Hypoxic stress enhances extension and branching of dorsal root ganglion neuronal outgrowth

Abstract It has been shown that painful intervertebral discs (IVDs) were associated with a deeper innervation. However, the effect of the disc's degenerative microenvironment on neuronal outgrowth remains largely unknown. The focus of this study was to determine the influence of hypoxia on dorsal root ganglion (DRG) neurite outgrowth. Toward this aim, the DRG‐derived cell line ND7/23 was either directly subjected to 2% or 20% oxygen conditions or exposed to conditioned medium (CM) collected from IVDs cultured under 2% or 20% oxygen. Viability and outgrowth analysis were performed following 3 days of exposure. Results obtained with the cell line were further validated on cultures of rabbit spinal DRG explants and dissociated DRG neurons. Results showed that hypoxia significantly increased neurite outgrowth length in ND7/23 cells, which was also validated in DRG explant and primary cell culture, although hypoxia conditioned IVD did not significantly increase ND7/23 neurite outgrowth. While hypoxia dramatically decreased the outgrowth frequency in explant cultures, it significantly increased collateral sprouting of dissociated neurons. Importantly, the hypoxia‐induced decrease of outgrowth frequency at the explant level was not due to inhibition of outgrowth branching but rather to neuronal necrosis. In summary, hypoxia in DRG promoted neurite sprouting, while neuronal necrosis may reduce the density of neuronal outgrowth at the tissue level. These findings may help to explain the deeper neo‐innervation found in the painful disc tissue. Highlights Hypoxia promoted elongation and branching of neurite outgrowth at single cell level, but reduced outgrowth density at tissue level, possibly due to hypoxia‐induced neuronal necrosis; these findings may help to explain the deeper neo‐innervation found in clinically painful tissues.


| INTRODUCTION
Chronic low back pain (LBP) has been reported as the leading cause of disability worldwide, considering its incidence, prevalence, and years lived with disability. 1 A major cause of chronic LBP is the degeneration of the intervertebral disc (IVD), 2,3 which is often associated with chronic inflammation, cell death, and extracellular matrix degradation. [4][5][6] Deviations in nutrient supply of IVD are frequently associated with IVD degeneration. 7,8 Ischemia-related risk factors such as aortic calcification, stenosis of lumbar arteries, smoking, and high serum cholesterol levels are consistently correlated with discogenic pain. 9,10 The association between permeability of the endplate and intervertebral disc degeneration has been described in human specimens since the seventies. 11 Recently, endplate perfusion was evaluated in adult patients using dynamic contrast enhanced MRI and a reduced endplate perfusion was found to be correlated with a higher degree of IVD degeneration. 12 One consequence of the decreased nutrient supply to the IVD is hypoxia, that is, an oxygen concentration below the metabolic requirements. 13 An increased metabolic consumption of oxygen by resident IVD cells upon degeneration was recently reported; 14 moreover, recruited inflammatory cells may further aggravate the hypoxic state. 15 Hypoxia bio-markers such as Hypoxia-Inducible Factor 1-alpha, Glucose Transporter Type 1, 3, and 9 were upregulated in clinical degenerative IVD samples compared to healthy controls. 16,17 While progress has been made to understand the mechanisms underlying disc degeneration, the most important clinical question concerns the pain related to disc degeneration. 18 The healthy IVDs are almost aneural, 19 but neo-innervation has been repeatedly observed in painful and degenerative disc tissue 20,21 which represents a prospective vital factor for the mechanism of chronic pain. However, the pathophysiological mechanism of the aberrant nerve sprouting in the context of IVD hypoxia remains unknown. The IVD exposed to hypoxia stress may therefore produce multiple factors stimulating the proximal sensory nerve structure named dorsal root ganglion (DRG), which is responsible for initiating nociceptive/pain input into the central nervous system for the sense of pain. 22 This influence mediated via IVD (hypoxia stressed IVD may release molecules that stimulate DRG) was defined as the indirect effect of hypoxia on DRG.
On the other hand, the degenerative IVD can compress 23 and chemically stimulate the DRG, 24 increase its endoneural fluid pressure and reduce its blood supply, 25 which is associated with the development of hypoxia in the DRG. Therefore, hypoxia may exert a direct influence on DRG. Although hypoxia in DRG has not been measured in discogenic pain patients so far, large animal models showed that exposure of nucleus pulposus tissue through IVD incision reduced blood flow in the nerve root and dorsal ganglion of dogs, therefore leading to DRG ischemia. 24,26 Furthermore, it has been shown that oxygen tension in DRGs was severely affected as a result of ischemia induced by aortic clamping. 27 In a word, hypoxia was shown to play multiple roles in IVD degeneration and IVD-related neuropathy in proximal DRG. The hypothesis of this study is that hypoxia may either directly or indirectly (via IVD) promote neurite sprouting which permits the neo-innervation of IVD and can possibly be associated to chronic discogenic pain. Therefore, the aim of this study was to investigate (a) the effect of hypoxiastressed IVD conditioned medium (CM) on DRG neurite outgrowth and (b) whether hypoxia can directly promote neurite outgrowth in DRGs.
To study the indirect effect of hypoxia on DRG (via IVD), hypoxia was modeled by culturing whole IVDs from bovine tails inside an incubator set at 2% oxygen level. Since oxygen tension in the bone marrow above the IVD endplate was reported to be around 6.4%, 28 2% oxygen around the cultured IVD organ would meet the definition of hypoxia. Likewise, since normal oxygen tension in rat DRG tissue was estimated to be between 3.9% and 5.5%, 29 2% oxygen would also represent a hypoxic condition for DRG cell culture. The DRG-derived ND7/23 cell line 30,31 was either cultured in hypoxia or in CM of hypoxia-stressed IVD. Findings were validated using rabbit dissociated DRG neurons and adult rabbit DRG explant cultures.

| MATERIALS AND METHODS
An overview of the experimental design used in this study is provided in Figure 1. Briefly, the effect of hypoxia on neurite outgrowth was initially tested using the DRG-derived neuronal cell line ND7/23 and was then validated using primary DRG neurons or DRG explants that were obtained from the lumbar spines (L2-5) of New Zealand white rabbits (n = 6, all female, 28

| Viability of DRG-isolated neurons at 2% and 20% oxygen
The DRG cell dissociation and culture were modified based on the previous report. 38 DRGs were dissected from two rabbit lumbar spines (L2-5) (New Zealand white female rabbits, 28 weeks old). The The viability and outgrowth of the DRG-isolated neuron culture were evaluated based on the protocol described by Gladman et al. 39 Briefly, DRG isolated cells were incubated with 3 μM ethidium homo- and Hoechst staining were identical to the ones used for the 4-day primary neuron culture. Images were acquired with EVOS FL Auto 2 Imaging System at ×20 magnification, 0.40 numerical aperture, and 6.8 mm working distance to visualize the neuronal soma and its outgrowth.
The selection of neurons for the evaluation of outgrowth was based on the following criteria: (a) the cell was identified as neuron by positive staining to NF200; (b) no sign of necrosis or apoptosis was detected; and (c) the outgrowths of the neurons were not contacting or intersecting with one another. The mask of the neuronal outgrowth was created using ImageJ with a threshold of 35-255. The "particle remover" plugin of ImageJ Fiji was used to eliminate the background.
The neurons were analyzed using the "Sholl" method for the length, density, and pattern distribution of outgrowth. 40  showed that hypoxic culture conditions increased the proportion of cells with outgrowth by 13.6% (P = .02; Figure 2C) and increased the median outgrowth length by 38.5% (P < .001) ( Figure 2D) at 3 days compared to normoxic cultures. Representative phase contrast images of the cells cultured at 2% and 20% oxygen are provided in Figure 2E,F.
3.2 | Hypoxia on IVD was not found to influence outgrowth of ND7/23 cell line When exposed to hypoxia, factors produced by the IVD were hypothesized to influence DRG neurite outgrowth. The results, however, did not seem to support this hypothesis. The IVDs treated with hypoxia showed no significant difference comparing to those treated by normoxia regarding the influence of their CM on neurite outgrowth of ND7/23. (Figure 2A,B).

| Hypoxia decreased DRG cell line viability
Viability was evaluated in a separate 2-day experiment using ethidium homodimer-1 staining (necrosis) and nuclei morphology (apoptosis) analysis. mRNA expression of anti-apoptosis (Bcl-2) and pro-apoptosis (Bax and Bad) genes 43 was evaluated using quantitative RT-PCR.
Results showed that hypoxia decreased the median proportion of viable cells by 3.0% (P < .001) without considering the detached dead cells, and reduced cell density by 84.7% (P < .001) ( Figure 4A).
Decreasing oxygen tension increased the median proportion of necrotic cells per field by 19.2% (P < .001) ( Figure 4A). Although by nuclei morphology analysis, the proportion of apoptotic cells per field was not significantly influenced by hypoxia (P = 1) ( Figure 4A), quantitative RT-PCR showed a significantly increase in anti-apoptosis gene expression in the hypoxic group (P < .001) (Bcl-2), while pro-apoptosis genes (Bax and Bad) were not significantly different between the culture groups (for Bax, P = .28) (for Bad, P = 1) ( Figure 4B). These results indicate the reduced viability in hypoxia is mainly due to higher necrosis.

| Hypoxia increased length but decreased frequency of outgrowth of 4-day-cultured DRG explants
DRG explant culture was used to verify the findings obtained with the cell line. To attain equal cultured DRG size between 2% and 20% oxygen, at each spinal level, DRG halves from one side were assigned to 2% oxygen, while halves from the other side were cultured under 20% oxygen (the assignment of the side was performed randomly).
After 4 days of culture, NF200 stained outgrowth density was lower under 2% oxygen as shown in Figure 5A-D. Removal of primary antibody resulted in the absence of positive staining (data not shown).
Quantification using the Simple Neurite Tracer plugin within ImageJ showed that outgrowth length was 28.6% higher under 2% oxygen (P < .001) ( Figure 5E), but total neurite outgrowth frequency per DRG was reduced by 90.9% at 2% as compared to 20% oxygen (P < .001) ( Figure 5F). F I G U R E 4 Viability of differentiated ND7/23 cells following a 3-day culture at 2% and 20% oxygen. A, Hypoxia induced a significantly higher proportion of necrotic cells, a significantly lower proportion of viable cells and a reduced cell density compared to culture at 20% oxygen (*P < .05 by Mann-Whitney test, n = 40 fields for both 2% and 20% of oxygen). B, 2% oxygen significantly increased the relative mRNA expression of the anti-apoptoticBcl-2 gene, while the pro-apoptosis genes Bax and Bad showed no significant difference. The Bad/Bcl-2 was even significantly lower under 2% oxygen (*P < .05 by Mann-Whitney test, n = 4 technical replicates for both 2% and 20% oxygen from 2 independent experiments). Data were normalized to normoxia control compared to 20% oxygen (P < .001) (Figure 6A,B,G). The proportion of apoptotic cells in 2% oxygen was 49.5% lower than in 20% oxygen (P < .001) ( Figure 6C-F,F). Overall, the proportion of viable neurons (non-necrotic and non-apoptotic) was similar between 2% and 20% oxygen (0.8% lower for 2% oxygen, P = .77) ( Figure 6I), but the necrosis/apoptosis ratio was significantly higher for the culture at 2% oxygen (39.3% higher, P < .001; Figure 6J). 3.7 | Hypoxia led to longer neurites with higher branching in cultured DRG primary neurons From the images of the primary neuron culture, the outgrowth appeared longer and denser at 2% oxygen ( Figure 7A,B). The "Sholl" analysis showed that the ending radius (which represents the radius of neuronal outgrowth) was 44.4% longer at 2% than at 20% oxygen (P = .049; Figure 7E). This finding was consistent with the results obtained for the explant cultures ( Figure 5).
The density of collateral branches was quantified by the number of intersections between the neurite arbor and concentric rings with regular radial increments centered in the neuronal soma ( Figure 7C).
The total number of intersections for each neuron was 2.13 times higher at 2% oxygen compared to 20% oxygen (P = .019) ( Figure 7F).
From the graph of the intersection distribution, it was evident that 2% oxygen always had a higher intersection number than 20% oxygen at various distances from soma, and this difference was even stronger at further distances from the soma ( Figure 7D). Thus, the intersections were separately evaluated at distances from the soma of 0,~30, 40-105, and 120-150 μm. Initial branch number (intersection number at 0 μm from soma) was 25% lower under 2% oxygen (P = .125) (Figure 7G), but at~30 μm from the soma, the outgrowth density at 2% oxygen was already 20% higher than at 20% oxygen (P = .084). The density of outgrowth was 20% higher at distances from the soma of 40 to 105 μm (P = .001) and became 140% higher at 120 to 150 μm from the soma for 2% oxygen (P < .0001). Hence, in regions more distant from the soma, more significant increases of outgrowth density were observed under 2% oxygen ( Figure 7H). The capability of collateral sprouting was represented by the ramification index parameter, which was 40% higher for 2% oxygen compared to 20% oxygen (P = .008; Figure 7I).

| DISCUSSION
Nerve ingrowth inside painful IVD tissue has long been observed, 21 but the underlying mechanism remains largely unknown. IVD degeneration is associated with nutrient deprivation and hypoxia stress in both IVD 17 and DRG 24 tissues. The hypothesis of this study was that the hypoxic stress in IVD and/or DRG causes aberrant neuronal outgrowth. In this study, the DRG-derived cell line ND7/23 was used to screen the effect of hypoxia. DRG cells were exposed to hypoxia either directly or indirectly, by application of CM collected from IVDs cultured under hypoxia. Although no influence of hypoxia in IVD was found on DRG neurite outgrowth, hypoxia in DRG clearly evidenced longer neurites which were then validated using primary DRG cell and DRG explant culture. Since the length of regenerated nerve fibers largely determines the depth of the innervation, our results suggest that hypoxia in DRG may contribute to the aberrant neurite sprouting.
Another aspect of neuronal outgrowth is collateral sprouting which may lead to a higher innervation density. For primary DRG neuron cultures, in-depth information could be obtained from the evaluation of neurite outgrowth pattern based on the "Sholl" method. Using fibers and lead to reduced tissue innervation. 48 Hence, it appears that F I G U R E 6 Viability of a 4-day primary culture of isolated DRG neurons. The neurons were isolated from the DRGs of two rabbit spines. A,B, Representative images of DRG cells cultured at 2% and 20% oxygen for 4 days. Immunofluorescent staining with NF-200 antibody was used to distinguish neurons from other cell types and ethidium homodimer-1 was used to detect necrosis, which is shown by the arrows in the image. Scale bars equal 200 μm. C-F, When combining Hoechst staining with immunofluorescence, neuronal apoptosis could be observed as indicated by arrow heads. Scale bars equal 100 μm. G, Under 2% oxygen a significantly higher proportion of necrotic cells in each field was observed than in cultures kept at 20% oxygen. H Significantly lower proportion of apoptotic cells per field was found at 2% oxygen. I, No significant difference was identified for the proportion of viable neurons at 2% and 20% oxygen (P = .769). J, The necrosis/apoptosis ratio per field was significantly higher for 2% than for 20% oxygen (***P < .001 by Mann-Whitney test, n = 210 and 223 fields for 2% and 20% oxygen, respectively) F I G U R E 7 "Sholl" analysis of DRG neurons cultured at 2% and 20% oxygen for 2 days. A,B, Pattern images of DRG neurons (immunofluorescent staining with NF-200 antibody). C, Schematic showing the "Sholl" analysis of the outgrowth pattern. Circled samplings intersect with the neurite arbors providing information on how far the outgrowth can reach (represented as ending radius) and the number of branches (represented as number of intersections) at different distance from the soma (~30, 40-105, and 120-150 μm). D, Number and distribution of intersections at different distances from the soma. Note that the curve representing the mean number of intersections at 2% oxygen (blue line) is always above that at 20% oxygen (red line). E, Statistics showed that the intersection radius at 2% oxygen was significantly longer than at 20% oxygen. F, The sum of intersections of the whole cell was also significantly larger for 2% than for 20% oxygen. G, The initial branch number showed no significant difference between 2% and 20% oxygen. H, The magnitude of the difference in intersection number between 2% and 20% oxygen increased with distance from the soma. I Ramification index, which is the maximum number of intersections per sampling divided by the initial branch number, was significantly higher at 2% than at 20% oxygen (for box plot, *P < .05, **P < .01, and ***P < .001 by Mann-Whitney test, n = 25 and 27 cells for 2% and 20% oxygen, respectively) stressful conditions can lead to either an increase or a decrease in tissue innervation. One evidence supporting the two-sided effect of stress on nerve density in tissues comes from the study of complex regional pain syndrome, where both neo-innervation (in hair follicles) and loss of innervation (in epidermal, sweat gland, and vessels) were simultaneously detected in skin biopsies. 49 The axonal density of regenerative sprouting has been shown to be associated with neuropathic pain. 50 an essential role in promoting IVD mediated neurite ingrowth. 56,57 Finally, in the degenerative IVD, local oxygen concentration and oxygen gradients in the proximity of neurite endings may have a stronger influence than the response of whole IVDs to hypoxia as assessed in this in vitro model.
Neuronal necrosis represents injury. Necrotic cells, losing plasma membrane integrity, will release their intracellular content, thereby provoking inflammation. 58 On the contrary, apoptosis is one kind of programmed cell death that has evolved to rapidly and efficiently eliminate unwanted cells, without eliciting inflammation and immune responses. 45 Based on our results, the ND7/23 cell line viability was reduced because of a higher proportion of necrotic cells but not due to apoptosis, since the expression of anti-apoptosis genes was even upregulated by hypoxia treatment. Similarly, a higher necrosis/apoptosis ratio was observed in primary DRG neuron cultures under hypoxic culture conditions. In vivo, the necrosis/apoptosis ratio has not been well characterized in DRGs yet, but the switch between necrosis and apoptosis represents an important tissue protecting mechanism against stress 59 and has been widely investigated in other tissues. For example, in vivo brain ischemia models showed that the neural protective agent Prothymosin α leads to a necrosis-apoptosis switch 60 and in vivo pancreatitis models showed a correlation between the injury effect of hypoxia/oxidative stress and an increased necrosis/ apoptosis transition. 61 The characterization of the mechanism of neuronal death, whether apoptosis or necrosis, will greatly improve our understanding of neuropathic components in chronic LBP. 62 According to International Association for the Study of Pain, neuropathic pain is defined as "pain that arises from a lesion or disease affecting the somatosensory system." 63 Indeed, neuronal apoptosis has been observed in in vivo neuropathic pain models, 64 but since neuronal apoptosis can be a consequence of nearby cell necrosis, 65 it remains unknown whether necrosis or apoptosis plays the main role in neuronal death related to neuropathic pain. Furthermore, the role of neuronal apoptosis in promoting pain remains uncertain, since mice deficient of pro-apoptosis gene Bax showed even higher nociceptive and spontaneous pain behavior. 64,66 Although the in vitro model used in our study offers the advantage of studying the sole effect of hypoxia (excluding other factors), in vivo characterization of DRG neuronal death (necrosis/apoptosis ratio) in IVD degeneration requires further investigations.
Screening using a cell line can be an easy and useful tool for preliminary evaluation. The ND7/23 is a hybrid line produced by PEGmediated cell fusion of mouse neuroblastoma (N18tg2) and rat DRG neurons that has been used to study nociception and neuronal outgrowth. 30,31 In the present study, the same trends of neuronal outgrowth and viability were found in the cell line and primary neurons.
Although the cell line was immortalized by a fusion with neuroblastoma, it may still function as a useful tool for preliminary screening provided proper validation using primary DRG and tissue explant culture.
Multiple mechanisms may have been involved in IVD neo-innervation and chronic pain, such as inflammation 56 and matrix degradation 67 in the affected tissue. In fact, these factors are closely linked to each other. For example, hypoxia is a consequence and a distinctive feature of inflammation and can itself induce inflammation. 68 These findings highlight the hypoxic microenvironment as a potential target in discogenic pain treatment among the multiple mechanisms involved.

| CONCLUSIONS
In summary, neuronal outgrowth could be influenced by both neuro-

DISCLOSURE STATEMENT
The authors declare no potential conflict of interest.