The present study showed that a rodent knee OA model induced by MIA exhibited tactile allodynia with decreased PWT and elevated proinflammatory mediators in local tissues. At the DRG level, the proportion of CGRP-ir DRG neurons innervating the knee joint significantly increased, unlike that of IB4-binding DRG neurons. Furthermore, the proportions of FG-labeled ATF3-ir and FG-labeled ATF3-ir GAP43-ir DRG neurons were significantly increased in the postinjection period. In the spinal cord, more proliferation of microglia were observed in the ipsilateral dorsal horn.
Histological and behavioral evaluation of the MIA-induced OA model
MIA injection has been reported to cause joint pathology via the inhibition of glycolysis, thereby targeting avascular cartilage and causing chondrocyte death [
30]. MIA-treated knees that were histologically evaluated in the present study exhibited osteoarthritic histological changes. The measurement of diameters of the knees increased significantly after day 7, which indicates the pathogenesis of knee OA in conjunction with the pathological findings The distribution of the weights of the animals showed no bias; therefore, the results of the von Frey tests indicate escape behavior and not lifting from physical loading. The significant decrease of PWT indicated tactile allodynia pathogenesis, consistent with a previous study [
28]. This decrease is considered to result from allodynia derived from central sensitization; the increase in afferent signaling from the joint nociceptors to the spinal cord neurons results in increased sensitivity of the spinal cord neurons to input from the joint, rendering the spinal cord neurons hyperexcitable [
31].
Some studies have noted "mirror pain" that could be observed as a referred pain in the contralateral limb [
32], whereas the results of the present study did not reveal significant evidence for this. This implies significant laterality of the pathogenesis of OA pain-related activities in the ipsilateral knees compared with that in the contralateral knees.
Cytokine expression in the local tissues of the knee joints
In the present study, the production of TNF-α and IL-6 showed significant temporal increases prior to the elevation of NGF production. MIA injection, by inducing chemical injury, might have induced the production, and inflammactivation of chondrocytes or synovial membranes, or both, is considered to play an important role as mentioned in the background. The evidence shows that normally quiescent chondrocytes, as well as synovial cells, respond to repetitive excess mechanical loading via stress-induced intracellular signals that mediate the production of proinflammatory mediators such as cytokines and cartilage-degrading proteinases [
33]. Another study found that synovial inflammation is a factor that likely contributes to dysregulation of chondrocyte function, favoring an imbalance between the catabolic and anabolic activities of chondrocytes in remodeling the extracellular cartilage matrix [
34]. Thus, the activated chondrocytes and synovial membranes should be the major sources of the elevated proinflammatory mediators, among which TNF-α produced from the activated chondrocytes or synovial membranes can induce IL-6 upregulation and neuropathic pain in the spinal cord and DRG [
35]. In parallel, another study found that IL-6 enhances the expression of TNF-α receptors [
36], thus these proinflammatory mediators act in concert to provoke further inflammation. NGF is generally reported to be involved in chronic inflammatory or neuropathic pain states [
37]. NGF is physically produced in articular structures and expressed in normal and OA synovial tissues, being increased in synovial inflammation, especially upon synovial tissue exposure to TNF-α [
14]. Thus, NGF production during inflammation might be involved in the modulation rather than in the induction of the joint inflammatory response.
The findings of the present study suggest that the elevated production of TNF-α and IL-6 in the acute phase after MIA injection may have induced elevated production of NGF and that these elevated proinflammatory cytokines are involved in the modulation of joint inflammation.
Evaluation of sensory innervation at different DRG and spinal cord levels
The present study showed that sensory neurons innervating the knee are predominantly CGRP-ir. The average numbers of the FG-labeled neurons did not significantly change between each time point, which indicates that there was no significant increase or decrease of the neurons innervating the joint. CGRP-ir DRG neurons are reported to be NGF-dependent and critical to hyperalgesic responses induced by inflammation [
38‐
40], which also means that these DRG neurons are "sensitive" to increased levels of NGF [
16‐
18]. This may explain the evidence for increased numbers of CGRP-ir DRG neurons in the MIA-injected knees in the present study, which may result from stimulation of the peripheral nerve fibers by increased peripheral NGF production initiated from local inflammation as proposed previously.
The present study showed evidence for a low percentage of IB4-binding DRG neurons (less than 4%). This approximately coincides with the findings of a previous study that reported the sensory innervation of hip joints [
29], while another study reported a complete absence of IB4-binding DRG neurons [
41]. However, IB4-binding DRG neurons may not be involved in pain generation without any overlaps between CGRP-ir neurons.
The numbers of both ATF3-ir and ATF3-irGAP43-ir FG-labeled DRG neurons showed significant temporal increases compared with their levels on the contralateral side. ATF3 is not thought be expressed during inflammation [
42], thus the increased expression of ATF3-ir in DRG neurons suggests the evidence of gradually progressive nerve injuries. The pain behavior evaluated in the hind paw using von Frey filaments is thought to be radiational pain derived from allodynia from the previous studies. Therefore, we should consider that it is not directly related to the neuronal damage indicating ATF3 or GAP43 immunoreactivity. However, the results of the present study suggest that both the behavioral changes and ATF3 production in DRG changes 7-14 days after the injection are related to each other.
Simultaneously, the significant gradual increase in the number of ATF3- and GAP43-ir DRG neurons implies a restoration process of the injured nerves in addition to the nerve injuries. The progressive nerve injury and regeneration may have resulted from the degradation of cartilage; the physical wearing of cartilage leads to exposure and degradation of subchondral bone, where sensory nerve ingrowth and pain-related mediators increase [
43]. Subsequently, the exposed nerve endings may become physically injured, which can accelerate nerve ingrowth into the subchondral bone. This can lead to the temporally increased expression of ATF3 and GAP43. The absolute values of the proportions of ATF3-ir and ATF3-irGAP43-ir DRG are not very high. However, we consider its significant existence in the ipsilateral DRGs to be important. In stronger neuronal models such as the sciatic nerve axotomy model, ATF3-ir and GAP43-ir are frequently and clearly observed [
44]. Therefore, we can hypothesize that MIA injection does not induce such drastic neuronal damage during the acute phase but the neuronal damage accompanying ATF3 immunoreactivity occurs as the neuronal damage proceeds.
Furthermore, it has been suggested that the proliferating microglia should also play a crucial role as is already mentioned in the background. Furthermore, increased levels of proinflammatory cytokines such as TNF-α and IL-6 produced after peripheral nerve injury have been reported to induce neuropathic pain by enhancing the activities of microglia in the spinal cord [
45,
46]. The present study showed significant proliferation of microglia in the ipsilateral dorsal horn in addition to the elevated levels of proinflammatory cytokines, which can contribute to the neuronal damage, which may lead to the pathogenesis of neuropathic pain. In addition, the changes in the dorsal horn of the spinal cord regarding microglia were significant only in the ipsilateral dorsal horn. In general, many changes in the nervous system can occur bilaterally even when there is a unilateral disease, which was not observed in the present study. One of the reasons for this is likely the post-injection period. Therefore, future investigations should be performed for longer periods
The findings above indicate that the MIA-induced OA pain results from inflammation, which initiates an inflammatory pain state, and subsequently, and neural damage gradually arising along with progressive cartilage degradation may induce a chronic state such as a neuropathic pain state. Also infrapatellar fat pad is reported to be a source of the pain in OA, thus we should take these reports into consideration [
47]. Considering translational research, this may have important implications for targeted analgesic therapies. In addition to antiinflammatory therapies such as NSAIDs, which may be mainly effective for the early treatment of OA, treatment for neuropathic pain may be effective for OA patients in metaphase.
The present study has some limitations. First, MIA-induced OA is a chemically induced model. To clarify the details, further investigation with other nonchemically induced pathological models, such as partial medial meniscectomy, should be conducted. Second, the behavioral test for tactile allodynia with von Frey hairs can be somewhat inaccurate and depends on the responses of the animals. To analyze the behavior of rats objectively and correctly, it may be helpful to use an apparatus to analyze their behavior by using other methods such as evaluating weight bearing in free-moving walking animals [
48] as well as response to heat or measuring direct pressure to knees [
49]. Third, the samples from each knee were complex structures consisting of synovium and capsules, and we did not investigate the distribution of proinflammatory cytokines in each structure. The distribution of receptors and their expression as well as cytokines themselves should be examined in future studies. Fourth, the present study examined up to 4 weeks after MIA injection. As the neuropathic pain state has a tendency to increase temporally, the behavior of the inflammatory pain state may be important; therefore, further investigation with a longer postinjection period may be needed. Lastly, we examined the knee joints to confirm the visual absence of yellow leakage of FG from and around the knee joints throughout the experiments. However, we did not examine this nor circulation using fluorescence microscopy. We think the effect of the leakage is counteracted by averaging the data. Further study may be necessary to clarify whether any leakage actually occurred.