Amino acid residue 149 of lecithin:cholesterol acyltransferase determines phospholipase A2 and transacylase fatty acyl specificity.

Human LCAT prefers phosphatidylcholine (PC) with sn-1-palmitoyl-2-oleoyl PC (POPC) as substrate for cholesteryl ester synthesis, whereas rat LCAT (which is 92% similar in amino acid sequence) prefers sn-1-palmitoyl-2-arachidonoyl PC (PAPC). Six recombinant human LCAT cDNA clones were constructed with unique clusters of rat sequence substitutions in the human background spanning the region encoding amino acids 121-296. Media from transfected COS cells expressing each of the constructs were assayed for LCAT cholesterol esterification (CE) or phospholipase A2 (PLA2) activity using substrate particles containing POPC or PAPC. The PAPC/POPC CE activity ratio of the cluster 1 construct (amino acids 149-158) was 1.3, resembling rat LCAT, whereas cluster 2-5 clones produced CE activity ratios <0.3, unchanged from human LCAT. The cluster 6 clone (Y292H/W294F) had an intermediate ratio (0.6). Similar results were observed for LCAT PLA2 activity. In additional studies, position 149 of human LCAT was changed to the rat sequence (hE149A) and compared to a triple mutation containing the remainder of the cluster 1 changes (G151R/E154D/R158Q). CE and PLA2 activity ratio for the hE149A construct was >1.7, similar to rat LCAT, whereas the triple mutation construct retained a ratio similar to human LCAT (<0.6). Thus, a single amino acid substitution (E149A) was sufficient to alter the fatty acyl specificity of human LCAT to that of rat LCAT, with an increase in activity toward PAPC. This is the first example of a point mutation in an enzyme with PLA2 activity that results in an increase in activity toward arachidonic acid.


Human LCAT prefers phosphatidylcholine (PC) with sn-1-palmitoyl-2-oleoyl PC (POPC) as substrate for cholesteryl ester synthesis, whereas rat LCAT (which is 92% similar in amino acid sequence) prefers sn-1-palmitoyl-2-arachidonoyl PC (PAPC). Six recombinant human
LCAT cDNA clones were constructed with unique clusters of rat sequence substitutions in the human background spanning the region encoding amino acids 121-296. Media from transfected COS cells expressing each of the constructs were assayed for LCAT cholesterol esterification (CE) or phospholipase A 2 (PLA 2 ) activity using substrate particles containing POPC or PAPC. The PAPC/POPC CE activity ratio of the cluster 1 construct (amino acids 149 -158) was 1.3, resembling rat LCAT, whereas cluster 2-5 clones produced CE activity ratios <0.3, unchanged from human LCAT. The cluster 6 clone (Y292H/W294F) had an intermediate ratio (0.6). Similar results were observed for LCAT PLA 2 activity. In additional studies, position 149 of human LCAT was changed to the rat sequence (hE149A) and compared to a triple mutation containing the remainder of the cluster 1 changes (G151R/E154D/R158Q). CE and PLA 2 activity ratio for the hE149A construct was >1.7, similar to rat LCAT, whereas the triple mutation construct retained a ratio similar to human LCAT (<0.6). Thus, a single amino acid substitution (E149A) was sufficient to alter the fatty acyl specificity of human LCAT to that of rat LCAT, with an increase in activity toward PAPC. This is the first example of a point mutation in an enzyme with PLA 2 activity that results in an increase in activity toward arachidonic acid.
Lecithin:cholesterol acyltransferase (LCAT 1 ; EC 2.3.1.43) is a 67-kDa glycoprotein that is responsible for cholesterol ester-ification in plasma (1). The enzyme displays two activities: a phospholipase A 2 (PLA 2 ) activity, which hydrolyzes the fatty acyl group from the sn-2 position of phosphatidylcholine (PC); and a transacylase activity, which catalyzes the transfer of the fatty acyl group from the acyl-enzyme complex to the 3-␤ hydroxyl group of cholesterol to form cholesteryl ester (CE) (2). The reaction is activated by apolipoprotein A-I (apoA-I), the major apolipoprotein of high density lipoproteins (HDL) (3).
LCAT has an important physiological role in the maturation of nascent, discoidal HDL to mature, spherical HDL, with the generation of a CE-enriched core (4,5). The enzyme has also been implicated in the reverse cholesterol transport pathway, which results in the net transport of cholesterol from peripheral tissues back to the liver for excretion (1,6). The central role of LCAT in these physiological processes is supported by the finding of plasma and tissue accumulation of free cholesterol and the presence of nascent HDL particles in plasma in LCAT-deficient states (7,8).
The enzymatic activity of LCAT is affected by the fatty acyl composition of HDL PC. When plasma HDL PC become enriched in long chain polyunsaturated fatty acids (i.e. carbon chain length Ͼ18 and number of double bonds Ͼ2) by dietary fat modification, the reactivity of LCAT is reduced (9,10). This effect can be reproduced using recombinant HDL (rHDL) as substrate particles for the LCAT reaction. rHDL are discoidal particles consisting of PC, cholesterol and apoA-I that are made by cholate dialysis (11,12). rHDL with PC species containing long chain n-3 or n-6 polyunsaturated fatty acids in the sn-2 position are less reactive than those containing 18:1 or 18:2 (13). The observed decrease in activity may be related to the molecular size of the PC substrate molecule. Using a monolayer apparatus, Parks et al. (13) observed an inverse relationship between the PLA 2 activity of human plasma LCAT and the mean molecular area of the PC substrate molecules. Pownall et al. (14) also demonstrated that a bulky substrate such as diphytanoyl PC reacted poorly with human LCAT. These studies suggest that PC substrate molecular size, which can be influenced by fatty acyl composition, is an important determinant of enzyme activity.
There is also a phylogenetic difference in LCAT reactivity. Portman and Sugano (15) were the first to demonstrate that the fatty acyl preference of rat plasma LCAT was different from that of human plasma LCAT. Subsequent studies with purified enzyme showed that rat LCAT was more reactive with PC substrates containing arachidonic acid compared to human LCAT (16 -19). This difference occurs in spite of a 87% identity between the rat and human LCAT protein sequence (20,21). The molecular explanation for the difference in reactivity is * This work was supported by National Institutes of Health Grants HL 49373 (to J. S. P.), HL 54176 (to J. S. P.), and HL 47864 (to R. A. A.) and by Grant-in-aid NC-95-SA-49 (to J. W.) from the North Carolina affiliate of the American Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The unknown. However, a recent study by Subbaiah et al. (22) using chimeric constructs of human and mouse LCAT cDNA suggested that the middle third of the LCAT protein (amino acids 130 -306) was responsible for conferring fatty acyl specificity. These authors have also reported that the positional specificity of human versus rodent LCAT is different; according to their data, human LCAT switches from using a sn-2 fatty acid to a sn-1 fatty acid when the PC substrate contains arachidonic acid in the sn-2 position (22)(23)(24). However, under the same experimental conditions, neither rat nor mouse LCAT display a change in positional specificity. Thus, little is known regarding the region of LCAT that is responsible for controlling fatty acyl substrate preference or positional specificity.
The purpose of the present study was to define the minimal region of LCAT necessary to confer fatty acyl specificity. We hypothesized that a discrete region of primary sequence is responsible for the fatty acyl specificity of LCAT. We used a comparative species approach that focused on the amino acid differences between human and rat LCAT over the middle third of the enzyme. Our results demonstrate that amino acid 149 is critical in determining the fatty acyl specificity of cholesteryl esterification and PLA 2 activity by LCAT.

EXPERIMENTAL PROCEDURES
Construction of Human-Rat-Human Chimeric LCAT-Rat LCAT cDNA was produced by reverse transcription of total hepatic RNA using a kit from Promega Co. (Madison, WI) followed by PCR amplification of the LCAT cDNA. The PCR reactions (100 l) consisted of 2 l of reverse transcription products as template, 1 M each of 5Ј sense and 3Ј antisense primers, 0.2 mM dNTPs, 10% Me 2 SO, 2 mM MgSO 4 , 1 unit of Vent R DNA polymerase (New England Biolabs, Beverly, MA), and reaction buffer. The primer oligonucleotides were designed from sequences published for rat LCAT (21) with the addition of an EcoRI site at the end of the 5Ј sense primer (5Ј-TAGAA TTCTG GGCTG TAATG GGGCT GCCTG-3Ј) and a HindIII site at the end of the 3Ј antisense primer (5Ј-CGTCA AGCTT AACAG TAAGT CTTTA TTC-3Ј). Amplification protocol involved heating to 94°C for 5 min, followed by 35 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min, and a final extension at 70°C for 10 min. The 1.3-kilobase pair product was isolated by agarose gel electrophoresis and purified on a Wizard™ PCR minicolumn (Promega Co., Madison, WI). A second round of amplification was then performed using the purified PCR product in an identical manner except for deleting Me 2 SO from the PCR buffer. The rat LCAT cDNA and pCMV5 expression vector (25) were digested with EcoRI and HindIII, and the purified products were ligated to generate pCMV.rL-CAT. The rat LCAT cDNA insert was sequenced by the dideoxy method using a Sequenase ® kit (U. S. Biochemical Co.).
A chimeric human-rat-human (HRH) LCAT was constructed by replacing the middle region of the human LCAT cDNA from the 5Ј KpnI site to the 3Ј PstI site with the corresponding region of rat LCAT cDNA; this region encodes a mature LCAT protein sequence from amino acids 120 to 296. A rat LCAT cDNA fragment of the middle region was obtained by PCR amplification using 0.04 pmol of pCMV.rLCAT as a template. A sense oligonucleotide primer (5Ј-TAGCAGGGTACCTGAA-CACAC-3Ј), which contained the KpnI site at the 5Ј end, and an antisense primer (5Ј-CGAGACTGCAGAAACATGTGC-3Ј), which contained an engineered PstI site, were used for PCR amplification. The PCR product band (527 base pairs) was purified by agarose electrophoresis and PCR minicolumn procedures and subcloned into the KpnI/PstIdigested pCMV.hLCAT. The sequence of rat LCAT cDNA insert was confirmed by the dideoxy method.
Site-directed Mutagenesis-Site-directed mutagenesis of hLCAT cDNA was accomplished by a megaprimer PCR procedure. A cluster (CL) mutagenesis strategy was used to introduce three to six point mutations within a region of the LCAT cDNA where differences between the human and rat LCAT amino acid sequence were clustered (Fig. 1). The first round of PCR included a 5Ј sense PCR primer with an EcoRI restriction site and a mutagenic antisense primer for H168Y and CL1-CL3, or a 3Ј antisense PCR primer with a BamHI site and sense mutagenic primers CL3-CL6. The sequences of mutagenic primers were as follows: CL1, 5Ј-CCTGC GAGCT TCTGG TAGTA CTCGT CCTGC TGGCG GGGCG CCAGC CGCC-3Ј; CL2, 5Ј-CCAGG ACTGG GGCTG GCGCA GCAGG AAATG GAGCA CGTG-3Ј; CL3, 5Ј-GCTTG ATGTT GGACA TGATG GGGAT CCGCT GGTTG TCACC TGAGG CCAAG ATCCG CATGG GCTTG-3Ј; CL4, 5Ј-TTCCC GCTCA CCATG TGTGG CCTGA G-3Ј; CL5, 5Ј-CCACA CCCAA CTTCA ACTAC ACAGG CCAGG ACTTC GAACG-3Ј; CL6, 5Ј-GAGGA AGGCT GGCAC ATGTT CCTGC AG-3Ј; and H168Y, 5Ј-AGGCA GCGTA CATCT CCTCC-3Ј. The PCR conditions were similar to those described above except 0.04 pmol of pUC.LCAT plasmid (kindly provided by Dr. John McLean, Genentech Inc., San Francisco, CA) was used as the template, the annealing temperature ranged from 45 to 50°C depending on the primer, annealing times were 0.5 min, and 20 cycles were used for amplification. For the second round of PCR, the ϳ500 -700-base pair product from the first round of PCR was agarose gel-isolated and PCR prep minicolumnpurified before use as a megaprimer with the appropriate sense or antisense primer. The thermocycling procedure was similar to the first round of PCR except the annealing temperature was raised to 70°C, and the time for the extension step was increased to 1.5 min. The PCR products were purified by a PCR minicolumn, digested with EcoRI and BamHI, and ligated into the pCMV5 expression vector. The entire coding sequence of the mutant constructs was confirmed by dideoxy sequencing.
In Vitro Expression of LCAT-Two g of pCMV.LCAT cDNA constructs were transiently transfected, using a DEAE-dextran method (26), into COS-1 cells grown in 35-mm tissue culture dishes. After transfection, the cells were washed three times with phosphate-buffered saline, switched to serum-free Dulbecco's modified Eagle's medium, and incubated an additional 72 h at 37°C. The medium then was collected, centrifuged at 500 ϫ g for 10 min, and immediately frozen at Ϫ70°C until assays were performed.
Assays for CE Formation and PLA 2 Activity by LCAT-LCAT incubations were performed in triplicate in 0.5 ml of buffer (10 mM Tris, 140 mM NaCl, 0.01% EDTA, 0.01% NaN 3 , pH 7.4) containing: rHDL (0.2 g of cholesterol for CE formation or 25-50 g of PC for PLA 2 activity), 0.6% bovine serum albumin (fatty acid-free; Sigma), 2 mM ␤-mercaptoethanol, and 20 -150 l of media or 5 l of plasma as a source of LCAT enzyme. Control incubations contained all constituents except the LCAT source. Tubes were gassed with N 2 and incubated at 37°C for 1 h for all PLA 2 reactions and for the CE formation from media samples. In the case of plasma samples used for measurement of CE formation, an incubation time of 10 min was used. Control experiments showed that CE formation was linear up to 2 h using our experimental conditions, provided CE formation was Ͻ25%. The reaction was stopped by Bligh-Dyer extraction solution (31) consisting of 1:2 chloroform:methanol with 25 g/ml added cholesterol and cholesteryl oleate (CE formation reactions) or 40 g/ml oleic acid (PLA 2 reactions). The phases were split by the addition of 0.29% NaCl (CE formation) or 0.05% H 2 SO 4 (PLA 2 activity). The radiolabeled products were separated by thin-layer chromatography and quantified by liquid scintillation counting (9).
HPLC was used to separate radiolabeled CE formed by LCAT using a C18 reverse phase column as described previously (32). Fractions were analyzed by liquid scintillation counting to monitor the elution profile of the radiolabeled CE.
ELISA for Quantification of LCAT Mass-Media samples were assayed in triplicate for LCAT mass using an ELISA procedure. A series of media volumes (2.5-10 l) was adjusted to 5 mM ␤-mercaptoethanol in Tris-buffered saline (200 l final volume) and incubated at 37°C for 2 h in microtiter plates. Purified human plasma LCAT (33) was used as standard for the assay. The wells were washed with water, blocked for 2 h at room temperature with 0.05% Tween 20 and 0.25% bovine serum albumin in Tris-buffered saline, and washed again. A 1:1000 dilution of anti-human LCAT chicken IgY in blocking buffer was applied to each well and was incubated at room temperature for 2 h. The IgY was purified from the eggs of chickens immunized with purified human plasma LCAT (34). The plate was washed, incubated for 2 h at room temperature with rabbit anti-chicken IgY conjugated with alkaline phosphatase (1:1000 dilution; Sigma), developed with p-nitrophenyl phosphate substrate (Sigma) and read at 405 nm after color development. The standard curve was linear from 1 to 30 ng/well and the intra-assay coefficient of variation was 14 -16%.
Data Analysis-DNA and protein sequence comparisons were made using the Genetics Computer Group (Madison, WI) and DNA Strider™ 1.2 (Christian Marck, Gif-Sur-Yvette, France) software. All data are presented as mean Ϯ standard error.

RESULTS
Our mutagenesis strategy was based on a comparative species approach that focused on differences in amino acid sequence between rat and human LCAT in the middle third of the protein. A comparison of the human and rat sequence is shown in Fig. 1. Overall, the two sequences share an 87% identity with 9 differences located between amino acids 1 and 120, 22 differences between amino acids 121 and 300, and 22 differences between 301 and 416. The differences in sequence from amino acids 121-300 could be roughly clustered into six regions with additional point differences as indicated in Fig. 1. Our experimental approach involved the mutation of all amino acids within these regions or clusters from the human sequence to that of the rat sequence and then in vitro expression of the mutant forms of LCAT for enzymatic characterization.
To determine whether the middle third of rat LCAT was responsible for conferring fatty acyl specificity as previously observed for mouse LCAT (22), we generated a HRH chimeric LCAT construct, in which the human sequence was exchanged for the rat sequence from amino acid residues 121-296. The HRH construct was transiently expressed in COS cells, and the medium was assayed for CE formation activity using rHDL containing POPC or PAPC. The results are shown in Table I. Human recombinant LCAT had one third the reactivity with rHDL containing PAPC compared to POPC. This resulted in a PAPC/POPC activity ratio of 0.33. Recombinant rat LCAT was equally reactive with both substrates, resulting in an activity ratio of 0.97. The HRH chimeric LCAT displayed a reactivity profile more similar to that of rat recombinant LCAT than human recombinant LCAT, resulting in an activity ratio of 1.21. Thus, amino acids 121-296 of the rat LCAT sequence were sufficient to confer the fatty acyl preference of rat LCAT for arachidonate, similar to the previous findings for mouse LCAT (22).
To better define the region of LCAT responsible for conferring fatty acyl specificity, we generated and transiently expressed a series of cDNA constructs that contained 2-4 amino acid substitutions of the rat sequence for the human sequence at six different sites or clusters over the middle region of the LCAT protein (see Fig. 1). The results of the cholesterol esterification assay are shown in Table II. Several controls are shown including a recombinant LCAT in which the active site serine has been mutated to alanine (S181A), human and rat wild-type LCAT, and human and rat plasma as sources of the enzyme. The activity of all of the recombinant LCAT proteins with rHDL containing POPC was similar; however, on average the values for CL6, HRH, and rat LCAT were lower than the others. The values for human and rat plasma were also similar to each other and, as expected, the S181A mutation eliminated enzymatic activity. The results with rHDL containing PAPC were somewhat different. CL1 and rat LCAT had activities that  1. Multiple sequence alignment of human and rat LCAT cDNA. GenBank™ sequences used for the alignment were M12625 (human; submitted 1986) and U62803 (rat; submitted 1996). Overall, there was 87% identity and 93% similarity between the two sequences. Differences in sequence are shown in bold type. Numbers over lines indicate the regions or clusters of the protein in which the human sequence was mutated to that of the rat sequence (see "Results").
were severalfold higher than those of the other recombinant LCAT proteins. In addition, the activity with rat plasma was 4-fold higher than that of human plasma.
The PAPC/POPC activity ratio is also shown in Table II. The CL1 activity ratio (1.29) is similar to that of HRH (1.20) and rat LCAT (0.97), whereas CL2-CL5 and the point mutation H168Y have activity ratios similar to that of human LCAT (0.32) and human plasma (0.51). Note that CL6 had an activity ratio (0.61) that was 2-fold greater than CL2-CL5, but less than half that of CL1. The activity ratio for human and rat plasma showed the expected 3-fold difference. These data demonstrate that CL1 region was critical in determining fatty acyl specificity for cholesterol esterification of LCAT.
The results of the PLA 2 assay are shown in Table III. The PLA 2 values for all of the recombinant LCAT proteins ranged from 1.6 to 3 nmol/h/g when assayed with POPC rHDL except for HRH, which was noticeably lower. When PAPC rHDL was used as substrate, the PLA 2 activity for CL1, CL6, and rat LCAT, on average, was higher than that of the other recombinant constructs. The PLA 2 activity in rat plasma was 9-fold greater than that of human plasma.
The PAPC/POPC activity ratio for PLA 2 hydrolysis is also shown in Table III. The ratio for CL1 (1.45) was nearly 5-fold greater than that of CL2-CL5 (0.28 -0.35). CL6 had an activity ratio that was 2-fold greater than CL2-CL5 and H168Y. The values for HRH and rat LCAT were similar to that of CL1. The activity ratios for recombinant human LCAT and human plasma were low (0.27-0.33) and similar to CL2-CL5. Thus, the CL1 region of LCAT is involved in determining the fatty acyl specificity of the PLA 2 step of the LCAT reaction.
Having identified CL1 as a critical region in determining fatty acyl specificity, we wished to determine the minimal sequence change necessary in CL1 to confer fatty acyl specificity. We performed a multiple sequence alignment of the CL1 region of LCAT using two primate sequences (human, baboon) and two rodent sequences (rat, mouse). The results are shown in Fig. 2. Our strategy was to identify amino acid residues that were conserved in the rat and mouse but were different in the baboon and human, which might give rise to the fatty acyl specificity difference between rodents and primates. Amino acid 158 was not conserved among the four species, whereas amino acid 154 was glutamic acid in the primates and aspartic acid in the rodents. Position 151 was a glycine in the primates but was not conserved in the rodents. However, position 149 was glutamic acid in the primates and alanine in the rodents. Thus, of the four amino acid variations in the CL1 region of LCAT, position 149 showed the greatest divergence in the primate versus rodent comparison, yet it did not change within the primate and rodent species analyzed. Based on this analysis, we generated a E149A mutation, in which position 149 of the human LCAT sequence was mutated to alanine. We also made a triple mutation, designated CL1_3, in which the amino acids at positions 151, 154, and 158 of the human sequence were changed to the rat sequence (i.e. G151R, E154D, and R158Q). The mutations were generated by megaprimer PCR as described for the cluster mutants and transiently expressed in COS cells. Table IV shows the results of the cholesteryl ester formation assay of the CL1 mutants. With rHDL containing POPC, the activity of rat and hE149A LCAT averaged less than that of human and CL1_3 LCAT. However, with rHDL containing PAPC, the activities of rat and hE149A LCAT were greater than human and CL1_3 LCAT. Fig. 3 shows the PAPC/POPC activity ratio for cholesteryl ester formation derived from the data in Table IV. The ratio for human LCAT (0.51 Ϯ 0.02) agrees closely with that of CL1_3 (0.58 Ϯ 0.03), whereas the ratio for rat LCAT (1.55 Ϯ 0.03) agrees closely with the value for hE149A (1.78 Ϯ 0.02). Thus, a single mutation at position 149 of the human enzyme was sufficient to change the fatty acyl specificity of LCAT cholesterol esterification from the human to the rat pattern. PLA 2 activity was also assayed from the same mutant constructs (Table V). As observed for the cholesterol esterification assay in Table IV, the PLA 2 activities for rat and hE149A   M12625 (1986)), baboon (L08633 (1993)), rat (U62803 (1996)), and mouse (J05154 (1990)) LCAT (GenBank™ accession numbers shown in parentheses) over the cluster 1 region. Differences in sequence among the four are shown in bold type.
LCAT were lower than those of human and CL1_3 LCAT when assayed with POPC, whereas the opposite trend was apparent when PAPC rHDL was used as substrate. Thus, the hE149A construct showed a higher activity for both cholesterol esterification and PLA 2 activity when assayed with rHDL containing PAPC compared to human and CL1_3 LCAT.
The PAPC/POPC ratio for PLA 2 activity is shown in Fig. 4 for the hE149A and CL1_3 mutant constructs. There was a 6-fold greater ratio for the hE149A recombinant LCAT (3.24 Ϯ 0.50) compared to human (0.47 Ϯ 0.07) or CL1_3 (0.47 Ϯ 0.14) LCAT. The ratio for rLCAT (3.19 Ϯ 0.13) was similar to that of hE149A LCAT. The overall pattern of the PLA 2 activity ratios (Fig. 4) was similar to that of the cholesterol ester formation ratios (Fig. 3).
Reports in the literature have described a PC positional specificity switch from the sn-2 to the sn-1 position when human LCAT encounters a polyunsaturated fatty acid in the sn-2 position of PC (22)(23)(24). To investigate this phenomenon with our LCAT mutant constructs, we separated the CE products formed by LCAT using HPLC. The results of a typical experiment are shown in Fig. 5. Only one major CE product was formed, regardless of the source of LCAT enzyme or whether the substrate PC was POPC or PAPC. The human and CL1_3 LCAT (Fig. 5, left panels) synthesize more cholesteryl oleate product with POPC rHDL than cholesteryl arachidonate product with PAPC rHDL. However, for rat and hE149A LCAT, the amount of cholesteryl arachidonate made with PAPC was similar to or greater than the amount of cholesteryl oleate made from POPC (Fig. 5, right panels). Although some cholesteryl palmitate was made in each reaction (peak at 24 -25 min), the absolute amount was small and about the same regardless of the source of enzyme. This amount of cholesteryl palmitate likely results from a small amount of palmitic acid contamination in the sn-2 position of the PC preparations (typically 5-7% based on snake venom PLA 2 digestion studies) rather than a utilization switch from the sn-2 to the sn-1 position. We also performed HPLC separations of CE from LCAT incubations using the CL1-CL6 mutants (data not shown). The elution profiles for CL1 and CL6 were similar to that of hE149A in Fig.  5, whereas the profiles for CL2-CL5 resembled that of CL1_3 in Fig. 5.

DISCUSSION
The species-specific fatty acyl preference of LCAT was first reported by Portman and Sugano (15) in 1964. Later reports confirmed that human LCAT has a higher reactivity with PC species containing oleic or linoleic acid in the sn-2 position compared to arachidonic acid, whereas rat LCAT displayed the opposite preference profile (16 -19). However, the molecular explanation for this well documented fatty acyl specificity difference between rat and human LCAT was not established. We hypothesized that a discrete region of primary sequence was responsible for the fatty acyl specificity of the enzyme, and the present study was undertaken to define the minimal region of LCAT necessary to confer fatty acyl preference. Using the previous observation of Subbaiah et al. (22), indicating that amino acid residues 130 -306 were involved in determining the fatty acyl preference of LCAT and a comparative species approach that focused on the amino acid differences between human and rat LCAT over this region, we identified a single amino acid residue in the human enzyme, the glutamic acid at position 149, that is responsible for conferring fatty acyl specificity. When E149 of human LCAT was mutated to an alanine, which is present in the rat sequence at position 149, the fatty   acyl specificity of the human enzyme was converted to that of rat LCAT, with no loss in reactivity toward rHDL containing POPC and an increased reactivity toward rHDL containing PAPC. The change in fatty acyl specificity for hE149A recombinant LCAT was observed for both PLA 2 and cholesterol esterification activities and reflected an increased reactivity with PAPC substrate particles compared to those containing POPC. To our knowledge this is the first example of a point mutation of an enzyme that results in such a striking change in fatty acyl specificity.
The molecular explanation for the increased reactivity of hE149A with PAPC rHDL is unknown, but may be related to the size of the active site pocket of the enzyme. Crystallographic studies have shown that pancreatic lipase (35) and secretory PLA 2 molecules (36, 37) contain a hydrophobic pocket in which the PC molecule binds prior to hydrolysis of the sn-2 fatty acid. While LCAT shares a lipase gene family active site motif (GXSXG) at the active site serine 181 and amino acid residues 151-174 form a putative interfacial binding region (6), few structural features of LCAT are known and a hydrophobic pocket for substrate binding has not been identified. A previous study has described an inverse relationship between the molecular surface area of the PC substrate molecules and the PLA 2 hydrolysis rate of PC monolayers by human plasma LCAT (13). These results suggested that the PC molecules with a greater molecular surface area, such as those containing long chain polyunsaturated fatty acids (38,39), might not fit well into the active site pocket of the enzyme. A similar conclusion was reached by Pownall et al. (14), who showed that 1,2-diphytanoyl, which has a larger molecular surface area than POPC, is a competitive inhibitor of human plasma LCAT, presumably because it has difficulty fitting into the active site of the enzyme. We hypothesize that the hE149A mutation results in a larger active site pocket that can more easily accommodate the larger PC molecules, such as PAPC, leading to increased activity with these substrates. Since crystallographic data of secretory PLA 2 molecules demonstrate hydrophobic amino acid residues in the interior of the binding pocket (40), it seems unlikely that the hE149 residue is part of the active site pocket. Rather, it seems more likely that the hE149A mutation results in a conformational change in the enzyme, which increases the size of the binding pocket. A systematic study of the hE149A and human wild-type LCAT enzymes using PC substrates with varying molecular surface areas is currently under way to test whether the size of the active site pocket plays an important role in determining enzyme reactivity.
The importance of size versus charge of the amino acid residue at position 149 of LCAT in conferring fatty acyl specificity is presently unknown. The hE149A mutation, which leads to increased reactivity with PAPC substrate, substitutes a smaller, hydrophobic, uncharged amino acid for glutamic acid in a region of the protein that is predicted to be hydrophilic based on Kyte-Doolittle analysis (41). Of the known LCAT sequences in the GenBank™ data base, the ones with glutamic acid at position 149 (human, baboon, rabbit, and pig) do not show a preference for arachidonic acid (42) compared to those that contain alanine at position 149 (rat and mouse). However, chicken LCAT, which has a fatty acyl specificity similar to human LCAT (42), contains a glycine at amino acid 149. Since glycine is more similar to alanine in size and hydrophobicity than glutamic acid (26), there is no obvious relationship between size and hydrophobicity of the amino acid at position 149 of LCAT and fatty acyl specificity. The effect of substituting different amino acid residues at position 149 on the fatty acyl specificity of LCAT is under active investigation.
An alternative explanation for the increased reactivity of hE149A toward PAPC rHDL could be an increased interfacial binding affinity or capacity compared to the wild-type human LCAT. LCAT must bind to the interface of its lipoprotein substrate particle and be activated by apoA-I before it hydrolyzes the sn-2 fatty acid of PC. Using a solid-phase binding assay or an activity inhibition assay, Bolin and Jonas (43) have shown that the apparent K m for cholesterol esterification of LCAT reflects interfacial binding affinity for the HDL particle surface, whereas the apparent V max for esterification reflects a preference of the molecular substrates (PC and cholesterol) for the active site of the enzyme. Furthermore, they showed that changing the phospholipid head group composition of the substrate HDL influenced the binding affinity of LCAT and the apparent K m and V max for cholesterol esterification. The substitution of alanine for glutamic acid at amino acid 149 could result in a conformational change in LCAT that could increase FIG. 5. HPLC separation of cholesteryl ester products synthesized by LCAT from rHDL containing POPC or PAPC, [ 3 H]cholesterol, and apoA-I. The LCAT reaction was performed and CE products were separated using a Microsorb C18 reverse phase column equilibrated with (1:1) isopropanol: acetonitrile (1 ml/min; 45°C). One-ml fractions were collected except from 20 to 25 min, when 0.5-ml fractions were collected for increased resolution of peaks. Each fraction was analyzed by liquid scintillation counting to monitor the elution profile of the radiolabeled CE. Authentic standards were monitored at 210 nm to identify the elution position of each CE species (FC, free cholesterol; C20:4, cholesteryl arachidonate; C18:1, cholesteryl oleate; C16:0, cholesteryl palmitate). The source of LCAT enzyme is shown at the top of each panel. hE149A represents a point mutation in which amino acid 149 of the human sequence was changed from Glu to Ala. CL1_3 represents a triple mutation in which the following changes to the human LCAT amino acid sequence were made: G151R, E154D, and R158Q. its binding to PAPC. A potential interfacial binding region has been identified (amino acids 151-174; Ref. 6) that is close to residue 149 in the primary sequence. Direct binding studies of human wild-type and hE149A LCAT to POPC and PAPC will be necessary to test this possibility.
Our studies revealed that both PLA 2 and cholesterol esterification activities were increased with hE149A LCAT when incubated with PAPC compared to POPC rHDL. However, the PAPC/POPC activity ratio for hE149A and rat LCAT was 2-fold greater for the PLA 2 assay (Fig. 4) compared to the cholesterol esterification activity (Fig. 3). These results suggest that the fatty acyl specificity of LCAT is determined at the PLA 2 step of the reaction and not at the transacylase step. After hydrolysis of the sn-2 fatty acyl group of PC by LCAT, an acyl-enzyme intermediate is formed with the active site serine 181 (2). The preferred acceptor of the fatty acyl group in the subsequent transacylation step is cholesterol, but water, lysoPC, diacylglycerol, and monoacylglycerol can also act as acceptors of the fatty acyl group (44). Thus, the transacylase step appears to be rather nonspecific with regard to acceptor molecules compared to the high specificity of the LCAT PLA 2 step, which discriminates not only the position of the fatty acyl group in PC (sn-2 versus sn-1) but also the type of fatty acyl group in the sn-2 position.
There are reports that human LCAT switches from using a sn-2 fatty acid to a sn-1 fatty acid when the PC substrate contains arachidonic acid or another long chain polyunsaturated fatty acid in the sn-2 position, whereas the rat enzyme does not (22)(23)(24). We investigated this possibility with our mutants by separating the LCAT synthesized CE species by HPLC. In all cases we detected only one major CE product, cholesteryl oleate or cholesteryl arachidonate for POPC and PAPC rHDL, respectively, regardless of the source of enzyme (Fig. 5). Although some cholesterol palmitate was apparent in each separation, the amount was similar for all incubations and was not dependent on the source of enzyme. This amount of cholesterol palmitate could result from a small amount of contaminating palmitic acid in the sn-2 position of the PC preparations rather than a positional switch of enzyme. Our PLA 2 assay results also support this conclusion. The radiolabeled POPC and PAPC preparations contained 99 Ϯ 0.9% (n ϭ 6) of the radiolabeled fatty acid in the sn-2 position as judged by snake venom PLA 2 digestion studies (9). Thus, the release of radiolabeled fatty acid in the LCAT PLA 2 assays (Tables III  and V) reflected hydrolysis at the sn-2 position and represented sn-2 positional specificity for LCAT. In addition, we have found no evidence for accumulation of radiolabeled lysoPC during our LCAT PLA 2 assays, which would occur if the enzyme were to have a positional switch from sn-2 to sn-1. 2 Therefore, taken together, our data do not support a positional switch of human LCAT or any of the mutant recombinant LCAT proteins generated in this study.
In summary, we have shown that a single amino acid substitution (E149A) is sufficient to alter the fatty acyl specificity of human LCAT to that of rat LCAT, with an increase in activity toward PC substrate containing arachidonic acid. Ar-omatic residues Tyr-292 and Trp-294 may also play a role in determining fatty acyl specificity. Furthermore, the fatty acyl specificity appears to be determined at the PLA 2 step of the reaction, not at the transacylase step. In spite of a striking change in fatty acyl specificity for hE149A LCAT, no evidence was found for a sn-2 to sn-1 positional switch.