The ubiquitin system: a critical regulator of innate immunity and pathogen–host interactions

Abstract

The ubiquitin system comprises enzymes that are responsible for ubiquitination and deubiquitination, as well as ubiquitin receptors that are capable of recognizing and deciphering the ubiquitin code, which act in coordination to regulate almost all host cellular processes, including host–pathogen interactions. In response to pathogen infection, the host innate immune system launches an array of distinct antimicrobial activities encompassing inflammatory signaling, phagosomal maturation, autophagy and apoptosis, all of which are fine-tuned by the ubiquitin system to eradicate the invading pathogens and to reduce concomitant host damage. By contrast, pathogens have evolved a cohort of exquisite strategies to evade host innate immunity by usurping the ubiquitin system for their own benefits. Here, we present recent advances regarding the ubiquitin system-mediated modulation of host–pathogen interplay, with a specific focus on host innate immune defenses and bacterial pathogen immune evasion.

INTRODUCTION

The highly conserved ubiquitin (Ub) protein consists of 76 amino acids and can be covalently attached to other proteins via an isopeptide bond formed between its C-terminal glycine residue and the ɛ-amino radical of lysine residues on substrates. Ubiquitination is carried out by the sequential actions of three enzymes. Initially, a Ub-activating enzyme (E1) catalyzes the formation of an energy-rich thioester bond between its active-site cysteine and the C-terminal glycine residue of Ub in an ATP-dependent manner. The activated Ub is then delivered to a similar cysteine residue in the active site of a Ub-conjugating enzyme (E2). Finally, a Ub ligase (E3) binds to both the Ub-charged E2 and the substrate protein, leading to the linkage of Ub to the target protein (Figure 1).¹ In addition to monoubiquitination, a protein can be modified with polymeric Ub chains that are composed of several Ub moieties. The presence of seven internal lysine residues (K6, K11, K27, K29, K33, K48 and K63) and the N-terminal methionine residue on Ub allows the assembly of eight homotypic and multiple-mixed conjugates.²

Figure 1.

The ubiquitination cascade and functional consequences of polyubiquitin chains with different linkage types. An E1 enzyme catalyzes the formation of a thioester bond between its active-site Cys residue and the C terminus of ubiquitin in an ATP-dependent manner. Activated ubiquitin is then transferred to the active-site Cys residue of an E2. The final step is mediated by an E3 ligase that either facilitates the direct transfer of the ubiquitin from E2 onto its substrate (RING E3 ligase) or forms a thioester intermediate with ubiquitin before transfer (HECT or RBR E3 ligase). The RING E3 can be a single chain or a large multiprotein complex that provides specificity and regulatory complexity. The substrate fate changes with different types of polyubiquitination. The K48- and K11-linked ubiquitin chains typically destine target proteins for degradation in the 26S proteasome, whereas the K63- and M1-linked chains are mostly implicated in nonproteolytic events such as signaling cascades. Notably, under autophagic conditions, the K63 linkage can act as a direct substrate for lysosomal destruction by binding to the adaptor p62. HECT, Homologous to E6AP C-Terminus; RBR, RING-In-Between-RING; RING, Really Interesting New Gene.

The human genome encodes 2 E1s, ~40 E2s and ~600 E3s. The E3 ligases comprise three major classes: Really Interesting New Gene (RING) E3s, Homologous to E6AP C-Terminus (HECT) E3s and RING-In-Between-RING (RBR) E3s.^{3, 4, 5} The majority of E3s contain a signature RING domain coordinating two Zn²⁺ in a cross-brace arrangement that functions as a docking site for Ub-loaded E2s. Some RING proteins harbor substrate-recognizing modules themselves and directly act as E3s (for example, tumor necrosis factor (TNF) receptor-associated factors (TRAFs) and inhibitor of apoptosis proteins (IAPs)), whereas others act in multisubunit E3 complexes that employ additional subunits for substrate binding (for example, anaphase-promoting complex/cyclosome and Skp1-cullin-F box protein (SCF)/cullin-RING-ligase (CRL)). In contrast to HECT and RBR E3s, E3s characterized by a RING domain possess no active-site cysteine and thus do not form a thioester intermediate, which means that the RING E3s function more as scaffolds to facilitate the transfer of a donor Ub to an acceptor lysine by aligning the catalytic factors for nucleophilic attack. Consequently, linkage specificity is largely conferred by the E2s rather than the RING E3s.³

HECT E3s feature the C-terminal HECT domain, which assumes a bilobal architecture, with the N-terminal lobe containing the E2-binding domain and the C-terminal lobe harboring the catalytic cysteine residue. Unlike RING E3s, the HECT E3s modify target proteins through a two-step reaction pathway in which the E2s first transfer the donor Ub to the catalytic cysteine residue on the HECT domain through a transthiolation reaction, retaining the high-energy thioester bond. HECT E3s then facilitate optimal orientation of the acceptor lysine toward the Ub-E3 thioester bond to enable nucleophilic attack. Consequently, linkage type is determined by the HECT E3s.⁴

The RBR E3s harbor a conservative catalytic region embedded with a RING1, an in-between RING and a RING2 domain. The RBR E3s catalyze ubiquitination via a RING/HECT-hybrid mechanism in which the recruitment of a Ub-charged E2 by RING1 facilitates the transfer of Ub to form a HECT-like thioester intermediate with a cysteine on RING2 before coupling Ub to its substrate.⁵ As with HECT E3s, RBR E3s control linkage specificity as exemplified by the linear Ub chain assembly complex (LUBAC), which exhibits a unique capacity to synthesize linear Ub chains.⁶

As a versatile post-translational protein modification, ubiquitination triggers diverse biological outcomes. The addition of a single Ub typically alters the activity, interaction or localization of the labeled protein, whereas the conjugation of polyubiquitin chains with different linkage types confers distinct functions because of dissimilar topology structures.⁷ For instance, the K48- and K11-linked polyubiquitin chains exhibit compact conformations and are usually direct substrates for 26S proteasome-mediated degradation, which comprises the canonical Ub–proteasome system. By contrast, when joined through M1 or K63, these conjugates adopt extended conformations and enable the reversible assembly of multiprotein complexes, which are largely implicated in various nonproteolytic events such as immune signaling cascades. In addition, homogeneous Ub chains linked via four other unconventional lysines have also been observed in cells. Several of these atypical linkages mediate the proteasomal degradation of substrates in a similar manner to K11 and K48 linkages; however, the reason for such redundancy remains enigmatic.⁸ Although previous research has primarily focused on the homogenously linked Ub chains, the mixed and branched Ub chains are emerging as crucial modifications bearing distinct physiological functions. For example, Meyer and Rape⁹ recently reported that the anaphase-promoting complex/cyclosome E3 ligase and its cognate E2s, Ube2c and Ube2s, collaboratively augment the rate of proteasomal destruction by decorating their substrates with branched K11- and K48-linked Ub chains.

The Ub status is decoded and translated into specific cellular outputs by various Ub receptors embedded with one or more Ub-binding domains (UBDs). UBDs differ in their structures, which encompass alpha-helical motifs (for example, ubiquitin-interacting motif (UIM)), zinc fingers (for example, Npl4 zinc finger (NZF)), pleckstrin-homology (PH) domains (for example, GRAM-like ubiquitin-binding in EAP45 (GLUE)), Ub-conjugating (Ubc)-related structures (for example, ubiquitin E2 variant (Uev)) and several other emerging Ub-binding structures such as the Src homology 3 and phospholipase A2-activating protein (PLAA) family Ub-binding (PFU) domains.¹⁰ Most UBDs, although structurally dissimilar, recognize the hydrophobic Ile44 patch present on the Ub surface, which is composed of Leu8, Ile44, Val70 and His68. Beyond this hotspot, several other motifs on the Ub molecule, such as the Phe4 patch, Ile36 patch, Asp58 patch and the TEK-box, can mediate either hydrophobic or hydrogen-bond interactions with UBDs. Remarkably, in addition to contacting a single Ub, UBDs have the capacity to distinguish between different polyUb linkage types by detecting the distance between successive Ub moieties, the sequence context surrounding the isopeptide bond and the free C termini of unanchored chains.

Akin to reversible phosphorylation, ubiquitination can be removed by deubiquitinases (DUBs), leading to the enhancement of protein stability or attenuation of Ub signaling and contributing to Ub homeostasis. DUBs are classified as five groups owing to their distinct signature catalytic domains, which include Ub-specific protease (USP) domains, ovarian tumor protease (OTU) domains, Ub C-terminal hydrolase domains, JAMM4 domains and Josephin domains.¹¹ All DUBs, except for the JAMM4 domain-containing DUBs, which are zinc-dependent metalloproteases, belong to the cysteine proteases. Despite folding into divergent architectures, the cysteine DUBs possess either a catalytic dyad (Cys and His) or triad (Cys, His, Asn or Asp) in their catalytic domains. It is noteworthy that DUBs exhibit specificity at various layers to discriminate between different linkage types or substrates, including between Ub and many Ub-like tags (for example, NEDD8 and SUMO), and to cleave the Ub chains from the terminus or from within the target protein. Strikingly, such specificity is not necessarily confined to particular DUB families. In fact, DUBs from the same family can exhibit dissimilar specificities. For instance, the 26S proteasome-localized USP domain-containing USP14 exhibits exo-activity, exclusively disassembling K48-linked chains from the distal end, yielding monoubiquitin, whereas the USP domain-containing CYLD possesses endo-activity, preferentially hydrolyzing K63-linked and linear chains to generate longer Ub chains through internal cleavage.

Here, we highlight the latest advances in the modulation of many aspects of host innate immune responses by the Ub system during infection, including inflammatory signaling, phagosomal maturation, autophagy and apoptosis. In addition, we will summarize the strategies adopted by pathogens to interfere with or to co-opt the Ub system to favor their own intracellular survival and propagation.

MODULATION OF INNATE IMMUNE SIGNALING BY THE UBIQUITIN SYSTEM

Accumulating data suggest that the ubiquitin system has a pivotal role in orchestrating signaling networks emanating from the pattern-recognition receptors.^{12, 13} Pattern-recognition receptors, such as Toll-like receptors (TLRs), RIG-I-like receptors (RLR) and NOD-like receptors (NLRs), provoke various host countermeasures in response to pathogen-associated molecular patterns (PAMPs) or pathogen-induced perturbation of cellular physiology. Upon engagement, most pattern-recognition receptors transduce signaling to promote the expression of pro-inflammatory cytokines, type I interferons (IFNs; IFN-α and IFN-β) and other innate immune genes by activating several immune signaling cascades including the nuclear factor-κB (NF-κB), mitogen-activated protein kinases (MAPK) and interferon regulatory factor 3 (IRF3) cascades. By contrast, certain ligated cytosolic pattern-recognition receptors initiate the formation of inflammasomes, which in turn activate the inflammatory caspases (caspase-1/4/5/11) to proteolytically process pro-IL-1β and pro-IL-18 and to evoke pyroptosis. Following secretion, these cytokines prime other immune cells to prepare them for defense against microbial infections. Given that the Ub system-mediated regulation of the TLR, RLR and TNF-α signaling pathways has been well documented in several previous reviews,^{12, 14} we will focus our discussion on NLR signaling.

NOD1/2 signaling

NOD1 and NOD2, two NLR family members, evoke NF-κB and MAPK activation upon sensing peptidoglycan-derived constituents.¹⁵ NOD1 contains an N-terminal caspase activation and recruitment domain (CARD) belonging to the death domain (DD) superfamily, a centrally located NOD followed by C-terminal leucine-rich repeats (LRRs). NOD2 shares identical domains with NOD1, except that it contains two N-terminal CARD modules.

When unstimulated, the LRRs occlude the NOD module to render NOD1/2 in a monomeric and autoinhibited state. Once ligated, NOD1/2 switches to an open conformation, undergoes self-oligomerization and binds the proximal adaptor receptor-interacting protein 2 through a homotypic CARD–CARD interaction, which in turn recruits cellular inhibitor of apoptosis protein 1/2 (cIAP1/2) to build a NOD1/2-associated signaling complex. Within this signaling hub, cIAP1/2 is activated to conjugate K63-linked chains to RIP2.¹⁶ The polyubiquitin chains then act like scaffolds to recruit two kinase complexes, specifically the TGFβ-activated kinase 1 (TAK1) complex and the canonical IκB kinase (IKK) complexes, through their respective ubiquitin-binding subunits, TAB2–TAB3 (TAK1-associated binding protein) and NEMO (NF-κB essential modulator, also termed IKKγ), which in turn activate the NF-κB and, putatively, the MAPK signaling cascades. Likewise, ITCH, a HECT E3 ligase, was also found to mark RIP2 with K63-linked chains following NOD2 activation.¹⁷ However, unlike cIAP1/2, ITCH-conjugated chains are required for optimal JNK and p38 activation but blunt NF-κB activation. Such a discrepancy might be attributed to the possibility that ITCH competes with cIAP1/2 to govern whether MAPK or NF-κB is the predominant cascade downstream of NOD2 activation. In addition, NOD2 activation leads to NEMO ubiquitination, which was shown to amplify NF-κB signaling and is partially accomplished by NOD2-activated TRAF6.¹⁸ Finally, a recent study suggests that the ubiquitin molecule might compete with RIP2 to associate with CARD of NOD1/2, thereby negatively affecting NOD1/2 signaling.¹⁹

Intriguingly, XIAP (X-linked Inhibitor of Apoptosis) was also recently observed to be present in the NOD2 signaling complex.²⁰ Unlike cIAP1/2, XIAP primarily synthesizes K63- and K48-independent atypical polyubiquitin chains on RIP2, which enable the incorporation of LUBAC into the NOD2 signaling complex. RIP2 was later uncovered to be the prime substrate of LUBAC in NOD2 signaling through quantitative proteomics analysis of the components obtained from M1-Ub-affinity purification.²¹ Although the UBD of TAB2/3, specifically the NZF domain, preferentially binds to K63-type chains, the UBD in NEMO, referred to as UBAN (ubiquitin binding in ABIN and NEMO), binds significantly more strongly to M1 chains compared with K63 chains. Thus, the synchronous addition of K63- and M1-linked chains on RIP2 may provide an optimal platform to facilitate the TAK1-catalyzed activation of the IKK complex.

For controlled and beneficial pro-inflammatory responses, ubiquitination must be counterbalanced by DUBs that are as critical as E3 ligases in regulating immune signaling. The linear ubiquitin chain-specific DUB OTULIN binds to the catalytic LUBAC subunit HOIP and selectively trims the M1-linked chains from RIP2, thereby attenuating NOD2 signaling.²¹ In addition, HOIP also interacts with CYLD, which limits the extension of both K63- and M1-linked chains on RIP2 to restrict NOD2 signaling.²²

Inflammasome signaling

In response to microbial infection, certain NLR family members, including NLRP3 and NLRC4, and the cytosolic DNA sensor AIM2 assemble into a multimeric canonical inflammasome complex to activate inflammatory caspase-1, which in turn mediates the maturation of the pro-inflammatory interleukin (IL)-1β and IL-18 as well as the induction of pyroptosis. Pyroptosis, distinct both morphologically and molecularly from apoptosis, occurs following the activation of inflammatory caspases and cleavage of gasdermin D, which induces membrane rupture and the leakage of cytosolic contents, thereby damaging the replication niche of intracellular pathogens and rendering them vulnerable to neutrophils.^{23, 24} Pyrin domain-harboring NLRP3 and AIM2 inflammasomes signal exclusively through the ASC (apoptosis-associated speck-like protein containing a CARD) adaptor to recruit and activate caspase-1. By contrast, the CARD domain-harboring NLRC4 inflammasome can signal to caspase-1 directly to elicit pyroptosis and indirectly through the adaptor ASC to augment IL-1β and IL-18 secretion.²⁵

Recently, the ubiquitin system has emerged as a critical regulatory mechanism underlying the pyrin domain-containing inflammasomes (Figure 2).²⁶ The NLRP3 is unique among the inflammasome sensors because, instead of one or two specific ligands, it responds to a wide range of stimulatory signals such as ATP, crystalline reagents and the microbial toxin nigericin. Two-step signaling, priming and activation is required for the NLRP3 inflammasome to be competent in processing pro-IL-1β and pro-IL-18 in macrophages. By contrast, the priming signaling alone can induce IL-1β maturation in monocytes where caspase-1 is constitutively activated.²⁷ Lipopolysaccharide (LPS) is best known to prime the NLRP3 inflammasome by inducing NF-κB activation and thus upregulating NLRP3 and pro-IL-1β protein levels.²⁸ In addition, the F-box protein FBXL2, a subunit of the SCF E3 ligase complex, serves as a sentinel inhibitor of NLRP3 by mediating its K48-linked ubiquitination and proteasomal degradation in resting human inflammatory cells. LPS priming prolongs the lifespan of NLRP3 by inducing and activating the FBXL2 inhibitor FBXO3.²⁹ Of note, apart from its protein levels, the activity of NLRP3 is also subject to post-translational regulation by ubiquitination. It has recently been shown that the DUB BRCC3 profoundly promotes inflammasome activation by deubiquitinating NLRP3, whereas the level of BRCC3 does not correlate with the abundance of NLRP3 after LPS exposure, hinting that non-degradative K63-linked ubiquitination might obstruct NLRP3 inflammasome activation in resting macrophages.³⁰

Figure 2.

Modulation of the NLRP3 inflammasome by the ubiquitin system. The NLRP3 inflammasome relies on both priming and activating signals to be competent in processing pro-IL-1β and pro-IL-18 as well as inducing pyroptosis in macrophages. LPS priming upregulates the production of NLRP3, pro-IL-1β and FBXO3 through activation of the NF-κB pathway. FBXO3 functions to decrease FBXL2 protein levels, which targets NLRP3 for K48-linked ubiquitination and proteasomal degradation. BRCC3 enhances inflammasome activation by removing the K63-linked ubiquitin chains from NLRP3. A20 deubiquitinates and thereby limits the maturation of pro-IL-1β. In response to various agonists such as ATP and crystals, NLRP3, ASC and pro-caspase-1 assemble into a multisubunit inflammasome complex in the cytoplasm. LUBAC and cIAP1/2-TRAF2 assist in NLRP3 inflammasome activation through linear and K63-linked ubiquitination of ASC and pro-caspase-1, respectively. By contrast, the K63-linked ubiquitination of ASC directs the NLRP3 inflammasome for degradation in the autophagosome by binding to the autophagic adaptor p62, ultimately resulting in the termination of NLRP3 inflammasome activation. ASC, apoptosis-associated speck-like protein containing a CARD; CARD, caspase activation and recruitment domain; cIAP1/2, cellular inhibitor of apoptosis protein 1/2; IL, interleukin; LRR, leucine-rich repeat; LUBAC, linear Ub chain assembly complex; NBD, nucleotide-binding domain; NF-κB; nuclear factor-κB; TNF, tumor necrosis factor; TRAF, TNF receptor-associated factor 6.

The ubiquitin system is also implicated in modulating other subunits of the NLRP3 inflammasome. The ASC can be linearly ubiquitinated by LUBAC, which enhances activation of the NLRP3 inflammasome independent of its positive role in NF-κB signaling.³¹ Nonetheless, it remains to be determined what signal triggers LUBAC-catalyzed modification of the ASC upon NLRP3 stimulation. Furthermore, the E3 ligases cIAP1/2 in association with the adaptor TRAF2 interact with and deposit K63-linked chains on caspase-1, thereby universally potentiating caspase-1 activation stimulated by various inflammasome agonists, such as ATP and cytosolic flagellin.³² Finally, A20 acts to limit spontaneous or exogenously stimulated NLRP3 inflammasome activation.³³ Mechanistically, in LPS-primed bone marrow-derived macrophages, pro-IL-1β was conjugated with polyubiquitin chains, which likely support the proteolytic cleavage of pro-IL-1β by encouraging higher-order oligomerization of inflammasome subunits. Ablation of A20 in macrophages sharply increased the ubiquitination level of pro-IL-1β in response to LPS alone.

Undoubtedly, the failure to obliterate unwanted inflammasomes would likely bring about excessive inflammation and cell death, which necessitates the ubiquitin system-mediated negative regulation of inflammasomes. A recent study showed that AIM2 or NLRP3 inflammasome activation is accompanied by autophagosome formation in macrophages, which functions to temper inflammation by clearing active inflammasomes. Specifically, the ASC might undergo K63-linked ubiquitination within the assembled inflammasomes, and the autophagic adaptor p62 would then assist in the delivery of inflammasome aggregates to autophagosomes for destruction.³⁴

FUNCTION OF THE UBIQUITIN SYSTEM IN PHAGOSOMAL MATURATION DURING PATHOGEN INFECTION

Phagocytosis, the uptake and clearance of microbial pathogens, exerts a crucial role in innate immune defense against infection. The nascent phagocytic vacuole is innocuous. These eventual antiseptic properties are gained following phagosomal maturation, which entails a succession of precisely choreographed membrane fusion and fission events to remodel the phagosomal membrane and contents, culminating in the biogenesis of phagolysosomes.³⁵ As with the endocytic pathway, the phagosome undergoes early, late and lysosome-fusion stages of maturation. The conversion between phagosomal stages is delicately coordinated by the Rab family GTPases, which comprise over 60 members in mammals.³⁶ Akin to other GTPases, the activity and localization of Rabs is tightly modulated by their cognate GEFs (guanine-nucleotide exchange factors), GTPase-activating proteins and guanine-nucleotide dissociation inhibitors.

Rab5 modulates early phagosomal morphogenesis by promoting the fusion of nascent phagosomes with early endosomes. Initially, the recruitment of Rabex-5, the cognate GEF of Rab5, to early phagosomes promotes Rab5 activation. The activated Rab5 then recruits its effector protein Rabaptin-5, which is capable of boosting the activity of Rabex-5, thereby constituting a positive feedback loop.³⁷ The association of Rabex-5 with early phagosomes is governed by cycling between ubiquitin binding and monoubiquitination.³⁸ Rabex-5 contains an A20-like ZnF that confers E3 activity and interacts with an Asp58-centered polar region on ubiquitin, an inverted UIM (IUIM) that binds to the Ile44 patch on ubiquitin, a central GEF catalytic core and a C-terminal coiled-coil (CC) domain comprising the Rabaptin-5-docking region.³⁹ Rabex-5 is recruited from the cytosol to the early phagosomes through its ZnF and IUIM domain-mediated interactions with the ubiquitinated membrane protein, as well as its CC domain-mediated association with other as yet undetermined factors. Following Rab5 activation and endosomal fusion or conveyance of cargo, Rabex-5 undergoes Ub-binding-stimulated monoubiquitination, leading to the migration of Rabex-5 from the phagosomes to the cytoplasm, where the covalently linked Ub moiety may bind intramolecularly or intermolecularly to its ZnF/IUIM motif. Rabex-5 deubiquitination, presumably carried out by the two endosomal DUBs, associated molecule with the SH3 domain of STAM (AMSH) and ubiquitin-specific processing protease Y (UBPY), abrogates this binding and allows the resumption of this cycle.⁴⁰ Together, ubiquitin system-mediated Rabex-5 regulation integrates cargo detection with endosomal fusion.

A hallmark of early-to-late phagosomal transition is the acquisition of Rab7 and concomitant loss of Rab5. The active Rab7, through its effectors, mediates further maturation of the phagosomes and centripetally directs late phagosomes along the microtubules to facilitate fusion with lysosomes. Parkin was observed to regulate the endolysosomal pathway (and putatively the phagolysosomal pathway) by targeting Rab7 for ubiquitination, thereby affecting the activity and protein level of Rab7. Specifically, parkin-mediated ubiquitination might increase the activity and membrane association of Rab7. Activated Rab7 demonstrates potentiated capacity for effector binding, which might in turn stabilize Rab7.⁴¹

Of note, the phagosomal membrane proteins can be targeted to intraluminal vesicles (ILVs), similar to the multivesicular body (MVB) formed in the early or late endosomal system, for degradation, which are generated after invagination and scission of the limiting membrane.⁴² Importantly, perturbation of this process adversely affects phagosomal maturation and pathogen eradication.⁴³ The ESCRT (endosomal sorting complex required for transport) machinery consists of four complexes, ESCRT-0, -I, -II and -III, which operate sequentially to mediate MVB formation via three distinct but connected steps: capturing the ubiquitin-tagged cargo, deforming the phagosomal/endosomal membrane and membrane abscission from the inside.⁴⁴ The ubiquitin system has a central role in ESCRT-mediated MVB biogenesis.⁴⁵ Specifically, ubiquitination specifies which membrane cargo should be delivered to the lysosomal lumen. Then, the ESCRT machinery sorts ubiquitinated cargo into ILVs. ESCRT-0, -I and -II complexes employ several distinct UBDs, including UIM (STAM1/2), double-sided UIM (Hrs), VHS (STAM1/2, Hrs), catalytically inactive Uev (Tsg101) and GLUE (EAP45), to capture and gather the ubiquitinated cargo into a specific membrane compartment. Finally, ESCRT-III with no UBDs primarily promotes vesicle budding and scission. In addition to recognizing ubiquitinated cargo, certain ESCRT proteins can themselves be ubiquitinated, leading to alterations in their stability, activity or localization. The ESCRT-0 subunit Hrs is monoubiquitinated, which maintains Hrs in an inactive conformation and precludes its recognition of the Ub-labeled membrane cargo owing to the intramolecular Ub–UIM interaction.⁴⁶ Similarly, disturbance of the E3 Mahogunin-mediated ubiquitination of the ESCRT-I subunit Tsg101 compromises endosome/phagosome-to-lysosome traffic.⁴⁷

Although ubiquitination provides a sorting signal for the influx of cargo into the MVB compartment, the cargo must be deubiquitinated before packaging into ILV to maintain stable cellular ubiquitin pools. In mammals, AMSH and UBPY are recruited to the endosome or phagosome not merely by the ESCRT-0 subunit STAM but also by the ESCRT-III complex. In contrast to AMSH, which exclusively removes K63-type chains, UBPY disassembles both K63- and K48-type chains. In addition to recycling Ub or cargo from the MVB pathway, these endosomal DUBs can reverse the autoinhibitory monoubiquitination of ESCRT proteins. For example, AMSH is capable of cleaving the ubiquitin moiety appended on Hrs.⁴⁸

Beyond ESCRT complexes, several other endocytic adaptors participate in MVB formation. Their stability and activity are similarly sensitive to modulation by the ubiquitin system. For example, CRL3^SPOPL mediates the degradative ubiquitination of the endocytic adaptor EPS15, which is needed for efficient ILV formation during endosomal/phagosomal maturation. CRL3^SPOPL may be activated at endosomes/phagosomes by the E3 ligase DCNL3, which is known to associate with membranes by virtue of a covalent lipid modification that conjugates NEDD8 to the CUL3 subunit.⁴⁹

INVOLVEMENT OF THE UBIQUITIN SYSTEM IN ANTIMICROBIAL AUTOPHAGY

Autophagy is an evolutionarily conserved homeostatic process in which cells eliminate protein aggregates, damaged or superfluous organelles, and microbial pathogens from their cytosol or utilize bulky cytosolic contents for metabolism.⁵⁰ A marked morphological feature of autophagy is the genesis of double-membrane autophagosomes, which sequester and transport the entrapped cargoes to lysosomes for decomposition. The process of de novo autophagosomal formation is primarily regulated by four complexes: the ULK1 protein kinase complex, the VPS34 lipid kinase complex, two transmembrane proteins, ATG9 and VMP1, and two ubiquitin-like protein conjugation machineries operating on ATG12 and LC3 (mammalian homolog of yeast ATG8). The first ubiquitin-like conjugation reaction facilitates the connection of ATG12 with ATG5, catalyzed by the E1-like ATG7 and E2-like ATG10. In complex with ATG16L1, ATG5–ATG12 in turn function as an E3-like ligase and collaborate with an E2-like ATG3 to facilitate the linkage of LC3 to phosphatidylethanolamine at the phagophore.

Substantial data indicate that autophagic initiation and progression is tightly modulated by the ULK1 and VPS34 complexes. The Ser/Thr protein kinase ULK1 assembles a stable complex with FIP200 (ATG17 in yeast) and ATG13, and its activity is controlled by the nutrient-sensitive kinase complex, mTORC1 (mammalian target of rapamycin complex 1). The Ser/Thr protein kinase mTOR exists in two distinct multisubunit complexes known as mTORC1 and mTORC2 in mammalian cells, with mTORC1 suppressing autophagy directly via the phosphorylation and deactivation of the ULK1, whereas mTORC2 promotes survival and suppresses autophagy indirectly via the phosphorylation and activation of AKT, which in turn activates mTORC1 along the PI3K–AKT–mTOR cascade.⁵¹ VPS34 is the sole class III phosphoinositide 3-kinase in mammals that functions in a stable complex containing p150 (VPS15 ortholog) and Beclin-1 to produce phosphatidylinositol 3-phosphate. Beclin-1 forms a binding platform for several other regulatory proteins with the capacity to either enhance (UVRAG, ATG14L, Bif1 and AMBRA1) or suppress (Bcl-2, TAB2/3 and Rubicon) the autophagic activity of VPS34.^{52, 53} Following mTOR suppression, activated ULK1 boosts the kinase activity of the ATG14L-harboring VPS34 complex through the phosphorylation of Beclin-1 on Ser14.⁵⁴

Pathogen infection-triggered selective autophagy is also termed xenophagy and functions to restrict a range of pathogens, such as Mycobacterium tuberculosis (Mtb) and Salmonella enterica serovar typhimurium. The ubiquitin system is widely implicated in the initiation of the autophagic pathway as well as the subsequent coupling of autophagosomes to bacterial targets (Figure 3).⁵⁵ The kinase activities of both mTOR and AKT are subject to ubiquitin system-dependent modulation. In response to nutrients, p62 brings TRAF6 to mTORC1 via its binding affinity for both raptor and TRAF6.⁵⁶ Then, the recruited TRAF6 marks mTOR with K63-linked ubiquitin chains, which are instrumental for the activation of mTOR. In addition, K48-linked ubiquitination has been documented to directly affect the stability or indirectly affect the kinase activity of mTOR. For instance, DEPTOR, an endogenous inhibitor for both mTORC1 and mTORC2, becomes phosphorylated by mTOR in cooperation with casein kinase I upon stimulation by growth signals, which create a phosphodegron for SCF^βTrCP binding, ultimately resulting in DEPTOR degradation and mTOR activation.⁵⁷ Finally, AKT undergoes TRAF6-catalyzed K63-type ubiquitination at the PH domain upon exposure to growth factor, which promotes AKT membrane adherence, phosphorylation and activation without influencing its phosphatidylinositol (3,4,5)-trisphosphate lipid-binding capability.⁵⁸

Figure 3.

The ubiquitin system regulates the xenophagy of bacterial pathogens. Both initiation of the autophagic pathway and the subsequent capture of bacterial targets is modulated by the ubiquitin system. On one hand, the E3 ligase TRAF6 supports K63-linked polyubiquitination and activation of mTORC1, ULK1 and Beclin-1-VPS34 complexes with the help of p62, AMBRA1 and TLR4, respectively, leading to the inhibition or activation of autophagy. On the other hand, cytosol invasion triggers the E3 ligase (for example, Parkin and LRSAM1)-mediated ubiquitination of the bacterial pathogen itself or the associated host components. Then, autophagic receptors such as p62, NDP52 and optineurin (OPTN) that bind to both ubiquitin and the autophagic marker LC3 direct ubiquitin-tagged bacteria to the autophagosome for elimination. The NDP52 is unique among the autophagic receptors because of its ability to bind both ubiquitin and galectin-8, another 'eat-me' signal that recognizes the exposed host glycans on the damaged phagosome. mTORC1, mammalian target of rapamycin complex 1; NDP52, nuclear domain 10 protein 52; TLR, Toll-like receptor; TNF, tumor necrosis factor; TRAF, TNF receptor-associated factor.

Recent data showed that the activity of ULK1 is also regulated by ubiquitination. In cells undergoing autophagy, AMBRA1, a binding partner for both Beclin-1 and ULK1, in complex with TRAF6 supports K63-linked polyubiquitination and consequent self-association, stabilization and activation of ULK1.⁵⁹ Notably, under normal conditions, mTORC1 phosphorylates AMBRA1 and precludes its recruitment of TRAF6. Because ULK1 activates AMBRA1 via phosphorylation, such AMBRA1-mediated modification of ULK1 likely operates as a positive feedback regulatory loop. Furthermore, the ubiquitin system targets Beclin-1 to modulate autophagic induction. Specifically, upon engagement by LPS, TLR4 recruits Beclin-1 to the receptor-bound signaling node where the activated TRAF6 decorates Beclin-1 with K63-linked chains, thereby disrupting the interaction between Bcl-2 and Beclin-1 and leading to the oligomerization of Beclin-1 and activation of VPS34.⁶⁰ In addition to its activity, the stability of Beclin-1 is also under the control of ubiquitination. For example, RNF216 binds to and destabilizes Beclin-1 through K48-linked ubiquitination, whereas USP19 maintains Beclin-1 protein levels through K11-specific deubiquitination.^{61, 62}

Upon induction, xenophagy, reminiscent of other selective autophagic routes, such as mitophagy, utilizes autophagic receptors that bind both a molecular tag attached to the bacterial cargo and the autophagic modifier LC3 to selectively bridge bacterial cargo to the autophagosomal membrane.⁶³ Ubiquitin appears to represent a common 'eat-me' signal in various selective autophagic pathways including xenophagy.⁶⁴ For example, both K48- and K63-linked ubiquitin chains were previously observed to accumulate around the vacuolar pathogen Mtb during infection in macrophages. Recent research partially sheds light on the underlying mechanism: the bacterial ESX-1 T7SS-mediated phagosomal permeabilization allows the cGAS-STING-dependent cytosolic DNA sensor pathway to access extracellular bacterial DNA and then triggers parkin-mediated K63-linked ubiquitination surrounding the Mtb-containing phagosome, which is required for the delivery of bacilli to autophagosomes.^{65, 66, 67} Despite these observations, it is still poorly understood how parkin is recruited and activated at the bacterial niche and which bona fide protein(s) are targeted by parkin, as well as the identity of other E3s responsible for K48-linked ubiquitination. Another well-defined example involves the LRR-harboring RING family E3 ligase LRSAM1, which targets Sa. typhimurium to the autophagic pathway in epithelial cells.⁶⁸ Notably, LRSAM1 directly senses and tags the cytosol-invasive bacteria with ubiquitin chains by virtue of its LRR and RING domains, respectively, in the absence of other assistant proteins. Unlike parkin, LRSAM1 predominantly synthesizes K6- and K27-linked ubiquitin chains. Furthermore, in addition to Sa. typhimurium, LRSAM1 was found to localize to multiple Gram-negative and -positive bacteria. Thus, LRSAM1 might represent a unique pathogen-recognizing E3 ligase that is active against a broad range of intracellular bacteria in the autophagic pathway.

The ubiquitinated coat deposited on bacteria or the surrounding vacuole are bound by three autophagic receptors, specifically, p62 (SQSTM1), NDP52 (nuclear domain 10 protein 52) and optineurin through their C-terminal UBA, ZnF or UBAN domains, respectively, resulting in the isolation of bacteria by autophagosomes.⁶⁹ In addition, NDP52 detects galectin-8 signaling, which interacts with the host glycans exposed on the impaired phagosomal membrane. It was proposed that NDP52 is transiently recruited to the impaired phagosome by galectin-8, whereas the persistent localization of NDP52 still requires ubiquitination.⁷⁰ Notably, these three autophagic receptors might perform non-redundant functions in xenophagy. A possible mechanism is that these receptors bind to different ubiquitin chains with dissimilar affinities. For example, p62 displays a preference for K63 chains over K48 chains, whereas Optineurin preferentially binds linear chains. Alternatively, there might be beneficial interactions among them. For example, NDP52 recruits TBK1 to the cytosolic Sa. typhimurium, which in turn enhances the affinity of Optineurin for LC3B through phosphorylation.⁷¹

APOPTOSIS REGULATION BY THE UBIQUITIN SYSTEM

Apoptosis is a vital cell suicide modality in innate immunity that can be triggered either along the extrinsic pathway by the pro-apoptotic TNF superfamily ligands or along the intrinsic pathway by several internal apoptotic insults, such as DNA damage and endoplasmic reticulum (ER) stress, leading to the clearance of the colonization of intracellular pathogens.⁷² The cell-surface death receptors (DRs), belonging to the TNFR superfamily and featuring the presence of a cytosolic DD, can be divided into two categories: TNFR1, DR3 and possibly DR6-binding TNFR1-associated death domain-containing protein and exerting pro-inflammatory and pro-survival functions, or Fas/CD-95, DR4 and DR5 engaging FADD (Fas-associated protein with DD) and exerting primarily pro-apoptotic functions. Furthermore, crosstalk might occur between these two opposing processes.⁷³

TNF-α-induced apoptosis can proceed via either a RIP1-dependent or a RIP-independent pathway.⁷⁴ The RIP1-dependent pathway is controlled by its ubiquitination status. Following the second mitochondrial activator of caspases (Smac)-induced cIAP autodegradation and CYLD/USP7/USP2a-mediated K63 deubiquitination of RIP1, RIP1 is liberated from TNFR1 signaling complex I and in turn recruits the FADD adaptor and the apical caspase-8 to build a cytosolic complex II, also dubbed the Death-Inducing Signaling Complex (DISC), which mediates apoptotic cell demise.^{75, 76} It is noteworthy that Smac is discharged from mitochondria via mitochondrial outer membrane permeabilization (MOMP), which suggests that TNF-α-induced RIP-dependent apoptosis might occur downstream of the intrinsic mitochondrial pathway. In addition, recent studies have indicated that the lack of linear ubiquitination in complex I accelerates complex II formation in TNF signaling.⁷⁷ Whereas CYLD directly binds LUBAC and removes linear chains, thereby sensitizing cells to TNF-provoked apoptosis, A20 binding to the linear chains via its ZnF7 domain antagonizes their removal, consequently blocking cell death.⁷⁸ Dissimilarly, the RIP1-independent apoptotic branch is controlled by the level of cellular FLICE Inhibitory Protein (c-FLIPL), which effectively blocks caspase-8 activation by competitively binding to FADD. The E3 ligase Itch activated by JNK1 phosphorylation acts to provoke apoptosis by directing c-FLIPL for proteasomal destruction.⁷⁹

In contrast to TNFR1, DR4, DR5 and Fas, upon stimulation by their cognate ligands, directly recruit the adaptor FADD and the initiator caspase-8 to establish a membrane-bound DISC. A recent report proposed that this assembled DISC recruits a Cul3-Rbx1 E3 ligase that deposits K63-linked chains on the small p10 catalytic domain of caspase-8.⁸⁰ The p62 adaptor then boosts caspase-8 aggregation within the intracellular ubiquitin-rich foci, thereby augmenting enzyme processing and leading to robust caspase-3 and caspase-7 activation. By contrast, TRAF2 mediates K48-linked ubiquitination of caspase-8 on its large p18 catalytic domain downstream of Cul3-Rbx1 at the DISC, thereby destining the activated caspase-8 for rapid degradation following autoprocessing and membrane separation.⁸¹ Thus, TRAF2 appears to place a stringent barrier that impedes apoptotic commitment upon DR4/5 or Fas ligation.

Mitochondria act as signaling hubs that integrate and translate diverse pro-survival and pro-apoptotic signals released from the cell nucleus or compartments during intrinsic apoptosis. The Bcl-2 family proteins, featuring one or more Bcl-2 homology (BH) domains, fine-tune this intrinsic apoptotic pathway by governing the mitochondrial outer membrane integrity.⁸² Substantial data have highlighted the key role of the ubiquitin system in controlling the dynamic balance between the protein levels of pro- and anti-apoptotic Bcl-2 proteins, which acts as a pivotal checkpoint for apoptosis. Upon stimulation by survival signals, BimEL, a pro-apoptotic BH3-only protein, undergoes Rsk1/2-executed phosphorylation in a conserved degron, which is enhanced by the Erk1/2-catalyzed phosphorylation of BimEL, thus facilitating SCF^βTRCP binding and its subsequent proteasomal degradation.⁸³ Conversely, the DUB USP27X binds BimEL upon its Erk1/2-dependent phosphorylation, removes the ubiquitin chains and stabilizes the phosphorylated BimEL.⁸⁴ Another example involves Bok (Bcl-2 ovarian killer), which acts as an unconventional effector of MOMP and triggers apoptotic cell death when the ER-associated degradation (ERAD) system is impaired during ER stress. Strikingly, unlike Bax and Bak, Bok harbors constitutive activity that is restricted by ERAD E3 ligase gp78-mediated degradative ubiquitination.⁸⁵ Finally, Mcl-1 is unique compared with other pro-survival Bcl-2 proteins because of its rapid turnover. Several E3 ligases, including MULE, SCF^βTRCP and SCF^FBW7, have been revealed to mark Mcl-1 for degradation in a cellular context-dependent manner.^{86, 87, 88} Conversely, USP9X binds and cleaves the K48-type chains from Mcl-1 to hamper the degradation of Mcl-1 by the 26S proteasome.⁸⁹

Beyond Bcl-2 family proteins, the ubiquitin system governs the stability, activity or localization of certain regulators of Bcl-2, such as p53, which induces apoptosis through transcription-dependent and -independent activation of Bcl-2 family members.⁹⁰ A critical negative modulator of p53 is MDM2 and its homolog MDMX.⁹¹ Despite bearing a RING domain, MDMX exhibits no intrinsic E3 activity. Rather, MDMX drives p53 degradation by forming a heterodimer with MDM2, the bona fide RING E3 ligase, which is mediated by a homotypic RING–RING interaction. Conversely, in response to genotoxic pressure, USP10 undergoes ATM-mediated phosphorylation and then migrates to the nucleus, where it stabilizes p53 through deubiquitination.⁹² Furthermore, basal levels of MDM2 can function alone or in homodimers to predominantly monoubiquitinate p53 in resting cells, leading to nuclear export of p53. Under apoptotic conditions, such cytoplasmic monoubiquitinated p53 translocates to the mitochondria, where it undergoes rapid deubiquitination and activation by the mitochondrial herpesvirus-associated ubiquitin-specific protease (HAUSP, also known as USP7).⁹³

Downstream of MOMP, the stability and activity of caspases is regulated by IAPs, which primarily include XIAP and cIAPs. Unlike cIAPs, XIAP acquires the ability to directly bind and restrain the initiator caspase-9 and the executor caspase-3/7.⁹⁴ In addition, XIAP promotes the K48-linked polyubiquitination of the processed caspase-3 instead of pro-caspase-3.⁹⁵ Analogous to XIAP, cIAP1 interacts with and ubiquitinates the fully mature caspase-7 and partially processed caspase-3, thereby facilitating their turnover.⁹⁶ Furthermore, IAPs regulate their abundance reciprocally, thereby ensuring optimal levels to allow cells to respond quickly to the ever-changing environment. For example, in a homogeneous interaction mediated by their respective RING domains, cIAP1 reduces the protein levels of XIAP by inducing its degradative ubiquitination.⁹⁷

INTERVENTION OF THE UBIQUITIN SYSTEM IN IMMNUE EVASION BY BACTERIAL PATHOGENS

Because the ubiquitin system is indispensable for eukaryotic cells to regulate diverse immune responses, many pathogens have developed multiple mechanisms to cunningly subvert this well-balanced program to favor their own survival. Bacterial pathogens inject diversified effector proteins into the host cellular environment through special secretion systems, including the type III secretion system (T3SS), T4SS, T6SS and T7SS, during infection.^{98, 99} After entry into the host cells, those effectors recompose various cellular processes to destroy host immune defenses and create an adaptive niche for bacterial intruders to survive and proliferate.^{100, 101} Notably, the ubiquitin pathways are sensitive targets of many bacterial effectors (Table 1). Here, we primarily focus on the detailed mechanisms employed by bacterial pathogens to manipulate the host ubiquitin system.

Table 1. List of bacterial effector proteins and their interactions with host cells.

Bacterial effector	Organism	Biochemical characteristics	Host targets	Cellular function	Ref
Mimicry of E3 ubiquitin ligase
SopA	Salmonella typhimurium	HECT-type E3 ligase	Unclear	Intervention in host inflammation	116
NleL	EHEC	HECT-type E3 ligase	Unclear	Modulation of actin-pedestal formation	134
NleG	EHEC, EPEC and Salmonella	RING/U-box-type E3 ligase	Unclear	Unclear	135
LubX	Legionella pneumophila	RING/U-box-type E3 ligase	CLK1	Promotion of bacterial intracellular survival	118
AnkB	L. pneumophila	F-box protein	K48-linked polyubiquitinated proteins, Trim21	Assimilation of host amino acids	125
LegU1	L. pneumophila	F-box protein	BAT3	Unclear	127
LegAU13	L. pneumophila	F-box protein	Unclear	Unclear	127
IpaH9.8	Shigella flexneri	NEL-type E3 ligase	NEMO	Inhibition of the NF-κB pathway	103
IpaH0722	S. flexneri	NEL-type E3 ligase	TRAF2	Inhibition of the NF-κB pathway	104
IpaH4.5	S. flexneri	NEL-type E3 ligase	p65	Inhibition of the NF-κB pathway	105
IpaH7.8	S. flexneri	NEL-type E3 ligase	GLMN	Promotion of inflammasome activation and cell death	106
SlrP	Salmonella typhimurium	NEL-type E3 ligase	TRX	Intervention in host cell death	113
SspH1	Sa. typhimurium	NEL-type E3 ligase	PKN1	Inhibition of the NF-κB pathway	114
SspH2	Sa. typhimurium	NEL-type E3 ligase	NOD1	Intervention in host inflammation	115
SidC/SdcA	L. pneumophila	C-H-D triad-containing E3 ligase	Unclear	Recruitment of host ER vesicles	128, 129
SdeA	L. pneumophila	E1/E2-independent E3 ligase; ADP-ribosyltransferase	Multiple ER-associated Rabs such as Rab33b, Rab1	Promotion of bacterial intracellular replication	131
Mimicry of deubiquitination enzyme
AvrA	Sa. typhimurium	DUB	IκB-α, β-catenin	Intervention in host inflammation	112
SseL	Sal. typhimurium	DUB	Ubiquitin aggregates around the SCV	Evasion of host autophagy	111
TssM	Burkholderia pseudomallei	DUB	TRAF3, TRAF6 and IκBα	Inhibition of type I IFN and NF-κB pathways	140
YopJ	Yersinia pseudotuberculosis	DUB	TRAF2, TRAF6, TRAF3, IκBα and TAK1	Inhibition of NF-κB and MAPK pathways	142, 145
YopP	Y. enterocolitica	DUB	TRAF6, NEMO, TAK1 and TAB1	Inhibition of NF-κB and MAPK pathways	141, 144
Other effector-mediated interference of the host ubiquitin system
OspI	S. flexneri	Glutamine deamidase	UBC13	Inhibition of NF-κB pathway	108
OspG	S. flexneri	Serine/threonine kinase	Ubiquitinated E2 such as UbcH5, UbcH7	Inhibition of NF-κB pathway	107
NleE	EHEC and EPEC	Methyltransferase	TAB2, TAB3	Inhibition of NF-κB pathway	137
NleB	Citrobacter rodentium	O-GlcNAc transferase	GAPDH	Inhibition of NF-κB pathway	146
Tir	EHEC and EPEC	Unclear	SHP-1 and SHP-2	Inhibition of NF-κB pathway	136
InlC	Listeria monocytogenes	Unclear	IKKα subunit of IκB	Inhibition of NF-κB pathway	150
LLO	Li. monocytogenes	Bacteria pore-forming toxin	UBC9	Inhibition of TGFβ signaling	152
PtpA	Mycobacterium tuberculosis	Protein tyrosine phosphatase with UIML domain	JNK, p38, TAB3	Inhibition of NF-κB and MAPK pathways	157
RavZ	L. pneumophila	Cysteine protease	Lipid-conjugated Atg8	Evasion of host autophagy	130
ActA	Li. monocytogenes	Unclear	Arp2/3 complex and Ena/VASP	Evasion of host autophagy	151
Cif	EHEC and EPEC	Glutamine deamidase	NEDD8, ubiquitin	Inactivation of host CRLs	138, 139
CHBP	B. pseudomallei	Glutamine deamidase	NEDD8, ubiquitin	Inactivation of host CRLs	138, 139
SopB	Sa. typhimurium	Inositol phosphate phosphatase	Unclear	Promotion of bacterial invasion and survival	121
SopE	Sa. typhimurium	GEF	Cdc42, Rac1	Promotion of bacterial invasion and survival	119
SptP	Sa. typhimurium	Protein tyrosine phosphatase	NSF, Syk	Promotion of bacterial invasion and intracellular replication	118, 120
InlA	Li. monocytogenes	Unclear	E-cadherin	Promotion of bacterial internalization	148
InlB	Li. monocytogenes	Unclear	c-Met	Promotion of bacterial invasion	149

Abbreviations: Arp2/3, actin-related protein 2/3; CLK1, CDC2-like kinase 1; DUB, deubiquitinase; EHEC, enterohemorrhagic Escherichia coli; Ena/VASP, enabled/vasodilator-stimulated phosphoprotein; EPEC, enteropathogenic E. coli; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GEF, guanine-nucleotide exchange factor; HECT, homologous to E6-AP C terminus; IFN, interferon; IKK, IκB kinase; INl, internalin; LLO, listeriolysin O; MAPK, mitogen-activated protein kinase; NEDD8, neural-precursor-cell-expressed developmentally down-regulated 8; NEL, Novel E3 Ligase; NEMO, NF-κB essential modulator; NF-κB; nuclear factor-κB; PKN1, protein kinase N1; RING, Really Interesting New Gene; SCV, Salmonella-containing vacuole; TAK1, TGFβ-activated kinase 1; TAB, TAK1-associated binding protein; TRX, thioredoxin; UIML, ubiquitin-interacting motif-like.

Shigella flexneri

As a prevalent pathogenic Gram-negative bacterium, S. flexneri utilizes its T3SS to invade and colonize human intestinal epithelial cells. Notably, in addition to S. flexneri, the majority of other Gram-negative bacterial pathogens are also equipped with T3SS to release a vast array of bacterial proteins into host cells. The core device of T3SS, termed the injectisome, is a 3.5-MDa syringe-like protein complex composed of stacked rings embedded in the bacterial membrane and a needle-like filament that extends into the extracellular space from the bacterial surface.⁹⁹ To some extent, the injectisome acts as a pipeline to transport pathogenic cargo from the bacterial cell to the host cell. Several of the T3SS effectors delivered by S. flexneri have been identified as Novel E3 Ligases (NELs), which are structurally distinct from the eukaryotic E3 ligases.¹⁰² Certain NEL family members, including IpaH9.8, IpaH0722 and IpaH4.5, have been shown to tamper with the host NF-κB cascade by ubiquitinating NEMO, TRAF2 and p65 for proteasomal degradation, respectively.^{103, 104, 105} Another NEL effector, IpaH7.8, possesses E3 ligase activity in vitro and targets an inhibitor of Cullin-based RING E3 ligase named GLMN for degradation, which augments inflammasome activation.¹⁰⁶ In addition, OspG and OspI, belonging to a different subsets of S. flexneri effectors, have recently been identified to suppress NF-κB activation by disrupting the ubiquitin system. Mammalian serine/threonine kinase imitator OspG binds to many ubiquitinated E2 proteins, including UbcH5 and Ubch7, in host cells, thereby enhancing its own kinase activity and preventing the ubiquitination and degradation of phosphorylated IκBα.¹⁰⁷ OspI inactivates Ubc13 by functioning as a glutamine deamidase, thus causing the inhibition of TRAF6 polyubiquitination.¹⁰⁸

Sa. typhimurium

Sa. typhimurium is an intracellular Gram-negative pathogenic agent that survives in a special vacuole called the Salmonella-containing vacuole (SCV) after internalization into the macrophages or epithelial cells.¹⁰⁹ Two different T3SSs, T3SS1 and T3SS2, encoded by pathogenicity island-1 (SPI-1) and SPI-2, respectively, are employed by Sa. typhimurium to secrete effector proteins. T3SS1 is favorable during the invasion stage, whereas T3SS2 is conducive to the development of SCV after bacterial entry into the host cell. Several of these effectors function to disturb the ubiquitin system.¹¹⁰ For example, the effector SseL is identified as a DUB and deubiquitinates the ubiquitin aggregates surrounding the SCV to help bacteria escape from autophagy.¹¹¹ Another DUB discovered in Sa. typhimurium is AvrA, which inhibits host inflammatory responses by deubiquitinating IκBα and β-catenin.¹¹² Similar to the IpaH family effectors in S. flexneri, Sa. typhimurium encodes three NELs: SlrP, SspH1 and SspH2. In particular, SlrP has an impact on host cell death by targeting mammalian thioredoxin.¹¹³ SspH1 decreases NF-κB activation by ubiquitinating the protein kinase N1,¹¹⁴ whereas SspH2 was shown to target NOD1 for monoubiquitination, thereby increasing the NOD1-dependent secretion of IL-8.¹¹⁵ Another E3 effector of Sa. typhimurium, SopA, is structurally similar to the eukaryotic homologous to E6-AP C terminus (HECT) E3 ligase and interacts with the host E2 enzyme UbcH7.¹¹⁶ Although the substrate in the host cells has not yet been demonstrated, it is believed that SopA-mediated ubiquitination is involved in the regulation of host inflammation.¹¹⁷ Intriguingly, Sa. typhimurium also utilizes the host ubiquitin system to process and regulate its own effector proteins. SopE and SptP harbor dissimilar functions, and their degradation is temporally differentially regulated by the ubiquitin system to ensure efficient bacterial invasion.^{118, 119, 120} Another example involves SopB, which is modified by monoubiquitination during a special invasion period or within the context of proper subcellular localization such that it systematically performs multiple functions.¹²¹

Legionella pneumophila

Reminiscent of Sa. typhimurium, the causative agent of a severe human pneumonia, L. pneumophila also establishes a special vacuole named the Legionella-containing vacuole (LCV) after intruding into host cells.¹²² L. pneumophila secrets various E3 ligase-mimic effectors via the Dot/Icm T4SS to maintain its intracellular survival.¹²³ T4SS, despite sharing structural similarity with T3SS, is extensively distributed in both Gram-negative and -positive bacteria and uniquely manages the translocation of not only effector proteins but also toxins and plasmids into host cells.⁹⁹ For example, the effector LubX contains two U-box domains that ubiquitinate the host CDC2-like kinase 1.¹²⁴ In addition, AnkB, LegU1 and LegAU13 exhibit structural similarity to the eukaryotic F-box-containing E3 ligase. AnkB is crucial for L. pneumophila to acquire nutrition as it locates to the LCV membrane and recruits K48-linked polyubiquitinated proteins, the degradation of which yields amino acids for the bacteria.¹²⁵ Another study showed that AnkB is also modified by Trim21-mediated K11-linked ubiquitination; however, this fails to change its stability.¹²⁶ LegU1 and LegAU13 are integrated into the host SCF E3 complex, with the former directly ubiquitinating BAT3, a host co-chaperone of Hsp70 that is implicated in ER stress modulation.¹²⁷ SidC and its paralog SdcA, identified in L. pneumophila, are capable of recruiting the ER-derived vesicles to build LCV.¹²⁸ A new study showed that SidC/SdcA is a novel type of bacterial E3 ligase with a special C46-H444-D446 catalytic triad that can form high-molecular-weight polyubiquitin chains.¹²⁹ In addition, the effector RavZ specifically deconjugates the lipid-attached LC3 marks and thereby prevents the autophagic pathway by functioning as a cysteine protease.¹³⁰ Finally, a peculiar yet interesting study recently reported that SdeA, belonging to the SidE family effectors of L. pneumophila, is a novel all-in-one ubiquitin conjugation enzyme, which solely mediates the ubiquitination of many host Rab GTPases by virtue of a putative mono-ADP-ribosyltransferase motif.¹³¹ Of note, this report not only demonstrates for the first time that ubiquitination can be catalyzed by bacterial effectors in the absence of E1 and E2 enzymes but also hints that ubiquitin can be activated by ADP ribosylation.

Enterohemorrhagic Escherichia coli and enteropathogenic Escherichia coli

Both enterohemorrhagic Escherichia coli (EHEC) and enteropathogenic E. coli (EPEC) are notable agents of the pathogenic E. coli family, whose adhesion to intestinal epithelial cells or invasion into host cells in some circumstances can induce multiple human bowel diseases.¹³² Two types of E3 effectors, NleL and NleG, are identified in HEHC. NleL shares a homologous C-terminal sequence with Sa. typhimurium SopA and is approved as a HECT E3 effector because of its catalytic activity to form polyubiquitin chains in vitro.¹³³ NleL modulates EHEC-induced actin-pedestal formation, which contributes to EHEC adherence to intestinal epithelial cells; however, the molecular target of NleG in host cells remains unclear.¹³⁴ As mimics of the eukaryotic RING/U-box E3 ligases, NleG family effectors are also identified in other pathogenic E. coli such as EPEC and Salmonella.¹³⁵ Both EHEC and EPEC deliver the effectors NleE and Tir to subvert the NF-κB pathway. NleE directly methylates the cysteine residue in the ZnF of TAB2/3 to hinder their ubiquitination, whereas Tir facilitates the binding of host protein tyrosine phosphatases SHP-1 and SHP-2 to TRAF6, which prevents the ubiquitination of TRAF6.^{136, 137} Finally, EPEC and EHEC also target host neddylation via the Cif effector to favor their survival. Similarly, Burkholderia pseudomallei encodes a Cif homolog effector named CHBP. These Cif family effectors function as novel glutamine deamidases and prefer to deamidate NEDD8 rather than ubiquitin and thus impair its conjugation to Cullin, which is an essential subunit of the host CRL E3 ligase complexes.^{138, 139}

Other Gram-negative bacterial pathogens

Three bacterial DUBs called TssM, YopJ and YopP have recently been identified from B. pseudomallei, Yersinia pseudotuberculosis and Y. enterocolitica, respectively. These DUBs analogously have a part in suppressing host inflammatory signaling pathways. TssM targets TRAF3, TRAF6 and IκBα to cleave both K48- and K63-type ubiquitin chains, thereby limiting the IRF3 and NF-κB cascades.¹⁴⁰ YopJ and YopP are homologous to each other and both exert multiple functions during bacterial infection. YopJ targets host TRAF2, TRAF6, TRAF3, IκBα and TAK1, whereas YopP targets TRAF6, NEMO, TAK1 and TAB1 to cleave ubiquitin moieties, resulting in interference of the NF-κB and MAPK cascades.^{141, 142, 143, 144, 145} In addition, the mouse pathogen Citrobacter rodentium, which imitates EPEC, delivers the effector NleB to O-GlcNAcylate, the host glyceraldehyde 3-phosphate dehydrogenase, thus disrupting its interaction with TRAF2 and arresting the NF-κB pathway.¹⁴⁶

Listeria monocytogenes

Li. monocytogenes is an important Gram-positive intracellular bacterial pathogen that causes the human disease listeriosis. Typically, certain internalin family proteins on the surface of Li. monocytogenes, such as internalin A (InlA), InlB and InlC, function to disorder the host ubiquitin system for efficient bacterial immune evasion and survival.¹⁴⁷ Specifically, InlA promotes the internalization of Li. monocytogenes by targeting E-cadherin on epithelial cells for tyrosine phosphorylation and subsequent ubiquitination mediated by Src kinase and Hakai E3 ligase, respectively.¹⁴⁸ InlB triggers the autophosphorylation and activation of the host receptor tyrosine kinase, c-Met, resulting in the ubiquitin ligase c-Cbl-dependent ubiquitination of c-Met and facilitating Li. monocytogenes invasion.¹⁴⁹ Briefly, InlC directly interacts with IKKα to impair phosphorylation and thus delay the destruction of IκB to suppress the NF-κB pathway.¹⁵⁰ Another surface protein of Li. monocytogenes, ActA, functions as a recruiter for the actin-related protein 2/3 complex and Ena/VASP, thereby promoting bacterial evasion of the autophagic pathway.¹⁵¹ Notably, Li. monocytogenes also hijacks the host ubiquitin system by secreting a pore-forming toxin, listeriolysin O, which degrades the host SUMO E2 enzyme Ubc9, thereby decreasing SUMOylation and impairing downstream TGFβ signaling.¹⁵²

Mycobacterium tuberculosis

Mtb is an excellent example of a human bacterial pathogen that successfully survives in phagosomes and prevents their normal maturation after engulfment by macrophage cells.¹⁵³ Although Mtb encodes a structurally ubiquitin-like protein called RpfB and even possesses a prokaryotic ubiquitin-like protein system (Pup system) that is functionally analogous to the prototypical ubiquitin system, thus far no direct evidence has demonstrated the capacity of Mtb to resist host immune elimination by utilizing its ubiquitin-like degradation system.^{154, 155} This is likely because the highly unstructured Pup is unable to act as a post-translational modification or to mediate signaling pathways. Therefore, Mtb inevitably resorts to other tactics to disturb the host immune system. Recently, our research group first reported that PtpA, a secreted Mtb tyrosine phosphatase, subtly co-opts the host ubiquitin system to facilitate Mtb survival, which was methodically summarized by Chen.¹⁵⁶ Our data showed that Mtb PtpA is activated to directly dephosphorylate p-JNK and p-p38 by binding to the host ubiquitin via a brand new UIM-like region, resulting in the repression of innate immunity.¹⁵⁷ Furthermore, we observed that PtpA interferes with the association between TAB3 and K63-linked ubiquitin chains by competitively interacting with the NZF domain of TAB3, thus subverting NF-κB activation. Our discovery that Mtb commandeers the ubiquitin system provides novel insight into the Mtb–macrophage interaction and reveals potential pathogen–host interface-based targets for the treatment of tuberculosis.¹⁵⁸

CONCLUSIONS

Mounting evidence has highlighted the essential regulatory role of the ubiquitin system in host–pathogen interactions. We are, however, far from fully appreciating both the molecular mechanisms and the physiological functions of these processes. As more members of the host immune defense pathways are identified as being regulated by the ubiquitin system, we are faced with a challenging mission to dissect the underlying molecular details. Which bona fide protein targets, E3s or DUBs are responsible for the observed phenotypes? How are these E3s and DUBs regulated under both normal and infected conditions? Considering that a specific protein is, more often than not, modulated by several different E3s and/or DUBs, attention should be paid to clarifying whether these different enzymes function redundantly or whether they function distinctly in a cell type and context-dependent manner. Even after thorough characterization of these biochemical events, the importance of evaluating the physiological relevance of in vitro findings before drawing definitive conclusions about a mechanism cannot be overemphasized. Similarly, there are still many gaps in our comprehension of ubiquitin system-relevant pathogen immune-evasion strategies. How are these pathogen-derived E3s and DUBs regulated after injection into the host environment? How do those multifunctional effector proteins orchestrate their diverse activities in physiological situations? Significantly, such knowledge will provide informative clues for the development of effective therapeutic approaches as well as vaccine strains to curb infectious diseases.