Prophylactic inhibition of neutrophil elastase prevents the development of chronic neuropathic pain in osteoarthritic mice

All experiments were performed on male C57BL/6 mice (Charles River Laboratories Inc., QC, Canada) or wild-type (PAR2+/+) and PAR2-deficient (PAR2−/−) mice raised in-house (original breeders developed on a C57BL/6 background from Jackson Laboratories, Bar Harbor, ME, USA). Following arrival at the animal care facility, mice were allowed to acclimate for at least 1 week. Animal weights ranged from 20 to 34 g (8–14 weeks old). Mice were housed at 22 ± 2 °C on a 12:12 h light/dark cycle with standard lab chow and water available ad libitum. All animal care and experimental procedures complied with the ARRIVE and Canadian Council for Animal Care guidelines (http://​www.​ccac.​ca/) and were approved by the Dalhousie University Committee on Laboratory Animals. A total of 255 animals were used in the experiments described here.

The proteolytic activity of neutrophil elastase was determined using the substrate Neutrophil Elastase 680 FAST (NE 680), a commercially available fluorescence agent that is optically silent upon injection and produces a fluorescence signal after specific clevage by neutrophil elastase [19, 41]. Animals were anesthetized (2–4% isoflurane; 100% oxygen at 1 L/min) and the hair was removed from both hind limbs. NE680 (1 nmol/25 μl) was injected subcutaneously over both knee joints and the mice were immediately placed in a prone position in the imaging chamber of an In-Vivo Xtreme imaging system (Bruker Corporation, Billerica, MA, USA). The substrate was excited at 650 nm wavelength and the emitted fluorescence was captured at 700 nm wavelength for a 2.5-s exposure time; the field of view was 10 cm and the lens aperture (fSTOP) was kept at 2. For analysis, identical regions of interest were drawn around both the ipsilateral (arthritic) and contralateral (naïve) joint, and fluorescence intensities were calculated for each knee. The fluorescence level of the contralateral knee joint was subtracted from that of the inflamed ipsilateral knee joint to account for any release of neutrophil elastase in the skin as a result of the injection itself.

Intravital microscopy (IVM) was used for the assessment of leukocyte trafficking within the synovial microcirculation, as previously described [19]. Mice were anesthetized using urethane (25% stock solution; 0.2–0.3 ml intraperitoneal (i.p.)) and a surgical plane of anesthesia was confirmed by the absence of a hindpaw withdrawal reflex. Mice were placed in a supine position on a surgery board, and a heated blanket (SoftHeat HP710-24-3P-S Electric Heating Pad, Kaz Inc., Southborough, MA, USA) was used to maintain core body temperature at 37 °C. The trachea was exposed by making a small longitudinal incision in the skin and was cannulated with PE-60 tubing (Clay Adams, Parsippany, NJ, USA) to allow unrestricted breathing. The right carotid artery and jugular vein were then isolated and cannulated with PE-10 tubing (Clay Adams, Parsippany, NJ, USA) filled with heparinized saline (1 U/ml). The carotid artery cannula was connected in series to a pressure transducer (Kent Scientific Corporation, Torrington, CT, USA), and mean arterial pressure was recorded on a differentially amplified blood pressure monitor (BP-1; World Precision Instruments, Sarasota, FL, USA). The fluorescent dye, rhodamine 6G (0.05%; 0.06 ml), was injected through the jugular vein cannula to stain leukocytes immediately before measurement. Lastly, the skin covering the knee joints (∼ 1 cm long × ∼ 0.5 cm wide) was removed and all superficial fasciae removed to allow an unrestricted view of the joint microvasculature. The surface of the knee was perfused intermittently with warm (37 °C) physiological buffer to prevent tissue desiccation. The knee joint microvasculature was visualized under incident fluorescent light using a Leica DM2500 microscope with a HCX APO L 20X objective and a HC Plan 10X eyepiece (Leica Microsystems Inc., Richmond Hill, ON, Canada; final magnification ×200). Straight, unbranched, postcapillary venules (diameter 20–50 μm) were selected for analysis. Recordings of 1 min duration were made using a BC-71 AVT camera (Horn Imaging, Aalen, Germany). Two leukocyte properties were assessed: (i) rolling leukocytes that move along the venular endothelium with a velocity less than the free-flowing cells in the same vessel and (ii) adherent leukocytes within a 100-μm length of venule which cling to the endothelial wall for at least 30 s. Three videos were captured from three different venules per knee joint and the values averaged.

Secondary allodynia was assessed by application of von Frey hair filaments to the plantar surface of the ipsilateral mouse hindpaw using a modification of the Dixon’s up-down method [42]. Animals were placed in elevated Plexiglas chambers on metal mesh flooring and were allowed to acclimate for 20 min. Once the exploratory or grooming behaviour ceased, a von Frey hair filament was applied perpendicular to the plantar surface of the hindpaw (avoiding the toe pads) until the hair started to bend, and the hair was held in place for 3 s. If the mouse showed withdrawal, shaking, or licking of the hindpaw, a positive response was recorded and the next lower strength hair was applied; if there was no response, the next higher strength hair was applied up to a maximum cut-off level which corresponded to a 4-g bending force. After the first difference in response was observed, four more measurements were made and the pattern of responses was converted to a 50% withdrawal threshold calculated using the following formula: 10[Xf + kδ]/10,000; where Xf = value (in log units) of the final von Frey hair used, k = tabular value for the pattern of the last six positive/negative responses, and δ = mean difference (in log units) between stimuli.

Saphenous nerve sections were removed from all pain experiment animals and g-ratios measured. The saphenous nerve preparation and G-ratio calculation were carried out as previously described [33]. Mice were euthanized at the end of the study with sodium pentobarbital (1000 mg/kg, i.p.), and a section of the ipsilateral saphenous nerve above the knee was collected and fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate for at least 24 h. After fixation, nerves were rinsed three times using 0.1 M sodium cacodylate buffer. The nerve samples were fixed again with 1% osmium tetroxide for 2 h, washed with distilled water, and kept in 0.25% uranyl acetate at 4 °C overnight. The next day, nerve samples were dehydrated with a graduated series of acetone (50%, 70%, 95%, 100% acetone and lastly by dried 100% acetone beads). Nerve samples were subjected to infiltration with epon araldite resin (3:1 ratio of dried 100% acetone to resin for 3 h, 1:3 ratio of dried 100% acetone to resin overnight, then 100% epon araldite resin 2 times for 3 h). The samples were then embedded in 100% epon araldite resin and placed in 60 °C oven for 48 h to cure. Sections (100 nm thick) were cut using a Reichert–Jung Ultracut E Ultramicrotome with a diamond knife and were placed on copper grids then stained with 2% aqueous uranyl acetate for 10 min followed by two rinses with distilled water for 5 min. These rinsed samples were kept in lead citrate for 4 min and rinsed again with distilled water and left to air dry. The samples were viewed using a JEOL JEM 1230 Transmission Electron Microscope (JEOL, Japan) at 80 kV, and one good quality image was captured per animal using Hamamatsu ORCA-HR digital camera at ×2500 magnification. G-ratio, analyzed using the G-ratio plugin on the ImageJ software (https://​imagej.​nih.​gov/​ij), is the square root of the internal axonal area (without myelin) divided by the whole axonal area (including myelin), and is a measure of myelin thickness. All nerve fibers (36 to 113 per section) present in a captured image were measured and were averaged to give a mean g-ratio value for each animal.

Mice were anesthetized (2–4% isoflurane; 100% oxygen at 1 L/min), and an acceptable plane of anesthesia was confirmed by failure to produce a hindpaw withdrawal reflex. The right knee joint was shaved and baseline knee joint diameter was measured using a digital caliper (Control Company, Friendswood, TX, USA). MIA (0.3 mg/10 μl) was injected into the intra-articular space of the knee joint using a 30-G needle inserted through the patellar ligament. The knee was then manually extended and flexed for 30 s to disperse the MIA throughout the joint. The inflammatory changes (joint diameter and leukocyte trafficking) and behavioral pain were assessed at baseline and on days 1, 3, 7, 10, and 14 post-MIA injection.

The proteolytic activity of neutrophil elastase in the inflamed knee joints was determined on days 1 and 14 post-MIA injection. Pharmacologically, the bioactivity of neutrophil elastase can be limited by inhibitors such as sivelestat and serpinA1. Sivelestat (or ONO-5046) is a synthetic, potent, and selective inhibitor of neutrophil elastase with an IC50 of 44 nM [43]. SerpinA1 (or alpha-1 antitrypsin) is a member of the serine proteinase family which primarily inhibits the activity of neutrophil elastase [44] but can also inhibit other serine proteinases including trypsin, chymotrypsin, proteinase 3, and cathepsin G [16]. The dose of sivelestat was selected on the basis of a previous study [19]. SerpinA1 has a long half-life of 15.5 h in mice [45]. The dose of serpinA1 was selected on the basis of pilot experiments in which serpinA1 blocked exogenous neutrophil elastase-induced joint inflammation and pain (data not shown).

A separate cohort of MIA-injected animals was treated on day 1 with the neutrophil elastase inhibitor sivelestat (50 mg/kg i.p.), and proteolytic activity was assessed 4 h later. Another group of MIA-injected animals received treatment with serpinA1 (10 μg i.p.) administered 15 min before and 12 h after MIA injection, and the proteolytic activity was assessed 24 h after injection of MIA.

To assess the functional contribution of neutrophil elastase in the model of experimental OA, two separate cohorts of animals were injected with MIA. The first cohort received treatment with a synthetic inhibitor of neutrophil elastase, sivelestat (50 mg/kg i.p.), administered 10 min before and 240 min after MIA injection on day 0 and once on days 1 to 3. The second cohort received treatment with an endogenous inhibitor of neutrophil elastase, serpinA1 (10 μg i.p.), administered 15 min before and 12 h after MIA injection. Following treatment, inflammation and pain were assessed at several time points over the 2-week time course.

Involvement of PAR2 in MIA-induced inflammation and pain was assessed using the PAR2 antagonist GB83 (5 μg i.p.), administered 10 min before and 120 and 240 min after MIA, on day 0 and once on day 1 (60 min before assessment). MIA was also injected into the knee joints of PAR2 knockout mice. Inflammation experiments focused on day 1, as this is the peak of the MIA-induced inflammatory effect. Behavioral pain was assessed on days 1, 3, 7, 10, and 14 post-MIA injection.

Sodium monoiodoacetate, rhodamine 6G, and urethane were obtained from Sigma-Aldrich (St. Louis, MO, USA). Sivelestat (neutrophil elastase inhibitor; 4-[[[2-[[(carboxymethyl)amino] carbonyl]phenyl]amino]sulfonyl] phenyl ester 2,2-dimethyl-propanoic acid, monosodium salt, tetrahydrate) was obtained from Caymen Chemicals (Ann Arbor, MI, USA). SerpinA1 (neutrophil elastase inhibitor) was obtained from Abcam, Inc. (Toronto, ON, Canada). GB83 (PAR2 antagonist; N-((S)-3-cyclohexyl- 1-((2S,3S)-1-(2,3-dihydrospiro[indene-1,4′-piperidine]-1′-yl)- 3-methyl-1-oxopentan-2-ylamino)-1-oxopropan-2-yl) isoxazole-5-carboxamide) was obtained from Axon Medchem (Groningen, The Netherlands). Neutrophil Elastase 680 FAST was purchased from PerkinElmer (Waltham, MA, USA). Sodium monoiodoacetate, sivelestat, and rhodamine 6G were dissolved in saline. GB83 was dissolved in vehicle (1:1:8 DMSO/cremophor/saline). Physiological buffer (composition—135 mM NaCl, 20 mM NaHCO3, 5 mM KCl, 1 mM MgSO4*7H2O, pH = 7.4) was prepared in-house.

The proteolytic activity of neutrophil elastase within the knee joint was increased on day 1 after MIA injection, and this effect was significantly reduced by treatment with either sivelestat or serpinA1 (Fig. 1b, P < 0.001, one-way ANOVA with Tukey’s test, N = 5–7). By day 14, the proteolytic activity of neutrophil elastase was low compared to day 1 (Fig. 1b, P < 0.0001).

Compared to baseline, intra-articular injection of MIA produced a significant increase in knee joint diameter on days 1, 3, 7, and 10 (Fig. 2a, P < 0.001, one-way ANOVA, N = 6–28). The number of rolling (Fig. 2b, P < 0.01, N = 24) and adherent (Fig. 2b, P < 0.01, N = 27) leukocytes increased significantly on day 1 as compared to baseline then decreased over the remainder of the 2-week time course. Treatment with sivelestat (on days 0–3) blocked the increase in knee diameter (Fig. 2a, P < 0.0001, two-way ANOVA, N = 5–28 per time point), number of rolling leukocytes (Fig. 2b, P < 0.01), and number of adherent leukocytes (Fig. 2b, P < 0.001) across the time course. Likewise, treatment with serpinA1 (on day 0) also blocked joint edema (Fig. 2a, P < 0.0001), number of rolling leukocytes (Fig. 2b, P < 0.05), and number of adherent leukocytes (Fig. 2b, P < 0.001) across the time course. Example images showing the inhibitory effect of neutrophil elastase blockade on leukocyte trafficking are shown in Fig. 2c.

Intra-articular injection of MIA caused a significant decrease in hindpaw mechanosensitivity, indicative of secondary allodynia. This pain appeared on day 1 and persisted to day 14 post-injection (Fig. 3, P < 0.0001, one-way ANOVA, N = 8–9 per time point). Moreover, early blockade of neutrophil elastase using sivelestat or serpinA1 decreased MIA-induced allodynia throughout the time course (Fig. 3, P < 0.0001, two-way ANOVA, N = 8–9 per time point).

The PAR2 antagonist GB83 (administered on days 0 and 1) significantly blocked the day 1 MIA-induced increase in knee joint diameter (Fig. 4a, P < 0.05, one-way ANOVA with Dunnett’s post hoc test; N = 9), and the number of adherent leukocytes (Fig. 4b, P < 0.001). In PAR2 knockout mice, MIA failed to cause a day 1 increase in knee joint diameter (Fig. 4a, P < 0.05), number of rolling leukocytes (Fig. 4b, P < 0.05), or number of adherent leukocytes (Fig. 4b, P < 0.01, N = 9). Intravital images showing the involvement of PAR2 in leukocyte trafficking during this acute phase of MIA are shown in Fig. 4c.

In pain assessment experiments, intra-articular injection of MIA caused significant hindpaw allodynia which appeared on day 1 and persisted to day 14 post-injection in wild-type mice (Fig. 5a, P < 0.0001, N = 5–10). Intra-articular injection of saline had no effect on hindpaw mechanosensitivity over the 2-week time course in wild-type (Fig. 5a) or PAR2 knockout (Fig. 5b) mice. In PAR2 knockout animals, MIA caused a mild secondary allodynia on days 1, 3, and 14 (Fig. 5b, P < 0.05, N = 10). However, the secondary allodynia observed in PAR2 knockouts was significantly less severe as compared to wild-type mice (Fig. 5, P < 0.0001), suggesting knockout of PAR2 partially prevents the development of pain caused by MIA.

Injection of MIA resulted in a significant increase in G-ratio values as compared to saline-injected controls (Fig. 6, P < 0.01, one-way ANOVA with Dunnett’s post hoc test; N = 8), indicating demyelination of the saphenous nerve fibers. Early blockade of neutrophil elastase using sivelestat (Fig. 6, P < 0.05) or serpinA1 (Fig. 6, P < 0.01) blocked MIA-induced demyelination.

Compared to saline-injected wild-type mice, MIA resulted in significant demyelination of the saphenous nerve fibers, as evidenced by an increase in G-ratio (Fig. 7a, P < 0.01, unpaired t test, N = 8). No such demyelination was observed in PAR2 knockout mice (Fig. 7b, P = 0.81, N = 5–10).