If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Pharmacology, Dalhousie University, 5850 College Street, Halifax, Nova Scotia, B3H 4R2, CanadaDepartment of Anesthesia, Pain Management & Perioperative Medicine, Dalhousie University, 5850 College Street, Halifax, Nova Scotia, B3H 4R2, Canada
Address correspondence and reprint requests to: S. Laverty, Département de Sciences Cliniques, Faculté de Médecine Vétérinaire, Université de Montréal, 3200 Rue Sicotte, St-Hyacinthe, Québec, J2S 2M2, Canada. Tel: 1-450-7788100.
To measure the nerve fiber density in synovial membranes from healthy and OA equine joints and to investigate the relationship between synovial innervation and OA severity, synovial vascularity and synovitis.
Twenty-five equine metacarpophalangeal joints were collected post-mortem. The joints were dissected and the macroscopic lesions of the articular cartilage were scored. Synovial membrane specimens (n = 50) were harvested, fixed, sectioned and scored histologically. Immunohistochemical staining and immunofluorescence with S-100 protein, that identifies nerve fibers, and ⍺-actin, that stains vascular smooth muscle, were also performed on site-matched specimens and the relationships between these tissues was interrogated.
The nerve fiber density was higher in the superficial layer (≤200 μm) of the synovium when compared to the deeper layer in control equine joints (mean difference (95% C.I.): 0.054% (0.018%, 0.11%)). In osteoarthritic joints, synovial innervation decreased in the superficial layer with increasing macroscopic OA score (β (SEM), 95% C.I.: −0.0061 (0.00021), −0.0011, −0.00017). The blood vessel density was also higher in the superficial layer of the synovium compared to the deep layer in the control (mean difference (95% C.I.): 1.1% (0.36%, 2.3%)) and OA (mean difference (95% C.I.): 0.60% (0.22%, 1.2%)) equine joints. Moreover, considering all synovial specimens, higher nerve fiber density in the deep layer positively correlated with blood vessel density (β (SEM), 95% C.I.: 0.11 (0.036), 0.035, 0.18).
The reduction in nerve fiber density with advanced cartilage degeneration suggests that peripheral neuropathy is associated with equine OA. Whether this link is associated with neuropathic pain, requires further investigation.
Osteoarthritis (OA) is a slowly progressive degenerative joint disease that leads to chronic pain and disability in all species. Although pain is the primary symptom of OA, and the principal reason for medical consults, surprisingly, the tissue origin and determinants of joint pain remain incompletely understood. There is a huge, pressing, unmet, clinical need to improve the understanding of the pathophysiology and mechanisms of OA joint pain in order to develop targeted, safe and effective drugs to alleviate symptoms in human and animal OA patients.
Pain may be classified broadly into three clinical phenotypes: nociceptive, inflammatory and neuropathic. In joints, nociceptive pain is a physiological response to abnormal movement which results in hyperactivity of articular nociceptors
. OA pain has in the past been considered to be primarily nociceptive; however, an increasing amount of emerging evidence suggests that OA pain also has a neuropathic component and may explain, in part, reported ineffective pain management in patients with this disease
Hallmark OA joint pathology includes articular cartilage degeneration and erosion, subchondral bone sclerosis and resorption, with periarticular and central osteophytes and synovitis. In recent years a strong correlation has been recognized between joint pain and the presence of MRI synovitis in human OA patients
, strongly suggesting that synovitis is a key player in pain perception in OA. The peripheral nervous system is major driver of OA inflammation and alleviation of this synovitis can reduce the development of chronic joint pain
Healthy synovial membrane is composed of a single layer of synoviocytes overlying a loose connective tissue subintima containing blood vessels and nerves. Two main types of nerves are involved in synovial homeostasis: the afferent fibers are responsible for the neurotransmission of mechanosensory information and in conjunction with sympathetic post-ganglionic nerves, these sensory nerves also regulate articular blood flow
, could provide an interesting insight into the relationships between synovial innervation and vascularity.
We hypothesized that the innervation of the equine synovial membrane changes in OA joints. The objectives of the present study were to measure the nerve fiber density in synovial membranes from healthy and OA equine joints and to investigate the relationship between innervation, vascularity, synovitis and joint degeneration.
Joint collection and characterization
Metacarpophalangeal joints were collected from horses at an abattoir. The metacarpophalangeal joint was chosen as it is one of the most frequently affected with equine OA
. The joints were harvested immediately post mortem and transported to the laboratory at 4C for dissection, macroscopic assessment and synovial membrane harvest (within approximately 6 h of death). The joints were opened and scored for severity of macroscopic OA based on articular cartilage changes. Indian ink was applied to the articular surfaces to enhance visualization of cartilage fibrillation. Eight regions in each joint including the distal metacarpus (n = 4 quadrants), first phalanx (n = 2 sides) and sesamoid articular surfaces (n = 2) were scored (Figure S1, Supplementary information online). The scoring system used consisted of 0 = normal; 1 = superficial lesion; 2 = linear fibrillation/erosion; 3 = focal fibrillation/erosion zone; 4 = linear ulceration to subchondral bone; 5 = focal ulceration to subchondral bone <1 cm; 6 = focal ulceration to subchondral bone >1 cm. For the purposes of this study, we elected to arbitrarily categorize the joints into control and OA status based on evidence of visible alterations in the articular cartilage integrity (fibrillation and erosions) compatible with macroscopic OA. Eight sites within the joint were scored and lesions scored from 0 to 2 were considered to be part of normal wear and tear in equine athletes and deemed suitable as controls for an OA joint. Joints with cumulative macroscopic scores ≤16 were categorized as control joints and >16 as OA joints for further statistical analysis.
Synovial membrane specimens were excised from the dorsal pouch (lateral and medial; n = 50), fixed in formalin (10%) and embedded in paraffin. Five micrometer sections were cut with a microtome (Thermo Fisher Scientific HM340E Rotary Microtome). Consecutive sections were then employed for histological staining and immunohistochemistry to allow a comparison of the microscopic synovitis score and the nerve fiber and vascular density (Fig. 1).
Sections were stained with hematoxylin-eosin-phloxine-saffron (HEPS) for histological scoring. A score 0 (absent) or 1 (present) was attributed for each of the following parameters: synoviocyte proliferation/hypertrophy, inflammatory infiltrate, villous hypertrophy, proliferation of blood vessels and cartilage/bone detritus. The maximum synovitis score obtained for a section was five. Sections with a synovitis score of three or higher were grouped for further analysis. The evaluation was performed by a board-certified pathologist (C.G.) with extensive experience in musculoskeletal pathology assessment and who was blinded to the OA score in the joint of origin.
Immunohistochemistry S100 & ⍺-actin
An indirect immunohistochemical technique was employed for the detection of S100-immunoreactive nerves. A peripheral nerve was employed as positive control (Figure S2, Supplementary information online). The sections were deparaffinized, rehydrated and then blocked with normal goat serum (dilution 1:10) in PBS, 1% w/v BSA for 30 min. The S100 reaction was performed by incubating the sections overnight with the primary antibody diluted 1:2500 in PBS and 1% w/v BSA (rabbit S100 from Dako®, Denmark). A normal rabbit serum was employed as primary antibody for negative control (Figure S3, Supplementary information online). Following PBS washes (×3), a biotinylated goat anti-rabbit secondary antibody (Vector Laboratories, California, USA), diluted 1:200 in PBS, 1% w/v BSA was then applied for 45 min. A second cycle of three PBS washes was performed followed by the addition of an avidin-biotin complex labeled with alkaline phosphatase (Vector Laboratories, California, USA) for 45 min. Following another wash cycle with PBS (X3), the final reaction was revealed by the addition of vector red (Vector Laboratories, Burlington, ON, Canada). The tissue sections were then washed with water and counterstained with Harris hematoxylin (Thermo Fisher Scientific, Kalamazoo, USA) for 2 min, differentiated in 1% acid alcohol (20 s), washed in running tap water (1 min), exposed to saturated lithium carbonate solution for 20 s and washed again in running tap water (5 min). The sections were mounted with micromount medium (Leica, Biosystem, Richmont, IL, USA). Antibodies to S100 protein have previously been employed as a general marker of synovial nerve fibers in the human knee
. The sections were prepared as described with the exception that ⍺-actin staining was performed by incubating the sections for 1 h with monoclonal anti-actin, α-smooth muscle – clone 1A4, ascites fluid (Sigma–Aldrich, St-Louis, Missouri, USA) at a concentration of 1:1000 in PBS and 1% w/v BSA.
Immunofluorescence (S-100 and ⍺-actin) was performed on a selected subset of sections to investigate the spatial relationships between the synovial nerves and vessels. Dewaxed sections were subjected to antigen retrieved by heating in the presence of 0.01-M sodium citrate buffer (pH 6.0) 30 min. After pretreatment, the sections were blocked with 2% goat serum (Vector Laboratories, Burlington, ON, Canada) diluted in PBS solution, then incubated overnight at 4°C with the S100 antibody (diluted 1:2500 in PBS and 1% w/v BSA) (rabbit S100 from Dako®, Denmark). Following washes (×3) with PBS, the second antibody anti-α-SMA (dilution, 1/250) (Sigma–Aldrich, Oakville, ON, Canada) was applied (1 h) at room temperature (approx. 22°C). After several washes, slides were incubated with fluorescent dye-conjugated goat anti-rabbit IgG antibody (dilution, 1/1,000) and fluorescent dye-conjugated goat anti-mouse IgG antibody (dilution, 1/500) (Thermo Fisher Scientific, Burlington, ON, Canada) for 1 h at room temperature in the dark. Diamidino-2-phenylindole dye (0.5 μg/mL) (Thermo Fisher Scientific, Burlington, ON, Canada) was applied as a nuclear counterstain. Slides were washed 3 times with PBS solution, then coverslips were added with a solution containing 30% glycerol and 0.4% gelatin (wt:vol) (Thermo Fisher Scientific, Burlington, ON, Canada). Images were acquired with an imaging microscope (Axio Imager M1 equipped with an AxioCam MRm, Zeiss Canada, North York, ON, Canada). Specificity of the staining was confirmed with the appropriate isotype antibody (mouse IgG2a, Sigma–Aldrich, Oakville, ON, Canada) as a control substance for detection of anti-α-SMA antibody and rabbit IgG (Vector Laboratories, Burlington, ON, Canada) as a control substance for detection S100 antibody.
Quantification of nerve fiber density
All the sections were examined with bright field light microscopy and then digitalized (Leica DM4000B microscope coupled with an Allied Vision Prosilica GT1920C camera and the software Panoptiq – magnification: ×200). The entire slides were analyzed using ImageJ® software. Quality inclusion criteria were positive staining of the section and the presence of synoviocytes at the surface or synovium integrity. Two parallel synovial layers (superficial; 0–200 μm from the joint cavity and deep; 200–400 μm) were traced on the saved images
. A point counting method with a 10-pixel grid spacing (7.1 μm grid spacing, 51 μm2 of area) was employed to measure the total nerve fiber area in each layer (Fig. 2). The 7.1 μm grid spacing was chosen as approximately 83% of the S-100 synovial positive fibers have been reported to be less than 10 μm
. The nerve fiber density was calculated as the ratio between nerve fiber area and the layer area. Seven specimens were examined by two different counters (an experienced and a non-experienced counter) to ensure the repeatability of the counting method. Two meetings were organized, one where the experienced counter explained the point counting method, showing an example in a specimen, and a second meeting to adjust the counts. The results of the nerve fiber density were compared after these two meetings.
Quantification of vascular density
The vessel density was also measured in the same superficial and deep synovial layers by a non-experienced counter under supervision of an experienced counter. The total vessel area was determined by circling each vessel and calculating the sum of the area. The vessel density was calculated as the ratio between total vessel area and layer area.
The intra-class correlation coefficient was used to determine agreement for nerve fiber density assessments between the two raters. A linear mixed model with horse id as a random factor (joints nested in horses) and rater identity (two categorical levels) and sample depth (two categorical levels) was employed to compare the counts to identify any rater effect on nerve fiber density in the two different layers. The influence of synovial sample side (medial or lateral) on the histological synovitis score (four categorical levels) was assessed with a Cochran–Mantel–Haenszel test for repeated measures. Mixed linear models with horse id as a random factor (joints nested in horses) were used to determine the association between nerve fiber or vessel density and histological synovitis scores (four categorical levels) and macroscopic OA scores (quantitative scores). For the mixed models, visual inspection of the residual values did not reveal marked deviations from model assumptions. Nerve fiber density and vessel density were expressed as percentages for further description and the arcsine square-root transformation of these percentages was used for analysis. Back-transformed means from the model estimates are provided below. Statistical significance was set at P < 0.05.
Information on specimen inclusion is provided in the flow chart (Fig. 1). Twenty-two metacarpophalangeal joints were included from 12 horses (seven gueldings and five females). The mean (±SD) age of the donors was 11.6 (±5.9) years old. The cumulative macroscopic OA score from included joints ranged from 4 to 46 with a mean (±SD) of 23.8 (±12.7). Macroscopic OA score of control joints are provided in Table S1, Supplementary information online. Some synovial sections were eliminated when judged to be of inadequate quality: three because of absence of synoviocytes on the histological sections and three joints because of no signal detection on S-100 immunostaining despite repeated immunostaining (up to five times). However, some nerve fibers were observed in HEPS-staining of synovial membrane specimens of these three joints. A suspicion of damage of synovial membrane specimens was consequently suspected during storage. Details are provided in the flow chart. Synovial specimens (n = 41) from 22 metacarpophalangeal joints were included in the final analyses.
Histological synovitis score
The histological score of synovitis varied from 0 to 4 with a mean (±SD) of 1.6 (±1.2). There was no difference in the synovitis score between lateral and medial samples revealing that the sample location (lateral or medial) did not affect the score.
Nerve fiber density
The peripheral nerve, used as S100 positive control, stained positively (Figure S2, Supplementary information online) and negative control stained negatively (Figure S3, Supplementary information online). The intra-class correlation coefficient between the two observers for nerve fiber density measurement was 98.1% in the superficial layer and 95.9% in the deep layer. Furthermore, the side location (lateral or medial) of the harvested synovial sample did not influence nerve fiber density. Both medial and lateral specimens were included in all further analyses.
The mean nerve fiber density was higher in the superficial layer of the synovium compared with the deeper layer in control joints (0.13% v. 0.018%; mean difference (95% C.I.): 0.054% (0.018%, 0.11%); P < 0.0001) (Fig. 3). The synovial innervation significantly decreased (β (SEM), 95% C.I.: −0.0061 (0.00021), −0.0011, −0.00017; P = 0.008) with increasing macroscopic OA score in the superficial layer alone. The horses' age was added as a fixed factor in the mixed linear model analysis, and did not influence this relationship. In other words, the density reduction of the superficial nerve fibers was only associated with OA severity and not age. Nerve fiber density was similar across the various synovitis scores in the superficial and in the deeper layers (Fig. 4).
The mean blood vessel density was higher in the superficial synovial layer compared with the deeper layer in control joints (4.9% v. 1.4%; mean difference (95% C.I.): 1.1% (0.36%, 2.3%); P < 0.0001). In OA joints, the mean blood vessel density was also higher in the superficial synovial layer compared with the deeper layer (4.1% v. 1.6%; mean difference (95% C.I.): 0.60% (0.22%, 1.2%); P < 0.0001) (Fig. 5). Furthermore, the vessel density in the superficial layer alone decreased with increasing macroscopic OA score, although this was not statistically significant (β (SEM), 95% C.I.: −0.0016 (0.00077), −0.00034, 0.000014; P = 0.052). When all results (control and OA) were pooled and analyzed, the nerve fiber density in the deep synovial layer alone also was positively associated with vessel density regardless of OA severity and synovitis (β (SEM), 95% C.I.: 0.11 (0.036), 0.035, 0.18; P = 0.005) (Fig. 4).
Representative images of control and OA joints and their corresponding cartilage damage, synovial tissue, S100 and ⍺-actin immunostaining are provided in Fig. 6.
Two synovial membrane specimens were selected for immunofluorescence to illustrate co-localization of vessels and nerves: a control without synovitis (score 0) and a representative sample with vascular proliferation. The control sample had few vessels and nerve fibers. In contrast, the sample with vascular proliferation had an increased nerve fiber density in the deep layer, and nerve fibers adjacent to blood vessels (Fig. 7). This figure illustrates the positive correlation between vessel density and nerve fiber density in the deep synovial layer.
This study provides evidence that there is a greater nerve fiber density in the superficial layers of healthy equine synovial membrane and that it decreases with increasing cartilage degeneration in the joint. The vascularity is also greater in the superficial layer, but, in contrast, increases with nerve fiber proliferation in the deep layer alone. The greater concentration of nerves toward the surface (within 200 μm) of the normal equine synovium is in agreement with observations of healthy human synovial membranes
. The later studies employed an alternative nerve marker, PGP9.5, a general nerve fiber marker. Taken together, these studies in different species, confirm that synovial innervation is plastic and changes with progression of both experimental and spontaneous OA. Furthermore, this nerve alteration could potentially contribute to OA neuropathic pain but this will require additional studies to test any potential link.
Several investigators have also previously identified a reduction in nerve fiber density in other joint structures including the posterior cruciate ligament in human OA knees after arthroplasty
. Peripheral ectopic impulse generation could arise from multiple sites on axons or in neuromas in dysfunctional nerve fibers. These ectopic impulses involve opening of voltage-gated sodium channels and the resultant spontaneous discharge results in abnormal sensations as well as hyperalgesia and allodynia
There have already been numerous, and sometimes contradictory results from investigations on the role of neuropathic pain mechanisms in OA to date. Studies of human knee and hip joints with OA have revealed a contrasting increase in synovial sensory nerve fibers with OA
. The apparent discrepancy between investigations may, in part, be explained by either different stages of the disease studied in spontaneous OA or an inadequacy of experimental animal models to recapitulate the many phenotypes of clinical OA. Animal models of OA often mimic only one aspect of the disease such as instability in surgical anterior cruciate ligament transection (ACLT) models
. Furthermore, the expression of OA pathology in animal models of OA is often rapid (weeks) whereas in clinical OA it is usually years and more chronic and advanced disease in terms of pathology.
The mechanisms that elicit changes in the synovial neural networks in OA remain unknown. The nerves may degenerate as a consequence of OA or, alternatively, the pathogenesis and progression of OA may be facilitated by morphological alterations or reduction of the nerves. A recent study has showed that intra-articular lysophosphatidic acid concentration increased with OA severity and intra-articular injection of this molecule caused joint nerve demyelination, joint damage and pain
A higher nerve fiber density in the deep layer correlated with higher vascularity in the present study when control and OA specimens were pooled. The samples were combined because there was no statistical association between vessel density and OA severity. Sympathetic fibers are found in the deep synovial layer in human knee synovium, located near blood vessels
. A proangiogenic role of the synovial neuropeptides CGRP and Substance P (SP), that are released from nerve endings, has been proposed in addition to their stimulation of vascular receptors for blood flow regulation
Inhibitors of angiogenesis and inflammation could be effective to reduce both structural damage and pain in OA. Anti-inflammatory drugs (Dexamethasone and Indomethacin) and a specific angiogenesis inhibitor (PPI-2458) have all been shown to reduce pain behavior, synovial inflammation, and synovial angiogenesis 35 days after meniscal transection in a rat OA model
. Angiogenesis could also be inhibited directly by antiangiogenic agents, such as vascular endothelial growth factor (VEGF) blockers and receptor tyrosine kinase inhibitors, which have been developed with the aim of reducing tumor neovascularization
. Dual inhibition of angiogenesis and nerve growth could be an attractive therapeutic strategy for patients with OA. The β-Nerve Growth Factor inhibitor tanezumab has proved efficacious in reducing pain in human patients with OA
It is acknowledged that the current study has some limitations. As the samples were harvested in an abattoir, a precise clinical history was unavailable so the link between the innervation changes we observed and pain perception requires additional study. Although S-100 protein is a reliable marker of synovial nerve fibers, it does not permit discrimination between the sensory and sympathetic fibers
. The use of nociceptor-specific markers such as SP and CGRP, could have provided additional information on this point, as these neuropeptides are released from joint primary afferents leading to peripheral sensitization
. However, the selectivity of these antibodies for equine synovium was unsatisfactory for our goal to quantify neuropeptidergic innervation (data not shown). Nevertheless, synovial fluid from horses with joint disease (OA and osteochondrosis) contain a higher concentration of SP than healthy horses
. Moreover a higher synovial concentration of SP correlated with positive response of intra-articular anesthesia of the metacarpophalangeal joint in horses, supporting its role in joint pain generation
. The lack of correlation between the histological synovitis score and nerve fiber density that we measured in horses may be explained by the presence of a less severe synovitis score or a lesser disease duration in the equine specimens we examined.
Many questions arise from this and similar studies. Namely, what contribution does the synovium have to OA pain and what is its importance relative to other joint structures such as the subchondral bone, collateral ligaments and meniscus in disease. Furthermore, it is imperative to identify the mechanisms which cause degeneration of the normal synovial neural network. These studies underpin that the nervous system should be included in the paradigm of the joint being considered as an organ.
The present investigation is the first quantitative study of synovial innervation and vascularity of control and OA equine joints and contributes to knowledge relevant to the pathophysiology of joint pain. In summary a reduction of synovial nerve fibers with progression of spontaneous cartilage destruction in joints is surprising, particularly in light of the link between joint synovitis and OA pain. Further studies are required to confirm the neuropathic origin of OA pain in horses.
SL conceived and designed the study and wrote the manuscript. RP participated in study design, evaluated the histological specimens and wrote the manuscript. CG evaluated the histological specimens and revised the manuscript. HR performed the histological and immunohistochemical analyses. IH evaluated the histological specimens. MPB collected the specimens. GB performed statistical analyses. JJMC participating in study design and revised the manuscript.
Conflict of interest
None of the authors has competing interests.
Role of the funding source
The study was funded by “Fonds en santé équine de l'Université de Montréal” and Zoetis.
Sheila Laverty's laboratory is currently funded by the National Science and Engineering Council of Canada (NSERC) (RGPIN/03836-2014), Quebec Cell, Tissue and Gene Therapy Network (The Cell) (RQ000521).
We would like to thank Melodie Schneider for her help in the counting method.
Appendix A. Supplementary data
The following are the supplementary data related to this article: