Resolvin D3 is dysregulated in arthritis and reduces arthritic inflammation

Animals

Male C57Bl/6 mice (8 wks old) were obtained from Charles River Laboratories (Newton, MA). All animal experiments were in accordance with the Harvard Medical Area Standing Committee on Animals (protocol no. 02570). For some experiments male C57Bl/6 (12 wks old) were obtained from Charles River, Kent or ALX/fpr2/3 knockout (KO) mice generated on a C57Bl/6 background (and back-backcrossed 10 times) in-house at the William Harvey Research Institute, Barts and the London School of Medicine, Queen Mary University of London, UK, as described (12). All animal experiments were performed in accordance with institutional guidelines

LC-MS-MS-based LM metabololipidomics

Deidentified serum from healthy volunteers and RA patients were obtained from DX Biosamples (San Diego, CA). Their demographics are shown in Supplemental Table 1. RA patients were diagnosed with stage III RA by board certified physician and were treated with standard therapies (Supplemental Table 1). For LM profiling, ice-cold methanol containing deuterated internal standards d₈-5S-HETE, d₄-LTB₄, d₅-LXA₄, d₄-PGE₂ and d₅-RvD2 representing each chromatographic region of identified LM (500 pg each) were added to samples prior to processing to facilitate quantification and assessment of sample recovery. LC-MS-MS based metabololipidomics were performed as described in (13). RvD3 was prepared and qualified by LC-MS-MS and spectroscopic analysis as in (11) and second synthetic RvD3 was obtained from Cayman Chemicals (Ann Arbor, MI). Also, the complete stereochemistry of RvD4 was recently determined (14). Multiple reaction monitoring (MRM) with signature ion pairs, Q1 (parent ion) - Q3 (characteristic daughter ion) for each molecule. Data acquisition was performed in negative ionization mode and identification was conducted in accordance with published criteria (13), with minimum of 6 diagnostic ions.

K/BxN Serum Transfer Inflammatory Arthritis Model

For self-resolving and. delayed-resolving arthritis: C57BL/6 male mice were administered arthritogenic K/BxN serum (100 µl, i.p.) at days 0 and 2 for self-resolving arthritis, and with additional challenge on day 8 for delayed-resolving arthritis. Clinical scores were assessed (score criteria per paw: 0, no signs of inflammation; 1, inflammation in one of the following aspects: individual phalange joints, localized wrist/ankle or swelling on surface of paw; 2, inflammation on two aspects of paw; 3, major swelling on all aspects of paw; maximum score = 12 per mouse) to monitor disease development. Joints were harvested at indicated time intervals, and LM metabololipidomics performed as described. In some experiments, mice were taken to moorFLPI-2 Laser Doppler imaging (Moor Instruments, Wilmington, DE) at day 15. For RvD3 treatments: Mice (, C57Bl/6 or ALX/fpr2/3 KO mice) were administered vehicle (0.1% ethanol) or RvD3 (100 ng) in 100 µl of saline i.p. daily, starting from day 0 until completion of experiments at day 6. Clinical scores were assessed as above and paw edema assessed by water displacement plethysmometry (Ugo Basile, Comerio, Italy). On termination of the experiments, arthritic paws and joints were collected for histology staining and LM metabololipidomics.

Partial least-squares discriminant analysis

Partial least squares-discriminant analysis (PLS-DA) was performed using SIMCA 13.0.3 software (Umetrics, San Jose, CA) following mean centering and unit variance scaling of LM amounts. The score plot shows the systematic clusters among the observations (closer plots presenting higher similarity in the data matrix). Loading plots describe the magnitude and the manner (positive or negative correlation) in which the measured LMs/SPMs contribute to the cluster separation in the score plot (15).

Histological Analysis

Joints were fixed and decalcified for 2 weeks prior to embedding in paraffin. Sections (8 µm) were stained with Haemotoxylin & Eosin and standard light microscopy used to determine the degree of neutrophil infiltration.

Ligand Selectivity Using Electric Cell-Substrate Impedance Sensing (ECIS)

hALX-transfected chinese hamster ovary (CHO) cells were plated at 1×10⁵ cells/well in HAM-F 12 buffer containing 10% FCS and G418 (37°C) for 36 h. Then the medium was changed to serum free HAMs media and vehicle or RvD3 (100 nM) were added and impedance across the cell monolayer measured using an ECIS (Applied Biophysics, Troy, NY) to monitor ligand-receptor interactions as in (16).

Statistics

Data are presented as mean ± s.e.m. The criterion for statistical significance was P < 0.05 using Student's t test (two groups), one-way ANOVA (multiple groups) or two-way repeated measure ANOVA (two groups over time) followed by a post hoc Bonferroni test using GraphPad Prism 6.

SPMs are temporally regulated in joints during self-limited inflammatory arthritis

To investigate whether SPMs are temporally regulated in joints during inflammation-resolution we used an established murine model relevant to human arthritis that is driven by the innate immune response (17, 18). Mice injected with K/BxN arthritogenic serum (100 µl/mouse, i.p.) at day 0 and day 2, developed a self-resolving polyarthritic inflammatory response that reached a maximum clinical score at day 11, and subsequently declined (Fig. 1a). The reduction in clinical scores corresponded with loss of leukocytes from the joints (Fig. 1b), a key histological feature of resolution (1).

Using liquid chromatography tandem mass spectrometry (LC-MS-MS)-based metabololipidomics we next investigated the temporal LM-SPM profiles in self-resolving joints. LM identified, include E and D-series resolvins, protectins, maresins from the omega-3 (i.e. EPA and DHA) bioactive metabolomes, and lipoxins as well as prostanoids and LT from the arachidonic acid bioactive metabolome (Fig. 1c, Supplemental Fig. 1). All LMs were identified in accordance with published criteria as illustrated for RvD3 (see Material and Methods; Fig. 1d; Supplemental Fig. 1).

Distinct LM profiles were identified in joints for each time interval prior to (day 0; naïve) and after induction of self-resolving arthritis using principal component analysis (PCA; Fig. 1d).) There were higher levels of RvD4, MaR1, RvE1 and 15R-LXA₄ in naïve compared with arthritic paws. Following serum challenge there was an increase in pro-inflammatory eicosanoids including PGE₂, PGD₂ and TXB₂, LTB₄, with a contaminant decrease in most SPMs identified (Fig. 1d; Supplemental Fig. 1) demonstrated by a downward shift in the 4- and 8-day clusters (arthritis) on the score plot (Fig 1d, left). This was followed by an anti-clockwise shift in the 16 d cluster (resolution) back towards the 0 day cluster (Fig 1d). Of note, levels of specific SPM, including the D-series resolvins RvD1, RvD2 and RvD3 increased in arthritic joints during the resolution phase (day 16; Supplemental Fig 1). These findings demonstrate that endogenous SPM pathways are active in murine joints that are temporally and differentially regulated during self-resolving inflammatory arthritis.

Shorter resolution intervals and specific SPM in self-resolving arthritis

Since select SPMs were temporally regulated in self-resolving joint inflammation, we investigated their levels during delayed-resolving inflammatory arthritis to determine which of these tissue protective molecules may be directly involved in resolving joint inflammation. First, to provide quantitative analyses of resolution (i.e. remission) of joint inflammation we applied resolution indices, as are widely used to calculate the loss of neutrophils from site of inflammation (19), using clinical scores to quantify the resolution interval (R_i). Here, the R_i is the time interval between when clinical scores reach maximum (T_max; day 8) and the time it reaches 50 % remission (T₅₀; day 12), giving an R_i of 4 d (Fig. 2a). Additional challenge with K/BxN-serum at day 8 prolonged arthritis with delayed resolution, with a T₅₀ of 17 d, or an R_i of 9 d (Fig. 1a). At day 15, mice were taken for laser Doppler imaging to further assess the degree of joint inflammation-resolution in comparison with unchallenged mice (without K/BxN serum, Fig. 2b). Increases in tissue perfusion monitored via laser Doppler were consistent with increases in joint inflammation. These results indicate that resolution of inflammatory arthritis is delayed following an additional K/BxN-serum challenge, which may be associated with an altered LM-SPM profile.

We next assessed the local endogenous LM-SPM profile in self-resolving joints and compared it to the delayed-resolving joints. Differences in local paw joint LM-SPM profiles were assessed using PLS-DA (Fig. 2c). To identify which variables were responsible for the separation between the clusters (Fig. 2c, left), the variable influence on the projection (VIP) score >1.0 was used to select the most significant variables contributing to the cluster separation. The self-resolving cluster was characterized by higher levels of RvD3, MaR1 and PD1 (VIP >1.0;Fig. 2c, right), whereas, the delayed-resolving cluster was associated with higher levels of pro-inflammatory LM, such as TxB₂, and PGE₂ (Fig. 2c right). Taken together, these findings indicate that delayed resolution dysregulates endogenous resolution programs in arthritic joints, increasing local pro-inflammatory mediators and diminishing SPM.

Dysregulated LM-SPM profiles in human rheumatoid arthritis patients

Toward human translation, we also compared the LM-SPM profile in serum from healthy individuals and patients with RA. All LM were identified in accordance with published criteria as illustrated for RvD1 and RvD3 (Fig. 3a). Quantification using MRM gave significantly lower levels of RvD3 (4.3 ± 1.4 vs. 17.7 ± 2.3 pg/mL, p=0.008) and PGD₂ (46.8 ± 9.8 vs. 95.7 ± 13.9 pg/mL, p=0.046), with a similar trend for RvD4 (1.1 ± 0.3 vs. 3.3 ± 0.9 pg/mL, p=0.083), found in serum from RA patients compared with healthy individuals (Supplemental Table 1). It is important to note that a larger sample size is needed to confirm these. Following identification using the LC-MS-MS fractions with matched mass spectra diagnostic ions and retention time of authentic standards, the data was subjected to multivariate analysis. The PLS-DA score plot demonstrated a clear separation between the healthy and RA serum clusters (Fig. 3b, left). The loading plot, indicating the magnitude and the manner (positive or negative correlation) each of the LM-SPM contribute to the cluster separation, showed that serum from healthy individuals was characterized (VIP score ≥ 1.0) by higher RvD3, RvD4, RvE3, 15R-LXA₄ and PGD₂ levels, while the RA serum was associated with higher TxB₂ levels (Fig. 3b, right). Hence, these results in human RA patients together with the in vivo findings in murine arthritis suggest that select SPMs, such as RvD3, are deregulated in arthritis.

RvD3 limits arthritis severity and joint inflammation in mice

Having found that RvD3 was dysregulated both in delayed-resolving arthritic joints (Fig. 2c) and in human RA serum (Fig. 3), we next sought to determine its pharmacological actions in inflammatory arthritis. Mice were challenged with K/BxN-serum (100 µl, i.p.) on day 0 and 2. Daily administration of RvD3 (100 ng/mouse, i.p.) starting at day 0 significantly reduced clinical scores by 35–60% (Fig. 4a) and hind paw edema by 50–85% (Fig. 4b) during the course of the experiments, with a significant reduction observed as early as day 2 (Fig. 4a,b). Histological analysis showed fewer leukocytes at day 6 in joints from arthritic mice administered RvD3 (Fig. 4c). Hence, these results demonstrate that RvD3 alleviates arthritis progression, reducing edema and limiting leukocyte numbers in murine K/BxN-serum induced arthritis.

RvD3 reduces local eicosanoid levels in murine inflammatory arthritis

Since RvD3 reduces eicosanoid production in acute peritonitis (11), we next assessed its ability to regulate local joint eicosanoid levels in murine inflammatory arthritis. We preformed LC-MS-MS based LM metabololipidomics with murine arthritic joints (day 6) to identify potential mechanisms underlying the anti-arthritic actions of RvD3. Treatment with RvD3 resulted in significant (p< 0.01) reduction in local joint levels of LTB₄ (34.7 ± 10.9%), PGE₂ (67.7 ± 6.9%), PGD₂ (54.3 ± 15.9%) PGF_2α (61.0 ± 10.1 %) and TxB₂ (50.0 ± 3.6 %; Fig. 4d). We also identified the non-enzymatic 8-iso-PGF_2α, indicative of oxidative stress, which was also reduced in RvD3 treated mice (Fig. 4e). Taken together, these results indicate that RvD3’s potent actions limiting paw joint inflammation include regulating local eicosanoids, characteristic of SPMs (1).

The arthritis-protective actions of RvD3 are lost in ALX/FPR2 deficient mice

We next tested whether RvD3 activates the pro-resolving receptor ALX/FPR2, which mediates the actions of LXA₄ and RvD1 (1). Expression of this receptor in transfected CHO cells demonstrated that RvD3 at 100 nM directly activated this receptor (Fig. 5a). Since mice lacking ALX/FPR2 exhibit a more pronounced and prolonged joint response compared to WT counterparts (20), we next investigated whether the arthritis-protective actions of RvD3 were altered in these mice. We found that, in ALX/FPR2 deficient mice, RvD3 did not reduce clinical scores, hind paw edema or LTB₄ levels (Fig. 5b–d). Thus, taken together, these results suggest that RvD3 activates ALX/FPR2 in vivo in mice that, at least in part, mediates RvD3 protective actions in this murine model of experimental arthritis.