IFN-αβ-Mediated Inflammatory Responses and Antiviral Defense in Liver Is TLR9-Independent but MyD88-Dependent during Murine Cytomegalovirus Infection¹

Host defense against acute viral infection requires induction of innate immune responses to rapidly limit viral replication and virus-induced pathology. The hepatotropic virus murine CMV (MCMV)⁴ induces an innate cytokine and chemokine network that is critical for promoting the development of appropriate antiviral responses (1). Previous studies have demonstrated that IFN-αβ-mediated effects promote production of the chemokine CCL2 from resident liver cells (2). In addition, CCL2/CCR2 interactions were required for recruitment of CCL3-producing macrophages and induction of NK cell inflammation and IFN-γ delivery to support control of infection and limit virus-induced disease in the liver (2, 3, 4, 5). Thus, the essential functions of IFN-αβ in antiviral immunity and promotion of inflammatory responses have been well defined (1, 2, 5, 6, 7). Nevertheless, the cellular sources and upstream mechanisms promoting IFN-αβ production in liver during MCMV infection have not been determined.

Because of their ability to release IFN-α and other proinflammatory cytokines, plasmacytoid dendritic cells (pDCs) are integral in the activation of innate, and regulation of adaptive, immune responses that function in the recognition and eradication of virus-infected cells (8, 9, 10, 11, 12). The pDCs have been isolated ex vivo during MCMV infection and characterized from various tissues such as spleen and blood (8, 10, 13). It is also highly likely that the recently described liver pDCs (14) will function as major determinants of hepatic immunity. However, pDCs within the liver have not been evaluated during MCMV infection, and their in vivo contribution to IFN-α production has not yet been determined.

Several studies have demonstrated that APCs including pDCs recognize and respond to viral challenge primarily through TLRs (15, 16, 17, 18, 19, 20). It has also been established that murine pDCs uniquely express TLR7 and TLR9 (17, 18, 19, 20, 21) and transduce intracellular signals through the adaptor molecule MyD88 to promote rapid and significant release of IFN-α (18, 22, 23). TLR7 expression promotes selective responsiveness to ssRNA viruses such as Influenza and vesicular stomatitis virus, whereas DNA containing viruses including HSV-1, HSV-2, and MCMV use TLR9 (22, 24, 25, 26, 27, 28, 29, 30). Moreover, IFN-α production by pDCs in blood and spleen has been shown to be dependent upon a pathway mediated by TLR9 and MyD88 during MCMV infection (13, 30). Compartmental differences have been described with respect to spleen and liver requirements for establishing localized defense against infection with MCMV (31, 32). Therefore, mechanisms mediating IFN-α production in other tissues may not necessarily reflect those that are required in the liver.

In this study, the cellular sources of IFN-α production and the effects of TLR9 and MyD88 in promotion of innate inflammatory responses and antiviral defense in the liver were determined following MCMV infection. The results demonstrate that hepatic pDCs are indeed major contributors to IFN-α expression in vivo. However, in striking contrast to the spleen, IFN-α release from hepatic pDCs is largely TLR9-independent but dependent on MyD88 functions for efficient production. Furthermore, responses downstream of IFN-α, including production of CCL2, CCL3, and IFN-γ and macrophage and NK cell accumulation, were also dependent upon expression of MyD88 but not TLR9. The results also demonstrate a requirement for MyD88 in limiting virus-induced liver pathology and viral replication, whereas TLR9 antiviral effects were minimal at this site. Taken together, this study defines contrasting and tissue-specific functions for MyD88 and TLR9, and suggests that the tissue environment or source of pDCs influences the pathways leading to IFN-αβ production and initiation of inflammatory responses promoting antiviral defense at a site of active infection.

TLR9-deficient (TLR9^−/−) mice (23) and MyD88-deficient (MyD88^−/−) mice (33) (both in the C57BL/6 × 129Sv mixed background) as well as heterozygote littermate controls were obtained from Dr. W.-M. Chu (Brown University, Providence, RI). Breeding pairs of homozygous TLR9^−/− mice in the C57BL/6 background were provided by Dr. B. Beutler (The Scripps Research Institute, La Jolla, CA) (34), and were kept and bred under pathogen-free conditions at Brown University. Specific pathogen-free C57BL/6 mice were purchased from The Jackson Laboratory. Male and female mice were used between 5 and 6 wk of age. Mouse handling and experimental procedures were conducted in accordance with institutional guidelines for animal care and use.

Stocks of Smith-strain MCMV salivary gland extracts were prepared (5). Mice were either uninfected or infected i.p. on day 0 with a moderate dose of 5 × 10⁴ PFU/mouse. For viral titer determination, organs were weighed, homogenized in cold supplemented RPMI 1640 medium (Invitrogen Life Technologies), and supernatants were obtained following centrifugation. Duplicate samples of serially diluted samples were used to infect monolayers of NIH 3T3 fibroblasts (American Type Culture Collection), and viral titers were determined by standard plaque assay as previously described (2, 35).

Liver and splenic leukocytes were prepared as previously described (2, 3, 4, 5). For generation of conditioned medium, leukocytes were plated without additional stimulation in flat-bottom microtiter plates at 10⁶ cells/well in RPMI 1640 supplemented with 10% heat-inactivated FBS (HyClone Laboratories). After 24 h of incubation at 37°C, supernatants were collected and used for cytokine analyses.

For DC enrichment, total liver leukocytes were pooled and incubated with CD11c-FITC and Ly6C-PE (BD Biosciences), followed by a double magnetic bead separation using a FITC multisort kit and an AutoMacs machine according to manufacturer’s protocol (Miltenyi Biotec). DCs were sorted into Ly6C⁺ (pDC) and Ly6C⁻ (non-pDC) populations (9, 10, 26, 30) and plated at the lowest common denominator for generation of conditioned medium. The flow through was also collected to recover CD11c⁻ cells. All sorting resulted in at least 82% purity of the desired population.

Conditioned medium supernatants were tested for IFN-α, CCL2, CCL3, and IFN-γ using commercial sandwich ELISA (PBL Biomedical Laboratories; R&D Systems). Limits of detection were as follows: CCL2 (8–10 pg/ml), CCL3 (18–25 pg/ml), IFN-γ (20–25 pg/ml), and IFN-α (5–10 pg/ml). IFN-αβ was also measured in sorted liver cell populations using an in vitro biological assay as previously described (36). Briefly, serial dilutions of liver leukocytes were incubated with L929 cells (American Type Culture Collection) for 24 h, after which vesicular stomatitis virus was added. Protection against vesicular stomatitis virus pathology was evaluated, with 50% protection being equivalent to 1 U/ml IFN. Limit of detection for IFN-αβ was 8 U/ml.

Cell surface staining was performed as previously described (2) using the following Abs: F4/80-PE and CD11b-FITC to identify macrophages; NK1.1-PE and TCR-β-allophycocyanin to distinguish NK cells. All Abs were diluted in the presence of anti-CD16/CD32 mAb (clone 2.4G2) to block nonspecific binding of Abs to the receptor of the Fc portion of the Ig. Isotype control Abs were used to correct for background fluorescence and set analysis gates. All Abs were obtained from BD Biosciences, except for anti-F4/80-PE, which was purchased from Serotec. Intracellular IFN-α staining was performed on liver and splenic leukocytes isolated from uninfected or 1.5 day MCMV-infected mice. Briefly, leukocytes were plated at 10⁶ cells/100 μl, and incubated for 4 h with brefeldin A (Sigma- Aldrich), followed by blocking and staining for surface expression of pDC Ag-1 (PDCA-1; Miltenyi Biotec). Samples were fixed overnight and then permeabilized using a Cytofix/Cytoperm kit (BD Biosciences). Cells were then stained with 1 μl/100 μl IFN-α-FITC (PBL Biomedical Laboratories) diluted in Cytoperm buffer for 30 min at 4°C. Stained cell samples were acquired using a FACSCalibur and analyzed using CellQuest software (BD Biosciences).

Liver lobes were fixed in 10% buffered formalin and paraffin-embedded for sectioning. Tissue sections were stained with H&E and microscopically observed. Overall liver damage was assessed by measuring alanine aminotransferase (ALT) levels in serum samples using a colorimetric kit according to the manufacturer’s instructions (Biotron Diagnostics).

Statistical significance of experimental results was analyzed by two-tailed Student’s t test where indicated, and considered at a value of p ≤ 0.05.

During MCMV infection, maximal levels of IFN-α protein are induced by 36–40 h (or 1.5 days) after infection (9, 36). Moreover, this cytokine is produced in the liver (5). As the pDC subset of DC has been established as an important source of IFN-α in spleen during infection with MCMV (8, 9), hepatic pDCs were evaluated for their in vivo production of IFN-αβ by bioassay in 40 h MCMV-infected mice. Conditioned medium prepared from pDC- (CD11c⁺Ly6C⁺), non-pDC- (CD11c⁺Ly6C⁻), or CD11c⁻-enriched liver leukocytes (Fig. 1,A) demonstrated significant increases in IFN-αβ production in pDCs (Fig. 1 B). In contrast to CD11c⁻ cultures, IFN-αβ was above the limit of detection in culture supernatants from non-pDCs, but at levels that were 10-fold less than what was observed in the pDC population. Thus, pDCs and non-pDCs contribute to IFN-αβ production, but pDCs are the principal producers of high levels of the cytokines in the liver.

Effective activation of DCs through TLR9-dependent mechanisms has been shown to support antiviral responses (25, 26, 30, 34). In particular, TLR9 was demonstrated to promote splenic production of IFN-α by pDCs during MCMV infection (30). To ascertain whether TLR9 mediates IFN-α production in the liver, leukocyte conditioned medium was prepared from livers of uninfected or 40 h MCMV-infected TLR9^−/− or their heterozygous wild-type (WT) littermates. ELISA was used to assess cultured supernatants for IFN-α production. The results demonstrate comparable levels of secreted IFN-α in liver leukocytes isolated from infected WT and TLR9^−/− mice (Fig. 2,A). In contrast, splenocyte culture supernatants derived from TLR9^−/− mice exhibited significantly reduced levels of IFN-α as compared with WT mice (Fig. 2,B), concurrent with previous observations (30). Similar contrasting effects of TLR9 on IFN-α production were observed in liver and spleen leukocyte conditioned medium generated from infected C57BL/6-TLR9-deficient mice and their WT counterparts (data not shown). IFN-α was not detected in liver or spleen of uninfected mice (Fig. 2 and data not shown). Taken together, these results suggest that although TLR9 signaling contributes to production of IFN-α in spleen, it is not essential in inducing production of the cytokine in the liver during infection with MCMV.

Our previous studies have demonstrated that IFN-αβ-mediated effects were required for CCL2 production and the recruitment of CCL3-producing macrophages that induce NK cell inflammation and IFN-γ delivery to the liver during MCMV infection (2, 3, 5). To assess the contribution of TLR9-dependent responses on production of chemokines and subsequent effective recruitment of inflammatory cells, liver leukocytes were prepared from uninfected or 40 h MCMV-infected heterozygous WT littermates and TLR9^−/− mice. Assessment of culture supernatants by ELISA demonstrated comparable levels of induced CCL2 (Fig. 3,A) and CCL3 (Fig. 3,B) proteins in the livers of infected WT control and TLR9^−/− mice, as well as similar levels of IFN-γ protein (Fig. 3,C). Concurrent with these observations, WT and TLR9^−/− mice showed a 2-fold increase in the frequency of accumulated macrophages (F4/80⁺CD11b⁺) and NK cells (NK1.1⁺TCRβ⁻) in response to infection when compared with respective uninfected mice (Fig. 3, D and E). Similar profiles of chemokine induction and inflammatory cell accumulation were observed in liver leukocytes of 40 h MCMV-infected C57BL6-TLR9^−/− animals and their WT counterparts (data not shown). Together, these results demonstrate that TLR9 signaling is not essential for promotion of IFN-αβ-dependent chemokine responses that mediate inflammatory cell recruitment to the liver during MCMV infection.

MyD88 has been shown to be required for IFN-α secretion from pDCs in the spleen and resistance to MCMV infection (13, 30, 34). To examine the contribution of MyD88 to IFN-α production in liver, ELISA was used to evaluate cytokine production in leukocyte conditioned medium prepared from uninfected or 40 h MCMV-infected heterozygous WT littermates and MyD88^−/− mice. IFN-α production was induced in liver samples from WT and MyD88^−/− mice after infection (Fig. 4,A). However, whereas IFN-α responses in WT mice reached values of 99 ± 24 pg/ml, MyD88^−/− mice only exhibited values of 20 ± 8 pg/ml. Thus, MyD88 deficiency resulted in a profound 80% reduction in the production of IFN-α when compared with WT mice. The effect of MyD88 deficiency on MCMV-induced production of IFN-α in the spleen was also evaluated. Spleen leukocyte conditioned medium prepared from MyD88^−/− mice produced significantly less IFN-α (31 ± 8 pg/ml) when compared with their WT counterpart cytokine values of 72 ± 15 pg/ml at 40 h MCMV infection (Fig. 4 B). Uninfected mice did not have detectable levels of the cytokine. These results indicate that IFN-α production in liver and spleen is dependent on signaling through MyD88.

To determine whether MyD88-dependent responses mediate induction of the chemokines involved in promoting inflammatory cell trafficking to liver, CCL2 and CCL3 secretion was measured in liver leukocyte conditioned medium prepared from uninfected or 40-h MCMV-infected WT and MyD88^−/− mice using ELISA. Uninfected mice demonstrated comparable levels of chemokines (Fig. 5, A and B). However, infected WT mice had a pronounced 3- and 5-fold elevation in CCL2 (Fig. 5,A) and CCL3 (Fig. 5,B) proteins, respectively, when compared with uninfected mice. In contrast, chemokine induction in MyD88^−/− mice was significantly affected such that CCL2 and CCL3 levels were markedly reduced by 50% and 80%, respectively, when compared with infected WT mice. MCMV-infected MyD88^−/− mice also produced significantly less IFN-γ when compared with WT counterparts (Fig. 5 C).

The effects of these reduced chemokine responses on inflammatory cell recruitment was evaluated in liver leukocytes prepared from uninfected or 40-h MCMV-infected WT and MyD88^−/− mice using flow cytometric analysis. The results show that MCMV-infected WT mice had a 2- and 3-fold increase in the frequency of accumulated macrophages and NK cells in the liver, respectively. In stark contrast, effective accumulation of these inflammatory cells was markedly impaired in infected MyD88^−/− mice (Fig. 5, D and E). As the total liver leukocyte population for infected WT and MyD88^−/− mice was comparable at 4.5 ± 0.02 × 10⁶ and 4.6 ± 0.01 × 10⁶, respectively, the observed differences in inflammatory cell recruitment are not attributed to changes in total cell yields. Furthermore, as uninfected liver leukocytes prepared from WT and MyD88^−/− mice generated comparable levels of CCL2 protein when stimulated ex vivo with recombinant IFNα, MyD88 is not itself directly required for CCL2 production. Collectively, these results demonstrate that unlike TLR9 signaling, MyD88 and the downstream induction of IFN-α is required for effective chemokine responses and inflammatory cell recruitment to liver during infection with MCMV.

It is well established that innate immune responses are essential in the early control of MCMV replication (7). In the liver, previous studies have demonstrated increased viral burdens and profound virus-induced pathology between days 3 and 5 of infection in mice deficient in CCL2 and CCL3 production, in addition to macrophage and NK cell recruitment (2, 3, 4, 5). Therefore, the effects of TLR9 and MyD88 on liver pathology and antiviral defense were evaluated. To first assess overall hepatic damage, expression of the liver enzyme ALT was measured in serum samples from heterozygous littermate controls and TLR9^−/− or MyD88^−/− mice that were either uninfected or infected with MCMV for 5 days. Uninfected or 3 and 5 day MCMV-infected C57BL/6 controls and C57BL/6-TLR9^−/− mice were also included. The results show that uninfected mice within each group exhibited comparable baseline levels of circulating ALT (Fig. 6, A–C). By day 3 or 5 after infection, significant elevations in ALT levels were detected in each group of mice. However, whereas all TLR9^−/− mice had similar levels of the enzyme when compared with their control counterparts (Fig. 6, A and B), MyD88^−/− mice reached values that were 2-fold higher than the ALT values observed in corresponding control mice (Fig. 6 C), indicating augmented liver disease in the absence of MyD88.

To evaluate the contribution of TLR9 and MyD88 to antiviral defense, liver and spleen samples were prepared from control animals and TLR9^−/− or MyD88^−/− mice. Mice were infected with 5 × 10⁴ PFU MCMV for 5 days, and viral burdens were evaluated. Results show viral titers in liver that were comparable between control and TLR9^−/− mice (Fig. 6,D). Additionally, although the viral burden in the spleen of TLR9^−/− mice was 1 log higher as compared with control mice, the increase was not statistically significant. In striking contrast, viral titers in liver and spleen were increased by 2 log in MyD88^−/− mice when compared with those in control mice (Fig. 6 E).

Because TLR9 has been shown to promote control of virus replication in spleen before day 5 of MCMV infection (13, 30, 34), viral burden was measured in liver and spleen samples from C57BL/6 and C57BL/6-TLR9^−/− mice on days 3 and 5 following MCMV infection. Interestingly, viral titers in the livers of both groups of mice remained comparable at day 3 and 5 of infection (Fig. 6,F). Yet, TLR9^−/− mice had splenic viral titers that were 2 log higher at day 3 of infection when compared with their counterpart controls at this time point (Fig. 6 G). However, by day 5 the viral titers were dramatically reduced by 64% in TLR9^−/− mice when compared with viral titers observed on day 3 of infection. Moreover, in host survival studies wherein controls and TLR9^−/− mice were challenged with 10⁵ PFU of MCMV, both groups of mice survived beyond 30 days of infection with this virus dose (data not shown). Thus, MyD88 promotes comparable antiviral defenses in liver and spleen. However, although TLR9 does contribute to the initial control of virus replication in the spleen, its effects in the liver are marginal.

To further characterize the role of TLR9 and MyD88 responses on antiviral defense, H&E-stained liver sections were prepared from paraffin-embedded liver sections from WT controls (Fig. 7, A and D), MyD88^−/− (Fig. 7, B and E), and TLR9^−/− (Fig. 7, C and F) mice that were either uninfected or infected with MCMV for 5 days. All uninfected mice appeared histologically similar (Fig. 7, A–C). Infected control mice did show evidence of small clusters of tightly compact nucleated cells in distinct regions throughout the liver (Fig. 7,D). These inflammatory foci have been shown to contain primarily NK cells (3, 5) and have been associated with antiviral defense (1, 2, 3, 37). Furthermore, infected hepatocytes with characteristic cytomegalic inclusion bodies were not prominently visible, and significant overt pathology was not evident in the control mice. In sharp contrast, large areas of necrotic lesions and significant numbers of cytomegalic inclusion bodies were readily observed in liver sections of MyD88^−/− mice (Fig. 7,E, and inset). Additionally, inflammatory foci were not as prominent. Interestingly, the liver pathology of infected TLR9^−/− mice (Fig. 7,F) was remarkably similar to the pathology observed in infected controls (Fig. 7 D). Collectively, these results clearly indicate a role for MyD88 but not TLR9 in promoting antiviral defense and limiting the development of virus-induced liver pathology.

Having characterized opposing TLR9 responses for IFN-α production in the spleen and liver under conditions of MCMV infection, the direct effects of TLR9 expression on pDCs for production of intracellular IFN-α in these tissues was evaluated. Spleen and liver leukocytes were prepared from control or TLR9^−/− mice that were uninfected or infected with MCMV for 1.5 days. The frequency of IFN-α-producing pDCs was then determined by intracellular expression of the cytokine. The results show induced expression of IFN-α in total spleen cells from control and TLR9^−/− mice upon infection (Fig. 8,A). However, the increase in the frequency of total IFN-α-producing cells observed in TLR9^−/− mice was 2-fold less than the increase observed in control mice. As the spleen cell numbers for total IFN-α⁺ leukocytes for control and TLR9^−/− mice increased to 9 ± 1 × 10⁵ and 5 ± 0.05 × 10⁵ from the uninfected values of 1 ± 0.1 × 10⁵ and 1 × 10⁵ ± 0, respectively, there was also a significant (p ≤ 0.04) 2-fold reduction in the absolute number of total IFN-α⁺ spleen cells in TLR9^−/− mice when compared with controls. Furthermore, the frequency of pDCs expressing IFN-α (IFN-α⁺PDCA-1⁺) in the spleen was profoundly reduced in TLR9^−/− mice when compared with their counterpart controls (Fig. 8, B and E). Because the total IFN-α⁺PDCA-1⁺ number for control and TLR9^−/− mice increased to 12 ± 1 × 10⁵ and 4 ± 0.2 × 10⁵ from the uninfected values of 5 ± 2 × 10⁵ and 2 ± 0.5 × 10⁵, respectively, a significant (p ≤ 0.004) 3-fold decrease in the absolute number of IFN-α-expressing pDCs was observed in the spleens of TLR9^−/− mice as compared with controls. In marked contrast, the proportion of total (Fig. 8,C) and PDCA-1⁺ (Fig. 8, D and F) liver leukocytes expressing IFN-α was comparable between control and TLR9^−/− mice during infection. Moreover, the total IFN-α⁺ liver leukocytes in control and TLR9^−/− mice increased likewise to 1 ± 0.1 × 10⁵ and 1 ± 0.04 × 10⁵ from the uninfected values of 1.3 ± 0.4 × 10⁴ and 1.9 ± 0.5 × 10⁴, respectively. The absolute number of IFN-α⁺PDCA-1⁺ liver cells was also remarkably similar in infected control and TLR9^−/− mice, with respective values of 1.09 ± 0.1 × 10⁵ and 1 ± 0.2 × 10⁵. Taken together, the results demonstrate that expression of TLR9 is required in the spleen, but not the liver, for efficient production of pDC-derived IFN-α protein in response to MCMV infection.

In this study, we demonstrate that pDCs secrete IFN-α in response to MCMV infection in the liver, and that MyD88 but not TLR9 signaling mediates this effect. Unsurprisingly, responses in the spleen depended on both TLR9 and MyD88 signaling pathways, as the absence of either of these molecules lead to impaired IFN-α production at this site. The results also show that TLR9 was not essential in the promotion of IFN-α-mediated inflammatory cascades in the liver. CCL2, CCL3, and IFN-γ production, as well as macrophage and NK cell recruitment to this site was not impaired in TLR9^−/− mice. In contrast, expression of MyD88 was critical in promoting chemokine and cytokine production, and inflammatory cell recruitment. Concurrent with these observations, TLR9-deficient mice were highly resistant to virus-induced hepatic disease and effectively controlled virus replication in the liver. However, MyD88 deficiencies resulted in overt liver pathology and enhanced susceptibility to virus infection. Taken together, these results support a role for MyD88 responses in promoting resistance to MCMV. However, the results also clearly demonstrate that TLR9 is dispensable for pDC recognition of the virus in the liver, assessed through the efficient production of IFN-α and downstream protective inflammatory responses. Furthermore, our study suggests that the infected tissue site uniquely contributes to the process of virus sensing by modulating the function of TLR signaling pathways.

The critical role of pDCs in producing IFN-α in response to MCMV infection in spleen and blood has been well documented (10, 11, 30, 34). In this study, pDCs in the liver were identified as major contributors of IFN-α in response to MCMV infection. This observation is not surprising because this particular subset of DCs has been shown to secrete up to 100-fold more IFN-α than other cell types in various models of infection (8, 12, 38, 39, 40, 41). Several studies have indicated that the expansive repertoire of TLRs on pDCs facilitates their recognition of RNA and DNA viruses and promotes IFN-αβ synthesis (16, 17, 18). Murine pDCs contain abundant levels of TLR7 and TLR9 within endosomal compartments. TLR7 recognizes ssRNA sequences, whereas TLR9 recognizes dsDNA and CpG oligonucleotides (22, 23, 28, 29, 42, 43). Both of these TLRs use the adaptor molecule MyD88 for subsequent production of IFN-α and chemokines (13, 22, 30). In the spleen, it has been shown that pDCs can sense MCMV infection via TLR9- and MyD88-dependent mechanisms, culminating in the rapid induction of IFN-α (9, 30, 34). However, it is becoming increasingly clear that induction of IFN-α during virus infection can occur in the absence of TLR9 (44, 45, 46).

It has been demonstrated during infection with HSV-1 that IFN-α can be induced using both TLR9 dependent and independent pathways (44). Splenic pDCs were shown to be entirely dependent on TLR9 for activation with HSV-1, whereas freshly isolated bone marrow pDCs responded to infection with IFN-α production in the absence of TLR9. It was concluded that the pDC tissue source was a major determining factor for the requirements of TLR9 for cytokine production. Our results support this conclusion, although the mechanisms involved in promoting tissue-specific TLR9 dependent and independent antiviral responses have not been elucidated. Recent studies have indicated that pDCs and other DC populations in the liver are less mature and immunogenic, providing evidence that spleen and liver DCs are quite different in nature with respect to their ability to promote immune functions (14, 47). Therefore it is possible that the differentiation state, level of maturity, and interaction of hepatic DCs with resident macrophages or other immune cell populations influence how TLR9, or other molecules that remain to be defined, contribute to viral recognition and establishment of hepatic antiviral defenses.

The data presented clearly demonstrates a TLR9-independent pathway to IFN-α induction and ultimately antiviral defense in the liver during infection with MCMV. However, this path to host protection is completely dependent upon MyD88 signaling, as MyD88^−/− mice make significantly less cytokines, including IFN-α, as compared with WT mice. Moreover, infected WT and MyD88^−/− mice exhibited comparable proportions and absolute numbers of pDCs (data not shown), thus the diminished IFN-α production in the absence of MyD88 function is not attributed to the ineffective recruitment of pDCs in the liver. In contrast, MyD88^−/− mice were impaired in their ability to effectively recruit inflammatory macrophages and NK cells and deliver IFN-γ to the liver. Consequently, virus-induced pathology and enhanced susceptibility to virus infection was evident in these mice. In an effort to resolve the disparities observed with respect to TLR9 and MyD88, we examined the roles of multiple TLRs including TLR2 and TLR7. Mice deficient in either TLR2 or TLR7 displayed levels of IFN-α, CCL2, CCL3, and IFN-γ proteins in the liver that were comparable to counterpart WT mice. Moreover, protective hepatic inflammatory responses were not significantly affected in the absence of TLR2 or TLR7 (data not shown). The TLR2 results are in agreement with recent studies that exclude a role for TLR2 in control of MCMV infection (13), although other herpesviruses appear to be recognized by this receptor (48, 49). In addition, major involvement of TLR3 and TLR4 signaling in response to MCMV infection has also been excluded (13). Taken together, it does not appear that TLRs contribute to antiviral defenses at least individually. In addition, we cannot eliminate the possibility that a combination of TLRs is required for appropriate induction of proinflammatory cytokines that would require MyD88 signaling. The MyD88 adaptor molecule is also required for transduction of signals downstream of the receptors for IL-1 and IL-18 (33). However, impaired cytokine production and inflammatory cell recruitment to the liver were not observed in MCMV-infected mice genetically deficient in the receptor for IL-1 (data not shown). As IL-18 does not play a major role in hepatic immunity or survival during MCMV infection (31), it was not directly examined in this study.

Studies have shown that MyD88 function is also required for activation of IFN regulatory factor (IRF) family members including IRF-7 (50, 51). Within the pDC subset in particular, MyD88 has been shown to form a complex with IRF-7 that is required for a robust induction of IFN-α (51). In addition, differential expression of IRF-7 in peripheral tissues including liver has been shown to regulate local production of IFN-α (52). As stated previously, we examined multiple well-established signaling pathways that use MyD88 in an effort to resolve the requirement of this adaptor in the liver. However, none of the molecules examined appeared to be essential for IFN-α production and induction of inflammatory cell trafficking at this site under conditions of MCMV infection. Therefore, it is possible that an unknown mechanism exists that results in MyD88-IRF-7 complexes that are unique to the liver and required for localized induction of IFN-α during infection. Ongoing studies are currently investigating this hypothesis.

The results presented concur with other studies in establishing that a diminished but still evident systemic cytokine response occurs in the absence of TLR9 under moderate infection conditions (13, 30). It is highly probable that hepatic pDCs contribute substantially to the residual production of IFN-α observed in spleen and serum through a pathway that is TLR9-independent but still dependent on MyD88 signaling. This highlights the need to preserve IFN-αβ-mediated immunoregulatory responses during a virus infection.

In conclusion, these studies identify a TLR9-independent but MyD88-dependent pathway that is used by pDCs in the liver to promote IFN-α production and recruitment of key innate inflammatory cells to establish hepatic immunity in response to MCMV infection. The work also suggests that the infected tissue site uniquely contributes to the process of virus sensing and regulation of localized antiviral defenses.

We thank Paula Weston and Michele Gardner from the Molecular Pathology Core Facility at Brown University for their assistance with histological sample preparations. We also acknowledge Dr. Bruce Beutler, The Scripps Research Institute, La Jolla, CA, for the gift of C57BL/6-TLR9^−/− mice.

The authors have no financial conflict of interest.