Vascularization of the dorsal root ganglia and peripheral nerve of the mouse: Implications for chemical-induced peripheral sensory neuropathies

In the present study, we have shown that the CBRA of the DRG has approximately a seven fold higher density of the CD31+ blood vessels (which are mostly capillaries) than the NFRA of the DRG or the proximal and distal regions of the sciatic nerve. Interestingly, previous reports have shown similar striking regional differences in the expression of tight junction proteins and the presence of a functional blood-nerve barrier in the NFRA of the DRG but not in the CBRA of the rat DRG [32]. Previous studies have also shown that the blood vessels that vascularize the CBRA of the DRG have large fenestrations when compared to the peripheral nerves [33‐35]. Thus, when large molecules such as albumin [32, 36] or horseradish peroxidase [35, 37] are injected into the rat tail vein these molecules remain in the vessels in the NFRA of the DRG and the vessels of the peripheral nerve but avidly leak out of the vessels that vascularize the CBRA of the DRG. Thus, the dense vascularization, lack of the full repertoire of tight junction proteins and the highly fenestrated blood vessels that are present in the CBRA of the DRG makes this area of the peripheral nerve uniquely open and accessible to a variety of low and high molecular weight agents [33]. Thus, even intravenous administration of fluorescently labeled Evans-blue albumin (MW 68,000) results in dense deposition of this tracer immediately adjacent to the plasma membrane of the soma of rat DRG neurons [32].

Together, these results suggest that a blood-nerve barrier is largely lacking in aspects of the peripheral nerve that house the cell bodies of sensory neurons. In addition to this relatively unrestricted access, there are high metabolic demands placed on the cell bodies [38] that support the long axons that innervate the most distal extremities. This may in part explain why intravenous administration of potentially neurotoxic agents such as methyl mercury show preferential access [33, 39], accumulation [39‐41] and toxicity [33, 41‐43] to sensory neurons when compared to motor neurons whose cell bodies are housed within the blood brain barrier within the ventral horn of the spinal cord.

Neurotoxicity is a major side effect of many commonly used anti-neoplastic and anti-HIV agents [11‐19, 44]. Previously, the dose limiting toxicity of many chemotherapeutic agents was hypersensitivity and neutropenia, but since the former can now be treated with antihistamines or steroids and the latter with granulocyte colony-stimulating factor [45], peripheral neuropathy is the de facto toxicity that limits the administration of many commonly used anti-neoplastic agents [11‐14]. This is significant as the current trend in oncology is towards more aggressive chemotherapy as is evident in recent studies demonstrating that in many chemotherapeutic regimens increased dose is associated with a clear increase in patient survival [13, 22‐24].

In previous studies we showed that intravenous administration of paclitaxel, a commonly used chemotherapeutic agent, leads to the development of peripheral neuropathy characterized by injury of neuronal and non-neuronal cells in the DRG [46, 47]. To define the location of the cells that showed the first sign of injury we administered this agent intravenously and then looked for the appearance of markers of cell injury in the rat DRG, peripheral nerve and spinal cord at 1, 3, 6 and 10 days following initial intravenous infusion of paclitaxel [42]. Using this strategy, it was found that at day 1 post-infusion there was a marked up-regulation of activated transcription factor-3 (ATF3), wich is a marker of cell injury/regeneration [48, 49] in a subpopulation of large and small DRG neurons. In contrast, markers of cell injury in the proximal or distal aspects of the sciatic nerve were not observed until 10 days post-infusion of paclitaxel, suggesting that one of the initial sites of paclitaxel induced injury was in the CBRA of the DRG. Interestingly, by 10 days post-paclitaxel infusion there were clusters of satellite cells in the DRG which have been suggested to be a "tombstone" of dead sensory neuron cell bodies [47]. Similar "tombstones" have been observed in DRGs obtained at autopsies of cancer patients treated with cisplatin [50] and in AIDS patients treated with antiretroviral agents [15, 51], which also frequently produce a predominantly sensory chemotherapy-induced peripheral neuropathy (CIPN) [16‐18, 20, 21]. Together these preclinical and clinical studies demonstrate that there are signs of injury and death of the cell body of sensory neurons of rats and patients receiving chemotherapeutic drugs and that direct drug-induced injury to the cell body and its supporting cells may participate in the generation of CIPN.

Defining the mechanisms and developing new therapies to attenuate or prevent therapy induced neuropathy in cancer and HIV patients are both needed and possible. Thus, in many cases these neuropathies are frequently the dose limiting toxicity that limits the ability of clinicians to pursue an aggressive chemotherapeutic regimen that will in large part determine the survival of the patient [52‐54]. For example, in patients with metastatic colorectal cancer, oxaliplatin therapy offers clear dose-related benefits in promoting disease free survival [55]. However, in many cases optimal and aggressive dosing of oxaliplatin is limited by CIPN which results in reduction or cessation of chemotherapy resulting in decreased survival rates for these patients [52‐54, 56, 57].

Therapy induced peripheral neuropathy presents a unique opportunity in neuropathobiology in that preemptive therapy can be commenced before the chemotherapeutic regimen begins. Previous studies in preclinical models of stroke and spinal cord injury have shown the significant advantages in preemptive versus post-injury administration of neuroprotective strategies [58]. However, the clinical reality in stroke and spinal cord injury is that neuroprotective therapies can usually only be initiated after the neuronal injury has occurred [58]. In contrast, anti-neoplastic or anti-HIV therapies offer a unique opportunity in that the precise time and extent of neuronal injury induced by the chemotherapeutic agent is known allowing preemptive trials to be conducted in both the preclinical and clinical setting.

Currently cancer and AIDS survivors constitute the largest and most rapidly expanding group of patients that have peripheral neuropathy and neuropathic pain [19, 20, 44, 59]. Thus, developing a mechanism based understanding of how these chemotherapeutic agents induce this primarily peripheral sensory neuropathy and developing mechanism based therapies to prevent and/or attenuate these neuropathies offers a significant opportunity to impact both the survival and quality of life of these patients.

Experiments were performed on a total of 12 adult male C3H/HeJ (C3H) mice (Jackson Laboratories, Bar Harbor, Maine) and 12 adult C57 B6.Cg-Tg(Thy1-YFP)16Jrs/J mice (Jackson Laboratories, Bar Harbor, Maine), weighing 20–25 g. The transgenic mice constitutively express yellow fluorescent protein (YFP) in motor and sensory neurons under the control of neuron-specific regulatory elements from the Thy1 gene [28, 29]. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Minnesota.

Mice were sacrificed with CO₂ and perfused intracardially with 20 ml of 0.1 M phosphate buffered saline (PBS) followed by 20 ml of 4% formaldehyde/12.5% picric acid solution in 0.1 M PBS. For whole-mount DRG preparations, the L3-L5 DRG of Thy1-YFP mice and C3H mice were harvested together with its adjacent spinal roots and nerve, post-fixed for 4 h in the perfusion fixative and then processed for immunohistochemistry (see below). For sectioned tissue, the lumbar (L3-L5) DRG, sciatic nerves, and lumbar spinal cord of C3H mice were removed, post-fixed for 4 h in the perfusion fixative, and cryoprotected for 24 h in 30% sucrose in 0.1 M PBS all at 4°C and then processed for immunohistochemistry. Longitudinal sciatic nerve sections (1.5 cm segment) were obtained at mid thigh level approximately 1.0 cm proximal to the trifurcation.

Qualitative analysis indicated no differences in the CD31+ blood vessel density in L3 to L5 DRG. Thus for quantification and presentation purposes only L4 DRG were used. The L4 DRG and attached nerves were incubated for 60 min at room temperature (RT) in a blocking solution of 3% normal donkey serum in PBS with 0.3% Triton-X 100 and then incubated overnight at RT in primary antisera against the platelet endothelial cell adhesion molecule CD31, a marker of endothelial cells present in blood vessels [26, 27] (monoclonal rat anti-mouse CD31, 1:500, BD Pharmingen, San Diego, CA). YFP immunofluorescence in Thy1-YFP mice was readily visible under an epifluorescence microscope, thus no amplification of the signal was performed. Tissue was washed in PBS and incubated for 3 h at RT with a secondary antibody (Cy3 anti-rat; 1:600, Jackson ImmunoResearch, West Grove, PA). Finally, the DRG and attached nerves were washed 3 × 10 min in PBS, mounted on gelatin-coated slides, dried, dehydrated via an alcohol gradient (70, 90, and 100%), cleared in xylene, and coverslipped with DPX.

Serial frozen sections of L4 DRG, lumbar spinal cord (coronal sections) and sciatic nerve (longitudinal sections) were cut at 15 μm on a cryostat and mounted onto gelatin-coated slides for immunohistochemical analysis. In this case, sectioned tissues were incubated overnight at RT in primary antisera against CD31. Secondary antibodies conjugated to various fluorescent markers (Cy2 1:200, Cy3 1:600; Jackson ImmunoResearch, West Grove, PA) were used and further immunohistochemical steps were performed as described above. To confirm the specificity of the primary antibody, controls included preabsorption with the corresponding synthetic peptide or omission of the primary antibody.

Laser scanning confocal microscopy of the whole mount preparations was performed with a BX-61 microscope equipped with the Fluoview 1000 imaging software 5.0 (Olympus America Inc, Melville, NY). Confocal z-series at 0.5 μm intervals were acquired for each observation area and filtered by two-frame Kalman low-speed scans. YFP was excited by a laser wavelength of 488 nm, and emissions were detected using a 522-nm emission filter. Sequential acquisition mode was used to reduce bleed through. Z-series of the different experiments were imported from the Olympus Fluoview format to the Imaris Pro Software 5.7.0 (Bitplane AG, Zurich, CH). Image threshold and channel pseudocolors were adjusted, and 3D reconstruction was performed in the Surpass module. Images were cropped with Adobe Photoshop CS and thereafter assembled in Adobe Illustrator CS.

Quantification of blood vessels was performed by determining the number of CD31 positive (+) blood vessels per unit area [30, 60]. Only sectioned tissue from C3H mice was used to perform this quantification. Microvessel profiles were identified using criteria described by Weidner and colleagues [61] where the presence of a vessel lumen was not required to identify vessel profiles. Only CD31+ blood vessels that were 2–10 μm in diameter were counted and CD31+ branched blood vessels were counted as one vessel. The number of CD31+ blood vessels was determined in cell body-rich area (CBRA) and nerve fiber-rich area (NFRA) [32]. Images of the DRG sections were captured on an Olympus BX51 epifluorescent microscope fitted with an Olympus DP70 digital camera and areas of interest were determined using Image Pro Plus version 3.0 software (Media Cybernetics, Silver Spring, MD). The number of CD31+ blood vessels per outlined area from 4 sections per ganglion spaced 100 μm apart was averaged for each animal and results were expressed as total number of CD31+ blood vessels per unit area (mm²). Random 250 μm × 250 μm areas of the sciatic nerve and medial dorsal horn of the spinal cord were viewed at ×400 magnification and the number of CD31+ blood vessels was quantified. Distal (approximately 1.2–1.5 cm distance from tribifurcation) and proximal areas (approximately 2.2–2.5 cm distance from tribifurcation) of the sciatic nerve segment were used for quantification. Two areas were counted per section of 4 sections of spinal cord and sciatic nerve (proximal and distal) spaced 100 μm apart. Results are expressed as the mean number of CD31+ blood vessels per unit area (mm²) ± SEM. Statistical differences were determined using ANOVA followed by Tukey post hoc test. p < 0.05 was considered significant.