Biosynthesis of phospholipids is of central importance in the normal functioning of biological membranes. S. cerevisiae
has served as an outstanding model system allowing the fundamental mechanisms of eukaryotic membrane production to be defined. While genetics have clearly indicated the existence of both a mitochondrial and a extra-mitochondrial route for de novo PE production, clear phenotypes only existed for the mitochondrially localized Psd1 pathway (Birner et al., 2001
). Here, we provide evidence that the second route of PE production catalyzed by the phosphatidylserine decarboxylase Psd2 is required for normal function of a vacuolar membrane protein. This functional requirement is mediated by the need for correct PE levels in the vacuolar membrane that support normal activity of Ycf1 and possibly other proteins.
Psd2 represents a relatively unusual type of phosphatidylserine decarboxylase enzyme and is found in plants but not in bacteria or mammals (Voelker, 1997
). Psd1 homologues are present in mammalian mitochondria and at the inner membrane of bacteria (reviewed in Vance, 2003
). Proteins exhibiting high sequence similarity to Psd2 have been described in plants (Nerlich et al., 2007
), suggesting the possibility that a common role for control of internal membrane PE exists in these organisms. Ycf1 homologues have been described in plants that localize to internal membranes and are required for cadmium resistance (Tommasini et al., 1996
Although Psd1 and Psd2 catalyze the same reaction, the PS substrate is apparently presented to each enzyme quite differently. Psd1 is located close to the site of PS synthesis, closely resembles the E. coli
Psd enzyme and has not been shown to require any other enzyme to carry out its role. Conversely, Psd2 is nearly twice the size of Psd1 and not as closely related to the bacterial enzyme (Trotter et al., 1995
). Psd2 also requires the presence of the Pdr17 PITP to be able to productively interact with PS as even elevated dosage of PSD2
is unable to bypass a pdr17
Δ mutation (). Previous work has demonstrated that membrane association of Psd2 requires the presence of an amino-terminally located C2 domain (Kitamura et al., 2002
). Together, these data suggest that although Psd2 and Pdr17 form a complex on the membrane, the independent membrane association of both proteins is likely required to permit complex formation.
The finding that Psd2 was localized to the endosomes was unexpected. Previous assignment of Psd2 to a Golgi/vacuole distribution (Trotter and Voelker, 1995
) was based strictly on biochemical fractionation experiments which did not provide the resolution to define the endosomal location we show here. We anticipated finding Psd2 on the vacuolar membrane because we found that loss of this enzyme triggered a cadmium-sensitive defect centered on reduction of Ycf1 function. The endosomal enrichment of Psd2/Pdr17, coupled with the demonstration that loss of this complex lowered vacuolar PE levels, indicates that phospholipid levels of the vacuolar membrane may be controlled by the direct action of this enzyme complex on the endosome. This suggestion is directly supported by our measurements of PE content of internal membranes. Modulation of endosomal PE levels is eventually communicated to the vacuolar membrane through vesicular transport. The indirect control of phospholipid content by Psd2 is reminiscent of the effect of the P-type ATPase Drs2 on plasma membrane asymmetry (Chen et al., 2006
). Drs2 is localized to the Golgi membranes (Chen et al., 1999
), yet its loss causes defects in distribution of phospholipids on the plasma membrane.
Changes in PE levels have been extensively documented to lead to problems in folding of membrane proteins (recently reviewed in Dowhan and Bogdanov, 2009
). Most of these studies have been carried out using reconstituted membrane proteins in an in vitro setting. Our experiments provide in vivo demonstration of the importance of phospholipid composition in regulating membrane protein function. An alternative view of the consequences of loss of Psd2 on Ycf1 function is that elevated PS levels act to inhibit transporter activity. Based on the extensive documentation of the stimulatory effect of PE on folding and subsequent activity of membrane proteins (Bogdanov et al., 1999
; Zhang et al., 2003
; Hakizimana et al., 2008
), we favor the positive effect of PE levels increasing Ycf1 activity rather than high PS levels inhibiting function of this ABC transporter.
Another important distinction between previous work carried out primarily on prokaryotic permeases and stimulation of their folding by PE (Wang et al., 2002
), comes from the fundamental differences in the biogenesis of prokaryotic and eukaryotic membrane proteins. The exceptionally well-studied prokaryotic lactose permease, in which the positive effect of PE on folding has been demonstrated (Bogdanov et al., 1999
), is inserted into its final membrane destination as it is being synthesized. This is quite different from Ycf1 that follows the typical itinerary of eukaryotic membrane proteins consisting of initial biosynthesis and insertion into the membrane of the endoplasmic reticulum, transit through the Golgi and only then reaching its vacuolar membrane site of function (Wemmie and Moye-Rowley, 1997
). Our data suggest a possible means of regulation of Ycf1 as the unique phospholipid composition of the vacuole compared with other internal membranes is evidently required for full activity of this transporter. This would have the result of keeping Ycf1 activity relatively low as it transits the secretory pathway en route to its functional residence in the vacuolar membrane.
Our previous work on Psd1 demonstrated that this phosphatidylserine decarboxylase protein has another function that in mitochondrial-nuclear signaling in addition to its enzymatic activity (Gulshan et al., 2008
). Psd1 also closely resembles its bacterial counterpart Psd (Clancey et al., 1993
; Trotter et al., 1993
) that has been carefully studied in enzymological terms. Although Psd1 and Psd2 carry out the same conversion of PS to PE, they execute this reaction in dramatically different ways. All available data indicate that Psd1 requires no other protein to function as a phosphatidylserine decarboxylase, whereas Psd2 clearly requires Pdr17 (Wu et al., 2000
). Additionally, loss of Psd1 C-terminal processing prevented enzymatic function but had no detectable effect on steady-state level of the resulting mutant protein (Gulshan et al., 2008
). Conversely, introducing this same C-terminal processing block into Psd2 led to a reduction in the level of mutant protein. Perhaps the different subcellular distributions of Psd1 and Psd2 explain the differential dependence on normal C-terminal processing for protein expression.