Genetic Complexity, Structure, and Characterization of Highly Active Bovine Intestinal Alkaline Phosphatases*

Mammalian alkaline phosphatases (APs) display 10–100-fold higher k cat values than do bacterial APs. To begin uncovering the critical residues that determine the catalytic efficiency of mammalian APs, we have compared the sequence of two bovine intestinal APs, i.e. a moderately active isozyme (bovine intestinal alkaline phosphatase, bIAP I, ∼3,000 units/mg) previously cloned in our laboratory, and a highly active isozyme (bIAP II, ∼8,000 units/mg) of hitherto unknown sequence. An unprecedented level of complexity was revealed for the bovine AP family of genes during our attempts to clone the bIAP II cDNA from cow intestinal RNAs. We cloned and characterized two novel full-length IAP cDNAs (bIAP III and bIAP IV) and obtained partial sequences for three other IAP cDNAs (bIAP V, VI, and VII). Moreover, we identified and partially cloned a gene coding for a second tissue nonspecific AP (TNAP-2). However, the cDNA for bIAP II, appeared unclonable. The sequence of the entire bIAP II isozyme was determined instead by a classical protein sequencing strategy using trypsin, carboxypeptidase, and endoproteinase Lys-C, Asp-N, and Glu-C digestions, as well as cyanogen bromide cleavage and NH2-terminal sequencing. A chimeric bIAP II cDNA was then constructed by ligating wild-type and mutagenized fragments of bIAP I, III, and IV to build a cDNA encoding the identified bIAP II sequence. Expression and enzymatic characterization of the recombinant bIAP I, II, III, and IV isozymes revealed averagek cat values of 1800, 5900, 4200, and 6100 s−1, respectively. Comparison of the bIAP I and bIAP II sequences identified 24 amino acid positions as likely candidates to explain differences in k cat. Site-directed mutagenesis and kinetic studies revealed that a G322D mutation in bIAP II reduced its k cat to 1300 s−1, while the converse mutation, i.e. D322G, in bIAP I increased its k cat to 5800 s−1. Other mutations in bIAP II had no effect on its kinetic properties. Our data clearly indicate that residue 322 is the major determinant of the high catalytic turnover in bovine IAPs. This residue is not directly involved in the mechanism of catalysis but is spatially sufficiently close to the active site to influence substrate positioning and hydrolysis of the phosphoenzyme complex.

Alkaline phosphatases (APs) 1 are dimeric, zinc-containing nonspecific phosphomonoesterases that exist throughout speciation, from Escherichia coli to man (1). Cloning of AP cDNAs from a variety of species and comparison of their primary structures has revealed a high degree of sequence conservation and even a 25-30% similarity between E. coli and mammalian APs (2,3). The three-dimensional crystallographic structure of the E. coli AP homodimer has been solved (4,5) and the reaction mechanism inferred (6). Efforts to crystallize any of the mammalian APs have so far been unsuccessful. Conservation of the residues that comprise the catalytic zinc binding site, as well as the substrate binding residues, suggests that the reaction mechanism is conserved throughout evolution (5). However, there are significant structural differences between the E. coli and mammalian APs including several loop regions that have undergone deletion and/or insertion.
Mammalian APs display the unique kinetic property, not shared by their bacterial ancestors, of being inhibited stereospecifically by L-amino acids and peptides through an uncompetitive mechanism (7,8). Using human placental AP as a paradigm for mammalian APs, we and others have established that residues within a surface loop unique to mammalian APs are responsible for the differential uncompetitive inhibition by L-amino acids (9 -12), the heat stability properties (13), and protein-protein interaction specificities exhibited by some mammalian APs (13). A major property of APs that remains to be explained in terms of structure is the large variability in catalytic activity displayed by mammalian APs, which have 10 -100-fold higher k cat values than E. coli AP (14). Because among mammalian APs, the intestinal isozyme has the highest specific activity, the bovine intestinal APs (IAPs) represent a potentially useful system for addressing this question. Besman and Coleman (15) demonstrated the existence of two IAP isozymes in the cow intestine, i.e. calf IAP and adult bovine IAP, by sequencing the amino termini of chromatographically purified AP fractions. We have previously reported the cloning and biochemical characterization of the recombinant adult bovine IAP, presently designated bIAP I (16). In this study we report the sequence and characterization of the calf IAP (bIAP II) and two novel bIAP isozymes (bIAP III and bIAP IV (Gen-Bank TM accession nos. AF052226 and AF05227)) and present evidence for the existence of an unprecedented level of complexity in the cow AP gene family. Sequence comparisons and site-directed mutagenesis have unequivocally identified a Gly residue at position 322 as the crucial residue that determines the high specific activity of bIAP II.

MATERIALS AND METHODS
Cloning-A gt 11 cDNA library prepared from adult cow intestine (CLONTECH Laboratories, Palo Alto, CA) was screened using a 1,075-bp HindIII fragment from the 5Ј end of the bIAP I cDNA (16) as a probe. Clones isolated from this library were used to screen an EMBL-3 SP6/T7 genomic library prepared from adult cow's liver (CLONTECH Laboratories). An unamplified ZAP II cDNA library was prepared from mRNA isolated, using the Trisolv TM reagent, from the small intestine of one adult cow using oligo(dT) as primer (Stratagene, San Diego, CA) and screened using the 1,075-bp HindIII fragment of the bIAP I cDNA as a probe. Probes were radiolabeled using a random primed DNA labeling kit (Boehringer Mannheim). Phage DNA was prepared as described previously (17) for gt 11 and EMBL-3 SP6/T7 clones. In vivo excision of ZAP II clones was performed according to the manufacturer's protocol (Stratagene). Genomic clones were characterized by Southern blot analysis as described previously (18). EcoRI cDNA fragments from gt 11 clones and various restriction fragments from clones derived from the other libraries were subcloned into the KSϩ vector (Stratagene). Plasmid DNA was prepared by the alkaline lysis procedure (18). Sequencing was performed using Sequenase according to the manufacturer's protocol (Amersham Pharmacia Biotech). Oligos used for sequencing bIAPs III and IV are as follows. 1s, GCC AAG AAT GTC ATC CTC; 1a, GAG GAT GAC ATT CTT GGC; 2s, GGT GTA AGT GCA GCC GC; 2a, GCG GCT GCA CTT AGA CC; 3s, AAT GTA CAT GTT TCC TG; 3a, CAG GAA ACA TGT ACA TT; 4s, CCA GGG CTT CTA CCT CTT; 4a, AAG AGG TAG AAG CCC TGG; 5s, ACC AGA GCT ACC ACC TCG; 5a, AAG CAG GAA ACC CCA AGA; and 6s, CTT CAG TGG CTT GGG ATT; 6a, AAT CCC AAG CCA CTG AAG. Nucleic acid sequences were analyzed using the MacVector sequence analysis program (International Biotechnologies, Inc., New Haven, CT).
Determination of the Amino Acid Sequence of bIAP II-Approximately 500 g of purified, highly active (approximately 7800 units/mg) cow intestinal AP was dissolved in 450 l of 6 M guanidinium hydrochloride, 0.25 M Tris, 1 mM EDTA, pH 8.5, followed by the addition of 30 l of mercaptoethanol. After reduction for 30 min at 100°C, cysteines were alkylated by addition of 35 l of vinylpyridine, and the mixture was incubated for 45 min at room temperature in the dark. The reaction mixture was then immediately desalted by a short reversed phase HPLC on a Aquapore RP300 column (30 ϫ 2.1 mm, Applied Biosystems, Weiterstadt) using a steep gradient of acetonitrile in 0.1% trifluoroacetic acid to elute bound enzyme. Protein-containing fractions were evaporated to dryness. In order to deglycosylate the enzyme, 125 g of AP were dissolved in 15 l of distilled water, 6 l of incubation buffer (250 mM Na 2 HPO 4 , 50 mM EDTA, pH 7.2), and 15 units of endoglycosidase F/N-glycosidase F (Boehringer Mannheim, Penzberg). The mixture was left overnight at 37°C and subsequently used for cleavages. Reduced and alkylated AP was enzymatically cleaved with different enzymes (i.e. endoproteinase Lys-C, endoproteinase Asp-N, endoproteinase Glu-C, and trypsin (Boehringer Mannheim, Penzberg) according to the instructions given on the data sheets for individual enzymes. Cyanogen bromide cleavage was performed with 10% (w/w) CNBr in 70% (v/v) formic acid for 8 h. After dilution with water the solution was reduced in volume on a SpeedVac concentrator (Savant) and applied onto reversed phase HPLC. Carboxypeptidase Y (8 ng/l) digestion of the COOH-terminal tryptic peptide was performed for 4 min, and the released peptides were analyzed by matrix-assisted laser desorption/ ionization mass spectrometry using a Bruker Reflex III instrument according to the instructions of the manufacturer. 2,5-Dihydroxybenzoic acid (10 mg/ml) in acetonitrile/water (50/50, v/v) was used as a matrix. Peptides derived from enzymatic or chemical cleavages were separated by reversed phase HPLC on a LiChrospher C18 sel B column, 125 ϫ 2 mm (Merck, Darmstadt) using a 0.1% trifluoroacetic acid/ acetonitrile solvent system. Flow rate was 300 l/min. The effluent was monitored by UV at 206 nm, and the fractions were collected manually. Mass determination of the peptides was performed on an API III electrospray mass spectrometer (PE-Sciex, Langen) according to the instruction of the manufacturer. Amino acid sequencing was performed using a 492A protein sequencer (Applied Biosystems, Weiterstadt) according to the instructions of the manufacturer.
Construction of bIAP II cDNA-To construct a cDNA coding for bIAP II, wild-type restriction fragments and site-directed mutagenized PCR fragments of cDNAs bIAP I, III, and IV were assembled, creating the L1N8 (three fragments) and INT 1 (nine fragments) intermediate cDNA constructs. INT 1 and bIAP III then served as the templates for sitedirected mutagenesis and fragments from these were assembled into a complete INT 2 (eight fragments) cDNA. Restriction fragments from INT 2 and site-directed mutagenized fragments of INT 2 were then assembled into the INT 3 (five fragments) cDNA and finally the bIAP II (four fragments) cDNA. Site-directed mutagenesis was performed as described previously (19) using BsaI as the restriction enzyme that cuts at a distance from its recognition site (GGTCTCN1/N5). All PCR products were sequenced to verify the absence of secondary mutations, and all constructs were verified by sequencing and restriction digestion. The sequence of the oligonucleotide primers used for amplifying the sitedirected mutagenized fragments are as follows, with the name of the The following ligation reactions were performed using, in all cases, the pcDNA-3 (Invitrogen, San Diego, CA) expression vector. The fragments are numbered according to the PCR reaction number given above or by the name of the wild-type or chimeric cDNA, followed by the restriction enzymes used to create the cohesive termini of that fragment. L1N8 ϭ pcDNA-3/ EcoRI-XbaI ϩ 1/EcoRI-BsaI ϩ 2/BsaI-BamHI ϩ bIAP I/BamHI-XbaI.
bIAP II Mutagenesis-Ten additional constructs were made to identify the residue(s) responsible for the differential kinetic properties of bIAPs I and II. All constructs were subcloned in pcDNA-3/EcoRI-XbaI. E210Vϩ, GGT CTC ATG TTT CCT GTG GGG ACC CCA GAC; E236A, GGT CTC CTG CCA TGC CTG CAC CAG GTT. Using these and previously listed oligos the following eight PCR reactions (a-h) were carried out using bIAP II as template: a, 1s, I133MϪ; b, S142Aϩ, M205KϪ; c, 1s, A142SϪ; d, V210Eϩ,330Ϫ; e, E210Vϩ,330Ϫ; f, M180Kϩ,E236AϪ; g, 236ϩ,330Ϫ; h, S142A,K205MϪ. The products of these were subcloned and sequenced, and then fragments isolated for the following ligations: Production and Characterization of the Recombinant Enzymes-All cDNAs (bIAP I, bIAP II, bIAP III, bIAP IV, and corresponding mutants) were cloned into the pcDNA-3 expression vector (Invitrogen, San Diego, CA), transfected into Chinese hamster ovary cells and stable transfectants were selected by growing the cells in the presence of 500 g/ml geneticin (Life Technologies, Inc.). Recombinant APs were extracted from the stably transfected Chinese hamster ovary cells as described previously (20). To measure k cat , microtiter plates coated with 0.1 g/ml high affinity anti-bovine AP monoclonal antibody (Scottish Antibody Production Unit, Lanarkshire, Scotland) were incubated with increasing concentrations of enzyme and the activity of bound enzyme was measured as the change in absorbance at 405 nm over time at 20°C upon addition of 30 mM p-nitrophenylphosphate (pNPP) as substrate in 1.0 M diethanolamine buffer (pH 9.8), 1 mM MgCl 2 , and 20 M ZnCl 2 . The p-nitrophenol concentration formed was calculated using an extinction coefficient of 10,080 liters mole Ϫ1 cm Ϫ1 . Commercial preparations with known specific activities (Biozyme Laboratories, 7,822 units/mg, and Boehringer Mannheim, 3,073 units/mg) as well as purified bIAP II (8,600 units/mg) were used as standards. The concentration of enzyme at those dilutions that saturated the antibody (E 0 ) was calculated from a standard curve of activity versus known enzyme concentrations using identical assay conditions. The maximal rate of substrate conversion (V max ) was then divided by E 0 to calculate k cat . To calculate K m , substrate concentration was varied between 0.25-2.0 mM pNPP, and the initial reaction rate was measured at 20°C over a time interval of 10 min. Regression plots of [pNPP]/V versus [pNPP] (Hanes plots) to the x intercept equaled ϪK m . Dividing the standard error of the predicted y value for each x in the regression by the slope of the regression gave the standard error of the K m . V max Ϯ S.E. was calculated by dividing K m Ϯ S.E. with the y intercept Ϯ S.E., using the appropriate equations to obtain V max S.E. Heat stability curves were produced by incubating extracts at 45-75°C in 5°C increments, 10 min each as described previously (16). The activity of each sample was then determined as above, and residual activity calculated as the percentage remaining compared with the nonheated sample. The temperature at which 50% residual activity remained (T 50 ) was calculated from the residual activity versus temperature plots.

RESULTS AND DISCUSSION
Cloning of bIAP III and IV-We set out to determine the structure of the fetal intestinal AP (bIAP II) defined by Besman and Coleman (15) as possessing a LIPAEEEN amino-terminal sequence since we knew that this amino-terminal sequence was found in purified high activity intestinal AP preparations (range 7,000 to 8,000 units/mg) available commercially (Biozyme Laboratories and Boehringer Mannheim GmbH). We screened a commercial 5Ј Stretch gt 11 bovine small intestinal cDNA library (CLONTECH Laboratories) with a 1,075-bp Hin-dIII fragment of the bIAP I cDNA that contains sequences included in exons I through VIII of the bIAP I gene. Twelve cDNAs were isolated that represented different size spots and intensity of hybridization on the filters. The fragments were subcloned and completely sequenced. Four of these cDNAs were identical to the bIAP I sequence as previously published (16). Four clones represent a new tissue-specific AP isozyme gene homologous, but not identical, to bIAP I, although the clones were unspliced. The largest of these clones (2,561 bp) aligned to the bIAP I gene 570 bp 5Ј of the start codon and extended to exon eight, and identifies the isozyme referred to here as bIAP VII (GenBank TM accession no. AF052230). Two cDNAs represented another variation of the bIAP I sequence and represented unspliced clones, the largest (783 bp) contained exon I to exon III sequences which define the bIAP V isozyme (GenBank TM accession no. AF052228). One cDNA (clone VIII) appeared to be yet another tissue-specific AP transcript. This clone is 1,642 bp long, is also unspliced, aligns with bIAP I from intron 2 to intron 8, and defines the bIAP VI isozyme GenBank TM accession no. AF052229). All these novel clones encode predicted amino-terminal sequences that are different from the expected LIPAEEEN sequence. Fig. 1 shows differences in the deduced first 80 amino-terminal amino acids of the newly identified isozymes bIAP III, bIAP IV, bIAP V, bIAP VI, bIAP VII in comparison with the corresponding residues of bIAP I (16) and of bIAP II as determined below. Still another cDNA was isolated that represents a new tissue nonspecific AP molecule. This partially spliced cDNA clone aligns with bovine kidney AP (21) (starting at residue 8) in exon II and extends to exon IX. This appears to be a different TNAP (TNAP-2, GenBank TM accession no. AF052231) molecule expressed in the bovine intestine. Northern blot analysis was performed on RNA samples isolated from different portions of the cow intestine of a single animal, and the sample with the highest expression was chosen for the construction of a new cDNA library in ZAP II vector (Stratagene). Adjacent segments of the intestine were used for enzyme purification. The entire unamplified library (1.0 ϫ 10 6 independent recombinant clones) was screened with the 1,075-bp HindIII bIAP I probe, and 65 clones were isolated and sequenced. All clones corresponded in sequence to one or the other of two novel bIAP cDNAs designated bIAP III and bIAP IV. The sequence of the 2,460-bp bIAP III cDNA is shown in Fig. 2 as well as the differences found in the coding region of the 2,536-bp bIAP IV cDNA. Neither of these full-length novel bIAP cDNAs coded for an amino-terminal LIPAEEEN sequence, while at the protein level it was clear that the LI-PAEEEN sequence was the major component of the purified preparation from the same intestinal region. We had to conclude that the bIAP II sequence was either "toxic" to the bacterial cells used for library construction or "unclonable" for some other reason.
Our previous work on the cloning of bIAP had revealed the structure of the bIAP I gene and also of a transcribed pseudogene (R201) (16). Southern blot analysis using the bIAP cDNA as a probe had revealed a complex pattern of bands and only two of them could be accounted for by the cloned bIAP I gene and the R201 pseudogene (16). Human and mouse APs are the two best characterized gene families, and both display the FIG. 1. Amino acid differences in the first 80 residues of the bIAP I, II, III, IV, V, VI, and VII sequences as determined from cDNA cloning. The single-letter code is used for the amino acids and a hyphen indicates a residue identical to that in the sequence of bIAP I.

FIG. 2.
Complete nucleotide sequence of the 2,460-bp bIAP III cDNA and deduced amino acid sequence. Nucleotide differences found in the coding region of the bIAP IV cDNA are written above the nucleotide sequence and those mutations that translate into amino acid differences are spelled out under the deduced amino acid sequence of bIAP III. Nucleotide differences in the 5Ј-and 3Ј-untranslated regions of the bIAP III and bIAP IV cDNA are not shown. same degree of genetic complexity. Human APs are encoded by three tissue-specific AP loci, i.e. GCAP (22), PLAP (23), and IAP (24) and one TNAP locus (25), and the mouse AP genes include two active tissue-specific AP genes, i.e. embryonic and IAP (26), one pseudogene (26), and one TNAP gene (27). Both the human and mouse tissue-specific genes are highly homologous and are each comprised of 11 exons contained in less than 5 kilobase pairs of DNA while the single TNAP gene in both species is composed of 12 exons occupying 40 -50 kilobase pairs of genomic DNA. The rat AP gene family has not been as well characterized, although a TNAP gene (28) and two different IAP cDNAs (29), both coding for glycosylphosphatidylinositol (GPI) anchored isozymes, have been cloned. The presence of at least seven IAP genes in the cow intestine and the existence of two tissue nonspecific AP loci in this species is, therefore, unprecedented. Since the cow has also been shown to have genes encoding at least two embryonic AP isozymes (30) the number of AP genes in this species is likely to be 10 or more.
Determination of the Amino Acid Sequence of the Highly Active bIAP II Isozyme-We resorted to determining the amino acid sequence of bIAP II using purified commercial preparations of high activity calf intestinal AP (Biozyme AP and Boehringer Mannheim GmbH). Direct NH 2 -terminal analysis revealed that purified AP preparations were heterogeneous. The two-dimensional electrophoresis pattern was far more complex than expected for a purified preparation and varied between the different AP preparations (data not shown). After deglycosylation, the two-dimensional gel electrophoresis complexity of the protein pattern remained almost unchanged, suggesting that the observed heterogeneity was mainly caused by differences in the primary structure of the APs, likely due to heterogeneity in the purified AP preparations, rather than by different glycostructures. Peptide maps were generated from the most active AP preparation by cleavage with endoproteinase Lys-C, Asp-N, Glu-C, trypsin, and cyanogen bromide after reduction and alkylation using both native and deglycosylated enzyme. The generated peptides were separated and isolated by reversed phase HPLC. Electrospray mass analysis of each fraction was performed, and the peptides were sequenced by Edman degradation and compared with our previously published amino acid sequence of bIAP I (16). The complete primary structure of bIAP II is given in Fig. 3. As expected, this isozyme possessed the LIPAEEEN amino-terminal sequence. where the main isoleucine was accompanied by a low amount of methionine (less than 20%). In position 210, the main glutamic acid was accompanied by a low amount of valine (less than 20%). In position 410 no phenylthiohydantoin amino acid residue could be detected in the Edman degradation of Asp-N cleavage derived peptide (384 -414), indicating a glycosylation of this position. The peptide (387-420) derived from the Lys-C digest of the deglycosylated material shows in position 410 an aspartic acid. In position 122 no amino acid could be detected, except in the case of peptides derived from deglycosylated material, where an aspartic acid was observed in this position. Thus, the high activity bIAP II isozyme was positively identified as being glycosylated at position 122 and 410, while glycosylation at position 249 remains a possibility. It should be noted that the double signals at positions 133, 205, 210 and 380 are compatible with the presence in the mixture of bIAP II together with either bIAP I, or IV, but given the fact that the starting APs preparation had a very high specific activity, it is most likely that the double signal is contributed by bIAP IV. The double signal at positions 251 and 252 cannot be explained by bIAP I, III, or IV and there is not enough sequence information to determine whether bIAP V, VI, or VII or still other isozymes may also be present.
The site of attachment of the GPI anchor has only been identified in the case of human PLAP to be Asp 484 (31,32). The cDNA of PLAP encodes a sequence of 29 amino acid downstream from Asp484. The sequence of this COOH-terminal peptide varies greatly between GPI anchored proteins (33). Even within the AP isozyme family, this COOH-terminal region, as deduced from cDNA cloning, has the least amount of sequence conservation (2). In the case of bIAPs, the deduced COOH-terminal sequences of bIAP I, III, and IV are considerably similar, so that it can be predicted that all these isozymes would be anchored through an equivalent amino acid position. Previously, we have shown that recombinant bIAP I is GPIanchored and it can be extracted either as a membrane-bound form retaining its GPI anchor or as a soluble isozyme (16). Our present work suggests that the GPI anchor site may be Ala 480 in the case of bIAP II since this was the COOH-terminal amino acid detected in mature bIAP II. This would suggest that bIAP I may be GPI anchored at Thr 480 and bIAP III and IV at the corresponding Ser 480 .

Construction of a Chimeric bIAP II cDNA, Expression and Characterization of the Recombinant bIAP I, II, III, and IV
Isozymes-The availability of the bIAP I, III, and IV cDNAs and the primary amino acid sequence of bIAP II made it possible to devise a scheme to construct a chimeric bIAP II cDNA to enable the expression and characterization of the recombinant enzyme. The strategy, depicted in Fig. 4, entailed ligating wild-type and mutagenized fragments of bIAP I, III, and IV. In the process of constructing the bIAP II cDNA, several intermediate constructs, encoding functional isozymes, were produced, i.e. L1N8, INT 1, INT 2, and INT 3, that would prove useful for the identification of amino acid residues influencing the catalytic activity of these isozymes (see below). It should be noted that since the protein sequencing work identified only the residues found in the fully processed mature bIAP II molecule, the NH 2 -terminal signal peptide necessary for intracellular trafficking to the cytoplasmic membrane was contributed by a cDNA fragment from the bIAP IV cDNA and the COOH-terminal GPI signaling sequence was contributed by a fragment  from the bIAP I cDNA (as depicted in Fig. 4).
The kinetic characterization of the recombinant bIAP I, II, III, and IV isozymes revealed differences in their catalytic properties as shown in Table I. bIAP II and bIAP IV possess very similar k cat values (5900 and 6100 s Ϫ1 ), about 3.4 times higher than that for bIAP I (1800 s Ϫ1 ), but even bIAP III had a k cat value 2.4 times higher than that of bIAP I. There are considerable differences with respect to the heat stability of the isozymes. bIAP I is the most heat stable of all four isozymes displaying 13°C higher T 50 than bIAP IV. bIAP II, and III have almost identical T 50 values, about 7°C lower than that of bIAP I.
Site-directed Mutagenesis Identifies a D322G Exchange as Responsible for the 3.0-fold Increase of Specific Activity in bIAP II Compared with bIAP I-The more than 3-fold difference in activity between bIAP I and II, and the fact that these isozymes differ in only 24 residues (Fig. 5), provided an experimental system to attempt to uncover the residue(s) that determine the differences in k cat between these isozymes. Expression and characterization of the intermediate chimeric enzymes, L1N8, INT 1, INT 2, and INT 3 enabled us to rule out the role of 11 putative candidate residues. The L1N8 mutant enzyme displayed a k cat comparable to bIAP I, ruling out mutations at position 2, 4, and 8, while both the INT 1 and INT 2 mutants displayed k cat values comparable to bIAP II, ruling out mutations at residues 380, 411, 416, 420, 427, 453, and 480. Similarly the k cat values of the INT 3 chimeric construct was similar to that of INT 1, INT 2, and bIAP II, ruling out an effect of the N192Y substitution. To identify which of the remaining 13 residues were responsible for determining the high specific activity, the bIAP II cDNA was used as a template to mutate each position for the corresponding bIAP I residue. We tested single point mutants carrying, respectively, N122K, I133M, A142S, K180M, M205K, E210V, E236A, G322D, and I332G mutations, as well as a combined (A289Q,A294V,Q297R,L299V) bIAP II mutant (Table I).
As can be seen in Table I and in Fig. 6a, the G322D mutation was, single-handedly, able to convert the kinetic properties of bIAP II into those of bIAP I. The changes included a 3-fold decrease in k cat and K m to values comparable to those of bIAP I (Table I). The converse mutation gave entirely consistent results, since by introducing a D322G mutation into bIAP I, the k cat and K m were increased in the resulting mutant to values comparable to that of bIAP II itself. Similarly, the introduction of an S322G substitution in bIAP III increased its k cat value to 5900 s Ϫ1 while the S322D mutation reduced its k cat value to 1200 s Ϫ1 , comparable to the k cat values of (Asp 322 )bIAP II and bIAP I. The difference in heat stability between bIAP I and bIAP II appears to be due to the combined effect of more than one substitution, since both the (Gly 322 )bIAP I and the (Asp 322 )bIAP II mutants display stability curves that are intermediate between those of bIAP I and II isozymes (Fig. 6b). The D322G substitution had a small destabilizing effect, i.e. approximately 3°C in T 50 (Table I), on the bIAP I isozyme, while the introduction of the G322D substitution into bIAP II causes a comparable increase in stability but not reaching the FIG. 5. Amino acid differences between the bIAP I, II, III, and IV isozymes. Only the differing residues are shown. Boxed or shaded residues on the bIAP I, bIAP III, or bIAP IV identify those sequences that were derived from DNA segments from bIAP I, bIAP III, or bIAP IV, respectively, in order to build the bIAP II molecule. The shaded residues are positions that did not need further modification in order to code for the bIAP II sequence, while residues that are boxed needed to be mutagenized into those substitutions found in the bIAP II isozyme. These 18 boxed residues correspond to the 18 arrowheads presented in Fig. 4. An asterisk identifies each of the 24 positions that differ between bIAP I and bIAP II. Despite the fact that no three-dimensional structure is available for any of the mammalian APs, our sequence comparisons indicate that residue 322 is located 2 amino acids away from a sequence absolutely conserved in APs throughout evolution, i.e. 311 EGGRIDHGHH 320 , that contains three crucial ligands (Glu 311 , Asp 316 , and His 320 ) coordinating to the active site zinc and magnesium ions. While further experimentation will be necessary to understand the detailed mechanistic effect of the D322G substitution, we can conclude from our data that the additional Asp at position 322 in bIAP I is impairing the hydrolysis of the phosphoenzyme complex during catalysis. Based on the general reaction scheme of APs (Scheme 1) (12,20), the expression for k cat equals k cat ϭ 1/(1/k 2 ϩ 1/k 3 ). Hence a rise in k cat can occur only as a consequence of an improved enzyme phosphorylation (k 2 ) or an improved hydrolysis of the phosphoenzyme complex (k 3 ). Our kinetic data reveals that a rise in k cat in (Gly 322 )bIAP I is paralleled by a proportionally identical rise in K m resulting in a constant k cat /K m ratio. Since this ratio equals k 1 /(1 ϩ k -1 /k 2 ), it follows that k 2 is not affected. Hence the enhanced catalytic activity in bIAP II and other Gly 322containing bIAP mutants in comparison with bIAP I results from an increase in k 3 . In other words, Asp 322 does not restrict phosphate positioning in the bIAP I active site pocket and does not impair the covalent phosphoenzyme complex formation, but it impairs the subsequent changes in coordination of the phosphate group during its hydrolysis from the active site Ser. The D322G substitution present in bIAP II relieves this interference.
Concluding Remarks-The present study has revealed an unprecedented level of complexity for the bovine IAP gene family and has shown that residues not directly participating in the mechanism of catalysis, but spatially close to the active site, are capable of influencing substrate catalysis contributing to variations in k cat in mammalian APs. Our findings also provide a rational explanation for the heterogeneity found in different purified commercial preparations of calf IAP. These heterogeneities result from difficulties in isolating a pure IAP isozyme from a tissue that may be expressing seven or more IAP genes. Furthermore, if more than one of the IAP genes are expressed in the same cell, heterodimers are likely to form which, as reported (20), will display noncooperative allosteric behavior where the stability and the catalytic properties of each monomer are controlled by the conformation of the second subunit. It follows that purified preparations of cow IAPs can show variations in the thermal stability, composition, and catalytic properties due to the combined effect of random heterodimer formation and multiplicity of expressed AP transcripts.