Glutamylation of Bacterial Ubiquitin Ligases by a Legionella Pseudokinase

Abstract

Legionella pneumophila encodes a family of phosphoribosyl ubiquitination ligases (SidE) essential for the bacterium to establish successful infection. Four independent studies now show that the SidE family of ubiquitin ligases are regulated by a novel mechanism of glutamylation via a pseudokinase-like Legionella effector, SidJ, in an ATP- and calmodulin-dependent manner.

Legionella pneumophila is a Gram-negative intracellular pathogen of freshwater amoebas. However, inhalation of Legionella-contaminated aerosols can cause infection of human alveolar macrophages and lead to a severe form of pneumonia known as Legionnaires’ disease. During infection, Legionella secretes more than 300 effector proteins via the Dot/Icm type IV secretion system to gain control over multiple host cellular processes, thereby converting the phagosome into the Legionella-containing vacuole (LCV), a specialized membrane-bound organelle amenable for its intracellular proliferation. One host cellular pathway subverted by Legionella is the ubiquitination pathway. So far, more than ten Legionella effectors have been identified as ubiquitin E3 ligases, including proteins containing the conserved For U-box domain and others akin to HECT-type E3 ligases.

In addition to these ubiquitin ligases, which utilize the canonical host ubiquitin-activating enzyme E1 and ubiquitin-conjugating enzyme E2, recent studies have identified the Legionella effector SidE and three homologs (SdeA, SdeB, and SdeC) as a family of unconventional ubiquitin ligases. These ligases act independently of ATP or E1 and E2 enzymes, but require a molecule of NAD⁺ [1–3]. This novel type of ubiquitination, catalyzed by the SidE family ligases, is described as phosphoribosyl ubiquitination (PR-ubiquitination). In the reaction, SidE first consumes a NAD⁺ molecule to generate an ADP-ribosylated ubiquitin (ADPR-Ub) by its mono-ADP-ribosyl transferase (mART) domain. Then, the phosphodiesterase (PDE) domain of SidE completes the PR-ubiquitination reaction by conjugating the phosphoribosyl-ubiquitin (PR-Ub) group from ADPR-Ub to a serine residue in targets. PR-ubiquitination is essential for Legionella optimal intracellular growth as the mutant strain carrying the deletion of all four SidE family members exhibited substantial growth defects in its amoeba host [4]. In the canonical ubiquitination pathway, ubiquitin can be cleaved from ubiquitinated substrates by deubiquitinases (DUBs). Similarly, PR-ubiquitination can also be reversed by Legionella PR-Ub-specific DUBs, which remove the PR-Ub moiety from PR-ubiquitinated substrates [5] (Figure 1). Furthermore, PR-ubiquitination is spatiotemporally regulated during Legionella infection. SidE proteins function at the LCV during early infection but are delocalized at a later stage. This has been attributed to the translocation of a Legionella metaeffector, SidJ [4]. SidJ is also able to suppress toxicity induced by overproduction of SidE family proteins in both yeast and mammalian cells [4,6], suggesting that SidJ may directly inhibit the PR-ubiquitination activity of SidE family ligases. However, the molecular mechanism underlying the regulation of SidE family ligases by SidJ had remained elusive.

Figure 1. The Phosphoribosyl Ubiquitination (PR-ubiquitination) Pathway and Its Regulation by SidJ.

The SidE family effectors catalyze PR-ubiquitination using a molecule of NAD⁺. DupA and DupB are two PR-Ub-specific deubiquitinases (DUBs) that cleave the phosphoribosyl-ubiquitin (PR-Ub) moiety from PR-ubiquitinated substrates. Upon activation by calmodulin (CaM), the metaeffector SidJ catalyzes the glutamylation of SidE family ligases and hence inhibits the PR-ubiquitination activity of SidE. An unknown effector may exist in Legionella to revive SidE family ligases by removing glutamate modification from the ligases. Abbreviations: GLU, glutamate; NAM, nicotinamide; PP_i, pyrophosphate.

Four recent independent studies now reveal that SidJ catalyzes an unusual type of post-translational modification to a key catalytic residue of the SidE family ligases in a calmodulin (CaM)-dependent manner [7–10]. These four papers describe either the X-ray crystal or the cryo- electron microscopy (EM) structure of SidJ in complex with CaM, with SidJ binding to CaM via a C terminal IQ motif-containing alpha helix. The structures also reveal that SidJ contains a bilobed kinase-like domain that retains a majority of the characteristic kinase catalytic motifs. Surprisingly, no phosphorylation events could be detected using conventional in vitro kinase assays with individually expressed and purified proteins. However, in a recent paper in Science [7], Black et al. noted that coexpression of SidJ, SdeA, and CaM in Escherichia coli resulted in a mobility shift of SdeA on SDS-PAGE. The authors further revealed that the total PR-ubiquitination signals were drastically reduced in cells expressing both SidJ and SdeA. In the Nature paper by Bhogaraju et al., the authors found that coexpression of SidJ and SdeA in mammalian cells resulted in inhibition of ubiquitin modification by SdeA [8]. In another Nature paper, Gan et al. observed that purified SdeA from mammalian cells cotransfected with SdeA and SidJ failed to PR-ubiquitinate its substrate Rab33b [9]. Each of these findings suggests that SidJ is directly modifying SdeA. Indeed, mass spectrometry analyses performed by all four groups revealed that SdeA is polyglutamylated at the mART catalytic residue E860 in a SidJ-dependent manner. All four groups further demonstrated that, as a direct consequence of this modification, ADP-ribosylation of ubiquitin catalyzed by the SdeA mART domain is obstructed, and thus the initiation of PR-ubiquitination is blocked (Figure 1).

The discovery of SidJ as a CaM-activated glutamylase raises several intriguing questions. First, how is CaM able to activate SidJ? In the paper by Sulpizio et al., the authors propose that CaM-binding may stabilize a loop, which is equivalent to the activation loop in protein kinases, in an active conformation via a network of interactions involving the CaM N-lobe. Consistent with this idea, deletion of a N terminal peptide from SidJ, which directly mediates the contacts with the CaM N-lobe, substantially impairs SidJ activity [10]. Second, the crystal structures reported by Black et al. and Sulpizio et al. revealed a migrated nucleotide-binding pocket bound with an AMP molecule. This pocket was postulated by Black et al. as a catalytic site mediating the second step of the glutamylation reaction, while Sulpizio et al. showed evidence in support of an allosteric regulatory role for this pocket. To ascertain the exact mechanism of this migrated nucleotide-binding pocket requires further examination. Another important question is whether SidJ also modifies host targets during Legionella infection. Bhogaraju et al. found that glutamylation signals remained on the LCV when cells were infected by a Legionella mutant strain lacking all four sidE family members, indicating that SidJ may target additional proteins for glutamylation. Using a mass spectrometry approach, several host proteins were identified as potential glutamylation targets by SidJ [8]. Forthcoming studies of these host targets of SidJ will likely provide insights into the functions of SidJ beyond the regulation of SidE family enzymes.

Remarkably, these four recent papers on SidJ have unveiled an archetypal example of a pseudokinase-like bacterial effector protein catalyzing protein glutamylation in the context of PR-ubiquitination pathway regulation. Investigations are needed to elucidate how the preference for glutamate by SidJ is achieved and how the substrate specificity is determined by SidJ to selectively modify residue E860 of SdeA. Furthermore, based on multiple examples where Legionella employs effectors to regulate or reverse the activity of other effectors, it would not be surprising if this pathogen uses yet another effector to counteract SidJ by removing glutamate modification from SidE ligases and/or host proteins.