A New Link Between Virus Cell Entry and Inflammation: Adenovirus Interaction With Integrins Induces Specific Proinflammatory Responses

Replication-defective or conditionally replicating human adenoviral (HAdV) vectors have shown considerable promise as gene or vaccine delivery vehicles, particularly for the treatment of cardiovascular diseases or cancer. Nevertheless, the propensity of HAdV vectors to elicit a robust proinflammatory response following systemic administration limits their use in the clinic.¹ As a result of the potential severity of virus-induced inflammation and accompanying toxicity,² the majority of HAdV-mediated gene delivery studies have used local rather than systemic delivery of the viral vector. For example, the efficiency of HAdV-mediated gene delivery has been evaluated following direct injection into a tumor mass³ or into the confined spaces of certain organs such as the bladder⁴ so as to limit exposure of the viral vector to immune cells that mediate cytokine production. However, the disadvantage of local injection is that vector spread throughout the target tissues is often suboptimal. Furthermore, local administration of HAdV vectors is not intended to reach distant sites in the body that may contain migrating tumor cells. Thus, to exploit the advantages of systemic HAdV administration, it will be necessary to obtain a better understanding of how this virus provokes host inflammatory responses.

There are several known signaling pathways by which HAdV vectors induce proinflammatory cytokines, although the precise contribution of each of these pathways to the development of immunotoxicity remains to be elucidated. One signaling pathway involves Toll-like receptor 9 (TLR9),^5,6 a sensor of the innate immune system that recognizes the double-stranded DNA genome of the virus in cell endosomes. Entry of adenovirus particles into human plasmacytoid dendritic cells, a minor subset of the monocyte population in peripheral blood, causes TLR9 activation, and this leads to significant production of type 1 interferon a (IFN-1a). This interferon response is consistent with the ability of TLR signaling to induce the expression of a wide range of proinflammatory cytokines.⁷ Because plasmacytoid dendritic cells represent only a small fraction of the cells that can interact with HAdV, it is not surprising that other antigen-processing cells, including tissue macrophages, possess distinct signaling pathways that lead to cytokine production. For example, a cytosolic supramolecular complex known as the NLRP3 inflammasome is also capable of recognizing viral DNA in macrophages and, in so doing, triggers the activation and secretion of cytokine interleukin 1b (IL-1b).⁸ Recently, Nociari et al. also reported that interferon signaling in macrophages involves the activation of several cell kinases, including Jun N-terminal kinase and TBK1.⁹ This occurs in an interferon regulatory factor 3 –dependent manner and requires host–cell interactions with the viral capsid as well as with the viral genome using cytosolic DNA sensor(s).

In murine models of HAdV vector administration, systemic delivery results in a substantial amount of the virus being targeted to liver hepatocytes as well as resident macrophages known as Kupffer cells.^10,11,12,13 In addition, a smaller proportion of the virus, depending on the dose administered, also finds its way to the spleen. Di Paolo and colleagues recently reported in the journal Immunity that HAdV elicits a signaling response in CD169-negative/MARCO-positive macrophages, located in the marginal zone of the spleen, that leads to IL-1a production.¹⁴ They also observed the proteolytic cleavage of pro-IL-1a by neutral proteases.¹⁵ This allows transport of the NTP of the cytokine to the nucleus with concomitant secretion of the mature IL-1a molecule. HAdV infection of marginal zone macrophages was also shown to cause a significant increase in the expression of a defined set of other proinflammatory cytokines and chemokines, including CXCL1, CXCL2, MCP-1, IL-6, and IL-1b. The transcription of cytokine and chemokine messenger RNA occurred quite rapidly (~10 minutes after vector administration) and, except for IL-1a and IL-1b, also required a functional IL-R1 signaling pathway (Figure 1). These authors provided compelling evidence—using integrin-deficient mice as well as an HAdV vector that lacks an integrin-binding motif (arginine-glycine-aspartate) in the virus penton base protein—that integrin avb3 plays a prominent role in cytokine and chemokine production. Because av integrin interactions with the HAdV penton base arginine-glycine-aspartate sequence have previously been shown to mediate adenovirus internalization,¹⁶ the implication of this finding is that virus internalization is a prerequisite for optimal cytokine production. However, given that a complete abrogation of IL-1a production was not observed in integrin-deficient animals, this could indicate that other cell entry pathways also contribute to the proinflammatory response. Perhaps of greatest interest, the authors showed that a temperature-sensitive mutant adenovirus (ts1) that enters cells normally but fails to carry out membrane penetration^17,18 also induced suboptimal levels of cytokine production at both the RNA and protein levels. An interpretation of these data is that the major signals for optimal cytokine induction are not elicited at the virus-internalization step but instead are induced at the subsequent step of HAdV-mediated endosome disruption. This finding is reminiscent of the recent observations of Fejer and colleagues, who reported that interferon production in splenic myeloid dendritic cells also requires HAdV escape from cell endosomes.¹⁹

Figure 1.

Schematic diagram of adenovirus (Ad)-induced cytokine production in splenic macrophages as proposed by Di Paolo and co-workers.¹⁴ (1) Ad penton base interaction with integrin avb3 promotes virus internalization into early endosomes. (2) Partial disassembly of the virion in the low-pH environment of the endosome allows release of the viral membrane lytic protein that disrupts the lipid bilayer. (3) Membrane disruption acts as a signal for increased transcription of pro-interleukin-1a messenger RNA (pro-IL-1a mRNA) as well as increased cytokine protein production. (4) Proteolytic cleavage of pro-IL-1a by neutral proteases results in nuclear localization of the N-terminal (NTP) fragment and secretion of the mature IL-1a protein. (5) binding of IL-1a to its receptor (IL-1R1) induces further signaling that produces a defined set of cytokines and chemokines (see text). MCP-1, monocyte chemoattractant protein-1; mRNA, messenger RNA.

A major question arising from these recent investigations is how HAdV-mediated endosome disruption leads to cell signaling and cytokine production. Although additional details of these processes remain to be uncovered, endosomal membrane fragmentation by various nonenveloped viruses could represent a way for the host immune system to distinguish between normal cellular and pathogenic events. The studies by Di Paoli and co-workers also raise the possibility of modifying the HAdV capsid (i.e., penton base) so as to limit interactions with integrins and/or the downstream endosomal membrane penetration step, thereby damping potentially harmful inflammatory responses. This will probably be a tricky endeavor, in that the efficiency of HAdV entry and gene delivery depends in large part on the key step of endosome penetration that is in turn regulated by integrin association. The dose and precise route of vector administration are additional variables that may contribute to the overall pattern of cytokine responses. Finally, it is worth noting that the majority of studies on innate immune responses to HAdV have been performed using murine models, and it will thus be important to correlate the findings obtained with these inbred animals with those occurring in nonhuman primates or humans.