant showed such poor growth that it was impossible to culture in liquid medium. Recombinant strains for the clpP1 operon (Fig. 2B), glnE (Fig. 2C), and the senX3-regX3 operon (Fig. 2D) showed reduced growth, reflecting a requirement for protein turnover, control of nitrogen assimilation, and gene regulation by RegX3, respectively. The P_tet-glnA1 knockdown strain grew normally even without tetracycline (Fig. 2C), although the deletion strain requires l-glutamine (14). Similarly, the P_tet-glnE strain was viable (Fig. 2C) and did not require l-methionine sulfoximine and l-glutamine for growth as described for the glnEΔ strain (4). These data suggest that a level of expression from the noninduced promoter occurs. In all cases, addition of tetracycline did not restore normal growth, suggesting that expression is still relatively low in the induced state. In contrast, growth of the P_tet-pknL (Fig. 2B), P_tet-glnA1 (Fig. 2C), and P_tet-regX3 (Fig. 2D) strains was normal even without tetracycline.

Fig. 2.

Growth of recombinant knockdown strains in axenic culture. Strains were cultured in 3 ml liquid medium in 16-mm-diameter glass tubes containing a 12-mm magnetic stirrer and stirred at 120 rpm. Open symbols, no tetracycline; closed symbols, 200 ng/ml tetracycline. Data are the averages of three independent recombinant strains. Error bars are omitted for clarity. (A) dprE1 operon. (B) clpP1 operon and pknL. (C) glnA1 and glnE. (D) senX3 operon and regX3. OD₅₈₀, optical density at 580 nm; WT, wild type.

We looked at the ability of each recombinant strain to grow in human THP-1 monocyte cells (8). The fadD32 knockdown was severely compromised for growth (Fig. 3 A); bacterial numbers were significantly lower than those of the wild type (P < 0.01) and the effect of fadD32 depletion was clearly bactericidal. All other knockdown strains grew to the same level as did the wild type in resting macrophages (data not shown).

Fig. 3.

Growth of recombinant knockdown strains in the macrophage infection model. Strains were used to infect THP-1 cells at a multiplicity of infection of 1:1 (8). Bacteria were harvested from monocytes, and CFU were determined. Macrophages were activated by the addition of 100 units/ml of IFN-γ 24 h prior to infection. (A) Resting macrophages. (B) Activated macrophages. Open symbols, no tetracycline; closed symbols, addition of tetracycline. Data are the means and standard deviations from three independent samples.

More pronounced phenotypes were seen in activated macrophages (Fig. 3B); GlnA1 and GlnE knockdown strains failed to increase in numbers, demonstrating that the effect of depletion was static. The clpP1 operon knockdown had an interesting phenotype (Fig. 3A and B): in resting macrophages, the number of CFU recovered was similar to that of wild type, but colonies were much smaller, suggestive of cell damage resulting in an extended lag or recovery phase during outgrowth. Macrophage infections are limited to 7 days (at which point macrophage viability declines), so the infection phase could not be extended to determine if damaged cells would be killed eventually. In activated macrophages there was a significant decrease in bacterial numbers at 7 days, confirming that ClpP1-ClpP2 depletion reduced virulence. This may also account for the strain appearing more attenuated in the presence of tetracycline than without tetracycline, since CFU in this case cannot tell the complete story regarding killing by the macrophage. Alternatively, tetracycline may induce expression of genes which sensitize the ClpP1-ClpP2 knockdown to macrophage-mediated killing.

The data from the macrophages were broadly consistent with the culture data: the fadD32 knockdown was severely attenuated in both systems, clpP1 operon and glnE knockdown strains showed reduced growth in both systems, and the pknL and regX3 strains were unaffected in either. The glnA1 mutant was attenuated in macrophages only, suggesting that its role is more important during infection; the senX3-regX3 knockdown strain was fully virulent but showed reduced growth in medium.

We looked at the level of expression of glnA1, glnE, and pknL in the wild-type and recombinant strains (Table 2 and Table 3). The relative expression levels were highest for glnA1, which is produced to a high level (6, 13). The relative level of expression of native pknL was low. For all three genes, the expression levels from P_tet were lower than the native levels; very little induction was seen in the presence of tetracycline, although for pknL a small increase was seen. These data confirm that the strains are true knockdowns but show that P_tet is not fully functional for induction. glnA1 expression was reduced in the wild-type strain in the presence of tetracycline; this may reflect changes in the cell wall and secreted proteins resulting from changes in efflux, as tetracycline induces multiple efflux pumps and other genes (2).

Table 2.

Primers and probes used in this study^a

Primer or probe	Sequence
glnA1qF	CATCATCAAGAACACCGCC
glnA1qR	TTCCACAGCGACTGATGAC
glnA1qP	CAGAACGGCAAAACGGTCACGTTCATGC
glnEqF	TGCGGGAGGAACTCAAGAAG
glnEqR	CCAGCAGCGGTTGATAGAAG
glnEqP	GTGCGGGTGTCGAAGTTACACACCAAA
pknLqF	TGGTCTACGAGCTGCTAAC
pknLqR	TCGATTACAGCACTGGCAC
pknLqP	TCGATTGCCTACCAACGGCTTGATGCTGAC
sigAqR	AGCAAAGTGAAGGACACC
sigAqF	GAAATCCAGCAATACGCCC
sigAqP	AGCCAGCTTGGCGATTTCATCGAAGACAGC

^aPrimers for quantitative PCR (qF and qR) and probes (qP) are listed for each gene.

Table 3.

Expression levels of target genes under the control of the native promoter and P_tet^a

Gene	Wild type		Recombinant Ptet
	With tetracycline	Without tetracycline	With tetracycline	Without tetracycline
glnA1	9.6 ± 2.6	54.8 ± 6.7	0.5 ± 0.2	0.8 ± 0.5
glnE	15.7 ± 2.1	8.8 ± 3.3	0.04 ± 0.03	0.5 ± 0.05
pknL	3.4 ± 0.5	3.5 ± 0.5	2.1 ± 0.2	1.04 ± 0.3

^aStrains were grown in liquid medium in the presence of 0 or 200 ng/ml tetracycline. RNA was prepared as described elsewhere (12) and subjected to reverse transcription quantitative-PCR analysis. Standard curves were generated for each primer-probe set using genomic DNA and used to calculate copy number for cDNA reactions. Copy number was corrected by subtracting background from genomic DNA in the samples (no reverse transcription reaction). Data were normalized to sigA and are the averages and standard deviations of a minimum of three independent samples.

Taken together, these data are informative in predicting the best targets for intervention to achieve cell death. GlnA1 and GlnE are poor targets, since growth is not completely restricted even when mRNA levels were reduced to <5%, suggesting that complete inhibition is required to kill the cells. PknL is a poor target, since reduction of the naturally low level of expression had no effect.

The growth phenotypes of our recombinant strains demonstrated that knockdown of both the operons involved in cell wall biosynthesis had the largest effect on growth and viability. The recombinant strains for the fadD32 and dprE1 operons were characterized in more detail. Recombinant strains expressing fadD32 from P_tet gave rise to colonies with a distinct morphology (Fig. 4 A). Growth was slow, taking over 8 weeks to form colonies of an appreciable size, and it was impossible to passage strains. The P_tet-fadD32 strain had increased sensitivity to several antibiotics: the MIC for rifampin was reduced 4-fold to 0.25 μg/ml and that for ethambutol was reduced 2-fold to 1 μg/ml. Cells became sensitive to ampicillin with a MIC of 1 μg/ml (wild type, >32 μg/ml).

Fig. 4.

Characterization of vulnerable cell wall targets using recombinant knockdown strains. Colony morphology of the wild-type (A) and P_tet-fadD32 operon knockdown (B) strains. Species identity was confirmed by 16S rRNA locus amplification and sequencing.

The P_tet-dprE1 knockdown strain was severely compromised for growth in culture (Fig. 2A). Western analysis confirmed that expression levels were reduced (not shown). The P_tet-dprE1 strain demonstrated increased sensitivity to compounds targeting the cell wall; the MIC for benzothiazinone, which targets DprE1 (9), was reduced 2-fold to 0.25 ng/ml and that for ethambutol was reduced 4-fold to 0.5 μg/ml.

Conclusion.

We developed a rapid method for assessing gene/operon vulnerability in which promoter replacement is achieved by recombination; this should prove useful for rapid evaluation of novel targets or construction of knockdown strain panels. Targeting cell wall biosynthesis had the most severe phenotypic effect, but depletion of protein turnover and gene regulation also compromised growth.