Optimisation and validation of immunohistochemical axonal markers for morphological and functional characterisation of equine peripheral nerves

Background: Horses are affected by various peripheral nerve disorders but defining their aetiology and pathophysiology is hampered by limited understanding of asso ciated morphological and pathological changes and involvement of specific axonal types. Objectives: To investigate the hypothesis that selected antibody markers, used in conjunction with various tissue processing methods, would enable identification of axons with different functional modalities within a range of equine peripheral nerves. Study design: Optimisation and validation study. Methods: A range of antibodies were evaluated immunohistochemically via fluores cence confocal microscopy in cadaver equine nerve samples of primary motor, mixed or primary sensory functions (recurrent laryngeal, phrenic and plantar digital) within formalin-fixed paraffin-embedded (FFPE) and formalin-fixed frozen (FFF) tissues sub jected to different antigen retrieval protocols. Results: Immunohistochemistry of FFPE-derived nerve samples with selected anti bodies and specific antigen retrieval methods enabled identification of myelinated and unmyelinated axons, cholinergic, sympathetic and peptidergic axons. The recur rent laryngeal and phrenic nerves are composed of myelinated cholinergic (motor), myelinated sensory fibres, unmyelinated adrenergic (sympathetic) axons and un myelinated peptidergic (sensory) axons. In contrast, as expected, the plantar digital nerve had no myelinated motor fibres being mainly composed of myelinated sensory fibres, unmyelinated sympathetic and unmyelinated peptidergic sensory axons. reveal characterisation morphological changes in neuropathies


| INTRODUC TI ON
Horses are affected by various neuropathies; these include, the axonal degeneration seen in peripheral nerves of horses with equine motor neuron disease, 1 motor neuropathy observed in recurrent laryngeal neuropathy (RLN), 2 sensory disorders such as equine trigeminal-mediated head shaking syndrome 3 and autonomic dysfunction seen in equine grass sickness. 4 The aetiology of many of these disorders remains poorly characterised, due in part to limitations in our understanding of pathophysiology. The latter can be facilitated by detailed characterisation of associated pathological changes but sophisticated methods for evaluating equine nerves are lacking. Typically, the morphology of equine nerves is examined in resin-embedded samples: the method is useful for assessing the severity of axonal loss and other pathological features, 5 but it is somewhat laborious and limited to specialised laboratories; importantly, the method does not enable identification of specific nerve fibres as motor, sensory or autonomic. Therefore, optimising a panel of neuronal markers to label different axonal subtypes might shed light on the pathophysiology of some equine neuropathies.
Nerve fibres are classified according to different criteria based on their size (small, medium or large diameter), their myelination (myelinated or unmyelinated), their function (sensory or motor) and their neurochemical constituents (cholinergic, adrenergic, nitrergic or peptidergic). 6 Study of the structural and functional complexity of the peripheral nervous system components requires validation of neuronal markers to investigate the pattern of tissue innervation, the associated physiological mechanisms as well as their alterations in many neuromuscular diseases. Various antibodies have been validated to label and characterise nerve fibres in dogs 7 and cats. 8 In horses, a few reports have demonstrated the neurochemical characterisation and the nature of innervation of some tissues including the small intestine, 9 nasal mucosa 10 and in equine spinal ganglia, 11 but to our knowledge, no studies have examined the neurochemical characterisation of axons within whole equine nerves by immunohistochemistry.
The aim of this study, therefore, was to optimise and validate a selection of markers, different processing techniques (formalin-fixed frozen [FFF] and formalin-fixed paraffin-embedded [FFPE]) and different antigen retrieval protocols (pressure cooker and microwave) for immunohistochemistry of equine peripheral nerves in order to provide a research tool for (in future) studying the associated pathological changes in certain equine neurological disorders. In particular, we hypothesised that immunohistochemical identification of motor fibres could be achieved by the use of an antibody to choline acetyltransferase (ChAT), 12 and that sensory unmyelinated axons would be labelled with antibodies to calcitonin gene-related peptide (CGRP) or substance P. 13 In addition, since autonomic sympathetic and parasympathetic fibres are immunoreactive to tyrosine hydroxylase (TH) and vasoactive intestinal polypeptide (VIP), respectively, 14 we hypothesised that antibodies to these proteins would be suitable for detection of autonomic fibres. This manuscript, therefore, describes the optimisation and validation of antibody markers for assessment of individual axonal types in a selection of equine nerves.

| Sample collection
Tissue samples were collected from five adult ponies and Thoroughbred horses (age range 3-8 years; all geldings) that were subjected to euthanasia as part of separate studies approved by the Home Office (PED82E67D) and local Animal Welfare Ethical Review Board. All horses and ponies were determined to be clinically and neurologically normal, and prior to euthanasia, Thoroughbred horses underwent resting and unsedated laryngoscopy 15 ; only horses with grade 1 or 2.1 laryngeal function were included as they are normal or minimally (subclinically) affected by RLN. 16

| Tissue processing
Two methods for tissue processing were used to assess optimal staining quality of various antibodies.

| 1-Formalin-fixed paraffin-embedded (FFPE) tissue
After fixation in 10% neutral buffered formalin for 24 hours, portions of the fixed nerves were subsequently embedded in wax (to enable transverse sectioning) as is routine for paraffin-embedded sections. Paraffin sections of 4 µm were cut and then dried in 37°C oven overnight before storage at room temperature until used.
Antigen retrieval was performed on paraffin-embedded tissue using two different heat-mediated antigen retrieval protocols: the first method achieved antigen retrieval with a domestic pressure cooker. Briefly, slides were dewaxed in xylene for 20 minutes and rehydrated in ethanol (twice in 100% and then 70% in distilled water for 2 minutes each). Finally, slides were incubated for 10 minutes in a pressure cooker containing 1 L of 10 mM sodium citrate buffer solution (pH 6.0) (diluted in distilled water) that had been boiling, by being placed on a hot plate at 270°C. Ultimately, this antigen retrieval protocol was adopted for immunohistochemistry of TH and ChAT.
The second antigen retrieval protocol involved the use of a domestic microwave (800 W). In brief, after dewaxing and rehydration, slides were placed in a slide rack, with the slides labelled and back-to-back oriented. After rehydration, slides were transferred to a slide box containing distilled water and left to rinse for 3 minutes, while a slide box containing 300 ml of 10 mM sodium citrate buffer solution (pH 6.0) was heated in the microwave at a power of 100% for 3 minutes. Once heated, the slide rack was added to the sodium citrate solution in the microwave at 50% power for 5 minutes.
Finally, the slides treated with these two heat-mediated antigen retrieval procedures were allowed to cool down to room temperature for 30 minutes in the same sodium citrate buffer. Then slides

Primary antibody
Host species Product code

| Immunohistochemistry
Primary antibodies used in this study are summarised in Table 1.
All secondary antibodies were IgG species-specific Alexa Fluor conjugates raised in donkey (Invitrogen (http://www.invit rogen. com) and Abcam) for wavelengths 488 (green) and 594 nm (red) and 647 nm (far red) and used at a concentration of 1:1000 (diluted in PBS).
The method of double or triple labelling is summarised in Table 2.
Briefly, after appropriate antigen retrieval of paraffin-embedded tissue, each section was circled with a hydrophobic pen (ImmEdge™ Hydrophobic Barrier Pen, H-4000) (Vector Laboratories) and blocked with 10% donkey serum in PBS for 1 hour at room temperature. Excess serum was then removed and primary antibodies were applied (as triple or double antibody mixtures) and incubated at 5°C in a cold room as follows: triple labelling with rabbit anti-NF200 5 , rabbit anti-beta III tubulin (B3T) (Abcam) and rat anti-myelin basic protein (MBP) (Abcam) was performed to stain myelinated and unmyelinated nerve fibres. A combination of rabbit anti-NF200 and rabbit anti-beta III tubulin (B3T) was used due to the relative dif-  The results of staining of formalinfixed paraffin-embedded (FFPE) tissue showed specific immunoreactivity for all markers tested. In contrast, some other markers (anti-ChAT, anti-TH and antisubstance P antibodies) did not achieve immunostaining on formalin-fixed frozen (FFF) tissues. Some markers (anti-ChAT and anti-TH antibodies) worked only with pressure cooker retrieval and the remaining markers achieved good staining quality with microwave retrieval initially used to confirm the specificity of antibodies to CGRP and substance P, ChAT and TH respectively (results not shown). Double and triple antibody labelling reactions were initially tested independently also to ensure the specificity of these markers; further, all primary antibodies were omitted in negative control reactions to ensure no nonspecific binding of the secondary antibodies (results not shown).
Tissue sections were incubated overnight with the primary antibody mixtures, and then washed three times in PBS over 10 minutes.
Secondary antibody mixtures were applied for 1 hour in a humidified chamber at room temperature, followed by three washes with PBS before adding a Slow fade diamond antifade Mountant (Invitrogen, S36963) (http://www.invit rogen.com). Slides were coverslipped and then examined by confocal microscope (Leica SP8) with at least three fascicles scanned from each nerve.

| RE SULTS
The same pattern of staining for each marker was observed in each nerve from all horses/ponies examined (n = 5). The comparative results of different nerve preparations and antigen retrieval ChAT-positive (horizontal arrows in Figures 3 and 4, B-2), some ChAT-negative myelinated fibres were also seen (arrowheads in Figure 4, B-2). In addition, unmyelinated fibres were either TH- (Figures 3 and 4 Our results also demonstrated that the fibre composition of the phrenic nerve is similar to the RLns: it contains myelinated motor and unmyelinated sympathetic and sensory fibres. The afferent innervation of the diaphragm is thought to provide sensory perception of breathing, which has a powerful impact on the regulation of motor output. 30 The extensive expression of TH-positive sympathetic fibres is mainly found in larger clusters between the myelinated fibres of the phrenic nerve in contrast with the more dispersed distribution in the RLns. This morphology might indicate that these fibres run in parallel throughout the entire length of the phrenic nerve and branch off distally within different regions of the diaphragm. While the sympathetic innervation of equine skeletal muscles has not been studied, in humans, sympathetic innervation regulates the tonicity of blood vessels in skeletal muscles, 31 and there is strong evidence F I G U R E 6 Serial sections of formalinfixed paraffin-embedded plantar digital nerve of a Thoroughbred horse stained with antibodies specific for NF200 and B3T (green) and MBP (red) (A-1 and A-2); TH (green), ChAT (absent red staining) and MBP (blue) (B-1 and B-2); CGRP (green) and collagen V (red) (C-1 and C-2). In of sympathetic innervation of intrafusal and extrafusal muscle spindle receptors in human skeletal muscles. 32 The immunostaining in the plantar digital nerve showed that all myelinated fibres in this nerve were ChAT-negative, and those that were unmyelinated are TH-or CGRP-positive. The absence of ChAT in this nerve branch supports the specificity of this marker.
Here, TH-and CGRP-positive axons confirm the presence of sympathetic and sensory axons in this nerve branch respectively.
We postulate that ChAT-negative myelinated fibres in this nerve branch are sensory myelinated fibres that innervate sensory receptors within the hoof. The release of CGRP from sensory fibres has a powerful vasodilator effect on digital arteries and veins. 33 Researchers 34 revealed that degeneration of both myelinated and unmyelinated C-fibres occurs in horses with laminitis (detected by electron microscopy), but they did not reveal the specific modality of the degenerating axons in this disorder. Consequently, application of IHC on digital nerves from horses with laminitis might help clarify the pathogenesis of this common disorder in the future.
In summary, this study provides a validation of neuronal antibody markers that could be used to study the normal composition of equine peripheral nerves and their pathological changes in various disorders, including RLN and laminitis.