Ligand-independent TLR signals generated by ectopic overexpression of MyD88 generate local and systemic anti-tumor immunity

Abstract

Although critical for initiating and regulating immune responses, the therapeutic use of individual cytokines as anti-cancer immunotherapeutic agents has achieved only modest clinical success. Consequently, many current strategies have focused on the use of specific immunotherapeutic agonists that engage individual receptors of innate immune networks, such as the Toll Like-Receptor (TLR) system, each resulting in specific patterns of gene expression, cytokine production and inflammatory outcome. However, these immunotherapeutics are constrained by variable cellular TLR expression and responsiveness to particular TLR agonists, as well as the specific cellular context of different tumors. We hypothesized that overexpression of MyD88, a pivotal regulator of multiple TLR signaling pathways, could circumvent these constraints and mimic coordinated TLR signaling across all cell types in a ligand independent fashion. To explore this hypothesis, we generated an adenoviral vector expressing MyD88 and demonstrate that Ad-MyD88 infection elicits extensive Th1-specific transcriptional and secreted cytokine signatures in all murine and human cell types tested in vitro and in vivo. Importantly, in vivo intratumoral injection of Ad-MyD88 into established tumor masses enhanced adaptive immune responses and inhibited local tumor immunosuppression, resulting in significantly inhibited local and systemic growth of multiple tumor types. Finally, Ad-MyD88 infection of primary human dendritic cells, tumor associated fibroblasts, and colorectal carcinoma cells elicited significant Th1-type cytokine responses, resulting in enhanced tumor cell lysis and expansion of human tumor antigen-specific T-cells. Thus, Ad-MyD88 initiated robust anti-tumor activity in established murine tumor microenvironments and in human contexts, suggesting its potential effectiveness as a clinical immunotherapeutic strategy.

Introduction

Immunotherapy strategies frequently target either single tumor antigens or alternatively, non-specifically induce anti-tumor immunity through non-specific approaches, such as the systemic delivery of purified cytokines^1,2. The development of gene transfer techniques has expanded focus on individual effector cytokine, chemokine, or receptor genes delivered to specific tissues, such as GM-CSF gene delivery³. These advances focused attention on the need for therapeutic synergy obtained from coordinated sets of cytokines,⁴ typically induced after stimulation of different pattern recognition receptors (PRRs).

One of the best characterized PRR families is the toll-like receptor (TLR) family which play an instrumental role in the generation of inflammatory and adaptive responses to wide range of pathogen or danger associated molecular patterns (PAMPs and DAMPs)⁵. TLRs are activated by ligand-mediated dimerization of different TLR family members, which in turn recruit adaptor proteins to initiate a unique collection of signaling pathways. Thus, TLR signaling elicits a pattern of cytokines and chemokines that result in a characteristic immune response, not often achieved by single cytokine delivery⁶. Significantly, as different TLR agonists are combined and multiple TLR receptors are engaged, cellular immune responses often become more synergistically complex and in some cases, polarize cells towards a Th1 phenotype⁷.

Although many different TLR-specific agonists are currently being tested⁸, most are being used in isolation thus negating the synergistic advantages afforded by the stimulation of multiple pathways. Furthermore, the effectiveness of single TLR agonists is highly dependent on the responsiveness of tumor cells to TLR agonists,^9,10 which is dependent upon cellular TLR expression that varies by tissue and can be repressed in certain tumor microenvironments¹¹. We hypothesized that an alternative strategy to overcome these obstacles could be to stimulate coordinated TLR signaling in multiple cell types by overexpressing adaptor proteins, which serve as pivotal signaling scaffolds for all TLRs.

TLR signaling interfaces with diverse sources of input enabling responses to a large diversity of PAMPs, and generates broad responses through a conserved core consisting of a limited number of adaptor genes¹². Chief among these is MyD88, which is critical for signaling in all TLRs (with the exception of TLR3). Upon recruitment, MyD88 homo-dimerizes and initiates a broad cascade of coordinated immune responses from a wide range of TLR combinations^13,14. Over-expression of MyD88 has been shown to mimic these responses, as MyD88 accumulates and homo-dimerizes in various cellular compartments to elicit cellular innate immune responses^15,16. Multiple studies have confirmed that MyD88 expression is critical in initiating cellular immune responses to a wide range of pathogens in a variety of different cell types⁶. While MyD88-mediated inflammation has been shown to be pro-tumorigenic in certain models of tumorigenesis, it has also been proven critical for initiating multiple types of anti-tumor immunity^17,18.

Given the importance of MyD88 in mediating immune responses, we hypothesized that specific overexpression of MyD88 by adenoviral vectors could initiate broad innate cellular and subsequently adaptive immune responses in the variety of cell types found in the tumor miroenvironment. This strategy would overcome lack of TLR expression or responsiveness, initiate extensive TLR signaling, and be independent of TLR ligands. We further hypothesized that delivery of this gene in vivo would stimulate proinflammatory responses and alter the immunosuppressive tumor microenvironment to produce therapeutic anti-tumor immunity in vivo.

Materials and Methods

Vector preparation

Mouse and Human MyD88 (ATCC plasmid#7502654 and 10700610) were cloned into E1- and E3- shuttle plasmids, and used to generate [E1-E3-] Ad vectors using pAdEasy¹⁹. Dual-expressing Ad vector were generated by inserting CMV-driven LacZ into the E3 region of pAd-MyD88 constructs. Ad vector stocks were evaluated for replication-competent adenovirus (RCA) via real-time PCR and tittered using the AdEasy Titer Kit (Stratgene, La Jolla, CA).

In vitro tumor cell lines and procedures

CT26.CL25 (LacZ expressing mouse colon carcinoma), CT26.WT (mouse colon carcinoma), B16-F10 (mouse melanoma), and 4T1 (mouse breast carcinoma) cells were obtained from the American Tissue Culture Collection and cultured accordingly. CpG ODN: 1826 (5′-TCCATGACGTTCCTGACGTT) (purchased from the Coley Pharmaceutical Group) and was used in vitro at a concentration of 5μg/ml. Primary Colon Cancer cells and tumor associated fibroblasts from surgically resected tumors, C57/Bl6 wild-type mouse embryonic fibroblasts (MEFs), and HLA-A*0201 PBMCs were kind gifts from Dr. David Hsu, Dr. Joseph Nevins, and Dr. H. Kim Lyerly respectively (Duke University, Durham, NC). Mouse Dendritic Cells were prepared by culturing bone marrow progenitors in GM-CSF (10ng/ml) and IL-4 (10ng/ml) for 5 days while human dendritic cells were prepared from PBMCs cultured in GM-CSF (10 ng/ml) and IL-4 (10 ng/ml) for 5–7 days²⁰. Mouse and human DCs were stained using CD80, CD86, CD83, CD40, HLA-DR, and CD11c antibodies (BD Biosciences, Becton Lake, NJ). Tetramer staining was performed using labeled CD3, CD4, CD8, MART1, and pp65 tetramer mixes (BD Biosciences, Becton Lake, NJ). Chromium-release assays were performed by culturing activated T-cells (10 day culture in 600U/ml of IL-2) with targets and incubating with chromium labeled target cells for 5 hours. Western blots were performed using MyD88 antibodies (ab2064) and Beta-Actin antibodies (ab8227) from Abcam (Cambridge, MA). Assessment of cell viability and proliferation were performed using a 3-[4, 5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazoliumbromide (MTT) assay.

Animal Procedures

C57BL/6J, BALB/c, SCID-B6.129S7-Rag1(tm1Mom), and NOD CB17-Prkdc SCID/J mice were purchased from Jackson Laboratory (Bar Harbor, ME). All tumors were injected at the indicated doses by subcutaneous administration in PBS. Tumor size was measured at the indicated time points by caliper measurement and tumor volume calculated as (small diameter)² × (large diameter/2). Footpad viral injections were performed by anesthetizing animals with a ketamine/xyaline/atropine mix and injecting via the footpad. The CpG ODN 1826 (5′-TCCATGACGTTCCTGACGTT-3′) was injected intralesionally as a single dose of 20 μg in PBS. Intratumoral viral injections were performed by direct injection of virus into the tumor in PBS. Animal studies were approved by the Duke University Institutional Animal Care and Use Committee.

Microarray Methods

RNA was extracted using TRI-Reagent (Molecular Reagents Center, Cincinnati, OH), purified using an RNeasy Kit (Qiagen, Valencia, CA), and assessed for quality using both UV-spectroscopy and a Agilent Lab-on-a-Chip 2100 Bioanalyzer (Agilent Technologies, Palo Alto, USA). RNA was directly labeled and hybridized using the MO36k 70-mer probe sets from Operon (Mouse Genome set v. 4.0) by the Duke University microarray core. Analysis was performed using Genespring v7.2 and DAVID v6.0²¹ with datasets deposited at NCBI’s Gene Omnibus Express (GEO) in a MIAME-compliant form (along with complete details of all procedures and analyses) as accession GSE18957 and GSE 190006 using the platform GP6524.

Quantitative rt-PCR and Cytokine/Chemokine Assays

RNA was reverse transcribed using iScript (Bio-Rad, Hercules, CA) and real-time PCR was performed using FastStart Universal SYBR Green (Roche, Indianapolis, IN) with intron-spanning primers, normalizing to the geometric mean of housekeeping genes (Beta-Actin, HMBS, and GAPDH), and calculate relative expression differences with the comparative CT method²². Serum and supernatants were assayed using Bio-Rad, Bio-Plex 23-Plex mouse and 28-Plex human cytokine kits, according to the manufacturer’s recommendations.

Statistics

Statistics for ELISA, qrt-PCR, ELISPOT, and MTT assays were performed using two-tailed homoscedastic Student’s t-test with Bonferroni multiple testing correction for Bio-Plex ELISA samples. For animal studies, statistical differences were calculated with a mixed effects regression model using autoregressive covariance.

ELISA and ELISpot procedures

Anti-adenovirus and anti-LacZ antibodies were quantified using ELISA. Plates were coated with 1×10⁹vp or 2μg of LacZ (Sigma, St. Louis MI) per well in a bicarbonate solution (200 mM NaHCO₃, 81 mM Na₂CO₃, pH 9.5) overnight at 4°C and washed 5 times before secondary sheep anti-mouse IgG H+L antibody (Jackson Immunoresearch Laboratories, West Grove, PA) and O-phenylenediamine dihydrochloride substrate (Sigma, St. Louis, MO) application.

Alloantigen-primed IFN-gamma-producing T-cells were quantified using an Enzyme Linked Immuno-Spot (ELISpot) assay. Capture and detectionanti-IFN-gamma monoclonal antibodies were purchased from MABtech (USA, Mariemont, OH). Stimulatory LacZ peptide (TPHPARIGL, 1ug/ml) and GFP peptide (HYSTQSAL, 1ug/ml) were purchased from GenScript (Piscataway, NJ), while LacZ protein (50μg/ml) and Amyloglucosidase protein controls (50ug/ml) were purchased from Sigma (St. Louis, MO). HIV irrelevant overlapping peptide mixes were purchased from BD Biosciences (San Jose, CA). Plates were developed with 3-amino-9-ethyl-carbazole [AEC] (Sigma-Aldrich, St. Louis MO), and the number of spots per well was determined using a KS ELISpot Automated Reader System with KS ELISpot 4.2 Software (Carl Zeiss, Inc., Thornwood, NY).

Results

Although contemporary immunotherapeutic strategies utilize single defined molecular interventions^23,24, we sought to deliver the TLR adaptor protein MyD88 as a means to initiate broad innate and adaptive anti-tumor immune responses within the tumor microenvironment.

Ad-MyD88 infection elicits unique patterns of immune stimulatory gene expression in bmDCs and MEFs and enhances adaptive immune responses in vivo

We first investigated the transcriptional effect of Ad-mediated mouse MyD88 overexpression on mouse bone-marrow derived dendritic cells (bmDCs). Ad-MyD88 infection resulted in significant MyD88 protein expression as determined by Western blot (Fig. S1) and microarray assessment revealed that ~3.1% of gene transcripts were significantly dysregulated by Ad-MyD88 compared to Ad-GFP^25,26 or mock infected counterparts (Fig. 1A), mainly in gene groups involving inflammation and myeloid development (using DAVID analysis).

Figure 1. Ad-MyD88 infection elicits Th1 inflammatory profiles in primary mouse dendritic cells and fibroblasts.

A) Microarrays performed for bmDCs and MEFs treated or infected as in A) (n=5). Hierarchical clustered heat map of significant genes. Gene groups affected by Ad-MyD88 overexpression from both MEFs and bmDCs represented in proportional pie chart B) and C) Assessment of significantly different cytokines and chemokines at 16hpi in bmDCs (n=4, MOI=2000) and MEFs D) (n=4, MOI=150). Where indicated * and ** represent p<.05 and p<.01 respectively in comparing Ad-MyD88 and CpG treated cells to Ad-GFP infected controls and # and ## represent p<.05 and p<.01 respectively in comparing Ad-MyD88 to CpG treated cells. Error bars represent SD.

To determine if non-immune cells infected with Ad-MyD88 would show similar responses, primary mouse embryonic fibroblasts (MEFs) were infected with Ad-MyD88 and transcriptional assessment revealed that an even larger fraction of genes (~4.3%) were significantly dysregulated compared to controls (Fig. 1A–B, S1), mainly in gene clusters involved in inflammation (using DAVID analysis). While revealing differences in specific gene signatures between cell types, these results demonstrated a conserved MyD88 inflammatory response in two primary cell types relevant to the tumor microenvironment.

When all differences between control infected and Ad-MyD88 infected MEFs and bmDCs were assessed together by 1-way ANOVA (p=.05 with Benjamini and Hochberg Multiple Testing Correction), it revealed that ~8.0% of the total transcriptome dysregulated by MyD88 overexpression. Functional analysis of these dysregulated genes revealed significant induction of the expression of cytokine/chemokine, lymphocyte activation, actin cytoskeleton, negative regulation of TGF-beta as well as leucine-rich repeat gene clusters (Fig. 1C and Table S2). Quantitative rt-PCR (qrt-PCR) of multiple genes in MEFs and bmDCs (Table S1) confirmed these differences and comparison with a MyD88-dependent CpG TLR9 agonist (ODN 1826) determined that Ad-MyD88 largey elicited greater transcriptional responses compared to CpG stimulation (Table S1). These transcriptional patterns indicated that MyD88 overexpression elicited a unique and broad Th1-biased immune stimulatory response in different cell types.

In addition to mRNA differences, alterations in protein expression were determined by Bio-Plex ELISA analysis of supernatants from Ad-MyD88 infected and CpG treated cells. These assays showed significantly higher secretion of proinflammatory and Th1 related cytokines and chemokines in Ad-MyD88 infected bmDCs (Fig. 1C and S2) and Ad-MyD88 infected MEFs (Fig. 1D and S2). Compared to CpG, we found that Ad-MyD88 infected cells secreted significantly different levels of multiple cytokines and chemokines dependant upon cell type (Fig. 1C–D and S2), suggesting differences in the mechanism of MyD88-dependent activation between these treatments. Collectively, these results demonstrate that Ad-MyD88 infection induces a selective Th1 inflammatory profile in different primary murine cell types in vitro in comparison to control Ad or CpG stimulated counterparts.

To determine if Ad-MyD88 could enhance adaptive immune responses, as has previously been reported for a MyD88-plasmid based strategy²⁷, we injected mice with a co-mixture of Ad-MyD88 and Ad-LacZ or with a dual LacZ and MyD88 expressing Ad versus control Ad vectors. Expression of MyD88, either as an Ad co-mixture or in the dual-expressing Ad, elicited significantly higher numbers of LacZ specific IFN-γ secreting functional T-cells by ELISPOT (Fig. 2A), indicating an Ad-MyD88 Th1 adaptive immune phenotype. In contrast, MyD88 overexpression did not significantly enhance IgG responses to LacZ or adenoviral antigens (Fig. 2B and 2C)

Figure 2. Ad-MyD88 infection elicits enhanced activation of adaptive immune responses in vivo.

A) C57/BL6 mice were vaccinated (2.6×10¹⁰ total viral particles (vp)) via footpad and adaptive immune responses assessed at 2 wpi by ELISPOT (using control PMA, LacZ (50μg/ml), control Amyloglucosidase protein (50μg/ml) or mock stimulation (n=5). B) and C) ELISA was performed on vaccinated mice (at 14 days post-injection) to determine LacZ (B) and Ad (C) specific IgG antibodies in mice treated as above (n=5). Where indicated * and ** represent p<.05 and p<.01 respectively in comparison to mock infected controls and # and ## represent p<.05 and p<.01 respectively in comparison to Ad control infected counterparts. Error bars represent SD.

Ad-mediated MyD88 overexpression alters gene expression in tumor cells and suppresses growth of established tumors

In contrast to the characteristic patterns of gene expression we observed in various untransformed cell types, we hypothesized that the genetic instability and potential loss of TLR signaling mediators in different tumor types could significantly hinder immune MyD88 pathway activation²⁸. Therefore, we compared Ad-mediated MyD88 overexpression with CpG (ODN 1826) treatment in murine tumor cell lines CT26.CL25, 4T1, and B16-F10. As before, western blots revealed that MyD88 was strongly overexpressed in Ad-MyD88 infected tumor cells (Fig. S3) and different types of tumor cells were highly responsive to MyD88 overexpression with multiple inflammatory genes being strongly activated (Fig. 3A). In contrast, these tumor lines were not highly responsive to CpG stimulation (Fig 3A). Significantly, Ad-MyD88 infection elicited enhanced expression of multiple cytokines and chemokines in supernatants (ELISA, Fig. 3B, Fig. S4), thus indicating largely intact functional signaling pathways downstream of MyD88 in different tumor types.

Figure 3. Ad-MyD88 activates innate immune responses in tumor cells and strongly suppresses tumor growth in vivo.

A) CT26.CL25, B16-F10, and 4T1 cells were infected (MOI=200) with Ad-MyD88, Ad-GFP, CpG treated (5μg/ml), or mock treated and transcription assessed at 48 hpi by quantitative real-time PCR (qrt-PCR). The average fold change is displayed by color coded gradient (red for higher expression and green for lower expression, n=3). B) Using conditions as in (A) supernatants were harvested from CT26.CL25 infected cells and cytokines/chemokines assessed at 24hpi (n=3). Several representative cytokine/chemokines are displayed. C) BALB/c mice with CT26.CL25 tumors were intralesionally treated as shown at 9–11 dpi (indicated by arrow) and tumor growth measured over time (n=5, error bars represent SE). D–F) BALB/c bearing D) CT26.WT (no LacZ), E) 4T1, or F) B16-F10 tumors were vaccinated at 8–12 dpi (indicated by arrow) as in C) (n=5, error bars represent SE). Where indicated * and ** represent p<.05 and p<.01 respectively in comparison to mock controls and # and ## represent p<.05 and p<.01 respectively in comparison to Ad control or CpG treated counterparts.

We next compared the therapeutic efficacy of Ad-MyD88 to a TLR9 agonist (CpG 1826)²⁹ that could activate innate responses in primary cells but not in tumor cells. As established tumors are therapeutically resistant³⁰, 11 day old (mean tumor volume ~50mm³) CT26.CL25 LacZ+ tumors were injected intralesionally with Ad-MyD88, CpG, or vehicle or non-CpG controls (Fig 3C). Tumor masses injected with Ad-MyD88 had significantly retarded growth in contrast to the minor effects of CpG. To determine if these responses could be enhanced by targeting the tumor-specific antigen, LacZ, we performed intralesional injections of Ad-MyD88, Ad-LacZ, Ad-MyD88-LacZ, Ad-GFP, co-mixtures of these vectors, or vehicle buffer alone. Notably, we found that Ad-LacZ injection did not significantly affect tumor growth (Fig. 3D), even though the tumors continued to express LacZ (data not shown). In contrast, all tumors injected with a MyD88 expressing vector had significantly reduced growth, independent of Ad-LacZ administration (Fig. 3D). To eliminate the immunologic effects of LacZ expression, we tested a non-LacZ expressing model. As before, a single intralesional injection of Ad-MyD88 into established (day 12, ~50mm3) non-LacZ expressing CT26.WT colon carcinomas resulted in significant growth suppression (Fig. 3E)

We extended these studies to different tumor types, testing the relatively non-immunogenic B16-F10 melanoma³¹ and 4T1 breast carcinoma lines^32,33. Established 4T1 and B16-F10 tumors (Day 8 and 9, ~50mm³) were injected intralesionally with Ad-MyD88 or Ad-GFP. As in the CT26 model, a single injection of Ad-MyD88 significantly suppressed the growth of both 4T1 and B16-F10 tumors (Fig. 3E). In vitro infection of these tumors revealed that Ad-MyD88 did not affect tumor growth (Fig. S5), thus demonstrating that anti-growth effects were not mediated by alteration of cellular growth properties. Thus, Ad-MyD88 elicits immune responses in tumors of distinct histologies that translate into repressed growth in vivo.

Ad-MyD88 injection elicits systemic and local immunity which suppresses tumor growth through T-cells and NK cells

To determine if Ad-MyD88 had a systemic effect on tumor growth, intralesional injection of Ad-MyD88 was tested in a bilateral CT26.CL25 tumor model. In addition to repression of growth in the Ad-MyD88 injected tumor, we observed growth repression in the contra-lateral uninjected CT26.CL25 tumor (Fig. 4A) suggesting a systemic response against tumor antigens^34,35. This was confirmed by ELISPOT analysis of splenocytes which demonstrated enhanced T-cell responses against LacZ in mice with Ad-MyD88 injected tumors (Fig. 4B). However, there was no enhancement of LacZ-specific IgG antibodies in Ad-MyD88 injected mice (Fig. S6).

Figure 4. Intralesional injection of Ad-MyD88 represses tumor growth by systemically enhancing Th1 responses and alleviating immuno-supression in injected tumors.

A) Mice bearing CT26.CL25 bilateral tumors were treated by a right flank tumor injection of the indicated Ad vectors 12 dpi (n=6, error bars represent SE). B) ELISPOT assays using LacZ, GFP, or control irrelevant peptides were performed on mice from A) (n=6, error bars represent SD). C) Qrt-PCR was performed to assess the transcription of immuno-suppressive genes from infected CT25.CL26 cells (MOI=200, 48hpi) and in injected tumor masses as described in (A) (n=5–6, error bars represent SD). D) and E) CT26.CL25 tumor bearing B6.129S7-Rag1(tm1Mom) mice (D) and NOD CB17-Prkdc SCID/J (E) were vaccinated with the respective Ad vectors or mock treated (at 50mm³ indicated by arrow) and tumor growth measured (n=5, error bars represent SE). Where indicated * and ** represent p<.05 and p<.01 respectively in comparison to mock infected controls and # and ## represent p<.05 and p<.01 respectively in comparison to Ad control infected counterparts.

While we demonstrated Ad-MyD88 elicited genes associated with Th1-type inflammation and TGF-beta repression in vitro (Fig. 1A–B), it was unclear if these responses could alter local tumor immunity in vivo. Since larger established tumors are known to be immunosuppressive³⁶, we investigated if Ad-MyD88 could alter this phenotype by assessing transcription levels of several known immunosuppressive genes after Ad-MyD88 infection of CT26.CL25 cells in vitro or intralesional injection in vivo. Surprisingly, we found that intralesional administration of Ad-MyD88 was able to repress the expression of 4 immunosuppressive genes (CSF-1, IL-10, PGE2, TGF-beta) tested in our panel (Fig. 4C)^23,37. This transcriptional suppression of immunosuppressive genes was not seen after tumor cell infection in vitro suggesting that the in vivo effect is indirectly mediated through stromal cell infection resulting in an altered inflammatory environment which promotes anti-tumor immunity^30,38,39.

As Ad-MyD88 could significantly enhance systemic T-cell responses in vivo, we hypothesized that Ad-MyD88-mediated tumor growth repression was largely T-cell mediated. To test this hypothesis, established CT26.CL25 tumors were intralesionally injected with Ad-MyD88 or Ad-GFP in mice deficient for T-cells. In comparison to immunocompetent mice, the degree of Ad-MyD88 tumor repression was highly diminished (p=.04), but still significant compared to Ad-GFP injected T-cell deficient control mice (Fig. 4D). These results suggest that T-cells play critical role in Ad-MyD88 mediated immunity but are not the sole effectors. The broad cytokine and chemokine repertoire elicited by MyD88 overexpression also suggested that NK cells could be playing a role in Ad-MyD88 mediated immunity^40,41, and this was confirmed in mice deficient for both T-cells and NK cells, where Ad-MyD88 had no effect on tumor growth (Fig. 4E).

Systemic delivery of Ad-MyD88 elicits activation of innate immune responses in vivo that are temporally regulated and well tolerated

As intratumoral viral administration can be partially systemic via vascular uptake⁴², we next investigated the effect of systemic Ad-MyD88 delivery in an in vivo model, exploiting the ability of adenovirus to effectively transduce hepatocytes in vivo after intravenous delivery²⁶. In this model, mice were i.v. injected with a therapeutically relevant dose (7.5×10¹⁰vp; equivalent to approx. 2.1×10¹⁴ vp in a 70 Kg human) of Ad-MyD88, Ad-GFP control, or vehicle control with hepatocytes harvested at 6 or 24hpi for transcriptional assessment by qrt-PCR. Mice tolerated these injections well and qrt-PCR assessment revealed the expression of multiple effector and regulatory immune genes peaked at 6hpi (Fig. 5A). Assessment of cytokine and chemokines at 1, 6, and 24 hpi, revealed that multiple cytokines/chemokines in the serum of Ad-MyD88 infected mice also peaked at 6hpi (Fig. 5B and 5C) with only G-CSF still significantly induced compared to controls at 24hpi (Fig. 5B). Comparison of genome copy number by real-time PCR at 6 and 24hpi revealed no difference in the infectivity or elimination of these viruses (data not shown). These results thus demonstrate that while Ad-MyD88 can strongly induce local and systemic innate immune responses in vivo after i.v. delivery, systemic exposure to very high doses of Ad-MyD88 and its induced cytokines/chemokines are well tolerated and still subject to immune regulatory control mechanisms.

Figure 5. Systemic administration of Ad-MyD88 elicited temporally enhanced innate immune responses.

A) C57/BL6 mice (n=6) were intravenously injected with indicated vectos (7.5×10¹⁰ vp) of and assessed for liver transcriptional dysregulation by qrt-PCR (Average dysregulated in comparison to mock presented). B) and C) Serum at 1, 6, and 24 hpi was assessed for inflammatory cytokines and chemokines respectively (n=6, error bars represent SD). Where indicated * and ** represent p<.05 and p<.01 respectively in comparison to mock infected controls and # and ## represent p<.05 and p<.01 respectively in comparison to Ad control infected counterparts.

Expression of human MyD88 in adenoviral vectors elicits proinflammatory patterns of gene expression in human cells, enhances tumor lysis, and the expansion of tumor antigen-specific T-cells

To determine if these results were relevant to human cancers, we tested human MyD88 overexpression in different primary and transformed human cells. Primary human dendritic cells (hDCs), tumor associated fibroblasts cultured from primary human colon cancer metastases (CRC-TAFs), multiple colon cancer cell lines, as well as two primary colorectal tumors surgically resected from patients (CRCs) were infected with a human MyD88-expressing Ad-MyD88 and immune pathway activation examined by qrt-PCR, as previously performed in mouse cells. Infection of primary immune and non-immune human cells with Ad-MyD88 vectors elicited a characteristic pro-inflammatory gene expression profile (Fig. 6A), although as before, responses did vary between cell types. Assessment of multiple cytokines and chemokines after hDC infection confirmed Ad-MyD88 induction of 12 different pro-inflammatory cytokines (Fig. 6B, data not shown), as well as the DC immune stimulatory molecules CD80, CD86, and CD40, and the DC maturation marker CD83 (Fig. S7A). Similarly, MyD88 elicited significant secretion of Th1 related cytokines, including CXCL10, in CRC-TAFs and tumor cells (Fig. S7B–G and data not shown). Of particular significance, we found that Ad-MyD88 infected primary CRCs secreted high levels of the major Th1 cytokines TNF-alpha, IFN-gamma, and CXCL10 compared to controls (Fig. 6B). Collectively, these results reveal that overexpression of human MyD88 strongly induces a pro-inflammatory immune-response profile in all human cell types and thus suggests that the use of Ad-MyD88 could significantly impact immune responses in the tumor microenvironment in human patients.

Figure 6. Ad-MyD88 activates innate immune responses in multiple types of human cells and enhances human tumor cell lysis as well as the expansion of human tumor antigen-specific T-cells.

A) Human dendritic cells (hDCs), primary Colorectal cancer cells (CRCs), and CRC tumor associated fibroblasts (CRC-TAFs) were infected Ad-MyD88, Ad-GFP, or mock infected (hDC MOI=2000, CRC MOI=10000, CRC-TAF MOI=2000)and gene expression assessed by qrt-PCR. The average fold change (from mock treated cells) is displayed by color coded by gradient (red for higher expression and green for lower expression, n=4). B) Representative cytokine and chemokine secretion from hDCs (upper panels) and primary CRC057 tumor cells (lower panels) at 16hpi treated as in (A) (n=4). C) Ad-MyD88 infection of COL205 cells enhances tumor cell killing by activated T-cells. COLO205 cells were infected Ad-LacZ, Ad-MyD88, or mock infected (MOI=10000) and cultured with IL-2 activated T-cells at indicated E:T ratios (n=6). D) Ad-MyD88 enhances expansion of MART1-specific CD8+ cells. Human peripheral blood monocytes (PBMCs) were treated as indicated (MOI=2500) and stimulated with MART1 peptide (100ug/ml). At 3dpi, IL-2 was added (60U/ml) and cells were expanded for 7 additional days (10dpi) and assessed by tetramer staining was performed using labeled CD3, CD4, CD8, antibodies and a MART1-specific tetramer. These experiments were independently performed five times in triplicate and representative scatterplots are shown. In all experiments, error bars represent SD, and where indicated * and ** represent p<.05 and p<.01 respectively in comparison to mock infected controls and # and ## represent p<.05 and p<.01 respectively in comparison to Ad control infected counterparts.

As Ad-MyD88 infection had induced the expression of multiple bioactive cytokines in different cell types, as well as the expression of co-stimulatory markers in DCs, we hypothesized that these and other effector mechanisms could have an impact on tumor cell lysis and T-cell expansion. Infection of a HLA-A*0201+ colorectal cancer line (COLO205) with Ad-MyD88 followed by co-incubation with IL-2 activated HLA-A*0201+ normal donor T-cells revealed that Ad-MyD88 infected cells were indeed more sensitive to tumor cell lysis in comparison to Ad-LacZ infected cells (Fig. 6C). To ascertain if Ad-MyD88 could enhance the expansion of tumor antigen specific T-cells, peripheral blood monocytes (PBMCs) were infected with Ad-MyD88 in the presence of the self-tumor antigen peptide MART1(27-35). Assessment of cytokines in culture supernatants revealed that Ad-MyD88 infected cells secreted higher levels the Th1-like cytokines GM-CSF, IFN-gamma, and TNF-alpha (Fig. S8). In addition, we found a 10 fold higher induction of MART1(27-35) peptide-specific T-cells in these cultures using tetramer staining (Fig. 6D). Collectively, these results demonstrate that Ad-MyD88 can elicit robust Th1-skewed innate immune responses in a wide range of cells present in a human colorectal tumor microenvironment, as well as, functionally inducing greater tumor cell lysis and expansion of tumor antigen-specific T-cells after infection of human cells in vitro.

Discussion

The lack of effective of immunotherapeutic strategies in patients with advanced cancer is in large part due to the inefficiency of generating immune responses coupled with the powerful immunosuppressive and immunomodulatory environment present in patients. Using adenoviral vectors that expressed MyD88, we sought to deliver coordinated TLR-like signals to stimulate innate host immune responses and provoke adaptive anti-tumor immunity. Our results demonstrate that Ad-MyD88 infection elicits a profound induction of inflammatory genes and a unique Th1-biased gene signature in multiple primary and transformed murine and human cell types in vitro. This was in contrast to the use of a specific TLR ligand, CpG which elicited a different Th1-biased signature in primary murine cells but did not elicit robust responses in transformed murine cells. In vivo, we found that a single intratumoral injection of Ad-MyD88 enhanced adaptive immune responses and suppressed local and systemic growth of multiple tumor types, in contrast to a single injection of a CpG TLR9 agonist. Finally, Ad-MyD88 infection of primary human cells in vitro elicited Th1-type responses, enhanced tumor cell lysis, and the expansion of tumor antigen-specific T-cells, suggesting that the clinical delivery of human Ad-MyD88 into a colorectal tumor microenvironment could effectively parallel the in vivo effectiveness observed with the CT26 mouse colorectal cancer model.

Previous approaches at cytokine therapy of cancer largely used single agents leading to elevated systemic cytokine levels which were linked to dose limiting toxicities ^34,43–48. The local delivery of a TLR specific signaling adaptor protein confers two major advantages. First, delivery of the TLR adaptor to multiple cell types in the tumor microenvironment leads to a pattern of cytokine secretion and innate immune activation, regardless of TLR receptor repertoire in normal and malignant cells. Second, it activates a consistent pattern of specific cytokines to elicit robust immunity. The expression of this pattern of cytokines at coordinated levels by Ad-MyD88 could also minimize the need for high systemic levels of individual cytokines which may be associated with toxicities^34,45–48. Thus, while cytokine-induced toxicity remains a concern in the use Ad-MyD88 that must be addressed in animal toxicology studies, the activation of a coordinated set of cytokines in multiple cell types may indirectly counteract systemic toxicities while permitting maximal immune responses.

Finally, our study demonstrated the effectiveness of human MyD88 expression in inducing cellular immunity in primary tumor cells and fibroblasts from metastatic colon carcinomas. While the importance of MyD88 and TLRs in humans is largely unknown⁴⁹, with only a few studies identifying patients with functional defects in TLR genes that translate into hindered immunity to different pathogens⁵⁰, our study found significant concordance between the responses elicited by mouse MyD88 expression in murine cells in vitro and in vivo, and human MyD88 overexpression in different types of human primary and transformed cells. These data suggest a strong functional role for MyD88 in human immunity and could further suggest the utility of MyD88-overexpression-based approaches in therapies requiring immune modulation, including infectious disease and autoimmunity.

In summary, this investigation reveals that MyD88 overexpression activates TLR pathways in established models leading to the effective generation of innate and adaptive responses that alter immunity to allow for effective tumor immunotherapy. The capacity of human Ad-MyD88 to elicit similar immune response profiles and effective immune stimulation in vitro in human cells thus indicates that Ad-MyD88 warrants further development and clinical translation as a novel means of enhancing anti-tumor immunotherapy.

Abbreviation

Ad

adenovirus

ANOVA

Analysis of Variance

FDR

Benjamini-Hochberg False Discovery Rate

dpi

days post-infection

ELISA

Enzyme-linked Immunosorbance Assay

GO

Gene Ontology

moi

multiplicity of infection

MyD88

Myeloid Differentiation Factor 88

Th1

T-helper type 1

TLR

Toll-Like Receptor

wpi

weeks post-infection