Characterization of idiopathic chronic diarrhea and associated intestinal inflammation and preliminary observations of effects of vagal nerve stimulation in a non-human primate

Abstract

Background:

Diarrhea is commonly associated with irritable bowel syndrome, inflammatory bowel disease, microscopic colitis, and other gastrointestinal dysfunctions. Spontaneously occurring idiopathic chronic diarrhea is frequent in rhesus macaques, but has not been used as a model for the investigation of diarrhea or its treatment. We characterized this condition and present preliminary data demonstrating that left vagal nerve stimulation provides relief.

Methods:

Stool consistency scores were followed for up to 12 years. Inflammation was assessed by plasma C-reactive protein, [¹⁸F]fluorodeoxyglucose (FDG) uptake, measured by postitron emission tomography (PET), multiplex T cell localization, endoscopy and histology. The vagus was stimulated for 9 weeks in conscious macaques, using fully implanted electrodes, under wireless control.

Key results:

Macaques exhibited recurrent periods of diarrhea for up to 12 years, and signs of inflammation: elevated plasma C-reactive protein, increased bowel FDG uptake and increased mucosal T helper 1 T-cells. The colon and distal ileum were endoscopically normal, and histology revealed mild colonic inflammation. Application of vagal nerve stimulation to conscious macaques (10 Hz; 30 sec every 3 hours; 24 hours a day for 9 weeks) significantly reduced severity of diarrhea and also reduced inflammation, as measured by FDG uptake and C-reactive protein.

Conclusions & Inferences:

These macaques exhibit spontaneously occurring diarrhea with intestinal inflammation that can be reduced by VNS. The data demonstrate the utility of this naturally occurring primate model to study the physiology and treatments for chronic diarrhea and the neural control circuits influencing diarrhea and inflammation that are not accessible in human subjects.

Graphical Abstract

Vagus nerve stimulation reduced diarrhea and bowel inflammation. Affected macaques had diarrhea, enhanced bowel fluorodeoxyglucose (FDG) uptake, elevated plasma CRP and increased mucosal Th1 type T cells.

1. INTRODUCTION

Diarrhea is a prominent symptom in a wide spectrum of gastrointestinal conditions ranging from disorders of gut brain interaction, such as irritable bowel syndrome (IBS),^1–3 to inflammatory conditions such as inflammatory bowel disease (IBD) and microscopic colitis. It can be chronic and debilitating. Diarrhea is a hallmark of IBD and occurs in nearly 80% of patients with IBD.^{4, 5} as well as being a feature of microscopic colitis.⁶ Treatment of diarrhea is based on the underlying cause. Available drugs for diarrhea treatment are typically peripherally-restricted opioid agonists such as loperamide (Imodium), eluxadoline (Viberzi) and diphenoxylate-atropine (Lomotil), the serotonin receptor type 3 receptor antagonist alosetron (Lotronex) and the gut-restricted antibiotic, rifaximin (Xifaxin). Each of these treatments is variably effective and can cause serious side effects.⁷ Moreover, opioid-based treatments are largely discouraged for use both due to side effects and the potential for abuse.

Water and electrolyte secretion in the gastrointestinal tract is under neural control, and in some cases diarrhea has been linked to abnormal over-activity of secretomotor neurons.^8–10 A possible way of targeting neural control in the gastrointestinal tract, vagal nerve stimulation, (VNS) has been investigated for its anti-inflammatory effects,^{11, 12} and has been shown to reduce diarrhea that is associated with intestinal inflammation in rodents.^{13, 14} Spontaneously occurring severe recurrent episodes of diarrhea have been described in non-human primates (NHPs), including rhesus macaques.^{15, 16} This condition presents in the absence of a specific pathogen, and consequently has been termed idiopathic chronic diarrhea (ICD).¹⁵ ICD in NHPs would appear to be a better animal model for the investigation of diarrhea therapy than provocative models, such as use of castor oil, because of its chronic and spontaneously occurring nature. Here we have characterized ICD and evaluated the effects of VNS on its clincal presentation and associated intestinal characteristics.

2. MATERIALS AND METHODS

2.1. Animals

Rhesus macaques (Macaca mulatta) were obtained from the Wisconsin National Primate Research Center (WNPRC). Animals affected with chronic diarrhea were identified from WNPRC electronic health records and selected for the study in consultation with the veterinary staff. Inclusion in the study was confirmed by a subsequent high stool consistency score (SCS). The WNPRC carried out a microbial analysis to confirm that diarrhea did not result from known bacterial or other pathogenic infections (Campylobacter, Shigella, Salmonella, Giardia lamblia, Trichomonas, and Balantidium coli). Control subjects were selected based on clinical histories that showed lack of diarrhea or any other health problems. The animals were housed individually, in adjacent cages in the same room, and received food and water ad lib. For animals implanted for VNS, the diets were matched. Animals with ICD were provided with a daily fiber supplement (Inulin, 2 g) before the start of the study. This supplement was continued and was added to the daily diet of the control subject at the time of study initiation. Any additional treatments or procedures were also matched between the control and ICD animals. All procedures were approved by the University of Wisconsin-Madison Animal Care and Use Committee (G006085).

2.2. Stool consistency score (SCS)

We quantified the occurrence of diarrhea by computing a SCS from daily observations carried out by the WNPRC animal care staff, who were trained by WNPRC’s veterinary staff to judge the quality of the feces, and were blinded to this study. The fecal output of every macaque was graded and recorded using a scale from 0–6, from the time of weaning to the time at which the study ended. The scores represent the following: 0 = normal solid/formed feces; 1 = soft less formed feces; 2 = soft less formed feces and unformed diarrhea; 3= soft less formed feces and watery liquid diarrhea; 4 = unformed diarrhea; 5 = unformed diarrhea and watery liquid diarrhea; 6 = watery liquid diarrhea.

2.3. Blood collections

Blood samples were collected from the saphenous or femoral vein of animals previously trained to allow blood collection while awake and gently restrained, following WNPRC standard operating procedures. Blood was collected 1–3 times per week into K2EDTA vacutainer tubes and, immediately following collection, whole blood samples were centrifuged at 1500 rpm to separate red cells from serum. Sera were stored at −80°C for later analysis of C-reactive protein (CRP) using an NHP-specific ELISA, according to manufacturer instructions (CRP-3, Life Diagnosis, Inc., Westchester, PA).¹⁷

2.4. Positron emission tomography

For [¹⁸F]fluorodeoxyglucose (FDG) positron emission tomography (PET) imaging, sedated NHPs (10mg/Kg ketamine, i.m.) were transported to the adjacent imaging facility via a dedicated corridor. In the facility, the animals were intubated for isoflurane anesthesia and received a catheter in the saphenous vein for FDG administration. A computed tomography (CT) scan of the entire body was followed by a bolus injection of FDG (3.5–5 mCi). PET imaging consisted of a series of 6–12 min whole body scans, performed at 15 min intervals over 90 min. At the end of the scan, the animals were transported back to their home cages and allowed to recover from the anesthesia.

2.5. FDG-PET/CT data analysis

PET images were obtained on a GE Discovery PET/CT 710 (D710) scanner (GE HealthCare, Piscataway, NJ, USA). The PET and CT images were exported from the scanner in Digital Imaging and Communications in Medicine (DICOM) format and imported into ITK SNAP software (version 3.8.0, www.itksnap.org) to segment the gut from duodenum to anus. Segmentation was performed manually based on anatomical landmarks in the CT images and the gut as a whole was labeled as the volume of interest (VOI). The signal from a region of the biceps was used as a benchmark control. All PET images were converted into standardized uptake value (SUV) units by normalizing the activity concentration to the dosage of FDG injected, time of injection relative to scan start, and the subjects’ body weight. SUV was evaluated at the same time across all subjects and quantitation of SUV was based upon the whole body acquisition, corresponding to 60–75 min post FDG injection. The PET signal within the segmented VOI and control region was quantified by importing the PET images and segmentation into MATLAB 2022 (Mathworks, Natick, MA), and run through a custom script (Appendix 1). The quantified SUV for the intestinal VOI was normalized to the control biceps region for the final measurement of the SUV ratio (SUVR).

2.6. Endoscopy and Histology

Endoscopies were carried out in the morning. In the afternoon prior to endoscopy, the animals were sedated with ketamine (10 mg/Kg, i.m.) and Golytely^® (polyethylene glycol 3350 with added electrolytes, Braintree Laboratories, Inc., Holbrook, MA) mixed in warm water (up to 50 mL/Kg) was administered via the orogastric or nasogastric route using a feeding tube. A video endoscope (VET-1335, Endoscopic Image Processor VS1000 (LG-200, MDS Inc., Valrico, FL) was inserted via the anus into the colon under direct visualization and advanced to the terminal ileum to evaluate the colonic and small intestinal mucosa and to obtain biopsies using biopsy forceps. The colonoscopies were carried out by a single, board-certified gastroenterologist (SS). Pinch biopsies were collected during endoscopy from the terminal ileum, right colon (cecum and ascending colon), left colon (descending colon and sigmoid colon) and rectum, placed in labeled tissue cassettes and fixed in >20 volumes of 10% normal buffered formalin. Tissue samples were processed according to standard histological techniques and embedded in paraffin. Sections were cut at 5 μm thickness using a HistoCore BIOCUT microtome (Leica Biosystems, Deer Park IL), followed by standard protocols for deparaffinization and staining with hematoxylin and eosin. Sections were evaluated by a board-certified pathologist (KAM) blinded to the clinical diagnosis of the animal.

2.7. Localization utilizing iterative bleaching extends multiplexity (IBEX)

Paraffin-embedded NHP tissue sections from formalin-fixed paraffin embedded (FFPE) colonic biopsies were deparaffinized and dehydrated according to standard protocols. Antigen retrieval was performed in 10 mM citrate buffer (pH 6.0) at 95 °C for 30 min. A modified version of the IBEX protocol,¹⁸ where tissue sections were bleached to remove sample autofluorescence, and to allow iterative staining was performed with 4.5% H₂O₂ and 24 mM NaOH for 2 h at room temperature under LED light exposure. Validation was performed to assess that this procedure resulted in bleaching of Alexa Fluor488, Alexa Fluor 647, and eFluor570, eFluor615, and that the bleaching process preserved antibody reactivity. Blocking to prevent non-specific antibody binding was performed with 5% BSA, 5% Hu-FC block (Invitrogen) in 0.1% TBST for 1 h at room temperature. Tissue sections were then incubated with primary conjugated antibodies (Table 1) overnight at 4°C. Tissues were incubated with non-conjugated primary antibodies, washed extensively and incubated with Alex Fluor 488-conjugated goat anti-mouse IgG1 secondary antibody (Cat #A21121) for 1 h at room temperature. After extensive washing, sections were incubated with nuclear dye DAPI (TBS TritonX100 0.5% v/v) for 10 min and coverslip mounted using Fluoromount G (Invitrogen). Confocal imaging was performed on a Leica SP8 STED 3X microscope with a 40X NA1.3 objective. Acquisition of the entire tissue section was achieved by tiling, whereby adjacent areas are imaged with a 10% overlap and digitally merged using the Leica software. Multiple focal planes (Z-stack) were acquired for each tile and were comprised of 16 to 20 Z-sections 0.7 μm apart. After imaging, coverslips were gently removed by floating in 1x PBS, with samples subjected to bleaching as indicated above. Slides were blocked as before, and a second round of antibody staining was performed.

Table 1.

List of antibodies used in this study for IBEX on NHP tissues

Antigen	Clone	Suitable for IBEX	Fluorophore	Supplier	Cat No.	Dilution
CD3	SP34–2	Y	Alexa Fluor 488	BD Biosciences	557705	1:20
FOXP3	236A/E7	Y	eFluor 570	eBioscience	41-4777-82	1:20
T-bet	4B10	Y	Alexa Fluor 647	BioLegend	644804	1:20
CD14	M5E2	Y	Alexa Fluor 488	BD Biosciences	561706	1:50
CD163	10D6	Y	Unconjugated	LS Bio	C87534	1:20
CD192/CCR2	48607	Y	PE	R&D Systems	FAB151P	1:20

2.8. Confocal microscopy data processing

Image registration for each successive round of immunolabelling in IBEX was performed by aligning the DAPI-stained nuclei. To register the image volumes, the middle plane of each stack in the DAPI channel was extracted as a new image. With StackReg from Pystackreg, a rigid body transformation was used to find the transformation matrix that translates and rotates image 2 to match image 1. This transformation matrix was then applied to all channels and planes of image 2. A new matrix was created for the final image with the same x,y,z dimensions and the total number of channels. This matrix was populated with the data from image 1 and the transformed image 2 channels except the DAPI channel. This was saved as a tiff image and viewed with Napari viewer.¹⁹ Counting cells of interest was performed by manual identification of cell types of interest and Cellpose²⁰ as a Napari plugin to identify all nuclei allowing for normalization to cell number across tissue samples.

2.9. Vagus nerve stimulation: surgery and experimental timeline

The timeline for device implantation and VNS treatment is shown in Figure 1. Following subject selection, FDG-PET imaging was carried out to determine whether subjects exhibited intestinal inflammation. After elevated FDG uptake was confirmed, the subjects were implanted for VNS with a bipolar, 2 mm inner diameter cuff electrode (LivaNova, PerenniaFLEX Model # 304; LivaNova, London, UK) around the left cervical vagus nerve following the methodology adapted for NHPs²¹. The simulating electrode was connected to a pulse generator (Livanova Sentiva Model # 1000). A subcutaneous pocket was created by extending the incision used to implant the stimulating cuff towards the thorax, on the left side, just inferior to the neck. The pulse generator was placed in the pocket and connected to the electrode wires. The connections to the electrode were tested by measuring impedance. Two rubber anchors were used to secure the leadwire to the muscles of the neck. The pocket was closed off from the neck incision to keep the pulse generator from shifting.

FIGURE 1. Experimental timeline.

Rhesus macaques underwent surgical implantation of electrodes and the pulse generator, following which diet was matched for a minimum of one week before baseline FDG-PET imaging, endoscopy, histology samples, and blood were collected. VNS treatment was for 63 days (D0 to D62) with further sampling in the week following VNS.

Following the surgery, the diets of the ICD and control animals were matched and baseline data collection, including blood collection, was started. Upon completing data acquisition for this stage, VNS was initiated at Day 0 (D0 in Figure 1), and the measurements were repeated during the 63 days of treatment. Post-treatment measurements were subsequently obtained.

2.10. Electrical stimulation

Electrical stimulation consisted of symmetrical bipolar pulses, 250 μsec per phase, delivered at 10 Hz for 30 sec every 3 hours, 24 hours a day. The stimulus frequency was based on a study in human that showed reduction of clinical signs in Crohn’s disease with intermittent 10 Hz stimulation²², a stimulus frequency also shown to be effective in reducing intestinal inflammation in rats.²³ The amplitude of the stimulation was 275 μA for 26 days, followed by 350 μA for 16 days, and 500 μA in the final 21 days of treatment. Stimulation was monitored and controlled using the manufacturer provided LivaNova tablet and wand system, which allowed stimulation parameters to be changed and impedance to be monitored non-invasively via Bluetooth by placing the wand on the subjects’ chest in close proximity to the implanted pulse generator.

2.11. Statistics

All experimental data were analyzed using MATLAB (The Mathworks, Natwick, MA). To assess the difference in stool consistency scores (SCSs) between control and ICD animals, six control and six ICD animals were identified from the colony. SCSs for each animal were averaged across its recorded history. These average values were then used to compare SCS across groups using a one-way ANOVA (p < 0.05) to test for the main effect of disease status on SCS. Three of each of the six control and ICD animals were used for further measurements of inflammation. We applied the Mann Whitney (Wilcoxan rank sum) single-tailed test to test the hypothesis that the median of control is less than the median of ICD for the PET and ICD signals.

One control and two ICD subjects then went on to receive VNS. A nested two-way ANOVA (p < 0.05) was carried out to evaluate the main effect of stimulation, the main effect of group (control vs. ICD), and the interaction between stimulation and group on FDG-PET and CRP. This also tested for differences between subjects within each group.

3. RESULTS

3.1. Characteristics of idiopathic chronic diarrhea (ICD) in rhesus macaques

The ICD affected animals exhibited relapsing and remitting diarrhea most of their lives, whereas non-affected animals experienced transient increases in SCS. The time course of the disease was tracked over 8 to 12 years in six ICD and six control animals (Figure 2). The SCS data, recorded daily with the 0–6-point scale, were first smoothed using a 30-day sliding window. Subsequently, the envelope of each subject’s SCS was computed by locating the peaks in the smoothed function and then fitting a piecewise cubit polynomial to the resulting data (Figure 2A). Stool scores from healthy controls are plotted in blue and animals afflicted with ICD are plotted in red. Subjects diagnosed with ICD had significantly higher SCS values compared to non-affected control animals (Figure 2B, nested ANOVA, F(1,35200) = 6411.34, p <0.0001). There was also significant variability of subjects within each group (nested ANOVA, F(1,10) = 506, p < 0.0001). Two ICD subjects (r14012 and r11079), and one control (r13082), were selected for studying the effects of VNS on intestinal inflammation.

FIGURE 2. Stool consistency scores in 6 unaffected NHPs and 6 NHPs affected with ICD, followed for 8 to 12 years.

A: Stool consistency scores (SCS), recorded daily with the 0–6-point scale. Data from healthy control subjects (blue) and ICD afflicted animals (red) are plotted as a function of time (years). B: Average SCS values computed from each animal’s history (shown in A) for healthy control and ICD afflicted NHPs. Data presented as mean ± 95% confidence intervals, open circles represent each subject’s average SCS. *** Control and ICD significantly different, p<0.0001.

3.2. Noninvasive PET imaging and CRP measurements in ICD affected rhesus macaques

We assessed intestinal metabolic activity using FDG-PET / CT imaging in three healthy control and three ICD subjects. Using this approach, identification and image segmentation of the gastrointestinal tract was performed allowing comparison of FDG uptake between control and ICD subjects (Figure 3A & B). Significantly increased FDG uptake was observed by PET in the gastrointestinal tract of ICD afflicted compared to healthy control subjects, qualitatively (B) and quantitatively (C, ANOVA, F(1,4) = 8.73, p = 0.042, control v. ICD). Substantially elevated FDG signals were observed in images of the stomach, small intestine, right and left colonic flexures and transverse colon. There was also an increased signal in the liver (Figure 3).

FIGURE 3. Inflammatory signals in ICD affected NHPS, compared to unaffected controls.

FDG-PET images in (A) control, unaffected NHP, compared to (B) an NHP with ICD. The color map was standardized between all subjects and indicates the uptake of FDG, with black and blue representing low uptake, and red indicative of high uptake. C: Quantification of the FDG uptake in the segmented complete gastrointestinal tract, weighted relative to the signal in the biceps muscle (standardized uptake value relative to biceps, SUVR) for healthy control and ICD affected animals (n=3 animals/group). The control was significantly less than ICD (p<0.05, Mann Whitney (Wilcoxan rank sum) test. (D) The plasma CRP concentration was significantly greater in animals with ICD compared to control (both n = 3). The control was significantly less than ICD (p<0.0076, Mann Whitney (Wilcoxan rank sum) test.

Inflammation was further assessed by quantification of serum C-reactive protein (CRP), a biomarker of inflammation. Consistent with animals afflicted with ICD having ongoing inflammation, serum CRP levels were significantly, and nearly three-fold, higher compared to control animals (ANOVA, F(1,35) = 19.32, p < 0.0001). There was also significant variability of subjects within each group (nested ANOVA, F(4,35) = 4.94, p = 0.00289).

3.3. Assessment of intestinal inflammation by endoscopy and histopathological analysis

To determine the extent of intestinal inflammation at a macroscopic level, colonoscopy with ileal intubation was performed on healthy and ICD affected animals. Complete colonoscopy with intubation of approximately 10 cm of the terminal ileum was confirmed by photodocumentation. Despite recurrent diarrhea (high SCS), elevated CRP values and increased FDG signal compared to control, endoscopy revealed normal mucosa from the distal colon to the terminal ileum in ICD compared to control animals. There was no evidence of mucosal ulceration, scaring or ischemic injury in these tissues. During this procedure mucosal biopsies were obtained from specified discrete sites including the ileum, right colon, left colon, and rectum.

Histopathological analysis of these biopsies by a pathologist blinded to clinical diagnosis revealed that as expected the colonic mucosa from healthy NHPs showed normal crypt architecture from the luminal surface to the muscularis mucosae (Figure 4A–C). At higher magnification, the lamina propria contained lymphocytes, plasma cells and scattered eosinophils with consistent spacing noted between the colonic crypts (Figure 4B). The colonic mucosa of ICD afflicted animals exhibited regions with basal lymphoplasmacytosis, alterations in crypt spacing (Figure 4D, F), and occasional distorted architecture including crypt branching (Figure 4G). In some animals with ICD, granulomas were identified (Figure 4E). Surface mucin was less apparent in the regions of inflammation. There was no evidence of metaplasia, cryptitis or crypt abscess formation. These findings demonstrate a patchy, mucosal inflammation in the ICD affected NHPs.

FIGURE 4. ICD affected Rhesus macaques exhibit low-level inflammation by histopathology.

A-C: Colon histology of control, unaffected, NHPs in which mucosal architecture was normal. At higher magnification (rectangle in “A”) the regions between glands were narrow and contained small numbers of lymphocytes, plasma cells and eosinophils (B: ellipse), and (C) surface goblet cells can be seen. D-G: There was mild expansion of the lamina propria by an increase in inflammatory cells in colon biopsies from idiopathic chronic diarrhea (ICD) NHPs. Branching of crypts was also apparent (F, arrows). This lamina propria expansion results in increased spacing between the crypts (circled in G). A poorly-formed granuloma was noted in a NHP with ICD and likely represents a reparative process to a previously injured/damaged crypt (circled in E). Intra-epithelial lymphocytes are indicated by arrows in H. The box in D is enlarged in E, and that in F is enlarged in G. Musc. muc. = muscularis mucosae.

3.4. Vagal nerve stimulation reduces ICD symptoms and reduces clinical inflammation

To assess the efficacy of VNS on ICD symptoms, stimulation of the left vagus nerve was performed using 250 μsec bipolar pulses, delivered at 10 Hz for 30 sec every 3 hours over a period of 9 weeks. The conscious NHPs were observed while VNS was applied. There were no observable reactions, (no flinching, no apparent distress and no vocalization) of the NHPs to the stimulation. Thus, stimulation that diminished manifestations of ICD, described below, had no observable behavioral effects.

The SCS observations over time for the ICD and control subjects are shown in Figure 5. VNS treatment suppressed diarrhea in affected NHPs, as indicated by reduced SCS scores. There was a significant main effect of VNS across all animals compared to the equivalent epoch of time immediately preceding stimulation (nested two-way ANOVA, F(1,610) = 6.583, p = 0.0105). The interaction of group (control vs. ICD) and VNS was also significant (two way nested ANOVA, F(1,610) = 15.056, p = 0.0001), indicating that the main effect of VNS differed between the two groups and was driven primarily by the reduction of SCS in the ICD subjects (Figure 5B).

FIGURE 5. Effect of VNS on the stool consistency score (SCS) from ICD affected subjects and a control macaque.

A, C, E: Daily SCS in rhesus macaques from 4.8 to 9.5 years (left panels), and time prior to, during and after VNS on an expanded time scale (right panel). VNS lasted 63 days (grey shaded area). B, D and F: The SCS data for the 9 weeks pre-VNS, 9 weeks of VNS, and the 9 weeks post-VNS are shown as violin plots. In C, note that during the 63 days of VNS, SCSs were 0 or1, except for a single day, when the macaque was administered intragastric Golytely. Other than at times of VNS, scores ranged between 0 and 6. See text for statistics.

We next investigated whether the markers of inflammation, FDG uptake and plasma CRP were decreased by VNS in the ICD subjects. VNS reduced inflammation as indicated by decreases in both the FDG signal in the gastrointestinal tract (Figure 6A), and in plasma CRP (Figure 6B). Consistent with the FDG uptake data presented in Figure 3, the control subject differed significantly in FDG uptake overall compared to the two ICD subjects (nested two way ANOVA, F(1,3) = 80.09, p = 0.0029). The interaction of group (control vs. ICD) with vagal stimulation was significant (nested two-way ANOVA, F(1,3) = 76.95, p = 0.0031). It is interesting that VNS reduced FDG uptake in the ICD subjects, but actually slightly increased it in the control (Figure 6A). The decreases in FDG uptake are consistent with the effect of VNS on SCS in the ICD subjects.

FIGURE 6. Chronic VNS reduces intestinal FDG signal and plasma CRP in subjects with ICD.

In two NHPs with ICD, vagal nerve stimulation reduced intestinal inflammation and plasma CRP levels. A) FDG uptake (standardized uptake value relative to biceps, SUVR) in the GI tract of control and ICD afflicted animals before VNS treatment, and from two scans during VNS, at day ~33 (375 μA stimulation), and ~55 (during 500 μA stimulation). B) CRP levels from samples taken during baseline period and throughout the VNS epoch for the same subjects. Data are mean ± 95% CI (note in some cases the CI falls within the symbol). See text for statistical comparisons.

Similarly, the groups differed significantly in CRP levels, with the ICD subjects having higher CRP levels compared to the control (nested two-way ANOVA, F(1,79) = 17.45, p = 0.0001). There was a significant effect of stimulation across all animals (nested two-way ANOVA, F(1,79) = 6.17, p = 0.0151), with VNS reducing CRP in the ICD subjects. The interaction of subject and VNS was also significant (two way nested ANOVA, F(1,79) = 14.63, p = 0.0003), indicating that the main effect of VNS was different between the two groups and driven primarily by the reduction of CRP in the ICD subjects.

3.5. Immunological profiling in rhesus macaques afflicted with ICD

To further investigate the increased immune cell numbers and immunopathology observed in ICD afflicted animals (subjects r14012 and r11079), we used a modified iterative bleaching extends multiplexity (IBEX) protocol to perform iterative immunolabelling and profile immune cell types in Rhesus macaque colonic tissues. This procedure was performed on colonic tissue sections from healthy control (Figure 7A – left panels) and ICD afflicted animals (Figure 7A- right panels) to identify T-cells (DAPI⁺ CD3⁺), that were T helper1 or regulatory T-cells by expression of the lineage-specific transcription factors; T-box expressed in T cells (Tbet+) or Foxp3⁺respectively. Although CD3⁺ T-cells and regulatory T-cells were identified, quantification of these subsets revealed only CD3⁺ Tbet⁺ T-cells were significantly increased (Figure 7B–D). Using IBEX, we identified and quantified these T-cell populations in a single subject pre-VNS, during VNS, and after treatment (Figure 7E). These data highlight the utility of IBEX to track complex immune phenotypes from endoscopic biopsies over time in the same subject.

FIGURE 7. ICD affected Rhesus macaques exhibit increased CD3⁺ T cells in biopsies from the colon.

A: Representative images of colonic tissues isolated from healthy control (left) and ICD affected subjects (right). Regions from the upper panels are reproduced in the lower panels at greater power with and without nuclear (DAPI) staining. Arrows: CD3⁺ FoxP3⁺ T cells; S: serosal side; L: Luminal side. Quantification of CD3⁺ T cells (B), CD3⁺ FoxP3⁺ T cells (C) and CD3⁺ Tbet⁺ T cells (D) between healthy control (n=2) and ICD affected subjects (n=2). E: Quantification of T cell populations in an ICD affected subject before, during and after VNS treatment. All quantifications were normalized against total number of nuclei in each tissue section.

4. DISCUSSION

Chronic recurring diarrhea is a prominent animal health issue requiring veterinary care in colonies of non-human primates.²⁴ The condition is not attributable to parasitic or bacterial infection at the time of disease manifestation.¹⁵ Chronic diarrhea is also a common problem for human patients and one for which there are no good animal models.

4.1. Nature of the disease

The cohort of macaques at the Wisconsin National Primate Research Center (WNPRC) that we have investigated exhibit diarrheal disease that has a history similar to that previously described in other groups of NHPs.^{15, 16, 24} Diarrhea in affected macaques at WNPRC was remitting and relapsing and lasted for many years, as long as 12 years in the macaques we have investigated. The NHPs also had plasma CRP levels about 3-fold greater than those of unaffected controls and showed FDG signals above control in the gastrointestinal tract measured as a whole. The histological changes, including the presence of rare crypt-rupture-associated granulomas, and increased numbers of lymphocytes within the surface epithelium, resembled human microscopic colitis, which is also associated with watery diarrhea.⁶ Other studies have also identified the histological changes in NHPs with chronic diarrhea as microscopic colitis.¹⁶ Endoscopy revealed that ICD afflicted animals had a macroscopically normal colonic mucosa that was free of ulceration, petechiae, or cobblestone appearance common to ulcerative colitis or Crohn’s disease. Similar normal endoscopy with increased intraepithelial lymphocytes has been observed in human patients in the setting of microscopic colitis.⁶ However, greater signals of inflammation were seen in FDG-PET images of the stomach, small intestine, liver and right and left colonic flexures than would be expected of microscopic colitis. The condition that we investigated has some parallels with IBS with diarrhea (IBS-D), in that it arose spontaneously in a sub-group of individuals and fluctuated in intensity over time. Mild intestinal inflammation has been reported in patients with IBS-D.^25–27 Using a modified iterative bleaching multiplexity (IBEX) technique¹⁸ that allows immunophenotyping in situ, we identified increased helper1 T-cell subsets in the macaque colon. Increased numbers of mucosal T cells also occur in IBS-D.^{26, 28} However, CRP is often elevated in patients with either IBD or IBS, and thus is not a differentiating marker.^{29, 30} Thus, our findings suggest that inflammation in ICD animals recapitulates some aspects of the pathology of microscopic colitis and has some features that resemble IBS-D.

4.2. Measures of inflammation

Fluorodeoxyglucose PET imaging effectively reveals the location of disease activity because inflamed tissue has high FDG uptake due to its elevated metabolic activity and glucose utilization.³¹ Thus, FDG-PET can be used to localize and quantify the extent of inflammation, including inflammation of intestinal tissue.^{32, 33}

We utilized FDG-PET imaging, which in human populations is useful to assess and locate regions of inflammation non-invasively with a high sensitivity and specificity, compared to more invasive methods.^32–34 In the current study, FDG-PET revealed enhanced signals in the stomach and small intestine. In the colon, strong signals were observed at the left and right colonic flexures and transverse colon, but was not apparent in other regions.

Plasma CRP was elevated 3-fold in affected NHPs. CRP, produced in the liver in response to cytokines arriving from the gut via the portal veins, facilitates complement binding to damaged and foreign cells, thus acts as a first line of defense against infections when the mucosa of the intestine is compromised.³⁵ It is an indicator of the overall degree of gastrointestinal inflammation.³⁴

4.3. Identification of T-cell subtypes

Although prior immunological assessments have been conducted on NHPs with ICD, many of these assays did not assess the affected colon or are the result of flow cytometric profiling of an entire section of bowel. With this in mind, we modified the IBEX technique¹⁸ to allow multiparameter identification of immune cell types in the lamina propria. Using this approach, we identified specific subsets of T-cells based on surface expression of CD3 and transcription factors critical to T helper lineage specification in endoscopic biopsy samples from healthy and ICD-afflicted animals. Specifically, ICD-afflicted NHPs had increased numbers of CD3⁺ Tbet⁺ T-cells indicative of Th1 T-cells, suggesting increased expression of cytokines such as IFNγ in the lamina propria. Although Th1 T-cells and cytokines have not been identified previously, it is possible that this reflects the source of the sample material. For example, the lack of IFNg mRNA transcripts were reported only in whole tissue mesenteric lymph node and spleen samples. Profound reductions in cytokine production have been observed at the single T-cell level with ex vivo restimulation and intracellular cytokine staining in ICD afflicted animals. These deficits in not only IFNγ but also IL-4, IL-13, and TNFα production shown by flow cytometry,³⁶ have been further substantiated by ex vivo cytokine release assays.³⁷ Together these findings could suggest that, while there are increased numbers of specific Th1 T-cells, these cells are in an exhausted state and non-functional. This conjecture has not been formally addressed and requires further investigation.

A benefit of IBEX, in contrast to flow cytometry where the entire sample is consumed, is that new slides of archived tissues can be generated to provide additional insight. Prior to this study, IBEX had been developed for use on mouse and human tissues capable of identifying 40 discrete cellular proteins, with the number of community-validated antibodies growing over time. While other high-parameter commercialized approaches have been developed, including the CODEX/phenocycler system, there has been limited use in NHP tissues since this system requires custom conjugation of antibodies and extensive validation.³⁸ To our knowledge, our study is the first to apply IBEX allowing for high parameter identification of cell types in the NHP mucosal immune system during inflammation.

4.4. Effects of vagal nerve stimulation

In the current study, the presenting gastrointestinal dysfunction was watery diarrhea that occurred spontaneously, and was associated with inflammation, primarily of the small intestine. VNS applied intermittently (10 Hz for 30 sec every 3 hours), substantially reduced diarrhea, inflammation monitored by total bowel FDG-PET and plasma CRP. In rats, VNS also significantly reduced diarrhea provoked by trinitrobenzene sulphonic acid (TNBS) treatment¹³ and improved stool quality.¹⁴ In addition, there is evidence that VNS enhances mucosal integrity in the colon,³⁹ which could contribute to diarrhea reduction, by reducing leakiness and promoting the generation of new enterocytes that contribute to re-absorption. The diarrhea in affected NHPs was watery, which suggests increased fluid secretion that exceeded re-absorptive capacity in the colon. Water and electrolyte transport across the lining epithelium of the small and large intestine is controlled through enteric secretomotor reflexes that are under inhibitory regulation from sympathetic neurons.^40–42 A major purpose of this sympathetic neural control is to protect whole body water and electrolyte balance, the need for which is illustrated by the life-threatening effects of pathogens, such as rotavirus, cholera, and pathogenic Escherichia coli that disrupt the neural control of fluid exchange.⁸ We have yet to investigate the mechanism through which VNS reduced watery diarrhea in the affected macaques. It could be through efferent vagal pathways that are deduced to have protective effects on mucosal integrity and inflammatory pathways in the colon,^{39, 43} or through activation of the previously described cholinergic anti-inflammatory pathway (CAIP), or other non-CAIP neuroimmune circuits⁴⁴. For example, there is convincing evidence that VNS activates vagal afferents that through central reflex pathways activate sympathetic noradrenergic neurons that exert anti-inflammatory effects.^{45, 46}

4.5. Limitations of these investigations

This is a preliminary study in which the numbers of rhesus macaques available for detailed study, and in particular the numbers that could be implanted with vagal nerve stimulators, was limited. Only one set of stimulus parameters was used, and future studies should investigate different parameters, including different stimulus frequencies, amplitudes and pulse durations, and different duty cycles. The study also leaves unresolved whether the anti-inflammatory effects were due to activation of vagal efferent or afferent pathways, or perhaps involved both.

Conclusions.

These studies indicate that long-standing chronic diarrhea in rhesus macaques (idiopathic chronic diarrhea) is associated with histopathological changes, inflammation in the upper abdominal gastrointestinal tract, and elevated plasma CRP. The condition is relieved by VNS, indicating that engagement of the neural mechanisms that influence colonic fluid transport may provide a treatment for chronic watery diarrhea. However, before this possibility is realized further investigation of mechanism needs to be undertaken.

Key Points:

• Idiopathic chronic diarrhea in non-human primates is an under-utilized model of spontaneously occurring diarrheal disease.

• Animals with ICD have intestinal inflammation indicated by independent measures including FDG uptake, T cell localization, histopathology, and serum C reactive protein.

• Cervical vagal nerve stimulation (VNS) attenuates diarrhea and reduces inflammation in afflicted animals.

% Appendix_1_. PET_analysis_script.m

% 20230522
% Written by Alan McMillan
% Dept. of Radiology
% School of Medicine and Public Health
% U of Wisconsin-Madison
% abmcmillan@wisc.edu
PET_file = ‘PET_13082_postVNS.nii.gz’;
label_file = ‘13082_postVNS_segmentation.nii.gz’;
% load PET and labels
PET = niftiread(PET_file);
PET_info = niftiinfo(PET_file);
labels = niftiread(label_file);
% need to add a static correction for each subject’s scan to convert from
% PET units (Bq/cc) into SUV units
% https://www.radiantviewer.com/dicom-viewer-forum/how-is-suv-calculated-from-dicom-pet-ct-image-files/946/
% D = Injected dose in millicuries
% W = Weight in kilograms
% T = Time from injection to start of F-18 PET scan in minutes
D = 4.15
W = 11.25
T = 90
PET_SUV = PET / (D*37000000 * 2^(-T/109.8)) * (W*1000);
% save SUV as Nifti file
PET_SUV_file = strrep(PET_file, ‘.nii’, ‘_SUV.nii’);
PET_SUV_info = PET_info;
PET_SUV_info.Datatype = ‘single’;
niftiwrite(PET_SUV,PET_SUV_file,PET_SUV_info);
% extract ROI data
bowel = PET_SUV( labels == 1 );
bicep = PET_SUV( labels == 2 );
% bowel signal
fprintf(‘SUV_max( bowel ) = %g\n’,max(bowel))
fprintf(‘SUV_min( bowel ) = %g\n’,min(bowel))
fprintf(‘SUV_mean( bowel ) = %g\n’,mean(bowel))
fprintf(‘SUV_std( bowel ) = %g\n’,std(bowel))
fprintf(‘\n’)
% bicep signal
fprintf(‘SUV_max( bicep ) = %g\n’,max(bicep))
fprintf(‘SUV_min( bicep ) = %g\n’,min(bicep))
fprintf(‘SUV_mean( bicep ) = %g\n’,mean(bicep))
fprintf(‘SUV_std( bicep ) = %g\n’,std(bicep))
fprintf(‘\n’)
% bowel:bicep ratio
fprintf(‘bowel to bicep ratio = %g\n’,mean(bowel)/mean(bicep));
fprintf(‘\n’)
% print all as CSV lines
fprintf(‘%g,%g,%g,%g,%g,%g,%g,%g,%g\n’,…
        max(bowel),min(bowel),mean(bowel),std(bowel), …
        max(bicep),min(bicep),mean(bicep),std(bicep), …
        mean(bowel)/mean(bicep));

DATA AVAILABILITY STATEMENT

Data is included in the paper or is available from the first author (LCP) on reasonable request.