Localisation analysis of nerves in the mouse pancreas reveals the sites of highest nerve density and nociceptive innervation

Neuropathy and neuro‐inflammation drive the severe pain and disease progression in human chronic pancreatitis and pancreatic cancer. Mice, especially genetically induced‐mouse models, have been increasingly utilized in mechanistic research on pancreatic neuropathy, but the normal “peripheral neurobiology” of the mouse pancreas has not yet been critically compared to human pancreas.


| INTRODUC TI ON
Pancreaticcancer(PCa)andchronicpancreatitis(CP)arecharacterizedbyprominentalterationsofintrapancreaticnerveslikenerve hypertrophy, neuro-inflammation, and neural invasion, which are now to give rise to severe neuropathic abdominal pain. [1][2][3] Mouse models, particularly genetically induced mouse models, have been recently and increasingly introduced into the study of these neuropathic alterations. [4][5][6] However, these studies do not yet seem to take into account sufficiently the anatomic differences in the innervation of the mouse pancreas vs human pancreas. Indeed, humanpancreasisacompactsolidorgan,whichislocalizedinthe retroperitoneum. 7 Conversely, mouse pancreas is a "mesenteric" type pancreas, which is not as compact and rather scattered in the adjacentsmallintestinalmesenteryandisthuslocalizedintraperitoneally. 7 For correct comparison and reporting of nerve alterations in mouse models and for purposes of simulating and studying human disease, it is imperative to know the localization and distribution of nerves in the mouse pancreas. Importantly, to date, despite some studies that compared the ganglia and the fine nerve fiber distribution, 8,9 there has been no study that investigated the differences in the nerve trunk anatomy between the mouse and the human pancreas.
In the present study, we performed a systematic morphological analysis of the nerves in correlation with their location in the pancreas.Here,weshowthatmousepancreaticnervetrunksaresolely located around peripancreatic lymphoid structures and around vascular complexes. Furthermore, we reveal a greater density of nerves intheheadandanteriorbody(corpus)ofthemousepancreas,when compared to the remainder of the organ. Among pain-transmitting nerve fiber subtypes, we show that the distribution of nerves which containpain-relatedneuropeptideslikesubstanceP(SP),CGRP,vasoactive intestinal peptide (VIP), or NOS, does not vary between different regions of the mouse pancreas and that they constitute around 10% percent of the total innervation. Overall, our study provides a reference for studying the mouse pancreatic innervation within the frame of morphological, anatomic, mechanistic, or neurochemical code studies.

| Systematic tissue harvesting and analysis
Methods of tissue harvesting can be deciding for the subsequent analysis of structures in the same tissue. Therefore, in this study, weappliedastandardizedharvestingmethodforcompleteembedding of the mouse pancreas with the adjacent organs. In 8-weekold C57BL/6J mice, we performed a median laparotomy and first grabbed the stomach and transected it distal to the pylorus. We then held the spleen and freed it from its retroperitoneal attachments. We then held the freed duodenum and the spleen, so that the pancreatic corpus was the only remaining part of the pancreas attached to mesentery of the small intestine. Here, the pancreas could be bluntly separated away from the mesentery under a stereomicroscope to definitely avoid collection of any mesentery structures. For the dorsal attachments, we also verified under a stereomicroscope the exclusion of any retropancreatic structures like lymph nodes.
The resected mouse pancreas-duodenum-spleen bloc was immediately fixed in 4% paraformaldehyde followed by paraffin embedding, as described previously. 1 The paraffin embedding of the pancreas was performed by strictly adhering to its normal anatomic location ( Figure 1A-C). Thus, we were able to preserve theanatomicorientationandsubsequentlyperformedacomplete front to back, that is, anterior to posterior sectioning of the organ.

Key Points
• Genetically engineered mice are increasingly used to study neuropathy and neural invasion due to pancreatic cancer. However, neuro-anatomy of mouse pancreas has not yet been critically compared to human pancreas.
• Here,weshowthatnervetrunksinthemousepancreas were exclusively located around intrapancreatic lymphoid structure and vessels, and the density of nerve trunks and particularly of sensory nerves was highest in the pancreatic head.
• The present study will serve as a reference for mouse pancreatic nerve trunk anatomy.

| Wild-type and knockout mice
Male C57BL/6J and B6.129S4-Bdnftm1Jae/J mouse strain was also purchased from the Jackson Laboratory (herein termed "Bdnf±mice").Themiceweresacrificedattheageof8weeksfor histological analysis. All animals were housed for at least 1 week prior to experimental use in micro isolators under specific pathogen-freeconditions,accordingtoFederationofLaboratoryAnimal ScienceAssociationsandinstitutionalrecommendations.

| Statistics
Resultsareexpressedasmean±standarddeviation(SD).Onlytwogroup analyses were performed, which were carried out using the unpaired ttest.Alltestsweretwo-sided,andaP value of <.05 was considered to indicate statistical significance. F I G U R E 1 Systematic harvesting andwhole-tissueembeddingofthe mousepancreas.A,B,Toensureanalysis of the complete organ and to preserve the anatomic orientation of the mouse pancreas(n=5),weexplantedthe mouse pancreas en bloc together with the adjacent duodenum and spleen, and strictly avoided collection of the neighbouringmesentery.C,Theparaffinembeddedwholeduodenum-pancreasspleen bloc was completely sectioned in anantero-posteriorfashion.D,Mouse intrapancreatic nerves were nearly solely foundintwoniches,ie(a)theperivascular regions,and(b)theperilymphoidareas. Theremainingregions,includingthe intrapancreatic septae, did not include nerves, which is in contrast with human pancreas

| Nerve trunks in the mouse pancreas are exclusively localized in perilymphoid and perivascular niches
Due to the widespread application of mouse models in the study of pancreatic diseases, it is becoming increasingly important to correctly understand the anatomy of the mouse pancreas. For this purpose, we systematically investigated the mouse intrapancreatic nerve anatomy and first performed a complete harvesting and paraffin embedding of the pancreas, strictly adhering to its normal anatomic location ( Figure 1A-C). This way, we were able to preservetheanatomicorientationandsubsequentlyperformeda complete front to back, that is, anterior to posterior sectioning of theorgan.Thisway,wegeneratedaround400slidesfromasinglepancreas,whereevery5thslidewasimmunostainedwithpanneuronal markers such as S100 or PGP 9.5 ( Figure 1A-C When looking closely at the morphology of these nerves, there was also a major difference between mouse and human pancre-

| Most nociceptive fibers are in the head of the mouse pancreas
In the next step, we placed our focus on the distribution of the nociceptive nerve fibers, since pain represents the cardinal and most severe symptom of exocrine pancreatic diseases like PCa and CP.
Here, we first looked at the amount of SP+ or CGRP+ nerve fibers within nerve trunks in the head, corpus, and tail of the mouse pancreas ( Figure 3A). Here, we found a remarkably higher propor-  Figure3C).Theseobservationssuggestedontheone F I G U R E 2 Distribution of nerve trunks inthemousepancreas.A-B,PGP9.5or S100wereusedaspan-neuralmarkers. Intrapancreatic nerve trunks in the mouse pancreasappearassmall-to-largecaliber clusters of several nerves at one of the above mentioned two particular locations. In contrast, human pancreas contains several singular nerve trunks within the normal parenchyma, between the acinar cells.C-D,Comparisonofthenervearea and nerve density in the head, corpus, and tailofthemousepancreas.E,Analysisof the differences in the nerve area at the anterior(A)vsposterior(P)surfacesof the head, corpus, and tail of the mouse pancreas. Unpaired t test

| Loss of Bdnf alters the nitrergic, but not the VIPergic, innervation of the mouse pancreatic head and corpus
In the final part, we looked at the amount of VIP+ or nitrergic, that is, nNOS-containing nerve fibers in the normal mouse pancreas ( Figure 4A) and compared these amounts to mice that were heterozygously knocked out for brain-derived neurotrophic factor (BDNF, here termed the Bdnf +/− mice), since homozygous Bdnf −/− mice are hardly alive until the adult age. 12 We recently reported that the amount of nitrergic fibers in the mouse pancreas is higher in the Bdnf +/− mice. 10

PancreaticdiseaseslikePCaandCPexhibitremarkableneuroplas-
tic changes that are closely linked to disease progression and pain status. 3,19,20 Understanding of the mechanism behind these prognostically relevant nerve alterations is highly dependent on the appli- intrapancreatic sympathetic ganglia. 26,27 These fibers project from the prevertebral ganglia and enter the pancreas either within mixed autonomic nerves or directly. 28 Furthermore, in humans, the body and the tail of the pancreas are known to be innervated from nerves fibers that arise from the celiac plexus and enter the pancreas along the branches of the splenic archery and the transverse pancreatic artery. 26 On the other hand, the pancreatic head receives the majority of the nerve fibers from the nerve plexus along the hepatic artery, the portal vein, and the inferior pancreaticoduodenal artery. [29][30][31] Thus, the entrance of nerves along neurovascular stalks is a well-known phenomenon fromthehumanpancreas.Theonlyotherstudythatanalyzedthe anatomy of mouse pancreas previously reported that the distribution of sympathetic nerve fibers in the mouse pancreas is more homogenous between the three different parts of the pancreas when compared to human pancreas. 32 However,ourstudyclearly showed that there is a concentration of nerves in the pancreatic headandcorpus,whencomparedtothetail.Consideringtheadditional difference between the anterior and posterior surface of the pancreas particularly in the corpus region, we underline that any analysis of pancreatic innervation in mouse models should pay strong attention anatomic region of tissue collection. We propose that due to lack of any significant difference between the anterior and posterior part, the choice of the pancreatic head for analysis may be more accurate for the purpose of comparison.
Another key finding of our study is the specific localization of intrapancreatic nerves around two major sites in the mouse pancreas. The first site; that is, the perivascular area, probably corresponds to the intrapancreatic continuation of the neurovascular stalks that are derived from the extrapancreatic, retroperitoneal regions. The second localization, that is, the vicinity of lymph-node like modular structures, is a novel finding that deserves attention. Lymphoid cells and immune cells are known to be regulated in their activity and differentiation by neural signals. 33,34 The close anatomic relationship that we hereby report for the first time maybe an indicator of such in your writing me on a regulation in the pancreas. When one considers the extremely highfrequencyofneuritisasneuro-inflammationinhumanCPand PCa, 1,35 it is imaginable that the proximity of nerves to lymphoid cell conglomerates may be an anatomic factor that predisposes to intrapancreaticneuro-inflammation.
It should also be underlined that the lymphoid structures that we detected in the present study are strictly intrapancreatic structures, as, due to the complete serial sectioning of the pancreas in a defined direction, we were able to exclude any structures that were not surrounded by pancreatic parenchyma.
All the lymphoid structures that were encircled by intrapancreatic nerves were also localized within the pancreas. A possible explanation for this phenomenon, but also for all the differences in the innervation of the mouse pancreas when compared to the human pancreas, lies in the differences in their intra-abdominal location. Indeed, human pancreas is a "compact, dense type" pancreasthatislocatedintheretroperitoneum.Conversely,the mousepancreasisaso-called"mesenterictype"pancreas,which isquitediffuselydistributedandembeddedintheattachedsmall intestinemesenteryandthathasanintraperitoneallocalization. 7 Due to this basic difference in the anatomy of human vs mouse pancreas,itisestimatedthemousepancreashasasimilarlymphnode drainage as the small intestine, which carries multiple lymph nodes in its mesentery. has not yet been shown; in fact, we could recently provide evidence for the lack of a role for these neuropeptides in the promotion of humanCP-associatedpain. 10 Still,thelevelsoftheSPreceptorsneurokinin-1 and neurokinin-2 receptor, 36 andCGRPlevelsintheintrathecal space have been found to associate with pain in human and ratCP. 37 Basedonourresults,itseemsthatthemajorityofSP-and CGRP-containing nerve fibers are located in the pancreatic head.
From human studies, it is known that the highest density of nerves is detectable in the pancreatic head, which is assumed to represent one of the reasons for the effectiveness of pancreatic head resection for relievingCP-orPCaassociatedpain. 3 Our present study implies a quite homogeneous distribution of VIPergic nerve fibers in the three main pancreatic regions.
Furthermore, it seems that the loss of VIPergic fibers in the pancreasisaccompaniedbyanincreaseofnitrergic,thatis,nNOS-containingnervefibersinthepancreaticheadandcorpus.Considering the recently discovered, potentially key role of nitrergic fibers in CP-associated pain, 10 we hereby underline the importance of the anatomicregionofthepancreaswhenanalyzingthenitrergicinnervation in the mouse pancreas.
The current study certainly harbors also some limitations.
Importantly, the deduced conclusions relate to our observations on nerve "trunks", thereby possibly omitting the distribution of small fiber networks, intrinsic, peri-islet neurons, and intrapancreatic ganglia that are inherently present in the human and mouse pancreas. 9,46,47 Second, we currently have no quantitative information related to differences in the 3D structure of nerve trunks and how they transverse into and through the pancreas. Therefore, future studies should increasingly apply 3D reconstruction and imaging technologies 8,9 for comparative analyses on human and mouse pancreas.Third,welimitedourstudytotheanalysisofneuropeptides likeVIP,SP,CGRP,andnNOSwithregardtosensationandpain,yet these analyses can certainly be expanded to include further neuronalsubgroups,includingTRPV1-,PACAP,5-HT,TRPA1-orTRPV4containing fiber subclasses. 48,49 Inconclusion,thepresentstudyprovidedadetailedquantitative illustration of the innervation and nociceptive fiber distribution in the mouse pancreas. Importantly, intrapancreatic nerves in mice appear asclustersofnumeroussmall-to-largenervetrunksaroundtwospecial niches, that is, the "perivascular" and "intrapancreatic perilymphoid"areas.Therefore,mouseintrapancreaticnervesexhibitmajor morphological differences when compared to human intrapancreatic nervesandareverydifficulttoanalyzewithineffortsofmurinemodellingofhumanpancreaticdisease(Table2).However,mousemodels will certainly continue to be of major benefit for understanding the molecular, genetic, and cellular repertoire of pancreatic diseases duringtheirdevelopmentandprogression.Still,aone-to-onetransfer ofconclusionsfromanatomic-histologicalobservationsinthemouse model to human disease should be avoided. In addition to calling attention to these major differences, we also hope that these observations will serve as a guide for researchers who study the role of innervation in pancreatic disease generation and progression.

CO N FLI C T O F I NTE R E S T
None.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data in the manuscript are available on request from the CorrespondingAuthor.