Engineered Zinc Finger Protein–Mediated VEGF-A Activation Restores Deficient VEGF-A in Sensory Neurons in Experimental Diabetes

All studies and procedures were licensed under the U.K. Animals (Scientific Procedures) Act 1986. Diabetes was induced in adult male Wistar rats (250–300 g; Charles River, Margate, U.K.) via intraperitoneal injection of freshly dissolved STZ (55 mg/kg in sterile saline), administered after an overnight fast. Diabetes was verified 3 days after STZ injection using a strip-operated reflectance photometer (OptimumPlus; MediSense, Whitney, U.K.) to measure the animal's tail vein blood glucose levels. Rats with blood glucose less than 15 mmol/l were excluded from the study. Age- and weight-matched rats were used as nondiabetic controls. The VEGF-A activating ZFP-TF VZ+434 has previously been described (22). Plasmids were formulated at a concentration of 2 mg/ml in 5% poloxamer 188 (BASF, Florham, NJ), 150 mmol/l NaCl, and 2 mmol/l Tris (pH 8.0; Sangamo BioSciences).

Three weeks after induction of diabetes, control and STZ-induced diabetic rats were anesthetized with isoflurane (2% in oxygen) and were either injected with a total of 250 μg VZ+434 at two sites in their left gastrocnemius/soleus muscle, or left untreated.

After injection of VZ+434, additional groups of treated and untreated diabetic rats also had their left sciatic nerve exposed and ligated with polypropylene monofilament nonabsorbable sutures (Prolene; Ethicon) at mid-thigh level, proximal to the site of injection. The wound was closed and animals recovered under observation. Rats were killed at selected time points after injection with VZ+434 (1, 2, 7, and 14 days), or for the sciatic nerve ligation studies at 1 day after ligation.

Ipsilateral and contralateral muscle, sciatic nerves, and lumbar (L)4/5 DRG were removed. Tissue designated for Western blot analysis was snap frozen in liquid nitrogen and stored at −40°C until processing. Sciatic nerve and DRG samples destined for immunohistochemical studies were postfixed in 4% paraformaldehyde overnight at 4°C, then cryoprotected in 30% sucrose in 0.1 M phosphate buffer for 24 h at 4°C. Tissues were then frozen in OCT embedding medium (VWR, Lutterworth, U.K.) and stored at −40°C until processing.

For sensory testing experiments, STZ-induced diabetic rats were either untreated or treated with 250 μg VZ+434 at 4, 6, and 8 weeks after STZ injection, in two sites in their left gastrocnemius/soleus muscle (n = 6–10 per group). Behavioral responses of STZ-induced diabetic rats and age-matched control rats to tactile stimulation were measured at baseline (before STZ injection), and at 4, 8, and 12 weeks after STZ injection.

Western blotting was conducted as previously described (25). Blots were incubated overnight at 4°C with primary antibodies: anti–VEGF-A (1:500; Santa Cruz Biotechnology), anti–total extracellular signal–related kinase (ERK; 1:2000; Cell Signaling Technology, Danvers, MA), anti–phospho AKT and anti–total AKT (1:2000; Cell Signaling Technology), and anti–βIII-tubulin (1:1,000; Sigma). Blots were washed and incubated in horseradish peroxidase–linked anti-rabbit IgG or anti-mouse IgG (1:2000; Cell Signaling Technology) for 1 h at room temperature. Protein bands were visualized using LumiGLO and Peroxidase Reagent (Cell Signaling Technology). The films were scanned, and the total pixel intensity for each band was calculated using SigmaScan Pro5 software (SPSS, Chicago, IL). Graphs and statistical analyses were prepared using GraphPad PRISM 4 software. Data were analyzed using either one-way ANOVA followed by either the Dunnett post hoc test, Student paired t test, or Student one-sample t test as appropriate. Data are expressed as mean ± SD.

Longitudinal sections (12-μm) of sciatic nerve or L4/5 DRG were thaw-mounted onto Superfrost Plus Slides (VWR). Sections were washed with PBS, and then incubated in blocking buffer (10% normal donkey serum, 0.2% Triton-X₁₀₀ in PBS) for 1 h at room temperature. Slides were incubated overnight at either 4°C or room temperature with primary antibodies (in blocking buffer): anti–VEGF-A (1:500; Santa Cruz Biotechnology), anti–calcitonin gene-related peptide (CGRP; 1:500; Sigma), anti–neurofilament 200 (NF200; 1:400; Chemicon), anti-S100β (1:500; Sigma), or anti–phospho (p) AKT (1:100; Promega). Slides were washed and then incubated with either cyanin 3–conjugated donkey anti-rabbit and/or fluorescein isothiocyanate–conjugated donkey anti-mouse IgG as appropriate (1:200; Jackson ImmunoResearch) for 1 h at room temperature. In addition some slides were colabeled with fluorescein isothiocyanate–conjugated isolectin B4 from Griffonia simplicifolia (20 μg/ml; Vector Labs, Peterborough, U.K.). Sections were mounted in Vectorshield containing DAPI (Vector Labs). Immunofluorescence was visualized using a Leica MPS60 DMR fluorescent microscope. Digital images were acquired using WASABI software (Hamamatsu Photonics, Herts, U.K.) and processed using Adobe Photoshop CS (Adobe Systems).

The threshold for high VEGF-A immunoreactivity (IR) was determined by calculating the mean staining intensity of four neurons per animal that were deemed to be positively stained (SigmaScan Pro). This mean intensity was then taken as the threshold value to determine the percentage of VEGF-A-IR neuronal profiles. All neuronal profiles were traced and mean VEGF-A-IR and Feret diameter of each profile were automatically calculated (SigmaScan Pro) and data were used for cell size distribution analysis. Cell profiles from four randomly selected sections (each 100-μm apart) of ipsilateral and contralateral L4/5 DRG (processed on the same slide) from four animals in each treatment group were measured. The method of recursive translation was applied to these results to correct for overestimation of both large and small-diameter cell profiles (26).

VZ+434 plasmid DNA levels were measured in duplicate reactions containing 100 ng of DNA and TaqMan Universal PCR Mix (Applied Biosystems, Foster City, CA). Reactions were analyzed using an ABI 7300 Real Time PCR System (Applied Biosystems). Plasmid copy numbers (PCNs) were determined from a standard curve prepared from a 10-fold serial dilution of VZ+434. The background PCN level was determined from the mean of VZ+434 levels in tissues from the right (contralateral) side plus 2 SD.

At 1 week prior to, and at 4, 8, and 12 weeks after induction of diabetes, the behavioral response to mechanical stimulation of the hind paw was assessed using a dynamic plantar aesthesiometer (Ugo Basile, Comerio, Italy). Paw withdrawal responses were measured in both right and left hind paws in all animals, and are reported as the mean of three measurements per paw, taken at least 10 min apart. Testing was repeated over two consecutive days, with results recorded from the second day of testing. The assessor was blinded to the VZ+434-treated and untreated STZ groups.

Using Western blot and immunocytochemical analysis, we compared the expression of VEGF-A in L4/5 DRG of control rats and rats 3 weeks after induction of STZ-induced diabetes (see Table 1 for details of blood glucose levels and body weight). VEGF-A levels were significantly reduced in L4/5 DRG from rats 3 weeks after STZ injection (control rats: 16.6 ± 3.2 versus diabetic rats: 3.9 ± 1.6 arbitrary units [AU], P < 0.0001, Fig. 1 A and B).

VEGF-A-IR was clearly detected within the cytoplasm of a population of small- to medium-diameter sensory neurons in the DRG of control rats (Fig. 1,C, arrows) and also in motor neurons in the ventral horn of the spinal cord (data not shown). Fewer VEGF-A-IR neurons were observed in DRG from STZ-induced diabetic rats (Fig. 1 D and E, control rats: 6.5 ± 2.7% VEGF-A-IR neurons versus diabetic rat: 1.9 ± 2.2% VEGF-A-IR neurons; P < 0.01). Immunofluorescence was abolished by omission of the primary antibody, indicating the VEGF-A-IR to be primary-antibody specific (data not shown).

A single intramuscular dose of the VEGF-A–activating zinc finger protein transcription factor VZ+434 resulted in a rapid and significant increase in VEGF-A protein expression in ipsilateral L4/5 DRG (Fig. 2). This increase was evident as early as 24 h after injection (left: 5.2 ± 1.2 versus right: 1.0 ± 0.2 AU, P < 0.05); VEGF-A levels were normalized by 1 week after injection (Fig. 2,A and B). VEGF-A-IR was detected in contralateral (Fig. 2,C) and ipsilateral (Fig. 2,D) L4/5 DRG within a population of small- to medium-diameter sensory neurons (Fig. 2,C and D; arrows). Intramuscular treatment with VZ+434 generated a significant increase in VEGF-IR in small-diameter neurons in the ipsilateral compared with contralateral DRG. VEGF-A-IR was significantly elevated in neurons with diameters between 18.75 and 22.5 μm compared with contralateral DRG (Fig. 2 E and F, P < 0.05).

To further characterize the population of VEGF-A-IR neurons in ipsilateral L4/5 DRG, we performed dual immunofluorescence experiments with anti–VEGF-A (Fig. 3,A, C, and E) and three neurochemical phenotypic markers: CGRP (Fig. 3,B), to label small-diameter nociceptive peptidergic neurons; isolectin B4 (IB4; Fig. 3,D), a marker for nonpeptidergic small-diameter sensory neurons; and NF200 (Fig. 3,F), a heavy chain neurofilament protein that labels predominantly large-diameter proprioceptive and mechanoreceptive sensory neurons. VEGF-A-IR was observed predominantly in the CGRP-IR and IB4-positive populations, whereas very few VEGF-A-IR neurons colocalized with NF200-IR (Fig. 3 E and F).

To investigate the mechanism by which intramuscular VZ+434 normalized VEGF levels in the DRG of diabetic rats, we measured VZ+434 DNA in ipsilateral and contralateral muscle and L4/5 DRG samples from diabetic rats, 24 h after a single intramuscular dose of VZ+434 (Fig. 4).

The background PCN of VZ+434 was set at 2,700 (derived from mean plus 2 SD from contralateral tissue samples). Ipsilateral muscle contained significantly more copies of VZ+434 than contralateral muscle (mean values: left 407168 vs. right 993 copies/100 ng DNA, P < 0.05, Mann-Whitney U test, Fig. 4). No significant left-right tissue difference in VZ+434 levels in DRG samples was observed (Fig. 4), suggesting that plasmid uptake is restricted to the site of injection, resulting in upregulation of VEGF-A in the muscle, consistent with previous observations (27). Importantly, the results demonstrate that axonal transport of VZ+434 plasmid is not responsible for the increased VEGF-A levels in the DRG.

Axonal transport of VEGF-A protein in the sciatic nerve of diabetic rats was assessed using immunocytochemistry and Western blotting techniques (Fig. 5). Ligation of the sciatic nerve of diabetic rats injected distally with VZ+434 caused an increase in VEGF-A-IR both proximal and, notably, distal to the ligature site (Fig. 5,A). VEGF-A-IR was clearly observed in axonal profiles; a region is highlighted and shown as inset (Fig. 5,A and B, arrows). VEGF-A-IR did not colocalize with S100-IR Schwann cells (Fig. 5 C).

Ligation of the sciatic nerve of untreated diabetic rats showed very little axonal accumulation of VEGF-A-IR (Fig. 5,E), restricted to the regions of nerve immediately adjacent to the ligature, a marked contrast to the wide distribution of VEGF-A-IR in the VZ+434-injected rats, particularly throughout the distal nerve (Fig. 5,A, arrows). VEGF-A-IR was absent from the contralateral nerve (Fig. 5 D).

Ligation of the sciatic nerve of control rats (nondiabetic), either injected with VZ+434 (Fig. 5,G) or left untreated (Fig. 5 F), revealed an accumulation of VEGF-A-IR proximal and distal to the ligature site.

These observations were confirmed by Western blot analysis of VEGF-A protein levels in nerve segments taken proximal and distal to the ligature (schematic diagram illustrated in Fig. 5,H). Protein levels were quantified by densitometric analysis. Increased VEGF-A was observed particularly in all distal segments and segments immediately proximal to the ligature site in diabetic rats injected with VZ+434 (Fig. 5,I). In contrast, sciatic nerve samples prepared from control rats injected with VZ+434 showed a more even distribution of VEGF-A throughout the whole nerve (Fig. 5,J), with the greatest accumulation observed in the proximal segment adjacent to the ligature (Fig. 5 J). Together these data demonstrate that VEGF-A undergoes axonal transport to the DRG and that intramuscular administration of VZ+434 is sufficient to overcome the deficit in VEGF-A levels observed in this tissue in this STZ model of diabetes.

Diabetes was associated with a significant decrease in basal VEGF-A protein levels in the DRG (Fig. 1 B and supplementary Fig. 1A and B, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-1526/DC1), and VZ+434 administration restored VEGF-A to control levels (supplementary Fig. 1A and B). However, this increase was markedly reduced by ligation of the sciatic nerve in three of five animals tested (mean change across all five animals was not significant) (supplementary Fig. 1A and B). Taken together with the absence of plasmid DNA encoding VZ+434 in the DRG, these data suggest that the observed normalization of VEGF-A levels in the DRG after VZ+434 treatment occurs via retrograde axonal transport of VEGF-A protein in the sciatic nerve.

To analyze the functional consequences of elevated VEGF-A levels in the DRG, we investigated intracellular signaling pathways potentially responsible for the VEGF-A–mediated neuroprotective effects in diabetic neuropathy. Specifically, we investigated activation of p38 MAPK, ERK1/2, and PI3-K pathways using phospho-specific antibodies. Neither the levels nor the phosphorylation status of the MAPKs p38 and ERK1/2 was significantly altered in the DRG of VZ+434-injected diabetic rats at any time point studied (e.g., 24 h after injection: p38: ipsilateral 1.2 ± 0.51 versus contralateral 1.0 ± 0.73 AU, P > 0.05, n = 4; ERK1/2: ipsilateral 1.56 ± 0.25 versus contralateral 1.0 ± 0.34 AU, P > 0.05, n = 4).

In contrast, an increase in AKT phosphorylation was observed in the ipsilateral L4/5 DRG 48 h after VZ+434 injection (left: 5.4 ± 2.6 versus right: 1.0 ± 0.4 AU, P < 0.05, n = 4), and activation was maintained for 1 week (left: 2.0 ± 0.6 versus right: 1.0 ± 0.6 AU, P < 0.01, n = 4; Fig. 6,A and B). This increase in pAKT in the ipsilateral compared with contralateral DRG was confirmed using immunocytochemical analysis (Fig. 6,C–F). Ipsilateral and contralateral DRG were colabeled with antibodies against VEGF-A (Fig. 6,C and E) and pAKT (Fig. 6,D and F). A threefold increase in neurons expressing pAKT-IR was observed in the ipsilateral DRG compared with the contralateral DRG (Fig. 6 D and F arrows; left: 10% pAKT-IR neurons versus right 3% pAKT-IR neurons, P < 0.05). Furthermore, of the pAKT-IR neurons 61% coexpressed VEGF-IR and of the VEGF-IR neurons 85% coexpressed pAKT-IR. Because pAKT-IR is observed in a greater number of neurons than those possessing VEGF-IR, this may suggest a paracrine action of VZ+434-generated and -transported VEGF-A within the DRG.

To determine whether the VZ+434-mediated increase of VEGF-A in small- to medium-diameter neurons has protective effects in experimental diabetic neuropathy, we investigated the functional efficacy of VZ+434 on the development of mechanical hypersensitivity and/or hyposensitivity in STZ-induced diabetic rats (Fig. 7).

Eight weeks after the induction of diabetes, hind paw thresholds to mechanical stimulation were significantly reduced compared with pre-STZ thresholds (Fig. 7,A) and age-matched control rats (Fig. 7,B). Similarly, paw withdrawal response thresholds from the contralateral hind limb of STZ-induced diabetic rats treated with VZ+434 were reduced compared with both pre-STZ values (Fig. 7,A) and age-matched control rats (Fig. 7,C). However, VZ+434 treatments provided significant protection against mechanical allodynia in the left hind paw 8 weeks after STZ injection (Fig. 7 A and D). This suggests that the VZ+434-mediated increase of VEGF-A in the DRG protects against the mechanical hypersensitivity observed in STZ-induced diabetes.

This study demonstrates that VEGF-A protein levels in the DRG are reduced in STZ-induced diabetic rats at early stages of the disease (3 weeks after STZ injection). This deficit was corrected by intramuscular administration of VZ+434, a plasmid that encodes a zinc finger protein-transcription factor capable of simulating endogenous production of VEGF-A. There was no evidence of VZ+434 plasmid DNA being transported to, or transcribed within, the DRG, rather VEGF-A protein was axonally transported to sensory neurons in ipsilateral L4/5 DRG 24 h after injection, and was associated with altered signal transduction, via the PI3-K pathway. Treatment with VZ+434 provided unilateral protection against mechanical allodynia 8 weeks after STZ injection.

VEGF-A was localized to small- to medium-diameter sensory neurons in the DRG of control rats, confirming previous studies (2,5), and was downregulated 3 weeks after induction of diabetes. The time course of VEGF-A expression in the DRG after induction of diabetes has not been systematically characterized to date, and evidence suggests a dynamic regulation over the course of the disease. VEGF-A-IR has been reported to be upregulated in the DRG 12 weeks after STZ injection in mice (28), however we found no change in VEGF-A protein levels 12 weeks after STZ injection in rats (data not shown).

Receptors for VEGF-A have previously been characterized in sensory neurons (8,9,29). Neurons appear to lack VEGFR-1, which is largely expressed by endothelial cells (30,31). Conflicting evidence exists about the neuronal expression of VEGFR-2 in sensory neurons. Sondell et al. (2,5,6) describe a postnatal decrease in the percentage of VEGFR-2-IR neurons in the DRG, with less than 10% of sensory neurons expressing VEGFR-2 in adult mouse DRG. These VEGFR-2–positive neurons were of the large-diameter RT97-positive population and the small-diameter CGRP-IR population; no IB4-positive neurons expressed VEGFR-2 (6). In contrast, other groups did not detect VEGFR-2 in sensory neurons (31). Neuropilin-1 is highly expressed in sensory neurons (6,31,32); however, the neuropilin receptors lack defined signaling motifs and appear to act as a coreceptor for VEGFR-2.

A decrease in VEGF-A in sensory neurons may be a functional defect associated with the pathogenesis of diabetic neuropathy. Indeed, clinical studies by Quattrini et al. identified a significant reduction in VEGF-A levels and endothelial cell dysfunction, and a loss in intraepidermal nerve fibers, in skin biopsy samples taken from patients with increasing severity of diabetic neuropathy (33). Diabetic neuropathy is a heterogeneous disease with a widely varying pathology including a neurotrophin deficit. A deficit in nerve growth factor (NGF) expression by sensory neuron targets and retrograde transport to the DRG is well established in experimental diabetic neuropathy (34,–36). Indeed, the reduced availability of NGF in experimental diabetic neuropathy provided the rationale of NGF administration as a potential treatment for the disease, and normalizes key molecular and functional aspects of the neuropathy (37,38). It has not been established whether glial cell–derived neurotrophic factor production is similarly impaired in diabetes, however therapy with exogenous glial cell–derived neurotrophic factor in experimental diabetes normalized cutaneous innervation in STZ-induced diabetic mice (39,40). VEGF-A has demonstrated neurotrophic functions in both central and peripheral neurons (4,–6), and a number of studies have demonstrated a neuroprotective role for VEGF-A in experimental diabetic neuropathy after administration of VEGF-A by gene transfer (16,17,24). Due to the paradox surrounding VEGF-A in relation to both macrovascular and microvascular complications in diabetes, targeted local delivery of VEGF-A is crucial for the treatment of diabetic neuropathy.

We have previously demonstrated that unilateral intramuscular injection of VZ+434 prevents the sensory and motor NCV deficits in the ipsilateral sciatic nerve of STZ-induced diabetic rats (24), and suggested a direct neuroprotective role of VEGF-A in experimental diabetic neuropathy. In this current study, we extend these observations to demonstrate a further neuroprotective effect of VZ+434, namely protection against mechanical allodynia. After intramuscular VZ+434 there is a correction in retrograde axonal transport of VEGF-A in the sciatic nerve to the DRG. As with the NCV change, this effect was restricted to the neurons ipsilateral to the injection. We also observed accumulation of VEGF-A proximal to the ligature, in accordance with other studies showing VEGF axonal transport to be bidirectional (6,16,41).

The mechanism of uptake of VEGF-A by sensory afferent terminals present in the gastrocnemius muscle is unclear. NGF release from cultured smooth muscle cells is elevated in the presence of contractile stimuli (42). It is feasible that, in our study, VEGF-A release from the injected skeletal muscle occurs in this way, aided by the everyday movement of the rats. We postulate that VEGF-A uptake occurs via binding to VEGFR-2; it is well established that upon ligand binding VEGFR-2 is internalized with caveolin-1 and dynamin-2 and undergoes intracellular trafficking and nuclear localization, via microtubules and the endocytic pathway (43,,,–47).

VZ+434-induced VEGF-A activates the cytoplasmic Ser/Thr kinase AKT in L4/5 DRG ipsilateral to the VZ+434 injection. The downstream targets of VEGF-A/AKT activation are currently under investigation. The PI3-K pathway is activated by many growth factors including IGFs, insulin, and NT-3 in sensory neurons (48,,–51), all of which are reduced in diabetic neuropathy (13,52,53). Furthermore, VEGF-A increases pAKT in primary cultures of dissociated rat sensory neurons (data not shown). A study using the PI3-K inhibitor, LY294002, on mice DRG explants implicated this pathway in sensory neuron survival, as LY294002 caused a dose-dependent cell death (54). PI3-K promotes neuronal cell survival by modulation of Bcl-2 and thus mitochondria function, and regulates energy metabolism in sensory neurons. In addition, activation of the PI3-K/AKT pathway increases axonal caliber and terminal branching of sensory neurons in vitro (55) and thus plays an important role in axon morphological differentiation.

Because phosphorylation of AKT occurs in a greater number of neurons than those possessing VEGF-A-IR after VZ+434 administration, this may support the notion of both an autocrine and paracrine action of transported VEGF-A within the DRG. This could also explain how expression of VEGF-A, by small- to medium-diameter neurons in the DRG, may ameliorate the NCV deficits measured in large-diameter myelinated axons in diabetic neuropathy (24), as increased pAKT-IR is observed in all populations of sensory neurons, however further experiments would be necessary to confirm this. Alternatively, NCV deficits may be ameliorated by another, as yet unidentified, mechanism such as restoration of nerve blood flow.

Our current work suggests that VEGF-A is able to modulate the PI3-K signaling cascade in the L4/5 DRG, which may underlie the neuroprotective action of VZ+434 in the management of diabetic neuropathy.

E.J.P. and B.D.-J. were supported by the Medical Research Council (U.K.) and the University of Manchester. N.J.G. was supported by Research Council UK and British Pharmacological Society (U.K.).

S.K.S. and R.S. are paid employees of Sangamo BioSciences. S.K.S. is Director of Preclinical Development and receives company incentive stock options and has been involved in the research and development of reagents described in the article. No other potential conflicts of interest relevant to this article were reported.

We thank Sangamo Biosciences, Inc., for the VZ+434 plasmid DNA and study support and Philip Gregory and Steve Zhang for review and helpful comments on the article.