Aldehyde accumulation in Mycobacterium tuberculosis with defective proteasomal degradation results in copper sensitivity

ABSTRACT Mycobacterium tuberculosis is a major human pathogen and the causative agent of tuberculosis disease. M. tuberculosis is able to persist in the face of host-derived antimicrobial molecules nitric oxide (NO) and copper (Cu). However, M. tuberculosis with defective proteasome activity is highly sensitive to NO and Cu, making the proteasome an attractive target for drug development. Previous work linked NO susceptibility with the accumulation of para-hydroxybenzaldehyde (pHBA) in M. tuberculosis mutants with defective proteasomal degradation. In this study, we found that pHBA accumulation was also responsible for Cu sensitivity in these strains. We showed that exogenous addition of pHBA to wild-type M. tuberculosis cultures sensitized bacteria to Cu to a degree similar to that of a proteasomal degradation mutant. We determined that pHBA reduced the production and function of critical Cu resistance proteins of the regulated in copper repressor (RicR) regulon. Furthermore, we extended these Cu-sensitizing effects to an aldehyde that M. tuberculosis may face within the macrophage. Collectively, this study is the first to mechanistically propose how aldehydes can render M. tuberculosis susceptible to an existing host defense and could support a broader role for aldehydes in controlling M. tuberculosis infections. IMPORTANCE M. tuberculosis is a leading cause of death by a single infectious agent, causing 1.5 million deaths annually. An effective vaccine for M. tuberculosis infections is currently lacking, and prior infection does not typically provide robust immunity to subsequent infections. Nonetheless, immunocompetent humans can control M. tuberculosis infections for decades. For these reasons, a clear understanding of how mammalian immunity inhibits mycobacterial growth is warranted. In this study, we show aldehydes can increase M. tuberculosis susceptibility to copper, an established antibacterial metal used by immune cells to control M. tuberculosis and other microbes. Given that activated macrophages produce increased amounts of aldehydes during infection, we propose host-derived aldehydes may help control bacterial infections, making aldehydes a previously unappreciated antimicrobial defense.

mpa mutant suppresses its NO-sensitive phenotype (5). We tested if the disruption of log could also suppress the Cu-sensitive phenotype of an mpa mutant. We confirmed that a Δmpa::hyg mutant, hereafter referred to as the mpa mutant, is sensitive to Cu (Fig.  1A, left panel, center bars; see Table 1 for all strains used in this work). Disruption of log (log::MycoMarT7) in the mpa mutant strain (hereafter referred to as an mpa log mutant) restored WT Cu resistance (Fig. 1A, left panel, right-side bars). Given that disruption of log prevents pHBA accumulation in the mpa mutant (5), this result suggested accumulation of pHBA in PPS mutants contributes to the Cu sensitivity of these strains.
Previous work by our lab found two additional mutations that suppress the NO sensitivity of an mpa mutant (5). These suppressor mutations are in glpK, which encodes a probable glycerol kinase, and in the promoter of cei, which encodes a possible secreted alanine-rich protein involved in cell membrane integrity (21). We found the mpa glpK (Δmpa::hyg glpK::MycoMarT7) double-mutant had increased Cu resistance compared to the parental mpa strain (Fig. 1B, center bars). GlpK phosphorylates glycerol, resulting in the production of several aldehydes including MG and glyceraldehyde-3-phosphate. Thus, disruption of glpK might reduce the overall aldehyde burden in the mpa mutant, mitigating the negative effects of pHBA accumulation in this strain. In contrast, the mpa ceip (Δmpa::hyg ceip::MycoMarT7) mutant was more sensitive to Cu than the mpa mutant ( Fig. 1B, right side bars). Given that defects in cei increase membrane permeability (21), it is possible its reduced expression makes the bacterial cytosol more accessible to Cu.
We next tested if exogenously added pHBA could sensitize WT M. tuberculosis to Cu. We used 1.2 mM pHBA for all experiments in this study because this concentration, which is non-toxic on its own, can robustly synergize with NO to sterilize M. tuberculo sis cultures (5). We preincubated WT M. tuberculosis with pHBA in minimal media for 24 hours before exposing the bacteria to Cu, reasoning that pre-treatment with pHBA would allow for transcriptional or other changes that affect Cu resistance. Additionally, pHBA added concomitantly with CuSO 4 appeared to inhibit Cu-dependent killing for unclear reasons. As previously reported, pHBA alone had no effect on CFUs recovered (Fig. 1C, striped bars) (5). Pre-incubation of bacteria with pHBA reduced CFU after Cu treatment compared to Cu treatment alone (Fig. 1C, compare dark gray and black bars between untreated and pHBA-treated). Thus, endogenously produced or exogenously added pHBA sensitized M. tuberculosis to Cu.

Expression of Cu-responsive genes was reduced in pHBA-treated M. tubercu losis
Microarray analysis of PPS mutants lacking either mpa or pafA resulted in the identification of five promoters controlled by the DNA binding protein RicR (15). RicR dissoci ates from DNA in the presence of Cu, leading to the expression of genes required for robust Cu resistance and virulence (15,18). Another Cu responsive operon, the CsoR operon, is also repressed in PPS mutant strains. Like RicR, CsoR releases repression of its operon after binding to Cu (13,22). The CsoR operon includes ctpV, which encodes a cation transporter that contributes to Cu resistance and virulence (17). Given that PPS mutants are Cu sensitive and have reduced expression of two Cu responsive systems, we hypothesized pHBA accumulation in PPS mutants was responsible for the gene expression changes. To test this hypothesis, we performed RNA-Seq on WT M. tubercu losis treated with pHBA ( Fig. 2; Table S1). We found that of 3,979 genes analyzed, six genes were significantly differentially upregulated more than twofold, and 35 genes were significantly differentially downregulated by more than twofold. Functional gene set enrichment analysis revealed significant downregulation of processes related to diverse metal ions, including Cu and iron, oxidative stress, and sulfur metabolism (Table  S2). Remarkably, there was considerable overlap in the gene expression profiles of PPS mutants versus WT M. tuberculosis (15) and pHBA-treated bacteria versus untreated WT bacteria (Table 2 and 3, gray rows), which is reflected in the large overlap of functional gene sets with significantly altered expression (Table S2). The substantial overlap in the transcriptomes supported our hypothesis that pHBA contributes to the transcriptional were incubated for 10 days in the indicated CuSO 4 concentrations. "Input" (white bars) indicates CFU at the beginning of the experiment, and "0" indicates how many bacteria were present after 10 days of incubation without added CuSO 4 (striped bars). Bars represent mean with SD. "<LOD" indicates below the limit of detection (100 CFU). Significant differences were calculated, comparing CFU of strains to the first strain on the x-axis at the same CuSO 4 concentration using an unpaired t-test with *P < 0.05; ***P  Table S1). While the RicR regulon encodes eight genes, only two (excluding ricR) have been implicated in Cu resistance: mmcO and mymT, with a mymT mutant having the strongest Cu-sensitive phenotype (18,19). To follow up the transcriptional analysis, we sought to determine if protein levels of either of these key Cu resistance proteins were reduced in pHBA-treated bacteria or a PPS mutant. Given the low abundance and small size (53 amino acids) of MymT in WT M. tuberculosis (15,19), we instead looked at MmcO. MmcO is a membrane-associated Cu oxidase that is hypothesized to convert Cu(I) into less toxic Cu(II) based on its high similarity to other Cu oxidases (16,18). We assessed levels of MmcO in Cu-treated mpa, mpa log, and ricR null (ΔricR::hyg) strains relative to a WT parental strain. As previously reported, the ricR null mutant over-produces MmcO relative to WT bacteria; in contrast, the mpa mutant had approximately two-fold less MmcO relative to WT bacteria (Fig. 3A). Consistent with the suppressive effect of the log mutation on Cu sensitivity in a PPS mutant (Fig. 1A), the mpa log strain had WT levels of MmcO (Fig. 3A).
We next determined if exogenously added pHBA could affect MmcO levels. In the absence of Cu, pHBA treatment alone resulted in low MmcO levels in all strains ( Even with the addition of Cu, the mpa strain showed low levels of MmcO, nearly half the level observed in Cu-treated WT bacteria (Fig. 3B, lanes 2 v. 5), and the addition of pHBA had a minor effect on the already low MmcO levels in the mpa mutant (Fig. 3B, lanes 5 v. 6). MmcO levels in the mpa log strain were similar to that of the WT strain (Fig. 3B, lanes 2v. 8, and lanes 3 v. 9). Together, these results indicate that the presence of endogenously produced or exogenously added pHBA reduced the levels of at least one RicR-regulated gene product, which supports our transcriptional data. Up to this point, our data support a model whereby PPS-defective M. tuberculosis are hypersusceptible to Cu due to a reduction in Cu-responsive gene expression caused by pHBA. However, we could not rule out the possibility that pHBA directly disrupts the function of one or more of the Cu-responsive proteins. A way to test this hypothesis is to measure the effect pHBA has on a ricR null mutant, which constitutively expresses high levels of all of the RicR regulon genes. We performed a Cu sensitivity assay on a ricR mutant pretreated with pHBA. Because ricR mutants are highly resistant to Cu (15), we needed to use a higher Cu concentration than in previous experiments to kill the bacteria. At the highest CuSO 4 concentration used, pHBA robustly sensitized the ricR mutant to Cu (Fig. 4A, black bars), suggesting that repression of the RicR regulon was not the only mechanism by which pHBA sensitized M. tuberculosis to Cu. Given that the metallothionein MymT plays a dominant role in Cu resistance in M. tuberculosis, we hypothesized that pHBA could affect the function of MymT. Metallothio neins are found in all domains of life and protect cells from the potentially harmful effects of free Cu (23). MymT uses cysteines to bind reduced Cu [Cu(I)] (19). Given that aldehydes can readily form hemiacetals with cysteines and disable protein function (24), we hypothesized that pHBA reacts with cysteines in MymT, preventing its ability to bind and sequester toxic Cu. Cu metallothioneins form solvent-shielded Cu(I)-thiolate cores   (19,25). Thus, a reduction in luminescence would indicate decreased Cu binding. We first measured MymT Cu-thiolate cores from WT M. tuberculosis by fractionating whole-cell lysates of Cu-treated bacteria and measuring luminescence emitted from each fraction after excitation by UV light. A sharp peak of luminescence was observed in the fractionation profile that could be attributed to MymT, given that this peak was absent from lysates of a mymT null mutant (Table 1 Fig. 4B). We next examined the effect of pHBA on MymT luminescence in the ricR mutant. We observed a reduction of the MymT luminescence peak in lysates collected from the ricR mutant treated with pHBA compared to that of untreated bacteria, strongly suggesting pHBA affected the ability of MymT to bind Cu (Fig. 4C). To ensure equivalent MymT levels in ricR mutant lysates treated with or without pHBA, we pooled and concentrated the three fractions corresponding to the maximal MymT luminescence peak for immunoblot analysis. We a Results are grouped by locus name or transcriptional regulator. All rows except csoR were more than twofold downregulated in PPS degradation-deficient strains (15). pHBA, para-hydroxybenzaldehyde; WT, wild-type. found MymT levels were high and equivalent in ricR mutant fractions, whether or not they were pHBA-treated (Fig. 4C, lower panel). Interestingly, we consistently detected a smaller species of MymT in the pHBA-treated samples. It is possible this smaller species was caused by a direct modification of MymT by pHBA, an idea that remains to be tested. We also tested if MymT luminescence was reduced in a M. tuberculosis PPS mutant compared to in the parental strain. The MymT luminescence peak in the mpa strain was significantly lower, and this reduction was restored to near WT levels in the mpa log double mutant (Fig. 4D). These data support the hypothesis that the accumulation of pHBA reduced either the amount or activity, or both, of MymT in a PPS mutant.

A metabolic aldehyde sensitized M. tuberculosis to Cu
A recent hypothesis put forward by the Darwin and Stanley labs proposes host cellderived aldehydes of metabolism contribute to bacterial control during infections (26). Macrophages undergo an increase in aerobic glycolysis, known as the Warburg Effect, following infection with M. tuberculosis and other pathogens (27)(28)(29)(30). A by-product of aerobic glycolysis is MG, also known as pyruvaldehyde, which has been detected at millimolar concentrations in M. tuberculosis-infected mouse macrophages (31). Unlike with pHBA, MG pretreatment resulted in increased Cu resistance, leading us to hypothe size that MG preincubation specifically induced either an aldehyde or Cu resistance pathway, an idea we are testing. Nonetheless, when simultaneously added to bacteria with Cu, MG at normally non-toxic levels synergized with Cu to robustly kill M. tubercu losis (Fig. 5A, right panel, gray and black bars). Furthermore, MG treatment reduced luminescence of MymT in a ricR mutant (Fig. 5B). Overall, this result indicates that a physiologic aldehyde that is present during infections has the potential to sensitize M. tuberculosis to Cu by disrupting one or more Cu-responsive proteins.

DISCUSSION
In this work, we sought to test if the Cu-sensitive phenotype of an M. tuberculosis PPS mutant was due to an accumulation of the aldehyde pHBA. We found that elimination of the PPS substrate Log, which is the source of pHBA, restored Cu resistance to WT (D) Fractionated M. tuberculosis lysates from WT, mpa, and mpa log strains. Inset: quantification of fold change between two independent experiments. All quantifications were analyzed for significance using an unpaired t-test with *P < 0.05; **P < 0.01. Unlabeled bars did not show significant differences.
Research Article mBio levels. Furthermore, addition of pHBA to WT M. tuberculosis cultures was sufficient to sensitize bacteria to Cu. Cu sensitization in both PPS mutants and pHBA-treated WT M. tuberculosis was likely due to the reduced expression of genes needed for Cu resistance. We also showed that the Cu-binding function of MymT was altered in the presence of pHBA, possibly by disrupting cysteines in the protein. Finally, we showed that MG, an aldehyde produced by activated macrophages, also sensitized M. tuberculosis to Cu in vitro. Collectively, we propose a model whereby pHBA and other aldehydes can directly or indirectly disable Cu sensing by RicR, leading to the constitutive repression of the RicR regulon, and also disrupt Cu binding by MymT, preventing its ability to confer Cu resistance (Fig. 6).

Research Article mBio
In this study we also showed that a transposon disruption in glpK, which encodes a glycerol kinase, suppressed the Cu sensitive phenotype of a PPS mutant. This mutation also suppresses NO sensitivity (5). Along these lines, several independent studies recently reported that glpK mutations are often found in clinical M. tuberculosis isolates (32)(33)(34). These mutations are reversible and confer decreased susceptibility to antituberculosis drugs, suggesting phase variation occurs in response to antibiotic pressure on the bacteria. The Alland group proposed that the activation of a general stress response following a block in glycerol metabolism provides increased antibiotic tolerance (32). Similar to our hypothesis that a glpK mutation reduces aldehyde levels in PPS mutants, the Sassetti lab proposed that a block in glycerol metabolism, which produces MG, could protect bacteria by lowering endogenous aldehyde burden; this idea suggests aldehydes could sensitize bacteria to antibiotics (33).
In addition to the RicR regulon, the metal-responsive CsoR operon and Zur regulon were similarly regulated between pHBA-treated WT bacteria and PPS mutants. CsoR represses the expression of ctpV, which is a cation transporter involved in Cu resistance (17) and is repressed in PPS mutants and WT bacteria incubated with pHBA. CsoR and RicR are paralogs that use cysteines to sense Cu (13,18); thus, it is possible that pHBA also disrupts the ability of CsoR to bind Cu, resulting in the constitutive repressed production of CtpV.
The Zur regulon was upregulated in pHBA-treated M. tuberculosis (Table 3) and in PPS mutants (15). Unlike the Cu regulators, Zur is a zinc-responsive regulator that binds to and represses its promoters in the presence of excess zinc (35). Zur-regulated genes are thus induced in low zinc and implicated in zinc uptake (35). The ESAT-6 cluster 3 (ESX-3) genes in the Zur regulon are also regulated by iron-dependent repressor (IdeR) (36). However, no other IdeR-dependent genes were differentially expressed between untreated and pHBA-treated M. tuberculosis, similar to what we observed in PPS mutants (15). Notably, unlike the Cu regulators and Zur, IdeR does not use cysteine to coordinate iron.
In addition to the Zur regulon, several genes were induced in pHBA-treated M. tuberculosis and not in PPS mutants ( Table 3). Some of these genes, e.g., mesT and Rv0195, are upregulated in hypoxia or after damage to the mycomembrane (37)(38)(39). It is possible that these genes are specifically induced in response to extracellular aldehyde exposure and not by endogenously produced aldehyde. Alternatively, the amount of pHBA we used for RNA-Seq cultures induced a transcriptional response that would not be achieved by what are likely much lower pHBA concentrations found in PPS mutants.
We also observed several uncharacterized operons with gene expression patterns shared between PPS mutants and pHBA-treated WT bacteria, including: the cysDNC operon, which encodes a sulfate-activating enzyme complex that is implicated in virulence, oxidative stress, sulfate limitation, and sensing of exogenous cysteine (40); Rv0762c-Rv0771, which includes a gene for a putative aldehyde dehydrogenase (aldA) and is predicted to be regulated by a probable ArsR-like metalloregulatory transcrip tional repressor (Rv0576) (41,42); and Rv3249c-Rv3252c that is predicted to be controlled by WhiB4, a redox-responsive, iron-sulfur cluster-containing transcriptional repressor ( Table 2; Table S2) (43). Rv3249c-Rv3252c encodes the rubredoxins RubA and RubB and a putative monooxygenase AlkB. Relevantly, many of these genes are predicted to be regulated by metal-binding proteins that use cysteines to coordinate their respective metals (42,43). Because aldehydes can form adducts with thiols in proteins, it is possible that cysteine-dependent metal coordinating proteins are particularly sensitive to aldehyde exposure.
Our data also suggest that aldehydes have a direct effect on the function of MymT. We observed that pHBA and MG reduced MymT luminescence, suggesting these aldehydes directly disrupted Cu binding to MymT. An alternative explanation is that pHBA quenched MymT luminescence rather than directly preventing MymT Cu binding; however, a smaller species of MymT formed only after pHBA treatment, suggesting this aldehyde directly affected the structure of MymT. This smaller species of MymT may have formed a more compact conformation due to a covalent interaction with pHBA, allowing it to migrate through SDS-PAGE gels more quickly. Alternatively, pHBA-modified MymT could have adopted a conformation that exposed it to peptidases, resulting in partially degraded MymT. Unlike MymT, we could not test whether or not pHBA directly affected MmcO activity given that MmcO is a membrane-anchored protein without an established activity assay. Although MmcO is not predicted to use cysteines to function, its activity may nonetheless be affected by aldehydes, a hypothesis that remains to be tested.
A recent study by the Glickman lab identified an integrated system involving Rip1 protease and the PdtaS/R two-component system in M. tuberculosis that senses and mediates resistance to Cu and NO (44). The NO sensitivity of a rip1 mutant is attributed to a block in chalkophore biosynthesis. Chalkophores bind to Cu with high affinity (44)(45)(46), and a follow-up study supports a model that mycobacteria require chalkophores to acquire Cu during low Cu conditions, i.e., this system is not for Cu resistance per se but for Cu acquisition (47). While it remains to be determined how Cu and NO resistance is conferred by Rip1, it is unlikely that aldehydes are involved given that known Cu resistance genes are not repressed in a rip1 mutant. Importantly, these data suggest there are additional ways for M. tuberculosis to be sensitized to NO and Cu.
An active anti-microbial role for aldehydes in vivo has yet to be established. In macrophages, aldehydes are produced at low levels during cellular metabolism (48,49), but a shift to aerobic glycolysis following infection with M. tuberculosis leads to an increase of aldehydes such as glyceraldehyde-3-phosphate and MG (27-29, 31, 50-52). Induction of aerobic glycolysis plays a role in infection control given that the inhibition of this pathway in mouse macrophages leads to loss of some interferon-γ-dependent control of M. tuberculosis growth (27). Thus, aldehydes produced during this shift to glycolysis might contribute to antibacterial activity for the host.
More broadly, aldehydes may also have a role in the defense against other microbes. For example, a recent report by the Portnoy lab showed that in Listeria monocytogenes, MG activates transcription of glutathione (GSH) synthase-encoding gene gshF, leading to increased GSH production and thereby activating the master virulence regulator PrfA (53). A L. monocytogenes mutant that lacks a gene-encoding glyoxalase A, a key enzyme in MG detoxification, has decreased GSH levels in vitro and is attenuated for infection. Together, these results suggest that MG is an important cue for some pathogens to turn on virulence programming.
Our data provide the first evidence that aldehydes can sensitize M. tuberculosis to Cu. Importantly, despite well-established evidence of their toxicity, the antibacterial mechanisms of aldehydes are relatively uncharacterized; thus, our study may begin to provide new insight into how aldehydes can target and inactivate bacterial pathways needed for survival.
For CuSO 4 solutions, stock solutions were made by dissolving the appropriate amount of CuSO 4 powder (Fisher Scientific) in water and filter-sterilizing with a 0.45-µm filter. For aldehyde solutions, 50 mM pHBA was made with pHBA powder dissolved in water and filter-sterilized using a 0.45-µm filter. MG (5.5 M) was diluted in sterile water just before use. Both aldehydes were purchased from Sigma-Aldrich, Inc.

RNA-Seq
M. tuberculosis cultures were grown in 7H9c media and treated with 1.2 mM pHBA at an optical density at 580 nm (OD 580 ) of 1.0 (late logarithmic to early stationary phase) or left untreated. Twenty-four hours later, RNA was purified as previously described (15). RNA was isolated from three biological replicate cultures. Library preparation and Illumina HiSeq Sequencing were performed by GENEWIZ, LLC. Sequence reads were mapped to the M. tuberculosis H37Rv genome sequence (RefSeq identifier GCF_000195955.2) using bwa v0.7.17 (54) and sorted using samtools v1.9 (55). An average of 97.8% of reads were mapped to the reference genome, indicating high quality of samples. Given a locus of interest, socAB, was not included in the assembly annotation from RefSeq, we added it to the annotation files at genomic coordinates NC_000962.3:1,933,937-1,934,497. Using the alignment files generated for each sample, the featureCounts command in Subread v2.0.1 (56) was used to count the reads mapping to each gene in the reference. Read counts per gene and sample (feature counts) were loaded into R v4.2.0 (57) for further analysis using the package DESeq2 v1.36.0 (58). DESeq2's normalization function was used to normalize the counts to make expression levels more comparable between the different samples. To compare gene expression changes between pHBA-treated and untreated samples, the Wald test was used to generate P-values and log 2 fold changes. Genes with an adjusted P-value of <0.05 and fold change of >2.0, when comparing pHBA-treated to untreated M. tuberculosis, were considered differentially expressed genes (Table S1). Raw sequencing data files are available in a PATRIC public workspace: (https://www.bv-brc.org/workspace/ginalimon@bvbrc/Limon_Darwin_RNA-Seq). The volcano plot was generated using python v3.10.6 and package bioinfokit v2.1.10 (59).

Gene set enrichment analysis
Gene Ontology (GO) annotations for assembly M. tuberculosis H37Rv (Genbank ID: AL123456.3) were obtained via the QuickGO REST API by querying all locus tags in the assembly's annotation. Gene set enrichment analyses were carried out for each of the three GO with topGO v.2.44.0 using the weight01 algorithm and the Fisher statistic, using a P-value significance threshold of 0.05. For genes whose expression was deemed significantly upregulated (21 genes) or downregulated (38 genes) in a previous publication, the gene universe consisted of all genes targeted by the probes of the deployed microarray (GEO ID GPL4292). For genes whose expression was deemed significantly altered in this publication, the gene universe consisted of all genes annotated in assembly M. tuberculosis H37Rv (Genbank ID: AL123456.3).

Cu sensitivity assay
Cu sensitivity assays were performed as previously described (15,18). Briefly, M. tuberculosis strains were grown in 7H9c to an OD 580 = 0.5-1.0. Bacteria were washed once with Sauton minimal media with no added Cu and collected using a low-speed centrifugation (150 g) to remove clumped cells. Supernatants containing mostly unclumped bacteria were diluted to OD 580 = 0.08 in Sauton minimal media. One hundred ninety-four microliters of this diluted culture was transferred to 96-well plates; 6 µL of the appropriate stock concentration of CuSO 4 or aldehyde or both was added to the desired final concentration. Plates were incubated at 37°C for 10 days; after which, cultures were diluted and inoculated onto 7H11 agar Y-plates. Plates were incubated for 14-21 days before enumerating CFU. As previously reported (15), we used a range of CuSO 4 concentrations due to variability of Cu sensitivity between experiments. Each experiment was done at least twice in technical triplicate.

M. tuberculosis lysate preparation for immunoblotting
For all blots, bacteria were grown in Sauton minimal media with no added Cu. For MmcO blots, at OD 580 = 0.5-1.0, cultures were treated with 1.2 mM pHBA for 4 hours before treatment with 50 µM CuSO 4 . Bacteria were harvested 24 hours after the addition of CuSO 4 . Bacterial densities were measured and equivalent cell numbers were collected based on the OD 580 of the cultures. For example, an "OD 580 equivalent of 1" indicates the OD 580 of a 1-mL culture is 1.0. For most assays, 5 OD 580 units were collected and washed once with Dulbecco's phosphate-buffered saline (Corning, Inc with 0.05% Tween-80) to remove BSA in 7H9c media. Bacterial pellets were then resuspended in 300 µL TE buffer (100 mM Tris-Cl, 1 mM EDTA pH 8.0), and transferred to bead-beating tubes with 200 µL zirconia beads; tubes were beaten for 30 seconds three times, with icing for 30 seconds in between in a mini bead beater (all materials from Bio-Spec). One hundred fifty microliters of lysate was transferred into new tubes with 50 µL 4✕ SDS sample buffer (250 mM Tris pH 6.8, 2% SDS, 20% β-mercaptoethanol, 40% glycerol, 1% bromophenol blue) and boiled at 100°C for 10 minutes. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto nitrocellulose membranes. Membranes were blocked in 3% milk or BSA prior to incubation with polyclonal rabbit antibodies as indicated in the figure legends. For loading controls, the same membranes were stripped with 0.2 N NaOH as described elsewhere (60) and re-blocked before incubation with antibodies to M. smegmatis PrcB (61).

Construction of a ΔricR::hyg mutant
We made a M. tuberculosis ricR deletion mutant strain using a previously described method (15). Briefly, pYUB854 (20) was used to clone sequences encompassing ~700 bp upstream (5′) and ~700 downstream (3′) of the ricR gene. The 5′ and 3′ sequences, including the start and stop codons, respectively, were cloned to flank the hygromycin resistance cassette in pYUB854. The plasmid was digested with PacI, and approximately 1 µg of linearized, gel-purified DNA was used for electroporation into M. tuberculosis. Mycobacterium tuberculosis strains were grown to an OD 580 of ~0.4 to 1, washed, and resuspended in 10% glycerol to make electrocompetent cells as described in detail in (62). Bacteria were inoculated onto 7H11 agar with 50 µg/mL hygromycin as needed; a no-DNA control electroporation was done to control for spontaneously antibiotic-resist ant mutants. Two weeks after plating, colonies were picked and inoculated into 200 µL 7H9c with antibiotics and then inoculated into 5-mL cultures for further analysis. Mutants were confirmed by PCR and sequence analysis.

MymT luminescence from M. tuberculosis lysates
We adapted a previously reported protocol for measuring Cu(I)-thiolate core lumines cence for use on filtered M. tuberculosis lysates (19). M. tuberculosis cultures were grown in Sauton minimal media to an OD 580 = 0.3-0.5 and treated with pHBA to a final concentration of 1.2 mM as needed, and 50 µM CuSO 4 4 hours later. Twenty-four hours after Cu addition, 12 OD 580 equivalent cell numbers were harvested by centrifugation and washed twice with buffer (10 mM HEPES, 150 mM NaCl pH 7.4). Bacteria were resuspended in 700 µL of the same buffer and lysed by bead beating as described for preparing lysates for immunoblotting. Lysates were then centrifuged for 7.5 minutes at 20,000 g, and the supernatants were passed through a 0.2-µ spin filter twice before application onto a Superose-6 10/300 GL column (Cytiva). Fractions were transferred to a UV-grade 96-well plate (Corning), and luminescence was measured with excitation at 280 nm, emission at 595 nm, and a cutoff of 325 nm.
For immunoblotting proteins in fractionated lysates, 200 µL of fractions correspond ing to the three at the peak fluorescence collected using method above were stored at -20°C. Fractions were then thawed on ice, pooled, and concentrated in a 0.5-mL centrifugal filter (Amicon). Samples were boiled in SDS sample buffer for 10 minutes before separation on 15% SDS-PAGE gels and transferred onto nitrocellulose membranes. Polyclonal MymT antibodies used for immunoblotting were a kind gift from Ben Gold and Carl Nathan (19).

ADDITIONAL FILES
The following material is available online.