Peptides derived from serum amyloid A prevent , and reverse , aortic lipid lesions in apoE / mice

Macrophages (M ) at sites of acute tissue injury accumulate and export cholesterol quickly. This metabolic activity is likely dependent on the physiological function of a major acute-phase protein, serum amyloid A 2.1 (SAA2.1), that is synthesized by hepatocytes as part of a systemic response to acute injury. Our previous studies using cholesterol-laden J774 mouse M showed that an N-terminal domain of SAA2.1 inhibits acyl-CoA:cholesterol acyltransferase activity, and a C-terminal domain enhances cholesteryl ester hydrolase activity. The net effect of this enzymatic regulation is to drive intracellular cholesterol to its unesterified state, the form readily exportable to an extracellular acceptor such as HDL. Here, we demonstrate that these domains from mouse SAA2.1, when delivered in liposomal formulation, are effective at preventing and reversing aortic lipid lesions in apolipoprotein E-deficient mice maintained on high-fat diets. Furthermore, mouse SAA peptides, in liposomal formulation, are effective at regulating cholesterol efflux in THP-1 human M , and homologous domains from human SAA are effective in mouse J774 cells. These peptides operate at the level of the foam cell in the reverse cholesterol pathway and therefore may be used in conjunction with other agents that act more distally in this process. Such human peptides, or small molecule mimics of their structure, may prove to be potent antiatherogenic agents in humans. —Tam, S. P., J. B. Ancsin, R. Tan, and R. Kisilevsky. Peptides derived from serum amyloid A prevent, and reverse, aortic lipid lesions in apoE / mice. J. Lipid Res. 2005. 46: 2091–2101. Supplementary key words atherosclerosis • cholesterol • acute-phase proteins • apolipoprotein E-deficient mice • high density lipoprotein Acute tissue injury commonly results in local cell death and the generation of large quantities of cell membrane fragments rich in cholesterol. As part of the reactive acute inflammatory process, macrophages (M ) are mobilized to such sites of injury, ingest these fragments, and acquire a considerable cholesterol load. A removal mechanism is required to mobilize this cholesterol either for reuse or excretion. Our past results suggested that a physiological role of one isoform of a major acute-phase protein synthesized by the liver in response to tissue injury, mouse serum amyloid A 2.1 (mSAA2.1), is the regulation of M cholesterol export (1–3). Fragmentation of this protein into peptides that span its entire length has revealed that the N-terminal region, mSAA2.1 1–20, is a potent in vitro and in vivo inhibitor of M ACAT (3). A separate region at the C terminus, mSAA2.1 74–103 , contains a domain that enhances the in vitro and in vivo activity of neutral cholesteryl ester hydrolase (CEH) (3). In combination, these two peptides drive stored cholesteryl esters into their unesterified form, which, in the presence of a functional cholesterol transporter and an extracellular cholesterol acceptor such as HDL, is rapidly exported from the M (2). These results suggested that such peptides may be useful in mobilizing cholesterol from M at sites of atherogenesis. To examine this possibility, we prepared liposomal formulations of the active SAA2.1 peptides and tested their ability to prevent, or cause the regression of, aortic lipid lesions in apolipoprotein E-deficient (apoE / ) mice maintained on a high-fat atherogenic diet. To determine whether these peptides, as liposomal formulations, are effective only in mouse M , we examined the influence of these peptides on 1 ) ACAT, CEH, and cholesterol efflux activities in cholesterol-laden THP-1 human M , and 2 ) the effect of homologous human peptides on cholesterol-laden J774

Acute tissue injury commonly results in local cell death and the generation of large quantities of cell membrane fragments rich in cholesterol. As part of the reactive acute inflammatory process, macrophages (M ) are mobilized to such sites of injury, ingest these fragments, and acquire a considerable cholesterol load. A removal mechanism is required to mobilize this cholesterol either for reuse or excretion. Our past results suggested that a physiological role of one isoform of a major acute-phase protein synthesized by the liver in response to tissue injury, mouse serum amyloid A 2.1 (mSAA2.1), is the regulation of M cholesterol export (1)(2)(3). Fragmentation of this protein into peptides that span its entire length has revealed that the N-terminal region, mSAA2.1 1-20, is a potent in vitro and in vivo inhibitor of M ACAT (3). A separate region at the C terminus, mSAA2.1 74-103 , contains a domain that enhances the in vitro and in vivo activity of neutral cholesteryl ester hydrolase (CEH) (3). In combination, these two peptides drive stored cholesteryl esters into their unesterified form, which, in the presence of a functional cholesterol transporter and an extracellular cholesterol acceptor such as HDL, is rapidly exported from the M (2). These results suggested that such peptides may be useful in mobilizing cholesterol from M at sites of atherogenesis. To examine this possibility, we prepared liposomal formulations of the active SAA2.1 peptides and tested their ability to prevent, or cause the regression of, aortic lipid lesions in apolipoprotein E-deficient (apoE Ϫ / Ϫ ) mice maintained on a high-fat atherogenic diet. To determine whether these peptides, as liposomal formulations, are effective only in mouse M , we examined the influence of these peptides on 1 ) ACAT, CEH, and cholesterol efflux activities in cholesterol-laden THP-1 human M , and 2 ) the effect of homologous human peptides on cholesterol-laden J774 mouse M .

En face quantitation of aortic lipid lesions
Animals were euthanized by CO 2 nacrosis and exsanguinated by cardiac puncture into heparin-coated syringes for plasma lipid analyses. Total plasma, LDL, and HDL cholesterol and triglyceride levels were determined with Roche modular automated instruments in the clinical laboratories of Kingston General Hospital. The aortas were perfused with 10 ml of PBS via the left ventricle and teased free from the body but left attached to the heart. The adventitial adipose tissue was removed, and the aortas were opened longitudinally, pinned out as described (4), washed with 60% isopropanol for 3 min, stained with Oil Red O (0.4% in 60% isopropanol) for 3 min, rinsed in 60% isopropanol for 3 min, and then fixed in 10% formalin for 2 min. Once fixed, the aortas were stored in 10% formalin until the lipid lesions were quantified. Quantification of the percentage of the aortic surface occupied by Oil Red O-positive lesions was performed with a program and apparatus from MCID M2 Imaging Research, Inc. (St. Catherines, Ontario, Canada) as described previously (5).

Preparation of red blood cell membranes as a source of cholesterol
To mimic the ingestion of cell membrane fragments by M at sites of tissue injury, red blood cells (RBCs) were used as a source of cholesterol, as described previously (2). Similar quantities of cholesterol, 175 g, were used in all experiments. The concentration of cholesterol in the membrane preparations was determined as described previously (2).

Preparation of HDL and acute-phase HDL, and purification of apoA-I and SAA isoforms
HDL and acute-phase high density lipoprotein (AP-HDL) were isolated from normal and inflamed mice, respectively, using sequential density flotation as described previously (6,7). The isolation, separation, and purification of apoA-I, SAA1.1, and SAA2.1 from acute-phase mouse plasma was performed as described previously (6,7). The purity of the isolated proteins was established by mass spectrometry and N-terminal sequencing.

Preparation and characterization of apolipoprotein-lipid complexes (liposomes)
Each of the intact proteins (apoA-I, SAA1.1, and SAA2.1) or the various synthetic peptides listed above were reconstituted with lipids to form liposomes using the procedure described by Jonas, Kezdy, and Wald (8), as detailed previously (2,3). When assessing the effects of the various apolipoproteins or peptides, these were always used as liposomes. Free peptides have no effect in culture or in vivo.

Cell culture
Mouse J774 M (TIB-67) and human THP-1 monocytes (TIB-202) were obtained from the American Type Culture Collection (Manassas, VA). J774 cells were cultured on six-well tissue culture plates at 10 6 cells/well and grown to 90% confluence in 2 ml of DMEM supplemented with 10% FBS. The medium was changed three times per week. THP-1 cells were maintained in RPMI-1640 medium containing 10% FBS according to the instructions supplied by the American Type Culture Collection. These monocytes were differentiated into M with 100 nM phorbol myristate acetate. The cells were seeded onto six-well tissue culture dishes at 10 6 cells per well and maintained in medium containing phorbol myristate acetate (100 nM). Media were replaced every 2 days, and experiments were started after 7 days in culture, when the cells morphologically were M .

Rates of hydrolysis of cholesteryl ester in THP-1 cells
Rates of hydrolysis of radiolabeled cholesteryl ester in THP-1 cells were determined exactly as described previously with J774 cells in the presence of 2 g/ml Sandoz 58-035 (an ACAT inhibitor) to prevent the reesterification of liberated [ 14 C]oleate and free cholesterol (2,3). At various times under the different culture conditions, cellular lipids were extracted and analyzed for cholesteryl ester radioactivity as described above.

Cholesterol efflux in tissue culture and in vivo
THP-1 and J774 cells were laden with cholesterol by incubating with RBC membrane fragments that had been equilibrated previously with 0.5 Ci/ml [ 3 H]cholesterol (2,3). Cholesterol pools were allowed to equilibrate for 18 h in culture, and efflux was examined after treatment with the different isoforms of SAA, or their peptides, as described previously (2,3). To determine cholesterol export in vivo, experiments were conducted as described and validated previously (2). Briefly, J774 M were laden with RBC membranes and [ 3 H]cholesterol as described above, washed with PBS/BSA, and then detached from the culture dishes. Five million cells in 200 l of DMEM were injected into mice via the tail vein. At various times thereafter, ‫ف‬ 25 l of blood was collected from the tail vein of each animal into heparinized capillary tubes, then centrifuged for 5 min in an Adams Autocrit Centrifuge to separate RBCs from plasma. Cholesterol efflux was determined by measuring the appearance of [ 3 H]cholesterol in plasma by scintillation spectrometry. To study whether the export of cholesterol from these injected J774 cells to plasma is influenced by mSAA2.1 1-20 , mSAA1.1 1-20 , or hSAA1.1/2.1 1-23 , 100 l of liposomes containing 15 g of one of these peptides was injected intravenously, and at various times after this injection, ‫ف‬ 25 l of blood was collected from the tail vein of each animal and the plasma was analyzed by scintillation spectrometry.

Protein determinations
Protein concentration was determined by the method of Lowry and coworkers (11), with the aid of a Bio-Rad protein assay kit.
Mice on the low-fat diet had 67 Ϯ 7% less area occupied by lipid lesions relative to mice on the high-fat diet. Among the groups on the high-fat diet, those treated with protein-free liposomes or liposomes containing mSAA2.1 1-20 , mSAA2.1 74-103 , or mSAA2.1 1-20 ϩ mSAA2.1 74-103 had, respectively, 14 Ϯ 19%, 45 Ϯ 8%, 41 Ϯ 13%, and 73 Ϯ 5% less area occupied by lipid lesions relative to mice on the high-fat diet. The value with protein-free liposomes is not significantly different from that of the mice on the highfat diet itself. The value with mSAA2.1 1-20 ϩ mSAA2.1 74-103 is equivalent to that of mice on the low-fat diet. The P values for the groups of mice treated with the various peptides are Ͻ0.05 and are indicated in the legend to Fig. 1. These data indicate that peptides mSAA2.1 74-103 , mSAA2.1 1-20 , and particularly mSAA2.1 1-20 ϩ mSAA2.1 74-103 are effective at inhibiting aortic lipid accumulation in apoE Ϫ/Ϫ mice.

Regression of aortic lipid lesions in apoE ؊/؊ mice by ACAT-inhibiting and CEH-enhancing peptides derived from mouse SAA2.1
In the regression mode, apoE Ϫ/Ϫ mice were first placed on the high-fat diet for a period of 4 weeks. They were then divided into four groups, all of which continued on the diet for an additional 16 days; group 1 received no treatment, group 2 was treated with mSAA2.1 1-20 , group 3 was treated with mSAA2.1 74-103 , and group 4 was treated with mSAA2.1 1-20 ϩ mSAA2.1 74-103 . In each case, the peptides were administered as liposomes once every 4 days (four doses). A fifth group consisted of apoE Ϫ/Ϫ mice on standard lab chow for 44 days. The results are illustrated in Fig. 2.

Effect of liposomes containing SAA peptides on plasma lipid parameters
The effect of the high-fat diet, and treatments, on plasma lipid parameters (triglycerides and total HDL and LDL cholesterol concentrations) are indicated in Table 1. There was a 3-to 5-fold increase in plasma cholesterol pa-rameters when the mice were placed on the high-fat diet; however, there was no apparent effect of the peptides singly, or in combination, on these parameters, despite the fact that these peptides clearly affected the degree of the aortic lipid lesions. A comparison of the plasma lipid parameters between mice that were on the high-lipid diet for 16 days versus 44 days suggests a modest increase in HDL cholesterol in the latter group but a reduction in the other plasma lipid values in the groups fed this diet for longer periods of time. The reasons for these changes are not obvious but may relate to a reduction in the intake of diet over time (there was no difference in body weight between the mice in the various groups; data not shown), perhaps an increase in HDL production, or adaptations to a high-fat diet that are not immediately apparent. Nevertheless, it is important to note that the increase in HDL cholesterol and the decrease in LDL/HDL cholesterol ratio did not protect against the development of lipid lesions unless liposomes containing SAA peptides were also added.

Effect of mouse AP-HDL and liposomes containing SAA2.1 or SAA2.1 peptides on human THP-1 M
Our previous studies with J774 M demonstrated that mSAA2.1 had an ACAT-inhibitory domain at its N terminus and a CEH-enhancing domain at its C terminus (2, 3). Operating individually or in concert, these peptides have proven remarkably effective in culture and in vivo at promoting the rapid efflux of cholesterol from cholesterolladen cells (2,3). To determine whether these results are peculiar to mouse cells, we examined the effect of mouse AP-HDL, mSAA2.1, and mSAA2.1 peptides (each as liposomes) on ACAT and CEH activity and cholesterol export with a human M cell line, THP-1. Figure 4 illustrates the baseline ACAT activity of THP-1 cells, the effect of feeding these cells mouse erythrocyte membrane fragments (as a source of cholesterol), and the subsequent influence of HDL or AP-HDL on such activity. As shown previously with J774 cells (2), loading the cells with cholesterol markedly increased THP-1 ACAT activity,  which was not significantly reduced by subsequent exposure to HDL. However, subsequent exposure to AP-HDL resulted in a 60% reduction in the increased ACAT activity. Figure 4 also shows that this ACAT-inhibitory property of AP-HDL resides in mSAA2.1, analogous to that shown previously with J774 cells (2,3). Furthermore, the domain responsible for the reduction in ACAT activity, as with J774 cells, resides in mSAA2.1 [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] but not in peptides composed of mSAA2.1 21-50 , mSAA2.1 51-80 , or mSAA2.1 74-103 .
The effects of AP-HDL, mSAA2.1, and peptides spanning the entire length of mSAA2.1 on CEH activity in THP-1 cells are shown in Fig. 5. Figure 5A illustrates the rate of breakdown of cholesterol oleate in the presence of medium alone, HDL, and AP-HDL. Only AP-HDL enhanced the esterase activity. Figure 5B shows that the enhanced esterase activity demonstrated in Fig. 5A was quantitatively attributable to mSAA2.1, because protein-free liposomes, and those containing apoA-I or mSAA1.1, had no such effect. Furthermore, the SAA2.1 domain responsible for the enhancement of CEH activity resides only in residues 74-103. These results are completely analogous to those seen previously with murine J774 cells (2,3).
The effect of these peptides on cholesterol efflux from human THP-1 cells is illustrated in Fig. 6. These cells were laden with cholesterol and then pretreated for 4 h with HDL or the designated liposomes, washed free of the pretreatment, and placed into fresh medium containing HDL as the common cholesterol acceptor for the different pretreatments. At various times thereafter, the efflux medium was collected and the quantity of radiolabeled cholesterol released into the medium was determined. The response of the THP-1 cells was similar to that seen previously with J774 cells (2,3). Those cells exposed to medium alone, or HDL alone, had the lowest rate of cholesterol efflux. Pretreatment with SAA2.1 peptides that correspond to the ACAT-inhibiting domain or the CEHenhancing domain released cholesterol more rapidly and in greater quantity than the control treatments. Pretreatment with both peptides promoted the most rapid efflux of cholesterol, and in greatest quantity.
The potential therapeutic value of increasing CEH activity by transgenic means has been alluded to by others (24). Furthermore, in culture, SAA can promote lipid efflux mediated by ABCA1 (2,25), an ATP binding cassette transporter, and unlike apoA-I, SAA can also promote lipid efflux in an ABCA1-independent manner (25). The potential therapeutic properties of mouse SAA (and homologous human) peptides is amply illustrated by the ability of these peptides to substantially prevent, and reverse, aortic lipid lesions in apoE Ϫ/Ϫ mice maintained on a high-lipid diet. As shown in Fig. 1, mice receiving peptides mSAA2.1 1-20 ϩ mSAA2.1 74-103 at the time they started on the high-lipid diet had aortas similar in appearance to those of such mice on the low-fat diet. Furthermore, because the nonnormalized mean aortic area occupied by lipid lesions after 16 days on the high-lipid diet was 1.05% and that after 44 days was 1.85%, the groups commencing treatment after 28 days on the high-lipid diet must have values between these two numbers. Moreover, the group receiving mSAA2.1 1-20 ϩ mSAA2.1 74-103 beginning on day 28 had 0.54% of their aortas occupied by lipid lesions after 44 days, a value at least 50% below that at the start of their treatment, indicating lesion regression. Although we did not include additional controls with irrelevant peptides in the prevention and regression experiments described here, irrelevant peptides in liposomal formulation were examined for their effects on M cholesterol export in vitro and in vivo in previously published work (3). Neither mSAA2.1 21-50 nor mSAA2.1 51-80 promoted cholesterol efflux from cholesterol-laden cells that had been previously established in mice (3). Furthermore, protein-free liposomes failed to prevent the development of aortic lipid lesions in apoE Ϫ/Ϫ mice on atherogenic diets. In the face of these results, we chose not to include additional negative controls.
It should be emphasized that the effect of these peptides is rapid, long-lasting [a single intravenous injection appears to be effective for 96 h or more (3)], and takes place even in the face of a 3-to 5-fold increase in total plasma cholesterol. Furthermore, these peptides do not appear to significantly alter plasma lipid parameters. In this respect, synthetic small molecule ACAT inhibitors have also been shown to have this effect without reducing plasma cholesterol levels (26)(27)(28). This is not surprising, because the administration of these peptides, as liposomes, would target M primarily and not the total body synthesis of cholesterol. It is precisely these cells, as foam cells, and their stored cholesterol, that play a crucial role in the pathogenesis of aortic lipid lesions. They are at the "beginning" of the reverse cholesterol pathway as it relates to cholesterol-laden M. Agents (SAA peptides, or small molecule mimetics thereof) that are targeted to and prompt these cells to release their cholesterol to a natural acceptor (e.g., HDL), small though these amounts may be www.jlr.org Downloaded from relative to total circulating cholesterol, may have a profound influence on the progression of atherogenesis.
Our current data and previously published work beg the question: how does SAA exert its antiatherogenic effects in vivo? And why is there a need for liposomal formulations in the prosecution of this work?
The physiological role of SAA2.1 is one directed at cholesterol efflux/recycling from M at sites of acute tissue injury. The induction of SAA is not a physiological response to atherogenesis in particular. Nevertheless, as we demonstrate in the present work, the mechanism by which SAA2.1 promotes M cholesterol efflux suggests that its active domains can be used for antiatherogenic purposes.
With regard to how SAA exerts its antiatherogenic effects in vivo, we have previously demonstrated that 1) acute-phase mouse HDL (HDL/SAA) has a significantly higher affinity for mouse peritoneal M than normal mouse HDL (21), and 2) M from inflamed animals have a significantly increased number of binding sites for HDL/SAA (21). We and others (22,23) have shown that HDL/SAA is readily and rapidly taken up by M, probably through a receptor-mediated mechanism. Furthermore, such HDL/SAA, when taken into cholesterol-laden M, inhibits M ACAT activity, enhances CEH activity, and promotes substantial cholesterol efflux from these cholesterol-laden cells (2,3), and these effects are mediated through SAA2.1. Therefore, our working hypothesis is that this is the physiological role for SAA, and it is in this manner that SAA would have antiatherogenic properties in vivo. As is apparent, SAA in vivo, as part of HDL, does not require a liposomal formulation for its action.
Although there are many reasons for requiring liposomal preparations in our studies, one was to determine which HDL/SAA apolipoprotein(s) exerted the observed M effects. Other reasons focused on potential therapeutic considerations (see below).
With regard to the HDL/SAA apolipoproteins, both SAA isoforms when delipidated and purified are very insoluble in aqueous media and aggregate even at concentrations between 0 and 6 g/ml (data not shown). To determine which of apoA-I, SAA1.1, or SAA2.1 exerted the observed M effects on ACAT, CEH, and cholesterol export activities, these proteins were purified and, for solubility reasons, reconstituted in HDL-like liposomes (2). The results showed conclusively that SAA2.1, rather than apoA-I or SAA1.1, exercised the observed effects. SAA2.1 was then shown to contain two domains, one of which affected M ACAT activity and the other of which affected CEH activity (3).
Although, as indicated above, liposomal formulations are not required for HDL/SAA's effects on ACAT, CEH, and cholesterol export activities in vivo, they are needed for the following reasons: 1) to solubilize the purified SAA isoforms for direct study; 2) because human SAA1.1 and SAA2.1 are both amyloidogenic (40), and thus intact SAA is not useful as a potential therapeutic agent (these findings directed us to the active domains as potential therapeutic agents); 3) because naked SAA2.1 peptides are rapidly cleared from the circulation and have no effect on cholesterol export in vivo (data not shown); 4) because liposomes target the relevant peptides to M; 5) because full-length human SAA2.1 (104 residues) in addition to being amyloidogenic is difficult to manufacture economically and with the required purity for human use; 6) and because the active domains are much shorter and therefore easier to manufacture and formulate for delivery.
Our previous and present results question the concept that SAA is a proatherogenic protein. This concept arose because correlations exist between small increases in plasma SAA concentration and poor clinical outcomes in patients with unstable angina (41,42). Furthermore, SAA levels increase with age, and patients with the highest SAA levels are more likely to manifest cardiovascular disease (43,44). Similar correlations have been shown in mice on