The main finding of this study is that the pharmacological blockade or genetic knockout of σ1R prevents the increased incidence of atypical axonal mitochondria in saphenous nerve myelinated fibers and the neuropathic pain signs associated with the administration of paclitaxel in mice. These findings suggest, for the first time, an involvement of the σ1R in the paclitaxel-evoked mitochondrial abnormalities that appear to be important in the pathophysiology of paclitaxel-induced neuropathy.
We confirm here that paclitaxel induces cold and mechanical allodynia in WT mice as previously reported [5, 28, 29]. However, when activation of σ1R was hindered, through a genetic or pharmacologic approach, the development of paclitaxel-induced allodynia was completely prevented, suggesting a key role for the σ1R in this type of neuropathic pain. These results are in agreement with those of previous studies demonstrating that σ1R-KO mice [23, 24, 30] and WT animals pretreated with σ1R antagonists [20–22, 28, 31] showed a marked reduction of pain in different models that activate central sensitization mechanisms. In addition, it has been reported that the spinal σ1R system contributes to diabetic neuropathic pain in mice . Therefore, the present and previously published behavioral data strongly support the involvement of σ1R in modulating pain, especially neuropathic pain.
Our paclitaxel treatment schedule induced an increase in the frequency of atypical mitochondria in A-fibers of mouse saphenous nerve. These atypical mitochondria were always swollen and/or vacuolated (area > 0.20 μm2; diameter > 500 nm). These criteria are very similar to those used by authors who also found an increased incidence of atypical axonal mitochondria in peripheral nerves of rats with paclitaxel- [7, 8], oxaliplatin-  or bortezomib-induced neuropathy . An increase in swollen and/or vacuolated mitochondria has also been reported in the peripheral nerves  and DRGs  of animals with diabetic neuropathy and in the sural nerves of patients with painful peripheral neuropathy induced by 2’3’-dideoxycitidine (ddC) and HIV infection . Hence, these mitochondrial structural alterations may be a common characteristic of these types of neuropathy.
Our finding of a low incidence of atypical mitochondria in naive mice appears to be a common observation in normal animals fixed with aldehydes [7, 35]. In the present study, we processed all nerves using the same methodology; the mitochondria conserved their double membrane, their neighboring microtubules were well preserved, and the mitochondria from other cells (Schwann cells, fibroblasts, endothelial cells) were normal. Consequently, it is highly unlikely that the paclitaxel-induced increase in atypical mitochondria was due to an unsuitable fixation rather than to a neurotoxic effect of paclitaxel. We also found an increase, although not statistically significant, in atypical mitochondria in the saphenous nerves of mice treated with paclitaxel-vehicle, similar to that reported previously . This is not surprising, given that one of its main components, Cremophor EL, can directly damage mitochondria .
It is known that paclitaxel-induced neuropathy is associated with a hypersensitization of A-fibers without affecting C-fibers in mice , and that paclitaxel mainly impairs the functionality of large myelinated (A-β) fibers in humans . In agreement with these data, we found that paclitaxel induced a significant increase in atypical axonal mitochondria in A-fibers but not in C-fibers of saphenous nerves from WT mice. The time-course of the mitochondrial and behavioral alterations was similar, and both were evident on day 10 and resolved by day 28. An increase in axonal atypical mitochondria (also in parallel with the time-course of behavioral changes) was previously reported in paclitaxel-treated rats [7, 8]. In these studies, however, both myelinated and unmyelinated fibers were affected and the behavioral and mitochondrial changes peaked later (27 days) than in the present study. The lack of significant effect of paclitaxel in unmyelinated fiber mitochondria of WT mice in our study could have been due to the greater percentage of atypical mitochondria in unmyelinated than myelinated fibers in control condition or to the greater variability in such a percentage in unmyelinated fibers (which would have affected the probability of reaching statistical significant differences). However, we do not think that this is the case since in WT-mice paclitaxel-treatment increases 7 times the percentage of atypical mitochondria in A-fibers but only 1.5 times in C-fibers. Because the percentage of atypical mitochondria in naive WT mice C-fibers is around 10% it would have been perfectly possible to increase this value also 7 times without reaching a plateau (since 90% of C-fibers mitochondria are typical in naive animals and therefore are susceptible to become atypical as a consequence of paclitaxel treatment) but, in fact, we do not observed such effect. Instead, differences between species may explain the discrepancy between C-fiber mitochondrial alterations induced by paclitaxel in rats [7, 8] and in mice (present study), because important variations in the primary afferent unmyelinated neurochemistry between mice and rats have been previously described [38, 39]. In fact, when similar protocols of paclitaxel treatment are used, the neuropathic pain induced by paclitaxel peaked earlier and it is of shorter duration in mice [5, 29, 40] than in rats [7, 8]. Moreover, a different involvement of spinal cord microglial activation by paclitaxel in mice and rats has been reported [29, 41, 42]. Nevertheless, paclitaxel did not produce alterations in the mitochondria of Schwann cells in rats  nor in mice (present study). Although the details of the paclitaxel-induced neuropathy seems not to be the same in rats and mice, it is interesting to note that in both species paclitaxel induces qualitatively similar behavioral and mitochondrial changes, which suggest that these are core characteristics of paclitaxel neuropathy independently of the species considered. Therefore, our results in mice and those previously reported by Bennett’s group in rats suggest that paclitaxel-induced neuropathic pain may result from an impairment of axonal mitochondria. In fact, functional impairment of mitochondria was recently reported in peripheral nerves from paclitaxel- and oxaliplatin-treated rats .
Previous studies in mice [45–47] and rats [48–50] found evidence of axonal degeneration or alterations in Schwann cells or microtubules after paclitaxel administration. We did not observed any of these structural irregularities, probably because the single and cumulative doses used here were markedly below those administered in the above-mentioned studies (single dose, 2 mg/kg in the present study vs. 5–50 mg/kg in the others; cumulative dose, 10 vs. 20–280 mg/kg). This explanation is supported by the absence of these structural anomalies in other studies using similarly low doses [7, 43] to those tested in the present study.
Genetic inactivation (σ1R-KO mice) or pharmacological blockade (σ1R antagonist) of the σ1R prevented paclitaxel-induced mitochondrial abnormalities and neuropathic pain signs. This suggests that the σ1R must be present and play a key functional role in the development of paclitaxel-induced painful neuropathy and atypical mitochondria. Therefore, the prophylactic effect of σ1R antagonists such as BD-1063 (present work; ) and S1RA  on the development of paclitaxel-induced cold and mechanical allodynia may be related to the prevention of these mitochondrial abnormalities. These data support the proposal of selective σ1R antagonists as a novel approach to the treatment of neuropathic pain .
It has been suggested that the mechanisms by which paclitaxel cause the mitochondrial abnormalities may derive from its binding to the β-tubulin associated with the voltage-dependent anion channel (VDAC) . VDAC is the most abundant protein in the mitochondrial outer membrane  and, under certain situations (e.g., mitochondrial Ca2+ overload), may open the mitochondrial permeability transition pore (mPTP) and eventually produce mitochondrial alterations, including mitochondrial swelling and the release of accumulated Ca2+
. Thus, paclitaxel has been found to induce these effects in vitro
[53–55]. Another possible explanation of paclitaxel-induced mitotoxicity is the indirect regulation of mPTP through the binding of paclitaxel to bcl-2, reversing the function of bcl-2 as a blocker of mPTP opening . Taken together, these data suggest that paclitaxel may induce mPTP opening by binding to the β-tubulin joined to VDAC and/or to bcl-2, which would induce mitochondrial swelling and increase the release of mitochondrial Ca2+ to the cytoplasm. The σ1R modulates VDAC function  and tonically regulates the expression of bcl-2 proteins  and consequently may also indirectly regulate mPTP opening, inhibiting mitochondrial swelling and Ca2+ release. Further studies are warranted to test this hypothesis.