Anti-angiogenic effects of interleukin-12 delivered by a novel hyperthermia induced gene construct

Abstract

Purpose: Interleukin-12 (IL-12) is a pro-inflammatory cytokine possessing anti-cancer and anti-angiogenic properties. This study quantitatively assessed the anti-angiogenic effect of IL-12 delivered using an adenoviral vector with murine IL-12 placed under control of a heat shock promoter. This approach limits systemic toxicity by restricting IL-12 delivery locally to the tumour. The kinetics of the downstream cytokines interferon-γ (IFN-γ) and interferon inducible protein-10 (IP-10) and other molecules affecting angiogenesis, vascular endothelial growth factor (VEGF) and plasminogen activator inhibitor-1 (PAI-1) were also studied.

Materials and methods: 4T1 tumours were grown in Balb/C mice and the AdhspmIL-12 construct was injected intra-tumourally. The tumours were heated after 24 h using a water bath. At various time points post-heating the tumours were collected and quantitatively assessed for cytokine production and vascularity.

Results: A significant reduction was seen in the tumour vasculature of the treated group vs. the control group mice. Systemic effects of IL-12 were limited to generalized immunostimulation. No hepatoxicity was noted.

Conclusions: This study suggests that IL-12 can be effectively delivered using a gene-based approach with a heat shock promoter. This results in quantitatively measurable anti-angiogenesis and general immunostimulation. The complex inter-play of other pro- and anti-angiogenic factors (IFN-γ, IP-10, VEGF and PAI-1) was also studied.

• Anti-angiogenesis
• interleukin-12
• hyperthermia

Introduction

Interleukin-12 (IL-12) is a potent pro-inflammatory cytokine possessing anti-cancer properties. A review on IL-12 by Colombo and Trinchieri [1] outlines its anti-tumour, anti-metastatic, anti-angiogenic properties and the ability to elicit long-term anti-tumour immunity. Voest et al. [2] first studied the role and possible mechanisms of action of IL-12 in anti-angiogenesis. The effect of IL-12 against basic fibroblast growth factor (bFGF) induced corneal neovascularization in mice was evaluated. IL-12 (1 µg per day) was given intra-peritoneally for two cycles of 5 days each with a 2-day interval. They found significant reduction in bFGF induced corneal neovascularization in the IL-12 treated mice. This effect was lost when the mice received a neutralizing IFN-γ antibody, suggesting that IFN-γ was an important mediator in the anti-angiogenic effects of IL-12. They proposed that IFN-γ has two major effects; (a) it induces production of interferon inducible protein-10 (IP-10), a member of the –C–X–C– chemokine family, which is directly anti-angiogenic and (b) IFN-γ inhibits metalloproteinase production. Metalloproteinases break down the extra-cellular matrix allowing new capillary sprouts to grow. IFN-γ induces production of soluble anti-angiogenic factor interferon-inducible protein 10 (IP-10) [3]. IP-10 is expressed in activated mononuclear cells, keratinocytes, fibroblasts, endothelial cells and T-cells. It was found to inhibit bFGF-induced neovascularization of Matrigel injected subcutaneously into athymic mice. IP-10 also suppressed endothelial cell differentiation into tubular capillary structures in vitro [4]. It has been suggested that the balance between the local concentrations of pro-angiogenic and anti-angiogenic factors within a tumour dictates whether angiogenesis occurs. Since angiogenesis is required for tumours and metastases to grow, angiogenesis inhibition is thought to be an important mechanism by which IL-12 controls tumour growth. Hence, IL-12 is proposed to act through the activation of IFN-γ/IP-10 axis. Other proposed mechanisms implicate direct natural killer (NK) cell mediated endothelial cytotoxicity [5] and downregulation of the VEGFR-3 receptor [6].

Local and systemic administration of IL-12 protein has been studied in various murine models [7–10] and in Phase I/II human trials [11], [12]. IL-12 protein therapy has largely been limited by dose-dependent toxicity [13]. Human recombinant IL-12 is associated with severe toxicity, even at doses as low as 1 µg kg⁻¹ per day [14]. Common toxicities included fever/chills, fatigue, nausea, vomiting and headache. Routine laboratory changes included anaemia, neutropenia, lymphopenia, hyperglycemia, thrombocytopenia and hypoalbuminemia. Dose limiting toxicities included oral stomatitis and liver function test abnormalities, predominantly elevated transaminases. Also, poor accumulation of IL-12 in solid tumours after intravenous injections (tumour:organ ratio <1) may partly explain its minimal anti-tumour effects [15].

Local expression of IL-12 may avoid systemic toxicity of recombinant interleukin-12 administration [16–18]. Intra-tumoural injections of adenoviral vectors or IL-12 plasmid DNA, naked or in complex with cationic lipid, have been attempted to deliver therapeutic IL-12 with the rationale that IL-12 will be produced only in the tumour, thereby reducing systemic side effects [19]. A few reports have indicated the efficacy of this approach [20–22], however, elevated systemic transgene levels were observed as the adenovirus can reach the circulation infecting other organs [23–25]. The promoters in most previous reports are constitutively active, such as cytomegalovirus (CMV) based promoters. This combination makes it likely that intra-tumoural injection of this constitutively active adenovirus approach will still result in toxicity [26], [27].

To achieve localized gene expression, an adenoviral vector was developed with murine IL-12 placed under control of a heat inducible promoter, hsp 70B. This promoter is highly inducible at temperatures between 39–42°C and large fold-inductions in downstream transgene expression can be achieved [28–30]. Heating the tumour after local injection of the adenoviral gene-therapy vector leads to activation of the hsp promoter and subsequent local IL-12 production, thus providing spatial and temporal control over transgene expression. The feasibility of combining radiotherapy, hyperthermia and heat inducible murine IL-12 gene therapy in a non-immunogenic B16.F10 melanoma line was studied [31]. It was concluded that hyperthermia-regulated gene therapy in combination with radiation is feasible and therapeutically effective in murine tumours with no apparent systemic toxicity [31], [32].

This study assessed the dynamics and inter-play of five cytokines known to affect tumour angiogenesis (IL-12, IFN-γ, IP-10, VEGF and PAI-1) and quantitatively measured anti-angiogenesis using lectin intra-vital perfusion (FITC-conjugated Lycopersicon esculentum lectin) and vascular antigen staining (CD31). A significant reduction in vasculature was noted in the Ad hsp murine IL-12 + hyperthermia treated group as compared to the controls. Indications of the anti-metastatic and generalized immunostimulatory properties of IL-12 were also noted with little or no evidence of systemic toxicity.

Materials and methods

Animals

BALB/c mice, 9–12 weeks of age, weighing 20–25 g, were obtained from the Laboratory Animal Resources at Colorado State University.

Design of vector

The AdEasy system (Stratagene, La Jolla, CA) was used to construct the AdhspmIL12. The two sub-units (p35 and p40) of the murine IL-12 were amplified and sequence-verified. They were then connected into one gene expression unit by use of a flexible linker sequence (Gly4Ser). A 400-bp hsp70B promoter was then used to control the expression of the modified murine IL-12 gene. The whole hsp-mIL12 gene expression cassette was then transferred into an adenovirus shuttle plasmid. The shuttle plasmid was cotransfected into 293 cells to derive AdhspmIL12. Amplification and isolation of the virus was achieved following standard protocols [33], [34].

Tumour cell injection

4T1 cells were obtained from American Type Culture Collection (ATCC); 5 × 10⁵ cells in 50 µl of PBS were injected into the left hind legs of the mice. Eight-to-ten days later the tumours reached a volume of 200 mm³. Daily tumour measurements were done using calipers. The tumour volume was calculated using the formula (short axis)² × long axis × π/6 mm³. Mice were euthanized at a tumour volume of ∼1000–1200 mm³. This study was approved by the Colorado State University Animal Care and Use Committee.

Treatment groups

Tumour bearing mice were divided into the following treatment groups: (1) Control group receiving intra-tumoural normal saline injection 50 µl; (2) Intra-tumoural injection of 10⁸ pfu of Ad LacZ (empty vector) suspended in 50 µl of normal saline + HT; (3) Intra-tumoural injection of 10⁸ pfu of AdhspmIL-12 suspended in 50 µl of normal saline (No HT); (4) Intra-tumoural normal saline injection 50 µl + hyperthermia (HT); and (5) Intra-tumoural injection of 10⁸ pfu of AdhspmIL-12 suspended in 50 µl of normal saline + HT.

All intra-tumoural injections were delivered 24 h prior to delivering HT through a single entry point. Multiple tracks were made in the tumour to achieve a more homogeneous distribution.

Hyperthermia

The mice were anaesthetized using 2% isoflurane and the tumours heated with the tumour bearing leg immersed in a water bath. The tumours were heated at 41°C for 60 min.

In vivo hypoxia and vascular markers

Hypoxyprobe-1 (Chemicon International, Temecula, CA) was injected intra-peritoneally at a dose of 75 mg kg⁻¹ dissolved in 100 µl of normal saline 90 mi before euthanasia. Lycopersicon (tomato plant) lectin (Vector Laboratories, Burlingame, CA) conjugated with FITC was used as a marker for functional vasculature. It was injected via the tail vein in a dose of 100 µg in 100 µl of normal saline 3 min prior to euthanasia.

Tissue harvesting

At 6 h, 24 h and 5 days post-hyperthermia mice were euthanized by cervical dislocation under anaesthesia. Within 1 min of death, part of the tumour was collected in a cryotube (for RT-PCR) and part embedded in Tissue-Tek optimum cutting temperature compound (OCT, for imaging). Both parts were immediately frozen in liquid nitrogen. Lungs, liver and spleen were collected over the next few minutes and preserved in formalin. Haematoxylin and eosin (H&E) stains were done on these for histological examination.

RNA isolation and cDNA synthesis

Tumour samples were lysed and RNA isolated using TRIzol Reagent (GibcoBRL) as per manufacturer's protocol. RNA purity was assessed spectrophotometrically. Total isolated RNA was treated with DNase I (Invitrogen) to remove any genomic DNA. cDNA synthesis was then carried out using Superscript II RNase H⁻ Reverse Transcriptase (Invitrogen, Carlsbad, CA). Each reaction mixture contained 10 µl of RNA solution to which was added 4 µl First strand buffer (5×), 1 µl dNTP (10 mM), 1 µl DTT (0.1 M), 0.25 µl RNase out (40 U µl⁻¹), 0.25 µl SuperScript II (200 U µl⁻¹), 2 µl of random hexamers (300 ng µl⁻¹) and 1.5 µl DNase, RNase free water giving a total of 20 µl per reaction mixture. This was incubated at 42°C for 50 min following which 30 µl of DNase, RNase free water was added to the mix and the enzymes inactivated by placing the reaction tubes on a 95°C heat- block for 5 min. The cDNA was then stored at −20°C until the time of RT- PCR.

Real time PCR

Real-time reverse transcriptase PCR was employed to detect and quantitatively express the production of murine genes of interest and GAPDH mRNA. Sequences for the primers and probes were obtained from the literature [35–38] and purchased from MWG Biotech (High Point, NC). The reporter dye attached covalently at the 5’ end was FAM (6-carboxyfluorescin) and the quencher bound to the 3’ end was TAMRA (6-carboxytetramethylrhodamine). Real time PCR was performed using the Applied Biosystems ABI Prism 7000 (Foster City, CA). The amplification protocol was: 2 min at 50°C, 10 min at 95°C, 45 cycles of 15 s at 95°C and 60 s at 60°C.

Cycle threshold values were obtained from the ABI software and were exported to Microsoft Excel and the 2^−ΔΔCt method was used to determine the relative expression of the genes of interest [39]. Briefly, a standard housekeeping gene (e.g. GAPDH, β-actin, β₂ microglobulin) is chosen as an internal control gene. This serves to normalize the amount of cDNA loaded for each reaction. An untreated control was selected as the ‘calibrator’ and the relative expression data is obtained as the fold change in gene expression normalized to the chosen endogenous reference gene and relative to the untreated control.

Prior to using the 2^−ΔΔCt method for relative quantification, its use was validated for the primer and probe sequences and PCR conditions being employed in the experimental conditions. The methods and acceptable results are described in the literatu [39], [40]. Briefly, the target cDNA was serially diluted and real-time PCR performed. Cycle threshold values for the serially diluted samples are plotted on a graph with the log dilution on the x-axis and cycle threshold values on the y-axis. The slope of the line is used to obtain the efficiency of PCR amplification using the formula: Efficiency = 10^(−1/slope) − 1.

Imaging for functional vasculature

Five micron thick frozen sections of the tumours were stored in the dark at −80°C. Tumour sections were imaged for FITC fluorescence using a Zeiss Axioplan 2 microscope (Carl Zeiss) with the KS 400 image analysis software. An area of the tumour away from the periphery was chosen for image capture. Images were captured at 100× magnification. Five-to-six separate images from four-to-five mice per group were captured for further image analysis. The x- and y-coordinates of each image were recorded to enable the same area to be imaged after CD31 and hypoxyprobe staining.

Immunohistochemistry

Immunohistochemical staining was performed using standard manual techniques on a vertical slide staining system (Credenza, Thermo Electron Corporation, Philadelphia, PA). Cryosections were fixed in −20°C acetone for 10 min and then allowed to dry at room temperature for 30 min. All reagents and kits were purchased from Vector Laboratories, Burlingam, CA unless otherwise noted. In all cases a control section was processed in which the primary antibody was replaced with antibody diluent. For monoclonal antibodies another negative control was run with an isotype similar non-specific antibody replacing the primary antibody at a similar concentration.

Hypoxyprobe/CD 31 staining

Identification of hypoxic regions were in fixed sections stained with the horseradish peroxidase based M.O.M. immunodetection kit with a hypoxyprobe 1-Mab1 (Chemicon International, Temecula, CA) as the primary Ab at 1:50 dilution in antibody diluent (DakoCytomation, Carpinteria, CA). DAB substrate was used to visualize the immunoreactive complexes. Tumour vessels were identified in sections incubated overnight in a rat monoclonal anti-mouse CD31 primary antibody (Clone MEC 13.3, BD Pharmingen, San Diego, CA) at 1:100 dilution in antibody diluent. A secondary biotinylated rabbit anti-rat IgG antibody at 1:100 dilution was applied and followed with a horseradish peroxidase streptavidin at 1:250 dilution in PBS. A VIP peroxidase substrate kit was used for visualization. Sections were dehydrated through ascending serially concentrated alcohol baths and mounted in a xylene-based permanent mounting media.

IL-12/IP-10 staining

Fixed sections were incubated overnight in primary antibody, rat monoclonal anti-mouse IL-12 (Clone C15.6, Biosource, Camarillo, CA) at 1:100 dilution in antibody diluent or were incubated in rabbit anti-mouse IP-10 (Peprotech, Rocky Hill, NJ) at 1:200 dilution in antibody diluent. A biotinylated secondary rabbit anti-rat IgG or goat anti-rabbit IgG at 1:100 dilution was applied and followed with a horseradish peroxidase streptavidin at 1:250 dilution. DAB substrate was used to visualize the immunoreactive complexes. Sections were dehydrated through ascending serially concentrated alcohol baths and mounted in a xylene-based permanent mounting media.

Image analysis

The saved images were quantitatively analysed for the parameters of interest using Scion Image (NIH). The ‘density slice’ was fixed for a colour of interest (FITC–green, CD31–blue, hypoxyprobe–orange/red) and the area occupied by the stain was obtained from the software expressed as pixels. All images for a particular colour were analysed using the same ‘density slice’.

Statistics

Pixel areas for FITC, CD31 and hypoxyprobe staining were compared between the treatment groups using the ANOVA test with pairwise comparison between groups.

Examination of lungs, liver and spleen

The lungs harvested from the control mice and the AdhspmIL-12 + HT treated mice were paraffin embedded, sectioned 10 µm thick and examined for the presence of metastases. Random sections of both lungs from the five mice in each group were stained using haematoxylin and eosin. The total number of metastases counted in 10 lungs each (right and left lungs) from the control and treated groups were compared.

The livers of the mice in all groups were paraffin embedded, sectioned 10 µm thick and examined for evidence of any hepatocellular changes, specifically hepatocellular necrosis. The spleens of all mice were similarly sectioned and examined for any evidence of generalized immunostimulation.

Results

Cytokine RT-PCR

Real time PCR was performed to quantitatively detect IL-12, IFN-γ, IP-10, PAI-1 and VEGF mRNAs with GAPDH as the housekeeping gene. Initially, experiments were performed to detect efficiency of amplification for the genes of interest. After ascertaining that the PCR amplification efficiency was in the 90–95% range using the primer and probe sequences, the real time PCR for various treatment groups was performed. Results are shown in Figure 1. Each bar represents the average ± standard error of mean of cytokine expression in tumours from five mice with each sample run in triplicate. The expression levels in the treatment groups are normalized to control group (group 1).

Figure 1. Relative expression of mRNAs in the treated groups 2–5c normalized to group 1 (each bar represents samples from five mice each in triplicate ± SEM). (a) Interleukin-12 (IL-12); (b) Interferon-γ (IFN-γ); (c) Interferon inducible protein-10 (IP-10); (d) Plasminogen activator inhibitor (PAI-1); and (e) Vascular endothelial growth factor (VEGF). Group 1: Normal saline injection, Control. Tumour and tissue samples collected at the same time as treatment group (5 days post-HT); Group 2: Ad LacZ + HT. Tumour and tissue samples collected 5 days post-HT; Group 3: AdhspmIL-12, no HT. Tumour and tissue samples collected at the same time as treatment group (5 days post- HT); Group 4: Normal saline injection, HT only. Tumour and tissue samples collected 5 days post-HT; Group 5: AdhspmIL-12 + HT (Group 5a: tumour sampled 6 h post-HT; Group 5b: tumour sampled 24 h post-HT; Group 5c: tumour sampled 5 days post-HT).

Maximum IL-12 mRNA levels were seen in the tumour samples collected 6 h post- HT (group 5a). By 24 h post-HT (group 5b) the level had reduced by ∼2.5 logs (Figure 1a). IL-12 mRNA was detectable up to 5 days post-HT (group 5c). IL-12 mRNA was also elevated in group 3, where the AdhspmIL-12 had been injected intra-tumourally but no hyperthermia had been delivered. This suggested a baseline level of hsp induction by factors other than hyperthermia. Peak IFN-γ mRNA levels were noted 24 h post- HT (Figure 1b) corresponding to the time-point at which the IL-12 protein levels (16–20 h after IL-12 mRNA peak) would likely be highest. IP-10 mRNA also followed the same trend. However, as opposed to IL-12 and IFN-γ mRNA levels, which were still high in group 5c, the IP-10 levels had returned to baseline by 5 days post-treatment (Figure 1c). PAI-1 mRNA levels were higher in group 4 (HT only) than in other groups where hyperthermia was combined with adenoviral infection (groups 2 and 5) (Figure 1d). VEGF levels were elevated in all treated groups (Figure 1e). There were highly varying fold-increases in mRNA within the groups, as represented by the large error-bars. In groups 5a–c, VEGF levels appeared to be inversely related to the IFN-γ mRNA levels PAI-1 too.

Tumour growth curves

After the tumours reached sizes of ∼3–4 mm in diameter, daily tumour measurements were done using calipers and volumes calculated. Relative tumour volumes are shown in Figure 2. A statistical analysis using the Kruskal-Wallis test revealed a statistically significant difference between the control and AdhspmIL-12 + HT group (p = 0.017) and between AdhspmIL-12 alone and AdhspmIL-12 + HT (p = 0.02). There was no statistically significant difference between any other groups compared pairwise.

Figure 2. Tumour growth curves. The relative tumour volumes for the treatment groups are shown. A statistically significant difference was noted between the control and virus + hyperthermia groups (p = 0.017) and between AdhspmIL-12 alone vs. AdhspmIL-12 + HT groups (p = 0.02). However, on pairwise comparisons using the Kruskal-Wallis test no significant difference was found between any other group.

Image analysis for vascular area

Five-to-six tumour sections from four-to-five mice per group were examined at 100× for Tomato lectin-FITC fluorescence and CD31 staining (Figure 3). Image analysis was done and the results are shown in Figure 4.

Figure 3. Tomato lectin-FITC and CD31 & hypoxyprobe staining. (a) Tomato lectin-FITC. (b) Same coordinates on the slide imaged for CD31 and hypoxyprobe staining. Area stained by CD31 is larger than FITC indicative of greater presence of the antigen than is actually patent functional vasculature. (c) Image from a control tumour section showing the patent vasculature in green. (d) IL-12 treated tumour 5 days post-hyperthermia showing reduced vasculature. (e) Section from a control group tumour showing large areas staining positive for CD31 antigen with small areas of hypoxia. (f) Section from a treated group tumour showing reduced CD31 staining and larger areas of hypoxia.

Figure 4. Area occupied by FITC and CD31 staining. (a) Percentage of field positive for Tomato lectin-FITC fluorescence, suggestive of functional vasculature. (b) Percentage of field staining positive for CD31 vasculature. The groups mentioned in the graphs are as follows: (1) Normal saline injection, Control. Tumours collected at the same time as treatment group (5 days post-HT); (2) Ad LacZ + HT. 5 days post-HT; (3) AdhspmIL-12, no HT. Same time as treatment group (5 days post-HT); (4) Normal saline injection, HT only. 5 days post-HT; (5) AdhspmIL-12 + HT. 5 days post-HT. Reduction in tumour vasculature was statistically significant (p < 0.05) in group 5 as compared to all other groups and for group 3 when compared to Control group (group 1).

The area occupied by functional patent vasculature (FITC staining) is less than the area staining for total vascular antigens (CD31 staining). In the control group it was 1.05 ± 0.85% and 3.5 ± 1.75% for FITC and CD31, respectively, while in the treatment group (group 5c) it was 0.26 ± 0.22% and 0.84 ± 0.62% for FITC and CD31.

A statistically significant (p < 0.05) decrease in tumour vasculature was seen in group 5 as compared to all other groups for functional patent vasculature (FITC) and for total vascular antigen (CD31). In group 3, where the vector had been injected but the tumour was not heated, the difference was also significant as compared to controls.

Image analysis for hypoxia

Hypoxyprobe staining was done to assess areas of tumour hypoxia in the different groups. Figure 5 shows the results. In the control group and groups 2–4 the tumour hypoxic area was 9.5–11.5% with standard deviations of 6–7%, however in the AdhspmIL-12 + HT group, the hypoxic area was 19.8 ± 7.5%. This increase was statistically significant (p < 0.05).

Figure 5. Area occupied by hypoxyprobe staining. The hypoxic area was higher for the AdhspmIL-12 + HT treated group as compared to all other groups. The groups mentioned in the graphs are as follows: (1) Normal saline injection, Control. Tumours collected at the same time as treatment group (5 days post- HT); (2) Ad LacZ + HT. 5 days post-HT; (3) AdhspmIL-12, no HT. Same time as treatment group (5 days post-HT); (4) Normal saline injection, HT only. 5 days post-HT; (5) AdhspmIL-12 + HT. 5 days post-HT.

Cytokine staining

The IL-12 stain is cytoplasmic and multi-focally distributed throughout the tumour section (Figure 6) in the AdhspmIL-12 + HT group. In the other groups (Control, Ad LacZ + HT, AdhspmIL-12 alone and HT alone) no IL-12 was detected in the immunohistochemically stained tumour sections. An estimate for the area staining positive for IL-12 was 5–10% for the tumours collected 24 h post-HT and 10–20% for the tumours collected 5 days post-HT.

Figure 6. Interleukin-12 staining. The red circles outline some of the areas staining positive for IL-12 in a haematoxylin background in the AdhspmIL-12 + HT group. This image is from one of the tumours collected 5 days post HT (100×). In the control, Ad LacZ + HT, AdhspmIL-12 alone and HT alone groups no IL-12 was detected in the immunohistochemically stained tumour sections.

Lung metastasis

To assess the anti-metastatic properties of IL-12, random histopathologic sections of both lungs from all five mice in the AdhspmIL-12 + HT group were compared to the five mice from the control group. No lung metastasis was identified in the five mice in the AdhspmIL-12 + HT group compared to a total of nine metastases noted in 10 individual lungs of the control group. Most lung metastases were aggregates of a few tumour cells, however, some of them appeared as large sheets of tumour cells replacing the normal lung parenchyma, as shown in Figure 7.

Figure 7. Lung metastases. Metastases were detected in the lungs from the mice in the control group. The metastases were seen as sheets of tumour cells replacing the normal lung architecture. (a) Lung metastasis at 40× magnification. (b) Lung metastasis at 200× magnification.

Liver and spleen histology

Mice livers were examined for pathological changes reflective of IL-12 toxicity with H&E stained sections. Liver tissue was normal with no hepatic necrosis noted in the control or the treated groups.

The spleen in the IL-12 treated mice had multiple coalescing large secondary follicles with large expanded germinal centres. Lymphoid follicles in the spleens of control mice were less prominent, smaller and had no or smaller germinal centres (Figure 8). This suggested generalized immunostimulation in the mice where the IL-12 was expressed.

Figure 8. Spleen histology. (a) Spleen from a control group mouse (40×). (b) Spleen from a mouse euthanized 5 days post-HT showing enlarged germinal centres (the yellow arrows outline one such germinal centre). Many follicles appear to be coalescing together.

Discussion

Tumour anti-angiogenesis was studied using a hyperthermia-inducible gene construct to deliver IL-12 intra-tumourally. This construct has been designed to achieve a better spatial and temporal control over IL-12 expression. The aim was to have high tumour levels of the cytokines with limited systemic circulation to prevent generalized haematologic and hepatic toxicity. The vector was injected into the tumour and 24 h allowed for adenoviral infection in order to achieve maximum possible efficiency of infectio [32], [41]. Hyperthermia was then delivered to the tumour bearing leg using a water bath as previously reported[31], [32], [42].

Interleukin-12 mRNA levels have been reported to be maximally induced at 4–12 h after stimulation or induction [43–45]. In in vitro studies the time point of maximal IL-12 mRNA levels had been found to be 6 h post- hyperthermia [30]; thus this time point was selected for tumour sample collection in the AdhspmIL-12 + HT group to measure cytokine levels. Based on the in vitro experiments and from data in the literatu[30], [41], [42], the protein levels peak ∼16–20 h post-induction and a few hours following the mRNA peak. Hence, it was reasoned that the downstream cytokines IFN-γ and IP-10 would show their peak mRNA levels in the tumour sample collected 24 h post-HT. This temporal relationship was confirmed in the tumour tissue samples of this study. The 5 days post-HT tumour sample was collected for IHC for analysis of anti-angiogenesis. It was expected that the cytokine levels would be returning toward baseline by this time. However, the IL-12 and IFN-γ mRNA levels were still ∼1000-fold and 8-fold higher than baseline. In contrast the IP-10 mRNA levels had reached baseline at day 5.

PAI-1 levels were maximally induced in the group receiving HT only (group 4). In a prior report, Roca et al. [46] studied the anti-angiogenic properties of hyperthermia. They heated cultured human endothelial cells at 39°C, 41°C, 43°C and 45°C for 60 min and determined cell viability immediately and 16 h after heating. cDNA expression array was used to study the gene expression profile of endothelial cells after heat shock. It was determined that hyperthermia activates a gene response involving the transcription of plasminogen activator inhibitor (PAI-1), a regulator of the plasminogen activation pathway involved in vascular thrombosis, metastasis diffusion, inflammation and angiogenesis. A dose-dependent inverse relationship between IFN-γ and PAI-1 has been reported [47–49]. A similar inverse relationship was present in this study and would explain why the hyperthermia alone group, in which no IFN-γ was induced, had higher levels of PAI-1 expression than the AdhspmIL-12 + HT group.

Elevated levels of VEGF mRNA were seen in all the treatment groups. Any anti-angiogenic measure such as hyperthermia [50] or interleukin-12 could possibly result in increased VEGF to counter the anti-angiogenesis [51]. An inverse relationship has been noted between IL-12 and VEGF [52–54] and also IFN-γ and V [55], [56]. Maximum IL-12 protein levels, which would be expected ∼18–24 h post-HT, coincide with the lowest VEGF mRNA levels in groups 5a–c.

For tumour growth studies the 4T1 tumours growing in Balb/C mice were treated with HT alone or AdhspmIL-12 + HT and compared with the control group. A statistically significant reduction in tumour growth was observed in the AdhspmIL-12 + HT group as compared to the control group. The tumour growth curve for the Ad LacZ + HT was similar to that for HT alone. Intra-tumoural injection of AdhspmIL-12 alone resulted in a decreased rate of tumour growth as compared to the controls. However, a statistically significant difference was seen only when high levels of IL-12 were produced by hyperthermia in similarly injected tumours. The relative volumes in the AdhspmIL-12 + HT treated mice were lower than those in the HT only group, however this difference was not statistically significant. Similarly, the difference between HT only and control was not statistically significant. A similar study [32] using this construct in B16F10 melanoma cell line tumours had also shown a delayed tumour growth in the IL-12 treated group. The reduction in tumour vasculature in group 4 may be responsible for the delayed tumour growth. In vitro studies done on 4T1 cells did not show evidence of appreciable reduction in percentage of viable cells using the trypan blue dye exclusion test when heated at 39–42°C. Only when heated at 43°C, the cell viability was reduced to ∼75% (data not shown).

For studying patent functional vasculature FITC-conjugated Lycopersicon esculentum (Tomato plant) lectin was injected 3 min prior to euthanasia. Lectins are rapidly distributed through all patent vessels within a circulation period of ∼60 s in mice [57]. They bind to the glycoproteins expressed on the endothelial cells [58]. CD31 staining, routinely used to study angiogenesis and anti-angiogenesis in tissue sections, over-estimates vascularity of a tumour [59–62] as it stains endothelial cells in the tumour sections even in the absence of a patent vascular lumen. In this study, the area of the tumour section occupied by functional vasculature (∼1%) was one-third of the area staining positive for total vasculature (∼3%). CD31 positive area in the tumour have been reported as ∼3.5% for A431 human squamous cell carcinomas and B16F10 melanomas [63] to ∼6.5–7% for neuroblastomas [64] and LLC tumours [63].

The vascular percentage of the treated group was 24% and 25% of the control group by CD31 and FITC studies, respectively. These values are close to those reported in literature for IL-12 induced anti-angiogenesis in Balb/C and C57CL/6 mice using Matrigel assay [5] and in Balb/C mice in a breast carcinoma model [52]. A significant reduction in vascularity was also noted in group 3, where the construct had been injected but hyperthermia had not been delivered. Approximately a 500-fold increase in IL-12 mRNA levels and a 20-fold and 16-fold increase in IFN-γ and IP-10 levels, respectively, were also seen. This suggested that some factor, other than HT, was leading to hsp induction in the tumour micro-environment although in a less efficient manner.

Various methods to measure tumour hypoxia have been described in the literature and the percentage of hypoxic cells, as assessed by hypoxia marker drugs, have been found to range from 0–50% with most tumours having hypoxic areas between 10–15% [65]. The level of hypoxia in 4T1 tumours, using Hypoxyprobe-1, was found to be 16 ± 7% [66]. In the tumour sections, using the same compound, the tumour hypoxic area was 9.7 ± 7.6% to 11.6 ± 7.3% for groups 1–4. However, in treatment group 5 (AdhspmIL-12 + HT) the hypoxic area had increased to 19.8 ± 7.5%. This difference was statistically significant and reflective of the reduction in tumour vascularity.

Adenoviruses are efficient vectors for use in gene therapy protocols. In vitro they have been found to have a high percentage infection of cells exposed [67], [68]. Infection efficiencies of 90–95% were found in the in vitro experiments using a GFP expressing adenoviral vector. However, the in vivo efficiencies, by intra-tumoural injections, are much lower. In malignant glioma cells, Puumalainen et al. [69] reported adenoviral infection as low as <0.01–11%. Adenoviral infections are highly dependent on the coxsackievirus and adenovirus receptor (CAR) status [70], [71]. Gu et al. [72] found variations in CAR status among human musculoendothelial tumours. They found that CAR mRNA was expressed at highest levels in osteosarcoma, Ewing's sarcoma, neurofibroma and schwannoma; at intermediate levels in bony exostosis, giant cell tumour, liposarcoma, synovial sarcoma, malignant peripheral nerve sheath tumour and haemangioma; and at low levels in alveolar soft part sarcoma and desmoid. In this study, immunohistochemical staining for IL-12 showed very sparse and patchy distribution suggestive of very low transfection efficiency in vivo. The distribution was more widespread on the 6th day post-HT as compared to 24 h post-HT. Still, only ∼10–15% of the cells appeared to be producing IL-12. IP-10, the downstream cytokine, showed a comparatively larger distribution in the tumour sections. This could be from the paracrine effects of IL-12 and IFN-γ.

It was surprising to find IL-12 protein in the tumours present and better distributed on the 6th day post-induction. The plasma half-life of IL-12 has been reported to be ∼10 h [73]. This is much longer than the half-lives of other cytokines that generally have t_1/2 of less than 30 min. Its persistence in the tumour for up to and possibly beyond 6 days may be due to a reduced rate of clearance from the tumour, resulting in a longer effective half-life and/or initiation of a positive feedback loop for cytokine production.

The anti-metastatic properties of interleukin-12 have been extensively studied in the literature [1] and are attributed to generalized immunostimulation and anti-angiogenic properties of IL- [2], [5]. Rakhmilevich et al. [74] delivered IL-12 using a gene-gun mediated approach and found a significant reduction in the number of lung metastases from 4T1 tumours with an increase in the survival time of the mice. In the treated group no metastases were detected in the lungs of the five mice examined as compared to a total of nine individual metastases in five control mice. The number of metastases is much lower than those reported in other studies [75], [76] where a total of 30–40 individual metastases were counted 3–4 weeks after 4T1 cell line inoculation in the control group. In this study only one random H&E section of both lungs from each mouse was examined for metastases ∼3 weeks post-inoculation and, thus, the number of metastases in both groups (control and treated) would be low.

Administration of recombinant IL-12 can cause liver toxicity in the form of hepatic necrosis [77] or steatosis [78]. No liver pathology was noted using the localized method of delivery in this study. The spleens from the IL-12 treated group showed evidence for generalized immunostimulation in the form of coalescing follicles with enlarged germinal centres. Splenic lymphoid hyperplasia has been reported in squirrel monkeys systemically treated with recombinant human IL-12 [79]. This feature was not observed in the spleens of the control mice.

In conclusion, using the AdhspmIL-12 construct, one was able to deliver IL-12 intra-tumourally and demonstrate its potent anti-angiogenic effects and delayed tumour growth with limited systemic toxicity. This study also investigated the role of other pro- and anti-angiogenic factors and their complex inter-play in influencing tumour angiogenesis.