Identification of differentially expressed mRNA in prokaryotic organisms by customized amplification libraries (DECAL): the effect of isoniazid on gene expression in Mycobacterium tuberculosis

Abstract

Understanding the effects of the external environment on bacterial gene expression can provide valuable insights into an array of cellular mechanisms including pathogenesis, drug resistance, and, in the case of Mycobacterium tuberculosis, latency. Because of the absence of poly(A)+ mRNA in prokaryotic organisms, studies of differential gene expression currently must be performed either with large amounts of total RNA or rely on amplification techniques that can alter the proportional representation of individual mRNA sequences. We have developed an approach to study differences in bacterial mRNA expression that enables amplification by the PCR of a complex mixture of cDNA sequences in a reproducible manner that obviates the confounding effects of selected highly expressed sequences, e.g., ribosomal RNA. Differential expression using customized amplification libraries (DECAL) uses a library of amplifiable genomic sequences to convert total cellular RNA into an amplified probe for gene expression screens. DECAL can detect 4-fold differences in the mRNA levels of rare sequences and can be performed on as little as 10 ng of total RNA. DECAL was used to investigate the in vitro effect of the antibiotic isoniazid on M. tuberculosis, and three previously uncharacterized isoniazid-induced genes, iniA, iniB, and iniC, were identified. The iniB gene has homology to cell wall proteins, and iniA contains a phosphopantetheine attachment site motif suggestive of an acyl carrier protein. The iniA gene is also induced by the antibiotic ethambutol, an agent that inhibits cell wall biosynthesis by a mechanism that is distinct from isoniazid. The DECAL method offers a powerful new tool for the study of differential gene expression.

Figure 1

Schematic representation of DECAL. (A) Generation of CAL. A cosmid library is screened for clones that contain ribosomal DNA sequences. Nonribosomal cosmids are digested into similar-sized fragments, gel-purified, ligated to PCR adapters, and PCR-amplified. (B) Positive selection and hybridization. Reverse-transcribed RNA samples are hybridized to a ribosomal DNA-free CAL, washed, then amplified to generate PCR probes. (C) The probes are labeled and hybridized to replicate colony arrays of genomic plasmid libraries. Colonies that hybridize with differing intensities to two PCR probes are selected for evaluation of differentially expressed sequences.

Figure 2

Hybridization of PCR probes to genomic DNA and plasmid digests. DECAL was performed by using RNA extracted from M. tuberculosis H37Rv cultures that were either untreated (INH−) or treated with isoniazid 1.0 μg/ml for 18 hr (INH+). (A) Radiolabeled INH− cDNA (before positive selection with CAL) and radiolabeled INH− and INH+ PCR probes (after positive selection with CAL and amplification) were hybridized to H37Rv genomic digests. The cDNA hybridized almost exclusively with a single band of ribosomal DNA. The INH− PCR probe and INH+ PCR probe both hybridized to multiple sequences in the M. tuberculosis chromosomal digests, but showed no hybridization to the ribosomal band. (B) Southern blots of M. tuberculosis H37Rv genomic DNA digested with PvuII and PstI digests of six plasmids (P1–P6) that hybridized differentially to the PCR probes on colony array screening. Southern blots were hybridized with radiolabeled INH− PCR probe (Upper) or INH+ PCR probe (Lower). The INH− PCR probe hybridized exclusively to P6. The INH+ PCR probe hybridized almost exclusively to P1 and preferentially to P2 and P3. P4 and P5 did not hybridize differently to the two probes and are unlikely to code for isoniazid-induced genes.

Figure 3

Induction of iniA after treatment with different antibiotics. Autoradiographs of a Northern blot containing RNA from M. tuberculosis cultures treated either with no antibiotics, 0.01 μg/ml isoniazid, 0.1 μg/ml isoniazid, 1 μg/ml isoniazid, 5 μg/ml ethambutol, 5 μg/ml streptomycin, and 5 μg/ml rifampin. The blots were hybridized first with an iniA DNA probe (Upper) to examine iniA induction; the blot was then stripped and rehybridized with a 16S probe (Lower) to confirm equal RNA loading.

Figure 4

RT-PCR of differentially expressed genes. (A) RNA was extracted from log-phase M. tuberculosis strain Erdman either without (lanes 1–3) or with (lanes 4–6) isoniazid added to the bacterial cultures for the last 18 hr. RNA from both cultures was equalized by comparison of the 23S band intensity. RT-PCR using three 10-fold dilutions of each RNA and either iniA, asd, or 16S specific primers was performed. Induction of iniA and suppression of asd by isoniazid are demonstrated. The amount of 16S RT-PCR product is similar for equivalent dilutions, indicating equal amounts of starting RNA. Lanes 7 and 8 are minus RT controls, and lane 9 is a negative PCR control. (B) Lack of iniA induction in an isoniazid-resistant strain. Cultures of isogenic BCG strain ATCC35735, which is susceptible to isoniazid (lanes 1–6), or ATCC35747, which is resistant to isoniazid (lanes 7–12), were incubated either in the presence or absence of isoniazid for the last 18 hr. Three 10-fold dilutions of RNA extracted from each culture were tested by RT-PCR for iniA induction. Induction is seen only in the INH-susceptible strain. Lanes 13–16 are minus RT controls, and lane 17 is a negative PCR control containing no added template.

Figure 5

Limits to distinguishing differences between samples. Ten-fold decreasing amounts of in vitro-transcribed M. tuberculosis inhA mRNA (tube 1 contained 1 × 10¹¹ molecules; tube 2, 1 × 10¹⁰ molecules; tube 3, 1 × 10⁹ molecules; tube 4, 1 × 10⁸ molecules; tube 5, 1 × 10⁷ molecules; tube 6, no added molecules) and 4-fold increasing amounts of ask/asd mRNA (tube 1 contained no added molecules; tube 2, 4 × 10⁶ molecules; tube 3, 1.7 × 10⁷ molecules; tube 4, 6 × 10⁷ molecules; tube 5, 2.5 × 10⁸ molecules; tube 6, 1 × 10⁹ molecules) were added to six tubes. Each tube also contained 1 μg of BCG total RNA. DECAL was performed separately for each tube. The PCR probes then were hybridized to six Southern blots containing ask/asd DNA, inhA DNA, and M. tuberculosis H37Rv genomic digests. Autoradiography exposure was equalized to the hybridization intensity of the H37Rv bands.

Figure 6

Applying DECAL to small amounts of starting material. Ten-fold decreasing amounts (1 × 10⁹ and 1 × 10⁸ molecules) of inhA mRNA and 4-fold increasing amounts (1 × 10⁸ and 4 × 10⁸ molecules) of ask/asd mRNA were added to two tubes, each containing 1 μg of BCG total RNA. The tubes were reverse-transcribed with biotin random primers, and serial 10-fold dilutions of the cDNA (equivalent to 1 μg, 100 ng, and 10 ng of starting RNA) were subjected to the DECAL method. The resulting PCR probes were hybridized to duplicate Southern blots of a genomic M. tuberculosis H37Rv digest, inhA DNA, and ask/asd DNA to assess for the presence of detectable differences in inhA and ask/asd signal. Autoradiography exposure was equalized to the hybridization intensity of the H37Rv bands.