Aberrant Pain Phenotypes Emerge Following Prenatal Hypoxic-Ischemic Injury in a Rabbit Model of Cerebral Palsy

Abstract

Cerebral Palsy (CP) is the most common motor disability in childhood, and the most frequent comorbidity is pain. Rabbit kits subjected to prenatal hypoxia-ischemia (HI) exhibit allodynia and an expansion of nociceptive afferents in the lumbar spinal cord at postnatal day (P5). In this study, we examined how HI alters the development of multiple sensory modalities and its effect on psychosocial measures and C-fiber distribution in the spinal cord. To do this, we performed an HI surgery to occlude blood flow to fetal New Zealand White rabbits for 40 minutes, or a sham surgery. We performed von Frey, Hargreaves, and a cold allodynia test at P1, P5, P11, and P18. Additionally, we performed open field, a two-texture preference test, and immunofluorescence assays at P18. HI kits exhibit altered development and allodynia in von Frey and Hargreaves and minor decreased sensitivity in cold allodynia. HI kits spend less time on the aversive side of the two-texture preference apparatus and more time in the center of an open field but a higher ratio of that time immobile. This is accompanied by changes in the distribution of C-fibers in the dorsal horn of the cervical and lumbar spinal cord. A principal components analysis revealed altered nociception and psychosocial changes are important for differentiating between control and HI kits but not distribution of C-fibers. Overall, HI rabbits kits exhibit altered sensory development, allodynia, anxiety-like behavior, and changes to the distribution of nociceptive afferents in the dorsal horn of the spinal cord.

Introduction

Cerebral palsy (CP) is an umbrella term for non-progressive movement and postural disorders arising from perinatal injury to the nervous system [19]. CP is the most common physical disability in childhood, with a prevalence of approximately 1–4 per 1000 live births [28; 37], but the lived experience of people with CP extends beyond motor dysfunction [19]. Pain is a major health issue in CP, strongly associated with reduced quality of life [29; 52; 66; 67]. Up to 76% of children and young people with CP experience pain, and about one in three meet criteria for chronic pain [31; 44; 45]. Children with CP report that pain interferes with sleep and their ability to have fun [44] and are over 2.5 times more likely to report psychological symptoms with severe pain [55]. This burden does not diminish with age. A recent meta-analysis of over 1,200 adults with CP reported pain prevalence at 70% [66].

The mechanisms underlying CP-related pain are heterogeneous and incompletely understood. Converging evidence indicates the etiology of chronic pain in CP extends beyond musculoskeletal pathology. Adults with CP exhibit higher prevalence of all pain phenotypes compared to the general population, and the co-occurrence of two or more pain types is much more common [47]. A subset of individuals with CP report neuropathic qualities such as burning or stabbing sensations, and spinothalamic sensory abnormalities correlate with greater pain intensity, implicating central pain mechanisms [17]. Quantitative sensory testing in children with CP reveals patterns of mechanical and thermal hypoesthesia and mechanical hyperalgesia, suggesting that CNS injury disrupts both discriminative and nociceptive somatosensory processing [8; 51]. This signals that the injured nervous system itself is an active contributor to pain experience, not merely a passive substrate [7; 47].

Despite this, the pathological mechanisms driving altered pain and sensory processing remain poorly characterized. This knowledge gap is consequential: early developmental windows represent critical periods during which aberrant neural connectivity may be amenable to targeted intervention [7; 42]. Progress requires experimental animal models that reproduce the sensory and pain phenotypes in addition to the motor features of CP. The established rabbit model of prenatal hypoxia-ischemia (HI) reliably recapitulates the movement disorder aspects of CP [21; 22; 24; 63], and our group has shown that neonatal HI rabbits exhibit mechanical allodynia and thermal hyperalgesia accompanied by expansion of calcitonin gene related peptide (CGRP)-positive C-fibers in the lumbar spinal cord [49]. Whether this model captures the multimodal sensory dysfunction characteristic of human CP, and whether these changes evolve across the neonatal period in a manner consistent with clinical trajectories has not been systematically examined.

The purpose of this study was to characterize sensory responses across multiple modalities throughout postnatal development in HI rabbits, and to determine whether behavioral changes are associated with reorganization of nociceptive fibers in the dorsal horn at postnatal day (P)18. We hypothesized that HI rabbits would exhibit alterations in nociception and psychosocial behavior corresponding to maladaptive changes in afferent distribution within the spinal cord. We found that HI rabbits exhibited altered development, hypertonia, altered nociception, anxiety-like behavior, and altered nociceptive afferent area in the dorsal horn.

Methods

Subjects.

All experiments were performed in accordance with the United States’ National Institutes of Health’s Guide for Care and Use of Laboratory Animals and approved by the University of Rhode Island’s Institutional Animal Care and Use Committee. Male and female New Zealand White rabbits were used in all experimental groups, and all rabbits were born in-house at the University of Rhode Island. The control group comprised a mix of naïve kits that underwent no surgery (n=12) and kits that underwent sham surgery in utero (n=13) which included sedation but no hypoxia-ischemia. Naïve control and sham kits came from 8 litters and were collapsed into a single control group, as separate statistical analyses on each behavioral test revealed inconsequential differences between these conditions, and our primary interest was to examine the effects of the HI surgery compared to uninjured conditions. A comparison of the sham and naïve controls is included in supplemental table 1 and supplemental table 2. The experimental hypoxia-ischemia (HI) group had 24 kits from 3 litters. Two dams provided multiple litters; one dam provided both a sham and an HI litter, and the other provided 1 control litter and 2 sham litters. Kits were monitored at least daily for hydration and visible milk in the stomach. If these factors were unsatisfactory, kits were orally administered supplemental rabbit milk replacer (Wombaroo Food Products, Mount Barker, South Australia). Some kits received this milk replacer as a consequence of litter size and/or failure to thrive. There were 2 stillborn control kits, 2 stillborn sham kits, and 0 stillborn HI kits. Kits were counted as still born ifthey were dead upon birth or if they died less than 24 hours after birth. Two HI kits died before the 18-day testing period was completed for unknown reasons. All rabbits were housed on a 12h light/dark cycle in a temperature and humidity-controlled room in standard Allentown rabbit enclosures. Dams were provided a nest box lined with hay before their due date. Water, chow (Envigo 2031; Inotiv, Inc., Lafayette, IN), and hay were available ad libitum in the home cages.

Hypoxia-ischemia surgery.

Pregnant New Zealand White dams at E22/E23 (~70% gestation) underwent HI procedures as previously described [22], including sedation with intramuscular administration of ketamine/buprenorphine/dexmedetomidine, (5/0.05/0.025 mg/kg, respectively) or ketamine/xylazine (25/1.5 mg/kg). They were then administered 5% or less isoflurane in O₂. A Fogarty arterial embolectomy catheter (Edwards Lifesciences Cat # 120602FP or 120403FP) was inserted into the femoral artery up to the level of the bifurcation of the descending aorta. Catheter placement at the bifurcation was confirmed via ultrasonography. The balloon catheter was inflated with saline under continuous Doppler monitoring until blood flow was blocked. This occluded blood flow to the uterine horns and the rabbit fetuses. After 40 minutes, the balloon was deflated and removed. The femoral artery was tied off and the incision was closed. Dams recovered and were returned to their home cages. For sham surgeries, pregnant dams were placed under anesthesia without artery occlusion. Importantly, all dams that underwent surgery recovered and gave birth normally on the expected due date +/− one day.

Behavioral Testing.

All kits underwent behavioral testing at P1, P5, P11, and P18. Care was taken to follow testing order that began with basic weight and length measurements, followed by locomotion/anxiety in the open field and reflexive movement with the righting reflex test. Then, mechanical and hot temperature sensitivities were assessed using von Frey and Hargreaves’, respectively. Hypertonia was determined with the modified Ashworth scale, further modified for rabbits[22]. Afterwards, non-evoked mechanical sensitivity was assessed using the two-texture preference test and cold temperature sensitivity was assessed using a cold allodynia test. von Frey, Hargreaves’, and the cold allodynia test were performed on all paws. The order of paws being tested was randomly determined by the experimenter. For von Frey, Hargreaves’, and cold allodynia tests, kits were allowed to acclimate to each of these testing environments for 5–10 minutes until they were quiet and still but not sleeping. Experimenters remained consistent throughout the behavioral testing per litter to limit inter-rater variability. Detailed descriptions of behavioral assessments are outlined below.

Open Field Test.

Behavior in an open field was tested at P18 as they reliably have their eyes open by this age[34]. This test was performed in a novel environment, the experimenter would leave the room, and anxiety-related measures were specifically collected[54]. Kits were placed in the center of a large, square open field (1m × 1m × 0.35m) with a soft foam mat lining the floor. Kits freely explored for 5 minutes. The test was recorded with an AUKEY PC-LM3 camera (AUKEY, Shenzhen, China) and analyzed using ANY-Maze software (Stoelting Co., Wood Dale, IL). For video analysis of anxiety-like behavior, the open field was divided into a surround zone which accounts for 68% of the apparatus and a center zone which accounts for 32% of the apparatus using ANY-maze (Stoelting Co., Wood Dale, IL). Signs of anxiety were measured using the amount of time kits spent immobile or in the center of the open field [54; 57]. Immobility detection was set to 65% and the minimum immobility period was set to 2000 ms. We examined measures associated with anxiety-like behavior (time spent, distance traveled, and immobility in the center).

Righting Reflex.

Reflex development was tested using an adapted righting reflex test and was recorded with an AUKEY PC-LM3 camera (AUKEY, Shenzhen, China). Kits were held supine in a box (45cm × 30.5 cm for P1 and P5 kits, 51 cm × 51 × 30.5 cm for P11 kits, and 1m × 1m × 35 for P18 kits) until experimenters let go of the kits and let the kits right themselves to a prone position. This test was repeated 10 times with 10 seconds between each trial and times for each kit were averaged. Post-hoc scoring was conducted in a blinded manner to determine righting time and behavioral differences in righting such as the kit struggling to right itself. Trials were excluded if the video skipped, if the kit was not secured in a supine position before the trial, or if the kit fell asleep during the trial.

von Frey Test.

Mechanical sensitivity was tested using the von Frey test. Testing methodology was adapted from a previous publication from our lab [49]. Rabbits were placed in a chamber with a plastic mesh gridded (8 mm × 8 mm) bottom that allowed access to the plantar surface of the paws of the rabbits. Starting with the 1g filament, a von Frey filament (Stoelting Cat # 58011) was applied perpendicular to the plantar surface of the paw in the ascending method. The filament was applied 5 times to each paw with a 30 second pause between each application of the filament [49]. The number of paw withdrawal reflexes was recorded as well as subsequent supraspinal responses (e.g. licking or looking at the paw) [16; 69]. The experimenter applies the next weaker filament when responses to 1g are noted, or the experimenter applies the next stronger filament until the paw responds to at least 3 of the 5 applications or is non-responsive to the 26g filament. The paw withdrawal threshold is considered the lowest von Frey filament that elicits a paw withdrawal response in at least 3 of the 5 applications.

Hargreaves’ Test.

Heat sensitivity was tested using the Hargreaves’ test. Testing methodology was adapted from a previous publication from our lab [49]. Using the Hargreaves apparatus (Model 336G Plantar/Tail Stimulator Analgesia Meter from IITC Life Science, Woodland Hills, CA), the rabbits were placed in a chamber on top of a tempered glass plate. From below, a 4×6 mm noxious radiant light beam was applied to the plantar surface of the paw. The radiant heat increased gradually from rest (20–30 °C) to a maximum of 83 °C at 8 seconds. The paw withdrawal latency was recorded as well as supraspinal responses (e.g. licking or looking at the paw). Three trials were collected for each paw and were averaged to yield one score per paw. In addition, the total number of paws tested where the rabbit kit displayed a supraspinal response was quantified (e.g. licking or looking at the paw) [16; 69]. The light beam was automatically shut off at 8 seconds (~80°C). If the kit did not withdraw its paw within 8 seconds, the paw was labeled as nonresponsive for that trial and was given a score of 9 seconds. A paw was considered “responsive”, if the paw withdrawal latency was < 9 seconds for at least one of its three trials. Hargreaves’ testing was only performed at P1, P5, and P11 because kits’ paws are insulated by fur at P18 and in our experience they no longer withdraw from the heat stimulus within 8s.

Ashworth Test.

Categorization of motor deficits was performed using a modified Ashworth scale that was further modified for rabbits as previously described [22]. Briefly, joints in the fore- and hindlimbs were rotated, and their tone was evaluated separately by two experimenters. Kits were considered to be “motor affected” if the muscle tone score was ≥3 in one or more evaluated joint(s) according to both experimenters. After prenatal HI, some kits exhibit hypertonia, while others are unaffected [22].

Two-Texture Preference Test for Non-Evoked Mechanical Sensitivity.

Non-evoked mechanical sensitivity was tested with an adapted two-texture preference test[33]. Rabbits were placed into a square open field apparatus (1m × 1m × 0.35m) with clear plastic walls. Then, the experimenter would leave the room. The floor was half covered by a soft foam mat (same as what was used for open field) and half covered by outdoor turf. This turf is a rough, potentially aversive texture. The rabbit kit was placed in the center of the open field apparatus and allowed to explore for 5 minutes while being recorded using the same setup as above. From ANY-maze, we extracted measures related to texture preference (time, distance traveled, and immobility on each side).

“von Freeze” Test for Cold Sensitivity.

We designed a new test to assess sensitivity to cold stimuli [11] inspired by the cold plantar assay[12; 13]. In this test, we placed rabbits in a chamber with a plastic mesh gridded (8 mm × 8 mm) bottom that allowed access to the plantar surface of the paws. Once the rabbit was still, a micro-popsicle was gently applied to one of the paws. The application of the micro-popsicle was enough to contact the skin, but not enough to lift or poke the paw. The paw’s withdrawal latency was recorded as well as supraspinal responses (e.g. licking or looking at the paw)[16; 69]. In addition, the total number of paws tested where the rabbit kit displayed a supraspinal response was quantified. The paw was labeled nonresponsive if it did not withdraw within 20 seconds and was given a score of 21 seconds.

Analysis of Nociceptor Distribution via Immunofluorescence.

P18 rabbits were euthanized with intraperitoneal injection of ~1mL/kg Euthasol solution (Covetrus, Portland, ME) before tissue harvest at P18. Transcardial perfusion was performed using chilled 0.01 M phosphate buffered saline (PBS; pH 7.4) until tissue blanching was observed. 4% paraformaldehyde (PFA) in PBS was then infused at a constant flow rate of 10–12 mL/min using a Gilson Minipuls 3 perfusion pump (Gilson Inc., Middleton, WI). The perfused spinal cords were dissected, post-fixed in PFA at 4°C overnight, washed in 0.01M PBS, and then submersed in 30% sucrose in PBS for cryoprotection. C7–C8 and L3–L5 blocks were isolated from the rest of the spinal cord, as they contain sensory afferents that innervate the plantar surface of the fore- and hindpaw, respectively [50; 62]. These blocks were frozen at −80°C until sectioning. For sectioning, spinal cord blocks were embedded in OCT compound (Fisher Scientific, Pittsburgh, PA). Transverse sections were sliced at −20°C using a cryostat (Leica Microsystems, Wetzlar, Germany) and collected in serial at 25 μm thickness. To avoid repeat regions of analysis, every 10^th section was used. Sections were mounted on Superfrost plus microscope slides (Thermo Fisher Scientific, Waltham, MA) and stored at −20°C until they were stained.

Sections were labeled with isolectin B₄ (IB4) and an antibody to detect CGRP to identify non-peptidergic and peptidergic nociceptive afferent fibers, respectively [18]. For staining, sections were dried at 37°C for several hours to improve section adherence to the slide. For antigen retrieval, sections were treated with Dako Antigen Retrieval Solution (Agilent Pathology Solutions, Santa Clara, CA) in a vegetable steamer at 95°C for 10 minutes. Sections were rinsed in 0.1 M oxidized phosphate-buffered saline (OX-PBS; 0.1 M PBS with 0.1 g/L Thimerosal) twice for 5 minutes each and then incubated in a blocking solution (10% normal goat serum, 0.2% Triton X-100 in OX-PBS) for 1 hour at room temperature. This was followed by an incubation in rabbit-to-rabbit blocking reagent (ScyTek Laboratories, West Logan, UT) for 1 hour at room temperature. Sections were then incubated with 1:500 anti-α-calcitonin gene-related peptide rabbit primary antibody (Peninsula Laboratories Cat# T-4032, RRID:AB 518147; BMA Biomedicals, Basel, Switzerland) and Lectin from Bandeiraea simplicifolia Biotin Conjugate (Sigma-Aldrich cat. L2140; MilliporeSigma, Burlington, MA) in a solution of 5% normal goat serum and 0.2% Triton X-100 in OX-PBS for 48 hours at room temperature. After three 5-minute washes in OX-PBS, sections were incubated in goat anti-rabbit cross-adsorbed secondary antibody conjugated for Alexa Fluor 488 (Molecular Probes Cat# A-11008, RRID:AB_143165; Thermo Fisher Scientific, Waltham, MA) and streptavidin conjugated for Alexa Fluor 594 (Jackson ImmunoResearch cat. 016-580-084; Jackson ImmunoResearch Labs, West Grove, PA) for 2 hours at room temperature. After three more 5-minute washes in OX-PBS, sections were cover slipped (Corning Cat No. 2990–245, Corning, NY) using DAPI Flouromount-G (Southern Biotech, Birmingham, AL) mounting medium. After cover slipping, slides were stored at room temperature in microscope slide boxes.

Stained sections were imaged at 20x (Leica DM6 B microscope; Leica Microsystems, Wetzlar, Germany). Depending on tissue availability, two to three representative images of the right and left dorsal horn of each animal were imaged. The proportional area of CGRP and IB4 were calculated within the cap of the dorsal horn (laminae I-II) using a moment-preserving thresholding algorithm [65] available through the Auto Threshold plugin on ImageJ (NIH, Bethesda, MD). The proportional area of colocalized CGRP- and IB4-positive area was determined using the Colocalization plugin on ImageJ (NIH, Bethesda, MD). The proportional area of CGRP in the deep dorsal horn (lamina III) was calculated with the same method. Representative images for publication were imaged at 40x with a Stellaris 5 confocal (Leica Microsystems, Wetzlar, Germany). Samples were excluded from analysis if the tissue could not be recovered for staining. 3 HI kits were excluded from cervical analysis, and 3 HI, 3 control, and 1 sham kit were excluded from lumbar analysis. Values from the analyzed sections of each dorsal horn were averaged and the averages are represented.

Data Analysis and Statistics.

All graphical analysis and statistical analysis for weight, length, two texture preference, open field, and immunofluorescence was performed using GraphPad Prism 10.6.1. Data analyzed using GraphPad Prism at multiple timepoints (weight, and length) were analyzed using a two-way repeated measures ANOVA model with a Geisser-Greenhouse correction. Tukey’s multiple comparisons test was used to analyze the difference between control and HI kits at each timepoint. Data analyzed at singular timepoints (two-texture preference test, open field test anxiety measures, immunofluorescence, and principal components) was assessed for normality using QQ plots and the Shapiro-Wilk test. Data that passed normality was analyzed with a Welch’s t-test. Data that did not pass normality was analyzed with a non-parametric Mann-Whitney test.

Disparities in behavioral measurements from each limb or side of the spinal cord occur in people with CP [15] as well as in our model [22]. Thus, data acquired from each paw or dorsal horn was treated as an individual data point rather than averaged to generate a single value for each kit. von Frey, Hargreaves’, and von Freeze were analyzed with the RStudio (version 2023.06.0+421) packages lme4, lmerTest, and emmeans to allow for clearer interpretation of changes over time in our longitudinal models. We performed linear mixed effects analyses with fixed effects for age and group and a random intercept for individual paws of rabbits to account for the repeated measurements on each paw from P1 to P18. In some cases, visual analysis of graphed means warranted testing linear spline mixed effects models with knots at specific time points. In these cases, initial models to test model fit were performed using maximum likelihood estimation. Models were compared by running an ANOVA and examining AIC values. If models were significantly different, the model with the lowest AIC value was chosen. When there was no statistically significant difference between the models, the simpler linear mixed effects model without the term to account for the change in slope was chosen. Linear spline mixed effects models were used in the von Frey forepaws model, with a knot (the inflection point) at P11 and the Hargreaves hindpaws model, with a knot at P5. In these cases, an additional age term was added to account for the change in slope after the knot. After choosing the best model, analyses were performed using restricted maximum likelihood estimation. From these models, we could extract the change in response for each one unit increase in time, the effect of HI on responses at P1, and the effect of HI on the change in response over time. To assess group differences, estimated marginal means were calculated and pairwise comparisons between groups at each timepoint were performed with Bonferroni’s adjustment to the p-values[30].

To explore the relationships amongst the immunofluorescence and behavioral data, principal component analysis (PCA) was performed using the R (version 4.5.2) package syndRomics [64]. The dataset used for PCA included 1) von Frey, Hargreaves’ and von Freeze pain scores at P18 with all paws averaged per animal, 2) Deep and superficial dorsal horn CGRP+ fiber area, IB4+ fiber area, and colocalization area of CGRP+ and IB4+ fibers, with all paws averaged per animal, 3) the ratio of time spent and immobility between the center and surround in the open field test and 4) the ratio of time spent and immobility on the aversive side versus the non-aversive side of the two-texture preference assay. Data was imputed and then standardized so that the mean was 0 and the standard deviation was 1 to ensure that each variable contributed equally to the analysis. Group differences in PC scores were analyzed using Welch’s t test.

Results

Prenatal HI causes reduced length and weight and delayed eye opening

Rabbits were measured for length and weight at each timepoint (Figure 1). At P1 and P5, there was no significant difference in weight or length. However, at P11, HI and control kits started to differ. HI kits weighed less (p=0.0002). Smaller size and weight has been described in this model[22] and for rabbits from larger litters[48]. In addition, at P11, we checked to see if the kits’ eyes were open, as this is approximately the age for this developmental milestone in our experience and others have reported similarly[34]. A majority of control kits (63%) had at least one eye open, while only 36% of HI kits had at least one eye open (Figure 1C). At P18, control kits weighed more (p<0.0001) and were longer (p=0.0022) than HI kits (Figure 1A–B). All kits had their eyes open at P18. Overall, HI kits exhibited delayed developmental milestones.

Figure 1: HI rabbit kits exhibit developmental differences compared to control kits.

(A) Average weights of kits plotted across time. HI kits were lighter compared to control kits at P11 (p=0.0002) and P18 (p<0.0001) but not P1 (p=0.6729) or P5 (p=0.1854). (B) Average lengths of kits plotted across time. HI kits were shorter at P18 compared to control kits (p=0.0022), but there was no difference in lengths at P1 (p=0.6888), P5 (p=0.1624), or P11 (p=0.4637) between HI and control kits. (C) The percentage of kits that had 0, 1, or 2 eyes open at P11. Statistics: Control N = 25, HI N = 24. (A) 2-way ANOVA (Group: p<0.0001, F=22.98, Time: p<0.0001, F=437.8, Group × Time: p<0.0001, F=22.47). (B) 2-way ANOVA (Group: p=0.0422, F=4.363, Age: p<0.0001, F=794.1, Group × Age: p=0.0024, F=5.672). **p<0.01, ***p<0.001, ****p<0.0001.

Prenatal HI caused motor deficits

Motor deficits were evaluated in rabbits at each timepoint. Hypertonia was assessed with Ashworth scoring tailored to rabbits [22]. Of the rabbits exposed to prenatal HI, 54% were considered motor affected (Figure 2). This demonstrates the HI surgery generated motor deficits consistent with previous studies [22; 56].

Figure 2: HI rabbit kits exhibit hypertonia but no changes to righting reflex compared to controls.

(A) Percentage of kits that were motor affected. (B) Average righting reflex latency over time. Control and HI kits exhibited similar righting times at P1 (p=0.1127), P5 (p=0.8985), P11 (p=0.1644), and P18 (p=0.0795). (C) The percentage of trials that were considered successful (without struggle) at each time point in control and HI kits. Statistics: Control N = 25, HI N = 24, (B) Mixed-effects analysis (Group: p=0.0699, F=2.82, Age: p<0.0001, F=346.7, Group × Age: p=0.2642, F=1.294). *p<0.05, **p<0.01, ****p<0.0001.

Reflex development was assessed with a righting reflex test at each timepoint. A normal righting reflex is observed to be a single, smooth, continuous rolling motion to right their body from supine to prone position. The head typically turns first and the body follows. Kits were timed in their ability to right themselves to prone from a supine position. Control and HI kits took the same amount of time to right themselves at P1 (p=0.1127), P5 (p=0.8985), P11 (p=0.1644), and P18 (p=0.0795) (Figure 2B). Importantly, at early postnatal timepoints, the righting movement was not a smooth movement but often exhibited a rocking motion before rolling over. The movement became more fluid over time in both control and HI kits. At P1, many HI kits struggled to right themselves (58%), and even more control kits struggled (88%) (Figure 2C). At P5, fewer control and HI kits struggled (44% and 50%, respectively). At P11, even fewer control kits struggled (12.5%), though the difference between P5 and P11 was less pronounced in HI kits (50% vs 33%). At P18, none of the control or HI kits struggled to right themselves. These results indicate a similar righting reflex development over time in control and HI kits.

Prenatal HI causes mechanical allodynia and alters mechanical nociceptive development

After prenatal HI, rabbits displayed aberrant development as well as mechanical allodynia compared to uninjured control kits. At P1, HI kits have a lower average paw withdrawal threshold than control kits in the forepaws (P<0.0001) and hindpaws (p<0.0001), indicating mechanical allodynia (Figure 3). This is partially due to control kits being generally insensitive to von Frey filaments; only 56% of forepaws (Figure 3B) and 56% of hindpaws (Figure 3E) are responsive to any filament applied at P1. In contrast, 96% and 92% of fore- and hindpaws in HI kits are responsive, respectively (Figures 3B, E). In addition, control kits exhibited supraspinal responses to mechanical stimulation in fewer forepaws as compared to HI kits (48% vs 71%; Figure 3C), indicating increased perception of the stimulus in HI kits. In contrast, control and HI kits both exhibited supraspinal responses in a relatively high percentage of hindpaws (76% vs 90%; Figure 3F), indicating relatively higher cortical awareness of hindpaw mechanical stimulation compared to forepaw stimulation. These results indicate that HI kits are more sensitive to mechanical stimulation and are more often aware of forepaw stimulation at P1 compared to control kits.

Figure 3: HI rabbit kits are hypersensitive to mechanical stimuli at P1, P5, and P11 in the von Frey test.

(A) Average forepaw withdrawal thresholds plotted across time. HI kits exhibited lower forepaw withdrawal thresholds compared to control kits at P1 (p<0.0001) and P5 (p<0.0001), but not P11 (p=0.4192) or P18 (p>0.9999). (B) Percentage of forepaws that are considered responders (withdrew at least 3/5 times at 26g or before) or nonresponders (did not withdraw at least 3/5 times during testing) at each time point in control and HI kits. (C) The percentage of forepaws that have at least one supraspinal response during testing plotted across time. (D) Average hindpaw withdrawal thresholds plotted across time. HI kits exhibited lower hindpaw withdrawal thresholds compared to control kits at P1 (p<0.0001), P5 (p<0.0001), and P11 (p=0.0.0046) but not P18 (p=0.1516). (E) Percentage of hindpaws that are considered responders or nonresponders at each time point in control and HI kits. (F) The percentage of hindpaws that have at least one supraspinal response plotted across time. Statistics: Control N = 50 paws, HI N = 48 paws, (A) Linear spline mixed effects model (Group: p<0.0001, β=−20.02, Age: p<0.0001, β =−2.31, Age after P11: p=0.0042, β=2.03, Group × Age: p<0.0001, β =2.52, Group × Age after P11: p=0.0003, β=−3.64). (D) Linear mixed effects model (Group: p<0.0001, β=−18.42, Age: p<0.0001, β=−1.96, Group × Age: p<0.0001, β=1.33). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

At P5, HI kits still had a lower paw withdrawal threshold than control kits in the forepaw (p<0.0001) and hindpaws (p<0.0001), indicating continued mechanical allodynia in HI kits (Figure 3). Control and HI kits exhibited supraspinal responses in a similar number of forepaws (44% vs 40%; Figure 3C) and hindpaws (84% vs 79%; Figure 3F), indicating similar perception of the mechanical stimuli at P5, despite differences in mechanical sensitivity (Figure 3A,D). Compared to P1, control kits were more responsive at P5 in the forepaws (56% vs 74% responders) and the hindpaws (56% vs 76% responders) (Figure 3B,E). HI kits had similar levels of responsiveness at P1 vs P5 in the forepaws (96% vs 88%) and hindpaws (92% vs 89.58%) (Figure 3B,E).

At P11, HI kits had significantly lower average paw withdrawal thresholds than sham kits in the hindpaws (p=0.0046) but not forepaws (p=0.4192) (Figure 3A,D). At P11, control and HI kits were generally responsive to the von Frey filaments in the forepaws (98% vs 90%; Figure 3B) and in the hindpaws (88% vs 94%; Figure 3E). In addition, compared to P5, kits exhibited supraspinal responses in more forepaws in both control (44% vs 76%) and HI (40% vs 79%) kits (Figure 3C). In contrast, compared to P5, kits exhibited supraspinal responses in a similar percentage of hindpaws in control (84% vs 82%) but not HI (79% vs 94%) kits (Figure 3F). These results indicate increased mechanical sensitivity in HI kits, but similar mechanical awareness in control and HI kits at P11.

At P18, there were no differences in withdrawal thresholds between the groups in the forepaws (p=1.0000) and hindpaw (p=0.1516) (Figure 3A,D). Control and HI kits were generally responsive to the von Frey filaments in the forepaws (96% vs 98%; Figure 3B) and in the hindpaws (100% vs 96%; Figure 3E). In addition, compared to P11, kits exhibited supraspinal responses in more forepaws in both control (76% vs 96%) and HI (79% vs 94%) kits (Figure 3C,F). In contrast, compared to P11, kits exhibited supraspinal responses in a similar percentage of hindpaws in both control (82% vs 88%) and HI (94% vs 92%) kits (Figure 3F).

Importantly, the developmental trajectory of mechanical nociception is altered in HI kits. From P1 to P18, there is an overall increase in mechanical sensitivity in the forepaws of control kits. On average, there is a 2.31g decrease in paw withdrawal thresholds each day as the control kits age until they reach P11 (P<0.0001). After P11, there is a change in their developmental trajectory, and paw withdrawal thresholds only decrease by 0.28g each day until P18 (p=0.0042). HI kits begin development with an average paw withdrawal threshold 20.02g lower than controls (p<0.0001), around 12.26 g on average. As HI kits age, from P1 to P11, there is slight increase in paw withdrawal thresholds each day, only 0.21g/day (P <0.0001). After P11, HI kits have a 1.4g decrease in paw withdrawal thresholds each day (P=0.0003) until P18, which is steeper than controls. In the hindpaws, the results are similar. On average, from P1 to P18, control kits have a 1.96g/day decrease in their paw withdrawal thresholds (p<0.0001). HI kits are initially more sensitive in their hindpaws with an average paw withdrawal threshold at P1 that is 18.42g lower than controls (p<0.0001); however, the rate of decline in paw withdrawal thresholds in HI kits is slower, only a 0.63g/day (P<0.0001) decrease. These results indicate that typical rabbit nociceptive development involves an increase in mechanical sensitivity over time accompanied by variation of mechanical awareness in the forepaws. Comparatively, HI kits exhibit increased sensitivity, responsiveness, and awareness of mechanical stimuli at several timepoints and changes to their development of mechanical nociception. Overall, HI kits exhibit mechanical allodynia and accelerated mechanical nociceptive development as compared to control kits.

Prenatal HI causes heat-evoked thermal hyperalgesia and alters thermal nociceptive development

HI kits exhibit increased thermal sensitivity in response to heat as they age. At P1, there is no difference in paw withdrawal latency between control and HI kits in the forepaws (p=0.9805) (Figure 4A) and the hindpaws (p=1.0000) (Figure 4D). Control and HI kits are responsive to heat stimuli, with responsiveness in 98% of forepaws (Figure 4A) and 100% of hindpaws (Figure 4D) in both groups. Both control and HI kits exhibit few supraspinal responses in response to forepaw stimulation (14% and 21%, respectively) (Figure 4C). Both groups exhibit relatively more supraspinal responses in the hindpaws (42% and 60%, respectively) (Figure 4F).

Figure 4: HI rabbit kits are hypersensitive to heat stimuli in the Hargreaves’ test at P11.

(A) Average forepaw withdrawal latencies plotted across time. HI kits exhibited a lower forepaw withdrawal latency compared to control kits at P11 (p=0.0008) but not P1 (p=0.9805) or P5 (p=0.0544). (B) Percentage of forepaws that are considered responders (withdrew at 8s or before) or nonresponders (did not withdraw at 8s or before) at each time point in control and HI kits. (C) The percentage of forepaws that have at least one supraspinal response during testing plotted across time. (D) Average hindpaw withdrawal latencies plotted across time. HI kits exhibited a lower hindpaw withdrawal latency compared to control kits at P11 (p=0.0281), but not P1 (p>0.9999) or P5 (p=0.7604). (E) Percentage of hindpaws that are considered responders or nonresponders at each time point in control and HI kits. (F) The percentage of hindpaws that have at least one supraspinal response plotted across time. Statistics: Control N = 50 paws, HI N = 48 paws, (A) Linear mixed effects model (Group: p=0.9805, β=−0.006, Age: p=0.1303, β=−0.040, Group × Age: p=0.013, β=−0.10). (D) Linear spline mixed effects model (Group: p=0.3633, β=−0.26, Age: p=0.0003, β=−0.26, Age after P5: p<0.0001, β=0.64, Group × Age: p=0.1330, β=0.15, Group × Age after p5: p<0.0222, β=−0.33). *p<0.05, **p<0.01, ***p<0.001

At P5, there is no difference in paw withdrawal latency between control and HI kits at this timepoint in the forepaws (p=0.0544) (Figure 4A) or the hindpaws (p=0.7604) (Figure 4D). Both control and HI kits continue to be responsive to heat stimuli in both the forepaws (96% and 100% responsiveness, respectively) (Figure 4B) and in the hindpaws (100% responsiveness for both) (Figure 4E). Both control and HI kits continue to exhibit few forepaw supraspinal responses (8% and 16%, respectively) (Figure 4C), and relatively more hindpaw supraspinal responses (36% and 38%, respectively) (Figure 4F).

At P11, HI kits have a lower withdrawal latency in the forepaws (p=0.0008) (Figure 4A) and the hindpaws (p=0.0281) (Figure 4D). Both control and HI kits continue to be responsive to heat stimuli in the forepaw (98% and 88%, respectively) (Figure 4B), with few supraspinal responses (12% vs 8%, respectively) (Figure 4C). However, hindpaw responsiveness decreases in both control (100% vs 74%) and HI (100% vs 73%) kits between these ages (Figure 4E). In addition, hindpaw supraspinal responses decrease between P5 and P11 in both control (36% vs 16%) and HI kits (38% vs 25%) (Figure 4F).

HI kits exhibit altered development of thermal nociception. In the forepaws, the paw withdrawal latency of control kits does not change from P1 to P11 (p=0.1303). However, this is not true for HI kits. Although there is no difference in the control and HI kits’ average withdrawal latencies at P1 (p=0.9805), HI kits do have 0.14s/day decrease in their forepaw withdrawal latencies (p=0.013) from P1 to P11. In the hindpaws, the paw withdrawal latency of control kits decreases 0.26s/day between P1 and P5 (p=0.0003), and from P5 to P11, it increases 0.38s/day. There was no difference at P1 between HI and control kits (p=0.3633) and no difference in the age-related decrease (p=0.1330) from P1 to P5. However, from P5 to P11, HI kits have a smaller age-related increase compared to controls; HI kits’ withdrawal latencies increase 0.20s/day from P5 to P11. This indicates that HI kits develop increased heat sensitivity over time, have an altered developmental trajectory of thermal nociception, and develop heat-evoked thermal hyperalgesia by P11.

Prenatal HI has limited effects on cold sensitivity

To evaluate cold thermal sensitivity after HI, we used an adapted cold plantar assay [12; 13], dubbed “von Freeze”. HI kits exhibit limited alterations to thermal nociception in response to the cold stimulus as compared to control kits. There is no difference in paw withdrawal latency between control and HI kits at P1 in the forepaws (p=0.4942) (Figure 5A) or the hindpaws (p=0.8482) (Figure 5D). Control and HI kits both generally respond to the cold stimulus in the forepaws (96% for both) (Figure 5B) and in the hindpaws (82% and 92%, respectively) (Figure 5E). Both control and HI kits have few supraspinal responses to the cold stimulus at this age in the forepaws (10% and 8%, respectively) (Figure 5C) and the hindpaws (14% and 27%, respectively) (Figure 5F).

Figure 5: There are minor differences between controls and HI in response to cold stimuli in the von Freeze test.

(A) Average forepaw withdrawal latencies plotted across time. HI kits exhibited similar forepaw withdrawal latencies compared to control kits at P1 (p=0.4942), P5 (p=0.2260), and P11 (p=0.0683), and P18 (p=0.1085). (B) Percentage of forepaws that are considered responders (withdrew at 20s or before) or nonresponders (did not withdraw at 20s or before) at each time point in control and HI kits. (C) The percentage of forepaws that have at least one supraspinal response during testing plotted across time. (D) Average hindpaw withdrawal latencies plotted across time. HI kits exhibited longer hindpaw withdrawal latencies compared to control kits at P11 (p=0.0473) and P18 (p=0.0261), but not P1 (p=0.8482) and P5 (p=0.5591). (E) Percentage of hindpaws that are considered responders or nonresponders at each time point in control and HI kits. (F) The percentage of hindpaws that have at least one supraspinal response plotted across time. Statistics: Control N = 50 paws, HI N = 48 paws, (A) Linear mixed effects model (Group: p=0.4950, β=−0.73, Age: p<0.0001, β=0.29, Group × Age: p=0.490, β=0.−0.07). (D) Linear mixed effects model (Group: p=0.848, β=−0.23, Age: p=0.470, β=−0.06, Group × Age: p=0.107, β=0.19). *p<0.05

At P5, there is no difference in paw withdrawal latency between control and HI kits at P5 in the forepaws (p=0.3442) (Figure 5A) or the hindpaws (p=0.2792) (Figure 5D). Both control and HI kits generally continued to respond to the cold stimulus in the forepaws (94% and 96%, respectively) (Figure 5B) and the hindpaws (82% and 73%, respectively) (Figure 5E). Both control and HI kits continued to have few supraspinal responses in the forepaws (10% and 6%, respectively) (Figure 5C) and the hindpaws (24% and 13%, respectively; Figure 5F).

At P11, there is no difference in paw withdrawal latency between control and HI kits in the forepaws (p=0.0683) (Figure 5A), but HI kits have higher paw withdrawal latencies in their hindpaws (p=0.04773) (Figure 5D). Both control and HI kits generally continued to respond to the cold stimulus in the hindpaws (80% and 88%, respectively) (Figure 5E) and HI kits continued to respond in the forepaws (90%) (Figure 5B). In contrast, control kits had a reduction in responsiveness in the forepaws between P5 and P11 (94% vs 74%) (Figure 5B). Both control and HI kits had a greater amount of supraspinal responses in the forepaws (26% and 25%, respectively) (Figure 5C) and the hindpaws (38% and 25%, respectively) (Figure 5F) as compared to P5.

At P18, there is no difference in paw withdrawal latency between control and HI kits in the forepaws (p=0.1085) (Figure 5A); however, HI kits have a higher paw withdrawal latency in their hindpaws compared to controls (p=0.0261)(Figure 5D). Both control and HI kits generally continued to respond to the cold stimulus in the forepaws (68% and 75%, respectively) (Figure 5B) and the hindpaws (74% and 67%, respectively) (Figure 5E). Both control and HI kits continued to have some supraspinal responses in the forepaws (18% and 27%, respectively) (Figure 5C) and the hindpaws (28% and 15%, respectively) (Figure 5F).

When examining the developmental curves of cold sensitivity using von Freeze, there are no differences between control and HI. In the forepaws, control kits have a 0.29s increase/day (p<0.0001) in their withdrawal latency from P1 to P18. There are no differences at P1 between HI and controls (p=0.495), and there is no difference in the developmental curve in the forepaws between HI and control kits (p=0.490). In the hindpaws, the results are similar. There are no changes to hindpaw withdrawal latencies over time in the controls (P=0.470). In addition, there is no difference between control and HI hindpaw withdrawal latencies at P1 (p=0.848), and there is no difference in the age-related change of HI kits compared to controls (p=0.107). Together, these data indicate prenatal HI may lead to minor differences in cold sensitivity over time; however, this needs to be examined further.

Prenatal HI increases P18 non-evoked mechanical sensitivity

Non-evoked mechanical sensitivity was assessed via a modified two-texture preference assay at P18 [33]. Kits were allowed to explore a 1m × 1m open field apparatus, first with the floor covered completely with a smooth foam mat (these results are described in the following section), and then another test was performed where half of the floor was covered with the same soft foam mat, and the other half was covered with outdoor turf. The foam mat was considered to be “non-aversive,” as it was soft and smooth, while the outdoor turf was considered to be “aversive,” as it had a rough texture and caused mechanical stimuli similar to von Frey filaments.

When on the two-texture open field, both control and HI kits explored both sides (Figure 6). Control and HI kits traveled the same distance on the aversive side (p=0.5055) and the non-aversive side (p=0.0563), implying similar locomotion (Figure 6 C, D). However, HI kits spent less time on the aversive side (p=0.0002) and more time on the non-aversive side (p=0.0147) as compared to control kits, implying a relative preference for the non-aversive side (Figure 6 E, F). In addition, when compared as a ratio, the time on the aversive : time on the non-aversive side for control kits was an average of approximately 1:1, indicating no preference for either side (Figure 6G). In contrast, the same ratio in HI kits was approximately 1: 2 and was significantly different from controls (p=0.0008), indicating a strong preference for the non-aversive side (Figure 6G). Qualitatively, the majority of control rabbits readily explore the area freely with little hesitation. Controls appear unbothered by the aversive texture, spending, on average, close to half the time on the turf. Only a few control rabbits licked their paw or kicked their hindpaw(s) behind them when hopping, which may be inferred as nocifensive behavior. This is in stark contrast to the behavior of HI rabbits. The HI rabbits were more hesitant to stand on the turf. At the beginning of testing, many of the HI rabbits would approach the turf, place their forepaws on it, pause, and then change direction. HI rabbits would often, either while on the turf or right after being on the turf, display the nocifensive behaviors mentioned above (licking or kicking).

Figure 6: HI kits prefer the non-aversive texture during the two-texture preference test.

Representative paths and immobility episodes of (A) control and (B) HI kits in the two-texture open field apparatus. (C) Control and HI kits travel the same distance on the aversive side (t=0.6721, df=38.27, p=0.5055, Welch’s t test). (D) Control and HI kits travel the same distance on the non-aversive side (t=1.960, df=43.72, p=0.0563, Welch’s t test). (E) HI kits spend less time on the aversive texture as compared to control kits (t=4.11, df=46.63, p=0.0002, Welch’s t test). (F) HI kits spend more time on the non-aversive texture as compared to control kits (t=2.936, df=38.33, p=0.0056, Welch’s t test). (G) The ratio of time spent on the aversive : non-aversive side is decreased in HI kits as compared to control kits (U=137, p=0.0008, Mann-Whitney test). (H) HI kits spend a smaller percentageof their time on the aversive side immobile as compared to control kits (U=151, p=0.0024, Mann-Whitney test). (I) Control and HI kits spend the same percentage of their time on the non-aversive side immobile (U=207, p=0.0640, Mann-Whitney test). (J) The ratio of the percent of time spent immobile on the aversive : non-aversive side is the same in control and HI kits (t=0.1655, df=33.49, p=0.8696, Welch’s t test). Control N = 25 kits, HI N = 24 kits, *p<0.5, **p<0.01, ***p<0.001

In addition to time spent and distance traveled on each surface, the percentage of time spent immobile on each surface was measured to determine whether the kits avoid the increased tactile input provided by the astroturf surface. Avoidance of the rougher surface may imply mechanical allodynia (normally innocuous stimuli are perceived as painful). Compared to control kits, HI kits spent a smaller percentage of their time on the aversive side immobile (p=0.0024) (Figure 6H). In contrast, HI and control kits spent the same percentage of their time on the non-aversive side immobile (p=0.064) (Figure 6I). When compared as a ratio, the percent time immobile on the aversive side : the non-aversive side was approximately 1:1 in both control and HI kits, and there was no difference between the groups (p=0.8696) (Figure 6J). This indicates HI rabbits may respond to aversive textures in their environment by actively moving away from them. Overall, HI kits show a strong preference for the non-aversive side of the apparatus, indicating non-evoked mechanical sensitivity.

Prenatal HI increases context-dependent immobility in the open field at P18

Psychosocial components are an important aspect of the pain experience, and the link between chronic pain and some psychosocial components, specifically mood disorders, is bidirectional[40]. Negative affect and anxiety are often associated with persistent pain[1; 27; 40]. Anxiety components of the chronic pain experience can be inferred by comparing the amount of time the kit spends in the center zone to the time spent in the surround zone of an open field[54]. Anxiety can also be inferred by immobility in situations that animals may find frightening, such as the center of an open field[39]. To measure anxiety-like behaviors, P18 kits were allowed to explore a 1m × 1m open field. ANY-maze software (Stoelting Co., Wood Dale, IL) was used to divide the apparatus into a center zone and a surround zone (Figure). The kits’ time spent, immobility, and distance traveled in each zone was recorded with ANY-maze software.

When in the open field, HI and control kits locomoted similarly. Both control and HI kits traveled the same distance in the center zone (p=0.5963) (Figure 7C) and the surround zone (p=0.8622) (Figure 7D). HI kits spent more time in the center zone (p=0.0257) (Figure 7CE and less time in the surround zone (p=0.0257) (Figure 7F). The proportion of time spent in the center zone versus the time spent in the surround zone was greater in HI kits (p=0.0257) (Figure 7G). Because immobility is an indicator of anxiety and kits begin testing in the center of the open field, we examined immobility in the center and surround zones to determine if HI kits respond to being in the center by becoming immobile. We found HI and control kits spent the same percentage of time immobile in the center zone (p=0.1265) (Figure 7H) and surround zone (p=0.7578) (Figure 7I). However, the percentage of time immobile in the center : the percentage of time in the surrounding area was greater in HI kits as compared to control kits (p=0.0233) (Figure 7J) which could indicate a greater amount of anxiety-like behavior in HI kits.

Figure 7: HI kits spend more time in the center of the open field test.

Representative paths and immobility episodes of (A) control and (B) HI kits in the two-texture open field apparatus. (C) Control and HI kits travel the same distance in the center zone (t=0.5337, df=42.15, p=0.5963, Welch’s t test). (D) Control and HI kits travel the same distance in the surround zone (U=279, p=0.8622, Mann-Whitney test). (E) HI kits spent more time in the center zone than control kits (U=180, p=0.0257, Mann-Whitney test). (F) HI kits spent less time in the surround zone than control kits (U=180, p=0.0257, Mann-Whitney test). (G) The ratio of time spent in the center : surround zones was larger in HI kits than control kits (U=180, p=0.0257, Mann-Whitney test). (H) Control and HI kits spent the same percentage of their time in the center zone immobile (U=213.5, p=0.1265, Mann-Whitney test). (I) Control and HI kits spent the same percentage of their time in the surround zone immobile (t=0.3102, df=45.37, p=0.7578, Welch’s t test). (J) The ratio of time spent immobile in the center : surround zone was higher in HI kits as compared to control kits (t=2.349, df=44.67, p=0.0233, Welch’s t test). Control N = 24 kits, HI N = 24 kits, *p<0.05

Prenatal HI differentially alters nociceptive afferent density in the cervical and lumbar spinal cord

Nociceptive primary afferent fibers transmit pain and temperature information from the periphery. Therefore, we examined the effect of prenatal HI on nociceptive afferent distribution and density in the dorsal horn at P18. Figure 8A–A” shows representative images of CGRP+ and IB4+ fibers in the cervical dorsal horn of control kits. CGRP+ fibers terminate in lamina I and outer layer of lamina II, while IB4+ fibers terminate in the inner layer of lamina II in the dorsal horn of the spinal cord, with little overlap [10; 23]. Interestingly, CGRP+ nociceptive afferents also terminated in the deep dorsal horn in the control kits (Figure 8B–B”). Previously, we found that CGRP+ fibers terminated in lamina III, and only after prenatal HI was there a more substantial presence in the lumbar deep dorsal horn in P5 kits [49]. This differential nociceptive afferent distribution in control kits at P18 suggests postnatal distribution of nociceptive primary afferent fibers in rabbits.

Figure 8: HI rabbit kits have decreased density of CGRP+ nociceptive primary afferents in the C7–8 dorsal horn at P18.

Representative images of the cervical dorsal horn labeled with calcitonin gene-regulated peptide (CGRP; A, B) or labeled with isolectin B-4 (IB4; A’, B’). The proportional area of CGRP-positive tissue significantly decreased in the superficial dorsal horn in HI kits as compared to control kits (E t=2.991, df=79, p=0.0037, Welch’s t test). CGRP area was also measured in the deep dorsal horn (C-C”, D-D”). but there was no change between HI and control tissue (F t=0.5374, df=80.5, p=0.5925, Welch’s t test). There is a trending increase in the proportional area of IB4-positive tissue in HI kits as compared to control kits (G t=1.959, p=0.0538, Welch’s t test). Colocalization of CGRP-positive and IB4-positive fibers (A”, B”) was unchanged between groups (H U=894, p=0.2995, Mann-Whitney test). Data are represented as mean ± SD; individual data points represent averages of 2–3 separate images from each left and right dorsal horn per animal. Sham and naïve kits are pooled as control kits, with circles representing naïve kits and triangles representing sham kits (n=12 naïve, n=13 sham, and n=21 HI kits). **p<0.01.

Cervical nociceptive afferents are distributed similarly in HI kits as in control kits (Figure 8C–C”, D–D”). CGRP+ nociceptive afferents terminated in lamina I, II_o, and III and IB4+ nociceptive afferents terminated in lamina II_i (Figure 8C–C”, D–D”). Surprisingly, the proportional area of superficial CGRP+ nociceptive afferents decreased after HI as compared to control kits (p=0.0037) (Figure 8E). There was a trending increase in IB4+ fiber density in HI kits compared to controls (p=0.0538) (Figure 8G). There were no group differences in CGRP+ (p=0.5925) fiber density in the deep dorsal horn (Figure 8F).

Lumbar nociceptive primary afferent distribution in control kits was similar to the cervical cord at P18 (Figure 9A–B). However, in HI kits, the proportional area of deep dorsal horn CGRP+ (p=0.0256) (Figure 9F) and IB4+ fibers (p=0.043) (Figure 9G) but not superficial CGRP+ (p=0.1079) (Figure 9E) fibers increased as compared to control kits. This is consistent with our previous finding that deep dorsal horn CGRP+ was increased in the lumbar cord after HI, though at P5 we did not see increased IB4+ fibers increased after HI [49]. In addition, peptidergic and non-peptidergic nociceptive afferent populations maintain distinct topographical distribution in both control and HI groups in the cervical (p=0.2995) and lumbar (p=0.6215) spinal cord, as evidenced by little to no colocalization (Figure 8H, 9H). This contrasts with our previous finding of increased colocalization of CGRP+ and IB4+ fibers after HI at P5 [49]. The differences are likely due to the developmental trajectory. Overall, our findings demonstrate differential effects of prenatal HI on P18 nociceptive afferent area at different spinal cord levels that correspond to forepaw and hindpaw dermatomes.

Figure 9: HI kits have increased density of IB4+ and deep dorsal horn CGRP+ nociceptive primary afferents in the L4–5 dorsal horn at P18.

Representative images of the lumbar dorsal horn labeled with calcitonin gene-regulated peptide (CGRP; A, B) or labeled with isolectin B-4 (IB4; A’, B’). The proportional area of CGRP-positive tissue in the superficial dorsal horn remained unchanged in HI kits as compared to control kits (E t=1.627, df=77.26, p=0.1079, Welch’s t test). But the proportional CGRP area increased in the deep dorsal horn (C-C”, D-D”) in HI as compared to control tissue (F U=537, p=0.0256, Mann-Whitney test). The proportional area of IB4 positive tissue also increased in HI kits as compared to control kits (G t=2.062, df=68.86, p=0.043, Welch’s t test). Colocalization of CGRP-positive and IB4-positive fibers (A”, B”) was unchanged between groups (H U=786, p=0.6215, Mann-Whitney test). Data are represented as mean ± SD; individual data points represent averages of 3 separate images from each left and right dorsal horn per animal. Sham and naïve kits are pooled as control kits, with circles representing naïve kits and triangles representing sham kits (n=9 naïve, n=12 sham, and n=21 HI kits). *p<0.05

Principal component analysis reveals clusters of variables that distinguish control from HI kits

Principal component analysis was conducted to reduce dimensionality across the behavioral and immunofluorescence dataset. Four principal components collectively accounted for 67.3% of the total variance (Figure 10). The variables contributing most strongly to each PC (loading coefficient ≥ 0.45) are summarized in Table 1.

Figure 10: Two-texture preference, open field anxiety, and nociception variables distinguish HI from control kits.

Principal component analysis revealed new uncorrelated variables called principal components (PCs). (A) The relative contributions of variables within PC 1–4. PC 1 has high contribution from Hargreaves’ scores and immunofluorescence variables, PC 2 has high contribution from two-texture preference and open field anxiety-related variables, PC 3 has high contribution from two-texture preference and open field anxiety-related variables as well as Hargreaves’ scores, and PC 4 has high contribution from von Freeze scores. (B) PC scores are plotted by experimental group. HI kits had increased PC 2 scores compared to control kits (t=2.391, df=35.17, p=0.0223, Welch’s t test), and decreased PC 3 scores (t=3.255, df=43.76, p=0.0022) and PC 4 scores (t=2.108, df=46.01, p=0.0405) as compared to control kits, while there was no difference between HI and control kits in PC 1 (t=1.578, df=37.98, p=0.1228 Welch’s t test). *p<0.05, **p<0.01

Table 1:

Loading values for each variable in the first four principal components

Variables	PC1 (26.3%)	PC2 (15.8%)	PC3 (14.3%)	PC4 (10.9%)
von Frey	0.24230151	−0.421297312	0.38577025	0.54680893
von Freeze	0.24484420	0.295907482	0.28708226	0.52712405
Time spent Aversive:Non-Aversive	0.04856346	0.201916863	0.73146317	0.19279680
Immobility Aversive:Non-Aversive	−0.02260012	−0.02260012	0.57614582	−0.68327170
Time spent Center:Surround	−0.08845090	0.845593408	0.11997564	0.06586015
Immobility Center:Surround	−0.14312412	0.668642681	−0.39251710	0.02098420
IB4+ fiber area	0.76092623	0.207551991	−0.11869539	−0.07658858
Hargreaves’	−0.51522738	0.346520037	0.47491008	−0.13730564
CGRP and IB4 colocalization	0.90768931	0.011754197	0.02497562	−0.22972749
Superficial DH CGRP+ fiber area	0.71660216	−0.049745918	0.22785032	−0.17810361
Deep DH CGRP+ fiber area	0.74628181	0.315944768	−0.11091560	0.06184419

PC1 was dominated by immunofluorescence variables and Hargreaves’ withdrawal latency, suggesting that spinal nociceptive signaling and heat sensitivity are tightly associated as a distinct dimension of the HI phenotype. PC2 and PC3 were driven primarily by open field and two-texture preference variables, respectively, reflecting the anxiety-related and texture aversion behavioral dimensions. Hargreaves’ withdrawal latency also loaded onto PC3, indicating partial overlap between heat sensitivity and texture aversion behavior. PC4 captured mechanical sensitivity, with von Frey and von Freeze thresholds loading positively alongside an inverse contribution of two-texture preference immobility.

When PC scores were compared between control and HI kits (Figure 10B), PC1 scores did not differ between groups (p = 0.1228). In contrast, HI kits had significantly higher PC2 scores (p = 0.0223), lower PC3 scores (p = 0.0022), and lower PC4 scores (p = 0.0405) relative to controls. The separation of HI kits from controls across PC2, PC3, and PC4, but not PC1, indicates that behavioral measures of anxiety-like behavior, texture aversion, as well as mechanical and cold sensitivity distinguished the groups, while the immunofluorescence-dominated PC1 did not.

Discussion

Here, we used a pre-clinical HI rabbit model of CP, a faithful model of the injuries to the developing central nervous system[26] and motor deficits[21; 22; 61], to longitudinally assess sensory alterations across early postnatal development and their association with psychosocial and anatomical changes commonly associated with neuropathic pain. We found that HI rabbits, when tested longitudinally throughout postnatal development, display increased mechanical nociceptive sensitivity and a trajectory consistent with accelerated development. Additionally, HI kits have increased thermal nociceptive sensitivity to heat, but not cold, stimuli. Consistent with these findings, at P18, HI kits display altered texture preferences compared to control kits and an increase in anxiety-like behavior. These behavioral changes are accompanied by altered topographical distribution of nociceptive primary afferent fibers in the spinal cord, including an expansion of peptidergic and nonpeptidergic nociceptors. Multivariate statistics revealed that variables composed primarily of two-texture preference, open field anxiety, and nociception variables distinguish HI from control kits.

It is imperative to utilize validated pre-clinical models of disease to better understand etiopathology. We have expanded on our previous work showing that the hypersensitivity of HI kits at P5 [49] is retained through postnatal development, and the developmental pattern of nociceptors and nociception resembles precocious development (anatomically, CGRP in the deep dorsal horn of the lumbar cord at P5 in HI resembles P18 in controls; and behaviorally, heightened mechanical nociception at younger ages in HI rabbits resembles controls at older ages). We hypothesize this is due to the brain injuries and weakened corticospinal tract underlying CP [25; 53]. With less descending control in HI rabbits, nociceptive afferents likely have less synaptic competition and dominate sensorimotor development. The range of ages that we studied in rabbits is roughly similar to children from neonatal to toddler stages based on the development of the corticospinal tract, the ability to weight support and locomote. This is consistent with literature in clinical populations showing already by ages 8 to 12, more children with CP self-report pain than the general population[46]. In a follow-up study 5 years later including many of the same participants, 74% of adolescents with CP between the ages of 13 and 17 reported pain[45]. Future studies in HI rabbits will assess behavioral and anatomical plasticity in older juvenile and adult rabbits.

Mechanical nociception was reliably altered in HI kits, as was thermal nociception in response to heat and cold. These sensory modalities are mediated by distinct mechanisms, which may point to changes in specific nociceptor populations or channels associated with mechanosensation, heat sensation, or cold sensation such as Piezo channels[68], TRPV1 channels[41] or TRPM8 channels[41], respectively. Interestingly, Black et al. associated TRPM8-positive fibers with proprioception, suggesting a potential functional link between nociception and the hyperreflexia that marks spasticity [6]. Further analysis of the molecular changes that may be responsible for the mechanical and heat-related hypersensitivity / cold-related hyposensitivity we observe in HI kits is warranted.

The two-texture preference test is a complex task involving cortical regions responsible for perception and execution of motor responses [3] and is a strong validation of our von Frey results. The aversive astroturf-like texture provides mechanical stimulation to the plantar surface of the paws in a similar manner as von Frey. Both control and HI kits displayed nocifensive behaviors (i.e., licking the paw) after locomoting on the astroturf; however, HI kits had more immobile events at the interface of the two textures compared to control kits. Importantly, analysis of locomotor ability revealed that both groups performed similarly, suggesting that HI rabbits may be avoiding the texture due to pain rather than a hesitancy to move or explore.

The sensory changes in HI kits are accompanied by alterations in anxiety-like behavior in the open field at P18. HI rabbits spent a greater proportion of time in the center of the open field compared to controls, which would traditionally indicate reduced anxiety; however, they have a greater ratio of immobility time in the center zone to immobility time in the surround zone with some indicators of freezing behavior (no visible movement besides respiration), a behavior consistent with heightened anxiety[9; 36; 39]. The pre-weaning age of our subjects may also account for this inconsistency, as the behavioral expression of anxiety differs between juveniles and adults[4; 14; 32; 58; 60].

In the open field, control rabbits initially display cautious exploratory behavior, moving slowly and extending onto their hindpaw toes before each hop, behaviors previously characterized as “fear” or “caution” in rabbits [39]. As they acclimate, controls transition to faster, more confident movement, termed “boldness” [39]. HI kits show a marked shift in this pattern. Upon placement in the center, HI kits display risk assessment behaviors, delaying exploratory initiation, remaining at or near the placement location for a prolonged period; some move in circles while others freeze for several seconds, exhibiting no visible movement except that required for respiration, before initiating locomotion. These observations are consistent with a heightened fear response and align with the broader biopsychosocial dimensions of chronic pain.

Despite capturing the largest proportion of variance, PC1 did not distinguish HI from control kits, suggesting that spinal nociceptive signaling and heat sensitivity — while altered in HI animals — are not the primary axes of group divergence at P18, possibly reflecting heterogeneity in the degree of spinal sensitization across HI animals. In contrast, HI kits diverged from controls along three independent behavioral dimensions: open field anxiety (PC2), texture aversion (PC3), and mechanical sensitivity (PC4). The co-loading of Hargreaves’ withdrawal latency onto PC3 alongside texture aversion variables suggests partial overlap between heat sensitivity and aversive texture avoidance, consistent with a shared dimension of sensory hypersensitivity that manifests across modalities. Collectively, these results indicate that the HI phenotype at P18 is multidimensional, encompassing sensory, affective, and motivational components consistent with a chronic pain state.

Our findings provide high-fidelity validation of the HI rabbit model of CP and reinforce its translational alignment with the clinical condition. We demonstrate that HI rabbits exhibit hypersensitivity to both mechanical and heat stimuli. Cleanly separating general hypersensitivity from allodynia or hyperalgesia remains difficult; however, these sensory changes, taken together with the two-texture preference and open field data, suggest that HI rabbits may be experiencing a pain state. This interpretation aligns with the clinical literature: people with CP have been shown to exhibit mechanical hypoesthesia, thermal hypoesthesia, and mechanical hyperalgesia, and 23% of individuals with CP in the same study met criteria for allodynia[8]. Disambiguating nociception, hyperesthesia, and pain within the HI rabbit model remains an important direction for future work. Our findings are also consistent with what has been reported in rodent CP models, where HI-induced allodynia and hypersensitivity on the von Frey and Hargreaves tests have been documented in an age- and sex-dependent manner and linked to cytoarchitectural changes in supraspinal regions associated with somatosensation[20; 35].

This study could be limited by the fact that HI litters included mildly affected kits and unaffected kits, and no severely affected HI kits. However, prior work did not reveal robust differences between motor-affected and motor-unaffected HI kits in mechanical and thermal sensitivity[49]. Pain is present across Gross Motor Function Classification System levels, but the painful dermatomes vary. For example, individuals able to ambulate are more likely to report foot pain than those who do not walk[2; 5; 38; 43; 44]. Thus, it will be important to assess nociceptive and psychosocial pain in severely affected HI rabbits in future experiments. Second, due to the size of rabbit litters and the number of tests performed on each postnatal day it was unreasonable to have one experimenter perform all of the tests or have the same person perform each test on each rabbit each day. This may have increased the variability since individual animals have been shown to respond differently in sensory tests depending on the person performing the test [59]. We accounted for this by randomizing who was performing the test across litters and timepoints, so all litters and groups were treated similarly. Third, some conclusions regarding the pain experience in HI kits are limited by the reliance on segmentally mediated nociceptive withdrawal reflexes to infer pain perception. To address this, we recorded supraspinal responses during von Frey, Hargreaves, and von Freeze testing to differentiate withdrawal responses localized to the spinal cord from those perceived at supraspinal levels. Moreover, the two-texture preference test assessed the perceived aversiveness of a rough, astroturf like walking surface. Together these assessments provide insight into the connectivity of local spinal cord circuits to supraspinal, cortical centers involved in pain perception. Behavioral assays such as the formalin test will be important for disambiguating hyperreflexia from perceptive pain in HI rabbits but were beyond the scope of the present study.

Conclusions

In conclusion, these data emphasize the clinical relevance of the HI rabbit as a model of CP pain and point to future work following pain behaviors in these rabbits into adulthood. This work and the research that builds upon it will lead to increased awareness of pain as a comorbidity of CP, a better understanding of nociception in the context of CP, and the development of more efficacious treatments for said pain.