Preferential decarboxylation of hydrophilic phosphatidylserine species in cultured cells. Implications on the mechanism of transport to mitochondria and cellular aminophospholipid species compositions.

In baby hamster kidney and other cultured cells the majority of phosphatidylethanolamine (PE) is synthesized from phosphatidylserine (PS) in a process which involves transport of PS from the endoplasmic reticulum to mitochondria and decarboxylation therein by PS decarboxylase. To study the mechanism of this transport process, we first determined the molecular species composition of PE and PS from baby hamster kidney and Chinese hamster ovary cells. Interestingly, the hydrophilic diacyl molecular species were found to be much more abundant in PE than in PS, suggesting that hydrophilic PS species may be more readily transported to mitochondria than the hydrophobic ones. To study this, we compared the rates of decarboxylation of different PS molecular species in these cells. The cells were pulse labeled with [3H]serine whereafter the distribution of the labels among PS and PE molecular species was determined by reverse phase high performance liquid chromatography and liquid scintillation counting. The hydrophilic PE species contained relatively much more 3H label than those of PS, which indicates that they are more readily decarboxylated than the hydrophobic ones. Control experiments showed that differences in [3H]PS and -PE molecular species profiles are not due to (i) incorporation of 3H label to some PE species via alternative pathways, (ii) differences in degradation or remodeling among species, or (iii) selective decarboxylation of PS molecular species by the enzyme. Therefore, hydrophilic PS species are indeed decarboxylated faster than the hydrophobic ones. The rate of decarboxylation decreased systematically with hydrophobicity, strongly suggesting that formation of so called activated monomers, i.e. lipid molecules perpendicularly displaced from the membrane (Jones, J. D., and Thompson, T. E. (1990) Biochemistry 29, 1593-1600), is the rate-limiting step in the transport of PS from the endoplasmic reticulum to mitochondria. The formation of activated monomers and thus the rate of transfer is probably greatly enhanced by frequent collisions between the two membranes which tend to be closely associated. The present data also provides a feasible explanation why hydrophilic molecular species in these cells are much more abundant in PE as compared with PS, its immediate precursor.

In many mammalian cells in culture most of phosphatidylethanolamine (PE) 1 is synthesized from phosphatidylserine (PS) in a process which involves transfer of PS from the endoplasmic reticulum (ER) to mitochondria, decarboxylation by PS decarboxylase (PSD) located in the inner mitochondrial membrane (1) and transfer of the nascent PE back to ER (2)(3)(4)(5)(6)(7). Transport of PS to mitochondria (i) is ATP-dependent (8), (ii) does not seem to require soluble cytoplasmic components, including lipid transfer proteins (8,9) or the cytoskeleton (8), (iii) takes place only between autologous ER membrane and mitochondria (10), and (iv) involves mainly newly made PS (11,12). Based on these facts, it has been suggested that the transport occurs upon collision of the ER and mitochondrial membranes (8,11). Supporting this, it has been shown that fragments of ER-related membrane (MAM, mitochondria-associated membrane) enriched in PS synthase and other enzymes of lipid biosynthesis cosediment with mitochondria from hepatocytes (13), CHO cells (12), and yeast Saccharomyces cerevisiae (14). The close association of MAM with mitochondria is expected to increase the probability of productive collisions.
To further elucidate the mechanism of PS transfer from ER to mitochondria, we have here compared the rates of decarboxylation of newly synthesized PS molecular species in cultured cells. Intriguingly, the results show that the hydrophilic PS species are decarboxylated much faster than the more hydrophobic ones. Since the rate of PS translocation to mitochondria, rather than decarboxylation itself, is rate-limiting (15), this finding strongly suggests that the hydrophilic PS species are preferentially transported to mitochondria. This, in turn, indicates that the rate-limiting step in the overall transfer process is the desorption of the PS molecule from the ER membrane. Such a mechanism would be in accordance with the previously proposed collision-based model and would explain why the diacyl-PE molecular species in these cells are, on the average, much more hydrophilic than the PS species. dishes or flasks (Nunc, Roskilde, Denmark) in Dulbecco's modified Eagle's medium (41965--039) supplemented with 10% fetal bovine serum, 2 mM glutamine, penicillin (200 units/ml), and streptomycin (200 g/ml) under 5% CO 2 at 37°C. CHO-K1 cells were cultured similarly except that minimal essential ␣-medium (22561-021) was used instead. All cell culture media and supplements were purchased from Life Technologies, Inc. (United Kingdom).

Labeling of PS and PE Molecular Species by [ 3 H]Serine in Intact
Cells-Confluent monolayers of BHK or CHO cells were incubated in a serum-free medium for 12 h, washed twice with PBS, and then labeled for 15 min at 37°C in serine-free minimal essential medium (21090-022) containing 15 Ci of [ 3 H]serine. In some experiments, ␤-chloroalanine (1 mM), an inhibitor of serine palmitoyltransferase (16) was added to the medium 2 h prior to the pulse. After the pulse, the cells were washed twice with PBS and chased at 37°C in serum-free medium containing serine (1 mM) and ethanolamine (1 mM) for up to 22 h. They were then washed twice with PBS, scraped into PBS, pelleted by lowspeed centrifugation and, after addition of bovine brain PS carrier (50 nmol/dish), the lipids were extracted according to Bligh and Dyer (17). The lipid extract was subjected to TLC using chloroform/methanol/ acetic acid/water (25:15:4:1) as solvent. The bands corresponding to PS and PE were scraped off and the lipids were eluted from silica with chloroform, methanol, 0.2 M acetic acid (1:1:0.2).
To analyze the 3 H labeling of PS and PE molecular species, these lipids were converted to trinitrophenyl (TNP)-derivatives essentially as described previously (18). The yield of TNP-PE and TNP-PS was 95 Ϯ 5 and 91 Ϯ 9%, respectively. Control experiments with standards showed that there was, as expected, no differences in the degree of derivatization of individual PE or PS molecular species. The TNPderivatives were purified by TLC using chloroform/acetone/methanol/ acetic acid/water (10:4:2:2:0.5) as solvent and extracted from silica as above. TNP-PS and TNP-PE molecular species were then separated on an Ultrasphere 4.6 ϫ 250-mm ODS (5 m) column (Beckman Inc., CA) using 30 mM choline chloride in methanol, 25  phosphate buffer (pH 7.4) containing 1.5 mM Triton X-100 and 1 mM EDTA was added and the mixture was sonicated for 4 min on a bath sonicator to assure the complete dispersion of the lipid. After addition of crude rat liver PSD (0.9 mg) solubilized (21) in the same buffer, the reaction mixture was incubated in a shaking water bath at 37°C and aliquots were removed and mixed with chloroform/methanol (1:2) at 5, 10, 15, and 20 min of incubation. Lipids were then extracted and [ 3 H]PS and [ 3 H]PE were separated with TLC as described above, except that the TLC plate was first developed with acetone to remove ␤-mercaptoethanol (originating from the enzyme preparation) which interferes with the subsequent derivatization step.
Determination of the Molecular Species Compositions of Cellular PS and PE-Four different methods were used to determine the molecular species composition of cellular PS and PE. First, PS and PE were converted to TNP-derivatives and the molecular species separated by reverse phase HPLC (RP-HPLC) as above and quantitated based on absorbance at 340 nm (18). Second, PE was hydrolyzed to diglycerides by incubation with phospholipase C and the diglycerides formed were converted to benzoyl derivatives. These were separated into diacyl, alkenyl-acyl, and alkyl-acyl subclasses, each of which was then fractionated into molecular species on a RP-HPLC column. All these steps were carried out essentially as detailed in Ref. 22. Third, the isolated PS and PE classes were subjected to electrospray mass spectroscopic analysis on a Micromass Quattro II tandem quadrupole mass spectrometer (Micromass, Altrincham, UK). This method allows both qualitative and quantitative analysis of the individual species (23). Fourth, gas chromatography analysis (24) of the fatty acid methyl esters of the total PE and PS fractions was performed. Fig. 1 shows the reverse phase HPLC separation of PS and PE molecular species from BHK cells as TNP-derivatives. The identity of the species eluting in different FIG. 1. Reverse phase HPLC analysis of PS and PE molecular species from BHK cells. BHK cells were grown to confluence in a complete medium and harvested into PBS with a cell scraper. The lipids were extracted, PS and PE fractions isolated with TLC, converted to TNP-derivatives and separated into molecular species on a C18 reverse phase column as outlined under "Experimental Procedures." The detection was at 340 nm. The retention times were normalized using the 18:0/18:1 peak as the reference. RRT, relative retention time. The major molecular species eluting in each peak are listed in Table I peaks are given in Table I. Obviously, the distribution of mass among PS species is very different from that of PE. In PS (Fig.  1, top panel) the major species are: 18:0/18:1 (peak 15), 18:1/ 18:1 (peak 8), and 16:0/18:1 ϩ 16:1/18:0 (peak 7) which constitute about 43, 20, and 17% of all PS species, respectively. Notably, there are only minor amounts of the more rapidly eluting species (peaks 1-6). In PE (Fig. 1, middle panel) those rapidly eluting species are relatively much more abundant than in PS, while the reverse is true for the slowly eluting ones. (Note that below we will refer to the rapidly eluting species as "hydrophilic" and to the slowly eluting species as "hydrophobic" for convenience. See below for justification of this correlation.)

Molecular Species Profile of PS Is Distinct from That of PE in BHK and CHO Cells-
To study to which extent these differences derive from the fact that PE from BHK cells contains significant amounts of plasmalogens, TNP-PE was subjected to acid treatment to cleave the plasmalogens. As shown in Fig. 1, bottom panel, the profile is, as expected, somewhat different from that of untreated TNP-PE, but the prevalence of the hydrophilic species, as compared with PS ( Fig. 1, top panel), is essentially retained. Analogous results were obtained with CHO cells (data not shown). The identity and relative abundance of PS and PE molecular species were confirmed by electrospray mass spectroscopy (23) and in the case of PE, also by RP-HPLC analysis of diacyl, alkenyl-acyl, and alkyl-acyl classes as benzoyl diglycerides (see "Experimental Procedures").
Since in BHK and CHO cells most, or even all of PE is derived from PS by decarboxylation (2)(3)(4)(5), it is striking that the molecular species profiles of PS, the precursor, is so different from that of the product, i.e. PE. This, intriguingly, suggest that the hydrophilic PS species may be decarboxylated more rapidly than the hydrophobic ones in these cells.  (25). Confluent monolayers were used to minimize recycling of the [ 3 H]ethanolamine moiety among PE species (26). After the incubation, PS and PE were isolated by TLC, converted to TNP-derivatives and separated into molecular species as outlined above.

TABLE I PS and PE molecular species in BHK cells
The molecular species compositions of PS and PE were determined by reverse phase HPLC analysis of the TNP-derivatives as well as with electrospray mass spectroscopy. PS and PE fractions were isolated, converted to TNP-derivatives and analyzed by RP-HPLC as specified under "Experimental Procedures." Electrospray mass spectroscopic analyses gave results consistent with the chromatographic data. The PE composition was confirmed by reverse-phase HPLC analysis of the molecular species as benzoyl diglycerides, PE composition was further studied by RP-HPLC analysis of the diacyl, alkenyl-acyl and alkyl-acyl subclasses of benzoyl derivatives (see "Experimental Procedures").

Peak
Diacyl species Alkenyl-acyl species a top panel, shows the distribution of 3 H radioactivity among PS molecular species immediately after the pulse. The most notable feature is that the more hydrophobic species contain most of the label and thus the distribution of radioactivity is similar to the distribution of mass ( Fig. 1, top panel). The distribution did not change significantly during a 7.5-h chase (Fig. 2, second panel from the top) suggesting that all species have similar turnover times. However, the amount of label in the more hydrophilic species (peaks 1-9) appears to be somewhat reduced as compared with the hydrophobic 18:0/18:1 species (peak 15). Fig. 2, third panel from the top, shows the distribution of the 3 H label among PE species immediately after the pulse. The radioactivity profile is quite similar to the distribution of mass among cellular diacyl PE species (Fig. 1, middle panel) and is therefore very different from that observed for PS (Fig. 2, top panel). The rapidly eluting (hydrophilic) PE species contain relatively much more label than in the case of PS. The distribution of the label among PE species changed somewhat after 7.5 h of chase (Fig. 2, bottom panel). First, it is apparent that the hydrophilic species (1-7) now contain relatively more label than at the end of the pulse. Second, some species (marked with the asterisk) had lost most of the label at this time, indicating that they turn over much more rapidly than the others. However, these latter species are not labeled via decarboxylation of [ 3 H]PS but via an alternative pathway (see below). Analogous results were obtained with CHO cells (data not shown). Based on these findings, it would be tempting to conclude that hydrophilic PS species are preferentially decarboxylated in these cells. However, several alternative explanations have to be excluded before such a conclusion may eventually be drawn.

Some Minor PE Species Are Labeled from [ 3 H]Serine via the Sphingosine-Phosphoethanolamine
Pathway-It is known that [ 3 H]serine radioactivity can incorporate to PE also via an alternative pathway that involves (i) incorporation of serine to sphingosine (ii), cleavage of sphingosine to phosphoethanolamine, (iii) formation of CDP-ethanolamine, and (iv) reaction of the CDP-ethanolamine with diglyceride to form PE (27). To asses the contribution of this pathway to labeling of PE species, the pulse-chase experiment was repeated in the presence of ␤-chloroalanine, an inhibitor of serine palmitoyltransferase (16,28). TLC analysis showed that the fraction of 3 H label in sphingomyelin, which was 16 -19% of total lipid radioactivity in control cells, decreased to zero in the presence of ␤-chloroalanine. This compound thus effectively blocks the metabolism of [ 3 H]serine via the sphingosine-phosphoethanolamine pathway in BHK cells. Fig. 3, top panel, shows the distribution of 3 H radioactivity among PE species in ␤-chloroalanine-treated and control cells. The amount of radioactivity in peaks 12 and 13 (marked with the asterisk) is markedly decreased in ␤-chloroalanine-treated cells. Interestingly, the label in peaks 12 and 13 also exhibited fast turnover in the pulse-chase experiment described above (Fig. 2). As expected, ␤-chloroalanine had no effect on the distribution of 3 H label in PS (Fig. 3, bottom panel). Analogous results were obtained with cycloserine, another inhibitor of serine palmitoyltransferase (data not shown). These data show that some PE species are labeled significantly via the sphingosine-phosphoethanolamine pathway and should thus be excluded when comparing the rates of decarboxylation of different PS molecular species in BHK cells.
A Significant Fraction of the 3 Bligh and Dyer (17). In the case of PS, 5.4 Ϯ 0.5% of the radioactivity was found in the lower chloroform phase. This shows, in accordance with previous data (2), that an insignificant amount of the 3 H label is incorporated to the diglyceride moiety of PS. In contrast, 20.3 Ϯ 2.3% of the 3 H label in PE was found in the lower phase, thus demonstrating a significant labeling of the diglyceride moiety.
To determine how the diglyceride label distributes between the [ 3 H]PE species, the [ 3 H]diglycerides derived from PE were converted to benzoyl derivatives and separated into molecular species by RP-HPLC. As shown in Fig. 4, a major part of the label was restricted to three diacyl diglyceride peaks. Based on comparison with authentic benzoyl diglyceride standards, these peaks were found to correspond to peaks 1-4, 7, and 8, respectively, in Fig. 1, second panel   remodeling rates of different PS and/or PE species, rather than from different rates of decarboxylation. To study this, the distribution of the 3 H label among PS and PE molecular species was determined at 0, 0.5, 1, 2, 7, and 22 h after the pulse. As shown in Fig. 5, top panel, there is a loss of radioactivity from all PS peaks during the chase. Interestingly, however, the loss appears to be more rapid from the peaks (1-6 in particular) containing the more hydrophilic species. Correspondingly, the hydrophilic PE species appear to be labeled more rapidly than the hydrophobic ones. Peaks 12 and 13 deviate from this general pattern, but this can be fully accounted by the fact that most of the label in these peaks is not derived via decarboxylation of [ 3 H]PS but via another pathway (see above). When the radioactivity in the corresponding PS and PE species are summed (Fig. 5, bottom panel), the total 3 H radioactivity is found to be essentially constant over the 22-h chase for all species, except for the sphingosine-derived species 12 and 13. The stability of the 3 H label in PS ϩ PE is in accordance with previous studies with BHK and CHO cells (2,3,12). These data strongly suggest that the marked differences in the 3 H labeling profiles of PS and PE molecular species (Fig. 2) are not due to differences in remodeling/degradation rates of PS or PE species, but indeed reflect a more rapid decarboxylation of the hydrophilic PS species. Note that the distribution of 3 H radioactivity in Fig. 5, bottom panel, indicates the relative proportions of different PS molecular species synthesized in BHK cells. Fig. 6 shows the decay of radioactivity for different [ 3 H]PS peaks, excluding those in which a major fraction of the radioactivity was not derived via [ 3 H]PS decarboxylation. Note that some peaks that are minor or not well separated have been grouped together. The plots indicate that in each case there are apparently two decay components, a fast and a slow one. However, the fractional contribution of these components (indicated in figure panels) varies significantly. For instance, the fast component accounts for 20, 12, and 2.5% of the total decay in peak groups 1-4, 7-9, and 14 -15, respectively. Also the halftime of the slow component varies significantly being approximately 10, 18, and 32 h, respectively. The half-time of the fast component could not be determined for all species due to the limited number of data points in the relevant region. We conclude that there are two, kinetically distinct pools of PS molecules in BHK cells: one which consists predominantly of the more hydrophilic PS species and is characterized by a rapid conversion to PE, and another, consisting mostly of the more hydrophobic species and characterized by a much slower conversion to PE. This conclusion is consistent with previous studies suggesting, in a more qualitative manner, the presence of two pools of PS with different decarboxylation kinetics in cultured cells (3,11,12,31).

PS Decarboxylase Does Not Discriminate between [ 3 H]PS Molecular Species-
To exclude the possibility that the more efficient decarboxylation of the hydrophilic PS species simply reflects selectivity of the PSD enzyme, rather than a more rapid transfer to mitochondria, the total [ 3 H]PS fraction was isolated from BHK cells incorporated in Triton X-100 micelles and incubated with solubilized PSD from rat liver 2 for up to 20 2 Rat liver enzyme was used, because the low specific activity of the [ 3 H]PS available required more PSD activity than what was feasible to obtain from BHK or CHO cells. It is, however, very likely that PSD from rat behaves similarly to PSD from hamster, a closely related species.

FIG. 4. Reverse phase HPLC analysis of [ 3 H]diglycerides derived from [ 3 H]PE. BHK cells were labeled 15 min with [ 3 H]serine
and chased for 7 h. [ 3 H]PE was then isolated and subjected to phospholipase C treatment, the diglycerides formed converted to benzoyl derivatives and separated on a reverse phase column (see "Experimental Procedures"). The major diacyl, alkenyl-acyl (PL), and alkyl-acyl (ALK) species are indicated (cf. Table I). The identity of the individual species were established by first separating the diacyl, alkenyl-acyl, and alkylacyl subclasses by normal phase HPLC followed by reverse phase HPLC analysis of each subclass (22). A very similar profile was obtained at 1 h of chase (not shown). The experiment was reproduced twice with similar results. RRT, relative retention time. profile is now very similar to that of [ 3 H]PS, thus strongly suggesting that PSD does not act on some PS species preferentially. There is somewhat more radioactivity associated with the most hydrophilic PE species as compared with PS. This could be due to that these hydrophilic [ 3 H]PS species, because of higher solubility to the aqueous phase, equilibrate more rapidly between micelles containing PSD and those devoid of it, and are thus more rapidly decarboxylated. Such hydrophobicitydependent transfer of phospholipids between micelles has been demonstrated (32). Also supporting the absence of PSD selectivity, the distribution of radioactivity among PS and PE species remained essentially constant at all times when the decarboxylation progressed from 0 to 50% (not shown). Nevertheless, the differences between [ 3 H]PS and [ 3 H]PE profiles seen in Fig.  7 are minor as compared with what was observed with intact cells (Fig. 2). DISCUSSION The present study shows that hydrophilic molecular species are much more abundant in PE as compared with PS, its immediate precursor, in BHK and CHO cells. Pulse-chase experiments with [ 3 H]serine revealed that the hydrophilic PE species contained proportionally much more label than those of PS. These findings strongly suggested that the hydrophilic PS species are decarboxylated considerably faster than the hydrophobic ones in intact cells. This could be confirmed by excluding alternative explanations, including: (i) labeling of particular PE species via the sphingosine-phosphoethanolamine pathway, (ii) selective incorporation of radioactive [ 3 H]serine metabolites to the diglyceride moiety of some PE species, (iii) differential remodeling/degradation of PS/PE species, or (iv) selectivity of the decarboxylase enzyme itself. In addition, the possibility that [ 3 H]ethanolamine released from [ 3 H]PE would have reincorporated via the CDP-ethanolamine pathway to PE species seems very unlikely because the chase medium contained a large excess of unlabeled ethanolamine and second, an insignificant amount of radioactivity was associated with the plasmalogen PE species (cf., Figs. 1-3) which are synthesized via the CDP-ethanolamine pathway (33).
Why would the rate of decarboxylation of a newly synthesized PS molecule in the cell depend on its hydrophobicity? There is strong evidence that the rate of transport to mitochondria, rather than decarboxylation itself, determines the overall rate of PS decarboxylation in cells (15). Therefore, we suggest that transport of PS molecular species from the site of their synthesis, i.e. ER, to mitochondria is dependent on their hydrophobicity. Such hydrophobicity-dependent translocation can be explained by assuming that the rate-limiting step in the transfer process is the formation of "activated" or "transitionstate" monomers, i.e. lipid molecules perpendicularly displaced from a membrane (32,34). The probability of such a displacement decreases strongly with increasing lipid hydrophobicity, which in turn depends mainly on the length and degree of saturation of the acyl chains (35). The significance of this effect is demonstrated by noting that addition of 2 methylene units to an acyl chain, or removal of one double bond from it, can decrease the rate of interparticle transfer by a factor of 5 to 10 (36 -38). This results in a predictable, systematic correlation between the rate of transfer and lipid hydrophobicity, which can be determined from the retention time on a reverse phase column (39). To see if such dependence is observed also for the present case, the half-time of decarboxylation versus the apparent hydrophobicity of PS molecular species (peaks) from BHK cells was plotted. As shown by Fig. 8, the plot is essentially linear. Analogous results were obtained with CHO cells (not shown). Since it is assumed that the rate of decarboxylation is directly proportional to the rate of PS transport to mitochondria (see above), these findings strongly suggest that activated lipid monomers are indeed involved in the translocation of PS molecules from ER to mitochondria.
This result may seem unexpected in the view that transport of natural phospholipids by such spontaneous efflux mechanism is generally considered as a very slow process (half-time tens of hours). However, it should be noted that formation of activated monomers, and hence the rate of translocation, can increase remarkably at higher membrane concentrations. Thus, the half-time for the transfer of palmitoyloleoyl-phosphatidylcholine between lipid vesicles decreased from 46 to 7 h when the acceptor vesicle concentration was increased from 2.5 to 40 mM (34). The latter half-time is similar to that for the transport of newly synthesized PS from ER to mitochondria (40, 41, this study). The enhancement of transfer at higher membrane concentrations has been attributed to stabilization of the transition state monomers by their interaction with the acceptor membrane (34). In this respect, it is noteworthy that according to morphological (42,43) and biochemical studies (13,44) the ER (MAM) and the outer mitochondrial membranes are closely associated. This means that the local concentration of the donor and acceptor membranes is very high, thus favoring collision-enhanced intermembrane lipid transfer. It should also be noted that the ER (MAM) and the mitochondrial outer membrane are relatively rich in PE (13,45), and this lipid greatly enhances lipid translocation via the collisional mechanism (46). 3 The involvement of membrane collisions in PS translocation from ER to mitochondria has been proposed previously (8,11).
An alternative model for collisional lipid transfer suggests that a transient fusion or hemifusion of the two membranes occurs upon collision of membranes, thus allowing the lipid mixing by lateral diffusion (47). Since lateral diffusion is essentially independent of lipid hydrophobicity (48), this model is clearly not compatible with the present data showing marked dependence of the translocation rate on lipid hydrophobicity (Fig. 8). The absence of cardiolipin, a major component of the outer mitochondrial membrane (45), from the ER is also incompatible with this model. In contrast, the absence of cardiolipin from ER is fully consistent with the transition state monomer model, since desorption of this lipid with four acyl chains from a membrane is predicted to be an extremely slow process (38).
Although the present data are in good accordance with spontaneous, collision-enhanced transfer of PS from ER to mitochondria, they by no means preclude the involvement of proteins. In this respect, it is of interest that the rate of transport mediated by the nonspecific lipid transfer protein (ns-LTP) is negatively correlated with the hydrophobicity of the lipid substrate (49,50). Ns-LTP has been found to be associated with isolated mitochondria (51) and detected immunologically in mitochondria of intact cells (52)(53)(54). This protein also enhances transfer of PS in vitro (15,55,56) and is capable of net transfer of lipids (57). On the other hand, several studies have indicated that ns-LTP may not be involved in transport of PS to mitochondria (9,11,15,58). Perhaps the strongest evidence for this is that decarboxylation of newly synthesized [ 3 H]PS, and thus apparently its transport to mitochondria, was not affected in a CHO cell mutant lacking ns-LTP (9). However, since individual PS or PE molecular species were not separated in that study, the possibility remained that the mutant compensates for the loss of ns-LTP by synthesizing more of the hydrophilic PS species which would translocate more readily via the nonmediated process (see above). Our preliminary compositional and [ 3 H]serine labeling studies with the same CHO mutant provide evidence that such compensatory changes in PS molecular species distribution synthesis may indeed take place. Although it is not clear if these changes are related to PS transport from ER to mitochondria, further studies seem to be required to decide whether or not ns-LTP plays a role in PS transport to mitochondria.
We cannot presently exclude the possibility that the disparate decarboxylation efficiences of different PS species is due to the presence of two or more distinct compartments of newly synthesized PS molecules in the cell. Thus, there could be a compartment close to mitochondria (e.g. MAM) where mainly hydrophilic PS species would be synthesized, and another more distant compartment where the more hydrophobic species would be produced. If so, preferential decarboxylation of the hydrophilic species could simply reflect their more favorable location regarding transport to mitochondria. We consider, however, this model improbable for two reasons. First, PS synthesis appears to be largely confined to the ER in rat liver (41,59). Hence, those putative compartments should represent different regions of ER, e.g. MAM and the bulk ER. However, as the ER membrane is obviously continuous and the lateral diffusion of lipids very fast, it is difficult to see why rapid mixing of the PS species synthesized in those two regions would not take place. Such mixing is expected to be complete within a few minutes. Considering that the half-time for the transport of newly synthesized PS from ER to mitochondria is several hours (40,41, this study), this two-compartment model could not explain the observed differences in the rate of decarboxylation between PS molecular species. Second, the data given in Fig. 8 would be compatible with this model only if one assumed that equilibration of PS species is strictly dependent on their hydrophobicity. This does not seem feasible considering that lateral diffusion of lipids is only very weakly dependent on their structure (48).
Previous studies have shown that newly synthesized PS molecules are decarboxylated much more efficiently than the pre-existing ones, thus suggesting the presence of a least two metabolically distinct pools of PS (11,12). This is strongly supported by the present study showing that the kinetics of decarboxylation of PS species has two components: a fast one with a half-time of about 20 min and a slow one with a halftime of 10 -32 h (Fig. 6). A feasible explanation for the existence of these kinetic components would be that efficient translocation of PS molecules from ER (or MAM) to mitochondria can occur only for a relatively short period of time after their synthesis. Based on the data shown in Fig. 6 we suggest that the width of this time window is 10 -20 min. This time is probably determined by the rate of PS removal from the ER membrane in transport vesicles which are constantly budding from the ER membrane. The half-time of transport of phospholipids from the ER is approximately 10 min (60). For the total PS pool, the half-time of the fast and slow components were determined to be 22 min and 28 h, respectively, and approximately 12% of the total pool of the newly synthesized PS molecules was estimated to be transported to mitochondria during the fast process in BHK cells. The latter value is in accordance with the conclusion that only a small fraction of the total PS pool is turning over rapidly in hetapatocytes (31). Eventually, however, most or even all of PS synthesized may be converted to PE as indicated by the fact that the combined PS ϩ PE radioactivity remains essentially constant for 22 h after the pulse (Fig. 5, bottom panel; Refs. 2 and 3) despite that the [ 3 H]PS radioactivity has decreased more than 50% at this time (Fig. 5, top panel; Refs. 2 and 3).
A major fraction of the PS removed from the ER in transport vesicles will probably end up to the plasma membrane. A recent investigation indicated that almost 70% of the total PS pool in BHK cells is located in the plasma membrane, and about 35% of the inner leaflet phospholipid is PS (61). This remarkable enrichment to the plasma membrane indicates that PS cannot move readily from this site to mitochondria. Supporting this, the movement of PS from the plasma membrane to mitochondria is an energy-independent process (62), suggesting that diffusion through the cytoplasm is the mechanism of transport. We propose that the slow component of decarboxylation observed here (Fig. 6) represent translocation of plasma membrane PS to the mitochondria by such spontaneous diffusion. The apparent increase in half-time of this slower process with increasing PS hydrophobicity supports this mechanism. In addition, we have recently observed that transport of pyrenelabeled PS species from the plasma membrane to mitochondria in BHK cells decreases logarithmically with increasing species hydrophobicity so that for the most hydrophobic ones hardly any transfer was detectable. 4 It is tempting to speculate that the high average hydrophobicity of PS species is physiologically relevant. It may allow the cell to maintain a high concentration of this lipid in the plasma membrane where it could be needed for such crucial processes as exo-and endocytosis (63).
It has been amply demonstrated that transport of PS from the ER to mitochondria is an ATP-dependent process (8,10,15,58). ATP is probably required to sequester Ca 2ϩ to a compartment in the ER, apparently to support PS synthesis which requires relatively high levels of Ca 2ϩ (64). However, it appears that ATP is required also for the translocation process, possibly to bring the MAM and the outer mitochondrial membrane more closely together (58) thereby enhancing the probability of collision and thus lipid translocation.
There is good evidence that a major fraction of PE derived from PS via decarboxylation in mitochondria is transported rapidly back to ER (11,40). The mechanism of this transfer is not known. However, if PS is transferred between ER and mitochondrial membranes via the activated monomer-collision process as suggested above, it is logical to assume that the same mechanism is employed for translocation of PE in the opposite direction. The prevalence of hydrophilic species in PE derived from PS in mitochondria (see above) should obviously be advantageous for this process to occur efficiently. Supporting a common transport mechanism for PS and PE, movement of newly made PE from mitochondria to ER is also ATP-dependent as indicated by studies with yeast (65).
In conclusion, the present data strongly suggest that hydrophilic PS species are more rapidly decarboxylated, and thus also transported to mitochondria after the synthesis in ER than the hydrophobic ones. This is best explained by assuming that formation of so called activated lipid monomers, enhanced by intermembrane collisions, is the rate-limiting step in the transfer process. This mechanism explains why hydrophobic species are much more prominent in PE as compared with PS, its immediate precursor. Preferential conversion of the hydrophilic PS species to PE also means that the average hydrophobicity of the remaining PS species will be significantly higher than that of those originally synthesized. This could be important for the maintenance of high PS concentrations in the inner leaflet of the plasma membrane.