Defining the importance of phosphatidylserine synthase 2 in mice.

Phosphatidylserine synthase 1 (Pss1) and phosphatidylserine synthase 2 (Pss2) produce phosphatidylserine by exchanging serine for the head groups of other phospholipids. Pss1 and Pss2 are structurally similar (approximately 32% amino acid identity) but differ in their substrate specificities, with Pss1 using phosphatidylcholine for the serine exchange reaction and Pss2 using phosphatidylethanolamine. Whether Pss1 and Pss2 are both required for mammalian growth and development is not known, and no data exist on the relative contributions of the two enzymes to serine exchange activities in different tissues. To address those issues and also to define the cell type-specific expression of Pss2, we generated Pss2-deficient mice in which a beta-galactosidase marker is expressed from Pss2 regulatory sequences. Histologic studies of Pss2-deficient mice revealed very high levels of beta-galactosidase expression in Sertoli cells of the testis and high levels of expression in brown fat, neurons, and myometrium. The ability of testis extracts from Pss2-deficient mice to catalyze serine exchange was reduced by more than 95%; reductions of approximately 90% were noted in the brain and liver. However, we found no perturbations in the phospholipid content of any of these tissues. As judged by Northern blots, the expression of Pss1 was not up-regulated in Pss2-deficient cells and tissues. Testis weight was reduced in Pss2-deficient mice, and some of the male mice were infertile. We conclude that Pss2 is responsible for the majority of serine exchange activity in in vitro assays, but a deficiency in this enzyme does not cause perturbations in phospholipid content or severe developmental abnormalities.

Phosphatidylserine is an aminophospholipid that constitutes 5-10% of mammalian membrane phospholipids (1). In mammals, phosphatidylserine is synthesized by a pair of enzymes, phosphatidylserine synthase 1 (Pss1) 1,2 and phosphatidylserine synthase 2 (Pss2) (2)(3)(4), located primarily within the mitochondria-associated membrane fraction of the endoplasmic reticulum (5,6). The two enzymes are structurally related, with 32% amino acid identity, and both are predicted to contain several transmembrane domains (2,3,6,7). Pss1 and Pss2 generate phosphatidylserine by catalyzing the exchange of serine for the head group of another phospholipid, but the two enzymes differ in their substrate specificities. Pss1 uses phosphatidylcholine for the exchange reaction (8,9), whereas Pss2 uses phosphatidylethanolamine (2-4, 6, 10). In vitro, Pss1 is capable of catalyzing the exchange of ethanolamine and choline in addition to serine; Pss2 is capable of catalyzing the exchange of ethanolamine but not choline. However, these ethanolamine and choline exchange reactions are not thought to be physiologically important for the in vivo synthesis of phosphatidylethanolamine (11) or phosphatidylcholine (12).
The physiologic "rationale" for the existence of two different phosphatidylserine synthases is unclear. No one knows whether mammalian growth and development require both enzymes, since no one has yet developed mice lacking either of the two genes. However, two groups have produced Chinese hamster ovary (CHO) cell lines lacking Pss1 by selecting for cells that required ethanolamine or phosphatidylserine for growth (8,9). Extracts from the mutant CHO cells manifested a ϳ50% decrease in serine exchange activity, suggesting that Pss2 accounts for a significant portion of the serine exchange activity in that cell type. Pss1 deficiency did not have a significant impact on cell growth when the cells were grown in the presence of ethanolamine or phospholipids, but the cells grew slowly, and their phospholipid content was perturbed when the cells were grown in the absence of ethanolamine, phosphatidylserine, or phosphatidylethanolamine (7)(8)(9)13).
Thus far, no one has developed cell lines that lack Pss2 expression, although a mutant CHO cell line that expressed reduced levels of Pss2 activity was generated (10). It is difficult to predict whether or not Pss2-deficient cells would be viable and healthy. On the one hand, one could argue that the exist-ence of Pss1 would make Pss2 expression superfluous. On the other hand, a large fraction of phosphatidylserine synthesis in CHO cells is due to Pss2, and it certainly would not be unreasonable to surmise that Pss2 would be crucial for cellular phospholipid homeostasis. It is also difficult to make a priori predictions about whether mice lacking Pss2 would be viable and, if so, whether there would be any pathology in tissues expressing high levels of the enzyme. Pss2 is expressed in a variety of different organs as judged by Northern blot analysis (6,7), but no information exists on which cell types express high levels of the gene. Finally, there are no biochemical data on the contribution of Pss2 to total serine exchange activity in different mammalian tissues.
The purpose of this study was to define the physiologic importance of Pss2 in mammals. To address this issue, we produced Pss2-deficient mice in which the expression of a marker gene, ␤-galactosidase (␤-gal), was driven by the regulatory elements of the Pss2 gene. The characterization of the Pss2deficient mice allowed us to fill in a number of gaps in our knowledge of Pss2. First, we demonstrated that mice lacking Pss2 survive development and are viable, although males have small testes and occasionally testicular atrophy. We were able to define the impact of Pss2 deficiency on serine exchange activities in different tissues and were able to demonstrate, by ␤-gal staining, which cell types express high levels of Pss2. Finally, we investigated whether or not Pss2 deficiency perturbed the phospholipid composition of tissues and cells.

EXPERIMENTAL PROCEDURES
Generation of Pss2-deficient Mice-A mouse embryonic stem cell line (KST314, strain 129P2/OlaHsd) containing an insertional mutation in Pss2 was identified in a gene-trapping screen (14). The gene-trapping vector, pGTITMpfs, was designed to interrupt genes that encode proteins with an N-terminal signal sequence and to create an in-frame fusion with the ␤-geo reporter gene (14). The embryonic stem cell line was used to generate male chimeric mice, which were bred with C57BL/6 mice to establish heterozygous (Pss2ϩ/Ϫ) and homozygous (Pss2Ϫ/Ϫ) knockout mice. Mice were genotyped by quantifying neomycin phosphotransferase II (neo) gene dosage in genomic DNA with a quantitative PCR assay (described on the BayGenomics Web site at baygenomics.ucsf.edu/protocols). Genotyping was also performed by quantifying neo gene dosage with Southern blots; for these studies, BamHI-digested genomic DNA was hybridized with pGTITMpfs that had been linearized with HindIII. All mice described here had a mixed genetic background (ϳ50% C57BL/6 and ϳ50% 129/OlaHsd). The mice were weaned at 21 days of age, housed in a barrier facility with 12-h light/dark cycle, and fed a chow diet containing 4.5% fat (Ralston Purina, St. Louis, MO).
Northern Blot Analysis-Sites of Pss2 and Pss1 expression were determined with mouse multiple-tissue poly(A) ϩ RNA blots (CLON-TECH, Palo Alto, CA). The protein-coding sequence of the Pss1 cDNA was amplified from a mouse liver cDNA library (CLONTECH) with oligonucleotides 5Ј-ATGGCGTCCTGCGTGGGGAGCAGG-3Ј and 5Ј-CAGCCGATGAAGAGGATTCTACACC-3Ј and cloned into pCRII (Invitrogen). A 1.1-kb Pss1 cDNA probe was produced by removing the insert with EcoRI. A 1.4-kb Pss2 cDNA was cloned into pCRII (7); the insert was released by EcoRI digestion. Probes were labeled with 32 P by random hexamer priming. The Northern blots were exposed to x-ray film for 12 h at Ϫ80°C.
Northern blots were also produced with total RNA isolated from brain, liver, and testis from Pss2ϩ/ϩ and Pss2Ϫ/Ϫ mice. RNA was isolated with the Tri Reagent RNA isolation kit (T9424; Sigma). Total RNA (20 g) was separated by electrophoresis on a 1% agarose/formaldehyde gel and then transferred to a Nytran SuPerCharge membrane (S00486; Schleicher & Schuell) and hybridized with the 1.4-kb Pss2 cDNA probe. The same blots were also probed with the 1.1-kb Pss1 cDNA probe and a neo probe (produced by cleaving pKSloxPNT with HindIII) (16).
Assays of Phosphatidylserine Synthase Activity-Tissue samples (ϳ100 mg) were homogenized with a Polytron homogenizer in a buffer containing 50 mM Tris-HCl (pH 7.4), 5 mM MgCl 2 , 1.0 mM EDTA, and 100 mM NaCl. The tissue homogenate was centrifuged for 5 min at 600 ϫ g, and the supernatant fluid was used for measurement of serine exchange activity. The reaction mixture (200 l) contained 160 l of tissue sample with 300 g of protein, 20 l of a buffer (100 mM CaCl 2 , 40 mM hydroxylamine, and 250 mM HEPES, pH 7.4), and 20 l of the radiolabel (a [ 14  . The reaction was allowed to proceed for 20 min at 37°C and was terminated by adding 5 ml of chloroform/methanol (2:1, v/v). A total of 1.5 ml of water was then added to each tube, and the tubes were centrifuged at 1,000 ϫ g for 5 min. The upper phase was aspirated and discarded. The lower phase was washed three times with 2.0 ml of methanol/water (1:1). The phospholipid products were extracted by the method of Bligh and Dyer (17), and radioactivity was measured.
Metabolic Labeling of Phosphatidylserine in Embryonic Fibroblasts-Immortalized fibroblasts from Pss2ϩ/ϩ and Pss2Ϫ/Ϫ embryos were grown to 80 -90% confluence in 60-mm dishes in the presence of 15% fetal bovine serum. The cells were then incubated for up to 6 h with 3 Ci/ml [3-3 H]serine. Lipids were extracted by treatment of the cells with isopropyl alcohol/hexane (2:3, v/v) and separated by thin layer chromatography in the solvent system chloroform/methanol/acetic acid/ formic acid/water (70:30:12:4:2, v/v/v/v/v). Phospholipids were visualized by exposure to iodine vapor and phosphatidylserine identified by comparison to authentic phosphatidylserine. Radioactivity in phosphatidylserine was determined by scintillation counting.
Examination of Tissues for ␤-Gal Activity-Cryostat sections (10 m thick) of frozen tissues from mice were collected on Superfrost Plus microscope slides (Fisher). The sections were dried, fixed (2% formaldehyde in phosphate-buffered saline (PBS) supplemented with 0.2% glutaraldehyde, pH 7.2, for 5 min at room temperature), and rinsed with PBS. The sections were then incubated with an X-gal (Invitrogen) staining solution for 2-6 h at 37°C. The X-gal staining solution was prepared by mixing 1 ml of a stock solution of X-gal (40 mg/ml in dimethyl sulfoxide) with 40 ml of PBS containing 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide trihydrate, and 2 mM MgCl 2 . After staining, the slides were rinsed with PBS, washed twice for 2 min with deionized water, and counterstained for 1-2 min with Nuclear Fast Red. The slides were washed again with PBS and deionized water and coverslipped with Gelmount (Biomeda, Foster City, CA).
Phospholipid Composition of Mouse Tissues and Cultured Cells-Lipids were extracted from tissue homogenates or fibroblasts by the method of Bligh and Dyer (17). The lipid-containing lower fraction was removed and washed three times with methanol/water (1:1, v/v). The lower phase was dried under a flow of nitrogen, and lipids were dissolved in chloroform containing 0.05% (w/v) butylated hydroxytoluene. Lipid composition was determined with a high performance liquid chromatography system (System Gold; Beckman) with a 4.8 ϫ 100-mm silica column (Allsphere Silica; 3 m; Alltech, Nicholasville, KY) coupled to an evaporative light scattering detector (ELSD 2000; Alltech). The mobile phase consisted of isopropyl alcohol/n-hexane/water run as a ternary gradient (58:40:2 to 52:40:8, v/v/v) in 10 min, followed by a stationary phase (52:40:8, v/v/v) for 35 min. Nouveau software (Beckman) was used for instrument control, data acquisition, and data analysis. Calibration curves were obtained for the different phospholipids by injection of known quantities of lipid standards.
Lipid Composition of Microsomes-Livers and testes were removed from Pss2ϩ/ϩ and Pss2Ϫ/Ϫ mice and homogenized in a motor-driven Dounce homogenizer in 4 volumes (w/v) of buffer containing 250 mM mannitol, 5 mM HEPES (pH 7.4), 0.5 mM EGTA, and 0.1% (w/v) bovine serum albumin. Unbroken cells and nuclei were pelleted by centrifugation (600 ϫ g for 10 min). Mitochondria were then pelleted with a second centrifugation (10,300 ϫ g for 10 min). The supernatant fluid from the second centrifugation was subjected to ultracentrifugation for 1 h at 100,000 ϫ g; the resultant pellet was designated as microsomes. Lipids were extracted from the microsomes (17) and separated by thin layer chromatography in the solvent system chloroform/methanol/acetic acid/ formic acid/water (70:30:12:4:2, v/v/v/v/v). Phospholipids were identified by exposure to iodine vapor and comparison with standards. Bands corresponding to phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, and phosphatidylinositol were scraped from the plate, and the amount of each phospholipid (nmol/mg of protein) was determined by measurement of lipid phosphorus (19).

Generation of Pss2
Knockout Mice-Pss1 and Pss2 are both expressed in a broad and overlapping group of organs, as judged by multiple-tissue Northern blots (Fig. 1, A and B). Pss2 is expressed at particularly high levels in the testis, but Pss1 is also expressed there. Pss1 is highly expressed in liver and heart, but Pss2 is also expressed in those tissues.
To determine whether normal expression levels of both of the phosphatidylserine synthases are required for mammalian development and for viability of adult mice, we generated Pss2 knockout mice with an embryonic stem cell line containing an insertional mutation in intron 2 of Pss2. The mutation results in the production of a fusion transcript containing Pss2 sequences encoding the first 72 amino acids of the enzyme (including the first transmembrane segment) spliced to the CD4 transmembrane domain and ␤-geo sequences from the gene trap vector (a fusion between Escherichia coli lacZ (encoding ␤-gal) and neo) (20,21). The production of the Pss2-␤-geo fusion transcript is under the control of the Pss2 regulatory sequences.
Pss2Ϫ/Ϫ mice develop normally, appear healthy, and grow at the same rate as littermate control mice. The plasma cholesterol, triglyceride, and phospholipid levels were no different than in wild-type littermates (data not shown). Female Pss2Ϫ/Ϫ mice exhibited normal fertility, as did the majority of the males, but ϳ10% of the male Pss2Ϫ/Ϫ mice (n ϭ 33 examined) were infertile or subfertile. The absence of a severe phenotype was not due to "leakiness" of the insertional mutation (e.g. production of a normal Pss2 transcript from the mutant allele). No normal Pss2 transcript (2.4 kb) was detectable in knockout mice (Fig. 1C). As expected, a ϳ10-kb Pss2-␤-geo fusion transcript could be observed on a Northern blot hybridized with a neo probe (Fig. 1D). The knockout of Pss2 did not result in increased Pss1 mRNA expression, either in fibroblasts cultured from Pss2Ϫ/Ϫ embryos (Fig. 1E) or in the testis or brain of adult Pss2Ϫ/Ϫ mice (Fig. 1F).
Phosphatidylserine Synthase Activity Is Reduced in Pss2deficient Mice-To explore the contribution of Pss2 to total phosphatidylserine synthase activity in mammalian tissues, we measured the ability of tissue extracts to catalyze the exchange of radiolabeled serine, radiolabeled ethanolamine, and radiolabeled choline into phospholipids. Serine exchange activity was ϳ95% lower in extracts from Pss2Ϫ/Ϫ testes than in extracts from Pss2ϩ/ϩ testes ( Fig. 2A). Serine exchange was also reduced substantially in the brain, heart, and liver of Pss2Ϫ/Ϫ mice, although to a lesser degree than in the testes ( Fig. 2A). The ability of tissue extracts from Pss2Ϫ/Ϫ mice to exchange ethanolamine was also markedly reduced (Fig. 2B). Interestingly, choline exchange, which is indicative of Pss1 activity, was increased by about 2-fold in Pss2Ϫ/Ϫ tissues Phospholipid Content of Membranes from Pss2ϩ/ϩ and Pss2Ϫ/Ϫ Tissues-Pss1 deficiency in CHO cells reduced phosphatidylserine synthase activity, and phospholipid composition of the mutant cells was perturbed when the cells were grown in ethanolamine-or phospholipid-depleted medium (6,8,9,13). To determine whether Pss2 deficiency affected tissue phospholipid composition, we measured the phospholipid content of the liver, brain, and testis of Pss2Ϫ/Ϫ mice and littermate Pss2ϩ/ϩ controls. No significant differences were observed (Fig. 3). We also measured the phospholipid content of microsomes from the liver and testis of Pss2Ϫ/Ϫ and Pss2ϩ/ϩ mice. Again, no significant differences were observed (Fig. 4).
We considered the possibility that the normal phospholipid composition of Pss2Ϫ/Ϫ tissues might have been due to an equilibration of phospholipids at the whole-animal level, perhaps as a result of the delivery of phospholipids to tissues by plasma lipoproteins. To test that possibility, we examined serine exchange activity and phospholipid content of Pss2ϩ/ϩ and Pss2Ϫ/Ϫ fibroblasts grown for 72 h in lipoprotein-deficient serum. Serine exchange in Pss2Ϫ/Ϫ fibroblasts was reduced by ϳ90% (Fig. 5A); however, no significant perturbation in cellular phospholipid composition was noted (Fig. 5B). To test whether the reduced serine exchange activity in extracts from Pss2Ϫ/Ϫ fibroblasts corresponded to a decrease in phosphatidylserine synthesis in intact cells, we determined the incorporation of radiolabeled serine into phosphatidylserine. Phosphatidylserine synthesis was reduced in Pss2Ϫ/Ϫ fibroblasts (Fig. 5C).
Cell Type-specific Expression of Pss2-The cell types responsible for high levels of Pss2 expression are unknown. To address this issue, we examined ␤-gal expression (reflecting sites of Pss2 expression) in 20-day mouse Pss2Ϫ/Ϫ embryos and in adult Pss2Ϫ/Ϫ mice. In embryos, the highest levels of ␤-gal expression were in brown adipose tissue over the dorsal thoracic cage (Fig. 6A, arrowheads) and in peripheral nerves (Fig.  6A, arrows). As expected, the brown adipose tissue stained intensely with Oil Red O (Fig. 6B). In adult mice, ␤-gal staining was intense in the brain, uterus, and testis. In the brain, ␤-gal expression was confined to neurons and was particularly prominent in the Purkinje cells of the cerebellum (Fig. 6, C and D) and in pyramidal neurons in the CA1 region of the hippocampus (Fig. 6E). High levels of ␤-gal expression were also observed in the myometrium of the uterus (Fig. 6F).
In testes of 4 -6-month-old Pss2Ϫ/Ϫ mice (n ϭ 25), we observed two histologic patterns. In the majority of the Pss2Ϫ/Ϫ mice (n ϭ 22), testis histology appeared normal by light microscopy, with small numbers of Leydig cells in the interstitium, Sertoli cells surrounding the spermatic ducts, and abundant spermatogonia and maturing spermatocytes (Fig. 7A). In the remaining three Pss2Ϫ/Ϫ mice, the testes were severely atrophic (ϳ30% of normal size), and histologic examination revealed hyperplasia of Leydig cells and contracted spermatic ducts with a thick layer of Sertoli cells and no spermatocytes (Fig. 7B). We never observed atrophic testes in Pss2ϩ/Ϫ (n ϭ 9) or Pss2ϩ/ϩ littermate controls (n ϭ 27). Regardless of whether the histology of Pss2Ϫ/Ϫ mice was normal or abnormal, ␤-gal staining of Sertoli cells was intense (Fig. 7, A and B). ␤-Gal staining was detectable in Sertoli cell projections adjacent to spermatocytes, but no staining of spermatogonia or spermatocytes was observed. Higher power images revealed punctate ␤-gal staining within Leydig cells, but the staining intensity was invariably lower than in Sertoli cells.
The finding of testis atrophy in 3 of 25 Pss2Ϫ/Ϫ mice but in none of the controls suggested that Pss2 deficiency might cause overt testis pathology but with incomplete penetrance. If this were the case, we hypothesized that a more thorough examination of male Pss2Ϫ/Ϫ mice might uncover subtle abnormalities in testis size or function, even in animals with normal testis histology. To test this possibility, we compared testis weights in male Pss2Ϫ/Ϫ mice (n ϭ 30, all with normal testis histology) and in littermate Pss2ϩ/ϩ mice (n ϭ 27). These studies revealed a small (ϳ13%) but highly significant (p Ͻ 0.01) reduction in testis weight in Pss2Ϫ/Ϫ mice (Fig. 8A).
Sertoli cells normally produce inhibin B, which negatively regulates serum FSH levels (22). Thus, Sertoli cell dysfunction leads to low serum inhibin B levels and high serum FSH levels (22). We suspected that subtle Sertoli cell dysfunction in Pss2Ϫ/Ϫ mice might be accompanied by increased serum FSH levels. Indeed, this was the case; serum FSH levels in Pss2Ϫ/Ϫ mice were significantly higher than in littermate controls (p Ͻ 0.01) (Fig. 8B). DISCUSSION Over the past 10 years, a series of biochemical and genetic studies have established that mammals have two different enzymes for generating phosphatidylserine, Pss1 and Pss2 (2)(3)(4)7). These two enzymes have significant sequence similarity but differ in their substrate specificities. In this study, we generated Pss2-deficient mice and used those mice and derivative cell lines to address a number of mysteries surrounding the mammalian phosphatidylserine-synthesizing enzymes. Our studies have added important new information to the field. First, it seems clear that both enzymes are not required for development. The observations that homozygous Pss2-deficient mice are viable, grow normally, and maintain normal tissue phospholipid compositions indicate that Pss2 is largely dispensable for development, at least under laboratory conditions. Second, we defined, by ␤-gal staining, the cell types that express Pss2 at high levels. The highest levels of Pss2 expression were located in the brown fat during development and in the Sertoli cells of the testis in adult mice. Third, we demonstrated that Pss2 accounts for the majority of serine exchange activity in mammalian tissues. Fourth, we demonstrated that Pss2 is not required for the maintenance of normal phosphatidylserine levels in cultured fibroblasts, even when the cells are deprived of ethanolamine and exogenous phospholipids.
Testicular atrophy was noted in ϳ10% of male Pss2Ϫ/Ϫ mice, and male mice lacking overt testicular atrophy had smaller testes than littermate controls. We suspect that these testes phenotypes could relate to a borderline capacity to synthesize phosphatidylserine in Sertoli cells, which provide nutritional support to the germ cells within the spermatic ducts (23,24). Aside from the reduced testis weight, borderline Sertoli cell function was also suggested by the increased serum FSH levels. Interestingly, we could not document any signifi-

FIG. 4. Phospholipid composition of microsomes from liver (A) and testis (B).
Liver and testis from Pss2ϩ/ϩ mice (black bars) and Pss2Ϫ/Ϫ mice (white bars) were homogenized, and microsomes were isolated by differential centrifugation. PS, phosphatidylserine; PE, phosphatidylethanolamine; PC, phosphatidylcholine; PI, phosphatidylinositol. Values are means Ϯ S.D. from four Pss2ϩ/ϩ mice and five Pss2-/-mice. cant perturbations in phospholipid composition in extracts of whole testes, although serine exchange activity was reduced by Ͼ95%. It is conceivable that phosphatidylserine levels could be reduced in a subset of the cells in the testes, such as the Sertoli cells, but we suspect that any such reductions would be slight and might well be buried within the variation inherent in measuring phospholipids.
Extracts from Pss2-deficient tissues manifested an unequivocal increase in choline exchange activity. Choline exchange is mediated by Pss1 (4, 6, 8). A simple explanation for the finding would have been an increase in Pss1 expression in response to the deficiency in Pss2. However, as judged by Northern blots, Pss1 mRNA expression was entirely normal, both in mouse tissues and in cultured fibroblasts (Fig. 1, E and F). We favor an alternative explanation, that the increased level of choline exchange in Pss2Ϫ/Ϫ tissues reflects a posttranscriptional increase in Pss1 activity. Kuge et al. (25) demonstrated that Pss1 activity is reduced by increased levels of phosphatidylserine in membranes, and they even identified a specific amino acid within Pss1 (Arg-95) that is crucial for the end product regulation of the enzyme. Thus, Pss2 deficiency could result in low levels of phosphatidylserine in some regulatory pool (i.e. membranes in the vicinity of Pss1), resulting in increased Pss1 enzymatic activity and thereby explaining the higher levels of choline exchange activity.
Serine exchange activity was reduced by ϳ90% in Pss2deficient fibroblasts, but the phospholipid composition was normal, even when the fibroblasts were grown in ethanolamine- free and phospholipid-depleted medium. Serine exchange activities were similarly depressed in brain, testis, and liver, and once again, there were no perturbations in tissue phospholipids. The simplest interpretation of these observations is that the low levels of serine exchange activity in Pss2Ϫ/Ϫ mice (i.e. due to Pss1) are sufficient to maintain normal levels of phosphatidylserine and phosphatidylethanolamine. However, two additional factors might be relevant in explaining why phosphatidylserine and phosphatidylethanolamine levels are not altered in Pss2-deficient tissues. First, the ϳ90% decrease in serine exchange activity in Pss2-deficient tissues was measured with an in vitro enzymatic activity assay, and that type of assay might not accurately reflect the rate of phosphatidylserine synthesis in the cell. The in vitro assay is performed under optimal conditions for which the amount of enzyme is limiting, whereas in the intact cell other factors (e.g. the supply of substrates, the presence of inhibitors, and the concentration of ions) might limit the rate of phosphatidylserine synthesis. Thus, although Ͼ90% of serine exchange activity in tissue extracts was eliminated by Pss2 deficiency, we cannot be sure that the rate of phosphatidylserine synthesis was invariably reduced to a comparable degree in each cell type and tissue. Second, and more importantly, a common homeostatic response to an increased rate of phospholipid synthesis is a concomitant increase in the rate of phospholipid degradation (6, 7, 23, 24).
Decreasing the rate of phosphatidylserine synthesis could result in a lower rate of phosphatidylserine and/or phosphatidylethanolamine degradation, facilitating the maintenance of constant phospholipid levels.
The metabolic labeling experiments in fibroblasts (Fig. 5C) suggested that the synthesis of phosphatidylserine is reduced by Pss2 deficiency, in keeping with the reduction in serine exchange activity. However, caution is advised in interpreting those fibroblast experiments. The cell lines used for those experiments were derived from different embryos (which were not on an inbred background) and had undergone independent immortalization events. In interpreting this experiment, one must consider the possibility that extraneous genetic differences, aside from the Pss2 mutation, could have had an indirect effect on phosphatidylserine synthesis rates. In the future, we believe that it will be essential to perform a comprehensive analysis of phosphatidylserine biosynthesis and turnover rates in multiple independent fibroblast cell lines and in multiple tissues from Pss2-deficient mice.
Our current observations contrast with earlier results with Pss1-deficient CHO cells (8,9). In the CHO cells, serine exchange activity was reduced by ϳ50% (using the same assay), but there were highly significant decreases in phosphatidylserine and phosphatidylethanolamine levels when the cells were grown in ethanolamine-free and phospholipid-depleted media (7)(8)(9)13). Why is the phenotype of Pss1 deficiency in CHO cells so different? We speculate that the differences might be explained by the different substrate specificities of the two enzymes. We suspect that Pss2 deficiency is well tolerated because Pss1 is still capable of generating phosphatidylserine from phosphatidylcholine. Even when Pss2-deficient cells are deprived of ethanolamine, phosphatidylserine continues to be produced from phosphatidylcholine, and phosphatidylethanolamine can still be generated by decarboxylation of phosphatidylserine. In contrast, in the setting of Pss1 deficiency, Pss2 can generate phosphatidylserine, but only if phosphatidylethanolamine is available for the exchange reaction. In the absence of exogenous ethanolamine, when phosphatidylethanolamine cannot be generated from the CDP-ethanolamine pathway, there is no way to synthesize significant amounts of phosphatidylserine or phosphatidylethanolamine. In the setting of Pss1 deficiency, phosphatidylethanolamine can only be generated from phosphatidylserine, and phosphatidylserine can only be generated from phosphatidylethanolamine. Thus, Pss1-deficient cells are dependent on exogenous ethanolamine to replenish their phosphatidylserine and phosphatidylethanolamine stores.
The availability of Pss2-deficient mice will make it possible to address a host of issues in the future. For example, we are intrigued by the high levels of Pss2 expression in neurons within the brain. Large amounts of phosphatidylserine are normally imported into mitochondria and then converted to phosphatidylethanolamine (26 -29). It would be interesting to determine whether the neurons of Pss2-deficient mice might be more susceptible to injury in response to metabolic conditions that limit the production of energy from mitochondria (e.g. hypoxemia or hypoglycemia). The current studies also raise several other issues for future experimentation. High on the list is how mice would respond to a deficiency in Pss1. For example, we would like to know whether Pss1 is dispensable for embryonic development of mammals, as was the case for Pss2. If Pss1-deficient mice survived development, we would like to determine how they would respond to a diet containing low levels of ethanolamine. We would also be interested in whether a single Pss1 knockout allele would elicit pathologic findings in Pss2Ϫ/Ϫ mice. Answers to these questions will probably emerge over the next few years.