A Possible Role for Plasmalogens in Protecting Animal Cells against Photosensitized Killing*

cells 12- acid (P12) membrane lipids. of Pl2-labeled excitation of the pyrene moiety serum albumin, and harvested using a single phase Bligh and Dyer solvent mixture (28). Lipids were extracted and total fluorescence determined. The results represent the total uptake of P12 by cells expressed as nanomoles/mg cellular protein, and each point is the average value of three determinations. The time course of PI2 uptake (2 pM) was followed in the presence (+) or absence (-) of 20 pM 1-0-hexadecyl-sn-glycerol supplementation.

about half of the choline-linked phospholipid is also recovered as the plasmalogen variant (3).
Plasmalogens were discovered 64 years ago, but no unique, biological functions have been ascribed to them (1,2). Like other phospholipids, plasmalogens can form bilayers when dispersed in water (10). Although plasmalogens do differ from diacylglycerophospholipids in certain, subtle physical properties (10)(11)(12) and in their sensitivity to mild acid hydrolysis (lo), it is not known how relevant these differences are under physiological conditions.
In order to study the biological significance of plasmalogens, we have isolated CHO mutants with a 10-fold reduced plasmalogen content (9). The primary lesion in these strains appears to be in the biogenesis of peroxisomes (9). The measurable level of N-ethylmaleimide-resistant dihydroxyacetone phosphate acyltransferase is reduced 50-fold and that of alkyl synthase is reduced &fold (9). These two enzymes (Fig. 2) catalyze the first steps of plasmalogen biosynthesis (1, 2, lo), and they are both localized in peroxisomes in wildtype cells (13,14). Since subsequent enzymes of plasmalogen biosynthesis, including the alkylglycerol salvaging reactions (Fig. 2), are localized in the endoplasmic reticulum (1,2), it is possible to restore normal plasmalogen levels without restoring peroxisomes to the mutants (see below) by supplementing the growth medium with l-O-hexadecyl-sn-glycerol (Fig. 2). Biochemically, the plasmalogen/peroxisome-deficient CHO mutants (9) resemble diploid human fibroblasts from patients with Zellweger syndrome (16-18), but recent studies suggest that the CHO mutants do complement certain common Zellweger lines (18). The CHO system has the advantage that the cells can be cloned and can proliferate indefinitely.
We now report that plasmalogen-deficient CHO mutants are several orders of magnitude more sensitive than normal CHO cells to conditions of photodynamic stress, a treatment that results in the generation of singlet oxygen and various reactive radical species (19)(20)(21). For this purpose, we have labeled cells with a pyrene-containing fatty acid sensitizer (22), designated P12 (Fig. l ) , followed by irradiation with long wavelength ultraviolet light (>300 nm). We have found that l-O-hexadecyl-sn-glycerol supplementation of the mutant cells restores considerable resistance to this treatment, suggesting that it is the absence of plasmalogens (and not of peroxisomes) that is, at least in part, responsible for P12/UV hypersensitivity of the mutants. These findings have led US to explore the hypothesis that plasmalogens, by means of their vinyl ether linkage, can function as scavengers of reactive oxygen species, thereby possibly protecting other targets (membrane proteins, polyunsaturated fatty acids) from photodynamic damage. Chemical evidence for the remarkable lability of plasmalogens during P12/UV treatment of cells is presented in the accompanying paper (39). The results de-

Mutants Hypersensitive to
Photodynamic Stress 11591

EXPERIMENTAL PROCEDURES
MateriaL~-12-(l'-Pyrene)dodecanoic acid (P12) was purchased from Molecular Probes (Junction City, OR). a-Monopalmitin, 1-0hexadecyl-sn-glycerol, and 2-0-hexadecyl-sn-glycerol were purchased from Serdary (London, Ontario). All other lipids and fine chemicals were purchased from Sigma. [ Cells and Culture Conditions-The Chinese hamster ovary cell line, CHO-K1 (wild-type strain), was obtained from the American Type Culture Collection. Strains ZR-78,ZR-82, and ZR-87 were isolated as plasmalogen/peroxisome-deficient derivatives of CHO-K1, as previously described (9). All cell lines were grown at 37 "C, in a 5% COZ, 95% air atmosphere, in Ham's F12 medium, containing 10% fetal bovine serum (Gibco), supplemented with glutamine (1 mM), penicillin G (100 units/ml), streptomycin (73.5 units/ml), and insulin (0.5 IU/ml). P12 was added to the growth medium by dilution of a concentrated stock solution (20 mM) prepared in dimethyl sulfoxide. The final concentration of dimethyl sulfoxide to which the cells were exposed was less than 0.2% v/v, and its presence did not influence the results. When adding P12 to living cell monolayers, complete growth medium containing five times more than the desired final concentration of P12 was prepared. An appropriate amount of this material was then added to the cell cultures to achieve the P12 concentrations indicated in the experiments. 1-0-Hexadecyl-sn-glycerol was added to the growth medium in the same way as P12, except that the 1-0-hexadecyl-sn-glycerol stock solution was prepared in ethanol.
PI2/UVKilli~-Cells growing in plastic tissue culture dishes were incubated for the appropriate times in the presence of various concentrations of P12. Prior to irradiation, the medium containing PI2 was aspirated, and the cells were washed twice with medium and then incubated in fresh medium (containing serum) but lacking P12. Cells were irradiated following the medium change by illumination for 5 min, unless otherwise indicated, with a Transilluminator (Ultraviolet Products Inc., San Gabriel, CA), placed underneath at a distance of 2 cm. A glass plate, 1 mm thick, was positioned between the tissue culture dish and the UV source to exclude ultraviolet light with wavelengths below 300 nm. The intensity at 365 nm was 1200-1400 microwatts/cm2 when measured through the plastic dish with a 365 nm Blak-Ray Ultraviolet Meter (Ultraviolet Products Inc.). Cells that had not been labeled with P12 could be irradiated for at least 30 min under these conditions without loss of viability.
Determination of Cell Viability-Two methods were used to determine the effect of P12/UV treatment on cell viability. 1) Cells were seeded at low density (100-200 cells/60-mm plate), allowed to attach for 12-24 h and treated with P12 and UV irradiation, as described above. Colonies resulting from surviving cells were counted after staining with Coomassie Blue, following outgrowth of the cells for 7-10 days. 2) Cells were seeded at a higher density (5 X lo3 cells/l6mm plate), allowed to attach for 12 h, and treated with PlZ/UV. Cell viability was evaluated 24 h later by measuring the incorporation of 37 "C (25).
[3H]thymidine into acid-precipitable material during a 2-h pulse at Conditions for Photosensitization Experiments under N2"Phosphate-buffered saline (PBS) (26) was degassed under aspirator vacuum, and bubbled with nitrogen under sterile conditions for 45 min with fast magnetic stirring. Cells were grown in plastic tissue culture dishes and treated with increasing concentrations of P12 for 18 h. After washing with medium, the cells received 3 ml of sterile PBS (Nz-purged or normal) prior to photosensitization on top of the UV box as described above. The dishes containing N2 purged PBS were covered with a Lucite chamber, which was purged continuously with nitrogen, 10 min before and during UV irradiation. Subsequently, the buffer was removed by aspiration and replaced with 3 ml of fresh medium. Colonies resulting from surviving cells were counted after staining with Coomassie Blue, following outgrowth of the cells for 7 days.
Incorporation and Distribution of PI2 and Radioactive Fatty Acids-Cells were seeded at the concentrations indicated in the text in 20-ml glass scintillation vials (5 cmZ) sterilized by heating for 18 h at 130 "C. After incubation in P12-containing medium under the various conditions indicated, the medium was removed, and cells were washed twice with 10 ml of ice-cold PBS (27), containing 2 mg/ml fatty acid-free bovine serum albumin. Next, 3.8 ml of a single phase Bligh and Dyer (28,29) mixture, consisting of chloroform:methanol:PBS (1:2:0.8, v/v), was added directly to the cells. (Parallel vials containing untreated cells were used for protein determinations. These cells were washed after P12 labeling with PBS lacking bovine serum albumin, and 0.8 ml of 1 N NaOH was used to solubilize the cells prior to protein determination (30).) When larger amounts of cellular material were required for subfractionation of lipid species, cells were grown to mid-exponential in 100 mm diameter tissue culture dishes, treated with P12, and harvested as above, with the exception that the cells were scraped from the dish in 5 ml of PBS with a rubber policeman, pelleted by centrifugation, resuspended in 0.8 ml of PBS, and then added to ch1oroform:methanol (1:2) to form the single phase Bligh and Dyer mixture (28,29). After addition of 0.3 mg of mouse liver phospholipids as carrier, 1 ml of PBS and 1 ml of chloroform were added to form the two-phase Bligh and Dyer system. The lower phase was collected and the upper, aqueous phase was washed once with 2 ml of fresh, pre-equilibrated lower phase. The two lower phases containing the cellular lipids were combined, dried under a stream of nitrogen and redissolved in 2 ml of chloroform:methanol (96:5). The P12 content of the lipid fraction was determined by measuring its fluorescence in a JY3 Jobin-Yvon spectofluorometer using an excitation wavelength of 344 nm and an emission wavelength of 378 nm. Absolute values were obtained by comparison of the sample values with those of P12 standard curves. Background fluorescence, obtained by extraction of cells that had not been treated with P12, was typically less than 5% of sample values. Accumulation of radioactive fatty acids by cells was monitored by liquid scintillation counting of a portion of the lipids, extracted as described above, using Patterson and Green scintillation mixture Total lipid extracts were separated into neutral and polar-lipid fractions using silica gel chromatography (32,33). Individual phospholipid species, including plasmenylethanolamine, were separated using two-dimensional thin layer chromatography as described previously (8,9). Radiolabeled lipid species were located by autoradiography and identified by their migration with standards (8,9).
Fluorescence Microscopy":, X 10' cells were seeded in 5 ml of medium in 60-mm tissue culture plastic dishes containing sterile glass coverslips. The cells were allowed to attach for 3-4 h, and each plate received 50 p1 of medium containing 200 p~ P12. After a 20-h incubation at 37 "C, the coverslips were removed, washed once in medium, and twice in PBS. After the cells were fixed in 4% formaldehyde in PBS for 1 h, a drop of glycerol was placed on top of the cell layer on the coverslip. The coverslip was then overturned and mounted on a glass slide. The edges of the coverslip were sealed to prevent dehydration of the samples. The cells were viewed in a Vario Orthomat Leitz microscope, equipped with a BP 350-410 excitation filter, an RKP 455 beam splitter, and an L P 445 emission filter, using a 63.5 X fluo-oil objective. Color slide photographs were made using 200 ASA Ektachrome films (Eastman Kodak). (31).

UVSensitivity of Pl2-labeled Wild-type and Mutant Cells-
Human leukemic cells labeled with 12-(1'-pyrene)dodecanoic acid (P12) are killed by long wavelength ultraviolet light (22). Excitation of the pyrene photosensitizer (25) under aerobic conditions presumably leads to the generation of singlet oxygen (Type I1 chemistry), a molecule that rapidly destroys histidine, tryptophan, methionine, and many other susceptible targets (19)(20)(21). In addition, excitation of the photosensitizer may initiate radical species (Type I process) (19)(20)(21). In the case of CHO-K1 cells (Fig. 31, labeling with 2 & I (or more) P12 for 20 h sensitized the cells to a 5-min irradiation with long wavelength UV light. The extent of killing was dependent upon the concentration of P12 used to label the cells (Fig. 3), the intensity of irradiation, and the time of exposure to UV light (data not shown). Three independently isolated CHO mutants defective in plasmalogen biosynthesis Effect of nitrogen purging on the viability of wildtype and mutant cells exposed to photosensitization. Wild-type and mutant cells growing in medium at 37 "C were labeled for 18 h with 2 pM P12. After removing the medium and washing, the cells were exposed for 4 min to UV light, either in normal sterile PBS or in sterile nitrogen-purged PBS, as described under "Experimental Procedures." The buffer was immediately aspirated and replaced with 3 ml of growth medium containing serum. Colonies were allowed to grow for another 7 days, and were subsequently stained (34) and counted. The results are expressed as the percentage of colonies counted relative to unlabeled control cells, and each point represents the average value of three determinations. and peroxisome assembly (9) were much more sensitive to P12/UV treatment than wild-type cells (Fig. 3). For instance, at 2 M M P12 there was 90-100% killing of the mutant cells, but only 20% killing of wild-type cells (Fig. 3). Labeling of mutant ZR-82 and wild-type cells with 2 PM P12 for various times (data not shown) indicated that 8 h of P12 labeling were suffiicient to render the mutant cells hypersensitive to photodynamic killing. Fig. 4 shows that both wild-type and mutant cells were more resistant to the P12/UV treatment when the system was purged with nitrogen in order to reduce the concentration of dissolved oxygen, as described under "Experimental Procedures." For example, the plating efficiency of wild-type cells (Fig. 4, panel B ) at 4 p~ P12 and 4 min UV was reduced to 50% in aerated buffer, but it was greater than 90% in nitrogenpurged PBS. Similarly, the plating efficiency of mutant cells at 2 p~ P12 and 4 min UV was 42% in normal PBS but 82% in nitrogen-purged PBS (Fig. 4, panel A). Nitrogen purging itself for 10-20 min had no toxic effect on wild-type or mutant cells. These data indicate that, besides P12 and UV light, oxygen is required to obtain full lethality, as expected for a true photodynamic effect (21).
Use of Photosensitization as a Selection Procedure-Mixing studies were performed in which P12/UV treatment was used to select wild-type cells out of a large population of mutant cells (Fig. 5). When 1 X lo4 mutant cells (ZR-82) were treated with 2 p~ P12, followed by the usual UV irradiation, no colonies resulted (Fig. 5, panel B ) , although one or two colonies occasionally appeared upon repeated attempts (not shown). When 20 wild-type cells were added to 1 X lo4 mutant cells, P12/UV treatment consistently resulted in the appearance of 5-10 colonies (Fig. 5, panel A), equivalent to the efficiency of plating of 20 wild-type cells without any P12/ UV treatment (Fig. 5, panel C). When surviving colonies obtained from such a mixing experiment (Fig. 5, panel A) were examined for the recovery of peroxisomal dihydroxyacetone phosphate acyltransferase, using colony autoradiography

FIG. 5. Selection of wild-type cells from a population of mutants. Cells were plated out in 100-mm dishes, labeled with 2 pM
P12 for 20 h at 37 "C and irradiated for 5 min after washing and incubation in fresh medium, as in Fig. 3. Cells were overlaid the following day with polyester cloth (35). After 9 days, the polyester cloths, containing the immobilized colonies, were frozen at -20 "C. The immobilized colonies were screened for peroxisomal dihydroxyacetone phosphate acyltransferase using colony autoradiography (9), and subsequently stained with Coomassie Blue (34, 35). Panel   cells accumulated 2-4 times more P12 than wild-type cells. Enhanced P12 incorporation was observed both in the neutral and the phospholipid fraction. Most of the P12 (83.5 and 87.6% in the wild-type and mutant, respectively) was found in the neutral lipids, which consisted primarily of triglycerides and cholesterol esters in these cells (data not shown). Unesterified P12 was also detected.
The accumulation and subcellular distribution of P12 were also examined using fluorescence microscopy (Fig. 7). Wildtype cells labeled with 2 p~ P12 (Fig. 7, panel A ) appeared uniformly fluorescent, while the mutant cells (Fig. 7, panel B ) contained areas of intense fluorescence, presumably neutral lipid droplets.
The enhanced uptake of P12 by the mutants may account, to some extent, for their hypersensitivity to P12/UV treatment. As shown below, however, another important factor responsible for the P12/UV hypersensitivity appears to be the absence of plasmalogens. The reasons for the enhanced P12 accumulation by the mutants are unknown, but the phenomenon could be explained by the absence of the peroxisomal poxidation system (15), which may be essential in degrading pyrene-containing fatty acids. It is known that P12 and related analogs are not activated to the corresponding carnitine derivative, essential for transport into the mitochondria, while still @-oxidized by isolated peroxisomes (36). The fact that there was no difference between wild-type and the various mutant cells in their ability to take up [l-"CC]palmitate or radiolabeled arachidonate (data not shown) provides indirect support for the view that the @-oxidation of P12 in wild-type cells occurs mainly in peroxisomes.
Bypass of the Plasmalogen Deficiency of the Mutant Cells by Alkylglycerol Supplementation-Addition of an alkylglycerol, such as 1-0-hexadecyl-sn-glycerol, to the medium resulted in the restoration of the plasmalogen pool in the mutant cells (Fig. 8), presumably because of efficient alkylglycerol salvage (Fig. 2). The extent to which the plasmenylethanolamine pool was restored could be varied over a 10-fold range and depended on the concentration of l-0-hexadecyl-sn-glycerol in the medium (Fig. 8). In the presence of -10 p~ l-o-hexadecylsn-glycerol, the plasmenylethanolamine content of the mu- Cells were seedegI at a density of 200 cells/60-mm plate and allowed to attach for 24 h. P12 was added at a final concentration of 2 phi for 12 h at 37 "C. The medium was changed, and the cells were irradiated for 6 min. Cells were overlaid with polyester cloth md glass beads (35) 24 h after Irradiation. Polyester cloths were removed after 8 days, and the polyester-immobhd colonies, arising from the surviving ceb, were stained. When the medium was supplemented with 1-0-hexadecyl-sn-glycerol (HG), it was added 24 h prior to and during the incubation with P12.

TABLE I Effect of I -0-hexadecyl-sn-glycerol supplementation
on Pl2lUV-induced cell killing Cells were seeded in 24-well plastic dishes (16 mm diameter) at a density of 5 X lo3 cells/well and allowed to attach overnight. Cells were then labeled at 37 "C for 12 h with 2 p~ P12, and washed with fresh medium, followed by UV irradiation for 5 min. 1-0-Hexadecylsn-glycerol-supplemented cells were grown in the presence of this material for 48 h prior to and during P12 exposure. Cells were allowed to grow for another 24 h and were subsequently labeled with [methyl-3H]thymidine (2.5 pCi/ml) for 2 h. Finally, the amount of 3H incorporated into acid-insoluble material was determined (25). sn-glycerol concentration in the medium (Fig. 9) and was observed throughout the usual range of P12 concentrations (Fig. 3) used to photosensitize the cells (data not shown).
Supplementation of the mutants with 20 ~L M l-o-hexadecylsn-glycerol did not inhibit or alter the enhanced P12 uptake characteristic of the mutants, as judged either by quantitation of P12 uptake (Fig. 6) or by fluorescence microscopy (data not shown). Furthermore, supplementation did not alter palmitate incorporation (data not shown), nor did it restore peroxisome biogenesis. Taken together, these results suggest that it is the absence of plasmalogens, in conjunction with the excess accumulation of P12, and not the absence of peroxisomes, that renders the mutant cells P12/UV hypersensitive.
Growth of the mutants in the presence of the isomer 2-0hexadecyl-sn-glycerol did not restore plasmalogens (data not shown), and it did not protect them against photodynamic killing (Table 11). Surprisingly, a-monopalmitin did afford significant protection against P12/UV treatment (Table 11) without restoration of plasmenylethanolamine (data not shown). Protection by a-monopalmitin may be explained, in part, by its ability to inhibit P12 uptake by as much as 40-50% (see below).

DISCUSSION
We have discovered a phenotype associated with plasmalogen deficiency in CHO cells that suggests a special antioxidative function for plasmalogens, not previously appreciated. When mutant CHO cells lacking plasmalogens are labeled with the photosensitizer, P12, they are rendered much more susceptible to killing by long wavelength UV light than are P12-labeled, plasmalogen-containing cells. UV irradiation of the pyrene moiety of P12 under aerobic conditions generates singlet oxygen and various reactive radical species (19)(20)(21). The genetic evidence presented here makes it necessary to consider mechanisms by which plasmalogens might protect cells from reactive oxygen species.
Whatever the molecular basis for the P12/UV hypersensitivity may be, this phenotype will be very useful for further genetic studies. The possibility of selecting for wild-type (i.e. P12-resistant) revertants and transfectants, coupled with the rapid screening afforded by colony autoradiography (9), should greatly facilitate the identification of the genes involved in peroxisome and plasmalogen biogenesis. A broader analysis of PlP/UV-resistant strains derived either from the mutants or from wild-type cells should also provide new insights into the mechanisms by which animal cells protect themselves against oxidative stresses.
Supplementation of the mutant cells with 1-0-hexadecylsn-glycerol permitted us to evaluate the relative importance of peroxisomes and plasmalogens in bringing about the P12/ UV hypersensitivity. Since supplementation restores plasmalogens without restoring functional peroxisomes, it is reasonable to conclude that plasmalogen deficiency is not the cause of the failure of peroxisome biogenesis in the mutants. On the other hand, the substantial restoration of P12/UV resistance accompanying the restoration of the plasmalogen pool in 1-0-hexadecyl-sn-glycerol-supplemented mutants supports the view that plasmalogens play a role in protecting cells against P12/UV-induced damage.
The finding that a-monopalmitin also restores resistance to P12/UV treatment (Table 11) without restoring the plasmalogen pool is puzzling. However, we have observed that amonopalmitin, in contrast to 1-0-hexadecyl-sn-glycerol (Fig.  6), reduces P12 uptake by CHO cells by 40-50%. In addition, it is likely that a-monopalmitin causes the cells to accumulate triglyceride droplets, which could sequester the photosensitizer. The possibility that a-monopalmitin induces the synthesis of some other protective molecule must also be considered.
The observation that the mutant cells can utilize exogenous 1-0-hexadecyl-sn-glycerol to restore their plasmenylethanolamine pool demonstrates that the normal topography of plasmalogen biosynthesis (ie. the localization of the first two enzymes of Fig. 2 in peroxisomes) (13)(14)(15) is not actually required for the regulation of cellular lipid composition. The reasons for the participation of peroxisomes in plasmalogen biosynthesis (13)(14)(15) in wild-type cells remain unclear. The ability of the mutant cells to utilize exogenous 1-0-hexadecylsn-glycerol when endogenous plasmalogen biosynthesis is deficient resembles the enhanced, specific utilization of exogenous phosphatidylcholine (37) or phosphatidylserine (38) observed in conditional mutants defective in phosphatidylcholine and phosphatidylserine biosynthesis, respectively. The mechanisms by which cells regulate the uptake of exogenous lipids, depending upon their need for them, require much further investigation. In the case of 1-0-hexadecyl-sn-glycerol supplementation it is especially striking that the total ethanolamine-linked phospholipid content remains constant (i.e. 19-2096) over a 40-fold range of plasmenylethanolamine to phosphatidylethanolamine ratios (Fig. 8). This finding indicates the existence of very precise mechanisms in CHO cells for the regulation of polar headgroup composition.
The apparent protection of cells against PlP/UV-induced photodynamic damage by plasmalogens raises the question of how such protection might be achieved. In the accompanying manuscript (39), we provide evidence for the hypothesis that a direct chemical reaction between oxygen and the vinyl ether moiety of plasmalogens occurs in P12/UV-treated cells. Although the reactivity of vinyl ethers toward singlet oxygen in chemical model systems is well known (19, no), the participation of plasmalogens in such reactions has not previously been considered. Whatever the mechanism, we demonstrate that plasmalogens are sensitive targets for reactive oxygen species in photosensitized CHO cells, and based on the phenotype of our mutants we speculate that plasmalogens might be endowed with the ability to protect other targets in membranes, such as the tryptophanyl, histidyl, or methionyl residues of proteins (19,21). Our hypothesis does not exclude additional functions for plasmalogens, including previous theories based on physical properties (10)(11)(12).