Interaction Energies between Tetrahydrobiopterin ... - ACS Publications

J. Phys. Chem. B 2007, 111, 9651-9654

9651

Interaction Energies between Tetrahydrobiopterin Analogues and Aromatic Residues in Tyrosine Hydroxylase and Phenylalanine Hydroxylase Meghan E. Hofto, Jessica N. Cross, and Mauricio Cafiero* Department of Chemistry, Rhodes College, 2000 North Parkway, Memphis, Tennessee 38112 ReceiVed: March 30, 2007; In Final Form: June 4, 2007

The phenylalanine residues 300 and 309 in the enzyme tyrosine hydroxylase are known to aid in the positioning and binding of tetrahydrobiopterin (BH4) to the enzyme active site. The residues phenylalanine 254 and tyrosine 325 similarly aid in binding BH4 in phenylalanine hydroxylase. BH4 is a cofactor necessary for enzyme function, and mutations in these residues have been shown to cause a decrease in enzyme function. We examine the pairwise interactions between each aromatic residue and BH4 using second-order Moller Plesset theory and density functional theory to determine the amount of binding due to these aromatic residues. Further, we perform in silico point mutations of these residues to determine if several likely mutations can cause a decrease in protein function. Our results show that dispersion dominates these interactions, and electrostatics alone is not enough to bind the BH4.

Introduction Tyrosine hydroxylase (TyrOH) is an enzyme that catalyzes the hydroxylation of tyrosine to L-DOPA, which is then used in the biosynthesis of neurotransmitters. Mutations in TyrOH that affect the binding site can lead to neurological disorders, including depression and Parkinson’s disease. Phenylalanine hydroxylase (PheOH) is an enzyme that catalyzes the hydroxylation of phenylalanine into tyrosine, for use by TyrOH in the production of DOPA. Mutations in PheOH are implicated in the disease phenylketanuria (PKU). Tetrahydrobiopterin (BH4) is an aromatic cofactor molecule used in both of the hydroxylation reactions mentioned above; this molecule is held in place partially via π-stacking and T-shaped aromatic interactions with aromatic enzyme residues 300 and 309 in TyrOH and residues 254 and 325 in PheOH. Crystal structures for wild-type and mutant (F300A, F309A) TyrOH bound to a BH4 analogue have been reported,1,2 and the effects of the mutations on the protein function have been found. Crystal structures of PheOH bound with BH4 have also been reported.3 In the current work, we compute the interaction energies between the cofactor and the 300 and 309 residues using second-order Moller Plesset theory (MP2) and several density functional theory (DFT) methods. Since the wild-type residues at 300 and 309 (TyrOH) and 254 and 325 (PheOH) are aromatic (phenylalanine and tyrosine) and BH4 contains an aromatic ring, the interaction energy is dominated by dispersion-type interactions. Our results for interaction energies agree well with the experimentally determined protein efficiencies and disease symptoms for the wildtype and mutant enzymes. We also apply our methods to several other likely mutations and predict the degree to which binding may be affected. Our MP2 results are used as a standard in this work, even though MP2 is known to overestimate aromatic interactions when compared to coupled cluster calculations [CCSD(T)]. Excellent work by Sinnokrot and Sherrill et al.4-7 and Tsuzuki et al.8 has shown the difference between MP2 and CCSD(T) to be on the order of 0.5 to 1 kcal/mol in the types of situations * Corresponding author. E-mail: [email protected].

we are studying. Since we are interested in differences in binding energies upon mutation rather than absolute energies, we consider the MP2 energies to be of sufficient accuracy. We also examine several DFT methods for their ability to model the medium range dispersion-like interactions in TyrOH. Our previous work9-12 has shown that HCTH40713 and, in some cases, SVWN14,15 provide good results for aromatic interactions, despite the fact that DFT, being a local model, does not rigorously model dispersion. Again, as we are attempting to predict differences in interaction energies upon mutation, the inaccuracy in DFT should not change the results. Zhao and Truhlar16,17 have also studied the ability of DFT methods to predict weak interactions, and their results, on different types of systems from those studied here, are similar to ours. Many computational protein studies are done using the DFT method B3LYP,18 which, while excellent in many situations, is generally very bad for the electrostatic and medium range dispersionlike interactions in protein/ligand binding. We include B3LYP in the current work for comparison. Hartree-Fock (HF) results with a reasonably large basis set are also reported as a measure of the pure electrostatic interactions. Computational Methods The crystal structure of wild-type rat TyrOH bound to BH4 was used to isolate the molecule pairs Phe300/BH4 and Phe309/ BH4. The crystal structure of PheOH was used to isolate the molecule pairs Phe254/BH4 and Tyr325/BH4. For each pair, hydrogens were added to the crystal structures and the amino acid residue in each was capped with OH to produce neutral, closed shell amino acid/BH4 complexes. The positions of the hydrogens and the OH were optimized using B3LYP/6-31G while keeping the heavy atoms fixed. The counterpoisecorrected19 interaction energies were then computed by difference between the complex and the single molecules using HF and MP2 with the 6-311G(d,p) basis set20 and DFT with the 6-311++G(d,p) basis set,20,21 where DFT ) B3LYP, HCTH407, and SVWN. We acknowledge that the addition of diffuse functions to the basis set used for the MP2 calculations would increase the absolute accuracy of the calculated interaction

10.1021/jp072518w CCC: $37.00 © 2007 American Chemical Society Published on Web 07/20/2007

9652 J. Phys. Chem. B, Vol. 111, No. 32, 2007

Hofto et al.

Figure 1. Optimized structures of wild-type residues with BH4.

energies, but our previous work on aromatic interactions9 suggests that they are not necessary in our current work. This previous work shows a comparison of T-shaped and sandwichtype dispersion interactions between aromatic molecules both with and without diffuse functions. The addition of diffuse functions increases the interaction energies of both sandwich and T-shaped complexes and increases the interactions of sandwich complexes relative to T-shaped complexes, but, for the purpose of comparing interactions within one type of complex at a time, which is what we do in the current work, the addition of these functions does not affect the conclusions drawn. The Phe309/BH4 and Tyr325/BH4 complexes form displaced T-shaped complexes, while the Phe300/BH4 and Phe254/BH4 complexes form displaced sandwich complexes. For the PheOH residue, a BH4 analogue lacking protons on the two nonaromatic-ring nitrogen atoms was used. For convenience, we refer to this molecule as BH4 in the following paragraphs. For each complex, the amino acid residue was mutated by changing the side chain and re-optimizing the newly added atoms while leaving the BH4 molecule and the amino acid backbone atoms fixed. The mutations included changing phenylalanine in each case to alanine, glycine, leucine, isoleucine, tyrosine, and methionine (in the case of Tyr325, we substitute phenylalanine in the above list). The interaction energies between the mutated side chains and the BH4 molecule were computed as above. The mutation model we use assumes that the active-site structure does not change significantly upon mutation. Since the wild-type residues are both phenylalanine, mutations to other large, nonpolar residues (Leu, Ile) likely do not change the active-site geometry very much, and therefore our method should be a good approximation. In the other cases, the activesite structure may relax more, and our model may not be as accurate. All calculations were done using the PQS software.22

TABLE 1: Counterpoise-Corrected Interaction Energies between BH4 and the Residues at Position 300 in TyrOH (in kcal/mol)a BH4 with Phe300 Ala300 Leu300 Ile300 Gly300 Met300 Tyr300

HF

B3LYP

HCTH407

SVWN

MP2

%

5.18 0.82 12.70 4.39 -0.30 4.97 5.50

3.64 0.49 8.93 2.76 -0.28 3.1 3.51

2.65 -0.58 10.8 3.00 -0.77 3.63 2.35

-5.43 -1.92 -4.19 -6.39 -0.57 -7.03 -5.71

-6.43 -1.49 -0.10 -4.29 -0.84 -4.46 -6.87

100 23 1 67 13 70 107

a DFT calculations use the 6-311++G** basis set, and HF and MP2 calculations use the 6-311G** basis set.

Sandwich Complex Results: Phe300 and Phe254 Figure 1 shows the optimized structures of the four wildtype residues with the BH4 analogue molecules. Coordinates of the structures for the wild type and all mutants are available from the authors. Phenylalanine 300 in TyrOH forms a sandwich-type conformation with BH4. Ellis et al. show that Phe300 is important in both steps of the enzymatic reaction (i.e., for the hydroxylation of both BH4 and tyrosine).2 The mutation F300A (phenylalanine to alanine) experimentally reduces the production of DOPA by a factor of 2.5. Our results for interaction energies between various residues at the 300 position and BH4 are shown in Table 1. First, we see that the wild-type enzyme with Phe300 binds BH4 quite strongly, -6.43 kcal/mol, according to MP2 (our standard in this work). The HF results for this same pair (repulsive by 5.18 kcal/mol) show that electrostatic interactions alone are not enough to produce the binding interaction. This makes sense, because in the sandwich arrangement the π-clouds of each ring group are directed toward each other and create repulsion. The difference between MP2 and HF, which can be attributed largely to dispersion, is ∼11.6 kcal/mol. The mutation of Phe300 to Ala300 simply removes the phenyl ring from the amino acid residue, while leaving a methyl group

Energies in BH4 Analogues and Aromatic Residues in place. By comparing the HF wild-type results with the Ala300 results, we can see that the methyl group alone is much less repulsive: it contributes only 20% of the total electrostatic repulsion. This reduced repulsion is due to the removal of the large π-cloud and the increased distance from the BH4 to the residue. The MP2 results for this mutation show that the methyl group contributes only about 20% of the total attractive dispersion energy of Phe300. This data correlates very well with the experimental result that this mutation causes a significant drop in production of L-DOPA. The mutation of Phe300 to Gly300 removes all side groups and is, in essence, a measure of the interaction between the residue backbone and the BH4 molecule. We can see that the backbone of the residue alone is bound to BH4 both electrostatically (HF) and by dispersion (MP2), although very weakly in both cases. Here it seems that electrostatics dominates the dispersion forces. This mutation would likely lead to a drop in the production of DOPA. The mutation to leucine is an interesting case. It would be expected that this mutation, being from one nonpolar group to another nonpolar group, would be rather benign, but in fact there is a large electrostatic repulsion (HF) between the residue and BH4. This may be because the symmetric leucine side chain forces a methyl group closely over the “front” ring of BH4, which includes three nitrogen atoms in the ring, and an oxygen hanging off the ring. These electronegative atoms may cause a strong repulsion on the close methyl group. The dispersion interaction (MP2) is just large enough to cancel the electrostatic component. This mutation would likely lead to a drop in the production of L-DOPA. The isoleucine and methionine mutations behave very similarly. Electrostatically (HF), they are both slightly less repulsive than the wild-type Phe300, and they are both bound by dispersion (MP2) at about 70% of the strength of the wild type. This is likely because both side chains feature a flexible structure that is bent away from the “front” ring of BH4. The geometry optimization puts these chains out over the back end of the molecule, over the dihydroxypropyl group. The tyrosine mutation leads to electrostatic and dispersion binding almost identical to that of phenylalanine, with slightly increased dispersion due to the extra electrons on the oxygen. All of these seem to retain enough binding to imply regular function of the protein. The mutation of the phenylalanine 254 residue in PheOH to an isoleucine residue is implicated in the disease PKU. Therapy for PKU can include flooding the system with excess BH4, thus implying that if enough BH4 is present, whatever problem the enzyme has can be overcome. This suggests the possibility of a simple decrease in the binding energy of BH4 as the modus operandi of this disease-causing mutation. Much like Phe300 in TyrOH, the Phe254 residue forms a sandwich complex with the BH4 molecule, and the interactions are expected to follow analogous trends. The interaction energies between BH4 and residues at the 254 position are shown in Table 2. As in TyrOH, Phe254 forms a strong complex with BH4, with an interaction energy of -5.54 kcal/mol (MP2), with a difference between HF and MP2 of about 10 kcal/mol, again as in TyrOH. The behavior upon mutation to alanine is nearly identical to that of TyrOH, with both the HF repulsion and the MP2 attraction decreasing because of the removal of the phenyl ring. In fact, the only real differences between the two residues in the two enzymes are in the leucine and isoleucine mutations. While TyrOH showed a small drop in MP2 interaction energy for the isoleucine mutation, PheOH showed a dramatic drop in

J. Phys. Chem. B, Vol. 111, No. 32, 2007 9653 TABLE 2: Counterpoise-Corrected Interaction Energies between BH4 and the Residues at Position 254 in PheOH (in kcal/mol)a BH4 with

HF

B3LYP

HCTH407

SVWN

MP2

%

Phe254 Ala254 Leu254 Ile254 Gly254 Met254 Tyr254

5.61 0.71 3.14 6.71 0.00 0.73 6.12

3.76 0.58 2.20 4.83 0.10 0.32 4.34

2.84 -0.90 0.97 4.97 -0.65 -1.47 3.15

-6.44 -2.27 -5.12 -4.95 -0.33 -5.26 -5.59

-5.54 -1.53 -3.55 -1.27 -0.59 -4.27 -5.02

100 35 79 77 5 82 87

a DFT calculations use the 6-311++G** basis set, and HF and MP2 calculations use the 6-311G** basis set.

TABLE 3: Counterpoise-Corrected Interaction Energies between BH4 and the Residues at Position 309 in TyrOH (in kcal/mol)a BH4 with Phe309 Ala309 Leu309 Ile309 Gly309 Met309 Tyr309

HF

B3LYP

HCTH407

SVWN

MP2

%

-1.55 -1.34 -0.649 -1.38 -1.33 -1.07 -2.26

-2.68 -2.47 -1.94 -2.46 -2.43 -2.19 -3.23

-2.69 -1.95 -1.5 -2.34 -1.86 -2 -3.5

-10.6 -8.11 -9.57 -8.55 -8.01 -8.23 -10.5

-6.66 -3.93 -4.93 -4.61 -3.74 -4.31 -6.74

100 59 74 69 56 65 101

a DFT calculations use the 6-311++G** basis set, and HF and MP2 calculations use the 6-311G** basis set.

interaction energy for this same mutation. This dramatic drop supports the fact that this mutation causes PKU symptoms. Of the DFT methods examined for these sandwich complexes, both the hybrid B3LYP and the heavily parametrized HCTH407 fail to predict the correct behavior of the interaction energies when compared to the results of MP2. Both seem to add a small amount of attractive interaction to the pure electrostatic interaction, though neither does a good job. The SVWN functional does show a very good ability to predict the trends. With the exception of a very high binding interaction for the leucine mutation in TyrOH and the isoleucine mutation in PheOH, SVWN seems to follow the MP2 numbers fairly well and would be an excellent qualitative substitute for MP2. This supports our previous finding that SVWN can mimic medium range dispersion-like interactions between benzene and polyaromatic rings very well.9,12 T-Shaped Complex Results: Phe309 and Tyr325 Table 3 shows our results for the Phe309/BH4 complex. Ellis et al. found that the mutation of phenylalanine to alanine at the 309 site reduces the production of DOPA by 40%.2 The authors believe that the production of DOPA proceeds in two steps: the hydroxylation of BH4, followed by the hydroxylation of tyrosine into DOPA. While DOPA decreases with the F309A mutation, the amount of BH4 intermediate does not decrease; thus, the authors hypothesize that Phe309 does not affect the first step, it only indirectly affects the second step (hydroxylation of tyrosine into DOPA). Our MP2 results show that there is a large decrease in interaction energy between the 309 residue and the BH4 upon mutation to alanine; furthermore, this change in interaction energy (41% with MP2) corresponds closely to the 40% decrease in DOPA production. We believe that the 309 residue must be quite important in binding the BH4 both before and after its own hydroxylation. Phe309 is farther away from the bound BH4 molecule than Phe300, and therefore the long-range interactions are less dramatically affected by mutations. All of the mutations examined maintain an attractive electrostatic interaction (from

9654 J. Phys. Chem. B, Vol. 111, No. 32, 2007

Hofto et al.

TABLE 4: Counterpoise-Corrected Interaction Energies between BH4 and the Residues at Position 325 in PheOH (in kcal/mol)a BH4 with

HF

B3LYP

HCTH407

SVWN

MP2

%

Tyr325 Ala325 Leu325 Ile325 Gly325 Met325 Phe325

2.52 0.25 1.73 1.69 0.22 2.78 2.32

1.69 0.31 1.36 1.32 0.21 1.40 1.52

1.05 -0.47 0.26 0.34 0.00 1.51 0.93

-3.05 -0.15 -2.03 -1.82 0.10 -5.10 -3.23

-1.78 -0.33 -0.54 -0.65 -0.07 -1.22 -1.85

100 18 31 37 4 69 104

a DFT calculations use the 6-311++G** basis set, and HF and MP2 calculations use the 6-311G** basis set.

HF) of about 1-2 kcal/mol and an attractive dispersion interaction (MP2-HF) of about 2-4 kcal/mol. It would seem that no mutation in this set would have an effect greater than the experimentally observed 40% drop in DOPA production for the alanine mutation. The DFT methods B3LYP and HCTH407 behave very similarly to HF, while SVWN overestimates the MP2 results. In this case, either HCTH407 or B3LYP would be a better mimic of MP2 than SVWN. Table 4 shows the results for the Tyr325/BH4 complex. Although it fills an analogous role for PheOH to Phe309 in TyrOH, the results are strikingly different. Tyr325 binds with about a third of the interaction energy of Phe309, though the trends on the mutations are similar. None of the mutations show a dramatic difference, with the exception of glycine, which shows negligible interaction. The differences in the interaction energies are likely because the positions of the residues in the two enzymes are reversed, with one binding BH4 from the left and the other from the right. Furthermore, in this case, it seems that SVWN is again the best choice for a DFT method that mimics the MP2 numbers very well. HCTH407 fails to find attractive interactions with the exception of the alanine mutation. Acknowledgment. This work was supported by a Rhodes College faculty development grant. Supporting Information Available: Interaction energies between tetrahydrobiopterin analogues and aromatic residues

in tyrosine hydroxylase and phenylalanine hydroxylase. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Goodwill, K. E.; Sabatier, C.; Stevens, R. C. Biochemistry 1998, 37, 13437-13445. (2) Ellis, H. R.; Daubner, S. C.; McCulloch, R. I.; Fitzpatrick, P. F. Biochemistry 1999, 38, 10909-10914. (3) Andersen, O. A.; Stokka, A. J.; Flatmark, T.; Hough, E. J. Mol. Biol. 2003, 333, 747. (4) Sinnokrot, M. O.; Sherrill, C. D. J. Phys. Chem. A 2006, 110, 10656-10668. (5) Sinnokrot, M. O.; Sherrill, C. D. J. Am. Chem. Soc. 2004, 126, 7690-7697. (6) Sinnokrot, M. O.; Sherrill, C. D. J. Phys. Chem. A 2003, 107, 83778399. (7) Ringer, A. L.; Sinnokrot, M. O.; Lively, R. P.; Sherrill, C. D. Chem.-Eur. J. 2006, 12, 3821-3828. (8) Tsuzuki, S.; Honda, K.; Uchimaru, T.; Mikami, M.; Tanabe, K. J. Am. Chem. Soc. 2001, 124, 104-112. (9) Van Sickle, K.; Culberson, L. M.; Holzmacher, J. L.; Cafiero, M. Int. J. Quantum Chem. 2007, 107, 1523-1531. (10) Godfrey-Kittle, A.; Cafiero, M. Int. J. Quantum Chem. 2006, 106, 2035-2046. (11) Hofto, M. E.; Godfrey-Kittle, A.; Cafiero, M. J. Mol. Struct.: THEOCHEM 2007, 809, 125-130. (12) Hofto, L. R.; Van Sickle, K.; Cafiero, M. Int. J. Quantum Chem. 2007, in press. (13) Boese, A. D.; Handy, N. C. J. Chem. Phys. 2001, 114, 54975503. (14) Slater, J. C. Quantum Theory of Molecules and Solids: Vol. 4, The Self-Consistent Field for Molecules and Solids; McGraw-Hill: New York, 1974. (15) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 12001211. (16) Zhao, Y.; Truhlar, D. G. J. Chem. Theory Comput. 2005, 1, 415432. (17) Zhao, Y.; Truhlar, D. G. J. Chem. Theory Comput. 2006, 2, 10091018. (18) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. (19) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553-566. (20) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650-654. (21) Clark, T.; Chandrasekhar, J.; Schleyer, P. V. R. J. Comput. Chem. 1983, 4, 294-301. (22) PQS, version 3.1; Parallel Quantum Solutions: Fayetteville, AR, 2004.