¹Applied Tumor Virology Program F0100 and INSERM U375, Deutsches Krebsforschungszentrum, Heidelberg 69120, Germany

²Laboratory of Inflammation and Signal Transduction, IRF Mario Negri, Milan 20157, Italy

Correspondence to: Dr Nathalia A Giese, Applied Tumor Virology Program F0100 and INSERM U375, Deutsches Krebsforschungszentrum, Heidelberg 69120, Germany. E-mail: n.giese@dkfz-heidelberg.de

We have previously shown that the growth of human tumor xenografts in immunodeficient mice can be efficiently suppressed upon infection with the autonomous parvovirus H-1 or with cytokine-transducing derivatives thereof. To further evaluate the benefits of implementing parvoviruses in cancer gene therapy, we have created a new recombinant vector, MVMp/IP-10, transducing the immunoactive, antiangiogenic chemokine IP-10, and used this virus to treat syngeneic tumors grown in immunocompetent mice. Intratumoral/intraperitoneal administration of only 3´10⁷ replication units of MVMp/IP-10 per animal strongly inhibited the progression of established H5V cell-induced vascular tumors, a highly malignant mouse model for human cavernous hemangioma and Kaposi's sarcoma. Retardation of recurrent tumor growth and suppression of life-threatening metastatic dissemination to internal organs were accompanied by a striking delay in hemangioma-associated mortality. Parental MVMp did not have a significant effect under these conditions up to the dose of 10¹⁰ infectious units/animal, but had strong antihemangiosarcoma activity when used to infect H5V cells ex vivo prior to implantation. In all cases, virus therapy was very well tolerated. Virus-induced suppression of hemangiosarcoma was dependent on host T cells and associated with intratumoral persistence of IFN-expressing cytotoxic lymphocytes, and led to the reduced expression of hepatic plasminogen activator inhibitor-1 (PAI-1), a metastasis-linked marker. This proof of principle study demonstrates that MVMp/IP-10 can aid the treatment of vascular tumors and that autonomous parvovirus-based vectors can be considered potent tools for cancer gene therapy purposes. Cancer Gene Therapy (2002) 9, 432-442 DOI: 10.1038/sj/cgt/7700457

hemangiosarcoma; metastasis; parvovirus MVMp; IP-10; IFN; PAI-1

Virus-based anticancer therapies involve the use of viruses either as replicating oncolytic agents, or as recombinant vectors for gene transfer.¹ The autonomous parvoviruses MVMp and H-1 belong to a group of small (~5 kb) nonintegrating single-stranded DNA viruses. Their oncotropic and oncotoxic properties make them good candidates for both types of applications.² Malignant transformation of a number of rodent and human cells correlates with their increased capacity to amplify and/or express parvoviral DNA, and with hypersensitivity to parvovirus-induced killing.³ These features contribute to the intrinsic oncosuppressive capacity of these viruses, but they are often insufficient to promote full inhibition of oncogenesis.^4,5 Therefore, we are developing strategies to reinforce the antineoplastic activity of parvoviruses through incorporation of therapeutic transgenes.⁶ Although the MVMp- and H-1-based vectors produced so far are unable to generate progeny virions, they retain their genuine parvoviral promoters and replication origins, and encode the pivotal nonstructural protein, NS1.⁷ These features allow amplification of the viral genome, efficient expression of inserted transgenes, and limited oncotoxicity.^7,8 Parvovirus vectors are expected to be relatively safe, being poorly expressed in quiescent cells and most nontransformed cells.^3,8 Furthermore, the MVMp and H-1 viruses are weakly immunogenic and rarely pathogenic to adult animals: the infections they cause can be abortive or persistent, but they are usually clinically nonapparent.^9,10 We have recently demonstrated that wild-type H-1 virus can reduce the tumorigenicity of HeLa xenografts in immunodeficient mice, and that recombinant derivatives transducing various cytokines (IL-2, MCP-3) are more efficient antineoplastic agents than the parental virus.^11,12 Our current focus is on assessing the antitumor activity of autonomous parvovirus vectors in immunocompetent hosts, and on identifying target malignancies for parvovirus-based anticancer therapy.

In the present work, MVMp, known to target transformed cells of both human and murine origin, and recombinant MVMp derivatives were tested for their ability to protect mice against a very aggressive mouse hemangiosarcoma. Antiangiogenic therapy is increasingly recognized as a powerful strategy for impeding the growth of various types of tumors and for circumventing acquired resistance to anticancer drugs. Vascular tumors represent an extreme'case of neovascularization and comprise lesions ranging from Kaposi's sarcoma associated with human'herpesvirus-8 (HHV-8) to infantile hemangiomas of unknown etiology.^13,14 Kaposi's sarcoma is a major cause of mortality in AIDS patients, while hemangiomas, the most common tumors of childhood, are disturbingly disfiguring and may induce life-threatening complications. Because conventional treatments of vascular tumors such as irradiation and surgery are unsatisfactory, an intensive search for new therapeutics is in progress.¹⁴ Murine hemangioma models have recently been developed, allowing the evaluation of new antiangiogenic compounds.^15,16,17 One of the most potent anticancer cytokines, IL-12, has been shown to suppress progression of experimental hemangiomas.^18,19 IP-10 (interferon-inducible protein with molecular weight of 10 kDa, recently classified as CXCL10) has been identified as a major IL-12-induced downstream effector of antitumor reactions, and found to inhibit the growth of several experimental tumors with varying efficacy, depending on the model and method of delivery.^{20,21,22,23,24,25,26} Although the mechanism of IP-10-induced tumor suppression has not yet been fully elucidated, both antiangiogenic and immune system-mediated effects appear to be essential to the antitumor activity of this chemokine in vivo.^27,28 Interestingly, IP-10 can act as an inverse agonist of the G-protein-coupled receptor homologue encoded by Kaposi's sarcoma-associated HHV-8, and thereby inhibit the constitutive signaling occurring in transformed cells²⁹ These facts prompted us to attempt to combine, in a recombinant vector MVMp/IP-10, the immunomodulating/angiostatic properties of IP-10 with the oncotropic/oncostatic features of MVMp, and to test the protective effect of MVMp/IP-10 against malignant hemangiosarcoma (H5V) implanted into syngeneic immunocompetent mice. In the present preclinical study, wild-type MVMp was found to inhibit H5V tumor progression, provided that infection was carried out ex vivo to ensure that most cancer cells were hit. When administered in vivo, MVMp did not protect effectively against H5V, in contrast to its recombinant derivative MVMp/IP-10. The latter was able to slow down the growth of relapsing primary umors and to drastically suppress the formation of multiple metastases.

Materials and methods

Viruses

Minute virus of mice (prototype strain MVMp) was produced and titrated by plaque assay using A9 indicator cells as described previously.³⁰ Virus titers are expressed in plaque-forming units (PFU) per milliliter. Vector DNA clone pdBMV800⁷ was used to produce the recombinant derivatives MVMp/GFP and MVMp/IP-10. The molecular clone of MVMp/IP-10 was obtained by inserting the 752-bp EcoRV-SacI fragment of pBluescript mIP-10 (a generous gift from J Farber, NIH) between the SacI and Klenow polymerase-filled MluI restriction sites of pdBMV800. MVMp/GFP contains the 730-bp "humanized" GFP coding sequence, placed under the control of the MVMp P38 promoter.¹² Infectious particles were produced by cotransfecting 293T/17 cells with the corresponding recombinant molecular clone and a helper plasmid encoding the parvoviral structural proteins.⁷ The wild-type and recombinant MVMp viruses, purified by CsCl gradient centrifugation followed by G25 Sephadex gel filtration, were diluted in MEM (Life Technologies, Eggenstein, Germany) prior to use. Recombinant viruses were titrated by infected cell hybridization assays on A9 cells,⁷ using an NS1- or IP-10-specific DNA probe. Recombinant parvovirus titers routinely reached 3-5´10⁷ replication units (RU)/mL.

Cell cultures

The cell lines 293T/17 (transformed human embryonic kidney cells; American Type Culture Collection, ATCC), A9 (mouse fibroblasts, ATCC), and H5V [polyoma virus middle T antigen (PymT)-transformed mouse endothelial cells]¹⁷ were maintained in DMEM (Life Technologies) supplemented with 2 mM l-glutamine, 100 g/mL penicillin, 100 U/mL streptomycin, and 10% (293T and H5V cells) or 5% (A9 cells) fetal calf serum. Prior to infection, A9 or H5V cells were seeded at a density of either 2´10³ cells/well into 96-well plates (for survival tests and enzyme-linked immunosorbent assays, ELISAs), 4´10⁴ cells/well into six-well plates (for FACS analyses), or 3´10⁶ cells/plate into 10-cm Petri dishes (for animal experiments), and incubated for 24 hours. After infection with MVMp, MVMp/GFP, or MVMp/IP-10, the cells and culture supernatants were harvested at the indicated time points and analyzed for cell survival or transgene expression. The multiplicity of infection (MOI) is expressed in PFU (wild-type virus) or RU (recombinant vectors) per cell.

Determination of cell survival

The tetrazolium salt, 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT), test for identifying living cells (Sigma-Aldrich Chemie, Deisenhofen, Germany) was used to compare the growth kinetics of uninfected and infected cultures. The extent of cell death in treated cultures was assayed by measuring the release of lactate dehydrogenase (LDH) according to the CytoTox96Ò test (Promega Deutschland, Mannheim, Germany).

Measurement of IP-10 production

To determine levels of secreted murine IP-10, an ELISA was established using goat antirat Ig (Becton Dickinson, Heidelberg, Germany) to coat the plates, purified rat antimouse IP-10 monoclonal antibodies cocktail A102-5/A102-6 (Becton Dickinson) to capture IP-10, and biotinylated goat antimouse IP-10 polyclonal antibodies BAF466 (R&D Systems, Wiesbaden, Germany) to detect IP-10. Recombinant murine IP-10 protein (Becton Dickinson) was used as a standard.

FACS analysis

For detection of GFP expression, MVMp/GFP-infected cells were harvested by trypsinization and washed in PBS containing 2% FCS, prior to analysis with a FACScan flow cytometer (Becton Dickinson).

Reverse transcription polymerase chain reaction (RT-PCR) analysis

Total RNA was prepared from samples of isolated tissues (kept at -80°C) using TrizolÒ (Life Technologies) according to the manufacturer's instructions. DNAseI-treated RNA was converted to cDNA, amplified, and analyzed according to RT-PCR protocols detailed previously.^31,32 The primers used for HPRT, CD4, CD8, NK 1.1, IL-2, IFN, IL-12p40, VEGF, and bFGF cDNA amplification have been described elsewhere.^31,33,34 The primers used for other markers were: PAI-1 (511 bp): 5'-GCT GGT GAA TGC CCT CTA CTT-3' and 5'-CAC TGT GCC GCT CTC GTT TAC-3'; PAI-2 (320 bp): 5'-AGG CAC AAG CAG GAG ATA AAA-3' and 5'-GAC AGC ATT CAC CAG CAC CAT-3'; uPA (600 bp): 5'-CTA GGC CTG GGG AAA CAC AAT-3' and 5'-GTC TGA ACC AAA CGG AGC ATC-3'; tPA (679 bp): 5'-GTA CTG CTG CTT TGT GGA CTG G-3' and 5'-AGG AGG CCT GGG ATG TGG TGA G-3'; CXCR3 (290 bp): 5'-CCT GAG CAG CAC GGA CAC CTT C-3' and 5'-GCA GAC AGA GAC CCC ATA CAA C-3'; Granzyme B (253 bp): 5'-ACT CAA ACA CGC TCA AAG A-3' and 5'-ATC CAG GAT AAG AAA CTC G-3'; CD40L (282 bp): 5'-ATC TGT GCT TTT TGC TGT GT-3' and 5'-TTA CTG TTG GCT TCG CTT AC-3'; CTLA4 (438 bp): 5'-ATT CAC CAT CAC ACA ACA CT-3' and 5'-GGG GCA TTT TCA CAT AGA CC-3'; IP-10 (381 bp): 5'-TGA GCA GAG ATG TCT GAA TC-3' and 5'-TCG CAC CTC CAC ATA GCT TAC AG-3'; viral transcripts (511 bp, unspliced RNA; 414 bp, the major form of spliced NS1 RNA): 5'-ACG CTC ACC ATT CAC GAC ACC GAA A-3' and 5'-ATC ATA GGC CTC GTC GTG CTC TTT-3'.The metastasizing H5V cells were detected using PymT-specific primers (439 bp): 5'-TTC TGA GCA ACC CGA CCT AT-3' and 5'-CTT CTT AGG TGG CGT TGC AT-3') and probe: 5'-GCC ACT CCT ATC CCC CAA CCC-3'. PCR products were analyzed by 2% gel electrophoresis and visualized either by direct staining with ethidium bromide or SybrGreenII (Roche Diagnostics, Mannheim, Germany) or by Southern blot hybridization and ECL detection (Amersham Pharmacia Biotech Europe, Freiburg, Germany), depending on the level of expression. PCR signals were measured^31,32 and ranked as described in the legend to Figure 4, with hypoxanthine phosphoribosyl transferase (HPRT) expression serving as a reference for equalizing the cDNA input for PCR. Viral transcripts in H5V cultures and tissues of infected animals were also quantitated using LightCyclerÔ real-time QRT-PCR technology (Roche Diagnostics).³⁵

Tumor model and virus treatments

The mouse H5V hemangiosarcoma model has been described previously.¹⁷ A total of 5´10⁶ H5V endothelioma cells/animal were subcutaneously implanted into the right hind flanks of syngeneic female C57Bl/6 mice (CRWiga, Sulzfeld, Germany). Disease progression was monitored on the basis of the appearance of cavernous hemangiomas at the injection site. Tumor size was measured with a metric electronic caliper and calculated according to the formula 0.785´width´length. Sudden death occurred frequently in tumor-bearing mice, being due to metastases and internal hemorrhage. Therefore, the mice were regularly checked for life-threatening symptoms such as a weakness, a drop in body temperature, and shivering, and moribund animals were killed. At necropsy or autopsy, the internal organs were inspected visually for the presence of metastatic foci and processed for histopathological (hematoxylin-eosin staining of formaldehyde-fixed tissue sections) or RT-PCR analysis. Virus (MVMp, MVMp/IP-10) or mock (dilution medium) treatment of animals was initiated in two ways: either H5V cells were infected with virus and then injected into mice (ex vivo approach), or mice with established tumors were exposed to daily intratumoral, then intraperitoneal, injections (in vivo approach). The virus doses and treatment regimens are described in the main text. The animals were kept in laminar flow safety cabinets (BDK Luft- und Reinraumtechnik, Sonnenbühl, Germany). Procedures for animal handling, treatment, and care were in compliance with institutional and governmental guidelines.

Results

In vitro susceptibility of transformed endothelial H5V cells to infection with MVMp and its recombinant derivatives

Wild-type MVMp was first tested for its ability to infect and kill H5V cells in vitro. H5V cultures proved competent for MVMp uptake and viral gene expression, accumulating 2´10⁶ transcripts/g total RNA within 24 hours after virus inoculation (MOI=10 PFU/cell), as measured by QRT-PCR. Using the MTT assay to identify living cells, we found that MVMp infection led to a dose-dependent reduction of the rate of H5V culture growth (Fig 1A). To determine the role of cell death in this inhibition, the LDH content of the culture medium was measured upon infection. Cytoplasmic LDH is freed by dying cells and constitutes a reliable marker of cytolysis. As shown in Figure 1A, increased release of LDH was detected in infected H5V cultures, indicating that MVMp is toxic to these cells. Yet, this cytotoxicity was pronounced only at high MOIs (100 PFU/cell), possibly reflecting a low capacity of H5V cells for MVMp multiplication. In fact, the output/input virus ratio corresponded to 0.2 in MVMp-infected H5V cultures (as compared to 10⁵ in the highly sensitive A9 producer cell line; data not shown). Furthermore, wild-type MVMp inhibited the growth of H5V cultures to about the same extent as the recombinant vectors MVMp/GFP and MVMp/IP-10, which lack the capsid genes and are unable to yield progeny viruses (Fig 1B). Though defective, the recombinant parvoviruses used in this study retain the pivotal NS1 gene and are proficient in DNA replication and gene expression.^6,7,8,11 We conclude that the H5V cell populations showed little ability to support a productive MVMp infection, but that even at low MOIs, viral metabolism takes place in at least a fraction of infected cells, causing growth to be significantly delayed in treated cultures.

In agreement with this view, H5V cell cultures infected with the recombinant vectors MVMp/GFP or MVMp/IP-10 were able to support NS1 and transgene expression. Two days after infection with MVMp/GFP at MOI of 1 RU/cell, the proportion of GFP-expressing cells in H5V cultures reached 10±3%. Stained cells were detected up to 7 days after infection (data not shown). As illustrated in Figure 1C, MVMp/IP-10-treated H5V cultures produced substantial amounts of the chemokine, even when infected at a low multiplicity (1 RU/cell). Endogenous production of IP-10 protein was below the level of detection in mock-, MVMp- and MVMp/GFP-infected H5V cultures (<1 ng/mL, as determined by ELISA), although the corresponding mRNA was detected by RT-PCR (see below, Fig 4A). Interestingly, transduced IP-10 did not increase the intrinsic toxic effect of MVMp on H5V cells as growth inhibition was the same whether induced by MVMp/IP-10 or MVMp/GFP (Fig 1B). This finding is in keeping with our failure to detect expression of the IP-10 receptor CXCR3 in H5V cells by RT-PCR (see below, Fig 4A). These various results prompted us to use H5V endothelioma as a model for IP-10 transduction and assessment of the chemokine's ability to potentiate MVMp-based treatment of vascular tumors.

Reduced tumorigenicity of H5V cells infected ex vivo with MVMp

We first examined whether MVMp could reduce the tumorigenic potential of H5V cells in vivo. H5V hemangiosarcoma is an immunogenic MHC class I⁺ (H-2^b) tumor. Progression of the disease in terms of remission, relapse, and metastasis depends strongly on the number of engrafted cells.¹⁷ Implants of up to 2´10⁵ cells appear to be fully rejected, but as the number of inoculated cells increases, so does the fatal recurrence of tumors among treated mice. In the present study, 5´10⁶ cells were inoculated per mouse, resulting in highly malignant hemangiosarcomas in 100% of the animals. As shown in Figure 2, A-C, H5V-injected mice did not succumb to the primary tumors, which developed at the injection site and subsequently regressed. Instead, the animals died later from local relapse and development of metastases at multiple distant sites. Wild-type MVMp was tested for the ability to interfere with this sequence of events when administered ex vivo into H5V cells (MOI=50 PFU/cell) prior to their subcutaneous injection into syngeneic C57Bl/6 mice. Figure 2B shows the effects of this treatment on tumor formation at the implantation site. The treatment did not prevent the appearance of primary hemangiomas: all animals implanted with virus-infected cells developed tumors, albeit of smaller size than those derived from mock-treated cells. These primary tumors regressed in both treated and mock-treated animals. In contrast, a striking difference was observed between the two groups at the subsequent stage of tumor relapse. Viral treatment led to a significantly longer remission period: 43±7 days for MVMp-exposed mice, compared to 15±1 days for mock-treated mice (Wilcoxon rank-sum test, P<.001). In the control group, furthermore, relapsing tumors grew aggressively, whilst in the MVM-infected group, they stagnated at the site of implantation (Wilcoxon rank-sum test, P<.01). Recurrent tumor nodules from MVMp-infected H5V cells regressed occasionally, and three of five animals displayed no discernable hemangioma by the end of a 3-month observation period, though H5V cell-containing tumor remnants were detected at autopsy (data not shown). Even more remarkably, the histological analysis of internal organs after autopsy showed that all mice implanted with mock-treated H5V cells carried multiple deforming cavernous metastases in the liver, spleen, uterine tubes, peritoneal cavity, and ovaries, whereas only one mouse implanted with MVMp-infected cells carried a single metastatic nodule in the peritoneal cavity (Fig 2C). This represents very significant MVMp-dependent suppression of metastases (chi-square test, P<.01). In keeping with these observations, all control mice succumbed to hemangiomas within 3 months postimplantation, whereas only one of six mice in the MVMp group was found dead on day 74 (Fig 2A, Mantel-Haenszel test, P<.01). In summary, ex vivo MVMp infection at a moderate multiplicity drastically reduces the tumorigenicity of H5V cells, as shown by the more effective host control of tumor growth, suppression of metastatic expansion of primary hemangiomas, and prevention of mortality.

Treatment of established tumors with MVMp and MVMp/IP-10

The strong antineoplastic effect of MVMp in ex vivo experiments encouraged us to test the efficacy of the virus against established H5V tumors. A total MVMp dose of 10¹⁰ PFU/mouse was administered in vivo to tumor-bearing animals according to the following regimen: starting on day 6 postimplantation, a single intratumoral injection each day for 5 days, then a 2-day pause, followed by a single intraperitoneal injection each day for 5 days, another 2-day pause, then a final 5-day course of intraperitoneal injections. The reason for changing the route of administration in the course of the experiment was to take into account the regression of the primary tumors and the spread of H5V cells to distal organs. Under these conditions, the wild-type virus had no significant effect on tumor growth, formation of metastases, and disease outcome (data not shown). This result prompted us to generate the recombinant vector MVMp/IP-10 and determine whether the antineoplastic effect of the parvovirus was reinforced through the additional action of transduced IP-10. Treatment of established hemangiomas was initiated 5 days after implantation of H5V cells into C57Bl/6 mice, and consisted of a total dose of 3´10⁷ RU of MVMp/IP-10 per mouse, administered according to the above-mentioned regimen. For comparison, other groups of mice received an equivalent inoculum of wild-type MVMp or medium alone. Whereas the wild-type virus had little effect on hemangioma progression, MVMp/IP-10 provided significant antitumor protection. As illustrated in Figure 3A, mock-treated mice survived the tumor cell implant no longer than 15 weeks, while half of the MVMp/IP-10-infected mice responded to the therapy and stayed alive significantly longer (Mantel-Haenszel test, P<.05). Strikingly, one third of MVMp/IP-10-treated animals survived until the end of the experiment at 6 months postimplantation, i.e., at least twice as long as controls. As shown in Figure 3B, the growth of recurring hemangiomas in the long-term survivors was more restrained than in control mice (Wilcoxon rank-sum test, P<.01 between MVMp/IP-10 and mock groups; P>.1 between MVMp and mock groups), indicating a potent antitumor effect of the recombinant virus. Another remarkable feature of the MVMp/IP-10 treatment was the significant suppression of metastases (Fig 3C, chi-square test, P<.05 between mock- and MVMp/IP-10-infected groups by 3 months postimplantation; and Fig 3D). We conclude from these experiments that MVMp/IP-10 gene therapy has a clear-cut therapeutic effect on hemangioma-bearing mice, even when relatively low doses of vector are administered in vivo. Interestingly, the treatment of hemangiosarcomas with recombinant MVMp/IL-2 was unsuccessful (data not shown), pointing to the specificity of IP-10 interference with hemangioma progression. It should also be mentioned that macroscopic and histological examinations of mice revealed no deleterious side effects attributable to parvovirus-based treatments, even at the high dose of 10¹⁰ PFU/animal and after multiple injections (data not shown).

Potential mechanism of MVMp and MVMp/IP-10 action against hemangiosarcoma

MVMp vectors fail to alter the expression of malignancy-associated molecules in H5V cells: The parvovirus-mediated antihemangioma effects described above might involve multiple mechanisms ranging from direct action of the virus on H5V cells (cytotoxicity or changes in the expression profile of surface and secreted proteins) to indirect interference with disease development. To approach this question, we first investigated whether MVMp or MVMp/IP-10 could directly affect the capacity of H5V cells to produce known markers of vascular tumor progression.^36,37,38 To this end, RT-PCR analyses were carried out using H5V cell cultures to measure the expression of plasminogen activator proteolytic system (PAS) components (tissue-type and urokinase-type plasminogen activators, tPA and uPA, and their inhibitors, PAI-1 and PAI-2), as well as angiogenic markers basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF). As illustrated in Figure 4A (left column), all these markers were produced constitutively by H5V cells, but differed in their level of expression ¾ high (PAI-1, uPA), low to moderate (VEGF, bFGF, PAI-2), or extremely low (tPA), as reproduced in six independent experiments. Importantly, infection of H5V cell cultures with either MVMp or MVMp/IP-10 did not cause any significant alteration in this expression profile, as measured 5 or 24 hours postinfection (data not shown). This argues against the hypothesis that these putative determinants of the H5V cell malignant phenotype are targets for the MVMp-mediated antineoplastic activity, although we cannot rule out a direct interference of the virus with the expression of other relevant yet undetermined H5V cell markers of malignancy.

Suppression of H5V tumor formation by MVMp vectors requires T cells: As shown above, (recombinant) MVMp had a direct influence on H5V cell populations, namely the reduction of their growth in culture. This effect might contribute to inhibiting tumor development, especially under conditions in which most transformed cells become infected (ex vivo approach). Yet, the (recombinant) MVMp-induced suppression of H5V culture growth was limited and transient (Fig 1B), suggesting that it may not be sufficient on its own to account for the protection of mice against hemangioma development. This was confirmed through experiments using T-cell-deficient mice (data not shown). When C57Bl/6 nu⁺/nu⁺ mice served as recipients for H5V cell implantation, animals inoculated with MVMp-infected cells did develop tumors more slowly than the ones receiving mock-treated implants. Furthermore, in the former group, the primary lesions were restricted to small, isolated foci. However, this inhibition was of short duration, and all mice in both groups eventually developed tumors and metastases, and succumbed to the disease. This contrasts with the long-term antihemangioma protection given by MVMp in immunocompetent animals under ex vivo conditions (Fig 2). Hence, the direct effects of MVMp on H5V cells (cytotoxicity or still unknown phenotypic changes) may lead to a delay in tumor development, but have to be enforced by host T cells in order that ex vivo infection with the parvovirus results in a suppression of hemangioma progression. Likewise, the MVMp/IP-10 vector did not protect nude mice against progression of H5V-type vascular tumors (data not shown). Altogether, these observations point to the involvement of immune system effector cells as mediators of the antihemangioma activity of (recombinant) MVMp.

Stagnation of tumor growth is associated with long-term intratumoral persistence of activated lymphocytes, expressing IFN To investigate the immune component of the antineoplastic activity of MVMp and MVMp/IP-10, tumor specimens derived from treated immunocompetent mice were further analyzed for the expression of markers of effector cell recruitment and activation. The tumor expression profile was characterized by RT-PCR for the various transcripts listed in Figure 4A. In mock-treated mice, the early phase of primary tumor development was associated with expression of markers indicative of T-cell infiltration (CD4, CD8) and, to a variable extent, T and NK cell activation (CXCR3, CTLA4, IFN, granzymes A and B) (data not shown). This is in agreement with previous observations showing that host defense mechanisms involving, more particularly, T cells are responsible for the regression of primary H5V tumors.¹⁷ However, this antineoplastic reaction was rapidly exhausted and failed to eradicate H5V cells, as apparent from the recurrence and progression of mock-treated tumors. As illustrated in Figure 4A, these recurring control tumors showed an overall reduction of the expression of above-mentioned immune parameters (in particular CD8, CXCR3, granzyme B, and IFN) and accumulated proteolytic (uPA, tPA, PAI-1, PAI-2) and proangiogenic (VEGF, bFGF) factor transcripts. Infection with MVMp or MVMp/IP-10 treatment led to strong virus gene expression in primary tumors (Table 1), but apart from that had no significant effect on the expression pattern of these tumors during the initial stage of lesion formation and regression (data not shown). In contrast, the (recombinant) virus treatment was associated with striking changes in immune profile of recurring tumors (Fig 4A), compare the right and middle columns), although no viral transcripts were detected anymore during this later stage (1-2 months after the last virus injection) in residual tumors or other organs (data not shown). Virus-induced stagnation of recurring tumors was accompanied by their persistent infiltration with T and NK cells (CD4, CD8, NK 1.1 hallmarks). These cells appeared to be fully activated (CXCR3, CD40L, CTLA4, IL-2) and cytotoxic (granzymes A and B, perforin, IFN). Especially conspicuous was the high production of IFN transcripts in virus-treated tumor remnants, as compared to mock-treated growing lesions (Fig 4B). Accumulation of IFN can thus be assumed to contribute to suppression of tumor progression, this cytokine being both directly cytotoxic to H5V cells³⁹ and capable of stimulating the granzyme/perforin-mediated cytotoxic activity of effector immune cells.⁴⁰

Suppression of distant metastases is associated with reduction of PAI-1 expression in livers: Hemangioma progression is known to involve the recruitment of host endothelial cells into lesions, which has been reported to depend on the release of proangiogenic and proteolytic factors by transformed and stromal cells.^37,38,41 This prompted us to determine whether the (recombinant) MVMp-induced suppression of metastases was associated with changes in the expression of these factors in distant target tissues. To this end, RNA expression patterns were determined in C57Bl/6 mice carrying hepatic metastases (H5V cell-implanted animals subjected, or not, to in vivo treatment with MVMp) or free of metastatic foci (control mice without H5V cell implantation, and H5V cell-implanted animals treated ex vivo with MVMp or in vivo with MVMp/IP-10). Among the markers studied, components of PAS showed a significantly enhanced expression during disease progression. This applied in particular to PAI-1 transcripts, whose continuous accumulation constituted a hallmark of hemangioma metastases in the liver (Fig 4B). Interestingly, the suppression of hepatic metastases in H5V tumor-bearing mice, as a result of parvovirus treatment, correlated with a markedly reduced level of liver PAI-1 transcripts. It is presently unclear whether the source of hepatic PAI-1 overproduction in the presence of H5V metastases lays in the tumor cells colonizing the liver or in the hepatic parenchyma reacting to this process. However that may be, the question arose whether (recombinant) MVMp inhibited the formation of liver metastases by preventing hepatic cells and/or H5V tumor cells, settled in liver, from generating a local microenvironment favorable to the recruitment of new endothelial cells and cancer progression. Although it cannot be ruled out, this possibility is not supported by the evidence available to date. On one hand, MVMp and its recombinant derivative had no significant effect on the ability of H5V cell cultures or tumors to produce proangiogenic (VEGF, bFGF) and proteolytic (PAI-1 and 2, uPA, and tPA) molecules (see above and Fig 4A). On the other hand, MVMp gene expression in liver could only be detected at low level soon after virus inoculation (Table 1) and became undetectable at later times, in particular in liver tissues that were free of metastatic foci (data not shown). Altogether, these data lead us to assume that the (recombinant) MVMp-dependent inhibition of hepatic PAI-1 production in tumor-bearing mice is a consequence rather than a cause of the virus-induced suppression of metastases formation, resulting from elimination of malignant H5V cells liable to invade distal organs.

Discussion

Over the past years, a number of viruses have been used for cancer gene therapy purposes.¹ Rodent parvoviruses typified by the MVMp and H-1 viruses are especially promising bases for the generation of viral vectors with antineoplastic activity, due to their capacity for replicating in transformed human cells and their oncotropic and oncosuppressive properties, which can be demonstrated in the absence of harmful side effects.^2,3,4,5,10 This prompted us and others to attempt to reinforce the genuine antineoplastic potential of these parvoviruses by supplementing them with therapeutic transgenes.^8,11,12 The present work constitutes a proof-of-principle study showing for the first time that a recombinant parvovirus can exert an antitumor activity after being administered in vivo, under conditions in which the wild-type virus has little protective effect. This was demonstrated in a highly aggressive metastatic hemangiosarcoma mouse model, using a vector based on the autonomous parvovirus MVMp to carry a therapeutic transgene IP-10. A number of properties of this vector system are worth noting. (a) The efficacy of this vector was quite remarkable in that a dose as low as 3´10⁷ RU of virus per mouse caused at least a doubling of the survival time in responding animals (about a hundred times less than usual doses of adenoviral vectors, including Ad/IP-10).²⁶ Because recombinant parvoviruses are comparable with other viral vectors with regard to their efficacy in expressing transgenes, as measured in cell cultures,^{11,12,20,26,42,43} it can be suggested that the oncotoxicity of parvoviruses may potentiate the effects of transgenes carried by their recombinant derivatives (see below). (b) Neither deleterious effects of MVMp/IP-10 attributable to unspecific transgene expression, nor side effects of parental MVMp were observed, even after multiple virus injections up to a 1000-fold higher dose than the one used for the recombinant. (c) Although no total cure was achieved, responding to therapy mice stayed alive for more than 6 months, which represents an over 100 days increase in their life expectancy (the equivalent of 10 years of human lifespan). The antitumor protection provided by MVMp/IP-10 is quite impressive because H5V model is an especially aggressive case of experimental hemangiosarcomas, which are usually refractory to complete cure by various treatments,^16,39,44,45 with IL-12 therapy being the most successful one.¹⁹

The antineoplastic activity of IP-10 is not without precedents, although the use of different delivery systems and tumor models precludes published data from being quantitatively compared with the present ones. The most powerful protection was observed using plasmocytoma cells stably transfected with an IP-10-expressing plasmid.²⁰ This study demonstrated that IP-10 has an antitumor potential which is mediated by T cells, but did not address the question of the in vivo delivery of IP-10 into preexisting malignancies. Application of IP-10 to established tumors gave variable results. When administered directly as IP-10 protein to immunocompromised mice bearing human tumor xenografts (Burkitt lymphoma, non small cell lung cancer)^21,22 or as vaccine in the form of tumor antigen/IP-10 fusion protein to immunocompetent mice bearing syngeneic tumors (38C-13 lymphoma),²³ IP-10 induced significant but transient antitumor reactions, without complete tumor eradication. The route of IP-10 delivery had a clear influence because treatment with a plasmid encoding the fusion protein failed to induce any protection.²³ In another study, an adenovirus-based vector carrying IP-10 proved to be more efficient than an equivalent recombinant transducing endostatin, in its ability to extend the life expectancy of B16.F10 melanoma-bearing mice, leading to full tumor eradication in one third of treated animals.²⁶ Two other attempts at using adenovirus-delivered IP-10 to treat mammary and colorectal adenocarcinomas and fibrosarcoma showed that IP-10 had moderate antineoplastic activity on its own, but a striking ability to potentiate antitumor reactions initiated by IL-12 or T-cell adoptive therapy.^24,25

The latter observation allows to suggest that the ability of IP-10 to foster ongoing immune reactions could be responsible for MVMp/IP-10-induced suppression of H5V hemangiosarcomas. It is known that lesions arising from H5V cells are immunogenic and serve as targets for a host defense reaction, which allows tumors of a small or medium size to be eradicated, but becomes overwhelmed and fails to prevent tumor recurrence when primary implants are too large.¹⁷ In particular, anti-H5V hemangiosarcoma responses were found to be potentiated by IFN and IL-12, while they were abrogated by anti-T-cell antibodies.^17,19,39 Our analysis of H5V lesions expression profiles indicated that inhibition of disease development in infected mice was associated with the persistent infiltration of stagnating tumors with activated effectors (CTL and NK) cells. Thus, we hypothesize that the parvovirus-mediated toxicity of NS1 for H5V cells stimulated intrinsic immune reaction, and that this response was intensified when the virus expressed IP-10 transgene in addition to its own NS1 and NS2 genes. The present study showed that MVMp delayed the growth of H5V cell cultures and exerted an oncosuppressive effect in the absence of transgene, although this effect was only apparent when MVMp was administered in vitro to H5V cells prior to their implantation, i.e., under conditions maximizing the fraction of tumor cells that are hit by the virus. It is thus possible that MVMp was able to reduce the load of implanted proliferating tumor cells in the ex vivo protocol. On the other hand, the cytotoxic action of MVMp is likely to be ineffective when the virus is administered in vivo and reaches only some of the tumor cells, especially because these are scattered between host endothelial cells that are recruited into established hemangiomas.^15,17 The major parvoviral cytopathic effector, NS1,⁴⁶ which is encoded by administered vectors, may potentiate effects of immunoactive transgene in following ways: (a) the NS1-mediated oncotoxicity might be expected to stimulate the release of tumor antigens, and (b) the NS1 ability to dysregulate cellular gene expression⁴⁶ might promote an immunogenic phenotype (costimulatory or adhesion molecules) of H5V cells. In turn, expression of IP-10 by transduced H5V cells could aid more efficient targeting of tumor cells by anti-H5V effectors because NK and activated T cells are both known to display the IP-10 receptor CXCR3 and to be sensitive to the activating and/or mobilizing properties of the chemokine.^47,48,49

Apart from affecting immune reactions, MVMp/IP-10 could also inhibit neovascularization in different ways, thereby contributing to the suppression of H5V lesions development. First, part of the protective effect provided by the parvovirus may be related to the decrease of the number of the H5V cells able to proliferate and recruit host endothelial cells, as suggested by the vector-induced inhibition of the H5V cells growth in cultures. Second, mobilization of activated lymphocytes into tumors could lead to a bystander effect and damage recruited host endothelial cells, besides H5V cells.⁵⁰ Furthermore, the MVMp/IP-10 may be endowed with a specific anti-endothelial activity. This possibility is raised by the recently reported capacity of human IP-10 for inhibiting endothelial cell growth, although the expression of the IP-10 receptor CXCR3 by normal and transformed endothelial cells is still a matter of controversy.^27,51,52 H5V cell cultures failed to express CXCR3 and were not sensitive to the MVMp-mediated production of IP-10 (present work and ²⁷). Hence, the CXCR3 expression detected in H5V tumors in vivo can tentatively be assigned to infiltrating activated T and NK cells.⁴⁷ Yet, the IP-10 receptor status of H5V cells and recruited host endothelial cells in tumors from implanted animals remains to be determined.

Apparently, primary tumors constitute likely candidate sites for maintaining a lasting immune response, induced during parvovirus treatment. To fully understand the mechanism of MVMp/IP-10-mediated protection, a second site needs to be considered for the virus-induced amplification of an antitumoral immune response ¾ lymphoid organs. The lymph nodes draining primary H5V lesions were found to sustain a significant level of virus-driven gene expression, besides the tumors themselves (data not shown). Whether invading tumor cells and/or components of lymphoid tissues are responsible for this extratumoral vector gene expression is currently under investigation. However that may be, the fact that tumor-draining lymph nodes are targets for MVMp-directed gene expression is especially interesting in the view of using this virus to transduce immunomodulating molecules. Lymph nodes constitute gathering sites for the effector cells that cooperate in eliciting an immune response. Thus, should H5V cells express IP-10 receptors in vivo (see above), the chemokine produced in DLN might also promote antitumoral defense by attracting primary tumor cells to the spot where the immune reaction is built up.⁵³

Besides validating the use of parvoviral vectors, this study substantiates the application of IP-10 to vascular tumor therapy. The chemokine IP-10 is recognized as a powerful anti-angiogenic agent and major downstream mediator of the potent antitumor cytokine IL-12.^27,54,55 Yet, to the best of our knowledge, IP-10 has not been tested to date against hemangiomas in clinical or preclinical trials. There is an urgent need for novel therapies of vascular tumors because currently applied treatments are not satisfactory or associated with substantial side effects.^14,56 The present work provides the first indication that hemangiosarcoma suppression can be achieved by means of IP-10. This strategy is further supported by the fact that no deleterious side effects associated with IP-10 applications have been reported so far (present investigation and ^{20,21,22,23,24,25,26}).

In conclusion, this study validates the use of a vector based on the autonomous parvovirus MVMp to carry a therapeutic transgene IP-10 in tumor cells and achieve a significant protection against preestablished tumors upon in vivo administration. Although further assessment is clearly required, this proof-of-principle study put MVMp and related parvoviruses up for inclusion in the series of viral vectors that deserve to be considered for cancer gene therapy applications.

Acknowledgements

We are grateful to JM Farber (NIH, Bethesda, MD, USA) for the mouse IP-10 cDNA clone and IP-10 primer sequences used in RT-PCR. We thank Search-LC (Heidelberg, Germany) for assistance in performing LightCyclerÔ QRT-PCR. We also thank the DKFZ Department of Cellular and Molecular Pathology (Director H-J Gröne) for performing histopathological examinations; and C Cziepluch and HC Morse (III) for critical reading of the manuscript. This work was supported by EU Commission Grants BIO 4 CT97-2167 and QLRT-2000-01010.

1 Kirn DH, McCormick F. Replicating viruses as selective cancer therapeutics. Mol Med Today 1996; 2: 519-527. Article MEDLINE

2 Cornelis JJ, Haag A, Kornfeld C et al. Autonomous parvovirus vectors. In: Cid-Arregui A, Garcia-Garrancá A, eds Viral Vectors: Basic Science and Gene Therapy Natick, MA: Eaton Publishing, 2000, 97-118.

3 Rommelaere J, Cornelis JJ. Antineoplastic activity of parvoviruses. J Virol Methods 1991; 33: 233-251. MEDLINE

4 Faisst S, Guittard D, Benner A et al. Dose-dependent regression of HeLa cell-derived tumours in SCID mice after parvovirus H-1 infection. Int J Cancer 1998; 75: 584-589. Article MEDLINE

5 Guetta E, Graziani Y, Tal J. Suppression of Ehrlich ascites tumors in mice by minute virus of mice. J Natl Cancer Inst 1986; 76: 1177-1180. MEDLINE

6 Russell SJ, Brandenburger A, Flemming CL et al. Transformation-dependent expression of interleukin genes delivered by a recombinant parvovirus. J Virol 1992; 66: 2821-2828. MEDLINE

7 Kestler J, Neeb B, Struyf S et al. Cis requirements for the efficient production of recombinant DNA vectors based on autonomous parvoviruses. Hum Gene Ther 1999; 10: 1619-1632. Article MEDLINE

8 Dupont F, Avalosse B, Karim A et al. Tumor-selective gene transduction and cell killing with an oncotropic autonomous parvovirus-based vector. Gene Ther 2000; 7: 790-796. MEDLINE

9 Haag A, Wayss K, Rommelaere J et al. Experimentally induced infection with autonomous parvoviruses, minute virus of mice and H-1, in the African multimammate mouse (Mastomys coucha). Comp Med 2000; 50: 613-621. MEDLINE

10 Kimsey PB, Engers HD, Hirt B et al. Pathogenicity of fibroblast- and lymphocyte-specific variants of minute virus of mice. J Virol 1986; 59: 8-13. MEDLINE

11 Haag A, Menten P, Van Damme J et al. Highly efficient transduction and expression of cytokine genes in human tumor cells by means of autonomous parvovirus vectors; generation of antitumor responses in recipient mice. Hum Gene Ther 2000; 11: 597-609. Article MEDLINE

12 Wetzel K, Menten P, Opdënakker G et al. Transduction of human MCP-3 by a parvoviral vector induces leukocyte infiltration and reduces growth of human cervical carcinoma cell xenografts. J Gene Med 2001; 3: 326-327. Article MEDLINE

13 Marchuk DA. Pathogenesis of hemangioma. J Clin Invest 2001; 107: 665-666. MEDLINE

14 McGarvey ME, Tulpule A, Cai J et al. Emerging treatments for epidemic (AIDS-related) Kaposi's sarcoma. Curr Opin Oncol 1998; 10: 413-421. MEDLINE

15 Williams RL, Risau W, Zerwes HG et al. Endothelioma cells expressing the polyoma middle T oncogene induce hemangiomas by host cell recruitment. Cell 1989; 57: 1053-1063. MEDLINE

16 Liekens S, Verbeken E, Vandeputte M et al. A novel animal model for hemangiomas: inhibition of hemangioma development by the angiogenesis inhibitor TNP-470. Cancer Res 1999; 59: 2376-2383. MEDLINE

17 Garlanda C, Parravicini C, Sironi M et al. Progressive growth in immunodeficient mice and host cell recruitment by mouse endothelial cells transformed by polyoma middle-sized T antigen: implications for the pathogenesis of opportunistic vascular tumors. Proc Natl Acad Sci USA 1994; 91: 7291-7295. MEDLINE

18 Wang C, Quevedo ME, Lannutti BJ et al. In vivo gene therapy with interleukin-12 inhibits primary vascular tumor growth and induces apoptosis in a mouse model. J Invest Dermatol 1999; 112: 775-781. Article MEDLINE

19 Vizler C, Rosato A, Calderazzo F et al. Therapeutic effect of interleukin 12 on mouse haemangiosarcomas is not associated with an increased anti-tumour cytotoxic T-lymphocyte activity. Br J Cancer 1998; 77: 656-662. MEDLINE

20 Luster AD, Leder P. IP-10, a -C-X-C- chemokine, elicits a potent thymus-dependent antitumor response in vivo. J Exp Med 1993; 178: 1057-1065. MEDLINE

21 Arenberg DA, Kunkel SL, Polverini PJ et al. Interferon-gamma-inducible protein 10 (IP-10) is an angiostatic factor that inhibits human non-small cell lung cancer (NSCLC) tumorigenesis and spontaneous metastases. J Exp Med 1996; 184: 981-992. MEDLINE

22 Sgadari C, Angiolillo AL, Cherney BW et al. Interferon-inducible protein-10 identified as a mediator of tumor necrosis in vivo. Proc Natl Acad Sci USA 1996; 93: 13791-13796. MEDLINE

23 Biragyn A, Tani K, Grimm MC et al. Genetic fusion of chemokines to a self tumor antigen induces protective, T-cell dependent antitumor immunity. Nat Biotechnol 1999; 17: 253-258. Article MEDLINE

24 Narvaiza I, Mazzolini G, Barajas M et al. Intratumoral coinjection of two adenoviruses, one encoding the chemokine IFN-gamma-inducible protein-10 and another encoding IL-12, results in marked antitumoral synergy. J Immunol 2000; 164: 3112-3122. MEDLINE

25 Palmer K, Hitt M, Emtage PC et al. Combined CXC chemokine and interleukin-12 gene transfer enhances antitumor immunity. Gene Ther 2001; 8: 282-290. MEDLINE

26 Regulier E, Paul S, Marigliano M et al. Adenovirus-mediated delivery of antiangiogenic genes as an antitumor approach. Cancer Gene Ther 2001; 8: 45-54. MEDLINE

27 Angiolillo AL, Sgadari C, Taub DD et al. Human interferon-inducible protein 10 is a potent inhibitor of angiogenesis in vivo. J Exp Med 1995; 182: 155-162. MEDLINE

28 Farber JM. Mig and IP-10: CXC chemokines that target lymphocytes. J Leukocyte Biol 1997; 61: 246-257. MEDLINE

29 Geras-Raaka E, Varma A, Ho H et al. Human interferon-gamma-inducible protein 10 (IP-10) inhibits constitutive signaling of Kaposi's sarcoma-associated herpesvirus G protein-coupled receptor. J Exp Med 1998; 188: 405-408. MEDLINE

30 Cornelis JJ, Becquart P, Duponchel N et al. Transformation of human fibroblasts by ionizing radiation, a chemical carcinogen, or simian virus 40 correlates with an increase in susceptibility to the autonomous parvoviruses H-1 virus and minute virus of mice. J Virol 1988; 62: 1679-1686. MEDLINE

31 Svetic A, Finkelman FD, Jian YC et al. Cytokine gene expression after in vivo primary immunization with goat antibody to mouse IgD antibody. J Immunol 1991; 147: 2391-2397. MEDLINE

32 Giese NA, Gazzinelli RT, Actor JK et al. Retrovirus-elicited interleukin-12 and tumour necrosis factor-alpha as inducers of interferon-gamma-mediated pathology in mouse AIDS. Immunology 1996; 87: 467-474. MEDLINE

33 Marti HH, Risau W. Systemic hypoxia changes the organ-specific distribution of vascular endothelial growth factor and its receptors. Proc Natl Acad Sci USA 1998; 95: 15809-15814. Article MEDLINE

34 Kimura K, Ando K, Ohnishi H et al. Immunopathogenesis of hepatic fibrosis in chronic liver injury induced by repeatedly administered concanavalin A. Int Immunol 1999; 11: 1491-1500. MEDLINE

35 Giese T. Development of quantitative RT-PCR tests for the expression of cytokine genes on the LightCycler. In: Meuer S, Wittwer C, Nakagawara K, eds Rapid Cycle Real-Time PCR: Methods and Applications Berlin: Springer, 2001, 251-261.

36 Andreasen PA, Egelund R, Petersen HH. The plasminogen activation system in tumor growth, invasion, and metastasis. Cell Mol Life Sci 2000; 57: 25-40. Article MEDLINE

37 Sabapathy KT, Pepper MS, Kiefer F et al. Polyoma middle T-induced vascular tumor formation: the role of the plasminogen activator/plasmin system. J Cell Biol 1997; 137: 953-963. MEDLINE

38 Bielenberg DR, Bucana CD, Sanchez R et al. Progressive growth of infantile cutaneous hemangiomas is directly correlated with hyperplasia and angiogenesis of adjacent epidermis and inversely correlated with expression of the endogenous angiogenesis inhibitor, IFN-beta. Int J Oncol 1999; 14: 401-408. MEDLINE

39 Dong QG, Graziani A, Garlanda C et al. Anti-tumor activity of cytokines against opportunistic vascular tumors in mice. Int J Cancer 1996; 65: 700-708. Article MEDLINE

40 Young HA, Hardy KJ. Role of interferon-gamma in immune cell regulation. J Leukocyte Biol 1995; 58: 373-381. MEDLINE

41 Bajou K, Noel A, Gerard RD et al. Absence of host plasminogen activator inhibitor 1 prevents cancer invasion and vascularization. Nat Med 1998; 4: 923-928. MEDLINE

42 Dranoff G, Jaffee E, Lazenby A et al. Vaccination with irradiated tumor cells engineered to secrete murine granulocyte-macrophage colony-stimulating factor stimulates potent, specific, and long-lasting anti-tumor immunity. Proc Natl Acad Sci USA 1993; 90: 3539-3543. MEDLINE

43 Addison CL, Arenberg DA, Morris SB et al. The CXC chemokine, monokine induced by interferon-gamma, inhibits non-small cell lung carcinoma tumor growth and metastasis. Hum Gene Ther 2000; 11: 247-261. Article MEDLINE

44 Lannutti BJ, Gately ST, Quevedo ME et al. Human angiostatin inhibits murine hemangioendothelioma tumor growth in vivo. Cancer Res 1997; 57: 5277-5280. MEDLINE

45 O'Reilly MS, Boehm T, Shing Y et al. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 1997; 88: 277-285. MEDLINE

46 Vanacker JM, Rommelaere J. Non-structural proteins of autonomous parvoviruses: from cellular effects to molecular mechanisms. Semin Virol 1995; 6: 291-297.

47 Loetscher M, Gerber B, Loetscher P et al. Chemokine receptor specific for IP10 and mig: structure, function, and expression in activated T-lymphocytes. J Exp Med 1996; 184: 963-969. MEDLINE

48 Taub DD, Sayers TJ, Carter CR et al. Alpha and beta chemokines induce NK cell migration and enhance NK-mediated cytolysis. J Immunol 1995; 155: 3877-3888. MEDLINE

49 Taub DD, Lloyd AR, Conlon K et al. Recombinant human interferon-inducible protein 10 is a chemoattractant for human monocytes and T lymphocytes and promotes T cell adhesion to endothelial cells. J Exp Med 1993; 177: 1809-1814. MEDLINE

50 Yao L, Sgadari C, Furuke K et al. Contribution of natural killer cells to inhibition of angiogenesis by interleukin-12. Blood 1999; 93: 1612-1621. MEDLINE

51 Salcedo R, Resau JH, Halverson D et al. Differential expression and responsiveness of chemokine receptors (CXCR1-3) by human microvascular endothelial cells and umbilical vein endothelial cells. FASEB J 2000; 14: 2055-2064. MEDLINE

52 Romagnani P, Annunziato F, Lasagni L et al. Cell cycle-dependent expression of CXC chemokine receptor 3 by endothelial cells mediates angiostatic activity. J Clin Invest 2001; 107: 53-63. MEDLINE

53 Ochsenbein AF, Sierro S, Odermatt B et al. Roles of tumour localization, second signals and cross priming in cytotoxic T-cell induction. Nature 2001; 411: 1058-1064. Article MEDLINE

54 Sgadari C, Angiolillo AL, Tosato G. Inhibition of angiogenesis by interleukin-12 is mediated by the interferon-inducible protein 10. Blood 1996; 87: 3877-3882. MEDLINE

55 Tannenbaum CS, Tubbs R, Armstrong D et al. The CXC chemokines IP-10 and Mig are necessary for IL-12-mediated regression of the mouse RENCA tumor. J Immunol 1998; 161: 927-932. MEDLINE

56 Barlow CF, Priebe CJ, Mulliken JB et al. Spastic diplegia as a complication of interferon Alfa-2a treatment of hemangiomas of infancy. J Pediatr 1998; 132: 527-530. MEDLINE

Figure 1 Susceptibility of H5V cells to infection by wild-type and recombinant MVMp viruses. A,B: Cell viability was assayed using MTT and LDH tests. Cultures were analyzed on day 3 (A) or at different times (B) following injection with indicated viruses. The MOI was varied (A) or set at 1 RU/cell (B). Results of MTT reduction measurements are expressed as percentages of the value obtained for the uninfected control at the corresponding time point. Results of LDH assays (amounts of enzyme released in the culture medium) are expressed as percentages of the total LDH content of the corresponding well. LDH release by uninfected cells was less than 5%. These data were confirmed using different culture conditions and alternative methods for assessing viability. C: IP-10 production by MVMp/IP-10-transduced cells was determined by ELISA. Culture supernatants were harvested at the indicated times following infection (MOI=1 RU/cell), and their IP-10 contents were measured. The data shown are mean±SEM from three independent experiments.

Figure 2 Reduction of the tumorigenicity of H5V endothelioma cells as a result of their ex vivo infection with MVMp. C57Bl/6 mice were injected subcutaneously with 5´10⁶ mock-treated or MVMp-infected (MOI=50 PFU/cell) H5V cells. Disease progression and outcome were determined as described in Materials and Methods by assessing the survival of implant-receiving animals (Kaplan-Meier survival curves, A), the growth of tumors (average tumor area±SEM, B) at the implantation site, and the absence of metastases at autopsy (as determined by macroscopic and histopathological examination of tissues, C). This experiment included five and six mice in the control and MVMp-treated groups, respectively.

Figure 3 Protection of mice against hemangiosarcoma progression as a result of their in vivo treatment with MVMp/IP-10. C57Bl/6 mice were injected subcutaneously with 5´10⁶ H5V cells. After tumor establishment (5 days postimplantation), the animals were exposed to a regimen of daily intratumoral MVMp/IP-10 injections (5 days, then a 2-day pause) followed by daily intraperitoneal injections (5 days, a 2-day pause, then again 5 days). The total dose was 3´10⁷ RU/mouse. A: Animal survival. B: Tumor growth. C,D: Metastases were determined as in Fig 2). Data are compiled from three experiments comparing mock-treated (n=12), MVMp-infected (n=13), and MVMp/IP-10-infected (n=13) animals.

Figure 4 Expression of markers of immune activation and angiogenesis in parvovirus-infected H5V tumor-bearing mice. C57Bl/6 mice were subjected to ex vivo or in vivo infection protocols described in the legends to Figures 2 and 3. Two to 3 months after H5V cell implantation, the mice were autopsied and checked for primary tumor growth and metastases. Organs and tumor remnants were isolated and analyzed (three to six specimens per group). A: Expression of PAS components, angiogenic factors, and immune markers was measured by RT-PCR in cultured H5V cells (left column) and in H5V tumors that were either progressing (middle column: mock treatment or infection with MVMp in vivo) or stagnating (right column: infection with MVMp ex vivo, or MVMp/IP-10 in vivo). Specific PCR products were obtained, visualized, and subjected to densitometric analysis as described in Materials and Methods. Signal intensities were quantitated using NIH Image 1.62 software, and the levels of normalized transcript expression (arbitrary OD units) were ranked as undetectable, low (values <200); moderate (values 201-400), and high (values >401). B: RT-PCR signals obtained for tumor-associated IFN and liver-associated PAI-1 in control and H5V cell-implanted mice treated according to the ex vivo or in vivo protocol.

Table 1 Parvovirus NS1 gene transcription in H5V tumor-bearing mice