(A) In this model, Ptdlns (PI) is either delivered directly to PtdIns4K (PI4K) or to the vicinity of the signalling complex and subsequently channelled through the complex. (B) In the second model, Ptdlns remains bound on PITP through the sequential phosphorylation and Ptdlns(4,5)?2 (PIP2) bound to PITP is the preferred substrate for PLC. PIP5K, phosphatidylinositol phosphate
5-kinase; PC, phosphatidylcholine; IPs, inositol-1,4,5-trisphosphate.
so that cellular responses are fine-tuned and optimised, thereby enhancing specificity, selectivity and speed while preventing unintended cross-talk between different pathways.
A good example is the protein InaD, which has five PDZ domains and regulates phototransduction in D. melanogaster photoreceptor cells (Shieh and Zhu, 1996), Using distinct PDZ domains, InaD interacts with three components o f the light-triggered PI cascade for phototransduction in the fruit fly visual system: the transient receptor potential (TRP) Ca^^ channel, PLCp and PKC. InaD apparently acts as a scaffolding protein to assemble these proteins into a signalling complex so that (i) efficient activation o f the TRP channel by PLCp in response to rhodopsin stimulation and (ii) deactivation o f TRP through phosphorylation by PKC can occur (Tsunoda et aL, 1997). InaD mutants defective in a particular PDZ domain produced incomplete complexes lacking the appropriate target protein and displayed corresponding defects in their physiology (Tsunoda et aL, 1997). In null mutants, signalling complexes were not formed, the three signalling molecules were mislocalized, and their protein levels markedly reduced. Thus, InaD is a modular, multivalent adaptor protein that not only links multiple signalling components within the same cascade but is also required for their stability and subcellular localization. These studies also demonstrated that it is the location o f a transduction molecule in the cell and not its presence per se that promotes effective signalling (Scott and Zuker, 1998; Tsunoda et aL, 1997).
An important implication o f the above-mentioned ‘co-factor m odel’ is that all intermediates in the PI signalling pathway are isolated from their counterparts in the plasm a membrane, resulting in metabolic compartmentation. This could help explain the empirical phenomenon o f agonist-responsive and -unresponsive pools o f cellular Ptdlns (Monaco and Gershengom, 1992). A pool is defined as a population o f PPIs which can be distinguished biochemically following cell-labelling experiments or fractionation o f cell lysates. Different lipid pools do not necessarily represent synthesis or localization at different intracellular membranes but pools may be on the same membrane and possibly sequestered by membrane proteins.
The agonist-responsive Ptdlns pool is the fraction o f total cellular Ptdlns that can be hydrolysed in response to receptor agonist, while the agonist-unresponsive pool is apparently resistant to hydrolysis, even after prolonged stimulation. It is presumed that distinct Ptdlns pools serve different cellular functions and may employ different PI kinase isoforms to mediate and regulate their metabolism. For example, overexpressing the yeast
PtdlnsPK Mss4p did not completely suppress the G2/M boundary blockage caused by
mutation o f stt4, a yeast PtdIns4K gene (Yoshida et aL, 1994b). This suggests that there may be a branch point in the signalling pathway between Stt4p and Mss4p and/or that Stt4p-dependent G2/M cell progression uses a P td ln s4 f pool different from that used by
Mss4p.
The existence o f distinct Ptdlns pools will also require segregation o f the remaining intermediates within the Ptdlns cycle. The DAGK a-isozym e has been found to associate with the activated EGFR (Schaap et aL, 1993) and may be a component o f a PI signalling complex associated with the EGFR (Payrastre et aL, 1991). DAG molecules which were generated randomly in the plasma membrane by an exogenous bacterial Ptdlns-specific PLC were not accessible to receptor-activated DAGK (van der Bend et aL, 1994). This topological restriction o f DAGK activity suggests that DAGK phosphorylates only DAG generated as a result o f receptor stimulation and is in line with the concept o f an agonist- responsive Ptdlns pool in which receptor-induced DAG is channelled from endogenous PLC to the DAGK aetive site. The data also imply the compartmentation o f DAG and possibly PtdOH. Such segregation o f intermediate metabolites is also reflected in experimental data which suggest that hydrolysis and resynthesis o f Pis occurred in a ‘closed’ PI cycle, whereby Ptdlns synthesized in response to agonist-induced depletion is preferentially sensitive to subsequent agonist-induced hydrolysis (Monaco and Gershengom, 1992, and references therein).
Compartmentation o f Pis and their derivatives have also been observed in the nucleus (D'Santos et aL, 1999; Vann et aL, 1997). For example, there may be at least two distinct pools o f DAG in the nucleus: one generated from Pis and the other from PtdCho, and the former was suggested to be the pool that mediates PKC translocation into the nucleus (Neri et aL, 1998). In addition, PI kinases and lipases are localized to different nuclear sites (Payrastre et aL, 1992). PtdIns4K-IIip was concentrated at lamina-pore complexes in the nucleus o f NIH 3T3 cells (de G raaf et aL, 2002), while PtdlnsfK s and PLCs were found in the inner nuclear compartment (Boronenkov et aL, 1998; Cocco et aL, 2000; Zini et aL, 1993).
Despite years o f studies, the physical manifestation o f these PI pools remains enigmatic, and different data have both supported and questioned their existence (Monaco and Gershengom, 1992). Nevertheless, at least three mechanisms have been proposed to help explain the metabolic compartmentation o f Pis. Firstly, EGF-induced PI tumover has
been suggested to take place in caveolae, which are discrete detergent-insoluble plasma membrane structures that are enriched in caveolin (Parton, 1996). Caveolae appear to contain most, if not all, o f the molecules involved in PI signalling, including receptors, PI kinases and Pis. Secondly, Pis in agonist-responsive pools have predominantly stearyl and arachidonyl acyl chains, while those in agonist-unresponsive pools contain other types o f acyl chain (Lee et aL, 1991). This is in line with studies showing that PITPs and DAGKs discriminate between different PI acyl groups (Tang et aL, 1996; Wirtz, 1991). Finally, the in vitro enzymatic activities o f PtdIns4K, PtdlnsPK and PLC isozymes differ with different modes o f substrate presentation, for example, in the presence o f carrier lipids, detergents and PtdlnsTP. These three mechanisms are not mutually exclusive and pave the way to a testable hypothesis for PI compartmentation.
M ost o f the above data support the concept that PITP plays an important role in the compartmentalised synthesis o f P td ln sf] in the cell, although the mechanism remains unresolved. The main objective o f this thesis is therefore to investigate and test the central tenets o f the co-factor model o f PITP function in PI signalling, that is, (i) the ability o f PITP to interact with PI signalling complex(es), and (ii) the viability o f PITP-bound Ptdlns as a direct substrate to PI kinases.