Spatial transcriptomic alterations of the dorsal horn in dogs with neuropathic pain

Supplemental Digital Content is Available in the Text.

IBA1- and GFAP-positive cells in dogs with syringomyelia and neuropathic pain may show transcriptomic changes reflecting altered reactivity, upregulated inflammatory processes, and downregulated immune functions.

Abstract

Introduction:

Microglia and astrocytes are believed to play a central role in the pathogenesis of neuropathic pain (NeP). These glial cells are commonly identified by the expression of ionized calcium-binding adaptor molecule 1 (IBA1) for microglia and glial fibrillary acidic protein (GFAP) for astrocytes. Under pathological conditions, astrocytes and microglia undergo both morphological and transcriptional changes, which may promote shifts in functions that can have both protective and detrimental effects on the surrounding neuroparenchyma. Like humans, the dog breed Cavalier King Charles Spaniels (CKCSs) suffers from heritable syringomyelia (SM) which in both species is associated with NeP.

Objectives:

To investigate the potential role of IBA1 and GFAP-positive cells in the dorsal horn in CKCSs with SM and NeP.

Methods:

Using NanoString GeoMx technology, we conducted spatial transcriptomic analyses on spinal cord dorsal horns from CKCSs with SM and NeP.

Results:

Several differentially expressed genes were identified in dogs with SM and NeP. Cells positive for IBA1 showed upregulation of inflammatory genes as well as a downregulation of immune functions, while GFAP-positive cells indicated different states of reactivity. Pathway analyses indicated that the PI3K-Akt signaling pathway may be involved in the generation of NeP in CKCSs with SM.

Conclusion:

These findings provide new insights into the complex molecular changes in dorsal horn IBA1 and GFAP-rich areas in the presence of NeP and SM. The findings of this study serve as a foundation for future research that may facilitate new understandings of NeP mechanisms.

1. Introduction

Chronic neuropathic pain (NeP) is a debilitating condition causing negative impact on quality of life.¹⁵ The pathogenesis of NeP is not fully understood, but microglia and astrocytes are recognized as key contributors.^4,7,13

Microglia interact with neurons and modulate synaptic function and plasticity. Once activated, they release proinflammatory cytokines, which promote neuroinflammation^8,41 and may play a role in the development of chronic pain.⁸ Astrocytes help maintaining synaptic function, regulating neurotransmitters, and secreting growth factors and other neuromodulatory molecules. At the cellular level, they are typically identified by the expression of glial fibrillary acidic protein (GFAP), while microglia are marked by ionized calcium-binding adaptor molecule 1 (IBA1). Under pathological conditions, astrocytes and microglia undergo both morphological and transcriptional changes.^8,12,23,24 This may have both protective and detrimental effects on the surrounding neuroparenchyma.¹² Activation of astrocytes and microglia in sensory pathways may enhance nociceptive signaling, potentially leading to abnormal pain experiences, characteristic of NeP, such as allodynia and hyperalgesia.³

Like humans, the dog breed Cavalier King Charles Spaniel (CKCS) suffers from heritable syringomyelia (SM) with NeP, making it an interesting model for investigations of NeP mechanisms.^18,29 Both dogs and humans with SM can be symptomatic or asymptomatic.^10,42 In symptomatic subjects, NeP is a dominant clinical sign.^22,32 In dogs, clinical signs of NeP include yelping in pain, hypersensitivity to touch of the dermatomes associated with the syrinx, scratching toward neck and shoulder, a preference for head elevation during sleep, and sleep disruption.^32,42 We have previously reported that spinal cord changes include a clear pattern of ipsilateral changes in the dorsal root entry zone (sensory area), characterized by deafferentation and reorganization of first-order axons into deeper laminae in cases with lateralized fictive scratching.²⁹ In cases where the syrinx affects the lateral border of the dorsal horn, gray matter, oedema, and white matter degeneration affects tracts of importance for noxious processing.²⁹

Spatial transcriptomic analysis allows RNA sequencing while maintaining cellular integrity and preserving tissue structure,⁴⁵ and is in this study used to explore the potential role of IBA1 and GFAP-positive cells in the dorsal horn in CKCSs with SM and clinical signs of NeP.

2. Methods

2.1. Animals

Spinal cord tissue from 13 euthanized privately owned CKCSs were used in the study. Based on physical, neurological, and magnetic resonance imaging investigations, and neuropathology, dogs were categorized as follows:

Group I—cases with SM and clinical signs of NeP (symptomatic): four intact females and two neutered females, one intact male and one neutered male.

Group II—subclinical (asymptomatic) cases with SM and no signs of NeP: two intact females.

Group III—controls with no SM and no signs of NeP: three intact females.

Seven dogs in group I, one dog in group II, and three dogs in group III participated in a previous study.²⁹

2.1.1. Phenotype characterization

All Danish dogs had a full physical and neurological examination and magnetic resonance imaging scan. Cases additionally had standard hematology, biochemistry, and thyroid profiles. A particular set of clinical signs characterize CKCSs with SM and NeP.^32,42 An extensive structured questionnaire using dichotomous questions designed to target such signs was used to detect owner-reported clinical signs of SM. Clinical signs that appeared while examining dogs, confirming the owners' observations, were also recorded. Similar investigations were used for the two British dogs.

All dogs had gross necropsy and histopathological interpretation of the spinal cord.

For ethical reasons, cases #1 to 8 were treated with pain relief medicines (Table 1).

Table 1.

Signalment, age at death, signs recognized as clinical markers of syringomyelia and pain in Cavalier King Charles Spaniels, and treatment for cases with MRI and histopathological confirmed syringomyelia.

	Case #1	Case #2	Case #3	Case #4	Case #5	Case #6	Case #7	Case #8
Sex	Female	Female	Female	Female	Female	Male	Female	Male
Neutering status	Intact	Intact	Neutered	Intact	Intact	Neutered	Neutered	Intact
Age at death	2.5 y	8.2 y	12.4 y	5.4 y	9.1 y	10.8 y	8.3 y	3.3 y
Clinical signs
Aversion against wearing collar/harness	X	X	X	X		X		Not wearing collar or harness
Scratching neck area (with skin contact)	X Left side worse if excited					X Right side	X Right side
Phantom scratching neck area (without skin contact)			X Right side	X Left side		X Right side	X Right side
Aversion against being touched or groomed in the neck region	X	X	X	X		X	X	X
Withdrawn from people and other pets	X	X	X	X		X	X	X
Pain vocalization with changing to certain body positions			X	X	X	X	X	X
Prefer to rest and sleep with head elevated		X						X
Sleep disruption		X	X	X				X
Pain treatment		GabapentinNSAID	Gabapentin	Pregabalin	Pregabalin	PregabalinNSAID	NSAID	PregabalinPrednisolone

Besides the treatment appearing from the table, dog #3 and dog #8 also received opioid treatment. MRI, magnetic resonance imaging; NSAID, non-steroidal anti-inflammatory drugs.

Dogs were donated for postmortem research by their owners after written consent. Dogs were euthanized due to treatment-resistant SM-related pain or to other non-neurological disease causing a poor quality of life. The age of dogs in group I ranged from 1.5 to 12.4 years, in group II 11.4 to 15.6 years, and in group III 0.8 to 8.8 years.

The study was approved by the Ethics Committee at the Department of Clinical Veterinary Science at University of Copenhagen (File Number 2014-5) and Animal Experiments Inspectorate under the Ministry of Food, Agriculture, and Fisheries (License Number 2006/561-1145, 2012-15-2934-00700 and 2016-15-0201-01074). All dogs were treated ethically as patients in accordance with The Danish and European legislations on the protection of animals (Act. No. 61 of January 19, 2024) and animals used for scientific purposes (Directive 2010/63/EU).

2.2. Tissue sampling

Spinal cords (C1–C8) were sampled within 90 minutes after euthanasia as described by Agerholm.¹ Spinal cords were immersed in 4% phosphate neutral-buffered formaldehyde (VWR Chemicals, Søborg, Denmark), stored at room temperature for 48 hours, and dehydrated using a Thermo Shandon Citadel 1000 tissue processor (Shandon Diagnostics Ltd., Chesire, United Kingdom). The cords were embedded in paraffin in a metal mold and kept at 40°C for 1 hour before cut into serial 5-mm transversal sections and placed in plastic cassettes and additional paraffin added.²⁹

2.3. RNAscope

RNA quality of the tissues was evaluated using RNAscope before initiating the transcriptomic protocol (see supplementary materials, http://links.lww.com/PR9/A356). The results were evaluated using the RNAscope scoring guidelines of Advanced Cell Diagnostics, Inc. The guideline is semiquantitative and based on the number of dots per cell, correlating to the number of RNA copies. The results are given as scores from 0 to 4 with zero meaning no staining or less than 1 dot per 10 cells and 4 corresponding to >15 dots per cell or more than 10% of the dots in clusters.

2.4. Digital spatial profiling

2.4.1. Tissue preparation

Most tissue sections were selected at the caudal part of the syrinx, as this area is subject to maximum stress and may therefore offer a more accurate representation of the tissue activity (Fig. 1).¹¹ Four samples from dogs in group I were sampled rostrally because of severe tissue destruction of the caudal region. Tissue sections (5 μm thick) were mounted on Superfrost Plus slides and incubated overnight at 37°C, followed by a two-hour incubation period at 60°C. Slides were deparaffinized in xylene 3 times for 5 minutes each, followed by rehydration in ethanol 100% and 95% for 2 × 5 minutes each and 5 minutes, respectively. Slides were subsequently washed in 10x phosphate-buffered saline (PBS) with a pH of 7.4 (BioNordika, Herlev, Denmark, cat. no. BN-53100). Slides were transferred to 99°C bio-water (BioNordika, cat. no. BN-51100) for 10 seconds, before being incubated at 99°C in 1x Tris-EDTA (Invitrogen, Waltham, MA, cat. no. 00-4956-58) for 30 minutes and immediately moved to 1x PBS afterward. To expose RNA targets, slides were incubated in 0.5 μg/mL Proteinase K solution (Invitrogen, cat. no. AM2546) for 15 minutes, preheated to 37°C, and washed in PBS for 5 minutes. Immediately after, the slides were washed in 4% phosphate neutral-buffered formaldehyde (NBF, Sigma Aldrich, Saint Louis, MO, cat. no. HT501128) for 5 minutes followed by 2 × 5 minutes in NBF Stop buffer (Tris base, Sigma Aldrich, cat. no. T1503, glycine, Sigma Aldrich, cat. no. G7126, and bio-water) and 5 minutes washing in 1x PBS. The RNA probes (GeoMx Canine Cancer Transcriptome Atlas) and Buffer R (licensed NanoString reagent) were preheated to room temperature, briefly vortexed, and spun down before applied to the slides as a hybridization solution alongside the probes. Gene Frames (Invitrogen, cat. no. AB0578) were applied to cover the tissue, and slides were incubated overnight in a hybridization chamber in a HybEZ II Hybridization System (Advanced Cell Diagnostics Inc., Newark, CA) at 37°C. The following day, Gene Frames were removed, and the slides were washed twice for 25 minutes in a Stringent Wash solution consisting of equal parts 4X saline-sodium citrate (SSC) (Sigma-Aldrich, cat. no. S6639) and 100% formamide (AppliChem, Darmstadt, Germany, cat. no. A2156), followed by 2-minute washes in 2X SSC solution twice. Next, the tissue was blocked with Buffer W (GeoMx RNA Slide Prep Kit for FFPE, Nanostring Technologies, Seattle, WA, cat. no. 121300313) for 30 minutes in a humidity chamber at room temperature. Unconjugated polyclonal rabbit antibody against IBA1 (2:1000, FujiFilm WAKO, Osaka, Japan, cat. no. 019-19741) was applied, and slides were incubated in a hybridization chamber for 1 hour at room temperature. Two washing steps in 2X SSC for 1 minute and 4 × 3 minutes were performed. Secondary Alexa Flour 647-conjugate goat antirabbit antibody (1:400, Invitrogen, cat. no. A21245) was applied, and slides were incubated for 30 minutes. This was followed by similar washing steps in 2X SSC. Finally, a rabbit Alexa Fluor 594-conjugate GFAP monoclonal antibody (1:150, Invitrogen, cat. no. A-21295) and nucleic acid stain SYTO 83 (1:125, Invitrogen, cat. no. S11364) were applied, followed by 1 hour incubation and 2 similar washing steps in 2X SSC. Slides were kept in 2X SSC at 4°C until the following day.

Figure 1.

T1-weighted midline sagittal MRI scan of the neck of a Cavalier King Charles Spaniel, illustrating the approximate region where the syrinx was sampled. For tissue examination, the caudal region of the syrinx is selected (white line) to represent the least chronic part of the syrinx, where it is believed to most accurately reflect the activity of the syrinx. In few of the dogs with syringomyelia and neuropathic pain (n = 4), the rostral section of the syrinx was sampled (orange line) due to severe tissue destruction caudally. MRI, magnetic resonance imaging.

A negative control was performed for all tissue samples.

2.4.2. NanoString GeoMx digital spatial profiling slide processing and selection of regions of interest

Slides were loaded into the slide holder of the GeoMx Digital Spatial Profiler. Each slide was covered with Buffer S (GeoMx RNA Slide Prep Kit for FFPE, Nanostring Technologies, cat. no. 121300313) and a new 96-well collection plate was installed. Exposure time was 300 to 400 milliseconds for Texas Red (GFAP) and Cy5 (IBA1) channels, and 200 milliseconds for Cy3 channels (SYTO83). Regions of interest (ROIs) were selected for each tissue section covering both dorsal horns using the build-in software of the GeoMx Digital Spatial Profiler (Fig. 2). The maximum size of a ROI was 660 × 785 µm. Regions of interest were further divided into individual segments, from which RNA probes were collected. It was not possible to collect ROIs based on spinal laminae, as there would be too few positive cells per ROI. Therefore, transcriptomic data from superficial and deep laminae cannot be separated. Unlike single-cell RNA sequencing, spatial transcriptomic technology collects probes from specific segments within ROIs called areas of illumination (AOIs), identified using fluorescent-based segmentation from fluorescent antibody signals. Choosing a cell-specific antibody as a marker enhances the likelihood of collecting RNA from specific cell types, although RNA from any cell beneath the antibody-defined cell may also be collected. This spatial aspect allows for the investigation of the environment surrounding the area of interest.

Figure 2.

Regions of interest (ROIs) in a caudal tissue section from a dog with syringomyelia and clinical signs of neuropathic pain. Both dorsal horns were included in the selection of ROIs (indicated by white lines). Yellow fluorescence indicates GFAP-positive structures and red fluorescence indicates IBA1-positive structures. GFAP, glial fibrillary acidic protein; IBA1, ionized calcium-binding adaptor molecule 1.

Regions of interest were collected from both dorsal horns to obtain the most accurate representation of activity in the spinal cord segment. As the syrinx progresses, dorsal horn tissue is destroyed or modified.²⁹ This destruction may occur in one or both dorsal horns as well as clinical signs that may be uni- or bilateral,³² highlighting the importance of collecting from both dorsal horns. Each ROI was manually drawn in polygonal shapes of varying sizes ranging from 54.6 × 10³ μm² to 434 × 10³ μm² and further divided into 2 antibody-guided AOIs: GFAP-positive regions and IBA1-positive regions. Glial fibrillary acidic protein areas of illumination were collected first using the following parameters: Erode 1 μm, N-Dilate 2 μm, hole size 8 μm², and particle size 5 μm². The parameters were similar for IBA1 except for hole size, which was set to 10 μm². Segmentation parameters were defined to exclude cells exhibiting concurrent GFAP and IBA1 expressions. The area size and nuclei of the AOIs varied for each ROI, with IBA1 area ranging from 1053 μm² to 17055 μm², and GFAP area ranging from 4229 μm² to 76040 μm². The nuclei count varied from 30 to 330 and from 43 to 1024 for IBA1 and GFAP, respectively.

2.4.3. Library preparation and sequencing

Next-generation sequencing (NGS) was initiated on an Illumina NovaSeq 6000 platform according to the specifications in the GeoMx-NGS User Manual (10153-05, NanoString Technologies). FASTQ files generated by the Illumina sequencer were converted to digital count conversion files using the GeoMx NGS pipeline and software.

2.4.4. GeoMx quality control

Quality control of AOIs with figures including principal component analysis plots are shown in Supplementary Figures S1-S3, http://links.lww.com/PR9/A356.

2.4.5. Data analysis

Data were subjected to normalization and quality control using R software version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria), the GeoMx NGS pipeline, and the standR package.²⁵ The threshold for cell count was adjusted to 25 and the library threshold to 1000. A linear-mixed model included in the standR package was used to identify differentially expressed genes between dogs from groups I and II and between groups I and III. These results were followed by a modified t test and Benjamini–Hochberg multiple testing adjustments for a False Discovery Rate < 0.05.²⁵ To test for pseudoreplication, the intra- and interindividual correlations have also been calculated with the R function: duplicateCorrelation. The correlations were built into the statistical model; however, as the MA-plots looked almost identical (Supplementary Figures S4-S5, http://links.lww.com/PR9/A356), only the original data are presented.

For statistical comparisons between the three clinically defined groups, the Kruskal–Wallis test was used for independent variables. The significance level was set at P < 0.05. The test was performed using IBM SPSS Statistics version 29 software.

2.4.6. Pathway analysis

Pathway analyses were performed with a default confidence level of 0.4 and a high confidence level of 0.9 to identify processes and pathways with a high degree of certainty. The Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes, and Reactome databases were used with the Canis lupus familiaris genome as a reference. The results were filtered for relevance to the central nervous system (CNS).

3. Results

3.1. RNAscope and immunohistochemistry

The RNAscope protocol yielded a score of 2, corresponding to 4–9 mRNA dots per cell and very few or no clusters. Immunohistochemistry showed that dogs with SM displayed higher intensity of GFAP fluorescence than control dogs (Fig. 3). The intensity also seemed more pronounced in dogs with SM and NeP than in those without clinical signs.

Figure 3.

Immunofluorescence-stained spinal cord tissue section from a control dog, a dog with syringomyelia and no neuropathic pain, and a dog with syringomyelia and neuropathic pain. The image shows (A) a control dog (no syringomyelia and no clinical signs of neuropathic pain), (B) a dog with syringomyelia but no clinical signs, and (C) a dog with syringomyelia and clinical signs of neuropathic pain. An artefact due to trauma to the cord during sampling is present on the right side of the cord (B) (asterisk). Yellow fluorescence indicates GFAP-positive structures, and red fluorescence indicates IBA1-positive structures. Dogs with syringomyelia (B and C) displayed higher GFAP fluorescence intensity than control dogs (A). The intensity also appeared more pronounced in dogs with syringomyelia and neuropathic pain (C) than in those without clinical signs (B). GFAP, glial fibrillary acidic protein; IBA1, ionized calcium-binding adaptor molecule 1.

3.2. Identification of differentially expressed genes related to neuropathic pain

A principal component analysis plot confirmed that mRNA expression differed within GFAP-positive and IBA1-positive AOIs (Fig. 4). Although both AOIs were present in all groups of dogs, they seemed more prominently expressed in dogs with SM and clinical signs of NeP.

Figure 4.

2D-PCA plot demonstrating the expression of GFAP and IBA1 Areas of Illumination in the entire study population. The plot illustrates that the 2 Areas of Illumination (GFAP and IBA1) differed in each clinically defined group: controls, syringomyelia with no clinical signs of neuropathic pain, and syringomyelia with clinical signs of neuropathic pain. The difference in expression between GFAP and IBA1 dots indicates that the 2 areas represent different cell groups. GFAP, glial fibrillary acidic protein; IBA1, ionized calcium-binding adaptor molecule 1; PCA, principal component analysis; SM, syringomyelia.

To assess gene expression in IBA1 and GFAP-positive cells in dogs with SM and signs of NeP, we analyzed the differentially expressed (DE) genes in this group of dogs compared with those in dogs with SM and no clinical signs as well as with the control group. Ninety-nine genes were differentially expressed when comparing dogs with clinical signs of NeP with those without clinical signs, 79 of which were related to IBA1-positive areas (Tables 2 and 3, Fig. 5). When comparing dogs with signs of NeP with the control group, 22 DE genes were found, with only 2 genes related to IBA1-positive areas (Tables 4 and 5, Fig. 6): UBB and GLUL. Both genes showed downregulated expression, with no upregulated DE genes identified in this comparison.

Table 2.

Differentially expressed genes related to glial fibrillary acidic protein–positive areas in the dorsal horn of dogs with syringomyelia and clinical signs compared with dogs with syringomyelia and no clinical signs.

GFAP-positive AOIs
Gene	Log (fold change)	Average expression	P Raw	P Adj
GLI1	0.518	9.07	4.81E-04	0.028
FZD3	0.386	9.70	6.70E-04	0.037
LCN2	0.360	9.85	7.84E-04	0.041
B2M	−0.562	11.2	6.96E-07	3.28E-04
DLA-12	−0.494	10.4	8.71E-07	3.28E-04
UBC	−0.419	11.4	1.05E-06	3.28E-04
VIM	−0.714	10.1	1.66E-06	3.88E-04
CD81	−0.431	11.9	3.02E-06	5.67E-04
HSPB1	−0.403	10.6	9.69E-06	0.002
RPL7A	−0.385	10.6	1.90E-05	0.003
LOC607874	−0.565	12.6	2.27E-05	0.003
PHGDH	−0.382	10.6	7.92E-05	0.008
FABP7	−0.371	11.1	8.17E-05	0.008
GFAP	−0.573	14.0	1.24E-04	0.011
CLU	−0.534	11.7	1.68E-04	0.013
SLC14A1	−0.377	9.73	2.36E-04	0.017
FGF1	−0.399	9.57	3.05E-04	0.020
NTRK2	−0.411	10.1	3.66E-04	0.023
SERPINE2	−0.303	10.3	8.39E-04	0.041
FGFR3	−0.314	10.6	9.07E-04	0.043

Upregulated genes are shown in bold, and downregulated genes are shown in italic.

AOIs, areas of illumination; GFAP, glial fibrillary acidic protein.

Table 3.

Differentially expressed genes related to ionized calcium-binding adaptor molecule 1–positive areas in the dorsal horn of dogs with syringomyelia and clinical signs compared with dogs with syringomyelia and no clinical signs.

IBA1-positive AOIs
Gene	Log (fold change)	Average expression	P Raw	P Adj
EIF4EBP1	0.865	9.62	3.39E-06	1.92E-04
IL11	0.910	9.21	5.56E-06	2.89E-04
BMP2	0.801	9.68	1.49E-05	6.65E-04
MAP3K7	0.800	9.66	1.64E-05	6.87E-04
COL4A4	0.690	9.73	7.94E-05	0.003
SMO	0.633	10.1	1.55E-04	0.005
POLR2D	0.674	9.83	2.13E-04	0.007
TDO2	0.665	9.61	2.36E-04	0.007
LTA	0.677	9.52	2.51E-04	0.007
FOXA2	0.704	9.46	3.18E-04	0.009
CXCR1	0.690	9.54	4.17E-04	0.012
COL24A1	0.601	9.87	4.77E-04	0.012
ITGB8	0.640	9.75	4.88E-04	0.012
LIFR	0.611	10.1	4.94E-04	0.012
LOC102156626	0.606	9.69	4.98E-04	0.012
HHIP	0.577	9.82	5.26E-04	0.012
MAPK1	0.485	10.6	6.22E-04	0.014
CD40LG	0.595	9.69	7.12E-04	0.015
RAG2	0.579	9.68	7.30E-04	0.015
CLEC1B	0.605	9.62	7.89E-04	0.016
NEIL1	0.586	9.76	8.05E-04	0.016
PPP2R2C	0.577	9.65	9.52E-04	0.018
S100A8	0.601	9.57	9.69E-04	0.018
SKP1	0.577	9.59	9.77E-04	0.018
FGFR2	0.558	9.94	0.001	0.018
NUMBL	0.630	9.46	0.001	0.020
PLA2G4E	0.583	9.47	0.001	0.020
CD320	0.558	9.83	0.001	0.021
FGFR3	0.540	10.6	0.001	0.021
SH2D1A	0.566	9.49	0.002	0.025
PLCB4	0.570	9.58	0.002	0.025
CHEK2	0.537	9.65	0.002	0.025
CCNE2	0.580	9.60	0.002	0.027
NOG	0.551	9.67	0.002	0.027
MTNR1A	0.577	9.33	0.002	0.031
MMRN2	0.538	9.59	0.002	0.031
RASA4B	0.534	9.53	0.002	0.034
NDUFA4L2	0.517	9.78	0.003	0.037
ATRX	0.512	9.80	0.003	0.037
SPRED1	0.479	9.95	0.003	0.037
CCL26	0.567	9.34	0.003	0.042
PI3	0.569	9.33	0.003	0.044
NT5E	0.528	9.90	0.004	0.046
ITPK1	0.497	9.92	0.004	0.047
LOC100855618	0.511	9.93	0.004	0.047
ALKBH3	0.513	9.70	0.004	0.048
HSD11B1	0.545	9.59	0.004	0.049
APLNR	0.543	9.63	0.004	0.049
CD74	−1.27	10.8	2.64E-23	2.47E-20
B2M	−1.11	11.2	3.72E-15	1.75E-12
LOC607874	−1.18	12.6	3.16E-14	9.87E-12
DLA-12	−0.985	10.4	1.35E-12	3.15E-10
DLA-DRA	−0.870	10.1	5.51E-11	1.03E-08
FCER1G	−0.810	10.3	1.57E-10	2.45E-08
C1QA	−0.618	10.7	7.35E-08	9.84E-06
RPL7A	−0.648	10.6	1.88E-07	2.21E-05
FCGR2B	−0.811	9.71	2.22E-07	2.31E-05
VIM	−1.05	10.1	2.75E-07	2.57E-05
HLA-DRB1	−0.693	10.0	4.93E-07	4.11E-05
GFAP	−0.818	14.0	5.27E-07	4.11E-05
S100A6	−0.647	10.9	1.21E-06	8.70E-05
CD63	−0.614	10.5	1.75E-06	1.17E-04
LGALS1	−0.622	10.6	2.10E-06	1.31E-04
LYZ	−0.727	9.77	3.48E-06	1.92E-04
APOE	−0.638	15.0	6.68E-06	3.24E-04
DLA-DQB1	−0.546	10.8	6.91E-06	3.24E-04
TXNIP	−0.573	10.7	1.69E-05	6.87E-04
C1QB	−0.510	10.3	4.20E-05	0.002
CSF1R	−0.549	10.2	5.12E-05	0.002
CD81	−0.416	11.9	9.71E-05	0.003
SLC2A1	−0.531	10.5	1.40E-04	0.005
GNAS	−0.500	12.4	4.74E-04	0.012
CLU	−0.580	11.7	6.64E-04	0.015
THY1	−0.500	11.0	6.77E-04	0.015
C3	−0.469	10.1	0.001	0.018
APP	−0.415	11.4	0.002	0.024
UBC	−0.343	11.4	0.002	0.025
DLA-DQA1	−0.437	9.55	0.004	0.049
FGL2	−0.414	10.0	0.004	0.049

Upregulated genes are shown in bold, and downregulated genes are shown in italic.

AOIs, areas of illumination; IBA1, ionized calcium-binding adaptor molecule 1.

Figure 5.

Volcano plots depicting differentially expressed genes in GFAP and IBA1-positive areas in the dorsal horn of dogs with syringomyelia and clinical signs of neuropathic pain compared with dogs with syringomyelia and no clinical signs of neuropathic pain. Upregulated genes are shown in red, and downregulated genes are shown in blue. GFAP, glial fibrillary acidic protein; IBA1, ionized calcium-binding adaptor molecule 1.

Table 4.

Differentially expressed genes related to glial fibrillary acidic protein–positive areas in the dorsal horn of dogs with syringomyelia and clinical signs of neuropathic pain compared with dogs without syringomyelia and no clinical signs (controls).

GFAP-positive AOIs
Gene	Log (fold change)	Average expression	P Raw	P Adj
CLU	0.551	11.7	3.95E-06	7.40E-04
COL11A2	0.444	9.78	4.62E-05	0.006
SERPING1	0.415	10.2	5.59E-05	0.006
CD74	0.391	10.8	1.90E-04	0.015
B2M	0.358	11.2	4.22E-04	0.028
SERPINE2	0.312	10.3	5.82E-04	0.036
CD44	0.364	10.5	6.43E-04	0.036
C1QB	0.350	10.3	6.51E-04	0.036
UBB	−0.546	12.8	6.79E-12	6.37E-09
NDUFA13	−0.396	11.3	2.60E-07	1.08E-04
TPI1	−0.385	11.6	3.47E-07	1.08E-04
PPP2R1A	−0.365	11.1	1.36E-06	3.18E-04
PKM	−0.312	11.8	2.87E-05	0.005
PEBP1	−0.267	12.5	5.05E-05	0.006
ALDOA	−0.331	11.1	1.08E-04	0.010
NEFL	−0.538	12.0	1.21E-04	0.010
UQCRQ	−0.326	10.3	2.33E-04	0.017
S100B	−0.297	11.0	7.17E-04	0.037
SPTAN1	−0.268	11.1	8.09E-04	0.040
ATP5F1D	−0.251	11.1	8.46E-04	0.040

Upregulated genes are shown in bold, and downregulated genes are shown in italic.

AOIs, areas of illumination; GFAP, glial fibrillary acidic protein.

Table 5.

Differentially expressed genes related to ionized calcium-binding adaptor molecule 1–positive areas in the dorsal horn of dogs with syringomyelia and clinical signs compared with dogs without syringomyelia and no clinical signs.

IBA1-positive AOIs
Gene	Log (fold change)	Average expression	P Raw	P Adj
UBB	−0.516	12.8	2.67E-08	2.50E-05
GLUL	−0.400	11.4	3.12E-05	0.015

Both genes were downregulated, and no upregulated genes were identified.

AOIs, areas of illumination; IBA1, ionized calcium-binding adaptor molecule 1.

Figure 6.

Volcano plots depicting differentially expressed genes in GFAP and IBA1-positive areas in the dorsal horn of dogs with syringomyelia and clinical signs of neuropathic pain compared with dogs without syringomyelia and no clinical signs (controls). Upregulated genes are shown in red, and downregulated genes are shown in blue. GFAP, glial fibrillary acidic protein; IBA1, ionized calcium-binding adaptor molecule 1.

3.3. Pathway analysis

Pathway analyses were performed for IBA1 and GFAP-related DE genes comparing SM dogs with and without clinical signs of NeP (Figs. 7 and 8) as well as comparing SM dogs with clinical signs and control dogs (Fig. 9). Pathway analyses were similar for all comparisons at both confidence levels.

Figure 7.

Pathway analysis in IBA1-positive areas for dogs with syringomyelia and clinical signs of neuropathic pain compared with dogs with syringomyelia and no clinical signs. Downregulated processes (blue) are mainly related to immune responses. Upregulated processes (red) are related to cellular communication and cytokine production. IBA1, ionized calcium-binding adaptor molecule 1.

Figure 8.

Pathway analysis in GFAP-positive areas for dogs with syringomyelia and clinical signs of neuropathic pain compared with dogs with syringomyelia and no clinical signs. There are no upregulated processes in GFAP-positive areas of illumination (AOIs) between dogs with syringomyelia and neuropathic pain and dogs with syringomyelia and no neuropathic pain. Downregulated processes (blue) mainly involve fibroblast growth factor receptor 3 (FGFR3). FGFR3, fibroblast growth factor receptor 3; GFAP, glial fibrillary acidic protein.

Figure 9.

Pathway analysis in GFAP-positive areas for dogs with syringomyelia and clinical signs of neuropathic pain compared with controls. Upregulated processes (red) include intracellular pathways related to apoptosis, whereas downregulated processes (blue) are mainly related to cellular metabolism. GFAP, glial fibrillary acidic protein.

3.3.1. Syringomyelia with neuropathic pain compared with syringomyelia with no neuropathic pain

In IBA1-positive AOIs from dogs with SM and clinical signs of NeP, there were several up- and downregulated differences compared with dogs with SM and no clinical signs (Fig. 7). Downregulated processes were mainly related to immune responses. Upregulated processes were related to cellular communication and cytokine production. The process of “signal receptor binding” seemed both up- and downregulated, covering different sets of genes. The genes related to an upregulation of “signal receptor binding” were LTA, IL11, LOC100855618, MAP3K7, CD40LG, CCL26, SPRED1, LIFR, BMP2, CD320, ITGB8, and SMO. The genes linked to downregulation were APOE, APP, CD81, C3, DLA-12, GNAS, THY1, FGL2, and CD74. There were no upregulated processes in the GFAP-positive AOIs in dogs with SM and NeP compared with dogs with SM and no clinical signs of NeP. In dogs with clinical signs of NeP, there was a downregulation of intracellular cascades, especially those related to fibroblast growth factor receptor 3 (FGFR3) (Fig. 8).

3.3.2. Syringomyelia with neuropathic pain compared with controls

Upregulated processes in GFAP-positive AOIs included intracellular pathways related to apoptosis as well as regulation of the complement cascade and regulation of cell-to-cell adhesion. Downregulated processes are mainly related to cellular metabolism (Fig. 9). No upregulated pathways or biological or molecular processes were found when comparing IBA1-positive AOIs from dogs with SM and clinical signs of NeP with the control group.

3.3.3. Evaluation of sex-related influences

To assess whether the sex of the dogs influenced the results, the two male dogs were omitted from the sample of dogs with SM and NeP, and the analysis repeated. For both GFAP and IBA1-positive AOIs, the total number of DE genes decreased. A new downregulated DE gene was identified when comparing GFAP-positive cells in SM-dogs with and without clinical signs of NeP, but pathway analysis indicated no differentially regulated processes (Supplementary Material Table S1, http://links.lww.com/PR9/A356). For IBA1-positive cells, seven new upregulated DE genes were identified (Supplementary Material Table S2, http://links.lww.com/PR9/A356). Pathway analysis revealed several upregulated and downregulated processes and pathways, most of which aligned with those in Figure 7. Upregulation was primarily associated with cytokine activity, signal transduction, and cell communication including new pathways such as regulation of MAPK and ERK1/ERK2 cascades, AMP signaling, and NF-kappa B. Downregulation was predominantly linked to reduced immune processes, consistent with Figure 7.

Compared with the control group, the total number of GFAP-related DE genes increased, while the total number of IBA1-related DE genes decreased (Supplementary Material Table S3-S4, http://links.lww.com/PR9/A356). For GFAP-positive cells, pathway analysis identified multiple regulated processes, most being upregulated. These processes were primarily associated with the immune system and stress response (Fig. 9). Only three downregulated processes were observed, related to Parkinson disease, respiratory electron transport, and the RAF/MAP kinase cascade. For IBA1-positive cells, pathway analysis showed no differentially expressed processes between the two clinical groups.

4. Discussion

Pathway analyses confirmed that our ROIs were mainly selective for astrocytes and microglia. There were no upregulated pathways or processes for GFAP-positive cells in dogs with SM and NeP compared with dogs with SM and no NeP. Downregulation of the MAPK signaling pathway suggests a reduction in activated astrocytes, which has previously been suggested to increase the recruitment of microglia to the site.^2,26 Astrocytes regulate synaptic transmission and are involved in synaptic plasticity, glutamate regulation, and creation of long-term potentiation.³¹ This process seems to be downregulated in SM dogs with NeP. Other downregulated pathways are related to FGFR3 binding and signaling. Fibroblast growth factor receptor signaling has been suggested as a requirement to maintain astrocytes in a nonreactive state.^21,46 Downregulation of FGFR3-related pathways may potentially disrupt this balance, leading to increased astrocyte reactivity.

Microglia are considered the innate immune cells of the CNS. Our pathway analysis of IBA1-positive AOIs showed downregulation of innate immune functions, including regulation of complement cascades. Downregulation also included the processes “positive regulation of response to stimulus,” “regulation of immune response,” and “positive regulation of immune response.” These findings align with a previous study on immune-challenged mice, where similar downregulation patterns have been observed and linked to inflammatory neurological disorders associated with subsequent depressive behavior.¹⁶ This is further supported by the upregulation of the pathway “long-term depression.” This suggests a microglial transcriptomic dysregulation with a downregulation of microglial immune function and an increased inflammatory profile in dogs with SM and NeP compared with dogs with SM and no NeP. This is supported by the upregulation of different cytokine activities and processes, TGF-β signaling, and signal transduction, indicating the presence of active microglia. We observed an increase in the PI3K-Akt signaling pathway, which may contribute to altered microglial activity and increased neuroimmune responses. Dysfunctional signaling in the PI3K-AKT pathway is believed to play a significant role in the initiation and maintenance of neuroinflammation and uncontrolled activation of this pathway may have detrimental effects on the cellular environment.⁹ This pathway is closely linked to microglial activity, as microglia possess various receptors that modulate the PI3K-AKT signaling cascade.⁹ Dysregulation of cell surface receptor expression or functionality that affects this pathway disrupts microglial activity, promoting neurodegenerative processes and sustained neuroinflammation.⁹ The PI3K-AKT signaling pathway was upregulated alongside an upregulation in cytokine activity and cytokine receptor binding and interactions in dogs with SM and NeP compared with dogs with SM and no NeP. The chronic presence of proinflammatory microglia and a dysregulated PI3K-AKT pathway may possibly contribute to the onset and progression of neurodegeneration observed in these dogs.

The differentially expressed genes in GFAP-positive AOIs suggest the presence of different astrocytic cell states in dogs with SM and NeP. Upregulated genes (LCN2 and FZD3) indicate reactive astrocytes.^12,23,49 Lipocalin-2 (LCN2) upregulation has been proposed to be essential for development of chronic itch by modulating astrocyte–neuron interactions in the spinal cord.^19,36 In CKCSs with clinical SM, fictive scratching directed toward the dermatome corresponding to the syrinx-affected region of the superficial dorsal horn is a recognizable clinical sign.^32,35 The spinal sensory motor pathways involved in fictive scratching may be influenced by astrocytic LCN2 upregulation. However, other genes linked to reactive astrocytes (B2M, VIM, HSPB1) were downregulated.^23,49 FABP7, a gene important for fatty acid metabolism and myelin structure, was downregulated, which could potentially affect fatty acid processing and myelin integrity in SM dogs with NeP, affecting neuronal function and communication.³⁹

In IBA1-positive AOIs, genes linked to inflammation, chemokines, and leukotriene A (ATRX, LTA, CXCR1, S100A8), along with MAPK1 and PI3, were upregulated, as confirmed by pathway analysis.^16,28,38 Downregulated genes were linked to immune function and the complement system (CD74, FCER1G, C1QA, C1QB, C3).^5,6,48,49 Genes associated with phagocytic activation and reactivity (CD74, C1QA, APOE, CSF1R, FGL2) were also downregulated, as supported by the pathway analysis.^{24,27,37,43,49} These findings may reflect the chronicity of the spinal cord disease SM. The colony-stimulating factor 1 receptor (CSF1R) is predominantly expressed by microglia in the CNS and is essential for microglial development and maintenance.⁹ Abnormal CSF1R expression can impair microglial function and has been implicated in the pathogenesis of neurodevelopmental and neurodegenerative diseases.⁹ Furthermore, aberrant PI3K-AKT signaling driven by abnormal CSF1R expression has been suggested to contribute to the onset of neurodegenerative diseases.⁹ These abnormal processes may be involved in the pathogenesis of SM with concurrent NeP in CKCSs, as our results show an upregulation of PI3 concurrent with a downregulated expression of CSF1R in IBA1-positive cells in these dogs compared with dogs with SM and no NeP. Elevated levels of PI3 protein have previously been observed in a rat model of NeP, suggesting a link between increased PI3 levels and NeP.⁴⁷

Compared with the control group, dogs with SM and NeP showed upregulation of reactivity genes (SERPING1, CD74, B2M, CD44, C1QB) in GFAP-positive AOIs.^23,34,37,49 CD44 upregulation suggests potential affection of the dorsal horn extracellular matrix in CKCSs with SM. Pathway analysis indicated a more stressful cellular environment in dogs with SM and NeP, characterized by upregulated pathways related to apoptosis regulation and the complement cascade, while metabolic and homeostatic pathways were downregulated. CD44 is involved in cell interactions and extracellular matrix stability and may lead to excessive hyaluronan accumulation in Alzheimer disease, causing perivascular and perineural infiltration, blood–brain barrier disruption, decreased blood flow, and compromised perineuronal nets.³⁴ Hyaluronan regulates astrocytic morphology and glutamate transporter function,^17,33 and microglia may degrade perineuronal nets in peripheral neuropathies.^30,40 In IBA1-positive AOIs, two genes were downregulated in dogs with SM and NeP compared with the control group: UBB, a housekeeping gene known to be downregulated in inflammatory microglia, and GLUL, which is involved in converting glutamate to glutamine.^20,38

There may be sex-dependent genetic differences between the individual dogs, and therefore, we did a sensitivity analysis by excluding the two male dogs. This did not lead to major changes in the overall conclusion.

We acknowledge certain limitations to our study: the GeoMx platform lacks single-cell resolution, meaning that gene expression data may reflect contributions from multiple cell types within GFAP- and IBA1-positive regions, rather than exclusively from astrocytes and microglia. The GeoMx Canine Cancer Transcriptome Atlas, a novel set of RNA probes designed for investigations of canine neoplasia, includes genes that are not necessarily related to the central nervous system. It is, however, the only panel of RNA probes specifically designed for canines, which increases the reliability of our findings and minimizes the risk of false-negative results due to species differences. Another restriction to our study is the small number of dogs participating in the study reducing statistical power. For ethical reasons, most cases used in this study received medication for pain control with gabapentin or pregabalin, which might influence the transcriptomic profile of spinal cord cells, potentially reducing cytokine expression in microglia.

Future studies should consider single-cell transcriptomic analysis to gain more comprehensive insights into the cellular environment, including neurons in the dorsal horn of dogs affected by SM with and without NeP and multiplexed protein mapping and cellular profiling to evaluate the functional consequences of the transcriptomic alterations.

Our study provides new information about molecular mechanisms contributing to the pathogenesis of NeP. The spontaneous SM-NeP model in CKCSs offers an interesting supplement to experimental rodent models due to anatomical and physiological similarities.^14,44

Disclosures

N.B.F. has received consultancy fees from PharmNovo, Vertex, NeuroPN, Nanobiotix, Neurvati, and Samiona, has undertaken consultancy work for Aarhus University with remunerated work for AKIGAI, Biogen, Merz, and Confo Therapeutics, and has received grants from IMI2PainCare an EU IMI 2 (Innovative medicines initiative) public-private consortium and the companies involved are: Grunenthal, Bayer, Eli Lilly, Esteve, and Teva, outside the submitted work. The other authors declare no conflicts of interest.