et al., 1996b; Shi et al., 1997). The mannose phosphate receptor (MPR) was thought to constitute the main pathway for granzyme entry into the cytoplasm (Motyka et al., 2000). However, more recent work has demonstrated that free granzymes do not require MPR to enter the target cell (Trapani et al., 2003). It has also been shown that a maximum of only 20% of grB is mannose- 6 phosphorylated, indicating that MPR-independent pathways must exist (Bird et al., 2005). Other studies have suggested that grB is predominantly complexed with the proteoglycan, serglycin, and that the entire complex is taken up into endosomes of the target cell via the MPR pathway (Veugelers et al., 2004). Although still controversial, recent studies have provided evidence against this idea (Bird et al., 2005; Raja et al., 2005), and cells deficient in M6R expression were as efficiently killed by CLs through granzyme-mediated apoptosis as M6R expressing cells, indicating that whether complexed to serglycin or not, granzymes do not require the M6R to mediate target cell entry (Trapani et al., 2003). In the absence of receptor-mediated endocytosis, granzymes have been postulated to enter the cell through fluidphase endocytosis (Bird et al., 2005; Trapani et al., 2003). After internalization, grB remains trapped in endocytic vesicles and cannot exert its apoptotic effects unless perforin or another pore-forming toxin is also present (Browne et al., 1999; Froelich et al., 1996b).

More recently, a direct role for perforin in grB entry into the target cell was proposed. Sublytic perforin levels caused calcium influx into the target cell, which triggered membrane repair and coincided with the endocytosis of granzymes (Keefe et al., 2005). These results suggest a requirement for the synchronous application of both perforin and grB to mediate target cell apoptosis. This proposed “endosomolytic” function of perforin, although popular as a hypothesis, has as yet not been supported by significant evidence. The different mechanisms by which perforin may facilitate granzyme entry into the target cell are shown in Figure 10-1.

2.5. Activation of death pathways by granzymes

Once released inside the target cell, granzymes are capable of processing various intracellular substrates, resulting in cell death. Granzymes are serine proteases that belong to the chymotrypsin superfamily and share common characteristics with chymotrypsin-like enzymes (Henkart et al., 1987). One of the key features of serine proteases is a triad of conserved residues (histidine, aspartic acid, and serine) at their catalytic site (Kraut, 1977; Murphy et al., 1988). A total of 11 granzymes have been identified in mice (A-G, K, L, M, and N), but only 5

Table 10-2. Chromosomal localization of functional granzyme gene subsets

Note: The granzyme genes are distributed to three loci, with each subfamily constituting a broad type of substrate specificity, either trypsinlike (tryptase), chymotrypsin-like (chymase), or cleavage after methionine (metase) activity.

Source: This table is a modified version of (Trapani, 2001) and (Grossman et al., 2003).

exist in humans (A, B, H, K, and M). Granzymes in both humans and mice are grouped functionally and genetically on the basis of their genes, localizing to one of three chromosomal loci, as summarized in Table 10-2.

Granzymes have very specific substrate specificities and are clearly processing (nondegradative) enzymes. Some granzymes (grA, K) cleave at basic residues (lys, arg) and others at bulky nonpolar residues (phe, trp) and therefore have trypsin-like (“tryptase”) or chymotrypsinlike (“chymase”) activity, respectively. Similarly, specificity for asp residues (grB) or met residues (grM) results in “aspase” and “metase” activity, respectively. The disparate substrate specificity suggests that granzymes may trigger specific death pathways and/or possess quite different additional functions. A key emphasis in this chapter is to describe the biological substrate preferences of different granzymes directly resulting in cell death.

3. GRANULE-BOUND CYTOTOXIC PROTEINS

The components of the secretory granules present in the CL can be categorized according to their proposed functions (summarized in Table 10-3). Some of the granule components are discussed in greater detail below.

3.1. Perforin

As described above, perforin plays a critical role in CL biology, primarily through its ability to form a pore in lipid membranes. However, very little is known about the mechanism of perforin pore formation and the specific perforin domains involved in this process. This is due, at least in part, to difficulty in expressing significant

quantities of perforin for structural studies. No structure exists for perforin, and few assays that probe structure/function relationships have been devised. Human perforin is a 67-kDa protein and consists of 555 amino acids, including its 21-residue signal peptide. There are some predicted functional domains interspersed throughout the protein, which are based on a combination of direct comparisons with complement proteins, perforin peptide experiments, and more recent perforin mutagenesis studies.

The membrane-interacting domain of both complement and perforin is commonly referred to as the membrane attack complex/perforin domain (MACPF) because of sequence similarity and proposed functional homology (Kwon et al., 1989; Lowrey et al., 1989; Shinkai et al., 1988). Crystal structures of the first proteins containing a MACPF domain, human C8α (Hadders et al., 2007; Slade et al., 2008) and Plu-MACPF (a putative toxin synthesized by the bacterium Photorhabdus luminescens), have recently been solved (Rosado

(MacDermott et al., 1985; Stevens et al., 1989; Stevens et al., 1987)

(Dupuis et al., 1993)

et al., 2007; Rosado et al., 2008). Crucially, the structural data revealed homology with cholesterol-dependent cytolysins (CDCs), a large family of bacterial poreforming toxins whose molecular mechanism is better understood. CDCs do not possess alpha helices capable of membrane spanning, but instead are thought to form a “pre-pore” before insertion to enable membraneinteracting domains to be revealed (Dang et al., 2005; Tilley et al., 2005). This mechanism of pore formation sees polymerization occurring after membrane binding and before membrane insertion. Based on the high degree of structural similarity predicted between perforin and these toxins, further studies are warranted to

determine whether perforin also inserts into membranes via a similar mechanism.

3.2. Granulysin

Granulysin is a cytolytic member of the saposin-like family of lipid-binding proteins (Clayberger and Krensky, 2003; Munford et al., 1995). Mice do not have a granulysin gene, and this cationic molecule is only present in the secretory granules of human CLs. It has lytic activity against various microbes, including Gram-positive and -negative bacteria, fungi, and parasites (Pena and Krensky, 1997). Granulysin has also been speculated to