Many lines of evidence support the importance of TNF in the development of PcP. Pc
stimulates TNF production and release from alveolar macrophages in both immunocompetent and immuno-suppressed mice and from monocytes and macrophages in culture (20
). Release of TNF from alveolar epithelial cells in response to the organism has also been demonstrated (22
). In the reconstituted SCID model of PcP, the onset of reduced compliance and hypoxia is temporally related to peak TNF mRNA in lung tissue and TNF protein in bronchoalveolar lavage fluid, in association with the influx of neutrophils, macrophages and lymphocytes. TNF protein is likewise increased in CD4+
T-cell depleted, Pc
infected mice in a CD8+
T-cell dependent manner (9
). Our previous studies using TNFR deficient mice demonstrated that maximal Pc
-induced chemokine production, lung injury and pulmonary dysfunction required intact TNFR signal transduction (9
). The current study demonstrates that TNFRs on cells resistant to split dose irradiation are sufficient to mount an inflammatory response to Pc
, comparable or in excess of that generated when all cells express the receptors. Limitation of TNFR distribution to marrow-derived cells improved control of Pc
burden and reduced Pc
Pulmonary function and inflammatory markers were analyzed in the TNFR chimeric mice when the sickest treatment group reached clinically peak disease, approximately four weeks after Pc
treatment in the chimeric mice; at and beyond this point mortality increased markedly in the KO to WT and WT to WT chimeras. Each of the transplanted mice that expressed TNFRs either on radiosensitive marrow derived cells or radio-resistant parenchymal cells or both (WT to WT, WT to KO or KO to WT) demonstrated some evidence of pneumonia at this time-point. However, the greatest mortality, decrement in pulmonary function, weight loss and inflammatory cell recruitment occurred in those mice expressing TNFRs on radio-resistant but not on marrow derived cells. TNFR null chimeras (KO to KO) were least injured despite having a relatively greater Pc
burden than the remaining chimeras. Interestingly, in vitro Pc
-dependent induction of chemokines in alveolar epithelial cells occurs independently of TNFRs (23
), which may explain the inflammation and physiologic impairment observed even in the TNFR null mice. In vivo, TNFR deficient models indicate that maximal inflammation that correlates with clinically significant loss of weight, pulmonary function and in death requires parenchymal TNFR responses. TNFR expression limited to marrow derived cells was sufficient to generate Pc-stimulated lung injury and to augment loss of pulmonary function and weight loss but was much less apt to cause mortality, and was associated with reduced MCP-1, KC and TNF production, intra-alveolar cell death (by LDH) and recruitment of lavagable cells. As previously observed, the severity of lung injury was not directly related to Pneumocystis burden (24
). Reduction in lung compliance weakly correlated with Pc
burden but only in the presence of parenchymal TNFR function, which may reflect inhibition of surfactant production or function (9
). The global TNFR null mice carried the greatest Pc
burden but the least evidence of injury consistent with a role for TNF in clearance of Pc
as previously demonstrated (9
). In the current study, the parenchymal cell null, marrow WT animals maintained the lowest burden of organism suggesting that TNF stimulates immune cells to remove the organism. It has also been suggested that adherence of Pc
to lung epithelial cells enhances proliferative growth of the organism (26
). The absence of parenchymal TNFRs may reduce Pc
-epithelial cell adherence, reducing the organism's growth potential and, potentially, leaving it more accessible to immune cell clearance. Further study is necessary to clarify this observation.
Accumulation of inflammatory cells and LDH in the bronchoalveolar space was parenchymal cell TNFR dependent. Although each of the chimeras developed some degree of lung injury there was a marked increase in numbers of inflammatory cells recruited to the lavagable air space in WT recipient mice, particularly when the bone-marrow derived, recruited inflammatory cells did not express the TNFRs. No increase in total lavagable cell counts was detected in the absence of parenchymal TNF signal transduction. The differential of the BALF cells was altered in response to Pc; infiltration by neutrophils and lymphocytes occurred in all chimeras but was exacerbated by the presence of parenchymal TNFRs. Flow cytometry suggested marked increases in the percentage of lymphocytes that are CD8+ in all Pc exposed groups with no difference between chimera groups. However, when considered as absolute number of lavagable cells, maximal increase in recruitment of CD8+ T-cells was seen in the TNFR WT recipient chimeras. This is consistent with a significant role for parenchymal TNFR signal transduction in the recruitment of these cells that have been shown to mediate Pc-induced lung injury in this CD4+ T-cell depleted model. It is noted that despite markedly increased inflammatory cell infiltration mediated by parenchymal TNFRs, the KO to WT chimeras were unable to control the Pc burden any better than the fully WT mice. This is consistent with the failure of sensitized CD8+ T cells to control organism burden, however the failure of enhanced macrophage numbers in this chimera to control the Pc again supports the role of TNF stimulation of the immune cells in Pc clearance
A significant contribution of parenchymal cell TNF responses to lung injury induced by pathogens has been previously demonstrated. For example, CD8+
T-cell recognition of alveolar cells expressing a specific viral antigen triggered MCP-1 and MIP-2 expression by the lung epithelial cells in large part due to T-cell transmembrane TNF (tmTNF) and the presence of TNFRsf1a on the epithelium (27
). In addition, a study of alveolar macrophages in patients with acute respiratory distress syndrome (ARDS) demonstrated enhanced tmTNF correlated with severity of disease although soluble TNF concentrations in the BAL did not (29
), suggestive of the importance of inflammatory cell bound TNF interacting with parenchymal receptors. Since tmTNF is an active signaling molecule, the expression of TNFRs on parenchymal cells constitute a mechanism for intercellular communication with tmTNF expressing immune cells. In this way, TNF expressed by inflammatory cells may have enhanced capacity to either induce cytotoxicity or to stimulate pro-inflammatory protein production by the structural cells of the lung. In the case of Pc
, the epithelial cells are anchors for the organism and so are best positioned to stimulate or amplify a local host response. Prior studies also suggest that CD8+
T-cell mediated Pc
-induced lung damage is dependent on MHC class I expression by radio-resistant cells (6
). Intracellular Pc
antigen processing or presentation by alveolar epithelial cells has not yet been demonstrated but is feasible. Intercellular TNF-TNFR interactions may enhance such lymphocyte-epithelial cell interactions.
As well as TNFα, the TNF receptors also bind and transduce signal of the homotrimeric ligand, lymphotoxin alpha, previously known as TNFβ. LTα also signals by forming heterotrimers with LTβ that bind the LTβ receptors. Removal of the TNF receptors therefore prevents signal transduction by homotrimeric LTα, as well as by TNFα. The phenotypes observed in the current study are then the result of manipulating both TNFα and LTα activity. Due to the strong relationship previously established between TNFα and progression of PcP it is thought likely that it is primarily the effect of TNFα signal transduction via TNF receptor activity that has been altered in this study. No regulation or role of LTα in PcP has been clearly demonstrated to date. Whether or not PcP is altered by LTα binding of TNF receptors on parenchymal or bone marrow cells however remains to be studied potentially by anti-ligand antibody or ligand specific knockout transgenic models.
The TNF sf1a receptor has been demonstrated to be necessary for normal development and maintenance of splenic B cell follicles and germinal centers (30
). In mice deficient in either lymphotoxin-alpha (LT-alpha) or the type I tumor necrosis factor (TNF) receptor, but not the type II TNF receptor, germinal centers failed to develop in peripheral lymphoid organs. While the chimeras in the current study have not been tested for germinal centers, it is highly unlikely that a reduction or failure of antibody production in response to Pc
accounts for the differences in PcP between the CD4+
Tcell depleted chimeras studied given the known role of these lymphocytes in producing anti-Pc antibodies. Even when Pc-infected SCID mice are given Pc-sensitized lymphocytes, CD4+
T cells are required for an antibody response to be generated (25
). In addition, non-immunized, CD4-depleted mice, directly analogous to the animals tested in the current study, do not make an antibody response to Pc
). Therefore, none of the chimeras tested in this study would have had antibody response to Pc
regardless of presence or absence of TNF receptors.
Macrophage predominance persisted and was amplified in the BALF of KO to WT mice, in contrast to the other chimeras. This occurred in association with exaggerated production of the CC chemokine MCP-1, documented by increased tissue mRNA as well as BAL protein concentrations. Since MCP-1 is a chemoattractant for lymphocytes as well as macrophages, it is possible that parenchymal response to TNF results in enhanced induction of MCP-1 from epithelial cells, for example, that in turn would increase recruitment of these cells. This model would be consistent with the viral model in which CD8+
T-cell tmTNF stimulates alveolar epithelial cells to produce MCP-1 (34
). It is also possible that activation of parenchymal cells by Pc
-induced TNF causes release of other chemo-attractants that enhance the recruit of inflammatory cells that become the source of MCP-1. Further analysis of the current model will determine the source of the CC chemokine. Previous studies suggest that CD8+
T-cells, even stimulated by specific recognition of alveolar epithelial cells, do not produce MCP-1 while the target cells do (28
). In contrast, in situ hybridization in the SCID mouse model of PcP demonstrated the primary location of MCP-1 mRNA to be the type II epithelial cells (23
). Maximal MCP-1 production occurs in the absence of hematogenous cell TNFRs so direct TNF stimulation of macrophages is not the source.
The alveolar macrophage numbers and MCP-1 induction was not as marked in the WT to WT chimera as in the KO to WT which is suggestive not only of a role for TNFR stimulation of parenchyma in production of this chemokine in response to Pc but also of a suppressive effect of TNFRs in the marrow derived cells. One potential explanation for immune cell TNFR mediated suppression of Pc-induced lung inflammation is that in the absence of TNFRs, inflammatory cells recruited to the lung fail to undergo TNF induced apoptosis, an important system of regulation of inflammation. Further studies of cellular turn-over in the current model are indicated.
MCP-1 and MIP-2 have both been implicated in the pathogenesis of PcP having been shown to be induced by Pc
and by TNF, as well as correlating with severity of disease in WT and TNFRKO mice. In this Pc
-chimera model however, independent regulation of the two chemokines was observed. In vitro studies with Pc
stimulated primary alveolar type II cells suggested that MCP-1 was induced by Pc
directly and that this induction was dependent on both NF-κB and JNK activity (23
). Preliminary data suggests a synergistic induction of MCP-1 from lung epithelial cells exposed to both Pc
and TNF (data not shown). In contrast, MIP-2, similarly induced by direct Pc
interaction with epithelial cells was unaffected by JNK inhibitors, consistent with differential regulation of the two chemokines and perhaps greater dependence of MIP-2 gene expression on NF-κB of which TNF is a most potent stimulant. Considering the RNA measurements, while maximal MCP-1 induction was dependent on parenchymal cell signaling in the absence of immune cell receptors, MIP2 induction was maximal when both cell compartments could respond to TNF, inducing a three-fold increase over that in the mixed chimeras.
One or both of the two distinct receptors, TNFRsf1a and TNFRsf1b receptors have been identified on the majority of lung cells tested, including type II pneumocytes, although their relative ratio varies by cell type and can be altered by stimulation (35
). In most studies, TNFRsf1a is constitutively expressed while TNFRsf1b expression is inducible. The predominant receptor, by mRNA, in mouse and human lung is TNFRsf1a, although TNFRsf1b is induced by many stimuli including TNF delivery and Pc
). A differential role of the two receptors in PcP has not yet been clarified. Like TNF, the TNFRs are also targets for matrix metalloproteinases. Soluble TNFRs (sTNFR) may act as a reservoir of soluble TNF or as circulating inhibitors of both soluble and tmTNF. In this study, we demonstrated that Pc
infection induces shedding of these receptors and accumulation of soluble (s)TNFRsf1a and TNFRsf1b both in serum and BAL of mice. Whether soluble TNFR levels may be useful as biomarkers reflecting the severity of PcP, or may have a physiological role in disease, is not yet determined. Interestingly, BMT chimera experiments suggest that in the basal, healthy state, the majority of circulating sTNFR sf1a originates from parenchyma while bone marrow derived cells appear to be the source of >50% of sTNFRsf1b. In PcP, both sTNFRsf1a and sTNFRsf1b originated from both parenchymal and marrow derived cells. Although the quantity of soluble TNFR present in serum or BALF have been shown to mirror the severity of other diseases, the regulation and function of TNFR solubilization remain unclear. sTNFRs may be involved in controlling the TNF response during the generation of an immune response. The KO to WT chimeras had reduced sTNFR and enhanced inflammatory response relative to WT to WT. It is possible that the lack of soluble receptors acting as competitive inhibitors contributes to the enhanced injury documented in the mixed chimeras. Alternatively, expression of membrane-bound TNFRs may create a feedback loop that regulates TNF transcription. Increased TNF production was documented in this study but only in the KO to KO mice, suggesting that TNF feedback on either marrow derived or parenchymal cells is sufficient to regulate the ligand.
The current study demonstrates that TNFRs on parenchymal cells, those resistant to split dose irradiation, are sufficient to mount an inflammatory response to Pc even if marrow derived cells are receptor null, comparable or in excess of that generated when all cells express the receptors. The relative importance of epithelial, endothelial or mesenchymal cell TNFR responses, as well as the source of the stimulating TNF, remains to be determined. Limitation of TNFR distribution to marrow-derived cells improved control of Pc burden and reduced the injurious host response, best demonstrated in this study by improved survival and cellular recruitment as compared to normal receptor expression. Limitation of TNFR distribution to parenchymal cells markedly worsened the inflammatory response and resulting injury. The results of this study support the notion that therapeutic interventions that inhibit parenchymal cell TNF signal transduction, while maintaining or enhancing immune cell TNFR responses, analogous to the WT to KO chimeras, could be effective both in limiting Pc burden and in limiting the injurious host inflammatory response. Alternatively, global anti-TNF treatment, if given as an adjunctive therapy to the currently used and effective anti-Pc drugs, could also have benefit for patients with severe PcP by reducing the immune aspects of PcP-related lung injury.