Coexistence of chronic hyperalgesia and multilevel neuroinflammatory responses after experimental SCI: a systematic approach to profiling neuropathic pain

General physical condition and signs suggestive of spontaneous pain after SCI

All animals lived through the experiment period except for one rat in the SCI group that was euthanatized due to a lower body skin lesion resulting from excessive licking possibly triggered by allodynia (note: another rat was removed from the study due to inadequate injury); light hematuria in several SCI rats resolved within 3–5 days. The group mean time to restore a reflex bladder in SCI rats was 4.3 ± 1.8 days (n = 12), which was typical for this model of injury [26] and much shorter than in rats with contusion SCI [27]. Control animals had neither postsurgical complication nor functional deficit. Thus, the group size of SCI and control animals was n = 12 and 10, respectively (Fig. 1A, B).

Changes in body weight were recorded to monitor wellbeing of the rats. There was an anticipated degree of body weight loss within the first week following T10 compression. Specifically, relative to the pre-SCI level, the mean percentage of body weight reduction was 0% for the laminectomy group, and 4.2% for the SCI group (range: 0–8.1%; n = 12), which was far below 15% set as a criterion to remove animals from the study. All SCI rats recovered their body weight to levels before surgery after the first week p.i., and afterward, continuously gained weight at a rate normally observed in rats following similar surgical procedures (data not shown) [14, 26].

To detect possible presence of spontaneous pain, a central component of clinical NP [5, 7], signs suggestive of stress and/or pain, which included porphyrin staining, reduced grooming, licking or scratching of an intact body area, vocalization, etc., were scored based on housing checking chart records using a formula we developed (see Additional file 1: Table S1 for specifics including references). In the range of 0–38, a score ≥ 5 attained from a non-invoked animal suggested likely presence of pain. Whereas all control rats (n = 10) had 0 points during weeks 2–8 p.i., the SCI rats (n = 12) showed scores in a range of 5–10 and 2–5 in weeks 2–5 and 6–8, respectively, with the group median score consistently being 5, indicating that some mild degree of pain might exist since the average body weight gains were not disturbed.

Post-SCI reduction of hindlimb coordinated motor functionLocomotion

Evaluation based on the BBB scale revealed marked deficits in hindlimb locomotion in rats after T10 SCI; in contrast, all rats after laminectomy showed no locomotor dysfunction (i.e., BBB score = 21; Fig. 1C). SCI rats demonstrated distinct reduction of hindlimb locomotor ability that was most profound at day 1 p.i. (Fig. 1C). However, compared to the typical sign of spinal shock (i.e., complete flaccid paralysis: BBB score = 0) that occurs during the first 24 h after lower thoracic contusion and T10 severe compression (50 g × 5 min) [26, 27], the SCI rats did not manifest spinal shock syndrome (i.e., group mean BBB score/day-1 p.i.: 3.1 ± 1.4; n = 12; Fig. 1C). Thereafter, hindlimbs showed a gradual recovery of locomotion that plateaued 3–4 weeks later. At the end of the study, the SCI group's mean BBB score was 16.1 ± 0.9 points (versus 21 ± 0 of the control group), showing a chronic deficit level that was less severe than that observed in the previous study using the same compression regimen (i.e., BBB scores of ~ 12) [14].

Incline plane test

Compared to the control group, the mean maximum degree where SCI rats could stabilize their body postures while facing-downward, mainly by the hindlimbs was significantly lower in 1 day p.i. (37.6 ± 2.0°/SCI versus a normal range of 55.7 ± 0.8°/control, p < 0.05; two-way repeated measures ANOVA with Sidak’s post hoc test). As seen in the locomotor changes, incline plane performance of the SCI rats improved more over the first 3 weeks before plateauing around 4–8 weeks p.i. (Fig. 1D). At 8 weeks p.i., the SCI group’s mean maximum incline plane degree was 49.3 ± 1.3°, which remained significantly lower than the 57.2 ± 0.9° of the control group (n = 12/SCI or 10/control; p < 0.05 to < 0.001; two-way repeated measures ANOVA with Sidak’s post hoc test).

In this study, the moderate T10 compression formula resulted in group average BBB scores and incline plane angles that were much closer to that of mild compression (20 g × 5 min) reported earlier [14, 26]. Thus, the term of “moderate compression” was used to specify the physical regimen of the quasi-static compression per se (35 g × 5 min). This discrepancy was probably caused by subtle differences in surgical details such as laminectomy size produced by different operation performers.

Post-SCI abnormalities of sensory functionsNeurological reflexes

The contact righting reflex was evaluated to assess the ability of the spinal cord to coordinate with the brainstem, peripheral afferent and efferent system, and neuromuscular junctions to correct abnormal body postures [14, 26, 33,34,35]. Furthermore, postural (placing) and spinal reflexes were tested [14, 26]. These data were presented as the percentage of rats in either group that had normal reflexes [14].

All SCI rats lost the ability to perform normal righting reflex (Fig. 2A), proprioceptive positioning reflex of paw placing (Fig. 2B), and withdrawal response to hindlimb extension (Fig. 2C), at 1 day p.i. There were noticeable recoveries of the three categories of reflexes in the following weeks with 80% SCI rats showing normal righting and paw placing reflexes (Fig. 2A, B), and extension reflex (Fig. 2C) by 3 and 4 weeks p.i., respectively. Conversely, withdrawal reflexes induced by brief pressuring (i.e., pressure withdrawal reflex; Fig. 2D) and pinching (i.e., nociception withdrawal reflex; Fig. 2E) of the hindpaw were detectable (despite severe reductions) 1 day p.i. (i.e., no spinal shock). Their recovery rates, however, were much slower and incomplete in comparison to those of the righting (Fig. 2A), placing (Fig. 2B), or extension reflex (Fig. 2C). For example, by 8 weeks p.i., only ~ 33% rats with SCI showed a normal nociception withdrawal reflex (Fig. 2E), which was much lower than that of righting, placing, or extension reflex recovery (i.e., 100%; Fig. 2A–C), revealing a protracted abnormality of nociceptive signal-induced neurological reflexes in rats with T10 SCI.

Mechanical hypersensitivity

Mechanical allodynia, a representative symptom of NP, is a hypersensitivity triggered by innocuous stimuli similar to light touch. Clinically, the standardized VFT is used to detect a mechanical sensory threshold in the quantitative sensory testing (QST) for pain in patients [36]. In this study, mVFT, which emulated light touch, was performed to detect evoked mechanical hypersensitivity (Fig. 2F) [14, 37]. Before SCI, the group withdrawal sensitivity threshold averaged 8.2 ± 0.6 g/control and 7.5 ± 0.4 g/SCI in the forepaws (p > 0.05; Fig. 2G), and 8.6 ± 0.4 g/control and 9.0 ± 0.8 g/SCI in the hindpaws (p > 0.05; Student’s t test; Fig. 2H). In contrast to a stable array of sensitivity thresholds attained from the control group, at day 1 p.i., the mean withdrawal sensitivity threshold drastically increased to 11.1 ± 1.5 g and 45.8 ± 5.1 g in the forepaws and hindpaws of SCI rats, respectively (Fig. 2G, H), suggesting that there was severe loss of neurological functions close to that of spinal shock. Thereafter, the group mean withdrawal threshold, compared to the pre-SCI level and control values, significantly decreased in the forepaws (i.e., above-injury level allodynia), starting 2 weeks after SCI (4.7 ± 0.7 g/week 2 p.i., p < 0.05; 3.9 ± 0.5 g/week 5 p.i., p < 0.01; and 3.8 ± 0.9 g/week 8 p.i., p < 0.01; n = 12; Fig. 2G); the hindpaws of SCI animals also had significantly decreased mean withdrawal sensitivity threshold (i.e., below-injury level allodynia; 4.1 ± 0.6 g/week 2 p.i., p < 0.01; 3.6 ± 0.4 g/week 5 p.i., p < 0.01; and 4.0 ± 0.3 g/week 8 p.i., p < 0.01; n = 12; Fig. 2H). Lastly, at-injury level mechanical allodynia measured in the dorsal dermatomes adjacent to the T10 injury epicenter (Fig. 2I) exhibited a comparable pattern of significant augmentation of hypersensitivity, compared to the control group (0.8 ± 0.3 g/week 2 p.i., p < 0.01; 0.5 ± 0.2 g week 5 p.i., p < 0.01; and 0.4 ± 0.1 g week 8 p.i., p < 0.01; n = 12/SCI or 10/control; two-way repeated measures ANOVA with Sidak’s post hoc test; Fig. 2I). For all three levels tested, there were no significant changes in withdrawal sensitivity thresholds in the laminectomy control group. These data demonstrated that SCI rats displayed subacute and chronic hypersensitivities to mechanical stimuli above-, below-, and at-injury level.

Thermal hypersensitivity

Thermal hypersensitivity threshold is an important parameter in the QST for diagnosing thermal allodynia, another important symptom of clinical SCI NP [38,39,40,41]. SCI and control rats first underwent the bilateral hot plate test for which the pre-SCI and pre-laminectomy mean hindlimb latency was 8.8 ± 0.8 s (n = 12) and 8.0 ± 0.9 s (n = 10; p > 0.5, Student’s t test), respectively (i.e., baseline values; Fig. 2J). While the control group showed a steady mean response latency, the group mean latency increased dramatically to 19.1 ± 3.1 s in the injured rats at day 1 p.i., relative to the baseline value (i.e., 8.8 ± 0.8 s), indicating a transient severe reduction of neurological function during the first 24 h p.i. Subsequently, the group mean latency of SCI animals gradually decreased, with the average threshold becoming significantly lower (i.e., ↑thermal hypersensitivity) than the baseline values and those of the control group, starting in week 3 p.i. (Fig. 2J; n = 12/SCI or 10/laminectomy; p < 0.05; two-way repeated measures ANOVA with Sidak’s post hoc test).

In the more sensitive unilateral hindlimb hot plate test [39], although an overall resemblance to the bilateral hot plate test data was observed, this test detected significantly reduced group mean latencies 1 week earlier, starting day 14 p.i., which lasted till the end of the study, compared to the control group (Fig. 2K; n = 12/SCI or 10/control; p < 0.05; two-way repeated measures ANOVA with Sidak’s post hoc test). Furthermore, the SCI group mean threshold of this test trended even lower than the bilateral hot plate test results (Fig. 2K), confirming the test was more sensitive.

Histopathological impacts of moderate T10 compression

Whereas no discernible lesions were found in the laminectomy control tissues, all injured spinal cords exhibited apparent damages in the white matter (dorsal, lateral, and ventral funiculi) and gray matter (dorsal and ventral horns, and intermediate and central regions) at the injury epicenter, forming cavities and lesion areas that further extended rostrally and caudally to demarcate the lesion volume (Fig. 3A). Epicenter coronal sections also contained a high number of basophilic inflammatory cells (i.e., subcellular structures containing nucleic acids stained dark blue/violet by hematoxylin) and intra-cavity loose webs of non-neural tissues (Fig. 3A). Farther away from the epicenter, sections sampled from T8–9 spinal cord that was ~ 5 mm rostral to the compression site (Fig. 3B), exhibited discernible demyelination (i.e., hypostained by solvent blue) and infiltration of inflammatory cells in both the white matter and gray matter (Fig. 3C).

Fig. 3figure 3

Histopathological outcomes. A Coronal spinal cord sections in 20 µm thickness post solvent blue/H&E stain showed histopathological defects in the SCI group only (scale bar: 1 mm). Injured spinal cord sections at and around the epicenter exhibited damages in the white matter and gray matter (e.g., cavities and lesion areas). The injured tissues also contained a large number of basophilic inflammatory cells (stained by hematoxylin) and intra-cavity loose webs of non-neural tissues. B, C T8 spinal cord sections that were ~ 5 mm rostral to T10 injury site (B; framed areas were magnified in C), had apparent demyelination (i.e., hypostained by solvent blue) and infiltration of inflammatory cells in both white matter and gray matter (C). D, E Area measurement of representative coronal sections at the injury site (i.e., 0 mm) and 1–3 mm rostral and caudal to it, revealed that compared to the control group, T10 compression significantly reduced the mean residual white matter (D) and gray matter areas (E) at the epicenter, and in spinal cord loci 1 and 2 mm adjacent to it (p < 0.05; two-way repeated measures ANOVA with Sidak’s post hoc test; n = 7/group)

Measurements of digital images of representative coronal sections at the injury site and 1–3 mm rostral and caudal to it, demonstrated that relative to the control group, T10 compression significantly reduced the mean residual white matter (Fig. 3D) and gray matter areas (Fig. 3E) at the epicenter, and in loci 1 and 2 mm adjacent to it (p < 0.05; two-way repeated measures ANOVA with Sidak’s post hoc test; n = 7/group). Taken together, the morphological evidence showed chronic neuroparenchyma loss and lesions, and presence of inflammatory cells in the epicenter and regions bidirectionally adjacent to it after T10 compression [14, 26].

Chronic NIF, NTM, NML, and NPL changes at multiple nuclei/regions of the CNS in rats with SCI NP

Our previous work demonstrated that lumbar and cervical NIF co-existed with NP-like behaviors in the early chronic phase of SCI [14, 26]. However, whether NIF after a T10 injury may extend into other thoracic cord levels such as T8 where more sympathetic preganglionic neurons reside [27], and different brain centers involved in pain signaling relay and regulation remained to be systematically investigated [14, 17, 42, 43].

Changes of NIF, NTM, NML, and NPL markers in injured spinal cords

To assess NIF responses in T8 neuroparenchyma, IRL of GFAP (astrocyte activation and reactive astrogliosis), Iba-1 (microglia/macrophage activation), TNFα (proinflammatory cytokine), and/or iNOS (proinflammatory polarization marker of microglia/macrophages and an inflammation mediator) expressions were evaluated in spinal cord sections sampled from 5 mm rostral to the injury epicenter (SCI group) or T10 level (i.e., laminectomy site; control group) (Fig. 4A; see Fig. 4B for the labeling of the framed areas that were examined: corresponding higher magnification images were in Fig. 4C–Q). Both the dorsal horn (DH) and dorsal column (DC) had significantly increased expression of GFAP (DH: Fig. 4C, D; DC: Fig. 4G, H) and iNOS (DH: Fig. 4C, D1; DC: Fig. 4G, H1) in injured spinal cords, compared to the control tissue (DH: Fig. 4C1; DC: Fig. 4G1). Also significantly elevated in injured spinal cords were expressions of TNFα (DH: Fig. 4E, F; DC: Fig. 4I, J) and Iba-1 (DH: Fig. 4E, F1; DC: Fig. 4I, J1; controls: Fig. 4E1/DH and Fig. 4I1/DC; n = 4/group; p < 0.01; Student’s t test). In addition, compared to the control group, the lateral column (LC) and ventral funiculi (VF) of SCI tissues showed significant IRL augmentations of GFAP (LC: Fig. 4K, L; VF: Fig. 4O, P), iNOS (LC: Fig. 4K, L1; VF: Fig. 4O, P1), TNFα (LC: Fig. 4M, N; VF: Fig. 4Q, R), and Iba-1 (LC: Fig. 4M, N1; VF: Fig. 4Q, R1; controls: Fig. 4K1, M1 for LC/Fig. 4O1, Q1 for VF; n = 4/group; p < 0.01; Student’s t test).

Fig. 4figure 4

Neuroinflammation in spinal cord tissue rostral to the epicenter. A Immunoreactivity level (IRL) of GFAP (reactive astrogliosis), Iba-1/(microglia/macrophage activation), TNF-α (proinflammatory cytokine), and/or iNOS (marker of proinflammatory microglia/macrophage; inflammation mediator) expressions were evaluated in coronal sections sampled from T8 spinal cord (SCI group) or T10 level (control group). B Specific areas in each section that were examined by IHC assay (for specific data point, please see individual images accordingly labeled in CQ). In the dorsal horn (DH; CF1) and dorsal column (DC; GJ1), there were significantly increased expressions of GFAP (DH: C, D; DC: G, H) and iNOS (DH: C, D1; DC: G, H1) in injured spinal cords, compared to the control tissue (DH: C1; DC: G1). Also significantly heightened in the dorsal spinal cord were expressions of TNFα (DH: E, F; DC: I, J) and Iba-1 (DH: E, F1; DC: I, J1; controls: E1/DH and I1/DC; n = 4/group; p < 0.01; Student’s t test). In addition, compared to control sections, the lateral column (LC) and ventral funiculi (VF) of SCI tissues showed significant IRL augmentations of GFAP (LC: K, L; VF: O, P), iNOS (LC: K, L1; VF: O, P1), TNFα (LC: M, N; VF: Q, R), and Iba-1 (LC: M, N1; VF: Q, R1; controls: K1, M1 for LC and O1, Q1 for VF; n = 4/group; p < 0.01; Student’s t test; scale bars: 40 µm/C, G; 60 µm/K, O)

Next, expression levels of calcitonin gene-related peptide (CGRP), a main neurotransmitter released from the C and Aδ sensory fiber terminals onto the substantia gelatinosa neurons, which facilitates NP/pain development [44, 45], and p75 neurotrophin receptor (p75NTR), a key receptor to regulate peripheral and central nociceptive neurons [46], were evaluated with IHC assays of T8 spinal cord sections (SCI group) or at T10 level (control). Relative to control tissues (Fig. 5A1/CGRP; Fig. 5C1/p75NTR), significantly heightened group mean IRL of CGRP (Fig. 5A, B) and p75NTR (Fig. 5C, D) were detected primarily in Rexed Laminae (RL) I and II (CGRP: right inset in Fig. 5A; p75NTR: inset in Fig. 5C), deeper zones of DH (CGRP: left inset in Fig. 5A), and the dorsal roots of SCI tissues (CGRP: Fig. 5H1, J; p75NTR: Fig. 5H2, J1; Fig. 5I/control; p < 0.01, n = 4/group, Student’s t test). For SCI NP-triggered NPL and NML alterations [14], IHC analysis uncovered significantly increased group mean IRL of Homer-1a, a molecular marker for DH neuronal plasticity (Fig. 5E, G) [14] and serotonin (5HT, a neuromodulator; Fig. 5E1, G1) in RL-I and II, relative to the control group (Fig. 5F; p < 0.05, n = 4/group, Student’s t test).

Fig. 5figure 5

Changes of neurotransmission, neuroplastic, and neuromodulatory molecules in injured spinal cords. IRL of calcitonin gene-related peptide (CGRP, a pain-related neurotransmitter) and p75 neurotrophin receptor (p75NTR, a regulator of nociceptive neurons) were significantly higher in spinal cord sections 5 mm rostral to the epicenter (i.e., ~ T8 spinal cord; SCI group; A and B/CGRP; C and D/p75NTR), compared to T10 level of the control (A1/CGRP; C1/p75NTR); the IHC signals were primarily located in Rexed Laminae (RL) I and II (CGRP: right inset in A; p75NTR: inset in C) and deeper zones of DH (CGRP: left inset in A) and in the dorsal roots (CGRP: H1, J; p75NTR: H2, J1; I/control; p < 0.01, n = 4/group, Student’s t test). Also disclosed by IHC analysis were significantly increased mean IRL of Homer-1a, a marker of DH neuronal plasticity (E, G) and serotonin (5HT), a neuromodulator (E1, G1) in RL-I & II of the SCI group, relative to the control group (F; p < 0.05, n = 4/group, Student’s t test). Furthermore, statistical linear regressions uncovered that IRL of CGRP and p75NTR were negatively correlated with the sensitivity threshold (unit: second) of the bilateral hot plate test (CGRP: p = 0.02, R2 = 0.83, K; p75NTR: p < 0.01, R2 = 0.94, L) in the SCI group; changes in IRL of Homer-1α and 5HT also correlated negatively with the sensitivity threshold of the bilateral hot plate test in SCI animals (Homer-1α: p < 0.01, R2 = 0.72, M; 5HT: p < 0.01, R2 = 0.85, N; scale bars: 100 µm)

Lastly, to determine possible functional impacts associated with expression changes of the afore-described molecules p.i., statistical linear regressions were conducted. The results showed that IRL of CGRP and p75NTR were negatively correlated with the thermosensitivity threshold (unit: second) determined by the bilateral hot plate test (CGRP: p = 0.02, R2 = 0.83, Fig. 5K; p75NTR: p < 0.01, R2 = 0.94, Fig. 5L) for the SCI group. Moreover, changes in IRL of Homer-1α and 5HT also correlated negatively with the thermosensitivity threshold derived from the bilateral hot plate test in animals with SCI (Homer-1α: p < 0.01, R2 = 0.72, Fig. 5M; 5HT: p < 0.01, R2 = 0.85, Fig. 5N). These data suggested that increased DH expressions of CGRP and p75NTR might have worsened SCI NP, which could have triggered stronger NPL and NML responses (e.g., higher expressions of Homer-1α and 5HT, respectively, in RL-I and II of DH).

Chronic NIF, neuronal hyperactivity, and serotonin reduction in the brainstem

Situated in the medulla oblongata at the junction of the cervical spinal cord and brainstem, neurons in the GrN (Fig. 6A) relay somatosensory information from the lower half of the body and the legs to the thalamus and participate in maintaining NP [47, 48]. In this study, significantly higher group mean IRL of Iba-1 (Fig. 6B, D) and TNFα (Fig. 6B1, D1) presented in brainstem coronal sections containing GrN of SCI rats compared to those of the control group (Fig. 6C). IHC double stains of the same level sections showed significantly elevated mean IRL of GFAP (Fig. 6E, G1) and iNOS (Fig. 6E1, G) in the SCI rats relative to the control group (Fig. 6F; n = 4/group; p < 0.05, Student’s t test). Because c-Fos expression in secondary sensory neurons in the GrN could be augmented by electrical or nerve injury stimulation of c-fibers to emulate nociceptive afferent signals [49], numbers of cFos+ neurons in GrN were quantified via IHC co-localization of cFos, and nuclear markers of NeuN and DAPI. The mean number of neurons producing cFos was significantly higher in SCI animals than that of controls (Fig. 6H/SCI versus Fig. 6I/control; statistics in Fig. 6K, n = 4/group; p < 0.05, Student’s t test). Orthogonal slicing confirmed co-localization of cFos, NeuN, a highly specific marker of neuronal nuclei (Fig. 

留言 (0)

沒有登入
gif