A protein in the pathogenic bacterium Legionella pneumophila has been found to attach the modifying molecule ubiquitin to human proteins, using a mechanism that, surprisingly, does not involve cellular E1 and E2 enzymes. See Letter p.120
Ubiquitin is a polypeptide of 76 amino acids that, when covalently attached to substrate proteins, results in either modulation of the protein's function or its destruction by the cell's proteasome machinery. Since its discovery in the late 1970s, conjugation of ubiquitin to substrate proteins has been shown to have an essential role in controlling almost all cellular processes, including cell division, DNA repair and protein synthesis1. The mechanism of ubiquitination is universally conserved from yeast to humans and typically proceeds through a three-enzyme cascade. Yet in this issue, Qiu et al.2 (page 120) report that the bacterial protein SdeA ubiquitinates several human Rab proteins without engaging any of this cellular ubiquitination machinery.
During standard cellular ubiquitination (Fig. 1a), the ubiquitin-activating enzyme (E1) activates the carboxy terminus of ubiquitin in a process that costs one ATP molecule (the cellular energy 'currency'). The activated ubiquitin is then transferred from E1 to the ubiquitin-conjugating enzyme (E2). Finally, the ubiquitin ligase enzyme (E3) catalyses the transfer of ubiquitin from E2 to lysine amino-acid residues in the substrate protein, with or without an intermediary step of E3 self-modification3,4.
Bacteria do not possess this ubiquitination system, but some pathogenic bacteria have evolved toxic proteins (effectors) that resemble members of the system, which they use to modulate host-cell processes to facilitate their intracellular survival and multiplication5. The pathogenic bacterium Legionella pneumophila uses about 10% of its genome (about 300 genes) to encode effectors that help it to divide and evade host-defence mechanisms6. Most L. pneumophila effector proteins have an enigmatic domain architecture that makes it difficult to predict their biochemical function on the basis of sequence similarity with other proteins, but a few effectors have been shown to carry out sophisticated biochemical modification of human proteins6,7.
Effector proteins of the SidE family (SdeA, SdeB, SdeC and SidE) were previously shown to be essential for the virulence of L. pneumophila against its natural host amoeba8. By protein-sequence analysis, Qiu et al. found a mono-ADP ribosyltransferase (mART) motif in all members of this family. They show that the mART motif is essential for SdeA-mediated toxicity in both yeast and mammalian cell culture. Unexpectedly, however, purified SdeA exhibited no detectable ADP-ribosylation activity, indicating that it might have a different biochemical function.
To investigate further, the authors turned to Rab proteins, which are major targets of L. pneumophila effectors6. They found that co-expression of SdeA with various Rab proteins in human cells led to the covalent modification of two of these proteins, Rab1 and Rab33, which are associated with the intracellular membrane structure known as the endoplasmic reticulum. This modification depended on the mART motif of SdeA and was also seen during infection of human cells with L. pneumophila containing wild-type SdeA, but not when SdeA had a mutated mART motif.
Mass spectrometry revealed ubiquitin peptides in the modified Rab proteins but not in the unmodified ones, suggesting that SdeA ubiquitinates Rab proteins during L. pneumophila infection. However, ubiquitination of Rab33 by SdeA was not detected in an in vitro reaction performed in the presence of E1, ATP and various E2s, suggesting that the standard cellular enzyme cascade does not mediate this reaction. The authors then tested the ability of SdeA to modify Rab33 in the presence of both untreated and boiled human cell lysate, and observed ubiquitination in both cases, indicating that a non-protein cofactor is crucial for this process (proteins are denatured by boiling). The molecule NAD is the natural cofactor for the ADP-ribosylation mediated by other mART-containing proteins9 — indeed, adding NAD but not ATP and/or magnesium ions (cofactors involved in standard ubiquitination) to reaction mixtures containing only SdeA, ubiquitin and Rab33 resulted in the ubiquitination of Rab33.
These observations mark the first report of substrate ubiquitination that is independent of E1 and E2 (Fig. 1b). Although the mechanistic details of SdeA-mediated ubiquitination are yet to be resolved, Qiu et al. present glimpses of the reaction intermediates (uncovered by mass spectrometry), which, as expected, differ from E1-dependent ubiquitination. In E1-catalysed activation, ubiquitin's carboxy terminus is modified by adenylation at the expense of an ATP; this is followed by the transfer of ubiquitin to the active-site cysteine residue of E1 and release of an AMP molecule10. By contrast, SdeA seems to catalyse the addition of ADP-ribose to the arginine-42 residue of ubiquitin with the help of NAD, releasing nicotinamide. The modified ubiquitin is subsequently transferred to the substrate protein through an unknown mechanism that results in the release of AMP (Fig. 1b).
In another deviation from the normal ubiquitination mechanism, SdeA shows no detectable difference in ubiquitination of Rab33 when using wild-type ubiquitin, ubiquitin lacking the two C-terminal glycine residues, or ubiquitin lacking all the surface lysine residues. It thus remains to be seen which residues of ubiquitin and Rab33 participate in the covalent linkage that is catalysed by SdeA. The authors also observed forms of Rab33 with multiple ubiquitin attachments. This may be explained by conjugation of multiple mono-ubiquitins or by the formation of polyubiquitin chains. Detailed structural and biochemical studies are required to address these points.
Qiu and colleagues find that SdeA-mediated ubiquitination of Rab33 has only a moderate effect on the protein's activity, and is not sufficient to explain the potent toxic effect of SdeA in cells. It is possible that more substrates exist for SdeA in vivo, and an unbiased screen will be needed to search for these. Undoubtedly, many researchers will also be curious about whether other proteins carry out ubiquitination independently of E1 and E2. Prime suspects for testing could be the bacterial-toxin-related mammalian proteins that contain mART motifs11. Qiu et al. have set the stage for exciting research that promises to uncover further ubiquitin chemistry with potentially far-reaching implications.Footnote 1
Hershko, A., Ciechanover, A. & Varshavsky, A. Nature Med. 6, 1073–1081 (2000).
Qiu, J. et al. Nature 533, 120–124 (2016).
Schulman, B. A. & Harper, J. W. Nature Rev. Mol. Cell Biol. 10, 319–331 (2009).
Joazeiro, C. A. P. & Hunter, T. Science 289, 2061–2062 (2000).
Maculins, T., Fiskin, E., Bhogaraju, S. & Dikic, I. Cell Res. 26, 499–510 (2016).
Hubber, A. & Roy, C. R. Annu. Rev. Cell Dev. Biol. 26, 261–283 (2010).
Xu, L. & Luo, Z.-Q. Microbes Infect. 15, 157–167 (2013).
Bardill, J. P., Miller, J. L. & Vogel, J. P. Mol. Microbiol. 56, 90–103 (2005).
Barkauskaite, E., Jankevicius, G. & Ahel, I. Mol. Cell 58, 935–946 (2015).
Haas, A. L. & Rose, I. A. J. Biol. Chem. 257, 10329–10337 (1982).
Glowacki, G. et al. Protein Sci. 11, 1657–1670 (2002).
About this article
Cite this article
Bhogaraju, S., Dikic, I. Ubiquitination without E1 and E2 enzymes. Nature 533, 43–44 (2016). https://doi.org/10.1038/nature17888
Signal Transduction and Targeted Therapy (2020)
Biochemical Pharmacology (2019)
International Journal of Molecular Sciences (2019)
Exploring the Structural Mechanism of Covalently Bound E3 Ubiquitin Ligase: Catalytic or Allosteric Inhibition?
The Protein Journal (2018)