Implications of the innate immune response to adenovirus and adenoviral vectors

Abstract

Adenovirus (AdV) is a common cause of respiratory illness in both children and adults. Respiratory symptoms can range from those of the common cold to severe pneumonia. Infection can also cause significant disease in the immunocompromised and among immunocompetent subjects in close quarters. Fortunately, infection with AdV in the normal host is generally mild. This is one reason why its initial use as a gene-therapy vector appeared to be so promising. Unfortunately, both innate and adaptive responses to the virus have limited the development of AdV vectors as a tool of gene therapy by increasing toxicity and limiting duration of transgene expression. This article will focus on the innate immune response to infection with wild-type AdV and exposure to AdV gene-therapy vectors. As much of the known information relates to the pulmonary inflammatory response, this organ system will be emphasized. This article will also discuss how that understanding has led to the creation of new vectors for use in gene therapy.

Up until the advent of gene therapy, interest in adenovirus (AdV) and its related diseases was primarily the concern of neonatologists and those involved in the care of immunosuppressed patients. Thus, the focus was on vaccination and early attempts at therapeutics. This changed with the introduction of the idea that “good DNA could be used to replace defective DNA in those who suffer from genetic defects”, first proposed by Rogers in 1970 [1]. However, it was not until 20 years later that the first successful human implementation of this new science occurred [2]. After this initial success, this promising new field continued to blossom for almost a decade until the unfortunate death of a volunteer in a gene-therapy research trial [3]. The volunteer was an 18-year-old patient with ornithine transcarbamylase deficiency. Although there were other potential confounding factors, he was the first research volunteer whose death was attributed to gene therapy. The research volunteer had been administered a large dose (6 × 10¹¹ viral particles/kg) of an early-generation AdV vector, which triggered a massive immune response associated with multiorgan failure and eventual death [4]. This case disrupted the field of gene therapy and led to a governmental investigation of both the scientists involved and the university, ultimately, resulting in stricter oversight and overall reform of the field [5].

Until the volunteer’s death, the use of AdV vectors was considered the most promising gene-transfer technology because they were easily produced, were not mutagenic and were capable of transducing a wide variety of cells [6]. The concern that the death of the volunteer may have been owing to activation of the innate immune response by the large dose of AdV vector used has been partially responsible for attempts to create newer generations of vectors, including modification of both the DNA and the capsid of the virus [7,8]. Their attempts have been guided by the evolving knowledge of the immune response to natural infection by AdV.

The host response to AdV infection involves both the innate and the adaptive immune systems [9]. The innate immune response governs the host’s immediate response to infection and it has evolved to rapidly detect and eliminate numerous pathogens. The innate immune response is fixed, and detection is limited because the receptors of the innate immune system are fully encoded in the genome and do not undergo somatic recombination. However, the adaptive immune response is not under this constraint. Owing to receptor gene somatic recombination, the response of the major players of the adaptive immune response – the antigen-presenting cell, the T lymphocyte and the B lymphocyte – can be focused to optimally recognize the specific pathogens to which they are exposed. The preservation of the immunologic memory that results from this process assists in prevention of reinfection or serious illness upon re-exposure to the same or similar pathogens [9].

In the case of AdV vectors for use in gene therapy, current evidence suggests that innate immune responses are responsible for the dose-limiting toxicity and that the adaptive immune responses, with elimination of transduced cells, limit the duration of transgene expression. Using information derived from wild-type infection, and from experience using earlier-generation vectors, scientists have endeavored to create a new ‘gutless’ vector, which has almost all viral DNA removed in order to attempt to minimize the adaptive immune response. Indirectly, this may decrease the initial innate immune response, as a lower dose of vector may be needed for prolonged gene expression. Several excellent reviews regarding the adaptive response to AdV, particularly gene-therapy vectors, are available [10,11]. However, this article will summarize the current understanding of the innate immune response to AdV infection in the respiratory system. Primarily, the focus will be on how the innate immune response is activated by AdV. Finally, this article will address how gutless vectors have been modified in order to avoid provocation of that response. It is noted, and will be discussed, that this modification may only indirectly decrease acute toxicity by reducing the initial dose needed to facilitate long-term expression. Additional adjunctive therapies used to diminish immune responses to AdV will also be discussed.

Adenovirus structure & serotypes

Adenovirus was first cultured and reported as a distinct viral agent in 1953 [12]. AdV infections cause a broad spectrum of illness in immunocompromised and in immunocompetent hosts. AdV is the cause of a large number of acute febrile respiratory syndromes among military recruits [13]. It is also a cause of ocular, respiratory and gastrointestinal infections in the general population. Structurally, AdV is nonenveloped, regular and icosahedral. It is estimated to be 65–90 nm in diameter [14–16]. The virion has 20 triangular faces and 12 vertices, with 240 hexons and 12 pentons. Each vertex contains a penton complex consisting of the penton base and fiber proteins. Inside the capsid is a single molecule of dsDNA with a molecular weight of 2.0–2.4 × 10⁷ (or ~35 kb in length). Antigenic determinants that are important in the serologic classification of AdV are inherent to the hexon, penton base and fiber. All human AdV serotypes share a crossreacting group antigen that is carried on the inner surface of the hexon capsomere [17]. To date, 55 serotypes of human AdV have been recognized [18–20] and grouped into seven species (formerly termed ‘subgenera’) on the basis of their hemagglutinating properties and biophysical and biochemical criteria and, more recently, DNA sequence homology: species A, B (subdivided into subspecies B1 and B2), C, D, E, F and G [20,21]. Subgroup identification is of more than simply academic interest, as the type of infections caused by the virus depends on the subgroup to which the virus belongs [22,23]. This is also of relevance to the material that follows, as the innate immune response may also depend on the AdV subgroup studied [24,25].

Innate immune system

There are many components of the innate immune system that act to prevent infection upon exposure to infectious agents, including viruses. These can be generally classified into mechanical defenses, anti-infective chemical defenses, cellular defenses and chemokine/cytokine defenses. A detailed description of these components and their role in AdV infection follows.

Mechanical defenses

Mucociliary clearance and coughing are the two mechanical methods of pathogen clearance in the lung. Mucociliary clearance is a process that involves distal to proximal transport of a bilayer of mucus and a periciliary fluid by cilia, collectively referred to as the airway surface fluid. The composition of the bilayer allows infectious particles to be bound in the surface mucus, while the cilia are able to function in a lower-density layer. This bilayer increases the efficiency of transport and achieves clearance of infectious particles from the peripheral airways in relatively short periods of time [26]. Clearance of bacteria from peripheral airways by mucus transport may require up to 6 h. Bacterial replication is inhibited for 3–6 h by various endogenous antimicrobial factors, including lactoferrin, lysozyme and defensins [27]. Conversely, viruses require cellular machinery to reproduce, and so mucosal inhibitors act by interfering with the ability of these organisms to infect cells.

There are several inhibitors of viral infectivity present in mucus. The structure of mucus contains multiple sialic acid residues. These residues compete with cellular receptors used by viruses, such as influenza virus, and, therefore, decrease virus infectivity. Mucus protease inhibitor prevents the action of airway tryptase clara, which enhances influenza A virus infectivity [28]. The role of these substances in protection from AdV infection has not been established, but it is likely that sialic acid residues in mucus inhibit viral infectivity in a similar manner since sialic acid is involved in the attachment of some subgroup D AdV types to epithelial cells [29–31]. Also, an artificially synthesized sialic acid derivative inhibits the subgroup D AdV 37 to corneal epithelia cells [32]. With regards to subgroup C AdV vectors, the mucus bilayer is perhaps a limiting factor for optimal AdV gene transfer [33]. In addition, there is charge-dependent repulsion between AdV-5 group C capsid protein and cell surface sialic acid, which can be diminished and, thus, infectivity enhanced by neuraminidase [29,34–36]. Therefore, the effect of sialic acid on AdV infection depends on the AdV type and cellular environment, in some cases enhancing and in others inhibiting this process.

Adenovirus has been shown to be inefficient in targeting proximal human and murine airway epithelium [37]. One important reason for this inefficiency appears to be the lack of appropriate AdV receptors on ciliated airway epithelial cells [38]. These interactions will be further discussed later in this article.

Chemical defenses

Antimicrobial peptides

Generally, particles of less than 5 μm bypass the mechanical defenses and gain access to the terminal airways. Therefore, additional defense mechanisms are required to maintain lung sterility. Important classes of antimicrobial compounds are the cationic antimicrobial peptides (CAMPs). These are peptides present intracellularly in cellular phagocytic components of the innate immune system, but they are also secreted into the airway by epithelial cells. They contain three major classes of molecules: the defensins produced by epithelial cells, neutrophils, monocytes and macrophages and dendritic cells [39,40], the cathelicidins produced by neutrophils, mast cells and epithelial cells [41–43], and the thrombocidins, which are produced by platelets [44].

Defensins

Defensins are single-chain, strongly cationic peptides, with molecular weights of 3–4.5 kDa [45]. They are divided into two classes, α- and β-defensins, based on their chemical structure. The α-defensins HNP-1–4 are produced by neutrophils, while the α-defensins HD-5 and -6 are both produced by the intestinal Paneth cells, but HD5 is also found in the female reproductive tract and has been detected in nasal and bronchial epithelial cells [46]. β-defensins HBD-1–3 are produced by epithelial cells of many organ systems, while HBD-4 is primarily expressed in the testis and gastric antrum [47–53]. These molecules have a broad spectrum of microbicidal activity against Gram-positive and Gram-negative bacteria, mycobacteria, fungi and certain enveloped viruses [45,54,55]. The antibacterial (but not antiviral) activity of defensins is dependent upon low salt concentrations [51,53,56,57]. This has implications in patients with cystic fibrosis in whom the salt concentration is altered owing to the genetic defect.

Although AdV is nonenveloped, defensins have activity against this virus. The human α-defensin HNP-1 is present in bronchoalveolar lavage (BAL) fluid and inhibits AdV-5 vector infection in vitro by 95% [58]. The α-defensin HD5 as well as the β-defensin HBD-1 also reduce AdV-5 vector infectivity in vitro by three- to five-fold [59]. Some of the defensins, especially HBD-2, are upregulated by various cytokines, including IL-1β [51,60]. Likewise, production of cytokines, such as IL-6 and IL-8, can be induced by defensins [61]. This provides for amplification of the antiviral response during infection. Defensins inhibited infectivity of wild-type AdV. The mechanism of this inhibition has been shown to be due to α-defensin HD5 and HNP1 blockade of AdV uncoating events, including the release of the endosomalytic protein VI, thus preventing genome exposure and nuclear entry [57,62,63].

Cathelicidins

Cathelicidins are antimicrobial peptides found exclusively in mammals. In vitro and in vivo studies indicate that that they are effector molecules of mammalian innate immunity. Although many different cathelicidins have been identified in mammals, only one has been identified in humans. The first human cathelicidin was cloned from cDNA isolated from human bone marrow as FALL-39 [64]. It is identical to the independently identified antimicrobial domain of hCAP-18 produced by neutrophils [65]. This peptide is cleaved to form a C-terminal fragment with antimicrobial activity, known as LL-37 [66]. The C-terminus of this peptide fragment binds and inactivates lipopolysaccharide [67]. LL-37/hCAP18 is present in neutrophil granules [65], and is produced by bone marrow and testis [64], epithelial cells of the skin [68], and respiratory [43] and squamous epithelia of human mouth, tongue, esophagus, cervix and vagina [69].

Cathelicidins have activity against bacteria, fungi and viruses, which varies by the species studied [43,70]. Data support a role for cathelicidins in the inhibition of orthopox virus (vaccinia) replication both in vitro and in vivo [71]. The lack of plasmacytoid dendritic cell recruitment, together with the missing upregulation of cathelicidin LL37 in atopic dermatitis lesions, is considered relevant for an atopic dermatitis patient’s susceptibility to eczema vaccinatum [72]. LL-37 has some activity against herpes simplex virus (HSV)-1 and -2, in vitro, but there are no reports for testing against AdV [73]. These peptides may have an indirect activity against AdV as they are upregulated by IL-6 [69] and IFN-γ [74], and cathelicidins are chemotactic for neutrophils, monocytes and T cells [74,75]. However, LL-37/hCAP18 was not upregulated in human colon epithelial cells stimulated with cytokines playing a role in intestinal inflammatory responses, including TNF-α, IL-1α, IFN-γ and IL-6 [76]. Thus, LL-37/hCAP18 appears to be differentially regulated among different cell types. Cathelicidins are probably partially responsible for the inflammatory infiltrate seen during AdV infection, since there is evidence that cathelicidins are upregulated by cytokines induced by AdV, and because cathelicidins are chemotactic.

Thrombicidins

Thrombicidins are antimicrobial peptides released from platelets stimulated with thrombin or found in platelet extracts. They are also released when platelets are exposed to infectious agents. Several of these peptides are either structurally similar to, or identical to, known chemokines [77]. Studies of peptides have demonstrated in vitro, and in some cases in vivo, activity against a few bacterial species and fungal species [44,77–79]. However, there are as yet no studies of the direct antiviral activities of thrombocidins.

Pulmonary surfactants

Although pulmonary surfactants’ primary role is to facilitate lung expansion by reducing surface tension, they also play a role in lung defense against infections. Two surfactant proteins, SP-A and SP-D, have antimicrobial activity. In this regard, they are classified as collectins or collagenous C-type lectins. These proteins consist of oligomers of trimeric subunits the subunits contain C-type lectin carbohydrate-recognition domains [80]. These domains bind to a number of different organisms, including bacteria and fungi, and decrease virulence both by inhibiting infectivity and by enhancing phagocytosis by cellular components of the innate immune system [80–84].

The lung collectins show specific interactions with various respiratory viruses. Purified SP-A and SP-D inhibit infectivity and hemagglutination activity of influenza A virus in vitro [85,86]. Collectins also inhibit the infectivity of respiratory syncytial virus (RSV) [87] and rotavirus [88]. SP-A also promotes the phagocytosis of HSV by rat alveolar macrophages [89].

Animal studies using SP-A- or SP-B-deficient mice have verified an in vivo effect on clearance of several of these microbes. Clearance of bacterial pathogens, as well as viral pathogens, such as RSV and influenza A virus, was deficient in SP-A-deficient mice [90]. SP-D enhanced the phagocytosis and pulmonary clearance of RSV in mice [91]. SP-D-deficient mice showed decreased viral clearance of certain strains of influenza A virus, and increased production of inflammatory cytokines in response to viral challenge. Viral clearance of strains of influenza A virus and the cytokine response were both normalized by the coadministration of recombinant SP-D [92]. Of particular interest to this article, clearance of AdV DNA from the lung and uptake of fluorescent-labeled AdV vector by alveolar macrophages was decreased in SP-A-deficient mice. Supplementation with SP-A enhanced viral clearance and inhibited lung inflammation during pulmonary adenoviral infection, confirming the importance of SP-A in antiviral host defense [93].

Cellular defenses

Macrophages

Cellular defenses of the innate immune system include both resident cells and those recruited from the bloodstream. Alveolar macrophages are the most prominent resident cells that engulf and kill infectious agents. They also provide a link to the adaptive immune system and function as antigen-presenting cells, although less effectively than other cells [94–96]. Macrophages have a pivotal role in immune surveillance of the respiratory tract, initiation of anti-infective inflammation and regulation of potentially harmful inflammatory responses. Alveolar macrophages make up 85% of cells recovered in BAL [97]. Alveolar macrophages play an important role in the elimination of AdV vectors from the lung. Alveolar macrophages recovered from the mouse lung 30 min after intratracheal administration of an AdV vector showed large amounts of vector genome, whereas much less was evident in alveolar macrophages recovered after 24 h [98]. In a murine model, AdV labeled with a fluorescent dye was very rapidly (~1 min) localized within the alveolar macrophages [99]. Although best studied in epithelial cells [100], internalization of AdV by mononuclear phagocytes is believed to occur by integrin receptor-mediated endocytosis [101,102].

Alveolar macrophages were also identified as a cell source of initial cytokine signaling by AdV. At 30 min after infection, alveolar macrophages expressed mRNA for TNF-α and IL-6, but airway epithelial or vascular endothelial cells did not. Blockage of virus uptake prior to internalization in the alveolar macrophages completely blocked TNF-α expression [99]. Other products released by macrophages include IL-12 (activates NK cells), LTB4, toxic oxygen species, IL-1 (T-cell stimulatory and proinflammatory cytokine) and antibacterial products, such as lactoferrin, lysozyme and defensins [99,103]. TNF, IL-1 and IL-6 are considered primary activators of the immune system. Although this activation is mostly caused by response to detection of the capsid proteins, AdV and vector DNA alone activate the immune response via Toll-like receptors (TLRs) [104]. Mice with impaired TLR9 receptors have significantly reduced IL-6 secretion compared with wild-type mice when exposed to HD AdV vectors [105].

Dendritic cells

Cells of similar lineage, the dendritic cells, also function in the respiratory tract, and are more like tissue macrophages present in other organ systems. They are less efficient at phagocytosis, but more efficient at antigen presentation than alveolar macrophages. They are indispensable to the initiation of the adaptive immune response. Tobacco smoke decreases the number of dendritic cells in lung tissue, and chronic tobacco exposure impairs the immune response against AdV [106]. This may be relevant as smokers appear to be at risk for pneumonia due to wild-type AdV [107]. Antigen presentation by murine dendritic cells is NF-κB dependent, and NF-κB inhibition blocks dendritic cell AdV antigen presentation and has a marked immunosuppressive effect in vivo [108]. Preferential activation of dendritic cells and macrophages may account for AdV inflammatory responses in vivo. Elimination of tissue macrophages and splenic dendritic cells in vivo considerably reduced the early release of IL-6, IL-12 and TNF-α, and significantly blocked the specific cellular immune response to AdV [109]. It is known that viral gene expression is not necessary for dendritic cell activation [110,111]. Viral capsid proteins, the penton, hexon and fiber, have all been shown to trigger activation of dendritic cells [112–114]. However, dendritic cells are also activated through TLRs during exposure to AdV infection [104]. When peripheral blood monocyte cells with or without plasmacytoid dendritic cells and purified plasmacytoid dendritic cells were exposed to AdV, only the purified plasmacytoid dendritic cells produced large levels of IFN-α, and this production was blocked when the cells were treated with chemicals to block TLR activation [115]. It also appears that the production of interferon (IFN) by dendritic cells leads to their own maturation [116]. Which mechanism is most responsible for dendritic cell activation by AdV is currently under study, but evidence is accumulating that TLR-independent activation of dendritic cells by AdV is important [117,118].

Neutrophils

Neutrophils are also important in the response to infectious agents, and are probably important to AdV. They internalize AdV in the presence of complement or IgG, and are attracted to the lung following infection with AdV [119]. This is partially due to release of various cytokines, including IL-8, by macrophages. IL-8 is a major neutrophil chemotactic factor in the lung [120]. Intravenously administered AdV significantly increased leukocyte rolling and adhesion in the liver within minutes of transduction. P-selectin, α₄-integrin and E-selectin are important in this process, as blockade of these receptors inhibited leukocyte rolling and subsequent adhesion. Depletion of circulating neutrophils eliminated leukocyte rolling and adhesion [121]. MIP-2 antibody and neutrophil depletion diminished hepatic injury in human AdV vector-exposed mice, as determined by both reduced serum aspartate aminotransferase and alanine aminotransferase levels and histology [122]. This suggests that early tissue injury is largely owing to chemokine production and neutrophil recruitment. One strategy to prevent AdV-mediated inflammation and, therefore, tissue injury would be to interfere with chemokine or neutrophil function.

Epithelial cells

The airway epithelial cells not only function as a passive barrier to infectious particles but also actively participate in the innate immune response to foreign antigens. This is particularly important in AdV infections because the epithelium is the major site for AdV replication. The immune response of epithelia to infection and antigen exposure consists of an increase in the release of antimicrobial peptides into the lumen of the airways, and the release of chemokines and cytokines into the submucosa, which initiate an inflammatory reaction. This results in the recruitment of phagocytes that help clear the invaders, and of dendritic cells and lymphocytes, which facilitate an adaptive immune response. Activation of TLRs on epithelial cells has now been shown to be involved in the regulation of expression of a variety of genes, including those encoding cytokines, chemokines and antimicrobial peptides [123]. TLR-mediated responses to AdV have been demonstrated in murine corneal epithelium [124]. However, AdV infection of the cornea resulting in keratitis was similar between wild-type mice and mice without TLR9. Also, the virally induced expression of chemokines and leukocyte infiltration could be blocked with RGD peptides. These results suggest that viral capsid–host cell integrin interactions are more important than TLR-mediated responses in the murine corneal epithelium [125].

There are other mechanisms whereby components of the innate immune system, in this case the lung epithelia, respond to AdV infection. AdV vector infection of A549 respiratory epithelial cells induces a significant expression of ICAM-1 and increased CD18-dependent adhesion of activated neutrophils [126].

This may be partially responsible for the initial inflammation seen in animal models of AdV infection, although direct stimulation of cytokines both from lung epithelia and alveolar macrophages also plays a role. Intratracheal administration of AdV genotypes 3p and 7h in mice caused a robust inflammatory response in the lung, which initially consisted of neutrophilic and monocytic alveolar infiltration. There was also mild peribronchial and perivascular inflammation. At early time periods after exposure, neutrophils were the predominant cell type recovered by alveolar lavage. This correlated with increased levels of neutrophilic chemokines, such as MIP-2 and KC (mouse IL-8 homologs) in lung tissue. Elevation of other pro-inflammatory cytokines, such as IL-1β, TNF-β, IFN-γ and IL-12, were also noted [127].

It is likely that direct infection or stimulation of distal epithelia is responsible for many of these changes, as the peribronchial inflammation was minimal and because proximal bronchial human airway epithelial cells are resistant to infection by AdV, at least as determined for AdV-5 vectors [38]. Infection of epithelia by AdV involves two steps. First, the globular head of the AdV fiber attaches to either the coxsackie and AdV receptor, the MHC class I α-2 domain or the CD46 receptor depending on the AdV subtype [128–131]. Then the penton base of AdV binds to the α_νβ₃ and α_νβ₅ integrins or α_νβ₁ integrins, which are important for virus internalization [102,132,133]. The resistance of the bronchial epithelia to AdV infection appears to be due to a lack of the appropriate receptors on the apical membrane. Coxsackie and AdV receptor and MHC I are polarized to the less accessible basolateral membrane [134], and α_νβ₃ and α_νβ₅ integrins are also minimally expressed on the apical plasma membrane of proximal airway epithelial cells [135]. Several comprehensive reviews of AdV receptors are available [136,137].

It is important to note that these restrictions are not likely to be as important in the distal human lung, specifically, in the alveolar epithelium. Human alveolar epithelial cell lines are easily infected by both type 5 and type 7 wild-type AdV [24]. Also, human lung tissue slices support wild-type AdV 7 virus replication, infection of the alveolar epithelia and an innate immune response, as evidenced by increased IL-8 release [138].

Cytokines & AdV

Adenovirus induces different clinical manifestations, ranging from unapparent infection or benign upper respiratory tract disease to necrotizing bronchiolitis or even disseminated fatal disease in immunocompromised patients. AdVs are cytopathic in permissive cell cultures and the inflammation in affected areas frequently extends beyond the necrotic epithelia, where AdV inclusions are most prominent. This suggests that additional mechanisms might be involved in the pathogenesis of tissue damage [139–141]. Activation of the immune system and the generation of numerous chemokines and cytokines clearly play a role in activation of inflammation and, therefore, could play a major role in the pathogenesis of tissue damage. Although there is a distinction between chemotactic cytokines (chemokines) and nonchemotactic cytokines by their ability to induce cell migration, their functions are interrelated. For example, chemokines not only attract inflammatory cells, they also activate them, and may cause them to proliferate. Also, while nonchemotactic cytokines may also have activating and proliferative effects, they also indirectly stimulate chemotaxis by stimulating the release of chemotactic cytokines. Both chemotactic and nonchemotactic cytokines have been shown to play an important role in pulmonary host defense. These include TNF-α, IL-1, IL-6, IL-8, IL-10, IL-12, IFN-γ and granulocyte colony-stimulating factor [142]. There is evidence that the cytokine response by itself is important in the consequences of infection with AdV. Increased concentrations of IL-6, IL-8 and TNF-α were associated with hypoperfusion, febrile peaks, tonic–clonic seizures and septic shock, and were significantly associated with the severity of AdV infection in children [143]. The activation of the innate immune response also occurs in the absence of virus gene transcription [24,144,145]. The alveolar macrophages are proposed to be the source of initial cytokine signaling after acute AdV respiratory tract infections. In a murine model, 30 min after infection, alveolar macrophages but not airway epithelial or vascular endothelial cells expressed mRNA for TNF-α and IL-6 [99]. The signal for initiation of TNF-α expression during AdV exposure to RAW264.7, a macrophage cell line, requires virion internalization and probably occurs during or subsequent to endosome acidification, as blockage of these specific steps of passage of the virus completely blocked TNF-α expression [99]. There is some uncertainty as to whether endosome acidification is required, as the agent to prevent this process, chloroquine, inhibits signaling pathways known to be important for cytokine production [146]. Other cells of the immune system also clearly participate in the initial innate immune response, as chemokines, particularly IL-8, are induced in airway epithelial cells by AdV in culture and in human tissue [24,25,138]. The innate immune response of alveolar epithelial cells does not appear to require triggering by macrophages, as IL-8 is induced in pure epithelial cell cultures in the absence of macrophages and IL-8 is induced in a human slice model where macrophages are not prominent in the preparation [24,25,138].

IL-1

IL-1 plays a role in the early inflammatory response seen after AdV infection. In a mouse model, intranasal inoculation of wild-type AdV-5 produced pneumonia, even though the virus did not replicate. Assays showed the appearance of IL-1, TNF-α and IL-6 in mouse lungs concomitant with the development of an early-phase infiltration with lymphocytes and monocytes/macrophages [147]. IL-1 is also the major mediator of a very early inflammatory response to AdV vector in the brain [148]. IL-1β may also play a role in progressive fibrosis and tissue remodeling in mice through the induction of profibrotic cytokines PGDF and TGF-β1 [149]. A key event in virus-induced inflammation is the local activation of endothelial cells, as indicated by the expression of adhesion molecules, such as ICAM-1, VCAM-1 and E-selectin. Fluids from AdV type 37-infected respiratory and ocular epithelial cells activated ICAM-1 and, to a lesser extent, VCAM-1 expression in cultured human umbilical vein cells [150]. Blocking studies with anticytokine antibodies implicated IL-1α as the epithelial cell-derived factor that activated ICAM-1 expression [150]. AdV-infected A549 respiratory cells treated with an IL-1 receptor antagonist (IRAP) expressed 75% less IL-8 than untreated cells, whereas IRAP pretreatment of TNF-α-stimulated cells did not affect IL-8 production [151]. Thus, IL-8 production by AdV vector-infected cells occurs partly through IL-1 activation that can be downregulated by IRAP. It is not known whether IL-1 is important in IL-8 induction by wild-type AdV. Finally, AdV vectors have been shown to induce IL-1 in macrophages through interaction with β₃ integrins. Induction of IL-1 does not involve traditional known signaling pathways [152].

IL-6

IL-6 is another cytokine that has been associated with the immune response following AdV infection. IL-6 is produced by vascular endothelial cells, mononuclear phagocytes, fibroblasts and activated T lymphocytes in response to a variety of stimuli and is referred to as the global response marker [153]. Nijsten et al. reported that IL-6 levels increase earlier in the acute-phase response than those of other proteins, such as C-reactive protein and α-1 antitrypsin, and are related to the generation of fever [154]. IL-6 may act as a hepatocyte-stimulating factor and induces various acute-phase proteins in the liver cells [155]. When classified according to the clinical findings and outcome, IL-6 was detected in the groups with severe and fatal AdV infections in children, and not in the group classified as moderate [143]. Increased concentrations of IL-6 were associated with hypoperfusion, febrile peaks, tonic–clonic seizures and septic shock in these children. High serum values are associated with the severity of AdV infection. In a study comparing 106 children with AdV, influenza or RSV infections, the IL-6 concentration was greater than 50 pg/ml in 78.6% of patients with AdV infections, compared with 15.8% in patients with influenza and none in patients with RSV infections. IL-6 was again associated with the severity of AdV infection and correlated with serum concentrations of C-reactive protein in the group with AdV infection [156]. Presumably, IL-6 plays a dual effect during viral infections – it may stimulate immune defenses against infected cells and may participate in tissue damage [157].

IL-8

The importance of IL-8 in acute AdV infection is suggested by the consistent finding of prominent neutrophilic peribronchial and alveolar infiltration early in the disease process in a number of animal models [158–161]. Inflammation also occurred under conditions in which the virus did not replicate [147]. This is important, as IL-8 is the major neutrophil chemotactic factor in the lung [120]. There are data suggesting that IL-8 has a role in the pathogenesis of naturally acquired acute AdV infection in humans. In children with AdV pneumonia, serum IL-8 levels correlate with clinical outcome [143]. The highest levels are seen in patients with a fatal outcome and approach values seen in patients with sepsis. Both TNF-α and IL-1, produced by alveolar macrophages, are potent inducers of IL-8 production by several cell types, including alveolar macrophages, type II epithelial cells and lung fibroblasts [120]. AdV strains and vectors vary in their ability to induce IL-8 and cause lung disease. Wild-type AdV-5 does not induce IL-8 release in vitro or cause pneumonia in immunocompetent humans. Both E1A-deficient AdV-5 vectors and wild-type AdV-7 induce IL-8 release and pneumonitis [13,24,162–166]. It is likely that AdV gene expression is not necessary for IL-8 induction, as induction occurs prior to AdV protein expression in AdV-7-infected cells, and AdV protein expression is minimal in gene-therapy vectors. Therefore, it is likely that induction occurs through the interaction of surface proteins with cellular receptors, which are known for AdV-5 but not entirely known for AdV-7. This could result in signaling pathway induction resulting in IL-8 gene expression. We, and others, have demonstrated p42/44 MAPK activation with exposure of cells to both virus types. We have confirmed that activation of this kinase is essential for IL-8 induction by AdV-7 [25]. The role of other signaling pathways in IL-8 induction by AdV is not known. Also unexamined is the mechanism whereby modification of wild-type AdV-5 for gene therapy results in a vector that is capable of inducing IL-8. A similarly modified AdV-7 replication-deficient vector has been constructed, but has not been tested for IL-8 or signal pathway induction.

Work by Hogg and coworkers suggests that latent AdV may act as a coactivator of the IL-8 gene, predisposing to chronic inflammation and the development of chronic obstructive pulmonary disease (COPD) in some patients. In their studies, increased amounts of AdV–DNA, specifically the AdV–E1A proteins, were detected in the lungs of patients with COPD compared with age-matched asymptomatic patients with similar smoking exposures [167,168]. Additional studies showed that quantitative regional expression of the AdV–E1A proteins in lung epithelial tissue also correlated with the number of inflammatory cells and the amount of lung destruction as determined by computed tomography [169]. Invitro studies, using lung epithelial cell lines, demonstrated that stable transfection with AdV–E1A proteins enhanced lipopolysaccharide-induced expression of IL-8 through NF-κB [170,171]. Thus, the presumed mechanism of AdV enhancement of COPD is amplification of the cytokine response, specifically IL-8 to external stimuli, such as lipopolysaccharide or cigarette smoke. There remains uncertainty about the role of AdV in COPD pathogenesis, as a recent study has shown a low incidence of E1A RNA expression and E1A DNA in patients with COPD [172].

IL-12

IL-12 is a heterodimeric protein consisting of two subunits (p35 and p40) and plays an important role in promoting Th1-type immune responses. It serves as the major signal for IFN-γ expression from T cells and NK cells [173]. Therefore it plays a role in linking the innate and adaptive immune systems.

IL-12 is induced during both wild-type AdV infection and AdV vector exposure. In a mouse model, after infection with mouse wild-type AdV, message for the p40 component of IL-12 was transiently increased shortly after infection [174]. IL-12 was expressed mainly by macrophages. AdV vector also stimulates IL-12. AdV vector infection of mouse dendritic cells stimulated IL-12 release, or IL-12-dependent dendritic cell activation of NK and T cells [175–177]. As well as this, an increased IL-12 response to AdV vector infection in BALB/c mice correlated with increased AdV vector clearance [178]. However, knockout of the IL-12 gene did not significantly affect AdV removal from the host [179]. Therefore, although IL-12 may be important in innate immune activation and the type of subsequent adaptive immune response to AdV infection, IL-12 does not by itself appear to modulate clearance of AdV.

TNF-α

TNF-α, initially named for its ability to trigger the necrosis and involution of certain tumors [180], is now widely recognized as a mediator of the host response to infection. TNF-α is rapidly produced following either antigen-specific or nonspecific stimulation of the alveolar macrophages, and has been designated as an early alarm response cytokine [142]. TNF-α and IL-1 do not appear to affect polymorphonucleocyte chemotaxis directly, but both of these cytokines are potent inducers for the production of certain CXC chemokines, such as IL-8, by alveolar macrophages, type II epithelial cells and other cells of the airways [85,120]. These chemokines are the major chemoattractants for polymorphonucleocyte recruitment into the lung during pulmonary infection and inflammation [120,181–184]. A study evaluating the expression of TNF-α and IL-1β after AdV infection of human monocytes (which are nonpermissive for AdV infection) and macrophages [185] showed that the production of both cytokines was enhanced in monocytes. However, in macrophages, a slight enhancement of TNF-α was seen and no IL-1β was detected. The results suggest that cellular genes might be activated in nonpermissive cells where no viral gene products can be detected. TNF-α signaling through both the p55 and p75 TNF receptors plays important roles in the magnitude of the humoral immune response. Absence of TNF-α or the p55 receptor significantly attenuates the antibody response to AdV [186].

Induction of TNF-α by modified AdV may occur during therapeutic use of the virus. Replication-competent AdVs with various E1 modifications designed to restrict their replication to tumor cells are being evaluated as oncolytic agents in clinical trials. In mouse models, such oncolytic AdV showed greater dose-dependent hepatotoxicity than E1-deleted AdV vectors following intravenous administration. Hepatotoxicity correlated with expression of wild-type E1A in the liver with increases in viral DNA levels. This was correlated with rapid induction of TNF-α to high levels and with rapid elevation of serum alanine aminotransferase. Hepatotoxicity was significantly reduced for an AdV with deletions in the region E1A (dl01/07) or a virus lacking E1A. The results suggest a mechanism for hepatotoxicity involving virus-induced production of local TNF-α release and E1A-mediated sensitization of hepatocyte killing [187]. However, AdV hepatotoxicity does not absolutely require AdV gene expression as AdV vectors in which the viral gene expression was completely suppressed caused hepatic toxicity in animal models [145]. Thus, both AdV capsid and AdV proteins are capable of causing cell damage during AdV vector exposure.

Interferons

Interferons are a family of cytokine mediators that influence the quality of cellular immune responses and amplify antigen presentation to specific T cells. IFNs were first discovered by Isaacs and Lindenmann in 1957 by the induction of IFNs in chick cells by influenza virus [188]. In 1967, it was observed that chick embryo fibroblasts inoculated with human AdV produced IFNs [189]. There are three major classes of IFN: IFN-α, IFN-β (type I IFN) and IFN-γ (type II IFN). AdV induces the production of high levels of type I IFN both in vitro and in vivo.

Type I IFN induction by AdV is the result of two mechanisms involving TLR-dependent and TLR-independent pathways. TLR9 mediates production of IFN by plasmacytoid dendritic cells by AdV and is dependent on MyD88. By contrast, conventional dendritic cells and macrophages produce IFN after detection of cytosolic AdV DNA through TLR-independent pathways. These TLR-independent pathways are triggered when cytosolic AdV–DNA stimulates the transcription factor IRF3 and mediates the IFN and proinflammatory response [118]. The high levels of type I IFN produced are critical for both the innate and adaptive immune responses against AdV vectors. This was demonstrated when AdV induction of IL-6 and IL-12 and T-cell activation were reduced in mice that were deficient for type I IFN receptors [190].

While type I IFNs are secreted by virus-infected cells, IFN-γ is produced by T cells and NK cells and plays an important role in cell-mediated immunity against a broad spectrum of intracellular pathogens. IFN-γ activates several leukocyte functions, including stimulation of respiratory burst, antigen presentation and priming TNF-α release by macrophages. AdV-7 at multiplicity of infection equal to one induced IFN in human leukocyte cultures [191]. Exposure of cells to IFN induces an antiviral state in which the replication of a wide variety of both DNA and RNA viruses is inhibited [192]. During this antiviral state, a set of cellular genes, termed IFN-stimulated genes, are transcriptionally induced. AdV infection of HeLa cell cultures induced the transcription of these genes, without IFN synthesis or the synthesis of new AdV proteins [193]. However, the E1A gene products specifically suppressed this transcription. The dual effect of AdV on the expression of the IFN-stimulated genes may represent an example of action and evolutionary reaction between virus and host.

Attempts to modify AdV vectors to diminish immune responses

Adenovirus vectors are commonly used for gene therapy. Their use was popularized because, like wild-type AdV, they have weak mutagenicity, the ability to transduce a wide variety of cells [6], and early generations were easily produced. However, early-generation vectors had limited transgene-carrying capacity and were highly immunogenic, which prevented long-term transgene expression and required high-dose administration to provide even short-term transgene expression [194].

The need for further modification of the first-generation AdV vectors (FG-AdV) became apparent after early animal studies, which exhibited local and systemic toxicity. However, lesser, although still important, limitations of FG-AdV as an agent of gene therapy were already known to exist, including adaptive immunogenicity that prevented long-term transgene expression, and a limitation in the transgene size due to viral packaging constraints.

These shortcomings of the FG-AdV vector led to its subsequent modification and the development of second- and third-generation vectors. These vectors had less residual expression of AdV gene products to which the inflammation seen with the first-generation vectors was attributed. As they caused less toxicity in lower animals, trials of high-dose administration in humans were started with a third-generation vector, an E1-deleted, E2A temperature-sensitive AdV-5. This culminated in the unfortunate death of the 19th patient in the trial.

Perhaps partially in response to this sad event have additional modifications of the vector been performed. Attempts to further reduce residual AdV gene expression resulted in the development of the helper-dependent AdV vector (HD-AdV), also known as the gutless or ‘high-capacity’ vector [195,196]. HD-AdV vectors have almost all viral DNA deleted except for two elements, the inverted terminal repeats and the packaging domain (ψ) [7]. With the deletion of almost all the viral DNA, the carrying capacity of HD-AdV vectors was substantially increased. Also, in animal models, the toxicity was decreased, while the duration of transgene expression relative to the earlier-generation AdV vectors was enhanced [197–199]. It is thought that there are two phases of AdV toxicity, an initial response caused by the innate immune system, and a second response due to the adaptive immune system [200,201]. In the preceding animal models, it was the second phase of toxicity that was reduced in the gutless vector-treated animals but the initial phase of toxicity persisted. This suggests that the toxicity of gutless vectors is primarily reduced by removal of the viral coding genes that trigger the adaptive immune response and less owing to a reduction in the innate immune response triggered by viral capsid proteins [202]. However, in one study, the first-generation but not the gutless vector induced acute liver injury in Rag1-deficient mice (lacking mature T and B cells). This suggests that AdV DNA may, in fact, enhance innate immune responses to the virus [196].

However, with the deletion of the viral genes, a production system had to be created that would enable the vectors to replicate. That production system consists of the HD-AdV vector, a producer cell line and a helper virus. One such producer cell line that is used is HEK 293 [203]. It provides the E1 gene that the HD-AdV vector is missing. The helper virus is an E1-deleted FG-AdV. It coinfects with the vector and provides the other viral genes necessary for vector production [204].

Since a helper virus is required for replication, contamination of the HD-AdV vector preparations with helper virus can occur [205], and the contamination can be hazardous. In order to reduce this contamination, several approaches have been tried. The most common approach involves the use of a recombination system such as Cre/loxP. Cre is a recombinase that recognizes a specific DNA sequence, the loxP site. If Ψ of the helper virus is placed between two loxP sites, it can be removed and, thus, prevent packaging of the helper genome into the viral capsid [206].

Other approaches to prevent contamination have included the use of hybrid helper viral systems with either HSV or baculovirus and the helper virus. By combining HSV with a helper virus to make a hybrid helper virus, hybridized helper virus produced during production of the final vector product can be removed. This is done by using known HSV extraction techniques to remove HSV–AdV hybrid virus. The baculovirus system relies on the property of this virus to transduce mammalian cells without replication. Thus, by the use of this baculovirus–AdV hybrid, replication of the helper virus is prevented [207,208]. However, the popularity of the hybrid viral systems has been limited owing to low yield of HD-AdV production and contamination with replication-competent AdV. A different approach relies on size constraints for packaging of the AdV genome. It is based on the use of a protein IX (pIX)-deleted helper virus [209]. Deletion of pIX results in virions that are heat labile, with capsids that accommodate only 35 kb, of viral DNA. If the viral DNA is larger than 35 kb it cannot be packaged. Therefore, by using a pIX helper virus larger than 35 kb, packaging of the helper virus should be limited and contamination prevented. Unfortunately, levels of contamination with this system are no different than with the Cre/loxP system. In addition, a kinetic approach has been employed [210]. By inserting a specific DNA sequence, attB, between the 5′ inverted terminal repeats and the Ψ, the lifecycle of the helper virus is delayed by 14 h. This delay allows helper virus to be harvested before any significant contamination with helper virus has occurred. This strategy is also advantageous because one can still obtain normal titers of helper virus by allowing infection to continue. Additional sophisticated techniques, such as adding ‘stuffer DNA’ to prevent the formation of undesired contaminating vectors by exploiting packaging limitations of AdV, have been developed [206,211]. Unfortunately, no helper virus removal strategy has yet been successful in eliminating all contamination [13].

Of concern regarding the strategy of modifying AdV vector toxicity by removing larger portions of the internal genome is that the viral particles induce the innate immune response without gene expression. In this regard, it has been demonstrated that viral gene expression is not required in order to induce toxicity. This was demonstrated by using AdV vectors irradiated by UV light with or without psoralen, which is designed to inhibit transcription but not damage viral surface proteins. AdV treated in this manner induces cytokine production and inflammation in mouse liver, lung and corneal tissue [122,125,212]. Several mechanisms for this directly proinflammatory effect of AdV have been proposed [152,213,214]. It should be noted that the AdV genome may contribute to acute toxicity based on the results previously discussed in Rag1-deficient mice exposed to the so-called gutless vector [196].

Conclusion

Protection from wild-type AdV infection involves both the innate and adaptive immune response. The innate immune system is complex, utilizing mechanical and cellular defenses. Activation of these cellular defenses triggers the release of cytokines and chemokines, which amplify the immune response through the recruitment and activation of other inflammatory cells. These cells then assist with elimination of infectious virus and/or bridge the gap between innate and adaptive immunity.

Much of the acute damage due to AdV infection appears to be owing to deregulation of the immune response. Thus, controlling this response may prove beneficial in suppressing the damage that occurs. This potential therapeutic avenue could be dangerous; however, if it enhanced overall viral replication and increased the infectious burden of the host. This therapeutic goal represents only a slight divergence from the goals of a gene therapist, who does not have to be greatly concerned with replication of the vector.

The current goal of the gene therapist is to create an AdV vector that avoids detection by the immune system and causes no deleterious side effects. Current HD-AdV vectors have been modified in order to achieve such a purpose. They have had much of their viral genome removed in order to avoid precipitating a toxic immune response and to maintain transgene expression. With those deletions come the challenge of maintaining the vector’s ability to transduce the host, and producing sufficient quantities of a radically modified vector. Progress has been made with the use of progenitor cells and helper viruses, but despite progress, trials have yet to demonstrate that these modifications have met all the aforementioned goals. A list of ongoing trials is available at [301].

It is now known that further efforts will need to be directed at modulating innate immune events that occur during early stages of vector–cell interaction. This is owing to the fact that direct stimulation of the innate immune system by AdV proteins occurs.

Future perspective

Attempts to lessen the innate immune response continue, with the focus being on three types of approaches: the first being pretreatment of the host, the second relying on modification of the AdV capsid or internal genome, and, finally, modifying the method or route of delivery.

Host pretreatment includes the use of immunosuppressive drugs such as glucocorticoids to globally suppress the immune system, or selective immunosuppression using, for example, with TNF-α blockers. Studies are needed to determine if pre-emptive treatment to avoid or suppress the innate immune response will be successful.

Modifications of the AdV vector include genetic modification and/or removal of AdV DNA, which has already been described in detail in this review, covalent modifications of the AdV capsid using synthetic polymers such as polyethylene glycol, AdV targeting techniques, display of immunoevasive proteins on the AdV capsid and, finally, chimeric AdV vectors.

Results using synthetic polymers, such as polyethylene glycol, have been promising in preliminary studies in mice, and studies in primates are in progress [215–217]. AdV targeting tries to maximize the transduction of the target tissue and minimize the transduction of other tissues by either using tissue-specific promoters or modification of AdV capsid proteins to enhance tissue-specific vector tropism [218]. Display of immunoevasive peptides on the AdV capsid has met with some success in mice, and work in this area is ongoing, including mutating cellular binding sites on the AdV capsid [8,219]. Also, chimeric vectors that display a combination of AdV capsid proteins from multiple AdV serotypes are in development [220–227].

Finally, modification of the method of delivery has been attempted. In the case of hepatic vector delivery, use of a balloon catheter in the vena cava to increase vector contact with the target organ, and decrease dissemination has shown promise in a nonhuman primate model [228].

This is, of course, a brief and incomplete summary of recent techniques that have been used. For a more detailed review of this topic, the reader is referred to the excellent review by Seregin [229].

Additional work will be needed in order to reduce the immune response to AdV or AdV vectors. Perhaps this is finally a chance for the information acquired during wild-type AdV infection studies to be applied to the field of AdV gene therapy. For example, inhibiting interactions of wild-type AdV with integrin receptors by co-administering homologous peptides decreases acute inflammation during experimental keratitis and it is conceivable that similar strategies could be used during gene therapy [125]. This ‘crosspollination’ of ideas, from those studying wild-type infection to those interested in AdV gene therapy, may finally lead to the discovery of techniques suitable for the repair of genetic deficiencies in the laboratory, and also in the clinic.

Executive summary.

• Wild-type adenovirus (AdV) infections cause a broad spectrum of illness in both immunocompromised and immunocompetent hosts.

• Innate immune responses to wild-type AdV are important in acute infection and to acute inflammation that occurs during AdV vector gene therapy.

• Modulation of the early-generation AdV vectors may directly or indirectly decrease acute toxicity and improve long-term transgene expression.

• Greater understanding of the innate and adaptive immune response to AdV and AdV vectors may improve the prospects of successful gene therapy.