A new HDL mimetic peptide that stimulates cellular cholesterol efflux with high efficiency greatly reduces atherosclerosis in mice.

Here, we report the creation of a single-helix peptide (ATI-5261) that stimulates cellular cholesterol efflux with Km molar efficiency approximating native apolipoproteins. Anti-atherosclerosis activity of ATI-5261 was evaluated in LDLR−/− and apolipoprotein (apo)E−/− mice ∼5–7 months of age, following 13–18 weeks on a high-fat Western diet (HFWD). Treatment of fat-fed LDLR−/− mice with daily intraperitoneal injections of ATI-5261 (30 mg/kg) for 6 weeks reduced atherosclerosis by 30%, as judged by lesion area covering the aorta (7.9 ± 2 vs.11.3 ± 2.5% control, P = 0.011) and lipid-content of aortic sinus plaque (25 ± 5.8 vs. 33 ± 4.9% control, P = 0.014). In apoE−/− mice, the peptide administered 30 mg/kg ip on alternate days for 6 weeks reduced atherosclerosis by ∼45% (lesion area = 15 ± 7 vs. 25 ± 8% control, P = 0.00016; plaque lipid-content = 20 ± 6 vs. 32 ± 8% control, P < 0.0001). Similar reductions in atherosclerosis were achieved using ATI-5261:POPC complexes. Single intraperitoneal injection of ATI-5261 increased reverse cholesterol transport from macrophage foam-cells to feces over 24–48 h. In summary, relatively short-term treatment of mice with the potent cholesterol efflux peptide ATI-5261 reduced substantial atherosclerosis. This was achieved using an L-amino acid peptide, in the presence of severe hypercholesterolemia/HFWD, and did not require daily injections or formulation with phospholipids when administered via intraperitoneal injection.


Animals
All procedures were approved by institutional Animal Welfare Research Committee. Seven-or eight-week-old LDLR Ϫ / Ϫ and apoE Ϫ / Ϫ mice (male, C57BL/6) from Jackson Laboratories (Bar Harbor, ME) were housed in a pathogen-free environment (12-h light/dark cycle) and given free access to food and drinking water.

Atherosclerosis studies
Mice were acclimated 1 week on chow diet before receiving high-fat Western diet (HFWD) (Harlan TekLad TD.88137) to promote atherosclerosis. LDLR Ϫ / Ϫ mice were fed HFWD for 7 weeks, then injected (intraperitoneally) with peptide (30 mg/kg body weight) daily for 6 weeks in the continued presence of HFWD. ApoE Ϫ / Ϫ mice were fed the HFWD for 18 weeks before injections (intraperitoneal) of ATI-5261 or ATI-5261:POPC complexes for 6 weeks. ApoE Ϫ / Ϫ mice were fed chow diet during peptide intervention. After anesthesia with isofl urane, mice were euthanized and their hearts and whole aortas were perfused with saline. Hearts were embedded in OCT compound (Tissue-Teck), frozen on dry ice, and stored until sectioning. Serial 10 m sections (every fi fth from the middle of the ventricle until the appearance of the aortic valve and every second section from the appearance to the disappearance of the aortic leafl ets) were collected on poly-Dlysine coated slides, stained with Oil Red O and hematoxylin, and counterstained with Fast Green ( 10 ). Quantifi cation of atheromatous lesions were performed by computerized analysis (Image Pro Plus, Version 6.0; Media Cybernetics, Inc.), and expressed as average % lesion area occupied with lipid from six sections/ mouse. The descending thoracic-and abdominal-aorta (up to bifurcation of common iliac arteries) were stored overnight in Histochoice fi xation, split open longitudinally, and stained with Oil Red O. The percentage of aortic surface covered by atheroma was determined by computer-assisted planimetry ( 10 ).

Toxicology, lipids, and in vivo RCT
Aspartate-and alanine-aminotransferase activities (AST and ALT, respectively) in plasma were quantifi ed by IDEXX Laboratories, Inc. (Maine) to assess liver toxicity/necrosis. Plasma total-cholesterol levels were determined using commercial kits. For aminotransferase and cholesterol measurements, blood was drawn from the retro-orbital plexus at termination of atherosclerosis studies, i.e., 2 h after fi nal injection of ATI-5261 and ATI-5261:POPC complexes (30 mg/kg dose). Ability of ATI-5261 to stimulate RCT in vivo was assessed as described ( 4 ). The outcome of this assay was used to identify the peptide injection schedule (daily or alternate days) for atherosclerosis studies; thus RCT was evaluated in atherosclerotic mice (i.e., not exposed to 6 weeks peptide intervention).

Statistics
Data were analyzed by ANOVA and Student's t -test, using unequal variance and two-tail test.

RESULTS
Members of our group recently found that the C-terminal (CT) domain of apoE was a potent mediator of ABCA1 cholesterol effl ux and that the entire CT domain (aa216-299) was required for activity ( 28 ). This previous study also compared with native apolipoproteins on a molar basis (17)(18)(19)(20). Other peptides that have been described target the macrophage cholesteryl ester cycle and LCAT activation to modulate HDL cholesterol effl ux and RCT (21)(22)(23).
In contrast to anti-oxidant activity, comparatively little has been done to optimize cholesterol effl ux potential of HDL mimetic peptides ( 19 ). Of particular interest is stimulating cholesterol effl ux via the membrane ABCA1, which enhances RCT and protects against macrophage foam-cell formation and atherosclerosis ( 24,25 ). Many apolipoproteins stimulate ABCA1 cholesterol effl ux, suggesting common determinants may govern the effl ux process ( 26 ). The nature of these determinants is not known with certainty, but class A amphipathic ␣ -helices are thought to be involved ( 17,27,28 ). Apolipoproteins are largely composed of various types of amphipathic ␣ -helices having polar and nonpolar surfaces ( 29 ). It is generally believed that two (or more) of these ␣ -helices linked via proline are required to mediate cholesterol effl ux effectively ( 17,28,30 ). Consequently, there is a defi ciency of small peptides for stimulating cholesterol effl ux with high potency. Difficulties designing peptides with secondary structure also limits scientifi c and clinical pursuits ( 31 ).
Here we report the design of a single-helix peptide that stimulates cholesterol effl ux with high effi ciency. This was achieved by engineering a class A ␣ -helix from a short segment derived from the C-terminal domain of apoE. The synthetic peptide displayed exceptional ␣ -helicity and solubility characteristics, stimulated cholesterol effl ux similar to native proteins on a molar basis, and reduced atherosclerosis in hyperlipidemic mice.

METHODS
Peptide, CD spectroscopy, lipid binding, and cholesterol effl ux assays ATI-5261 was synthesized from L-amino acids and capped with N-terminal acetyl and C-terminal amide groups (Biosynthesis Inc., TX). Lyophilized peptide ( ‫ف‬ 95% pure) was routinely dissolved to ‫ف‬ 3-4 mg/ml PBS (pH = 7.4), but stock solutions of >140 mg/ml could easily be prepared. The latter represents ‫ف‬ 100-to 1000-fold greater solubility than current HDL mimetic peptides ( 22 ). Complexes of ATI-5261 and POPC were prepared by cholate dialysis ( 32 ). Peptide concentrations were determined by absorbance at 280 nm. Mean hydrophobicity and amphiphilicity were calculated as described ( 27 ). Circular dichroism (CD) spectroscopy was carried out on a Jasco 810 spectropolarimeter at 25°C, using lipid-free peptide in 10 mM phosphate buffer (pH = 7.4) and an average of four scans (20 nm/min/scan) ( 33 ). Cholesterol effl ux activities were determined as before ( 27,28 ). Effl ux effi ciency ( K m ) was calculated using the Michaelis-Menten equation (Graph-Pad Prism4) and 4 h data with 0.1 to 10 g ATI-5261/ml and 1 to 20 g apoA-I/ml. Activity of lipid-free ATI-5261 to lyse human RBCs was evaluated as described ( 34 ). Percent hemolytic activity was calculated from absorbance readings at 540 nm, using maximum values obtained in the presence of 1% triton X-100.
ATI-5261 was selected over other peptides because of superior effl ux effi ciency and high aqueous solubility ( 27 ). For the present studies, plasma half-life of 125

I-ATI-5261 in apoE
Ϫ / Ϫ mice and The engineered peptide (designated ATI-5261) possessed an ␣ -helical content of ‫ف‬ 70-80% in the absence of lipid, which was ‫ف‬ 2-fold higher than the original aa238-266 segment ( Fig. 1 ). The mean hydrophobicity and hydrophobic moment of ATI-5261 were Ϫ 0.22 and 0.24, respectively, versus values of Ϫ 0.34 and 0.14 for the original aa238-266 sequence. Therefore, ATI-5261 displayed greater hydrophobicity and amphiphilicity characteristics compared with the original sequence from which it was designed. The aa238-266 peptide failed to stimulate cholesterol effl ux from cAMP treated macrophages ( Fig. 2A ), revealed that relatively hydrophobic segments (aa260-299) containing a class G ␣ -helix conferred effl ux effi ciency, whereas the early portion (aa216-237) of the fi rst helical segment of the CT domain possessed acidic residues endowing class A structure. Presently, we tested whether these features of hydrophobic character and class A acidic residues could be compressed into a single relatively small ␣ -helix to create a peptide that stimulates cholesterol effl ux with high effi ciency. A segment corresponding to the central (aa238-266) CT region was used as a template to design the effl ux peptide ( Fig. 1 ).

Fig. 1.
Creation of ␣ -helix peptide ATI-5261. Top bar represents the CT domain of apoE (aa216-299). The aa238-270 segment possesses features useful for constructing a model class A ␣ -helix. These features include numerous sites for intra-helical ionic interactions (i+4) at the lipid-water interface ( 35 ) and potential for designing a favorable nonpolar surface for binding lipid. Underlined qlutamine (Q) and alanine (A) residues were replaced with underlined tryptophan (W) and phenylalanine (F) to accommodate strongly hydrophobic amino acids at these sites within a shorter sequence. The polar Q21 was replaced with leucine (L) to expand nonpolar surface-area (ATI-5261wheel and -net diagrams) and I13 replaced with phenylalanine (F) to further increase hydrophobicity. Q residues on the polar surface were replaced with smaller alanine (A) residues, generally good for ␣ -helices. The A4 to serine (S) substitution was intended to minimize hydrophobic character of the polar surface. Glutamate (E) was used at positions 15 and 19 to create an alignment of acidic residues down the center of the polar surface, thereby endowing class A structure. E19 also created an additional site for salt-bridge formation with R23. Right side of the fi gure compares the original apoE CT sequence with that of ATI-5261 (amino acid changes in bold). Net-charge of the original aa238-266 sequence = 0 and net-charge of ATI-5261 = -1. Also shown are Far-UV CD spectra and % ␣ -helicity for the original apoE CT seq. aa236-266 (top scan) vs. ATI-5261 (bottom scan, mean ± SD, n = 3), using 62 M concentrations of peptides in 10 mM phosphate buffer (pH = 7.4).
to native apolipoproteins ( Fig. 2C ). The peptide stimulated cholesterol effl ux at concentrations where apoA-I was largely ineffective ( Fig. 2D ), reaching maximal effl ux at 3 g peptide/ml ( Fig. 2D ). As a result, ATI-5261 stimulated cholesterol effl ux with a K m molar effi ciency approxi-consistent with previous results using peptides based on the apoE CT domain ( 28 ). In contrast, ATI-5261 stimulated high levels of cholesterol effl ux from macrophages treated with cAMP and low levels of effl ux in the absence of ABCA1 induction ( Fig. 2A, B ). This behavior was similar schedules, respectively. The latter was verifi ed in a second apoE Ϫ / Ϫ mouse study (18 weeks HFWD), where 10 mg/kg injected ip on alternate days (6 weeks) reduced the lipid content of aortic -sinus plaque (26 ± 5 vs. 36 ± 4 control, respectively, n = 8/group, P = 0.00088) in continued presence of HFWD.
Complexes of ATI-5261:POPC (7-8 nm) possessed ‫ف‬ 4fold greater cholesterol effl ux capacity than the lipid-free peptide ( Fig. 2E, F ). This was associated with an increased V max for activity obtained with complexes versus the lipidfree peptide ( Fig. 2G ). The former was likely attributed to phospholipid and ABCA1-independent mechanisms. However, cholesterol effl ux to complexes increased substantially with cAMP treatment of cells, suggesting a portion of effl ux was attributed to ABCA1. This was verifi ed using ABCA1 expressing HeLa cells (16 ± 0.4 vs. 22 ± 3% effl ux/24 h from nonexpressing and ABCA1-expressing cells, respectively, 50 g complexes/ml, n = 3). The ABCA1 component of effl ux was not attributed to contamination of the complexes with lipid-free peptide ( Fig. 2E , right). Therefore, ATI-5261 formulated with POPC was not prevented from mediating ABCA1 effl ux, as expected for apoA-I on HDL ( 36 ). The peptide-POPC formulation, however, stimulated cholesterol effl ux with lower efficiency versus lipid-free ATI-5261, as judged by a higher K m for activity ( Fig. 2G ).
Daily intraperitoneal treatments of atherosclerotic LDLR Ϫ / Ϫ mice with lipid-free ATI-5261 for 6 weeks reduced atherosclerosis by ‫ف‬ 30% ( Fig. 3 ). Total-cholesterol concentrations in plasma were similar in control and peptide treated mice at termination (2552 ± 586 and 2809 ± 436 mg/dl, respectively), indicating that ATI-5261 reduced substantial atherosclerosis in the presence of severe and persistent hypercholesterolemia.
ATI-5261 stimulated macrophage RCT in apoE Ϫ / Ϫ mice over 24-48 h, as judged by an increase in fecal [3H]sterol secretion ( Fig. 4A ). The latter suggested ATI-5261 may reduce atherosclerosis if provided at 48 h intervals, as opposed to daily injections. This was tested by treating apoE Ϫ / Ϫ mice with either daily ip injections of ATI-5261 (15 mg/kg) or injections every other day (30 mg/kg) for 6 weeks. Both protocols signifi cantly reduced plaque lesions ( Fig. 4B ), producing 20 and 47% reductions in atherosclerosis with daily and alternative-day injection   and tandem repeats, indicating that effi cient stimulation of cholesterol effl ux is not dependent on multiple ␣ -helices as previously thought. ATI-5261 was engineered by introducing amino acid substitutions to a "template" segment derived from the CT domain of apoE. Negatively charged glutamine residues were introduced into this segment at positions 15 and 19 (endows class A structure), together with hydrophobic amino acids (A, W, F, and L) at various positions indicated in Fig. 1 . These changes were accompanied by a marked increase in ␣ -helicity, hydrophobic moment, and hydrophobicity. The former is consistent with secondary structure being a key determinant for effl ux activity ( 17,37 ). The molecular determinants governing ATI-5261 structure and activity are not fully understood at present. However, we have found that ATI-5261 apparently does not form high molecular weight aggregates in solution (data not shown), as previously described for peptide 4F ( 22 ). This suggests that numerous factors may infl uence the cholesterol effl ux activity of HDL mimetic peptides, including secondary structure, physical state of the peptide in solution, as well as charge and/or hydrophobic characteristics. Consequently, identifying determinants that confer cholesterol effl ux activity likely requires detailed and systematic mutagenesis experiments that take into account changes in the biophysical properties, aggregation tendency, and ␣ -helical behavior of peptides ( 19 ).
Peptide 18A mimics class A amphipathic ␣ -helices of apoA-I ( 12 ). This peptide stimulates cholesterol effl ux from ABCA1 expressing cells, but high concentrations are required for activity compared with apolipoproteins on a molar basis ( 17,20 ). It is often discussed that18A is unable to inhibit atherosclerosis development in mice, contrasting its D-4F and 5F analogs ( 12,22,23 ). However, 18A stimulates cholesterol effl ux in vitro similar to its 4F family members ( 20 ). This suggests that anti-atherosclerosis activity of these peptides does not correlate with a cholesterol effl ux mechanism. Rather, the anti-atherosclerosis activity of D-4F appears to correlate with anti-oxidative function ( 12,16,23 ).

DISCUSSION
ATI-5261 constitutes a novel peptide that stimulates cellular cholesterol effl ux with high potency. The cholesterol effl ux response saturated with increasing concentrations of lipid-free ATI-5261, displaying a hyperbolic relationship typical of apoA-I and E ( 28 ). This relationship is indicative of a cooperative process that governs cholesterol effl ux stimulation. Noteworthy, ATI-5261 lacks proline residues mice, and shows greatly improved ␣ -helicity/solubility characteristics, suggesting the peptide may be useful therapeutically. Obviating the need for full-length apolipoproteins/peptide formulations simplifi es production, reduces costs, and may enable wide-spread applications of HDL mimetic technologies. More detailed mechanistic studies of ATI-5261 may prove useful for optimizing dose-regimens and biomarkers for human effi cacy studies.
Our studies adopted dietary and treatment protocols similar to those used in studies of ETC-216, i.e., apoA-IMilanophospholipid complexes ( 10 ). This comparison revealed ATI-5261 reduced atherosclerosis to roughly the same extent as ETC-216 in the apoE Ϫ / Ϫ mouse model. Moreover, ATI-5261 did not exhibit hemolytic activity and was highly soluble in aqueous buffers, suggesting the peptide may be useful for acute treatment of atherosclerosis and/or as a chronic therapy.
The underlying mechanism(s) by which ATI-5261 exerts its anti-atherosclerosis effects in vivo are incompletely understood at present. The peptide was designed to stimulate cholesterol effl ux with high potency and thus is likely to be effi ciently lipidated in vivo. Thus the lipid-free peptide administered intraperitoneally likely associates with endogenous phospholipid before entering the plasma compartment. This may improve peptide pharmacodynamics and explain why the lipid-free peptide reduced atherosclerosis to the same extent as ATI-5261:POPC complexes in the present studies. Therefore, it is currently not known whether the lipid-free peptide would reduce atherosclerosis to the same extent as ATI-5261:POPC complexes if administered intravenously, which represents one potential route for administration in humans. Interestingly, ATI-5261 proved highly effective at reducing atherosclerosis when administered every other day. This could refl ect peptide mechanism-of-action and/or increased stress associated with daily injections/handling of mice, which may infl uence response to treatment.
Complexes of ATI-5261:POPC stimulated ABCA1 cholesterol effl ux, providing evidence the peptide bound to lipid was not prevented from interacting with membranes of ABCA1 expressing macrophages. The latter could involve peptide exchanging from the complexes to the cellsurface. Moreover, spiking of mouse serum with peptide (2:1 mol ratio relative to apoA-I) increased by ‫ف‬ 40% cholesterol effl ux activity of the subsequently diluted serum (1% fi nal) from cAMP treated J774 cells (4.2 ± 0.6 vs. 5.9 ± 1%/4 h control vs. ATI-5261), but had little impact on cholesterol effl ux from J774 cells not induced for ABCA1 response (2.2 ± 0.16 vs. 2.4 ± 0.3%/2 h for control vs. ATI-5261). These results suggest ATI-5261 is capable of exerting effects on cholesterol effl ux in a biological milieu in the presence of lipid and align with the peptide increasing macrophage RCT/fecal [3H]sterol secretion in mice. With regards to the latter, ATI-5261 and [3H]foam-cells were injected simultaneously via the intraperitoneal injection route for assessment of RCT. Therefore, it will be important to test whether a similar response occurs following intravenous infusion of peptide or using different methods of delivery. In addition, ATI-5261 may bind oxidized lipid and thereby exert anti-oxidant effects ( 42 ). It is also feasible that the anti-atherosclerosis effects of ATI-5261 could be independent of lipid transport, perhaps involving ABCA1-mediated activation of JAK2/STAT3 signaling to suppress infl ammation ( 43 ).
In summary, ATI-5261 possesses potent cholesterol effl ux activity, reduces substantial atherosclerosis in