*2.1. NEDDylation in Liver Fibrosis*

NEDDylation is a reversible ubiquitin-like PTM, characterized by the covalent conjugation of NEDD8. The pathway of the NEDDylation process involves NEDD8 specific enzymes, such as E1 activating enzymes (NAE1 and UBA3); E2 conjugating enzymes (UBE2M/UBC12 and UBE2F); E3 ligase enzymes, which catalyze NEDD8 transference to the target protein (MDM2, RBX1, FBXO11, RNF7, CBL, DCUN1D1, and DECUN1D2); and deneddylase enzymes (SENP8/NEDP1, ATXN3, USP21, CPS5, UCHL1, and UCHL3) [47,48]. Noteworthy, NEDD8 is synthetized as a precursor and must be activated at the C-terminal Gly76 mainly by NEDP1 [49] in order to be integrated inside the NEDDylation cycle and conjugated to the lysine residue of target proteins [50]. The conjugation of NEDD8 can modify its target protein in different ways, such as inducing conformational changes, changing its subcellular localization, enzymatic activation, or inhibition, competing with other Ubls or inducing its stability [51,52].

The mechanisms that trigger the deregulation of NEDDylation are not well understood, but it has been reported that the levels of NEDD8 are increased under stress conditions in vitro [53]. In fact, alterations in the NEDDylated protein levels have been described in different pathological conditions, such as neurodegenerative disorders [54] and cancer [48,55,56]. Focusing on the liver context, patients with HCC and intrahepatic cholangiocarcinoma, as well as mouse models of HCC, showed a significant increase in the global NEDDylation proteome and NEDDylation intermediates [55–59]. In addition, under diverse stress conditions, the canonical pathway of NEDDylation via NAE1 changes, being NEDD8 conjugation predominantly mediated by the Ube1 E1 ubiquitin enzyme [53]. Likewise, in HCC, where NEDDylation levels are enriched, NEDP1 protein levels disappear promoting the inhibition of ATPase activity of HSP70 and, thus the apoptosis resistance of cancer cells. Hence, these result shows how the tight regulation of the NEDD8 cycle can modulate vital cellular functions like apoptosis [60].

Regarding liver fibrosis and NEDDylation, Zubiete-Franco et al. described for the first time an increase in the global NEDDylated proteome in patients with liver fibrosis as well as in mouse models of CCl4- and BDL-induced liver fibrosis [61]. Importantly, NAE1-specific inhibition in these mouse models showed a reduction in the liver damage associated with decreased apoptosis, inflammation, and fibrosis. These results were explained by the effect of NEDDylation inhibition in the different hepatic cell subtypes. The decrease in inflammation after NEDDylation inhibition can be explained in part by the incapacity of Cullin-1 and SCFβTrCP (E3 Ligase) to ubiquitinate and degrade IKBα, promoting NF-kB stabilization in the cytoplasm [47,48]. Interestingly, in this work the authors describe how NEDD8 levels increase in activated HSCs, and consequently neddylation inhibition could directly block its activation. Indeed, after NAE1 inhibition, HSCs show an increase of cell death partly mediated by c-Jun accumulation, a target of cullin degradation. On the other hand, it has been described that Casitas B-lineage lymphoma (c-Cbl) acts as a NEDD8 Ligase promoting TGF-β signaling and stabilization of the type II receptor (TβRII) in blood cells [62]. In agreement with this line of evidence, other authors have shown very recently that the in vivo inhibition of the transcription factor SRSF3 NEDDylation, associated with its prevention of degradation, protects mice from fibrosis [63].

In conclusion, the NEDDylation inhibition is a key mechanism to down-regulating the inflammatory response, further reducing cell damage and subsequent liver fibrosis, in addition to specifically targeting HSC death.

### *2.2. SUMOylation in Liver Fibrosis*

SUMOylation is another ubiquitin-like PTM that consists in the covalent addition of one or multiple SUMO subunits to Lys residues usually located on the SUMO consensus motifs of target proteins. SUMOylation occurs as a hierarchically organized process catalyzed by the E1 activating enzyme, the E2 conjugating enzyme, and an E3 SUMO ligase [64]. The extension of the SUMO chain is possible thanks to a specialized type of E3 ligase family of enzymes known as E4 SUMO elongases [65].

To date, five SUMO isoforms have been described in humans, being SUMO 1, 2, and 3 the most ubiquitous. SUMO modifiers are similar in size and structure to ubiquitin, but show little sequence homology compared to ubiquitin. SUMO 2 and 3 share approximately 97% identity, whereas SUMO 1 is only 50% identical in sequence. SUMO isoforms differ in several aspects, such as in the E3 ligase preference or the ability to form SUMO chains on the substrate proteins. Moreover, different functions and mechanisms of regulation within the cell would be expected since SUMO2/3 conjugation becomes more relevant under stress conditions [64]. The SUMO E1 activating enzyme is composed by the SAE1 and UBA2 heterodimer, while Ubc9 is the only E2 SUMO conjugating enzyme recognized. Conversely, a huge range of E3 SUMO ligases exist, which are grouped in the canonical PIAS family and non-canonical E3 ligases such as RanBP2 or Cbx4, thus conferring specificity to the process [64]. SUMO-mediated modification can be reversed by the action of deSUMOylating enzymes, which are also involved in the maturation of the SUMO precursor protein. SENPs belong to the most common family of protein deSUMOylases but, unrelated DESI1, DESI2, and USPL1 SUMO proteases exist as well [66]. Since SUMOylation is mostly restricted to the nucleus, it is not a surprising fact that SUMO is involved in many nuclear processes such as DNA damage response, genome integrity, transcription regulation, as well as preservation of protein stability and modulation of subcellular localization of the substrate proteins [67,68].

SUMOylation is a highly dynamic process enabling fast global changes in the SUMO status of the proteome in response to internal and external stimuli, often stress such as heat shock, nutrient depletion, genotoxic or oxidative stress [69–72]. This rapid adaptation is possible thanks to several mechanisms of regulation that can control SUMOylation levels. In addition to deSUMOylases, the SUMO-targeted ubiquitin ligase (STUbl) enzymes can modify global SUMOylation levels by binding to SUMO chains on proteins and poly-ubiquitinating them, eventually leading to their proteasome-mediated degradation. Moreover, a crosstalk between SUMOylation and other PTMs, such as ubiquitination or phosphorylation, has also been reported to affect the SUMOylation status [73,74]. The localization of the SUMO enzymatic machinery constitutes an additional critical factor for the modulation of the SUMOylation levels [64].

Hence, controlled SUMOylation is required for normal cell behavior. According to proteomics studies, between 1000 and 3000 human proteins are modified by SUMO. The identified SUMOylated proteins are implicated in almost all cellular processes [66]. A deregulation in SUMOylation dynamics has been associated with fibrotic disorders occurring in the heart, lung, and kidney, amongst other diseases [75–77]. And there is increasing evidence that SUMOylation might play a regulatory role in liver fibrosis too [78–80].

A recent study referred to Ubc9, the only existing SUMO E2 conjugating enzyme, as a potential therapeutic target for the prevention and treatment of liver fibrosis. Protein and mRNA expression levels of Ubc9 were described to be significantly upregulated in the LX-2 liver fibrosis in vitro model, and in the HepG2 and SMMC-7721 HCC cell lines. Interestingly, shRNA-mediated silencing of Ubc9 expression in activated LX-2 cells resulted in a decreased expression of α-SMA and type I collagen fibrosis markers, as well as a diminished secretion of IL-6 and TNF profibrotic cytokines. Additionally, downregulation of Ubc9 blocked cell cycle progression and promoted activated LX-2 cell cycle arrest in G2 phase. Importantly, an induction of apoptosis in activated LX-2 cells was detected after Ubc9 expression knockdown, mainly attributed to the abrogation of the canonical NF-κB signaling pathway, which is also a known target of SUMOylation [78].

Another piece of work placed the deSUMOylating enzyme SENP2 as a critical protein to attenuate CCl4-induced liver fibrosis in mice by inducing activated HSC apoptosis via suppression of Wnt/β-catenin signaling program. SENP2 protein and mRNA expression levels were found to be decreased both in vitro and in vivo in activated hepatic stellate cells (HSCs) during the CCl4-induced liver fibrosis mouse model, being those levels restored after removal of the damage stimulus. On the one hand, in vitro SENP2 overexpression resulted in a decreased α-SMA and COL1A1 protein expression in a TGF-β-activated hepatic stellate cell line. Moreover, increased expression of SENP2 reduced cell viability, favored cell cycle arrest in G0/G1 phase and induced apoptosis of the in vitro TGF-β-activated HSCs. On the other hand, siRNA-mediated silencing of SENP2 in TGF-β-activated HSCs induced α-SMA and COL1A1 protein expression, stimulated cell proliferation, and reduced apoptosis. Finally, the expression of the Wnt/β-catenin pathway members was downregulated upon SENP2 overexpression in TGF-β-activated HSCs, thus suggesting a therapeutic role of SENP2 in liver fibrosis [79].

Although it has not been specifically studied in the context of liver, various members of the TGF-β/Smad canonical pathway, which is common to fibrotic processes, have been found to be SUMOylated [66]. TGF-β type I receptor (TRβI/ALK5), whose phosphorylation and activation are mediated by TGF-β, is SUMOylated further enhancing the activation and modulation of the downstream Smad signaling cascade [73]. Furthermore, TGF-β signal transducers Smad proteins are also postranslationally modified by SUMOylation. For example, Smad4 SUMOylation protects it from its ubiquitination and subsequent proteasomal degradation [81]. Interestingly, Smad nuclear interacting protein 1 (SNIP1), a transcription repressor for both TGF-β and NF-κB signaling pathways, is a SUMO substrate. SNIP1 inhibits the TGF-β signaling by hampering the recruitment of p300 coactivator to the Smad complex, whereas SNIP1 SUMOylation attenuates its inhibitory effect on the TGF-β response further facilitating the expression of PAI-1 and MMP2 [82]. In summary, it is suggested that interfering in the SUMOylation of these proteins could be a potential strategy for the treatment of diseases induced by aberrant TGF-β signaling, which not only includes liver fibrosis but also HCC. Nevertheless, more focused research is needed regarding the impact of the TGF-β/Smad pathway SUMOylation in the particular context of liver fibrosis.

Conversely, a study highlights the importance of SUMOylation for liver fibrosis regression. Reduced glutathione (GSH) is implicated in many cellular processes including fibrogenesis. GSH protects against oxidative stress, which activates HSCs. Thus, high levels of GSH would maintain HSC in a quiescent state, and this requires SUMOylation of Nrf2 and MafG, which facilitate heterodimerization and activation of the antioxidant response element (ARE) located in the promoter region of many genes involved in the antioxidant defense, such as the GSH synthetic enzymes [80].

Finally, it has also been demonstrated that SUMO 1 and SUMO2/3 could play a role as autoantigens during PBC, since autoantibodies to these proteins have been detected in the sera of patients suffering from this autoimmune disease. Nonetheless, further research is needed in order to understand how the development of SUMO autoantibodies can lead to autoimmunity in PBC [83].

Overall, SUMOylation is a highly dynamic process which can have both beneficial and pathological consequences in the cellular physiology depending on the protein substrate, cell type, or context. Therefore, inhibition of global SUMOylation might not always be an ideal therapeutic strategy due to potential unforeseeable secondary effects. Alternatively, a more realistic rationale would involve the discovery and development of small molecules or peptidomimetics that block the protein–protein interactions between specific E3 SUMO ligases or SENPs and their substrates that are known to be altered in a diseased state.
