Oncolysis by viral replication and inhibition of ... - Wiley Online Library

869

Oncolysis by Viral Replication and Inhibition of Angiogenesis by a Replication-Conditional Herpes Simplex Virus that Expresses Mouse Endostatin John T. Mullen, M.D.1 James M. Donahue, M.D.1 Soundararajalu Chandrasekhar, Sam S. Yoon, M.D.1 Wenbiao Liu, M.D.2 Lee M. Ellis, M.D.2 Hideo Nakamura, M.D., Ph.D.1 Hideki Kasuya, M.D., Ph.D.1 Timothy M. Pawlik, M.D., M.P.H.1 Kenneth K. Tanabe, M.D.1

Ph.D.

1

1

Division of Surgical Oncology, Department of Surgery, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts.

2

Department of Surgery and Cancer Biology, University of Texas M. D. Anderson Cancer Center, Houston, Texas.

BACKGROUND. In preclinical models, infection of tumors by oncolytic strains of herpes simplex virus 1 (HSV-1) resulted in the destruction of tumor cells by viral replication and release of progeny virion that infected and destroyed adjacent tumor cells. However, complete tumor regression was rarely observed. METHODS. To augment the antitumor effect of viral oncolysis, a replication conditional HSV-1 mutant (HSV-Endo) was constructed in which the murine endostatin gene was incorporated into the HSV-1 genome. RESULTS. Replication of HSV-Endo effectively destroyed several colon carcinoma cell lines in vitro. Secretion of endostatin by HSV-Endo–infected HT29 human colon carcinoma cells was confirmed by Western blot analysis. The secreted endostatin was biologically active as assessed in a chick chorioallantoic membrane assay. Importantly, endostatin production at the site of viral replication did not inhibit viral replication. Direct injection of HSV-Endo into flank tumors caused tumor destruction, and some of the HSV-Endo–treated flank tumors completely sloughed. Immunohistochemical staining of the tumors revealed a decreased number of blood vessels in the HSV-Endo–treated group versus the control group. CONCLUSIONS. The oncolytic HSV-1 mutant HSV-Endo provided a two-pronged therapy; namely, inhibition of angiogenesis and direct tumor cell destruction by viral replication. Cancer 2004;101:869 –77. © 2004 American Cancer Society. KEYWORDS: herpes simplex virus 1, oncolysis, angiogenesis, endostatin, colon carcinoma.

T

Supported by Grants CA76183, CA71345, GM07035, and DK43352 from the National Institutes of Health; the Claude E. Welch Research Fellowship; the Marshall K. Bartlett Research Fellowship; and the Carl Ockerbloom Research Fund. Address for reprints: Kenneth K. Tanabe, M.D., Division of Surgical Oncology, Massachusetts General Hospital, Cox Building 626, Boston, MA 02114; Fax: (617) 724-3895; E-mail: [email protected] Received April 22, 2004; accepted May 13, 2004.

o our knowledge, the majority of cancer gene therapy applications to date have employed replication-defective viruses as vectors, whereby antineoplastic effects result solely from delivery and expression of therapeutic transgenes.1 The effectiveness of this approach is limited because only a small subpopulation of the target cells are transduced. An alternative strategy exploits viral replication for tumor cell lysis, or oncolysis. Viral oncolysis is an efficient mechanism for tumor destruction, as infection of tumor cells leads to rapid cell death and the release of progeny virion that can then infect neighboring tumor cells.2 Nonetheless, herpes simplex virus 1 (HSV-1)–mediated viral oncolysis alone is not sufficient to completely eradicate tumors in many animal models. In fact, administration of the wild-type HSV-1 strains does not cure mice with diffuse colon carcinoma liver metastases, even at doses associated with significant toxicity. Therefore, we attempted to design oncolytic HSV-1 vectors that deliver therapeutic transgenes to augment their efficacy. First-generation mutants, such as rRp4503 and HSV1yCD,4 harbor suicide transgenes that encode

© 2004 American Cancer Society DOI 10.1002/cncr.20434 Published online 30 June 2004 in Wiley InterScience (www.interscience.wiley.com).

870

CANCER August 15, 2004 / Volume 101 / Number 4

enzymes that activate cytotoxic precursors to their active metabolites. These metabolites then kill neighboring, uninfected tumor cells, a process termed the “bystander effect.”5 These first-generation vectors have proven effective as antitumor agents, yet the cytotoxic metabolites these viruses activate in turn inhibit viral replication.6 In addition, tumor cell resistance to these metabolites is common. To our knowledge, treatment with these vectors has not produced complete tumor regression in animal models. We sought to employ a transgene whose protein product does not antagonize viral replication, to which tumor resistance would not develop, and which had previously been shown to cause tumor regression in animal models. We chose the transgene for murine endostatin. Endostatin is a potent inhibitor of angiogenesis,7 a process critical for the growth and metastasis of tumors. We previously demonstrated that murine endostatin, constitutively expressed by stable transfectants of renal and colon carcinoma cell lines, inhibits tumor formation in the flank, lung, and liver.8 However, in this study we evaluated the efficacy of endostatin transferred into tumor cells before their inoculation into mice. Our goal was to determine the efficacy of an oncolytic HSV-1 mutant that delivers the gene for murine endostatin against established tumors. In the current study, we describe the construction of a replication-conditional HSV-1 mutant that delivers the murine endostatin transgene. Our results demonstrate that the virus effectively destroys tumor cells in preference to normal cells and simultaneously expresses mouse endostatin. The secreted endostatin is biologically active in vivo, as assessed in a chick chorioallantoic membrane (CAM) assay. Endostatin expression does not inhibit replication of the virus. Intratumoral viral replication combined with endostatin secretion inhibits tumor growth and causes tumor sloughing in mice.

MATERIALS AND METHODS Cell Lines and Viruses The human colon carcinoma cell lines HT29, NCIH508, LoVo, and LS174T, as well as the African Green Monkey kidney cell line Vero, were obtained from the American TypeCulture Collection (Rockville, MD). The mouse colon carcinoma cell line, MC26, was obtained from the National Cancer Institute Tumor Repository (Frederick, MD). All cell lines were maintained in Dulbecco-modified Eagle’s medium, 10% fetal bovine serum, 100 U/mL penicillin, and 100 ␮g/mL streptomycin. Primary human hepatocytes were prepared as described.9 Primary human keratinocytes were kindly provided by James Rocco (Massa-

chusetts Eye and Ear Infirmary, Boston, MA). The HSV-1 vector, hrR3 (kindly provided by Sandra Weller, University of Connecticut, New Haven, CT), is comprised of an insertion of the lacZ gene into the UL39 locus of the parent virus, KOS (kindly provided by Donald Coen, Harvard Medical School, Boston, MA). The viruses were propagated and titered on Vero cells. Heat inactivation of viruses was performed as described.10

Construction of HSV-Endostatin The recombining plasmid, pKpX2-endo-afpI, was constructed as follows. cDNA specimens encoding green fluorescent protein were excised from the pQBI25-fC1 plasmid (Quantum Biotechnologies, Carlsbad, CA) with SpeI and NotI and inserted into pcDNA3.1 (Invitrogen, Carlsbad, CA). The resulting expression cassette, with the cytomegalovirus (CMV) promoter upstream and the polyA tail downstream of the green fluorescent protein gene, was excised as a PmeI fragment. This was then inserted into the StuI site of pKpX2 (kindly provided by E. Antonio Chiocca, Massachusetts General Hospital, Boston, MA), which contains the UL39 gene, to produce pKpX2-afpI. The 1.9kilobase (kb) NruI fragment of pEndoSTHB,8 which contains the mouse endostatin cDNA downstream of the CMV promoter and the murine immunoglobulin (Ig) kappa chain signal peptide and upstream of a c-myc epitope and polyhistidine (His) tag, was subcloned into the EcoRV site of pKpX2-afpI to create the plasmid pKpX2-endo-afpI. XbaI-linearized pKpX2endo-afpI and KOS DNA were cotransfected into Vero cells with LipofectAMINE/PLUS (Gibco; Gaithersburg, MD) according to the manufacturer’s instructions. Vero cell lysates were then freeze-thawed three times to release infectious viruses and then replated onto fresh Vero cells with an agarose overlay. Recombinant viruses were identified as green fluorescent plaques under fluorescence microscopy and were isolated and plaque purified three times on Vero cells. HSV-endostatin (HSV-Endo) virus stocks were then prepared by infection of Vero cells and titered by standard plaque assay in duplicate.

Southern Blot Analysis Viral DNA was isolated after lysis of infected Vero cells with 0.5% sodium dodecyl sulfate (SDS) and proteinase K (500 ␮g/mL) by repeated phenol-chloroform extraction and ethanol precipitation. DNA was digested with NruI, BamHI, or SpeI, separated by agarose gel electrophoresis, and transferred to a nylon membrane (Amersham Corporation, Arlington Heights, IL). Probes to ICP6 (0.7 kb BamHI fragment of pKpX2), green fluorescent protein (180 bp SpeI-NheI

HSV-1 Oncolysis and Angiogenesis/Mullen et al.

fragment of pKpX2-afpI), and mouse endostatin (560 bp BamHI-HindIII fragment of pEndoSTHB) were labeled, hybridized to the membrane, and detected with an enhanced chemiluminescence (ECL) system (Amersham Corporation) as described by the manufacturer.

Western Blot Analysis of HSV-Endo–Conditioned Supernatant Fluid A total of 8 ⫻ 105 HT29 cells were plated onto 60-mm plates and infected the following day with HSV-Endo and the control vector, hrR3, at a multiplicity of infection (MOI) of 0.5. One milliliter of conditioned, serumfree supernatant fluid was harvested 24 hours later and concentrated in a Microcon 10 microconcentrator (Amicon, Beverly, MA) to 20 ␮L and subjected to electrophoresis under reducing conditions on a 4 –20% Tris-glycine gel (Novex, San Diego, CA). Proteins were transferred onto a poly(vinylidene difluoride) membrane (Hybond-P; Amersham Pharmacia, Arlington Heights, IL) and probed with an anti-c-myc mouse monoclonal antibody (Sigma Chemical Company, St. Louis, MO), followed by a horseradish peroxidaseconjugated antimouse Ig and a signal amplification step with the enhanced ECL kit (Amersham). The experiment was performed twice.

Chick CAM Assay Fertilized white Leghorn eggs were obtained at 3– 6 days of age and maintained in a humidified incubator at 37 °C until used for the assay at 9 days of age. The CAMs were separated from the shells, creating false air sacs, and small windows were cut directly over the false air sacs to provide access to the underlying CAMs. Sterile filter discs were then saturated with basic fibroblast growth factor (bFGF) with or without the addition of concentrated media conditioned by either hrR3 or HSV-Endo–infected tumor cells. The discs were placed onto the CAM in an area with a low density of preexisting blood vessels, and the window was sealed with sterile tape. After 72 hours of incubation, the filter discs with their attached CAM tissue were dissected free from the embryos and observed under a stereomicroscope. This experiment was performed twice.

Viral Replication and Cytotoxicity Assays Viral replication assays were performed as described.11 Briefly, 1 ⫻ 106 cells were infected with 2 ⫻ 106 plaque-forming units (pfu) of virus for 2 hours, at which time unadsorbed virus was removed by washing with a glycine–saline solution (pH ⫽ 3.0). The supernatant fluid and cells were harvested 40 hours after infection, exposed to 3 freeze-thaw cycles to re-

871

lease virions, and titered on Vero cells. The results represent the mean of three independent experiments. Viral cytotoxicity assays were performed as described.12 Briefly, cells were plated onto 96-well plates at 5000 cells per well for 36 hours. Virus was added at MOI values ranging from 0.0001 to 10 and incubated for 6 days. The number of surviving cells was quantitated using a colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Experiments were performed in triplicate.

Animal Studies Animal studies were performed in accordance with guidelines issued by the Massachusetts General Hospital subcommittee on research animal care. For in vivo viral replication studies, 1 ⫻ 106 MC26 cells in 100 ␮L of Hank’s balanced salt solution (HBSS) were injected subcutaneously into the bilateral flanks of syngeneic BALB/c mice (Charles River Labs, Wilmington, MA). When the tumors reached an approximate volume of 300 mm3, 1 ⫻ 108 pfu of HSV-1 in 100 ␮L HBSS was injected into the flank tumor. Each of the bilateral tumors on a given mouse were treated in the same fashion so as to provide duplicate results, and the experiments were repeated a minimum of two times. The mice were euthanized 48 hours later and the flank tumors were resected, weighed, minced, and homogenized in HBSS containing collagenase A (1 mg/mL; Boehringer Mannheim, Mannheim, Germany). The samples were incubated for 1–2 hours at 37 °C, subjected to 3 freeze-thaw cycles, centrifuged at 3000 revolutions per minute to remove cellular debris, and the supernatant fluids were recovered and titered on Vero cells (n ⫽ 4 tumors per viral strain). To assess the therapeutic efficacy of HSV-Endo injected into flank tumors, a single cell suspension of 1 ⫻ 106 MC26 cells in 100 ␮L HBSS was injected into the right flanks of BALB/c mice. After 6 days, when the tumors reached an average volume of 50 mm3, the mice were randomly divided into treatment groups of 7 mice per group. Intratumoral injections of 1 ⫻ 108 pfu HSVEndo, hrR3, or heat-inactivated HSV-Endo in 100 ␮L media were performed on Days 6, 8, 10, and 12. Tumor width (W) and length (L) were measured by Vernier caliper every 2– 4 days. The tumor volume (TV) was determined by the following formula: TV ⫽ (L ⫻ W2)/2. This experiment was performed three times. A similar experiment was performed on separate groups of mice (n ⫽ 7 mice per group). However, in these experiments, the mice were euthanized on Day 22 (when the flank tumors were developing necrosis) and the tumors were harvested for immunohistochemical analysis for CD31 expression. The animal

872


experiments for immunohistochemical studies were performed twice.

Immunohistochemistry Tumor specimens frozen in optimum cutting temperature (Miles, Inc., Elkhart, IN) solution were sectioned (8 –10 ␮m thick), mounted on positively charged Superfrost slides (Fisher Scientific Company, Houston, TX), and air-dried for 30 minutes. Tissue specimens were fixed in cold acetone (5 minutes), 1:1 acetone/ chloroform (5 minutes), and acetone (5 minutes), and then washed with phosphate-buffered saline (PBS) 3 times for 3 minutes each. Specimens were then incubated with 3% hydrogen peroxide in PBS (v/v) for 12 minutes at room temperature to block endogenous peroxidase. Sections were washed 3 times for 3 minutes each with PBS (pH ⫽ 7.5) and incubated for 20 minutes at room temperature in a protein-blocking solution comprised of PBS supplemented with 1% normal goat serum and 5% normal horse serum. Frozen sections were incubated with the primary antibody directed against mouse CD31/PECAM-1 (Pharmingen, San Diego, CA, 1:100 dilution in proteinblocking solution) overnight at 4 °C. Sections were then rinsed 3 times for 3 minutes each in PBS and incubated for 10 minutes in protein-blocking solution before the addition of peroxidase-conjugated secondary antibody. The secondary antibodies used to stain for CD31 (peroxidase-conjugated goat anti-rat IgG [H ⫹ L], Pharmingen) were diluted 1:200 in proteinblocking solution. After incubation with the secondary antibody for 1 hour at room temperature, the samples were washed and incubated with stable diaminobenzidine (Research Genetics, Huntsville, AL) substrate. Staining was monitored under a bright-field microscope, and the reaction was stopped by washing with distilled water. The sections were counterstained with Gill’s no. 3 hematoxylin (Sigma) for 15 seconds and mounted with Universal Mount (Research Genetics). Control specimens were treated with a similar procedure except that the primary antibody was omitted.

Quantification of Tumor Microvessel Density To quantify tumor microvessel counts, photographs of tumor microvessels (stained for CD31) were obtained from 5 random 0.159-mm2 fields of each slide at ⫻ 100 magnification using a Sony 3-chip camera (Sony, Montvale, NJ) mounted on a Zeiss universal microscope (Carl Zeiss, Thornwood, NY). The pictures were converted to grayscale using Adobe Photoshop (Adobe Systems, Inc., San Jose, CA), and the vessels were quantitated with NIH Image 1.62 (available from URL: http://rsb.info.nih.gov/nih-image/Default.html [accessed January 20, 2004]).

RESULTS Construction of a Replication-Conditional HSV-1 Mutant Expressing Murine Endostatin and Green Fluorescent Protein (HSV-Endo) HSV-Endo was constructed by recombining sequences from the plasmid pKpX2-endo-afpI into the UL39 locus of the wild-type parent KOS (Fig. 1A). Plasmid pKpX2-endo-afpI contains transgenes expressing mouse endostatin and green fluorescent protein. The mouse endostatin gene was subcloned downstream of the CMV promoter and murine Ig kappa chain signal peptide to ensure constitutive expression and secretion of mouse endostatin by infected cells. Likewise, cDNA for green fluorescent protein was subcloned downstream of a second CMV promoter to ensure the immediate expression of this visible marker protein in infected cells. pKpX2-endo-afpI was cotransfected with KOS DNA into Vero cells, and the resulting recombinant plaques that appeared fluorescent green by ultraviolet microscopy were isolated. One clone, HSV-Endo, was plaque purified to homogeneity on Vero cells. The DNA structure of HSV-Endo (Fig. 1A) was confirmed by restriction endonuclease digestion and Southern blot analysis, using probes to the UL39, green fluorescent protein, and mouse endostatin genes (Fig. 1B). DNA prepared from HSV-Endo, digested with NruI, and hybridized to a UL39 probe, yielded a 4.6-kb fragment instead of the 0.9-kb fragment obtained with the wild-type KOS. This 4.6-kb fragment is expected when the 3.7-kb cassette containing murine endostatin and green fluorescent protein has integrated into the UL39 locus by homologous recombination. In addition, polymerase chain reaction amplification of mouse endostatin sequences from HSV-Endo indicated the presence of this gene within the viral genome (data not shown). Secretion of mouse endostatin into the supernatant of HSV-Endo–infected cells was demonstrated by Western blot analysis (Fig. 1C). hrR3 serves as an ideal control virus for experiments because it is identical to HSV-Endo in three respects. First, they are both replication-conditional HSV-1 mutants. Second, they are both defective in ICP6 (viral ribonucleotide reductase) expression by virtue of disruption of the UL39 locus. Third, they are both derived from the KOS strain. HT29 human colon carcinoma cells and Vero cells were infected with either hrR3 or HSV-Endo using an MOI of 0.5, and after 24 hours the serum-free supernatant fluid was concentrated and analyzed by SDS/ polyacrylamide gel electrophoresis. Distinct bands at 29 kilodaltons (kD), the expected size of the endostatin-c-myc-His fusion protein, were observed in the

FIGURE 1. Construction of herpes simplex virus-endostatin (HSV-Endo). (A) The UL39 gene of the wild-type virus KOS encodes the viral protein ribonucleotide reductase (ICP6). hrR3 contains an insertion of the lacZ gene into the UL39 locus. HSV-Endo contains an insertion in the UL39 locus of genes for green fluorescent protein and mouse endostatin (ME). The locations of the probes used for Southern blot analysis (UL39, green fluorescent protein, and ME) are shown. (B) Southern blot analysis performed on DNA prepared from the pKpX2-endo-afpI plasmid that was used for homologous recombination to derive HSV-Endo, and the viruses KOS and HSV-Endo. DNA digested with NruI and probed with an 800-base pair (bp) BamHI fragment of UL39 yields the 0.9-kilobase (kb) fragment from KOS that is expected in the absence of homologous recombination. The 4.6-kb fragment observed from HSV-Endo results from integration of the 3.7-kb sequence containing cDNA for green fluorescent protein and ME into the UL39 locus. DNAs digested with BamHI and hybridized with a probe to green fluorescent protein produced the expected 1.5-kb fragment in HSV-Endo, indicating the presence of the green fluorescent protein gene sequence in the mutant viral genome. Lastly, DNAs digested with SphI and hybridized with a probe to ME yielded the expected 900-bp and 1.4-kb fragments. (C) Western blot analysis of secreted ME-c-myc-His fusion protein. Conditioned supernatants from HT29 cells infected with hrR3 (lane 1) or HSV-Endo (lane 3) and from Vero cells infected with HSV-Endo (lane 2) were concentrated and then detected with an anti-c-myc monoclonal antibody. The size of the ME-c-myc-His fusion protein is estimated based on peptide sequence analysis to be Mr 29,000. (D) Chorioallantoic membrane assay. Conditioned, serum-free medium from HSV-Endo and hrR3-infected tumor cells was concentrated, filtered, and then applied to filter discs together with basic fibroblast growth factor (bFGF). After 72 hours of incubation, the filter discs were observed under a stereomicroscope and photographed. Left panel: bFGF only; middle panel: hrR3 and bFGF; right panel: HSV-Endo and bFGF.

874


HSV-Endo– conditioned supernatant fluids but not in the control.

HSV-Endo–Expressed Endostatin is Biologically Active In Vivo To test whether the endostatin secreted by HSV-Endo– infected cells is able to inhibit in vivo angiogenesis, we used a CAM assay. Conditioned, serum-free medium from HSV-Endo and hrR3-infected cells was concentrated and filtered to remove any contaminating viral particles and then applied to filter discs together with bFGF. We observed inhibition of angiogenesis in CAMs treated with HSV-Endo– conditioned medium compared with the control CAMs treated with either hrR3-conditioned medium or bFGF alone (Fig. 1D).

HSV-Endo Replicates More Efficiently in Colon Carcinoma Cells than in Normal Cells and Endostatin Expression Does not Inhibit Viral Replication HSV-Endo harbors an insertional inactivation of the UL39 gene, and so it is defective in its expression of viral ribonucleotide reductase (ICP6). Accordingly, we presumed that, similar to hrR3, HSV-Endo should replicate more robustly in cells with high levels of the complementing mammalian ribonucleotide reductase.12 We have previously shown that ribonucleotide reductase expression is substantially greater in colon carcinoma cells than in cultured primary hepatocytes12 and that the ICP6-defective HSV-1 mutant hrR3 replicates preferentially in colon carcinoma cells rather than in hepatocytes.11 To test whether HSVEndo similarly replicates more efficiently in colon carcinoma cells than in normal cell types, we compared the replication of HSV-Endo with that of KOS and hrR3 in primary cultures of human hepatocytes and human keratinocytes and in HT29 human colon carcinoma cells. HSV-Endo replication in HT29 cells was as robust as that of the wild-type strain KOS and identical to that of hrR3 (Fig. 2A). In contrast, HSV-Endo replication was 4 log orders less in human keratinocytes and human hepatocytes than in human colon carcinoma cells. Similar to other ICP6-defective HSV-1mutants, HSV-Endo replicates preferentially in cancer cells compared with normal cells. Although HSV-Endo replicates as well as KOS and hrR3 in HT29 colon carcinoma cells in vitro (Fig. 2A), an in vitro assay may not demonstrate the full effect of endostatin on viral replication. We therefore developed an in vivo viral replication assay in which the interactions between viral replication and endostatin production can be assessed. We injected KOS, hrR3, or HSV-Endo into MC26 mouse colon carcinoma tumors growing in the flanks of mice. We allowed viral replication to proceed for 48 hours, at which point we

FIGURE 2. Combined herpes simplex virus-endostatin (HSV-Endo)–mediated oncolysis and endostatin expression. (A) The titer of infectious virion recovered was determined 40 hours after infection of colon carcinoma cells, hepatocytes, and keratinocytes with KOS, hrR3, or HSV-Endo. (B) The titer of infectious virion recovered was determined 48 hours after inoculation of flank tumors with KOS, hrR3, or HSV-Endo. Open bar: KOS; shaded bar: hrR3; solid bar: herpes simplex virus-endostatin.

harvested and homogenized the flank tumors to recover virus for titering. Titers of ICP6-defective HSV-1 recovered after direct intratumoral inoculation are maximal 48 hours after inoculation, and are undetectable 10 days after intratumoral inoculation. We observed no significant difference between hrR3 and HSV-Endo in their replication in tumors (Fig. 2B), indicating that endostatin secretion does not significantly affect replication of ICP6-defective HSV-1 mutants.

HSV-Endo Destroys Colon Carcinoma by a Two-Pronged Approach: Oncolysis and Inhibition of Tumor Angiogenesis We measured the oncolytic effects of HSV-Endo in vitro by a standard cytotoxicity assay, in which a panel of human and mouse colon carcinoma cell lines were infected with increasing titers of virus, after which cell survival was determined. All colorectal carcinoma cell lines that were evaluated exhibited appreciable susceptibility to HSV-Endo–mediated oncolysis (Fig. 3). As expected, the cytopathic effect of HSV-Endo in vitro was identical to that of its wild-type parent, KOS, as well as to hrR3. To examine the antitumoral efficacy of HSV-Endo in vivo, MC26 mouse colon carcinoma tu-


875

FIGURE 3.

Herpes simplex virus-endostatin (HSVEndo)–mediated cytotoxicity in vitro. Colon carcinoma cells were plated in 96-well plates and infected with HSV-Endo at the indicated multiplicity of infection values. Cell survival was determined by MTT assay 6 days later. Representative results of experiments performed in triplicate are shown. Open triangles: MC26; solid circles: NCIH 508; open diamonds: LoVo; open circles: LS 174T; open squares: HT29.

mors were established in the flanks of syngeneic, immunocompetent BALB/c mice, and intratumoral virus injection was started on Day 6, when the tumors were approximately 50 mm3. Three groups of mice (n ⫽ 7 mice per group) each received 4 intratumoral injections of virus on Days 6, 8, 10, and 12 after tumor implantation. One group received HSV-Endo and the control groups received either hrR3 or heat-inactivated HSV-Endo (HI-Endo). To enable us to identify the antineoplastic effects of endostatin, we specifically selected a dose for which hrR3 alone had minimal efficacy. By size criteria, no differences were initially noted among the treatment groups (Fig. 4). However, the macroscopic morphology of the tumors was quite different among the groups. HSV-Endo–treated tumors were soft and becoming centrally necrotic— characteristics not captured by size measurements. This continued until roughly the 23rd day after tumor implantation (and 17 days after viral administration), at which time some HSV-Endo–treated tumors completely sloughed. Thus, whereas tumor size measurements do not reflect a difference among the groups until Day 23, a clear difference existed between HSVEndo–treated tumors and control tumors in the days leading up to the change in size from tumor slough. After tumors sloughed, new tumors regrew by 7–10 days later, indicating the persistence of viable tumor cells. It is not surprising that the rate of growth of these new tumor nodules paralleled that of control tumors, because neither HSV-1 nor transgene product is present 4 weeks after initial treatment. As assessed by tumor measurements on Day 27, HSV-Endo reduced tumor growth an average of 88% and 83% relative to the control groups, HI-Endo and hrR3, respectively. We observed the same results when the experiment was repeated.

FIGURE 4. Antitumor activity of herpes simplex virus-endostatin (HSV-Endo). MC26 tumors growing in the flanks of BALB/c mice were injected with hrR3, HSV-Endo, or heat-inactivated HSV-Endo on Days 6, 8, 10, and 12 after tumor implantation. On Day 30, P ⫽ 0.0002 for HSV-Endo versus heat-inactivated hrR3; P ⫽ 0.0008 for HSV-Endo versus hrR3; and P ⫽ 0.026 for hrR3 versus heat-inactivated HSV-Endo (Mann–Whitney U tests). Open diamonds: heatinactivated herpes simplex virus-endostatin; open circles: hrR3; open triangles: herpes simplex virus-endostatin. We performed a similar experiment but harvested the tumors on Day 22 when they were growing at the same rate as hrR3-treated tumors to evaluate blood vessel counts. The tumors were analyzed by immunohistochemistry for CD31 expression, and vessel counts were significantly lower in HSV-Endo–treated tumors compared with hrR3-treated tumors (60 ⫾ 5 vs. 74 ⫾ 4; P ⬍ 0.05 by the Student t test). Assessment of vessel

876


counts at later time points (e.g., after the tumors had sloughed) was not reliable because the sloughed tumors were necrotic.

DISCUSSION Our laboratory has been interested in the development of replication-conditional HSV-1 mutants for viral oncolysis. We initially demonstrated that the ICP6defective HSV-1 mutant hrR3 preferentially replicates in colon carcinoma cells versus normal liver cells and effectively destroys colon carcinoma liver metastases.11 The antineoplastic efficacy is not dependent on host immunity11 and is independent of p53 status of the cancer cells.13 This virus, however, only inhibits tumor growth and does not lead to tumor regression or complete eradication. In fact, even the administration of the wild-type HSV-1 KOS produces limited antitumor effects (data not shown). Oncolysis alone by attenuated HSV-1 mutants may not be sufficient to achieve complete tumor regression. It is logical to examine delivery of therapeutic transgenes by replication-conditional HSV-1 mutants.14 The virus, rRp450, expresses the rat cytochrome P450 2B1 transgene, which initiates bioactivation of the prodrug cyclophosphamide into its active cytotoxic metabolite.3 Administration of rRp450 in combination with the prodrug cyclophosphamide more effectively destroys tumors than rRp450 alone.6 HSV1yCD is an ICP6-defective HSV-1 mutant that expresses the transgene encoding yeast cytosine deaminase.4 Cytosine deaminase converts the prodrug, 5-fluorocytosine (5-FC), to the cytotoxic agent, 5-fluorouracil. The administration of HSV1yCD in combination with 5-FC more effectively destroys tumors than HSV1yCD alone. However, replication of both these HSV-1 mutants is inhibited to some degree in the presence of their respective prodrugs by the cytotoxic metabolites that their transgene products generate. In addition, tumor resistance to these cytotoxic metabolites is common. Moreover, in animal models, we were unable to demonstrate tumor regression (as opposed to slower tumor growth) using these mutants together with their prodrugs.4,6 Accordingly, we sought to identify a transgene whose protein product does not antagonize viral replication, to which tumor resistance would not develop, and which had previously been shown to cause tumor regression in animal models. The murine endostatin transgene meets all of these criteria. Endostatin is a 20-kD fragment of collagen XVIII and is a potent inhibitor of angiogenesis.7 Angiogenesis is a critical process in the growth and metastasis of tumors and constitutes an important point in the control of cancer progression. This pro-

cess is regulated by proangiogenic and antiangiogenic factors. For tumors to grow and metastasize, they must progress through the “angiogenic switch” in which the balance favors angiogenesis.15 One strategy to shift the balance against tumor-induced angiogenesis is to deliver an endogenous inhibitor of angiogenesis, such as endostatin, to the tumor microenvironment. We have previously shown that mouse endostatin, constitutively expressed by stable transfectants of renal and colon carcinoma cell lines, inhibits the formation of tumors in different organ environments.8 Others have demonstrated growth inhibition of established tumors using replication-defective viruses such as adenovirus16,17 and adeno-associated virus18 to deliver the murine endostatin transgene. Our goal was to determine the antitumor efficacy of combination therapy with murine endostatin gene delivery in the context of an oncolytic, replication-competent viral vector. To that end, we constructed a replication-conditional HSV-1 that harbors the murine endostatin transgene (HSV-Endo). As expected, HSV-Endo behaves like hrR3 in terms of its oncolytic effect. Both of these mutants replicate preferentially in neoplastic cells versus normal cells by virtue of their inactivation of the gene encoding viral ribonucleotide reductase (UL39). On infection and replication within tumor cells, however, HSV-Endo–infected cells secrete mouse endostatin. We demonstrated that endostatin production at the site of viral replication does not inhibit viral replication, and the oncolytic effect of HSV-Endo is not antagonized by the local production of endostatin. This is in contrast to oncolytic vectors that harbor suicide transgenes that activate cytotoxic metabolites as a means to augment tumor cell death. The cytotoxic metabolites may cause “bystander” cell killing but may also inhibit viral replication, thereby antagonizing the viral oncolytic effect. In addition, secretion of endostatin by an oncolytic, replicationcompetent viral vector attacks two tumor compartments simultaneously, namely, tumor cells and endothelial cells. Approximately one-third of HSV-Endo–treated tumors sloughed, and we have confirmed this to be the case in repeated experiments. In contrast, we have never observed this type of response after numerous experiments involving treatment with HSV-1 mutants that are identical to HSV-Endo with respect to the absence of ICP6 but that do not express endostatin. This observation is consistent with secreted endostatin having been responsible for the antineoplastic effects observed in the tumors treated with HSV-Endo. Endostatin is known to inhibit angiogenesis, and the observed reduction in blood vessel counts in HSV-


Endo–treated tumors relative to hrR3-treated tumors is consistent with this mechanism of action. HSV-1 replication in tumors is not detectable beyond 1 week after viral administration, and the half-life of endostatin is extremely short. Tumor neovascularization is dependent on a balance of proangiogenic and antiangiogenic factors, and the transient expression of endostatin during viral replication within the tumors appears to have altered this balance. Tumors become necrotic in areas that have outgrown their blood supply. In the current study, we described the construction of a novel oncolytic HSV-1 that harbors the murine endostatin transgene. To our knowledge, the current study represents the first report of combined gene therapy with endostatin delivered in the context of a replication-competent viral vector. Efforts to make the therapeutic response to HSV-Endo more durable are currently underway.

7.

8.

9.

10.

11.

12.

13.

REFERENCES 1.

2. 3.

4.

5.

6.

Rosenberg SA, Blaese RM, Brenner MK, et al. Human gene marker/therapy clinical protocols. Hum Gene Ther. 2000;11: 919 –979. Martuza RL. Conditionally replicating herpes vectors for cancer therapy. J Clin Invest. 2000;105:841– 846. Chase M, Chung R, Chiocca EA. An oncolytic viral mutant that delivers the CYP2B1 transgene and augments cyclophosphamide chemotherapy. Nat Biotech. 1998;16:444 – 448. Nakamura H, Mullen JT, Chandrasekhar S, Pawlik TM, Yoon SS, Tanabe KK. Multimodality therapy with a replicationconditional herpes simplex virus 1 mutant that expresses yeast cytosine deaminase for intratumoral conversion of 5-fluorocytosine to 5-fluorouracil. Cancer Res. 2001;61:5447– 5452. Freeman SM, Abboud CN, Whartenby KA, et al. The “bystander effect”: tumor regression when a fraction of the tumor mass is genetically modified. Cancer Res. 1993;53: 5274 –5283. Pawlik TM, Nakamura H, Yoon SS, et al. Oncolysis of diffuse

14.

15. 16.

17.

18.

877

hepatocellular carcinoma by intravascular administration of a replication-competent, genetically engineered herpesvirus. Cancer Res. 2000;60:2790 –2795. O’Reilly MS, Boehm T, Shing Y, et al. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell. 1997;88:277–285. Yoon SS, Eto H, Lin CM, et al. Mouse endostatin inhibits the formation of lung and liver metastases. Cancer Res. 1999;59: 6251– 6256. Yoon SS, Carroll NM, Chiocca EA, Tanabe KK. Cancer gene therapy using replication-competent herpes simplex virus type 1. Ann Surg. 1998;228:366 –374. Lelie PN, Reesink HW, Lucas CJ Inactivation of 12 viruses by heating steps applied during manufacture of a hepatitis B vaccine. J Med Virol. 1987;23:297–301. Yoon SS, Nakamura H, Carroll NM, Bode BP, Chiocca EA, Tanabe KK. An oncolytic herpes simplex virus type 1 selectively destroys diffuse liver metastases from colon carcinoma. FASEB J. 2000;14:301–311. Carroll NM, Chiocca EA, Takahashi K, Tanabe KK. Enhancement of gene therapy specificity for diffuse colon carcinoma liver metastases with recombinant herpes simplex virus. Ann Surg. 1996;224:323–330. Yoon SS, Carroll NM, Chiocca EA, Tanabe KK. Influence of p53 on herpes simplex virus type 1 vectors for cancer gene therapy. J Gastrointest Surg. 1999;3:34 –38. Ichikawa T, Chiocca EA. Comparative analyses of transgene delivery and expression in tumors inoculated with a replication-conditional or -defective viral vector. Cancer Res. 2001;61:5336 –5339. Folkman J. Role of angiogenesis in tumor growth and metastasis. Semin Oncol. 2002;29:15–18. Feldman AL, Restifo NP, Alexander HR, et al. Antiangiogenic gene therapy of cancer utilizing a recombinant adenovirus to elevate systemic endostatin levels in mice. Cancer Res. 2000;60:1503–1506. Sauter BV, Martinet O, Zhang W, Mandeli J, Woo SLC. Adenovirus-mediated gene transfer of endostatin in vivo results in high level of transgene expression and inhibition of tumor growth and metastases. Proc Natl Acad Sci USA. 2000; 97:4802– 4807. Shi W, Teschendorf C, Muzyczka N, Siemann DW. Gene therapy delivery of endostatin enhances the treatment efficacy of radiation. Radiother Oncol. 2003;66:1–9.