Efficacy of nonviral gene transfer of human hepatocyte growth factor (HGF) against ischemic-reperfusion nerve injury in rats

Ischemic neuropathy is common in subjects with critical limb ischemia, frequently causing chronic neuropathic pain. However, neuropathic pain caused by ischemia is hard to control despite the restoration of an adequate blood flow. Here, we used a rat model of ischemic-reperfusion nerve injury (IRI) to investigate possible effects of hepatocyte growth factor (HGF) against ischemic neuropathy. Hemagglutinating virus of Japan (HVJ) liposomes containing plasmids encoded with HGF was delivered into the peripheral nervous system by retrograde axonal transport following its repeated injections into the tibialis anterior muscle in the right hindlimb. First HGF gene transfer was done immediately after IRI, and repeated at 1, 2 and 3 weeks later. Rats with IRI exhibited pronounced mechanical allodynia and thermal hyperalgesia, decreased blood flow and skin temperature, and lowered thresholds of plantar stimuli in the hind paw. These were all significantly improved by HGF gene transfer, as also were sciatic nerve conduction velocity and muscle action potential amplitudes. Histologically, HGF gene transfer resulted in a significant increase of endoneurial microvessels in sciatic and tibial nerves and promoted nerve regeneration which were confirmed by morphometric analysis. Neovascularization was observed in the contralateral side of peripheral nerves as well. In addition, IRI elevated mRNA levels of P2X3 and P2Y1 receptors, and transient receptor potential vanilloid receptor subtype 1 (TRPV1) in sciatic nerves, dorsal root ganglia and spinal cord, and these elevated levels were inhibited by HGF gene transfer. In conclusion, HGF gene transfer is a potent candidate for treatment of acute ischemic neuropathy caused by reperfusion injury, because of robust angiogenesis and enhanced nerve regeneration.

Introduction Ischemic neuropathy is common among patients with lower extremity peripheral artery disease [1,2]. The incidence of critical limb ischemia caused by peripheral artery disease has been estimated to be approximately 500-1000 per million per year [3][4][5][6]. Clinical neurological deficits have been reported to occur in 22-88% of critical limb ischemia cases [7]. Sensory symptoms, such as numbness, painful paresthesia and burning [8,9], may last for a long time in spite of reestablishing an adequate blood flow [10,11]. The incidence of ischemic neuropathy may be expected to increase in line with the current and projected increases in the elderly and obese populations. However, effective therapies for ischemic neuropathy are not established.
Gene transfer via retrograde axonal transport is a possible way of expressing identified transgenes at targeted locations within the nervous system. Although the molecular mechanism of retrograde transport of HGF is not fully understood, we have achieved successful gene transfer into the rat peripheral nerves via retrograde axonal transport from intramuscular injections with either (a) with hemagglutinating virus of Japan (HVJ)-liposomes containing plasmids encoded with luciferase or ß-galactosidase, or (b) with HVJ envelope encoded with hepatocyte growth factor (HGF) [12,13].
In the search for an agent to improve the symptoms of ischemic neuropathy, HGF is a potent candidate because of its powerful neurotrophic and angiogenic properties. We have investigated whether nonviral retrograde gene transfer of HGF might improve ischemic neuropathy using the rat model of ischemic-reperfusion nerve injury (IRI).

Rats
All animal procedures were carried out in compliance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Committee on the Ethics of Animal Experiments of the National Defense Medical College (Protocol Number: 07021). One hundred-eight male Wistar rats, weighing 220-280 g, age 8-9 weeks were analyzed in this study. Animals were housed in a temperature-controlled room with a 12-hour light-dark cycle. During the procedure, all efforts were made to minimize suffering, to minimize the number of animals, to utilize alternative techniques, if available, to relieve pain and /or distress as necessary, and to increase the level of monitoring if rats exhibit the sign of morbidity. Rats were euthanized by CO 2 inhalation followed by cervical dislocation.

Study design
The method used to induce IRI was as described by Nukada and McMorran [36]. Briefly, in rats under sodium pentobarbital anesthesia the major arteries supplying the right hindlimb; abdominal aorta, common iliac, femoral, and superficial circumflex iliac arteries, were exposed and occluded for 4 h using microvascular clips (TKS-1, 40g; Bear Medic Corporation, Ichikawa, Japan), and then reperfused.
Immediately after reperfusion, HVJ envelopes (100 μl) containing 100 μg human HGF plasmid DNA, were injected into the proximal one-third of the tibialis anterior muscle of the right hindlimb over a few minutes via a 27-gauge needle. HGF gene transfer was repeated at 1, 2 and 3 weeks after IRI. Rats were divided randomly into 4 groups as follows: (1) IRI with HGF gene transfer (IRI+HGF group), (2) IRI without HGF gene transfer (IRI group), (3) similar surgery to expose arteries but no IRI (sham group), and (4) no surgical procedures (control group).

Construction of plasmid DNA and preparation of the HVJ envelopecomplex vector
To produce an HGF expression vector, human HGF cDNA (2.2 kb) was inserted between the NotI sites of a pcDNA 3.1 (−) vector. The HVJ envelope-complex vector was prepared using an HVJ envelope vector kit (GenomOne-Neo; Ishihara Sangyo, Osaka, Japan) as described previously [13].
In our previous studies, we have demonstrated that HGF expression in ipsilateral sciatic nerve and dorsal root ganglia (DRG) was high on day 3 and decline on Day 7 after the first gene transfer, and was high again day 3 after the second gene transfer [29]. We have also confirmed the specific primers we used against rat HGF mRNA did not cross-react with human HGF mRNA [29].

Behavioral studies
Mechanical allodynia and thermal hyperalgesia of the right hind paw were measured at 14 defined time-points until 8 weeks post-operation (1, 2, 3, 5, 7, 10, 14, 17, 21, 24 days, and 4, 5, 6, 7 weeks). Methods have been reported previously [29,37,38]. In brief, mechanical allodynia was determined by quantifying the withdrawal threshold of the soles of right hind paw in response to Von Frey hairs. Von Frey hairs were applied perpendicularly to the plantar surface of the hind paw, sufficient force being used to bend the filament. Brisk withdrawal or paw flinching was taken as a positive response.
Thermal hyperalgesia was assessed using a Hargreaves apparatus (Plantar Test; Ugo Basile, Italy). A radiant heat stimulus was applied to the midplantar area of the hind paw, and the time between initial heat and paw withdrawal was recorded. Each testing for mechanical allodynia and thermal hyperalgesia was repeated five times, and the mean value was used.

Laser Doppler flux and thermographic studies of the hind paw
Blood flow and thermographic measurements in the hind paw were measured at baseline and 1, 3, 6 and 8 weeks after IRI, as in our previous study [29]. Blood flow was using a laser Doppler flowmeter (PIM II: Perimed AB, Järfälli, Sweden), and thermographic measurements was performed using a TH9100 ML camera (NEC Sanei, Tokyo, Japan).

Neurometer measurement of current stimulus threshold
The stimulus thresholds were measured using a Neurometer according to the method of Kiso et al. [39]. Briefly, a small electrode (GT100-30; Neurotron Inc., Baltimore, USA) for stimulation was attached to the plantar surface of the hind paw. A skin patch dispersion electrode (TE174D; Fukuda Denshi, Tokyo, Japan) was attached to the lower back of the rat. In rats lightly restrained using Ballman cages (Natsume, Tokyo, Japan), transcutaneous nerve stimuli -each comprising three sine-wave pulses at 5, 250, or 2,000 Hz-were applied to the plantar surface, using an animal response test mode of the Neurometer Current Perception Threshold (CPT) (Neurotron Inc., Baltimore, MD) [40]. Delivery of sinusoidal electrical stimuli at 5, 250, or 2,000 Hz allows evaluation of responses primarily in small unmyelinated (C), small myelinated (Aδ), and large myelinated (Aβ) fibers, respectively [41][42][43][44][45]. The minimum current intensity (mA) at which each rat vocalized was defined as the CPT. The data presented herein were obtained from three consecutive measurements, and the CRT was measured before surgery and at 1, 3, 6 and 8 weeks after surgery.

Nerve conduction study
Rat was on a heating pad, and the right hindlimb skin temperature was kept 35 ± 1ºC. The right sciatic nerve was exposed from the sciatic notch to the knee under general anesthesia with sodium pentobarbital. The nerve was stimulated using hook electrodes at the sciatic notch and knee. Recording needle electrodes were placed in the 1st and 4th interosseus muscles of the foot. Motor nerve conduction velocity (MNCV) and compound motor action potential (CMAP) were recorded using Neuropack instrumentation (Nihon Koden, Tokyo, Japan) as described previously [46]. Nerve conduction studies of the right sciatic nerve were carried out before surgery, and at 1, 3, 6, and 8 weeks after surgery.

Histological studies of the sciatic nerve
After measurements of MNCV and CMAP at 3, 6 and 8 weeks post-surgery, bilateral sciatic and tibial nerves were taken from the upper-thigh to ankle level under sodium pentobarbital anesthesia (n = 6 at each time-point for each group) [36]. All specimens were fixed overnight in 2.5% glutaraldehyde in 0.1 M phosphate buffer solution (pH 7.4). Subsequently, they were postfixed with 1% osmium tetroxide, dehydrated, and cut into consecutive 2-3 mm blocks before embedding in epoxy resin. Semi-thin sections (1 μm in thickness) were stained with methylene blue. Ultra-thin sections were stained with EM stainer followed by lead citrate and examined with a transmission electron microscope (H7600, Hitachi, Tokyo, Japan). Morphometric analysis of the number and diameter of myelinated nerve fibers and the number of endoneurial microvessels in sciatic and tibial nerves, was determined using a NIH image analysis.

RT-PCR
The mRNA expressions for P2X2, P2X3, and P2X4 receptors, and TRPV1 (transient receptor potential vanilloid receptor subtype 1) in right sciatic nerve, right DRG (L5), and spinal cord were examined at 1 and 3 weeks post-operatively by semiquantitative RT-PCR as described previously [29]. Sequence information used in the current study is listed in S1 Table.

Statistical analysis
Values for each parameter are expressed as means ± SD. Fisher's protected least-significantdifference test was applied to the data when significant F-ratios were obtained in an analysis of variance (ANOVA), and significance was set at p<0.05. The statistical significance of the difference between curves representing repeated measures of behavioral function (allodynia or hyperalgesia) was assessed using a repeated measures analysis (SPSS Inc., Chicago, IL, USA).

Mechanical and thermal thresholds
On day 1 post-surgery, right hind paw withdrawal threshold using Von Frey hairs in both IRI +HGF and IRI groups was significantly decreased when compared with pre-surgical levels, and was also significantly less than in control and sham groups (Fig 1A). The mechanical threshold in IRI+HGF group became significantly greater than in IRI group on week 1 postsurgery, and thereafter increased with time ( Fig 1A). The recovery of threshold in IRI+HGF group was faster than in IRI group, and at week 5 post-surgery, threshold in IRI+HGF group became not significantly different from those in controls. The mechanical threshold of left hind paw in IRI+HGF and IRI groups was reduced, although statistically not significant. Raw data of these thresholds of bilateral hind paws in IRI+HGF, IRI, control and sham groups were shown in S2 Table. Thermal threshold was significantly reduced at day 1 post-surgery in both IRI and IRI +HGF groups when compared with those in control and sham groups ( Fig 1B). However, the recovery of threshold was much faster in IRI+HGF group than in IRI group. At week 2 postsurgery, there was no significant differences of thresholds between IRI+HGF group and controls. Thermal threshold in IRI+HGF group was significantly greater from week 2 to 6 postsurgery than in IRI group ( Fig 1B). There were no significant changes of behavioral studies in the control and sham groups.

Laser Doppler flux and thermographic studies
Laser Doppler flux of the right hind paw in IRI+HGF group was not significantly different at all time-points from those in controls. Blood flow in IRI group was significantly decreased at weeks 1 and 3 post-surgery than those in IRI+HGF group (Fig 2). Blood flow in the left hind paw was not significantly different between experimental groups (Fig 2). Skin temperature of the right hind paw in IRI+HGF group was significantly greater than in IRI group at weeks 1, 3, and 6 post-surgery, although it was reduced in both IRI+HGF and IRI groups when compared with pre-surgical levels (Fig 3). Skin temperature in the left hind paw of both IRI+HGF and IRI groups was significantly reduced at week 1 when compared with pre-surgical level (Fig 3). In control and sham groups, there were no significant changes of blood flow and thermographic measurements.

Neurometer measurement of current stimulus threshold
Regardless of whether stimuli were delivered at 5, 250, or 2,000 Hz, the threshold at the right hind paw was significantly reduced in both IRI+HGF and IRI groups when compared with pre-surgical level. However, the threshold in IRI+HGF group was significantly greater than in IRI group at 1, 3, 6 and 8 post-surgery (Fig 4). In the left hind paw, thresholds at all current stimuli were lowered in both IRI+HGF and IRI groups at 1 week post-surgery, and at 2,000 Hz stimulus, thresholds in IRI+HGF group were significantly greater than in IRI group at weeks 1 and 3 (S1 Fig). In control and sham groups, no significant changes have been observed.

Nerve conduction study
MNCVs and CMAP of the right sciatic nerves were significantly reduced in both IRI+HGF and IRI groups when compared with pre-surgical levels ( Fig 5). However, MNCVs in IRI +HGF group were significantly faster than in IRI group at weeks 3 and 6 post-surgery ( Fig  5A). CMAPs amplitude in IRI+HGF group was significantly greater than in IRI group (Fig 5B) at week 3. Recovery of slowed MNCVs and lowered CMAP amplitudes were faster in IRI +HGF group than in IRI group, correlated with the recovery of foot and toe drop.

Histological study of the sciatic nerve
Semi-thin transverse sections at the lower-thigh level of right sciatic nerve obtained at week 3 post-surgery revealed a loss of large myelinated nerve fibers in both IRI+HGF and IRI groups, although the number of large myelinated nerve fibers was apparently greater in IRI+HGF group than in IRI group (Fig 6). Small myelinated nerve fibers observed in both groups were mainly regenerating nerve fibers. Nerve fibers with disproportionately thin myelin relative to axon area were observed (S2 Fig), suggesting regenerating nerve fibers. Nerve fibers with axonal degeneration were scattered in both IRI and IRI+HGF groups, although more prominent in IRI group (Fig 6). Unmyelinated nerve fibers were apparently normal under the electron microscopy ( S2 Fig). Endoneurial edema and vascular changes were not apparent.

PLOS ONE
Morphometric analysis of myelinated nerve fibers at the lower-thigh level of right sciatic nerve at 3weeks after surgery revealed that the number of myelinated nerve fibers, and mean diameter and area of myelinated nerve axons were significantly less in both IRI+HGF and IRI groups than in controls (Table 1). However, axon area and diameter were significantly greater in IRI+HGF group that in IRI group ( Table 1). The diameter histogram of myelinated nerve fibers confirmed an increased number of small myelinated nerve fibers in both IRI+HHGF and IRI groups (S3 Fig). Peaks of axon diameters in both groups occurred at around 1.5μm, while the peak of controls was approximately 3.5μm.
In sciatic nerves at the lower-thigh level, fascicular area at 3 weeks after surgery was not significantly between experimental groups (Fig 7 and S3 Table). The density of endoneruial microvessels in both right and left sciatic nerves was significantly greater in IRI+HGF than in IRI, sham and control groups (Fig 7, S4 Fig and S3 Table). Transverse semi-thin sections of the contralateral (left) sciatic nerves at the upper-thigh level in IRI+HGF and IRI groups were showed in S5 Fig to highlight a contrast of the number of endoneurial microvessels. A significant increased density of endoneruial microvessels in IRI+HGF group was observed in bilateral tibial nerves as well ( S6 Fig and S4 Table).

RT-PCR
We detected mRNA expressions for P2X2, P2X3, P2X4, and P2Y1 receptors, and for TRPV1, in right sciatic nerve, right DRG (L5) and spinal cord at weeks 1 and 3 post-surgery. We calculated ratios for the expression levels of the various receptor mRNA to those of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA. In sciatic nerves, ratios obtained for P2X3 receptor mRNA, P2Y1 receptor mRNA, and TRPV1 mRNA in IRI group were all significantly greater than in IRI+HGF and control groups at week 1 (Fig 8). The ratios of P2X3 and P2Y1  mRNAs at week 3 were not significantly different between three groups. Ratios obtained for TRPV1 mRNA at weeks 1 and 3 in IRI+HGF group were significantly less when compared with those in IRI group, although were significantly greater than in control group (Fig 8). No PLOS ONE significant differences were found among 3 groups in the ratios obtained for P2X2 receptor mRNA and P2X4 receptor mRNA (Fig 8). In DRG, the mRNA levels were measured by mixing together all the specimens from 6 rats at each time-point, because of the small size of rat DRG. The ratios obtained for P2X3 and P2Y1 receptors and TRPV1 mRNAs in DRG at weeks 1 and 3 showed a similar trend as those of sciatic nerves; lower ratios in IRI+HGF than in IRI group (S7 Fig). In spinal cord, the ratio for TRPV1 mRNA in IRI+HGF group was significantly less when compared with those in IRI group at weeks 1 and 3, although mRNA expressions of P2X2, P2X3, P2X4 and P2Y1 receptors were not significantly different between three groups (Fig 8).

Discussion
We have investigated whether nonviral retrograde gene transfer of HGF might improve symptoms of ischemic neuropathy using a rat model of IRI, and demonstrated that HGF gene transfer resulted in 1) significant improvement of mechanical allodynia and thermal hyperalgesia, decreased skin blood flow and temperature, lowered plantar thresholds of electrical stimuli in the hind paw, and sciatic nerve conduction parameters, 2) enhanced nerve regeneration in the sciatic nerve histologically, 3) significant increases of the density of endoneurial microvessels in bilateral sciatic and tibial nerves, and 4) reduction of the elevated levels of P2X3 and P2Y1 receptor mRNA and TRPV1 mRNA in the sciatic nerve. These results suggest that repeated retrograde gene transfer of HGF may improve symptoms of ischemic neuropathy. Several well-defined underlying molecular mechanisms of retrograde transport of neurotrophic factors have been identified [47], and the efficacy of intramuscular injection over the calf muscle of HGF has been demonstrated. In our previous study, we demonstrated expression of human HGF-like immunoreactive protein in the rat sciatic nerve by repeated intramuscular injections of nonviral gene transfer of human HGF into the hindlimb muscle [29]. In fact, in the DRG and sciatic nerve at 3 days after the first and second HGF gene transfers, the levels of rat HGF mRNA and protein were increased 1.2-to 2.5-fold when compared with those in controls. We have also shown the efficacy of HGF upon neuropathic pain in a rat model of CCI [29]. Intramuscular injections of HGF have also been successful for treatment of models of streptozocin-induced diabetes [48] and crush injury [27]. Clinically, intramuscular administration of VM202, plasmid DNA expressing two isoforms of HGF, resulted in significant symptomatic relief in painful diabetic neuropathy [49]. In a double-blind, placebo-controlled study, Kessler at al. confirmed that repeated intramuscular injections of HGF are effective in reducing pain in patients with painful diabetic neuropathy [50].
There have been electrophysiological and histological studies of the ischemic neuropathy caused by IRI [36,[51][52][53]. However, the mechanism causing sensory symptoms, e.g. numbness, paresthesia, burning and pain, is still not completely understood. In this study, we confirmed that mechanical allodynia and thermal hyperalgesia were evident within 1 day after IRI and persisted for at least 6 weeks. Repeated HGF gene transfer was successful in enhanced recover of these mechanical allodynia and thermal hyperalgesia after IRI. Using an animal response test mode of the Neurometer CPT, we also found that all nerve fibers including small unmyelinated (C), small myelinated (Aδ), and large myelinated (Aβ) fibers, were adversely affected after IRI and repeated HGF gene transfer led to an early full recovery to the pre-surgical level in all three fiber types.
By analyzing the morphometrical parameters of endoneurial microvessels and myelinated nerve fibers in sciatic nerves at 3 weeks after surgery, we found that 1) the number of endoneurial microvessels was significantly increased in IRI+HGF group than in IRI, control and sham groups, and 2) axon area and diameter of myelinated nerve fibers in IRI+HGF group were significantly greater when compared with those in IRI group. Neovascularization by HGF gene transfer enhanced nerve regeneration, and an increased axon area in IRI+HGF group was due to the presence of intensified regenerating nerve fibers. Because ischemia causes delayed nerve regeneration [54,55], regenerating nerve fibers in IRI group were associated with a decrease in axon area.
A significant increased density of endoneurial vessels was observed in the contralateral sciatic and tibial nerves as well in IRI+HGF group. Because the transient ligation of abdominal aorta, at the level of distal to left iliolumbar artery and proximal to right illiolumar artery, during an ischemic surgery, transient ischemia at the mid-thigh level of sciatic nerve which is the watershed zone along the length of sciatic-tibial nerves [56,57], occurred in the contralateral hind limb [36,53]. Other results in the current study including behavioral studies, skin blood flow and temperature, and Neurometer measurements, also suggested mild ischemia in the contralateral side. Because s, the potent angiogenic property of HGF spread into the contralateral hind limb.
Levels of P2X3 and P2Y1 receptors, and TRPV1 mRNAs in sciatic nerves and DRG were elevated at 1 week after IRI, and these elevations were depressed by repeated HGF gene transfer. It is known that the P2X3 receptor is expressed by unmyelinated sensory nerve fibers and mediates thermal hyperalgesia, whereas P2Y1 receptors are expressed by myelinated sensory nerve fibers and mediate mechanical allodynia [58][59][60]. TRPV1 is essential for tissue injury-induced thermal hyperalgesia [61]. Our results suggest when HGF protein is induced via nonviral retrograde gene transfer, it may act directly on the nervous system to exert effects leading to improvements in mechanical and thermal thresholds [16,[22][23][24][25]42]. Intriguingly, we failed to find increases in P2X2 or P2X4 mRNA levels after IRI. P2X2/3 heteromerics show a relation to mechanical allodynia induction [62]. Tsuda et al. reported that mechanical allodynia requires activation of the P2X4 receptor, and nerve injury-induced pain depends upon ongoing signaling via the P2X4 receptor [59]. Indeed, in our previous study using a rat model of CCI, we demonstrated preventive effect of repeated HGF gene transfer on an induction of mechanical allodynia and elevations in the levels of P2X2, P2X4, and P2Y1 receptor mRNAs [29].

Conclusion
We have demonstrated that transfer of the human HGF gene into the rat nervous system by repeated intramuscular injection of HVJ envelope resulted in potent angiogenesis and subsequently enhanced nerve regeneration in ischemic nerves. These results support HGF transfer as a promising candidate for treatment of acute ischemic nerve injury caused by reperfusion injury.