Heat shock repressor HspR accumulates in a M. smegmatis bpa deletion mutant

It was previously reported that the heat shock repressor HspR accumulates in a bpa knockout strain in M. tuberculosis and is a Bpa-dependent proteasomal degradation substrate in vitro (Jastrab et al, 2015). To explore if this observation could be corroborated in the soil bacterium M. smegmatis, we generated a M. smegmatis Δbpa deletion mutant and performed a proteomic analysis comparing the wild-type M. smegmatis strain with the bpa knockout strain. Each strain was cultivated in biological triplicates under both standard and heat shock conditions. The bacterial proteomes were analyzed using a label-free, quantitative proteomic approach based on data-independent acquisition mass spectrometry. When assessing the relative abundance of HspR, we observed that under heat shock conditions, HspR showed a fivefold increase in abundance in the bpa knockout strain compared with the wild-type strain. In fact, even under standard conditions, HspR still accumulated threefold in the knockout over the wild-type strain (Fig 1A and B). This suggests that Bpa-dependent proteasomal degradation controls the levels of HspR in M. smegmatis, both under standard and heat shock conditions.

In addition to HspR, we identified other proteins that accumulated in the bpa knockout strain, both under standard conditions and during heat shock, under the chosen criteria (>twofold abundance increase, <0.05 adjusted P-value). However, among the 22 proteins identified under standard conditions and the 13 proteins identified during heat shock, HspR is the only protein identified for both (Tables S1 and S2). Furthermore, under heat shock, HspR stands out as the most significant hit with the highest fold-change. It can be concluded that HspR is the main Bpa-dependent substrate under heat shock conditions also in M. smegmatis.

Deletion of bpa in M. smegmatis causes sensitivity to heat shock and oxidative stress during growth in liquid culture

Based on our observation that HspR accumulated significantly in a heat-shocked M. smegmatis Δbpa mutant strain, we aimed to determine if the bpa knockout strain experienced any growth defects under this stress. We conducted growth experiments with wild type, bpa knockout, and bpa complemented M. smegmatis strains in minimal medium at 37°C or at 45°C. To induce heat shock, the cultures were initially grown at 37°C until reaching the logarithmic phase, diluted, and then subjected to a temperature shift to 45°C. Under standard conditions, the knockout strain displayed no growth impairment. However, when exposed to heat shock conditions, the bpa knockout strain exhibited significant growth impairment compared with both the wild type and bpa complemented strains. This resulted in a roughly twofold increase in the generation time for the bpa knockout strain (9.4 h) compared with either the wild type (4.9 h) or the bpa-complemented strain (4.5 h) (Fig 1C).

In addition, aiming to explore another form of stress that could impact protein structure and function, we investigated the effects of oxidative stress on the bpa deletion strain in comparison to the wild-type strain by adding 5 mM H₂O₂ to cultures grown in minimal medium. Similar to the previous experiment, no growth defect was observed for the bpa knockout strain under standard conditions. However, under oxidative stress, the bpa knockout strain was no longer viable. This phenotype could be reversed by complementing the knockout strain with bpa under its native promoter (Fig 1D).

Taken together, the growth experiments suggest that the M. smegmatis mutant strain deficient in bpa is impaired under stress conditions known to impact protein structure and function.

HspR is a Bpa-specific proteasomal degradation substrate in vitro

Based on our proteomics analysis, we identified HspR as the primary Bpa-dependent substrate under heat shock conditions. Next, we explored degradation of HspR in vitro. Bpa lacks an ATPase motor domain that unfolds substrates for translocation into the proteasomal degradation chamber. We therefore hypothesized that Bpa-dependent proteasomal degradation requires inherent instability or disorder in the recruited substrates. Previous studies have shown that mycobacterial HspR undergoes partial unfolding during heat shock, leading to its dissociation from the natural operator and allowing for the transcription of vital heat shock chaperones such as DnaK (Bandyopadhyay et al, 2012). We postulated that DNA-bound HspR would be more stable against proteasomal degradation. To test this hypothesis, we conducted in vitro experiments using recombinantly produced proteins from M. tuberculosis. We measured the degradation of HspR in the presence or absence of the HAIR motif DNA. We found that in the absence of DNA, Bpa mediated the rapid turnover of HspR by the 20S CP, whereas the 20S CP alone was unable to degrade HspR (Fig 2A). However, when HspR was incubated with the HAIR motif before the assay, it exhibited significant stabilization against Bpa-mediated proteasomal degradation. These results support our hypothesis that DNA-bound HspR is protected from proteasomal degradation.

Previous studies have also demonstrated that DnaK binds to HspR as a co-repressor to repress its own transcription and the transcription of the other members of the dnaKJgrpE-hspR operon under standard conditions (Bucca et al, 2000; von Rosen et al, 2021). Therefore, we investigated whether DnaK could stabilize HspR against degradation by the Bpa–CP complex. In this experiment, we pre-incubated HspR with DnaK before the assay and observed no degradation of HspR during the recorded eight-hour time course. This contrasts with HspR alone, which was completely degraded within ∼1 h. These findings suggest that HspR, when in a stable ternary complex with its cognate DNA or the co-repressor DnaK, is either unable to bind to Bpa or cannot be engaged by the proteasome core. Furthermore, by monitoring the temperature-related intensity changes of fluorescently labeled Bpa in a capillary-based titration experiment with HspR, we demonstrated that when in a 1:1 complex with DnaK or its cognate DNA, HspR no longer interacts with Bpa (Fig S1A). Notably, the stabilization of HspR against Bpa-mediated proteasomal degradation by its cognate DNA is temperature-dependent, as observed in gel-based degradation assays (Fig S1B). At 30°C, HspR in complex with its cognate DNA is completely stabilized against Bpa-mediated proteasomal degradation. At 37°C, less than 50% of HspR is degraded after 8 h. At 42°C (heat shock temperatures), degradation of HspR by the Bpa–CP complex occurs even faster (Fig S1C), albeit still slower than in the complete absence of DNA at 37°C (Fig 2A).

To determine whether HspR is specifically targeted for proteasomal degradation by Bpa, we investigated whether the alternative proteasome interactor Mpa could facilitate the degradation of HspR (Fig 2B). Thus far, attempts to achieve in vitro binding of Mpa to the wild type 20S CP have been unsuccessful, necessitating the use of an “open-gate” variant of the 20S CP (20S CP^OG), lacking the first seven amino acid residues of the proteasomal α-subunits. By assembling Bpa or Mpa (in the presence of ATP) with the 20S CP^OG, fully saturating the proteasome to exclude unstructured substrates from entering the degradation chamber directly, we found that the Bpa–CP^OG complex rapidly degraded HspR. Under the chosen reaction conditions, the turnover of the substrate was nearly completed within the first 10 min of the reaction time course. In contrast, the Mpa–CP^OG complex was unable to degrade HspR, and the substrate remained stable against degradation over the entire eight-hour time course. Therefore, our results indicate that HspR is a Bpa-specific proteasomal degradation substrate. This finding is consistent with the observation that HspR levels increase in a M. tuberculosis bpa knockout strain but not in an mpa knockout strain (Jastrab et al, 2015).

HspR is conserved in actinobacteria

To explore the substrate determinants influencing turnover by the Bpa–proteasome complex, we first examined the conservation of HspR domains, which could potentially contribute to substrate recruitment and processing.

We conducted a multiple sequence alignment of HspR orthologs from 20 actinobacterial species using ClustalW (Fig 3A). The alignment showed high conservation of HspR in the DNA-binding helix-turn-helix (HTH) domain and in a short sequence stretch of ∼9 residues near the C-terminus (Fig 3A). Using AlphaFold, we then generated a structure prediction for a dimer of M. tuberculosis HspR, which predicted regions of disorder in the N-terminal region (1–15) and the C-terminal region (residues 116–125, red) (Fig 3B). The AlphaFold model indicated an α-helical fold for the conserved HTH domain (depicted in blue) and the less conserved C-terminal domain (C-dom) (Fig 3B). Based on the multiple sequence alignment, the AlphaFold model, and disorder predictions (Walsh et al, 2014), we constructed a domain organization model of HspR (Fig 3C), where the N- and C-terminal sequence regions are predicted to be disordered.

The conserved disordered C-terminal region of HspR is important for Bpa-mediated degradation

To investigate if the predicted disorder in the N- and C-terminal sequence regions serves as the primary determinant for Bpa-mediated proteasomal degradation, we generated truncated variants of M. tuberculosis HspR. These variants lacked either the non-conserved N-terminal stretch (HspRΔN15) or the conserved C-terminal region (HspRΔC9) (Fig 4A). We then recorded Bpa-mediated proteasomal degradation time courses for wild-type HspR (HspR WT) and the two HspR truncation variants (Fig 4B). We found that Bpa in conjunction with 20S CP^WT rapidly degraded HspR WT, achieving nearly complete turnover within 5 min under the chosen conditions. In contrast, the 20S CP^WT alone was unable to degrade HspR within an eight-hour time course under the same conditions (Fig 4B, top). When a hexapeptide carrying the GQYL proteasomal interaction motif, capable of inducing gate-opening by inserting itself between the α-subunits of the 20S CP, was used at sufficiently high concentration (200 μM), it supported proteasomal degradation of HspR WT almost as effectively as Bpa, resulting in near-complete substrate turnover after 30 min. Similarly, the permanently open proteasome variant 20S CP^OG, where the α-gate was truncated, efficiently turned over HspR within 15 min under the chosen conditions.

The Bpa-dependent proteasomal degradation of the C-terminally truncated HspR variant, HspRΔC9, occurred significantly slower compared to degradation of HspR WT. Complete degradation was achieved only after about 2 h under the chosen conditions (Fig 4B, middle). When the hexapeptide with the C-terminal GQYL motif was used instead of Bpa to induce gate-opening, the reaction was even slower, taking ∼8 h to complete. In the absence of Bpa or the hexapeptide, HspRΔC9 remained stable against degradation by 20S CP^WT throughout the entire time course. However, the genetically engineered open-gate variant, 20S CP^OG, degraded HspRΔC9 within 15 min, only slightly slower than the degradation of HspR WT by 20S CP^OG.

Bpa-mediated proteasomal degradation of the N-terminally truncated HspR variant, HspRΔN15, occurred even slightly faster than the degradation of HspR WT. This indicates that the N-terminal disordered region does not facilitate Bpa-mediated proteasomal degradation (Fig 4B, bottom). Substituting Bpa with the hexapeptide containing the GQYL interaction motif, which induces gate-opening of the 20S CP^WT, led to slower degradation of the HspRΔN15 variant, resulting in complete degradation of the substrate within 30 min compared with 5 min required for the Bpa-mediated reaction. In the absence of gate-opening, HspRΔN15 remained stable over the course of 8 h, whereas the engineered open-gate variant degraded HspRΔN15 within 5 min. The degradation behavior of the HspRΔN15 variant was thus comparable to what we observed for HspR WT (Fig 4C). Collectively, these results suggest that the disordered C-terminal region of HspR, but not its N-terminal region, acts as a degradation determinant for Bpa-mediated proteasomal degradation. However, the role of the C-terminal region in substrate binding or threading remained unclear at this stage of the study.

Disorder is not sufficient for Bpa-mediated proteasomal degradation

To further explore the role of disorder in Bpa-mediated proteasomal degradation, we examined two model substrates in their native and permanently non-native conformations (Fig S2A and B). Were conformational state of a protein to be the sole determinant for substrate recruitment, the non-native form of the model substrate should be readily recruited and degraded by the Bpa–CP complex, whereas the folded form of the substrate should remain stable against Bpa-mediated proteasomal degradation. One of the model substrates used was ribonuclease A (RNaseA) from the bovine pancreas. Native RNaseA is stabilized by four disulfide bridges between eight cysteine residues. To generate a non-native form of this model substrate, we unfolded RNaseA under reducing conditions and modified the cysteine residues by S-carboamidomethylation using iodoacetamide. Intact mass spectrometry confirmed the complete alkylation of all eight cysteine residues (Fig S2C). The secondary structure composition of folded and unfolded RNaseA was assessed by circular dichroism, with folded RNaseA exhibiting expected contributions of both α-helical and β-sheet secondary structure elements, which were absent in the carbamidomethylated RNaseA preparation (Fig 5A, left). The second conformational model substrate was FimA, the main subunit of the type 1 pilus rod from uropathogenic Escherichia coli. Protomers in the assembled pilus engage in a domain-swap interaction known as donor-strand complementation. FimA possesses an incomplete immunoglobulin-like fold lacking the C-terminal β-strand, which is supplied by the subsequent pilus subunit carrying the missing strand as an N-terminal extension (the donor strand) (Puorger et al, 2011). We used a variant called FimAa, where the missing N-terminal donor-strand is fused to the C-terminus, resulting in a self-complemented, highly stable monomer. Within the time frame of our experiments, FimAa can be considered permanently folded, as unfolding occurs on average only once every 10¹² yr. Another variant, FimAt, lacks the N-terminal extension, rendering it permanently unfolded, as indicated by the comparison of the CD spectra of FimAa and FimAt (Fig 5A, right) (Puorger et al, 2011).

To assess the conformational state as a determinant for substrate recruitment, we performed degradation experiments using the folded and non-native forms of the two model substrates. Our degradation time courses show that Bpa cannot facilitate the proteasomal degradation of the folded model substrates (Fig 5B). Both native RNaseA and the permanently folded FimA variant, FimAa, remained completely stable against Bpa-mediated proteasomal degradation (Fig 5B). Likewise, gate-opening with the GQYL-containing hexapeptide did not result in the degradation of native RNaseA or FimAa by the 20S CP. Even the engineered open-gate proteasome variant was unable to degrade either of the two natively folded substrates. These results provide further evidence for the principle of compartmentalization used by the proteasome and related degradation machines, wherein stably folded domains are unable to gain access into the degradation chamber.

In contrast, the engineered open-gate proteasome efficiently degraded the unfolded forms of RNaseA and FimAt, demonstrating that fully disordered proteins can diffuse through the ungated α-ring pore and be degraded. However, remarkably, we observed no degradation of the unfolded form of either model substrate by the Bpa–CP complex (Fig 5C). Substituting Bpa with the GQYL-containing hexapeptide led to slow degradation, with almost all FimAt degraded by the end of the eight-hour time course, and approximately half of the non-native RNaseA degraded within the same time frame. These results suggest that Bpa provides an additional layer of selectivity and that although conformational disorder is necessary for entry into the proteasomal degradation chamber, it alone is insufficient to render a protein a substrate for the Bpa-CP complex.

Having established that the disordered C-terminal stretch of HspR contributes to Bpa-dependent turnover of HspR (Fig 4B), next, we examined if we could render the unfolded FimA variant, FimAt, a substrate for Bpa-mediated proteasomal degradation by creating a C-terminal fusion with the C-terminal domain of HspR (FimAt^HspR-Cdom). However, the fusion did not significantly enhance the degradation of FimAt by the Bpa–CP complex (Fig 5D). Interestingly, we observed that FimAt^HspR-Cdom was degraded by 20S CP^WT when the GQYL-containing hexapeptide was used to open the gate, resulting in complete degradation within 30 min. This suggests that the HspR C-terminal disordered region facilitates diffusion into the proteasome chamber, but additional determinants are required for efficient Bpa-mediated proteasomal degradation.

The conserved disordered C-terminal stretch of HspR is a specific threading motif for Bpa-mediated proteasomal degradation

Our findings indicate that the C-terminal stretch of HspR plays a significant role in Bpa-mediated proteasomal degradation. For a more quantitative comparison of the degradations of HspR WT and HspRΔC9, we conducted gel densitometric analysis of their initial degradation rates (Fig S3A and B). The comparison showed that HspR WT is degraded an order of magnitude faster than the C-terminal truncation variant. The apparent initial velocity of degradation of wild-type HspR by the Bpa–proteasome complex under the chosen conditions is 0.65 μM/min, whereas that of the C-terminal truncation variant is 0.06 μM/min. However, the specific involvement of the C-terminal disordered region in substrate recruitment remained unclear. One explanation could be that this part of HspR serves as a direct binding determinant. Alternatively, the disordered region could act as a facilitator for threading, as its extended conformation and lack of tertiary interactions would enable diffusion of the C-terminal tail across the open proteasomal gate.

Previous studies suggested hydrophobic rings within the dodecameric core of Bpa are involved in hydrophobic interactions between substrates (Hu et al, 2018). The conserved C-terminal–disordered region of HspR contains several hydrophobic residues that could potentially participate in such interactions (Fig 3A). To investigate the involvement of the C-terminal region in HspR binding to Bpa, we fluorescently labeled Bpa and monitored the temperature-related fluorescence intensity changes in a capillary-based titration experiment with HspR WT and HspRΔC9 (Fig 6A). If the C-terminal region of HspR was critical for Bpa binding, we should observe a higher dissociation constant for HspRΔC9 than HspR WT. However, we found that the normalized fluorescence intensity changes were similar for HspR WT and HspRΔC9, with comparable calculated dissociation constants of 5.60 ± 0.97 μM and 3.62 ± 0.58 μM, respectively. These results suggest that HspR WT and HspRΔC9 bind Bpa with similar affinities, and the conserved C-terminal region of HspR does not significantly contribute to the binding to Bpa.

Although disorder appears to be important for Bpa-dependent substrate degradation, HspR WT possesses a significant amount of α-helical secondary structure, which is maintained in the truncated variant HspRΔC9 (Fig 6B). Previous studies have suggested that the C-terminal sequence stretch of HspR modulates its thermal stability (Bandyopadhyay et al, 2012). Therefore, we measured thermal unfolding transitions of the wild-type HspR and the HspRΔC9 variant (Fig 6C). Indeed, we observed that HspRΔC9 had a higher melting temperature (T_m = 320.9 ± 1.8 K/47.8 ± 1.8°C) than HspR WT (T_m = 314.8 ± 0.6 K/41.6 ± 0.6°C). However, we found that the degradation defect upon deletion of the C-terminal stretch was not temperature dependent (Fig S3C and D). Bpa-mediated proteasomal degradation of both HspR WT and HspRΔC9 was slower at lower temperature (30°C) compared with 37°C. Yet, the degradation defect of HspRΔC9 was not reverted when conducting the gel-based degradation assay at 42°C. We monitored the decrease in gel band intensity of HspR and HspRΔC9 and analyzed them using densitometry (Fig S3D).

Considering that the conserved, disordered C-terminal stretch of HspR is crucial for efficient protein turnover, we explored whether this degradation defect could be reversed by complementing HspRΔC9 in vitro with a random GS linker (GGS GSS GSG) or the C-terminal sequence region of an unrelated unfolded protein. We fused either the nine-residue GS linker or the last nine residues of the unfolded model substrate FimAt to the C-terminus of HspRΔC9 and tested these variants for Bpa-mediated proteasomal degradation (Fig 6D). However, neither HspRΔC9^GSlinker nor HspRΔC9^FimAtC9tail rescued the degradation defect of HspRΔC9. In addition to substituting the C-terminal stretch of HspR, we assessed individual residues in this region (W121, K122, R124, and R125) for their contribution to efficient substrate degradation (Fig 6D). Single- or double-point mutations of these residues (W121A,K122A, and R124A R125A) severely impaired HspR degradation. We monitored the decrease in gel band intensity of HspR and analyzed them using densitometry (Fig 6E and F). These results suggest that although the C-terminal region of HspR is not a major binding determinant, it exhibits a sequence preference that contributes to the facilitation of HspR degradation by the Bpa–CP complex. The data support a role for the C-terminal disordered region of HspR as a sequence-dependent threading determinant for Bpa-mediated proteasomal degradation.

The N-terminal HTH domain of HspR contributes to substrate recruitment for Bpa-mediated proteasomal degradation

Based on our findings indicating a negligible contribution of the C-terminal–disordered region of HspR to the affinity for Bpa, we next investigated which part of HspR is involved in the initial capture complex with Bpa before threading into the 20S CP. We designed peptides spanning the sequence of HspR and assessed the binding of these peptides to fluorescently labeled Bpa. In total, we probed eight 15-amino acid peptides covering residues 16–121 of HspR (Fig 7A and Table S3). Peptides containing either the full or partial epitope of the Bpa-interacting region were expected to exhibit binding, whereas peptides spanning regions of HspR that do not interact with Bpa were anticipated to show no binding. Given that we had previously established that the disordered N-terminal stretch of HspR (1–15) did not contribute to substrate recruitment, we excluded this sequence from our peptide screen.

To evaluate the binding interactions, we fluorescently labeled Bpa and monitored the temperature-related fluorescence intensity changes in a capillary-based titration experiment. The peptides were titrated against the fluorescently labeled Bpa, covering a concentration of 1 mM–0.49 μM. Initially, we performed a qualitative screen to identify peptides that induced a signal change (Fig 7B). Among the peptides tested, only peptide 1, which corresponds to residues 16–30 within the HTH domain of HspR, produced significant changes. We then quantified the binding of peptide 1 to Bpa by monitoring the temperature-related fluorescence intensity change. The analysis showed that peptide 1 bound Bpa with a dissociation constant of 0.33 ± 0.09 mM (Fig 7C).

To further probe which residues might be involved in the interaction between HspR and Bpa, we focused on conserved charged amino acids within the region spanning residues 16–30. The entire 15-residue stretch exhibits high conservation across actinobacteria (Fig 7D). Among these conserved residues, Glu19 and His24 were chosen as potential interaction candidates. In addition, R29, which we later demonstrate to be crucial for DNA binding, also was included. Although several hydrophobic residues are also strongly conserved, the AlphaFold structure prediction suggested their primary role in stabilizing the fold of the helix bundle. Consequently, we focused on the charged residues that were predicted to be more exposed and pointing outward, making them more likely to engage with other binding partners. We constructed mutants targeting these residues (HspR H24A, HspR E19K, HspR E19A, and HspR R29A). In case of E19, we included a mutation to a residue with opposite charge in addition to the alanine mutation.

Subjecting the HspR HTH variants of residues E19 and H24 to Bpa-mediated proteasomal degradation, we observed that mutation of H24 or E19 resulted in a notable degradation defect (Fig 7E). Unlike HspR WT, which was degraded within 5 min, it took 2 h to degraded about 90% of the HspR H24A variant. The glutamate variants exhibited an even slower degradation, without reaching complete degradation even after 8 h. To ensure that the observed defect was not due to a loss of structure, we recorded CD spectra of the HspR H24A, HspR E19A, and HspR E19K variants (Fig S4A–D), and found that they, like the wild-type protein, displayed a CD signature with the characteristic strong α-helical contribution. Although Bpa-mediated degradation was impaired, as expected, these HspR HTH variants could still be degraded by the hyperactive “open gate” proteasome (Fig S4E). These results indicate that residues in the N-terminal HTH domain of HspR play a role in recruiting the protein for Bpa-mediated proteasomal degradation through direct interaction with Bpa. Furthermore, temperature-related changes in fluorescently labeled Bpa in the capillary-based titration experiment showed that the HspR H24A variant bound Bpa less tightly than HspR WT, exhibiting a dissociation constant of 12.06 ± 4.71 μM compared with 3.56 ± 0.62 μM observed for the wild type (Fig S4F).

In contrast to the mutations in H24 or E19, the HspR R29A variant displayed no defect in Bpa-mediated proteasomal degradation, indicating that this positively charged residue does not participate in the recruitment process (Fig S5A). Interestingly, although the wild-type HspR is protected from Bpa-mediated proteasomal degradation in the presence of its cognate DNA (Figs 2A and S5A), the HTH variant R29A was still degraded by the Bpa–CP complex in the presence of DNA containing the HAIR motif (Fig S5A). We quantified the decrease in gel band intensity of HspR using densitometry (Fig S5B). Our results suggest that residue R29 is involved in the binding of HspR to the HAIR motif. Notably, the stabilization of HspR through binding to cognate DNA with the HAIR motif even confers protection against degradation by the hyperactive open-gate proteasome, 20S CP^OG. Our results suggest that an effect on overall conformational stability is responsible, rather than a local effect on the C-terminal–threading region because the experiment was conducted with the C-terminally truncated variant, HspRΔC9, which is efficiently degraded by the open-gate proteasome in absence of HAIR DNA but is rendered degradation-resistant in its presence (Figs S5C and 4B, middle gel).

The HspR co-repressor DnaK is a pupylation and proteasomal degradation substrate

Previous proteomic studies in mycobacteria have suggested that DnaK, a heat shock chaperone and the co-repressor of HspR, may be a target for pupylation (Festa et al, 2010; Poulsen et al, 2010). This implies a potential interplay between pupylation-dependent Mpa-mediated proteasomal degradation, ATP-independent Bpa-mediated proteasomal degradation, and the dnaKJgrpE-hspR operon. To demonstrate the pupylation of DnaK in vitro, we conducted pupylation assays (Fig S6A). In the presence of the PafA ligase and ATP, we observed near-complete pupylation of DnaK after 20 h. Pupylation of DnaK is less efficient compared to pupylation of the well-established pupylation substrate PanB. The presence of additional protein bands above 70 kD indicates the pupylation of multiple lysine residues in DnaK. Mass spectrometric analysis of DnaK–Pup identified pupylated lysine residues at K159, K469, K483, and K522 (source data Fig S6). Notably, K159, K469, and K483 are located in the nucleotide-binding domain, whereas K522 is found in the substrate binding domain. Next, we tested whether pupylated DnaK could serve as a proteasomal degradation substrate in vitro. We found that the Mpa–CP^OG complex could degrade DnaK–Pup in vitro, whereas the open-gate proteasome alone could not (Fig S6B). These findings provide experimental evidence that DnaK is a pupylation and proteasomal degradation substrate.

Generation of the bpa deletion strain in M. smegmatis

Deletion of bpa in M. tuberculosis (Jastrab et al, 2015) was previously shown to affect the expression of essential downstream genes. Because bpa (MSMEG_6365) in M. smegmatis is located upstream of the orthologous genes (MSMEG_6366, MSMEG_6367, MSMEG_6369) and they were shown to be essential in M. smegmatis (Dragset et al, 2019), we followed an analogous strategy for knockout construction, first relocating the downstream genes before generation of the knockout (Parish & Brown, 2008). We constructed an MC²155 SMR5 strain carrying MSMEG_6366/6367/6369 on kanamycin resistant integrative plasmid (pTTP1B, Addgene #91723) under the natural promoter. This strain was then used as the parental strain to further knockout bpa along with MSMEG_6366/6367/6369 from their original genomic position. Up- and down-stream regions (1,500 base pairs each) of MSMEG_6365–6369, amplified from MC²155 SMR5 genomic DNA, was cloned into a hygromycin-resistant suicide plasmid (pGOAL19, Addgene #20188). 1 μg of suicide plasmid was treated with UV light for 1 min before transformation into strain MC²155 SMR5:MSMEG_6366/6367/6369 and plating onto 7H10 agar supplemented with 50 μg/ml hygromycin at 37°C. Single crossovers (SCOs), identified as blue colonies by X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) underlay and overnight incubation at 37°C, were cultured in 7H9 at 37°C supplemented with 50 μg/ml hygromycin and grown to an OD600 of 0.7. Serial dilutions of 10⁰, 10⁻¹, and 10⁻² were plated onto 7H10 supplemented with 2% (wt/vol) sucrose and incubated at 37°C until colonies grew. Double Crossovers (DCOs) were identified as white colonies after X-gal underlay and overnight incubation at 37°C and were screened for bpa deletion by colony PCR. DCOs were restreaked onto 7H10 agar and single colonies of DCOs were further tested for absence of bpa by amplification of the genomic locus. DCOs that did not contain bpa (now confirmed bpa deletion strains) were grown in liquid 7H9 media to make glycerol stocks. The absence of bpa was also verified by Western blotting. One single DCO was used for this study and is referred to as the M. smegmatis bpa deletion strain. The bpa gene together with its natural promoter was amplified from MC²155 SMR5 genomic DNA and cloned into an apramycin resistant integrative vector (pFlag_attp, which was a gift from Markus Seeger, Addgene # 110095) to generate the complementation vector. The complemented strain was produced by co-transformation of this plasmid and a plasmid containing the integrase (pMA_int, which was a gift from Markus Seeger, Addgene # 110096) into competent bpa knockout M. smegmatis cells and plating onto 7H10 supplemented with 25 μg/ml kanamycin and 50 μg/ml apramycin. From a single verified colony, a glycerol stock of the bpa-complemented knockout strain was prepared.

Bacterial strains and growth experiments

Bacterial strains used in this study were M. smegmatis MC²155 SMR5 (WT), M. smegmatis MC²155 SMR5 bpa deletion strain constructed as described above (Δbpa) and the bpa-complemented M. smegmatis MC2155 SMR5 bpa deletion strain (Δbpa:bpa). Bacterial cultures (WT, Δbpa, and Δbpa:bpa) were grown in M9 minimal medium, supplemented with 2 mM MgSO₄, 0.2% (vol/vol) glycerol, and 0.05% (wt/vol) Tween-80. For the heat stress growth experiment, strains were grown in triplicates in 30 ml cultures at 37°C to logarithmic phase and then diluted to an optical density at 600 nm of 0.01. Cells were further grown at 37°C under standard conditions or 45°C for heat shock. Optical density was measured at 600 nm at the given time points and plotted.

For oxidative stress experiments, strains were grown in triplicates in 30 ml cultures in the same minimal medium as mentioned above to logarithmic phase and then diluted to an optical density of 0.1. Cells were further grown under standard conditions or supplemented with 5 mM of H₂O₂ for oxidative stress. Optical density was measured at 600 nm at the given time points and plotted.

Proteomics

25 ml bacterial culture were grown in 7H9 medium (Difco). Strains were grown at 37°C under standard conditions or stressed for 3 h at 45°C for heat stress. Cells were harvested and lysed by mechanical bead-beating in 20 mM HEPES–NaOH, pH 7.5, 150 mM KCl, 2 mM EDTA. Lysates were cleared by spinning for 5 min at 16,000g. Protein concentration of the lysates was determined by bicinchoninic acid assay (BCA Protein Assay Kit; Thermo Fisher Scientific) and diluted to 1 mg/ml. 50 μl of diluted lysate was mixed with sodium deoxycholate (DOC) to a final concentration of 5% (wt/vol). Proteins were reduced by incubation with 1,4-dithiothreitol (final concentration 12 mM) for 30 min at 37°C and alkylated by incubation with iodoacetamide (final concentration of 40 mM) for 45 min at room temperature in the dark. Samples were diluted with 0.1 M ammonium bicarbonate to a final concentration of 1% (wt/vol) DOC. Proteins were digested overnight with lysyl endopeptidase (Wako Chemicals) and sequencing grade porcine trypsin (Promega) at an enzyme:substrate ratio 1:100 at 37°C with constant shaking (800 rpm). The digestion was stopped by the addition of formic acid to a final concentration of 1% (vol/vol) (pH <3). Precipitated DOC was filtered out by centrifugation at 800g (Eppendorf rotor S-4x1000) for 5 min with a 0.2 μm PVDF membrane filter (Corning FiltrEX 96-well white filter plate). The peptide mixtures were loaded onto 96-well elution plates (Waters), desalted, and eluted with 80% (vol/vol) acetonitrile, 0.1% (vol/vol) formic acid. After elution, peptides were dried in a vacuum centrifuge, resolubilized in 0.1% (vol/vol) formic acid to a final concentration of 1 mg/ml and analyzed by mass spectrometry.

Data were acquired on an Orbitrap Eclipse Tribrid mass spectrometer in data-independent mode (DIA). The samples were separated using easy LC system by a 120 min linear gradient at a flow rate of 300 nl/min with increasing buffer B (95% (vol/vol) acetonitrile in 0.1% (vol/vol) formic acid) from 3% to 30% (vol/vol) on a 40 cm × 0.75 mm i.d. column (New Objective, PF360-75-10-N-5) packed in house with 1.9 μm C18 beads (Dr. Maisch Reprosil-Pur 120). The column was heated to 50°C. For DIA, a full-MS1 scan was acquired between 350 and 1,100 m/z at a resolution of 120,000 with an AGC target of 100%. Forty-one variable-width windows were used to measure fragmented precursor ions. DIA-MS2 spectra were acquired at a resolution of 30,000 and an AGC target of 400%. The first mass was fixed at 200 m/z and the normalized collision energy was set to 28. To maximize parallelization, a duty cycle time was 3 s.

The data were searched in Spectronaut version 14.11 (Biognosys), using a directDIA approach. The spectral library was created based on a pulsar search using the default setting and trypsin digestion rule. The data were searched against the UniProt FASTA database (M. smegmatis, strain ATCC 70084/mc(s)115, UP000000757, November 2018). The targeted data extraction was performed in Spectronaut version 14.11 with default settings except for the machine learning which was set to “across experiment” and the data filtering which was set to “Qvalue” (adjusted P-value). The FDR was set to 1% on peptide and protein level. Differential analysis on protein level was performed with Protti (Quast et al, 2022) package, using moderated t test and correcting for multiple hypothesis testing using Benjamini–Hochberg correction (Benjamini & Hochberg, 1995). Post processing and data visualisation were performed in R.

Multiple sequence alignment and structural modelling

HspR amino acid sequences from several actinobacterial species were compiled and aligned in ClustalW (Larkin et al, 2007; Madeira et al, 2019). The multiple sequence alignment was visualized in Jalview (Waterhouse et al, 2009) and conservation was calculated by amino acid identity between residues. A structural model of HspR was predicted in AlphaFold (Jumper et al, 2021; Varadi et al, 2022) using the full-length amino acid sequence from M. tuberculosis HspR. This structural model was used in corroboration with information gained from the PASTA 2.0 prediction software (Walsh et al, 2014) on regions of disorder, to construct a domain organization model of M. tuberculosis HspR.

Cloning, expression, protein purification, and DNA probe generation

All genes for constructs for heterologous expression in E. coli were amplified by PCR from genomic DNA from either M. tuberculosis H37Rv or M. smegmatis MC²155 SMR5 with Q5 polymerase (New England Biolabs). For protein expression, coding sequences were cloned into a pET28a vector under a T7 promoter by Gibson assembly. Proteasome constructs were cloned into a pETDuet vector under a T7 promoter to allow for co-expression of the α and β subunits, with a Strep-Tactin tag on the C-terminus of the β subunit. All truncations and fusions were either generated by Gibson Assembly (New England Biolabs) or blunt end-ligation using T4 ligase (New England Biolabs). HspRΔN15, HspR E19A, HspR E19K, and HspR C-terminal fusions were fused to a histidine tag containing maltose-binding protein (His₆MBP) at the N-terminus. All constructs were transformed into E. coli Rosetta cells for overexpression and induced using an autoinduction media system (Studier 2005). Cells were lysed with a high-pressure homogenizer (Microfluidizer M110-L, Microfluidics).

His₆MBP-HspR variants, His₆MBP-HspR fusions, His₆Bpa, and His₆DnaK were purified by Ni-NTA affinity chromatography. Proteasome variants were isolated by StrepTactinXT affinity chromatography. Bpa and proteasome variants were applied directly onto a Superdex 200 or Sepharose 6 size exclusion chromatography column, respectively, and purified in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 1 mM EDTA, 10% (vol/vol) glycerol. The other proteins were first cleaved with Tobacco Etch Virus endopeptidase (TEV protease) and further purified by reapplication to Ni-NTA affinity chromatography to remove the His₆ tag and/or application to amylose chromatography to remove the His₆–MBP fusion from the protein of interest. HspR variants were applied onto a heparin column and run in a linear gradient from 50 mM HEPES–NaOH, pH 7.5, 0 M–2 M NaCl, 5 mM MgCl₂, 5% (vol/vol) glycerol to isolate the protein of interest. Finally, proteins were then applied onto a Superdex 75 size exclusion chromatography column and purified in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 5% (vol/vol) glycerol.

FimAt fusion-variants, HspR WT, HspRΔC9, and HspR H24A were isolated from inclusion bodies under denaturing conditions. Cells were resuspended and lysed as mentioned above and then added to 0.8 x volume of Triton buffer (60 mM EDTA 1.5 M NaCl, adjusted to pH 7.0, 6% [vol/vol] Triton X-100), stirring at 4°C for 30 min. Inclusion bodies were harvested by centrifugation (20,000 rpm for 45 min) and washed five times in 100 mM Tris–HCl, pH 8.0, 20 mM EDTA. Inclusion bodies were solubilized in 5 ml of buffer (50 mM Tris, 6 M guanidinium chloride) per gram of inclusion bodies by stirring for 2 h at room temperature. Solubilized inclusion bodies were spun at 100g for 30 min at 20°C. The supernatant containing solubilized protein was refolded by dialysis against 50 mM Tris–HCl, pH 7.5, 5 mM MgCl₂, 0.5 M arginine, 10% (vol/vol) glycerol. Refolded proteins were further purified by ion exchange chromatography (50 mM HEPES–NaOH, pH 7.5, 50 mM to 1 M NaCl, 5 mM MgCl₂, 5% [vol/vol] glycerol) and gel filtration over a Superdex 75 size exclusion column in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 5% (vol/vol) glycerol. Mpa, PafA, PanB, and Pup were purified as described previously (Laederach et al, 2019) and purified FimAa and FimAt were kindly gifted by the Glockshuber laboratory. The HAIR motif was generated as previously described (Parijat & Batra, 2015). DNA was produced was by annealing 100 μM of forward (fw) and reverse (rv) primers at 95°C in 70 cycles, decreasing 1°C/cycle until 25°C. The primers were the following: HAIR motif oligomers (fw 5′-CCG TCG AGG CAA GCT TGA GCG GGG TGC ACT CAT CAT AGT GCA GGA AAG AA-3′); (rv 5′-TTC TTT CCT GCA CTA TGA TGA GTG CAC CCC GCT CAA GCT TGC CTC GAC GG-3′).

RNaseA alkylation

To permanently unfold model substrate ribonuclease A (RNaseA), lyophilized protein from bovine pancreas was purchased (Sigma-Aldrich) and dissolved at a concentration of 3 mg/ml in degassed denaturing buffer (6 M guanidinium chloride, 10 mM Na₂EDTA, 200 mM Tris–HCl, pH 8.6, 1 mM DTT) at room temperature for 30 min. In the dark, 300 μl reaction, containing 200 μl of the 3 mg/ml denatured and reduced RNaseA was incubated with 17 mM iodoacetamide (Sigma-Aldrich) in denaturing buffer at room temperature for 30 min. The denaturing buffer was exchanged by dialysis or by application over a PD10-desalting column (GE Healthcare) with 20 mM sodium phosphate buffer (pH 8.0). As cysteine alkylation with iodoacetamide results in addition of a carbamidomethyl group (57 D), full alkylation of all eight cysteine residues was confirmed by intact mass spectrometry.

Circular dichroism and thermal transitions

To analyze protein fold and stability, we used circular dichroism (CD) to measure secondary structure content. Model substrates were dialyzed into 20 mM sodium phosphate buffer (pH 8.0) before CD measurements. For thermal transitions, HspR variants were dialyzed into 20 mM sodium/potassium buffer, pH 7.5, 150 mM NaCl, 1 mM MgCl₂. CD spectra were recorded by measuring ellipticity (ϴ) in a 0.1 cm quartz cuvette (Hellma Analytics) placed in a Jasco J-710 spectropolarimeter (Brechbühler). For CD spectra, the measured ellipticity (ϴ) from five acquisitions was normalized to mean molar ellipticity in degree cm² dmol⁻¹:MRW=ϴ *100*kDc * d * Nwhere N is the number of amino acid residues, d is the cuvette pathlength in cm, and c is the protein concentration in mg/ml.

Thermal transitions by CD spectroscopy for HspR and variants were measured at 222 nm from 20°C to 90°C. The recorded ellipticity was first fitted with the following equation for a one-state thermal transition:Sobs=[(Sn∘+mn∙T)+(Su∘+mu∙T) exp(∆HmR∙Tm)(T−TmTm)][1+exp(∆HmR∙Tm)(T−TmTm)]where S_obs is the observed ellipticity, Su∘ is ellipticity of unfolded protein extrapolated to 0 K, m_u is the slope of the unfolded protein ellipticity, Sn∘ is the ellipticity of native protein extrapolated to 0 K, m_n is the slope of the native protein ellipticity, and T is the temperature in Kelvin, T_m is the melting temperature, ΔH_m is the enthalpy difference between native and unfolded protein at T_m, R is the gas constant, and T is the temperature in K. For thermal transitions, experiments were performed in independent triplicates and one representative transition is shown.

Next, ellipticity was converted to the fraction of folded HspR (f_N) according to the following equation taking into account the linear temperature dependencies of pre- and post-transition baselines, using the parameters determined by fitting with the equation mentioned above:fN=Sobs−(Su∘+mu∙T)(Sn∘+mn∙T)−(Su∘+mu∙T)where S_obs is the observed ellipticity, Su∘ is ellipticity of unfolded protein extrapolated to 0 K, m_u is the slope of the unfolded protein ellipticity, Sn∘ is the ellipticity of native protein extrapolated to 0 K, m_n is the slope of the native protein ellipticity, and T is the temperature in Kelvin.

Proteasomal degradation assays

4 μM substrate was incubated with 12 μM Bpa protomer or 200 μM hexapeptide (GTGQYL) and 0.4 μM assembled proteasome core particle (20S CP^WT or 20S CP^OG) in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTT, and time points were taken at the indicated intervals. As a negative control, Bpa was omitted from the reaction and as a positive control substrate was incubated with 20S CP^OG in absence of Bpa. To quantify degradation, band intensity was measured from triplicate gels via densitometry (GelAnalyzer 19.1) and plotted against time. Reactions were incubated at 37°C and time points were taken at the indicated intervals. To investigate the temperature dependence of certain degradation defects, reactions were, in addition to 37°C, incubated at 30°C and 42°C and indicated as such in figures.

To test if HspR is a Bpa-specific substrate, 3 μM Mpa_6, or Bpa₁₂ were incubated with 0.4 μM 20S CP^OG in the presence of 2.5 mM ATP for 30 min at room temperature to assemble the Bpa– or Mpa–proteasome complexes. After the pre-incubation, 2.5 mM additional ATP was added, and the reactions were started by adding 4 μM substrate. Reactions were conducted in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTT. Reactions were incubated at 37°C and time points were taken at the indicated intervals.

To measure the initial velocity of HspR degradation, we conducted the Bpa-mediated proteasomal degradation assay at closer time intervals in the early phase of degradation (0–12 min for HspR WT and 0–2 h for HspRΔC). Speed of degradation was assessed densitometrically by measuring band intensity decrease of HspR over time (GelAnalyzer 19.1). Reactions were incubated at 37°C, and time points were taken at the indicated intervals.

To determine if pupylated DnaK from M. tuberculosis is a proteasomal degradation substrate, 2 μM DnaK–Pup was incubated with 0.2 μM 20S CP^OG, 0.2 μM Mpa₆ in 50 mM HEPES–KOH pH 7.5, 150 mM NaCl, 10 mM KCl, 20 mM MgCl₂, 10% (vol/vol) glycerol, 10 mM ATP, 1 mM DTT. Reactions were incubated at 37°C and time points were taken at the indicated intervals.

For all degradation assays, time points were quenched in Laemmli buffer and reactions were visualized by running samples on SDS–PAGE. For all gel-based degradation assays, reactions were conducted in independent triplicates and one representative gel is shown.

Pupylation assays

6 μM of substrate was incubated with 12 μM Pup and 1 μM PafA in buffer P (50 mM HEPES–KOH, pH 8.0, 150 µm NaCl, 20 mM MgCl₂, 10% [vol/vol] glycerol, 1 mM DTT, 5 mM ATP). Reactions were incubated at 30 °C and time points were taken at the indicated intervals. Time points were quenched in Laemmli buffer and reactions were visualized by running sub samples on SDS–PAGE. For all gel-based pupylation assays, reactions were conducted in independent triplicates and one representative gel is shown.

Pupylated DnaK from M. tuberculosis was produced in batch by incubating 35 μM DnaK with 40 μM Pup, and 1 μM PafA in buffer P for 20 h at 30 °C. Pupylated DnaK was separated from other proteins via gel filtration on a Superdex 200 column.

Substrate-binding assays

Affinity of substrate for Bpa was investigated by measuring the temperature-related fluorescence intensity changes in a Monolith NT.115 from NanoTemper. First, Bpa lysine residues were labeled with a 20-fold excess of fluorescein. Bpa and dye were incubated for 1 h at room temperature, after which the mixture was spun at 14,000 rpm for 10 min at 4°C. Excess fluorescein was separated from labeled Bpa (Bpa-Fl) by gel filtration over a Superdex 200 analytical size exclusion column in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 10% (vol/vol) glycerol.

For binding experiments, 200 nM Bpa-Fl protomer was titrated with serial dilutions of HspRΔC9, HspR WT, or HspR H24A at indicated concentrations in 50 mM HEPES–NaOH, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 5% (vol/vol) glycerol, 0.01% (vol/vol) Tween-20. Binding of HspR WT to Bpa was either measured with HspR WT alone or in a 1:1 complex with DnaK or the HAIR operator DNA. Reactions were pre-incubated 10 min at room temperature and absorbed into premium capillary tubes (NanoTemper). The measurements on the Monolith NT.115 were performed at 20°C with 20% blue LED power and 20%, 40%, and 80% MST power, which corresponded to a temperature jump of 2–8°C. For each substrate, the binding titration was conducted at least in triplicate, and scans were compiled and analyzed with the NTAffinity Analysis v2.0.2 software using the T-jump model to calculate dissociation constants based on the Kd model (Jerabek-Willemsen et al, 2014). Error bars denote the SD.

Peptide-binding array

To identify regions of M. tuberculosis HspR contributing to Bpa binding, we designed peptides of 15 amino acids in length that covered the sequence of HspR excluding the N- and C- terminal stretches (Table S3). In total, eight peptides were ordered (Genescript) and dissolved in buffer M (50 mM HEPES–KOH, pH 7.5, 150 mM NaCl, 5 mM MgCl2, 5% [vol/vol] glycerol, 0.1% [wt/vol] pluronic 127, 2% [vol/vol] DMSO). Bpa lysine residues were labeled with threefold excess RED-NHS second generation dye using the RED-NHS second generation kit according to manufacturer guidelines (NanoTemper).

Binding experiments were conducted on a Monolith NT.115 instrument (Nanotemper). 250 nM Bpa-RED protomer was titrated with serial dilutions of 2 mM–30 nM peptide 1–7 in buffer M. Reactions were pre-incubated 10 min at room temperature and aspirated into premium capillary tubes (NanoTemper). The measurements were performed on a Monolith NT.115 (NanoTemper) at 25°C with 20% red LED power and 40% MST power. Scans were compiled and analyzed with the NTAffinity Analysis v2.0.2 software using the T-jump model to calculate dissociation constants based on the Kd model (Jerabek-Willemsen et al, 2014).