Beyond antibiotic resistance: the whiB7 transcription factor coordinates an adaptive response to alanine starvation in mycobacteria

SUMMARY

Pathogenic mycobacteria are a significant cause of morbidity and mortality worldwide. The conserved whiB7 stress response reduces the effectiveness of antibiotic therapy by activating several intrinsic antibiotic resistance mechanisms. Despite our comprehensive biochemical understanding of WhiB7, the complex set of signals that induce whiB7 expression remain less clear. We employed a reporter-based, genome-wide CRISPRi epistasis screen to identify a diverse set of 150 mycobacterial genes whose inhibition results in constitutive whiB7 expression. We show that whiB7 expression is determined by the amino acid composition of the 5’ regulatory uORF, thereby allowing whiB7 to sense amino acid starvation. Although deprivation of many amino acids can induce whiB7, whiB7 specifically coordinates an adaptive response to alanine starvation by engaging in a feedback loop with the alanine biosynthetic enzyme, aspC. These findings describe a metabolic function for whiB7 and help explain its evolutionary conservation across mycobacterial species occupying diverse ecological niches.

eTOC

Poulton et al. find that the mycobacterial antibiotic resistance gene, whiB7, regulates an alanine starvation response by coordinating a transcriptional feedback loop with the alanine biosynthetic gene, aspC. This extended function for whiB7 may help to explain the evolutionary conservation of this pathway across pathogenic and environmental mycobacteria.

Graphical Abstract

INTRODUCTION

Mycobacteria cause a wide range of diseases and include some of the oldest infections described in human history, leprosy and tuberculosis (TB). Both of these diseases are still prevalent today^1,2, and TB remains the leading cause of death from any single infectious agent.² Infections caused by non-tuberculous mycobacteria, including Mycobacterium abscessus and Mycobacterium avium are increasingly common, especially amongst immunocompromised individuals.³ Although these mycobacterial diseases vary in manifestation, they share the common feature of being difficult to treat.

The reasons why mycobacterial diseases are difficult to treat are complex and multifactorial.^4–7 However, one major difficulty is the high level of intrinsic drug resistance of mycobacteria.⁸ Intrinsic drug resistance can be attributed in part to the relatively impermeable mycobacterial outer membrane, which is composed of a thick layer of waxy mycolic acids and other glycolipids which slow the uptake of a wide range of compounds.^9,10 Further, mycobacteria have a large number of additional intrinsic drug resistance mechanisms that act via drug efflux¹¹, drug modification¹², target modification¹³, and target rescue.¹⁴ Although intrinsic drug resistance determinants vary across mycobacterial species, the whiB7 pathway has emerged as a conserved and key mechanism limiting the activity of different antibiotics against diverse mycobacterial pathogens.^15–18 WhiB7 is a transcriptional activator that promotes the expression of a suite of intrinsic drug resistance determinants including: 1) the tap drug efflux pump¹⁹; 2) eis, an acetyltransferase that modifies and inactivates aminoglycosides¹²; 3) erm, a ribosomal RNA methyltransferase whose activity prevents macrolide antibiotic engagement with the ribosome¹³; 4) hflX, a ribosome splitting factor that rescues drug-stalled ribosomes¹⁴; and many other genes.¹⁷

WhiB7 was originally discovered in the soil bacterium Streptomyces lividans.¹⁶ whiB7 was hypothesized to have originated as a self-protection mechanism in antibiotic-producing bacteria, and then been acquired and retained by an ancestral soil-dwelling actinobacteria to protect against antibiotics being produced by other soil bacteria.¹⁶ Why WhiB7 would be retained in pathogenic mycobacteria like Mtb and M. leprae, long after their progenitors left the antibiotic containing soil, has remained an open question.

WhiB7 expression is activated in response to and protects against a number of different antibiotics, particularly ribosome-targeting antibiotics like macrolides, lincosamides, and spectinomycin.¹⁶ In both Streptomyces and Mycobacteria whiB7 expression is controlled by an upstream open reading frame (uORF)-mediated transcriptional attenuation mechanism in its 5’ regulatory region.^20–22 Under non-stressed conditions, whiB7 transcription initiates at a distal upstream transcriptional start site (TSS) and proceeds through a short uORF. Efficient translation of the uORF during un-stressed conditions results in the formation of a rho-independent terminator which stops transcription prior to the start of the whiB7 ORF.²² During conditions of translation stress, for example the presence of a macrolide antibiotics, the uORF is inefficiently translated. This results in the formation of an antiterminator structure and transcriptional readthrough into the whiB7 ORF.²² Elevated WhiB7 levels then initiate a positive feedback loop by binding to and activating transcription from the whiB7 promoter, generating whiB7 mRNA levels as much as 1,000 times higher than basal whiB7 expression under unstressed conditions.^23,24

Beyond ribosome-targeting antibiotics, numerous groups have identified epistatic mutations that result in constitutive whiB7 expression and decreased drug sensitivity. Gomez et al. identified M. smegmatis mutants harboring mutations in rplO, rplY, and rplF that showed broad, low-level antibiotic resistance, in part due to constitutive whiB7 expression.²⁵ Schraeder et al. identified M. smegmatis mutants in the arginine biosynthetic genes argA and argD that led to constitutive whiB7 expression and promoted tolerance to aminoglycoside antibiotics.²⁶ Our group identified partial loss-of-function mutations in the essential translation factor ettA in Mtb clinical isolates that confer low-level resistance to multiple antibiotics, in part through constitutive whiB7 expression.²⁷ Lastly, through yet to be defined mechanisms, WhiB7 expression in Mtb is activated in response to host-derived stressors during ex vivo infection in macrophages.²⁸ Collectively, these findings suggest that alterations in various biological processes can lead to whiB7 induction.²³

Here, we set out to comprehensively define the biological pathways whose inhibition leads to whiB7 induction. By coupling a whiB7 expression reporter with a genome-wide CRISPRi screen, we identified ~150 genes whose inhibition results in constitutive expression of whiB7. In addition to the expected functions in ribosome biogenesis and translation, we identified a diverse set of hit genes involved in other cellular pathways including redox homeostasis and transcription. We show that the responsiveness of whiB7 to amino acid starvation is determined by the amino acid composition of the uORF, which explains species-specific differences in whiB7 induction signals. To identify physiological pathways beyond intrinsic resistance that may be dependent on whiB7, we performed a CRISPRi vulnerability screen. This screen identified a regulatory feedback loop where whiB7 can sense the deprivation of alanine to upregulate expression of the alanine aminotransferase, aspC, demonstrating a physiological role for whiB7 outside of its canonical role in intrinsic drug resistance.

RESULTS

Development of whiB7 reporter systems in mycobacteria

To better understand the complex set of signals that can induce whiB7 expression, we sought to globally identify genes whose inhibition results in constitutive whiB7 expression. To identify such genes, we built two whiB7 transcriptional reporters.²³ The entire whiB7 regulatory region (500 bp upstream of the whiB7 ORF start codon) was cloned upstream of either a zeocin resistance gene (zeoR) or the mScarlet fluorescent reporter (Fig. 1A). Consistent with faithful recapitulation of whiB7 regulation, treatment of the zeoR reporter strain with a subinhibitory concentration of the translation inhibitor clarithromycin produced phenotypic zeocin resistance (Fig. 1B). For reference, the zeocin resistance gene driven by the strong, constitutive EM7 promoter (PEM7) is shown (Fig. 1B). Faithful reporter function could be seen at the genetic level as well. CRISPRi-mediated knockdown of the essential translation factor, ettA, and the arginine biosynthetic gene, argD, resulted in whiB7 induction as evidenced by phenotypic zeocin resistance (Fig. 1C), consistent with prior results.^26,27 The mScarlet reporter strain showed high-level fluorescence in response to clarithromycin but not the cell wall biosynthesis inhibitor isoniazid (Fig. 1D). The mScarlet reporter strain strongly fluoresced in the ettA and argD knockdown strains, but not in non-targeting (NT) CRISPRi control strain (Fig 1E). Taken together, these results show that both genetic reporter systems can be used to identify conditions under which whiB7 expression is induced.

Figure 1. Development and validation of whiB7 reporters.

(A) Genetic architecture of PwhiB7 reporter constructs. Both reporters were cloned into single-copy, integrating plasmids.

(B) Zeocin resistance profiles of M. smegmatis strains (~1,000 colony forming units (CFU)/well) harboring the zeocin resistance gene driven by the indicated promoters. EM7 is a strong, constitutive promoter derived from the T7 promoter. Plates contain either no drug (left) or 100 ng/ml = 0.25X minimum inhibitory concentration (MIC) of the known whiB7 inducing drug, clarithromycin (right). The red rectangle marks clarithromycin-dependent growth of the M. smegmatis PwhiB7:zeoR strain in the presence of 10–20 μg/ml zeocin.

(C) Growth of the indicated M. smegmatis CRISPRi strains on agar plates containing zeocin at 0, 10, or 20 ug/mL. NT = non-targeting; KD = knockdown.

(D) Dose response curves (mean ± s.e.m., n = 3 replicates) of the PwhiB7:mScarlet reporter M. smegmatis strain for clarithromycin (top graph) and isoniazid (bottom graph). Drug dose-response curves (percent growth) are shown in black and mScarlet fluorescence (RFU) are shown in red.

(E) Normalized fluorescence values of the indicated M. smegmatis CRISPRi strains at 0, 12, and 24 hours after addition of ATc to activate CRISPRi (mean ± s.e.m., n = 3 replicates). Both ettA and argD are targeted with two different sgRNAs each, denoted #1 and #2. P300 is a strong, constitutive promoter. EV = empty vector.

(F) Normalized fluorescence values for the PwhiB7:mScarlet reporter strain (top row: M. smegmatis, bottom row: M. tuberculosis) in response to the listed panel of drugs. Fluorescence values are indicative of the highest value obtained at a sub-MIC concentration of each drug. Normalized RFU: M. smegmatis = −18.2 to 2,000; M. tuberculosis = 13.2 to ≥6,000. RIF = rifampicin; LVX = levofloxacin; EMB = ethambutol; INH = isoniazid; BDQ = bedaquiline; STR = streptomycin; AMK = amikacin; CAP = capreomycin; CAM = chloramphenicol; LZD = linezolid; TGC = tigecycline; FUS = fusidic acid; CLR = clarithromycin; AZT = azithromycin; LNC = lincomycin; CND = clindamycin; SPT = spectinomycin; BRT = bortezomib; ECU = ecumicin⁵⁰.

To further validate the sensitivity and specificity of the mScarlet reporter system, a larger panel of antibiotics were tested for their ability to induce whiB7 in both M. smegmatis and M. tuberculosis. Results were largely concordant between the two bacterial species, with several ribosome-targeting drugs including macrolides, lincosamides, and spectinamides resulting in potent whiB7 induction compared to antibiotics targeting other biological processes (Fig. 1F, Fig. S1). For reasons yet unknown, M. tuberculosis and M. smegmatis showed subtle differences in whiB7 induction by several drugs, including chloramphenicol, linezolid, and capreomycin, which may reflect species-specific differences in ribosome protection factors.

Genome-scale CRISPRi epistasis screen identifies a diverse set of genes whose inhibition induces whiB7 expression

To identify genes whose inhibition activates whiB7 expression, we transformed an M. smegmatis strain harboring the PwhiB7:zeoR reporter construct with a genome-scale CRISPRi library (Fig. 2A). This library contains 176,451 sgRNAs targeting 6,642/6,679 of all M. smegmatis genes and includes sgRNAs of varying predicted knockdown efficiencies to enable tunable knockdown of essential genes.²⁹ The library was grown for ~5 generations in the presence of anhydrotetracycline (ATc) to activate CRISPRi, induce target knockdown, and allow for whiB7 induction and production of the ZeoR selectable marker. After this “pre-depletion” phase, the library was plated in the presence of ATc with or without zeocin. After outgrowth, roughly 100-fold fewer colonies were observed on the zeocin-containing vs lacking plates, suggesting that only ~1% of all sgRNAs selected for whiB7 expression (Fig. 2B, Fig. S2A). Spontaneous zeocin resistance, i.e. zeocin resistance in the absence of ATc, was a negligible contributor to the overall number of colonies (Fig. 2B, Fig. S2A). Genomic DNA from colony forming units (CFU) from each set of plates was extracted and used to quantify sgRNA abundance by deep sequencing.

Figure 2. Genome-scale identification of whiB7-inducing CRISPRi strains in M. smegmatis.

(A) M. smegmatis whiB7 induction screen workflow.

(B) CFU quantification from the M. smegmatis screen described in panel A.

(C) Volcano plot showing log2 fold-change (L2FC) values and false discovery rates (FDR) for each gene in the M. smegmatis whiB7 induction screen. The expected hit genes ettA and argD are annotated. Hit genes (n=150) were defined as having a L2FC > 2 and FDR < 0.01.

(D) Functional categories of hit genes from the M. smegmatis whiB7 induction screen.

This screen identified 150 genes whose inhibition led to enrichment on zeocin containing plates (Fig. 2C, Supplemental Data 1). Included amongst the 150 hit genes were ettA and argD, whose inhibition was previously shown to activate whiB7 expression (Fig. 1C,E).^26,27 Genes essential for in vitro growth²⁹ were significantly overrepresented amongst hit genes, making up 105 out of 150 hit genes (70%) despite only comprising ~6% of the M. smegmatis genome, highlighting the utility of tunable CRISPRi to investigate phenotypes of essential genes. Amongst the hit genes, the most represented functional categories were genes involved in amino acid biosynthesis, tRNAs, and tRNA synthetases (Fig. 2D, Supplemental Data 1). Additionally, many hit genes have known or predicted functions in ribosome assembly, maturation, and function. Interestingly, a number of genes involved in transcription initiation, elongation, and termination were identified as hits, possibly reflecting defects in transcription-translation coupling that could perturb uORF translation and formation of the rho-independent terminator, although other mechanisms cannot be excluded.³⁰ Both succinate dehydrogenase and the cytochrome c oxidase were identified in this screen, potentially consistent with previous work demonstrating the responsiveness of whiB7 to redox stress.²³ Note that because of the potential polar effect of CRISPRi knockdown, whiB7 induction could be a result of knockdown specifically of the targeted hit gene, or a downstream gene in a potential operon. Further experiments would be necessary to distinguish direct from polar effects, and thus we present the results of the whiB7 induction screen without considering potential polar effects.

We next validated several hit genes from the various functional categories in both M. smegmatis and M. tuberculosis using individual sgRNAs. In accordance with the screen, knockdown of ettA, the alanine aminotransferase aspC, the transcription antitermination factor nusB, and the threonyl-tRNA synthetase thrS resulted in strong whiB7 induction as assessed by qPCR, whereas knockdown of the non-hit essential gene mmpL3 did not (Fig. S2B,C). Using the PwhiB7:mScarlet reporter system, we further validated a larger panel of hit and non-hit genes in both M. smegmatis (Fig S2D,E) and M. tuberculosis (Fig. S2F). Results with individual strains were largely consistent with the whiB7 induction screen and concordant between the two bacterial species. Consistent with the interpretation that inhibition of amino acid biosynthetic genes leads to whiB7 induction as a result of translation stalling within the uORF, we could chemically complement whiB7 induction in amino acid auxotrophs by provision of the appropriate amino acid (Fig. S2G). whiB7 expression was reversed in an argD CRISPRi strain by provision of arginine.²⁶ Similarly, whiB7 expression was reversed in the aspC knockdown strain by supplying exogenous alanine but not aspartate or glutamate, consistent with prior work demonstrating that contrary to its annotation, aspC acts as an alanine aminotransferase.³¹

Amino acid composition of the uORF dictates the responsiveness of whiB7 to amino acid starvation

Results from the M. smegmatis CRISPRi whiB7 induction screen suggested that auxotrophy for many different amino acids can result in whiB7 induction. Biosynthetic genes and/or tRNA/tRNA synthetase pairs related to 18/20 amino acids were identified as hits (Supplemental Data 1). Interestingly, pharmacological (6-FABA) and genetic (CRISPRi) inhibition of tryptophan biosynthesis³² activated whiB7 expression in M. smegmatis, but this effect was not seen in M. tuberculosis (Fig. 3A–C). This raised the question of what might underly these species-specific differences in whiB7 induction signals. While the WhiB7 protein sequence is highly conserved across mycobacteria²⁰, the whiB7 uORF displays substantial length and sequence variation across different species (Fig. 3D, Supplemental Table 1). Therefore, we interrogated whether coding sequence variation in the uORF could explain the differences in responsiveness to tryptophan deprivation between M. smegmatis and M. tuberculosis.

Figure 3: Modulation of the uORF coding sequence alters the response to amino acid deprivation.

(A-B) Dose response curves (mean ± s.e.m., n = 3 replicates) of the PwhiB7:mScarlet reporter strain for clarithromycin (translation inhibitor) and 6-FABA (tryptophan biosynthesis inhibitor), grown in the presence or absence of 1 mM tryptophan. (A) Top row = M. smegmatis; (B) bottom row = M. tuberculosis. Drug dose-response curves (percent growth) are shown in black and mScarlet fluorescence (RFU) are shown in red. The blue shaded region highlights the differential whiB7 induction of M. smegmatis and M. tuberculosis in response to tryptophan limitation by 6-FABA.

(C) Normalized fluorescence of the indicated PwhiB7:mScarlet M. smegmatis and M. tuberculosis CRISPRi strains 24 hours and 8 days after addition of ATc, respectively (mean ± s.e.m., n = 3 replicates). Statistical significance with respect to each non-targeting CRISPRi strain was calculated using a Student’s t-test; **P< 0.01, ****P< 0.0001, n.s. = non-significant.

(D) Amino acid composition (x-axis) of annotated or predicted whiB7 uORF sequences from the listed bacteria. A black X denotes lack of that amino acid within the indicated whiB7 uORF. M. tb = M. tuberculosis; M. ks = M. kansasii; M. av = M. avium; M. ul = M. ulcerans; M. mr = M. marinum; M. ft = M.fortuitum; M. ab = M. abscessus; M. sm = M. smegmatis; S. co = S. coelicolor; R. jo = R. jostii.

(E) Genetic architecture of the engineered M. smegmatis whiB7 uORF variants, transformed into the $\Delta$whiB7 M. smegmatis strain.

(F) whiB7 ORF mRNA fold-change of the indicated M. smegmatis CRISPRi strains 18 hours after addition of ATc (mean ± s.e.m., n = 3 biological replicates). The M. smegmatis CRISPRi strains harbor the whiB7 uORF variants as depicted in panel (E). whiB7 ORF mRNA fold change is relative to sigA and normalized to the respective non-targeting CRISPRi strain for each uORF variant. Statistical significance with respect to the WT whiB7 uORF strain was calculated for each knockdown mutant using a Student’s t-test; *P< 0.05, **P< 0.01, n.s. = non-significant. WT = wild-type.

The M. smegmatis whiB7 uORF encodes every amino acid except cysteine. In contrast, the M. tuberculosis uORF is lacking several amino acids, including tryptophan (Fig. 3D). According to the current model of whiB7 induction²², amino acid deficiency should stall uORF translation at specific codons when the ribosome is unable to load the cognate, charged aminoacyl-tRNA. Stalled translation then results in formation of an antiterminator structure and transcriptional readthrough into the whiB7 ORF.²² If a particular amino acid is not represented in the uORF (e.g. tryptophan), the inability to produce that amino acid should not stall uORF translation and thus not induce expression of whiB7. To test this model, we generated two variants of the M. smegmatis whiB7 uORF (Fig. 3E). A Trp-less uORF variant was generated where the single tryptophan residue was converted to alanine. A His-less uORF variant had all three histidine residues converted to alanine. None of these mutations are predicted to impact the formation of the whiB7 antiterminator (Fig. S3). We transformed the uORF variants into $\Delta$whiB7 M. smegmatis and then monitored the effects of tryptophan, histidine, and alanine deprivation on whiB7 ORF expression (Fig. 3F). Deprivation of tryptophan by knockdown of trpC (Fig. 3F) or 6-FABA treatment (Fig. S4A) induced strong whiB7 expression in the WT and His-less uORF strains but significantly less whiB7 expression in the Trp-less uORF strain. Conversely, deprivation of histidine by knockdown of hisH induced strong whiB7 expression in the WT and Trp-less uORF strains but did not induce whiB7 in the strain harboring the His-less uORF. Importantly, all strains showed high level whiB7 induction in response to aspC knockdown, since all three uORF variants harbored abundant Ala codons. The failure to induce whiB7 in the uORF variants was not a result of different levels of target gene knockdown (hisH, trpC, aspC) nor due to differences in basal whiB7 expression (Fig. S4A,B). Taken together, these results are consistent with the model that the ability of whiB7 to “sense” the depletion of particular amino acids is determined by the amino acid composition of the whiB7 uORF.

We next analyzed the whiB7 uORF sequence across ~50,000 clinical M. tuberculosis isolates²⁷ to determine whether heterogeneity in this region may influence basal whiB7 expression or the responsiveness of whiB7 to amino acid deprivation. Although we did not observe any instances where a mutation introduced a new amino acid not already present in the whiB7 uORF (Supplemental Data 2), we did identify several rare mutations that led to a robust increase in basal whiB7 expression (Fig. S4C,D).³³ These uORF mutations include a loss-of-start codon mutation as well as several distinct indels which likely alter uORF translation dynamics in such a way that promotes transcriptional readthrough into the whiB7 ORF. Importantly, none of the M. tuberculosis uORF mutations tested in these experiments (Fig. S3C) occur within the transcriptional antiterminator or terminator structures, suggesting that the effects of these mutations are due to altered translation dynamics of the uORF and not simply due sequence-based disruption of key RNA secondary structures (Fig. S3).

whiB7 mediates an adaptive response to alanine starvation

Although whiB7 expression can be induced by the depletion of many different amino acids, we observed that across actinobacterial species, the whiB7 uORF is universally rich in alanine (Fig. 3D). Depending on the species, 25–40% of the whiB7 uORF codes for Ala. Each uORF (Fig. 3D) was significantly enriched (p < 0.001) for alanine and only alanine relative to the respective proteome of each species.³⁴ The canonical role of whiB7 is as an inducible intrinsic resistance mechanism to translation inhibiting antibiotics. What then might account for the strong enrichment for a particular amino acid (Ala) in the whiB7 uORF? Inspired by classic attenuation mechanisms for E. coli tryptophan and leucine biosynthesis^35–37, as well as the observed responsiveness of whiB7 to alanine starvation (Fig. 3F, Fig. S2C–G), we hypothesized that whiB7 may play a more general physiological role in regulating alanine metabolism, independent of its canonical role in antibiotic resistance.

To further explore this potential physiological role for whiB7, we performed a genome-wide differential vulnerability screen (Fig. 4A).²⁹ Differential vulnerability screens identify genes that become more or less sensitive to CRISPRi inhibition between two (or more) genetic backgrounds or growth conditions.²⁹ To perform this screen, the same tunable sgRNA library²⁹ used to identify upstream whiB7 expression activators (Fig. 2A) was transformed into $\Delta$whiB7 M. smegmatis.²⁶ The resulting CRISPRi library was then passaged for approximately 30 generations in the presence or absence of ATc. Every 2.5 or 5 generations, we harvested genomic DNA and analyzed sgRNA abundance by deep sequencing. We then compared gene vulnerability between wild-type M. smegmatis²⁹ and the wwhiB7 mutant. Quantification of gene vulnerability revealed strong concordance between the two strains (R² = 0.963), although there were rare differentially vulnerable genes. We identified a total of 20 differentially vulnerable genes, 9 of which were more vulnerable in $\Delta$whiB7 $(\text{ΔVI}<-2)$ and 11 of which were less vulnerable $(\text{ΔVI}>2)$ in $\Delta$whiB7.

Figure 4: Identification of differential vulnerabilities in $\Delta$whiB7.

(A) Differential vulnerability screen in $\Delta$whiB7 M. smegmatis. This screen identifies genes that become more or less sensitive to CRISPRi inhibition between wild-type and $\Delta$whiB7 M. smegmatis. Please see the Materials and Methods section and Bosch et al.²⁹ for further details on screen analysis.

(B) Expression-fitness relationships for the five indicated M. smegmatis genes. The fitness cost (beta_E) is plotted as a function of predicted sgRNA strength (an estimate of the magnitude of target knockdown). Both aspC, asd, and hisC are more vulnerable in $\Delta$whiB7 M. smegmatis (tourquise).

(C) Growth of the indicated M. smegmatis CRISPRi strains monitored by spotting serial dilutions of each strain on the indicated media. Supplemental alanine and aspartate were added at 1 mM. Note that hypomorphic sgRNAs that are predicted to lead to intermediate levels of knockdown are shown for trpC and mmpL3, as strong sgRNAs leading to high-level knockdown would block growth of both wild-type and $\Delta$whiB7 M. smegmatis and not be relevant controls for differential vulnerability.

(D) Growth of the indicated M. tuberculosis dual-gene knockdown CRISPRi strains in 7H9 + ATc, with or without supplemental alanine (1 mM). Dual NT represents a CRISPRi plasmid encoding two non-targeting sgRNAs; NT + whiB7 KD represents a CRISPRi plasmid encoding a single non-targeting sgRNA and a whiB7 targeting sgRNA; aspC KD + whiB7 KD represents a CRISPRi plasmid encoding one sgRNA targeting aspC and a separate sgRNA targeting whiB7.

The gene which showed the greatest difference in vulnerability between WT and $\Delta$whiB7 was the alanine aminotransferase aspC, although more subtle differences were also observed for aspartate semialdehyde dehydrogenase asd and histidinol-phosphate aminotransferase hisC (Fig. 4B). All three of these genes were also identified as genes whose inhibition resulted in whiB7 induction (Fig. 3A,F, Fig. S2B–F, Supplemental Data 1). The differential vulnerability screen indicated that silencing of hisC, asd, and particularly aspC resulted in a much stronger fitness costs in $\Delta$whiB7 than wild-type M. smegmatis, potentially consistent with our hypothesis that whiB7 plays a physiological role in amino acid metabolism. Interestingly, the vast majority of other amino acid biosynthetic genes and essential processes were not differentially vulnerable (Supplemental Data 3, Fig. 4B), suggesting that whiB7 is specifically required to maintain fitness under physiological conditions in which AspC and to a lesser extent HisC and Asd activity is perturbed.

Using individual CRISPRi strains, we confirmed that aspC and asd were substantially more vulnerable in $\Delta$whiB7 compared to a wild-type M. smegmatis strain (Fig. 4C, Fig. S5A,B). These phenotypes were not a result of a CRISPRi polar effect, as the potential operonic genes downstream of both aspC and asd were not differentially vulnerable. This effect was specific, as silencing of the non-hit control genes trpC and mmpL3 was not associated with a greater fitness cost in $\Delta$whiB7 (Fig. 4C). The fitness defect could be reversed in the aspC CRISPRi strains by supplementing exogenous alanine, but not aspartate (Fig. 4C, Fig. S2G).³¹ These results are also consistent with AspC being the primary alanine biosynthetic enzyme in mycobacteria³¹, as CRISPRi mutants for the alanine dehydrogenase (Ald) show no phenotypic defects and do not show elevated whiB7 expression (Fig. S6A–C). Interestingly, amino acid supplementation exacerbated the growth defect associated with asd knockdown in both the wild-type and $\Delta$whiB7 (Fig. S5C), which may reflect the accumulation of toxic intermediates of aspartate metabolism that cannot be eliminated in the absence of Asd (Fig. S6D). We next validated aspC differential vulnerability in M. tuberculosis. Concurrent knockdown of both aspC and whiB7 resulted in stronger growth inhibition than knockdown of aspC alone. This effect could be reversed by alanine supplementation, but not glutamate or aspartate (Fig. 4D, Fig. S5D). Knockdown of whiB7 did not exacerbate the growth defect associated with trpC knockdown, demonstrating the specificity of this effect for aspC.

Why is aspC more vulnerable in $\Delta$whiB7? Previous RNA sequencing and ChIP sequencing data suggests that aspC is part of the whiB7 regulon in M. smegmatis¹⁷, M. abscessus^17,38, and S. coelicolor³⁹. Consistent with these results, we were able to confirm the interaction of WhiB7 with the M. smegmatis aspC promoter by chromatin immunoprecipitation (ChIP) RT-qPCR (Fig. 5A). These observations, combined with the increased vulnerability of aspC in $\Delta$whiB7, suggest that whiB7 and aspC may participate in a feedback loop. In this model, alanine depletion results in translational stalling in the alanine-rich whiB7 uORF. This results in the induction of WhiB7 and WhiB7-mediated upregulation of aspC, thus restoring alanine levels. In the absence of whiB7, alanine depletion is not sensed, aspC is not induced, and the cell fails to initiate an adaptive response to restore alanine levels. To test this model, we first targeted aspC by CRISPRi and measured the level of aspC knockdown in both WT and $\Delta$whiB7 strains. If WhiB7 is important to upregulate aspC expression under conditions of alanine starvation, then aspC knockdown should be stronger in the $\Delta$whiB7 strain. Consistent with the feedback model, we observed significantly greater aspC knockdown in the $\Delta$whiB7 knockout strain (Fig. 5B). No differences in target knockdown were observed for the control genes mmpL3 or trpC, indicating that CRISPRi efficiency does not differ between the wild-type and $\Delta$whiB7. Also consistent with the model, we observed that ettA knockdown strains, which serve as a model for robust, constitutive whiB7 expression²⁷, show increased expression of aspC (Fig. 5C). Induction of aspC in the ettA knockdown strains could specifically be reversed by knocking down or knocking out whiB7, in accordance with prediction that whiB7 is critical for aspC induction (Fig. 5C, D). Importantly, induction of aspC in the ettA knockdown strain was not affected by concurrent knockdown of the downstream whiB7 regulon gene, tap, consistent with a direct interaction between whiB7 and aspC. Taken together, these data support the conclusion that whiB7 plays an important role outside of canonical intrinsic drug resistance by regulating cellular alanine levels through a feedback loop with aspC (Fig. 5E).

Figure 5: whiB7 coordinates a feedback loop with aspC.

(A) ChIP RT-qPCR of the aspC promoter in the indicated M. smegmatis WhiB7 N-terminal 3X-FLAG strains. Fold-enrichment of aspC promoter qPCR signal relative to the control trpC promoter (WhiB7-independent) is indicated. whiB7 was induced either by CRISPRi knockdown of ettA or aspC (18 hours +ATc), or treatment with clarithromycin for 12 hours. Statistical significance with respect to the non-targeting CRISPRi strain or DMSO control was calculated using a Student’s t-test, **P< 0.01, ***P<0.001 n.s. = non-significant.

(B) Relative mRNA levels of the indicated genes in the indicated M. smegmatis CRISPRi strains 15 hours after addition of ATc (mean ± s.e.m., n = 3 biological replicates). mRNA fold-change for the indicated gene at the bottom of each pair of bar graphs was calculated relative to sigA and normalized to the respective non-targeting CRISPRi strain for WT M. smegmatis or the $\Delta$whiB7 strain. Grey = WT, blue = $\Delta$whiB7. Statistical significance between the WT and $\Delta$whiB7 strain was calculated using a Student’s t-test, **P< 0.01, ***P<0.001.

(C) Relative mRNA levels of the indicated genes in the indicated M. tuberculosis CRISPRi strains 5 days after addition of ATc (mean ± s.e.m., n = 3 biological replicates). ettA dual CRISPRi knockdown strains are shown for whiB7 and the downstream whiB7 regulon gene, tap, which serves as a negative control. mRNA fold-change for the gene indicated on the y-axis was calculated relative to sigA and normalized to the respective non-targeting CRISPRi dual non-targeting CRISPRi strain. Statistical significance for each strain was calculated with respect to the dual non-targeting (NT) strain for each of the indicated genes using a Student’s t-test, *P<0.05, **P< 0.01, ***P<0.001, ****P<0.0001.

(D) Relative mRNA levels of the indicated genes in ettA knockdown CRISPRi strains (M. smegmatis) 18 hours after addition of ATc (mean ± s.e.m., n = 3 biological replicates). mRNA fold-change for the gene indicated on the y-axis was calculated relative to sigA and normalized to the respective non-targeting CRISPRi strain for WT M. smegmatis or the $\Delta$whiB7 strain. Grey = WT, blue = $\Delta$whiB7. Statistical significance between the WT and $\Delta$whiB7 strain was calculated using a Student’s t-test, **P< 0.01.

(E,F) Proposed model (E) and mechanism (F) for the whiB7-aspC feedback loop.

DISCUSSION

whiB7-mediated drug resistance complicates treatment of mycobacterial diseases.^15,40–42 Here, we took a functional genomics approach to globally define the upstream signals that trigger expression of whiB7, as well as downstream physiological processes that depend on whiB7. We discovered 150 genes whose inhibition leads to constitutive whiB7 expression. Hit genes include expected processes critical for translation, including amino acid biosynthesis, tRNAs, tRNA synthetases, ribosome assembly, maturation, and function. We also identified numerous unexpected hit genes involved in diverse processes including transcription, central carbon metabolism, the electron transport chain, and uncharacterized genes. It will be important to determine how silencing of these unexpected hit genes results in whiB7 induction. Essential genes were significantly overrepresented amongst hit genes, highlighting the ability of tunable CRISPRi to investigate phenotypes of essential genes and more broadly demonstrating the utility of CRISPRi reporter screens to identify genes that functionally interact with a regulatory sequence. Downstream of whiB7, we identified a critical feedback loop between whiB7 and aspC that coordinates an adaptive response to alanine deprivation. AspC has no known role in promoting intrinsic drug resistance²⁷, thus demonstrating that whiB7 plays an important physiologic role outside of its canonical role in intrinsic drug resistance by regulating cellular alanine levels.

Our whiB7 induction screen results are consistent with the model that whiB7 senses general translation stress via the uORF.²² Our results build on this model of uORF-mediated translation sensing by showing that the specific amino acid composition of the uORF determines the sensitivity of whiB7 induction to amino acid deprivation. This more nuanced model suggests that species-specific differences in uORF composition may produce unique whiB7 induction patterns during stress conditions. This may be particularly informative for drug discovery efforts targeting amino acid biosynthetic enzymes or aminoacyl tRNA synthetases.^32,43,44 Drugs which activate whiB7 expression will likely antagonize aminoglycosides, macrolides, and potentially other classes of antibiotics.¹⁵ Thus, in order to avoid antagonistic drug combinations, new antimycobacterial compounds would ideally avoid activation of the whiB7 pathway.¹⁵ Accordingly, targeting leucine or tryptophan metabolism in M. tuberculosis will likely result in minimal whiB7 induction, whereas targeting histidine or alanine metabolism will likely result in high level whiB7 induction. The results presented here will be informative for predicting and avoiding these types of whiB7-mediated antagonistic drug interactions.

Although the whiB7 uORF displayed substantial sequence diversity across actinobacterial species, the conserved enrichment for alanine suggested that whiB7 might play a physiological role in alanine metabolism. Indeed, we found that the alanine biosynthetic enzyme aspC becomes substantially more vulnerable in the absence of whiB7, and we present evidence that whiB7 is capable of sensing alanine levels and buffering aspC knockdown via a feedback loop. Our data suggest a similar mechanism may operate for asd and the hisC operon, although future work is necessary to test this hypothesis. The model presented here shows parallels to the classic attenuation mechanisms described in E. coli amino acid biosynthetic operons. Several E. coli operons such as the tryptophan and leucine operons have a leader/uORF sequence rich in the corresponding amino acid.^35–37 Deprivation of that particular amino acid results in ribosome stalling in the leader peptide, resulting in antitermination and transcriptional readthrough into the downstream biosynthetic genes to restore amino acid levels. While the mycobacterial whiB7 uORF typically encodes a longer and more diverse peptide sequence that can broadly sense amino acid starvation, it appears to be particularly tuned to sense alanine deprivation. Why whiB7 evolved such an “Ala leader” mechanism tuned to sense and specifically regulate alanine levels, as opposed to some other amino acid, remains an interesting and open question. Given the centrality of alanine in bacterial metabolism, we cannot rule out functions for this feedback loop beyond the proteinogenic functions for alanine. For example, flux through AspC could enable alanine to serve as an indirect readout of pyruvate levels and the various metabolic pathways in which pyruvate participates. Further, L-alanine is a direct precursor to D-alanine, a core component of cell wall peptidoglycan. Future studies may investigate the broader metabolic and physiological consequences of whiB7-mediated aspC regulation.

Lastly, the results presented here may help further our understanding of the complex interactions between mycobacterial metabolism, the host immune response and antibiotic activity. In the context of TB, immune activation has been shown to induce drug tolerance.^45,46 Although the molecular basis for this phenomenon is likely to involve many distinct host and bacterial pathways, whiB7 induction may be an important contributor to host-induced drug tolerance.^19,26 Previous work showed that M. tuberculosis whiB7 is upregulated upon infection of both resting and IFN-γ-activated macrophages^28,47, which may promote rifampicin and aminoglycoside tolerance mediated by the Tap efflux pump.¹⁹ Our results may help define the host-imposed stressors and affected bacterial pathways that lead to intramacrophage whiB7 induction. Future studies could use the M. tuberculosis PwhiB7:mScarlet reporter to dissect whiB7 expression patterns across a diverse set of host immune cells including different subsets of macrophages, neutrophils, and dendritic cells, as well as scenarios involving drug treatment.^48,49 We hope that these studies will shed light on how distinct intracellular environments influence mycobacterial metabolism, thus shaping drug resistance and tolerance patterns.

Limitations of this study:

The work presented here was largely performed in the model bacterium, Mycobacterium smegmatis, with key findings being confirmed in virulent M. tuberculosis. As evidenced by our studies on the amino acid composition of the uORF, there may be species-specific differences in whiB7 activation signals that are not fully captured in our work. Similarly, previous work has demonstrated species-specific differences in the whiB7 regulon in M. abscessus¹⁷, presenting the possibility of additional physiological functions for whiB7 beyond those that we demonstrate in M. smegmatis and M. tuberculosis. CRISPRi vulnerability screens performed in a whiB7 knockout strain of M. abscessus, for example, may reveal unique physiological functions for whiB7 in that species. With respect to the CRISPRi screens, it is important to note that not all essential genes whose inhibition activates whiB7 expression may be recovered in the PwhiB7:zeoR screen. Essential hit genes require sgRNAs strong enough to reduce gene product activity sufficiently to induce whiB7, but not reduce fitness enough to prevent colony formation. Such false negatives, likely enriched for small genes for which fewer sgRNAs are available, may be recoverable using the PwhiB7:mScarlet reporter coupled with FACs to obviate the need for colony outgrowth. Finally, these studies were conducted using standard laboratory culture medium. Pathogenic and environmental mycobacteria encounter a number of stresses that are not accounted for under these culture conditions.⁴⁹ Performing analogous studies in infection models or conditions experienced by environmental species may uncover entirely new aspects of whiB7 biology.

SIGNIFICANCE

Mycobacterial pathogens are a significant cause of morbidity and mortality worldwide. Among these species is Mycobacterium tuberculosis, the world’s leading infectious killer. Mycobacteria have a high level of intrinsic antibiotic resistance, making infections challenging to treat. The conserved whiB7 stress response is a key contributor to mycobacterial intrinsic drug resistance. Although we have a comprehensive structural and biochemical understanding of WhiB7, the complex set of signals that activate whiB7 expression remain less clear. Here, we employed a genome-wide CRISPRi epistasis screen and identified a diverse set of 150 mycobacterial genes whose inhibition results in constitutive whiB7 expression. Many of these genes encode amino acid biosynthetic enzymes, tRNAs, and tRNA synthetases, consistent with the proposed mechanism for whiB7 activation induction by translational stalling in the uORF. We show that the ability of the whiB7 5’ regulatory region to sense amino acid starvation is determined by the coding sequence of the uORF. The uORF shows considerable sequence variation among different mycobacterial species, but it is universally and specifically enriched for alanine. Providing a potential rationalization for this enrichment, we find that while deprivation of many amino acids can activate whiB7 expression, whiB7 specifically coordinates an adaptive response to alanine starvation by engaging in a feedback loop with the alanine biosynthetic enzyme, aspC. Our results provide a holistic understanding of the biological pathways that influence whiB7 induction and reveal an extended role for the whiB7 pathway in mycobacterial physiology, beyond its canonical function in antibiotic resistance. These results have important implications for the design of combination drug treatments to avoid whiB7 induction, as well as help explain the conservation of this stress response across a wide range of pathogenic and environmental mycobacteria. Further, this work provides a functional genomic framework for identifying regulators of other inducible transcriptional pathways in mycobacteria.

STAR METHODS

Resource Availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Jeremy Rock (rock@rockefeller.edu).

Materials Availability

Key plasmids and the M. smegmatis CRISPRi library have previously been deposited on Addgene: Plasmids: plRL19 (#163634); plRL58 (#166886); and plRL61 (#163633). CRISPRi library: RLC11 (#163955). Specific plasmids and mycobacterial strains can be provided upon request.

Data and code availability

• Raw sequencing data are deposited to the NCBI Short Read Archive under project numbers PRJNA970266 and PRJNA970343.

• All source code and CRISPRi library FASTA files are publicly available online (Github: https://github.com/rock-lab/whiB7KO_screen_2023/ and https://doi.org/10.5281/zenodo.10384792).

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Bacterial strains:

M. smegmatis strains are modified derivatives of the mc² 155 strain. M. tuberculosis strains are modified derivatives of the H37Rv strain. E. coli strains used for DNA propagation and molecular cloning are all modified derivatives of the NEB 5-alpha strain.

Mycobacterial cultures:

All mycobacteria were grown at 37°C in Difco Middlebrook 7H9 broth or on 7H10 agar supplemented with 0.2% glycerol (7H9) or 0.5% glycerol (7H10), 0.05% Tween-80 and either 1X albumin-dextrose-catalase (ADC) for M. smegmatis or 1X oleic acid-albumin-dextrose-catalase (OADC) for M. tuberculosis and M. abscessus. For CRISPRi experiments, anhydrotetracycline (ATc) was used at 100 ng/ml. For plasmid selection kanamycin was used at 20 μg/mL and zeocin was used at 20 μg/mL. For the whiB7 induction screen in M. smegmatis zeocin was used at 10 μg/mL. M. tuberculosis cultures were grown standing in tissue culture flasks (unless otherwise indicated) with 5% CO₂, whereas M. smegmatis strains were grown in shaking conditions in sealed culture vessels. Amino acids were supplemented to the media at the indicated concentration.

Relative growth of individual CRISPRi strains was determined by spotting assay. Ten-fold serial dilutions (starting at 30,000 cells/spot) were plated on 7H10 with or without 100 ng/mL ATc. Plates were incubated at 37°C and imaged after 2 days.

Method details

Plasmid construction and cloning

CRISPRi sgRNAs were cloned into either plRL61 (M. smegmatis) or plRL58 (M. tuberculosis). Custom oligonucleotides (see supplementary material for exact top and bottom oligo sequences) with corresponding sticky ends were annealed and ligated (T4 DNA ligase) into BsmBI-digested CRISPRi plasmids. Ligation products were transformed into NEB 5-alpha chemically competent cells and selected on LB + kanamycin (50 μg/mL). Individual colonies were picked and the correct sgRNA sequence was confirmed by sanger sequencing before subsequent transformation into M. smegmatis and M. tuberculosis.

Reporter constructs and whiB7 variant constructs were cloned into plRL91 (Nat selection, M. smegmatis) or plRL125 (Zeocin selection, M. tuberculosis). plRL91 was digested with DraIII and plRL125 was digested with HpaI. After gel purification of digested vector, purified PCR fragments containing overlapping sequence of at least 20 nt were incorporated via Gibson assembly (NEB HiFi assembly) using 0.02 pmol of digested vector and 0.04 pmol of each PCR fragment. After selection on LB + ampicillin (plRL91) or LB + zeocin (plRL125) colonies were subjected to sanger sequencing and/or whole plasmid sequencing (plasmidsaurus). Sequence-verified clones were transformed into mycobacteria as described below.

Mycobacterial plasmid and library transformations

For standard transformations into M. smegmatis and M. tuberculosis, cultures of at least 25 mL were grown to late logarithmic phase (OD₆₀₀ of 0.6–1.0) and cells were pelleted at 4,000 x g (room temperature) for 10 min. M. tuberculosis cultures were treated with 200 mM glycine 16–24 hours prior to transformation. After pelleting, cultures were washed 3X in filter-sterilized 10% glycerol (equivalent to original culture volume). Following the third wash, cells were resuspended in 10% glycerol (5–10% of the original culture volume). For each transformation, 100 μL of competent cells were mixed with >100 ng of integrating plasmid (CRISPRi, whiB7 variant, etc.) and >100 ng of integrase-expressing “suicide” plasmid (plRL19, plRL40, plRL62, etc.). Cells were electroporated at 2.5 kV with a 2 mm cuvette and a resistance of 700 Ω. Electroporated cells were immediately recovered in 1 mL of complete 7H9 and cultured at 37°C for one doubling period (3 hours for M. smegmatis and 24 hours for M. tuberculosis) after which cells were pelleted at 4,000 x g for 5 min, resuspended in 100 μL of 7H9 and plated on selective 7H10 plates. Colonies were picked after three days for M. smegmatis and 21 days for M. tuberculosis.

For the transformation of the RLC11 M. smegmatis genome-wide CRISPRi library into the whiB7 zeocin reporter strain (figure 2) or the $\Delta$whiB7 M. smegmatis strain, 50 transformations were performed. The transformation protocol was identical to the one above with the following modifications. 400 mL of late logarithmic phase culture was concentrated down to 10 mL of final electrocompetent cells. The CRISPRi plasmid used to generate the RLC11 library contains the integrase so no plRL19 integrase plasmid was expressed in trans. Finally, all steps leading up to the electroporation were performed at 4°C. See sections below for information on how the library was harvested and prepared for screening following the initial transformation.

PwhiB7:mScarlet reporter assays

mScarlet fluorescence was measured using a Tecan spark plate reader with an excitation of 563 nm and and emission of 600 nm. For M. smegmatis assays, plates were cultured in the plate reader for 48 hours with shaking at 500 rpm. Fluorescence and optical density were measured every 30 minutes. For M. tuberculosis assays, plates were incubated under standing conditions and were read for fluorescence and optical density at 4 to 7 days post-plating. Normalized fluorescence was calculated by dividing the background-adjusted fluorescence value by the background-adjusted optical density value.

Antibacterial activity measurements:

All compounds (see key resources table) were dissolved in DMSO (VWR V0231) and dispensed using an HP D300e digital dispenser in a 384-well plate format using a 2-fold dilution series. DMSO did not exceed 1% of the final culture volume and was maintained at the same concentration across all samples. Cultures were growth synchronized to late logarithmic phase (~ OD₆₀₀ = 0.8) and then back-diluted to a starting OD₆₀₀ of 0.01. 50 μl cell suspension was plated in triplicate in wells containing the test compound. Plates were incubated standing at 37 °C with 5% CO2. OD₅₈₀ was evaluated using a Tecan Spark plate reader at 10–11 days post-plating and percent growth was calculated relative to the DMSO vehicle control for each strain. IC₅₀ measurements were calculated using a nonlinear fit in GraphPad Prism. For all MIC curves, data represent the mean ± s.e.m. for triplicates. Data are representative of two independent experiments.

Key Resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-FLAG M2 antibody	Sigma Aldrich	Cat# F1804
Bacterial and virus strains
E. coli NEB 5-alpha	New England Biolabs	C2987H
M. tuberculosis H37Rv	Christopher Sassetti (UMass Worcester)	NA
M. smegmatis mc2 155	Sarah Fortune (Harvard)	NA
$\Delta$whiB7 M. smegmatis mc2 155	Carl Nathan (Weill Cornell)	Schrader et al.26
Chemicals, peptides, and recombinant proteins
Amikacin	Sigma Aldrich	Cat# A0365900
Anhydrotetracycline	Fisher Scientific	Cat# AC233135000
Azithromycin dihydrate	Sigma Aldrich	Cat# PZ0007
Bedaquiline	AdooQ Biosciences	Cat# A12327–5
Bortezomib	Sigma Aldrich	Cat# 5043140001
Capreomycin Sulfate	Sigma Aldrich	Cat# PHR1716
Chloramphenicol		Cat# C-105
Clarithromycin	Sigma Aldrich	Cat# C9742
Clindamycin hydrochloride	GoldBio	Cat# C-175–10
Ecumicin	Sanghyun Cho (Univ. Illinois at Chicago College of Pharmacy)	Gao et al29
Ethambutol dihydrochloride	Sigma Aldrich	Cat# E4630
Fusidic acid sodium salt	Sigma Aldrich	Cat# F0881
Isoniazid	Sigma Aldrich	Cat# I3377
Kanamycin	GoldBio	Cat# K-120–50
Levofloxacin	Sigma Aldrich	Cat# 28266
Lincomycin hydrochloride	Sigma Aldrich	Cat# 62143
Linezolid	Sigma Aldrich	Cat# SML1290
PBTZ169	Cayman Chemical	Cat# 22202
PA-824 (Pretomanid)	Sigma Aldrich	Cat# SML1290
Rifampicin	TCI	Cat# R0079
Spectinomycin dihydrochloride	GoldBio	Cat# S-140–5
Streptomycin sulfate	GoldBio	Cat# S-150–100
Tigecycline hydrate	Sigma Aldrich	Cat# PZ0021
Vancomycin hydrochloride	Sigma Aldrich	Cat# V2002
Zeocin	Alfa Aesar	Cat# J67140–8EQ
6-FABA	Sigma Aldrich	Cat# 443026
Bacto™ casamino acids	Thermo Fisher	Cat# 223050
L-alanine	Sigma Aldrich	Cat# 1250–100GM
L-aspartic acid	Sigma Aldrich	Cat# A7219–100G
L-glutamic acid	Sigma Aldrich	Cat# G1501–100G
L-lysine	Sigma Aldrich	Cat# L5626–100G
L-methionine	Sigma Aldrich	Cat# M9625–100G
L-threonine	Sigma Aldrich	Cat# T8625–100G
L-tryptophan	Sigma Aldrich	Cat# T0254–100G
Critical commercial assays
NEB HiFi 2X Master Mix (Gibson assembly)	New England Biolabs	Cat# E2621L
NEB Q5 2X Master Mix (molecular cloning)	New England Biolabs	Cat# M0492L
NEBUltra II Q5 Master Mix (NGS PCR)	New England Biolabs	Cat# M0544L
NextSeq High Output Kit v2.5 (75 Cycles)	Illumina	Cat# 20024906
NovaSeq 6000 S2 kit	Illumina	Cat# 20028316
SuperScript IV First-Strand Synthesis System	Thermo Fisher	Cat# 18–091-050
SYBR™ Green PCR master mix	Thermo Fisher	Cat# 4364344
T4 DNA ligase	NEB	Cat# M0202T
BsmBI-HF restriction enzyme	NEB	Cat# R0739L
Deposited data
Raw sequencing data from the whiB7 induction screen (Fig. 2) and the $\Delta$whiB7 vulnerability screen (Fig. 4).	This study	SRA: PRJNA970266 and PRJNA970343
Oligonucleotides
sgRNAs and RT-qPCR primers (Table S2)	This study	NA
Recombinant DNA
plRL91 (plAW145) Empty complementation vector used in M. smegmatis for figure(s) 1B,E	This study	https://benchling.com/s/seq-l2e2KoT2t1rryGetcqEk?m=slm-buY07uBNcN8k2Ftj0UHx
plRL125 backbone used for cloning of M. tuberculosis mScarlet reporter construct (not used directly in experiments)	This study	https://benchling.com/s/seq-mslTEj8tckVmDo7agJvm?m=slm-3looaqQ5LvmWaqUuDvJB
plRL284 (plNP441) zeoR driven by EM7 promoter used in M. smegmatis for figure(s) 1B	This study	https://benchling.com/s/seq-keovdkto7gS8Pvlffwc1?m=slm-QxR93n3NeuO6SocRtVZh
plRL285 (plNP442) zeoR driven by M. smegmatis whiB7 + 500 bp region used in M. smegmatis for figure(s) 1A–C, 2A–C	This study	https://benchling.com/s/seq-BuEHqVYrpKdaPOgbM1LG?m=slm-YGryGwZumq25G32nLEqJ
plRL286 (plNP445) mScarlet driven by p300 promoter used in M. smegmatis for figure(s) 1E	This study	https://benchling.com/s/seq-mdAXDB8tz9PX5jircmcG?m=slm-z645sCvy9h2xhj3CayMJ
plRL287 (plNP446) mScarlet driven by M. smegmatis whiB7 + 500 bp region used in M. smegmatis for figure(s) 1A,D-F, S1A, S2C,D,F, 3A,B	This study	https://benchling.com/s/seq-xVp5VXajLMpA6ZUtM9CU?m=slm-gXTXKm1xNzchZnpYKKIW
plRL277 (plNP466) mScarlet driven by M. tuberculosis whiB7 + 500 bp region used in M. tuberculosis for figure(s) 1F, S1B, S2E, 3A,C	This study	https://benchling.com/s/seq-kaaIOaUpb4hbZUCmR4aC?m=slm-AGfDgX4AjN4Z3uamM1ru
plRL288 (plNP623) whiB7 driven by WT regulatory region used in M. smegmatis for figure(s) 3E,F, S4A–C	This study	https://benchling.com/s/seq-lgjYeAJP4iqefoWNX96K?m=slm-rvpRpiZHaybrmqgOoJ8C
plRL289 (plNP624) whiB7 driven by regulatory region with His-less uORF used in M. smegmatis for figure(s) 3E,F, S4A–C	This study	https://benchling.com/s/seq-KrKP84U6kGmwRF88Q354?m=slm-6Kt3A5yWd5SuOVxPCriT
plRL290 (plNP625) whiB7 driven by regulatory region with Trp-less uORF used in M. smegmatis for figure(s) 3E,F, S4A–C	This study	https://benchling.com/s/seq-JBPYxdqzMde5aW3c3WKq?m=slm-5yTQ43IYoqMZwsHo1IDr
plRL291 (plNP672) M. smegmatis whiB7 driven by Mtb regulatory region (+500 bp) used in M. smegmatis for figure(s) S4C,D	This study	https://benchling.com/s/seq-Td5eHorcOa20WK251v8H?m=slm-t2pWnjdRICdO4W8YA2Js
plRL292 (plNP687) M. smegmatis whiB7 driven by Mtb regulatory region (+500 bp) with Ala13Pro mutation in uORF used in M. smegmatis for figure(s) S4C,D	This study	https://benchling.com/s/seq-yoQpkJEjxgkuXqiQuUxK?m=slm-77arP3sIM71P9o9afYsg
plRL293 (plNP688) M. smegmatis whiB7 driven by Mtb regulatory region (+500 bp) with His22fs delG mutation in uORF used in M. smegmatis for figure(s) S4C,D	This study	https://benchling.com/s/seq-gJKRIhRdHgzMIiD6GPeL?m=slm-lofzbUxFg7018X3ZCi4U
plRL294 (plNP689) M. smegmatis whiB7 driven by Mtb regulatory region (+500 bp) with His22ProPro mutation in uORF used in M. smegmatis for figure(s) S4C,D	This study	https://benchling.com/s/seq-B8dUoil9DphongRVKHeX?m=slm-vSbXazg2ubnQcyiZ1vSC
plRL295 (plNP690) M. smegmatis whiB7 driven by Mtb regulatory region (+500 bp) with Val1Gly start lost uORF mutation in uORF used in M. smegmatis for figure(s) S4C	This study	https://benchling.com/s/seq-lgrqU4siiz9Vu3gnapEi?m=slm-L9pSJKkjCOFmWvnoGPoJ
plRL296 (plNP691) M. smegmatis whiB7 driven by Mtb regulatory region (+500 bp) with His22fs (insCTAT) mutation in uORF used in M. smegmatis for figure(s) S4C,D	This study	https://benchling.com/s/seq-zOy0ar8rwDLYOYhx33fo?m=slm-Wy14qhxXIg2IGvTmDtw3
plRL62 Tweety integrase suicide vector used in M. smegmatis for figure(s) 1A–F, S1A, 2A, S2C–F, 3A,B,E,F, S4A–C	This study	https://benchling.com/s/seq-LtN3e9WbixCmPd0VOCAa
plRL40 Giles integrase suicide vector used in M. tuberculosis for figure(s) 1F, S1B, S2E, 3A,C	This study	https://benchling.com/s/seq-2PVXvHq0CfNPVacwz1Ku
plRL19 (plJR1949) L5 integrase suicide vector used in M. smegmatais and M. tuberculosis for figure(s) 1C,E, S2A–F, 3A,F, S4B, 4C,D, S5A–C, 5A–D	Bosch et al. 30	Addgene Cat# 163634
plRL58 CRISPRi plasmid M. tuberculosis KD experiments (co-transformed with plRL19)	Bosch et al. 30	Addgene Cat# 166886
plRL61 CRISPRi plasmid M. smegmatis KD experiments (co-transformed with plRL19)	Bosch et al. 30	Addgene Cat# 163633
Software and algorithms
ChemiDoc imaging system	BioRad	Cat# 12003153
Prism 9	GraphPad	https://www.graphpad.com/features
Python	Python software foundation	https://www.python.org/downloads/
Rstan (version 2.19.3)	Stan Development Team, 2020	https://mc-stan.org/users/interfaces/rstan.html
SciPy (version 1.2.2)	Virtanen et al. 51	https://scipy.org/install/
Stan (version 2.19.3)	Stan Development Team, 2021	https://mc-stan.org/users/interfaces/
statsmodels (version 0.10.1)	Seabold and Perktold52	https://www.statsmodels.org/stable/install.html
Subread aligner (version 1.6.0)	Liao et al. 53	https://sourceforge.net/projects/subread/files/subread-2.0.6//
Spark® Plate reader	Tecan	https://lifesciences.tecan.com/
QuantStudio™ 5 qPCR system	Thermo Fisher	Cat# A28140
Vulnerability analysis pipeline (original code)	This paper and Bosch et al. 30	Github: https://github.com/rock-lab/whiB7KO_screen_2023/ and https://doi.org/10.5281/zenodo.10384792

whiB7 induction CRISPRi screen

The M. smegmatis strain harboring the PwhiB7:zeoR construct was transformed with the genome-wide CRISPRi library initially described in Bosch et al.²⁹ (Addgene 163955) with greater than 400-fold coverage. Transformants were plated on complete 7H10 agar with kanamycin at 20 μg/mL. After 3 days of growth, biomass was collected in 10% glycerol and single cell suspensions were created using the two dissociation cycles on a gentleMACS Octo Dissociator (Miltenyi Biotec #130095937) using the RNA_01 program and 12 gentleMACS M tubes (Miltenyi Biotec #130093236). After passaging in complete 7H9 and filtering through a pluriselect filter (10 μm) stocks glycerol stocks were frozen. For the screen one stock of the CRISPRi library was thawed into 25 mL of complete 7H9. After growth to late log-phase, 3 parallel cultures were started at an OD₆₀₀ of 0.025 in the presence of ATc. After 16 hours of pre-depletion, each culture was diluted to 1×10⁷ CFU was plated onto plates containing ATc with or without 10 μg/mL zeocin. In parallel, a non-ATc treated library culture was plated at the same density on 10 μg/mL zeocin without ATc. In parallel, serial dilutions of pre-depleted and non pre-depleted cultures were plated on smaller agar plates for CFU enumeration. All plates were counted and harvested after 3 days outgrowth. Biomass was from each plate was scraped into TE buffer and subjected to GentleMACS dissociation (M tube, 2X RNA cycle). Approximately 50 OD₆₀₀ units were then subjected to gDNA extraction via the TE-lysozyme method described below.

$\Delta$whiB7 vulnerability CRISPRi screen

The $\Delta$whiB7 M. smegmatis strain described in Schrader et al.²⁶, was transformed with the genome-wide CRISPRi library (Addgene 163955) with greater than 200-fold coverage. Transformants were plated on complete 7H10 agar with kanamycin at 12.5 μg/mL, due to potential increase in kanamycin sensitivity in the absence of whiB7. After 3 days of growth, biomass was collected and single cell suspensions were generated using the same process as the induction screen (see above). For the passaging time course, one stock of the CRISPRi library was thawed and expanded into 25 mL of complete 7H9 + kanamycin 12.5 μg/mL. Parallel, triplicate cultures were set up with or without ATc at 100 ng/mL. At each indicated timepoint (generations), cultures were harvested for genomic DNA isolation. Remaining culture was used for continued passaging. Each sample was then subjected to gDNA extraction via the TE-lysozyme method described below.

Genomic DNA extraction

Genomic DNA was isolated from bacterial pellets using the CTAB-lysozyme method as previously described.²⁹ After drug treatment 10–20 OD₆₀₀ units of the cultures were pelleted by centrifugation (10 minutes at 4,000xg) and were resuspended in 1ml PBS + 0.05% Tween-80. Cell suspensions were centrifuged again for 5min at 4,000×g, the supernatant was removed, and pellets were frozen until processing. For the DMSO-treated culture and the cultures treated with supra-MIC drug concentrations, 500 μL of the remaining culture was spread evenly on a 15 cm petri dish containing complete 7H10 + 0.4% activated charcoal. After 17–21 days of outgrowth, the biomass was scraped off of the plate using PBS + 0.05% Tween-80 and gDNA was processed the identically to the pellets obtained directly from liquid culture. Upon thawing, pellets were resuspended in 800 μl TE buffer (10mM Tris pH 8.0, 1mM EDTA) + 15mg/mL lysozyme (Alfa Aesar J60701–06) and incubated at 37 °C for 16h. Next, 70μl 10% SDS (Promega V6551) and 5 μl proteinase K (20 mg/mL, Thermo Fisher 25530049) were added and samples were incubated at 65°C for 30min. Subsequently, 100 μl 5M NaCl and 80 μl 10% CTAB (Sigma Aldrich H5882) were added, and samples were incubated for an additional 30min at 65 °C. Finally, 750 μl ice-cold chloroform was added and samples were mixed. After centrifugation at 16,100×g and extraction of the aqueous phase, samples were removed from the biosafety level 3 facility. Samples were then treated with 25 μg RNase A (Bio Basic RB0474) for 30min at 37 °C, followed by extraction with phenol:chloroform:isoamyl alcohol (pH 8.0, 25:24:1, Thermo Fisher BP1752I-400), then chloroform. Genomic DNA was precipitated from the final aqueous layer (600 μl) with the addition of 10 μl 3M sodium acetate and 360 μl isopropanol. DNA pellets were spun at 21,300×g for 30min at 4 °C and washed 2× with 750 μl 80% ethanol. Pellets were dried and resuspended with elution buffer (Qiagen 19086) before spectrophotometric quantification. The concentration of isolated genomic DNA was quantified using the DeNovix dsDNA high sensitivity assay (KIT-DSDNA-HIGH-2; DS-11 series spectrophotometer/fluorometer).

Library preparation for Illumina sequencing of CRISPRi libraries

Next generation sequencing was performed as follows. The unique barcoded region was amplified from 500ng genomic DNA with 16 cycles of PCR using NEBNext Ultra II Q5 master mix (NEB M0544L) as described in.²⁹ Each PCR reaction contained a unique indexed forward primer (0.5μM final concentration) and a unique indexed reverse primer (0.5μM). Forward primers contain a P5 flow cell attachment sequence, a standard Read1 Illumina sequencing primer binding site and custom stagger sequences to guarantee base diversity during Illumina sequencing. Reverse primers contain a P7 flow cell attachment sequence, a standard Read2 Illumina sequencing primer binding site and unique barcodes to allow for sample pooling during deep sequencing. Following PCR amplification, each ~230bp amplicon was purified using AMPure XP beads (Beckman–Coulter A63882) using one-sided selection (1.2×). Bead-purified amplicons were further purified on a Pippin HT 2% agarose gel cassette (target range 180–250bp; Sage Science HTC2010) to remove carry-over primer and genomic DNA. Eluted amplicons were quantified with a Qubit 2.0 fluorometer (Invitrogen), and amplicon size and purity were quality controlled by visualization on an Agilent 2100 bioanalyzer (high sensitivity chip; Agilent Technologies 5067–4626). Next, individual PCR amplicons were multiplexed into 10nM pools and sequenced on an Illumina sequencer according to the manufacturer’s instructions. To increase sequencing diversity, a PhiX spike-in of 2.5–5% was added to the pools (PhiX sequencing control v3; Illumina FC-110– 3001). Samples were run on the Illumina NovaSeq 6000 platform (single-read 1 ×85 cycles and 6 × i7 index cycles).

RNA extraction and RT-qPCR

Total RNA extraction was performed as previously described.²⁹ Briefly, ~2 OD₆₀₀ units of bacteria were added to an equivalent volume of GTC buffer (5 M guanidinium thiocyanate, 0.5% sodium N-lauroylsarcosine, 25 mM trisodium citrate dihydrate and 0.1 M 2-mercaptoethanol), pelleted by centrifugation, resuspended in 1 ml TRIzol (Thermo Fisher 15596026) and lysed by zirconium bead beating (MP Biomedicals 116911050). Chloroform (0.2 ml) was added to each sample and phases were separated by centrifugation. The aqueous phase was then purified by Direct-zol RNA miniprep (Zymo Research R2052). Residual genomic DNA was removed by TURBO DNase treatment (Invitrogen Ambion AM2238). After RNA cleanup and concentration (Zymo Research R1017), 3 μg RNA per sample was reverse transcribed into complementary DNA (cDNA) with random hexamers (Thermo Fisher 18–091-050) following the manufacturer’s instructions. RNA was removed by alkaline hydrolysis and cDNA was purified with PCR cleanup columns (Qiagen 28115). Next, knockdown or induction of target genes was quantified by SYBR green dye-based quantitative real-time PCR (Applied Biosystems 4309155) on a Quantstudio System 5 (Thermo Fisher A28140) using gene-specific qPCR primers (5 μM), normalized to sigA (rv2703) and quantified by the $\text{ΔΔCt}$ algorithm. All gene-specific qPCR primers were designed using the PrimerQuest tool from IDT (https://www.idtdna.com/PrimerQuest/Home/Index) and then validated for efficiency and linear range of amplification using standard qPCR approaches. Specificity was confirmed for each validated qPCR primer pair through melting curve analysis. All qPCR primers used in this study can be found in the Supplemental Plasmids and Primers Table.

Chromatin immunoprecipitation (ChIP) RT-qPCR

Chromatin immunoprecipitation of an N-terminally 3X-FLAG-tagged WhiB7 was performed according to the protocol described by Jaini et al.⁵⁴ Briefly, merodiploid strains expressing a second copy of WhiB7 (driven by the native promoter), either tagged or untagged, as well as the CRISPRi sgRNA of interest were cultured + ATc for 18 hours and grown to late log phase (OD₆₀₀ ~0.7) to allow for induction of whiB7. Alternatively, cultures (no CRISPRi) were treated with either DMSO or clarithromycin (to induce whiB7) for 12 hours. Protein-DNA cross-linking was carried out by adding formaldehyde to the cultures to a final volume of 1% and shaking at room temperature for 30 minutes. Cross-linking was quenched by the addition of glycine (250 mM final). Cells were pelleted and mechanically lysed by bead-beating (max speed) 3-times for 30 seconds (1 minute pause between cycles) in the presence of protease inhibitor. Clarified cell lysates (500 μL) were subjected to sonication using the Covaris S220 (Power = 140, Duty = 5, Cycles/Burst = 200) for 18 minutes with microTUBE 500 AFA tubes (Covaris 520185) and each sample was incubated with 5 μg anti-FLAG M2 antibody (Sigma Aldrich F1804) at 4°C overnight. Immune complexes were purified using 100 μL of protein G agarose slurry (Thermo Scientific 20398) for 2 hours. After washing, DNA protein complexes were eluted at 65°C with Tris-EDTA buffer + 1% SDS, 2 times and the separate eluates were pooled. Reverse cross-linking was performed with 1 mg/mL (final) proteinase K at 37°C for 2 hours followed by an overnight incubation at 65°C. DNA was purified using a Qiagen PCR cleanup kit (Qiagen 28104) and subsequently subject to RT-qPCR as described above. Fold-change was calculated using the $\text{ΔΔCT}$ method as described above. The $\text{ΔCT}$ value for each samples was based off of the CT value for the trpC promoter, which has not been shown to be regulated by whiB7, nor is it in close proximity to a whiB7-regulated gene. The $\text{ΔΔCT}$ was then calculated by comparing each experimental sample (i.e. aspC knockdown) to the relevant control strain (i.e. non-targeting CRISPRi). Samples derived from a merodiploid M. smegmatis strain expressing an untagged whiB7 did not show any appreciable signal above background.

Quantification and statistical analysis

Analysis of RT-qPCR data

Sample sizes are reported in each figure legend. Expression was normalized to sigA (rv2703) and fold-change was quantified by the $\text{ΔΔCt}$ algorithm. All statistical analyses were executed using GraphPad Prism 9 software. Means between groups were compared using two-tailed unpaired Student’s student’s t-test. Data are displayed as mean ± SEM. Statistically significance of data are indicated: ns, not significant, * P < 0.05, **P < 0.01, ***P < 0.001 and ****P<0.0001.

Analysis of whiB7 induction screen

Sequencing counts were obtained in the manner described by.²⁹ Counts were normalized for sequencing depth and an sgRNA limit of detection (LOD) cut-off was set at 20 counts in the ATc only condition. Only sgRNAs that made the LOD cut-off (i.e. counts > 20) were analyzed further. sgRNA counts were analyzed using MAGeCK ⁵⁵ (version 0.5.9.2) in python (version 2.7.16) comparing each + zeocin + ATc condition to the matched control (+ATc only) sample. Gene-level log2 fold change (L2FC) was calculated using the ‘alphamedian’ approach specified with the ‘gene-lfc-method’ parameter, which estimates the gene-level L2FC as the median of sgRNAs that are ranked above the default cut off in the Robust Rank Aggregation used by MAGeCK. Negative control sgRNAs were used to calculate the null distribution and to normalize counts using the ‘-- control-sgrna’ and ‘– normalization control’ parameters, respectively. MAGeCK gene summary output results can be found in Source Data 1. A gene was determined to be an enriched hit if it had a false discovery rate (FDR) < 0.01 and a log2 fold change (L2FC) > 2 in positive selection.

Differential vulnerability analysis of $\text{Δ}$whiB7 CRISPRi screen

Gene vulnerability in the WT and $\text{Δ}$whiB7 M. smegmatis strains was determined using the vulnerability model previously described²⁹ with some modifications. Briefly, under the updated model, read counts for a given sgRNA in the minus ATc conditions were modeled using a Negative Binomial distribution with a mean proportional to the counts in the plus ATc condition, plus a factor representing the log2 fold-change:

$$y_i^{-ATc}\sim NegBinom\left({\eta }_i,\varphi \right)$$

$${\eta }_i=\log \left(y_i^{+ATc}+{\lambda }_i\right)+TwoLine\left(x_i,{\alpha }_l,{\beta }_l,\gamma ,{\beta }_e\right)$$

where ${\lambda }_i$ is an sgRNA-level correction factor estimated by the model, $x_i$ represents the generations analyzed for the i-th guide, and the TwoLine function represents the piecewise linear model previously described, which models sgRNA behavior over time²⁹. The logistic function describing gene-level vulnerabilities was simplified by setting the starting point of the curve (K) equal to 0, representing the fact that weakest possible sgRNAs are expected to impose approximately no effect on bacterial fitness i.e.:

$$Logistic(s)=\frac{{\beta }_{max}}{(1+\exp (-H\cdot (S-M)))}$$

The Bayesian vulnerability models were run for each condition independently, and samples for all the parameters were obtained using Stan (4 chains with 1,000 warmup iterations, and 4000 samples total). Differential vulnerabilities were estimated by taking the difference of pairwise (guide-level) vulnerability estimates, resulting in posterior samples of the differential vulnerability expected for a gene, between the two strains. We estimate that this differential vulnerability model is sufficiently sensitive to detect a ~18% reduction in counts for a given gene between the two genetic backgrounds in which the screen is performed (Methods S1).

Supplementary Material

2

Methods S1 Vulnerability sensitivity model related to STAR methods

3

Supplemental Data 1 whiB7 zeo screen results related to Figure 2

4

Supplemental Data 2 whiB7 uORF variants related to Figure S4

5

Supplemental Data 3 whiB7 KO vulnerability related to Figure 4

HIGHLIGHTS.

• A CRISPRi reporter screen reveals cellular perturbations that induce whiB7 expression

• whiB7 senses amino acid starvation in a uORF sequence-dependent manner

• A genome-wide CRISPRi vulnerability screen identifies metabolic functions for whiB7

• whiB7 coordinates adaptation to alanine starvation through a feedback loop with aspC