Prophylactic Gene Therapy With Human Tissue Kallikrein Ameliorates Limb Ischemia Recovery in Type 1 Diabetic Mice

Procedures complied with the Guide for the Care and Use of Laboratory Animals (Bethesda, MD, Institute of Laboratory Animal Resources, National Academy of Sciences, 1996) standards. Type 1 diabetes was induced by streptozotocin (STZ; 40 mg/kg body wt i.p. daily for 5 consecutive days) (6) in male 2-month-old CD1 mice (Charles River, Comerio, Italy). Mice were considered diabetic only if they had fasting glycemia >250 mg/dl and overt glycosuria at 14 days from the first STZ injection. Persistence of diabetes was determined throughout the study. Body weight was recorded before STZ and every 14 days thereafter.

At 1 month after the first documentation of glycosuria, mice were anesthetized (2,2,2-tribromoethanol anesthesia, 880 mmol/kg i.p.; Sigma) to receive saline or adenoviral vectors carrying hTK gene (Ad.hTK; kindly provided by Dr. Julie Chao, Medical University of South Carolina, Charleston, SC) (7) or luciferase gene (Ad.Luc; each at 1 × 10⁸ plaque-forming units) into left adductor muscles.

After 14 days, mice were killed for determination of adductor microvascular density (n = 7, each group) or submitted to left hindlimb ischemia by femoral artery electrocoagulation (Ad.hTK, n = 10; Ad.Luc, n = 10; saline, n = 6). Age-matched nondiabetic mice injected with saline were studied for reference (n = 10).

Hindlimb blood flow was measured at 30 min and 4, 7, and 14 days after ischemia with a perfusion imager system (Lisca) (7). The ischemic-to-nonischemic foot blood flow ratio was calculated as an index of blood flow recovery.

After killing, the adductors of anesthetized mice were perfusion fixed and processed for evaluation of capillary density (6). For identification of arterioles, sections were stained with a mouse monoclonal antibody targeted to α-smooth muscle actin, as previously described (6). Arterioles were recognized as those vessels positive for α-smooth muscle actin and with one or more continuous layers of smooth muscle cells in their wall. Arteriole density per square millimeter of section was calculated as described previously (6). Proliferation and apoptosis were assessed in muscles harvested at 0, 3, and 14 days from ischemia (n = 5 per group). Proliferating cells were immunohistochemically identified by the use of a monoclonal antibody (Dako) against proliferating cell nuclear antigen (PCNA). The transferase-mediated dUTP nick-end labeling (TUNEL) assay revealed apoptosis as previously described (6). PCNA- or TUNEL-positive cell number was counted at ×1,000 magnification and expressed as the number of positive cells per square millimeter of section.

Additional experiments were performed to evaluate the protective potential of the type 1 diabetes reference drug insulin. To this aim, type 1 diabetes was induced in 2-month-old CD1 mice that were then randomly allocated into two groups receiving insulin (3 units daily Humulin U; Lilly) (6) or vehicle (n = 8 per group). Another set of nondiabetic mice was used as controls (n = 8). After 14 days, ischemia was induced in all three groups. Blood flow was measured at 30 min, 7 days, and 14 days. Mice were then killed for evaluation of adductor microvascular density.

Real-time quantitative PCR (ABI Prism 7000 sequence detection system software, version 1.0; Perkin Elmer) was used to determine eNOS and VEGF-A mRNA levels in muscles harvested from type 1 diabetic (untreated or preventively given Ad.hTK) or nondiabetic mice (n = 6 per group) before or 10 days after ischemia. Total RNA was isolated using TRIzol reagent (Invitrogen) and subsequently treated with DNase (Qiagen) to avoid DNA contamination. DNase-treated RNA was reverse transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen). The sequences of primers for amplification of murine eNOS (based on the cDNA sequences from the Genebank database NM-008713) were 5′ CCTTCCGCTACCAGCCAGA 3′ (forward) and 5′ CAGAGATCTTCACTGCATTGGCTA 3′ (reverse). VEGF-A (from Genebank sequence M95200) primers were: 5′ CCA GCG AAG CTA CTG CCG TCC A′ 3′ (forward) and 5′ ACA GCG CAT CAG CGG CAC AC 3′ (reverse). Primers for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (based on Genebank database sequence NM-008084) were 5′ CGT GGG GCT GCC CAG AAC AT 3′ (forward) and 5′ TCT CCA GGC GGC ACG TCA GA 3′ (reverse). ENOS or VEGF-A cDNA levels were normalized to GAPDH housekeeping gene cDNA levels.

Results are expressed as the means ± SE. Multivariate repeated-measures ANOVA was performed to test for interaction between time and grouping factor. In multiple comparisons among independent groups in which ANOVA and F test indicated significant differences, statistical value was determined according to Bonferroni’s method. Differences within and between groups were determined using paired or unpaired Student’s t test, respectively. P < 0.05 denoted statistical significance.

Spontaneous capillarization response to limb ischemia was impaired in type 1 diabetic mice (Figs. 1A and B). Insulin supplementation was able to control glycosuria (data not shown), but not to restore proper capillarization (Fig. 1C). Diabetes also blunted the increase in arteriole density in response to ischemia (Fig. 2A), with no compensation by insulin (Fig. 2B).

Endothelial cell proliferation (as assessed by PCNA staining) was similar in ischemic muscles of healthy or type 1 diabetic mice at 3 or 14 days after surgery (3.25 ± 0.52 vs. 2.92 ± 0.52 PCNA-positive endothelial cells/mm² at 3 days, P = NS) (data not shown). However, in diseased animals the effort to build up new vessels was frustrated by enhanced late activation of endothelial cell apoptosis (Fig. 3A). The effect was confirmed after normalization of the number of apoptotic endothelial cells by capillary density (37 ± 20 vs. 4 ± 3 apoptotic endothelial cells/1,000 capillaries in contralateral normoperfused muscles, P < 0.05). Myocyte apoptosis was likewise activated in ischemic muscles of type 1 diabetic mice (Fig. 3B).

Ad.hTK did not alter body weight or glycosuria as compared with other treatment groups. We found that hTK potentiates limb microvessels of type 1 diabetic mice. In fact, in adductors not subjected to ischemia, Ad.hTK increased capillary density (883 ± 60 vs. 547 ± 19 capillaries/mm² in contralateral, P < 0.001) and arteriole density (7.3 ± 1.4 vs. 2.8 ± 0.5 arterioles/mm² in contralateral, P < 0.05). In contrast, Ad.Luc did not alter adductor vascularity (587 ± 45 capillaries/mm² and 2.9 ± 0.8 arterioles/mm², P = NS vs. contralateral). Saline was also ineffective (data not shown).

Furthermore, prophylactic local administration of Ad.hTK restored the ability to mount a proper reparative neovascularization response to ischemia. Capillarization was increased by 40% in Ad.hTK-pretreated muscles, whereas the angiogenic response to ischemia was virtually absent in controls (Figs. 4A and B). Likewise, Ad.hTK increased arteriole density (Fig. 5).

As shown in Fig. 6, Ad.hTK-pretreated muscles displayed an increased number of PCNA-positive endothelial cells at 3 days from ischemia, indicative of activated endothelial cell proliferation (5.27 ± 0.27 vs. 2.92 ± 0.52 PCNA-positive endothelial cells/mm² in saline-injected muscles, P < 0.01); however, the effect was lost at 14 days (data not shown). Thus, hTK confers endothelial cells with improved proliferative potential at the early phases of the healing process, when the sprouting of new capillaries typically occurs. In addition, Ad.hTK pretreatment drastically reduced ischemia-induced endothelial cell and myocyte apoptosis (Figs. 7A and B). Results are further illustrated in Fig. 7C, showing TUNEL-positive cells in ischemic muscles of healthy (i) or type 1 diabetic mice given saline (ii) or Ad.hTK (iii).

In nonischemic muscles from type 1 diabetic mice, the expression of eNOS and VEGF-A was reduced to 10 and 28% the levels seen in healthy mice, respectively (P < 0.05 for both comparisons). ENOS and VEGF-A expression remained downregulated after induction of ischemia in type 1 diabetic mice injected with saline (data not shown). In contrast, Ad.hTK-pretreated muscles showed twofold increased eNOS mRNA levels after ischemia (P < 0.05) but no change in VEGF-A (P = NS).

Ad.hTK-pretreated type 1 diabetic mice displayed significantly improved limb hemodynamics, matching the recovery rate seen in nondiabetic mice (Fig. 8A). Insulin did not ameliorate blood flow recovery (Fig. 8B).

Type 1 diabetes is associated with accelerated atherosclerosis leading to premature myocardial, cerebral, and peripheral ischemia. Moreover, type 1 diabetic patients manifest a more severe course after vascular occlusion, including more frequent postinfarction angina, heart failure, cerebrovascular insufficiency, and nonhealing limb ulcers. The present study confirms that severe impairment in collateral growth contributes to the poor clinical outcome (3,4,10).

We explored the possibility that endothelial cell biology derangement may play a role in the deficit. In vitro conditions simulating the type 1 diabetes microenvironment reportedly reduce endothelial cell proliferation rate and branching angiogenesis (11,12). Excessive activation of apoptosis has been documented for endothelial cells cultured in high glucose or glycated collagen (13,14). Furthermore, seminal in vivo studies demonstrated that type 1 diabetes is responsible for excessive microvascular cell apoptosis in limb muscles (6) and retina (15).

Progression of diabetic microangiopathy follows a slow, inexorable course (6); however, as shown here, dramatic accelerations may occur after arterial occlusion. We documented that endothelial cell apoptosis is maintained at low levels in ischemic muscles of healthy mice. In sharp contrast, the cooperation of ischemia and diabetes dramatically increased endothelial cell death. Thus, uncontrolled apoptosis may jeopardize the attempt to build up functional collateralization.

It should be pointed out that apoptosis was mainly activated during late stages of the repair process. In this critical phase, the provisional network of endothelial tubes is, in part, dismantled and replaced by new intimal and medial smooth muscle cells instrumental in creating arteriolar connections and restoring proper circulation (16). However, arteriole remodelling could be seriously compromised if angiogenesis terminators are excessively upregulated, as reportedly occurs under conditions of hyperglycemia (17,18). Consistently, we found that reparative arteriogenesis is profoundly depressed in type 1 diabetic mice.

In previous studies, we demonstrated that diabetic microangiopathy is not an incurable condition. In fact, without the need of superimposed metabolic control, gene therapy with hTK was able to prevent and rescue microvascular rarefaction (6). Here, we report the discovery that hTK gene transfer protects type 1 diabetic muscles from the consequences of supervening ischemic insult. Adenovirus-mediated gene transfer preceded the induction of ischemia by 2 weeks. We know from previous studies that this time is sufficient for transgene expression to expire (7). Therefore, the beneficial action of hTK on postischemic recovery can be considered truly prophylactic. Ad.hTK-injected muscles displayed improved endothelial cell proliferative potential during early phases of postischemic recovery, and they displayed reduced apoptosis at vascular and myocyte levels during later stages of healing. Improved capillarization by Ad.hTK may benefit myocyte viability by increasing oxygen and nutrient diffusion to the muscle.

VEGF-A and eNOS expression was found to be reduced in type 1 diabetic muscles and not modulated by ischemia, which is in agreement with defective VEGF-A and NO production in diabetic animal models (3,19). Impaired eNOS expression was partially restored by hTK pretreatment. This effect could be relevant to protection from ischemia-induced apoptosis. In fact, activation of eNOS, leading to enhanced synthesis of NO, reportedly promotes endothelial cell survival via inhibition of pro-apoptotic caspases and induction of anti-apoptotic protein expression (20). Similar to NO, VEGF exerts inhibitory control of apoptosis (21). However, diabetes-related VEGF-A liabilities were not corrected by hTK. These findings favor the possibility that hTK-induced microvascular effects are mediated by a PI 3-kinase–NO–Akt mechanism independently of VEGF-A, as recently documented in a separate study (C.E., unpublished data). Other unforeseen mechanisms implicated in the pro-survival activity of Ad.hTK merit further investigation.

We documented that insulin—known to activate the pro-angiogenic and anti-apoptotic PI 3-kinase–Akt pathway in endothelial cells (22)—could not restore the proper neovascularization response to muscular ischemia in type 1 diabetic mice. Although metabolic control was seemingly achieved by the use of a standardized dose (6), it might be possible that earlier treatment or species-specific insulin should have been used to obtain significant microvascular effects.

The theory that inhibition of endothelial cell apoptosis may facilitate angiogenesis has been already introduced by others (23). Nevertheless, the present study is the first to document the feasibility of improving the phenotype of diabetic endothelial cells to let them regain resistance to ischemic damage. Prophylactic gene therapy with Ad.hTK seems particularly attractive for therapeutic exploitation because of its ability to improve survival and further stimulate endothelial proliferation, which is also required for neovascularization. However, much more work is necessary before proposing hTK gene transfer for clinical use, which includes testing therapeutic efficacy in different diabetic models, tailoring new vector systems, and testing more in depth the absence of adverse effects.

Altogether, our results support the therapeutic utility of prophylactic angiogenesis gene therapy with hTK for secondary prevention of type 1 diabetic vascular complications. In particular, prophylactic hTK gene transfer might be envisaged to benefit diabetic patients at risk of occlusion of limb arteries.

This work was supported by grants from the Juvenile Diabetes Research Foundation (2000/49) and the European Association for the Study of Diabetes (to P.M.) and from the Italian Health Minister (Ricerca Finalizzata 2002; to C.E.).

The authors thank Drs. Cosseddu and VanLinthout (both from INBB) for gene expression experiments.