A QM-MD simulation approach to the analysis of FRET processes in (bio)molecular systems. A case study: complexes of E. coli purine nucleoside phosphorylase and its mutants with formycin A

Abstract

Predicting FRET pathways in proteins using computer simulation techniques is very important for reliable interpretation of experimental data. A novel and relatively simple methodology has been developed and applied to purine nucleoside phosphorylase (PNP) complexed with a fluorescent ligand - formycin A (FA). FRET occurs between an excited Tyr residue (D*) and FA (A). This study aims to interpret experimental data that, among others, suggests the absence of FRET for the PNPF159A mutant in complex with FA, based on novel theoretical methodology. MD simulations for the protein molecule containing D*, and complexed with A, are carried out. Interactions of D* with its molecular environment are accounted by including changes of the ESP charges in S1, compared to S0, and computed at the SCF-CI level. FRET probability W F depends on the inverse six-power of the D*-A distance, R da . The orientational factor 0 < k(2) < 4 between D* and A is computed and included in the analysis. Finally W F is time-averaged over the MD trajectories resulting in its mean value. The red-shift of the tyrosinate anion emission and thus lack of spectral overlap integral and thermal energy dissipation are the reasons for the FRET absence in the studied mutants at pH 7 and above. The presence of the tyrosinate anion results in a competitive energy dissipation channel and red-shifted emission, thus in consequence in the absence of FRET. These studies also indicate an important role of the phenyl ring of Phe159 for FRET in the wild-type PNP, which does not exist in the Ala159 mutant, and for the effective association of PNP with FA. In a more general context, our observations point out very interesting and biologically important properties of the tyrosine residue in its excited state, which may undergo spontaneous deprotonation in the biomolecular systems, resulting further in unexpected physical and/or biological phenomena. Until now, this observation has not been widely discussed in the literature.

Fig. 1

E. coli PNP (ribbon representation) complexed with FA (yellow). The closest to FA tyrosine residue (Tyr160) is in magenta. Other tyrosines: Tyr27, Tyr52, Tyr72, Tyr186, and Tyr173 (going counterclockwise, when starting from the upper, left corner) are in red

Fig. 2

UV absorption (solid) and fluorescence emission (dash-dot) spectra of E. coli PNP mutatnts PNPF159Y, PNPF159A, and FA (magenta) in 50mM Hepes, pH 7: a PNPF159Y (blue), and neutral form of FA (magenta); b PNPF159A (blue), and neutral form of FA (magenta). Maximum intesities of the absorption and emission spectra of PNP (λ_ex 270 nm) and FA (λ_ex 295 nm) are normalized to unity

Fig. 3

Absorption a and emission after excitation at 280 nm b normalized spectra of wild type PNP (red), PNPF159Y (black), and PNPF159A (blue)

Fig. 4

Fluorescence emission (λex 280 nm) spectra of PNPF159Y (a), PNPF159A (b), 4 μM (FA4), and 12 μM (FA12) FA in 50mM Hepes buffer (pH 7), and fluorescence-emission difference (λex 280 nm) spectra after subtraction of FA at both concentrations (Dif Mix4-FA, Dif Mix12-FA), and after substraction of PNPF159Y (a), or PNPF159A b at both concentrations of FA (Dif Mix4-PNP, Dif Mix12-PNP)

Fig. 5

Fluorescence excitation (λem 340 nm) spectra of PNPF159Y (a), PNPF159A (b), 4 μM (FA4), and 12 μM (FA12) FA in 50mM Hepes buffer (pH 7), and fluorescence-excitation difference (λem 340 nm) spectra after subtraction of FA at both concentrations (Dif Mix4-FA, Dif Mix12-FA), and after substraction of PNPF159Y (a), or PNPF159A b at both concentrations of FA (Dif Mix4-PNP, Dif Mix12-PNP)

Fig. 6

Fluorescence-emission difference (λ_ex 280 nm) spectra after subtraction of different FA concentrations and PNPF159Y a or PNPF159A b in 50mM Hepes pH 7

Fig. 7

Fluorescence-excitation difference (λ_em 340 nm) spectra after subtraction of different FA concentrations and PNPF159Y a or PNPF159A b in 50mM Hepes pH 7.

Fig. 8

Fluorescence emission difference (λ_ex 280 nm) spectra of PNPF159Y a and PNPF159A b in 50mM acetate buffer, pH 5 after subtraction of the fluorescence emission spectra of FA in different concentrations and PNPF159Y a or PNPF159A b fluorescence emission spectra

Fig. 9

Structures, charges, and electric transition moments of tyrosine, its deprotonated form and formycin A. CHARMM charges in the ground state are in bold. Changes of the ESP charges after excitation from S_o to S₁ are on the right site of the CHARMM charges. Arrows denote electric transition moments between S_o and S₁, computed using the SCF-CI approximation with TURBOMOLE

Fig. 10

E. coli PNP wild-type. Phe 159 stabilizes position of FA in the binding site. For such configurations FRET is very effective

Fig. 11

The F159A mutant of E. coli PNP. The lack of the phenyl ring in the 159 position results in a weaker binding and higher mobility of the ligand. FA from time to time partially “flows-out” of the binding site which results in decreasing of FRET. Above are two characteristic configurations (left — optimal for FRET, right — not optimal one)

Fig. 12

Wild-type PNP from E. coli in complex with FA. Characteristic are one of the most effective in FRET configurations (left) and an inhibiting one (right)

Fig. 13

Representative snapshots from the MD simulations for the PNPF159Y mutant with Tyr160^(-)* and in complex with FA. Characteristics are one of the most effective in FRET configurations (left) and an inhibiting one (right)

Fig. 14

Fluorescence-emission difference spectra of 3.9 μM PNPF159Y mutant with 12 μM FA, relative to the arithmetic sum of the two components: with a decreasing pH values from pH 7 (Hepes buffer) to pH 6 and pH 5 (acetate buffer), and b decreasing concentrations of acetate buffer (pH 5)