Based on our observation of the recruitment of MDL-1+
leukocytes following ConA-induced liver injury and the report that DV binds and activates MDL-1 (6
), we set out to delineate the cellular and molecular mechanisms underlying the progression from tissue injury/SIRS to shock. In order to specifically isolate the MDL-1–mediated effects rather than those associated with a more complex infectious process, we employed inactivated DV and, even more specifically, an agonist anti–MDL-1 antibody, to remove the effects of DV particles binding to other known binding partners such as DC-SIGN (6
). Thus, our experimental system does not represent a model of DV infection but instead functions to specifically address the contribution of MDL-1 signaling in the progression from liver injury to shock. Using this system, we found that ConA-induced liver injury drives a hepatic infiltration of MDL-1+
immature myeloid cells of the granulocyte lineage and that triggering of MDL-1 on these cells, either by inactivated DV or an agonist antibody, results in lethal shock. An increase in circulating immature neutrophils (“left shift”) is a common component of the clinical syndrome of SIRS/sepsis. This work suggests that, in the appropriate context such as liver injury, these immature neutrophils may contribute to the pathogenesis of shock.
In our model, in the absence of ConA treatment, very few MDL-1+
cells were detected in the liver (Figure B) and administration of DV or agonist mAb caused no overt response (data not shown). This fact highlights the importance of ConA-induced liver injury for the recruitment of immature myeloid cells. This is in agreement with two recent studies reporting that liver inflammation induced by high-fat diet or Th1 cells may drive the accumulation of immature myeloid cells in the liver (40
). The precise role of these infiltrating immature myeloid cells is not clear, as they have been reported to have a proinflammatory or antiinflammatory response. Interestingly, mice treated with ConA for 4 hours showed a modest, 1.5-fold increase in serum ALT (Supplemental Figure 3A, bottom panel), suggesting that only minor liver injury is sufficient to trigger the recruitment of these immature myeloid cells.
ConA-induced hepatic injury closely resembles the pathophysiology of T cell–mediated liver diseases; it therefore has been used extensively as an animal model for autoimmune and viral hepatitis (12
). DV-infected mice showed significant liver damage, which also correlated with T cell activation and hepatic cellular infiltrate (43
). This is concordant with clinical data showing that in general more than 80% of dengue patients had elevated serum levels of liver enzymes (ALT, AST), and the mean levels of these enzymes were significantly higher in patients with DSS as compared with DF (44
). Dengue patients also displayed increased lymphocyte infiltration to the portal track (45
). We therefore believe that ConA-induced, T cell–mediated hepatitis is a valid model for virus-induced liver injury.
IFN-γ, TNF-α, IL-1β, IL-6, and G-CSF have previously been reported to be elevated in serum of DV-infected patients (47
). We found that ConA treatment also elevated serum levels of these proinflammatory molecules; and we were able to recapitulate the ConA effect by treating mice with recombinant G-CSF to recruit the MDL-1+
cells, as well as a low dose of IFN-γ, TNF-α, IL-1β, and IL-6 to prime the pathogenic cells. Subsequent triggering by DV or agonist mAb resulted in mice dying of shock, mediated primarily by MDL-1 activation and not due to cytokine priming. Therefore, results from this in vivo reconstitution experiment may provide further insight into the important role these cytokines play in the recruitment and priming of the MDL-1+
pathogenic cells during DV infection.
Given the profound sensitizing effect of ConA-induced liver injury on MDL-1–mediated shock, we asked whether this effect is unique to ConA or if liver injury in general would trigger infiltration of these immature myeloid cells and sensitize mice to shock. To address this question, we used the APAP-induced liver injury model, since APAP overdose is currently the most frequent cause of acute liver failure in both the United States and the United Kingdom (49
) and thus has significant clinical implications. Although APAP treatment caused both liver injury and hepatic infiltration of MDL-1+
cells in a time-dependent manner, MDL-1 stimulation at the time of maximal recruitment failed to cause death. One possible explanation is that insufficient numbers of pathogenic cells were recruited to the liver, as there were 7-fold fewer MDL-1+
cells after APAP- versus ConA-induced injury. This relative inability of APAP-induced injury to mobilize MDL-1+
cells can be attributed to the fact that there is significantly less G-CSF being released in response to the injury. In addition to the deficit in the recruitment of MDL-1+
cells, APAP-induced liver injury also generated markedly fewer inflammatory cytokines, including IFN-γ, TNF-α, IL-1β, and IL-6, which we have shown to be critical for cell priming. These findings are not entirely surprising, considering the very different mechanisms of action for liver injury induced by APAP (toxic metabolite–mediated) and ConA (T cell–mediated), which have been discussed in previous publications (51
). Fitting with our mouse model data, early epidemiological studies reported that APAP-induced liver damage had no potentiating effect on the DV-induced disease outcome in humans (53
Our observation that liver injury induced by both ConA and APAP is capable of triggering hepatic infiltration of MDL-1+
cells, albeit with different efficiencies, led us to hypothesize that the association between liver injury and development of DSS may be attributed to the recruitment of pathogenic cells. Increased numbers of immature neutrophils in the blood of DSS patients have been reported (55
); it would be of great interest to see whether these patients also have elevated liver enzyme levels in the blood.
Based on cell surface markers and morphology (19
), we have identified the pathogenic cells responsible for MDL-1–mediated lethal shock as immature myeloid cells. These cells also expressed the myeloid precursor marker CD33 (20
) and MHC class I, but not hematopoietic stem cell/progenitor cell marker CD34 (58
) or MHC class II. This unique phenotype of CD33+
MHC class I+
is consistent with that of promyelocytes (59
). We also observed strong G-CSFR staining on MDL-1+
cells, which is in accordance with two previous studies reporting that G-CSFR is highly expressed on CD33+
human bone marrow cells (63
) and provides further evidence that the MDL-1–expressing pathogenic cells are immature myeloid cells/promyelocytes. This is in agreement with clinical data showing an increased number of immature leukocytes in the blood of patients during DSS or septic shock (55
) and the detection of MDL-1+
cells in the spleen of septic patients (65
). In mice, CD11b+
immature myeloid cells are also expanded in the spleen during sepsis (66
immature myeloid cells can be further characterized, based on cell surface markers Ly6G and Ly6C, into granulocytic (CD11b+
) and monocytic (CD11b+
) populations. The immature myeloid cells expanded in our model were primarily granulocytic, and they expressed higher levels of MDL-1 than the smaller monocytic population. Interestingly, the Ly6G+
granulocytic subset of the CD11b+
cells is also the predominant population expanded in tumor-bearing mice (69
Immature myeloid cells have been reported to be mobilized from bone marrow into peripheral blood primarily by G-CSF (70
), which corroborates our finding that G-CSF, but not GM-CSF, KC, MIP-2, or MCP-1, acts as a critical proinflammatory molecule in MDL-1–induced shock by promoting the mobilization (and possibly the generation, egression, and maturation) of the pathogenic cells to the liver. In addition to the liver, these ring cells are also found in peripheral blood in response to ConA treatment (Supplemental Figure 7A), supporting the notion that immature myeloid cells egress from the bone marrow into the circulatory system in response to inflammation. We also asked whether MDL-1 itself is involved in myeloid cell migration by comparing the number of ring cells in WT and MDL-1–/–
mice treated with ConA. We observed similar ConA-induced migration of immature myeloid cells, in both Cytospin preparation and IHC staining, between WT and MDL-1–/–
mice (Supplemental Figure 7, A–D). Therefore, we concluded that ConA drives the recruitment of immature myeloid cells independent of MDL-1 and that the protection demonstrated in MDL-1–/–
mice is not due to impaired hepatic migration of the pathogenic cells.
Overproduction of NO is known to be an important factor in the pathogenesis of shock of various etiologies in humans. Compared with healthy individuals, patients with septic or anaphylactic shock have an elevated NO level in blood or exhaled air, respectively (75
). In addition, DSS patients have significantly higher serum NO than patients with DHF (77
). It is generally believed that during the early stage of shock, there is a transient release of low levels (nanomoles) of NO produced by eNOS, which is constitutively expressed primarily in endothelial cells and is activated by an increase in intracellular calcium. This is followed by sustained local production of a large quantity (micromoles) of NO via calcium-independent iNOS
gene induction in immune cells, including macrophages, in response to immunological stimuli such as inflammatory cytokine production. The iNOS-mediated overproduction of NO is assumed to be responsible for the cardiovascular failure and mortality associated with different kinds of shock. However, results from our study clearly indicated that eNOS is the primary enzyme responsible for lethal shock triggered by MDL-1 and the associated increase in serum NO. This may seem surprising, although it is not without precedent: our results are consistent with a recent study showing the exclusive role eNOS plays in anaphylactic shock (78
), as well as two reports implicating constitutive eNOS in septic and hemorrhagic shock (79
Although eNOS has originally been thought to be expressed primarily in endothelial cells, increasing evidence is revealing a much broader expression spectrum in leukocytes than previously thought. eNOS is also expressed in both mouse and human neutrophils, monocytes, and macrophages, as well as in human immature myeloid cells (81
). In the mouse model, on the other hand, the production of NO by CD11b+
immature myeloid cells has only been reported to be mediated by iNOS (67
). Our study is the first to our knowledge to identify a unique population of NO-producing murine immature myeloid cells expressing only eNOS, as we demonstrated that these cells are iNOS–
and that ConA treatment induces a time-dependent upregulation of eNOS
mRNA in the liver, whereas iNOS
transcript is undetectable (data not shown). The predominant role of eNOS in immature myeloid cells that we report is in agreement with a recent publication showing high eNOS expression in immature rat neutrophils, which is downregulated with maturation (89
). Our results showing that these eNOS-expressing immature myeloid cells are also MDL-1+
and are capable of producing NO with MDL-1 stimulation in vitro suggest a one-cell model scenario. The pathogenicity conferred by passive transfer and the protection achieved after in vivo depletion further highlight the importance of these cells. However, at this point we cannot eliminate the possibility of a two-cell model in which MDL-1–activated immature myeloid cells produce intermediate mediator(s) to stimulate endothelial cells to generate NO.
TNF-α has also been implicated as a crucial mediator for shock of different etiologies. In clinical data, elevated TNF-α levels were reported in blood of patients with septic, hemorrhagic, and anaphylactic shock (90
). High levels of TNF-α were also observed in sera of all patients with dengue infection, with the highest values found in patients with DSS (93
). Furthermore, the elevation of TNF-α in the serum has been reported to be strongly associated with clinical disease progression from DF to DSS (94
). Furthermore, anti–TNF-α antibody has been reported to be an effective treatment to reduce mortality associated with DV infection (96
). Previously, macrophages were reported to be the primary cell type expressing MDL-1 that DV interacts with to trigger TNF-α release (6
). The present study identifies immature myeloid cells as another MDL-1–expressing cell capable of releasing the inflammatory cytokine TNF-α in response to DV stimulation and playing a crucial role in DV-mediated pathobiology. Future studies measuring MDL-1 expression on the immature neutrophils that are increased in the blood of patients with DSS or septic shock, and the ability of these cells to produce TNF-α in response to MDL-1 stimuli, are expected to further establish the clinical relevance of these cells in shock. Our results showing that MDL-1+
cells expressed both eNOS and TNF-α led us to investigate the cross-regulation of these two mediators. In our MDL-1–mediated pathway, eNOS is upstream of TNF-α and is able to regulate TNF-α production by modulating TACE activity. This result fits into the paradigm of a previous study showing that NO activates TACE by nitrosation of the inhibitory cysteine switch motif in the TACE prodomain and cleaving the membrane-bound TNF-α (29
). In addition to binding its cognate receptor to induce an inflammatory response, TNF-α can activate eNOS activity via an Akt-dependent pathway to initiate an amplification process (97
). TNF-α is also able to terminate or dampen the signal by downregulating eNOS
mRNA by shortening its half-life as a negative feedback mechanism (99
). Although the hypothesis has not been proven, it is likely that this complex network of interplay between NO and TNF-α is part of the MDL-1–mediated pathway to fine-tune the biological response.
In addition to NO and TNF-α, we also observed elevated levels of multiple cytokines (IFN-γ, IL-1β, IL-6, IL-10) and chemokines (MCP-1, MIP-1α, IP-10) in the serum of mice that died of MDL-1–mediated shock (Supplemental Figure 8). This is in accordance with the results of 6 separate dengue patient studies done in Brazil, Cuba, India, Singapore, Thailand, and Vietnam that all showed increases in IFN-γ, IL-1β, IL-6, IL-10, MCP-1, MIP-1α, and IP-10 in the serum (48
). These data support the idea that a “cytokine storm” is involved in the pathogenesis of DSS (105
). It also demonstrates that our MDL-1–mediated shock model resembles the cytokine profile observed in DV-induced human disease. However, the contribution of each individual cytokine or chemokine to the pathogenesis of DSS, and the role MDL-1 plays in modulating these mediators, are well beyond the scope of this study but will warrant separate detailed investigation in the future.
The MDL-1–mediated signaling pathway is not clearly defined. Current understanding of signaling events downstream of DAP12 and DAP10 are primarily based on in vitro data obtained by triggering of receptor TREM-1 and NKG2D, respectively (106
). Therefore, we decided to use both genetic knockout mice and biochemical inhibitors to interfere with the signaling pathways in vivo to identify the MDL-1–specific downstream signaling events that are relevant to this disease model. Our results strongly indicated that activation of both adaptor protein DAP12 and DAP10, as well as the protein kinase Syk, PI3K, and Akt, are critical for MDL-1–induced shock. Our results showing an absolute requirement for both DAP10 and DAP12 signaling is somewhat surprising, given that previous studies in NK cells have demonstrated at least partial redundancy between these two adaptor molecules (108
). However, it appears that the relative contribution of the two adaptor molecules to the downstream function is highly dependent on the examined receptor and cellular function. This requirement for signals generated from both adaptor molecules may also indicate that a critical “threshold” of signal, possibly at the level of PI3K (110
), has to be reached to trigger shock. This hypothesis is supported by a recent study showing that both DAP10 and DAP12 are required for maximal TREM2-stimulated PI3K activity, with DAP10 mediating the recruitment of PI3K to DAP12 and the activation of downstream kinase Akt (111
). In our model of MDL-1–induced shock, DAP10 may be similarly responsible for amplifying DAP12-mediated PI3K activation, leading to a pathogenic activation of the downstream Akt/eNOS signaling response.
This is the first report to our knowledge to establish eNOS as a target of MDL-1, the triggering of this receptor resulting in eNOS activation and NO release, which in turn modulates another downstream cytokine, TNF-α. It has been suggested that an increase in intracellular calcium alone is not sufficient to affect the enzymatic activity of eNOS and that protein phosphorylation is also critical in regulating eNOS function (112
). Since Akt-mediated eNOS phosphorylation has been shown to play an important role in anaphylactic shock (78
), we asked whether activation of MDL-1 can stimulate Akt to phosphorylate eNOS. Our results indicated that (a) triggering of MDL-1 results in phosphorylation of Akt and eNOS, (b) transduction of MDL-1 signal requires physical interaction between the two proteins to form a signaling complex, and (c) complex formation is independent of MDL-1 activation. In endothelial cells, HSP90 has been suggested to function as a scaffold to facilitate the proximity of and association between Akt and eNOS (113
). As it is likely that such a scaffolding protein is involved in MDL-1–mediated signal transduction in immature myeloid cells, this should be further investigated.
The mortality rate for shock is high, with statistics associated with septic shock and DSS approximately 50% and 44%, respectively (117
). Volume replacement therapy is considered the only effective treatment to restore circulation. Several other treatment options have been tested and found not to be effective. Two clinical trials have been conducted using anti–TNF-α mAb for the treatment of septic shock, but the results were inconclusive, as one study showed reduced mortality, while the subsequent trial revealed no significant improvement in survival (119
). Another clinical trial using NOS inhibitor monomethyl-l
-arginine (L-NMMA) treatment for septic shock patients showed improved resolution of shock with reduction in serum NO levels and increased vascular tone (121
). Despite a positive outcome on blood pressure, a phase III clinical trial had to be prematurely terminated due to increased mortality (123
). Given the prominent role eNOS played in anaphylactic shock and the newfound importance of this enzyme in shock triggered by MDL-1 revealed by our study, it is plausible that a selective eNOS inhibitor could be used as a new therapy. TACE inhibitor is currently being tested in a clinical trial for the treatment of rheumatoid arthritis (124
). Our results showing that TAPI-1 treatment significantly prolongs the survival of MDL-1–triggered shock suggest the prospect of using TACE inhibitor as an alternative treatment for shock. Furthermore, our study indicates the crucial role that the PI3K/Akt pathway plays in lethal shock mediated by MDL-1, suggesting the application of PI3K/Akt inhibitors for DSS patients, in addition to their use in anticancer therapy (125
Our study suggests that this mechanism may play an important role in the progression from tissue injury/SIRS to shock in a variety of clinical settings where mobilization of immature granulocytic myeloid cells occurs, especially in the case of dengue infection, where progression to DSS has been associated with ongoing liver damage and the viral particle itself has been shown to activate MDL-1. Given that inhibitors of many components of this signaling pathway (MDL-1, Syk, PI3K, and Akt) are being developed as potential drugs in other therapeutic areas, it will be interesting to assess whether this mechanism contributes to the progression to shock in patients and whether inhibition of this pathway has therapeutic value.