Virus self-improvement through inflammation: No pain, no gain

It is becoming increasingly clear that there are circumstances where human cytomegalovirus (CMV) benefits from inducing a strong host inflammatory response. The virus encodes a number of immunomodulatory functions, including proinflammatory cytokines and chemokines, and, like other viruses that persist in the host, commits many viral functions to avoidance of the host immune response (1, 2). The work reported in this issue of PNAS by Zhu et al. (3) shows that prostaglandin (PG) E₂, a key mediator in the inflammatory response, is induced and performs an important function to support viral replication. These authors show that PG synthesis follows induction of cyclooxygenase-2 (COX-2) and cytoplasmic phospholipase A₂ transcriptional activation (4) during CMV infection. The increased level of PGE₂ in turn influences the expression of the critical viral transcriptional regulatory protein (IE2) and through this impact, strongly reduces replication and production of progeny (3).

COX-2 (also known as PG H synthetase) catalyses the first two steps in the synthesis of PG from arachidonic acid to PGH_2, an intermediate that is rapidly converted by additional enzymes to any of a number of physiologically active, tissue-specific mediators, including PGD₂, PGE₂, PGF₂, PGI₂, and the related prostanoid, thromboboxane A₂ (5). COX-2 is one of the most strongly induced host genes following CMV infection of cultured primary human fibroblasts (4). PG levels are well known to increase rapidly in response to a number of cell activation and inflammatory signals such as bacterial lipopolysaccharide and stimulation by proinflammatory cytokines interleukin-1β and tumor necrosis factor α (5). Exposure of cultured cells to a variety of microbial and viral pathogens causes a strong activation of this gene (6–8). In the course of natural infection, CMV interacts with cell types in the epithelial, endothelial, and hematopoietic lineages, all of which produce distinct sets of prostanoids that may now be considered candidate mediators of viral replication based on the importance of PGE₂ in cultured fibroblasts (3).

Prostaglandin (PG) E₂, a key mediator in the inflammatory response, performs an important function to support viral replication.

COX-2 is the critical site of action for antiinflammatory drugs like aspirin, indomethacin, and other nonsteroidal antiinflammatory drugs (NSAIDs), as well as for a growing list of specific COX-2 inhibitors (coxibs) like celecoxib (Celebrex) and rofecoxib (Vioxx). COX-2 inhibitors are in widespread use to control chronic inflammatory diseases, relieve vasodilation, reduce fever, and control pain (9) mediated by the prostanoids. Although not generally used to control viral disease in patients, NSAIDs have been shown to inhibit replication of other RNA and DNA viruses. Evidence that induced PGE₂ benefits CMV replication at a critical regulatory step suggests that other viruses and microbial agents may benefit from the impact of PG induction. This finding leads to the possible targeting this pathway as an auxiliary means of controlling infections.

Following exposure of cells to CMV, the synthesis and release of PGE₂ becomes maximally elevated within the first 24 h, well before release of progeny virus has begun. This induction initiates very early after exposure to virus (4, 10). Induction of this pathway, which can occur following exposure of cells to either infectious virus or virus particles, does not require any viral gene expression and occurs in nonpermissive cells (3, 11). This, in itself, is not surprising because diverse microbial agents seem to strongly activate COX-2 expression in mammalian cells (6–8). A structural component of CMV or the act of binding to cells may initiate this event (Fig. 1), and infectious virus yields less induction than inactivated virus particles (10), suggesting that a function expressed during infection limits the levels of activation. Soluble CMV envelope glycoprotein gB triggers a dramatic impact on the host cell (12), but is only a modest inducer of COX-2. The remarkable feature of CMV is of course that there is a specific role for the resulting increased prostaglandin levels. Inhibitors of both COX-2 and phospholipase A₂, an enzyme that drives release of arachidonic acid from membrane stores, block viral replication and reduce viral gene expression due to reduced levels of PGE₂ (3, 11, 13). Among the viruses that benefit from COX-2 function is another herpesvirus, herpes simplex virus, that was long ago shown to benefit from increased prostaglandin synthesis in cell culture, as well as in animal models (14), but this benefit has not shown up in the clinic. Increased replication resulting from proinflammatory consequences of increased PGE₂ fits with the success of CMV in proinflammatory disease settings, such as its frequent reactivation from latency and active replication in allograft transplant recipients (15).

Figure 1.

A model for increased prostaglandin E2 synthesis in CMV replication. Cell activation via binding or other effects of viral structural proteins leads to rapid transcriptional activation (in the nucleus) of two key cellular genes, encoding cytoplasmic phospholipase (cPLA) and cyclooxygenase-2 (COX-2). cPLA translocates to membranes of endoplasmic reticulum and outer nuclear membrane and directs arachidonic acid (AA* to AA) release from membrane lipids. In a step that is inhibited by NSAIDs or coxibs, COX-2 converts AA to the intermediate PGH₂, which is then converted to PGE₂ by prostaglandin E synthase (PGES). PGE₂ release from cells is via the prostaglandin transporter (PGT). Released PGE₂ may function in an autocrine or paracrine fashion on cells via one of four G-protein-coupled prostaglandin E (EP₁–EP₄) receptors on the plasma membrane, with the EP₂ receptor most abundant on human fibroblasts. The autocrine activation loop is depicted here. Intracellular signaling by PGE₂ may also contribute to autocrine activation. Activation of EP₂ results in activation of cAMP and downstream effects modulating transcriptional and posttranscriptional levels of transcripts encoding the key viral regulatory protein, IE2. (For more information, see ref. 16.)

How does PGE₂, the major PG produced in human fibroblasts and a major proinflammatory mediator in host tissues, likely alter the transcript levels for a key viral regulatory protein? PGE₂, like all prostanoid mediators, is short-lived, rapidly secreted from cells, and active only on proximal cell targets. These mediators may function in an autocrine manner, possibly without even being secreted from the cells producing them (16). As the major product of COX-2 activation in most cell types, including the fibroblasts used to study CMV, PGE₂ accumulates through the action of microsomal PGE synthase, which may be predicted from the current report also to be induced during viral infection. Evidence that PGE synthase and COX-2 are coordinately regulated in other systems (16) suggests that this may occur in virus-infected cells. Like other PGs, PGE₂ is rapidly transported from cells via a general PG transporter and acts locally via one of four G-protein coupled, seven transmembrane spanning PGE receptors (EP₁–EP₄; refs. 16 and 17). Positive-acting receptors in this group were first characterized by the relaxation response of smooth muscle mediated via cyclic adenosine monophosphate (cAMP) and may mediate a general stimulation via protein kinases and phospholipase C to produce both inositol-1,4,5-trisphosphate (IP₃) and diacylglycerol. PGE₂ is unique among prostanoids in mediating effects via an array of receptors encoded by four different genes, with additional receptor variants being generated by differential mRNA splicing. Although a precise understanding of the cell and tissue distribution of these receptors is still developing, they localize on the cell surface, cytoplasmic, and nuclear membranes of cultured cells. PGE₂ receptors mediate signaling in one of three ways, which are grossly oversimplified here. EP₁ signaling is via G_q and causes release of intracellular calcium stores. EP₂ and EP₄ mediate cellular responses by signaling via G_s to increase cAMP levels, whereas EP₃ inhibits cellular responses by signaling via G_i to decrease intracellular cAMP (16, 17). The diverse number of prostanoid ligands produced as a result of COX-2 activation, along with combinations receptors that may be encountered on a particular cell type, all suggest a very complex relationship exists in nature. This being said, PGE₂ signaling to CMV infected fibroblasts may be expected to follow similar themes from what has been surmised in other settings (Fig. 1).

Given that each cell type carries a distinct set of prostanoid receptors, it is worth considering which PGE₂ receptor(s) predominate on fetal human diploid fibroblasts. These cells receive PGE₂ signals predominantly via the EP₂ receptor, although they express low levels of EP₃ and EP₄ (18). Of course, the dramatic impact of CMV infection on other host gene expression (4, 10) may extend to these receptors. The type of receptor predominating on infected cells will need to be ascertained experimentally. Assuming that the major consequences of PGE₂-mediated activation by CMV infection of fibroblasts occurs through EP₂, the result would be a broad activation of cAMP with downstream consequences that might be similar to treatment with other cAMP or protein kinase inducers. The enhancer-promoter transcriptional element that controls expression of the CMV IE2 regulatory gene product is induced via cAMP response elements (19) and has been shown to be a target of PGE₂-mediated activation. As discussed in the Zhu et al. report (3), however, this gene gives rise to IE2 via differential splicing and other major product of the transcription unit, IE1, is not significantly affected by PG inhibitors. Although posttranscriptional regulation in this region may be occurring, there is as yet little insight into how conditions such as cAMP induction might alter the relative amounts of these two key immediate early gene products.

The relative roles of different PGE₂ receptors in various settings have been ascertained with specific antagonists such as EP₂ inhibitors 11-deoxy-PGE 2 or butaprost, as well as through the study of receptor-knockout strains of mice. EP₂ has been characterized for its role in smooth muscle relaxation in the pulmonary, intestinal, renal, and reproductive systems. These types of studies have served to support a central role of PGE₂-mediated signaling across a wide variety of physiological systems, including ovulation and uterine implantation, bone development, renal and gastrointestinal function, vascular hypertension, pulmonary function, and of course, the immune system (17, 20). PGE₂ is also a component of the immediate inflammatory response that interfaces with the INF system. Thus, modulation of just this one PG has predicted consequences far beyond its benefit to virus replication efficiency.

There is a growing range of diseases attributed to CMV infection (15). Cytomegalovirus infects a majority of adults without causing much disease but has been recognized as a major opportunist causing significant disease in the immunocompromised host. This virus remains the leading infectious cause of birth defects, a situation that results in thousands of hearing-impaired children being born every year in the U.S. The incidence of congenital infection has not been accurately determined in large studies, but is estimated to be in the range of one to 10% of all pregnancies (15). The combined knowledge that there is a dramatic impact of the virus on PGE₂ and that there is a clear role of this PG in uterine function during pregnancy means that a closer evaluation of this important disease spectrum is warranted. Infection in adults targets hematopoietic lineage cells that support CMV dissemination and life-long latency; whereas infection of the fetus targets additional cell types, including neurons that control sound perception. CMV is also a leading cause of acute as well as chronic problems in immunocompromised hematopoietic as well as solid organ allograft transplant recipients where it has been accused of increasing the severity of tissue rejection. Although antiviral drugs are used to control CMV, their toxicity at doses needed to inhibit replication continues to present difficulties. Vascular problems that follow CMV reactivation and active infection in allograft recipients have received continued attention and have contributed to the expectation that this virus may be the cause of more conventional chronic vascular disease. Again, the well known importance of prostanoid mediators in vascular physiology and inflammation make this an area that warrants further clinical and experimental investigation.

Finally, it is not just the induction of prostanoids that makes CMV a proinflammatory threat to the host. Whereas increased inflammation that results from the induction of host PG synthesis may be prolonged by viral infection, host cytokines and chemokines that are induced by viral infection also play very important roles in modulating inflammatory and immune responses. On top of these effects, the potent proinflammatory effects of a virus-encoded chemokine UL146/vCXC-1 (21) and a very active chemokine receptor, US28 (22), are likely to contribute (2, 19). Given the lack of a current means of preventing CMV infection in the population, the ramifications of infection by this large and genetically complex virus will continue to be felt. CMV, and its relatives that infect laboratory mammals, will continue to reveal the rich evolutionary relationship that balanced the interests of the host and the virus and will, it is hoped, yield some information on how these processes result in disease. The pathogenesis of CMV infection and disease has evolved through a battle of genetic one-upmanship between virus and host that has not yet been fully elaborated on either side.