Ubiquitination by a Mycobacterium protein that mimics E1 and E3 activities

Mycobacterium tuberculosis (Mtb) has evolved various strategies to co‐opt the host ubiquitin network to facilitate its proliferation. In the current issue of EMBO Reports, Liu and colleagues (Wang et al, 2021) demonstrate that the Mtb kinase PknG catalyzes ubiquitination by an unprecedented mechanism wherein the reaction starts by ATP hydrolysis occurring at the α‐phosphate position, leading to covalent attachment of the modifier to Lys82 of the E2 conjugation enzyme UbcH7. Ubiquitin is then delivered to host proteins important for immunity by a putative peptidase activity also embedded in PknG. This novel activity of PknG expands our understanding of protein ubiquitination mechanisms, which may be harnessed to identify potential therapeutics for fighting Mtb infection.

Mycobacterium tuberculosis has evolved various strategies to hijack the host ubiquitin network. A study in this issue now shows that the Mtb kinase PknG catalyzes ubiquitination by an unprecedented mechanism thereby inhibiting host innate responses.

Ubiquitination is a posttranslational modification which mostly is a three‐reaction cascade catalyzed by E1, E2, and E3 enzyme, respectively. Whereas the function of E1 is to activate ubiquitin (Ub) at the expense of ATP to produce a Ub‐AMP adduct, E2s mostly function as a middleman which dictates the type of ubiquitin chain to assemble, and E3s play the role in substrate selection (Hershko et al, 2000) (Fig 1). Ubiquitinated proteins undergo rapid changes in their abundance, activity, or cellular localization, which allow cells to effectively respond to intrinsic developmental cues or to insults elicited by infection or abiotic stress. Ubiquitination can be reversed by deubiquitinases to balance the signaling (Clague et al, 2012).

Figure 1. The biochemical basis of PknG‐induced ubiquitination and its comparison to canonical and phosphoribosyl ubiquitination.

(A) PknG catalyzes a two‐step ubiquitination reaction which starts by binding to UbcH7 and transferring Ub to the E2 to form UbcH7K82‐Ub in a process energized by ATP. Second, PknG catalyzes the transfer of Ub from UbcH7K82‐Ub to protein substrates, which requires a putative peptidase activity. (B) Canonical ubiquitination starts by the production of the E1‐Ub complex via the reaction intermediate AMP‐Ub at the expense of ATP. Ub is then transferred to a catalytic Cys residue in the E2 enzyme. The reaction is completed by an E3 ubiquitin ligase which selects substrate proteins for ubiquitination by various mechanisms. (C) Phosphoribosyl ubiquitination by SidEs starts by ADP‐ribosylation to produce the reaction intermediate ADPR‐Ub. The second step involves the cleavage of the phosphodiester bond in ADPR‐Ub by the phosphodiesterase (PDE) activity and the transfer of phosphoribosyl Ub to serine residues on substrate proteins.

Many infectious agents hijack host ubiquitination to create a niche conductive for their proliferation. Among these, a large cohort of bacterial virulence factors function as E3 ubiquitin ligases, which by cooperating with E1 and E2 enzymes, target‐specific host processes (Zhou & Zhu, 2015). Similarly, some virulence factors function as deubiquitinases (Zhou & Zhu, 2015) and others directly target ubiquitin itself or enzymes involved in ubiquitination (Song & Luo, 2019).

Mycobacterium tuberculosis (Mtb) has evolved sophisticated mechanisms to evade host immunity, including the delivery of virulence factors such as the Ser/Thr protein kinase PknG into the cytosol of infected macrophages (Walburger et al, 2004; Chai et al, 2020). A ∆pknG mutant fails to prevent the fusion of its phagosome with the lysosome, leading to premature death of engulfed bacteria in macrophages (Walburger et al, 2004). PknG appears to phosphorylate proteins of both bacterial and host origins (Baros et al, 2020), but the role of such modification in Mtb virulence largely remains undefined.

Using yeast two‐hybrid screening, Wang and co‐workers have now identified the ubiquitin conjugation enzyme UbcH7 (an E2) as a specific binding partner for PknG, and such binding requires residue Glu190 localized within a β‐grasp fold‐like region (Wang et al, 2021). The PknG_E190A mutant retains the kinase activity but can no longer bind the E2 enzyme. In subsequent experiments aiming at determining the potential role of PknG in ubiquitination, the authors unexpectedly found that in reactions without E1 enzyme, PknG induces an ATP‐dependent conjugation of Ub to UbcH7 (Fig 1A), which suggests that PknG functions as an E1 equivalent. The PknG_E190A mutant has lost the Ub transfer activity, indicating that the ability to interact with UbcH7 but not the kinase activity is essential for PknG‐catalyzed Ub transfer. In addition, residue Pro57 in PknG was found to be critical for the transfer of Ub to UbcH7.

There are at least two repercussions associated with PknG‐induced Ub conjugation to UbcH7. First, differing from E2 charging by canonical E1 which hydrolyzes ATP at the α‐phosphate position to form an unstable AMP‐Ub adduct, PknG hydrolyzes ATP at the γ‐phosphate akin to a typical kinase. Second, instead of Cys86, the active site in canonical ubiquitination, the Ub recipient site in UbcH7 for PknG‐induced transfer is Lys82, which forms an isopeptide bond with the carboxyl end of Ub to produce UbcH7K82‐Ub (Fig 1A). Consistently, the UbcH7_C86A mutant lacking the active Cys still functions in PknG‐induced reactions. As expectedly, UbcH7K82‐Ub cannot receive additional Ub from canonical E1, most likely due to steric interference.

Another twist in reactions catalyzed by PknG is its role in the transfer of Ub from UbcH7K82‐Ub to host proteins. In this step, PknG appears to function as a peptidase that cleaves the isopeptide bond between Ub and UbcH7 in UbcH7K82‐Ub and concomitantly induces the formation of K‐48 type polyubiquitin chains on TRAF2 and TAK1, two proteins in NF‐κB signaling (Fig 1A). Interestingly, the transfer of Ub from UbcH7K82‐Ub to its substrates by PknG requires Ile87. The observation that UbcH7K82‐Ub can be used to ubiquitinate TRAF2 and TAK1 by PknG_P57A, which is defective in the E2 charging activity, clearly established that ubiquitination catalyzed by PknG is a two‐step reaction. That the PknG_E190A mutant defective in binding UbcH7 fails to catalyze such transfer indicates that interaction with UbcH7 is important for the E3‐like activity.

The E1‐ and E3‐independent ubiquitination by PknG is reminiscent of phosphoribosyl ubiquitination catalyzed by SidEs from Legionella pneumophila. In this scenario, Ub is activated by ADP‐ribosylation to form the reaction intermediate ADPR‐Ub, which is used by a second enzymatic activity to modify target proteins (Qiu et al, 2016) (Fig 1C). Although the biochemical mechanisms differ, in each case, only one or no host protein is required to complete protein ubiquitination, which highlights the high efficiency of bacterial virulence factors in modulating host signaling.

Although these unusual activities associated with PknG are highly intriguing, a few important questions concerning the biochemical basis of the catalysis remain. First, how is ATP used in the reaction that links Ub to UbcH7? Is it involved in Ub activation or in making the side chain of Lys82 suitable for receiving the modifier? Structural analysis of PknG‐UbcH7 complexes will be instrumental in answering these questions. Second, what is the catalytic mechanism for the cleavage of the isopeptide bond in UbcH7K82‐Ub by PknG? Due to the property of its side chain, Ile87 is more likely to play a supporting role by providing the structural scaffold for the catalysis. Finally, in reactions without TRAF2 or TAK1, PknG does not cleave Ub from UbcH7K82‐Ub. How the substrates trigger the cleavage is another important question.

PknG has been long known to be required for Mtb virulence (Walburger et al, 2004). Its kinase activity has been explored to identify small molecule inhibitors (Scherr et al, 2007). Because the ubiquitin ligase activity plays a more important role in Mtb virulence than the kinase activity (Wang et al, 2021), inhibitors that target the ligase activity thus hold a great premise as therapeutics against Mtb infection. Given the uniqueness of the two biochemical reactions in PknG‐mediated ubiquitination, compounds that attack these activities may have fewer off targets in humans. Clearly, such compounds may be combined with kinase inhibitors to more completely abrogate the function of PknG for better therapeutic effects. In sum, the discovery made by Wang and co‐workers represents an outstanding example that not only extends our appreciation of novel mechanisms in enzymology, but also provides the platform for the search of cures for a disease that poses grave threats to public health around the world.

EMBO reports (2021) 22: e53006.