The role of post-translational modifications by ubiquitin/ISG15 tags in exposed proteins during NETosis: Immunogenic or tolerogenic signaling process in Systemic Lupus Erythematosus?


 Background: Neutrophil extracellular traps (NETs) from patients with Systemic Lupus Erythematosus (SLE) are characterized by lower ubiquitylation and myeloperoxidase (MPO) as a substrate. The structural and functional effect of such modification and if there are additional post-translational modifications (PTMs) are unknown.Methods: To assess the expression and functional role of PTMs in NETs of patients with SLE; reactivation, proliferation and cytokine production was evaluated by flow cytometry using co-cultures with dendritic cells (DC) and CD4+ from SLE patients and healthy controls. The impact of ubiquitylation on MPO was assessed by molecular dynamics. The expression of ISG15 in NETs was evaluated by immunofluorescence and Western Blot.Results: Fifteen patients with SLE and 10 healthy controls were included. In the co-cultures of CD4+ lymphocytes with DC stimulated with ubiquitylated MPO or recombinant MPO, a higher expression of IFNγ and IL17A was found in CD4+ from SLE patients (p < 0.05). Furthermore, with DC stimulated with ubiquitylated MPO a trend towards increased expression of CD25 and Ki67 was found in lupus CD4+ lymphocytes, while the opposite was documented in controls (p < 0.05). Through molecular dynamics we found the K129-K488-K505 residues of MPO as susceptible to ubiquitylation. Ubiquitylation affects the hydration status of the HEME group depending on the residue to which it is conjugated. R239 was found near by the HEME group when the ubiquitin was in K488-K505. In addition, we found greater expression of ISG15 in the SLE NETs vs controls (p < 0.05), colocalization with H2B (r = 0.78) only in SLE samples and increased production of IFNγ in PBMCs stimulated with lupus NETs compared to healthy controls NETs.Conclusion: The ubiquitylated MPO has a differential effect on the induction of reactivation and proliferation of CD4+ lymphocytes in patients with SLE, which may be related to structural changes by ubiquitylation at the catalytic site of MPO. Besides a lower ubiquitylation pattern, NETs of patients with SLE are characterized by the expression of H2B conjugated with ISG15, and the induction of IFNγ by Th1 cells.


Introduction
Neutrophil extracellular traps (NETs) play a pathogenic role in diverse autoimmune diseases, including systemic lupus erythematosus (SLE) (1). These NETs are brillar mesh structures decorated by nuclear chromatin and neutrophil granular peptides (i.e. myeloperoxidase, elastase, lactoferrin and LL-37) and are one of the most potent tools to combat microorganisms (2). Likewise, these structures are a potential source of autoantigens, whose externalization might lead to self-tolerance breakdown. Indeed, enhanced NETosis, diminished clearance or posttranslational modi cations in the protein cargo of these NETs are related to tissue damage in SLE (3). Diverse stimuli have been acknowledged, including those related to microorganisms, such as LPS, in ammatory stimuli (ie ROS, TNF-α and IL-8) as well as those considered as sterile, such as the antigen-antibody complexes (4). Indeed, Petretto A,et al (4) showed by proteomic analysis a differential protein cargo and post-translational modi cations, dependent upon the induction stimuli for NETosis, Interestingly the majority of modi ed peptides are derived from myeloperoxidase (MPO), which might be responsible for diverse biologic effects (4). This enzyme catalyzes the transformation of hydrogen peroxide in hypochlorous acid, hence MPO has a bactericidal function. MPO is key for the NETosis induction, since it acts synergistically with neutrophil elastase (NE) for the degradation of membranes and chromatin decondensation (5). Furthermore, MPO has been acknowledged as a dual modulator of immune responses, particularly when it is located at the extracellular milieu, where it is able to induce tissue damage mediated by increased oxidative stress (6). At the cellular level, MPO induces T cell proliferation in a dose dependent manner in antineutrophil cytoplasmic antibodies (ANCA) associated vasculitis and healthy donors (7), as well as having an immunosuppressive role, mediated by IL-10 (8) and by dendritic cell suppression through reduced CD86 and IL12 expression (6).
Post-translational modi cations (PTMs) have been related to NETosis induction and its pathogenic role in autoimmune diseases has been acknowledged, since it implies a mechanism for neoantigen expression. Indeed, citrullination of histone 2B during NETosis is a key element for pathogenic autoantibody responses in SLE (9). Furthermore, NETs from SLE patients are characterized by a differential PTM pro le, including histone hyperacetylation (10) and a de cient polyubiquitilation (K48/K63) pro le (11), which induce pro-in ammatory macrophage responses. Among PTMs with potential regulatory role in autoimmune pathogenic responses, the ISGylation, de ned as the conjugation of the ubiquitin-like-protein ISG15; is related to enhanced type I IFN responses (12) besides that, extracellular ISG15 also promotes the synthesis of IFNγ by T and B cells. Additionally, NETs are potent inducers of IFNα/β by plasmacytoid dendritic cells in SLE (13) and type I IFN responses are the hallmark of the molecular signature in SLE. Nonetheless, it has not been addressed if ISGylation is present in NETs from SLE patients as a potential source of ISG15 and a regulator of IFN responses.
Evidence suggests that PTMs, particularly ubiquitylation and ISGylation might play a role for enhanced NETosis in SLE and particularly, MPO was shown to be target of this PTM in NETs from SLE patients and healthy donors (11). However, the precise amino acid residue where ubiquitin tag is added is currently unknown, as well as its potential impact in cellular responses. Therefore, the aim of the present work is to address the role of PTMs related to ubiquitin and ubiquitin-like-protein, ISG15 from NETs and their effects in the regulation of cellular responses in SLE.

Materials And Methods
Patients and controls. We recruited 15 Mexican-mestizo adult patients with active SLE (SLEDAI > 6) according to the ACR criteria (14), who were followed-up in a tertiary care center (Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán). As a control group, ten age and gender matched healthy subjects were included. Exclusion criteria were applied to those patients with any kind of acute or chronic infection, pregnancy, puerperium and neoplasia. All healthy controls and SLE patients signed an informed consent before inclusion, and the protocol was approved by our Institutional Ethics Committee (Ref. 2152) in compliance with the Helsinki declaration. Laboratory features of SLE patients were assessed as follows: Auto antibodies of SLE patients (anti-dsDNA IgG, anti-nucleaosomes, anti-cardiolipin and anti-β2glycoprotein I) were assayed using a commercial ELISA.
In vitro ubiquitylation of MPO. A commercial Ubiquitylation Kit (Abcam™) (HeLa lysate) was used according to the manufacturer's instructions. The recombinant human MPO (rhMPO 10 µg) was incubated for 4 h at 37 ° in a water bath and stored at -80°. To con rm that the ubiquitylation reaction took place, a 4-15% polyacrylamide gel electrophoresis was performed with 50 µg of ubiquitylated MPO (UbMPO) and rhMPO. Subsequently, the transfer to a PVDF membrane was carried out for 1 h at 100v, then, the non-speci c binding sites were blocked with Starting Block solution (Thermo Fisher ™).
Overnight incubation was carried out with the primary anti-ubiquitin antibody (P4D1 Santa Cruz™) and followed by the secondary antibody HRP anti-Mouse for 1 h (Thermo Fisher™). Membranes were developed by enhanced chemiluminiscence with ECL Western Blotting Substrate (BioRad™). Samples were acquired with a digital image analyzer (Chemidoc MP™, Biorad™) and quanti ed by densitometry with ImageLab software (Biorad™). ( Figure S1).
CD4 + T cell isolation and CD4 + /MoDC cocultures: Both lupus and control CD4 + T cells were also isolated from total PBMCs by magnetic negative selection with microbeads according to the Human CD4 + Isolation Kit manufacturer instructions (Miltenyi Biotec™). CD4 + /MoDC cocultures were performed in a 10:1 ratio at 37 °C, and 4% CO 2 for 3 days to evaluate proliferation and for 7 days to evaluate reactivation and cytokine production from CD4 + T cells. The cocultures were set up as follows: a) Unstimulated CD4 + cells; b)CD4 + cells plus DCs activated only with LPS; c) CD4 + cells plus DCs activated with LPS and rhMPO; d)CD4 + cells plus DCs activated with LPS and ubiquitylated MPO; and e)CD4 + cells polyclonally stimulated with 50 ng/mL of αCD3 (HIT3a BD™) and 100 ng/mL of αCD28 (NA/LE BD™).
CD4 + T cell proliferation and activation markers assessment by ow cytometry: For the assessment of activation markers, after washing twice with 5% FBS in PBS and incubated with FVS700 to evaluate viability, PBMCs were stained with the following uorescent surface labelled-antibodies: (CD4-APC, CD25-PE, CD83-PE/Cy5, HLA-DR-PerCP) (All Bio Legend™). After that, the cells were xed with True-Nuclear™ 1X Fix Concentrate, then washed with FoxP3-Perm Buffer™ and nally incubated with labelled anti-FoxP3 BV421. For the proliferation assay, the same steps described above were performed and after xation and permeabilization, PBMCs were incubated with anti Ki67-APC (Bio Legend™) instead.
For intracellular cytokine staining, PBMCs were stimulated with PMA (Sigma™), ionomycin from Streptomyces conglobatus (Sigma™) and incubated with monensin (GolgiStop™, BD Horizon™) at 37ºC for 5 h. After washing twice with 5% FBS in PBS, PBMCs were stained with anti CD4-AF488 and then xed and permeabilized with Cyto x/cytoperm (BD™), and nally incubated with intracellular anti IFNγ-APC (Th1), IL4-PE (Th2), IL17A-BV421V (Th17) (Bio Legend™) ( Figure S2). Molecular Dynamics Methods. To assess the initial positions of all-atoms of the backbone of the MPO were taken from the cryogenic crystal structure isoform C, PDB:1CXP, chains A and C (17), while the ubiquitin coordinates were taken from the crystal structure, PDB:1UBQ, chain B (18). To approach the post-translational modi cations of MPO, the solvent accessible surface area (SASA) and the root mean square deviation (RMSD) t of MPO at lysine residues with K6, K48 and K63 of the ubiquitin, were used as criterium to identify exposed lysines. Larger values of SASA were used to identify susceptible residues for ubiquitylation (an enzymatic reaction that its de ned by an isopeptide bond among the amino group of substrate lysine chains and the carboxyl terminus G63 of ubiquitin).
To the generation of simulation trajectories, MPO was de ned as the center in the simulation box and the water molecules were added via the solvate plugin of VMD 1.9.1 program (19), lling the space around the protein after a 2 Å cutoff. To de ne the physiologic electrolyte concentration of NaCl at 0.15 M we used the ionized VMD plugin (19). Every system was committed to energy minimization for 10 K steps of conjugate gradient algorithm. Later, the crystallographic structure was relaxed in steps of 200 ps each, with positional constraints setting force constants of 25, 15, 10 5 3, 2 ,1 kcal/mol Å. After preequilibration, all simulation trajectories were produced with the NAMD 2.12 program (20). Con gurations in the isothermal-isobaric ensemble, at 300K and 1 bar conditions, were generated using Langevin dynamics scheme to maintain constant temperature (20), and a Nose-Hoover Langevin piston for pressure control (21,22). For computational e ciency, a multiple time-step integration scheme was set, with 1 fs for bond forces, 2 fs for short-range non-bonding interactions, and 4 fs for full electrostatic forces (23,24). Coulomb interactions were computed using particle-mesh smooth Ewald summation (25,26) with a tolerance of 10 − 6 for the direct part of the Ewald sum, a fourth-order interpolation scheme, and a grid spacing of ~ 1 Å, for each direction of the simulation box. All bonds for hydrogen atoms were constrained using the SHAKE algorithm (27). All-atom force eld parameters were used in this study. The CHARMM 36 force eld (28)(29)(30) with CMAP correction for the protein atoms (31), and the TIP3P water model (32). In house analysis scripts were developed for speci c calculations, such as hydration level at the active site, heme group contact maps. ISG15 detection in ex vivo NETs by Western Blot. After density gradients, neutrophils were isolated with dextran sedimentation as previously described (33). We assessed the spontaneous NET formation (without stimuli) and lipopolysaccharide (LPS)-induced NETosis with 1 µg/mL E. coli O111:B4 LPS (Sigma Aldrich ™) for 4 h. Subsequently, 300 µL of RPMI with micrococcal nuclease (0.01U/µL) were added, the supernatant was centrifuged and stored at -20 °C. The protein was quanti ed using the bicinchoninic acid method at a wavelength of 562 nm. 4-15% polyacrylamide gel electrophoresis was performed on the lysates from the NETs of patients with SLE and healthy controls. Subsequently, the transfer was carried out for one hour at 100v to a PVDF membrane. Nonspeci c binding sites were blocked with the Starting Block Thermo Fisher™ solution. The membrane was then incubated overnight with anti-ISG15 (Abcam ™), after 3 washes with TBS-tween, the secondary antibody HRP anti-Rabbit (Thermo Fisher™) was added for 1 h. Membranes were developed by enhanced chemiluminiscence with ECL Western Blotting Substrate (BioRad). Samples were acquired with a digital image analyzer (Chemidoc MP, Biorad) and quanti ed by densitometry with ImageLab software (Biorad). Data was expressed as normalized expression to MPO.
Assessment of ISG15 in ex vivo NETs by confocal microscopy. Neutrophils were incubated in RPMI without phenol red (Thermo Fisher ™), 1% FBS, and 1% 10 mM 4-(2-hydroxyethyl)-1piperazineethanesulfonic acid (HEPES) buffer. For confocal microscopy, neutrophils were seeded in 0.01% poly-L-Lysine (Sigma-Aldrich ™) coated coverslips at 37ºC during 1.5 h. After xation with 4% paraformaldehyde (Santa Cruz ™) (33) at 4ºC during 24 h, washing and blocking with 0.2% gelatin from porcine skin (Sigma-Aldrich ™) was performed. For indirect immuno uorescence, we used the following primary and secondary antibodies diluted in 0.2% gelatin from porcine skin: rabbit anti-human ISG15 (Abcam™), mouse anti-human H2B (Abcam™), donkey anti-rabbit Alexa Fluor 555 (Thermo Fisher™) and donkey anti-mouse Alexa Fluor 488 (Thermo Fisher™). Chromatin was stained with Hoechst 33342 (Thermo Fisher™) and coverslips were mounted on slides with ProLong® Gold Antifade Mountant (Thermo Fisher™) (34). The samples were acquired in an Eclipse Ti-E Nikon confocal microscope (Minato, Tokyo, Japan). The number of cells positive for H2B and nuclear staining (Hoechst) were considered producers of NETs. The mean uorescence intensity and the area percentage of NETs were quanti ed using Fiji free software (ImageJ ™) using the average of 6 elds at 40x, normalized for the total number of cells analyzed. The colocalization of ISG15 with histone 2B (H2B) was calculated by Pearson correlation coe cient and validated through the Costes method (35).
Statistical analysis. Quantitative variables were expressed as median with interquartile range (IQR) or minimum and maximum (min-max). The homogeneity of variances of each experiment was determined using the Brown-Forsythe and Bartlett tests. The non-parametric data were analyzed with the Kruskall-Wallis test and the adjustment was made using the Dunn multiple comparison test, as well as Mann-Whitney U test (sum of Wilcoxon ranges) to compare medians between groups with free software Past 3 version™ (United States). Pearson's linear correlation coe cient was used (36), and signi cance was veri ed using the Costes method (35). In all cases, signi cant differences were considered fora p value lower than 0.05. Statistical analysis and graph generation was performed with support of the GraphPad Prism 8 version ™ software (United States).

Results
Demographic and clinical features of patients with SLE. Table 1 shows the main characteristics of the fteen patients with active SLE included (SLEDAI > 6), of which 80% were women and 20% men. The median age was 27.18 years (19-35 years). The disease activity was measured using the SLEDAI score (37) with a median of 20 points (8-24), 90.0% had constitutional activity, 82% renal activity, 72.7% hematological activity and 63.6% mucocutaneous activity at the time of the blood draw. All had positive anti-nuclear antibodies in different patterns, 86.6% had anti-dsDNA IgG in positive titers with a median of 142 IU/mL (2.1-485 IU/mL), 33.3% had anti-nucleosomes with a median of 209 U/mL (11.4-660 U/mL). 86% of patients had hypocomplementemia. All patients used immunosuppressive treatment at the time of sampling. UbMPO induces a differential proliferation and activation pro le as well IFNγ and IL17A production. By means of intracellular cytokine staining, increased production of IFNγ (Panel A Fig. 1) and IL17A (Panel B Fig. 1) was found in SLE CD4 + T lymphocytes co-cultured with dendritic cells activated with rhMPO and UbMPO compared to the group in which the co-culture included DC activated only with LPS (p = 0.0286).
In SLE patients, a trend towards increased proliferation was observed only in the CD4 + lymphocytes stimulated with UbMPO/LPS in comparison with both, the lymphocytes activated with rhMPO/LPS (UbMPO/LPS vs rhMPO/LPS p = 0.1143) and with those activated with LPS in the absence of MPO (UbMPO/LPS vs LPS p = 0.1143) (Panel C Fig. 1).  Table 1).
Ubiquitylation of MPO is related to HEME group modi cations: The HEME group was attached to D94 of chain α, and E242 of chain β, using an ester bond among methyl groups of catalytic site and carboxyl side chains of the subunits (Panel A Fig. 2). Also, owed to its proximity to the HEME active site, the D98 was protonated, as suggested by Carpena, et al (38). We found three lysine: (K129-MPO, K488-MPO, and K505-MPO) as the best candidates for ubiquitylation and prepared four systems for detecting structural and conformational changes made by ubiquitylation (Panel B Fig. 2). Contact Maps at the active site. Through the analysis of relative frequencies in the contact map of the residues close to the HEME group, with a cutoff point for bond distance < 2 Å (Panel C of Fig. 2); differential pro les were found depending on the ubiquitin binding site to the MPO. In this way, R239 was found near the HEME group when ubiquitin was at positions K505 or K488 (Panel D Fig. 2).
Hydration of the active site. To investigate the dynamics of the water molecules in the active site, we measure the number of water molecules at 3 Å from the HEME group. Interestingly, ubiquitylation seemed to disrupt the amount of water in the active site (Panels E, F, G, H. Figure 2). Ub-K129-and Ub-K488-MPO showed an increased amount of water molecules in the active site relative to the MPO, while Ub-K505MPO decreased the amount of water close to the HEME group. This observation could be relevant to the enzymatic activity, as the substrates are chloride ions and hydrogen peroxide, which could increase the amount of water in the active site.
NETs from SLE patients are enriched in ISG15. We found a differential ISG15 expression characterized by higher amounts of ISG15  Fig. 3).
NETs are an extracellular source of ISG15 and H2B is one of the targets in SLE ISG15 was found both intra and extracellularly in the NETs of patients with SLE by confocal microscopy (Panel A Fig. 4), but not in the NETs of healthy controls (Panel B Fig. 4). Accordingly, we found increased expression of ISG15 in lupus NETs (both spontaneous and LPS-induced) compared to healthy controls (p = 0.0179) (Panel C, Fig. 4).
The colocalization analysis was performed in the area considered as a NET, by immuno uorescence for ISG15 and three potential substrates, H2B, HMGB1 or LL-37. We found that H2B showed the higher colocalization (r = 0.81) (Panel D Fig. 4) Enriched ISG15 NETs from SLE patients induce IFNγ production from diverse cellular subsets. To address the functional impact of ISG15 on the NETs from lupus patients, the production of IFNγ in subsets of T lymphocytes (CD4 + and CD8 + ) and NK cells was evaluated by multiparametric ow cytometry. Higher IFNγ production was found in total PBMCs that were stimulated with NETs from lupus patients that previously had shown to be enriched by ISG15 (Panel B Fig. 5) compared to cells stimulated with NETs from healthy controls (Panel C Fig. 5) (%IFNγ 6.5[6.31-10.1] vs 3.5[1.7-5.1]) p < 0.05). Figure 5 shows a higher production of IFNγ by stimulating different subsets of PBMCs with ISG15 enriched NETs from SLE patients, compared with NETs of healthy donors, lacking such post-translational modi cation, including CD4 + lymphocytes (Panel F Fig. 5), CD8 + lymphocytes (Panel G Fig. 5), and NK CD56 + lymphocytes (Panel H Fig. 5).

Discussion
In the present work, we demonstrated the presence of PTMs related to ubiquitin tag (total ubiquitin and ISG15), in NETs from SLE patients, and their impact in the reactivation, proliferation and polarization of CD4 + cells of SLE patients and controls. Furthermore, to our knowledge this is the rst report on the differential patterns of water molecules and conformational changes in residues near the catalytic site of MPO conjugated to ubiquitin. Interestingly, our data also suggest NETs as a source of ISG15 in SLE and shows that H2B is a main target of this PTM and its functional role in regulating IFNγ secretion by T lymphocytes in SLE.
The trend towards increased proliferative responses of CD4 + T lymphocytes from patients with SLE, as well as the polarization towards Th1 and Th17 subpopulations producing IFNγ or IL17A in the presence of components of NETs such as recombinant MPO and ubiquitylated MPO is consistent with a previous study in which CD4 + and CD8 + T lymphocytes from controls stimulated with supernatants of NETs induced in vitro, had higher levels of CD25, CD69, phosphorylation of ZAP70, secretion of IFNγ and IL17A; however, there was no increase in the percentage of proliferation with ethinyl deoxyuridine (EdU) (39). This difference in proliferation could be related to a differential threshold for activation and proliferation signals, which could be reached by the NET components and therefore able to induce the production of effector cytokines such as IFNγ and IL17A, but insu cient to fully activate T cell proliferation. In agreement to this, other studies have described proliferation of PBMCs from patients with vasculitis associated with anti-neutrophil cytoplasm antibody (ANCA) by increasing MPO concentrations to 10 µg/mL (7,40), which corroborates the requirement for a high concentration of MPO for the regulation of proliferation in T lymphocytes. Furthermore, the ability to produce IFNγ by PBMCs from patients with ANCA vasculitis at the dose we used for the activation of dendritic cells (5 µg/mL), had been previously described by ELISA and this production was not carried out when using lower doses (0.5 µg / mL) (8). This is the rst study addressing the biological effect of the ubiquitin tag in MPO in the context of systemic autoimmunity. Interestingly, the proliferative response of CD4 + lymphocytes depends on the ubiquitin tag as well as the disease state (ie healthy controls vs SLE). It has been described that the type of response that is mounted (proin ammatory vs anti-in ammatory) is mainly related to the context in which the antigenic determinant (immunogen or tolerogen) is seen, highlighting the microenvironment, the PTMs and the disease state as key elements for the functional effect of every antigen (41). It is feasible that the relationship of ubiquitin with a lower activation and proliferation of CD4 + T cells in controls is due to its ability to mask potentially dangerous antigens for the system (such as MPO in NETs) (42). Nonetheless, this relationship is in turn unbalanced in the context of SLE where both MPO with and without ubiquitin was able to induce polarization of CD4 + T cells towards a Th1 and Th17 effector phenotype. This is in agreement with previous data from our research group suggesting that lupus patients have a ubiquitin de ciency state involving diverse cellular subsets and processes, in which they express lower amounts of E3 ligases (key enzymes for ubiquitylation) in T cells (43,44) as well as diminished polyubiquitin K63 chains in NETs (11),, both with functional impact, such as resistance to anergy and altered macrophage responses, respectively. Accordingly, in healthy controls, the presence of ubiquitin tag involves a suppressive mechanism when it is bound to a molecule with a dual role in in ammatory signaling, such as MPO, which might be diminished in the context of systemic autoimmune diseases, such as SLE, explaining the shift towards Th1 and Th17 in CD4 + cells in patients and the tolerogenic role in controls, diminishing the activation and proliferative responses.
The molecular dynamics simulations allowed to describe the closeness of arginine 239 (R239) when ubiquitin was bound to K505 or K488 compared to native MPO. R239 is an important residue of the polypeptide chain lying near the distal face of the HEME group (45) and has been involved in the catalytic mechanism of MPO (46). There is evidence using molecular docking that the LGM2605 peptide, by its hydroxyl group, inhibits the enzymatic activity of MPO by blocking R239. The authors propose that while this position may not directly displace H 2 O 2 binding, blocking the substrate channel may slow H 2 O 2 access to HEME (47). Our results using contact map analysis allowed us to nd that R239 decreased its distance from the HEME group when ubiquitin was conjugated to K505 or K488 compared to native MPO.
This implies a lower probability of entry to the active site by the reaction substrates (Cl − and H 2 O 2 ).
One of the most relevant changes in the structural dynamics of the MPO was the differential hydration of the active site mediated by the ubiquitinated lysine residue. Water molecules are relevant to receptorligand recognition because they modify the geometry of the active site and contribute to binding a nity.
Our results propose that when ubiquitin conjugates to K129 or K488, the number of water molecules close to the HEME group increases, and the opposite occurs when ubiquitin binds to K505. This differential pattern has been demonstrated in states of activation of receptors associated with G proteins where the role of water molecules is relevant to de ne the direction and magnitude of the movements of these proteins (48). It has been reported that in squid rhodopsin, the conformation of its active state is accompanied by an increase in the number of internal water molecules (IWM) and that inter-helix water molecules directly affect water-mediated interactions in the activation process (49). In a model of recognition of vasopressin and oxytocin receptors, it was found that very few water molecules interact with the D2.50 domain of the active site and appear to be a fundamental to stabilize their structure and participate in the binding of ligands (50), which could be similar to what was found in the MPO, highlighting the potential functional translation of the conformational changes related to the ubiquitin tag.
Furthermore, the evolution of structures is consistent with functional speci city, and ubiquitin-like proteins are an obvious example of the pleiotropism inherent in their function, as is the case with the 15 kDa IFN-related gene. ISG15 is another PTM that has a great diversity of functions depending on its location. Recently it has been identi ed the lymphocyte function-associated antigen 1 (LFA-1) or CD11a/ αL as the putative receptor of ISG15 in NK cells inducing the activation of Src kinases and IFNγ and IL10 synthesis (51). However, currently, the release mechanism is unknown, secretory granules of neutrophils, lysosomes, exosomes, apoptotic bodies, and microvesicles released from infected macrophages have been suggested as probable mechanisms (52). Based on our results, we can propose a new noncanonical secretion mechanism of ISG15 by NETosis in patients with SLE. To our knowledge, this is the rst report of extracellular ISG15 as a component of NETs only in SLE patients, highlighting its role in this disease. It is feasible to propose NETosis as one of the non-canonical secretion mechanisms of extracellular ISG15, since the increased levels of type I IFN, and the molecular transcriptional signature of SLE are suggested to be key pathogenic players in the disease. (13) (53) In cells that respond to interferon, ISG15 as conjugated or free, the latter with "cytokine-like" activity with evidence supporting its ability to induce the production of IFNγ as well (54,55). Additionally, controversy exists about the pathogenic role of ISG15, since the overexpression of ISG15 in NETs of SLE patients could be both a consequence of the overstimulation by the IFNα/β signature, and a signal from NETs involved in the production of IFNγ by lymphocyte subpopulations (56,57). Our data show that indeed NETs from SLE patients are capable of inducing IFNγ production in CD4 + T lymphocytes as has been previously shown by Tillack KB, et al (39).
Finally, we show that the protein cargo of NETs from SLE patients is enriched in ISG15 and able to induce IFNγ production, which agrees with previous data by Iglesias-Guimarais, et al (58) which show that in addition to CD4 + lymphocytes, IFNγ production was also reported by other subpopulations such as CD8 + and NK cells, both known targets of action of extracellular ISG15 (58), which is also produced by lupus plasma cells (57). Therefore, it can be argued that NETs promote the activation of these cells with the subsequent production of IFNγ in response (among other stimuli) to free ISG15. (59,60) Thus, we propose ISG15 in NETs as a new component in the physiopathogenic scheme of autoimmune diseases such as SLE, with an effector correlate that includes the induction of Th1 lymphocytes with a proin ammatory potential ( Fig. 6. Graphical abstract). There is evidence that the de ciency of ISG15, through the ISGylation of USP18, can degrade this potent inhibitor of IFNα/β-mediated signaling and promote its accumulation (12), therefore ISG15 has even been proposed as a downregulator of type 1 IFN (59), and may have a different functional pro le depending on its binding to ISGylated substrates or its extracellular release, as shown in the present study and previous reports in which the administration of free ISG15 as an adjuvant, helps to enhance the production of IFNγ and an increase in the cytolytic activity of CD8 + T cells against papillomavirus (61).
More than 30% of the ISG15 targets were described by mass spectrometry as nuclear proteins (56). These data suggest histones (nuclear proteins that are part of NETs) as potential targets for this PTM (62,63).
The ubiquitin label in H2B (K123) has been found to affect chromatin dynamics and RNA polymerase passage to facilitate a robust transcription system (64), (65). So far, this has been the closest evidence that ubiquitin or some other ubiquitin-like protein is related to histone conjugation, speci cally H2B.
Therefore, we propose H2B as a candidate substrate for ISG15, whose conjugation might induce changes in structure and function as previously reported with other PTMs in chromatin, and a source of autoantigens in autoimmune pathologies (63).
Our study has many limitations. First, it is a transversal study with no tracing of patient's characteristics.
Also, we used effector CD4 + in the cocultures with DC, instead of naïve CD4 + , resulting in a reactivation pro le of ex vivo Th1 or Th17 instead of in vitro Th0 polarization. Nevertheless, it is the rst study to address the in vitro, in silico and ex vivo role of ubiquitin and NETosis-released-ISG15, their impact on cellular responses from SLE patients as well as the improvement in our knowledge of the pathophysiology of Lupus and autoimmunity.

Conclusions
In summary, UbMPO can induce the reactivation of Th1 and Th17 cells, producing IFNγ and IL17A; enhanced proliferation in SLE patients, as well as dampened activation and proliferation in healthy donors T cells, highlighting the presence of changes in the intrinsic dynamics of MPO when conjugated with ubiquitin. Besides, our data suggest that NETosis is a non-canonical release mechanism for ISG15, Consent for publication: The images and data contained in this manuscript are not related to a single individual and are entirely unidenti able.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Figure 1
UbMPO regulates CD4+ lymphocyte activation, proliferation and cytokine production. CD4+ lymphocytes from SLE patients or healthy controls were cocultured with LPS activated DCs with UbMPO or rhMPO.
Increased production of proin ammatory cytokines, IFNγ (A) and IL17A (B) was documented for CD4+ lymphocytes from SLE patients upon stimulation with UbMPO. A differential proliferative response was found between SLE and healthy controls, with increased proliferation from the lupus samples (C) and decreased proliferation from healthy controls (D) towards UbMPO stimulation. Lower activation of healthy control CD4+ lymphocytes upon UbMPO stimulation was found (E). * p=0.0421. **p=0.03038.
***p=0.0286.   Graphical abstract: NETosis as a non-canonical release mechanism of ISG15 in SLE. In lupus patients, the NETs protein cargo is consistent with augmented amount of ISG15, which colocalized with H2B and induce the activation of CD4+, CD8+ and NK to a Th1 subset that enhanced the production of IFNγ (A). Low concentrations of ISG15 were found in healthy controls NETs with a diminished colocalization with H2B and reduction of IFNγ production by CD4+, CD8+ and NK cells (B). Therefore, ISG15 is a ngerprint of the interferogenic signature, and its release is mediated through NETosis in lupus patients.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.