Sustained xanthine oxidase inhibitor treat to target urate lowering therapy rewires a tight inflammation serum protein interactome

Background: Effective xanthine oxidoreductase inhibition (XOI) urate-lowering treatment (ULT) to target significantly reduces gout flare burden and synovitis between 1–2 years therapy, without clearing all monosodium urate crystal deposits. Paradoxically, treat to target ULT is associated with increased flare activity for at least 1 year in duration on average, before gout flare burden decreases. Since XOI has anti-inflammatory effects, we tested for biomarkers of sustained, effective ULT that alters gouty inflammation. Methods: We characterized the proteome of febuxostat-treated murine bone marrow macrophages. Blood samples (baseline and 48 weeks ULT) were analyzed by unbiased proteomics in febuxostat and allopurinol ULT responders from two, independent, racially and ethnically distinct comparative effectiveness trial cohorts (n=19, n=30). STRING-db and multivariate analyses supplemented determinations of significantly altered proteins via Wilcoxon matched pairs signed rank testing. Results: The proteome of cultured IL-1b-stimulated macrophages revealed febuxostat-induced anti-inflammatory changes, including for classical and alternative pathway complement activation pathways. At 48 weeks ULT, with altered purine metabolism confirmed by serum metabolomics, serum urate dropped >30%, to normal (<6.8 mg/dL) in all the studied patients. Overall, flares declined from baseline. Treated gout patient sera and peripheral blood mononuclear cells (PBMCs) showed significantly altered proteins (p<0.05) in clustering and proteome networks. CRP was not a useful therapy response biomarker. By comparison, significant serum proteome changes included decreased complement C8 heterotrimer C8A and C8G chains essential for C5b-9 membrane attack complex assembly and function; increase in the NLRP3 inflammasome activation promoter vimentin; increased urate crystal phagocytosis inhibitor sCD44; increased gouty inflammation pro-resolving mediator TGFB1; decreased phagocyte-recruiting chemokine PPBP/CXCL7, and increased monocyte/macrophage-expressed keratin-related proteins (KRT9,14,16) further validated by PBMC proteomics. STRING-db analyses of significantly altered serum proteins from both cohorts revealed a tight interactome network including central mediators of gouty inflammation (eg, IL-1B, CXCL8, IL6, C5). Conclusions: Rewiring of inflammation mediators in a tight serum protein interactome was a biomarker of sustained XOI-based ULT that effectively reduced serum urate and gout flares. Monitoring of the serum and PBMC proteome, including for changes in the complement pathway could help determine onset and targets of anti-inflammatory changes in response to effective, sustained XOI-based ULT. Trial Registration: ClinicalTrials.gov Identifier: NCT02579096

Treatment of hyperuricemia with XOI drugs (principally by using allopurinol or febuxostat) is central to gout management (6, 7).However, effective XOI urate-lowering treatment (ULT) to target also paradoxically induces an elevated gout are burden early in treatment (6, 7, 9); remodeling of articular monosodium urate (MSU) crystal deposits and consequent release of free crystals are held partly responsible (10)(11)(12).Notably, changes in a subset of CD14 positive monocytes, overactivation of CD8 + T cells, and upregulate arachidonate metabolism also have been implicated perpetuating systemic gouty in ammation after ULT initiation (13).
MSU crystals stimulate gouty in ammation in large part by activating monocytes and macrophages, promoting NLRP3 in ammasome-mediated IL-1b release, and neutrophil in ux and activation that amplify the in ammatory cascade (1,14).C5 cleavage on the MSU crystal surface, and consequent C5b-9 complement membrane attack complex (MAC) assembly and membrane pore-forming activity play a major role in the model gouty arthritis in ammatory process (15,16).
Recent clinical trials have demonstrated that effective XOI urate-lowering treatment (ULT) to target eventually reduces gout are burden and synovitis between 1-2 years therapy (17)(18)(19).Importantly, ares decrease in this time frame despite total resolution of urate crystal deposits being far slower, and particularly di cult to achieve (10), and continuing systemic in ammation even in the periods between ares and in clinical remission (13).In clinical practice, this situation is associated with lack of clarity on how long anti-in ammatory gout are prophylaxis, typically using low dose colchicine, is necessary after initiating XOI-based ULT and achieving the serum urate target (9).
The data revealed the ability of proteomics to detect anti-in ammatory changes in cultured XOI-treated macrophages, and in response to sustained, effective XOI-based ULT in gout patient sera and PBMCs.Our results provide unbiased evidence that sustained XOI treat to target ULT in gout re-wires complement activation and other in ammatory pathways.

Subjects
As previously reported in detail (25), Cohort 1 human subjects were studied under informed consent, and with local IRB approval (at the Jennifer Moreno San Diego Veterans Affairs Medical Center) in a prospective study ancillary to the national, multi-site comparative effectiveness ULT trial VA CSP594 STOP GOUT (26).In that trial, gout patients were randomized to a treat to urate target ULT regimen using allopurinol or the more selective XOI febuxostat (23).Unless contraindicated, colchicine was prescribed as the primary anti-in ammatory gout are prophylaxis, with colchicine routinely stopped at 6 months ULT.Twenty consecutive patients meeting the 2015 ACR/EULAR gout classi cation criteria (27), and with current hyperuricemia, were recruited from the Rheumatology Outpatient Clinic at the San Diego site (25).Once again (25), the gout validation cohort (Cohort 2, n = 30)) was from the University of Nebraska Medical Center, in Omaha, NE research site, under informed consent and with local IRB approval.We previously characterized Cohort 1 gout patient metabolomic pro les at time zero and 12 and 24 weeks of treat to target ULT, done in a blinded way for the XOI used, and following the trial protocol (25).

Proteomics:
Sera were obtained from both cohorts, with peripheral blood mononuclear cells (PBMCs) also prepared from Cohort 1 samples.All subjects were clinically assessed by study physicians for palpable tophaceous disease and presence of active are or quiescent arthritis, with co-morbidities and current medications also recorded.
For serum collection, research personnel collected non-fasting blood samples into 10 ml BD Vacutainer Blood Collection Tubes containing spray-coated silica and a polymer gel to facilitate serum separation.
Following 30 min incubation at room temperature, tubes were centrifuged for 10 min at 2000×g and sera were transferred into 1.7ml tubes and immediately frozen and stored at − 80°C until analyses were performed.
For PBMC preparation, non-fasting blood samples collected into 10 ml BD Vacutainer K2 EDTA Plus Blood Collection Tubes were transferred to a conical tube containing equal volume of PBS (~ total 20 ml).
The samples were then layered over Sigma Histopaque®-1077 (20 mL) in 50 mL conical tubes at room temperature, followed by centrifugation at 400×g in a swinging bucket centrifuge for 30 minutes at room temperature with no brake.The white cellular layer containing PBMCs at the interface between the plasma and density gradient was collected and washed in PBS by dilution and centrifugation for 10 minutes at 250×g.PBMC pellets were immediately frozen and stored at − 80°C until analyzed.

Mass Spectrometry Proteomics:
Sample preparation for proteomic analyses of BMDMs and patient sera was done as we previously described in extensive detail (28), with slight modi cation to the sample digestion protocol, which used 10µg trypsin in 50mM TEAB at 47˚C for 3 hours.After protein extraction and trypsin digest, 50ug aliquots of samples were reserved for TMT pro-labeling (28).Bridge channels for downstream data analysis of serum samples, were prepped by combining 5µg of all samples; 50µg aliquots of our bridge sample were then prepared for each TMT-plex (5 total).

Mass spectrometry data acquisition
Serum and BMDM proteomic data were acquired as described in detail (28).In brief, serum and BMDM proteomic data were acquired through an Thermo Orbitrap Fusion equipped with a Thermoeasy nLC 1000.For Mass spectrometry data search, raw mass spectrometry les were searched using Proteome Discoverer 2.5.0.400.The SEQUEST algorithm was used for spectral matches of raw data with in silico generated protein databases.Serum samples were searched against the UniProt Homo sapiens proteome (05-06-2023) and BMDM samples were searched against the Mus musculus proteome (05-06-23).

Mass Spectrometry Metabolomics
Sample preparation of patient sera for metabolomics were essentially as previously described (28).In brief, for data Analysis, metabolite features were rst normalized to the intensity of value of the internal standard, sulfamethazine, in each sample and then multiplied by 1E6.Missing values (with peak intensities of 0) in metabolite features were set to NA.Then, features with more than 20% missing values per group (timepoint) were removed from analysis.Missing values in remaining features were imputed using K-Nearest Neighbor (KNN) imputation using the R package (1.68.0).Intensity values were then log2 transformed.
Principal coordinate analysis (PcoA) was conducted with metabolite features, using Bray-Curtis distance calculation in the R package.PERMANOVA analysis was conducted using categorical metadata and metabolite features using Bray-Curtis distance calculation in the ADONIS R package.Binary comparisons between timepoints were done through the R package using Students T-test.Volcano plots were created in GraphPad Prism.All other plots were made using package in R. MetaboAnalyst (5.0) was used for metabolite functional enrichment analysis using MS peaks ranked by Student's T test p-values.A p-value cutoff of 0.05 was used for the mummichog algorithm.

Statistical analyses:
Paired statistical analyses of gout patient serum and PBMC samples across two timepoints (UCSD cohort), and for three timepoints for sera (Nebraska cohort), were conducted to identify signi cantly altered proteins.Unpaired statistical analyses were conducted for the cultured mouse BMDM samples.Signi cantly altered proteins were calculated using a Wilcoxon matched pairs signed rank test using Graphpad Prism with a p-value cutoff of 0.1 (serum) or 0.05 (BMDM, PBMC).For multivariate Analysis, Principal Component Analysis (PCA) was conducted using the R package using all normalized protein features.Principal Coordinate Analysis (PCoA) was conducted using the R package using the Euclidean Distance Matrix (EDM) of normalized protein features.PERMANOVA analysis was used to calculate data in uence by metadata categories.
Gene Ontology enrichment analysis was conducted through input of signi cantly altered proteins in both diseases to their respective controls into Cytoscape.Protein interactome analysis was conducted through input of signi cantly altered proteins in both diseases to their respective controls into String-DB with an interaction con dence of 0.700 (high-con dence).

Validation of XOI treatment effects on purine metabolism and the serum metabolome
We previously validated XOI treatment effects on purine metabolism in Cohort 1 (25).Here, we conducted untargeted metabolomics on sera of gout patients on effective serum treat to target ULT in Cohort 2 subjects treated with either febuxostat or allopurinol for 48 weeks.We annotated metabolite features using the Global Natural Products Social Molecular Networking (GNPS) platform.Since timepoint signi cantly in uenced our paired proteomic data set, we conducted paired binary comparisons between timepoints.Comparison of baseline (BL) and proteomics endpoint 48wks of ULT revealed several signi cantly altered metabolites, with some signi cantly changed by 24wks ULT (Supplemental Fig. 2A).Functional enrichment analysis of all identi ed metabolite features, using MS1 peak information, validated serum metabolome changes in purine and pyrimidine metabolism in Cohort 2 in this study.These ndings were associated with signi cant changes in multiple other pathways, including arachidonic acid metabolism, and most pronounced for linoleate metabolism at 24 and 48wks ULT (Supplemental Fig. 1B).The new ndings for Cohort 2 reinforced previously published effects of XOI stats stats treatment on the serum metabolome and lipidome in gout patients of Cohort 1 and on the serum lipidome in both Cohorts 1 and 2 (25).

Effects of XOI treatment to urate target on the serum proteome
Global serum proteome changes before and at 48wks XOI-based ULT were found in gout patient Cohort 1 and the independent validation Cohort 2 (Fig. 2B), whose demographics and changes in serum urate are summarized (Fig. 2A, Supplemental Fig. 1A & 1B).We observed overall decrease in serum urate (sUA) levels after 48wks ULT, but relatively stable C-reactive protein (CRP) levels after ULT in both cohorts.
Examining each cohort independently from Baseline (BL) to serum proteomics Endpoint (48 wks of ULT;EP), we found 21 and 49 signi cantly changed proteins (p < 0.05, Wilcoxon signed-ranks test) for Cohort 1 and 2, respectively.Interactome analysis through STRING-db, was accompanied by "pindropping" known gouty-in ammation markers, known to be below the mass spectrometry detection limits (30), along with the signi cantly altered proteins from both cohorts.We identi ed 23 high con dence interacting proteins (Fig. 2C), which Gene Ontology enrichment analyses revealed to belong to 4 major categories: Innate immune response, humoral immune response, protein/peptide secretion, and posttranslation modi cation of proteins (Fig. 2D, Table 1).TNF There were 277 overlapping protein identi cations between both independent cohorts.There were signi cant in uences (PERMANOVA p < 0.10) from patient and timepoint on Cohort 1 and 2 results, respectively (Supplemental Fig. 1C&D).We subjected these proteins to interactome analysis, and observed 138 high con dence interacting proteins (Fig. 2E).Moreover, we identi ed 70 proteins that were similarly altered at 48wks ULT (Fig. 2E) in both cohorts.We also identi ed those proteins in our interactome that fell into innate or humoral gene ontology enrichment categories.Results showed rewiring of networked key in ammation mediators not detectable by conventional serum biomarker pro ling, including C8 cleavage products, VIM, PPBP/CXCL7, KRT16, TGFB1, IGF-I, and sCD44.These novel biomarkers of XOI ULT effects were clustered with central gout mediators including IL-1B, CXCL8, IL6, and C5, in a tight protein interactome.Results revealed a novel functionally important network of physically interacting serum proteins in gouty in ammation that was altered in response to ULT to target with XOI drugs.

XOI treatment to serum urate target effects on the PBMC Proteome
Last, to further characterize in vivo response to XOI-based ULT in gout, we isolated PBMCs from Cohort 1 patients.We identi ed 197 signi cantly altered proteins at 48wks ULT (p < 0.05, Fig. 3A), with 42 highcon dence (> 0.700) interacting proteins (Fig. 3B) We found these proteins in the PBMC interactome (listed in Supplemental Table 2) belonging largely to secretion, leukocyte, and neutrophil activation gene ontology pathways (Fig. 3C).Moreover, the KRT protein ndings for serum proteins were validated in the PBMC proteomics studies.
We next sought to understand how patient information associated to the PBMC proteome (Supplemental Fig. 2A) and found several cytokines to have signi cant in uence (Supplemental Fig. 2C, p-value < 0.1).
Interpatient correlation analysis identi ed two distinct proteome groups (Supplemental Fig. 2A-B), and statistical analysis identi ed proteins driving the separation of proteome group 1(n = 5) and 2 (n = 14).We analyzed samples separated by timepoint and identi ed the top scored proteins at Baseline and 48wks of ULT (Fig. 3D).We identi ed overlapping protein drivers of separation at both timepoints, and interactome analysis of identi ed driver proteins at both timepoints along with "pin-dropped" gout proteins (Fig. 3E) found strong and high con dence (> 0.700) interactions between known gout mediators and top identi ed proteins, particularly MMP9 and other proteins identi ed at 48wks ULT.Hence, PBMC proteome analysis further teased apart XOI-based ULT effects in gout patients while highlighting anti-in ammatory effects.

Discussion
Gout requires a unique approach to arthritis targets and biomarkers of the response to XOI-based ULT, due to variable phenotypes, and weaving of urate homeostasis, comorbidities, and in ammatory arthritis (1)(2)(3)(4)(5)8).In contrast to the genetics of urate biology, genome-wide association studies have identi ed few genetic coding variants potentially involved in gouty arthritis (31,32).Therefore, this biomarker study assessed the biomarker potential of proteomic pro ling of gout patient sera at 48wks sustained ULT to urate target with XOI that reduced both are burden and serum urate in two independent cohorts.Speci c serum proteomics ndings at 48wks XOI-based treat to target ULT, in both cohorts studied, included decreased C8A and C8G chains, which play a major role in complement C5b-9 MAC assembly and activity that, along with C5a generation, contribute substantially to the in ammatory process in model gouty arthritis (15,16,34).Paradoxically, we detected increase in serum of the NLRP3 in ammasome scaffolder and activation promoter VIM (vimentin) (35), of interest because early increase in gout ares is seen in XOI-based ULT (9), Increased serum sCD44 was noteworthy, since sCD44 inhibits macrophage phagocytosis of urate crystals and consequent NLRP3 in ammasome activation, by blocking crystal binding to transmembrane CD44 (36).
We also observed increase in serum of TGFB1, which promotes model gout are resolution by suppressing macrophage activation by crystals (37).Conversely, IGF-I, which cross-talks with and can synergize with TGF-beta, was decreased in serum at 48wks ULT (38).We detected decrease in serum of the phagocyte-recruiting chemokine PPBP/CXCL7 (39), and decreased lactoferrin, a neutrophil-released co-activator of the lubricin-degrading serine protease Cathepsin G (40).That nding was of note, since Cathepsin G is a major degrader of lubricin, which functions as a substantial constitutive suppressor of gouty in ammation and urate production by synovial resident macrophages (41).We also observed an increase in monocyte/macrophage-expressed keratin-related proteins (KRT9,14,16), further validated by Cohort 1 gout patient PBMC proteomics.KRT16 is implicated in monocyte to macrophage differentiation, and MMP-1 and innate immune responses to tissue damage in epithelia (42).
Last, STRING-db analyses of signi cantly altered proteins from both cohorts revealed that the tight serum protein interactome network altered by XOI-based ULT encompassed a core group of central mediators of gouty in ammation (including IL-1B, CXCL8, IL6, C5)(4).
Robustness of our ndings on effects of effective ULT on the serum protein interactome discovered here was buttressed by a group of parallel studies.First, in this context, previously published evidence in gout Cohort 1 that the ULT regimen altered the serum metabolome, and the serum lipidome in gout Cohorts 1 and 2, and effects of febuxostat on lipolysis in cultured adipocytes (25).Moreover, the current study demonstrated that the serum metabolome was signi cantly altered for purine and pyrimidine metabolism in Cohort 2, associated with signi cant changes in multiple other pathways, most pronounced for linoleate metabolism at both 24wks and 48wks ULT.Second, analyses of the Cohort 1 proteome of gout patient PBMCs identi ed 42 high-con dence interacting proteins belonging largely to secretion, leukocyte, and neutrophil activation gene ontology pathways.The KRT ndings for serum proteins were validated in the PBMC proteome.In addition, we found strong and high con dence (> 0.700) interactions between known gout mediators and EFS identi ed proteins, particularly in the proteins identi ed at 48wks of ULT, including MMP9.Whereas no signi cant difference in MMP9 abundance levels was identi ed between BL and 48wks of ULT, further study would be needed to validate signi cance of differences between PBMC proteome groups 1 and 2. The collective results of PBMC proteome analysis further teased apart the effects of XOI-based ULT in gout, and highlighted anti-in ammatory effects of XOI-based ULT on these leukocytes as a whole.
We employed in vitro studies that characterized effects of the selective XOI febuxostat on the proteome of cultured murine BMDMs stimulated by the major gouty in ammation driver IL-1b.Febuxostat suppressed multiple pro-in ammatory IL-1b-induced changes in the macrophage proteome.Analyses of gene ontology enrichment of proteins found in the macrophage protein interactome revealed that in vitro XOI treatment of activated BMDMs broadly reversed many pro-in ammatory responses.Notably, the most pronounced pathway changes were seen in classical and alternative pathway complement activation, which reinforced the impact of the ndings for XOI-treatment effects on C8A and C8G in the gout patient serum proteome.Febuxostat also altered lymphocyte-mediated immunity, brinolysis, and cytolysis gene ontology pathways in cultured macrophages in response to IL-1b.Our ndings in cultured macrophages and gout patient PBMCs were novel partly because previous studies have suggested that both hyperuricemia and urate crystals program elevated monocyte in ammatory responses in vitro and that hyperuricemia primes model gout in ammation in mice in vivo model gout (43)(44)(45).
A pro-in ammatory serum proteome signature was recently characterized in asymptomatic hyperuricemia (AH) by targeted proteomics (46).The approach used the Olink Target 96 In ammation Panel™ (46), distinct from the unbiased mass spectrometry-based approach utilized in the current study.
The methodology employed dual recognition by oligonucleotide-labelled antibody probe pairs and DNAcoupled quantitative PCR, designed to detect speci c immunoregulatory proteins below mass spectrometry detection limits (46).Upregulated serum immunoregulatory proteins in AH group included the mTOR effector 4E-BP1, IL-18R1, multiple growth factors, chemokines, members of the IL-6 cytokine and TNF superfamily, with a Th17 cell signature, and increases in in ammation-dampening IL-10 and FGF21 also identi ed (46).Using the same targeted serum proteomics approach, a small sub-study of 13 subjects before and 3 months into successful XOI-based treat to target ULT also revealed signi cant downregulation of LIF-R.CDCP1, IL-18, NT-3, IL10RB, CCL28, CCL11, and SLAMF1 (46).All of the differentially detected proteins in that targeted proteomics study, which were predominantly cytokines and growth factors, were below the detection limits of of our unbiased mass spectrometry serum proteomics approach (Sanchez, C, et al, unpublished observations).Therefore, the design, approach, and sample size of the current study were unique and provided distinct information on the molecular signature of XOI effects on hyperuricemia in gout.
Hyperuricemia increases blood monocyte population expansion in vivo in humans (44) However, monocytes, and other mononuclear leukocytes, are heterogeneous, and can be recruited into diseased or challenged tissues, and one limitation in this study is that monocytes are normally only a small fraction (ie, ≤ 10%) of PBMCs (47).PBMCs remain a source of highly informative biomarkers for acute and chronic in ammatory diseases, but also are highly heterogeneous (48), buttressing the limitation of this study that PBMCs only were obtained at the Cohort 1 site.This trial did not have a placebo or uricosuric treatment arm.Moreover, we did not study gout patient controls from the same clinical trial that failed to achieve serum urate target, However, the proportion of such subjects overall in the VA STOP GOUT trial was low (ie, ~ 20%) (19), and all those subjects were considered at least partially treated since they received XOI-based ULT.
In conclusion, a novel, functionally important network of physically interacting proteins in gouty in ammation was altered in response to sustained, effective XOI-based ULT.Potential clinical signi cance of the results, especially for data from the clinical trial, included that the treat to target XOIbased ULT regimen is associated with early increase in are activity before gout ares eventually decrease (9).Moreover, the current study provides further support for the use of serum proteomics, including approaches targeting the complement pathway and the in ammatory secretome, to provide biomarkers for responses to gout pharmacotherapy, and for characterization and prognosis of different clinical phenotypes in the disease (41,46,49,50).C. Gene ontology enrichment analysis of signi cantly altered proteins from both proteomic cohorts.

Abbreviations
Enrichment was conducted on Cytoscape with the Human Proteome as background.
D. Protein interactome of the detected overlapping proteins from both cohorts.Nodes are colored based on whether their abundance change was the same in both cohorts after 48wks of ULT, and shaped based on their direction of change after ULT.
E. Gene ontology enrichment analysis of overlapping proteins from both cohorts.Enrichment was conducted on Cytoscape with the Human Proteome as background.
to participate: Cohort 1 human subjects were studied under informed consent, and with local IRB approval at the Jennifer Moreno San Diego Veterans Affairs Medical Center.The gout validation Cohort 2, was from the University of Nebraska Medical Center, in Omaha, NE research site, under informed consent and with local IRB approval..