Dose-dependent Neuroprotection of VEGF₁₆₅ in Huntington's Disease Striatum

Abstract

Huntington's disease (HD) is a devastating neurodegenerative disorder caused by abnormal polyglutamine expansion in the huntingtin protein (Exp-Htt). Currently, there are no effective treatments for HD. We used bidirectional lentiviral transfer vectors to generate in vitro and in vivo models of HD and to test the therapeutic potential of vascular endothelial growth factor 165 (VEGF₁₆₅). Lentiviral-mediated expression of Exp-Htt caused cell death and aggregate formation in human neuroblastoma SH-SY5Y and rat primary striatal cultures. Lentiviral-mediated VEGF₁₆₅ expression was found to be neuroprotective in both of these models. Unilateral stereotaxic vector delivery of Exp-Htt vector in adult rat striatum led to progressive inclusion formation and striatal neuron loss at 10 weeks post-transduction. Coinjection of a lower dose VEGF₁₆₅ significantly attenuated DARPP-32⁺ neuronal loss, enhanced NeuN staining and reduced Exp-Htt aggregation. A tenfold higher dose VEGF₁₆₅ led to overt neuronal toxicity marked by tissue damage, neovascularization, extensive astrogliosis, vascular leakage, chronic inflammation and distal neuronal loss. No overt behavioral phenotype was observed in these animals. Expression of VEGF₁₆₅ at this higher dose in the brain of wild-type rats led to early mortality with global neuronal loss. This report raises important safety concerns about unregulated VEGF₁₆₅ CNS applications.

Introduction

Huntington's Disease (HD) is an autosomal dominant, fatal and progressive neurodegenerative genetic disease. The abnormality/mutation identified in 1993 is an expansion of the trinucleotide CAG repeat (>36 glutamines) 17 codons downstream of the initiation codon of the Huntingtin gene (Htt).¹ Patients with HD experience symptoms including chorea, involuntary muscle movements, cognitive impairment, personality, and behavioral changes at around 45–60 years of age. Death usually occurs 10–15 years following symptom onset. The expanded polyglutamine tract causes a toxic gain-of-function of the Htt gene, leading to death and dysfunction of the GABAergic medium spiny striatal neurons, which are the main neurons involved in early stage HD pathogenesis.² Abnormal polyglutamine aggregates, dysregulated transcription, disturbances of the proteolysis and proteosome protein degradation pathways, blockages of synaptic transmission, excitotoxicity, and mitochondrial disfunction have all been implicated in the neuronal pathology observed in HD (reviewed in ref. 3).

Currently there is no cure for HD and only symptomatic treatments are available. Research into anti-aggregating, autophagic, neurotrophic and gene knockdown therapeutic approaches is being pursued in HD. Molecular chaperones or heat shock proteins belong to a large family of proteins classified according to cellular localization, function and molecular weight.⁴ These proteins are induced in times of stress and are responsible for refolding misfolded protein, regulation of protein quality control, fusion of vesicles, signal transmission processes, apoptosis, and proteosomal degradation.⁴ Studies based on C. elegans, Drosophila melanogaster, S.cerevisiae, and rodent HD models have all illustrated the capability of these chaperones to suppress aggregate formation.^5,6,7,8,9 Neurotrophic factors, some downregulated by the transcriptional dysregulation in HD striatal neurons, such as brain-derived neurotrophic factor, glial-derived neurotrophic factor, and ciliary neurotrophic factor have been found to be neuroprotective in a variety of animal models of HD.^{10,11,12,13,14}

Initially discovered in 1983 and cloned in 1989, vascular endothelial growth factor A (VEGF-A) is a multifunctional cytokine that exerts a variety of effects on vascular endothelial cells by interaction with receptors VEGFR1 and VEGFR2, that together promote the formation of new blood vessels.^15,16 Thus, in addition to rendering microvessels hyperpermeable, VEGF-A (including VEGF189, 165a and b and 121 forms mainly) stimulates endothelial cells to migrate and divide, profoundly altering their pattern of gene expression, and protecting them from apoptosis and senescence (reviewed in ref. 17).

As knockout of VEGF causes mouse embryonic lethality, transgenic mice with an altered VEGF gene (VEGF δ/δ) were created in which the hypoxia response element, located in the VEGF gene promoter, was deleted. This enabled a reduction in VEGF expression in response to hypoxia while still allowing survival into adulthood. In these mice a delayed pathological phenotype resembling human amyotrophic lateral sclerosis (ALS) was observed.¹⁸ This selective spinal motor neuron death was attributed to a diminished perfusion from blood vessels. Supplying VEGF₁₆₅ via retrogradely transported lentiviral vectors (LVs) to spinal motor neurons of the SOD1^G93A mouse model of ALS resulted in significant amelioration of disease phenotype.¹⁹ These results confirmed the neurotrophic role of VEGF for motor neurons as subsequently substantiated by others.^20,21

Lentiviruses are widely used and acknowledged as a powerful tool in gene therapy studies due to valuable characteristics such as; their ability to transduce both dividing and non-dividing cells, large insertion capacity (8–10 kb), long-term and stable expression of the inserted transgene, and minimal induction of immune responses.²² More importantly, recombinant LVs can be used to simulate disease states in animals. Specifically for HD, de Almeida et al. and Palfi et al. have used LVs encoding the mutant Htt fragments to induce the disease state in targeted striatal regions of the brain of rodents and primates and also established models of HD in primary neuronal cultures.^23,24,25 Perrin et al. and more recently Martin et al. demonstrated that lentiviral-mediated expression of Hsp104/Hsp27 or stress activated kinase MSK-1 can attenuate striatal dysfunction in these HD disease models.^6,26

To investigate the possible neuroprotective effect of long-term VEGF-A expression in models of polyQ toxicity, we designed novel bidirectional vectors capable of both inducing in vitro and in vivo models of HD and transducing in specific therapeutic genes. Here, we report that VEGF₁₆₅ neuroprotects striatal neurons in these models but its effect is strictly dose dependent with higher doses inducing CNS neurodegeneration and neuroinfammation via pleiotropic vascular effects.

Results

Design and expression of lentiviral transfer plasmids

So as to be able to trigger Htt-induced aggregate formation and neurodegeneration in vitro and in vivo, we cloned 18Q or 82Q and coexpressed with enhanced green fluorescent protein (eGFP) in bidirectional LVs based on a previously published vector system²⁷ (Figure 1b). We also cloned similar vectors to test VEGF₁₆₅ and Hsp70, which is known to be capable of ameliorating polyQ toxicity, and coexpressed them with red fluorescent protein (RFP).^6,8 By using the bidirectional lentivector systems and distinct marker genes (Figure 1a), we could spatially distinguish transduction with different vectors. Western blot analysis demonstrated that wild-type and Exp-Htt fragments (18Q and 82Q) and full-length human proteins for VEGF₁₆₅ or Hsp70 of the predicted molecular weight were expressed from these vectors (Figure 1c–e). As VEGF₁₆₅ is secreted, a specific ELISA confirmed its presence in supernatants of 293T cells transduced with VEGF₁₆₅ expressing LVs (Figure 1f). High level expression of the marker gene from these vectors was further detected by FACS or fluorescent microscopy following either transfection or transduction of 293T cells (data not shown). To evaluate if VEGF₁₆₅ affects Exp-Htt expression (82Q), 293T cells were cotransduced with both 82Q and VEGF₁₆₅ expressing LVs. Western blot analysis confirmed that there was no alteration in Exp-Htt expression in the cotransduced cells (Figure 1g).

Figure 1.

Generation of bidirectional lentiviral constructs. (a) The HIV-1–based transfer vectors used in this study used a bidirectional promoter system based on the minimal cytomegalovirus promoter (minCMV) and the human phosphoglycerate kinase (hPGK) promoter, thus allowing co-ordinate expression of both reporter and test transgenes from the same vector. These were as originally reported by Amendola et al. (2005)²⁷ but differ in that there is no constitutive transport element (CTE) of the Mason-Pfizer virus included in the upstream cassette that is in antisense orientation to the vector LTR. This modification results in better in vivo expression (Mario Amendola, personal communication to V.T.). (b) We generated transfer vectors in which the human phosphoglycerate kinase (hPGK) promoter drives expression of either wild-type exon I huntingtin (Htt171-18Q) or mutant huntingtin (Htt171-82Q) fragments, human codon-optimized vascular endothelial growth factor 165 (VEGF₁₆₅), human codon-optimized heat shock protein 70 (Hsp70) or C-truncated low-affinity nerve growth factor receptor (ΔLNGFR). Both huntingtin fragments are 5′ myc tagged. The minCMV promoter drives the expression of the marker transgenes enhanced green fluorescent protein (Htt vectors and negative control 1) or red fluorescent protein (VEGF₁₆₅, Hsp70 vectors and negative control vectors 2 & 3). (c) Western blot analysis of 293T cell extracts transfected with either bidirectional 18Q and 82Q vectors (pCMV-eGFP.hPGK-Htt171-18Q and pCMV-eGFP.hPGK-Htt171-82Q) or positive control expression vectors (SIN-PGK-Htt171-18Q and SIN-PGK-Htt171-82Q, gift from Nicole Deglon) confirmed appropriate expression of 18 and 82Q huntingtin protein. Arrows indicate ~33 (195aa) and 50 kD (256aa) proteins respectively detected with anti huntingtin EM48 antibody. (d) Expression of Hsp70 was confirmed in 293T cell extracts transfected with pCMV-RFP.hPGK-Hsp70 and Hsp70 positive control (pCMVHsp70, a human non-codon optimized cDNA, gift from Professor RI Morimoto) producing a 70 kD expected protein as detected by anti-Hsp70 antibody. (e) Expression of VEGF₁₆₅ was confirmed in 293T cell extracts transfected with pCMV-RFP.hPGK-VEGF₁₆₅. Arrow indicates a 25 kD expected protein detected by anti-VEGF antibody just above a non-specific band also present in non-transfected and negative control samples (pCMV-RFP.hPGK-ΔLNGFR). Expression of DsRed was confirmed with DsRed antibody. Blots for GAPDH were included as loading controls. (f) An ELISA confirmed the presence of VEGF in the supernatant of 293T cells transduced with pCMV-RFP.hPGK-VEGF₁₆₅. (g) Coexpression of 82Q and VEGF₁₆₅ in 293T cells cotransduced with pCMV-eGFP.hPGK-Htt171-82Q and pCMV-RFP.hPGK-VEGF₁₆₅ at an MOI of 10.

VEGF₁₆₅ and Hsp70 are anti-aggregating in an 82Q SH-SY5Y model

We used human neuroblastoma SH-SY5Y cells, a neuronal cell line that exhibits moderate levels of dopamine beta hydroxylase activity and displays neuronal (MAP2⁺) characteristics (data not shown), which have been used previously to generate in vitro models of HD.²⁸ Undifferentiated SH-SY5Y cells were transduced with LV expressing 18Q (control), 82Q or 82Q in combination with LVs expressing VEGF₁₆₅, Hsp70 or truncated low-affinity nerve growth factor receptor (ΔLNGFR control) at an MOI of 25. Ninety-six hours after transduction, cell death, apoptosis, and Htt aggregate formation were assessed. Cell death was assessed by trypan blue staining (Figure 2a). SH-SY5Y cells transduced with 82Q vector alone or in combination with ΔLNGFR vector showed similar levels of cell death (18.31 ± 2.20% and 19.41 ± 2.58% respectively), whereas cells transduced with 82Q + VEGF₁₆₅ showed a significant reduction in cell death to 12.85%, an overall reduction of 29.81%, (*P < 0.05, Student's paired t-test, n = 4) demonstrating a clear neuroprotective effect of VEGF₁₆₅ in this HD model. A similar pattern was observed when investigating apoptosis. Adherent cells were incubated with Hoechst-33342 dye and assessed by fluorescent microscopy for cells demonstrating typical apoptotic morphology. Bright cells displaying condensation/fragmentation of the nuclear material (pyknosis) and formation of apoptotic bodies were classed as apoptotic (Figure 2b, top panels). The percentage of apoptotic cells was quantified from 10 fields of view (40× magnification) for each condition. SH-SY5Y cells transduced with 82Q or 82Q + ΔLNGFR LVs showed similar levels of 7.33 and 8.01% apoptotic cells respectively, whereas cells transduced with 82Q + VEGF₁₆₅ LVs showed a significant reduction in apoptotic cells to 4.65%, an overall reduction of 36.6% (***P < 0.001, Student's paired t-test, n = 4).

Figure 2.

Neuroprotective and anti-aggregating effects of vascular endothelial growth factor 165 (VEGF₁₆₅) in an in vitro SH-SY5Y model of Huntington's disease. (a) SH-SY5Y neuroblastoma cells were transduced with lentiviral vector expressing wild-type Htt (18Q), Exp-Htt (82Q) alone or in combination with lentiviral vector encoding human VEGF₁₆₅ (82Q + VEGF) or control lentiviral vector (82Q + ΔLNGFR). Ninety-six hours after transduction, cell death was assessed by trypan blue staining. Expression of VEGF with 82Q showed a significant reduction in trypan blue positive cells compared with 82Q alone (n = 4, mean ± SEM, *P < 0.05, Student's t-test). Equivalent levels of trypan blue positive cells were observed in the non-transduced and 18Q-transduced groups. (b) The experiment was repeated and cell apoptosis assessed by Hoechst dye staining. Bright staining indicated cells undergoing apoptosis (examples of apoptotic cells circled). Coexpression of VEGF with 82Q showed a significant reduction in Hoechst positive cells compared with 82Q alone (n = 4, mean ± SEM, ***P < 0.001, Student's t-test). (c) SH-SY5Y cells were transduced with lentiviral vector encoding wild-type Htt (18Q), Exp-Htt (82Q) alone or in combination with lentiviral vector encoding human VEGF₁₆₅ (82Q + VEGF), negative control lentiviral vector (82Q + ΔLNGFR) or positive control lentiviral vector encoding Hsp70 (82Q + Hsp70). Ninety-six hours after transduction, the formation of huntingtin inclusions was assessed by EM48 antibody and DAB staining. Quantitative analysis demonstrates that coexpression of VEGF with 82Q significantly reduced the number of huntingtin aggregates compared with 82Q alone (n = 4, mean ± SEM, *P < 0.05, Student's t-test). Scale bars represent 50 µm.

Transduced SH-SY5Y cells were assessed for Htt aggregates with EM48 antibody and DAB staining. Dark cellular staining indicated the formation of Htt aggregates (Figure 2c, top panels). SH-SY5Y cells transduced with 82Q or 82Q + ΔLNGFR LVs demonstrated a similar number of Htt aggregates (15.14 versus 15 average aggregates per field of view, 40× magnification), whereas cotransduction with 82Q + VEGF₁₆₅ LV resulted in a reduction in aggregate number to 10.89 (an overall reduction of 28.1%), demonstrating VEGF₁₆₅ is anti-aggregating in this model. A positive anti-aggregating effect of Hsp70 was also observed in this HD model.^6,8 Cotransduction of Hsp70 and 82Q vectors resulted in a reduction in aggregate formation by 73.6% as compared with the 82Q only transduced cells.

VEGF₁₆₅ is neuroprotective in an 82Q striatal neuronal culture model while Hsp70 is anti-aggregating

We further investigated the potential of our bidirectional lentiviral system to generate a long-term in vitro model of HD using rat primary striatal cultures.²³ Striatal neurons grown in primary culture (NeuN⁺/DARPP32⁺ data not shown), demonstrate a range of neurochemical phenotypes and cell–cell interactions which correspond closely to those present in the mature striatum in vivo.²⁹ Primary striatal neuronal cultures were transduced with 82Q LV alone or in combination with VEGF₁₆₅, Hsp70, or ΔLNGFR LVs. Controls were transduced with 18Q or ΔLNGFR LVs coexpressing eGFP or RFP reporters. Six weeks after transduction cells were stained with the neuronal marker NeuN to assess cell survival and the Htt aggregate marker EM48 (Figure 3). Neurons transduced with 82Q LV showed a significant reduction in number compared with those transduced with 18Q and ΔLNGFR LVs (5.71 versus 9.36 and 10.21 mean NeuN⁺ cells respectively). This cell death was partly restored when cells were coinfected with 82Q and VEGF₁₆₅. Cotransduction with VEGF₁₆₅ LV resulted in a significant increase of 28.8% in the number of surviving neurons compared with controls, demonstrating substantive neuroprotection from 82Q toxicity in this progressive in vitro HD model. In accordance with previous reports, Hsp70 was not found to be neuroprotective in cells cotransduced with 82Q and Hsp70;⁸ however, Hsp70 did significantly reduce the number of 82Q aggregates as detected with EM48 staining (5.61 average aggregates for 82Q alone versus 1.23 average aggregates for 82Q + Hsp70, a reduction of 78.2%). In the 82Q + VEGF₁₆₅–transduced neurons a reduction in aggregates was observed compared with 82Q alone (3.77 versus 5.61 respectively); however, this reduction was not significant when compared with the 82Q + ΔLNGFR control group (82Q + VEGF₁₆₅: 3.77 versus, 82Q + ΔLNGFR: 3.84 average aggregates).

Figure 3.

Neuroprotective effects of vascular endothelial growth factor 165 (VEGF₁₆₅) in an in vitro primary striatal neuronal culture model of Huntington's disease. Primary striatal neuronal cultures were transduced with lentiviral vectors encoding ΔLNGFR or 18Q as controls and 82Q alone or in combination with ΔLNGFR, Hsp70 or VEGF₁₆₅. Six weeks after transduction striatal cultures were fixed and stained for huntingtin aggregates (EM48 antibody) and NeuN followed by DAB staining. Huntingtin aggregates were observed in all groups expressing 82Q, but a reduction of aggregates was observed in cells cotransduced with 82Q and the positive control Hsp70. Neuronal survival was assessed with NeuN staining. Striatal neuronal cultures coinfected with 82Q and VEGF showed a significant enhancement of cell survival compared with 82Q alone (n = 3, mean ± SEM, ***P < 0.001, Student's t-test) and 82Q plus Hsp70 or ΔLNGFR control (n = 3, mean ± SEM, **P < 0.01, one way analysis of variance). Scale bar represents 50 µm.

LV PolyQ expression induces aggregation and neurotoxicity in the striatum in vivo

We next attempted to generate a lentiviral-mediated in vivo model of HD using our novel bidirectional LVs.²⁵ Rats were stereotaxically injected in the right striatum with 82Q vector to induce biochemical features of HD. Preliminary studies demonstrated striatal Htt aggregate formation without neurotoxicity at 5 weeks and increased aggregate formation with striatal neurodegeneration at 8 weeks after transduction (1.25 µl of 82Q LV, 6.78 × 10⁸ TU/ml, 1.11 × 10⁵ ng p24/ml) (Supplementary Figure S1). On the basis of these initial observations, we increased the 82Q LV dose (4 µl, 1.33 × 10⁹ TU/ml, 7.87 × 10³ ng p24/ml) and tested incubation times of 8, 10, or 12 weeks to achieve optimal neurotoxicity and Htt aggregate formation in the striatum (Figure 4). Targeted transduction of 82Q LV within the striatum was confirmed by immunohistochemistry with GFP antibody staining (Figure 4a). In this experiment, the 10-week time point demonstrated maximal DARPP-32 depletion (Figure 4a–e), with almost twofold increase in damaged area compared with the 8-week time point (10 week: 3.34 mm³ versus 8 week: 1.89 mm³). Extending the incubation period to 12 weeks did not increase the area of DARPP-32 depletion. The maximal number of Htt aggregates was also detected at the 10-week time point as detected with 2B4 antibody and did not increase with increasing incubation time (average number of aggregates for 82Q-transduced brains = 3,214 ± 1,289 n = 3 Figure 4f–i). For Htt aggregates detection in vivo, the 2B4 antibody provided less background than EM48 antibody staining. The huntingtin inclusions detected costained for ubiquitin and c-Myc antitag antibody (Figure 4j–m) and appeared to be predominantly nuclear following confocal microscopy analysis corresponding to previous reports (Figure 4n–q).²⁵ No DARPP-32 depletion or Htt aggregates were detected in control brains transduced with 18Q LV as expected (Figure 4b,f). These findings confirmed sufficient neurotoxicity and Htt aggregate formation at 10 weeks after transduction in this in vivo HD model.

Figure 4.

Generation of an in vivo model of 82Q striatal aggregation and toxicity. Wistar rats were transduced in the striatum with 18Q or 82Q expressing LV by sterotaxic injection and examined 8, 10, or 12 weeks after transduction (n = 6). (a) Representative coronal sections of the striatum demonstrating GFP expression, DARPP-32 depletion and 2B4 stained huntingtin aggregates from a GFP.82Q LV–injected rat (4× tiled images). (b–e) Evaluation of the neurotoxicity of 18Q (10 weeks) and 82Q (8, 10, and 12 weeks) on DARPP-32 stained striatal sections. As expected, no striatal degeneration was observed in the 18Q-transduced striatum. Most significant striatal degeneration was observed at the 10-week time point in the 82Q-transduced striatum. (f–i) No Htt inclusions were detected in the 18Q-transduced striatum with 2B4 antibody. The highest level of huntingtin inclusions were detected in the 82Q-transduced striatum at the 10-week time point. Scale bars represent 100 µm (j–m) DAPI, c-Myc (MYC) and ubiquitin triple staining indicates that huntingtin inclusions (c-Myc tagged) are ubiquitinated. Scale bar represents 25 µm. (n–q) DAPI, 2B4 and GFP triple staining indicates the presence of nuclear Htt inclusions within 82Q-transduced neurons. Scale bar represents 25 µm.

Coexpression of 82Q and low dose VEGF₁₆₅ attenuates neurodegeneration in the rat striatum

To determine whether VEGF₁₆₅ confers neuroprotection in our in vivo HD model, rats were coinjected with 82Q and VEGF₁₆₅ LVs into the right striatum and compared with an 82Q plus control ΔLNGFR coinjected group. Ten weeks after transduction, a marked decrease in neurodegeneration within the striatum, as determined by DARPP-32 immunohistochemistry, was observed in the 82Q + VEGF₁₆₅–coinjected group (Figure 5a). The total volume of DARPP-32 depletion was significantly smaller in the presence of VEGF (82Q + ΔLNGFR: 4.385 ± 0.56 mm³ versus 82Q + VEGF 2.79 ± 0.18 mm³; *P < 0.05, n = 4). VEGF expression reduced the volume of polyQ-induced DARPP-32 depletion by 36.4%. There was a similar reduction in striatal degeneration observed with NeuN staining (Figure 5a), a late marker of neuronal dysfunction in LV-based HD rat models.³⁰ The total volume of NeuN depletion for 82Q + ΔLNGFR was 2.554 ± 0.22 mm³ compared with 82Q + VEGF area of 1.564 ± 0.08 mm³, an improvement of 38.8%. We then analyzed the number of Htt aggregates present within the striatum of both groups and observed an overall decrease in 2B4 stained aggregates in the 82Q + VEGF₁₆₅ group compared with 82Q + ΔLNGFR control group (Figure 5b) (mean aggregates 3,668 ± 595.1 versus 6,701 ± 570.3, a 45.3% reduction in the presence of VEGF₁₆₅).

Figure 5.

Neuroprotective and anti-aggregating effects of low dose vascular endothelial growth factor 165 (VEGF₁₆₅) in an in vivo model of HD. (a) Immunocytochemical detection of DARPP-32 (top panels) and NeuN (bottom panels) was performed with specific antibodies (see Materials and Methods). 82Q-induced neurodegeneration confirmed by DARPP-32 and NeuN depletion is attenuated by VEGF expression. Graphical representation of the total volume depletion for DARPP-32 and NeuN for 82Q + ΔLNGFR control and 82Q + VEGF–infected striatums (n = 4, mean ± SEM, *P < 0.05, **P < 0.001, Student's paired t-test). (b) The number of striatal 82Q-induced Htt aggregates was significantly reduced by coexpression of VEGF (n = 4, mean ± SEM, *P < 0.05, Student's paired t-test, 82Q + ΔLNGFR control versus 82Q + VEGF infection). (c) Reactive astrocytes were assessed by GFAP staining (82Q + ΔLNGFR control versus 82Q + VEGF infection). Reactive astrocytes were observed in both groups around the sites of injection in the cortex, corpus collosum and the striatum (arrow heads). (d) Striatal 82Q vector spread as determined by GFP reporter expression is equivalent for both 82Q + ΔLNGFR control and 82Q + VEGF–transduced groups (n = 4, mean ± SEM). (e) Percentage weight increase of animals following 82Q + VEGF or 82Q + ΔLNGFR injection. Scale bars represent 200 µm.

VEGF₁₆₅ has been previously reported to activate astrocytes.^31,32 We therefore looked for activated astrocytes using the glial specific marker glial fibrillary acidic protein (GFAP). Activated astrocytes were observed in both groups in and around the site of injection/needle tract (cortex, corpus collosum, and striatum on ipsilateral side) and also along the corpus collosum on the contralateral side to similar extents (Figure 5c). A small increase in activated astrocytes was observed in the 82Q + VEGF₁₆₅ group near the epicentre of injection which could be attributed to activation by VEGF₁₆₅ (Figure 5c, left panels); however, extensive astrogliosis was not observed in the 82Q + VEGF₁₆₅ group. A similar anteroposterior spread of vector throughout the striatum, as characterized by eGFP expression, was observed between both groups indicating the differences in neurodegeneration and aggregate formation detected were not attributed to a difference in 82Q LV injected (Figure 5d) (anterior/posterior LV spread: 82Q + ΔLNGFR 2,280 µm versus 2,400 µm 82Q + VEGF).

Interestingly, rats receiving 82Q + VEGF gained more weight during the 10-week incubation compared with the 82Q + ΔLNGFR control group (Figure 5e), further emphasizing the positive effect of VEGF₁₆₅ in HD pathology. No adverse inflammatory response (as assessed by ED1 and CD4 markers), vascularization (Glut-1 marker) or vascular leakage (albumin staining) were detected (Figure 7b and data not shown). These findings suggest LV delivery of VEGF₁₆₅ may have potential therapeutic application in the treatment of HD.

Coexpression of 82Q and high dose VEGF₁₆₅ in the rat striatum causes overt toxicity and chronic inflammation

To see if the extent of VEGF₁₆₅-induced neuroprotection could be improved upon, we coinjected 82Q with a higher dose of VEGF₁₆₅ LV (tenfold higher titer) to increase the overall amount of VEGF₁₆₅ expression. A striking observation from these brains harvested 10 weeks after injection was that rats receiving the 82Q coinjection with the higher dose of VEGF₁₆₅ suffered from an enlarged/swollen brain (~25% increase in size), along with a hemorrhagic legion around the injection site which was not observed in the control group (Figure 6a) nor the 82Q + lower dose VEGF₁₆₅ group. Coinjection with high dose VEGF₁₆₅ also caused a reduction in overall body weight increase compared with 18Q, 82Q, and 82Q + ΔLNGFR control groups (Figure 6b). VEGF₁₆₅ is well described for its role in angiogenesis and vascular permeabilization.^17,33 We therefore assessed whether an overexpression of VEGF₁₆₅ in this system causes adverse vascularization and blood leakage. Endothelial glucose transporter-1 (Glut-1), a specific blood–brain barrier (BBB) marker protein was used to determine vascularization. Secreted VEGF₁₆₅ is known to increase glucose transport across the BBB.^34,35 Glut-1 positive microvessels were detected within the striatum around the injection site and in more distant areas of the striatum (Figure 6c) indicating vascularization and BBB disruption. The entire brain was then assessed for blood albumin protein leakage (Figure 6d). A striking observation following sectioning was a large intracranial cavity containing necrotic tissue originating from the site of injection. Extensive damage throughout the brain was observed affecting predominantly the striatum, corpus collosum, and interconnecting forceps major (Figure 6d). This was accompanied with an extensive amount of blood albumin leakage which appeared to be largely restricted to the ipsilateral side but extended throughout the brain from early anterior to posterior brain regions (Figure 6d dark staining). There was also an enlargement of the brain on the ipsilateral side. These findings indicate that lentiviral-mediated VEGF₁₆₅ overexpression has serious consequences for BBB permeability and tissue necrosis in the rat brain which override the neuroprotective effects observed at a lower dose.

Figure 6.

Acute striatal toxicity and vascularization with high dose vascular endothelial growth factor 165 (VEGF₁₆₅) in an in vivo model of HD. (a) Transduction with 82Q + high dose VEGF LV caused extensive brain swelling and hemorrhage. Arrows indicate site of injection and necrotic lesion. There was no adverse phenotype observed in the 82Q + RFP control–infected group (n = 6). (b) 82Q + high dose VEGF–injected rats gained less weight compared with 18Q, 82Q and 82Q + RFP control–infected groups. Rats receiving high dose VEGF LV alone lost body weight and perished before 4 weeks due to adverse affects (see Supplementary Figures S2 and S3). (c) Glut-1 staining demonstrated extensive vascularization in the striatum of the 82Q + high dose VEGF–injected brain on the ipsilateral side. (d) Vascular leakage was assessed by sectioning the entire brain and staining for albumin (dark staining). Extensive vascular leakage was present, predominantly on the ipsilateral side. Significant tissue damage was observed in the striatum and corpus collosum originating from the site of injection. (e) Neurodegeneration and astrogliosis were assessed by DARPP-32, NeuN, and GFAP staining, with DAPI counter staining. (i,ii) Extensive DARPP-32 loss (green staining) was observed in the ipsilateral striatum (str), whereas strong DARPP-32 staining was observed in the contralateral striatum. (i,iii) A large number of vacuoles were observed on the contralateral side. (iv,v) Extensive NeuN loss was observed in the ipsilateral striatum, along with extensive astrogliosis. Activated astrocytes (GFAP, green staining) were observed in the (vii,viii) ipsilateral corpus collosum (cc), (x,xi) layer2/3 cortex (L2/3c), and (xiii,xiv) forceps major (fm). (iii,vi,ix,xii) Astrogliosis was not observed on the contralateral side. Scale bar represents 100 µm.

We next assessed neurodegeneration and astrogliosis (Figure 6e). A large proportion of the striatum suffered extensive necrosis and the remaining striatum showed a dramatic reduction in DARPP-32 positive medium spiny neurons compared with the contralateral side (Figure 6e; i,ii). This was accompanied with a large reduction of NeuN⁺ cells in the ipsilateral striatum (Figure 6e; iv). NeuN staining in the striatum on the contralateral side appeared normal suggesting there was little or no neurodegeneration on the contralateral side (Figure 6e; iii). In the CNS, glial cells greatly out-number neurons, not only playing a supporting role in essential neuronal function but also acting as primary responders to CNS insults such as infection, injury and degenerative disease.³⁶ We observed a widespread infiltration of GFAP⁺ reactive astrocytes in the damaged striatum (Figure 6e; iv,v), corpus collosum (Figure 6e; vii,viii) and ventral regions of the brain including layer 2/3 cortex (Figure 6e; x,xi). The most concentrated region of reactive astrocytes was observed in the corpus collosum adjacent to the most severe area of damage affecting the corpus collosum and striatum (Figure 6e; vii). The corpus collosum connects the two hemispheres of the brain and also extends to the brain posterior connecting to the forceps major. No reactive astrocytes were observed on the contralateral side corpus collosum (Figure 6e; vi); however, they were detected extensively in the forceps major on the ipsilateral side (Figure 6e; xiii,xiv). The extensive amount of astrogliosis observed is most likely due to a combination of VEGF₁₆₅-induced activation and proliferation of astroglia in response to the large amount of neurodegeneration observed.

We then investigated whether there had been any inflammatory response to VEGF₁₆₅ overexpression by examining various inflammatory markers such as macrophages/monocytes (ED1), T helper cells, monocytes, macrophages and dendritic cells (CD4 glycoprotein) and cytotoxic T cells (CD8). We found significant macrophage/monocyte infiltration in the ipsilateral striatum, corpus collosum, forceps major and piriform cortex brain regions (Figure 7a; ii,vi,x,xiv), whereas none were observed on the contralateral side (Figure 7a; i,v,ix,xiii). VEGF₁₆₅ is a chemoattractant to monocytes and can induce their migration across collagen membranes and endothelial cell monolayers.³⁷ A similar pattern of infiltration was observed following staining with antibody against CD4 glycoprotein, with large amounts of CD4⁺ cells detected in ipsilateral striatum, corpus collosum, forceps major and to a lesser extent piriform cortex (Figure 7a; iii,vii,xi,xv). A small number of cytotoxic T cells (CD8⁺) were detected within the striatum near the area of damage but not in other regions (Figure 7a; iv,viii,xii,xvi). These findings are in direct contrast to low dose VEGF₁₆₅ where no ED1 or CD4⁺ cells were detected (Figure 7b).

Figure 7.

Extensive neuroinflammation with high dose vascular endothelial growth factor 165 (VEGF₁₆₅) in an in vivo model of HD. (a) 82Q + high dose VEGF–injected brains: widespread macrophages/monocyte (ED1, green staining) and CD4⁺ T cell (CD4, red staining) infiltration was observed in the ipsilateral side in regions such as the (ii, iii) striatum (str), (vi, vii) corpus collosum (cc), (x, xi) forceps major (fm), and (xiv,xv) piriform cortex (pc). (iv) Some cytotoxic T cells were also observed in the striatum close to the site of injection. No inflammatory markers were detected on the contralateral side (i,v,ix,xiii). (b) 82Q + low dose VEGF–injected brains: No macrophage/monocyte infiltration (ED1, green staining) or CD4⁺ T cells (CD4, red staining) were observed in the striatum (str) or corpus collosum (cc) of rats injected with either 82Q + ΔLNGFR or 82Q + low dose VEGF (i–viii). Scale bar represents 100 µm.

To ensure that the dramatic inflammatory responses observed here were not caused by LV contamination 18Q, 82Q, and VEGF₁₆₅ LV preps were tested for endotoxin, bacterial and mycoplasma contamination (see Materials and Methods). No contamination or abnormal endotoxin levels were detected in any of the LV preps. These results indicate the vascular properties of unregulated vector–mediated VEGF₁₆₅ expression override its neuroprotective ones at a certain dose leading to inflammation and neurotoxicity instead and thus raises important safety implications for unregulated CNS applications with this factor.

VEGF₁₆₅ overexpression in the striatum leads to neurotoxicity and death

As a control, three rats were injected with the same high dose of VEGF₁₆₅ LV unilaterally in the striatum. In the absence of Exp-Htt, VEGF₁₆₅ overexpression proved to have more severe adverse side effects including progressive weight loss, akinesia and cyphosis in all injected animals. This led to premature death in two of the animals after just 3 weeks. The third animal was euthanatized and processed shortly after. Post mortem examination revealed similar but far more global brain pathology to that observed in the VEGF₁₆₅ + 82Q coinjected brains with neurodegeneration, vascular leakage, and astrogliosis spreading further anteroposterior from the injection site ipsilaterally and being more prominent on the contralateral side (Supplementary Figure S2a,b). Neuroinflammation was also more prominent with a greater influx of macrophages/monocytes observed on the contralateral side (Supplementary Figure S3). We speculate that coexpression of mHtt reduces expression of VEGF₁₆₅ from the transduced striatum thus reducing its deleterious side effects.

Discussion

For this study, we used a novel bidirectional HIV-1 LV approach to study the therapeutic potential of the human isoform of VEGF-A, VEGF₁₆₅, to protect against neurodegeneration in both in vitro and in vivo models of HD. We successfully constructed efficient bidirectional LVs enabling us to set up in vitro and in vivo HD models and test new gene therapies in them. Their application produced similar effects to those seen with alternate vector systems.^23,25

Using LV-mediated overexpression of mHtt, we were able to generate two in vitro models of HD; a fast, aggressive HD model by infecting neuronal-like SH-SY5Y cells with a high MOI (25) of 82Q LV and a slower, more progressive primary striatal neuron HD model by transducing with 82Q LV at a lower MOI (10). In both models, we demonstrated a clear neuroprotective role of VEGF₁₆₅ against Exp-Htt–induced cytotoxicity. In the SH-SY5Y HD model, LV-mediated overexpression of VEGF₁₆₅ reduced cell death and apoptosis by a third, likewise VEGF overexpression in primary striatal neurons expressing htt171-82Q improved neuronal survival by a similar degree. These encouraging observations are in accordance with previous in vitro studies which highlight VEGF₁₆₅ as a neuroprotective growth factor in cortical, hippocampal, motor, retinal, and dopaminergic neurons following polyglutamine-induced toxicity or hypoxia.^{33,38,39,40,41,42,43,44}

In a generated in vivo HD model, we were able to replicate the neuroprotective role of VEGF₁₆₅ observed in the in vitro experiments, by cotransduction with a low dose of LV. Promisingly, striatal neurodegeneration, induced by LV-mediated mHtt overexpression, was attenuated by VEGF expression, as assessed by early and late markers of striatal degeneration, DARPP-32 and NeuN. These results highlight the therapeutic potential of lentiviral-mediated delivery of VEGF in HD to reduce Exp-Htt–induced neurodegeneration. The capacity of VEGF₁₆₅ as a therapeutic for other neurodegenerative diseases such as ALS has been previously explored with positive outcome. VEGF delayed neurodegeneration and prolonged survival in ALS mice,¹⁹ consistent with findings observed in SOD1^G93A/VEGF^+/+ double transgenic mice.⁴⁵ Intrastriatal delivery of VEGF by AAV vector has shown to be beneficial in models of Parkinson's disease, promoting survival of dopaminergic neurons and improving rotational behavior.⁴⁶ Notably, a significant astrocytic infiltration was observed in this study. In a more recent study, the delivery of VEGF₁₆₅ via injectable hydrogels, to rats receiving striatal injection of quinolinic acid to mimic neuronal loss and behavioral deficits characteristic of HD, was shown to be neuroprotective at 3 weeks.⁴⁷ The advantage of our HD model over others is the expression of mHtt in the striatum which replicates the human pathology of the disease more closely, resulting in both progressive striatal neurodegeneration and aggregate formation. Also, using LVs as a therapeutic delivery system for VEGF₁₆₅, or indeed other therapeutic genes, has the distinct advantage of providing long-term gene expression abolishing the need for repeat treatments. Although not directly comparable due to different delivery systems and disease models, these studies and our own findings demonstrate the favorable effects of VEGF as a therapeutic in neurodegenerative diseases including HD. Furthermore, no adverse side effects of VEGF expression such as vascular leakage, vascularization, or immune response were reported similar to our observations with a low dose of VEGF LV.

In addition to neuroprotection, we also observed an unexpected VEGF-dependent anti-aggregating effect in both the SH-SY5Y and in vivo HD models, which to our knowledge is the first observation of its type. The exact mechanism of how VEGF has reduced aggregate formation is unclear from our studies and requires further investigation. Interestingly, in other acute and chronic models of HD, the neurotrophic factors, ciliary neurotrophic factor and brain-derived neurotrophic factor, have also shown to be neuroprotective, ameliorating striatal degeneration,^13,48 but without modifying inclusion formation.

In this study, we tried to maximize the neuroprotective and anti-aggregating effects of VEGF₁₆₅ in our in vivo HD model by increasing the dose but dramatically, overexpression of VEGF₁₆₅ caused extensive, adverse side effects including tissue damage, hypertrophy, vascular leakage, vascularization, astrogliosis, and neuroinflammation which overshadowed the beneficial effects observed at the lower dose. Similar unanticipated pleiotropic effects following high dose administration of VEGF to the adult CNS, have been reported using different delivery systems. Administration of human recombinant VEGF₁₆₅ by osmotic minipump and subdural sponge placement has been shown to cause cerebral angiogenesis, extensive protein leakage, BBB permeability, and proliferation of astroglia.³⁵ Also chronic overexpression of VEGF₁₆₅ by intracerebral infusion via miniosmotic pump or intracerebral injection of an adenoviral vector encoding VEGF₁₆₅ increased BBB permeability and inflammation in wild-type rat brain.⁴⁹ In addition, astrocyte-derived VEGF-A was shown to mediate BBB disruption, lymphocyte infiltration, and tissue damage in CNS inflammatory disease.⁵⁰ These observations taken together with our findings demonstrate that VEGF₁₆₅ expression can be both beneficial and detrimental in the brain dependent on the dose and method of administration.

The mechanism by which VEGF₁₆₅ exerts its neuronal survival effects appears to be via signaling through VEGFR2, PI3-kinase/Akt, and MEK/ERK predominantly but also via regulation of potassium channels which are dependent on VEGFR1 signaling.³⁸ Interestingly, vasopermeability, cell proliferation, and angiogenesis are also triggered via VEGFR2 signaling but through activation of PLCγ. This suggests that a dose-dependent threshold of VEGF may exist which tips the balance from neurosurvival/protection to vasopermeability and neurotoxicity.

Materials and Methods

Cells. Cell lines, 293T cells (HEK 293T/17, ATCC CRL-11268), were grown in Dulbecco's modified Eagles medium (Sigma, St Louis, MO; D6546) supplemented with Penicillin/Streptomycin (Pen/Strep; Sigma; P4333), 4 mmol/l L-glutamine (Sigma; G7513), and 10% (v/v) new born calf serum (Sigma; N4762). SH-SY5Y (ECACC, 94030304) cells were maintained in 1:1 Ham's F12 (Sigma; N4888): Eagle's minimal essential medium (Sigma; M2279) supplemented with 2 mmol/l glutamine, 1% non essential amino acids, and 15% fetal bovine serum. Primary striatal cells were prepared as follows: timed-pregnant Sprague–Dawley rats (Charles Rivers Laboratories, Les Oncins, France) were killed by CO₂ inhalation, and embryos (E16) were collected in a Petri dish and placed on ice. Dissections were performed under a stereomicroscope in ice-cold dissection medium (DMX) containing Ca²⁺- and Mg²⁺-free phosphate-buffer saline (PBS), 0.6% D-glucose, 1% Pen/Strep (10,000 U/ml, 10,000 Ag/ml), and 10 mmol/l HEPES (Invitrogen, Carlsbad, CA). Ganglionic eminences or cerebral cortices were isolated, cut into pieces using forceps, and collected in a 15 ml Falcon tube. The medium and buffer were removed by decanting on ice. The tissue was homogenized by repeated pipetting with a fire-polished Pasteur pipette in DMX containing 1% bovine serum albumin (Sigma). Cells were centrifuged at 4 °C for 5 minutes at 1,000g and resuspended in 10 ml of glutamate-free Neurobasal medium containing 1% B27 (Invitrogen), 2% Pen/Strep (10,000 U/ml, 10,000 Ag/ml), 0.5 mmol/l L-glutamine, and 15 mmol/l KCl. Cells were plated at a density of 150,000 cells/cm² in multi-well dishes coated with 20 µg/ml poly-L-lysine (Sigma; P6407). The cultures were maintained in a humid incubator (5% CO₂ at 37 °C), and half of the medium was changed weekly. Cultures were kept for up to 6 weeks.

Cloning and generation of transfer vector plasmids. Transfer vector plasmids encoding the genes of interest (htt171-18Q, htt171-82Q, and VEGF₁₆₅) and reporter genes (either GFP or RFP) (Figure 1) were generated by cloning into the bidirectional promoter vector, pCCL.sin_PPT.SV40polyA.GFP.minhCMV.hPGK-ΔLNGFR.WPRE, which was kindly provided by Mario Amendola (San Raffaele Telethon Institute for Gene Therapy, HSR-TIGET, Milan, Italy). p889-HD-N171-Myc (18Q) and p926-HD-N171-Myc (82Q) plasmids encoding wild-type and mutant fragment of the N terminal (171aa) htt gene respectively were kindly provided by Nicole Deglon (Commissariat á l'Energie Atomique, Orsay Cedex, France). htt171-18Q and 82Q were subcloned into pBluesript II KS (+) to add additional restriction sites before cloning into the bidirectional transfer vector downstream of the hPGK promoter, in place of the ΔLNGFR gene to create pCCL.sin_PPT.SV40polyA.GFP.minhCMV.hPGKHTTN171Myc18Q.WPRE and pCCL.sin_PPT.SV40polyA.GFP.minhCMV.hPGKHTTN171Myc82Q.WPRE respectively. Codon optimized Human VEGF₁₆₅ (NCBI Ref. Seq. NP_001165097.1) and Hsp70 (NCBI Ref. Seq. NC_000005.9) incorporating a Kozak sequence to increase translational initiation and two stop codons to ensure efficient termination were synthesized by GENEart (Life Technologies, Paisley, UK). VEGF₁₆₅ or Hsp70 was subcloned into pCR2.1-TOPO to add additional restriction sites before cloning into final transfer vector downstream of hPGK. The red fluorescent protein from DsRed2-N1 (kind gift by Massimo Pizzato) was subcloned into pCR2.1-TOPO, then a shuttle vector p0.6(a) that carries the bidirectional promoter (kindly provided by Mario Amendola), before cloning into the final transfer vectors under the minhCMV promoter giving rise to pCCL.sin_PPT.SV40polyA.RFP.minhCMV.hPGK.hVEGF₁₆₅.WPRE and pCCL.sin_PPT.SV40polyA.RFP.minhCMV.hPGK.HSP70.WPRE. All constructs were verified by restriction enzyme digest and sequencing.

Protein analysis

Western blotting: To confirm transgene protein expression, 293T cells were transfected with each of the newly generated transfer plasmids using Lipofectamine LTX reagent (Invitrogen; 15338-100). Forty-eight hours after transfection, cell lysates were prepared by the addition of RIPA buffer plus inhibitor cocktail (Thermo, Loughborough, UK; 89908 and 78440) and protein levels assessed by the Bradford assay and normalized. Lysates were boiled in 1× Laemmli buffer (6× 259 mmol/l Tris pH 6.8, 10% (w/v) SDS, 40% (v/v) glycerol, 0.6 µmol/l dithiothreitol) and run on a 10% SDS-PAGE gel. Separated proteins were transferred onto nitrocellulose membrane (Hybond-ECL, GE Healthcare, Chalfont St Giles, UK), blocked with PBS/Tween-20 (0.1% (v/v)) with 5% (w/v) milk powder) and probed with antibodies to VEGF₁₆₅ (Santa Cruz, Santa Cruz, CA; sc-7269) or HSP70 (Santa Cruz; sc-59569), c-Myc (tag HTT) (Santa Cruz; sc-56634) and GAPDH (Santa Cruz; sc-47724). Antibody-recognized proteins were detected using a horseradish peroxidase-conjugated anti-rabbit IgG secondary (Sigma) and ECL reagent (Pierce, Loughborough, UK).

ELISA: To confirm the presence of VEGF₁₆₅ in the supernatant of 293T cells transduced with VEGF₁₆₅ expressing LV vector (MOI 10), an ELISA was performed 72 hours after transduction against human VEGF according to the manufacturer's protocol (R&D Systems, Abingdon, UK; DVE00).

Production and purification of LVs. HIV-1–based LVs were produced using a standard transient calcium phosphate transfection protocol.³⁴ Briefly, 12 × 150 mm tissue culture dishes were seeded with 1.4 × 10⁷ 293T cells per dish. The following day cells were transfected with 32 µg of the relevant bicistronic transfer vector plasmid (expressing either eGFP or RFP reporter gene and huntingtin or VEGF₁₆₅ or Hsp70 or lacking a transgene), 12.5 µg plasmid expressing the HIV-1 gag/pol gene (pMD2-LgpRRE), 6.25 µg plasmid expressing HIV-1 Rev (pRSV-Rev), 9 µg of the VSVg envelope, and 15 µg pAdvantage plasmid to enhance viral production. Sixteen hours transfection, media was replaced with fresh media supplemented with 10 mmol/l sodium butyrate. Thirty-six hours after sodium butyrate induction, media containing vector was filtered through a 45 µm filter and centrifuged overnight (Beckman, High Wycombe, UK; F500 rotor, 6,000g). The viral vector pellet was resuspended on ice with cold PBS and repelleted by ultracentrifugation (Beckman; SW-32 Ti rotor, 68,500g, 1.5 hours). The pellet was then resuspended over several hours in TSSM formulation buffer (20 mmol/l tromethamine, 100 mmol/l NaCl, 10 mg/ml sucrose, and 10 mg/ml mannitol). Vector preps were stored at −80 °C. LVs were tested for bacterial contamination by inoculating liquid broth with a 5 µl aliquot of vector and incubating at 37 °C while shaking for 21 days. LVs were also tested for mycoplasma (Lonza, Slough, UK; LT07-418) and endotoxin (Pierce; 88282) contamination using specific kits following manufacturer's protocols.

Titration of LVs. The functional titers of LVs carrying either eGFP or RFP reporter genes were determined by flow cytometry. Briefly, 293T cells were seeded in 12-well plates at a density of 5 × 10⁵ cells per well. 16–24 hours after seeding, cells in a single well were quantified using a hemocytometer. Cells in the remaining wells were transduced for 6 hours with a serial dilution of the vector in the presence of 8 µg/ml polybrene. Seventy-two hours after transduction, the percentage of GFP +ve cells was evaluated by flow cytometry and TU/ml calculated (TU/ml = no. cells on day of transduction × proportion of positive cells × dilution factor). The number of physical lentiviral particles present in the viral vector stock was estimated using a HIV-1 p24 Antigen ELISA (RETROtek from ZeptoMetrix, Buffalo, NY).

Cell models of HD

SH-SY5Y transductions: Eight-well glass chambered slides were precoated with poly-D-lysine and laminin as follows: Four hundred microliter of 50 µg/ml poly-D-lysine (Sigma, P6407) was added to each well and slides were incubated for 2 hours at room temperature. Wells were washed three times with tissue culture H₂O and allowed to air dry before the addition of 400 µl of 10 µg/ml laminin (Sigma, L-2020). Laminin coated slides were incubated overnight at 37 °C and washed three times with phosphate buffered saline before seeding cells. SH-SY5Y cells were seeded at 80,000 cells per well, 16–20 hours after seeding cells were transduced with LVs at an MOI of 25 and after 6 hours the medium was refreshed. After 48 hours, half of the medium for each well was refreshed. Ninety-six hours after transduction, cell death, apoptosis, and aggregate formation were assessed. To evaluate the percentage cell death, all cells from each treatment were stained with trypan blue solution (0.4%; Sigma). The number of stained and unstained cells were quantified by hemocytometer count from 5 × 1 mm² squares for each sample and the percentage cell death calculated. To quantify cells undergoing apoptosis, adherent cells were incubated with 5 µg/ml Hoechst 33342 (Sigma; B2261) for 15 minutes at 37 °C and then analyzed by fluorescent microscopy. Apoptotic cells were distinguished by bright blue staining. Images were captured at 40× magnification and the number of apoptotic cells quantified from 10 fields of view per sample.

Primary striatal neuron transductions: Primary rat striatal neurons were seeded at 1.2 × 10⁵ cells per well in poly-D-lysine coated eight-well chambered slides. Striatal cultures were transduced at DIV 1 with LVs at an MOI of 10 and medium replaced with conditioned medium after 6 hours. Cultures were kept for 6 weeks, with a weekly medium change, and cell survival and aggregates assessed by NeuN and EM48 immunohistochemistry.

Stereotaxic surgery. All animal procedures were approved by the local Ethical Committee and performed in accordance with United Kingdom Animals Scientific Procedures Act (1986) and associated guidelines. All efforts were made to minimize the suffering and the number of animals used. Adult male wistar rats of 300–350 g were used for stereotaxic experiments. The animals were housed under a 12-hour light/dark cycle (light phase, 7:00 p.m. to 7:00 a.m.) with food and water available ad libitum. Animals were supplied by Harlan, Blackthorn, UK.

Before surgery, all animals were deeply anesthetized by inhalation of a mixture of 3L oxygen and 3.5% isoflurane (Merial, Harlow, UK) and then received systemic analgesia. They were placed in a stereotactic frame (Taxic-6000; World Precision Instruments, Sarasota, FL) with the nose bar at 3.3 mm. The anesthetic mixture was changed to 1L oxygen and 2.2% isoflurane for the remainder of the operation. The scalp was cut and retracted to expose the skull. Craniotomy was performed by drilling directly above the target region, to expose the pial surface. One single injection was directed into the right striatum using the stereotactic coordinates relative to bregma: anteroposterior, 0.5 mm; mediolateral, 3.0 mm; dorsoventral, 5.0 mm. Each rat received 6.0 µl of LV via a 32G blunt needle using an infusion pump (UltramicropumpIII and Micro4 Controller; World Precision Instruments) at a flow rate of 0.2 μl/minutes over 30 minutes. The needle was then allowed to remain on site for an additional 5 minutes before slow retraction. At the completion of surgery, the skin was sutured. For the lower VEGF₁₆₅ dose experiments, rats received 4 μl 82Q LV (1.25 × 10⁹ TU/ml, 1.65 × 10⁵ ng p24/ml) and either 1 μl VEGF LV (1.89 × 10⁸ TU/ml, 5.21 × 10⁴ ng p24/ml) or 0.5 μl of ΔLNGFR control (4 × 10⁸ TU/ml, 1.64 × 10⁵ ng/ml) and made up to a total injection volume of 6 μl with TSSM. For the higher dose VEGF experiments, rats received 3 μl 82Q LV (1.4 × 10⁹ TU/ml, 2.94 × 10⁵ ng p24/ml) and either 2.2 μl VEGF LV (8.8 × 10⁸ TU/ml, 3.95 × 10⁵ ng p24/ml) or 3 μl of RFP control (8.19 × 10⁸ TU/ml, 1.33 × 10⁵ ng/ml) and made up to a total injection volume of 6 μl with TSSM.

Rats were euthanized 10 weeks after injection by intraperitoneal injection of an overdose of sodium pentobarbitone, transcardially perfused with 50 ml ice-cold saline (0.9% w/v NaCl) plus heparin (20 U/ml), followed by 250 ml of 4% paraformaldehyde (Sigma). Brains were removed and post-fixed for 4 hours in 4% paraformaldehyde, followed by cryoprotection in 10% glycerol and 20% sucrose in 1× PBS for at least 72 hours. Brains were subsequently embedded and frozen in OCT (Surgipath FSC22; Leica Microsystems, Wetzlar, Germany). Thirty micron coronal sections from the entire striatum were cut using a cryostat (Leica Microsystems), mounted on poly-L-lysine coated slides and stored at −20 °C.

Immunohistochemistry

Immunostaining of cell cultures: To evaluate the neuronal cell and aggregate numbers, cells were fixed for 15 minutes at room temperature with 4% paraformaldehyde in PBS. After washing for 5 minutes with PBS, cells were permeabilized for 30 minutes in permeabilization buffer (10% goat serum, TBS, 0.5% triton X-100), washed again and incubated with normal blocking serum (Vectastain ABC kit, PK-6200; Vector Labs, Peterborough, UK) for 20 minutes. Cells were then incubated at 4 °C overnight with either mouse anti-NeuN (Millipore; MAB377) or mouse anti-huntingtin protein monoclonal antibody (clone mEM48; Millipore, Watford, UK; MAB5374). After washing, the secondary antibody (biotinylated antibody; Vectastain ABC kit) was applied, and color was developed using the Vectastain ABC reagent and DAB (Vectastain ABC kit). Immunostaining was examined under Nikon Eclipse TS100 inverted system microscope.

Immunostaining of brain sections: Free floating brain sections were permeabilized and blocked for 1 hour in permeabilization buffer. Antibodies were prepared in blocking buffer (10% goat serum, TBS, 0.03% triton X-100) to the following working dilutions: eGFP (1:500; Abcam, Cambridge, UK; ab290), NeuN (1:200; Millipore; MAB377), DARRP32 (1:1,000; Abcam; ab40801), EM48 (1:400; Millipore; MAB5374), 2B4 (1:200; Millipore; MAB5492), ubiquitin (1:200, Dako Z0458), Glut-1 (1:250; Abcam; ab652), albumin (1:400, MPI cappel 55728), c-Myc (1:100; Santa Cruz; sc-56634), ED1 (1:200; Abcam; ab31630), CD4 (1:200; Abcam; ab33775) and CD8 (1:200; Abcam; ab33786). Sections were incubated in primary antibody for 48 hours, washed three times with TBS and incubated in blocking buffer for 30 minutes before incubation with goat anti-mouse or anti-rabbit secondary antibodies conjugated with alexa fluor 488 or 594 (1:400, Invitrogen) for 2 hours at RT. Sections were then washed five times with TBS before being mounted with polyvinyl alcohol mounting medium with DABCO anti-fade (Sigma, 10981) in presence of 0.05 µg/ml DAPI (Lonza) to label nuclear DNA.

Image analysis and counting. Images from fluorescent–labeled sections (striatum) were taken under an epifluorescent microscope (Eclipse 80i; Nikon, Tokyo, Japan) equipped with a motorized stage for automated tiling of adjacent fields (Prior Scientific, Cambridge, UK). For DARPP-32 depletion assessment, a series of sections throughout the striatum 180 µm apart (caudate and putamen) were used for immunostaining. Images were captured at 4× magnification. The area of DARPP-32 depletion in mm² for each section was calculated with a computer-assisted imaging program (Image Pro Plus 7.0; Media Cybernetics, Rockville, MD). Lesion areas in each section were determined as region poor in DARPP-32 staining relative to the surrounding area. The total volume DARPP-32 depletion was estimated using the following equation: volume = d(a₁ + a₂ + a₃….), where d is the distance between sections (180 µm), and a₁, a₂, a₃…. are DARPP-32-depleted areas for individual sections. For huntingtin aggregate counts, five fields of view were captured for each section at 4× magnification and then image analysis software (Image Pro Plus 7.0) was used to count aggregates at a defined threshold intensity. The aggregate count for each animal was determined and then averaged for each group.

Bioinformatics and statistics. The amino acid sequences for transfer vector plasmids were aligned using DNA dynamo software. Statistics were analyzed and graphed using GraphPad Prism software (Graphpad Software, La Jolla, CA). Data are expressed as mean ± SEM and were evaluated by either one-way analysis of variance or paired Student's t-test.

SUPPLEMENTARY MATERIAL Figure S1. LV HD in vivo model. Figure S2. VEGF overexpression in wild-type striatum induces overt toxicity, vascularization and death. Figure S3. VEGF overexpression in wild-type striatum induces extensive macrophage/monocyte infiltration.

Supplementary Material

Figure S1.

LV HD in vivo model.

Figure S2.

VEGF overexpression in wild-type striatum induces overt toxicity, vascularization and death.

Figure S3.

VEGF overexpression in wild-type striatum induces extensive macrophage/monocyte infiltration.