Snook et al. Journal for ImmunoTherapy of 19) 7:104RESEARCH ARTICLEOpen AccessSplit tolerance permits safe Ad5-GUCY2CPADRE vaccine-induced T-cell responses incolon cancer patientsAdam E. Snook1*† , Trevor R. Baybutt1, Bo Xiang1, Tara S. Abraham1, John C. Flickinger Jr1, Terry Hyslop2,Tingting Zhan1, Walter K. Kraft1, Takami Sato3 and Scott A. Waldman1†AbstractBackground: The colorectal cancer antigen GUCY2C exhibits unique split tolerance, evoking antigen-specific CD8 ,but not CD4 , T-cell responses that deliver anti-tumor immunity without autoimmunity in mice. Here, the cancervaccine Ad5-GUCY2C-PADRE was evaluated in a first-in-man phase I clinical study of patients with early-stagecolorectal cancer to assess its safety and immunological efficacy.Methods: Ten patients with surgically-resected stage I or stage II (pN0) colon cancer received a singleintramuscular injection of 1011 viral particles (vp) of Ad5-GUCY2C-PADRE. Safety assessment and immunomonitoringwere carried out for 6 months following immunization. This trial employed continual monitoring of both efficacyand toxicity of subjects as joint primary outcomes.Results: All patients receiving Ad5-GUCY2C-PADRE completed the study and none developed adverse eventsgreater than grade 1. Antibody responses to GUCY2C were detected in 10% of patients, while 40% exhibitedGUCY2C-specific T-cell responses. GUCY2C-specific responses were exclusively CD8 cytotoxic T cells, mimickingpre-clinical studies in mice in which GUCY2C-specific CD4 T cells are eliminated by self-tolerance, while CD8 Tcells escape tolerance and mediate antitumor immunity. Moreover, pre-existing neutralizing antibodies (NAbs) tothe Ad5 vector were associated with poor vaccine-induced responses, suggesting that Ad5 NAbs oppose GUCY2Cimmune responses to the vaccine in patients and supported by mouse studies.Conclusions: Split tolerance to GUCY2C in cancer patients can be exploited to safely generate antigen-specificcytotoxic CD8 , but not autoimmune CD4 , T cells by Ad5-GUCY2C-PADRE in the absence of pre-existing NAbs tothe viral vector.Trial registration: This trial (NCT01972737) was registered at on October 30th, 2013. words: Colorectal cancer, GUCY2C, Guanylyl cyclase C, VaccineIntroductionWhile checkpoint inhibitor and CAR-T cell therapieshave initiated a paradigm shift in the management ofsome cancers [1], there remains an unmet need forimproved treatment of colorectal cancer (CRC), the 4thleading cause of cancer and 2nd leading cause of cancer* Correspondence: [email protected]†A.E.S. and S.A.W. contributed equally to this work.1Department of Pharmacology and Experimental Therapeutics, ThomasJefferson University, 1020 Locust Street, JAH 368, Philadelphia, PA 19107, USAFull list of author information is available at the end of the articlemortality worldwide [2]. At the time of initial diagnosis,about two-thirds of CRC patients undergo surgical resection with curative intent, but 30–50% of these patients experience recurrence and die of their disease. Adjuvantchemotherapy only marginally improves survival in stageIII disease, and has no benefit in pN0 (stage I and II;lymph node negative) patients. Moreover, checkpointinhibitors such as nivolumab and pembrolizumab are effective only in microsatellite instable (MSI) CRC [3],reflecting their high density of mutation-associated neoantigens targeted by effector T cells [4, 5]. In contrast, The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (, which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication o/1.0/) applies to the data made available in this article, unless otherwise stated.

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104checkpoint inhibitors are ineffective against microsatellitestable (MSS) CRC which accounts for 85% of cases. Theseconsiderations underscore the clinical opportunity fornovel therapeutics, particularly immunotherapies, to prevent disease recurrence and improve survival in patientswith stage I-III colorectal cancer. In that context, immunotherapeutic paradigms in cancer may be most effectivein the prevention of recurrent metastases in patients withminimal residual disease [6]. Thus, emerging tumor vaccine paradigms that promote durable antitumor efficacywithout autoimmunity, represent a unique opportunity toimprove colorectal cancer outcomes.Guanylyl cyclase C (GUCY2C), a membrane-spanningreceptor synthesizing the second messenger cyclic GMP(cGMP), is selectively expressed by intestinal epithelialcells and a subset of neurons [7–10] and near-universallyoverexpressed in colorectal cancer. Indeed, GUCY2C hasbeen detected in 1000 CRC specimens, but not inextra-gastrointestinal parenchymal tissues or tumors [7,11, 12]. Moreover, within intestinal epithelial cells,GUCY2C is localized in apical brush border membranes,placing it outside the mucosal barrier [13]. The anatomicaland functional compartmentalization of GUCY2C hasbeen confirmed by RT-qPCR [13, 14], radioligand imagingand biodistribution [13], and immunotoxin [15], vaccine[16–20], and CAR-T cell [21, 22] treatment. Together, intestinal compartmentalization and near-universal expression by primary and recurrent colorectal cancer [14, 23,24], establish GUCY2C as an attractive target for immunotherapeutic prevention of colorectal cancer recurrence.Adenovirus (Ad5)-delivered GUCY2C-based vaccinesinduce antigen-specific CD8 T-cell and antibody responses in syngeneic mice [16–20, 25–27]. Mediated byCD8 T-cells rather than antibodies, these immuneresponses target colorectal cancer metastases in lungand liver in mouse models of prophylaxis and therapy[16, 18–20, 26, 27]. Immunization with GUCY2C-basedvaccines produces memory CD8 T-cell responses thatprovide durable protection against metastases in mice,modeling vaccination in CRC patients with minimumresidual disease [16–18]. Importantly, GUCY2C vaccination provides therapeutic efficacy in the absence ofautoimmunity [16–20].Beyond the safety and efficacy of GUCY2C vaccination,preclinical studies in mice demonstrated that self-tolerance,which limits the production of immune responses to selfproteins and subsequent autoimmunity, reduced vaccineinduced CD8 T-cell responses to GUCY2C, and eliminated GUCY2C-specific antibody and CD4 T-cell responses [18–20]. However, self-tolerance in mice didnot directly impact GUCY2C-specific CD8 T cells andantibody-producing B cells [18]. Rather, self-toleranceeliminated GUCY2C-specific CD4 T cells, which servean essential “helper” role in the production of CD8 Page 2 of 12T-cell and B-cell responses [18]. Thus, self-tolerance isuniquely “split” - eliminating GUCY2C-specific CD4 Tcells, while preserving functional pools of CD8 T and Bcells which can be activated with GUCY2C-independentCD4 T-cell help [18]. Indeed, inclusion of CD4 T-cellepitopes from influenza hemagglutinin (S1) or the synthetic CD4 T-cell epitope PADRE fully activatedGUCY2C-specific CD8 T and B cells, improving vaccineantitumor efficacy 750%, without autoimmunity [17, 18].Here, we translate observations of split tolerance toGUCY2C from animal models to humans in a phase Iclinical trial establishing selective CD4 T-cell tolerance asa key mechanism influencing cancer vaccine responses inhumans, and which may be leveraged to elicit antitumorimmunity without autoimmune toxicity.Materials and methodsStudy design and treatmentThis was a phase I study ( identifierNCT01972737) of stage I or II (pN0) colon cancerwithin 3 years of surgery and no clinical or laboratoryevidence of local or systemic recurrence. The studyprotocol and all amendments were approved by theThomas Jefferson University Institutional Review Board(IRB) and Institutional Biosafety Committee (IBC). Thestudy was conducted in accordance with the protocol,Good Clinical Practice guidelines, the ethical principlesoutlined in the Declaration of Helsinki, and the NIHGuidelines for Research Involving Recombinant or Synthetic Nucleic Acid Molecules. All patients providedwritten informed consent to participate.Ad5 vectors have well-established potency to induceantigen-specific immune responses in animal modelsand humans, as well as a long and impressive safety record. PADRE is a CD4 T-cell epitope that is active inthe context of most human HLA molecules [28] and isrequired to optimally induce GUCY2C-specific immuneresponses and efficacy in animal models [17, 18].Codon-optimized cDNA encoding human GUCY2Cresidues 1–429 with a C-terminal PADRE epitope(Fig. 1a) was cloned into the E1 region of pAd/CMV/V5 (Life Technologies, Carlsbad, CA) containing E1and E3-deleted human serotype 5 adenovirus (Ad5; Fig.1b). Ad5-GUCY2C-PADRE vector used for these studies was produced under GMP conditions at the BaylorCollege of Medicine in the Cell and Gene Therapy Vector Development Laboratory.In vitro GUCY2C-expression experiments (dose-response and time-course) were carried out in A549(ATCC, Manassas, VA) cells. Virus was added to the cultures at the indicated doses and culture supernatantswere collected at the indicated time points. RelativeGUCY2C levels were quantified in supernatants by western blot using MS7 mouse anti-GUCY2C monoclonal

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104Page 3 of 12Fig. 1 Ad5-GUCY2C-PADRE design and antigen expression. a GUCY2C is a membrane-spanning enzyme possessing an extracellular ligand-bindingdomain and intracellular cGMP-producing catalytic domain. The extracellular domain (ECD) of GUCY2C was employed in the vaccine design andincluded the PADRE epitope on its COOH-terminus. b GUCY2CECD-PADRE was inserted into the E1 region of E1/E3-deleted Ad5. c A549 cells weretransduced in duplicate with Ad5-GUCY2C-PADRE at an MOI of 10 to 10,000 for 48 h. GUCY2CECD-PADRE expression was quantified in supernatants byimmunoblot analysis. Densitometry was employed to quantify expression. d A549 cells were transduced in duplicate with Ad5-GUCY2C-PADRE at anMOI of 10,000 for 24–96 h and GUCY2CECD-PADRE expression was quantified in supernatants by immunoblot analysis. Densitometry (arbitrary units)was employed to quantify expression. Blots in c and d are representative of two experiments and graphs indicate the mean SD from 2 experimentsantibody [9, 21, 26] and HRP-conjugated goat anti-mousesecondary antibody (Jackson Immuno, West Grove, PA).Patients received a single intramuscular injectionof 1011 viral particles (vp) Ad5-GUCY2C-PADRE.Safety assessment was carried out in-clinic for 30min and via phone call for 1 week. Patients returnedfor in-clinic safety assessment and immunomonitoring blood collection on days 30, 90, and 180 daysafter immunization. This trial employed continualmonitoring of both efficacy and toxicity of subjectsas joint primary outcomes. All patients completedthe study.

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104Human subject ImmunomonitoringVenous blood was collected into BD Vacutainer GlassSerum Tubes for serum collection and BD Vacutainer CPT Mononuclear Cell Preparation Tubes with SodiumCitrate for peripheral blood mononuclear cell (PBMC)isolation. For serum collection, blood samples were incubated 30 min at 37 C, centrifuged, and supernatantswere transferred to cryovials and stored at 20 C.PBMCs were collected by centrifugation according tomanufacturer’s instructions. PBMCs were then washedwith 1x CTL-WASH buffer (Cellular Technology Limited, Cleveland, OH), counted using a Muse Cell Analyzer(Millipore, Darmstadt, Germany), and cryopreserved inCTL-Cryo ABC freezing medium (Cellular TechnologyLimited) according to the manufacturer’s instructions.Cryovials were frozen at 80 C overnight in a CoolCell LX Alcohol-free Cryopreservation Container (Biocision,Mill Valley, CA) before long-term storage in LN2.Page 4 of 12Technologies, Berlin, Germany); 2 and 10 μg/mL humanGUCY2C Peptide Mix (custom 15 mer/11 aa overlap library of human GUCY2C1–429, JPT Peptide Technologies); 1 μg/mL PADRE (BAP-251, EMC Microcollections,Tübingen, Germany). Plates were incubated overnight at37 C/5% CO2 followed by development according to themanufacturer’s instructions. Spots were quantified usingan ImmunoSpot S6 Universal Analyzer (Cellular Technology Limited). For CD4/CD8 depletion experiments,CD4 or CD8 T cells were negatively selected fromthawed PBMC samples by magnetic-activated cell sorting (MACS; Miltenyi Biotec, Bergisch Gladbach,Germany) prior to ELISpot analysis. Small aliquots ofPBMCs or CD4- or CD8-depleted PBMCs were stainedwith anti-CD4-PerCP (clone S3.5, Invitrogen, Carlsbad,CA) and anti-CD8-Alexa Fluor 700 (clone 3B5, Invitrogen) and analyzed on a BD LSR II flow cytometer.Analyses were performed using FlowJo software (FlowJo,LLC, Ashland, CA).GUCY2C-specific antibody quantification by ELISAHexahistidine-tagged human GUCY2CECD (amino acids 1–429) protein was produced in suspension HEK293 cells andpurified to 90% purity by immobilized metal affinity chromatography (GenScript, Piscataway, NJ). Nunc-ImmunoPolySorp plates (Nunc, Roskilde, Denmark) were coated for4 h at room temperature with human GUCY2CECD proteinat 10 μg/mL in coating buffer (Immunochemistry Technologies, Bloomington, MN). Plates were washed and freebinding sites were blocked with SynBlock (Immunochemistry Technologies) overnight at room temperature. Serumsamples were thawed and titrated in coated, washed platesfrom 1/20 to 1/2560 in 10% nonfat dry milk and incubated2 h at room temperature. Plates were washed and boundhuman antibody was detected with HRP-conjugated goatanti-human antibody (Jackson Immuno) for 2 h at roomtemperature. Following a final wash, Turbo TMB substrate(ThermoFisher Scientific Pierce, Waltham, MA) was addedand the plates incubated for color development, followedby determination of optical absorbance (POLARstarOptima plate reader, BMG Labtech, Cary, NC).Ad5 NAb titer quantificationMouse or human serum samples were heat-inactivatedfor 1 h at 56 C and then titered in duplicate from 1/20to 1/10240 in black 96-well tissue culture plates, 50 μL/wellfinal volume. 108 vp of Ad5-CMV-eGFP virus (Vector Development Lab, Baylor College of Medicine) was added toeach well of titered serum (50 μL/well of 2x109 vp/mL).105 A549 cells (ATCC) were then added to each well(100 μL of 106 cells/mL). For quantification of %neutralization by serum samples, controls included viruswith cells alone (0% neutralization) and cells alone (100%neutralization). Plates incubated 41 h at 37 C/5% CO2 before quantification of eGFP fluorescence (490 nm excitation, 510 nm emission) using a POLARstar Optima platereader (BMG Labtech). Sample fluorescence was normalized to controls described above and titers were determined using nonlinear regression as the serum dilutionproducing 50% neutralization (Prism v7, GraphPad Software, La Jolla, CA).Mouse Ad5 NAb studiesT-cell response quantification by ELISpotHuman IFNγ SC Enzymatic ELISpot plates (Cellular Technology Limited) were coated according to the manufacturer’s instructions. PBMC samples were thawed usingCTL Anti-Aggregate Wash (Cellular Technology Limited)according to the manufacturer’s instructions and 5 105 viable cells/well were plated in ELISpot plates in serum-freeCTL-Test Medium (Cellular Technology Limited) without a rest period. Antigens were prepared and added toPBMCs at the indicated final concentrations inCTL-Test Medium (Cellular Technology Limited)with 1% DMSO in all conditions: 1% DMSO (ATCC);1 μg/mL Ad5 Peptide Mix (PM-HAdV5, JPT PeptideAnimal studies were approved by the Thomas JeffersonUniversity Institutional Animal Care and Use Committee(IACUC). All mouse studies employed 10 week oldfemale BALB/c mice (Jackson, Bar Harbor, ME). Femaleswere nulliparous and not pregnant. To establish Ad5NAb titers, vehicle control (naïve) or control Ad5 (Ad5NAb High) were administered intramuscularly as two50 μL injections, one in each of the two hind limbs. Ad5exposure was repeated 21 days later to establish high Ad5NAb titers. Two weeks after completing the above Ad5exposures, serum was collected for Ad5 NAb titer determinations and animals were immunized with 108 IFUAd5-GUCY2C-S1 [18]. Two weeks later, serum and

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104Page 5 of 12splenocytes were collected and GUCY2C-specific antibodyand GUCY2C- and Ad5- specific CD8 T-cell responseswere quantified as previously described [16–20].Animal modelsStatistical analysisHuman antibody responsesResultsA mixed effect model assuming the interaction betweenserum dilution (1/20, 1/40, etc.) and immune status (pre-vaccination vs post-vaccination) with random effect ofreplications was applied and the one-sided comparison ofimmune status at different dilutions was determined. Thetiter was identified as the greatest dilution producing asignificantly higher signal than pre-vaccination serum atthe same dilution.Ad5-GUCY2C-PADRE is composed of an E1/E3-deletedrecombinant human type 5 adenovirus expressing the human GUCY2C extracellular domain (ECD; GUCY2C1–429)fused on its C-terminus to the universal CD4 T-helpercell epitope PADRE (Fig. 1a and b). Previous studies demonstrated that only the extracellular domain of GUCY2Cis a viable vaccine target reflecting the high sequenceconservation of the intracellular domains of guanylylcyclase family members and broad tissue distributionof guanylyl cyclases A, B, and G [20]. GUCY2CECD-PADRE and an upstream CMV promoter were clonedinto the E1 region of Ad5 (Fig. 1b). Replication-deficient Ad5-GUCY2C-PADRE vector was produced inHEK293 cells and purified by CsCl ultracentrifugationemploying GMP procedures at the Center for Cell andGene Therapy, Baylor College of Medicine. In vitrostudies confirmed dose-dependent (Fig. 1c) andtime-dependent (Fig. 1d) expression and secretion ofGUCY2CECD-PADRE protein by western blot.Human T-cell responsesFor GUCY2C-specific responses, data obtained from10 μg/mL GUCY2C was employed for analysis unlesspre-vaccination signals with 10 μg/mL GUCY2C were 50 spots/well, indicating a high level of non-specific activation at that concentration. In that case, data obtainedfrom 2 μg/mL GUCY2C was used for analysis. ModifiedDistribution Free Resampling (mDFR) algorithms [29]were applied to compare antigen-stimulated (test count)responses to DMSO (control count) at each day, as wellas the pairwise comparisons of the antigen-specificchanges (DMSO-subtracted) between day 0 (pre-vaccination; control count) and each post-vaccination timepoint (test count). The difference between the log of thetest count and the log of the background control count isreferred to as mDFR(eq), while the difference between thelog of the test count and twice the log of the backgroundcontrol count is referred to as mDFR(2x). A positiveantigen-specific response (antigen vs. DMSO) requiredthat antigen vs DMSO at time point X is P 0.05 andantigen-specific spots at time X 5. A positive vaccine-induced response at time point X (antigen-specific responseat time X vs time 0) required that antigen vs DMSO at Xis P 0.05 and antigen-specific response (antigen minusDMSO) at time X vs time 0 is P 0.05 andantigen-specific spots at time X 5. We refer to a result asstrongly significant if the mDFR(2x) P 0.05 and moderately significant if it is not strongly significant, but themDFR(eq) P 0.05. ELISpot responses in patients following CD4/CD8-depletion were compared by Two-wayANOVA with GraphPad Prism v7. For comparisons ofAd5 NAb High and Low patients, for each antigen(GUCY2C, PADRE, and Ad5), the mean difference ofantigen and DMSO between High patients and Low patients was compared. A mixed effect model assumingthe interaction between time and Ad5 NAb status(High vs. Low) with random effect of patients was applied and Low vs. High differences between each dayand day 0 were determined.Responses in animal models were compared by T-test orTwo-way ANOVA, as appropriate, with GraphPad Prism v7.Ad5-GUCY2C-PADRE vectorAd5-GUCY2C-PADRE safety profileTen colorectal cancer patients were enrolled and treatedwith 1011 vp Ad5-GUCY2C-PADRE. Additional file 1:Table S1 describes the baseline patient characteristics. Themedian age was 65 (49–76) years, patients were primarilyCaucasian (80%) and patients were distributed equally between male and female. All patients had stage I or II colorectal cancer previously treated with surgery but notchemo/radio/immuno-therapy. Treatment-related acutetoxicity was assessed in the clinic every 10 min for 30 minafter injection and by telephone on days 3 and 8 followingvaccination. Patients also returned to the clinic 30, 90, and180 days after vaccination for safety assessment. All patients completed the study. Adverse events (Table 1) weregraded according to The Common Terminology Criteriafor Adverse Events (CTCAE version 4.0). Mild grade 1/2toxicities included injection site pain and fever which areTable 1 Treatment-related toxicities occurring during the 6months following Ad5-GUCY2C-PADRE vaccinationToxicityGrade 1/2Grade 3/4Total (%)Chills/Rigor202 (20%)Dizziness101 (10%)Diaphoresis101 (10%)Injection site or arm pain/swelling202 (20%)Aches101 (10%)Fever101 (10%)

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104anticipated following a viral vector immunization. Nograde 3/4 toxicities occurred at any time during the6-month follow-up period after vaccination. Moreover,clinical laboratory assessments performed on days 30,90, and 180, including CBC with differential, comprehensive chemistry panel, and antinuclear antibody (ANA)titers, revealed no vaccine-related adverse events. Importantly, no adverse events related to toxicity in GUCY2C-expressing tissues were observed. GUCY2C is a self proteinexpressed on the luminal surface of small and large intestinal epithelia [7, 8], as well as anorexigenic hypothalamic and midbrain dopaminergic neurons [9, 10].However, consistent with mechanisms controlling immune compartmentalization [30], preclinical studies ofGUCY2C vaccination in mice [16–20] confirmed theabsence of Ad5-GUCY2C-PADRE-induced autoimmunity in intestine or brain.Ad5-GUCY2C-PADRE-induced immune responsesIn preclinical studies, immunization with Ad5GUCY2C-PADRE induced time- and dose-dependentGUCY2C-specific T-cell and B-cell responses and antitumor immunity mediated by CD8 T cells [17–20, 26].Here, GUCY2C-specific immune responses were quantified after Ad5-GUCY2C-PADRE administration by ELISAand IFNγ-ELISpot to quantify antibody and T-cell responses, respectively (Additional file 1: Table S2). T-cellresponses to PADRE and Ad5 also were quantified byIFNγ-ELISpot. Patient responses typically followed 1 of 4patterns and representative responses of each are shownin Fig. 2. All other patient responses are shown in Additional file 1: Figure S1. Patient 1001 had no pre-vaccineantibody responses to GUCY2C or T-cell immunity toGUCY2C, PADRE, or Ad5 and Ad5-GUCY2C-PADREvaccination did not induce responses to these antigens(Fig. 2a). Similarly, while no pre-vaccination responseswere observed in patient 1009, vaccination inducedAd5-specific T-cell responses, but not GUCY2C-specificor PADRE-specific responses (Fig. 2b). In contrast to thesepatients, patient 1008 possessed Ad5-specific T-cell responses prior to vaccination, and Ad5-GUCY2C-PADREvaccination increased those responses (Fig. 2c). Similarly,GUCY2C-specific T-cell responses also were induced byAd5-GUCY2C-PADRE vaccination, initially peaking onday 30, followed by a gradual decline through the final180-day time-point (Fig. 2c). However, this patient produced no PADRE-specific T-cell response or GUCY2Cspecific antibody response, recapitulating preclinicalstudies in mice in which GUCY2C-specific antibody responses require responses to exogenous CD4 “helper”T-cell epitopes, reflecting GUCY2C-specific CD4 T-celltolerance [17–20, 26]. Patient 1007 was the only patientthat produced a response by all three arms of adaptive immunity (Fig. 2d). That patient produced a PADRE-specificPage 6 of 12CD4 T-cell response, a GUCY2C-specific antibody response, and a GUCY2C-specific CD8 T-cell responsethat peaked between days 30 and 90, before declining overthe remainder of the study.GUCY2C-specific T-cell responses are exclusively CD8 T-cellPreclinical studies in mice revealed split tolerance toGUCY2C, eliminating CD4 T cells, but not CD8 T orB cells, which could be fully engaged with exogenousCD4 helper T-cell epitopes (S1 or PADRE) to produceantitumor immunity without autoimmunity [18–20]. Toextend that observation from mice to humans, CD4 orCD8 T cells were depleted from PBMCs of patients1007 and 1008 (GUCY2C responders, Fig. 2), prior toquantification of T-cell responses by IFNγ-ELISpot todetermine the cell type responsible for GUCY2C-specificresponses (Fig. 3). Depletion of CD8 , but not CD4 , Tcells (Fig. 3a, c) eliminated ELISpot responses in bothpatients (Fig. 3b, d). Thus, vaccine responses in coloncancer patients are mediated exclusively by CD8 T cells,recapitulating GUCY2C immunology in mice [18–20].Indeed, selective CD4 T-cell tolerance appears to be auniversal mechanism regulating GUCY2C-specific immunity, eliminating GUCY2C-specific CD4 T cells, butnot B cells or CD8 T cells, in C57BL/6 [17, 19, 20] andBALB/c [18, 20, 26] mice and in humans (Figs. 2 and 3).Ad5 neutralizing antibodies may limit Ad5-GUCY2CPADRE immunogenicity in patientsAdenovirus, including serotype 5 (Ad5) used in Ad5GUCY2C-PADRE, is a natural pathogen producing mildinfections in humans. Natural exposures induce Ad5neutralizing antibodies (NAbs) that inhibit future infections or gene delivery by recombinant adenoviruses,including Ad5-based vaccines, by preventing infection ofhost cells required for antigen expression and inductionof immune responses [31–34]. To determine the impactof Ad5 NAbs on Ad5-GUCY2C-PADRE immunogenicity,Ad5 NAbs were quantified in patient serum collectedprior to Ad5-GUCY2C-PADRE vaccination (day 0) usingan Ad5-GFP reporter virus inhibition bioassay (Fig. 4a).Titers ranged from 10 to 10,000 and an obvious pattern emerged in which 50% of the patients had titersbelow 200 (Ad5 NAb Low) and the other 50% were characterized by titers above 200 (Ad5 NAb High; Fig. 4b).Separating patients into Ad5 NAb Low and High cohortsrevealed a relationship between Ad5 NAb titer andGUCY2C-specific T-cell responses in which responseswere significantly greater in Ad5 NAb Low patients(Fig. 4c). PADRE-specific T-cell responses, which weregenerally low, showed no relationship to Ad5 NAb titer(Fig. 4d). Similar to GUCY2C-specific T-cell responses(Fig. 4c), Ad5-specific T-cell responses also were limited in the Ad5 NAb High group (Fig. 4e).

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104Page 7 of 12Fig. 2 Ad5-GUCY2C-PADRE-induced immune responses. Patient blood samples were collected before (day 0) and 30, 90 and 180 days after Ad5GUCY2C-PADRE immunization. GUCY2C-specific antibody titers were quantified by ELISA and GUCY2C, PADRE, and Ad5 -specific T-cell responses werequantified by IFNγ-ELISpot. ELISpot assays employed DMSO as an antigen-negative control. The statistical significance for T-cell responses at each timepoint (compared to DMSO) was determined by modified DFR(eq) or DFR(2x) after Westfall–Young max-T correction, and p-values 5% are shown inyellow [mDFR(eq)] or red [mDFR(2x)], respectively. The statistical significance of T-cell responses obtained for each post-vaccination time point(compared to day 0) were determined by a similar modified DFR-like permutation method with Westfall-Young max-T correction. RepresentativeGUCY2C non-responders (a and b) and responders (c and d) are shown. All other patient responses are shown in Additional file 1: Figure S1

Snook et al. Journal for ImmunoTherapy of Cancer(2019) 7:104Page 8 of 12Fig. 3 GUCY2C-specific CD8 , but not CD4 , T-cell responses. PBMCs from GUCY2C-responder patients 1007 (a and b) and 1008 (c and d)collected 30 days after Ad5-GUCY2C-PADRE administration were left unsorted or depleted of CD4 or CD8 T cells by MACS. PBMCs, CD4depleted PBMCs, and CD8-depleted PBMCs were analyzed by FACS to confirm depletion (a and c) and tested for GUCY2C-specific T-cellresponses by IFNγ-ELISpot (b and d). NS not significant, *** P 0.001, **** P 0.0001, Two-way ANOVA. Depletion efficienciesdetermined by FACS were 98% for CD4 T cells and 75–95% for CD8 T cellsMouse models confirm Ad5 NAb sensitivity of Ad5GUCY2C vaccinesbarrier to Ad5-GUCY2C-PADRE vaccination in humanpopulations.To confirm the impact of Ad5 NAbs on GUCY2C vaccination with Ad5 vectors in a mouse model, mice wereexposed to control Ad5 vector by two intramuscular immunizations, producing animals with high Ad5 NAbtiters ( 3000; Fig. 5a and b). Ad5-naïve mice or micewith high Ad5 NAb titers were then immunized withAd5-GUCY2C-S1 (a mouse GUCY2C vaccine analogousto Ad5-GUCY2C-PADRE [18]), and GUCY2C-specificantibody (Fig. 5c) and CD8 T-cell responses (Fig. 5d) werequantified. Consistent with previous mouse studies [31]and human responses to Ad5-GUCY2C-PADRE (Fig. 4),pre-existing Ad5 NAbs eliminated GUCY2C-specific antibody (Fig. 5c) and CD8 T-cell responses (Fig. 5d) in mice.Together, these data suggest that pre-existing Ad5 NAbimmunity eliminates Ad5-GUCY2C-PADRE viral particlesin vivo prior to entry into host cells, preventing subsequentgene expression and induction of host immune responses,establish

ited, Cleveland, OH), counted using a Muse Cell Analyzer (Millipore, Darmstadt, Germany), and cryopreserved in CTL-Cryo ABC freezing medium (Cellular Technology Limited) according to the manufacturer’s instructions. Cryovials were frozen at 80 C overnight in a CoolCell LX Al