Selectivity of oncolytic viral replication prevents antiviral immune response and toxicity, but does not improve antitumoral immunity

Abstract

Oncolytic infection elicits antitumoral immunity, but the impact of tumor-selective replication on the balance between antiviral and antitumoral immune responses has not yet been investigated. To address this question, we constructed the highly tumor-selective adenovirus Ad-p53T whose replication in target tumor cells is governed by aberrant telomerase activity and transcriptional p53 dysfunction. Telomerase-dependent or nonselective adenoviruses were constructed as isogenic controls. Following infection of mice with the nonselective adenovirus, viral DNA and mRNA levels correlated with strong stimulation of innate immune response genes and severe liver toxicity, whereas telomerase-/p53-specific replication did not trigger innate immunity and prevented liver damage. Compared to telomerase-dependent or unselective viral replication, telomerase-/p53-specific virotherapy significantly decreased antiviral CD8-specific immune responses and antiviral cytotoxicity in vivo. Consistent with our hypothesis, telomerase-selective replication led to intermediate results in these experiments. Remarkably, all viruses efficiently lysed tumors and induced a therapeutically effective tumor-directed CD8 cytotoxicity. In immunocompetent mice with extended lung metastases burden, treatment of subcutaneous primary tumors with Ad-p53T significantly prolonged survival by inhibition of lung metastases, whereas unselective viral replication resulted in death by liver failure. In summary, the degree of tumor selectivity of viral replication marginally influences antitumoral immune responses, but is a major determinant of antivector immunity and systemic toxicity.

Figure 1

Generation of highly tumor-specific, oncolytic adenoviruses: hTert/p53-dependent transcriptional targeting of adenoviral replication. (a) Illustration showing the concept of hTert/p53-dependent adenoviral replication for highly selective tumor cell targeting. After infection of normal cells, eventual background activity of the E1A controlling hTertgal-promoter is blocked by expression of the transcriptional repressor GAL4-KRAB in response to p53. In tumor cells, p53-dependent GAL4-KRAB response is absent, and full activity of the hTertgal-promoter is released to express E1A to levels required for productive viral replication and cell lysis. (b) Scheme that shows the artificial “hTertgal”-variant of the hTert-promoter. The hTert core promoter is flanked by clusters of GAL4-binding sites. (c) Susceptibility of the hTertgal-promoter for p53-dependent, GAL4-KRAB-mediated transcriptional repression in p53⁺ A549 and p53-mutant Huh7 cells. Luciferase reporter constructs under control of the hTertgal-, or the hTert-promoter, respectively, were co-transfected with prMinRGC-Gal4-KRAB. pSV40-LacZ was co-transfected as internal control. Forty-eight hours after transfection, luciferase signal was determined and normalized by β-galactosidase activity (mean ± SD). The hTertgal-promoter is effectively inhibited by p53-dependent expression of GAL4-KRAB. (d) Schematic drawing of generated adenoviruses differing in the quality of tumor-selective replication (Ad-control: nonselective, Ad-Tert: hTert-selective, and Ad-p53T: hTert/p53-selective).

Figure 2

p53-dependent replication of Ad-p53T in vitro. (a) Activation of a p53-responsive reporter was investigated in murine CMT64 cells with different p53 status. Cells were co-transfected with a prMinRGC-Luc reporter plasmid and pSV40-LacZ as internal control. Forty-eight hours after transfection, luciferase activity was determined and normalized by β-galactosidase activity (mean ± SD). The prMinRGC-promoter is highly active in p53⁺ CMT64 cells compared to human A549 cells as positive and p53-mutated Huh7 cells as negative control. Activation is strongly downregulated in p53-knockdown CMT64-miRp53 cells that express a microRNA against p53. (b) Gal4-KRAB and E1A levels were visualized by western blot analyses of different cell lines 24 hours postinfection (p.i.) with Ad-p53T and Ad-Tert. Equal loading was confirmed by β-actin blot (data not shown). After infection with Ad-p53T, GAL4-KRAB expression and consequent E1A repression were observable in p53⁺ cells but not in p53-dysfunctional cells. (c,d) Oncolytic efficacy was determined in vitro. Cells were infected with Ad-p53T, Ad-Tert, or Ad-control at different MOIs. All used cells are telomerase-active but differ in their p53 status. Seven to eight days following infection, the extent of cell lysis was visualized by crystal violet staining (c) and determined by cell viability assay (d, mean ± SD). (e) To investigate viral DNA replication, cells were infected with different adenoviruses. MOI of 5 was used for A549 and Huh7 cells, whereas MOI of 50 was used for CMT64 and CMT64-miRp53 reflecting the lower permissivity of these cells for adenoviral transduction. Four, twenty-four, forty-eight, and seventy-two hours postinfection, total DNA was isolated, and the adenoviral DNA content was determined by qPCR. (f) Ninety-six hours after infection, viral progeny per infected cell was determined (mean ± SD). (g) Primary human IMR-90 fibroblasts (telomerase-negative/p53-normal) were infected at MOI 0.1. At the indicated time points, total DNA was isolated and the viral DNA content was determined by qPCR (mean ± SD). Together, the results demonstrate that replication of Ad-p53T is regulated in a p53-dependent manner. Due to the complete absence of DNA replication in primary human fibroblasts, Ad-p53T can be regarded as highly tumor-selective, oncolytic adenovirus. *P < 0.05; **P < 0.01. ifu, infection forming units; MOI, multiplicity of infection; NS, not significant.

Figure 3

Highly tumor-selective replication after systemic infection in vivo prevents from liver damage and reduces antiviral cellular immune responses. (a,b) C57Bl/6 mice were treated by i.v. injection of 2.5 × 10⁹ ifu of adenoviruses as indicated. Seventy-two hours after infection, livers were explanted and examined either macroscopically (a) or by hematoxylin and eosin–stained tissue sections (b) for signs of liver damage (example shown as representative of n = 3 mice per group of two independent experiments). (c) Forty-eight hours and seventy-two hours after infection, blood samples were drawn, and serum transaminase activity was determined (mean ± SD; n = 3 per group). (d) Survival was determined following viral treatment as described above (n = 5 per group). (e) 1 × 10⁹ infectious particles were i.v. injected in C57Bl/6 mice. Total DNA was prepared from liver tissue at different time points after infection. Content of viral DNA genomes per total tissue DNA was determined by qPCR (mean ± SD, n = 4 mice in each group). (f) Total RNA was isolated from liver tissue at different time points after infection of i.v.-treated C57Bl/6 mice. Relative change of mRNA levels of innate immune response genes were determined by RT-qPCR (mean ± SD, n = 4 mice in each group). (g) Viral gene expression after i.v. treatment of C57Bl/6 mice was examined of total RNA isolations (mean ± SD, n = 3 mice in each group). (h) Antiviral cellular immune responses were investigated after systemic treatment of C57Bl/6 (for E1A and E1B) and DBA/2 (for hexon) mice with 1 × 10⁹ ifu of adenoviruses as indicated. At day 14 following injection, T cells were isolated from spleen and plated for ELISpot assays. Thirty-six hours after stimulation with major histocompatibility complex class I–restricted peptides, interferon-γ-releasing cells, specific for adenoviral E1A, E1B, or hexon, respectively, were counted (mean ± SD, n = 5 mice in each group). The results demonstrate that effective attenuation of Ad-p53T in normal tissue prevents from life-threatening hepatotoxicity after systemic vector delivery, and leads to decreased triggering of innate immunity and reduced antiviral cellular immune responses. *P < 0.05; **P < 0.01; ***P < 0.0001. ALT, alanine transaminase; AST, aspartate transaminase; ifu, infection forming units; NS, not significant; p.i., postinfection.

Figure 4

Tumor treatment with highly selective, oncolytic viruses allows for effective tumor lysis but leads to reduced viral burden in nontarget tissue. Subcutaneously grown CMT64-miRp53 tumors on C57Bl/6 mice were treated twice on subsequent days by intratumoral injection of 3 × 10⁹ ifu of indicated adenoviruses. At day 3 following initial treatment, tumors were isolated and histologically examined by bright field microscopy (complete overview is shown in a, magnification of the lytic centers is presented in b). (c) DAPI-stained tissue slices were investigated by fluorescence microscopy to detect lytic regions (green cells represent living tumor cells because CMT64-miRp53 cells are also transgenic for GFP; blue color: DAPI-stained nuclei. One slide is shown as representative example of n = 5 mice per group). (d,e) At 6, 24, 48, and 72 hours following second treatment, tumors and livers were harvested and DNA was prepared. Viral DNA replication in tumor (d) and liver (e) was quantified by qPCR (mean ± SD; n = 4 mice per group per time point). Together, the data show that all used viruses sufficiently lysed tumors following intratumoral application accompanied by a massive lymphocyte infiltration at the locus of oncolytic inflammation. Though comparable levels of viral replication could be observed in the tumors, the virus burden in nontarget tissue (liver) correlated well with tumor selectivity of the investigated viruses. *P < 0.05; **P < 0.01. ifu, infection forming units; NS, not significant; p.i., postinfection.

Figure 5

Oncolytic tumor treatment with highly selective adenoviruses drastically reduces antiviral immune responses, whereas antitumoral response was affected to a minor extent. (a) Subcutaneously grown CMT64-miRp53 tumors were treated twice by intratumoral injection with 3 × 10⁹ ifu of the indicated adenoviruses. After 14 days, T cells were isolated from spleens to investigate antigen-specific immune responses by enzyme-linked immunosorbent assay. Plated splenocytes were stimulated with major histocompatibility complex (MHC) class I–restricted peptides for E1A or E1B as viral antigens, or GFP as tumor-associated antigen. After 48 hours, antigen-specific interferon-γ-release was determined in the supernatants of stimulated cells (mean ± SD; n = 5 mice in each group). (b) An in vivo cytotoxic T lymphocyte assay was performed at day 7 following intratumoral treatment of s.c. CMT64-miRp53 tumors. Donor splenocytes were stained with CFSE and different MHC class I–restricted peptides. LacZ was used as control peptide (left peak), and E1A or GFP was used as virus-specific or tumor-specific peptide, respectively (right peak). Splenocytes were then i.v. injected into treated recipient mice. After 18 hours, splenocytes of recipient mice were prepared and analyzed by FACS to determine the peptide-specific cytotoxicity (mean ± SD; n = 5 mice in each group). The results demonstrate that stringent tumor selectivity of adenoviral replication strongly reduces antiviral immune responses. In contrast, the results suggest that tumor selectivity has only limited consequences on effective provocation of antitumoral immunity. *P < 0.05; **P < 0.01. CFSE, carboxyfluorescein succinimidyl esterl; GFP, green fluorescent protein; ifu, infection forming units; NS, not significant.

Figure 6

Highly tumor selectivity of Ad-p53T leads to a therapeutic benefit after high-dose virotherapy in animals with large burden of metastases. Primary s.c. tumors and lung metastases of CMT64-miRp53 cells were established in syngeneic C57Bl/6 mice and subsequently treated according to the therapeutic scheme in a. (b) At the time of treatment start, primary s.c. tumors were treated twice on two subsequent days by intratumoral injection of 3 × 10⁹ ifu. Therapeutic efficacy of induced systemic antitumoral immunity was investigated on noninfected lung metastases. For this purpose, mice were harvested at day 12 after initial treatment, lungs were inflated with paraformaldehyde (PFA) solution for fixation, and hematoxylin and eosin (H&E)–stained lung sections were microscopically investigated to determine the lung metastases burden (each image is shown as a representative example of n = 3 mice per group). (c) Survival of mice was monitored and determined (***P < 0.0001, log-rank test; n = 7 mice per group). All animals died due to lung metastases. However, the results show that oncolytic infection of the primary tumor led to significantly prolonged survival. (d) Therapeutic scheme of a high-dose tumor treatment model with extended metastases burden: an increased dose of CMT64-miRp53 cells were injected i.v. to induce lung metastases earlier as compared to the previous scheme. Subcutaneously grown tumors (induced at day 12) were threefold i.t. infected with high doses of infectious particles (1 × 10¹⁰ ifu). (e) Survival of mice was monitored after treatment according to the scheme in d (***P < 0.0001, log-rank test; n = 8 mice per group). (f) Explanted livers at day 5 after initial treatment (one liver as representative example of n = 4 mice per group) were macroscopically investigated for signs of liver injury. (g) At day 15, mice were harvested, lungs were inflated with PFA solution for fixation purposes and were examined macroscopically (each image is shown as a representative example of n = 3 mice per group). Extent of lung metastases burden was investigated on H&E-stained lung sections. The results show that viral infection of the primary tumor, irrespective of the degree of selectivity, induces systemic antitumoral immunity that is therapeutically effective against distant, noninfected lung metastases. The results obtained from a model that stringently addresses dose-related toxicity demonstrate that highly tumor-selective adenoviruses can be applied at increased doses to achieve therapeutic efficacy without lethal hepatotoxicity. ifu, infection forming units.