Details in Supplementary Material.

M. bovis BCG (ATCC35734), M. tuberculosis H37Rv (ATCC 27294) and M. bovis-mRFP-SsrA strains were maintained in Middlebrook 7H9 media (Difco) supplemented with 0.5 % (v/v) glycerol, 0.05% (v/v) Tween 80, and 10% (v/v) Middlebrook ADC (Albumin-Dextrose-Catalase) (Difco). Hygromycin B (Roche) was added when appropriate. Enumeration of bacteria was performed by plating on Middlebrook 7h10 (Difco) agar plates containing 0.5% (v/v) glycerol, and 10% (v/v) Middlebrook OADC (Oleic acid-Albumin-Dextrose-Catalase) (Difco). HepG2 cells (ATCC HB-8065) were cultured at 37°C with 5% CO₂ atmosphere in DMEM media (Gibco) complemented with 10% FBS heat-inactivated (Gibco), penicillin (100 U/mL, Gibco) and streptomycin (100 μg/mL, Gibco).

The plasmid pGMEH-p38-mRFP-SsrAec3 (Hygro^R) carries the mCherry RFP gene cloned downstream of the p38 mycobacterial promoter and fused to the ClpP1P2-specific SsrA tag as well as a hygromycin-resistance cassette. pGMEH-P38-che-ssrAec3 was a gift from Dirk Schnappinger (Addgene plasmid # 27059). The plasmid was electroporated into M. bovis BCG to generate M. bovis-mRFP-SsrA. Transformants were recovered on 7H10 agar supplemented with 50 μg/mL hygromycin (Sigma-Aldrich) and grown in 7H9 broth supplemented with the same concentrations of hygromycin. Pre-culture of M. bovis-mRFP-SsrA were then harvested at mid-log phase, diluted to OD₆₀₀ 0.2 in complete 7H9 media and dispensed into 96-well plates (200 μL/well) in presence of two-fold serially diluted compounds. M. bovis-p38-mRFP-SsrA untreated samples were used as negative control whereas M. bovis-p38-mRFP-SsrA treated with Bortezomib was used as positive control. Fluorescence signal acquisition was carried out after 24 h of incubation using a M200 Pro plate reader (Tecan). Red fluorescence was acquired under excitation/emission at λ = 587/630 nm. Relative fluorescence units were plotted as a function of drug concentration. The maximum fluorescence value obtained with Bortezomib was taken as the maximum (i.e., 100%) of ClpP1P2 inhibition. ClpP1P2 IC₅₀ (i.e., the concentration required to inhibit 50% of ClpP1P2 activity) was determined in three independent replicates.

Turbidity-based growth inhibition was performed to assess anti-mycobacterial potency of the compounds. M. bovis BCG or M. tuberculosis H37Rv pre-cultures were harvested at mid-log phase and diluted to OD₆₀₀ 0.05 in complete 7H9 media. Bacterial suspensions were then dispensed in 96-well plate (200 μL/well) in presence of two-fold serially diluted compounds and incubated for 5 days at 37°C under shaking (100 rpm). Cells were manually resuspended and OD was measured at 600 nm on M200Pro plate reader (Tecan). Percentage of growth was determined as compared to untreated control and plotted as a function of drug concentration. MIC₅₀ (i.e., the concentration that inhibit 50% of growth) was determined in three independent replicates. MBC was determined by colony forming unit (CFU) enumeration on agar. Briefly, 10⁶ CFU from mid-log phase pre-culture of M. tuberuculosis H37Rv were treated for 5 days with concentrations ranging from one to ten times the MIC₉₀ (Minimum inhibitory concentration required to inhibit 90% of bacterial growth) of each compound. Treated cultures were then plated on agar and CFU counted after 3–4 weeks of incubation at 37°C. CFU from an untreated inoculum were enumerated in similar manner as a control. MBC₉₉ (i.e., the concentration that induce a 100-fold reduction in CFU count) was recorded. MBC₉₉ was determined thrice independently.

Cell-based CT-like peptidase assay was performed using the Proteasome-Glo™ Cell-Based Assay Reagent (Promega) according to manufacturer's guidelines. Briefly, HepG2 cells (10⁴ cells/well) or M. bovis BCG log phase cultures were treated with the indicated compounds for 2 h followed by incubation with the luminescent substrate for 10 min. Luminescence was detected with a Tecan M200 Pro plate reader. Relative luminescence (RLU) was plotted as a function of drug concentrations and Proteasome IC₅₀ (i.e., the concentration required to inhibit 50% of the proteasome activity) was determined. Bortezomib was used as a positive control.

Chloromethyl ketones (CMKs) comprise a distinct class of covalent serine protease inhibitors (Bogyo and Wang, 2002). We speculated that a Bortezomib analog carrying an alternative CMK warhead may retain ClpP1P2 activity while reducing activity against the human proteasome. Thus, we synthesized a Bortezomib derivative replacing the boronic acid with a CMK warhead (Pyrazine-phenylalanine-leucine-chloromethyl ketone or Pyr-FL-CMK) and determined its potency against bacterial and human enzymes. To measure the intracellular ClpP1P2 inhibition, we employed a M. bovis BCG strain expressing a ClpP1P2-specific SsrA-tagged Red Fluorescent Protein (RFP) reporter as described previously (Moreira et al., 2015). This reporter strain allows the measurement of the intracellular inhibition of ClpP1P2 protease via the accumulation of the non-degraded tagged RFP protein (Figure 1A). Under undisturbed conditions, the ClpP1P2 complex recognizes the SsrA-tagged RFP and degrades the protein, resulting in low basal fluorescence. In presence of a ClpP1P2 inhibitor, RFP-SsrA is no longer degraded and accumulates, leading to increase in fluorescence. We report here the ClpP1P2 IC₅₀, i.e., the drug concentration required to inhibit 50% of activity. Bortezomib was used as a positive control. Bacterial and mammalian proteasome intracellular inhibition was measured using the whole-cell target-based Proteasome-Glo assay (Promega). This assay makes use of a proteasome chemotryptic activity-specific LLVY-tagged aminoluciferin (Gandotra et al., 2010). Under undisturbed conditions, the proteasome cleaves the LLVY tag, allowing the luciferase to oxidize the aminoluciferin generating luminescence. In the presence of a proteasome inhibitor, the LLVY cleavage does not occur. The tagged aminoluciferin cannot be used by the luciferase enzyme, preventing the emission of light (Figure 1B). We report here the proteasome IC₅₀, i.e., the concentration required to inhibit 50% of the proteasome activity as compared to an untreated control. Bortezomib was again used as a control of proteasome inhibition.

As shown in Table 1, the CMK analog of Bortezomib, Pyr-FL-CMK, retained its activity against the bacterial target ClpP1P2 with about a 15-fold loss of potency (25 vs. 1.6 μM). Surprisingly, it also retained its activity against the bacterial proteasome albeit with a similar loss of potency compared to Bortezomib (25 vs. 0.8 μM). In contrast, Pyr-FL-CMK was inactive against the mammalian proteasome when tested up to a concentration of 500 μM, while Bortezomib (Pyr-FL-BA) displayed a proteasome IC₅₀ of 5 nM. Mycobacterial ClpP1P2 and proteasome inhibition translated into whole cell activity against both the M. bovis BCG reporter strain as well as the actual pathogen M. tuberculosis with growth inhibitory concentrations (MIC₅₀) of 20 and 25 μM, respectively. Furthermore, Pyr-FL-CMK displayed bactericidal activity within two-fold of Bortezomib's MBC₉₉ (the minimum bactericidal concentration required to kill 99% of the bacterial population) of 100 vs. 50 μM. Finally, we assessed cytotoxicity of both Bortezomib and its CMK analog against mammalian cells using a standard MTS assay (Riss et al., 2004). We report here the CC₅₀ (cytocidal concentration required to kill 50% of the cells) for each compound. The CMK analog was not toxic to HepG2 cells at up to 500 μM while Bortezomib displayed a CC₅₀ of 250 μM. These results show that replacing Bortezomib's boronic acid with a chloromethyl ketone was sufficient to confer selective inhibition of the bacterial target over the mammalian proteasome. Surprisingly, the CMK analog not only retained ClpP1P2 activity but also mycobacterial proteasome activity (Table 1). Bacterial whole-cell growth inhibition, bactericidal potency were also retained while cytotoxicity was limited (Table 1).

To further support these findings, we interrogated a small library of 11 commercially available peptides harboring a chloromethyl ketone head group. We identified two CMK compounds active against mycobacterial ClpP1P2 and proteasome but inactive against the mammalian proteasome. As shown in Table 1, Carboxybenzyl-glycine-leucine-phenylalanine-CMK (Z-GLF-CMK) and Carboxybenzyl-glycine-glycine-phenylalanine-CMK (Z-GGF-CMK) inhibited ClpP1P2 and the proteasome at similar concentrations when compared to Pyr-FL-CMK while being inactive against the mammalian proteasome at concentrations up to 500 μM. For both compounds, target inhibition translated to whole cell bacterial growth inhibition. For Z-GLF-CMK, bactericidal activity was further characterized with an MBC₉₉ of 200 μM (four- and two-fold higher than Bortezomib and Z-FL-CMK, respectively). However, it appeared that Z-GLF-CMK and Z-GGF-CMK both exhibited some cytotoxic activity against HepG2 cells with CC_50s of 60 and 125 μM, respectively. These compounds have been associated with serine protease inhibition, namely chymotrypsin, granzyme B, and cathepsin G (Bogyo and Wang, 2002) and they may have off-target effects which could account for their cytotoxicity. Nevertheless, these two compounds further support that peptidyl-chloromethyl ketone inhibitors display selective inhibition of both bacterial ClpP1P2 and proteasome targets.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.