Adenovirus vector induced Innate Immune responses: Impact upon efficacy and toxicity in gene therapy and vaccine applications

Abstract

Extensively characterized, modified, and employed for a variety of purposes, Adenovirus (Ad) vectors are generally regarded as having great potential by many applied virologists who wish to manipulate and use viral biology to achieve beneficial clinical outcomes. Despite widespread functional prominence and utility, (i.e.: Ad based clinical trials have begun to progress to critical Phase III levels, it has recently become apparent that investigations regarding the innate immune response to Ads may reveal not only reasons behind previous failures, but also reveal novel insights that will allow for safer, more efficacious uses of this important gene transfer platform. Insights gained by the exploration of Ad induced innate immune responses will likely be most important to the fields of vaccine development, since Ad based vaccines are highly acknowledged as one of the more promising vaccine platforms in development today.

Adenovirus is currently known to interact with several different extracellular, intracellular, and membrane bound innate immune sensing systems. Past and recent studies involving manipulation of the Ad infectious cycle as well as use of different mutants have shed light on some of the initiation mechanisms underlying Ad induced immune responses. More recent studies using microarray based analyses, genetically modified cell lines and/or mouse mutants, and advanced generation Ad vectors have revealed important new insights into the scope and mechanism of this cellular defensive response. This review is an attempt to synthesize these studies, update Ad biologists to the current knowledge surrounding these increasingly important issues, as well point areas where future research should be directed. It should also serve as a sobering reality to researchers exploring the use of any gene transfer vector, as to the complexities potentially involved when contemplating use of such vectors for human applications.

Introduction

Adenovirus (Ad) based gene transfer has been an important part of the scientific landscape, contributing to a great number of studies, both of basic virology, as well as more focused and recent uses in gene therapy and vaccine applications. As of this date, more human clinical trials utilize recombinant Ad (rAd) based vectors than any other gene transfer platform. Furthermore, recent progress in utilizing Ad based vectors as a vaccine platform suggests greater uses of this gene transfer workhorse for expanding purposes, such as in HIV-AIDS, cancer immunotherapy approaches, and in vaccination for serious infections such as Ebola, and bird flu. Despite these successes, the scientific and bio-industrial communities have recently recognized that a great lack of knowledge exists regarding the mechanisms by which Ad vectors induce several classes of innate immune responses. Unfortunately this has, in some isolated cases, resulted in serious consequences. Clearly, a better understanding of the Ad induced innate immune responses will not only foster safer utilization of this platform, but such knowledge will also greatly improve future applications of this increasingly useful gene transfer platform.

Since very early in the evolutionary history of multi-cellular organisms, cellular and viral parasites have emerged to pose a significant challenge to a complex organism’s survival (Hoffmann et al., 1999). In mammals, this challenge was met by the primordial co-evolution of defensive networks that include both the innate and adaptive immune systems. The principal task of these systems was to evolve in a manner that allowed the host to rapidly detect the first encounter with a potential pathogen, coordinate both isolation and neutralization of the invader, and to create a “rapid-response” defense mechanism to be triggered in the event of repeat encounter (Beutler and Rietschel, 2003; Hoffmann et al., 1999; Medzhitov and Janeway, 2002). In light of this evolutionary arms race, the difficulty surrounding the conversion of viruses into vectors for gene transfer should come as little surprise.

The use of viral vectors for gene transfer is a relatively simple idea; insertion of desired genetic material into the viral genome, thereby allowing one to take advantage of the inherent ability of the virus to transduce cells for a desired therapeutic outcome. For the past twenty years, this strategy has been applied to the one of the most extensively studied and widely used viral vector systems, those based upon human adenoviruses (Ads). Named for the adenoid tissue from which it was initially isolated, the genus Adenoviridae encompasses a large family of non-enveloped, double-stranded DNA viruses. From this family, human subgroup C Ads have been the most extensively characterized, with two of its serotypes, 2 and 5, emerging as the viral platforms most commonly used for gene delivery and vaccination purposes. While work involving the genetic manipulation of the virus has been widely reported, much recent research has focused on analyzing Ad vector interactions with the innate immune system. We feel that a clearer understanding of this knowledge will help direct future research efforts attempting to improve the safety and/or more effective use of recombinant Ad-based vectors in widespread clinical applications. This review will therefore examine the innate immune repertoire elicited by Ads, or Ad-based vectors, both in vitro and in vivo, including their interactions with important extracellular and intracellular pathogen sensing mechanisms.

Ad INDUCED CELLULAR DEFENSE RESPONSES IN VITRO

More than 30 years ago, in vitro studies using adenovirus found it to be a potent inducer of ‘interferon’ in chick embryonic fibroblasts, setting the stage for the exploration of innate immune responses to adenovirus in vitro (Tarodi et al., 1977; Ustacelebi and Williams, 1972). Since that time there has been considerable research in this area, and it is now well-appreciated that in vitro Ad infection of cells initiates a broad series of events, inclusive of stress response pathways, metabolic changes, and intrinsic immune responses. The ability to generate high titers of Ad vector has facilitated in vitro analyses of Ad-specific defense responses, as lower Ad titers may activate similar responses, but at levels that may be undetectable with current technologies. As an aside, these considerations should be understood when comparisons are made between gene transfer (viral) vectors, as a “lack” of detection of immune responses could be due to inadequate assay sensitivity. Here we will examine the cellular defensive responses of both ‘non-immune’ cells and more specialized ‘immune’ cell types to Ad infection, highlighting common patterns as well as insights into the mechanism of this response. We will then examine recent work regarding the molecular mechanisms underlying these responses in an examination of the pathogen recognition receptor (PRR) contribution to adenoviral induced innate responses in vitro.

“Non-Immune” Cell Responses to Ad infection in vitro

Ad infection of what many would consider “non-immune” cells has been found to precipitate extremely rapid changes, with p42/MAPK phosphorylation and ERK signaling occurring 10–20 minutes after Ad infection in both HeLa (human cervical epithelial cells) and A549 (human respiratory epithelial cells), the latter of which was shown to be, in part, dependent upon fiber/CAR interactions, and independent of penton-RGD integrin interactions (Tamanini et al., 2006; Tibbles et al., 2002). Similar signaling events occur in Ad infected REC (murine renal epithelium-derived) cells, with p38 and ERK activation occurring 10 minutes post-infection. These early signaling events were directly linked to the subsequent expression of chemokine genes, as pharmacological inhibitors of p38 and ERK directly inhibited Ad triggered IP-10 expression (Liu and Muruve, 2003; Bowen et al., 2002; Bruder and Kovesdi, 1997; Tibbles et al., 2002). In a follow-up study, Liu et al., (2005) showed that within 10 minutes, RGD/integrin interactions resulted in PI3K dependent Akt and NF-κB activation and subsequent IP-10 transcription. Other genes induced by Ad infection included various leukocyte adhesion molecules, such as ICAM-1 and VCAM-1. In human embryonic derived 293 (HEK293) cells, UV-inactivated Ad vectors induce the expression of several chemokines, particularly RANTES, IP-10, and MIP-2 (Liu and Muruve, 2003). Furthermore, HeLa, A549, and the TGP61 mouse insulinoma cell line, show similar innate chemokine activation patterns, with Ad infections inducing expression of IP-10, RANTES, and in human cells, IL-8. In contrast to findings in macrophages, the induction of RANTES and IP-10 in REC epithelial cells and in HeLa cells occurred in the absence of a significant TNF-α and IL-1β induction. Nuclear translocation of NF-κB also figures prominently in Ad induced innate induction of cytokines and chemokines, both in cell culture and in vivo (Borgland et al., 2000; Bowen et al., 2002; Loser et al., 1998; Muruve et al., 1999). These data demonstrate that Ad infection activates broad cellular responses in what many would consider “non-immune cells”, such as epithelial and endothelial cells.

Furthermore, different non-immune cell types respond to Ad infection by the production of different sets of cytokines, although the specific repertoire of cytokines elaborated likely varies between cell types.

In contrast to the approach taken in the aforementioned studies, (where innate immune activation was probed through the investigation of single genes or kinases) several investigations have explored the effects of Ads on non-immune cells using a more global approach, involving microarray based transcriptome analysis. For example, Stilwell and Samulski, (2004) found that first generation recombinant Ads (FG-Ads), as well as ‘empty Ad capsids’ altered a significant portion of the cellular transcriptome. Specifically, 24 hours post-infection (hpi) of IMR-90 human primary lung fibroblasts, FG-Ads caused 3-fold or greater changes in the expression of 4.2% of the genes examined, while ‘empty’ Ad capsids caused expressional changes in 0.7% of the genes examined, at . Despite the fact that the ‘empty Ad capsids’ possessed a somewhat altered morphology, and lacked any encapsulated DNA, (thus no terminal protein), they still elicited a sizable number of transcriptome changes in this non-immune human diploid embryonic fibroblast cell line, constituting a putative “capsid induced” transcriptome signature.

In an investigation of tropism-modified Ads, Volk et al., (2005) found rather striking differences in the global gene transcription profiles induced by Ad transduction of a human melanoma cell line (M21) with 1) native FG-Ads, 2) an Ad5 virus displaying the Ad3 fiber knob domain (Ad5/3) which alters the interaction with the coxsackie-Adenovirus receptor (CAR), and 3) an Ad5 vector displaying a fiber knob with an additional RGD binding domain. All three vectors induced dysregulation of the cellular transcriptome at 30hpi, with the native Ad5 vector inducing the most dysregulation, followed by the Ad5/3 chimera and the RGD fiber knob mutant Ad. The common group of genes whose expression was dysregulated by the different Ads showed little common functionality, although a statistical grouping of over-represented genes that may have had consensus functions was not performed. This study highlights the importance of the adenoviral entry process upon cellular defensive responses elicited after infection, and should serve as an example of the complexities one may encounter when proposing to utilize detargeted, retargeted or alternative serotype Ad vectors.

In a similar comparative analysis, Martina et al., (2007) characterized the transcriptome response induced by infection of liver cells (Huh-7) with wild-type Ad5 (wtAd5), and compared this response to infections of cells with either FG-Ad5 (deleted for E1 and E3 genes) or a helper dependent Ad (HD-Ad). The HD-Ad contained full length stuffer DNA content, but retained only the ITRs of the original Ad genome (Parks et al., 1996). Based strictly upon fold-change analysis, at 4hpi wtAd5 dysregulated the expression of 905 Huh-7 genes, while the FG-Ad5 dysregulated 552 (61% of wt-Ad5), and the HD-Ad dysregulated 300 (33% of wt-Ad5) Huh-7 genes; at 12hpi Ad5 dysregulated the expression of 1649 Huh-7 genes, while the FG-Ad5 dysregulated expression of 527 (32% of wt-Ad5), and the HD-Ad dysregulated expression of 658 (40% of wt-Ad5) Huh-7 genes. Though the modified Ads dysregulated a sizeable number of genes relative to similar infections with wtAd5, only a small proportion of the genes overlapped based on functional or pathway analysis based methods. Importantly, these investigators also noted that although HD-Ad5 and FG-Ad5 both induced similar levels of transcriptome dysregulation (relative to the wild-type Ad5), only ≈8% of these changes were shared between the two. This lack of overlap indirectly suggests that specific DNA sequences contained within the adenoviral vector may impact upon other cellular responses, although future studies will be required to confirm this hypothesis.

Addressing the issue of Ad induction of cellular transcriptome responses over time, Dorn et al., (2005) assayed global gene expression changes at 12, 32, and 48hpi in HeLa cells after infection with wild-type Ad12. This investigation determined that maximal gene induction of interferon-related genes in Ad12 infected HeLa cells occurs at 12hpi, with strong global repression of most other genes occurring at the later time points. Although this study used a wild-type Ad (in contrast to the ablated vectors mentioned previously), it did demonstrate the importance of time as a factor in the course of the overall cellular innate immune response, with the complexities surrounding the transcriptome responses during a wild-type Ad infection being a sub-aspect of the total response.

These analyses began to identify a core of Ad dysregulated genes, although the architecture of the signaling cascade governing Ad innate immune responses (as reflected by global transcriptome responses) remained largely uncharacterized. Our group recently completed a study in Ad vector infected mouse embryonic fibroblasts (MEFs) in an attempt to identify “Ad gene response” networks (Hartman et al., 2007a). At 16hpi, we detected dysregulation of 22% of the total number of gene probes assessable by the array, 75% of which were upregulated at least 2-fold, and 33% that were upregulated 3-fold or higher. Possibly due to the fact that MEFs express a full Toll-Like Receptor (TLR) repertoire, major Ad-affected gene ontology categories included immune response and antigen processing genes. However, cell cycle, hydrolase activity, heat shock and cytokinesis genes were also over-represented. UV inactivation of the Ad or cyclohexamide treatment of the Ad infected MEF cells did not affect the expression of a subset of these genes, strengthening the notion that Ad virions initiate rapid host transcriptome responses independent of Ad transcription, DNA replication and/or protein expression. In many cases, UV-inactivated virus resulted in higher levels of gene upregulation suggesting Ad genome derived transcription products may actually repress capsid (and/or cytoplasmically sensed “Ad DNA”) -initiated responses.

To indirectly identify transcription factors responsible for Ad-induced or suppressed gene networks, we analyzed the promoters of genes that were dsyregulated by Ad, scanning for over-represented transcription factor binding sites. AP-1, NF-κB, Oct1, Stat1, and Gut-enriched Krueppel-like factor (KLF4) are but a few of the more than 30 different transcription factors we identified as involved in the induction of the Ad innate immune response in MEFs (Hartman et al., 2007a). Supporting these observations and predictions, a recent array analysis by Granberg et al., (2006) found KLF4 to be upregulated within 2hpi of Ad infection of human IMR-90 cells. Future studies will be needed to verify these predictions in human cells, and/or in vivo.

Immune Cell Responses to Ad infection in vitro

In parallel with non-immune cells, multiple groups have demonstrated defense responses after Ad infection of more typical “immune cells”. For example, in cultured dendritic cells (DCs), various Ad vectors have demonstrated inflammatory effects, with studies mainly focused on the induction of NF-κB and upregulation of maturation markers (Hirschowitz et al., 2000; Korst et al., 2002; Varnavski et al., 2003). While Ad induced maturation of DCs was originally a matter of some dispute, the great majority of studies support the view that Ad infection activates DCs of both murine and human lineages (Dietz and Vuk-Pavlovic, 1998; Hirschowitz et al., 2000; Korst et al., 2002; Lyakh et al., 2002; Miller et al., 2003; Varnavski et al., 2003). Different serotypes of Ad appear to retain this capacity although there appears to be differences in the strength of induction from different serotypes. For example, chimpanzee serotype derived Ad vectors demonstrate significant upregulation of both IL-6 and Type I interferons (IFNs) in cultured human and mouse DCs, to a greater degree compared to their Ad5 counterparts (Varnavski et al., 2003). Significantly, this was the first display of a clear induction of Type I IFNα from Ad-infected myeloid DCs. In addition, induction of Type I IFNs by the chimpanzee AdC68 was shown to be independent of viral transcription and replication, and was also found to substantially inhibit vector derived transgene expression (Hensley et al., 2007; Hensley et al., 2005).

Likewise, multiple studies have found macrophages to be affected by adenoviral infection. Exposure of primary mouse alveolar macrophage cells to Ad virions resulted in the rapid induction of multiple cytokines including TNF-α, IL-6, MIP-2, and MIP-1α (Zsengeller et al., 2000). This effect was dependant on adenoviral uptake but not necessarily adenoviral nuclear transduction. Exposure of RAW264.7 mouse macrophages to Ad virions stimulated the rapid upregulation of TNF-α, an effect that was abolished when endosomal trafficking and acidification was prevented by pre-treating Ad infected cells with chloroquine (Zsengeller et al., 2000). A similar result was observed if endosomal escape of the virus was abolished by denaturing the virus at high temperatures prior to its exposure to cells (Trevejo et al., 2001; Zsengeller et al., 2000). Studies using an Ad mutant unable to lyse the endosome confirmed a lack of IRF3 induction following infection of primary mouse macrophages (Nociari et al., 2007). Together, these data indicate that elements of innate immune signaling against Ads in macrophages are dependent on endosomal trafficking and/or rupture.

In human peripheral blood mononuclear cells (PBMCs, innate cell precursors), Ad transduction also results in the transcriptional upregulation of the IL-6 and RANTES genes, as well as steady increases of GM-CSF, MIP-1α, and IL-8 transcripts over a 96-hr. window (Higginbotham et al., 2002). As with macrophages, these responses are maintained despite UV-inactivation of the virus. A recent study showing IL-6 and TNF-α induction in mouse macrophages after exposure to calcium chloride precipitated empty adenoviral capsids (Cerullo et al., 2007) was in contrast to work demonstrating that IRF3, nor other innate immune response genes, are induced in mouse macrophages after exposure to empty capsids (Nociari et al., 2007). Thus, in certain contexts, empty Ad capsids appear capable of inducing immune responses, but this may not be the case for all cell types. Future studies should focus on differentiating the importance of Ad capsids from the DNA content of the capsids in the elaboration of defensive immune responses, in different immune cell types.

Studies of Ad interactions with PRRs in vitro

The discovery and characterization of several different families of pathogen recognition receptors (PRRs) in the past decade have allowed investigators to probe the mechanism surrounding the initiation of cellular defense responses to adenoviral infection in vitro. To this end, we will present the relevant studies and findings involving adenoviral interactions with the Toll Like Receptor (TLR) and RLR (RIG-I Like Receptor) families of PRRs.

Role of TLR receptors in Ad induced innate immunity in vitro

A prominent group of PRRs that has recently been proven to play a pivotal role in the recognition of pathogens is the TLR family. Following TLR activation, a complex effector network is stimulated, resulting in a host defense response that is characterized by elaboration of a cascade of cytokines and chemokines, with subsequent influx of respondent professional immune cells that attempt to eliminate the invading pathogen. These receptors recognize many types of PAMPs found within viral and/or bacterial pathogens, including Triacyl lipopeptides (TLR1), zymosan (TLR2/6), viral dsRNA/poly(I:C) (TLR3), LPS (TLR4), bacterial flagellin (TLR5), viral ssRNA (TLR7/8), bacterial and viral CpG DNA (TLR9) and a profilin-like molecule (TLR11) (reviewed in (Lee and Kim, 2007b)). Mouse Mammary Tumor Virus (MMTV), Respiratory Synticial Virus (RSV), Human Papilloma Virus (HPV), Sendai Virus, and Measles Virus (Bieback et al., 2002; Compton et al., 2003; Czarneski et al., 2003; Haynes et al., 2001; Lopez et al., 2004; Lund et al., 2003; Means et al., 2003; Rassa et al., 2002; Yang et al., 2004) and now adenoviruses are but a few examples of viruses shown to precipitate an immune response through TLR dependent mechanisms.

Several lines of investigation have examined the in vitro involvement of the TLR family in Ad innate immunity. In one such study, Philpott et al., (2004) showed that Ad induced CD86 expression in dendritic cells (DCs) was independent of MyD88 signaling. In contrast, recent evidence has indicated that TLR9, an endosomally localized TLR shown to recognize CpG DNA, was required for the induction of various cytokines in response to Ad infection (see Figure 1). Peritoneal macrophages infected by HD-Ads also produced high levels of IL-6 in a TLR9 dependent manner (Cerullo et al., 2007). Furthermore, HD-Ad, co-precipitated with calcium phosphate (CaPO₄) to achieve higher transduction of cells, induced MCP-1, IL-6 and TNF-α in peritoneal macrophages. However, only IL-6 and TNF-α inductions were TLR9 dependent, signifying a TLR9 independent mechanism for Ad induced MCP-1 production. In parallel, the induction of IL-6 and TNF-α following CaPO₄ facilitated infections by ‘empty capsid’ Ads was also TLR9 independent, suggesting the requirement of Ad genomic DNA for induction of these cytokines through TLR9, as well as the presence of a capsid induced immune response independent of TLR9 signaling.

Figure 1. Intracellular and membrane bound sensors of Adenovirus.

Toll-Like Receptors (TLRs) are membrane bound Pattern Recognition Receptors (PRRs) that recognize conserved molecular structures known as pathogen-associated molecular patterns (PAMPs) found on invading pathogens. (A) Based on the cell type, Ad has been shown to elicit innate immune responses through TLR9 and MyD88 dependent pathways. Ad induced stimulation of these pathways results in rapid activation of over 30 transcription factors including NF-κB, and IRF3, as well as release of numerous cytokines and chemokines. (B) Ablation of either MyD88 or TLR9 does not fully suppress these innate responses, suggesting redundant systems, such as other TLR proteins, or adaptors such as TRIF, exist that stimulate immune response pathways that are independent of MyD88 and/or TLR9. (C) Intracellular sensors of dsDNA, such as RIG-I, are also likely PRRs involved in Ad induced innate signaling in various cell types resulting in similar activation of downstream pathways including activation of IRF3 and release of Type I IFNs. The effector molecules, downstream of the activated receptors, are currently unknown, but are likely shared with pathways known to be activated by other PAMPs. NFκB, nuclear factor kappa B; TRIF, TIR domain-containing adaptor protein inducing interferon beta; MyD88, myeloid differentiation primary response gene-88; IRF, interferon response factor; IFN, interferon; IL, interleukin; MCP, monocyte chemoattractant protein; RANTES, regulated on activation, normal T-cell expressed and secreted; MIP, macrophage inflammatory protein; IP, interferon gamma inducible protein; MAVS , mitochondrial antiviral signaling.

Iacobelli-Martinez and Nemerow, (2007) showed that Ad5 virus pseudotyped with the fiber from Ad16 (Ad5.F16, a CD46 receptor utilizing Ad) induced higher levels of Type I IFNs in peripheral blood mononuclear cells (PBMCs) than did the CAR utilizing Ad5. In a subsequent analysis, their data indicated that in HeLa cells, Ad5.F16 induced Type I IFN secretion and NF-κB activation to a greater level than Ad5, and that these responses were TLR9, but not TLR7 dependent. These results demonstrate that the route of Ad entry may substantially contribute to the type of innate immune response initiated by the TLR pathways. Finally, while Ad5 induction of Type I IFNs was mediated by TLR9 in a MyD88 dependent manner in plasmacytoid DCs (pDCs), it was induced in a MyD88 independent manner in conventional DCs (cDCs), and peritoneal macrophages (Basner-Tschakarjan et al., 2006; Zhu et al., 2007). These studies indicate a cell type specific role for TLR pathways in the innate immune response to Ad infection.

These cell-type specificities are mirrored by findings investigating the role of the TLR adaptor known as TRIF. TRIF is known to play a significant role in the downstream signaling of TLR3 and TLR4. Nociari et al., (2007) demonstrated that Ad induction of IL-6 is significantly diminished in TRIF deficient macrophages compared to wild-type counterparts, but that it is only partially diminished in TRIF deficient cDCs. Furthermore, the Ad-mediated induction of IFNα, IP-10, and RANTES in macrophages or cDCs is not affected by TRIF deficiency. We demonstrated the MyD88 and TRIF dependent Ad induction of cytokines in human A549 cells, bolstering the view that TLR adaptors figure prominently in the cellular immune responses to Ad vector infection (Hartman et al., 2007a; Hartman et al., 2007b).

Role of the RLR receptor RIG-I and Ad DNA in Ad induced innate immunity in vitro

The intracellular receptors RIG-I and MDA-5, serve as dsRNA detectors that together recognize the genomes of encephalomyocarditis virus, Newcastle disease virus, Sendai virus, influenza virus, and vesicular stomatitis virus independent of TLR regulated pathways (Gitlin et al., 2006; Kato et al., 2005; Kato et al., 2006). This recognition typically results in the recruitment of, or interaction with, the mitochondrial MAVS protein (also known as IPS1, VISA, or Cardif), stimulation of TANK-related kinases, and subsequent phosphorylation and activation of Type I IFN specific transcription factors known as interferon response factors, specifically IRF3 and IRF7 (Kumar et al., 2006). Recent evidence suggests that a similar TLR independent mechanism exists for the recognition of dsDNA, using Ad as the model system. As alluded to earlier, in murine bone marrow derived DCs (BMDCs) or bone marrow derived macrophages (BMDMs), Ad infection resulted in MyD88 and TRIF independent increases in CD86, CD40 and CD80 expressing cells, transcription of a variety of cytokines and chemokines, and activation of IRF3, presumably due to Ad capsid penetration of the endosome, and exposure of the Ad dsDNA genome to a cytoplasmic sensor (Nociari et al., 2007). A recent article by Cheng et al., (2007) describes, for the first time, a direct role for RIG-I in sensing intracellular dsDNA. Contrary to previous reports (Ishii et al., 2006; Kumar et al., 2006; Sun et al., 2006), the data in this study indicate that expression of dominant negative RIG-IC (or knocking-down MAVS using siRNA) completely suppresses IFNβ promoter activity following poly(dAT:dAT) or poly(I:C) treatment in the human hepatoma cell line Huh-7. In an experiment specific for Ad, Cheng et al., (2007) also showed that transfection of Ad DNA into Huh-7 cells containing a loss of function point mutation in RIG-I (Huh-7.5.1) also resulted in loss of IFNβ transcriptional induction. These results support evidence previously published by our lab implicating MAVS in Ad sensing. We specifically showed that shRNA mediated knock-down of MAVS protein expression in human lung epithelial A549 cells suppressed IFNβ promoter activity following Ad infection (Hartman et al., 2007a).

Thus, while there is growing evidence of the role played by TLR and RLR families in the sensing of adenoviral infection, future studies are needed to ascertain the identity of other TLR and RLR family members involved, the cell type and response specificities involved, as well as the identity of other PRR families that may participate in sensing adenoviruses.

AD INDUCED INNATE IMMUNITY IN VIVO

While in vivo cellular transduction is a potentially more complicated process than in vitro studies of cellular transduction, (i.e. involving putative binding or coating of Ad particles with several blood-borne factors prior to interactions with several different cell types) initial forays investigating the in vivo cellular response to Ads have nonetheless noted many similarities to in vitro investigations. Furthermore, the mechanism of innate immunity in vivo must also be considered in the context of different PRR repertoires present within different cell types in vivo. Thus, this portion of the review will focus on the roles of 1) extracellular factors, 2) cellular factors, and 3) membrane-bound PRRs (specifically TLRs) during Ad induced innate immune responses in vivo.

Role of extracellular molecules in Ad induced innate immunity in vivo

To recognize and defend against different pathogens in various settings throughout the body, mammals have evolved a wide array of extracellular PRRs that serve to identify molecular structures conserved among various pathogens, structures referred to as pathogen-associated molecular patterns (PAMPs). Multiple types of extracellular PRRs, including pentraxin and collectin family members, as well as protein members of the complement pathways, both classical (C1q, C4) and alternative (Factor B, Factor D), opsonize pathogens to facilitate their phagocytosis and destruction, many times via interactions with cellular receptors (C1qR, CR1, CR2, CR3, CR4). In the context of adenoviral infection in vivo, Ad has been determined to interact with surfactant and complement in addition to interacting with pre-existing immunoglobulins.

Since breaching the keratinized epithelium found within the skin is a challenge for most viruses, entry is usually achieved by an alternate route, most notably through mucosal membranes. Since Group C human Ads primarily infect the upper respiratory tract, it was parsimonious to postulate an Ad interaction with surfactant, a family of C-type collections (C-type lections) primarily produced by the epithelial cells of the lung. In a 1999 study, Harrod et al., (1999) demonstrated greater inflammation and decreased lung clearance after intra-tracheal administration of FG-Ad5 in Surfactant-A (SP-A) deficient mice, as compared to their wild-type counterparts. Ad infected, surfactant deficient mice displayed greater levels of epithelial damage, higher numbers of infiltrating inflammatory cells, and higher levels of IL-6, TNF-α, and IL-1β in bronchoalveolar lavage fluid. Interestingly, increased levels of Ad vector derived DNA were found in the lungs of Ad infected SP-A deficient mice, relative to Ad infected normal mice. Other studies have noted this interaction, and documented the disruption of surfactant homeostasis after administration of Ad in a murine lung model (Ross et al., 1995; Zsengeller et al., 1997). Thus, SP-A appears to block Ad vectors in vivo, reducing their inflammatory profile through unproven but likely opsonizing effects.

One of the benefits of Ad based gene transfer vectors is their ability to successfully transduce large amounts of hepatic tissue following intravenous injection (Einfeld et al., 2001; Nicklin et al., 2005; Nicol et al., 2004; Shayakhmetov et al., 2005a; Shayakhmetov et al., 2004). This requires levels of virus several logs higher than would normally be found during a wild-type infection, a situation that likely amplifies Ad interactions with humoral proteins and PRR’s that have evolved to respond to much lower viral concentrations. This model, however, also allows detection of host-pathogen interactions that might not be detectable (but no less relevant) during wild-type infections. For example, our group and others have shown that Ad4 or Ad5 can activate the complement system in the presence or absence of pre-existing anti-Ad immunoglobulins through either the classical pathway, or through direct interactions with proteins of the alternative pathway respectively (Cichon et al., 2001; Jiang et al., 2004). Shayakhmetov et al., (2005a) demonstrated that fiber knob interactions with both blood factor IX and C4 binding protein were possible in vitro, and that ablating this interaction reduced serum levels of IL-6 and IFNγ 6hpi of this mutant virus in vivo.

In array based analyses of Ad induced liver transcriptome responses (comparing Ad injected wild-type and complement factor C3 (C3-KO) mice) Kiang et al., (2006) reported that of all liver genes dsyregulated 6 hours after Ad injection, 3 in 5 Ad upregulated genes, and 1 in 5 Ad downregulated genes were in part complement dependent for these transcriptional responses. Like those identified in MyD88 deficient mice (see below), the Ad induced, complement dependent genes identified in this analysis were largely representative of NF-κB, apoptosis, Th1, and IL-6 signaling pathways. This may indicate an interaction between complement and TLR pathways that has yet to be identified, but of potential relevance to several pathogens in addition to Ads per se. Supporting this notion, a subset of Ad induced plasma cytokines were also dependent on MyD88, and complement functionality. These include: KC, IL-5, G-CSF, GM-CSF, and IL-6. Future studies dissecting potential Ad complement TLR system interactions, as well as efforts to capitalize on this knowledge to improve the safety profile of Ad vectors in vivo are now underway.

The host has several other humoral innate immune mechanisms poised to rapidly identify, sequester, and neutralize low levels of pathogens not previously encountered. These “hair-triggered” guardians include non-specific, pre-existing “natural” immunoglobulins, lysozymes, C-reactive protein, collectins, pentraxins, serum amyloid, as well as mannose binding proteins. These molecules can bind pathogens and prevent their entry into host cells as well as opsonize them to facilitate their phagocytic digestion by innate immune cells. However, the roles of these systems in Ad mediated immune responses remain currently unknown.

Finally, most viruses, including Ads, elicit neutralizing, Ad specific antibodies after primary inoculation into a host (Jooss and Chirmule, 2003; Malkevitch et al., 2003; Santra et al., 2005). Binding of Ad vectors by specific immunoglobulins facilitates rapid clearance of the vector by cellular elements of the innate immune response, likely including Kupffer cells, and other cells of the reticuloendothelial system (RES), as well as activation of the classical arm of the complement pathway (Cichon et al., 2001). Recently, it has been shown that that Ad vectors may be more toxic in the presence of anti-Ad antibodies than in their absence, the etiology of which is unclear at the moment (Varnavski et al., 2005; Varnavski et al., 2002).

The limited studies preformed thus far indicate that adenoviral virions interact with and are opsonized by surfactant-A in the lungs and by complement in the bloodstream. Opsonization by surfactant facilitates viral clearances and results in a less inflammatory phenotype while opsonization by complement results in a more inflammatory phenotype in an intravenous setting. In this same vain, binding by pre-existing immunoglobulins results in a more rapid uptake by the RES and greater inflammatory effects in vivo. Thus, the effect of opsonization by extracellular factors depends on the type of extracellular factor involved and the context of infection. While only a few extracellular systems have been studied, it is likely that adenoviral particles interact with multiple types of extracellular PRRs to generate the particular innate and inflammatory responses observed in vivo.

Cellular elements of the in vivo innate immune response to Ads

The first investigations of the in vivo innate immune response to adenoviral vectors proceeded from gene therapy based approaches that used intravenous administration of type 5 adenoviral vectors to highly infect mouse hepatic tissue. Several studies have shown that Ad transduction into the murine liver results in the upregulation of many of the same interferon-related genes found in vitro (e.g. MCP-1, IP-10, RANTES, MIP-2, etc.) reaching peaks at 6–12 hours post-injection with a return to near-baseline by 24 hours post injection. Infection with UV inactivated and HD-Ad vectors showed similar gene induction patterns, however, induction of a second phase response at 5dpi and 7dpi required Ad derived viral gene transcription (Liu et al., 2003; Zaiss et al., 2002). Our studies using a microarray based approach to analyze in vivo Ad induced immune responses demonstrated that a large portion of liver tissue derived transcripts (∼15%) are affected by adenoviral infection, reaching peak levels at 6hpi. While many of the Ad affected genes belong to innate immune families, other groups of Ad-dysregulated genes include RNA regulatory, apoptosis, lysosome, and endocytic families of genes (Hartman et al., 2007b).

Whether or not Ad induced immune activation is correlated with hepatocyte transduction is currently unresolved, and use of hepatically detargeted Ad vectors has not provided much clarification. While some groups have reported a lack of hepatic transduction with penton RGD and fiber CAR binding double mutants in certain strains of mice and rats, others have reported high levels of in vivo hepatic transduction from these same viruses in other strains of mice, rats and in primates (Einfeld et al., 2001; Martin et al., 2003; Ni et al., 2005; Nicol et al., 2004; Smith et al., 2003a; Smith et al., 2003b). Recent evidence suggests that the HSPG binding motif present in the Ad5 fiber is sufficient for in vivo hepatic transduction in the absence of both CAR and RGD motifs (Kritz et al., 2007).

The emerging consensus favors the importance of the HSPG motif and potentially the size of the fiber used relative to Ad upregulated innate immune response genes. For instance, in DBA/2 mice infected with RGD-ablated vectors, significantly lower transcription levels of IP-10 and MIP-2 were noted, as compared to those receiving a similar dose of wild-type Ad vectors (Liu et al., 2003). In C57BL/6 mice, injection with short-shafted Ad vectors, which display fibers without HSPG motifs, also led to decreased liver transduction and significantly lower transcription levels of MIP-2 and IL-1α, in comparison to Ad vectors displaying wild type fibers (Shayakhmetov et al., 2004). Intriguingly, in mice injected with these vectors, IP-10 upregulation was similar but delayed, leading the authors to speculate that induction of this gene is not dependent upon hepatocyte transduction per se. Elimination of resident macrophages by pretreatment of mice with agents such as gadolinium chloride does not change upregulation of these genes in vivo, but in fact accentuates their upregulation, potentially through the increased hepatic transduction afforded by macrophage depletion (Liu et al., 2003). These findings provide evidence for the importance of stromal cell transduction in Ad induced innate immune responses.

The systemic elaboration of several cytokines and chemokines after Ad injection accompanies the induction of innate immune genes. The cellular origin of these systemic cytokines is not fully known, although evidence obtained thus far suggests prominent roles played by innate immune cells specifically, such as macrophages and dendritic cells. The importance of macrophages is most easily demonstrated in studying Ad’s natural tropism of the respiratory tract. In an in vivo study, Zsengeller et al., (2000) showed rapid accumulation of Ad5 vectors in alveolar macrophages 10 minutes after vector administration in the murine upper respiratory tract. This effect was strongly associated with the induction of inflammatory cytokines and chemokines (such as IL-6, TNF-α, MIP-2, and MIP-1α) produced by alveolar macrophages.

Other recent studies have shown the importance of dendritic cells in the Ad mediated induction of cytokines in vivo. Using intravenous injection of fiber modified Ad vectors, Koizumi et al., (2007) found that systemic induction of IL-6 (and to a lesser extent IL-12) correlated with high vector uptake by (but low transduction of) splenic cells. When several splenic cell types were assessed after Ad injection, only conventional B220-CD11c+ DCs were found to have significant induction of IL-6 and IL-12 transcription. However, macrophage populations were not assessed in this study, thus it is unclear if resident or trafficking nonresident macrophage populations are also responsible for cytokine production in the spleen.

Kupffer cells are the primary macrophages resident in the liver, and have the primary function of sequestering pathogens that enter the liver vasculature. Kupffer cells are known to play a large role in the sequestration of intravenously injected Ad vectors, thereby preventing hepatocyte transduction unless saturated by overwhelming Ad doses. Depletion of Kupffer cells prior to Ad injection results in correction of the non-linear dose threshold effect, i.e. lower dosage Ad infusions are more capable of efficiently transducing greater numbers of hepatocytes (Lieber et al., 1997; Worgall et al., 1997; Ziegler et al., 2002). This improved ability to transduce hepatocytes is accompanied by a decrease in Ad induced endothelial cell activation and/or cytokine release (Schiedner et al., 2003; Snoeys et al., 2006). Kupffer cell sequestration of Ad particles has also been confirmed in baboons, with injection of 1.2 × 10¹² particles/kg providing evidence of both Kupffer cell sequestration and cytokine induction (IL-6) but also some limited hepatocyte transduction. A 10 fold higher dose (1.2 × 10¹³ particles/kg) resulted in widespread liver transduction by the vector, but this dose was also lethal to the treated animals.

In addition to Kupffer cells, endothelial cells also sequester large amounts of Ad vectors following injection of higher doses in primates. These patterns of Ad vector sequestration and toxicity mimic those noted in mice, at near equivalent dosages per kilogram (Morral et al., 2002; Tao et al., 2001). Though technically not cells, blood platelets may figure prominently in these responses. It has been shown that Ads can infect platelets in vitro, and mice intravenously injected with Ad vectors can be found to contain Ads within their platelets as well (Faraday et al., 1999; Gillitzer et al., 2005; Stone et al., 2007). The significance of Ad interactions with platelets (i.e.: upon subsequent Ad induced innate or adaptive immune responses) awaits future investigations.

Finally, several groups have shown that coating of Ad using Poly-ethylene glycol (PEG) prior to injection can effectively reduce the production of several inflammatory cytokines such as IL-6, IL-12, and TNF-α, as well as Ad induced thrombocytopenia. Studies have also shown that “PEGylation” results in a reduced uptake of Ad vectors by Kupffer cells and macrophages in vitro (Croyle et al., 2002; Croyle et al., 2005; O'Riordan et al., 1999) These complexities point to future studies required to dissect out and assess the significance of the multi-faceted interactions occurring between Ads and blood proteins, platelets, macrophages, endothelial cells and respective parenchymal cells.

Role of cellular TLRs in Ad induced innate immunity in vivo

Unlike in vitro studies, to our knowledge there have been no published studies to date investigating the effect of other cellular PRRs (such as RLR or NOD family members) in Ad immunity in vivo. This area constitutes a gap in adenoviral immunity which future studies should address. As with in vitro investigation of TLRs, there have been few studies directly assessing the interaction of Ad and TLRs.

The first study investigating Ad interaction with TLRs in vivo was actually carried out before the discovery of TLRs. In 1991, Ginsberg et al., (1991) investigated the molecular pathogenesis of adenoviral induced pneumonia in mice, using several inbred mouse strains. They discovered that although adenovirus did not replicate in mouse lungs, pathogenic pneumonia would develop due to the inflammation induced by the high concentration of viral inoculum used. Curiously, they noted the highest level of inflammation in C57BL/6N and CBA/N mice and the lowest levels of inflammation in C57BL/10ScN mice, a strain with an endogenous deletion of TLR4 (Poltorak et al., 1998). Eight years later, Thorne et al., (1999) found that Ad infected C3HeB/FeJ mice (wild-type TLR4) elicited much greater Ad-induced lung inflammatory responses when compared to congenic, C3H/HeJ (TLR4 defective) mice. These studies implicate TLR4 as being involved in the innate immune response to Ad in the lungs, although the molecular characterizations of these interactions remain unexplored. In contrast, a different study showed that C3H/HeJ (TLR4 defective) mice intravenously injected with high doses of Ad exhibited some inflammatory reactions noted after high dose intravenous Ad injections of wild-type mice, although these limited studies did not include tandem comparisons to TLR4 wild-type congenic mice (Lieber et al., 1997).

Using a broader approach, our group recently completed a set of experiments aimed at delineating the role of the critical TLR adaptor MyD88 subsequent to intravenous administration of Ad vectors. For example, we defined the MyD88 dependent aspects of the Ad induced liver, transcriptome profile in vivo (Hartman et al., 2007b). We found that 15% of all assessable gene probes represented on the array were significantly dysregulated by Ad in the livers of wild type animals. In comparison, livers of Ad treated, MyD88 deficient mice exhibited dysregulation of only 4.4% of assessable gene probes. In this analysis, a significant number of TLR, MAPK, apoptosis and complement activation genes were identified as MyD88 dependent, Ad upregulated genes. The Ad induction of plasma cytokines and chemokines, such as KC, G-CSF, IL-6 and MCP-1 were also found to be partially MyD88 dependent at 1hpi, and Ad inductions of plasma MCP-1, MIP-1α, IL-5, IL-6, IL-12(p40), G-CSF, GM-CSF and RANTES were partially MyD88 dependent at 6hpi. Indirectly supporting this work, Shayakhmetov et al., (2005b) showed reduced liver toxicity as measured by ALT, as well as reduced plasma levels of IL-1β and IL-6 following injection of Ad into IL-1R1 deficient mice, a result also likely mediated by MyD88, as the IL-1R also utilizes MyD88 as an intra-cellular signaling adaptor.

Validating the importance of innate immune responses in directing downstream adaptive immune responses to pathogens, we have also demonstrated that in Ad injected MyD88 KO mice, anti-Ad IgG and transgene-specific IFNγ T-cell responses were significantly lower in Ad injected MyD88 deficient mice, as compared to Ad injected controls. TLR9 has also been implicated as a specific TLR involved in the immune response to Ad in vivo. In TLR9 deficient mice, induction of IL-6, IL-12 at 6hpi and IFNα at 12hpi, has been shown to be partially dependent on TLR9. Notably, the induction of KC and MCP-1 was not dependent on TLR9, and verified these cytokines’ dependence upon MyD88 functionality (Cerullo et al., 2007). In vitro work by a different group demonstrated that Ad infection induced IFN-α in a TLR9-MyD88 dependant fashion in pDCs but not other cell types (Zhu et al., 2007).

At present, the collective knowledge surrounding Ad interactions with alternate TLRs or intracellular receptors in vivo is relatively lacking. As mentioned, it is likely that other TLRs, along with TLR9, are also required for the complete induction of the innate immune response to Ads, based on the induction of multiple cytokines in TLR9 deficient mice, and the continued dysregulation of Ad induced immune pathways in Ad injected, MyD88 deficient animals. Evidence in vitro indicates that the RIG-I/MAVS pathway could be involved in these responses. These findings do support the notion that exploring and understanding the nuances of Ad vector interactions with the innate immune system can have far reaching implications, in this instance providing justification for studies attempting to understand how TLR responses to Ad vectors impact upon Ad vector efficacy in vaccine applications. Finally, Ad interactions with other cellular PRRs have not yet been investigated in vivo, further underscoring the large gaps in our knowledge of the mechanisms underlying adenoviral innate immunity in vivo.

Conclusion

Modified Ads are commonly used for a wide-range of scientific and therapeutic endeavors, such as vaccine platforms, oncolytic agents, or gene transfer vectors for targeted applications both in vitro and in vivo. This wide applicability makes understanding the innate immune response to these viruses particularly important for improving efficacy in each of these types of endeavors. This review highlights that the nature of Ad induced immune responses are complex, despite the limitations of current technologies used to define these responses. As improved technologies emerge, and more sophisticated animal models become available (i.e.: those with multiple gene modifications simultaneously present in a single animal) these complexities will likely broaden.

Collectively, the data reviewed in this manuscript begins to identify the scope of the Ad induced innate immune response, as it involves extracellular, intracellular, and membrane bound receptors, that make up a complex network of immune response pathways that have co-evolved to rapidly respond to invading pathogens such as Ads. These responses vary based on the type of entry Ad initiates, the type of DNA incorporated into the virion, as well as the extracellular milieu encountered by the virion before infection.of various cell types. As a consequence, these variables are both capable of limiting the efficacy of Ad based gene transfer vectors, as well as augmenting the efficacy of Ad based vaccine vectors. Furthermore, as pseudotyped Ads, as well as alternative serotype Ad based vectors become more widely utilized, it will also be likely that each of the vectors will not only offer unique insights, but also encounter new limitations , based in large part on their altered interactions with the innate immune system.

Because of these considerations, the Ad vector community specifically, and the gene transfer community at large, must continue to vigorously identify, investigate, and understand the innate immune systems that may be triggered upon introduction of foreign transgenes by any gene transfer platform. Without these types of investigations, future gene therapy trials may again encounter untoward or unforeseen outcomes attributable to innate immune responses, while therapeutic vaccination and cancer therapy trials utilizing vectors may show decreased efficacy secondary to misdirected and/or vector-limited immune responses. Continued efforts to understand the innate immune response will lead to greater wisdom in the use of all gene transfer based vectors, enabling future investigators to more effectively and pro-actively manipulate the innate immune system, to either minimize the toxicity, or accentuate the immune response desired after Ad vector mediated gene transfer specifically, and/or via other vectors generally.

Table 1.

Summary of in vitro analyses of Ad induced responses in infinite and finite lifespan cells derived from various tissues from both human and murine species

Cell Line	Cell Type	Species	Type
HeLa	Adenocarcinma epithelial cells	Human	Immortal
M21	Melanoma derived	Human	Immortal
HB2	Mammary epithelial cells	Human	Immortal
A549	Lung carcinoma cell line	Human	Immortal
Huh-7	Hepatoma cell line	Human	Immortal
PBMCs	Peripheral blood mononuclear cells	Human	Primary Culture
Myeloid DCs	Dendritic cells derived from PBMCs	Human	Primary Culture
IMR-90	Lung fibroblast	Human	Primary Culture
RAW264.7	Ascites derived macrophage	Murine	Immortal
REC	Renal epithelial cells	Murine	Immortal
TGP61	Pancreatic insulinoma	Murine	Immortal
BMDCs	Bone marrow derived dendritic cells	Murine	Primary Culture
MEF	Mouse embryonic fibroblasts	Murine	Primary Culture
PM	Peritoneal macrophages	Murine	Primary Culture
pDCs	Plasmacytoid DCs	Murine	Primary Culture
cDCs	Conventional DCs	Murine	Primary Culture
KC	Hepatic Kupffer cells	Murine	Primary Culture

Cell Line	Time Frame	Ad Induced Immune Response	References
HeLa	30mpi; 2, 16, 48hpi	ERK and MKK activation, IL-8 transcription (30mpi); NFκB activation (6hpi); RANTES (16hpi), IFNβ induction (48hpi)	Bruder et al., 1997; Bowen et al., 2002; Iacobelli-Martinez et al., 2007
*HeLa	12, 32, 48hpi	Array analysis revealed max induction of IFN regulated genes at 12hpi	Dorn et al., 2005
M21	30hpi	Array analysis finding 89 gene changes common to Ad5, Ad5-RGD mutant and Ad5/3 chimera infection	Volk et al., 2005
HB2	24hpi, 72hpi	Array analysis finding induction of 24 (24hpi), and 96 (72hpi) genes including IL-8, IL-6, NFκB	Scibetta et al., 2005
A549	20mpi, 16hpi, 24hpi	Activation of MAPK(20mpi), NFκB, AP-1 and Oct-1 (16hpi); MAVS, MyD88, TRIF required for IFNβ promoter activity	Hartman et al., 2007; Tamanini et al., 2006
Huh-7	4, 12, 24hpi	Induction of IFNβ transcript; array approach: WT-Ad5, HD-Ad5, FG-Ad5 shows differentially modulated pathways	Cheng et al., 2007, Martina et al., 2007
PBMCs	0 – 96hpi	TNFα, IL-1β, GMCSF - gradual increase; IL-6, RANTES- Max at 30hpi; IL-8- Max at 72hpi; 24hpi IFNα induction (24hpi)	Higginbotham et al., 2002, Iacobelli-Martinez et al., 2007
Myeloid DCs	24, 48hpi	IFNα induction; Increased DC surface markers CD40, CD80, CD86	Varnavski et al., 2003
IMR-90	2hpi, 24hpi	Array analyses identifying KLF4 induction (2hpi) and 3-fold induction of 4.2% of assessed genes (24hpi)	Stilwell et al., 2004; Granberg et al., 2006
RAW264.7	2hpi	TNFα release, blocked by chloroquine and heat denaturation	Zsengeller et al., 2000
REC	10mpi; 6hpi	PI3K dependent Akt activation; ERK and p38 activation (10mpi); IP-10 induction, NFκB activation (6hpi)	Tibbles et al., 2002; Borgland et al., 2000; Liu et al., 2003/2005
TGP61	1hpi, 6hpi	Induction of MIP-2 (1hpi), RANTES, IP-10 (6hpi) transcription	Muruve et al., 1999
BMDCs	18hpi, 36hpi, 48hpi	IFNα, IFNβ, IL-6, IL-12(p40), IL-1β induction; Increased DC surface markers CD86, CD54,CD40; IRF3 activation	Hirschowitz et al., 2000; Korst et al., 2002; Lyakh et al., 2002;
	5hpi	Induction of TNFα, IFNβ, CCL5, CCL1, CCL4, and IP-10 transcription	Cerullo et al., 2007, Nociari et al., 2007
MEF	16hpi	Array analysis revealed induction of 22% of assessed genes, induction of 30 TFs and MIP1a, IL-6, KC and RANTES	Hartman et al., 2007
PM	6hpi, 18hpi	IL-6, TNFα induction (6hpi); IFNα, IFNβ induction (18hpi)	Cerullo et al., 2007; Zhu et al., 2007
pDCs	18hpi	IFNα, IFNβ induction	Zhu et al., 2007
cDCs	18hpi	IFNα, IFNβ induction	Zhu et al., 2007
KC	18hpi	IFNα, IFNβ induction	Zhu et al., 2007

^*Wild-type Ad12 was used to infect HeLa cells as opposed to wild-type Ad5 or FG-Ad5

Table 2.

Summary of in vivo analyses of Ad induced innate immune responses

Strain	Genetic Background	Challenge	Virus	Dose	Time Frame	Ad induced immune resposne	Reference
BALB/c	Wild-Type	Intratracheal	FG-Ad5LacZ	4 × 1010 OPU (v.p.)	0.5, 3, 6 hours	Induction of IL-6, TNFα, MIP2, MIP1α in BALF; inductio of trascripts in AM	Zsengeller et al., 2000
C57BL/6	Wild-Type	Intravenous	FG-Ad5[E1-]	5 × 107 p.f.u.	3 days	cDNA array approach identified several induced transcripts	Haberberger et al., 2000
C57BL/6	Wild-Type	Intravenous	FG-Ad5/9 short shaft	1 × 1011 v.p.	0.5, 6, 24hpi	Reduced MIP2, IL1α transcripts; delayed IP-10 induction; Reduced plasma TNFα, IL-6, MCP1, IFNγ levels	Shayakhmetov et. al, 2004
C57BL/6	Wild-Type	Intravenous	FG-Ad5-Fiber Modified (7K)	1 × 1011 v.p.	3, 24, 48hpi	Induction of IL-6 correlates with splenic transduction	Koizumi et al., 2007
C57BL/6	Wild-Type	Intravenous	HD-Ad; FG-Ad5βGal - PEGylated	8 × 1010 v.p.	6hpi	Reduced serum IL-6, IL-12, TNFα induction; Reduced thrombocytopenia	Croyle et al., 2005
C57BL/6	Wild-Type	Intravenous	FG-Ad5LacZ	1.5 × 1011 v.p.	1, 6, 24hpi	Induction of serum KC (1hpi), IL-5, GCSF, GM-CSF, IL-6, IFNy, IL-1β, IL-12(p40) RANTES; Array approach identifies 15–19% of assessable liver genes dysregulated at 6hpi	Kiang et al., 2006; Hartman et al., 2007
C57BL/6	Wild-Type	Intravenous	FG-Ad5 + TNFα antibody	1 × 1011 v.p.	0.5, 6, 24hpi	Reduced plasma IL-6 levels (6hpi)	Shayakhmetov et. al, 2005
C57BL/6	Wild-Type	Intravenous	FG-Ad5 + IL-1α/β antibodies	1 × 1011 v.p.	0.5, 6, 24hpi	Reduced IL-1α, IL-1β transcripts; reduced plasma IL-6 (6hpi); reduced ALT	Shayakhmetov et. al, 2005
C57BL/6	Wild-Type	Intravenous	FG-AdLacZ	2 × 109 p.f.u.	6hpi, 12hpi	Induction of serum IFNα, IL-6	Zhu et al., 2007
C57BL6	Wild-Type	Intravenous	FG-Ad5mut no C4/FIX binding	1 × 1011 v.p.	6hpi	Reduced serum IL-6 and IFNγ induction	Shayakhmetov et. al, 2005
DBA/2	Wild-Type	Intravenous	FG-Ad5Luc	1 × 1011 v.p.	1hpi - 7dpi	Induction of biphasic MIP1β, MIP1α, MIP-2, MCP-1, IP-10, TNFα trancripts, induction of sustained RANTES transcript; Neutrophil Infiltration (1hpi)	Liu et al., 2003
DBA/2	Wild-Type	Intravenous	FG-Ad5LacZ	109 – 1011 v.p.	1, 6, 24hpi	Dose dependent induction of RANTES, MIP1β, MIP2, MCP-1, IP-10 transcripts	Zaiss et al., 2002
DBA/2	Wild-Type	Intravenous	FG-Ad5[E1-] - RGD mutant	1 × 1011 v.p.	1, 6, 24hpi	Reduced IP-10 and MIP-2 transcript induction; reduced Neutrophil Infiltration	Liu et al., 2003
Several	Wild-Type	Intranasal	Wild Type Ad5	1 × 1010 p.f.u.	1 – 7 days	Induction of lung and plasma TNFα, IL-1 and IL-6; induction of IL-1β transcript in lung	Ginsberg et al., 1991
C57BL6/129s	TLR9 Deficient	Intravenous	HD-Ad5LacZ	5 × 1012; 1 × 1013 v.p./kg	6hpi	Plasma IL-6, IL-12 reduced	Cerullo et al., 2007
C57BL6/129s	TLR9 Deficient	Intravenous	FG-Ad5LacZ	2 × 109 p.f.u.	12hpi	Reduced IFNα	Zhu et al., 2007
C3H/HeJ	TLR4 Deficient	Intratracheal	FG-Ad5LacZ	1 × 1010 v.p.	3, 12, 24hpi	Reduced neutrophil infiltration, TNFα, IL-6 in BALF compared to control C3HeB/FeJ	Thorne et al., 1999
C3H/HeJ	TLR4 Deficient	Intravenous	FG-Ad5hAAT	1 × 1010 TU	0 – 96hpi	Induction of TNFα and IL-6, however not compared to control C3HeB/FeJ	Lieber et al., 1997
C57BL/10ScN	TLR4 Deficient	Intranasal	Wild Type Ad5	1 × 1010 p.f.u.	1 – 7 days	Lower overall inflammatory response	Ginsberg et al., 1991
129/J	SP-A Deficient	Intratracheal	FG-Ad5Luc	1 × 109 p.f.u.	6, 24hpi	Increased TNFα, IL-6, MCP1, MIP1α, MIP2 in lung homogonate; increased TNFα, IL-1β, IL-6 in BALF	Harrod et al., 299
C57BL6/129s	MyD88 Deficient	Intravenous	FG-Ad5LacZ	1.5 × 1011 v.p.	1hpi, 6hpi	Array analysis revealed almost 50% Ad dysregulated genes are MyD88 dependent; plasma KC (1hpi), GM-CSF, GCSF, IL-5, IL-6, IL-12(p40), MIP1α, MCP1, RANTES (6hpi) reduced	Hartman et al., 2007
C57BL6/129s	MyD88 Deficient	Intravenous	FG-Ad5LacZ	2 × 109 p.f.u.	12hpi	Reduced IFNα	Zhu et al., 2007
C57BL6/129s	IL-1R Deficient	Intravenous	FG-Ad5βGal	1 × 1011 v.p.	0.5, 6 24hpi	Altered IL-1α, TNFα, MIP2 gene expression, Reduced plasma IL-1β, IL-6, ALT levels	Shayakhmetov et. al, 2005
129Sv	IFNαβR Deficient	Intravenous	FG-Ad5LacZ	2 × 109 p.f.u.	6hpi	Prolonged transgene expression, reduced neutralizing Ab titer	Zhu et al., 2007
C57BL6/129s	C3-Deficient	Intravenous	FG-Ad5LacZ	1.5 × 1011 v.p.	6hpi	Array analysis revealed 60% Ad upregulated, and 20% Ad downregulated genes are complement dependent; reduced plasma KC (1hpi), IL-5, GCSF, GM-CSF and IL-6 compared to wild-type controls	Kiang et al., 2006

FG, First Generation; HD, Helper-Dependent; p.f.u., plaque forming units; v.p., viral particles; TU, transducing units; OPU, optical particle units