Abstract

   Objectives: This study investigates the role of retinol binding protein
   4 (RBP4) in an articular context. RBP4, a vitamin A transporter, is
   linked to various metabolic diseases. Methods: Synovial fluid RBP4
   levels were assessed in crystalline arthritis (CA) patients using
   ELISA. RBP4’s impact on articular cell types was analysed in vitro
   through RT-PCR and flow cytometry. Proteomic analysis was conducted on
   primary human osteoarthritis chondrocytes (hOACs). Results: Synovial
   fluid RBP4 concentrations in CA patients correlated positively with
   glucose levels and negatively with synovial leukocyte count and were
   elevated in hypertensive patients. In vitro, these RBP4 concentrations
   activated neutrophils, induced the expression of inflammatory factors
   in hOACs as well as synoviocytes, and triggered proteomic changes
   consistent with inflammation. Moreover, they increased catabolism and
   decreased anabolism, mitochondrial dysfunction, and glycolysis
   promotion. Both in silico and in vitro experiments suggested that RBP4
   acts through TLR4. Conclusions: This study identifies relevant RBP4
   concentrations in CA patients’ synovial fluids, linking them to
   hypertensive patients with a metabolic disruption. Evidence is provided
   that RBP4 acts as a DAMP at these concentrations, inducing robust
   inflammatory, catabolic, chemotactic, and metabolic responses in
   chondrocytes, synoviocytes, and neutrophils. These effects may explain
   RBP4-related metabolic diseases’ contribution to joint destruction in
   various rheumatic conditions like CA.

   Keywords: arthritis, chondrocytes, gout, inflammation, crystal
   arthropathies

1. Introduction

   Retinol binding protein 4 (RBP4) is a hepatokine member of the
   lipocalin family which is characterized by its carrier function of
   small hydrophobic molecules in the blood. It was first described as a
   plasma protein that binds retinol (vitamin A) and transports it through
   the bloodstream [[44]1]. When we refer to RBP4, we must make a
   differentiation between the two isoforms of holo-Rbp4, when it is bound
   to retinol, and apo-Rbp4, when it is found alone. In its holo form,
   Rbp4 binds to another molecule called transthyretin (TTR) in plasma,
   which prevents its glomerular filtration and its binding to other
   molecules [[45]2,[46]3]. Currently, there are only two receptors that
   are known to be RBP4-specific which are as follows: “stimulated by
   retinoic acid 6” (STRA6) and “stimulated by retinoic acid 6-like”
   (STRA6L) [[47]4,[48]5,[49]6]. RBP4 signalling through these receptors
   (JAK/STAT pathway) [[50]7] has been associated with several
   inflammatory and metabolic diseases [[51]8,[52]9]. Nonetheless,
   RBP4–mediated pro-inflammatory effects have also been related to the
   signalling pathways of Toll-like receptors (TLRs), such as TLR4, at
   least in macrophages [[53]10,[54]11] and cardiomyocytes [[55]12].

   High levels of RBP4 have been associated with inflammation in metabolic
   disorders like insulin resistance [[56]9,[57]13], metabolic syndrome
   [[58]14,[59]15], fatty liver [[60]16], cardiovascular diseases like
   atherosclerosis [[61]17], and chronic inflammatory diseases such as
   psoriasis [[62]8]. Moreover, altered levels of RBP4 have been reported
   in patients with rheumatic diseases, including rheumatoid arthritis
   (RA) [[63]18,[64]19] and ankylosing spondylitis [[65]20]. RBP4 was also
   specifically related to RA co-morbidities in RA patients
   [[66]21,[67]22]. Furthermore, RBP4 has recently been described to be
   present within the joints of osteoarthritis (OA) patients, and its
   levels have been correlated with the production of matrix-degrading
   enzymes in OA chondrocytes [[68]23]. Nonetheless, RBP4 has not been
   studied in the context of crystalline arthropathies, which are a group
   of inflammatory arthritides associated with crystal deposition in the
   synovial space including gout, which is common in patients with obesity
   or metabolic alterations [[69]24], and pseudogout (calcium
   pyrophosphate dihydrate crystal deposition (CPPD)-mediated disease),
   which is associated with degenerative joint diseases [[70]25,[71]26].
   In this study, we aimed to determine the presence of RBP4 in synovial
   fluid from crystalline arthritis (CA) patients as well as its potential
   as an articular pro-inflammatory and pro-catabolic factor.

2. Materials and Methods

2.1. Immunoenzymatic Assay of RBP4

   Synovial fluid was obtained through arthrocentesis from twenty-seven
   people (twenty-one men and six women with an average age of 65.7 ± 14.9
   years) who had an acute knee effusion due to CA. All patients provided
   signed informed consent approved by the Research Ethics Committee of
   Santiago de Compostela, which is affiliated with the regional public
   health system (SERGAS) and has a reference number of 2016/258. Samples
   were treated with type I-testicular bovine hyaluronidase
   (Sigma-Aldrich, St. Louis, MO, USA) (1500 IU/mL), incubated for 30 min
   at 37 °C, and centrifuged for 10 min at 410× g. Synovial fluid levels
   of RBP4 were determined employing the human RBP4 ELISA kit
   (ThermoFisher Scientific, Madrid, Spain) and following the
   manufacturer’s instructions.

2.2. Patients’ Clinical Parameters and Drug Consumption

   The clinical parameters of the patients involved in this study of
   synovial and blood glucose levels, uric acid levels, and leukocyte cell
   count were obtained from their medical records. Patients’ drug
   consumption information was also obtained from their medical records.
   None of the patients were taking any corticosteroids or any medication
   that could affect uric acid levels at the moment of the extraction. The
   techniques used by the central hospital laboratory to obtain the
   clinical parameters were as follows: patients’ synovial and blood
   glucose levels were obtained through the glucose-hexokinase II method
   using an ADVIA Chemistry XTP system (Siemens, Munich, Germany); uric
   acid levels were obtained through the Fossati enzymatic reaction using
   an ADVIA Chemistry XTP system; and leukocyte cell count was obtained
   through flow cytometry using an ADVIA 2120i (Siemens).

2.3. Molecular Docking Analysis

   A molecular docking is used to get a computational approximation of the
   strength of interaction between two proteins or protein molecules. RBP4
   (PDB-3FMZ) and TLR4 with its co-receptor MD2 (PDB-3FXI) structures were
   obtained from the RCSB Protein Data Bank (PDB) [[72]27] and optimized
   using PyMol software 2.5.228 [[73]28]. Pydock [[74]29] was used for the
   molecular docking analysis. Results are shown in Gibbs free energy
   units (Kcal/mol) and ranked from lowest (strongest interaction) to
   highest (weakest interaction) for all the possible conformations.

2.4. Cell Isolation and Culture

   Primary human OA chondrocytes (hOACs) were isolated from cartilage
   samples from male patients (aged 65–71 years old) who underwent
   articular-related surgery and provided signed informed consent approved
   by the Research Ethics Committee of Santiago de Compostela (2016/258).
   A sample was processed using pronase and collagenase digestion and
   grown in DMEM/F12 with 10% fetal bovine serum (FBS), 50 μg/mL
   streptomycin, 4 mM L-glutamine, and 100 U/mL penicillin/streptomycin
   (P/S) (10%FBS/DMEM/F12) (all from Sigma-Aldrich). All the hOACs used in
   this study were used under passage three. ATDC5 cells, a chondrogenic
   cell line, were obtained from the Riken Bank (RCB Cat# RCB0565,
   RRID:CVCL_3894) and grown in DMEM/F12 with 5% FBS, 4 mM L-glutamine,
   100 U/mL P/S, 50 μg/mL streptomycin, 10 µg/mL human apo-transferrin,
   and 30 nM sodium selenite (Sigma-Aldrich) on adherent culture dishes
   (Corning, Glendale, AZ, USA). SW982 cells were obtained from ATCC (Cat#
   HTB-93, RRID:CVCL_1734) and grown in 10%FBS/DMEM/F12. Polymorphonuclear
   leukocytes (PMNs) were isolated from human blood samples from three
   healthy donors (one male and two females, 27.3 ± 1.0 years old) who
   provided signed informed consent approved by the Research Ethics
   Committee of Santiago de Compostela (2020/357). PMNs where isolated
   employing Ficoll (ThermoFisher Scientific), dextran (Sigma-Aldrich),
   and ACK buffers (Lonza, Porriño, Spain) and cultured in RPMI
   (Sigma-Aldrich) with 10% FBS, 1% P/S, and 4 mM L-glutamine.

2.5. RBP4 Stimulus

   ATDC5 and hOAC cells were seeded on 24-well plates (Corning),
   FBS-deprived for 18 h, and stimulated with 2.5–25 µg/mL mouse RBP4
   (Sino Biological, Beijing, China) for 48 h. Once we established an
   effective dose for RBP4, its synergy with IL1β (Sigma-Aldrich) was
   tested at the effective concentrations of 25 µg/mL (RBP4) and 1 ng/mL
   (IL1β). RBP4 inflammatory potential was also tested after 3 h
   pre-treatment with 1 µM CLI-095 (IBIAN Technologies, Zaragoza, Spain).

2.6. Gene Expression Analysis

   Cells were lysed using TRI Reagent (Sigma-Aldrich) and RNA was purified
   using E.Z.N.A. total RNA kit I (Omega Bio-Tek, Atlanta, GA, USA)
   according to the manufacturer’s instructions. RNA was treated with
   DNase I (Cientisol, A Coruña, Spain) and retro-transcribed employing
   the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems,
   New York, NY, USA). Gene expression levels of RBP4; interleukin 6
   (IL6); C-C motif chemokine ligand 2 (CCL2); intercellular adhesion
   molecule 1 (ICAM1); vascular cell adhesion molecule (VCAM); nitric
   oxide synthase 2 (NOS2); nicotinamide phosphoribosyltransferase
   (NAMPT); matrix metalloproteinases (MMPs) 1, 3, and 9; chemokine C-X-C
   motif ligands (CXCL) 1, 2, 8, and 12; and hypoxanthine
   phosphoribosyltransferase (HPRT) were assessed through a real-time PCR
   using iTaq Universal SYBR Green Supermix (BioRad, CA, USA) and primers
   (Sigma-Aldrich) listed in [75]Supplementary Table S1. Data analysis was
   performed using the MxPro software 4.1.0 (Stratagene, CA, USA).
   Relative quantification was obtained through the 2^−ΔΔCt method, with
   HPRT as the reference gene. Data were normalized against a non-treated,
   non-stimulated control for each experiment.

2.7. Proteomic Analysis

   Cellular proteome from hOACs cultured with or without RBP4 (25 µg/mL)
   for 48 h were studied through micro-liquid chromatography–mass
   spectrometry (micro-LC–MS/MS) using a hybrid quadrupole TripleTOF 6600
   (Sciex, CA, USA). Both proteomes were identified through the
   qualitative shotgun data-dependent acquisition (DDA) method
   [[76]30,[77]31], and protein levels were measured through the
   quantitative sequential window acquisition of all theoretical mass
   spectra (SWATH) method [[78]31,[79]32]. A 5% false discovery rate (FDR)
   and a p-value ≤ 0.05 were used to filter the dataset. FunRich software
   3.1.3 [[80]33] was used to determine proteome enrichment.

2.8. Annexin V-FITC Apoptosis Assay

   PMNs were seeded in non-adherent plates, treated with 25 µg/mL RBP4 for
   6 h, and stained them with annexin V (Sigma-Aldrich) and propidium
   iodide (ThermoFisher Scientific) according to the manufacturer’s
   instructions. Data were acquired with a BD FACSAria flow cytometer (BD
   Biosciences, Madrid, Spain) and analysed using FlowJo™ software 10.8.1
   for Windows (BD Biosciences).

2.9. Statistical Analysis

   Data were presented as the mean ± standard error of the mean (SEM) for
   three independent experiments. Statistical analyses were performed
   using GraphPad PRISM 8 software 8.0.2 (La Jolla, CA, USA).
   Statistically significant differences between experimental conditions
   were determined through Student’s t-test. A difference was considered
   significant if the p-value was ≤ 0.05.

3. Results

3.1. RBP4 Is Present in the Synovial Fluid of Patients with Crystalline
Arthritis

   In this study, we first aimed to see if there was any correlation
   between RBP4 levels in synovial fluid and other clinical parameters of
   patients with CA obtained from their medical history using the values
   closest to the date of sample extraction. RBP4 concentration in the
   synovial fluid of these patients was between 2.44 and 22.62 µg/mL, with
   a mean concentration of 11.74 ± 5.85 µg/mL (11.53 ± 2.17 µg/mL in gout
   patients and 11.94 ± 1.30 µg/mL in CPPD patients). We observed that
   RBP4 levels correlated positively with the total glucose in synovial
   fluid ([81]Figure 1A) as well as the blood glucose concentration
   ([82]Figure 1B). However, these correlations were not significant in
   women ([83]Supplementary Figure S1A). Among these patients, we also
   found that hypertensive patients had higher levels of synovial RBP4,
   with a mean of 18.22 ± 1.07 µg/mL vs. 7.96 ± 4.96 µg/mL in
   non-hypertensive patients (p = 0.0003); however, they were not
   significant in women once again ([84]Supplementary Figure S1B). When
   splitting these patients according to their specific pathology, this
   difference was only preserved in gout patients (p = 0.0058), who have a
   higher frequency of metabolic alterations, but not in pseudogout
   patients (p = 0.6291) ([85]Figure 1C). Additionally, we noticed that
   synovial fluid leukocyte count, a well-known marker of inflammation for
   this type of arthropathy, correlated negatively with serum uric acid
   levels ([86]Figure 1D) and also with synovial RBP4 levels ([87]Figure
   1E), although no significant correlation was found between uric acid
   and RBP4 levels ([88]Supplementary Figure S2).

Figure 1.

   [89]Figure 1
   [90]Open in a new tab

   Correlations found in RBP4 levels tested through an ELISA assay used
   for the analysis of synovial fluid from 27 crystalline arthritis
   patients (14 patients with gout and 13 patients with CPPD). There is a
   positive correlation between patients’ synovial RBP4 concentrations and
   total glucose in the synovial fluid (A) and blood glucose levels (B).
   There was no significant difference in synovial RBP4 levels in gout and
   CPPD patients (p = 0.88). However, when separating our patients based
   on hypertension diagnosis, we observed that patients’ RBP4 levels are
   higher in hypertensive patients when both considering all patients
   together and gout patients alone, but this difference is absent in CPPD
   patients (C). Patients RBP4 levels correlated negatively with the
   synovial leukocyte count (D), which also correlated negatively with
   uric acid blood levels (E). RBP4 (retinol binding protein 4), CPPD
   (calcium pyrophosphate deposition). (p = statistical significance; ** p
   < 0.01; *** p < 0.001).

3.2. RBP4 Activates the Inflammatory Response of Chondrocytes

   RBP4 has been described as exhibiting inflammatory activity in
   macrophages and endothelial cells [[91]10,[92]34]. Thus, we tested the
   inflammatory effect of RBP4 on the chondrogenic cell line ATDC5. We
   tested a range of RBP4 concentrations (2.5, 5, and 25 µg/mL) similar to
   the ones that we found in the synovial fluid of CA patients, which were
   similar to the ones described in OA patients [[93]23]. After 48 h of
   stimulation with RBP4, we observed that different genes related to the
   inflammatory response (IL6, CCL2, ICAM, VCAM, NOS2, NAMPT, MMP1, MMP3,
   and MMP9) were significantly upregulated in a dose-response manner in
   ATDC5 cells ([94]Figure 2A). After validating these results for hOACs
   ([95]Figure 2B), we decided to investigate whether RBP4 could
   potentiate the inflammatory responses elicited by IL1β, a cytokine
   typically found in different arthritides. As shown in [96]Figure 2C, we
   found that RBP4 synergically upregulated the IL1β-mediated upregulation
   of several inflammatory genes, such as CCL2, ICAM and NOS2.

Figure 2.

   [97]Figure 2
   [98]Open in a new tab

   RBP4 stimulation of ATDC5 cells for 48 h produced an upregulation of
   IL6, CCL2, ICAM, VCAM, NOS2, NAMPT, MMP1, MMP3, and MMP9 in a
   dose-response way (A). Human OA primary chondrocytes were treated with
   RBP4 to validate the previous results, obtaining the same upregulation
   of all the inflammatory genes in a dose-response way (B). RBP4 can
   synergize with IL1β, upregulating CCL2, ICAM, and NOS2 independently
   and also synergistically. NAMPT and MMP1 showed an additive response.
   VCAM, IL6, MMP9, and MMP3 were upregulated by both stimuli, but they
   had no synergic or additive effect together (C). RBP4 (retinol binding
   protein 4), IL6 (interleukin 6), CCL2 (C-C motif chemokine ligand 2),
   ICAM (intercellular adhesion molecule 1), VCAM (vascular cell adhesion
   molecule), NOS2 (nitric oxide synthase 2), NAMPT (nicotinamide
   phosphoribosyltransferase), MMP1 (matrix metallopeptidase 1), MMP3
   (matrix metallopeptidase 3), and MMP9 (matrix metallopeptidase 9), IL1β
   (Interleukin 1 Beta). (* p < 0.05; ** p < 0.01; *** p < 0.001; **** p <
   0.0001 versus control condition. # p < 0.05; ## p < 0.01, ### p <
   0.001; #### p < 0.0001 between the conditions marked).

3.3. TLR4 Is Partially Involved in the Inflammatory Response to RBP4

   TLR4 has been described to participate in RBP4-induced inflammation.
   Hence, we performed a computational recreation of the interaction of
   these proteins ([99]Figure 3A–C). After several thousands of iterations
   ([100]Supplementary Table S2), the performed molecular docking revealed
   that the interaction between TLR4 and RBP4 exhibited a very negative
   free energy, which suggests that RBP4 may interact with TLR4. To
   confirm this, we stimulated ATCD5 chondrocytes with RBP4 in the
   presence or absence of the specific TLR4 inhibitor CLI-095. As shown in
   [101]Figure 3D, CLI-095 pretreatment inhibits the expression of most of
   the inflammatory genes induced by RBP4, including IL6, ICAM, VCAM,
   NOS2, NAMPT, and MMP9.

Figure 3.

   [102]Figure 3
   [103]Open in a new tab

   Visual representation of TLR4 structure in green (A), RBP4 in red (B),
   and the best result of the molecular docking (C). Cli95 pre-treatment
   partially inhibits the effect of RBP4, downregulating IL6, ICAM, VCAM,
   NOS2, NAMPT, and MMP9 (D). RBP4 expression is not modulated by IL1β,
   RBP4, or both stimuli together. The inflammatory stimulus LPS alone or
   in combination with a well-known COX1-2 inhibitor (indomethacin) in
   different doses cannot modulate RBP4 expression (E). TLR4 (Toll-like
   receptor 4), RBP4 (retinol binding protein 4), IL6 (interleukin 6),
   CCL2 (C-C motif chemokine ligand 2), ICAM (intercellular adhesion
   molecule 1), VCAM (vascular cell adhesion molecule), NOS2 (nitric oxide
   synthase 2), NAMPT (nicotinamide phosphoribosyltransferase), MMP1
   (matrix metallopeptidase 1), MMP3 (matrix metallopeptidase 3), and MMP9
   (matrix metallopeptidase 9), IL1β (Interleukin 1 Beta), Indo
   (indomethacin). (* p < 0.05; ** p < 0.01; *** p < 0.001; **** p <
   0.0001 versus control condition. # p < 0.05; ## p < 0.01; #### p <
   0.0001 between the conditions marked).

3.4. RBP4 Expression Is Not Modulated by Inflammatory Stimuli of Chondrocytes

   In addition, we studied RBP4 expression under different inflammatory
   conditions in ATDC5 cells. Results obtained showed that LPS, IL1β,
   RBP4, and the combination of IL1β and RBP4 did not induce RBP4 mRNA
   expression. Interestingly, indomethacin, a well-known COX1/2 inhibitor,
   did not have any effect on RBP4 expression either ([104]Figure 3E).

3.5. RBP4 Induces Proteomic Changes Associated with Increased Glycolysis,
Decreased Mitochondrial Metabolism, and Reduced Anabolism in hOACs

   Considering that RBP4 induced the expression of inflammatory and
   catabolic genes in hOACs, we performed a proteomic analysis to achieve
   further insights into its effects on these cells. After 48 h of
   treatment with RBP4 (25 µg/mL), 160 proteins were significantly
   modulated ([105]Supplementary Table S3). Among the identified proteins,
   we found several enzymes involved in two interconnected metabolic
   pathways of glycolysis and the pentose phosphate pathway (PPP)
   ([106]Figure 4A). Regarding glycolysis, we observed that fructose
   bisphosphate aldolase A and C (ALDOA, ALDOC), triosephosphate isomerase
   (TPI), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and alpha
   enolase 1 (ENO1) were significantly augmented in hOACs after RBP4
   treatment. Regarding the PPP, 6-phosphogluconate dehydrogenase (PGD)
   was upregulated, while 6-phosphogluconolactonase (6PGL) was
   downregulated. There were also two proteins related to these pathways
   that were significantly increased after RBP4 treatment as follows:
   glucose 1,6-bisphosphate synthase (PGM2) and aldo-keto reductase (AKR).
   PGM2, which was increased 18.74 times, is an enzyme that produces more
   substrate molecules for the glycolytic pathway. In a similar way,
   isoforms B1, C1, and C2 of AKR, with fold induction changes of 4.36,
   2.16, and 1.95, provide more substrate molecules into both glycolysis
   and the PPP.

Figure 4.

   [107]Figure 4
   [108]Open in a new tab

   (A) Volcano plot showing a general view of the proteomic analysis, red
   dots represent the proteins that are significantly regulated by the
   RBP4 treatment while black dots are the proteins that are not
   significantly different. (B) RBP4 stimulation enhances glucose uptake
   through the GLUT1 receptor and produces a general upregulation of the
   glycolytic pathways, including ALDO A, ALDO C, TPI, GAPDH, and ENO1,
   which were enriched in RBP4-treated cells. In the pentose phosphate
   pathway, there was an enrichment of 6PGL and a depletion of PGD. Two
   proteins highly related to the previous pathways are PGM2 and AKR1.
   PGM2 generates more glucose 6-P in the glycolytic process, and AKR1 can
   produce more aldoses or more alcohol-enriched sugars that might
   increase the oxidative stress in the cell. (C) Graphic representation
   of the different collagen-related proteins that are significantly
   changed between control and RBP4-treated chondrocytes in the proteomic
   analysis. (D) All the interferon-related proteins are highly
   upregulated. RBP4 (retinol binding protein 4), GLUT1 (glucose
   transporter 1), ALDO A (fructose-bisphosphate aldolase A), ALDO C
   (fructose-bisphosphate aldolase C), TPI (triosephosphate isomerase),
   GAPDH (glyceraldehyde 3-phosphate dehydrogenase),
   6-phosphogluconolactonase (6PGL), ENO1 (enolase 1), PGD
   (6-phosphogluconate dehydrogenase), PGM2 (glucose 1, 6-bisphosphate
   synthase), AKR (aldo-keto reductase). (Green arrows represent
   downregulated proteins; Red arrows represent upregulated proteins. ** p
   < 0.01; *** p < 0.001).

   In the mitochondrial environment, we could identify two different
   events. Firstly, many proteins involved in mitochondrial metabolism and
   oxidative phosphorylation were significantly downregulated, including
   aspartate aminotransferase (AST), glutamate dehydrogenase 1 (GLUD1),
   ATP synthase F(0) complex subunit B1 (ATP5PB), and phosphoenolpyruvate
   carboxykinase (PCK2). On the other hand, two proteins involved in ROS
   scavenging, mitochondrial superoxide dismutase (SODM) and microsomal
   glutathione S-transferase (MGST1), were upregulated. Cytosolic enzymes
   involved in redox regulation, glutaredoxin-1 (GLRX1) and superoxide
   dismutase (SOD), were also increased. However, heme oxygenase 1
   (HMOX1), which is induced by oxidative stress and prevents the
   activation of catabolic pathways elicited by inflammatory mediators,
   was 50% downregulated after RBP4 treatment. It is also important to
   note that upon hOAC stimulation with RBP4, there was also a strong
   downregulation of multiple structural collagens, like type I, VI, and
   XII, and an upregulation of catabolic interstitial collagenase
   ([109]Figure 4C). Moreover, consistent with the inflammatory role that
   we described for RBP4, there was an upregulation of many
   interferon-related proteins ([110]Figure 4D).

   Employing the data obtained from the proteomic analysis about the
   modulated expression of diverse proteins, we performed different types
   of enrichment analyses. Biological pathway enrichment showed the
   upregulation of glucose transport and metabolism and the downregulation
   of insulin receptor signalling and citric acid cycle ([111]Figure 5A).
   In the cellular component enrichment, we observed that extracellular
   matrix proteins, including several collagens (types I, III, VI, XII,
   and XIV), were reduced ([112]Figure 5B). The clinical phenotype
   enrichment showed that the proteomic profile of the RBP4-treated hOACs
   exhibited significant similarities with arthritis and diabetes mellitus
   clinical proteomic profiles ([113]Figure 5C).

Figure 5.

   [114]Figure 5
   [115]Open in a new tab

   Different types of proteomic enrichments after RBP4 treatment. Pathway
   enrichment analysis (A) shows that there is a depletion of insulin
   receptor signalling cascade, fatty acid metabolism, and citric acid
   cycle pathways. It also shows an enrichment of glycolysis, urea cycle,
   interferon signalling, superoxide radical degradation, glucose
   transport, nicotinate metabolism, and glucocorticoid biosynthesis
   pathways. Cellular component analysis (B) shows a general depletion of
   collagens (types I, III, VI, XII, and XIV), extracellular matrix, and
   mitochondrial proton-transporting ATP synthase complexes. Clinical
   phenotype analysis (C) shows similarities between the proteomic profile
   of RBP4-stimulated chondrocytes and hepatic argininosuccinate synthase
   deficiency, arthritis, and diabetes mellitus. RBP4 (retinol binding
   protein 4).

3.6. RBP4 Induces the Expression of Inflammatory and Chemoattractant Genes in
Synoviocytes

   To provide a better context for RBP4’s impact on joints, we also tested
   its effect on the human synovial cell line SW982. In these cells,
   although RBP4 downregulated the mRNA expression of VCAM, it induced the
   mRNA expression of other inflammatory and catabolic factors like IL6,
   ICAM, NOS2, NAMPT, MMP1, MMP3, and MMP9. In addition, RBP4 also
   strongly induced the mRNA expression of key neutrophil chemoattractant
   genes from the CXCL family, including CXCL1, CXCL2, and CXCL8, while it
   slightly downregulated the homeostatic chemokine CXCL12 ([116]Figure
   6A).

Figure 6.

   [117]Figure 6
   [118]Open in a new tab

   Inflammatory genes expression on a human synovial cell line (SW982)
   after RBP4 treatment. IL6, ICAM, NOS2, NAMPT, MMP1, MMP3, and MMP9 were
   upregulated after RBP4 treatment, but VCAM was downregulated, while
   CCL2 showed no changes. Cellular adhesion-related genes CXLC1, CXCL2,
   and CXCL8 were upregulated, while CXCL12 was downregulated (A). A
   representative image of the flow cytometry experiment. RBP4 produced an
   activation of neutrophils and reduced apopotosis and the canonical
   death pathway of neutrophils, but it did not significantly increase the
   necrotic pathway. Each dot represents one cell and the colour scale
   goes form red (more density of cells) to blue (less density of cells)
   (B). Graphic representation of the neutrophil activation by RBP4
   experiment in figure B (C). RBP4 (retinol binding protein 4), IL6
   (interleukin 6), CCL2 (C-C motif chemokine ligand 2), ICAM
   (intercellular adhesion molecule 1), VCAM (vascular cell adhesion
   molecule), NOS2 (nitric oxide synthase 2), NAMPT (nicotinamide
   phosphoribosyltransferase), MMP1 (matrix metallopeptidase 1), MMP3
   (matrix metallopeptidase 3), MMP9 (matrix metallopeptidase 9), CXCL1
   (chemokine (C-X-C) motif) ligand 1), CXCL2 (chemokine (C-X-C motif)
   ligand 2), CXCL8 (chemokine (C-X-C motif) ligand 8), and CXCL12
   (chemokine (C-X-C motif) ligand 12). (* p < 0.05; ** p < 0.01; *** p <
   0.001; **** p < 0.0001).

3.7. RBP4 Promotes Neutrophils Activation

   Taking aim at the DAMP-like activity of RBP4 and its induction of
   neutrophil chemoattractant genes in synoviocytes, we studied the effect
   of RBP4 on the vitality of primary human PMNs by annexin V-FITC
   apoptosis assay. After 6 h stimulation with RBP4, PMN viability was
   significantly increased in comparison with the untreated cells, whereas
   the number of cells in early and late apoptosis was reduced
   ([119]Figure 6B,C).

4. Discussion

   In this study, we described the concentration of RBP4 in the synovial
   fluid of patients with CA, its positive correlation with glucose
   levels, and its negative correlation with synovial leukocyte count. In
   vitro, RBP4 activated neutrophils and exhibited inflammatory and
   chemotactic activity in hOACs and synoviocytes through TLR4 activation.
   Moreover, it induced proteomic changes in hOACs consistent with an
   inflammatory response, increased catabolism, and decreased anabolism,
   mitochondrial dysfunction, and significant glycolysis promotion. This
   supports the role of RBP4 as a DAMP not only in CA but also in other
   diseases like OA [[120]35].

   Rheumatic diseases are closely related to metabolic alterations like
   diabetes, non-alcoholic fatty liver, obesity, and metabolic syndrome
   [[121]36,[122]37]. RBP4 is an hepatokine that has been related to these
   diseases [[123]9,[124]14,[125]15,[126]16] and also to other pathologies
   related to inflammation [[127]8,[128]38] and articular tissue
   destruction [[129]18,[130]19,[131]20], including OA [[132]23]. RBP4 has
   inflammatory properties across diverse cell types
   [[133]10,[134]39,[135]40] and activates TLR4-mediated responses in
   adipocytes [[136]10]. Despite TLR4 being one of the most important
   receptors of the innate immunity and having a key role in the
   development of several rheumatic diseases [[137]41], there are no
   studies on the specific effect and signalling of RBP4 on different
   articular cells like chondrocytes or synoviocytes. Moreover, although
   the concentration of RBP4 in OA synovial fluid has been previously
   described, little was known about its concentration in CA synovial
   fluid. It is well known that crystalline arthritides have a bigger
   inflammatory component than OA. Nonetheless, the concentration range in
   these patients was between 2.44 and 22.62 µg/mL, which overlaps with
   the concentration range described in OA patients [[138]23], thereby
   suggesting that the articular inflammatory environment does not
   determine RBP4 synovial levels. In line with this, RBP4 levels were
   also not significantly different between pseudogout and gout patients.
   Consistent with this idea, neither inflammatory stimuli nor
   anti-inflammatory drugs were able to modulate the RBP4 mRNA expression
   in chondrocytes in vitro. Even though the levels of synovial RBP4 in
   healthy controls remain to be determined, its median plasma level has
   been determined in a cohort of 150 healthy individuals [[139]42]. Given
   that plasma RBP4 levels has been described to be almost 2.5 times
   higher than synovial levels [[140]23], it is possible to speculate that
   synovial RBP4 levels in healthy controls (15.5 µg/mL) in plasma
   [[141]42] might be lower than those in CA and OA patients.

   RBP4 is tightly related to several rheumatic disease comorbidities,
   like diabetes and cardiovascular pathologies [[142]43]. Therefore, we
   investigated whether the metabolic context of CA patients could be
   related to the synovial concentration of RBP4. We found a significant
   correlation between synovial RBP4 levels alongside total synovial
   glucose and the blood glucose concentration. This finding is consistent
   with previous reports, where RBP4 contributed to insulin resistance and
   glucose intolerance at the local and systemic levels
   [[143]13,[144]39,[145]44]. We also observed that CA patients with
   hypertension exhibited higher synovial RBP4 levels than normotensive
   patients. Supporting this, RBP4 systemic levels were described to
   correlate with the increased prevalence of hypertension [[146]43].
   Interestingly, this significant increase in RBP4 levels in CA patients
   with hypertension was observed in gout patients but not in pseudogout
   patients, which might be explained by the fact that gout is associated
   with more metabolic alterations than pseudogout [[147]25].

   Metabolic alterations drive the worsening of multiple rheumatic
   diseases. Hence, the increased synovial levels in CA patients with
   elevated glucose levels and hypertension led us to investigate its
   effects on articular cells in vitro. Consistent with the inflammatory
   effect of RBP4 in extraarticular tissues, the RBP4 concentrations
   observed in the synovial fluid of CA patients promoted the expression
   of multiple inflammatory and catabolic genes in cultured hOACs and
   synoviocytes. Interestingly, RBP4 synergized with IL1β in chondrocytes,
   thereby promoting the expression of several inflammatory genes (CCL2,
   ICAM, and NOS2). This has not been described before but may be in
   concordance with other studies that showed IL1β upregulation in
   adipocytes upon RBP4 stimulation [[148]10]. Overall, the strong
   inflammatory effect of RBP4 and its synergic activity with IL1β might
   explain the contribution of certain metabolic diseases to the
   development of rheumatic diseases.

   To achieve further insights into the mechanism of RBP4 inflammatory
   actions, we explored RBP4 potential binding to the TLR4 receptor
   through computational chemistry. Data obtained revealed that RBP4
   directly docks to the TLR4 receptor. This suggests a direct activation
   of this receptor by RBP4, which supports previous findings about RBP4
   signalling [[149]10]. Apart from the receptor itself, TLR4 agonist
   recognition can vary depending on the cell type [[150]45]. In
   chondrocytes, TLR4-specific inhibition with CLI-095 partially inhibited
   the inflammatory actions of this hepatokine, which points to RBP4 as an
   articular DAMP.

   Surprisingly, despite the fact that RBP4 promoted cellular inflammatory
   responses, we found that RBP4 levels correlated negatively with
   leukocyte count in the synovial fluid of CA patients. However, even
   though synovial leukocyte count is an inflammatory sign, all of these
   patients had a cell count of over 5000, which is considered a marker of
   an articular inflammatory process [[151]46]. Moreover, a reduced
   leukocyte count has been associated with the phagocytosis of urate
   crystals [[152]47]. In line with this, we found that uric acid levels
   also correlated negatively with leukocyte count. Furthermore, since the
   synovial tissue controls the number of infiltrating cells entering into
   the articular compartment, we explored the effect of RBP4 on human
   synoviocytes in vitro. RBP4 stimulation of these cells upregulated the
   expression of the neutrophil chemoattractant genes, while the
   homeostatic chemokine CXCL12 was significantly downregulated. This
   might produce leukocyte retention in the synovial membrane [[153]48],
   which would explain the negative correlation between RBP4 and the
   synovial fluid leukocyte count observed in CA patients. All these data
   together led us to hypothesise that part of these cells could still be
   in the joint, but in the synovial membrane rather than in the synovial
   fluid. Nonetheless, to discard a pro-apoptotic effect of RBP4 on the
   infiltrating cells, we tested the effect of RBP4 on cultured human
   primary neutrophils, which are the main infiltrating cells in CA
   joints. We observed that RBP4 stimulation of these cells promoted their
   activation, increasing their vitality and reducing their apoptosis,
   which further supports the role of this protein as a DAMP [[154]49].

   Cartilage degradation is a hallmark of the progression of multiple
   rheumatic diseases because it powerfully impairs articular function.
   Apart from RBP4 inflammatory actions in hOACs, we thus explored its
   effect on their metabolism upon in vitro stimulation. We found that
   RBBP4 induced proteomic changes associated with the promotion of the
   glycolytic pathway. This leads to an accumulation of pyruvate that has
   been described to be related to inflammatory response
   [[155]50,[156]51,[157]52] and also to cartilage degradation, as has
   been reported in OA cartilage [[158]53]. Moreover, proteomic results
   also indicated the development of a relevant mitochondrial dysfunction
   upon hOAC stimulation with RBP4. The mitochondrial alterations included
   the upregulation of ROS-related enzymes along with the downregulation
   of oxidative phosphorylation enzymes. Chondrocyte mitochondrial
   dysfunction is widely known to be a key step toward cartilage
   degradation [[159]54]. Consistent with RBP4 activation of TLR4, a
   decrease in mitochondrial activity together with an increase of ROS
   generation have been described as a result of the activation of the
   innate immune responses [[160]55,[161]56].

   Altogether, the metabolic and mitochondrial alterations observed in
   RBP4-stimulated hOACs suggest the presence of a pro-catabolic
   environment. In line with this, proteomic data as well as diverse
   enrichment analysis confirmed a strong inhibition of the expression of
   multiple protein matrix components. Moreover, a pathological enrichment
   analysis revealed that the proteomic profile elicited by RBP4 in hOACs
   was similar to those associated with diabetes and arthritis, which
   further supports that RBP4 might be a common link between systemic
   metabolic imbalance and articular inflammatory and catabolic events. Of
   note, synovial RBP4 levels have been observed to correlate with its
   systemic levels in OA patients [[162]23]. Supporting the key role of
   RBP4 in metabolic diseases, its therapeutic targeting has successfully
   controlled the development of hypertension and insulin resistance in
   animal models [[163]43]. Accordingly, considering the deleterious
   effects of RBP4 in diverse articular cells, we hypothesise that RBP4
   would be an appealing therapeutic target to control the impact of
   diverse RBP4-associated metabolic comorbidities in the joint
   destruction associated with CA and other arthritides.

   Despite all the data shown, we acknowledge that this study has some
   limitations. First, this study does not have an animal model that could
   have underpinned our results. However, we are confident that our
   results with human primary cells provide enough evidence to situate
   RBP4 as a DAMP. Finally, it is important to note that the difference
   between male and female sample numbers could be a limitation.
   Nonetheless, this imbalance is consistent with the grater prevalence of
   these diseases in males [[164]57].

   In conclusion, in this work we confirmed the presence of relevant
   concentrations of RBP4 in the synovial fluid of CA patients that are
   related to the metabolic and cardiovascular contexts of these patients.
   We provide evidence that RBP4 works as DAMP at these concentrations,
   inducing strong inflammatory and metabolic responses in chondrocytes,
   synoviocytes, and neutrophils. These effects might explain the
   contribution of certain metabolic diseases to joint destruction in
   diverse rheumatic diseases like crystalline arthritis.

Supplementary Materials

   The following supporting information can be downloaded at:
   [165]https://www.mdpi.com/article/10.3390/antiox13010124/s1, Figure S1:
   Correlations for RBP4 levels tested by an ELISA assay on synovial fluid
   with blood glucose levels of patients obtained from their medical
   records and separated by sex. In men (n = 21), the correlation remains
   significative while in women (n = 6) is close to significance (A). When
   separating our patients by its hypertension diagnosis and sex,
   hypertense men have higher RBP4 levels, on the other side, we have just
   one sample of non-hypertensive women which does not allow us to perform
   the correspondent test (B), * p < 0.05; Figure S2: Correlations for
   RBP4 levels tested by an ELISA assay on synovial fluid with uric acid
   levels of patients obtained from their medical records. There is not a
   significant correlation between both variables. Table S1. Primers used
   for both mouse and human for all the RT-PCRs performed. Table S2.
   Summary of the ten best results of the molecular docking, showing the
   different molecular interactions and their values together with a final
   score. Results are sorted from best to worse. Table S3. Proteomic
   analysis.
   [166]Click here for additional data file.^ (1.1MB, zip)

Author Contributions

   Concept and design, A.P.-P., E.F.-T., A.A.-P., A.C.-G. and A.J.-M.;
   analysis and interpretation of the data, A.P.-P., M.P.-R., M.G.-F.,
   M.L.-F. and S.B.B.; drafting of the article, A.P.-P., M.P.-R. and
   J.C.A.; critical revision of the article for important intellectual
   content, M.P.-R., A.A.-P. and R.G.; final approval of the article,
   M.P.-R., J.C.A., A.J.-M. and R.G.; provision of study materials or
   patients, A.C.-G., S.B.B., A.J.-M. and R.G.; statistical expertise,
   A.P.-P., M.P.-R., E.F.-T., A.A.-P., M.L.-F. and S.B.B.; obtaining
   funding, A.J.-M. and R.G.; administrative, technical, or logistic
   support, A.P.-P., M.P.-R., M.G.-F., A.C.-G. and J.C.A.; collection and
   assembly of data, A.P.-P., E.F-T., A.A.-P., M.G.-F. and M.L.-F. All
   authors have read and agreed to the published version of the
   manuscript.

Institutional Review Board Statement

   This study fulfilled all the ethical requirements and practice
   protocols for research of the Health Research Institute of Santiago de
   Compostela (IDIS) with code 2016/258.

Informed Consent Statement

   Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

   Data is contained within the article and [167]supplementary material.

Conflicts of Interest

   The authors declare no conflicts of interest.

Funding Statement

   This work was funded by the Instituto de Salud Carlos III (ISCIII)
   through provided pro-jects [grant numbers PT20/00009, FI20/00210,
   CM20/00186, RD21/0002/0025, PI22/00407, PI19/01446, PI16/01870,
   CPII20/00026, CP15/00007] and co-funded by the European Union through
   (EU). This publication is part of the ENDOTARGET project, that has
   received funding from the European Un-ion’s Horizon Europe research and
   innovation programme, with project number 101095084; the Spanish
   Ministry of Science through FECYT [grant number PRECIPITA PR250], the
   Mutua Madrileña Foundation [grant numbers MMA 2018 & MMA 2020], the
   Ministry of Universities [grant numbers FPU17/01706], and UDC Margarita
   Salas RSU.UDC.MS06; and funded by the European Union through
   NextGenerationEU. Support was also associated with personnel as
   follows: E.F-T. [MMA 2018], M.G-F. [FPU17/01706], A.A-P. [IDIS 2019],
   M.L-F. [FI20/00210]. A.P-P. [MMA 2020], A.C-G. [Rio Hortega
   CM20/00186], A.J-M. [SERGAS no grant number], S.B.B. [IDIS no grant
   num-ber], R.G. [ISCIII and SERGAS through the Miguel Servet II program
   CPII20/00026, PT20/00009] and M.P-R. [UDC Margarita Salas
   RSU.UDC.MS06].

Footnotes

   Disclaimer/Publisher’s Note: The statements, opinions and data
   contained in all publications are solely those of the individual
   author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI
   and/or the editor(s) disclaim responsibility for any injury to people
   or property resulting from any ideas, methods, instructions or products
   referred to in the content.

References