Residues Determining the Binding Specificity of Uncompetitive Inhibitors to Tissue-Nonspecific Alkaline Phosphatase

Authors


Abstract

Recent data have pointed to TNALP as a therapeutic target for soft-tissue ossification abnormalities. Here, we used mutagenesis, kinetic analysis, and computer modeling to identify the residues important for the binding of known ALP inhibitors to the TNALP active site. These data will enable drug design efforts aimed at developing improved specific TNALP inhibitors for therapeutic use.

Introduction: We have shown previously that the genetic ablation of tissue-nonspecific alkaline phosphatase (TNALP) function leads to amelioration of soft-tissue ossification in mouse models of osteoarthritis and ankylosis (i.e., Enpp1−/− and ank/ank mutant mice). We surmise that the pharmacologic inhibition of TNALP activity represents a viable therapeutic approach for these diseases. As a first step toward developing suitable TNALP therapeutics, we have now clarified the residues involved in binding well-known uncompetitive inhibitors to the TNALP active site.

Materials and Methods: We compared the modeled 3D structure of TNALP with the 3D structure of human placental alkaline phosphatase (PLALP) and identified the residues that differ between these isozymes within a 12 Å radius of the active site, because these isozymes differ significantly in inhibitor specificity. We then used site-directed mutagenesis to substitute TNALP residues to their respective homolog in PLALP. In addition, we mutagenized most of these residues in TNALP to Ala and the corresponding residues in PLALP to their TNALP homolog. All mutants were characterized for their sensitivity toward the uncompetitive inhibitors l-homoarginine (L-hArg), levamisole, theophylline, and L-phenylalanine.

Results and Conclusions: We found that the identity of residue 108 in TNALP largely determines the specificity of inhibition by L-hArg. The conserved Tyr-371 is also necessary for binding of L-hArg. In contrast, the binding of levamisole to TNALP is mostly dependent on His-434 and Tyr-371, but not on residues 108 or 109. The main determinant of sensitivity to theophylline is His-434. Thus, we have clarified the location of the binding sites for all three TNALP inhibitors, and we have also been able to exchange inhibitor specificities between TNALP and PLALP. These data will enable drug design efforts aimed at developing improved, selective, and drug-like TNALP inhibitors for therapeutic use.

INTRODUCTION

Alkaline phosphatases (E.C.3.1.3.1; (ALPs) are dimeric enzymes present in most organisms.(1) They catalyze the hydrolysis of phosphomonoesters with release of inorganic phosphate and alcohol.(2) In humans, three of the four isozymes are tissue-specific, the intestinal (IALP), placental (PLALP), and germ cell (GCALP) ALPs, whereas the fourth ALP is tissue-nonspecific (TNALP) and is expressed in bone, liver, and kidney. The first three isozymes are >90% homologous to each other at the protein level, with PLALP and GCALP differing only by 12 amino acid substitutions. In contrast, TNALP is only 50% homologous to the three tissue-specific human isozymes.(3)

The clearest evidence that ALPs are important in vivo has been provided by studies of human hypophosphatasia where a deficiency in the TNALP isozyme, caused by deactivating mutations in the TNALP gene,(4–6) is associated with defective bone mineralization in the form of rickets and osteomalacia.(7) The severity and expressivity of hypophosphatasia depends on the nature of the TNALP mutation.(8), (9) The mapping of hypophosphatasia mutations to specific 3D locations on the TNALP molecule has provided clues as to the structural significance of these areas for enzyme structure and function.(10) In bone, TNALP is confined to the cell surface of osteoblasts and growth plate chondrocytes, including the membranes of their shed matrix vesicles (MVs).(11–14) In fact, by an unknown mechanism, MVs are markedly enriched in TNALP compared with both whole cells and the plasma membrane.(15) It has been proposed that the role of TNALP in the bone matrix is to generate the inorganic phosphate needed for hydroxyapatite crystallization.(16–18) However, TNALP has also been hypothesized to hydrolyze the mineralization inhibitor PPi to facilitate mineral precipitation and growth.(19–21) Electron microscopy observations of MVs derived obtained from hypophosphatasia patients(22) or from TNALP knockout mice(23) reveal that they contain apatite-like mineral crystals but that extravesicular crystal propagation is retarded. This growth retardation could be because of either the lack of TNALP's pyrophosphatase function or the lack of inorganic phosphate generation. Our recent studies have provided compelling proof that a major role for TNALP in bone tissue is to hydrolyze PPi to maintain a proper concentration of this mineralization inhibitor ensuring normal bone mineralization.(24–26) We showed that crossbreeding the Enpp1−/− and the ank/ank mutant mice to TNALP knockout mice caused normalization of extracellular PPi levels and resulted in the correction of the soft-tissue ossification abnormalities.(25), (27) The increase in extracellular PPi concentrations led, in turn, to an increased expression of osteopontin. Correction of the soft tissue ossification was attributed to the combined upregulation of two inhibitors of calcification (i.e., PPi and osteopontin). Importantly, these studies have pointed to TNALP as a potentially useful therapeutic target for the treatment of soft tissue ossification abnormalities including ankylosis, osteoarthritis, and arterial calcification.

The fact that deletion of the TNALP gene results in the elevation of PPi and osteopontin concentrations in bone matrix and suppresses the soft tissue ossification in both Enpp1−/− and ank/ank mutant mice suggests that the chemical ablation of TNALP function may be a promising therapeutic strategy for soft tissue ossification. Since the early 1960s, several specific inhibitors of ALP isozymes have been uncovered. They include l-amino acids, such as L-phenylalanine, L-tryptophan, L-leucine, and L-homoarginine,(28–30) as well as some nonrelated compounds, such as levamisole(31) and theophylline (Fig. 1).(32) The inhibition is of a rare uncompetitive type,(33) but the exact binding site and precise mechanism of inhibition have only been elucidated for the binding of L-phenylalanine and L-leucine to the PLALP active site.(34–37) No information is available about the binding sites of the three commonly used inhibitors of TNALP, that is, L-homoarginine (L-hArg), levamisole, and theophylline. Furthermore, these inhibitors are not entirely specific for TNALP(28), (38) and have low binding affinity, requiring the in vivo administration of very high concentrations to achieve biological effects.(39–44)

Figure FIG. 1..

Chemical structure of the uncompetitive inhibitors (A) l-homoarginine, (B) levamisole, and (C) theophylline.

In this study, we compared the modeled 3D structure of TNALP with the structure of PLALP and pinpointed the residues that differ between the two isozymes in the active site area. We then clarified the location of the binding sites for the three TNALP inhibitors studied, and we also exchanged inhibitor specificities between TNALP and PLALP. These data will help in drug design efforts to synthesize improved, selective, and drug-like TNALP inhibitors suitable for therapeutic use for the treatment of soft tissue ossification.

MATERIALS AND METHODS

Site-directed mutagenesis

Mutant variants of TNALP or PLALP were constructed using either the QuikChange XL site-directed mutagenesis kit (Stratagene, San Diego, CA, USA) or by PCR using the overlap extension approach as before.(37) TNALP-FLAG or PLALP-FLAG cDNA, cloned into pcDNA3 mammalian expression vector (Invitrogen, Carlsbad, CA, USA), were used as templates in the mutagenesis reactions. Final DNA constructs were checked by automatic sequencing for the presence of the desired mutation and absence of unwanted secondary mutations.

The QuikChange XL kit was used to introduce the following mutations: F108, (F108, Q109), A120, E434, G434, Q434, and (D433, E434) in TNALP, and E107, R428, H429, and (R428, H429) in PLALP. The following primers, together with the analogous reverse complementary primers, were used according to the mutagenesis protocol from the manufacturer—in TNALP: F108, 5′-CTGTGTGGGGTGAAGGCCAATTTCGGCACCGTGGGGGTAAGCGC-3′; (F108, Q109), 5′-GTGTGGGGTGAAGGCCAATTTCCAGACCGTGGGGGTAAGCGCAGCC-3′; A120, AGCGCAGCCACTGAGCGTGCCCGGTGCAACACCACCCAG-3′; E434, 5′-CAGTCTGCTGTGCCCCTGCGCGAGGAGACCCACGGCGGGGAG-3′; G434, 5′-CAGTCTGCTGTGCCCCTGCGCGGCGAGACCCACGGCGGGGAG-3′; Q434, 5′-CAGTCTGCTGTGCCCCTGCGCCAGGAGACCCACGGCGGGGAG-3′; (D433, E434), 5′-CAGTCTGCTGAGCCCCTGGACGAGGAGACCCACGGCGGGGAG-3′ and in PLALP: E107, 5′-CCTGTGCGGGGTCAAGGGCAACGAGCAGACCATTGGCTTGAGTGC; R428, 5′-CGGCAGCAGTCAGCAGTGCCCCTGCGCGAGGAGACCCACGCAGGCG-3′; H429, 5′-CGGCAGCAGTCAGCAGTGCCCCTGGACCACGAGAACCCACGCAGGCG-3′; (R428, H429), 5′-CGGCAGCAGTCAGCAGTGCCCCTGCGCCACGAGACCCACGCAGGCG-3′.

To create other mutants, a PCR mutagenesis approach based on the overlap extension method was used. In TNALP, primers 5′-GAAGGCCAATGAGGCCACCGTGGGGGTA-3′ for A109,5′-GAAGGCCAATGAGCAGACCGTGGGGGTA-3′ for Q109, 5′-AGCGCAGCCACTGAGCGTGACCGGTGCAACACCACCCAG-3′ for D120, 5′-AGCGCAGCCACGGAGCGTAACCGGTGCAACACCACCCAG-3′ for N120, 5′-GCCGCCTACGCCCACTCGGCTGCCCGGGACTGGTACTCAGAC-3′ for A166, 5′-GCCGCCTACGCCCACTCGGCTAACCGGGACTGGTACTCAGAC-3′ for N166, 5′-ACGCCCACTCGGCTGACCGGGCCTGGTACTCAGACAACGAG-3′ for A168, 5′-ACGCCCACTCGGCTGACCGGAACTGGTACTCAGACAACGAG-3′ for N168, 5′-CTTCACATTTGGTGGAGCCACCCCCCGTGGCAACTCT-3′ for A371, 5′-CAGTCTGCTGTGCCCCTGGCCCACGAGACCCACGGCGGGGAG-3′ for A433, 5′-CAGTCTGCTGTGCCCCTGGACCACGAGACCCACGGCGGGGAG-3′ for D433, 5′-CAGTCTGCTGTGCCCCTGCGCGCCGAGACCCACGGCGGGGAG-3′ for A434, and 5′-CTGCTGTGCCCCTGCGCTCCGAGACCACGGCGGGGAGG-3′ for S434 were used, and their reverse complementary analog were initially used together with flanking primers 5′-CTGGCCATTGGCACCTGCCTTACTAACTCCTTAGTGCCAGAG-3′ and 5′-GTGGTCAATTCTGCCTCCTTCCACCAGCAAGAAGAAGCC-3′ (for mutants in the region 108–166) or 5′-CTCCTGACCCTTGACCCCCACAATGTGGACTACCTATTGGGT-3′ and 5′-GTGCTGGAATTCGGCTTCTGCAGTTACTTGTCATCGTCGTCC-3′ (for mutants in the region 371–434) in two PCR reactions for each mutant. The PCR products were then combined in an overlap extension PCR reaction with the flanking primers. These PCR products were next ligated into pCRII/TOPO vector (Invitrogen), and sequenced to confirm the mutations. The resulting intermediate constructs were digested with PflMI, DraIII and EcoRI, or Eco91I and Eco81I and ligated together with the corresponding fragments of the template plasmid TNALP-FLAG/pcDNA3, digested with the same enzymes.

In PLALP, primers 5′-CAAGGGCAACTTCGGGACCATTGGCTTGAGTGC-3′ for G108, 5′-TGCGGGGTCAAGGGCAACGAAGGTACCATTGGCTTGAGT-3′ for (E107, G108), 5′-GCAGCCGCCCGCTTTAGCCAGTGCAACACGAC-3′ for S119, and 5′-ACCAGTTGCGGTCCACCGTGTGGGCGT-3′ for D165 were used in a similar approach using flanking primers 5′-ATGGGCGGTAGGCGTGTACGGTGGGAGG-3′ (forward) and 5′-CCGGTCTCGATGCGACCACCCTCCAC-3′ (reverse). The overlap extension PCR products were ligated into pCRII/TOPO vector, digested with PflMI, and ligated together with the corresponding main fragments of the PLALP-FLAG/pcDNA3 plasmid. The mutants (E107, G429)PLALP and (G108, G429)PLALP were made with similar ligation reactions, but using the main fragments of the plasmid (G429)PLALP-FLAG/pcDNA3(37) instead of the wildtype PLALP-FLAG/pcDNA3.

Expression and purification of enzymes

The mutant constructs were transfected into COS-7 cells for transient expression. Either the DEAE-dextran mediated method or a commercial transfection reagent SuperFect (Quiagen) was used for the transfections. The conditioned media with secreted proteins were collected after 3 and 6 days, concentrated if needed by ultrafiltration on YM-50 Centricon columns (Amicon, Beverly, MA, USA), and purified by affinity chromatography with anti-FLAG M2 antibody gel (Sigma, St Louis, MO, USA) according to the protocol from the manufacturer.

Enzyme kinetics experiments

Relative specific activities of the mutants were measured as described previously.(37) In brief, samples of the enzymes were added to microtiter plates coated with M2 anti-FLAG antibody, and saturating activities with the substrate pNPP were measured in 1 M DEA/HCl buffer, pH 9.8, containing 1 mM MgCl2 and 20 μM ZnCl2. The determinations of Km and the inhibition studies were done in the same buffer, with varying concentrations of pNPP and/or inhibitors. Levamisole (l-tetramisole), L-homoarginine, L-phenylalanine (all from Sigma) and theophylline (Fluka) were used as reagents in the inhibition studies. Ki values for the uncompetitive inhibitors were obtained from the inhibition studies using 20 mM pNPP (saturating substrate concentration) as well as at 1 mM pNPP. The results of enzyme kinetics studies were analyzed by nonlinear regression using software Prism 3.02 (GraphPad Software). Variations in Ki within a factor of 2 were not considered functionally relevant.

Molecular modeling

The 3D model of TNALP was constructed with the PLALP structure as a template using software MODELLER 3.0 as described previously.(10), (45) The further analysis of the TNALP structure and its comparison with that of PLALP was carried out with SwissPDB viewer,(46) RasMol, and TURBO-FRODO. Molecular models of the interaction of the ligands with TNALP were achieved using a simulated annealing-based protocol(47) with the program X-PLOR (version 3.1).(48) The parameter set used for these calculations is the CHARMM version 22.(49) For the ligands, geometrical and nonbonded parameters were derived from ab initio quantum calculations with the program GAUSSIAN98.(50) During the simulated annealing-based protocol, the ligand was driven to the catalytic site of the enzyme using a single distance restraint between one atom of the ligand and the phosphorus atom of the P-Ser 93 residue using the noe function of X-PLOR. For each ligand, this protocol was reiterated 30 times to give a corresponding set of solutions.

RESULTS

Computer modeling and mutagenesis

We constructed the 3D model of TNALP based on the known structure of PLALP,(51) using the software MODELLER 3.0. The two structures were superimposed, and the amino acid differences in the active site area (the 12 Å region around the catalytic Zn1 ion) were pinpointed (Fig. 2). The differences between human TNALP and chicken TNALP were also mapped, because it has been reported that chicken TNALP is less susceptible to inhibition by levamisole than human TNALP.(52) In total, six positions with amino acid differences were found, which could be clustered into two groups. The first group, using the TNALP numbers, includes residues 433 and 434, and the second group includes residues 108, 109, 120, and 166 (Table 1). In addition, residues Asp168 and Tyr371 in TNALP were also investigated as possible sites of interaction with inhibitor molecules, as suggested by our previous data.(37)

Table Table 1.. Active Site Residues Studied
original image
Figure FIG. 2..

Active site region of (A) human TNALP and (B) human PLALP. The highlighted residues are those that differ between TNALP and PLALP, with the exception of Y371. The residues are colored according to their chemical properties: blue = basic, red = acidic, violet = aromatic, yellow = aliphatic, and green = uncharged polar. The zinc atom is shown in orange, the magnesium is in green, and the phosphate moiety is shown in yellow and red. The figures were produced with the use of TURBO-FRODO.

Based on this analysis, we embarked on a mutagenesis study, constructing the following mutants: F108, A108, Q109, (F108, Q109), A120, D120, N120, A166, N166, A168, N168, A371, A433, D433, A434, Q434, G434, E434, S434, and (D433, E434) in TNALP and E107, G108, (E107, G108), S119, R428, H429, (R428, H429), (E107, G429), and (G108, G429) in PLALP. The (G429)PLALP mutant was constructed previously.(37) Mutations were introduced into the TNALP or PLALP cDNA cloned into the pcDNA3 expression vector. In all mutants, the COOH-terminal signal for the attachment of the GPI-anchor was substituted with a hydrophilic FLAG peptide as described before.(9) Resulting proteins, expressed in COS-7 cells, were secreted to the medium and could be purified by anti-FLAG affinity chromatography. The GPI-anchored membrane-bound TNALP was also studied and confirmed to have properties not significantly different from the FLAG-tagged TNALP (data not shown). In addition to human TNALP, we also expressed the GPI-anchored form of the chicken TNALP isozyme and studied its inhibition properties.

All of the mutant proteins retained significant activity when expressed in COS-7 cells, as can be observed from their kinetic parameters kcat and Km values presented in Table 2, which summarizes all the kinetic data discussed in this paper. The most noticeable decrease in kcat was found in the (A109), (Q109), (F108, Q109), (N168), (A371), (A433), and (D433) mutants of TNALP, whereas the largest increase in Km was found in the (E434) and (D433, E434) TNALP mutants.

Table Table 2.. Kinetic Parameters and Inhibition Constants (±SD) of Wildtype and Mutant TNALP and PLALP Isozymes
original image

Inhibition by l-amino acids

l-Homoarginine (L-hArg), a homolog of arginine, is a specific uncompetitive inhibitor of TNALP.(53) On the other hand, L-phenylalanine (L-Phe) is a well-known selective inhibitor of PLALP.(28) In agreement with previous data, we found that both human and chicken TNALP are inhibited by L-hArg with a Ki of about 2 mM, at least 30 times more efficiently than PLALP. L-Phe inhibited PLALP 25 times better than TNALP. To clarify which amino acid residues in TNALP and PLALP were responsible for these marked differences in inhibition selectivity, we produced a large number of mutated variants of both isozymes, concentrating on those residues that are different between TNALP and PLALP in the active site area. The results are presented in Table 2 and also shown graphically in Fig. 3 to facilitate the identification of the changes. Among the TNALP mutations investigated, the biggest changes in L-hArg inhibition were found for substitutions Phe-108 and Ala-371, which decreased inhibition by 35- and 22-fold, respectively, in (F108)TNALP and (A371)TNALP. The mutants (A108)TNALP, (F108, Q109)TNALP, (E434)TNALP, and (D433, E434)TNALP displayed a lesser, but still noticeable, negative change in inhibition. Apart from these changes, a large number of substitutions at different positions (Ala-120, Asn-120, Asp-120, Ala-166, Asn-166, Ala-168, Asn-168, Gln-109, Asp-433, Ala-433, Ala-434, Ser-434, Gly-434, Gln-434, and Ser-434) had no effect on the inhibition.

Figure FIG. 3..

Inhibition constants of PLALP, TNALP, and their mutants. Bar chart representation of Ki as a function of the mutants for L-hArg, L-Phe, levamisole, and theophylline, measured at 20 mM pNPP. Native enzymes are identified as wildtype (wt), TNALP mutants (TNALP), and PLALP mutants (PLALP) and indicated by the amino acid substitution.

In contrast, several mutations made in the context of PLALP improved the inhibition by L-hArg, most significantly in (E107)PLALP, but also in the (H429)PLALP, (R428, H429)PLALP, (G429)PLALP, and (E107, G108)PLALP mutants. The double mutant (E107, G429)PLALP was inhibited by L-hArg as strongly as the wildtype TNALP enzyme. When the inhibition by L-Phe was tested, a reversed effect was found for mutations (F108)TNALP and (E107)PLALP, which now significantly improved or decreased the inhibition, respectively, compared with the wildtype enzymes. A similar effect was found for reciprocal substitutions (Q109)TNALP and (G108)PLALP.

On the contrary, the effects of substitutions at position 434 in TNALP, particularly (E434), on the L-Phe inhibition were parallel to those observed for L-hArg, and mutation (H429) in PLALP decreased Ki for both inhibitors, so this residue cannot explain the selectivity of inhibition. Also, the substitution Tyr-371 in TNALP reduced L-Phe inhibition significantly, as it did for L-hArg inhibition. Apart for a small decrease in inhibition found in (N168)TNALP, all other mutations had no effect on inhibition.

Inhibition by levamisole

Levamisole, the l-stereoisomer of tetramisole, is a potent uncompetitive inhibitor of TNALP. In fact, it is nearly 100 times more effective than L-hArg. It shows marked preference toward TNALP. In contrast with the data obtained for L-hArg, we found that levamisole inhibition depends very much on the nature of residue 434 in TNALP. Substitutions H434E (as in PLALP), H434Q (as in chicken TNALP), H434S (as in the intestinal isozyme), H434G (as in GCAP), or H434A all lead to a significant decrease in inhibition. The (E434)TNALP mutant displayed a greatly increased Ki value; in fact, it was inhibited about as weakly as PLALP. We also confirmed that wildtype chicken TNALP is much less inhibited by levamisole (about 8.5-fold) than human TNALP as previously reported.(52)

In contrast to the substitutions at His-434 in TNALP, mutations at positions 108 (Ala-108 and Phe-108), 120 (Ala-120, Asp-120, and Asn-120), 166 (Ala-166 and Asn-166), 168 (Ala-168 and Asn-168), and 433 (Ala-433 and Asp-433) did not affect the inhibition by levamisole. A 2-fold increase in Ki was observed in TNALP mutants containing the substitution G109Q and in PLALP mutants containing the mutation F107E. However, no effect was found for the reciprocal mutations Q108G in PLALP and E108F in TNALP or in the (A108)TNALP mutant, suggesting that the positions 108 and 109 (TNALP numbering) play only a minor role in levamisole inhibition. We also observed a 2-fold decrease in inhibition caused by the Asp-433 mutation in the double (D433, E434)TNALP mutant compared with the (E434)TNALP mutant and a similar effect in the reciprocal PLALP mutants. However, this effect was not seen in the single Ala-433 and Asp-433 substitutions, suggesting that in this case, residue 433 did not directly interact with the inhibitor, but rather indirectly influenced the interaction of residue 434 with levamisole. It is likely that the side chains of residues 433 and 434 in TNALP have high mobility, which makes it difficult to assess the exact nature of their interactions.

Besides the substitutions at His-434, the only mutation that significantly influenced the inhibition by levamisole was Ala-371, increasing the Ki value about eight times. This suggests that, together with His-434, Tyr-371 in wildtype TNALP forms the binding area for levamisole. We performed a double inhibition experiment using both L-hArg and levamisole on wildtype TNALP (Fig. 4). The parallel lines in the reciprocal plot of v versus inhibitor concentration suggest that the two uncompetitive inhibitors can only act independently on the enzyme (i.e., no simultaneous binding is possible and their effect is additive, although both inhibitors are stabilized in the active site pocket on different amino acid residues).

Figure FIG. 4..

Additive uncompetitive inhibition of TNALP. Plot of residual ALP enzyme activity (1/v) as a function of the concentration of levamisole, measured with 20 mM pNPP in the presence of increasing concentrations of l-homoarginine (▪, absent; ▾, 0.5 mM; ○, 2.5 mM).

Inhibition by theophylline

Theophylline (a 1,3-dimethyl derivative of xanthine) is a potent inhibitor of TNALP. Double reciprocal plots of v versus substrate concentration (Fig. 5A) confirmed that TNALP is inhibited by theophylline in an uncompetitive manner, but in agreement with findings by Glogowski et al.,(54) we found evidence of substrate inhibition. This did not preclude assessing Ki values during inhibition studies, as shown from secondary replots of the y-intercepts versus the concentration of theophylline, enabling correct assessment of the Ki (Fig. 5A, inset). Moreover, at [pNPP] = 1 mM, substrate inhibition was virtually absent, motivating us to study enzyme inhibition of TNALP and PLALP mutants both at 1 and 20 mM pNPP (Table 2). In general, lower apparent Kis were found with 20 mM pNPP, as expected. In agreement with published data, we thus found that theophylline is a much better inhibitor of TNALP than of PLALP. In addition, mutagenesis of TNALP, while reducing the affinity of theophylline for the resulting mutant, did not modify the uncompetitive nature of the inhibition, as shown for the (Q434)TNALP mutant (Fig. 5B). As with levamisole inhibition, the main determinant of sensitivity to theophylline is His-434. The H434E substitution (as in PLALP) causes a 50-fold reduction in inhibition, making the corresponding TNALP mutant even less inhibited than wildtype PLALP. The reciprocal E429H mutation in PLALP improves the apparent Ki for theophylline inhibition by 30-fold, nearly to the level of wildtype TNALP. Thus, strikingly, only this single substitution can account for all the differences in theophylline inhibition between TNALP and PLALP. Several other substitutions at position 434 have been investigated and displayed intermediate apparent Ki values between those of TNALP (25 μM measured at 20 mM pNPP) and PLALP (947 μM measured at 20 mM pNPP). One such mutation is H434S (as in the intestinal isozymes). Earlier experimental data showed that human intestinal ALP is inhibited by theophylline better than PLALP. In agreement with these data, we found that (S434)TNALP is inhibited noticeably better than (E434)TNALP, although worse than wildtype TNALP. Similarly, chicken TNALP, which has Gln-434, has a Ki value of 111 μM, measured at 20 mM pNPP.

Figure FIG. 5..

Uncompetitive nature of TNALP inhibition by theophylline. Double-reciprocal plot of ALP enzyme activity (v) vs. substrate concentration ([S]) during the inhibition of TNALP by increasing concentrations of theophylline. (A) Inhibition of wildtype TNALP by increasing concentrations of theophylline (▪, absent; ▴, 25 μM; ▾, 50 μM). The inset shows the replot of 1/V(max) vs. [Theophylline] (μM) that enabled correct assessment of Ki. (B) Inhibition of (H434Q)TNALP by increasing concentrations of theophylline (▪, absent; ▴, 250 μM; ▾, 500 μM; •, 750 μM). The inset shows the replot of 1/V(max) vs. [Theophylline] (μM) that enabled correct assessment of Ki.

Despite the fact that His-434 is clearly the most important residue for theophylline inhibition, we found that the inhibition can also be influenced by the nature of residues 433, 371, and 108. TNALP has Arg in position 433. Mutations R433A and R433D (as in PLALP) decreased theophylline inhibition two times. This effect was also noticeable in (D433, E434)TNALP and in the reciprocal (R433, H434)PLALP double mutant. The Y371A substitution only showed a 2-fold decrease in inhibition. Interestingly, theophylline inhibition in TNALP could be further improved by introducing Phe-108 (as in PLALP) instead of Glu in TNALP.

DISCUSSION

In PLALP, the best-studied mammalian ALP isozyme, two residues have been implicated in the binding of uncompetitive inhibitors to the active site. The first is Glu-429 (Gly-429 in GCAP), which determines the selectivity toward L-Leu between these two isozymes.(34), (35), (55) The second is Tyr-367, which we have shown to also have a significant effect on the inhibition by L-Phe and L-Leu.(37) As was suggested by computer modeling, this residue, perfectly conserved in all mammalian ALPs, constitutes part of the hydrophobic binding site for the side chain of the inhibitor. However, nothing is currently known about the residues that determine the specificity in inhibition in the TNALP molecule.

In this study, we have therefore produced a number of TNALP and PLALP mutants, the catalytic efficiency of which was only mildly affected. Therefore, these mutants were suitable to study uncompetitive inhibition by L-hArg, L-Phe, levamisole, and theophylline, because differences in Ki measured for different ALP mutants reflect affinity changes in the active site pocket for the different inhibitors, rather than mutation-dependent changes in kinetic parameters.(36) The first obvious conclusion relates to the crucial role of TNALP residue 108 in determining the selectivity of inhibition by amino acids. A single substitution, that is, E108F in TNALP, completely reversed the pattern of TNALP inhibition, rendering it more than 10 times more sensitive to L-Phe than to L-hArg inhibition. The reciprocal mutation, that is, F107E in PLALP, made the enzyme nearly equally sensitive to L-Phe and L-hArg. However the double (E108, G109)PLALP mutant displayed a 2-fold selectivity toward L-hArg compared with L-Phe. The importance of residue 108 in TNALP is further supported by analysis of the (A108)TNALP mutant, which displayed a 10-fold reduction in inhibition by L-hArg and a slight reduction in L-Phe inhibition. These data suggest that selectivity toward l-amino acid inhibition, in both TNALP and PLALP, is largely determined by residues 108 and 109. Docking studies for L-Phe in the PLALP active site(37) and of L-hArg in the TNALP active site (Fig. 6A) give additional support to this conclusion, because they show the side chains of amino acid inhibitors contacting with the pocket formed by residue 108 and its neighbors. Residue 108 lies within the same pocket as Tyr-371 (Tyr-367 in PLALP), which, as we have recently shown,(37) is important for the efficient inhibition of PLALP by L-Phe. If the suggested mode of binding of amino acid side chains in the active site of TNALP or PLALP is correct, one should expect reduced inhibition in the case of Ala-371 substitution in TNALP. The data obtained for the Ala-371 proves that this is true. When tested for either L-Phe or L-hArg inhibition, (A371)TNALP is one of the least efficiently inhibited mutants in this study.

Figure FIG. 6..

Calculated optimal docking of the inhibitors (A) l-homoarginine, (B) levamisole, and (C) theophylline into the modeled active site of TNALP. The figures were produced with the use of TURBO-FRODO.

In addition to residues 108, 109, and 371, another position that has a large effect on the inhibition is His-434 (Glu-429 in PLALP). In PLALP, the important effect of substituting Glu-429 has been clearly shown.(34), (35), (55) However, here we show that changes at His-434 in TNALP do not significantly affect the inhibition by either L-Phe or L-hArg (Table 2; Fig. 3). There is some experimental uncertainty in the determination of the Ki, as reflected by the SD values shown in Table 2. For this reason, we have not interpreted differences in Ki smaller than a factor of 2. Thus, we can state that mutagenesis of residue 434 in TNALP hardly affects inhibition by L-Phe or by L-hArg. These substitutions include H434Q (a substitution distinguishing chicken TNALP from human TNALP), H434S (as in the intestinal isozyme), H434G (as in GCAP), and H434A. The only exception was observed in the (E434)TNALP mutant. Here a noticeable increase in Ki for both L-Phe and L-hArg was observed, and the inhibition was further reduced in the (D433, E434)TNALP double mutant. Interestingly, the single substitutions at residue 433 did not affect the inhibition. In the reciprocal (H429)PLALP and (R428, H429)PLALP mutants, a 2-fold improvement in inhibition was observed for L-Phe, whereas the inhibition by L-hArg was increased 4- to 5-fold. Thus, all tested substitutions at residue 434 in TNALP or 429 in PLALP had parallel effects on the inhibition by L-Phe and L-hArg.

Summarizing the data for L-Phe and L-hArg inhibition, we can conclude that the binding of amino acid inhibitors occurs in the area E108-G109-Y371-H434 in TNALP (equivalent to F107-Q108-Y367-E429 in PLALP). The specificity of L-hArg inhibition in TNALP is most likely determined by electrostatic interaction of the positively charged side chain of the inhibitor with residue Glu-108 in the active site. The conserved Tyr-371 is also necessary for the binding, most likely by providing the hydrophobic area for the nonpolar alkyl part of the inhibitor side chain. In PLALP, the residues Phe-107 and Gln-108, together with Tyr-367, provide a hydrophobic pocket that accommodates the phenyl ring of phenylalanine. Finally, residue 434 in TNALP (429 in PLALP) modulates the inhibition. Here the negatively charged glutamate side chain is clearly unfavorable for the inhibition by either L-hArg or L-Phe. However, other substitutions at 434 show nearly the same level of inhibition as in wildtype TNALP. We conclude that residue 434 interacts with the carboxyl part of amino acid inhibitors, and the electrostatic effect is most important, while the specific nature of the side chain at 434 plays minor role.

Levamisole, the l-stereoisomer of tetramisole, is a well-known potent uncompetitive inhibitor of TNALP,(31) and it had been reported that chicken TNALP is about eight times less sensitive to levamisole than human TNALP.(52) Chicken TNALP has only a few substitutions in the active site area compared with human TNALP, one of them being glutamine at position 434. Our results showed that chicken TNALP is inhibited comparably, although slightly better than, the (Q434)TNALP variant of TNALP. The fact that such different amino acids as Gly or Glu at this position all showed a marked decrease in inhibition suggests that the effect is not caused by steric hindrance, but rather depends on the particular presence of the amino acid His at this position. A possible explanation for the role of His-434 is a stacking interaction between the flat hydrophobic groups of the protein and the ligand. The reciprocal E429H mutation in PLALP improved the inhibition by levamisole about 10 times, further confirming the crucial role of this residue. We conclude that the binding area for levamisole partially overlaps with the binding pocket for amino acid inhibitors, but is more spatially restricted than the latter. A more important difference between the two inhibitors is the nature of their interactions with the active site. In contrast with the inhibition by amino acids, the selectivity of levamisole toward the TNALP isozyme is nearly fully explained by the substitution at residue 434 (Table 2; Fig. 6B).

While our mutagenesis study has emphasized that His-434 is the main determinant of theophylline binding to TNALP, it is not possible to establish the exact orientation of the theophylline molecule in the active site. Data in the literature have documented the effect of several substitutions in the xanthine structure.(56) For example, substitutions at N7, C8, and N9 lead to compounds with no inhibitory activity (see Fig. 1 for numbering). One can conclude that N7 or N9 is involved in important interactions with the enzyme, probably through Zn1 ion in the active site. In contrast, several substitutions at N1 and N3 in theophylline gave active compounds with altered selectivities toward bovine TNALP or intestinal isozymes. For example, a negatively charged 1-carboxymethyl derivative of theophylline has a greatly decreased activity against the TNALP isozyme but enhanced activity against the intestinal isozyme, thus being a selective inhibitor of the latter. Interestingly, bovine intestinal isozymes have a positively charged residue (Arg or Lys) at the position homologous to Gly-109 in TNALP and a noncharged tyrosine at the position homologous to negatively charged Glu-108 in TNALP. These data, together with the positive effect of the Phe-108 substitution on theophylline inhibition in our study, suggest that the binding area for the large ring of theophylline might lie near TNALP residues 108/109 and that the nature of the substitutions at N1 (or N3) in theophylline should be complementary to the nature of residues 108/109 for effective inhibition to take place. Mutations at residue 434 lead to a wide range of theophylline inhibition levels, suggestive of close interactions between residue 434 and the inhibitor (Fig. 6C). The reason for the special significance of His-434 for the inhibition by theophylline (and levamisole) is not fully clear. As already discussed, a similar flattened structure of both inhibitors may help form favorable stacking interactions with His-434. In the PLALP structure, residues Asp-428 and Glu-429 have high side chain mobility, as judged by the B-factor values. The same should hold true for the homologous residues Arg-433 and His-434 in TNALP. At the same time, the increased mobility of the side chains of these residues may lead to a better adjustment during the binding of the inhibitor molecules. While there is a consensus about the uncompetitive nature of TNALP inhibition in the case of l-amino acids and levamisole, inhibition by theophylline has been found to be either uncompetitive or noncompetitive.(32), (38) We found theophylline inhibition of wildtype TNALP to be uncompetitive, but also found, in agreement with Glogowski et al.,(54) evidence of substrate inhibition at high substrate concentrations, complicating a correct assessment of the Ki values for theophylline inhibition. Apparent Ki values reported in Table 2 have therefore been determined both at high (20 mM) and low (1 mM) substrate concentrations.

In conclusion, by using four established uncompetitive inhibitors of ALPs, this study identified all amino acid residues, explaining the selectivity of inhibition of human TNALP by these inhibitors. Through drug design approaches, these findings will pave the way for the development of more specific and safe TNALP inhibitors for therapeutic use to treat pathological soft tissue mineralization disorders.

Acknowledgements

This work was supported in part by National Institutes of Health Grants DE 12889 and AR 47908. AK was supported by a stipend from Umeå University.

Ancillary