Evidence that cynomolgus monkey cholesteryl ester transfer protein has two neutral lipid binding sites.

Two inhibitors of cynomolgus monkey cholesteryl ester transfer protein were evaluated. One, a monoclonal antibody made against purified cynomolgus monkey cholesteryl ester transfer protein, was capable of severely inhibiting triglyceride transfer, but had a variable effect on cholesteryl ester transfer. At low antibody to antigen ratios, there was what appeared to be a stoichiometric inhibition of cholesteryl ester transfer, but at high antibody to antigen ratios the inhibition of cholesteryl ester transfer was completely relieved, even though triglyceride transfer remained blocked. Fab fragments of the antibody had no effect whatsoever on cholesteryl ester transfer, but were capable of completely blocking triglyceride transfer. The other inhibitor, 6-chloromecuric cholesterol, severely inhibited cholesteryl ester transfer with minimal inhibition of triglyceride transfer. When both inhibitors were added to the assay, both cholesteryl ester and triglyceride transfer were inhibited; an indication that the inhibitors did not compete for the same binding site on cholesteryl ester transfer protein. When the antibody was given subcutaneously to cynomolgus monkeys at a dose which inhibited triglyceride transfer in the plasma by more than 90%, there was no detectable effect on the high density lipoprotein (HDL) cholesterol level, but the HDL triglyceride levels decreased from 13 ± 2 to 1 ± 0 mol/mol of HDL (mean ± S.D.); an indication that the antibody uncoupled cholesteryl ester and triglyceride transfer in vivo. The 6-chloromecuric cholesterol could not be evaluated in vivo because it is a potent lecithin:cholesterol acyltransferase inhibitor. The fact that cholesteryl ester transfer can be inhibited without effect on triglyceride transfer and, conversely, that triglyceride transfer can be inhibited without effect on cholesteryl ester transfer indicates that these two lipids are not transferred by a single, non-discriminatory process.


MATERIALS AND METHODS
Reagents-HDL was isolated from cynomolgus monkey plasma by ultracentrifugation (d 1.063-1.225 g/ml; (9)) and then passed over a 2.6 ϫ 100-cm column of agarose beads (A-0.5M; Bio-Rad) to remove residual LDL and Lp(a). Lipoprotein-deficient plasma (LPDP) was prepared by adjusting the density of fresh plasma from cynomolgus monkeys to 1.225 g/ml with solid KBr and centrifuging at 285,000 ϫ g for 48 h. The top one-third of the tubes' contents was removed by slicing. The bottom fraction was dialyzed extensively against phosphate-buffered saline and stored at 4°C. Human LDL containing [4-14 C]cholesteryl oleate and [9,10-3 H]triolein (Amersham) was prepared as described previously (9).
Recombinant cynomolgus monkey CETP was purified from the culture media of Chinese hamster ovary cells that had been transfected with cynomolgus monkey CETP cDNA exactly as described previously (18). SDS-polyacrylamide gel electrophoresis and amino-terminal sequence analysis indicated that the CETP isolated by this method was Ͼ97% pure.
Monoclonal antibodies were prepared using the purified, recombinant CETP essentially as described previously (11). These antibodies were designated CMTP-1 and CMTP-2. CMTP-1 had no effect on either cholesteryl ester or triglyceride transfer when evaluated in the radioactivity-based assay described here or a fluorescence-based assay described previously (19). At certain concentrations (see "Results") CMTP-2 inhibited both cholesteryl ester and triglyceride transfer in both assays.
CMTP-2 antigen binding fragments were prepared and purified exactly as described by Margulies (20), except that the IgG was purified from the ascites using a commercially available kit (E-Z-SEP for ascites, Pharmacia Biotech, Piscataway, NJ). Analysis of the Fab preparation by SDS-polyacrylamide gel electrophoresis showed a major band of approximately 50 kDa and no detectable IgG.
The compound 6-chloromecuric cholesterol (U-617) was synthesized and purified by chemists at the Upjohn Co.
In Vitro Studies-The CETP assay used here has been described in detail previously (9) and is a modification of that originally described by Pattnaik et al. (21) and modified by Morton and Zilversmit (17). Briefly, radiolabeled lipids are incorporated into a "donor" lipoprotein and that donor is then incubated with an "acceptor" lipoprotein in the presence of a source of CETP. In the present study, cynomolgus monkey LPDP or purified recombinant cynomolgus monkey CETP was used as CETP source; human LDL labeled with [4-14 C]cholesteryl oleate and [9, H]triolein was used as the donor; and cynomolgus monkey HDL was used as the acceptor. Each incubation tube contained LDL, HDL, a source of CETP and incubation buffer (Tris-HCl (50 mM, pH 7.4) containing bovine albumin (1%), NaCl (150 mM), and NaN 3 (3 mM)). The amount of LDL-or HDL-associated lipid contained in the assay tube is shown in Table I. The final volume in the assay tube was 700 l. The reaction was started by the addition of CETP and the redistribution of radioactivity was followed by taking samples at various times, removing the LDL by precipitation, and quantifying the radioactivity in the supernatant (9).
In the absence of transfer protein, redistribution of radioactivity was essentially zero. In the presence of CETP, the distribution of the radioactive lipids between the LDL and HDL approached equal specific activity. This occurred regardless of whether the radioactive lipid was initially in LDL or HDL. The redistribution kinetics were those of a simple, two-pool, closed system, in which there was no appreciable net mass transfer. Fig. 1 shows a typical radioactivity redistribution pattern when LDL was the donor and HDL the acceptor. The HDL radioactivity is expressed as a fraction of the expected equilibrium value, where equilibrium is defined as the point at which the specific activity of a given lipid is the same in the donor and acceptor. Note that both the triglyceride and the cholesteryl ester radioactivity content of HDL were approaching a plateau by 8 h (Fig. 1), even though their mass ratios were different (Table I). That was taken as indication that the CETP-mediated redistribution of the radioactivity was behaving as predicted.
To evaluate CETP inhibition, that inhibitor was added at the indi-cated concentration and the assay mixture incubated for 3 h. The CMTP-2 and the CMTP-2 Fab fragments were contained in saline. The U-617 was contained in Me 2 SO. The final Me 2 SO concentration in the reaction mixture was 1% (v/v); a concentration that had no detectable effect on the assay. The plasma cholesterol and triglyceride concentrations and the HDL composition and Stokes' radius were determined as described previously (22).
In Vivo Studies-Adult, male cynomolgus monkeys obtained from the Upjohn Colony were used for these studies. In study 1, four monkeys received CMTP-2 and 4 others received CMTP-1. In study 2, three monkeys received CMTP-2. In each instance the antibody was dissolved in sterile saline and passed through a 0.45-m Millipore filter prior to injection. Thirty mg of protein in a final volume of 12 ml was injected into each monkey subcutaneously, between the shoulder blades. In study 1, 7-ml blood samples were taken just prior to administration of the antibody (t ϭ 0) and at 0.25, 1, 2, 3, 4, 7, 9, and 11 days after antibody administration. In study 2, a 24-ml blood sample was taken from each monkey 7 days prior to antibody administration and a second 24-ml blood sample was taken 48 h after antibody administration. All blood samples contained EDTA at 1 mg/ml and azide at 0.02% and were kept at 4°C until analyzed.
Statistics-The effects of subcutaneous administration of CMTP-2 on the plasma lipids of the cynomolgus monkey (Table II) were evaluated using a one-way, repeated measures analysis of variance and Dunnett's test for multiple comparisons (23). The effects of CMTP-2 treatment on the size and composition of HDL were evaluated using the paired t test (23).  (20,000 dpm) and [9,10-3 H]triolein (46,000 dpm) was incubated with cynomolgus monkey HDL (39 g of HDL protein) in the presence (squares) and absence (circles) of cynomolgus monkey LPDP for the times indicated. The reaction was stopped by precipitation of the LDL and the radioactivity in the supernatant (HDL) measured. The chemical composition of the LDL and HDL used in this assay is contained in Table I. Equilibrium was defined as the point at which the distribution of a given radioactive lipid between LDL and HDL was the same as that lipids mass. The data were analyzed by a non-linear least squares method using a simple first order reaction model. The good agreement of the theoretical curves with the experimental points was taken as support for the model. f, q, cholesteryl ester radioactivity; Ⅺ, E, triglyceride radioactivity. Fig. 2A shows the effect of increasing CMTP-2 concentration on the redistribution of radioactivity between LDL and HDL, as measured with cynomolgus monkey LPDP as the source of CETP. Triglyceride transfer was progressively inhibited with increasing amounts of the antibody, whereas cholesteryl ester transfer was progressively inhibited only at antibody concentrations Յ16 nM. Beyond this point, the inhibition decreased with increasing antibody concentration until at concentrations near 100 nM, cholesteryl ester transfer was almost normal. Subsequent studies (not shown) demonstrated that at CMTP-2 concentrations above 1 M, cholesteryl ester transfer was not inhibited whatsoever.

RESULTS
The icons in the lower portion of the graphs are intended to show the theoretical relationship between antigen (depicted as the gray globular structure) and antibody (depicted as the black y-shaped structure) at various antibody concentrations. If the affinity of the antibody for CETP is high, then at antibody concentrations below stoichiometry the antibody should be saturated with CETP, i.e. both antigen binding sites should be occupied. This form is depicted by the icon (a) in the lower left of Fig. 2A. As the antibody concentration increases, the degree of saturation decreases, such that at antibody concentrations well above stoichiometry only one of the antigen binding sites should be occupied. That form of the antibody-CETP complex is depicted by the icon (b) in the lower right portion of Fig. 2A.
We hypothesized that only the saturated form of the antibody ( Fig. 2A, icon a) is capable of inhibiting cholesteryl ester transfer, and the loss of inhibition evident at high antibody concentrations (Ͼ16 nM) coincides with the accumulation of the hemisaturated form of the antibody ( Fig. 2A, icon b). To test this premise, we produced antibody binding (Fab) fragments of CMTP-2 and evaluated the effects of those Fab fragments on cholesteryl ester and triglyceride transfer. The results of those experiments are shown in Fig. 2B. The CMTP-2 Fab fragments, like the intact antibody, were capable of completely inhibiting triglyceride transfer, but no inhibition of cholesteryl ester transfer was detectable, even at Fab concentrations near 1 M. These experiments not only support the hypothesis that both antigen binding sites on CMTP-2 must be occupied for cholesteryl ester transfer to be inhibited, but also demonstrate that CETP-mediated cholesteryl ester transfer can be uncoupled from triglyceride transfer.
To eliminate the possibility that a distinct protein was responsible for each type of neutral lipid transfer, the Fab studies were repeated using recombinant cynomolgus monkey CETP that had been purified from the media of Chinese hamster ovary cells expressing the protein. Fig. 3 shows the results of those experiments. Note that, as with LPDP, there was a selective inhibition of triglyceride transfer by the Fab fragments and no inhibition of cholesteryl ester transfer whatsoever. Thus, triglyceride transfer could be selectively inhibited regardless of whether LPDP, or purified, recombinant CETP was used as the source of transfer activity.
The question then arose as to whether the converse condition could be created, i.e. whether cholesteryl ester transfer could be  To determine whether the CMTP-2 Fab binding site on cynomolgus monkey CETP overlapped the U-617 binding site, monkey LPDP was preincubated with high concentrations of the Fab fragments after which the effects of U-617 on lipid transfer were measured. As expected, triglyceride transfer was inhibited greater than 95%; the effects on cholesteryl ester transfer are shown in Fig. 5. Note that the Fab fragments did not block the ability of the compound to inhibit cholesteryl ester transfer and, in fact, appeared to increase the potency of U-617 somewhat. That indicates that the site on CETP at which U-617 binds is different from that which binds the Fab.
Given the observations that the transfer of cholesteryl ester and triglyceride could be selectively inhibited in vitro, the question arose whether the same would occur in vivo. Previous studies (25)(26)(27) indicated that inhibition of cholesteryl ester transfer resulted in an increase in HDL-cholesterol levels, but it has not been firmly established what effect selective inhibition of triglyceride transfer would have on plasma lipoprotein levels. To see whether triglyceride transfer could be inhibited in vivo to a significantly greater extent than cholesteryl ester transfer, 30 mg of purified CMTP-2 were administered to each of four cynomolgus monkeys subcutaneously. Analysis of plasma samples taken from those monkeys at various times after antibody administration indicated that triglyceride transfer was inhibited approximately 90% from 24 to 96 h and that it was inhibited to a much larger extent than cholesteryl ester transfer (Fig. 6). Table II shows the effects of antibody administration on the plasma cholesterol and triglyceride concentrations in those monkeys. Note that there was not a significant effect of the antibody on either total or HDL cholesterol levels; however, there was a clear decrease in total plasma triglyceride concentrations as a result of antibody treatment, and that decrease appeared to be due largely to the decrease in the HDL fraction. In separate studies run in parallel with these, four other monkeys were injected each with 30 mg of CMTP-1, a monoclonal antibody to cynomolgus monkey CETP which had no detectable

FIG. 4. Effects of 6-chloromecuric cholesterol (U-617) on CETPmediated neutral lipid transfer.
The assay is the same as that described in the legend to Fig. 2, except that the compound U-617 was present at the concentrations indicated. U-617 was dissolved in Me 2 SO for addition to the assay mixture. The final Me 2 SO concentration never exceeded 1% (v/v); a concentration that has no detectable effect on CETP-mediated neutral lipid transfer. f, cholesteryl ester radioactivity; Ⅺ, triglyceride radioactivity.

FIG. 5. Effects of combining 6-chloromecuric cholesterol and CMTP-2 Fab on CETP-mediated cholesteryl ester transfer.
The assay is the same as that described in the legend to Fig. 4, except that only the effects on cholesteryl ester transfer are shown. f, effects of U-617 alone; Ⅺ, effects of U-617 ϩ Fab. The CETP source (100 l of cynomolgus monkey LPDP), Fab fragments (1 M final concentration), and assay buffer were mixed and incubated at 37°C for 2 h, after which the radiolabeled LDL, HDL, and the U-617 were added. The latter mixture was then incubated for an additional 3 h at 37°C. effect on either triglyceride or cholesteryl ester transfer in vitro, and in these studies no effect on total or HDL triglyceride or cholesterol concentrations was detected. That indicates that the change in HDL triglycerides evident in Table II was due to inhibition of CETP-mediated triglyceride transfer by CMTP-2.
To evaluate more carefully the effects of CMTP-2 administration on the structure of the HDL, three different monkeys were given 30 mg each of the antibody subcutaneously, and a blood sample taken 48 h later for HDL composition analysis. The results of those analyses are contained in Table III and show that treatment with the antibody for 48 h resulted in an HDL that was virtually devoid of triglyceride, but was otherwise unaffected. That was taken as evidence that triglyceride transfer was inhibited well out of proportion to cholesteryl ester transfer, and therefore, that the antibody also uncoupled cholesteryl ester and triglyceride transfer in vivo. It also indicated that essentially all of the triglyceride in the HDL of the cynomolgus monkey was transferred there from other triglyc-eride-rich lipoproteins (e.g. very low density lipoprotein and chylomicrons) by CETP. DISCUSSION The studies reported here demonstrate that the CETP-mediated transfer of cholesteryl ester can be inhibited without effect on triglyceride transfer and, conversely, triglyceride transfer can be inhibited without effect on cholesteryl ester transfer. From this we conclude that the two lipids are not transferred by a single, non-discriminatory process. Given that the same protein is responsible for the transfer of both types of neutral lipid, the question becomes, how is that discrimination possible? We would propose, based on currently accepted models of the transfer process (2, 3), that two specific, nonoverlapping binding sites are involved in the transfer of the two classes of lipid.
An underlying assumption here is that both of the inhibitors presented in this paper block transfer by interacting directly with CETP, and not by altering the donor or acceptor. In the case of U-617, we base that deduction on previous studies 2 which showed that U-617 is a competitive inhibitor which probably reacts with an unpaired cysteine in CETP. In the case of CMTP-2, we know that the antibody recognizes an epitope on CETP, since it was obtained by screening hybridomas using a direct binding assay. Finally, if either or both of the current models of the transfer process (2, 3) are correct, then the inhibitors must be interacting directly with the protein, for it would not be possible to selectively inhibit transfer of one or the other lipid by blocking the docking, dissociation, or diffusion steps.
To confirm that the selective inhibition detected using the in vitro assay was, in fact, real, and not an artifact of the assay, we evaluated the effect of the inhibitor of triglyceride transfer in vivo. The reasoning was that if selective inhibition is occurring, it should manifest itself as a selective change in the lipoprotein composition. For example, previous studies (25)(26)(27)(28)(29)(30) have shown that inhibition of cholesteryl ester transfer causes an increase in the plasma HDL-cholesterol levels. If, as indicated by the in vitro assay, CMTP-2 at high concentrations inhibits triglyceride transfer but not cholesteryl ester transfer, in vivo administration of the antibody should change the HDL triglyceride content with no effect on the HDL cholesterol levels. That is precisely what occurred: the triglyceride content of HDL was reduced more than 10-fold by antibody treatment while the cholesterol content of HDL remained essentially unchanged. Thus, CETP-mediated cholesteryl ester and triglyceride transfer can be uncoupled in vivo, in full agreement with the two-site hypothesis.
Others (16,17,31) have reported that CETP-mediated triglyceride transfer could be selectively inhibited, but it has FIG. 6. Evidence for disproportionate inhibition of triglyceride transfer by CMTP-2 in vivo. Thirty mg of the indicated antibody was administered subcutaneously to each of four cynomolgus monkeys and blood samples were taken at the indicated times. Plasma (100 l) from those blood samples was then used as the source of CETP in the standard assay and the ratio of the triglyceride and cholesteryl ester transfer activities determined in each plasma sample. Those ratios were normalized (i.e. divided by the ratio in the t ϭ 0 plasma sample) to make visual comparison easier. Ⅺ, plasma samples from monkeys that received CMTP-1; f, plasma samples from monkeys that received CMTP-2.

TABLE III
Composition of HDL before and after administration of monoclonal antibody CMTP-2 to cynomolgus monkeys CMTP-2 (30 mg) was administered subcutaneously to 3 male cynomolgus monkeys and blood samples for HDL isolation collected 48 h later (pretreatment blood samples were taken 1 week before antibody injection). The mean (ϮS.D.) Stokes' radius of the HDL was determined as described previously (22) and was 49.5 Ϯ 0.1 Å. No effect of CMTP-2 treatment on Stokes' radius was detected. Neither was there any effect of antibody treatment on the mean HDL density (1.123 Ϯ 0.006 g/ml; calculated as described by Schumaker (35)  been argued (32) that, because triglycerides have a larger molecular volume (1600 Å 3 ) than cholesteryl esters (1090 Å 3 ), a given inhibitor could partially obstruct access to a single binding site such that the cholesteryl esters could exchange but not the triglycerides. However, in these and previous studies, 2 U-617 was found to inhibit transfer of the smaller molecule but not that of the larger. Therefore, that explanation is not adequate in this instance. On the other hand, by postulating the existence of two neutral lipid binding sites, one can invoke the same principle to explain the selectivity of U-617 for cholesteryl esters; i.e. that a second, but smaller, neutral lipid binding site exists on CETP which is accessible to cholesteryl esters, but not triglycerides, and with which U-617 interacts specifically. If discrete binding sites exist, and each of the inhibitors used here is, in fact, selective for a distinct site, then the binding of one inhibitor should not reduce the effectiveness of the other. Incubation of CETP with CMTP-2 Fab fragments completely blocks triglyceride transfer, but does not interfere with the ability of U-617 to inhibit cholesteryl ester transfer. Thus, the binding sites of the two inhibitors do not overlap; a conclusion in full agreement with the two binding site hypothesis.
Finally, the two-binding site hypothesis is entirely compatible with the observation that CMTP-2 inhibition of cholesteryl ester transfer was reversible beyond a certain antibody concentration, but triglyceride transfer was not. We would propose that the site principally responsible for triglyceride transfer is the epitope recognized by the antibody (i.e. blocked by attachment of the antigen binding region of the antibody) and that the cholesteryl ester binding site is a separate site which does not react with the antibody, but is located such that access to it is hindered by a second CETP molecule bound to the same antibody ( Fig. 2A, icon a). Thus, at low antibody concentration (below 16 nM; Fig. 2A), both antigen binding sites are occupied and cholesteryl ester transfer is inhibited; however, as the antibody concentration increases beyond 16 nM and the unsaturated form of the antibody ( Fig. 2A, icon b) becomes the predominant species, access to the cholesteryl ester transfer site is restored and inhibition of cholesteryl ester transfer is relieved. The fact that CMTP-2 Fab fragments had no observable effect on cholesteryl ester transfer is taken as support for this explanation of the reversible transfer, and therefore, for the premise that topologically distinct, neutral lipid binding sites exist on cynomolgus monkey CETP.
If CETP does, in fact, contain a distinct binding site for each of the two substrates, then there would be the possibility that genetic drift could alter the protein such that the ratio of the specificities toward the two substrates would be different for different species. Such species to species differences should be less evident in the case of a single, non-discriminatory binding site. Comparing CETP activity from three species and using identical assay conditions for measuring those activities, we do, indeed, find substantial species-dependent differences in the substrate specificities. As shown in Table IV, the ratio of transfer rates, V CE /V TG , ranges from 0.76 for hamster CETP to 7.5 for the human CETP. Thus, substrate specificity is an intrinsic property of the protein and that specificity can vary consider-ably among species, suggesting divergent adaptation of the two sites.
There is abundant precedent for the existence of multiple lipid binding sites on a single protein. Probably the best example of this is plasma albumin, which has two major binding domains (designated subdomains IIA and IIIA (33)), and possibly multiple lipid binding sites within a subdomain (34). The experiments described here suggest that two or more neutral lipid binding sites may also exist on cynomolgus monkey CETP, but offer no information as to whether these represent distinct structural domains, or separate binding sites within a single domain. They do, however, suggest that the binding sites are dissimilar, since one appears to have a higher affinity for triglycerides and the other a higher affinity for cholesteryl esters.
Morton and Zilversmit (16) have argued against the two-site hypothesis, based largely on the observation that increasing the cholesteryl ester concentration in the donor particle decreased the rate of transfer of triglycerides (Ref. 16 , Table II) which they interpreted as indication that the two lipids were competing for a single site. However, since increasing the fraction of cholesteryl esters in the lipid core automatically decreased that of the triglycerides, these observations may simply reflect the fact that the transfer rate of each substrate depends on the concentration of that substrate in the donor, i.e. the triglyceride transfer rate may have decreased, not because of competition for the binding site, but simply because the triglyceride concentration in the donor was reduced. Thus, these data do not prove that these two lipids compete with each other for a single binding site. Nonetheless, we would agree with Morton and Zilversmit (16) that the two lipids are probably mutually exclusive (binding at one site precludes binding at the other), for otherwise it is difficult to explain the "net mass transfer" phenomenon. Mutual exclusivity could occur as a result of conformational changes induced in one site when the other is occupied; or, because of their spatial relationship, CETP may simply be unable to present both sites to the lipid donor simultaneously. In the latter case, one would have to hypothesize that, once either site becomes occupied, CETP dissociates, so that only 1 mol of a given lipid would be transferred at a time, otherwise net mass transfer could not occur.
The single most compelling evidence that two sites exist is the observation that exchange of either lipid can continue, even though transfer of the other is blocked. Several studies (16,17,24,31,37) 2 have demonstrated that that can occur. Ko et al. (24), for example, produced a monoclonal antibody against rabbit CETP that had an inhibitory pattern which was very similar to that observed with CMTP-2, i.e. at high antibody concentrations triglyceride transfer was completely blocked, but cholesteryl esters continued to exchange. Despite these results, those authors proposed that there was but one lipid binding site, and invoked the steric hindrance hypothesis (discussed above) to explain their observations. However, this seems unlikely given the observations that compounds such as U-617 and PD 140195 (37) can selectively inhibit transfer of the smaller of the two lipids. Ko et al. (24) also observed that, even though cholesteryl ester exchange continued after triglyceride exchange was blocked, net mass transfer of cholesteryl ester appeared to have been blocked. Thus, by blocking one site, one apparently renders CETP incapable of net mass transfer.
Although no direct evidence exists for the one or two lipid binding sites on CETP, many of the observations reported to date are difficult to reconcile with a single binding site, whereas all observations are consistent with the two-site hypothesis. The question of one or two sites is more than academic and leads to the physiologically important question of whether cholesteryl ester and triglyceride binding and transfer are mutually exclusive, or can a ternary CETP-CE-triglyceride complex be formed? We feel that the mutual exculsion of the two lipids from CETP is probably real but is due to the fact that CETP is unable to present both binding sites simultaneously to the lipid donor.