(Received for publication, February 10, 1997, and in revised form, March 20, 1997)
From the Department of Medicine, Harvard Medical School, and the Division of Rheumatology, Immunology, and Allergy, Brigham and Women's Hospital, Boston, Massachusetts 02115
The functional characteristics of leukotriene C4 synthase (LTC4S), which specifically conjugates leukotriene A4 with GSH, were assessed by mutagenic analysis. Human LTC4S and the 5-lipoxygenase-activating protein share substantial amino acid identity and predicted secondary structure. The mutation of Arg-51 of LTC4S to Thr or Ile abolishes the enzyme function, whereas the mutation of Arg-51 to His or Lys provides a fully active recombinant protein. The mutations Y59F, Y97F, Y93F, N55A, V49F, and A52S increase the Km of the recombinant microsomal enzyme for GSH. The mutation Y93F also markedly reduces enzyme function and increases the optimum for pH-dependent activity. The deletion of the third hydrophobic domain with the carboxyl terminus abolishes the enzyme activity, and function is restored by the substitution of the third hydrophobic domain and carboxyl terminus of 5-lipoxygenase-activating protein for that of LTC4S. Mutations of C56S and C82V alone or together and the deletion of Lys-2 and Asp-3 of LTC4S do not alter enzyme function. The direct linkage of two LTC4S monomers by a 12-amino acid bridge provides an active dimer, and the same bridging of inactive R51I with a wild-type monomer creates an active pseudo-dimer with function similar to that of the wild-type enzyme. These results suggest that in the catalytic function of LTC4S, Arg-51 probably opens the epoxide ring and Tyr-93 provides the thiolate anion of GSH. Furthermore, the monomer has independent conjugation activity, and dimerization of LTC4S maintains the proper protein structure.
Cysteinyl leukotrienes (LT)1 are smooth muscle contractile lipid mediators with profound biologic effects. They elicit direct bronchial smooth muscle contraction in many species including humans (1, 2) and increase vascular permeability through endothelial cell contraction at the post capillary venules (3, 4). The pathobiologic role of the cysteinyl leukotrienes in patients with bronchial asthma is established by the measured benefit observed after treatment with inhibitors of the biosynthetic pathway or receptor antagonists (5, 6).
The formation of LTC4 is initiated by cell activation with release of arachidonic acid from membrane phospholipid by the action of cytosolic phospholipase A2 (7). The released arachidonic acid binds to an integral perinuclear membrane protein, the 5-lipoxygenase-activating protein (FLAP), and is presented to 5-lipoxygenase (8-10). 5-Lipoxygenase metabolizes arachidonic acid in two sequential steps to form 5-hydroperoxy-eicosatetraenoic acid and then the epoxide intermediate LTA4 (11). The integral perinuclear membrane LTC4 synthase (LTC4S) conjugates LTA4 with reduced glutathione (GSH) to form the intracellular product, LTC4 (12, 13). After the carrier-mediated export of LTC4 (14, 15), the sequential cleavage of glutamic acid and glycine provides the extracellular, receptor-active derivatives, leukotrienes D4 and E4 (16, 17), respectively.
The cloning of the cDNA for human LTC4S from human leukemic cell libraries revealed that the nucleotide and deduced amino acid sequences had no significant homology with the GSH S-transferases (18, 19). Instead there was 31% amino acid sequence identity with human FLAP, which increased to 44% for the amino-terminal two-thirds of these proteins (amino acids 5-99 for LTC4 synthase and 9-103 for FLAP). Furthermore, these proteins had a related functional domain in that the inhibitors of the arachidonic acid-binding domain of FLAP (20, 21), termed FLAP inhibitors, suppressed the conjugation by LTC4S of LTA4 with GSH. There was also an apparent structural homology in that the deduced amino acid sequence for each protein predicted three putative hydrophobic transmembrane domains with two intervening hydrophilic loops of essentially the same size.
The deduced amino acid sequence of mouse LTC4S displayed 88% overall homology with the human enzyme, with only 9 of the 18 amino acid residues of the carboxyl terminus being identical (22). That the purified recombinant mouse LTC4S exhibited kinetics nearly identical with those of the purified recombinant human enzyme suggested that the carboxyl terminus of LTC4S is not critical for the conjugation function of the enzyme (22).
On the basis of mutagenic analyses by point mutations, deletions, and substitutions, we have identified a number of the functional characteristics of LTC4 synthase. Arg-51 in the first hydrophilic loop is required for function, most likely the opening of the epoxide ring of LTA4; and the activation of GSH to the thiolate anion is attributed to Tyr-93 by the shift in pH optimum with mutation. A number of residues, particularly tyrosines, are implicated in GSH binding as demonstrated by the increase in Km with mutation. The second and third hydrophobic domain tolerate partial and complete replacement, respectively, by the corresponding domains of FLAP, but they cannot be deleted. Finally, that a pseudo-dimer created by bridging inactive R51I and wild-type LTC4S with a 12-amino acid linker has function and a Km identical to those of the wild-type enzyme reveals that dimerization provides the protein folding essential to catalytic function.
COS-7 cells (American Type Culture Collection, Rockville, MD); Nu Serum Plus (Collaborative Research, New Bedford, MA); Taq polymerase and dNTP (Pharmacia Biotech, Uppsala, Sweden); MK-886 and FLAP cDNA (Dr. J. Evans, Merck Frosst, Pointe Claire-Dorval, Quebec, Canada); dimethyl sulfoxide (Me2SO), chloroquine, DEAE-dextran (Mr > 500,000), bovine serum albumin, GSH, Bis-Tris propane, fetal calf serum (Sigma); and acetonitrile and methanol (Burdick & Jackson, Muskegon, MI) were obtained as indicated. LTA4-methyl ester (LTA4-ME) (Dr. J. Rokach, Florida Institute of Technology, Melbourne, FL) was synthesized as described (23).
COS-7 cells were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum and 15 µg/ml gentamycin at 37 °C under a humidified atmosphere of 5% CO2 and 95% air.
Site-directed MutagenesisPolymerase chain reaction (PCR)
mutagenesis was carried out to generate the appropriate mutated
cDNAs. Primer I corresponded to nucleotides 1-18 (5 to 3
) of
LTC4S cDNA. Primer II, a 3
primer, corresponded to
nucleotides 597-622 (3
to 5
). Two primary PCR fragments were
produced by using primer I plus the inverse mutagenic primer (3
primer) and primer II plus the forward mutagenic primer (5
primer) for
30 cycles. The two primary PCR fragments were mixed, then denatured at
97 °C for 2 min, annealed at 55 °C for 2 min, and extended at
72 °C for 4 min for one cycle. The secondary PCR products with the
specific mutations were obtained after 30 additional cycles of PCR with
primers I and II.
To obtain hybrid A, a chimeric molecule of LTC4S and FLAP,
30-base mutagenic oligonucleotides were used in which the first 15 nucleotides corresponded to the LTC4S sequence and the
second 15 nucleotides to the FLAP sequence. The first primary PCR
fragment was generated with primer I and the reverse mutagenic
oligonucleotide with LTC4S cDNA as template, and the
second primary PCR fragment was obtained with the forward mutagenic
oligonucleotide and primer III (a 3 oligonucleotide corresponding to
nucleotides 739-710 of FLAP cDNA). To obtain hybrid B and hybrid
C, 6 and 12 PCR reactions with appropriate oligonucleotides,
respectively, were carried out to generate the primary products.
The mutated cDNAs were ligated into the pCR3 expression vector
(Invitrogen, San Diego, CA). The constructed plasmids bearing the
desired mutations were used to transform the Escherichia
coli strain Top 10F by the calcium chloride procedure (24) and
were sequenced to confirm the desired base substitutions and to ensure that no other mutations were introduced into the nucleotide sequence during PCR.
Plasmids were prepared with a Nucleobond isolation kit (Nest, Southboro, MA) or a QIAprep spin column (QIAGEN, Chatsworth, CA) and were sequenced as described by Sanger et al. (25) using dye-labeled dideoxy nucleotides as terminators. The samples were analyzed on an Applied Biosystems model 373A automated DNA sequencer (26).
Transfection and Functional Analysis of the Expressed ProteinsCOS-7 cells were transfected with either wild-type
LTC4S cDNA or the mutated LTC4S cDNA by
DEAE-dextran transfection (18). Three days after transfection, the
cells were harvested by treatment with trypsin and suspended in
microsomal buffer (50 mM HEPES buffer, pH 7.9, containing 2 mM EDTA and 10 mM -mercaptoethanol) at a concentration of 5 × 106 cells/ml. The cells were
lysed by sonication with a microtip sonicator (Branson, Danbury, CT) at
a setting of 7. Cell lysates were centrifuged at 100,000 × g for 60 min at 4 °C, and the pellets (microsomes) were
resuspended in microsomal buffer and solubilized by the addition of
0.4% Triton X-102, 0.4% sodium deoxycholate (DOC), and 10% glycerol
and stirring at 4 °C for 45 min.
The function of each mutant enzyme as well as the wild-type enzyme was first screened by the incubation of 1-10 µl of solubilized COS cell microsomes with 10 mM GSH and 20 µM LTA4-ME in 200 µl of 50 mM HEPES buffer containing 10 mM MgCl2, pH 7.6. The reactions were carried out at room temperature for 10 min and stopped by the addition of three volumes of methanol containing 200 ng of prostaglandin B2 (PGB2) as an internal standard. LTC4-ME levels in the samples were determined with high performance liquid chromatography (HPLC). When a mutated construct did not provide enzyme function, the expression of the mutated protein was confirmed by SDS-PAGE immunoblot analysis with rabbit anti-LTC4S antibody as described (13).
Reverse phase-HPLC was carried out with a model 126 dual pump system and model 167 scanning UV detector (Beckman Instruments) controlled by a Gateway 2000 P4D66 computer equipped with Beckman System Gold software. Samples were applied to a 5-µm 4.6 × 250-mm C18 Ultrasphere reverse phase column (Beckman Instruments) equilibrated with a solvent of methanol/acetonitrile/water/acetic acid (10:15:100:0.2, v/v), pH 6.0 (solvent A). Immediately after the sample was injected, the column was eluted at a flow rate of 1 ml/min with a programmed concave gradient (System Gold curve 6) to 20% solvent A and 80% solvent B (pure methanol) over 0.2 min with continuation isocratically for 5.8 min. Beginning at 6 min, solvent B was increased linearly to 100% over 0.2 min and was maintained at this level for 7 min more. The ultraviolet absorbance at 280 nm and the ultraviolet spectra were recorded simultaneously. The retention times for PGB2 and LTC4-ME were 7.4 and 8.9 min, respectively. LTC4-ME was quantitated by calculating the ratio of the peak area to the area of the internal standard PGB2.
To study the effect of pH on enzymatic conjugation by wild-type and mutant enzyme, 1-µl or 5-µl (Y93F) samples of solubilized COS cell microsomes were incubated with 10 µM LTA4-ME and 10 mM GSH in Bis-Tris propane buffer containing 10 mM MgCl2 at various pH levels. The reactions were carried out at room temperature for 2 min and were stopped by the addition of three volumes of methanol and 10 µl of acetic acid. Product generation was quantitated by reverse phase-HPLC as described above.
Purification and Kinetic Analysis of Selected Expressed ProteinsRecombinant LTC4 synthases were purified by
S-hexyl GSH-agarose affinity chromatography (22).
Solubilized microsomes were loaded onto a 2.5 × 1.5-cm open bed
of S-hexyl GSH resin equilibrated with 50 mM
HEPES, 1 mM EDTA, 5 mM -mercaptoethanol,
0.1% Triton X-102, 10% glycerol, pH 7.6 (buffer A) at 4 °C. The
column was washed sequentially with 5 volumes each of the following:
buffer A, 0.3 M NaCl, 20 mM GSH, 0.1% DOC;
buffer A, 2.5 mM S-hexyl GSH, 2.5 mM
S-octyl GSH, 0.1% DOC; and buffer A, 0.1% DOC. The enzyme was then eluted with five fractions of one volume each of 15 mM probenecid in buffer A containing 0.1% DOC. The enzyme
was purified nearly to homogeneity as determined by the presence in
silver-stained SDS-PAGE gels of a predominant 18-kDa protein band with
a minor 34-kDa band; only the 18-kDa band interacted with
anti-LTC4S antibody on immunoblot analysis. After
sequential concentration and dilution with a Microsep 10K centrifugal
concentrator (Filtron Technology Corp., Northborough, MA) to remove
probenecid, the protein concentration was estimated by staining a
SDS-PAGE gel of the purified enzyme and a standard amount of lysozyme
resolved in the same gel with Coomassie Blue dye and comparing their
intensities.
To examine the enzyme kinetics of recombinant wild-type or mutant LTC4S, 3-100-ng portions of purified enzyme were incubated in 200 µl of HEPES buffer, pH 7.6, containing 10 mM MgCl2 with either 20 µM LTA4-ME and various concentration of GSH or with 10 mM GSH and various concentrations of LTA4-ME for 2 min at room temperature. The reactions were terminated by the addition of three volumes of methanol containing 200 ng of PGB2. Samples were then analyzed for LTC4-ME by reverse phase-HPLC.
To determine the possible involvement of the two cysteine residues in the dimer formation required for function, we mutated Cys-56 to Ser and Cys-82 to Val in separate constructs and within the same construct. None of the mutations affected conjugation function of the transfected COS cell microsomes, indicating that neither cysteine residue is involved in dimer formation or in enzyme catalysis. The mutation of the single consensus N-glycosylation site at Asn-55 to Ala did not alter the electrophoretic mobility in SDS-PAGE, suggesting that this residue is not glycosylated in LTC4S. The mutation N55A did, however, increase the Km of GSH for the enzyme, suggesting that this residue participates in the GSH binding site.
Point Mutations of Charged ResiduesBecause a deletion of a
single charged amino acid, Asp-62, in FLAP had been associated with a
loss of binding function (21), initial point mutations were directed to
charged amino acid residues (Fig. 1). The mutations
E45Q, E47Q, E58Q, and R48S did not attenuate the conjugation function
of the transfected COS cell microsomes as compared with wild-type
construct in the same assay. However, the mutation of Arg-51 to
isoleucine or threonine abolished the functional activity of the
enzyme. Notably, activity was not diminished in the microsomes by point
mutation of Arg-51 to lysine or histidine, both of which have the
ability to function as basic side chains to donate a H+
ion. The active and inactive Arg-51 mutants were comparably positive for protein by immunoblot analysis of the microsomes.
To confirm the immunoblot analysis of the microsomal preparation and to
assess one parameter of protein folding, we examined the ability of the
R51T mutant protein to bind to a S-hexyl GSH-agarose column.
R51T bound to S-hexyl GSH and could be eluted from the column with probenecid in the same protocol as the wild-type enzyme (Fig. 2). Nonetheless, purified R51T was not able to
catalyze the conjugation reaction even at a protein concentration
30-fold higher than that of the wild-type enzyme (data not shown).
Point Mutations of the Tyrosine Residues
Because a tyrosine residue is known to catalyze the conjugation of the GSH with xenobiotics in cytosolic glutathione S-transferases (27, 28), the role of such residues was examined by point mutations of LTC4S. Mutations Y59F, Y93F, and Y97F each increased the Km of GSH by 5-15-fold in the microsomal assay (Table I). Only the mutation Y93F was associated with a reduction in function under standard assay conditions. None of the tyrosine mutants with an elevated Km for GSH exhibited a change in the Km for LTA4-ME, thereby revealing substrate specificity.
|
Kinetic analysis of the isolated recombinant Y93F established a greater
than 90% reduction in Vmax (Table I) and
confirmed the increase in Km for GSH. The catalytic
efficiency (Vmax/Km) of Y93F
(0.14) is 1/260 of that of the wild-type enzyme (36.6). The loss in
function with the Y93F mutant prompted an examination of the pH
dependence of the residual enzyme (Fig. 3) as an
indication that Tyr-93 was involved in thiolate anion formation. A
shift in the optimum to a higher pH was observed in the microsomal
assay with 5-fold more Y93F as compared with Y59F or the wild-type
construct. Furthermore, the pH-dependent conjugation curve
of Y93F resembled that of the non-enzymatic reaction.
Additional Point Mutations of the Hydrophilic Loops
The carboxyl terminus of the first hydrophilic loop is implicated in the binding of drugs, termed FLAP inhibitors, that interfere with the cellular presentation of released arachidonic acid to 5-lipoxygenase (9, 21). The cellular and subcellular conjugation function of LTC4S is inhibited by relatively high concentrations of a FLAP inhibitor. Eight of 11 amino acid residues in the carboxyl terminus of the first hydrophilic loop of LTC4S are identical to those of FLAP. Thirteen point mutations within the carboxyl terminus of this loop and the nearby second hydrophobic domain, namely L39F, E45Q, F46Y, E47Q, R48S, Y50F, Q53N, V54Q, F60Y, L62T, T66V, V69S, H75Q, and E76Q (Fig. 1), did not affect the function of the microsomal mutant enzyme from transfected COS cells. A single point mutation at the amino terminus of the first hydrophilic loop, Q25K, and point mutations within the second hydrophilic loop, R92S, R99S, S100V, R104S, Y109I, and S111V, did not affect the function of the microsomal enzyme. The A52S mutant of the carboxyl terminus of the first hydrophilic loop did exhibit an increase in Km for GSH (Table I).
Deletion MutationsNeither the deletion of the charged amino acids Lys-2 and Asp-3 at the amino terminus nor the deletion of the 14 amino acids at the carboxyl terminus altered the function of the microsomal recombinant enzyme expressed by the COS cells (data not shown). However, the deletion of the third hydrophobic domain with the carboxyl terminus did eliminate function even though protein was present by immunoblot analysis of the COS cell transfectant (data not shown).
Substitution of the Second or Third Hydrophobic Domain of FLAP for Comparable Regions of LTC4SBecause the elimination
of LTC4S function with deletion of the third hydrophobic
domain could reflect a conformational effect, the hydrophobicity
provided by that region was restored by substitution of the comparable
FLAP domain with 33 amino acid differences and 7 extra amino acids
added at the carboxyl terminus. This substitution to create a hybrid
LTC4S/FLAP protein (hybrid A, Fig.
4A) provided COS cell microsomes with enzyme
activity. The Km value for LTA4-ME is
comparable to wild-type enzyme. A hybrid created by the substitution of
the second hydrophobic domain of FLAP with 16 amino acid differences
for that of LTC4S (hybrid B, Fig. 4B) did not
function. However, substitution for the residues 66-81 (hybrid C, Fig.
4C) within this domain with residues 70-85 of FLAP
containing 10 amino acid differences plus a mutation P78L provided
active recombinant hybrid protein with the expected
Km value for LTA4-ME.
Effect of a Pseudo-dimer LTC4S on the Conjugation Function of the Enzyme
A covalent pseudo-homodimer of
LTC4S was created by connecting two LTC4S
cDNAs with 36 nucleotides that encode a 12-amino acid bridge
containing 6 histidine residues and a factor Xa recognition sequence.
When this construct was transfected into COS-7 cells, the expressed
protein migrated as a dimer both in gel filtration chromatography and
in SDS-PAGE (Fig. 5). The expressed microsomal protein
(WT/WT) is fully active in the microsomal conjugation of
LTA4-ME with GSH. Furthermore, the recombinant
pseudo-heterodimer of R51I LTC4S monomer
(R51I/WT) and wild-type monomer also retained enzymatic
function after gel filtration. Kinetic studies with solubilized
microsomes gave Km values for LTA4-ME
and GSH of 6.5 µM and 3.2 mM, respectively,
for the pseudo-heterodimer; these values are similar to those for the
wild-type enzyme (Table I).
The functional characteristics of LTC4S, a novel
perinuclear membrane protein that conjugates GSH to LTA4 to
form LTC4, the parent compound of the receptor-active
cysteinyl LTs, were assessed by mutagenic analysis. The initial
approaches were based on three features of LTC4S: the
conjugation of GSH in the absence of significant nucleotide or amino
acid sequence homology with the cytosolic GSH S-transferase
superfamily, the amino acid sequence identity and predicted secondary
structure homology with FLAP, and the evidence by gel filtration for
function as a homodimer but not as the constituent monomer. Point
mutations suggested the amino acid residues contributing to the GSH
binding site, the role of Tyr-93 in promoting the formation of thiolate
anion of GSH, and the requirement for Arg-51 or a replacement amino
acid with similar charge. Arg-51 is presumed to open the epoxide ring
of LTA4 for the conjugation with the thiolate anion of the
reduced GSH (Fig. 6). Substitutions of the corresponding
second and third hydrophobic domains, in part or in full, for those of
LTC4S and the functional integrity of covalent
pseudo-heterodimers composed of R51I and wild-type LTC4S
monomers revealed the conformation-dependent function of
the non-covalently linked natural dimer.
Because LTC4S showed a nearly identical predicted secondary structure to FLAP and a significant amino acid identity in a region at the carboxyl terminus of the first hydrophilic loop corresponding to the putative FLAP inhibitor-binding domain (18, 19, 29), we previously speculated that LTA4 would bind to the carboxyl terminus region of the first hydrophilic loop. This view was supported by the finding that a FLAP inhibitor, MK-886, albeit at a relatively high dose, inhibited the conjugation activity of LTC4S (18, 22). Because point mutations of Asp-62 of FLAP had established the role of a negatively charged residue at this position for binding a FLAP inhibitor to prevent the presentation of released arachidonic acid to 5-lipoxygenase, we mutated the corresponding Glu-58 of LTC4S (Fig. 1). Neither the point mutation E58Q nor mutations throughout the putative FLAP inhibitor binding domain-like region, i.e. L39F, E45Q, F46Y, E47Q, R48S, Y50F, Q53N, V54Q, and F60Y, impaired the conjugation function or the IC50 of MK-886 (data not shown). The finding that a residue opposite in charge to Asp-62 of FLAP, namely Arg-51 of LTC4S, is critical to the function of LTC4S implies that the putative FLAP inhibitor binding domain-like sequence of LTC4S is not likely to be the binding domain for LTA4 or for MK-886.
That the critical feature of the R51T or R51I mutations responsible for the abolishment of conjugation function is loss of a charged residue is supported by the finding of full activity for the R51K and R51H mutations. The levels of protein for both the Thr-51 and the Ile-51 mutant enzymes were comparable to that of wild-type protein as determined by SDS-immunoblot analysis of the transfected COS cell microsomes. Furthermore, the solubilized microsomal protein of the R51T mutant, although without function, bound to S-hexyl GSH-agarose during purification and eluted with probenecid in a manner similar to that of the wild-type enzyme; thus, the absence of function is not due to the lack of expression or improper folding of the conformational GSH binding site (Fig. 2). Proper folding of the inactive monomer is also indicated by the activity of the pseudo-heterodimer incorporating R51I with wild-type monomer (Fig. 5, Table I). The critical role of a H+-donating residue at position 51 is revealed both by the nearly normal function of the mutant enzymes of R51K or R51H (Table I) and by the failure of R48S, R92S, R99S, or R104S to attenuate function or substrate binding affinities. Inasmuch as LTA4 contains an epoxide ring susceptible to the attack by H+ ion, Arg-51 probably provides this catalytic function for conjugation with the thiolate anion of GSH (Fig. 6). It is noteworthy that the recently identified microsomal glutathione S-transferase II, which possesses some ability to conjugate LTA4 with GSH, also contains an Arg-51 (30).
Cytosolic GSH S-transferases are homo- or heterodimers in which each monomer contains a xenobiotic (lipophilic)-binding site, a GSH-binding site, and a tyrosine residue whose hydroxyl group enhances both the formation and stabilization of the thiolate anion of GSH to promote conjugation with the xenobiotic (27, 28). Point mutations Y50F and Y109I did not affect the conjugation function of the recombinant microsomal enzymes, whereas tyrosine mutations Y59F, Y93F, and Y97F at the carboxyl terminus of the first hydrophilic loop and the amino terminus of the second hydrophilic loop increased the Km for GSH (Table I). That V49F, A52S, and N55A of the carboxyl terminus of the first hydrophilic loop also increased the Km of the recombinant microsomal LTC4S for the GSH about 5-15-fold, again without altering the Km for LTA4-ME, places the GSH binding site at the carboxyl terminus of the first hydrophilic loop and the amino terminus of the second hydrophilic loop. These data imply that both of the hydrophilic loops are localized to the same (cytosolic) side of the membrane.
Because the Y93F mutant in the second hydrophilic loop of LTC4S not only increased the Km for GSH but also markedly decreased the catalytic efficiency (1/260 of wild type) (Table I), Tyr-93 most likely provides the hydroxyl group that enhances both the formation and the stabilization of the thiolate anion of GSH (Fig. 6). The mutation of the tyrosine residue in cytosolic GSH S-transferases critical for catalysis shifts the curve for the pH-dependent conjugation to resemble that of the non-enzymatic reactions (27); the same assessment was made for Y93F compared with Y59F and the wild-type LTC4S (Fig. 3). The Y93F mutation both attenuates conjugation function and shifts the pH dependence curve at least one unit to the optimum observed for the non-enzymatic reaction. Because the Y59F mutation increased the Km for GSH without loss of function or alteration in pH curve, Tyr-93 is the residue responsible for the enhancement of the thiolate anion formation, with resultant conjugation as the epoxide ring of LTA4 is opened via Arg-51. None of the point mutations altered the low micromolar Km for LTA4.
The importance of secondary and tertiary structures in the binding of substrates and in the catalytic function of LTC4S was examined through the construction of LTC4S/FLAP hybrid molecules (Fig. 4), the construction of covalent pseudo-dimers, and the deletion of selected domains. The deletion of the 14 amino acid residues of the carboxyl terminus did not alter the function of the enzyme expressed in transfected COS cell microsomes. This result is compatible with the earlier finding that the kinetic parameters of the recombinant human and mouse LTC4S were virtually identical, even though 9 of the 18 amino acid differences between these 150-amino acid polypeptides occurred in the carboxyl terminus (22). That deletion of the third hydrophobic domain with the carboxyl terminus of LTC4S abolished enzyme activity, whereas substitution with the comparable region of FLAP provided substantial function, indicates that a difference in 16 amino acids in this region still allows for maintenance of the needed tertiary structure. Although substitution of the complete second hydrophobic domain of LTC4S with that of FLAP was not acceptable, a partial replacement for a stretch of 16 amino acids (Fig. 4) created a functional microsomal enzyme (Table I). Taken together, these data argue that, although the sequence of the amino acids is less critical than their hydrophobicity in the third hydrophobic domain of LTC4S, the sequence itself rather than the hydrophobicity of the second hydrophobic domain is important for proper enzyme function.
Inasmuch as the point mutations of the cysteine residues C56S and C82V separately or jointly do not affect enzyme function, the dimer state needed for function is not based on a disulfide bond. The role of the companion monomer in a dimer was characterized by the functional integrity of a construct prepared with an inactive R51I monomer linked to a native monomer as compared with a dimer constructed with two native monomers. The covalent pseudo-dimers exhibited an apparent molecular mass of 36 kDa on gel filtration, as is characteristic of native (31) or the recombinant wild-type LTC4S. In contrast to the wild-type LTC4S, which migrates at 18 kDa on SDS-PAGE, both the pseudo-homodimer (WT/WT) and the pseudo-heterodimer (R51I/WT) migrate as 33-kDa proteins in SDS-PAGE immunoblot analysis (Fig. 5). That the pseudo-heterodimer retains enzymatic function and exhibits the same Km for GSH and LTA4-ME, respectively, as the wild-type enzyme (Table I) reveals that a single wild-type monomer has independent catalytic activity, which can only be expressed within the tertiary structure provided by a complementary monomer.
A model for LTC4S function based on the findings obtained
by mutagenic analysis would include the following: presentation of the
two hydrophilic loops central to catalytic function on the same side of
the membrane (Fig. 7), binding of GSH to a conformation binding site that involves a combination of residues in the carboxyl terminus of the first hydrophilic loop and the amino terminus of the
second hydrophilic loop, promotion of GSH thiolate anion formation and
stabilization by the hydroxyl group of Tyr-93 in the amino terminus of
the second loop, opening of the epoxide bond of LTA4 by the
H+ of the guanidinium group of Arg-51 in the carboxyl
terminus of the first loop, and conjugation of the activated
substrates, GSH and LTA4 (Fig. 6).
We thank Dr. Christopher Walsh of Harvard Medical School for suggestions about this manuscript and Joanne Miccile for secretarial assistance.