Secretory Sphingomyelinase, a Product of the Acid Sphingomyelinase Gene, Can Hydrolyze Atherogenic Lipoproteins at Neutral pH IMPLICATIONS FOR ATHEROSCLEROTIC LESION DEVELOPMENT*

The subendothelial aggregation and retention of low density lipoprotein (LDL) are key events in atherogenesis, but the mechanisms in vivo are not known. Previous studies have shown that treatment of LDL with bacterial sphingomyelinase (SMase) in vitro leads to the formation of lesion-like LDL aggregates that become retained on extracellular matrix and stimulate macrophage foam cell formation. In addition, aggregated human lesional LDL, but not unaggregated lesional LDL or plasma LDL, shows evidence of hydrolysis by an arterial wall SMase in vivo , and several arterial wall cell types secrete a SMase (S-SMase). S-SMase, however, has a sharp acid pH optimum using a standard in vitro SM-micelle assay. Thus, a critical issue regarding the potential role of S-SMase in atherogenesis is whether the enzyme

A critical event in early atherogenesis is the subendothelial retention of atherogenic lipoproteins, including LDL 1 (1,2), lipoprotein(a) (3), and triglyceride-rich lipoproteins (4). Retained lipoproteins likely trigger a series of biological responses, such as endothelial changes and recruitment of macrophages to the arterial wall, that are central to the initiation and progression of atherosclerosis (5).
Although the mechanism of lipoprotein aggregation in lesions is not known, several studies from our laboratories suggest a role for the enzyme sphingomyelinase (SMase). First, LDL treated with bacterial SMase forms lesion-like self-aggregates (20) due to enrichment in ceramide (7), the major product of SM hydrolysis; furthermore, these aggregates potently induce macrophage foam cell formation in vitro (16,20). Second, aggregated LDL from human atherosclerotic lesions, but not unaggregated lesional LDL or plasma LDL, shows evidence of hydrolysis by an extracellular SMase, and LDL retained in rabbit aortic strips ex vivo is hydrolyzed by an extracellular, cation-dependent SMase (7). Third, and most important, we have found that several cell types present in atherosclerotic lesions, namely macrophages (21) and endothelial cells, 2 secrete a Zn 2ϩ -activated SMase (S-SMase).
The cellular origins, secretion, and cation dependence make S-SMase a leading candidate for the arterial wall SMase that acts on retained lipoproteins. Nonetheless, two major issues regarding the relevance of S-SMase to atherogenesis needed to be addressed. First, mammalian SMases are much more selective than bacterial SMases in terms of the milieu in which the SM is presented to the enzyme (22). 3 Second, studies on the molecular origin of S-SMase have revealed that it is a product of the same gene, the acid SMase (ASM) gene, that gives rise to lysosomal SMase (L-SMase) (21). Not surprisingly, therefore, S-SMase shares with L-SMase a sharp acid pH optimum when assayed under standard in vitro conditions using detergentsolubilized SM-micelles as a substrate (21,23). Although it is possible that acidic enzymes are active in advanced atherosclerotic lesions, where local pockets of acidity may occur (24 -28), a role for such enzymes in pre-lesional or early lesional events would require activity at neutral pH with physiologic substrates.
In this context, the goal of the current study was to test whether S-SMase can hydrolyze LDL-SM, particularly at neutral pH. Herein, we show that S-SMase can hydrolyze and aggregate native LDL at acid but not neutral pH. LDL modified by several means that have been shown to occur or might occur during atherogenesis, however, is an excellent substrate for S-SMase at pH 7.4. Most importantly, LDL extracted from human atherosclerotic lesions is efficiently hydrolyzed by S-SMase at neutral pH, suggesting that LDL is modified in the arterial wall to increase its susceptibility to S-SMase. Our results support a role for S-SMase in the subendothelial hydrolysis of LDL-SM, perhaps leading to lipoprotein aggregation and lesion initiation and progression.
Isolation of LDL from Human Lesions-LDL was extracted from abdominal aortic aneurysm plaque material as described previously (4). Briefly, aortic plaque was removed from individuals as part of the standard reconstructive surgery for abdominal aortic aneurysms at the San Francisco Veterans Affairs Medical Center. Plaque material, which ranged in weight from 2 to 12 g, was obtained in the operating room and immediately placed into ice-cold 7 mM citrate buffer, pH 7.4, containing 15 mM NaCl, 3 mM EDTA, 0.5 mM butylhydroxytoluene, 1 mM phenylmethylsulfonyl fluoride, 1.5 mg aprotinin/ml, 2 mM benzamidine, and 0.08 mg of gentamicin sulfate/ml. Blood and adherent thrombus were removed by blotting with absorbent gauze, scrubbing with a small brush, and sharp dissection as necessary. Loosely retained lipoproteins were extracted by mincing the plaque into 0.5-1.0-mm 2 pieces and incubating them overnight on a Labquake shaker at 4°C in a nondenaturing buffer (0.1 M citrate, pH 7.4, with 1 mg of EDTA/ml, 0.3 mg of benzamidine/ml, 0.08 mg of gentamicin sulfate/ml, 10 g of aprotinin/ ml, 10 g of Trolox (an anti-oxidant)/ml, and 20 g of phenylmethylsulfonyl fluoride/ml). The extracted material was cleared of particulate matter by centrifuging at 800 ϫ g for 10 min, and 1.019 Ͻ d Ͻ 1.063-g/ml lipoproteins were isolated by sequential potassium bromide density ultracentrifugation (4,36).
[ 3 H]SM Labeling of LDL-Plasma LDL was labeled with [N-palmitoyl-9,10-3 H]SM as follows (7). ϳ3.5 mCi (63 nmol) of [N-palmitoyl-9,10-3 H]SM and 13 nmol of phosphatidylcholine (PC) were mixed in chloroform, and the solvent was removed first under a stream of nitrogen and then by lyophilization. The dried lipids were resuspended in 1 ml of 150 mM NaCl, 1 mM EDTA, 10 mM Tris-HCl, pH 7.5, and, to prepare [ 3 H]SM/PC liposomes, sonicated for three 50-s pulses at 4°C using a tapered microtip on a Branson 450 sonicator (setting 3). The liposomes were then incubated with 30 mg (by protein mass) of LDL, 50 g of partially purified phospholipid transfer protein, 100 units of penicillin, and 100 g of streptomycin for 18 h at 37°C under argon. LDL was then separated from the liposomes after phospholipid transfer by centrifuging the mixture at density ϭ 1.006 g/ml for 8 h at 35,000 rpm in a Beckman 50.3 rotor; the supernatant containing the liposomes was removed, and the LDL band at the bottom of the tube was harvested. The LDL solution was mixed with buffer containing 150 mM NaCl, 0.3 mM EDTA, pH 7.4, and centrifuged as before. This wash procedure was performed a total of four times, resulting in the removal of 95% of the unreacted [ 3 H]SM/PC liposomes. All lipoproteins were stored under argon at 4°C and were used within 2 weeks of preparation.
[ 3 H]SM-Emulsions-[ 3 H]SM-emulsions with a lipid composition similar to human LDL were prepared as follows: 5.4 mg of cholesteryl oleate, 0.48 mg of triolein, 1.08 mg of free cholesterol, 2.04 mg of phosphatidylcholine, 0.96 mg of sphingomyelin, and 50 Ci of [N-palmitoyl-9,10-3 H]sphingomyelin were all added in chloroform to a sonication vial, and the solvent was completely evaporated by exposure to a stream of nitrogen, followed by the high vacuum of a lyophilizer. The dried lipids were resuspended in 3 ml of buffer containing 150 mM NaCl, 0.3 mM EDTA, pH 7.4, and sonicated under a stream of argon at 40°C until translucent (approximately 90 min). The sonicated material was then centrifuged twice at 15,000 ϫ g to pellet any titanium shed from the sonication probe. SM-rich emulsions were prepared exactly as above except that 1.2 mg of phosphatidylcholine and 1.8 mg of SM were used, and sonication time was increased to 120 min. [ 3 H]SM-emulsions were enriched with apoCIII based on the method of Ahmad et al. (42). Briefly, [ 3 H]SM-emulsions (0.4 ml; 182 nmol of SM) were incubated with 200 g of apoCIII (22.5 nmol) for 2 h at 40°C. A portion of the emulsions was re-isolated from free apoCIII using ultrafiltration (29) as follows: the crude emulsion/apoCIII mixture was diluted to 2 ml with buffer containing 150 mM NaCl, 0.3 mM EDTA, pH 7.4, and ultrafiltered and concentrated to 0.2 ml using a Centricon 30 (molecular weight cut-off ϭ 30,000); the concentrated emulsions were then diluted to 2 ml, and the process was repeated 5 times. [ 3 H]SM-emulsions run through the same enrichment and re-isolation protocols in the absence of apoCIII served as the control for the experiments in Fig. 5A.
Ceramide Assay-Ceramide was measured from LDL lipid extracts using the method described by Schneider and Kennedy (43) and adapted by Preiss et al. (44). In this method, diacylglycerol (DAG) kinase phosphorylates ceramide and DAG using [␥-32 P]ATP. For ceramide measurement, the lipids were first incubated with 0.1 N KOH in methanol for 1 h at 37°C, which hydrolyzes DAG but not ceramide. The extracted lipids were dried under nitrogen and then solubilized in 5 mM cardiolipin, 7.5% octyl-␤-glucopyranoside, and 1 mM diethylenetriaminepentaacetic acid by bath sonication. This solution was then added to reaction buffer (50 mM imidazole HCl, pH 6.6, 50 mM NaCl, 12.5 mM MgCl 2 , 1 mM EGTA) containing sn-1,2-DAG kinase (0.7 units/ml). The reaction was initiated by the addition of [␥-32 P]ATP (final concentration ϭ 10 mM). After incubation at room temperature for 60 min, the reaction was stopped by lipid extraction with chloroform:methanol:HCl (100:100:1, v/v/v) and 10 mM EDTA. Ceramide 1-[ 32 P]phosphate in the organic phase was separated by TLC using chloroform:methanol:acetic acid (65:15:5, v/v/v) and visualized with autoradiography and identified by comparing with standards. The spots corresponding to ceramide 1-[ 32 P]phosphate were scraped and counted, and the mass was calculated by comparison with a ceramide standard curve. To rule out changes in the DAG kinase enzyme itself (cf. Ref. 45), C2:0 ceramide was added to some of the lipid extracts and shown not to undergo increased phosphorylation in samples in which LDL ceramide was found to be elevated.
Sphingomyelin (SM) and Phosphatidylcholine (PC) Assays-Lipid extracts (46) of lipoproteins were chromatographed by TLC using chloroform:methanol:acetic acid:H 2 O (50:25:8:4, v/v/v/v). Individual phospholipid subclasses were visualized by iodine vapor staining, and the SM and PC spots were identified by comparison with standards. The spots were scraped, extracted twice with chloroform:methanol (2:1), and assayed for phosphate content by the method of Bartlett (47).
LDL Oxidation Assays-LDL lipid peroxides were measured using the method of El-Saadani et al. (48). LDL (50 -150 g of protein), in a volume of no more than 100 l, was added to 1 ml of color reagent (0.2 M KH 2 PO 4 , 0.12 M KI, 0.15 mM NaN 3 , 2 g of Triton X-100/liter, 0.1 g of benzalkonium chloride/liter, 10 M ammonium molybdate, 20 M BHT, 25 M EDTA, pH 6.2) and incubated in the dark for 30 min at room temperature; light absorbance at 365 nm was then measured and lipid peroxides were quantified by comparison with a H 2 O 2 standard curve. Thiobarbituric acid-reactive substances (TBARS) were measured using a standard method (49). Briefly, LDL (100 l, 100 -200 g of protein) was mixed with 1 ml of 20% trichloroacetic acid and incubated on ice for 30 min. Following precipitation, 1 ml of 1% thiobarbituric acid was added, and the samples were heated at 95°C for 45 min. After cooling, the samples were centrifuged at 1000 ϫ g for 20 min, and the light absorbance at 532 nm was measured. TBARS were quantified by comparison with a malonaldehyde standard curve prepared using tetramethoxypropane. LDL electrophoretic mobility was assayed by loading 30 g of LDL protein onto a polyacrylamide 0.75-27% gradient gel (Lipogel; Zaxis, Hudson, OH) and electrophoresing in 0.1 M Tris base, 0.1 M boric acid, 20 mM EDTA (upper chamber, pH 8.7; lower chamber, pH 8.3) for 12 h at 100 V. The gel was then stained with Sudan black and the bands visualized by counter-staining with methanol:acetic acid: water (10:7: S-SMase-The source of S-SMase was serum-free conditioned me-dium from DG44 Chinese hamster ovary cells stably transfected with the human acid SMase cDNA (21,50). Our previous work demonstrated that S-SMase is the only detectable SMase secreted into the culture medium (21). Cells were plated in 100-mm dishes and cultured for 48 h in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 292 g of glutamine/ml, 100 units penicillin/ml, and 100 g of streptomycin/ml. The cells were then changed to low protein serum-free media for 12 h, washed 3 times with phosphate-buffered saline, and finally incubated for 18 h in fresh serum-free media (6 ml per 100-mm dish). The "18-h conditioned medium" was then collected, centrifuged at 1000 ϫ g to pellet any cells, and, except where indicated, ZnCl 2 (final concentration ϭ 100 M) was added to fully activate and stabilize S-SMase; this S-SMase-containing conditioned medium was then used fresh to treat LDL and lipid emulsions.

S-SMase Treatment of LDL and Emulsions-
The standard incubation mixture consisted of up to 50 l of sample (LDL or emulsions), 25 l of S-SMase-containing conditioned media (see above), and a volume of assay buffer (0.1 M Tris-HCl, pH 7.4, 0.04% azide; or, where indicated, 0.1 M sodium acetate, pH 5.5, 0.04% azide) to bring the final volume to 200 l. The reactions were incubated at 37°C for no longer than 16 h and then extracted by the method of Bligh and Dyer (46). For the samples containing [N-palmitoyl-9 -10-3 H]SM-LDL or [N-palmitoyl-9 -10-3 H]SM-emulsions, the lower organic phase was harvested, evaporated under N 2 , and fractionated by TLC using chloroform:methanol (95:5). The ceramide spots were scraped and directly counted to quantify [ 3 H]ceramide. In all other samples ceramide was determined as described above.
Statistics-Unless otherwise indicated, results are given as means Ϯ S.D. (n ϭ 3); absent error bars in the figures signify S.D. values smaller than the graphic symbols.

RESULTS
Hydrolysis and Aggregation of Native, Oxidized, and sPLA 2treated LDL by S-SMase-We first sought to determine whether S-SMase could act on intact LDL and, if so, whether it would cause LDL aggregation. Although treatment of native LDL with S-SMase at pH 7.4 caused very little SM hydrolysis, S-SMase hydrolyzed significant amounts (nearly 90%) of LDL-SM at pH 5.5 (Fig. 1A), leading to LDL aggregation (Fig.  1B). Thus, S-SMase can hydrolyze the SM in intact LDL without the need for detergent, leading to LDL aggregation.
Prior data from Callahan et al. (51) suggested to us that S-SMase might require low pH only for binding of SM to the enzyme and not for the hydrolysis process per se. We reasoned, therefore, that changes in the surface of lipoproteins might favor the binding of lipoprotein-SM to S-SMase, which might allow the enzyme to function at neutral pH. In particular, several modifications of LDL relevant to atherosclerosis, including oxidation (52), hydrolysis with phospholipase A 2 (8,30,53), and sphingomyelin enrichment, 4 alter the structure of the lipoprotein surface. In this context, we first examined S-SMase  Fig. 2A, solid  bars). Similar results were obtained when LDL-SM hydrolysis was assayed by measuring generation of ceramide mass (data not shown). Note that FeSO 4 and lipoxygenase caused similar degrees of LDL oxidation as indicated by the levels of thiobarbituric acid-reactive substances (TBARS) (Fig. 2B) and lipid peroxides (Fig. 2C) in the LDL.
To more precisely define the relationship between LDL oxidation and its susceptibility to S-SMase hydrolysis, we incubated [ 3 H]SM-LDL with FeSO 4 for increasing amounts of time and compared its extent of oxidation to its hydrolysis by S-SMase at pH 7.4. [ 3 H]SM-LDL oxidation, assayed by measuring lipid peroxides (Fig. 3, open circles) increased almost linearly with increasing time of incubation with FeSO 4 . Electrophoretic mobility of LDL as assessed by native polyacrylamide gradient gel electrophoresis, indicating addition of negative charge, also increased with time of oxidation (not shown). Most importantly, the susceptibility of [ 3 H]SM-LDL to S-SMase hydrolysis at neutral pH was, after an initial threshold level of oxidation, closely correlated with the extent of [ 3 H]SM-LDL oxidation (Fig. 3, closed squares). Note that oxidation of [ 3 H]SM vesicles that contained only SM did not lead to a substantial increase in susceptibility to S-SMase at neutral pH (data not shown). Moreover, acetylated LDL, another negatively charged form of LDL (39), was not more susceptible than native LDL to S-SMase at neutral pH (data not shown). Thus, neither oxidation of SM nor addition of negative charges can explain the increased susceptibility of oxidized LDL to S-SMase at neutral pH.
Lysophosphatidylcholine enrichment of LDL occurs during oxidation due to activation of an apparently latent phospholipase A 2 activity (54), and recent in vitro and in vivo data suggest that hydrolysis of lesional LDL by nonpancreatic secretory phospholipase A 2 type II (sPLA 2 ) is an atherogenic event (8, 30, 53). We therefore asked whether LDL treated with    (Fig. 4). Furthermore, exogenously added lysophosphatidylcholine stimulated S-SMase hydrolysis of [ 3 H]SM-LDL 2-3-fold at pH 7.4 (data not shown). Thus, while it is possible that several oxidation products are acting together, lysophosphatidylcholine appears to be particularly effective in stimulating S-SMase hydrolysis of LDL.

Lipoproteins Enriched in ApoCIII Are Susceptible to S-SMase-induced Hydrolysis and Aggregation at Neutral pH-
Recent data suggest that apoCIII enrichment of lipoproteins promotes atherogenesis (35,55,56), although the mechanism is not known. Moreover, Ahmad et al. (42) have shown that apo-CIII, but not other apolipoproteins, stimulates L-SMase hydrolysis of SM-liposomes. Given that L-SMase and S-SMase are such similar enzymes (21), we asked whether apoCIII could stimulate S-SMase hydrolysis of lipoprotein-SM at neutral pH. Because enriching native LDL with apoCIII in vitro proved extremely difficult, we first examined the effect of apoCIII on S-SMase hydrolysis of [ 3 H]SM-labeled emulsions containing lipids in the same proportions as in LDL. For this experiment we treated three different types of emulsions with S-SMase at pH 7.4: control emulsions not enriched with apoCIII (control), emulsions mixed with apoCIII (ϩ CIII, emulsion-bound ϩ free CIII), and emulsions mixed with apoCIII followed by re-isolation from free apoCIII (ϩ CIII, emulsion-bound only). Although S-SMase hydrolysis of the emulsions with bound plus free CIII was 2-fold greater than hydrolysis of the control emulsions, emulsions with only bound CIII were hydrolyzed approximately 10-fold more than control emulsions (Fig. 5A). Thus, apoCIII stimulates S-SMase hydrolysis of LDL-like emulsions, particularly in the absence of free apoCIII (see Ref. 42 and "Discussion").
To determine whether apoCIII stimulates S-SMase hydrolysis of lipoprotein-SM, we compared the hydrolysis of LDL from LDLr0 mice (32) with LDL from LDLr0 mice expressing a human apoCIII transgene (LDLr0/CIII) (35); as expected, LDL from the LDLr0/CIII mouse was enriched in apoCIII, whereas LDL from the LDLr0 mouse was not (data not shown). The data in Fig. 5B clearly show that LDL from LDLr0/CIII mice was significantly hydrolyzed by S-SMase at pH 7.4, whereas LDL from LDLr0 mice was hydrolyzed very little. Importantly, the turbidity at A 430 nm of LDLr0/CIII LDL, but not of LDLr0 LDL, increased 1.5-fold following treatment with S-SMase, indicating particle aggregation. Lipoprotein enrichment with apoCIII, therefore, increases its susceptibility to hydrolysis and aggregation induced by S-SMase. In data not displayed in this figure, we found that the turbidity at A 430 nm of LDLr0/CIII LDL, but not of LDLr0 LDL, increased 1.5-fold following treatment with S-SMase, indicating particle aggregation.

Plasma LDL from E0 Mice Is Efficiently Hydrolyzed by S-
SMase at Neutral pH-The E0 mouse develops widely distributed and complex atherosclerotic lesions like those seen in humans and thus has become a widely used model of atherosclerosis (33,34). A likely factor contributing to the extensive atherosclerosis in these mice is the atherogenicity of E0 lipoproteins, but the mechanisms whereby apoE-deficient lipoproteins lead to foam cell formation and other lesional events is not known. In the context of our current study, we tested whether plasma LDL from these mice was susceptible to S-SMase at neutral pH; as a comparison, we used plasma LDL from LDLr0 mice, which also develop atherosclerosis, although to a lesser degree than E0 mice (32)(33)(34). Although the basal ceramide contents of native plasma LDL from these two strains of mice are equivalent (Fig. 6A, hatched bars), treatment with S-SMase at pH 7.4 resulted in generation of 4 -5-fold more ceramide in LDL from E0 mice than in LDL from LDLr0 mice (Fig. 6A, solid bars), which resulted in aggregation of E0 LDL. 4 Jeong et al. 4 have recently shown that E0 LDL has a high SM:PC molar ratio (0.36) compared with LDLr0 LDL, which has a SM:PC ratio of 0.19. Since SM enrichment of lipoproteins may increase their substrate availability to S-SMase, one factor contributing to the increased susceptibility of E0 LDL to S-SMase might be its relatively high SM:PC ratio. To test the importance of the SM:PC ratio on S-SMase activity in a controlled system, [ 3 H]SM-labeled emulsions were prepared containing lipids similar to those in human LDL with a SM:PC ratio of either 0.5 (control emulsions) or 1.5 (SM-rich emulsions) and then treated with S-SMase at pH 7.4. Although the control emulsions, unlike native plasma LDL, were significantly hydrolyzed by S-SMase, SM enrichment resulted in a 2-fold stimulation of [ 3 H]SM-emulsion hydrolysis by S-SMase (Fig. 6B). Together, these findings demonstrate that SM enrichment of lipoproteins may markedly enhance their hydrolysis by S-SMase at neutral pH.
Human Atheroma-derived LDL Is Hydrolyzed by S-SMase at Neutral pH-A critical issue regarding the physiologic relevance of the above findings is whether LDL modified in the arterial wall shows greater susceptibility to hydrolysis by S-SMase than native LDL at neutral pH. To address this issue, we compared two different samples of lesional LDL with same donor plasma LDL for their susceptibility to hydrolysis by S-SMase at neutral pH. First, the data in Fig. 7 confirm our previous finding that lesional LDL is enriched in ceramide compared with plasma LDL (compare first pair of bars with second pair of bars) (7). Most importantly, however, treatment of lesional LDL with S-SMase at pH 7.4 generates nearly 10-fold more ceramide than treatment of plasma LDL (compare third pair of bars with fourth pair of bars). 5 LDL retained in the arterial wall, therefore, is an excellent substrate for S-SMase at neutral pH.
What are the critical modifications to lesional LDL that stimulate its hydrolysis by S-SMase? To begin addressing this question, we examined lesional LDL for the presence of the LDL modifications that we identified above as increasing susceptibility to hydrolysis by S-SMase at neutral pH in vitro. For oxidation, we compared the levels of TBARS and lipid peroxides in the lesional and plasma LDL samples from Fig. 7; levels in both the plasma and lesional samples were below the detection limit of our assay (Ͻ0.5 nmol/mg protein and Ͻ4 nmol/mg protein for TBARS and lipid peroxides, respectively). Similarly, we were unable to detect apoCIII in our two samples of lesional LDL (data not shown). We next asked whether the ratio of SM to PC was different for lesional and plasma LDL. The data in Table I show that whereas the PC level in lesional LDL was modestly reduced compared with plasma LDL, the SM content of lesional LDL was nearly 3-fold higher than that in plasma LDL (cf. Refs. 11-13). As a result, the SM:PC ratio for lesional LDL was 3-4-fold higher than the ratio for plasma LDL. Thus, lesional LDL is highly enriched in SM, a modification that stimulates S-SMase hydrolysis of LDL at neutral pH in vitro (see Fig. 6).

DISCUSSION
Aggregated LDL retained in vivo shows evidence of hydrolysis by a SMase in the subendothelium (7). Although the mammalian SMase responsible for this extracellular process has yet to be identified, S-SMase is a leading candidate since it is the only SMase that we have been able to detect in the conditioned media of cultured arterial wall cells (7). Furthermore, prelim- inary immunohistochemistry studies indicate that S-SMase is present in both the pre-lesional artery and in atherosclerotic lesions. 6 Macrophage-derived S-SMase (7) may be a source of the enzyme in established atherosclerotic lesions. Most recently, we have found that cultured human coronary artery endothelial cells secrete abundant amounts of S-SMase basolaterally, 2 making the endothelium a potential source of the enzyme in the pre-lesional artery.
S-SMase, however, has an acid pH optimum using a standard in vitro, detergent-based SM-micelle assay (21,23). Indeed, we found that S-SMase can hydrolyze and aggregate native plasma LDL but only at acid pH (Fig. 1). Although S-SMaseinduced LDL aggregation under these conditions may be relevant in more advanced lesions, where an acidic environment may exist in the vicinity of lesion macrophages (27,28) or as a result of hypoxia-induced metabolic acidosis (57), early lipoprotein aggregation likely occurs at a more neutral, physiologic pH. Thus, our finding that certain types of atherogenic lipopro-teins are substrates for S-SMase at neutral pH may have relevance to early atherogenesis. Callahan et al. (51) demonstrated that for L-SMase, which is closely related to S-SMase (21), only the affinity for SM in detergent micelles (i.e., K m ) is highly sensitive to changes in pH, whereas the maximal velocity (V max ) for SM hydrolysis is pH-independent. Thus, a plausible explanation for our findings is that changes in the lipid surface of the various forms of modified lipoproteins increased the accessibility of their SM to S-SMase. The overall concept that "acid" SMase can function at neutral pH may also have relevance to ceramide-mediated cell-signaling processes, such as apoptosis, that are triggered by the hydrolysis of cellular SM by a product or products of the acid SMase gene (58 -62).
What is the physiologic significance of subendothelial lipoprotein SM hydrolysis? One theoretical possibility is that SMase-treated lipoproteins may be able to deliver ceramide to cells and induce certain ceramide-mediated signal transduction events (63)(64)(65)(66). In fact, oxidized LDL-induced stimulation of smooth muscle cell proliferation may involve the ceramide signaling pathway (67). The major physiological link that we have been studying, however, is SMase-induced lipoprotein self-aggregation (16,20). In this regard, we have found three examples in which S-SMase hydrolysis led to lipoprotein aggregation as follows: native LDL at acid pH (Fig. 1B), LDL from LDLr0/CIII mice at neutral pH (see "Results"), and VLDL ϩ LDL and LDL from E0 mice at neutral pH. 4,5 In contrast, two of the lipoproteins that were hydrolyzed by S-SMase at neutral pH, namely oxidized LDL and sPLA 2 -treated LDL, did not show visible aggregation. In these two cases, the level of ceramide generated (ϳ0.05 and 0.013 nmol of ceramide/g of protein, respectively) was below the threshold level for lipoprotein aggregation in vitro (ϳ0.08 nmol of ceramide/g of protein) (7). Thus, more extensive exposure of these lipoproteins to S-SMase at neutral pH would likely be required for aggregation to occur. Furthermore, it is possible that other arterial wall factors, such as proteoglycans, collagen, and lipoprotein lipase (5,68,69), lower the threshold for ceramide-mediated lipoprotein aggregation in vivo. Interestingly, oxidized LDL treated with S-SMase at neutral pH contained similar levels of ceramide as aggregated lesional LDL (7) (also, compare Fig. 2A with Fig. 7), consistent with the idea that the threshold for ceramide-mediated aggregation may be lowered in vivo.
Our finding that S-SMase can hydrolyze and cause the aggregation of apoCIII-enriched lipoproteins at neutral pH is of particular interest since, through unknown mechanisms, apo-CIII has been linked to certain forms of coronary heart disease (55). In addition, overexpression of apoCIII in mice significantly increases atherosclerosis (35). Although our samples of lesional LDL (1.019 Ͻ d Ͻ 1.063 g/ml) did not contain apoCIII, it is possible that less dense lesional lipoproteins or lipoproteins isolated from other individuals or from earlier lesions might have shown evidence of apoCIII. A possible mechanism for the apoCIII effect described herein may be gleaned from the finding that S-SMase-mediated hydrolysis of apoCIII-containing SM-emulsions was greater in the absence of free apoCIII (Fig. 5A). These data are consistent with the idea that apoCIII may act as a bridge between S-SMase and membrane SM, thereby stimulating SM hydrolysis.
A property of E0 LDL that likely plays a role in its susceptibility to S-SMase is the high SM:PC ratio in these particles (Fig. 6B). The mechanism of lipoprotein-SM enrichment appears to be a combination of increased SM synthesis and decreased plasma SM clearance in the E0 mice. 4 There is evidence, however, that other factors may also be involved in the increased susceptibility of E0 lipoproteins. In particular, native human plasma LDL has an SM:PC ratio that is greater than 6 S. Marathe and I. Tabas, unpublished data.

FIG. 7.
Human lesional LDL is susceptible to hydrolysis by S-SMase at neutral pH. Human plasma LDL (plasma) and LDL extracted from human atherectomy specimens (lesion) were isolated from the same donor. LDL preparations from two different patients (Expt. 1, hatched bars and Expt. 2, solid bars) were incubated at a concentration of 100 g of protein/ml in the absence (Ϫ) or presence (ϩ) of S-SMase-containing conditioned media (125 l/ml) at pH 7.4 for 14 h at 37°C. Lipids were then extracted and assayed for ceramide mass. Incubation at 37°C in the absence of S-SMase did not affect the ceramide contents of plasma and lesional LDL. that in E0 LDL and equal to that in the "control" emulsions in Fig. 6B and yet is substantially less susceptible to hydrolysis by S-SMase at neutral pH than these particles. Since E0 LDL contains mostly apoB48 (33,34) and the emulsions contain no protein, it is possible that apoB100, the protein in human native plasma LDL, confers relative resistance to hydrolysis by S-SMase. If so, the degradation of apoB100 that is known to occur with LDL oxidation (52) may contribute to the enhanced susceptibility of oxidized LDL to hydrolysis by S-SMase. It is also possible that apoE itself may be an inhibitor of S-SMase hydrolysis, although we have recently found that apoE enrichment of the E0 lipoproteins does not block S-SMase-induced aggregation. 4 A critically important finding in this report was that LDL extracted from human atherosclerotic lesions was hydrolyzed 10-fold more than same donor plasma LDL by S-SMase at neutral pH (Fig. 7). These data indicate that lesional LDL is modified in vivo, either before or after retention in lesions (see below), to a form that renders it more susceptible to hydrolysis by S-SMase. Although it is difficult to determine the specific characteristics of lesional LDL that increase its susceptibility to S-SMase, we have observed, as have others (11)(12)(13), that lesional LDL is enriched in SM (Table I), which we have shown is a potent stimulator of S-SMase hydrolysis of LDL (Fig. 6B). The mechanism of SM enrichment of lesional LDL is not known. The possibilities include SM enrichment of LDL retained in the subendothelium (e.g. by increased arterial wall synthesis of SM (70) followed by transfer of the SM to retained LDL) or selective uptake of SM-rich LDL from the plasma. Another property of lesional LDL that may increase its susceptibility to S-SMase is oxidation. Although we did not find evidence of TBARS or lipid peroxides in our samples of lesional LDL, more sensitive and/or specific techniques, such as immunostaining for oxidized epitopes (11) or evidence of myeloperoxidase-induced oxidation (9), may be required.
In summary, S-SMase, a product of the acid SMase gene (21), can efficiently hydrolyze and aggregate modified LDL at neutral pH. Most importantly, LDL extracted from atherosclerotic lesions, but not plasma LDL, is significantly hydrolyzed by S-SMase at neutral pH, indicating that modifications present in vivo can stimulate S-SMase hydrolysis of LDL. In previous reports, we have shown that SMase treatment of lipoproteins promotes lipoprotein aggregation and retention to matrix, which induces macrophage foam cell formation (16,20), that lipoprotein-SM hydrolysis occurs in the subendothelium in vivo (7), and that S-SMase is secreted by both macrophages (21) and endothelial cells. 2 Based on the sum of these data, we propose a model in which S-SMase secreted by arterial wall cells hydrolyzes lipoproteins retained in the subendothelium, which, in turn, promotes lesion initiation and progression.