Phillip Ian Bird Building 13B Department of Biochemistry and Molecular Biology Monash University Clayton, VIC 3800 Australia Tel: +61 3 9905 3771 Fax: +61 3 9905 3726 e-mail: email@example.com
Serine proteases are important components of the immune system, playing a role in many processes including migration, phagocytosis and elimination of virally infected and cancerous cells. Members of the serpin superfamily regulate the activity of these proteases to limit tissue damage and unwarranted cell death. This review focuses on the role of intracellular (clade B) serpins in maintaining viability of both innate and adaptive immune cells.
Leukocyte and complement serine proteases play an important role as effectors of the immune system by killing invading pathogens or by destroying infected, abnormal or foreign cells. Members of the serpin superfamily of protease inhibitors regulate such proteases to prevent both tissue damage and the premature death of immune cells. In insects, serpins participate in toll receptor signaling (1, 2), a function not observed in vertebrates. Some pathogens produce serpins to subvert the immune response, such as the poxvirus serpin crmA, which inhibits caspases (3, 4).
This review will focus primarily on the vertebrate clade B serpins present in the nucleocytoplasm of cells of both the innate and adaptive immune systems. These include SERPINB1 [monocyte neutrophil elastase inhibitor (MNEI)], which inhibits neutrophil elastase and cathepsin G (5); SERPINB6 [proteinase inhibitor (PI)-6], responsible for inhibiting cathepsin G (6) and SERPINB9 (PI-9), a regulator of granzyme B (GrB) (7). It has been suggested that these serpins protect effector cells from premature death initiated by exposure to their own granule proteases (4). We will explore recent evidence that supports this idea and discuss emerging, broader roles of these intracellular serpins in the mammalian immune system, particularly in light of recent discoveries using genetically manipulated mice.
Serpin structure and inhibitory mechanism
Overwhelmingly, serpins act as inhibitors of serine or cysteine proteases, although some have evolved alternative, non-inhibitory roles. The superfamily is defined by a conserved tertiary structure consisting of 9 α-helices (denoted A-I) and three β sheets (denoted A–C). The regions important for protease inhibition are centered on β-sheet A and the reactive center loop (RCL). The RCL participates in the initial interaction with the target (cognate) protease, which recognizes it as a substrate and cleaves between two residues termed P1 (N-terminal of the cleavage event) and P1′ (C-terminal of the cleavage event). The P1 residue is a critical determinant of serpin specificity and mutation of this residue results in the loss or conversion of inhibitory activity. For example, the Pittsburgh mutation of α1-antitrypsin (SERPINA1) alters the P1 methionine to arginine, which results in inhibition of thrombin rather than elastase, and causes a fatal bleeding disorder (8).
Following RCL cleavage, the serpin–protease interaction may proceed along either of the two pathways (Figure 1). Along the ‘inhibitory pathway’, cleavage of the RCL leads to a rapid conformational change in the serpin, with the RCL inserting into β-sheet A to form a six-stranded antiparallel sheet. This structural change occurs before the deacylation step in the protease’s catalytic cycle. Therefore, the covalent bond between the serpin P1 and the protease’s catalytic serine remains intact, with the result that the protease is dragged to the opposite pole of the serpin during RCL insertion, distorting its active site (9). This traps both protease and serpin in an inactive sodium dodecyl sulphate (SDS) stable, covalently bound complex. Along the ‘substrate pathway’, the RCL insertion occurs after protease deacylation, and the serpin becomes a true protease substrate. The P1–P1′ bond is cleaved and the protease releases, remaining active, while the serpin is rendered inactive. Serpins may also be inactivated by RCL cleavage outside the P1–P1′ bond: this represents a potential mode of serpin downregulation by noncognate proteases.
The physiological importance of the serpin–protease interaction depends on whether the inhibitory or substrate pathway dominates, which may be influenced by cofactors (10) and can be assessed on the basis of two kinetic parameters, the association rate constant (Kass) and stoichiometry of inhibition (SI). The Kass defines the rate of RCL insertion, while the SI is a measure of how many serpin molecules are required to inhibit a single molecule of protease (and therefore how many serpin–protease interactions occur before the inhibitory pathway ensues). A physiologically significant interaction will occur with a Kass of 1 × 105 to 1 × 107 M/s and a SI close to 1 (11).
Serpin evolution and classification
The serpin superfamily currently contains over 1000 members and is represented in every kingdom of life. The superfamily has been divided into 16 clades based on phylogenic relationships (12). Serpins forming the nine clades (A–I) observed in higher animals cluster on the basis of function, rather than species (12). Of these, eight clades contain extracellular proteins, while clade B contains predominantly intracellular proteins. Such phylogenic classifications have been used to introduce a standardized nomenclature system that employs a SERPIN or SERP prefix followed by a clade designation letter, a numerical identifier and small case paralogue letter (13). Thus, the first described member of the human clade B serpins is designated SERPINB1. The orthologous mouse serpin is designated Serpinb1, but there are several paralogues of this serpin only present in the mouse which are designated Serpinb1b, Serpinb1c and Serpinb1d. (Use of upper or lower case letters in the descriptor follows guidelines of the appropriate species genome nomenclature committee. In this review we will use the clade designator and numerical identifier when discussing individual serpins.)
The plethora of serpins almost certainly reflects the large variety and range of proteases that need to be regulated. Indeed, it has been noted that proteases usually comprise around 1% of sequenced genomes, and that those involved in reproduction and immunity show signs of positive selection (14). The serpin scaffold and inhibitory mechanism is ideally adapted to rapid evolutionary change because a single substitution in the RCL can alter the efficiency of interaction with the target protease or convert it into recognizing a different protease entirely. The existence of rodent-specific paralogues in several serpin clades (especially clades A and B) is consistent with a larger number of proteases in these species but potentially complicates the use of these animals as experimental models because complementary or compensatory effects because of paralogues are difficult to exclude.
As discussed below, a number of serpins belonging to clades A, B and G contribute to inflammation and immunity and can be separated into two broad categories: those regulating extracellular processes and those regulating intracellular processes. Study of the extracellular serpins is well advanced and valuable insights into their pathophysiological roles have been gained by the availability and study of human mutants. By contrast, a human genetic deficiency in an intracellular serpin is yet to be described.
Extracellular serpins in inflammation and immunity
SERPINA1 (α1-antitrypsin) is a potent extracellular inhibitor of neutrophil elastase (15, 16). Produced in the liver, it is found in high concentrations particularly in the lung during the inflammatory response, where it inhibits extracellular neutrophil elastase (17). Inactivating mutations have been shown to cause chronic obstructive pulmonary disease because of elastase-induced lung tissue damage as well as liver disease and chronic inflammation (17, 18). Recent evidence suggests that A1 may act as a signaling molecule by modulating inflammation independently of its protease inhibitory function (19–21). Both inactive and native forms of A1 increase lipopolysaccharide (LPS)-mediated macrophage activation when incubated for a short period of time. However, when incubated over a long period of time (18 h), A1 reduces levels of proinflammatory mediators and increases anti-inflammatory cytokines (19, 20). A1 has a similar effect on B cells, which show anti-inflammatory actions after incubation with both native and polymerized A1 (21). These results suggest that A1 may play an important, additional role in immunity as an inflammatory modifier, outside of protease inhibition.
SERPINA3 (α1-antichymotrypsin) is a potent extracellular inhibitor of immune system chymases, including cathepsin G and mast cell chymases (22). Similar to A1, it is produced in the liver but found in highest concentration in the respiratory tract, where it prevents degradation of extracellular matrix by proteases released during inflammation. A3 has been implicated in chronic obstructive pulmonary disease and neurodegenerative disorders (23).
SERPINA9 (centerin) is found in naïve B cells within the germinal centers of secondary lymphoid organs and is increased in response to activation of these cells by CD40L (24). It is able to inhibit trypsin-like proteases efficiently, although some inhibitory activity against thrombin and plasmin has also been shown (25). Two splice variants of the protein are also suggested to be expressed intracellularly.
SERPING1 (C1 inhibitor) is a plasma protein that inhibits various proteases of the complement system, contact system and fibrinolytic system (reviewed in 26). It has anti-inflammatory functions by controlling protease activation, but also has a function outside of its inhibitory role. It binds LPS, limiting the activation of macrophages and preventing sepsis (27). Additionally, it has been shown to anchor to sites of inflammation by binding exposed selectin molecules, where it can limit leukocyte migration (28).
Intracellular serpins in inflammation and immunity
In humans, there are 13 clade B serpins. Ten SERPINB genes are found on chromosome 18, which is syntenic with mouse chromosome 1. The mouse locus contains 12 serpin genes as a result of duplications of B3 (29, 30). Serpins within this cluster that may have roles in immunity include SERPINB2 and SERPINB10. B2 (plasminogen activator inhibitor 2) is a predominantly intracellular serpin found in eosinophils, monocytes and macrophages (reviewed in 31). It was initially characterized as an inhibitor of the extracellular protease urokinase-type plasminogen activator and is highly expressed in leukocytes in response to inflammatory mediators. The intracellular target(s) of B2 unknown but it protects retinoblastoma protein from proteolysis (32), and can decrease cell proliferation and DNA synthesis (33), so it may influence cell growth and apoptosis. However, mice lacking B2 do not have any defects in leukocyte differentiation or recruitment (34). B10 (bomapin) is a nuclear serpin found in bone marrow in the developing blood cells of the monocyte lineage (35, 36). Mitogen-induced maturation of monocyte cell lines reduces or ablates bomapin expression, suggesting an involvement in monocyte development (36).
The remaining human clade B serpins SERPINB1, SERPINB6 and SERPINB9 – which are those largely involved in immunity – are encoded by genes located in a separate cluster on chromosome 6. In the mouse, the syntenic locus on chromosome 13 contains 15 serpin genes (37). These mouse chromosome 13 serpin genes can be grouped by homology to human B1 (four genes), B6 (six genes) or B9 (seven genes) (37).
SERPINB1 (human MNEI; mouse elastase inhibitor) is an inhibitor of neutrophil elastase present in cells of the myeloid lineage, particularly granulocytes and macrophages (5, 38). Neutrophil elastase, a protease also expressed by neutrophils and macrophages, is stored in the azurophilic granules of these cells (39). It participates in phagocytosis by degrading bacterial components and is also released in small amounts into the extracellular space during inflammation, where it is hypothesized to assist in leukocyte migration and directly affects pathogens (40). Excess extracellular elastase activity has been implicated in the degradation of extracellular matrix proteins leading to disease including emphysema and cystic fibrosis (CF) (41).
B1 inhibits both elastase-like proteases and chymotrypsin-like proteases through two distinct P1 residues (38, 42). Elastinolytic proteases, including neutrophil proteinase 3 and elastase, cleave after Cys344, while chymotryptic enzymes, including cathepsin G and mast cell chymase, cleave after the Phe343 (42). Interestingly, B1 inhibits these enzymes at an SI near 1, indicating most of these interactions will be physiological and suggesting that B1 may be a general inhibitor of azurophilic granule proteases (42). There is also an overlap in function between B1 and SERPINB6, as both inhibit chymotrypsin-like proteases including the primary target of B6, cathepsin G (6, 42, 43).
B1 is found intracellularly in the nucleus and cytoplasm (44, 45) and can be inactivated by oxidation of residues within the RCL (5). This suggests that B1 requires a redox-controlled environment – such as the cytosol – to function correctly and would have a short half-life outside the cell. As little as 2% of neutrophil elastase is secreted from the cell, which is consistent with it acting primarily at the level of the phagosome (46). Thus, it is likely that B1 acts primarily to protect the cell from proteases released into the cytoplasm during stress or phagocytosis. This view is supported by mice lacking B1, which have reduced survival when infected with the bacterial pathogen Pseudomonas aeruginosa (47). This is apparently because of decreased viability of neutrophils and increased proteolysis of extracellular proteins, which is consistent with B1 being required to protect neutrophils from their own granular proteases (47). However, despite the expression of B1 in all cells of myeloid lineage, macrophages are unaffected by B1 loss. This could be because of lower expression of target proteases or increased levels of another partially compensating serpin. As increased expression of B1 occurs in the B6 knockout mice (43), it is possible that B6 increases in B1 knockout mice. To completely clarify the roles of B6 and B1 in maintaining neutrophil and macrophage lifespan both would have to be removed from the system simultaneously.
Much work has been performed investigating the impact of B1 on control of extracellular neutrophil elastase released during chronic infections, particularly respiratory diseases. B1 and granular proteases are found in increased concentrations in the respiratory fluids of CF patients and in models of bronchial pulmonary dysplasia (48, 49). Mouse and rat models of CF have shown that treatment with recombinant B1 increases protection of tissue during inflammation and enhances bacterial clearance by neutrophils (50, 51). Because granulocyte necrosis results in the release of vast quantities of elastase that overwhelms other locally available inhibitors such as A1 and secretory leukocyte protease inhibitor, treatment with recombinant B1 may restore the balance between neutrophil function and tissue remodeling during inflammation (50, 51). However, as mentioned above, B1 is normally found inside these cells (5). Extracellular B1 is found either in complex with elastase or in an inactive, cleaved state, but not in an active conformation (49). Although exogenous B1 has potential as a therapeutic for some respiratory diseases, release of active endogenous B1 is unlikely to occur under normal physiological conditions.
SERPINB6 (human PI-6; mouse Spi-3) is an intracellular serpin expressed primarily in myeloid cells, platelets, endothelial and epithelial cells (6, 52, 53). Human and mouse B6 both contain an arginine residue at the P1 position (52, 53). Because of this specificity, B6 is a potent inhibitor of tryptic proteases including plasmin, thrombin and kallikrein-8; however, its most physiologically relevant target protease is cathepsin G, a neutrophil azurophilic granule protease (6, 53, 54). Cathepsin G is a chymase that participates in the destruction of phagocytosed bacteria and fungi and cleaves a wide variety of substrates, including the pro-apoptotic effector, caspase 7 (55). B6 efficiently inhibits cathepsin G at an SI of 1 and association rate constant greater than 107 M/s (6). B6 is also present in mast cells, where it interacts with β-tryptase (56).
Because of its nucleocytoplasmic localization, B6 is proposed to protect host cells from cathepsin G, which may be released into the cytoplasm when the cell is under stress, or following granule–phagosome fusion during an inflammatory reaction (6). In vitro, neutrophils from mice lacking B6 are less effective at killing the yeast pathogen Candida albicans; however, there is no difference in the response to systemic infection (43). Furthermore, no difference in development or migration of B6-deficient neutrophils is evident, despite involvement of granule proteases in both these processes (43). These findings may be explained by the overlapping inhibitory profiles of different clade B serpins. For example, B1 also inhibits cathepsin G, and its expression increases in B6 heterozygous- and homozygous-deficient mice in all tested organs, particularly the spleen (43). Thus, B1 and B6 compensate for each other to ensure cell survival.
SERPINB9 (human PI-9; mouse Spi6) is an intracellular inhibitor of the cytotoxic protease, GrB (7, 57, 58). GrB is expressed primarily in cytotoxic lymphocytes (CL) comprising cytotoxic T lymphocytes and natural killer (NK) cells (59). It is stored in granules until recognition of a target cell causes perforin-mediated delivery of GrB into the target cell cytoplasm. GrB cleaves a wide variety of cytosolic pro-apoptotic substrates, including Bid and the effector caspases 3 and 7 (reviewed in 60). Efficient inhibition of GrB by B9 is therefore important to protect effector and bystander cells from unintended apoptosis arising from exposure to mislocalized GrB. The ability of B9 to protect against GrB-mediated death has been shown by overexpression of either human or mouse B9 in cell lines. It confers resistance to purified GrB and perforin or to GrB delivered by CLs (57, 61–63).
Despite similarities in function, human and mouse B9 probably control GrB through different mechanisms. Human B9 is a direct, highly efficient inhibitor of human GrB, mainly because of a P1 glutamic acid, which is cleaved only by proteases with a preference for acidic residues (7). It reacts with human GrB with an SI of 1 and has Kass above 106 M/s, indicating that it is a fast physiological inhibitor (7). By contrast, mouse B9 has a nonacidic P1 residue (Cys), a Kass outside the defined physiological range with both mouse and human GrB, and a high SI (6 against mGrB and >10 against hGrB) (61). These data suggest that although mouse B9 is able to inhibit mGrB-mediated death, it does not directly interact with mGrB. Recent reports have indicated that mGrB is much less cytotoxic than hGrB and lacks the ability to cleave the key pro-apoptotic substrate Bid because of differences in its catalytic cleft topology (61, 64, 65). The differences between the mouse and human B9 RCLs can be explained in three non-mutually exclusive ways: the reduced cytotoxicity of mGrB (61) means that mouse B9 does not need to be as efficient as human B9; mouse B9 controls mGrB-mediated death by interacting with downstream pro-apoptotic targets; or mouse B9 is also required to control other rodent-specific GrB paralogues to prevent death.
GrB shares with caspases the ability to cleave substrates after acidic amino acids, primarily aspartic acid. A key feature of human B9 is its acidic P1 (glutamic acid) residue, raising the possibility that B9 also controls caspases. However, B9 is a poor inhibitor of caspases in vitro (57, 66). Conversion of the P1 residue to aspartic acid improves inhibition of caspases but reduces the rate of GrB inhibition by 100-fold (57). This suggests that the P1 residue of B9 bestows selectivity for GrB without affecting caspases (57), which are required to remove redundant cells from the system (67).
Overexpression of B9 in Jurkat cells does not prevent caspase-mediated death initiated by Fas ligation, but death is prevented by B9 containing a P1 Asp (57). By contrast, B9 reduces Fas-mediated death following overexpression in HeLa cells (68) and decreases tumor necrosis factor (TNF)-, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)- and Fas-induced death in transfected sarcoma cells, possibly by binding to pro-caspase 8 or 10 and interfering with their auto-activation (69). In vitro studies using purified components show that B9 can slowly inhibit caspase 1; however, the kinetics indicate a non-physiological interaction (70, 71). In cell-based systems, B9 expression in vascular smooth muscle cells reduces interleukin-1β release, presumably by regulating caspase 1 (70). On balance, these results suggest some capacity of human B9 to control cytokine production by regulating the inflammatory caspases-1, -4 or -5, and possibly an ability to control pro-apoptotic initiator caspases. However, the obvious problem is that these studies have relied on examining expression of human B9 in transfected cell lines, and there have been no convincing analyses carried out on endogenous B9 in primary cells or tissues. Furthermore, no studies have examined the interaction of mouse B9 with caspases, although its nonacidic P1 residue would suggest that caspases are not directly regulated by mouse B9.
It has been proposed that a key role of B9 is in maintaining cytotoxic T lymphocyte (CTL) effector cell lifespan. B9 is expressed primarily in cells that express high levels of GrB, such as CL (7, 66, 72). B9 levels in cultured resting T cells are low, and increase following exposure to general mitogenic activators concanavalin A (ConA) and phorbol myristate acetate (PMA) (72). However, no study has investigated B9 expression in T cells under physiological stimuli such as viral infection. B9 associates with the cytosolic face of secretory granules when CLs are active, and is proposed to inhibit GrB released into the CL cytosol during receptor ligation or degranulation (72). This release of GrB within CL is illustrated by B9–GrB complex formation in NK cell lines, as well as in CTL and NK cells activated through receptor cross-linking (72–74). In addition, overexpression of B9 in primary CTL enhances their ability to kill target cells, suggesting that B9 provides protection from mislocalized GrB released into the cytoplasm during degranulation (72). This model is supported by data obtained from B9 knockout mice, where lymphocytic choriomeningitis virus (LCMV) infection causes a high level of CTL apoptosis, resulting in lower numbers of effector cells, a reduction in the number of granules per CTL and a defect in ability to lyse targets effectively (75). This phenotype is rescued by crossing the B9-deficient mice with GrB-deficient mice (75). Thus, B9 plays an important, if not essential, role in protection and maintenance of CTL populations.
The role of B9 in the formation or maintenance of memory cells is presently unclear as studies using transgenic mice overexpressing B9 disagree with more recent studies using B9-deficient mice (76, 77). Overexpression of B9 has no effect on the formation of effector CTLs but leads to increased numbers of CTL memory cells formed after LCMV infection (77). By contrast, absence of B9 has no effect on formation or maintenance of memory CTL (76). This disparity suggests that overexpressed B9 may be targeting another protease with unintended effects on memory cell lifespan. As memory cells were defined differently in the two studies, it is also possible that the subsets of memory cells analyzed have different expression of GrB and therefore react differently to B9 levels. Crossing the B9 transgenic mice with GrB-deficient mice would resolve these questions.
B9 may also have a protective role in accessory cells. Human dendritic cells (DCs) express high levels of B9, which is proposed to protect them from lysis by CTL that can occur during peptide presentation (66, 72). DCs from both myeloid and lymphoid lineage have the highest resting B9 level of any cell type, and expression increases in response to inflammatory mediators such as LPS, single-stranded RNA and TNF-α(78). Mouse B9 is similarly increased in primary DCs in response to LPS and CD40L (79). Increased B9 expression in both human- and mouse-activated DCs correlates with increased resistance to GrB-mediated death compared with naïve DCs, suggesting that it is necessary to shield DCs from GrB released from CLs during the process of antigen presentation and activation (78, 79). Furthermore, in the absence of correct maturation, DCs fail to express B9 and are subsequently sensitive to CTL lysis on presentation of peptide (79). Lower expression levels of B9 have been detected in CD4+ T lymphocytes, macrophages, mast cells and neutrophils (7, 72, 80). This expression was originally thought to protect these cells from misdirected GrB released from adjacent CL in the process of degranulation. However, recent evidence has shown that GrB is also expressed in CD4+ T lymphocytes, mast cells and macrophages (81–83), and there is debate about its presence in neutrophils (84–86). Therefore, B9 may also serve to protect these cells from endogenous GrB rather than exogenous GrB. Mouse B9 has been implicated in the maintenance of granulocytes, preventing death by inhibiting granule neutrophil elastase (87). However, the SI of elastase and B9 is 15, indicating that the interaction may not be relevant in a physiological setting unless the inhibitor is expressed in vast excess. Thus, B9 may be acting indirectly, perhaps by modulating the necrotic pathway downstream of neutrophil elastase, rather than by directly inhibiting neutrophil elastase itself.
B9 is expressed at sites likely to be exposed to inflammatory cells as well as at sites of immune privilege. It has also been detected in the circulation (88). B9 has been noted in endothelial and mesothelial cells, and levels increase when exposed to inflammatory mediators (89). B9 has also been implicated in protecting hepatocytes during an immune response. These cells increase B9 expression in response to multiple inflammatory mediators, including interferon-α, -β and -γ, and have been shown to increase expression in a GrB-dependent manner (90, 91). It has been suggested that NK cells, which infiltrate the liver, have high levels of GrB and kill exclusively via the granule exocytosis pathway, and hence high levels of B9 are required to provide protection for hepatocytes (92). RNAi knock down of mouse B9 expression leads to excess death of hepatocytes and causes acute liver failure (93). Similarly, immune-privileged sites including testis, ovaries, placenta, embryonic stem cells and the lens of the eye express B9 (66, 89, 94). Knock down of mouse B9 in embryonic stem cells renders them susceptible to GrB-mediated death (95). B9 has also been detected in the circulation as a monomer and in complex with GrB following cytomegalovirus infection and is suggested to inhibit circulating GrB (88). However, given the exclusively cytoplasmic distribution of B9 and the correlation between circulating B9 levels and viral load, it is more likely that infection leads to necrosis of B9-expressing cells causing release of B9 and B9–GrB complex.
Finally, perhaps the most controversial area of research involving B9 is its involvement in tumor resistance to CL-mediated lysis. Overexpression studies in cell lines have shown that B9 protects transformed cells from GrB-mediated lysis. However, it is unknown what effect, if any, in vivo B9 expression by tumors has on CL lysis (57, 61, 63). B9 expression has been reported in many tumor types, including leukemias, lymphomas and endothelial cancers originating from multiple organs including the skin, breast and colon (63, 96). Despite this, expression of B9 by a tumor does not always correlate with poorer prognosis (96–99), and most studies have failed to correlate results with the expression of B9 in the originating tissue or consider whether loss of B9 correlates with a better prognosis. An important question remaining is whether B9 expression in primary tumors provides protection from CL lysis. A recent study from Godal and colleagues using primary human CTLs on tumor cells suggested that expression of B9 has little or no effect on susceptibility to CTL lysis (100). This indicates that factors such as other granzymes and death ligands in the CTL, or receptors, pro- and anti-apoptotic molecules in the target, also play important roles in tumor susceptibility.
Serpins function in the immune system to limit damage caused by proteases at extracellular sites of inflammation, and in the cytoplasm of effector, accessory and bystander cells. Emerging evidence from knockout models indicates that maintenance of effector cell number and function are dependent on protection by intracellular serpins from endogenous granule proteases. Further research is needed to define the roles these serpins play in human pathophysiology.
MSJM is the recipient of an Australian Postgraduate Award. Work in this laboratory is supported by the National Health and Medical Research Council of Australia.