Farnesyl diphosphate i GGTase-I
Fig. 1. Overview of the isoprenoid biosynthetic pathway and the structures of the prenyl groups attached to proteins by FTase and GGTase I.
was used in these types of experiments, the label was found to be incorporated into a number of cellular proteins, dubbed "prenylated proteins" (10,11).
The first prenylated mammalian protein identified was the nuclear protein lamin B (12,13). At about the same time, independent studies on the a-factor mating peptide of Saccharomyces cerevisiae revealed that this peptide was modified by a farnesyl isoprenoid (14). The realization that both lamin B and the a-factorpeptide contained a so-called "CAAX motif' at their carboxyl terminus (where "C" was the cysteine residue that served as the isoprenoid attachment site, "A" signified an aliphatic amino acid, and "X" denoted an undefined amino acid) prompted examination ofother proteins containing the motif to determine if they too were prenylated. Foremost among the CAAX-containing proteins examined were the products of the Ras family of protooncogenes. The discoveries that Ras proteins were modified by farnesylation and that the modification was required for the oncogenic forms of these proteins to transform cells triggered an immediate and widespread interest in this form of lipid modification (15-17). Subsequent studies have identified almost a hundred prenylated proteins in mammalian cells (1,5,18), and revealed that, in addition to the 15-carbon farnesyl moiety, the 20-carbon geranylgeranyl isoprenoid could also be attached to proteins (19,20) (Fig. 1).
For CAAX-containing proteins, prenylation is but the first step in a series of three posttranslational modifications that occur at the C-terminus of most of these proteins. The three C-terminal amino acids (i.e., the -AAX) are subsequently removed by a membrane-bound protease, and finally a membrane-bound enzyme methylates the newly-formed carboxyl group to produce a methylester at the C-terminus (21,22). Furthermore, in addition to the modifications at the CAAX motif, in some prenylated proteins other modifications such as palmitoylation and phosphorylation can occur in the C-terminal region just upstream of the CAAX motif (5).
Monomeric guanine nucleotide (GTP)-binding proteins (G proteins) such as Ras, Rap, Rho, and Rab comprise the largest set of prenylated proteins (5,23). Among these G proteins, Ras proteins have attracted particular attention because of the important role of Ras in carcinogenesis (24,25). The normal functions ofRas proteins are in cellular signal transduction pathways that are essential for cell growth and differentiation (25-27). Moreover, specific mutations in Ras proteins render them oncogenic, and such mutations are found in about 30% of all human tumors, including over 90% of human pancreatic cancer and 50% of human colon cancer (18,24). The dependence of the transformed phenotype on the constitutive activity of Ras prompted considerable speculation that blocking the Ras signaling pathway could provide a way to treat such cancers (24). Hence, the finding that farnesylation is absolutely required for oncogenic Ras function identified a specific point in the process, i.e., the attachment of the isoprenoid, for which development of specific inhibitors might provide an approach to this type of cancer chemotherapy (18,28-30).
2. FTASE AND GGTASE-I: THE CAAX PRENYLTRANSFERASES
The first identified protein prenyltransferase was protein farnesyltransferase (FTase), originally isolated from rat brain cytosol using an assay that followed the incorporation of radiolabel from 3H-FPP into a recombinant Ras protein (31). The finding that CAAX proteins containing methionine or serine at their C-terminus were farnesylated, whereas those ending in leucine were modified by a geranylgeranyl moiety (32-34), provided the initial evidence for the existence of a distinct enzyme that would catalyze the addition of geranylgeranyl to certain proteins in the CAAX class. Using an approach similar to that which led to the identification of FTase, an enzymatic activity capable of transferring the geranylgeranyl group from geranylgeranyl diphosphate to candidate proteins was identified (35,36). This enzyme, protein geranylgeranyltransferase type I (GGTase I), exhibited properties similar to those of FTase (see below). The C-terminal leucine residue was shown to be responsible for the specific recognition of substrate proteins by GGTase I by producing a Ras protein with a leucine-for-serine switch at the COOH-terminal position, a switch that converted the Ras protein from a FTase to a GGTase I substrate (35).
Mammalian FTase and GGTase I share many properties (37). Both enzymes are heter-odimers that contain a common subunit (designated the a-subunit) of48 kDa and distinct P-subunits of 46 kDa (FTase) and 43 kDa (GGTase I) (31,38,39). Both proteins are zinc metalloenzymes that operate through apparently quite similar kinetic and chemical mechanisms (see below). Both enzymes have been cloned, and sequence analysis has 1) confirmed that the a-subunits are the products of the same gene and 2) revealed that the P-subunitshad~35% sequence identity atthe amino acidlevel (40-42). The significance of the two enzymes sharing a common subunit is not yet clear, but the existence of an identical a-subunit andahighly homologous P-subunit for these two protein prenyltrans-ferases provided the initial evidence that discrete segment(s) of the P-subunit would be responsible for the remarkable substrate specificities of the enzymes.
Structural information just recently has begun to emerge on the CAAX prenyltrans-ferases. Data to date have come from analysis of mammalian FTase, whose X-ray crystal structure was determined at 2.2 A resolution (43). In this structure, which was of the free (i.e., unliganded) enzyme, the a-subunit was found to be folded into a crescent-shaped domain composed of seven successive pairs of coiled coils termed "helical hairpins," which contact a significant portion ofthe P-subunit. The existence ofrepeat motifs in this subunit was first predicted from sequence alignments of mammalian and fungal a-sub-units (44). The P-subunit was also found to consist largely of helical domains, with the majority of the helices arranged into an a-a barrel structure. One end of the barrel was open to the solvent, while the other end was blocked by a short stretch of residues near the C-terminus ofthe P-subunit. This arrangement results in a structure containing a deep cleft in the center of the barrel that possesses all of the features expected for the active site ofthe enzyme, including the aforementionedbound zinc ion (see Subheading 2.4.1.). Quite recently, crystal structures of the complex of FTase with its isoprenoid substrate FPP have been reported that provide a quite detailed snapshot of the binding site for this substrate on the enzyme (45,46) (see also Chapter 3).
2.2. Recognition of Substrates
Binding of isoprenoid substrates by both CAAX prenyltransferases is of very high affinity with KD values being in the low nM range (47-49); the initial realization of this property came from findings that the enzyme-isoprenoid complexes could be isolated by gel filtration (50,51). The use of photoactivatable analogs of both FPP and GGPP revealedthatthe analogs couldbe specificallycrosslinkedto the P-subunits ofFTase and GGTase I, respectively, upon activation (52-54); these and related findings with peptide substrates (see Subheading 2.2.2.) provided the initial evidence that the active sites for the enzymes were, as expected, predominately associated with the P-subunits. The recent crystal structures determined for FTase-FPP complexes have provided the formal proof of this hypothesis (45,46).
An early observation made with FTase was that the enzyme could bind both FPP and GGPP with relatively high affinity, although only FPP could serve as a substrate in the reaction (50). A more detailed study of the FPP binding properties of FTase revealed that there is in fact a significant difference in the binding of the two isoprenoids, with GGPP binding being some 15-fold weaker than that of FPP (49), although this still translates to an apparent affinity of ~100 nM for GGPP binding to FTase. A structural-based hypothesis why FTase exhibits such high affinity binding of GGPP to form a complex that is essentially catalytically inactive has been advanced (46). Briefly, this hypothesis—discussed in detail in Chapter 3—is that the depth of a hydrophobic binding cavity in the P-subunit acts as a ruler that discriminates between the two isoprenoids based on their chain length. No structural information is yet available for GGTase I, although this enzyme exhibits much higher selectivity; binding ofGGPP to the enzyme is several hundred-fold tighter than that of FPP (48,49).
Analogs of FPP have been identified that bind to FTase with high affinity but cannot participate in catalysis (55,56). These analogs have been quite useful in mechanistic studies of FTase, because they allow formation of an inactive FTase-isoprenoid binary complex (see Subheading 2.4.). Analogs of GGPP that should allow similar studies with GGTase I have also been described (57,58).
As noted in Subheading 1, mammalian cells contain a wide variety of proteins that are processed by CAAX prenyltransferases. Substrates of FTase include all four Ras proteins, nuclear lamins A and B, the y-subunit of the retinal trimeric G protein transducin, and a variety ofkinases and phosphatases (18,59-63). Known targets of GGTase I include most identified y-subunits of heterotrimeric G proteins and a multitude of Ras-related monomeric G proteins, including most members of the Rac, Rho, and Rap subfamilies
(1,5). All these protein substrates contain a Cys residue precisely four amino acids from the C-terminus. Furthermore, as noted in Subheading 2.1., the identity of the C-terminal residue (i.e., the "X" of the CAAX motif) determines which of the two enzymes will act on the protein. FTase prefers proteins containing Ser, Met, Ala or Gln, whereas Leu at this position directs modification by GGTase I (1,21). This property ofthe enzymes make it possible to predict with reasonable accuracy from its primary sequence which prenyl modification will be on a protein.
An important property of both FTase and GGTase I is that they can recognize short peptides containing appropriate CAAX motifs as substrates (31,36,64). A quite detailed analysis of specificity in recognition of Ca1a2X sequences by FTase indicates that the a1 position has a relaxed amino acid specificity, while variability at a2 and X are more restricted. Basic and aromatic side chains are tolerated at a1 but much less so at a2, whereas acidic residues are not well-tolerated at either position (64,65). Moreover, substitution at the a2 position by an aromatic residue in the context ofa tetrapeptide creates a molecule that has been reported to be not a substrate for FTase but rather a competitive inhibitor (66). One such peptide, CVFM, has served as the basis for design of peptidomimetic inhibitors of FTase (67-69).
Binding of peptide substrates to FTase has been examined by nuclear magnetic resonance (NMR) using a resonance transfer approach. One such study reported that the CAAX sequence of a peptide substrate adopts a Type I P-turn conformation when bound to the enzyme (70). However, a similar study of binding of a peptidomimetic inhibitor of FTase termed L-739,787 revealed a slightly different conformation most closely approximating a Type III P-turn (71). A note of caution here is that, in both cases, the binding ofthe pep-tide/peptidomimetic was examined in the absence of bound isoprenoid on the enzyme. The recent realization that the kinetic mechanism is most likely an ordered one in which isoprenoid binding precedes that of the peptide/protein substrate (47), and that the binding of the isoprenoid markedly increases the affinity for the peptide substrate (72) (see Subheading 2.3.) may have profound implications for this data, as the binding ofthe pep-tide/protein substrate to the enzyme-isoprenoid complex may be very different than its binding to free enzyme.
The zinc ion in both FTase and GGTase I is essential for the high affinity binding of the protein substrate (but not, however, for binding of the isoprenoid substrate) (48,73), and recent studies indicate a direct coordination of the thiolate of the Cys residue of the protein substrate with the metal ion during catalysis (72,74). Further evidence supporting a metal-substrate interaction in the enzymes comes from studies showing that the zinc ion can be replaced by cadmium, and the cadmium-substituted enzymes retain steady-state activity but have somewhat altered protein substrate specificities (48,75). The location of the zinc ion was determined in the crystal structure of FTase to be in the P-subunit near the interface with the a-subunit (43), consistent with the findings that both protein andpeptide substrates canbe crosslinkedto the P-subunit ofFTase (50,76), andthat short peptide substrates containing divalent affinity groups label both the a- and P-subunits upon photoactivation (76).
2.2.3. Cross Prenylation by CAAX Prenyltransferases: Is it Important?
Although FTase and GGTase I seem to be quite selective for their substrates, cross-specificity (i.e., modification of a protein by either enzyme) has been observed (36,65). However, whether this ability to modify alternate substrates is of physiological significance is still somewhat unclear. The most compelling evidence in this regard comes from studies in fungal systems. Yeast lacking FTase are viable, although they exhibit growth defects (77). Overexpression of GGTase I in these mutants can partially suppress the growth defects, suggesting that GGTase I can at some level prenylate substrates of FTase (78). Although yeast lacking GGTase I are not viable, the phenotype can be rescued by overexpression of two essential G protein substrates of the enzyme (78), suggesting that FTase can prenylate some substrates of GGTase I if the substrate proteins are overproduced.
In terms of mammalian CAAX proteins, two Ras isoforms—K-Ras4B and N-Ras— can serve as substrates for both FTase and GGTase I in vitro, although they are much better substrates for FTase (79,80). Although under normal conditions these two Ras isoforms seem to be modified solely by the farnesyl group, geranylgeranylation of the proteins can be detected in cells if FTase is inhibited (81,82). The primary determinant for this type of cross-prenylation appears to be the existence of a Met as the C-terminal residue of these proteins (80). Although these studies do not provide evidence to support the notion that cross-prenylation has significance under normal physiological conditions, it is certainly a concern in terms of the biology associated with FTase inhibition. The discussion of these concerns can be found in Chapters 5,13, and 15.
There do appear to be some mammalian proteins that can be normally modified by either isoprenoid. One such example is the Ras-related small G protein TC21, where the presence of Phe as the C-terminal residue apparently allows modification by either enzyme (83). Additionally, another Ras-related small G protein, RhoB, has been shown to be farnesylated as well as geranylgeranylated even though its C-terminal residue is Leu (84,85); farnesylation of this protein is most likely due to an ability to be processed by FTase, rather than an alternate activity of GGTase I (86). How RhoB gets recognized and farnesylated by FTase is not yet clear, although a Lys residue in the second position of the CAAX motif may be partly responsible. Whatever property is responsible for this "dual prenylation," the differently prenylated forms of RhoB apparently have unique functions, as suppressing the farnesylated population by treatment of cells with a FTase inhibitor suppresses RhoB-dependent cell growth (86).
Mammalian CAAX prenyltransferases are quite slow enzymes, with kcat values in the range of 0.05 s-1(31,87). Steady-state kinetics of mammalian FTase were initially interpreted as indicating a random-order binding mechanism in which either substrate could bind first (87). However, the failure to trap enzyme-bound protein or peptide substrate in transient kinetic experiments suggested that either substrate binding is actually ordered or that the dissociation rate constant of the protein/peptide substrate is so fast and the affinity is so weak that farnesyl diphosphate (FPP) binding first is the kinetically preferred pathway (47,87) (Fig. 2). Consistent with this functionally ordered mechanism, the affinity of FPP for FTase is in the low nM range; whereas the affinity of the peptide substrate in the absence of bound FPP is relatively weak but this affinity is increased several hundred-fold by the binding ofFPP analogs (72). The aforementioned pre-steady-state kinetic studies also revealed that the association of the peptide substrate with the FTase . FPP binary complex was effectively irreversible with a kassoc of2 x 105 M-1s-1 (47). While the rate constant for product formation could not be accurately determined in these studies, a lower limit of > 12 s-1 was established using protein fluorescence (47). A more precise determination of the rate constant for product formation has come from
Was this article helpful?