CD8 T lymphocytes control microbial infections in two (LCMV)(Bouley et al., 1995; Fiette et al., 1995; Finke et general ways: by secreting cytokines, such as IFN and al., 1995; Hwang et al., 1995; van den Broek et al., 1995a, TNF, and by lysing infected cells via the perforin or 1995b; Guidotti and Chisari, 1996; Topham et al., 1996; Fas/FasL pathways. These two functions differ in that Tay and Welsh, 1997). This indicates that cytokines can cytokine release can have both local and systemic conplay a role in limiting infections by both lytic and nonsequences, whereas cell lysis is limited to target cells lytic viruses. Although LCMV can be cleared noncytothat are in direct contact with the T cell. It has been pathically from some cells (eg, hepatocytes)(Guidotti generally assumed that for controlling viral infections, et al., 1999a), complete clearance of nonlytic viruses the cytolytic function of CD8 T cells far outweighed such as LCMV often requires perforin-mediated lysis of the contribution made by their release of cytokines. infected cells (Kagi et al., 1994; Walsh et al., 1994). In However, several studies have shown that exposure to contrast, perforin-mediated lysis may be less important cytokines can directly reduce viral replication and, in in controlling infection by lytic viruses, since these some cases, cytokines alone are able to “cure” infected agents inevitably destroy their host cells during their cells by inactivating viral replication in the absence of replicative cycle (Kagi et al., 1996). The remarkable findcell death (Guidotti and Chisari, 1996; Estcourt et al., ing that cytokines can abolish viral replication without 1998). This indicates that target cell lysis and cytokine cytolysis (eg, LCMV and HBV replication in hepatoproduction play complementary roles in controlling incytes)(Guidotti et al., 1996, 1999a, 1999b) demonstrates fection. However, T cell–mediated control of virus infecan example of the ultimate antiviral defense strategy; tion can be costly to the host, because it is almost the invading pathogen is eradicated without the expense inevitably accompanied by some degree of immunopa- of destroying the infected host cell. thology. In many viral infections, signs and symptoms How is viral replication blocked without harming the worsen even as viral titers decline, resulting from the infected cell? In vitro experiments have shown that dihost’s immune response to the virus rather than from the rect exposure to cytokines such as the interferons or destructive effects of the virus per se. This minireview TNF/induces cellular proteins that can decrease tranfocuses on cytokine release; the immunopathological scription and/or replication of several viruses including sequelae of perforin-mediated lysis, although some- influenza, VSV, HSV, adenovirus, encephalomyocarditis times serious (Gebhard et al., 1998), will not be further virus, and parainfluenza virus (Horisberger et al., 1983; discussed herein. Mestan et al., 1986; Wong and Goeddel, 1986; Kumar Production of inflammatory cytokines in response to et al., 1988; Staeheli and Pavlovic, 1991; Zhao et al., infection can cause immunopathology that ranges from 1996). Some of the most extensively studied cellular minor (eg, low-gradefever) tosevere (eg, organfailure proteins that mediate these responses are doubleand death). How do T cells maintain a balance between stranded RNA activated protein kinase (PKR), 2-5-oliantiviral efficacy and the hazards of unbridled cytokine goadenylate synthetase (2–5OAS), dsRNA-specific adenrelease? Antigen-specificTcellshavedevelopedregula- osine deaminase (dsRAD), and MxA (Katze, 1995; tory mechanisms that allow them to very rapidly turn on Boehm et al., 1997; Haller et al …