Evaluation of cancer virotherapy with attenuated replicative Semliki forest virus in different rodent tumor models

Abstract

Semliki Forest virus (SFV) is one of the latest candidates for a virotherapeutic agent against cancer, and recent studies have demonstrated its efficacy in tumor models. In the present study, we examined the antitumor efficacy of an avirulent SFV strain A7(74) and its derivative, a replication-competent SFV vector VA7-EGFP, in a partially immunodeficient mouse tumor model (subcutaneous A549 human lung adenocarcinoma in NMRI nu/nu mouse) and in an immunocompetent rat tumor model (intracranial BT4C glioma in BDIX rat). When subcutaneous mouse tumors were injected 3 times with VA7-EGFP, intratumorally treated animals showed almost complete inhibition of tumor growth, while systemically treated mice displayed only delayed tumor growth (intravenous injection) or no response at all (intraperitoneal injection). This was at least partially due to a strong type I interferon (IFN) response in the tumors. The animals did not display any signs of abnormal behavior or encephalitis, even though SFV-positive foci were detected in the brain after the initial blood viremia. Intracranial rat tumors were injected directly with SFV A7(74) virus and monitored with magnetic resonance imaging. Tumor growth was significantly reduced (p < 0.05) with one virus injection, but the tumor size continued to increase after a lag period and none of the treated animals survived. Three virus injections or T-cell suppression with dexamethasone did not significantly improve treatment efficacy. It appeared that the local virotherapy induced extensive production of neutralizing anti-SFV antibodies that most likely contributed to the insufficient treatment efficacy. In conclusion, we show here that SFV A7(74) is a potential oncolytic agent for cancer virotherapy, but major immunological hurdles may need to be overcome before the virus can be clinically tested. © 2007 Wiley-Liss, Inc.

Experimental forms of cancer therapy have been introduced to accompany traditional treatment modalities to improve efficacy, or to offer alternative ways to eradicate tumors that are difficult to treat by surgery or radio- or chemotherapy. Gene therapy with many viral vectors has been tested extensively during the last 2 decades with some degree of success.1 Oncolytic virotherapy is most commonly utilizing modified adenovirus- or herpes virus vectors, but also other types of viruses, e.g. the RNA viruses reovirus and Newcastle disease virus are considered oncotropic by nature. The oncolytic potency of these viruses has already been demonstrated in preclinical studies2, 3, 4 and even in clinical (phase I/II) trials.5 Thus, it appears that a wide variety of different viruses have features that make them good candidates for oncolytic virotherapy.

Alphaviruses, such as Semliki Forest virus (SFV) and Sindbis virus, are positive-strand RNA viruses that have been engineered for efficient production of heterologous proteins and utilized as gene-transfer agents in various applications.6, 7, 8, 9 Alphaviral vectors have recently been tested also in cancer targeting, and some of the results are extremely promising. For example, SFV10, 11 and Sindbis virus12, 13, 14 vectors have been shown to be highly effective against both xenografts and orthotopic tumors in mice. These results have encouraged us to study the oncolytic potential of a replicative SFV virus strain or a replication-competent SFV vector in relevant tumor models.

We have earlier reported the construction and utility of a replication-competent SFV vector, VA7-EGFP, based on the avirulent strain A7(74) of SFV. This vector, which is identical to the parental virus except that it contains another subgenomic unit expressing EGFP, was shown to transduce efficiently rodent and human malignant cells, replicate and spread in these cells, and rapidly induce prominent cell death.8 Furthermore, the antitumor capacity of VA7-EGFP was recently investigated in an immunodeficient SCID mouse model,15 showing dramatic regression of subcutaneous human melanomas, regardless of whether the vector was administered systemically or locally. On the other hand, SCID mice devoid of any type of adaptive immune response gradually developed neuropathological symptoms that eventually were lethal to the treated animals.

EGFP, enhanced green fluorescent protein; IFN, interferon; MOI, multiplicity of infection; pfu, plaque forming unit; p.i., postinfection; SFV, Semliki Forest virus.

In this study, we have assessed the oncolytic potential of both the VA7-EGFP vector as well as the parental strain A7(74) in more immunocompetent rodent models: (i) human lung adenocarcinoma grafted subcutaneously to nude mice and (ii) syngenic BT4C rat glioma cells implanted intracranially in BDIX rat brain. Our results indicate that the virus and the derived vector are safe oncolytic agents in partially or fully immunocompetent rodents. However, their efficacy appeared to be limited, as neither with systemic virus administration (nude mouse model) nor by repeated administrations or immunosuppression (rat glioma model) could complete tumor regression be achieved. Despite replication competence, virotherapy with vectors based on avirulent SFV may require optimization to overcome the restrictions posed by the immune system. However, the failure of viral oncolysis in this study may also reflect the inherent differences in the susceptibility of the cancer cell lines used, and the results presented here may thus not describe the full oncolytic potential of the virus. Further studies in other cancer models are highly warranted, and the role of the immune system in virus clearance must be elucidated in detail.

Material and methods

Viruses

The replication-competent, attenuated SFV A7(74) and the derived vector VA7-EGFP used in this study have been described earlier.8, 16 For virus production, 4 × 10⁶ baby hamster kidney (BHK)-21 cells were seeded onto Ø 10-cm cell culture plates. After 24 hr, the cells were transduced with multiplicity of infection (MOI) 1 in total volume of 7 ml. The virus-containing medium was collected after 30 hr, purified with 0.2-μm pore size filter and stored at −70°C. Titration of SFV A7(74) and VA7-EGFP was performed as described by Santagati et al.17 VA7-EGFP titer was also confirmed by flow cytometry (FACSCalibur, Becton Dickinson, San Jose, CA), measuring the number of GFP-positive BHK-21 cells 8-hr posttransduction.

Cell lines and in vitro experiments

BHK-21 cells (ATCC CCL-10) and BT4C rat glioma cells18 were grown in high-glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine and 50 μg/l gentamicin. Human lung adenocarcinoma cell line A549 (ATCC CCL-185), human squamous cell lung cancer SW-900 (ATCC HTB-59) and human large cell lung cancer cell line NCI-H661 (ATCC HTB-183) were cultured in media recommended by ATCC. All cell lines were adherent and grown at +37°C in the presence of 5% CO₂.

The transduction efficacy and oncolytic capacity of VA7-EGFP vector (MOI 0.1, 1 and 10) in A549, SW900 and NCI-H661 cells were tested on 12-well plates using 5 × 10⁴ cells/well. For testing of transduction efficiency, the cells were infected with VA7-EGFP and analyzed 8-, 24-, 48- and 72-hr postinfection (p.i.), using flow cytometry and fluorescence microscopy.

Rodent tumor models and in vivo experiments

The efficacy of virotherapy was determined in vivo in immunodeficient NMRI nu/nu mice using a subcutaneous A549 tumor model.19 In this experiment, the efficacy of virotherapy was compared when replication-competent VA7-EGFP (5.8 × 10⁸ pfu/ml) virus was injected, either systemically (i.v. and i.p.) 100 μl/animal or locally (i.t.) 30 μl/tumor. The first virus administrations were performed 10 days after tumor cell implantation, and thereafter at 7-day intervals (1 injection per tumor each time, 3 injections total). Each group had 4 animals (total 8 tumors/group). Also, the distribution of virus upon treatment was analyzed in the A549 subcutaneous tumor model. The tissue samples (serum, brain, tumor tissue) were collected 16 hr, 2 days, 4 days and 1 week after viral injections (i.p., i.v. and i.t.).

Inbred BDIX female rats (200–250 g) (Harlan, Netherlands) were anesthetized intraperitoneally with fentanyl-fluanisone (0.15 ml/100 g) (Janssen-Cilag, Hypnorm®, Buckinghamshire, UK) and midazolame (Roche, Dormicum®, Espoo, Finland) and placed in a stereotaxic apparatus (Kopf Instruments). BT4C rat glioma cells (10,000 cells in 5 μl of OptiMEM medium) were implanted intracranially in the right corpus callosum at the depth of 2.5 mm at the following coordinates; 1 mm caudal to bregma and 2 mm right to sutura sagittalis. Three weeks after cell inoculation, the avirulent SFV strain A7(74) (1 × 10⁹ pfu/ml) was injected, either systemically (i.v. and i.p.) 100 μl/animal or locally (i.t.) 20 μl/tumor. When the tumors reached the size meeting the euthanasia criteria defined by the Kuopio University animal ethics and welfare committee, the rats were anesthetized and perfused with 4% paraformaldehyde/PBS by transcardiac route for 10 min. Tissue samples were embedded in OCT compound (Miles Scientific, Elkhart, IN) and immunostained as described later.

MRI methods

MR imaging was carried out using Magnex 4.7 T magnet (Magnex Scientific, Abington, UK), equipped with actively shielded gradients (Magnex Scientific) interfaced to Varian ^UNITYINOVA console (Varian, Palo Alto, CA). Half volume coil with loop diameter 22 mm (High Field Imaging, Minneapolis, MN) was used as a receiver and a transmitter. Rats were fixed in a custom-built holder using a mouth bar and ear pins. T2-weighted multislice spin echo images were acquired from tumor bearing animals with parameters: time-to-echo 70 msec, time-to-repetition 2 sec, field of view 35 × 35 mm², number of points in the free induction decay was 256 complex points, number of phase encoding steps 128, number of acquisitions 2 and number of slices 11–17, depending on tumor size with slice thickness 1 mm without gap between slices.

Histological analyses

Blood was collected from CO₂-euthanized mice and the animals were perfused with physiological saline and 4% paraformaldehyde. Brains and tumors were collected, and after 24-hr incubation in 4% paraformaldehyde, the tissues were embedded in paraffin. Parallel sections (4 μm) were stained with hematoxylin and eosin, and virus was detected with a rabbit polyclonal SFV antibody (1:2,500) and a peroxidase anti-rabbit conjugate. To determine the virus titer in the serum (stored at −70°C), a plaque assay was performed on BHK-21 cells as described.17

Western blotting

The type I interferon (IFN) response induced by VA7-EGFP vector in nude mice subcutaneous tumors was analyzed by MxA expression. Tumors were collected at each time point (16 hr, 48 hr, 96 hr and 7 days) and stored at −70°C for protein extraction. Tumors were homogenized mechanically in standard buffer (25mM Tris-0.1 mM EDTA, pH 7.4) containing protease inhibitors aprotinin (Sigma) 5 μg/ml, leupeptin (Sigma) 5 μg/ml and 50 mM NaF (Baker B.V., Mallinckrodt, Germany), centrifuged at 10,500×g for 20 min, after which the supernatant was collected and stored on ice. Protein concentration was measured with the Bradford method at A₅₉₅. Western blotting against MxA protein was performed as described earlier by Rautsi et al.,20 using rabbit anti-MxA antibody (1:2,000 dilution, kindly provided by Professor Ilkka Julkunen, National Public Health Institute, Helsinki, Finland) and HRP-conjugated anti-rabbit IgG antibody (1:200,000 dilution, Amersham Biosciences, Little Chalfont, UK).

Detection of neutralizing antibodies

For detection of neutralizing SFV-antibodies, serum samples were collected from BDIX rats at the end point (24 days) of the treatment study and stored at −70°C. BT4C rat glioma cells were split on 48-well plates, 30,000 cells per well. Serum dilutions (1/200, 1/600 and 1/2,400) from 3 different animals per treatment group (3 × i.t. virus and 1 × i.t. virus plus dexamethasone) were mixed with VA7-EGFP vector (MOI 0.1), incubated for 1 hr at +37°C, and then added onto the cells. Each sample was analyzed in triplicate. Cells were harvested at 12- and 24-hr time points for determination of EGFP-positive cells by flow cytometry. In addition, the following controls were analyzed: (i) VA7-EGFP virus alone (negative control), (ii) VA7-EGFP virus + serum from nontreated rat (serum control), (iii) VA7-EGFP virus + anti-SFV rabbit serum (positive control).

Statistical analyses

The tumor sizes in the in vivo experiments were indicated as mean tumor size (weight or volume) ± SEM. One-way or 2-way analysis of variance with Bonferroni's post hoc test for multiple comparisons was used for statistical analysis. When comparing less than 3 groups, 2-tailed t test was used (software: GraphPad Prism 3.0, GraphPad software, San Diego, CA).

Results

Characterization of SFV A7(74) virus and VA7-EGFP vector in vitro

In this study, the avirulent SFV strain A7(74)16 and the derived EGFP expressing vector VA7-EGFP8 were used. The VA7-EGFP vector was utilized in cell culture experiments to verify the spreading and oncolysis in human and rat cancer cells, while both the vector and the parental A7(74) virus were used in antitumor animal studies.

It has been earlier shown that both A7(74) virus and VA7-EGFP vector infect and replicate in BT4C rat glioma cells in vitro.8 Infection with VA7-EGFP at MOI 1 resulted in 100% of SFV-positive cells at 24 hr, although the replication rate turned out to be lower than in human tumor cell lines. Importantly, these studies revealed that replication efficiency of the virus vector was comparable to that of the parental virus. In order to verify these results, A549 human lung adenocarcinoma cells, SW900 lung squamous cancer and NCI-H661 large cell lung cancer cells were infected with VA7-EGFP vector (MOIs 0.1, 1 and 10) and observed by microscopy 8, 24, 48 and 72 hr later. It appeared that the vector replicated efficiently in all lung cancer cell lines. Virtually, all live cells were GFP-positive at 48 hr p.i. with lower MOIs (0.1 and 1). Progressing cytopathic effect could be observed at 48 hr (p.i.) with MOI 0.1 or 1, whereas at MOI 10 the virus caused almost complete cell death (results not shown).

Oncolytic efficacy of VA7-EGFP in nude mouse subcutaneous lung cancer model

To evaluate the oncolytic potential of the replicative VA7-EGFP vector in a partially immunocompromized tumor model, we used human lung adenocarcinoma cell line A549 to form subcutaneous tumors in NMRI nu/nu mice. This mouse strain is devoid of thymus and mounts, thus, a very weak T-cell response, but develops almost normal antibody response and even slightly elevated levels of natural killer (NK) cells compared to fully immunocompetent mice. We have earlier characterized this model and shown that these tumors can be successfully treated by virus vector-mediated HSV-TK/ganciclovir gene therapy.19

In the first phase of the experiment, VA7-EGFP vector (1.7 × 10⁷ pfu in 30 μl) was injected into each tumor once a week for 3 weeks. The tumors were measured twice a week and weighed at the end of the experiment (day 21). The results are presented in Figure 1. Intratumoral injection of VA7-EGFP turned out to be very efficient in this model, almost completely preventing the tumor growth. The tumor size in the treatment group was significantly smaller (p < 0.001) than in the tumors of the control group (Fig. 1a). This was confirmed by measuring tumor weights that revealed statistically significant differences (p < 0.001) at the end of the experiment (Fig. 1b).

Figure 1. Treatment of subcutaneous mouse tumors with VA7-EGFP vector. Subcutaneous A549 tumors were implanted in NMRI nu/nu mice (2 tumors per animal) and 10 days postimplantation VA7-EGFP was injected intratumorally (i.t.) using 30 μl (1.7 × 10⁷ pfu) of the vector, or intravenously (i.v.) or intraperitoneally (i.p.) using 100 μl (5.8 × 10⁷ pfu) of the vector. Injection times are indicated by arrows below the x-axis. Control tumors received equal volume of saline. (a) Tumor growth curves with local (i.t.) injections, (b) tumor weights at the end of local treatment experiment, (c) tumor growth curves with systemic (i.p. and i.v.) injections, (d) tumor weights at the end of systemic treatment. Each data point represents a mean of 8 tumors, deviation markers (± SEM) are shown. Statistical analysis was carried out using 2-tailed t test, 1-way or 2-way analysis of variance with Bonferroni's post hoc test for multiple comparisons (***p < 0.001).

In the second phase, we injected the virus systemically using intraperitoneal (i.p.) and intravenous (i.v.) routes (100 μl, 5.8 × 10⁷ pfu). As shown in Figure 1c, the tumor growth was only slightly delayed with i.v. injections compared to controls, and the tumors sizes continued to increase, reaching the size of control tumors by the end of week 3 after treatment initiation. Intraperitoneal injections were even less efficient, and no significant difference in tumor growth between the treatment group and the control group was observed during the course of the experiment. These findings were confirmed by analysis of tumor weights at the end of the study (Fig. 1d).

Presence of SFV A7(74) virus in nude mouse tumors and brain

Nude mice do not develop clinical symptoms upon infection with avirulent SFV despite life-long persistence of the virus in the CNS.21, 22 As expected, no neurological deficits or altered behavior could be observed during the experiments in the present study, and the mice remained alert, ate and drank normally and did not lose weight.

When the infected mice (1 injection per animal, i.t.: 1.7 × 10⁷ pfu in 30 μl; i.p and i.v.: 5.8 × 10⁷ pfu in 100 μl) were analyzed for the presence of SFV, a typical transient viremia was observed (data not shown). Serum samples of all treated mice contained ≫ 3,000 pfu/ml at 16 hr p.i., regardless of the route of administration. At later time points (48 hr–7 days), no virus was detected in the blood. Histological analysis revealed SFV-positive foci in the subcutaneous tumors 16 hr p.i., irrespective of the route of administration (Table I). It should be noted, however, that of the 4 i.v.-treated animals only 1 had virus in the tumor. At later time points (days 2 and 4), virus could be detected in i.t.-treated tumors, whereas the i.v.- and i.p.-treated tumors were SFV-negative. The virus-positive areas were large in i.t.-treated tumors, whereas in tumors of systemically administered animals virus, spreading was markedly limited (Fig. 2). In some mice, virus could be detected by immunohistochemistry in the brain at day 4 p.i., and 1 week after injection all animals showed brain infection. The virus-positive cell clusters were small and no signs of cell damage could be observed (Fig. 2).

Figure 2. Histological analysis of mouse tumors and brain after administration of VA7-EGFP vector. Treated animals were formalin-perfused, and their tumors and brains were processed with standard histological techniques and stained with anti-SFV antibody. Typical SFV-positive foci from tissues of mice treated with 1 intratumoral (i.t.), intraperitoneal (i.p.) or intravenous (i.v.) injection are shown.

Table I. Presence of Virus in Nude Mouse Tissues After SFV A7(74) Injection

Tissue (route of administration)	Time postinjection
	16 hr	2 d	4 d	7 d
Tumor-bearing mice were injected with SFV A7(74) using intratumoral (i.t.), intraperitoneal (i.p.) or intravenous (i.v.) route, followed by perfusion and recovery of tumors and brain tissues at indicated time points. Tissue slices were stained with anti-SFV polyclonal antibody and investigated under microscope: +, positive staining in one animal; ++, positive staining in two animals; +++, positive staining in three animals; (+) weak staining; −, all samples negative.
Tumor (i.t.)	+++	+++	++	−
Tumor (i.p.)	+++	−	−	(+)
Tumor (i.v.)	+	−	(+)	−
Brain (i.t.)	−	−	+	+++
Brain (i.p.)	−	−	++	+++
Brain (i.v.)	−	−	+(+)	+++

To detect whether type I IFN response was associated with the lack of complete tumor eradication with VA7-EGFP therapy, we subjected the tumor tissues to Western blot analysis and used MxA protein expression as a marker for type I IFN response. The VA7-EGFP injections triggered a prominent MxA expression with both administration routes (Fig. 3). The difference between locally (i.t.) treated and systemically treated animals was the kinetics of the response. The MxA expression became detectable in the locally treated tumors already at 16 hr p.i., whereas in the systemically treated tumors, the response did not emerge before the 48-hr time point.

Figure 3. Type I IFN response in mice treated with SFV VA7-EGFP vector. The vector was administered intravenously (i.v., 5.8 × 10⁷ pfu in 100 μl) or intratumorally (i.t., 1.7 × 10⁷ pfu in 30 μl) 10 days postimplantation into nude mice bearing subcutaneous A549 tumors. The tumors were collected 16 hr, 48 hr, 96 hr and 7 days after virus administration. MxA protein accumulation was used as an indicator of the type I IFN response and studied by Western blotting. MW, molecular weight marker; CO, untreated tumor; +, positive control; arrow indicates the location of MxA protein (76 kDa).

Oncolytic capacity of SFV A7(74) in immunocompetent intracranial rat glioma model

To test the efficacy of SFV A7(74) virotherapy in fully immunocompetent animals, a model resembling more actual human cancers, we used the BDIX/BT4C syngenic rat glioma model.23 In our preliminary experiments, we treated 3 intracranial tumor-bearing rats, each with 1 × 10⁸ pfu of SFV A7(74) using either i.p. or i.v. injection route and followed the tumor volumes regularly by MRI analysis, until the predefined endpoints were reached and the animals were killed. Results showed that the tumors remained completely unresponsive, and therefore, the subsequent studies were carried out using intratumoral injections of the virus (data not shown).

Upon a single i.t. administration of 2 × 10⁷ pfu of SFV A7(74), the tumors showed significant (p < 0.05) growth retardation that lasted up to 10 days compared to untreated controls (Fig. 4a). Correlating with the decline in tumor volume, the MRI scans revealed centrally located areas of lower contrast in the treated tumors, indicating necrosis induced by virus replication. After this time point, however, the tumor sizes in the treatment group started to increase, and although the treated tumors never reached the size of the control tumors, their regrowth was faster than the steady increase in volume observed for the control tumors (Fig. 4b). As a consequence of this growth acceleration, the survival time of the treated animals was not significantly increased (Fig. 4c). At 17 days p.i., both the treated and untreated tumors had reached the size at which the euthanasia criteria were met and the animals were killed.

Figure 4. Direct administration of SFV A7(74) virus into intracranial rat gliomas. BT4C gliomas were induced to BDIX rats by intracranial injection of 10⁴ cells. The animals were monitored by MRI until their tumor sizes met the euthanasia criteria. One injection (20 μl, 2 × 10⁷ pfu) of SFV A7(74) was carried out with the aid of microinjection unit of a stereotaxic frame (control animals received equal volume of saline), followed by MRI measurements of the tumor sizes (every third day until day 10 and weekly after that). (a) MRI images of 2 animals representing the control group (upper row) and SFV A7(74) treated group (lower row). The animals were followed until day 31 after virus/saline injection. (b) Tumor growth curves. Each point represents a mean of 5 tumors, deviation markers (± SEM) are shown. (c) Effect of intratumoral SFV A7(74) injection on survival of the animals. All rats were observed daily and killed after the euthanasia criteria were met (see Material and Methods).

Given the possibility of immune-mediated clearance of virus, in order to enhance the efficacy of SFV A7(74) treatment, we repeated the experiment with 3 intratumoral injections (1 injection at 3 day intervals). Furthermore, to minimize interference from the cell-mediated immunity, a group of rats also received the T-cell suppressor dexamethasone (1 mg/animal/day for 7 days) prior to a single injection of virus. As shown in Figure 5, neither the multiple virus injections nor the dexamethasone treatment enhanced the oncolytic efficacy of the virus compared to treatment with a single injection (p > 0.05, comparison of 1 × injection to 3 × injections or dexamethasone treatment). All the treatment regimens yielded similar transient stop of tumor progression which, however, was followed by accelerated increase in tumor mass leading to eventual death of the animals. Thus, these data suggest a minor role of T cells in the virus clearance.

Figure 5. Effect of multiple virus injections or dexamethasone treatment on the efficacy of rat glioma virotherapy with SFV A7(74). BT4C gliomas were induced to BDIX rats by intracranial injection of 10⁴ cells. The animals were monitored by MRI until the mean tumor size met the euthanasia criteria. One (i.t. × 1) or 3 (i.t. × 3) injections of SFV A7(74) was carried out with the aid of microinjection unit of a stereotaxic frame. Control animals received equal volume of saline. One group (i.t. + dex) received 1 mg/animal of dexamethasone once a day for 7 days before a single virus injection. MRI measurement of the tumor sizes was accomplished every third day until day 10 and weekly after that. Each point represents a mean of 5 tumors, standard error of the mean (SEM) markers are shown.

To investigate whether the treated animals developed neutralizing antibodies against SFV, the rat serum samples collected upon killing the animals were analyzed with in vitro neutralization assay. When the vector (VA7-EGFP) was incubated for 1 hr with cell culture medium or control rat serum (from animals not exposed to SFV) and added onto BT4C cells, no difference in transduction efficiency, i.e. the proportion of EGFP-positive cells, could be detected (Fig. 6). However, when the vector was incubated with either of the serums (diluted 1/200) from the treated rats (3 × i.t. injection or 1 × i.t. injection combined with dexamethasone treatment), no EGFP expression was seen. Serum from uninfected rat did not neutralize virus at any concentration. Interestingly, with 1/600 dilution of the 3 × i.t.-treated animal serum, the inhibition of the VA7-EGFP replication appeared to be even more efficient than with the positive control serum (rabbit anti-SFV). The dilution 1/2,400 did not show any neutralizing effect with serum from 1 × i.t. + dex-treated animals, whereas the 3 × i.t. serum still induced 25% inhibition of the vector infectivity (p < 0.01). In conclusion, the appearance of high-titer neutralizing anti-SFV antibodies is most probably one of the contributing factors to the failure of the SFV virotherapy in this immunocompetent rat glioma model.

Figure 6. Detection of neutralizing anti-SFV antibodies in the rats subjected to virotherapy with SFV A7(74). The sera of the rats treated with the virus was collected at the end of the experiment and mixed with VA7-EGFP vector to demonstrate the neutralizing activity in the serum. BT4C cells were transduced with VA7-EGFP (MOI 0.1) vector that was incubated for 1 h at +37°C with different dilutions (1/200, 1/600 and 1/2,400) of the rat sera. The anti-SFV antibody-mediated neutralization can be seen as reduction of EGFP-positive cells 24-hr posttransduction. Positive control for neutralizing antibodies was anti-SFV serum (from rabbit), negative control serum was 1/200 diluted serum from untreated rats, and transduction control was VA7-EGFP alone. Statistically significant differences (***p < 0.001, **p < 0.01) were observed when comparing controls to the treated rat serum samples (3 × i.t. or 1 × i.t. + dexamethasone). Each bar represents mean from 3 individual rat serum samples ± SEM. Statistical analysis was carried out using 1-way analysis of variance with Bonferroni's post hoc test for multiple comparisons.

Discussion

The purpose of this study was to verify whether the attenuated SFV A7(74) virus and the derived vector VA7-EGFP would be potent and safe oncolytic agents in partially or fully immunocompetent rodent tumor models. Our previous studies have shown that this virus efficiently replicates in a number of different human tumor cell lines in vitro and in tumors derived from them.8, 15 The latter study15 revealed that human melanoma xenografts in SCID mice were efficiently eradicated by the VA7-EGFP vector, although the treatment in these severely immunodeficient animals lead to undesired neurological side effects. We expected no side effects in the present study as SFV A7(74) is avirulent in both normal immunocompetent animals as well as in nude mice.21, 22 On the other hand, it was not clear whether the immune system would allow the virus enough time to successfully find and destroy the tumors.

When NMRI nu/nu mice were treated with the VA7-EGFP vector, no neurological symptoms or other signs of neuropathology were observed at any time point after the vector administration. The vector gave rise to a transient and asymptomatic viremia regardless of the route of administration, and virus could be detected in the brains of treated mice up to 7 days p.i. We did not analyze the tissues beyond 1 week p.i., but it is likely that the virus is cleared and becomes undetectable by immunohistochemistry at later time points. Namely, other groups have reported a transient decline in CNS virus titers in infected nude mice,22, 24 although viral RNA could still be detected by in situ hybridization. The decline in virus antigen observed in the present study correlates with peaking of virus-specific IgM and the leakiness of the blood–brain-barrier that occurs during infection with avirulent SFV.21, 22, 24, 25, 26 We therefore believe that virus-specific IgM may have inhibited virus replication below the detection limit due to temporary access of serum components to the CNS. Although not analyzed in the present study, the A7(74) CNS titers have been shown to be restricted also in nude and SCID mice (several magnitudes lower than in WT SFV), indicating that the restriction of A7(74) replication is not mediated by cells of the adaptive immune system.22, 24

The efficacy of the virotherapy in the nude mouse model is demonstrated in Figure 1. However, as opposed to the SCID mouse tumor model data in our previous study,15 only the intratumoral administration route was effective. In i.t.-treated animals, the tumors grew very slowly, and their mean size at the end of the experiment was significantly (p < 0.001) smaller than the mean tumor size in the controls, whereas the i.v.-treated tumors only showed delayed growth, and the tumors in i.p.-treated animals grew identically to control tumors. As shown in Table I, with systemic administration routes, the vector is able to reach subcutaneous tumors and replicate therein to some extent. However, the SFV-positive foci in the tumors were smaller than in the i.t.-treated tumors (Fig. 2), suggesting on one hand that insufficient amounts of the vector homed to tumors, and on the other hand, that virus replication was rapidly overpowered by the functional components of the nude mouse immune system.

It is likely that clearance of the virus was mediated by the innate immune system, as all the i.v.- and i.p.-treated tumors were devoid of virus already at 2 days p.i. With respect to this explanation, the inefficient tumor eradication in the nude mouse A549 xenografts may be due to inherent features of the cell line, particularly in the case of systemic vector administration. The human A549 lung cancer cell line has been shown to be less susceptible to SFV transduction and cytotoxicity than many other tumor cell lines.15 A549 cells are known to elicit a vigorous type I IFN response upon virus infection,27 and SFV replication is efficiently inhibited by IFNs.28 Indeed, when the type I IFN response in the tumor tissues was examined with the aid of MxA Western blotting (Fig. 3), it turned out that these cells displayed a substantial induction of IFNs. Interestingly, we observed a vector administration route-dependent difference in the induction of the response. The MxA induction appeared and disappeared earlier in the tumors of i.t.-treated animals compared to animals that received the vector systemically. This difference in the MxA expression patterns makes sense, because the virus needs time to travel through the whole animal and home to the tumor when injected at a distant site. Although the response in locally treated tumors was fast and strong, it was still insufficient to prevent virus replication and subsequent tumor cell destruction. Alphaviruses, including SFV, are known to resist IFN response, as they can slow down protein translation via formation of specific capsid RNA haipin structures,29 which allows viral replication in the presence of active PKR. On the other hand, despite the slower onset of the IFN response in the i.v.-treated animals, the antitumor effect of the vector administered systemically was more modest. This suggests that there is a certain threshold amount of virus in the tumor that is required for a progressive replication and oncolysis. The critical amount is likely to be reached when the local route is used, whereas the systemic route may not provide enough viruses to home to tumors. Thus, it is likely that the type I IFN response, perhaps in concert with other components of the innate immune response, is one of the major hurdles against successful virotherapy.30

In nude mice, also virus-specific IgM plays an important role in controlling the infection, and although antibodies alone are unable to clear the virus from the CNS, they may remove it from the subcutaneous tumors.21, 22 It has been shown that the vasculature in many solid tumors is leaky for serum protein components, including immunoglobulins.31, 32 Moreover, despite the lack of functional T cells, nude mice retain NK cell activity, which may contribute to a premature abortion of the infection. NK cells clearly play a role in destroying alphavirus-transduced tumor cells, as has been demonstrated by Tseng et al.13 In contrast to our results, however, the presence of NK cells in their models enhanced the oncolytic capacity of a Sindbis virus vector. This discrepancy likely relates both to differences between the viruses and differences between the cancer cell lines used.

Results from the therapy trials in the rats with fully functional immune system pose even more complicated situation. Our initial tumor eradication experiments with SFV A7(74) using different administration routes, showed no growth retardation by the i.v. or i.p. treatments. The complete lack of response in systemically treated animals in the rat pilot study and the prominent role of the type I IFN response seen in the mouse tumor model suggested us to perform the experiments in immunocompetent rats with local virus injections only. It turned out that, after a single intratumoral injection of virus (Fig. 4), the tumor growth was halted for the first 10 days and the tumors remained significantly smaller compared to the control tumors. However, after this initial lag period, the tumors started to grow with a rate faster than that of the control group, and about 6 weeks after infection, the animals started to display signs of discomfort due to tumor burden and had to be killed. Thus, also i.t. treatment in this model was unable to inhibit tumor growth in the long run, and the overall survival rate in the treatment group did not significantly differ from that in the control group.

It is most likely that the virus had been cleared from the rats, as no virus was found in the blood, brains or the tumors at the end of the experiment analyzed by the plaque assay or immunohistochemistry. As also multiple virus administrations or treating the rats with the immunosuppressor dexamethasone prior to virus administration failed to increase treatment efficacy, vector clearance was probably mediated by T-cell independent mechanisms, i.e. antibodies and NK cells. Also, tumor regrowth initiated at 10 days p.i., precisely the time by which the immune-mediated clearance of the virus is known to be complete in adult immunocompetent hosts and the time at which virus-specific IgG synthesis starts to peak.21, 22 Since tumor regrowth occurred with the same kinetics in both the untreated and the T-cell suppressed rats, antibodies and the innate immune defense may have played significant roles in eliminating the virus. Indeed, our results implicated that neutralizing anti-SFV antibodies are the major contributors in the failure of SFV virotherapy. The data from the in vitro neutralization assay (Fig. 6) reveal that sera from the treated rats are loaded with antibodies that efficiently prevent the VA7-EGFP vector from transducing BT4C cells. The antibody titer appeared to correlate with the number of virus injections (3× induces more than 1×) and was insensitive to dexamethasone, as expected. Interestingly, serum from the rats injected 3 times with the virus contained even more neutralizing antibodies than our control anti-SFV serum raised in rabbits by immunizing with purified virus. It remains unclear, however, whether the potential innate resistance of the BT4C cell line to virus contributed to the failure of the virotherapy. It is likely that the innate immune system plays at least some role, as it has been recently shown by Fulci et al. that herpes simplex virus vector-mediated virotherapy is severely restricted by the components of innate immune response, and this inhibition can be overcome by immunosuppression with cyclophosphamide.33

In conclusion, we show here that virotherapy with attenuated replicative SFV A7(74) virus or the derived replication competent vector VA7-EGFP is safe in immunocompetent rodents, but the therapy regimens require optimization. The data in this report suggest that the immune system (both innate and humoral) poses a serious barrier to virotherapy with avirulent SFV. Such restrictions may, however, be overcome by using relevant forms of immunosuppression or by including therapeutic genes that increase the oncotoxicity of the virus.34 To establish whether the problems observed in the present study are universal or due to inherent differences in the tumor cell lines used, further studies with other cancer models are required.

Acknowledgements

Research and Science Foundation of Farmos, North Savo Cultural Foundation, North Savo Cancer Foundation and Kuopio University Foundation are gratefully acknowledged for supplying grants to A-M.M. Tor, Joe and Pentti Borg's Foundation is acknowledged for grants to AH. MV-K is a member of the Turku Graduate School of Biomedical Sciences.