Prime-Boost Immunization Eliminates Metastatic Colorectal Cancer by Producing High-Avidity Effector CD8⁺ T Cells

Colorectal cancer is the fourth most commonly diagnosed cancer and the second leading cause of cancer-related deaths in the United States (1). Current standard of care consists of surgical removal of primary tumor and adjuvant chemotherapy to treat metastatic disease. In the context of the low efficacy and high toxicity of existing adjuvant therapies, immunotherapies, particularly vaccines, for colon cancer are an emerging alternative reflecting their potential specificity and resulting low toxicity (2). However, cancer vaccine platforms with utility in patients are limited. Protein and peptide vaccines have been largely ineffective, although next-generation peptide vaccines may prove more efficacious (3). In contrast, although dendritic cell vaccines are generally immunogenic, including the U.S. Food and Drug Administration–approved Sipuleucel-T vaccine for prostate cancer (4), those treatments require expensive, personalized vaccines, limiting their use to a small subset of patients. Recombinant viral vectors offer great promise and have been approved as vaccine vectors for infectious diseases (5). Unfortunately, few viral vectors are suitable for humans (typically adenovirus [Ad5] and poxviruses), and these are limited by pre-existing and/or vaccine-induced neutralizing immunity (6). Thus, novel vaccine platforms are needed to optimally treat cancer in humans.

DNA vaccines provide a safe and inexpensive alternative vaccine strategy that is not limited by vector immunity. DNA vaccines have been examined for cancers and infectious diseases, with their primary advantages being stability, safety, and absence of neutralizing immunity, permitting multiple administrations (7, 8). Indeed, DNA vaccines have been approved for animal use to treat canine melanoma and infectious diseases (9, 10). However, low transfection rates and immunogenicity have limited the usefulness of DNA vaccine monotherapy in humans. These have been partially overcome by electroporation (EP), gene gun, nanoparticle, and ultrasound delivery, as well as by the inclusion of adjuvants (7, 8), but DNA vaccine success has been limited to combinations with other vaccine platforms in heterologous prime-boost regimens (11). Guanylyl cyclase C (GUCY2C) is an immunotherapeutic target in colorectal cancer, reflecting its limited expression in normal tissues and persistent expression in colorectal cancers (12–17). GUCY2C is confined to the apical surfaces of intestinal epithelial cells (18–21) and a subset of hypothalamic neurons (22, 23), areas that are structurally segregated from the systemic immune compartment (12). Further, its expression is maintained throughout colorectal tumorigenesis, and GUCY2C is found in >95% of colorectal cancer metastases (24, 25). Previously, we demonstrated that recombinant replication-deficient Ad5 expressing the extracellular domain of GUCY2C induces protective GUCY2C-specific CD8⁺ T cell responses, without toxicity, in a mouse model of metastatic colorectal cancer (13–15, 26).

In this study, we examined a prime-boost strategy consisting of GUCY2C-expressing DNA and adenoviral vaccines to enhance antitumor efficacy, identifying a strategy composed of sequential immunization with DNA and Ad5 (DNA+Ad5) that induces superior antitumor immunity and prevents disease progression in the majority of animals, without collateral autoimmunity. Moreover, the enhanced antitumor immune responses produced by DNA+Ad5 vaccination reflected the production of high-avidity effector CD8⁺ T cells. In the context of enhanced immune responses, superior antitumor immunity, and lack of toxicity, the DNA+Ad5 vaccination strategy identified in this article is poised for clinical translation to prevent recurrent disease in early-stage colorectal cancer patients.

BALB/c mice were obtained from Charles River. Animal protocols were approved by the Thomas Jefferson University Institutional Animal Care and Use Committee.

Except where indicated, all DNA and Ad5 GUCY2C–specific vaccine constructs used mouse GUCY2C fused at the C terminus to the influenza HA_107–119 epitope (known as site 1 [S1]), which was described previously (14). The Ad5 vaccine composed of mouse GUCY2C-S1 (Ad5–GUCY2C–S1) was described previously (14). The DNA vaccine construct used mouse GUCY2C-S1 cloned into pcDNA-DEST47 (Life Technologies). Plasmids pCCL20 (pEF1-mCCL20), pCCL21 (pCUB-mCCL21), and pMaxGFP were described previously (27). The control plasmid pcDNA 3.1 was from Life Technologies. For i.m. and intradermal (i.d.) EP, 50 μg of plasmid DNA in 50 μl of water was delivered into the gastrocnemius muscle of each leg or two distant skin sites per mouse, respectively. DNA delivery was followed by 10 electric pulses (field strength = 100 V/cm, pulse length = 20 ms, pulse interval = 1 s) delivered by an ECM 830 Square Wave Electroporation System (BTX Harvard Apparatus).

A total of 1 × 10⁸ IFU Ad5–GUCY2C–S1 was administrated to each gastrocnemius muscle by i.m. injection. For DNA vaccinations, the pCCL20 (25 μg) and pCCL21 (25 μg) plasmids in water were combined and injected with EP, followed 12 d later by immunization with 50 μg of GUCY2C-expressing or control DNA via EP. Prime-boost strategies (Ad5+Ad5, Ad5+DNA, DNA+Ad5) were administered with 21 d between vaccinations. For therapeutic studies (Supplemental Fig. 3), control or GUCY2C-expressing DNA vaccinations were administered without chemokines on day 3, and control or GUCY2C Ad5 was administered on day 10. Radiotherapy used a PanTak, 310 kVe x-ray machine to deliver 10 Gy radiation on day 3 to only the chest, protecting other organs with lead shielding. The checkpoint inhibitor Abs anti–CTLA-4 (clone UC10-4F10-11) and anti–PD-L1 (clone 10F.9G2; both from Bio X Cell) were administered i.p. (100 μg each) in PBS on days 1, 4, 7, 11, and 15.

Immune responses to the dominant GUCY2C-specific CD8⁺ T cell epitope GUCY2C_254–262 (26) were measured in splenocytes 14 d after the final immunization.

Multiscreen filtration plates (Millipore) were coated with 10 μg/ml anti-mouse IFN-γ capture Ab (BD Biosciences) overnight. Splenocytes (1 × 10⁶ per well) were stimulated with DMSO or 10 μg/ml GUCY2C_254–262 peptide for 24 h. Spots were developed with 2 μg/ml biotinylated anti–IFN-γ detection Ab (BD Biosciences) and 2 μg/ml alkaline phosphatase–conjugated streptavidin, followed by NBT/BCIP substrate (both from Pierce). Spot-forming cells were enumerated using computer-assisted video imaging analysis (ImmunoSpot v5; Cellular Technology).

Splenocytes were stimulated for 6 h with DMSO or GUCY2C_254–262 and Protein Transport Inhibitor Cocktail (eBioscience). Cells were stained with a LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Invitrogen) and anti-CD8α–PerCP–Cy5.5 (clone 53-6.7; BD Pharmingen), and intracellular cytokine staining was performed using a BD Cytofix/Cytoperm Kit (BD Biosciences) and the following Abs: anti-IFN-γ–allophycocyanin–Cy7 (XMG1.2), anti-TNF-α–PE–Cy7 (MP6-XT22) (both from BD Biosciences), and anti-MIP1α–PE (clone 39624; R&D Systems). Cells were fixed in 2% paraformaldehyde and analyzed on a BD LSR II flow cytometer. Analyses were performed using FlowJo software (TreeStar).

A total of 1 × 10⁶ splenocytes was plated per well in ELISPOT plates with various concentrations of GUCY2C_254–262 peptide for 24 h. ELISPOT plates were developed as described above.

Mice were immunized as described above and challenged i.v. with 5 × 10⁵ CT26 cells expressing mouse GUCY2C (15) in PBS 6 d later, and survival was measured longitudinally. Intravenous administration of CT26 cells is a well-established model of metastatic colorectal cancer, preferentially forming metastases in the lungs (13–15, 26, 28, 29).

Tissues were collected from mice 14 or 180 d after no treatment or treatment with GUCY2C-expressing DNA or DNA+Ad5 prime-boost vaccination. Tissues were fixed in formalin and embedded in paraffin. Sections were stained with H&E and scored for toxicity by a blinded pathologist (L.B.-B.). Scoring criteria are described in Supplemental Table I. Dextran sulfate sodium (DSS) studies were performed as previously described, with some modification for the BALB/c mouse model (13). Three weeks after completing control or GUCY2C DNA+Ad5 immunization regimens, DSS (Sigma-Aldrich) was administered ad libitum in the drinking water at 5% (w/v) for 8 d, followed by normal drinking water for the remainder of the experiment. Mice were weighed daily throughout the experiment, and three representative mice were euthanized on day 9 following DSS initiation for histopathologic analysis (by L.B.-B.). Negative area under the curve peaks were calculated on weight curves using GraphPad Prism v6, and absolute values were plotted.

All analyses used GraphPad Prism Software v6. T cell enumeration by ELISPOT was analyzed by one-way ANOVA (multiple comparisons) or the t test (single comparisons). Cytokine polyfunctionality used two-way ANOVA. TCR avidity measurements were analyzed by nonlinear regression [log(agonist) versus normalized response], and comparisons were made using the exact sum-of-squares F test (GraphPad v6). Nonlinear regression plots depict the regression results (line) with 95% confidence intervals (cloud) computed from results calculated from multiple independent experiments. Survival comparisons were analyzed by the Mantel–Cox log-rank test. The relationship between each immunoassay and tumor immunity was analyzed by F test of linear regression analysis. Autoimmunity comparisons used one-way ANOVA, and DSS comparisons used the t test.

We previously used i.d. administration of DNA vaccines combined with CCL20 and CCL21 chemokine adjuvanation codelivered by in vivo EP to enhance melanoma-specific immune response and local immunotherapeutic efficacy targeting dermal melanoma tumors (27). In this study, we examined DNA vaccine immunogenicity following i.m. administration to treat systemic colorectal cancer metastases characterized by dissemination to lung, liver, and brain. As expected, Ag expression assayed by GFP fluorescence was superior following i.m. administration compared with i.d. administration, in a time-dependent fashion (Fig. 1A, 1B). Increased Ag expression following i.m. administration suggests that this route may be favorable for DNA vaccine-induced immune responses. Moreover, DNA vaccine administration in the absence of EP eliminated Ag expression (Fig. 1C), consistent with literature suggesting a 100–1000-fold increase in transgene expression with EP (30, 31). Thus, transgene expression was optimized by i.m., rather than i.d., administration of DNA plasmid in the context of EP.

In the context of robust transgene expression following i.m., rather than i.d., DNA delivery (Fig. 1), we compared the immunogenicity of i.m. and i.d. vaccinations (Fig. 2A). The i.m. DNA vaccination with the syngeneic mouse colorectal cancer Ag GUCY2C produced systemic GUCY2C-specific CD8⁺ T cell responses, as quantified by ELISPOT, that were >10-fold greater than i.d. immunization (Fig. 2A), consistent with greater transgene expression in muscle (Fig. 1). However, adenoviral vector immunization against GUCY2C (Ad5) produced GUCY2C-specific CD8⁺ T cell responses that were >10-fold greater than those produced by i.m. DNA vaccination (Fig. 2B). Moreover, Ad5 vaccination produced an 18-fold increase in the median survival time of mice with GUCY2C-expressing colorectal cancer metastases in lung compared with DNA vaccination (Fig. 2C). Moreover, repeated administrations of DNA vaccination failed to improve GUCY2C-specific CD8⁺ T cell responses (Supplemental Fig. 1). Taken together, i.m. DNA vaccination produces GUCY2C-specific CD8⁺ T cell responses and antitumor responses, but those are not comparable to Ad5 vaccination, suggesting that DNA vaccination alone is not a viable strategy for GUCY2C-directed tumor immunotherapy.

Although GUCY2C-specific DNA vaccination alone was poorly immunogenic, we hypothesized that it could be combined with Ad5 in prime-boost regimens to improve antitumor immunity. To maximize GUCY2C-specific immune responses and antitumor immunity, different heterologous prime-boost strategies were explored. Combining GUCY2C-specific DNA and Ad5 vaccinations in either order (DNA+Ad5 or Ad5+DNA) increased the survival of mice with GUCY2C-expressing CT26 colorectal cancer metastases (Fig. 3A). Although both heterologous prime-boost strategies improved median survival time compared with Ad5 alone (43.5 d), DNA+Ad5 was superior to Ad5+DNA (111 versus 77.5 d, p < 0.05) and produced GUCY2C-specific memory responses in surviving mice (Supplemental Fig. 2). A truncated version of the DNA+Ad5 regimen (no chemokines; 7 d between boosting) also extended survival in a therapeutic setting when combined with local radiotherapy (Supplemental Fig. 3). Examination of GUCY2C-specific CD8⁺ T cell responses, as quantified by T cell number (ELISPOT; Fig. 3B) and polyfunctional cytokine production (Fig. 3C), revealed modest increases with DNA+Ad5 immunization but no increase with Ad5+DNA immunization. However, the functional TCR avidity of GUCY2C-specific CD8⁺ T cell responses (Fig. 3D) induced by DNA+Ad5 (0.30 μg/ml) and Ad5+DNA (0.82 μg/ml) were substantially increased compared with Ad5 immunization alone (1.92 μg/ml). Importantly, GUCY2C-specific CD8⁺ T cell numbers (Fig. 3B), polyfunctional cytokine responses (Fig. 3C), TCR avidity (Fig. 3D), and antitumor efficacy (Fig. 3A) were not improved by homologous Ad5 prime-boost (Ad5+Ad5) compared with Ad5 alone, confirming the role of heterologous prime-boosting in GUCY2C-specific CD8⁺ T cell enhancement.

In the context of little or no enhancement of GUCY2C-specific CD8⁺ T cell number and effector cytokine responses by heterologous prime-boost, yet substantial increases in TCR avidity and antitumor immunity, we hypothesized that TCR avidity enhancement underlies the improved antitumor efficacy of GUCY2C-specific prime-boost regimens. To test this hypothesis by manipulating TCR avidity while preserving the magnitude and cytokine profile of GUCY2C-specific CD8⁺ T cells, we created an Ad5 vaccine expressing only the dominant preprocessed GUCY2C-specific CD8⁺ T cell epitope (Ad5-GUCY2C_254–262). We showed previously that viral vectors expressing processed MHC class I epitopes result in high surface peptide–MHC density, producing low-avidity T cell responses (32). Indeed, the quantity (Fig. 4A) and effector cytokine profiles (Fig. 4B) of Ad5-GUCY2C_254–262–induced CD8⁺ T cell responses were comparable to DNA+Ad5 immunization, whereas the TCR avidity of Ad5-GUCY2C_254–262–induced T cells was ∼20-fold lower than DNA+Ad5 (5.88 versus 0.30 μg/ml, p < 0.0001). In turn, Ad5-GUCY2C_254–262 produced only an 11-d improvement in median survival compared with control immunization, whereas the majority of DNA+Ad5-immunized mice survived for >200 d (Fig. 4D). When comparing antitumor efficacy with T cell quantity (Fig. 5A), T cell cytokine polyfunctionality (Fig. 5B), and T cell avidity (Fig. 5C) across all tested vaccination combinations, only T cell avidity predicted tumor outcomes, revealing the previously unrecognized mechanism of functional T cell avidity enhancement mediating synergy of DNA+Ad5 prime-boost vaccinations.

Enhancing TCR avidity during adoptive T cell therapy for melanoma enhances antitumor immunity and autoimmunity, suggesting a strong relationship between efficacy and toxicity in T cell immunotherapy (33). However, immunologic compartmentalization between systemic and mucosal immune systems prevents autoimmunity in mice receiving Ad5-based GUCY2C vaccines, suggesting that the increased TCR avidity and antitumor immunity observed with DNA+Ad5 prime-boost vaccination would not come at the expense of increased autoimmunity (13–15). Indeed, mice immunized against GUCY2C with DNA or the DNA+Ad5 prime-boost strategy were free of acute (2 wk after final immunization) or chronic (6 mo after final immunization) autoimmune toxicity (Fig. 6A, 6B) in tissues associated with GUCY2C expression (small and large intestines, cecum, salivary gland, and hypothalamus), as well as tissues devoid of GUCY2C (heart, lung, liver, kidneys, and stomach). Moreover, oral administration of DSS, an established model of experimental colitis, resulted in equivalent clinical and histopathologic colitis and recovery in control and GUCY2C DNA+Ad5-immunized mice (Fig. 6C–E). Thus, GUCY2C DNA+Ad5 immunization did not exacerbate experimentally induced colitis, allowing clinical recovery. Together, DNA+Ad5 prime-boost immunization enhances TCR avidity, maximizing antitumor immunity without concomitant autoimmunity.

Despite emerging oncoimmunotherapeutics (34), cancer causes ∼8 million deaths worldwide each year, and the total economic burden of cancer, exceeding 1 trillion dollars annually, is higher than that of any other disease (35, 36), which highlights the need for effective, safe, and inexpensive strategies for its treatment in developed and developing regions around the world. In that context, vaccines have had one of the largest impacts on public health in human history, eradicating several infectious diseases and nearly eliminating morbidity and mortality associated with many others (37). However, cancer vaccines have been largely unsuccessful, reflecting, in part, their poor immunogenicity and a lack of biomarkers predictive of patient outcomes. In this article, we show that a strategy using heterologous prime-boost with GUCY2C-targeted DNA and Ad5 vaccines cures ∼50% of mice with metastatic colorectal cancer. More importantly, the efficacy of this strategy reflected increased TCR functional avidity compared with Ad5 vaccination alone. Indeed, TCR functional avidity, but not CD8⁺ T cell number, cytokine, or polyfunctional cytokine responses, was highly correlated (p < 0.0001) with outcomes, and experimental reduction of TCR functional avidity, while maintaining effector T cell numbers and cytokine profiles, eliminated vaccine efficacy.

These results contrast with previous studies, primarily in infectious disease, which suggest that polyfunctional cytokine responses are the key determinant in outcomes (38). Indeed, analysis of untreated patients with poor HIV control (progressors) and those with HIV control (nonprogressors) demonstrated that nonprogressors possessed higher functionality than progressors (39). More recently, suppression of HIV replication, through killing of HIV-infected targets by high-avidity CD8⁺ T cells (40), was found to be the exclusive determinant of HIV viremic control in progressors (41), suggesting that CD8⁺ T cell avidity, rather than polyfunctionality, may also underlie HIV control. Similarly, experimental manipulation of TCR avidity using adoptive transfer of T cells genetically engineered to express TCRs of varying affinities to the melanoma Ag gp100 demonstrated that antitumor efficacy was determined by TCR affinity (33). Of note, anti-melanoma efficacy and melanocyte-targeted autoimmunity were tightly coupled, suggesting that there may not be a window of TCR affinity in which selective tumor tissues, but not self-tissues, may be targeted (33). In that context, anatomical localization of GUCY2C to intestinal apical membranes (18–20) and restricted immune trafficking to intestinal mucosa by T cells induced by systemic vaccination (42–44) decouples antitumor activity and autoimmunity. Immunization with Ad5 alone produces antitumor immunity targeting metastatic colorectal cancer in lungs and liver, without concomitant autoimmunity (13–15). Similarly, TCR avidity and antitumor efficacy were improved ∼10-fold following DNA+Ad5 prime-boost, without acute (2 wk) or chronic autoimmunity (∼6 mo; Fig. 6). Thus, although TCR avidity may not be exploited to decouple antitumor immunity and autoimmunity (33), Ag and immune compartmentalization may be exploited to produce safe and effective immunotherapeutics targeting GUCY2C, as well as other Ags selectively expressed in mucosa and by metastatic cancer (cancer mucosa Ags) (12–15).

Although DNA+Ad5 produced superior antitumor efficacy through TCR avidity enhancement, the mechanism(s) mediating TCR avidity enhancement has not been defined and likely reflects selective enrichment of individual T cell clones possessing high TCR avidity and/or T cell–intrinsic TCR avidity enhancement. DNA vaccination may preferentially prime T cell clones of high TCR avidity by eliciting direct presentation and indirect (cross) presentation of encoded Ags (45, 46), which can be expanded by subsequent Ad5 boosts. The relative contribution of direct and indirect presentation of GUCY2C to the induction of CD8⁺ T cell responses following DNA or Ad5 immunization has not been explored, preventing predictions of their contributions to TCR avidity enhancement. Studies using cell/tissue-specific promoters, such as the CD11c promoter to limit expression to professional APCs (direct presentation) or the human desmin gene 5′ regulatory region to limit expression to myocytes (indirect presentation), may determine the role of direct and indirect Ag presentation in TCR avidity enhancement (47). Moreover, spectratype analysis (48) or next-generation sequencing (49) may reveal shifts in Ag-specific T cell populations following DNA+Ad5 immunizations, whereas TCR-transgenic models possessing a fixed TCR repertoire may prevent TCR avidity enhancement by DNA+Ad5 immunization. In that context, a recent study using OVA-specific TCR-transgenic mice suggests that TCR avidity enhancement following DNA+vaccinia prime-boost is independent of clonal selection or proximal TCR signaling and instead depends on a decreased TCR-activation threshold via a T cell–intrinsic MyD88 pathway (50). Although TCR avidity enhancement was dependent on MyD88 signaling, that study (50) used Rag^+/+ TCR-transgenic mice, permitting thymic recombination of endogenous TCR alleles, expanding the TCR repertoire, and permitting a contribution by T cell clone selection to TCR avidity enhancement. Defining the mechanism(s) mediating TCR avidity enhancement by DNA+Ad5 immunization may lead to the development of targeted adjuvants that enhance TCR avidity without requiring heterologous prime-boost strategies, reducing vaccination complexity and potentially decreasing the time required to generate high-avidity responses, a critical consideration in the development of therapeutic vaccines in cancer and infectious diseases.

TCR avidity enhancement following prime-boost is not limited to immunization combinations composed of DNA and Ad5. DNA+vaccinia (50), as well as fowlpox+vaccinia (51), prime-boost regimens similarly enhance TCR avidity. Moreover, we showed that heterologous viral vector immunization regimens targeting GUCY2C produce increasing antitumor efficacy with each successive boost (15). Together, they suggest that TCR avidity enhancement may be a generalizable outcome of heterologous prime-boost immunizations, independent of Ag target and vaccine design. In the context of ongoing clinical studies of Ad5-GUCY2C (52), GUCY2C-specific TCR avidity and clinical efficacy could be enhanced, without autoimmunity, using heterologous prime-boost strategies in patients with GUCY2C-expressing cancers, including colorectal and a subset of gastric, esophageal, and pancreatic cancers.

We thank Dr. Gilbert Kim, Dr. Egeria Lin, and Amanda Aing for assistance with processing the tissues collected for autoimmunity scoring and Dr. Peng Li for assistance with the evaluation of autoimmunity.

S.A.W. was the Chair of the Data Safety Monitoring Board for the Congestive Heart Failure Cardiopoietic Regenerative Therapy (CHART-1) Trial sponsored by Cardio3 BioSciences and the Chair (uncompensated) of the Scientific Advisory Board to Targeted Diagnostics and Therapeutics, which provided research funding that supported this work in part, and has a license to commercialize inventions related to this work. The other authors have no financial conflicts of interest.