Selective inhibition of soluble TNF using XPro1595 relieves pain and attenuates cerulein-induced pancreatic pathology in mice, with a possible to central pain processing

Treatment of acute pancreatitis remains a challenge, with therapy focused on supportive care and treatment 13 of the inciting etiology. Individuals with pancreatitis may experience severe upper abdominal pain, 14 although pain mechanisms in patients with pancreatitis are incompletely understood and likely 15 multifactorial, with possible pain processing occurring in the central nervous system, a process known to 16 be associated with the upregulation of inflammatory cytokines. Inflammation plays a prominent role in the 17 induction of acute pancreatitis, with the inflammatory cytokine tumor necrosis factor (TNF) both 18 exacerbating cell death, pro-inflammatory signaling and edema, as well as promoting reparative and 19 restorative mechanisms. This duality of function can be explained by different forms of the TNF ligand 20 preferentially activating different receptor subtypes, whereby the uncleaved transmembrane form of the 21 ligand (tmTNF) preferentially activates TNFR2 promoting restorative functions, but once cleaved the 22 soluble form of TNF (solTNF) preferentially activates TNFR1 promoting detrimental pathology. For this 23 reason, the traditional TNF inhibitors that inhibit both TNFR1 and TNFR2 have shown modest success in 24 patients, but come with numerous side-effects, including immunological dysfunction and heart failure. 25 Therefore, we sought to assess the effect of a novel selective inhibitor of solTNF (XPro1595) on pancreatic 26 pathology and associated neuropathic pain in a mouse model of acute pancreatitis, and observe its effect on 27 an area of the brain (the hippocampus) known to play a role in neuropathic pain processing. XPro1595 28 administration began after the initial peak in serum amylase to maximize clinical relevance. Administration 29 of XPro1595 prevented pancreatic immune cell infiltration, that subsequently prevented tissue disruption 30 and acinar cell death. These improvements in pathology were associated with a significant reduction in 31 mechanical hypersensitivity (neuropathic pain). XPro1595 treatment also prevented an increase in hippocampal astrocyte reactivity, associated with the of in this


INTRODUCTION 42
Pancreatitis is a leading cause for gastrointestinal disease-related hospital admissions, and is primarily an 43 inflammatory condition of the pancreas that can either be acute or chronic, lasting many years. Although 44 individuals with pancreatitis can progress to have quite severe complications, the majority of patients 45 endure only mild bouts of the disease with symptoms including upper abdominal epigastric pain, nausea 46 and vomiting (1,2). Acute pancreatitis can be broadly classified into interstitial edematous pancreatitis 47 with peripancreatic fat stranding and fluid accumulation, or necrotizing pancreatitis that as the name 48 suggests promotes necrosis of pancreatic parenchyma and/or peripancreatic tissue (2). These differences 49 in disease pathologies has driven the creation of a plethora of existing pre-clinical animal models, with no 50 single animal model displaying all aspects of acute or chronic pancreatitis. None-the-less these models 51 frequently promote a systemic inflammatory response by either non-invasive (administration of toxins, 52 transgenic mice or diet based) or invasive means (vessel/duct blockage or perfusion) (3). 53 54 Studies in patients and animals have identified the disease course and severity of pancreatitis is mostly 55 governed by inflammatory cells that drive local and systemic immune responses, of which a major 56 contributor is the upregulation of the inflammatory cytokine tumor necrosis factor (TNF). TNF production 57 promotes the induction of inflammatory genes, acinar cell death and recruitment of immune cells (4-8), 58 which prompted investigation of traditional TNF inhibitory therapies to prevent or reduce these pathologies 59 and associated symptoms. Early studies in rodents modulating TNF ligand and receptor activity using TNF 60 receptor fusion proteins or anti-TNF antibodies (e.g., Etanercept, Infliximab and Adalimumab) showed 61 promise with reduced pancreatic pathologies such as edema, inflammation, necrosis and vacuolization (9-62 12), and similar positive outcomes were seen in patients when TNF inhibitors were administered to treat 63 other cooccurring conditions (13,14). Unfortunately, the abundance of side-effects in these traditional TNF 64 inhibitors (including immunological dysfunction and even the induction of pancreatitis itself (15)(16)(17)(18)(19)), 65 combined with their apparent inability to reduce mortality in early studies performed on sepsis patients (20, 66 21) dampened enthusiasm for their further use in patients with pancreatitis. More than 2 decades on 67 however, additional meta-analysis' of patients with sepsis have revealed an overall improvement in survival 68 rates (22,23), when studies are sufficiently powered, and this may have encouraged the examination of 69 these traditional TNF inhibitors in patients with pancreatitis (24,25). None-the-less, these traditional 70 inhibitors still promote severe side-effects, and their use should be cautioned in patients. 71

72
The discrepancy between these paradoxical outcomes in rodents and humans warrants further investigation, 73 but is likely due to the differences in TNF receptor subtype functions, that have complicated the TNF field 74 until recently. TNF is first produced as a transmembrane protein (tmTNF) that preferentially activates TNF 75 receptor 2 (TNFR2: CD120b or p75/p80) (26), but once TNF is cleaved from the cell membrane to exist 76 in a soluble form (solTNF) it preferentially activates TNF receptor 1 (TNFR1: CD120a or p55/p60) (26). 77 Although the two receptors can trigger some common signaling pathways (27), TNFR2 activation generally 78 promotes beneficial outcomes such as cell survival, induction of neurogenesis, and promotion of CNS 79 autoimmunity (28-30), while TNFR1 activity generally promotes detrimental outcomes such as cell death, 80 aberrant neuronal plasticity, and exacerbation of the existing inflammatory response (28,31,32). 81 Therefore, being able to selectively block the activity of solTNF/TNFR1, while sparing the activity of 82 tmTNF/TNFR2 activity, would likely prove beneficial to patients, although traditional TNF inhibitors are 83 unable to distinguish between the different TNF ligand or receptor subtypes. For this reason, a novel 84 'second generation' TNF inhibitor was developed that selectively inhibits only the soluble form of TNF 85 (solTNF: XPro1595), with no known side-effects, and has been proven to be safe and well tolerated in 86 patients (33), with an ability to reduce neuroinflammation in the brains of Alzheimer Disease patients (34). 87 Indeed, within the pancreas TNFR1 activity is known to exacerbate cell death, inflammation and edema 88 (10), while TNFR2 promotes pancreatic regeneration (35,36). Therefore, assessing the outcomes 89 selectively inhibiting solTNF using XPro1595 in rodents with pancreatitis is highly clinically relevant. 90

91
Replicating the symptoms of acute pancreatitis in rodents, especially induction of visceral pain is 92 challenging, as pain is a subjective experience and therefore difficult to assess in animals. None-the-less, 93 animal models are useful to identify the possible underlying mechanisms of pain necessary to create 94 therapeutic interventions. Many pre-clinical acute and chronic models of pancreatitis promote increased 95 sensitivity of the pancreas to electrical stimulation, as well as promoting the induction of neuropathic pain 96 (detection of a stimulus not normally detectable) in regions both local (abdomen) and distant (hindpaw) to 97 the site of inflammatory origin (37-44), suggesting activation of both normal pain pathways and central 98 sensitization. Alterations to peripheral A and C fiber activity promote the induction of pain (45, 46), while 99 peripheral targets such as ion channels regulating neural sensitization (47,48), and release of chemical 100 mediators of inflammation such as pro-inflammatory cytokines (e.g. TNF/TNFR1) modulating these effects 101 (47,(49)(50)(51). Central plasticity also overlays these pain pathways, involving central brain structures such as 102 the hippocampus whereby reductions in synaptic protein expression can reduce LTP (52-54), leading to 103 loss of input to cortical pain sensing regions (e.g somatosensory and prefrontal cortices) (55, 56). 104 Importantly, established rodent models of neuropathic pain identify hippocampal TNF/TNFR1 activity 105 regulates the peripheral hypersensitivity (52,57,58). Therefore, the systemic inflammatory response in 106 individuals with pancreatitis that upregulates TNF/TNFR1 activity may play a substantial role in both the 107 induction of pancreatic pathology and neuropathic pain. 108 109 For these reasons, we sought to investigate the effects of using the novel non-opioid 'second-generation' 110 biologic XPro1595 to selectively bind and neutralize solTNF in a mouse model of pancreatitis. XPro1595 111 has a 17-hour half-life, can cross the blood-brain-barrier, has been successfully used in numerous pre-112 clinical inflammatory disease models (59-63), and is safe and well-tolerated in both cancer and Alzheimer 113 Disease patients (33,34,64,65). The use of XPro1595 therefore has a high potential for translation to the 114 clinic with the promise to improve outcomes in inflammatory pancreatic conditions. 115

METHODS 117
Animals 118 Male C57Bl/6J mice aged 2 to 4 months were used for the current study. Animals were housed in a 12-119 hour light/dark cycle with food and water ad libitum. Procedures related to animal use were approved by 120 the Virginia Commonwealth University Institutional Animal Care and Use Committee (in accordance with 121 NIH care and use of laboratory animals). Mice were anaesthetized with ketamine (75 mg/kg) and xylazine (14 mg/kg) by intraperitoneal injection, 139 prior to undergoing transcardial perfusion using approximately 15 ml PBS, followed by approximately 4 % 140 paraformaldehyde (Sigma). Tissue was dissected, stored in 4 % paraformaldehyde for 2 hours, 141 cryoprotected in 20 % sucrose in PBS for 48 hours, and then quickly frozen in OCT over isopentane on dry 142 ice, and stored at -80 o C. Serial frozen coronal sections were cut 40 µm thick through the pancreas and 143 hippocampus. Some slides underwent hematoxylin and eosin (H&E) staining to assess pancreas 144 inflammatory infiltrates and tissue integrity. Other slides underwent immunohistochemical analysis to 145 further assess inflammatory state. Slides containing sections of either pancreas or brain were permeabilized 146 with 0.2 % triton X-100 (Sigma) in 2 % fish gel in PBS solution and immunohistochemically labelled with Gold Antifade mounting medium containing DAPI (ThermoFisher). Sections were photographed (pancreas 151 at 10x and 40x; brain at 40x) with equal exposure on an Olympus CK-2 inverted microscope, connected to 152 a 3MP Amscope digital camera (MU300-CK) with Amscope Software version 3.2, prior to analysis using 153 NIH ImageJ version 1.52a. 154 155

Histological Analysis 156
To assess pancreatic immune cell infiltration 2 and 7 days post-induction of acute pancreatitis, photographs 157 of H&E stained pancreatic sections were semi-quantitated on an arbitrary scale within each field of view: 158 0 = no inflammatory cells present; 1 = some inflammatory cells present; 2 = many inflammatory cells 159 presesnt. At 7 days, within each H&E photograph we also semi-quantitated pancreatic tissue integrity 160 (spaces between acinar cell clusters: 0 = normal pathology; 1 = some spaces evident; 2 = large amount of 161 space evident appearing similar to a 'cracking' effect), acinar cell atrophy (0 = normal pathology; 1 = acinar 162 atrophy present but not immediately apparent; 2 = acinar atrophy prevalent throughout the tissue), and 163 integrity of intralobular duct (degree of invasion of spaces normally occupied by vessels within the large 164 pancreatic lobules: 0 = normal pathology, 1 = some invasion present but not immediately apparent, 2 = 165 large amount of invasion present). Values for each photograph were averaged per section, per animal, and 166 then per group. To assess the extent of circulating macrophage infiltration within the pancreas, the 167 immunohistochemistry images were quantitated for the level of IBA-1. Images were imported into ImageJ, 168 converted to gray scale and thresholded, and the area fraction of pixels positive for IBA-1 was quantitated. 169 For each photograph, IBA-1 expression level (arbitrary units) were measured in fractional areas to give an 170 average IBA-1 intensity. Values for each photograph were averaged per section, per animal, and then per 171 group. The same protocol was applied to brain sections containing hippocampus at 7 days post-induction 172 to quantitate astrocyte reactivity using the GFAP antibody. 173 174

Mechanical von Frey Hindpaw Neuropathic Pain 175
In a dimly lit room, a 10" x 19" extension window screen (Thermwell) was fully extended and placed atop 176 2 polystyrene boxes, with a desk lamp placed behind and just under the height of the screen, angled towards 177 the investigator. Four mice at a time were placed on top of the screen, with a 600 ml glass beaker (Pyrex) 178 placed over the top of each mouse to prevent escape. A disposable underpad was draped over the beakers 179 to minimize any light and/or movement stimulation. After 15 minutes acclimatization under the beaker, 180 hindpaw hypersensitivity was assessed by holding the von Frey filament (Bioseb) handle under the screen, 181 and slowly raising the end of the filament up through the screen to press against the under-side of the 182 mouse's hindpaw walking pad until a slight bend was observed in the fiber.
Continued 183 advancement/bending of the filament does not necessarily produce more force of application. The 184 investigator tested the lightest filament first, and sequentially tested up through the filament sizes until a 185 positive result was established. A positive result was the mouse noticing 3 out of 5 consecutive tests for 186 each filament, defined as the mouse withdrawing its foot, licking or shaking its foot, or rapidly moving its 187 body away from the stimulus. Once a positive result was established for each mouse, the testing was 188 concluded for that mouse for that day. The testing occurred as rapidly as possible to reduce restraint 189 distress, although it was noticed that mice would often fall asleep during testing, which required gentle 190 tapping from underneath the screen to wake up the animal. 191 192

Statistical Analysis 193
All data were assessed for homogeneity of variance, after which statistical analysis was performed. 194 Histological differences were assessed using the Student's t-test, and behavioral differences (intra-and 195 inter-group analysis) were assessed using two-way repeated measures analysis of variance with Student-   We next wanted to identify whether an early alteration in pancreatic inflammatory infiltrates had other 235 effects on pancreas pathology at a later timepoint. For this, 7 days after the induction of acute pancreatitis 236 we semi-quantitated pancreatic tissue integrity ('cracking' effect within small acinar clusters). We 237 identified that administering cerulein reduces tissue integrity within small cell clusters ( Figure 3A&C), 238 while XPro1595 treatment prevented this effect (significantly less than vehicle-treated pancreatitis mice, 239 and not different to non-pancreatitis mice) ( Figure 3B&C). We also semi-quantitated acinar cell atrophy. 240 We observed that vehicle-treated cerulein-induced acute pancreatitis promotes acinar atrophy by day 7 241 ( Figure 3D&F), compared to vehicle-treated non-pancreatitis mice ( Figure 3F). In contrast, the XPro1595-242 treated pancreatitis group did not display acinar atrophy (significantly less than vehicle-treated pancreatitis 243 mice, and not different to non-pancreatitis mice) ( Figure 3E&F). We further assessed the integrity of the 244 intralobular duct between large pancreatic lobules. Cerulein-induced pancreatitis promoted significant 245 disruption of pathology between the large pancreatic lobules, whereby acinar clusters within lobules often 246 invaded these spaces ( Figure 3G&I). This effect was not observed in the XPro1595-treated pancreatitis 247 group (not significantly different to non-pancreatitis mouse group) ( Figure 3H&I). 248 249

Selective inhibition of solTNF using XPro1595 attenuates cerulein-induced neuropathic pain 250
Acute pancreatitis often promotes severe and constant pain in the upper abdomen, which can extend around 251 to the back, and last for a few days. In accordance with this, many pre-clinical rodent models display both 252 pancreatic and referred neuropathic pain in both the abdomen and hindpaws (37-44). To assess the role of 253 solTNF in the induction of pancreatitis-induced neuropathic pain, we measured the level of hindpaw 254 mechanical hypersensitivity at 3, 5 and 7 days post-induction, in mice treated with and without XPro1595. 255 Vehicle-treated non-pancreatitis mice (i.e. vehicle S.C. and I.P.) were assessed to establish baseline 256 sensitivity response, set at 100%. We observed the hindpaw sensitivity of XPro1595-treated non-257 pancreatitis mice (i.e. XPro1595 S.C., and vehicle I.P.) was not different from vehicle-treated non-258 pancreatitis mice over the testing period (Figure 4), suggesting XPro1595 does not regulate hindpaw 259 hypersensitivity under non-pathogenic conditions. Next, we assessed the hindpaw sensitivity of mice with 260 ceruline-induced acute pancreatitis. We identified that vehicle-treated acute pancreatitis mice (i.e. Vehicle 261 S.C., and cerulein I.P.) displayed persistent hindpaw hypersensitivity, beginning from the first day of testing 262 (day 3) until the last (day 7). In contrast, the hindpaw hypersensitivity of XPro1595-treated pancreatitis 263 mice (XPro1595 S.C., and cerulein I.P.), was not significantly different to baseline control mice at any time 264 point tested, and was significantly better than vehicle-treated pancreatitis mice on days 5 and 7. We semi-quantitated hippocampal GFAP expression, as a marker of astroglial reactivity 7 days after 275 cerulean administration. We observed that cerulein-induced acute pancreatitis increased the tendency to 276 for astrocyte reactivity (GFAP expression) in the hippocampal CA1 region 7 days post-induction, compared 277 to non-pancreatitis mice ( Figure 5A,B&E), which was not apparent in the XPro1595-treated pancreatitis 278 group ( Figure 5C,D&E). 279

DISCUSSION 281
The onset of acute pancreatitis causes abdominal tenderness and pain, as well as nausea and vomiting that 282 coincides with a spike in the blood pancreatic enzymes amylase and lipase (72), and an inflammatory 283 response (69). Many causes of acute pancreatitis have been identified including pancreatic duct obstruction, 284 alcoholism and a genetic mutation, with management including removal of any obstruction/s, nutritional 285 regulation (including pancreatic enzyme supplementation and hydration), and pain management (73). 286 Pharmacologic interventions have targeted the inhibition of proteolytic enzymes using broad spectrum anti-287 protease inhibitors that showed variable outcomes in animals if delivered before disease onset (74,75). 288 Unfortunately, these inhibitors failed to show any effect in patients, possibly due to administration at a time 289 point after peak enzymatic activity, which may be unavoidable given the short peak in enzyme activity in 290 patients (76,77). Another pharmacological direction is to regulate the immune response, which also 291 displays different phases of pro-and anti-inflammation, but which may represent a more clinically relevant 292 timepoint. Animal studies inhibiting pro-inflammatory mediators including IL-6, and ICAM, or bolstering 293 anti-inflammatory mediators such as IL-10 have also shown variable successes (78-82), but enthusiasm for 294 their use in patients has diminished due to limited benefits observed in regulating the inflammatory response 295 (83-86). Notably however, early studies in rodents using TNF inhibitors showed promise with improved 296 pancreatic pathology (9-12), with positive outcomes also seen in patients (13, 14), although their abundance 297 of side-effects combined with their inability to reduce mortality in early studies on sepsis patients (20, 21) 298 dampened enthusiasm for their further use in patients with pancreatitis. More than 2 decades later however, 299 additional meta-analysis' reveal an overall improvement in survival rates in patients with sepsis when 300 studies are sufficiently powered (22,23). That combined with the development of a novel selective inhibitor 301 of solTNF (XPro1595) with no known side-effects, that has proven to be safe and well tolerated (33), with 302 an ability to reduce neuroinflammation in patients (34), warrants additional investigation. Indeed, our pre-303 clinical studies identify that selectively inhibiting solTNF by administrating XPro1595 in a clinically 304 relevant window reduces the course of the disease by limiting pancreatic inflammatory infiltrates. This 305 early reduction in inflammatory severity prevents subsequent pancreatic pathological damage, despite this 306 study being performed in a rodent model of pancreatitis (cerulein induction) that is known to resolve the 307 inflammatory response within 7 days. 308

309
One of our important findings in these studies is the improvement of neuropathic pain "referred pain" over 310 the course of the study in the XPro1595-treated mice, even after the resolution of the inflammatory response 311 in all groups (Day 7). This is important because it suggests that the early inflammatory response is a 312 significant contributor to the development of pancreatitis-induced pain, which can be alleviated by 313 selectively inhibiting solTNF. Several lines of evidence suggest TNF plays a key role in regulating pain. 314 First, TNF can regulate signaling along the traditional pain pathways at the level of the nociceptors (87, 315 88), dorsal root ganglion (50, 89), thalamus (90), and somatosensory cortex (90). In support of this, 316 administering infliximab to block TNF in patients with rheumatoid arthritis reduced pain-induced CNS 317 activity (90). Second, strong evidence supports of role of TNF throughout the limbic system to regulate 318 pain (52-54, 90, 91), possibly by incorporating emotional memories of pain. One region of the limbic 319 system that has gained a lot of attention is the hippocampus whereby hippocampal TNF/TNFR1 activity 320 regulates the severity of neuropathic pain (52-54, 91). Indeed, in our mouse model we observed a strong 321 tendency for upregulation of GFAP in the hippocampus 7 days after the induction of acute pancreatitis 322 suggesting that in our model hippocampal TNF/TNFR1 activity could be contributing to the induction of 323 neuropathic pain since reactive astrocytes are known synthesizes of excess TNF (92, 93). However, the 324 subcutaneous administration of XPro1595, while clinically relevant, prevents determination of the 325 molecular mechanisms involved. 326 327 328

CONCLUSION 329
Excess levels of the inflammatory cytokine TNF plays a prominent role in many inflammatory disease 330 pathologies, including the induction of pancreatitis. Attempts to use TNF receptor fusion proteins or 331 monoclonal antibodies to regulate this cytokines function have shown some successes clinically, but these 332 have been fraught with complications due to their numerous adverse side-effects, including drug-induced 333 acute pancreatitis. Our data provide support for the clinical use of a novel second generation TNF inhibitor 334 XPro1595 that selectively inhibits only the detrimental soluble form of TNF to prevent the disease sequelae, 335 while sparing the beneficial transmembrane form of TNF to allow reparative cellular mechanisms to remain. Committee. 347

Consent for publication 348
Not applicable. 349

Availability of data and materials 350
The datasets generated and/or analyzed during the current study are available from the corresponding 351 author on reasonable request. 352

Competing interests 353
The authors declare that they have no competing interests. 354

Funding 355
We gratefully acknowledge the grant support from the Virginia Commonwealth Neurotrauma Initiative 356 funding to KJD (FP00001476), and Virginia Commonwealth University Department of Surgery pilot 357 funding to KJD. Funding bodies did not contribute to the design of study or collection, analysis, 358 interpretation of data or writing of the manuscript. 359

Author's contributions 360
RR and MD performed the experiments and tabulated the data. KJD designed, planned, funded, analyzed 361 and interpreted the data. All authors read and approved the final manuscript. 362