TIMP-1 attenuates the development of cutaneous inflammation-induced hypersensitivity

Unresolved inflammation is a significant predictor for developing chronic pain, and targeting the mechanisms underlying inflammation offers opportunities for therapeutic intervention. During inflammation, matrix metalloproteinase (MMP) activity contributes to tissue remodeling and inflammatory signaling through proteolytic maturation of cytokines. MMP activity is regulated by tissue inhibitors of metalloproteinases (TIMPs) 1-4. TIMP-1 and -2 have known roles in pain, but only in the context of MMP inhibition. However, TIMP-1 also has receptor-mediated cell signaling functions that are not well understood. Here, we examined how TIMP-1-dependent cell signaling impacted inflammatory hypersensitivity and ongoing pain. We found that hindpaw injection of complete Freund’s adjuvant (CFA) increased keratinocyte-derived TIMP-1 that peaked 3 days following inflammation, when mechanical hypersensitivity began to emerge in WT mice. These data suggest that TIMP-1 expression inhibits the development of inflammatory hypersensitivity. To examine this possibility, we injected TIMP-1 knockout (T1KO) mice with a diluted CFA mixture to examine how subtle cutaneous inflammation affected behavioral hypersensitivity. T1KO mice exhibited rapid onset thermal and mechanical hypersensitivity at the site of inflammation that was absent or attenuated in WT controls. We also found that T1KO mice exhibited hypersensitivity in adjacent tissues innervated by different sets of afferents, and skin contralateral to the site of inflammation. Replacement of recombinant murine (rm)TIMP-1 alleviated hypersensitivity when administered at the site and time of inflammation. To examine the MMP-dependent and -independent mechanisms of rmTIMP-1, T1KO mice were administered full-length rmTIMP-1, the N-terminal region (TIMP-1(N)) with MMP-inhibitory properties, or the C-terminal region (TIMP-1(C)) that retains receptor signaling function. Each of the peptides prevented inflammatory hypersensitivity, suggesting that rmTIMP-1 acts through mechanisms that . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019;


INTRODUCTION
Tissue inhibitors of matrix metalloproteinases (TIMPs) and matrix metalloproteinases (MMPs) are released during tissue damage to facilitate tissue remodeling through degradation and reorganization of the extracellular matrix (ECM) (Gardner and Ghorpade, 2003;Nagase et al., 2006;Ries, 2014). During this process MMPs also engage an inflammatory response through proteolytic maturation of cytokines, and both of these activities is regulated through a 1:1 stoichiometric interaction with one of four tissue inhibitors of metalloproteinases (TIMP-1, -2, -3, -4) (Huang et al., 2011). The interaction between MMPs and TIMPs is tightly controlled, but research has shown that during tissue damage, dysregulation in the balance of MMPs and TIMP function can lead to pathological conditions such as arthritis, multiple sclerosis, Parkinson's Disease, cancer, and even chronic pain (Nakagawa et al., 1994;Nagase et al., 1999;Kouwenhoven et al., 2001;Yang et al., 2002;Brew and Nagase, 2010). Studies examining the role of MMPs in pain specifically have shown that increased MMP-2 and -9 activity contribute to increased painrelated behavior in response to injury that are reversed by MMP antagonism (Kawasaki et al., 2008;Ji et al., 2009;Li et al., 2016;Remacle et al., 2018). These findings contributed, in part, to the development of several small molecule drugs that directly target and inhibit MMP activity.
However, more than 50 clinical trials examining the efficacy of these drugs were discontinued due to the emergence of adverse events, including musculoskeletal pain (Cathcart and Cao, 2015;Martinho et al., 2016). While these trials indicated that specific targeting of MMP activity alone was not an effective strategy for pain treatment, they do suggest that additional mechanisms related to MMP activity may contribute to pain and its inhibition, a conclusion drawn from the very research that spawned the development of the MMP inhibitors; that endogenous inhibitors of MMP activity, such as TIMP-1 and -2 decrease pain-related behavior.
TIMP-1 has been best characterized as an endogenous inhibitor of MMP activity. Indeed, TIMP-1 regulates 14 of the 24 known MMPs (Baker et al., 2002;Gardner and Ghorpade, 2003;Nagase et al., 2006;Kawasaki et al., 2008), and has been shown to prevent the development of mechanical and thermal hypersensitivity following nerve damage (Kawasaki et al., 2008;Martinho et al., 2016). However, this identified role was characterized purely in the context of MMP inhibition. Interestingly, the structure of the protein allows it to function outside of its wellcharacterized role as MMP-inhibitor, an effect that is produced by binding of its N-terminal domain to MMPs. There is now mounting evidence that the C-terminal domain binds to membrane bound receptors, including CD63. Binding of TIMP-1 to CD63 engages intracellular signaling events that allow TIMP-1 to act as a trophic factor in the initiation of cell migration and differentiation (Gardner and Ghorpade, 2003;Jourquin et al., 2005;Thorne et al., 2009;Moore and Crocker, 2012;Claycomb et al., 2013).
Because TIMP-1 and MMPs can be up-regulated in concert with one another during tissue damage and repair, such as peripheral nerve injury (Parkitna et al., 2006;Huang et al., 2011;Kim et al., 2012;Yokose et al., 2012), it is difficult to disentangle how TIMP-1 regulates of tissue remodeling/repair and pain independent of one another.
Inflammation is a core component of the nerve injury process, and, in general, is a significant predictor of pain chronicity (Kehlet et al., 2006). Therefore, we used a model of cutaneous inflammation to examine the effects of TIMP-1 signaling on pain in the absence of frank tissue damage. We found that hindpaw injection of complete Freund's adjuvant (CFA) induced TIMP-1 expression in keratinocytes prior to the emergence of hypersensitivity in wildtype (WT) mice, suggesting that the release of TIMP-1 at the site of inflammation inhibited development of hypersensitivity. Supporting this conclusion, we found that TIMP-1 knockout (T1KO) mice exhibited robust hypersensitivity local and distal to the site of inflammation that was prevented with the administration of recombinant murine (rm)TIMP-1 and the individual N-and C-terminal constructs at the time of CFA-injection. This result suggests that cell-signaling mechanisms may also contribute to the antinociceptive effects of TIMP-1. In addition, inflammation in WT and T1KO mice did not result in the activation or expression of cutaneous MMP-9 or cytokines in a genotype-specific manner. Finally, we found that administration of rmTIMP-1 prevented ongoing inflammatory pain and evoked mechanical sensitivity in WT mice. Collectively, our data suggests that TIMP-1 regulates the allogenic properties of inflammation and that TIMP-1 may be a target for improving pain management.

MATERIALS AND METHODS
Animals: Experiments were conducted using 8-12-week-old (20-30 g) male WT (C57BL/6; Jackson Laboratories, Bar Harbor, ME) and T1KO mice that were group housed, and maintained in a temperature-controlled environment on a 12 hr light-dark cycle with free access to food and water. TIMP-1 knockout (T1KO) mice (Lee et al., 2005) were backcrossed onto a C57BL/6 background for greater than 13 successive generations and bred in-house as a homozygous line (Crocker et al., 2006). All studies were approved by the UConn Health Institutional Animal Care and Use Committee and treated in accordance with published NIH standards.

Complete Freund's Adjuvant (CFA):
To produce an acute, local inflammatory response, we subcutaneously (s.c.) injected the right hindpaw of mice with emulsified (50% in 10 μL) CFA (Sigma, St. Louis, MO). To assess primary hypersensitivity (i.e. at the site of inflammation) we administered CFA into the glabrous skin or ventral surface of the right hind paw. Conversely, secondary hypersensitivity was assessed in skin that was adjacent or contralateral to the site of inflammation. All samples were compared to naïve controls because in a pilot experiment we . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; found that vehicle injection alone caused increased sensitivity in T1KO mice ( Figure 1). While this result is interesting and suggests that subtle perturbations cause robust alterations in sensory thresholds, adding saline-treated mice confounds our ability to examine the effects of inflammation. Therefore, to interpret the effects of inflammation per se, naïve mice were used as comparison controls. The literature is also mixed on the use of vehicle controls in experiments using CFA, and our experiments are in line with previously published work (Allchorne et al., 2005;Jankowski et al., 2012;Imbe and Kimura, 2017).
von Frey testing: All mice were place into transparent Plexiglas chambers (radius = 32 mm, height = 108 mm) on an elevated mesh screen and were allowed to acclimate for a minimum of 1 hour before testing. To assess mechanical sensitivity, the plantar surface of the right hind paw was stimulated using von Frey filaments using the up-down method described by (Dixon, 1980). Nocifensive responses were counted as robust flexion responses, paw shaking, or paw licking and subtracted from individual baseline threshold to account for inter-subject variability. Data are presented as paw withdrawal thresholds (PWT; in grams).
Thermal hyperalgesia: Thermal hyperalgesia to radiant heat was assessed using a Hargreaves apparatus (Harvard Apparatus; Holliston, MA) (Hargreaves et al., 1988). Briefly, all mice were . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; placed in transparent Plexiglas chambers (radius = 32 mm, height= 108 mm) on top of a framed glass panel and were allowed to acclimate for a minimum of 1 hour before testing. Following the acclimation period, an infrared (IR) beam was aimed at the plantar surface of both hind paws. The intensity of the IR beam was chosen to produce average baseline paw withdrawal latency (PWL) of 15-20 seconds. Stimuli were presented 5 times in an alternating fashion between each hind paw with 5 minute intervals between successive stimulus exposures. PWLs collected from each paw were then averaged and analyzed. A 30 seconds exposure cutoff was employed to prevent tissue damage.
Conditioned Place Preference (CPP): CPP was used to assess ongoing pain in WT mice. A 3day single trial protocol was used. On day 1, all mice freely explored a 3-chamber CPP box for 15 minutes prior to injection with CFA or saline. Preconditioning (baseline) behavior was analyzed using automated software (ANYMAZE, Stoelting) to ensure there were no baseline differences in the time spent in any of the chambers. On day 2 (conditioning day), all mice received an intrathecal (i.t.) injection of saline (5 μL volume) under isofluorane anesthesia and upon waking (within 2 min) were confined into the pre-assigned pairing chamber for 30 min. They were then returned to their home-cages for 4 hours. All mice then received an intrathecal (i.t.) injection of clonidine (2 μg/uL; 5 μL volume) through lumbar puncture and upon waking were confined to the opposite pairing chamber for 30 min. Vehicle and clonidine paired chambers were randomly assigned and counterbalanced between animals. On day 3 (test day), mice were returned to the CPP apparatus and allowed to freely explore all chambers across 15 min. The total time spent in each chamber was assessed using automated software (ANYMAZE). Conditioning day was 20-24 hours following CFA injection as previous research indicates ongoing pain is observed at this time-point . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
Tissue collection: Mice were anesthetized with a lethal dose of ketamine and xylazine mixture (90/10 mg/kg, respectively), intracardially perfused with ice cold 0.9% saline, prior to the dissection of ipsilateral hairy skin, L2 -L3 DRG, and L2 -L3 spinal cord segments. Tissues were collected following the completion of behavior or at designated time-points for molecular analysis.
Enzyme-linked immunosorbent assay (ELISA): Protein was extracted through homogenization in ice-cold RIPA buffer/protease inhibitor cocktail and spun for 20 min at 4°C at 18,000 rcf. Each sample's total protein concentration was determined using Pierce BCA Protein Assay Kit

Immunohistochemistry (IHC):
Hairy skin (n=3/condition), glabrous skin (n=3/condition), and DRG (n=3/condition) from WT mice were excised and incubated in 0.06% brefeldin A (BFA) in serum free HBSS for 20 min at room temperature. Half of the samples were incubated in inflammatory soup (IS) (10uM; bradykinin triacetate, histamine dihydrochloride, serotonin hydrochloride, prostaglandin E2 dissolved in normal cerebral spinal fluid, pH 6.0) or serum-free media (Kessler et al., 1992;Hachisuka et al., 2016). Incubation in IS or serum-free media occurred for 24 hours (Kessler et al., 1992). Spinal cords from inflamed or naïve WT mice were isolated 24 hours following CFA treatment or from designated naïve controls. Prior to tissue collection, . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
MMP-9 colorimetric activity: Protein was extracted from the hind paw of wildtype and T1KO mice 1 day post-CFA injections (s.c., 10μL) or from designated naïve controls for a high throughput screening of MMP-9 activity. Gelatinase activity was measured using the SensolyteGeneric MMP colorimetric assay kit (Anaspec, Fremont, CA). Samples were run in duplicate and end-point enzymatic activity was analyzed using a glutathione reference standard.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; RT-qPCR: Total RNA from WT and T1KO was extracted from naïve and inflamed samples 1 day post CFA administration using a RNeasy Mini Kit (Qiagen, Venlo, Netherlands). To quantify cutaneous TIMP-2 and TIMP-4 mRNA expression, equal amounts of cDNA were synthesized using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA) and mixed with SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA) and 2uM of both forward and reverse primers (see Table 1). GAPDH was amplified as an internal control.
The threshold crossing value was noted for each transcript and normalized to the internal control.
The relative quantitation of each transcript was performed using the Ct method and presented as fold change relative to naïve WT expression.

Gene Forward Reverse
Timp2 Statistical analysis: All data were analyzed using one-way or mixed designs Analysis of Variance (ANOVA). Post hoc analyses were performed using Tukey's HSD, and statistical significance was determined using a p<.05. Statistical analysis was performed using SPSS (Version 25). Since ANOVAs rely on linear relationships among data, and not all effects can be resolved using linear based statistical tests, we used trend analyses (e.g., contrasts) to test for significant nonlinear relationships in some of our behavioral analyses. An added benefit of this approach is that trend analyses are more robust than ANOVAs (Tabachnick and Fidell, 2007). For the CPP data, clonidine induced CPP was assessed using a 2-way repeated measures ANOVA with post-hoc analysis of pre vs post conditioning time spent in the clonidine paired chamber for each treatment . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; group using Sidak's multiple comparisons test. Between groups analysis was performed on difference scores calculated as post-conditioning (-) pre-conditioning time spent in the clonidine paired chamber.

Cutaneous TIMP-1 expression is upregulated prior to the onset of inflammatory hypersensitivity
To determine whether cutaneous inflammation alters the expression of TIMP-1 in tissues along the peripheral sensory circuit, we injected emulsified CFA (10 μL, s.c.) into the hairy skin of the ipsilateral hind paw and collected spinal cord (SC; L2-L3), dorsal root ganglia (DRG; L2-L3), and hairy skin over the course of 7 days. We found that inflammation did not alter the overall expression of TIMP-1 protein in SC or DRG, all F's > 1.13, p>.05 (Figure 2 A, B). However, we observed a significant increase in cutaneous TIMP-1 protein 1, 3, 5, and 7 days following CFA administration, F(4,19) = 37.54, p<.01 ( Figure 2C). To confirm the above results, and to localize the cellular source of TIMP-1 expression, immunohistochemistry (IHC) was performed on DRG and skin samples incubated in vitro with or without inflammatory soup (IS) (Kessler et al., 1992), as well as spinal cords following in vivo inflammation. Although overall TIMP-1 expression levels were unaltered in the spinal cord and DRG following inflammation, we found that TIMP-1 colocalized with glial fibrillary acidic protein (GFAP) expressing cells following inflammatory stimulation, demonstrating that astrocytes (Figures 2D) and satellite glial cells ( Figure 2E) appear to increase levels of TIMP-1 expression following inflammation (Huang et al., 2011;Welser-Alves et al., 2011). We also found that TIMP-1 expression was upregulated in K14-positive keratinocytes in both hairy and glabrous skin following inflammatory stimulation ( Figure 2F; glabrous skin data not shown).
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; To associate the expression of cutaneous TIMP-1 with the development of mechanical hypersensitivity, we assessed paw withdrawal thresholds (PWT) on the plantar surface of the hind paw for 7 days following CFA injection into the dorsal, hairy skin. When we compared the temporal expression of TIMP-1 to the time course of behavioral sensitivity, we found that TIMP-1 protein levels peaked 3 days following CFA administration (see Figure 2C), at a time when mice developed mechanical allodynia, F(2,12) = 43.94, p< 0.05, Figure 2G. We also assessed TIMP-1 expression in the ventral surface of the hind paw, where we assessed mechanical stimulation and found that CFA-injection into hairy skin caused a concomitant increase in TIMP-1 in glabrous K14-positive keratinocytes (data not shown). Together, these data indicate that cutaneous inflammation induces the expression of TIMP-1 in keratinocytes at the time of inflammation and prior to the onset of mechanical allodynia.

Mice lacking TIMP-1 exhibit hyperalgesia in inflamed and uninflamed cutaneous tissues
To determine whether endogenous TIMP-1 expression is important for the normal progression of hypersensitivity, we used a global TIMP-1 knockout (T1KO) mouse strain. We first assessed behavioral responsiveness to radiant heat on the plantar surface of the hind paw following s.c. administration of a diluted CFA solution. We chose to use a diluted CFA solution because our preliminary experiments suggested that exposure to slight challenges significantly alters sensitivity in T1KO mice (see Figure 1). To ensure that any potential differences in responding to inflammatory stimulation were not due to preexisting differences in sensory thresholds between strains, we measured baseline responding to radiant heat and found no significant differences in paw withdrawal latencies (PWL), F(1, 31) = .47, p>.05 ( Figure 3A). Our analysis failed to detect any significant differences in responding between naïve and WT mice administered diluted (e.g., subthreshold) CFA ( Figure 3A). Interestingly, analysis did show that inflamed T1KO mice . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
Next, we assessed mechanical response thresholds (von Frey) following diluted CFA administration. Analysis of baseline responses to mechanical stimulation did not reveal any significant differences between genotypes F(1, 36) = .34, p < .05 ( Figure 3C). We did find that CFA administration reduced mechanical response thresholds in both genotypes F(1, 34) = 17.61, p < .01. However, T1KO mice exhibited greater mechanical hypersensitivity 1 day following CFA treatment, compared to WT controls, all Fs < 4.59, p <.05 ( Figure 3D). Thus, we concluded that the presence of TIMP-1 delays the onset of hypersensitivity, and that the absence of TIMP-1 results in a rapidly developing hypersensitivity.
To determine whether the rapid emergence of inflammatory hypersensitivity in T1KO mice was due to compensatory expression of Timp2 or Timp4 mRNA, we examined cutaneous expression of each transcript in naïve and inflamed T1KO mice. We found no significant differences in the basal expression of either transcript in WT or T1KO mice. However, Timp2 and Timp4 expression decreased in T1KO mice following inflammation ( Figure 4A, B), suggesting that genetic deletion of TIMP-1 does not result in a compensatory response from other TIMPs, all Fs < 11.65, p < 0.01.
Our current data demonstrate that cutaneous TIMP-1 is an early emergent protein following inflammation, and that in the absence of TIMP-1 the development of hypersensitivity is accelerated. Therefore, TIMP-1 signaling may also have important implications for regulating the development of inflammatory hypersensitivity in tissue adjacent to the site of inflammation that is innervated by afferent terminals that are different from those that innervate inflamed skin. To test this possibility, we assessed the development of hypersensitivity in the glabrous skin following . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; injection of diluted CFA into hairy skin in both T1KO and WT following baseline assessment of sensitivity. Again, we observed no genotype-specific differences in baseline reactivity, and because of this consistent finding, we will no longer present data depicting baseline behavioral reactivity. Analysis using an ANOVA revealed that T1KO mice, relative to WT mice, exhibited increased sensitivity to mechanical stimulation on the plantar surface of the hindpaw following inflammation of hairy skin that was not temporally dependent, all Fs < 4.41, p <.05, (Figure 5A, B). Additional analysis using trend analyses, however, revealed that inflamed T1KO mice exhibited increased inflammatory hypersensitivity 1 day following CFA treatment when compared to all other mice, F(1,57) = 11.55, p <.01 ( Figure 5A).
To examine whether inflammatory hypersensitivity on the plantar surface of the hindpaw could be prevented by administration of rmTIMP-1, and the potential signaling mechanism by which the effect of rmTIMP-1 alleviates hypersensitivity, T1KO mice received injections of 3 different recombinant constructs of the TIMP-1 protein, full-length rmTIMP-1 [TIMP-1(FL)], the truncated N terminus peptide [TIMP-1(N)] that retains MMP inhibitory function but no cell signaling capacity, or the truncated C terminus peptide [TIMP-1(C)]) that lacks MMP inhibitory capacity but retains its cell signaling function at the time of CFA administration. To limit the complexity of our experimental design, and to determine the optimal dose for the administration of each TIMP-1 construct, we conducted a pilot experiment using a small cohort of T1KO mice given 1, 10, or 100 ng/μL of TIMP-1 at the time of inflammation. We found that 10 ng/μL was effective at reducing inflammatory hypersensitivity (data not shown). We then administered a separate cohort of T1KO mice 10 ng/μL of TIMP-1(FL), TIMP-1(N), or TIMP-1(C) at the time of CFA administration. Mechanical hypersensitivity was assessed 24 hours later. While inflamed T1KO mice exhibited a significant reduction in mechanical thresholds, T1KO mice treated with . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The above data show that inflammation in one somatic region could lead to mechanical hypersensitivity in tissue distal to the site of inflammation, reminiscent of "mirror image pain" (Treede et al., 1992;Treede et al., 2015). To test this possibility, we inflamed one hindpaw and measured mechanical sensitivity on the opposite hindpaw for 7 days following CFA administration. We also examined whether treatment with rmTIMP-1 at the site and time of inflammation affected sensitivity. We found that inflamed T1KO mice exhibited contralateral mechanical hypersensitivity over the course of 7 days following CFA-injection relative to WT mice ( Figure 5D). Interestingly, this contralateral hypersensitivity was prevented by treatment with rmTIMP-1 in T1KO mice, F (4, 43) = 5.52, p < 0.05 ( Figure 5D).

The lack of TIMP-1 does not alter the expression of local inflammatory molecules
TIMP-1 is primarily known as a broad-spectrum MMP inhibitor, and because MMPs are known to contribute to hypersensitivity, we hypothesized that the absence of TIMP-1 may cause hypersensitivity due to elevated activity and expression of MMP-9 (Kawasaki et al., 2008;Brew and Nagase, 2010). Examination of hairy skin collected 1 day following CFA from WT and T1KO mice demonstrated that there was an inflammation-induced increase in both MMP-9 expression and activity, all Fs > 7.61, p <.05 but that these effects were not genotype-specific, all Fs < 2.05, p >.05 ( Figure 6A, B). The TIMP/MMP axis also regulates the proteolytic maturation of inflammatory molecules which can cause hypersensitivity (Pagenstecher et al., 1998;Ellis and Bennett, 2013). We next assessed whether the absence of TIMP-1 during inflammation caused . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; elevated expression in skin. Using ELISAs, we assessed the expression of cutaneous IL-1β, IL-6, TNF-, and IL-10 at 1 day following CFA-injection. Analysis revealed an inflammation-induced increase in IL-1β and IL-6 expression ( Figure 6C, D) all Fs > 12.94, p <.05. In comparison, analysis of TNF-α and IL-10 did not reveal any significant differences following inflammation or between genotypes, all Fs < 4.64, p > .05 ( Figure 6E, F). These data suggest that TIMP-1 does not affect the normal progression of hypersensitivity through the known MMP regulatory or cytokinedependent mechanisms.

Administration of recombinant TIMP-1 attenuates ongoing pain in WT mice.
Previous experiments demonstrate that the administration of rmTIMP-1 attenuates evoked mechanical and thermal hypersensitivity in T1KO mice. Here, we examined whether the administration of rmTIMP-1 also attenuated ongoing pain in WT mice using CPP (as described in Methods; (King et al., 2009); Figure 7A). Analysis of pre-compared to post-conditioning time spent in the chamber indicate an effect of drug (F (1, 30) = 7.269, p<0.05), with post-hoc analysis confirming an increase in post-conditioning time spent in the clonidine paired chamber compared to pre-conditioning time in vehicle treated mice (p<0.01) but not in rmTIMP-1 treated mice (p>0.05) ( Figure 7B). These observations indicate that mice administered CFA and TIMP-1 did not demonstrate clonidine-induced CPP ( Figure 7B, C). Because clonidine is only effective when pain is present (King et al., 2009), these results suggest that treatment with rmTIMP-1 attenuated ongoing inflammatory pain.

DISCUSSION
Although TIMP-1 is known to regulate ECM dynamics, inflammation, and wound healing, previous reports suggest that TIMP-1 may also play a role in the emergence of pain, particularly following nerve damage (Kawasaki et al., 2008;Ji et al., 2009;Huang et al., 2011). However, the . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; protective function of TIMP-1 in the development of inflammatory hypersensitivity required further investigation. Previous RNA sequencing results shows that TIMP-1 mRNA is present throughout the CNS and peripheral nervous systems (PNS) (Fagerberg et al., 2010). In the adult mouse model, TIMP-1 expression is low during basal conditions but can be induced following insult to nervous system (Crocker et al., 2006;Parkitna et al., 2006;McKelvey et al., 2015). Therefore, we first determined that cutaneous inflammation alters TIMP-1 expression proximal to the location of inflammation as well as distal locations associated with increased hypersensitivity.
Furthermore, cutaneous inflammation induced TIMP-1 protein levels within 24 hr, peaked at 3 days, and waned by day 7. These data compliment other studies that suggest that TIMP-1 is inducible, immediate factor involved in the rapid response to tissue damage (Crocker et al., 2006;Kawasaki et al., 2008).
TIMP-1 is expressed by a variety of cell types including astrocytes and oligodendrocytes, endothelial cells, fibroblasts, mast cells, and keratinocytes (Yokose et al., 2012;Claycomb et al., 2013). Using an in vitro model of inflammation, we determined that cutaneous TIMP-1 expression was induced in K14-positive keratinocytes within 24 hr. These results agree with prior work showing keratinocytes upregulate TIMP-1 following UVB photodamage (Yokose et al., 2012). Furthermore, given that keratinocytes augment nociceptive signaling through the release of neuroactive molecules, TIMP-1 may be released from keratocytes to attenuate nociceptive responsiveness immediately following tissue damage (Fagerberg et al., 2010;Yokose et al., 2012;Baumbauer et al., 2015;Edqvist et al., 2015;Moehring et al., 2018). However, the effect of TIMP-1 may depend on the type of insult experienced. A recently published gene array comparing gene expression profiles in rats following chronic constriction injury (CCI) of the sciatic nerve or s.c.
administration of CFA, and found that TIMP-1 mRNA was increased within 3 days following . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; nerve damage in the spinal cord and DRG but not following cutaneous inflammation (Parkitna et al., 2006). While we also reported no changes in overall TIMP-1 protein expression in the spinal cord or DRG following inflammation, we did observe increased TIMP-1 expression in GFAP expressing cells in both structures, suggesting that inflammation increases TIMP-1 expression in astrocytes and satellite glial cells (SGCs), respectively. Interestingly, previous work has shown that peripheral nerve injury increases TIMP-1 in SGCs (Parkitna et al., 2006;Huang et al., 2011), and that experimental autoimmune encephalomyelitis increases TIMP-1 in astrocytes (Pagenstecher et al., 1998;Claycomb et al., 2013).
To test whether local TIMP-1 expression is important for regulating the onset of inflammatory hypersensitivity, we used global TIMP-1 knockout mouse and assessed thermal and mechanical responding in response to inflammation. We found that inflamed T1KO mice developed exaggerated nocifensive responses to mechanical and thermal stimulation at the site of inflammation that was not observed in WT mice. This is important because it demonstrates that in the absence of TIMP-1, subtle perturbations can result in profound functional changes in sensory processing. Indeed, we observed such an effect when T1KO mice were administered s.c. injections of physiological saline. Our results further suggest that TIMP-1 may have an active role in regulating responding to injury and inflammation, specifically, and sensory stimulation, in general.
To determine whether T1KO mice develop a pathological pain-like phenotype following inflammation, we assessed mechanical thresholds in tissue adjacent (innervated by different sets of afferents) and contralateral to inflammation. We found that mice lacking TIMP-1 exhibited mechanical sensitivity in both adjacent and distal skin This gain of sensitivity in uninjured tissues is often specific to mechanical stimulation and is associated with the development of pathological pain (Treede et al., 1992;Treede et al., 2015), suggesting that TIMP-1 may have a yet to be defined . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; role in the regulation of normal somatosensation. In further support of this conclusion, hypersensitivity in both the tissue adjacent and distal to inflammation was attenuated when rmTIMP-1 was administered to T1KO mice at the site of inflammation. These data suggest that inflammation induces TIMP-1 expression in a coordinated fashion that influences the normal progression of inflammatory sensitivity. Consequently, as is the case for T1KO mice, dysfunction in TIMP-1 signaling, may lead to an aberrant process that encourages the transition from acute to chronic pain. Because TIMP-1 is primarily characterized as an inhibitor of MMP activity, and MMPs have a known role in driving pain following nerve injury (Kawasaki et al., 2008;Remacle et al., 2018). We hypothesized that disrupting the TIMP/MMP balance would cause increased hypersensitivity through MMP and proinflammatory cytokine signaling. However, we failed to observe any genotype-specific differences in the expression of MMP-9 or cytokines in the skin following inflammation. This suggested to us that TIMP-1 may be inhibiting development of hypersensitivity by a receptor-mediated mechanism. Supporting this, administration of the truncated C-terminal of TIMP-1 that retains cell signaling function ameliorates inflammatory hypersensitivity. However, given that the truncated MMP-inhibitory N-terminal domain also attenuated hypersensitivity, we cannot rule out the role of MMP-inhibition in the regulation of pain, nor do we posit that the effects of TIMP-1 are solely due to a cell signaling event. What we do propose is that TIMP-1 may act through both pathways to effectively attenuate pain progression. This latter point may help to explain, at least in part, why small molecule inhibitors of MMP activity have limited efficacy. By targeting the entire TIMP-1 pathway, we may be able to effectively manage pain progression. Moreover, the site of TIMP-1 administration may also determine efficacy. TIMPs and MMPs are ubiquitously expressed throughout all tissues, and may . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. have divergent responses to changing physiological environments. If administration of TIMP-1 occurs directly at the site of injury (or sensitivity), efficacy may be increased by precisely targeting the tissue affected by pain or injury, while simultaneously avoiding disrupting the TIMP/MMP balance in unaffected tissue.
Finally, demonstrating that TIMP-1 alleviates ongoing pain in WT mice, in conjunction with our previously described data, identifies TIMP-1 as a potential clinical target. One lingering concern was that administration of rmTIMP-1 may result in levels cutaneous TIMP-1 that were beyond normal physiological levels. However, it should be noted that this is a common occurrence when drugs are used to treat clinical conditions, including pain, and targeting TIMP-1 may a novel molecule for the development of therapeutics for pain management.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Figure 1.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

K14
TIMP-1 Merge IS F . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Figure 3
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Figure 5
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Figure 6.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/540724 doi: bioRxiv preprint first posted online Feb. 4, 2019; Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CF A treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CF A treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CF A treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. * TIMP-1 blocks CFA-induced ongoing pain in C57bl/6 male mice.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CFA treated mice did not significantly increase time spent in the clonidine paired chamber. A t-test between the vehicle and rTIMP-1 treated groups is not significant. Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CFA treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. A.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CF A treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CFA treated mice did not significantly increase time spent in the clonidine paired chamber. A t-test between the vehicle and rTIMP-1 treated groups is not significant. Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CF A treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. A.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CF A treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CF A treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. * TIMP-1 blocks CFA-induced ongoing pain in C57bl/6 male mice.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CFA treated mice did not significantly increase time spent in the clonidine paired chamber. A t-test between the vehicle and rTIMP-1 treated groups is not significant. Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CFA treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. A.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CF A treated mice did not significantly increase time spent in the clonidine paired chamber . A t-test between the vehicle and rTIMP-1 treated groups is not significant. * TIMP-1 blocks CFA-induced ongoing pain in C57bl/6 male mice.
Top graph: Comparison of pre-conditioning and post-conditioning time spent in the clonidine paired chamber show a significant increase in the postconditioning time in the Vehicle/CFA treated mice, but not the rTIMP-1/CFA treated mice. *p<0.05 vs pre-conditioning time.
Bottom graph. Difference scores demonstrate that vehicle/CFA treated mice show significantly increased time spent in the clonidine paired chamber (*p<0.05 vs null hypothesis of 0). In contrast, the rTIMP-1/CFA treated mice did not significantly increase time spent in the clonidine paired chamber. A t-test between the vehicle and rTIMP-1 treated groups is not significant. B.

Conditioned Place Preference Chamber
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Figure 1. Saline injection causes hypersensitivity in mice lacking TIMP-1.
Paw withdrawal thresholds (PWT) presented as change from baseline in grams following s.c.
injection of 0.9% saline. T1KO mice administered saline exhibited a significant reduction in mechanical thresholds 1 day following injection compared to WT controls (n=6/condition). * indicate significant differences compared to naïve controls, p<.05, and error bars depict SEM.  . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
(D) T1KO mice develop significantly reduced PWTs 1 day following CFA administration compared to WT controls (n=8-10/condition).  Hindpaw administration of CFA produces mechanical hypersensitivity on the paw contralateral to inflammation in T1KO relative to WT mice. Treatment with rmTIMP-1 attenuated contralateral hypersensitivity in T1KO mice. PWT are presented as change from baseline. n=8/condition, * represent significant differences relative to naïve controls, p<.05, and error bars depict SEM.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Figure 6. Inflammation does not alter pro-inflammatory molecules in a genotype-specific manner.
A) Cutaneous inflammation significantly increases MMP-9 protein expression in WT and T1KO skin 1 day following CFA administration (n=7/condition). B) Cutaneous inflammation increases MMP-9 activity in WT and T1KO hairy skin 1 day following CFA administration. C) Cutaneous inflammation significantly increases IL-1β protein expression in WT and T1KO hairy skin 1 day following inflammation. D) Cutaneous inflammation significantly increases IL-6 protein expression in WT and T1KO hairy skin 1 day following CFA administration. E) Cutaneous inflammation does not affect expression of TNF-α following CFA administration. F) Cutaneous inflammation does not affect expression of IL-10 protein in WT and T1KO skin following CFA administration. n=4/condition, * represent significant differences relative to naïve controls, p<.05, and error bars depict SEM. . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.