Protein Phosphorescence Quenching: Distinction between Quencher

Jul 2, 2010 - In theory, the external quenching rate can be estimated from the ... from the Internal Diffusion of Solutes of Increasing Molecular Size...
0 downloads 0 Views 635KB Size
Download PDF
J. Phys. Chem. B 2010, 114, 9691–9697

9691

Protein Phosphorescence Quenching: Distinction between Quencher Penetration and External Quenching Mechanisms Giovanni B. Strambini* and Margherita Gonnelli Consiglio Nazionale delle Ricerche, Istituto di Biofisica, 56124 Pisa, Italy ReceiVed: April 22, 2010; ReVised Manuscript ReceiVed: June 4, 2010

The accessibility of quenching solutes, Q, of various molecular sizes to buried Trp residues in proteins, as attested by dynamic quenching of their phosphorescence emission, is instrumental for probing structural fluctuations in these macromolecules. However, interpretation of quenching rates in terms of Q migration through the globular fold requires that alternative reaction pathways, such as long-range interactions with Q in the solvent, be ruled out. In theory, the external quenching rate can be estimated from the distance dependence of the through-space interaction by assuming compliance with the rapid diffusion limit regime. To validate the applicability of theoretical predictions to external quenching of protein phosphorescence, we compared the rate of quenching of the buried Trp residues of RNase T1 and parvalbumin by acrylamide and the bigger double-headed derivative bisacrylamide. The results showed that larger bisacrylamide is twice as efficient a quencher as acrylamide, implying that for these superficially buried residues the reaction is dominated by long-range interactions with acrylamide in the aqueous phase. To test the dependence of the quenching rate constant, kq, on solvent viscosity, quenching studies were extended to glycerol-water solutions ranging in bulk viscosity from 1 to 120 cP. Apart from an initial about 2-fold increase, kq was found to be independent of solvent viscosity, thus demonstrating that external quenching rigorously complies with the rapid diffusion limit regime. Experiments were extended to larger acrylamide derivatives to evaluate the impact of Q size on the external quenching rate. Introduction The dynamic features of protein structure are attracting growing attention from the progressive realization that for these macromolecules flexibility and function are intimately correlated. Of special interest are slow collective motions in the submicrosecond-second time domain as there is mounting evidence that structural fluctuations in this time scale are directly linked to enzymatic catalysis, signal transduction, and proteinprotein interactions.1-7 Notwithstanding the relevance to biological function, slow motions beyond the microsecond range are still poorly characterized, especially when they involve transitions to rarely populated states.1,8 Fluorescence and phosphorescence quenching of buried Trp residues by the diffusive penetration of small solutes through generally compact globular folds bear direct evidence on the conformational flexibility of these macromolecules. In principle, by varying the quencher (Q) size, it should be possible to characterize structural fluctuations over a wide range of amplitudes.9-13 Quenching studies measure the luminescence lifetime (τ) as a function of quencher concentration, [Q], and evaluate the bimolecular quenching rate constant, kq, from the gradient of the Stern-Volmer plot

1/τ ) 1/τ0 + kq[Q]

(1)

where τ0 is the unperturbed lifetime. Because of the long τ0 (milliseconds to seconds) of phosphorescence, the long-lived emission is suited for monitoring a wide range quenching rates, down to the second time domain. * Corresponding author: Tel +39 050 315 3046, Fax +39 050 315 2760, e-mail [email protected].

With chromophores free in solution and efficient quenching reactions, kq is close to the diffusion-limited rate constant kD ) 4πr0D (r0 is the sum of molecular radii and D is the sum of the diffusion coefficients). With Trp residues inside proteins kq is invariably smaller than kD and, depending on the size of Q and on the surface accessibility of the chromophore, may fall by several orders of magnitude.14 Naturally, for kq to reflect the slow migration of Q through the protein matrix, alternative quenching mechanisms must be ruled out. Particularly in the case of highly hindered diffusion through compact regions of the macromolecule, quenching by long-range through-space interactions with Q in the solvent may become a competitive pathway. In fact, if it is generally agreed that small diatomic and triatomic molecules (e.g., O2, NO, CS2) penetrate readily the protein matrix, for quenchers the size of acrylamide, or larger, interpretation of small kqs in terms of the frequency and amplitude of structural fluctuations may not always be warranted. It has also been suggested that acrylamide cannot penetrate the protein fold even in the long millisecond-to-second time scale and that the quenching reaction involves either longrange interactions with acrylamide in the solvent or partial unfolding transitions bringing the buried chromophore in proximity of the aqueous phase.15,16 Thus, for any Q-protein pair the possibility to estimate the magnitude of the external quenching rate is crucial for distinguishing between penetration and external quenching mechanisms and carry out protein permeability studies with quenchers of various sizes. External quenching consists of free diffusion of Q to the protein surface followed by a quenching interaction with the buried chromophore. In some cases it may also involve weak binding of Q to a protein site within interaction distance of the chromophore. The latter possibility may be distinguished from

10.1021/jp103615y  2010 American Chemical Society Published on Web 07/02/2010

9692

J. Phys. Chem. B, Vol. 114, No. 29, 2010

dynamic processes as it gives rise to a downward curving, saturating, Stern-Volmer plot. Potentially, dynamic quenching can be estimated employing a relationship derived from knowledge of the distance dependence of the quenching reaction, k(r), and the minimum separation of Trp from the aqueous interface, assuming that the process is under reaction control17 (i.e., that the reaction occurs in the rapid diffusion limit, RDL, regime). Although the latter condition is believed to apply to inefficient quenching of protein phosphorescence,15,18 rigorous experimental demonstration is still wanting. In this report we first establish that acrylamide quenching of the proteins RNase T1 and Ca-parvalbumin, whose Trp residue is buried 2-2.5 Å below the molecular surface, is fully dominated by through-space interactions with acrylamide in the solvent. Then, we show that kq is insensitive to a more than 100-fold increase in solvent viscosity, thus providing a rigorous demonstration that the reaction complies with the RDL regime. Estimation of the through-space rate constant, validated by the good agreement between theoretical and experimental kq, is then extended to larger acrylamide derivatives to evaluate the impact of molecular size on the external quenching rate. Experimental Methods All chemicals were of the highest purity grade available from commercial sources and unless specified to the contrary were used without further purification. N-Acethyltryptophanamide (NATA) was from Sigma-Aldrich (Italy) and prior to use was recrystallized twice from ethanol/water. Acrylamide (99.9% electrophoretic purity) was from Bio-Rad Laboratories (Richmond, CA). Ultrapure, MB grade N,N′-methylenebisacrylamide (BA) was from USB Corp. (Cleveland, HO). N-(Hydroxymethyl)acrylamide (HA) (>98% HPLC), N-[tris(hydroxymethyl)methyl]acrylamide (TA) (93%), and 2-acrylamido-2-methyl-1propanesulfonic acid sodium salt (SA) were from Sigma-Aldrich (Italy). Tris(hydroxymethyl)aminomethane (Tris), spectral grade glycerol, and NaCl Suprapur were from Merck (Darmstadt, Germany). Water was purified by reverse osmosis (Milli-RX 20; Millipore, Billerica, MA) and subsequently passed through a Milli-Q Plus system (Millipore Corp., Bedford, MA). The protein parvalbumin from whiting fish was purchased from Calbiochem Corp. (San Diego, CA), and Ribonuclease T1 (RT1) was a kind gift from Dr. Nick Pace, Texas A&M University. RNaseT1 was dialyzed into 100 mM Na-acetate buffer, 1 mM EDTA, and pH 5.5, and parvalbumin into Tris 50 mM, 10 mM CaCl2, and pH 8. Quenchers stocks were prepared daily by dissolving commercial supplies in the experimental buffer. For phosphorescence measurements in fluid aqueous solutions, NATA and protein samples were placed in appositely constructed, 5 × 5 mm2, quartz cuvettes with a leak proof capping designed to allow thorough removal of O2, by the alternating application of moderate vacuum and inlet of ultrapure N2.19 For viscous glycerol containing samples O2 removal required prolonged equilibration times, up to 3 h. In phosphorescence measurements, the final protein concentration ranged between 2 and 5 µM. The viscosity of glycerol solutions was determined by an Ostwald viscosimeter, placed in a bath at the required temperature, calibrated with glycerol-water (50/50, w/w) solution of known viscosity. Phosphorescence Measurements. Phosphorescence spectra and decays were measured with pulsed excitation on a homemade apparatus.19 The exciting light (λex ) 292 nm) was provided by a frequency-doubled Nd:YAG-pumped dye laser (Quanta Systems, Milano, Italy) with pulse duration of 5 ns

Strambini and Gonnelli and a typical energy per pulse of 0.5-1 mJ. For spectral measurements the emission, collected at 90° from the excitation, was dispersed by a 0.3 m triplet grating imaging spectrograph (SpectraPro-2300i, Acton Research Corp., Acton, MA) set to a band-pass of 2.0 nm. The emission was monitored by a backilluminated 1340 × 400 pixels CCD camera (Princeton Instruments Spec-10:400B (XTE), Roper Scientific Inc., Trenton, NJ) cooled to -60 °C. Excitation and fluorescence light were eliminated from the detection system by a mechanical chopper. For decay measurements the phosphorescence intensity was collected at 90° from vertical excitation through a filter combination with a transmission window of 405-445 nm (WG405, Lot-Oriel, Milano Italy, plus interference filter DTBlau, Balzer, Milano, Italy) and monitored by a photomultiplier (EMI 9235QA, Middlesex, UK). The latter was protected against fatigue from the strong excitation/fluorescence pulse by either a gating circuit or a mechanical chopper synchronized to the laser trigger, which closed the emission slit during the excitation pulse. The time resolution of the latter option depends on the chopper speed and for the experiments reported here was maintained constant to 30 µs. The photocurrent was amplified by a current-to-voltage converter (SR570, Stanford Research Systems, Stanford, CA) and digitized by a 16-bit high-speed (1.25 MHz) multifunction data acquisition board (NI 6250 PCI, National Instrument Italy, Milano, Italy) supported by LABVIEW software capable of averaging multiple sweeps. Typically, less than 100 sweeps were sufficient for a good signalto-noise ratio even for the shortest decays. Prompt Trp fluorescence from the same sample was simultaneously collected through a 310-375 band-pass filter combination (WG305 nm plus Schott UG11) and detected by a UV-enhanced photodiode (OSD100-7, Centronics, Newbury Park, CA). An analogue circuit was used to integrate the photocurrent, and its output was digitized and averaged by a multifunctional board (NI 6250 PCI, National Instrument Italy, Milano, Italy) utilizing LABVIEW software. The prompt fluorescence intensity was used to account for possible variations in the laser output between measurements as well as to obtain fluorescencenormalized phosphorescence intensities. A decrease in the latter would indicate the extent of any rapid quenching of phosphorescence that might occur during the dead time of these measurements, owing, for example, to protein bound quenchers. All phosphorescence decays were analyzed in terms of discrete exponential components by a nonlinear least-squares fitting algorithm (DAS6, fluorescence decay analysis software, Horiba Jobin Yvon, Milano, Italy). Quenching Experiments. Quenching experiments were carried out by measuring the phosphorescence lifetime, τ, of NATA and proteins samples as a function of the quencher concentration [Q] in solutions.14 The bimolecular quenching rate constant, kq, was derived from the gradient of the Stern-Volmer plot (eq 1). For each quencher concentration at least three independent samples were analyzed, and the reported results were obtained from the mean lifetime value. The reproducibility of phosphorescence lifetime measurements was generally better that 5%. Quenching Models. Quenching of the luminescence of Trp residues in proteins by small solutes (Q) in solution has been analyzed in terms of mechanisms that involve either Q diffusion to the protein surface followed by long-range interactions with the buried chromophore (external quenching) or diffusive penetration of Q through the protein matrix to its proximity. Pertinent analytical expressions for the quenching rate, obtained by solving Fick’s diffusion equation applying radiation boundary conditions, and their application range in the realm of protein

Protein Phosphorescence Quenching

J. Phys. Chem. B, Vol. 114, No. 29, 2010 9693

phosphorescence have been discussed in detail before.20 Briefly, by neglecting a transient time-dependent term arising from Q molecules already in proximity of the chromophore at the instant of excitation, the steady-state bimolecular quenching rate constant, kq, is given by

kq ) kDkI/(kD + kI)

(2)

where kD ) 4πaD is the Smoluchowski diffusion controlled rate constant, kI ) 4πa2B is the diffusion-independent maximum rate constant, and D is the sum of the separate diffusion coefficients of Q and protein in solution. B is an interaction strength parameter related to the distance dependence of the quenching interaction, k(r), and a is the distance (center-tocenter) of closest approach between Q and chromophore. Depending on which of the two rate constants is smaller, the process goes from the diffusion-controlled regime (kD , kI, kq ∝ D) to the reaction-controlled regime (kD . kI, kq ) kI), where kq is independent of diffusion. Assuming an essentially flat protein surface and an exponential interaction, k(r) ) k0 exp[-(r - r0)/re], it was shown that the bimolecular rate constant for external quenching of protein phosphorescence by Q relegated to the aqueous phase decreases with the thickness of the protein spacer (rp) separating the chromophore from the aqueous interface (rp + r0 ) a is the shortest center-to-center distance between the internal chromophore and Q at the aqueous interface) according to17 kq(rp) ≈ kI(rp) ) 2πN10-3[(rp + r0)re2 + 2re3]k0 exp(-rp/re) M-1 s-1

(3)

where N is Avogadro’s number and re is the attenuation length (units of re and rp in cm). Special cases of external quenching may take the form of binding of Q to the protein surface (with a dissociation constant Kd) or involve slow rate-limiting conformational transitions (e.g., partial unfolding with a rate k0) bringing the buried chromophore to the aqueous interface. In these cases kq will depend on [Q] and is given by12

kq ) k(r')/([Q] + Kd) kq ) k0 /([Q] + σ)

for binding

for protein gating

(4) (5)

where k(r′) is the quenching rate of Q bound at a distance r′ from the chromophore and σ is the [Q] corresponding to midpoint saturation of the gate. The latter contributions can be discriminated from kq(rp) in that in both cases kq depends on [Q], and consequently the process will give rise to nonlinear downward curving Stern-Volmer plots. Alternatively, for diffusive penetration of Q across a protein shell of thickness T, characterized by a diffusion constant Dp , D, the process is bound to be under diffusion control and in such cases the rate constant assumes the form of a corrected Smoluchowski equation20,21

kq ) 4πaDp(1 + a/T)

(6)

where now a is the radius of the impermeable protein core surrounding the chromophore.

Results and Discussion Quenching of NATA Phosphorescence by Acrylamide and Bisacrylamide (BA). The effectiveness of acrylamide and of its larger derivative N,N′-methylenebisacrylamide (BA) as quenchers of indole phosphorescence in aqueous solutions (10 mM Tris buffer, pH 7), at 25 °C, was investigated with the Trp derivative NATA. The addition of these quenchers to a solution of NATA (5 × 10-5 M) shortened the phosphorescence lifetime, maintaining the decay exponential throughout. The decrease of τ as a function of [Q] gave in each case a linear Stern-Volmer plot (Figure 1a) whose gradient yielded a bimolecular quenching rate constant of (1.9 ( 0.1) × 109 M-1 s-1 for acrylamide and (3.9 ( 0.1) × 109 M-1 s-1 for BA, respectively. Both values are smaller than kD, (5-7) × 109 M-1 s-1,22implying that the triplet state quenching efficiency per collisional encounter is smaller than 1. As the corresponding fluorescence quenching rate constant, Fkq, is practically identical to the diffusion rate constant kD,19,20 the quenching efficiency per collisional encounter is given by the ratio Pkq/Fkq. Fkq ) (6-7) × 109 M-1 s-1 for acrylamide22 and 5.5 × 109 M-1 s-1 for BA.23 Hence, the decreased efficiency for the excited triplet-state derived from the ratio Pkq/Fkq is ≈1/3 for acrylamide and ≈2/3 for BA. In general, the ratio represents the spin statistical factor governing the transition from encounter complex to products.24 Because in the case of double-headed BA the spin factor is expected to double, one would anticipate Pkq(BA) to be twice as large as P kq(acryl), as is observed experimentally. Quenching of Trp Phosphorescence in RNase T1 and in Parvalbumin. RNase T1 and whiting fish parvalbumin are single Trp proteins in which the aromatic chromophore is fully shielded from the solvent, the indole ring being separated from the aqueous interface by a 2-2.5 Å thick protein spacer (rp).25,26 Acrylamide and BA quenching of these proteins led to a progressive shortening of the phosphorescence lifetime of otherwise monoexponential decays. The results obtained in buffer, at 25 °C, are displayed in the Stern-Volmer plots of Figure 1b,c. Throughout, up to the limit of measurable quenching rates, (1/τ - 1/τ0) ≈ 104 s-1, the plots were found to be linear on [Q] (no saturation effects were observed), which is consistent with a dynamic quenching process. The magnitude of kq obtained from the gradient of these plots (Table 1) emphasizes that, with respect to NATA, the quenching rate of these proteins decreases by 4-5 orders of magnitude with both acrylamide and BA, suggesting that access of these solutes to the sites of W59 in RNase T1 and of W102 in parvalbumin is strongly inhibited. However, despite that the size of BA is more than twice that of acrylamide (Mw ) 154 cf. 71), we find for both proteins that kq(BA) ≈ 2kq(acryl). As the rate of penetration of the globular fold depends sharply on Q size, the only prospect for larger, double-headed, BA to be twice as effective as acrylamide is that the reaction be dominated by through-space interactions with Q in the solvent. This result establishes that in these proteins quenching by both acrylamide and BA is predominantly external and that diffusive migration pathways make an insignificant contribution to the rate. To get a rough estimate of the activation barrier to acrylamide and BA quenching of RNase T1, measurements were extended to supercooled solutions at -5 °C. As shown in Table 1, the value of ∆H((kq) derived from a two-point Arrhenius plot is about 6 kcal/mol, very similar between the two quenchers. For a through-space quenching reaction, ∆H((kq) represents the activation barrier of the quenching interaction (k(r)). The small magnitude of ∆H((kq) is consistent with an efficient electron exchange/transfer process27 and contrasts with the 11-25 kcal/

9694

J. Phys. Chem. B, Vol. 114, No. 29, 2010

Strambini and Gonnelli TABLE 1: Acrylamide and Acrylamide Derivatives Bimolecular Quenching Rate Constants, kq, for Quenching of NATA and Protein Phosphorescence in Aqueous Solutions sample NATA RNase T1

Parvalbumin

RNase T1

a

Figure 1. Lifetime Stern-Volmer plots relative to phosphorescence quenching by acrylamide (b) and BA (9) at 25 °C. (a) NATA in 10 mM Tris, pH 7; (b) Ca-parvalbumin in 50 mM Tris, 10 mM CaCl2, pH 8; (c) RNaseT1 in 100 mM Na-acetate buffer, 1 mM EDTA, pH 5.5. λex ) 292 nm.

mol28 observed for acrylamide quenching of proteins where the reaction involves diffusive penetration of the protein matrix.14 A reviewer correctly pointed out that partial unfolding transitions bringing internal chromophores to the aqueous interface would also lead to quenching reactions with Q in the aqueous phase, with little discrimination on Q size and charge. These major structural rearrangements are ruled out by multiple evidence. For stable compact protein folds, such as those of RNase T1 and Ca-parvalbumin, unfolding like transitions,

quencher acrylamide BA acrylamide BA HA TA SA acrylamide BA acrylamide BA HA TA SA

τ0 (ms) T ) 25 °C 2.5 24.5

3.5 T ) -5 °C 53.5

kq (M-1 s-1)a 1.9 × 109 3.9 × 109 5.7 × 104 1.3 × 105 5.3 × 104 3.3 × 104 2.1 × 104 3.0 × 104 6.7 × 104

∆H((kq) (kcal/mol)

5.8 6.1 6.3 6.7 7.0

1.9 × 104 4.1 × 104 1.6 × 104 9.2 × 103 5.5 × 103

The reproducibility of kq is generally better than 7%.

breaking down even part of the β-sheet structure enveloping the Trp residue would be expected to be rather slow, be characterized by relatively large activation barriers and, by representing the rate-limiting step of the process, lead to nonlinear Stern-Volmer plots (see eq 5). The quenching rate would reach a plateau, kq[Q] e k0 (k0, the unfolding rate), similar for various Qs but probably quite different among protein folds. Experiment shows that the Stern-Volmer plots, for both RNase T1 and Ca-parvalbumin, are linear up to quenching rates of 104 s-1, which implies that the unfolding transition rate would have to be unrealistic large, k0 > 105 s-1, for saturation effects not to be manifest in the plot. The similarity of kq between the two proteins would imply that k0/σ is similar (kq ) k0/σ in the quasilinear, low [Q], range), which would be a coincidence. Finally, an unfolding mechanism is not supported by the small activation barrier found for kq and is definitely ruled out by the invariance of kq over a large increase in solvent viscosity, as structural fluctuations in proteins are notoriously coupled to solvent viscosity (see viscosity effects below). By contrast, long-range interactions from Q in the solvent are independent of the structural dynamics of protein systems and, in the case of RNase T1 and Ca-parvalbumin, accurately account for the experimental magnitude of kq, for the linearity of Stern-Volmer plots, for the low activation barrier of kq, and for its total independence on solvent viscosity. Effect of Solvent Viscosity on Acrylamide Quenching of RNase T1. Under conditions of inefficient quenching, when kq , kD, eq 2 predicts kq ) kI, namely, that the reaction is under reaction control where the rate is independent of the diffusion coefficient and therefore of solvent viscosity. To verify the above hypothesis, acrylamide quenching of RNase T1 was carried at increasing solvent viscosity, employing glycerol/buffer mixtures. Lifetime Stern-Volmer plots were obtained for solutions varying from 10 to 70% (weight fraction) in glycerol content, at constant ionic strength and pH, at 25 and -5 °C. Throughout, the phosphorescence decay remained exponential, and 1/τ was linear with the acrylamide concentration. Because glycerol impurities made a significant contribution to the phosphorescence emission during the first 200-300 µs of the decay, the dead time of lifetime measurements increased relative to buffer, thus lowering the detection limit of quenching rates to ≈3 × 103 s-1. The dependence of kq on solvent viscosity, η, is

Protein Phosphorescence Quenching

Figure 2. Effect of solvent viscosity on the bimolecular quenching rate constant, kq, for acrylamide (solid symbols) and BA (empty symbols) quenching of RNase T1 phosphorescence in glycerol/buffer solutions at 25 °C (2, ∆) and at -5 °C (9, 0).

displayed in Figure 2. The results clearly indicate that the effect of glycerol on kq is modest at both temperatures and show that, instead of a potential slowing down of the reaction in viscous solutions, there is a roughly 2-fold enhancement of the rate constant. Apart from the initial rise in kq, in the 1-20 cP range, the data proved the invariance of the bimolecular quenching rate constant over 100-fold increase in solvent viscosity. This is a clear confirm that inefficient external quenching of protein phosphorescence complies with RDL regime and thus validates the use of eq 3 to estimate the contribution from long-range interactions with any Q in the solvent. The above conclusion is quite general to every protein with buried Trp residues because, compared with the very superficial Trp residues of RNase T1 and parvalbumin, external quenching is bound to be even less efficient when the thickness of the protein spacer increases. Conversely, a significant dependence of kq on solvent viscosity excludes an external quenching mechanism and, consequently, is more indicative of internal Q migration during structural fluctuations in the macromolecule. In fact, the roughly 7-fold (at 25 °C) and 45-fold (at -5 °C) reduction in kq, reported for acrylamide quenching of buried W314 of liver alcohol dehydrogenase in 68% glycerol solution,12 is consistent with the slowing down of quencher penetration by structural fluctuations becoming more sluggish in viscous media. Having established that external quenching in ordinary aqueous solutions complies with the rapid diffusion limit regime, the corresponding rate constant can now be calculated from the distance dependence of the quenching interaction, k(r), and the minimum thickness of the protein spacer, rp, separating the chromophore from the aqueous interface, by applying eq 3. For acrylamide, k(r) derived with model indole compounds in glasses and in liquid solutions is k(r) ) 3.5 × 1010 exp[-(r r0)/0.029] s-1, r in nm14 (relative to the initial report the preexponential term, k0, has been corrected for a error in its evaluation). The parameters k0 ) 3.5 × 1010 s-1 and re ) 0.029 nm predict a quenching rate constant of 8.3 × 104 M-1 s-1 for rp ) 0.2 nm, decreasing to 1.6 × 104 M-1 s-1 for rp ) 0.25 nm, a range in excellent agreement with the experimental rate of 5.7 × 104 M-1 s-1 for RNase T1 and 3.0 × 104 M-1 s-1 for parvalbumin. With BA, because of the 2-fold increase in the spin factor eq 3 predicts that the corresponding rate constant be twice as large as that of acrylamide, again in agreement with experiment. The about 2.3-fold increase of kq in glycerol solutions, observed at both 25 and -5 °C (data obtained at 10 °C also

J. Phys. Chem. B, Vol. 114, No. 29, 2010 9695 exhibit the same pattern), was unexpected and called for further enquiry on its possible origin. Interference from quenching impurities, increasing the apparent quenching rate in glycerol, is ruled out as τ0 was actually somewhat larger in viscous solutions. Other possibilities are (1) a slight reduction of the distance of closest approach (rp) between acrylamide in the solvent and W59, from glycerol-induced alterations of the hydration layer or of the local protein conformation, and (2) an increase in quenching efficiency associated with partial relaxation of the spin factor in viscous solutions. When intersystem crossing between spin states of different multiplicity of the encounter complex is fast relative to back-diffusion into separate reactants, the spin restriction governing the transition to products is predicted to relax24 and the spin factor increases above the value of 1/3, entailed by the ratio Pkq/Fkq ≈ 1/3 of NATA. To distinguish between the two alternatives, the quenching profiles in glycerol solutions were repeated with BA in place of acrylamide. The rationale is that while any alteration of rp will have the same consequence on kq(acryl) and kq(BA), spin relaxation is expected to have a much smaller impact on doubleheaded BA whose spin factor being closer to 1 (≈2/3) cannot increase much further. The results, displayed in Figure 2, demonstrate that in the case of BA there is no significant initial rise of the quenching rate and that kq(BA) is, within experimental error, practically constant up to the highest glycerol content, at both 25 and -5 °C. This finding clearly excludes a variations of rp as the cause for the initial jump of kq(acryl) in glycerol solutions. It is worth noting that, after the initial rise, kq(acryl) becomes very similar to kq(BA). Both the observations that in viscous solutions acrylamide and BA attain equal external quenching efficiency and that the initial rise of kq(acryl) occurs in the same viscosity range at 25 and -5 °C, but at different solvent composition, lend support to the hypothesis that the rate enhancement is linked to spin relaxation. While we are not aware of reports documenting spin relaxation in triplet state quenching reactions, we point out that the phenomenon should also apply to slow Q migration through the protein matrix and that, consequently, the increase in quenching efficiency should be taken into account when estimating the internal diffusion coefficient, Dp, from eq 6. External Quenching by Large Acrylamide Derivatives. One way to probe the frequency-amplitude spectrum of structural fluctuations in proteins is to determine the permeability of quenching molecules of various sizes, preferably carrying the same quenching moiety. With acrylamide several derivatives are commercially available where the quenching moiety is linked to neutral or charged groups of variable size (Figure 3). As diffusive penetration of the protein matrix is expected to drop sharply with increasing Q size competition from external quenching is expected to become more important the larger is Q. Having established that acrylamide quenching of RNase T1 is external and that eq 3 provides a valid estimate of its rate constant, we employ this protein and eq 3 to determine the impact of Q size on the external quenching rate of larger acrylamide derivatives, namely, N-(hydroxymethyl)acrylamide (HA, Mw ) 101), N-[tris(hydroxymethyl)methyl]acrylamide (TA, Mw ) 175), and 2-acrylamido-2-methyl-1-propanesulfonic acid sodium salt (SA, Mw ) 229). With larger quenchers kq is expected to decrease as the minimum separation between the acrylamide moiety of Q and the internal indole ring will increase with the radius of Q. In addition, owing to the roughness of protein surfaces, larger solutes may also approach less closely to the aqueous interface. Also, the quencher orientation in proximity of the protein surface may not be random, and with

9696

J. Phys. Chem. B, Vol. 114, No. 29, 2010

Figure 3. Space filling models and chemical formula of acrylamide and acrylamide derivatives employed in this study.

Strambini and Gonnelli the quenching interaction. Although the net effect of increasing the bulkiness of Q is to add to the minimum separation between the acrylamide moiety in the solvent and the internal chromophore (“rp” in the exponent of eq 3 becoming rp + ∆r, where ∆r ) rQ(effective) - r0(acrylamide)) the actual increment in ∆r, as pointed out above, will depend on various factors and is largely unpredictable. To stress that the impact of Q size on external quenching needs to be determined experimentally, we show that the simplest calculation of ∆r based on the increment of Q size, assuming spherical shape, is inadequate to estimate of kq. The increment of ∆r [rQ ) (3Mw/4πdN)1/3, N the Avogadro’s number, and d the density ≈1.2] is 0.36, 1.0, and 1.37 Å for HA, TA, and SA, respectively, and the corresponding value of kq is shown in Figure 4. The comparison between predicted and experimental rate constant indicates that the discrepancy becomes progressively larger with Q size, suggesting that quenching efficiency is maintained higher than predicted, presumably by factors partly compensating for the increase in Q radius. On the basis of eq 3, the decrease in experimental rate with Q size corresponds to an effective increase of ∆r of ≈0.0, 0.15, and 0.29 Å for HA, TA, and SA, respectively. Although these increments may not rigorously apply to other proteins, they represent best parameters for a closer prediction of external quenching rates in novel proteins. The activation barrier for these quenchers, derived from the two-point Arrhenius plot (Table 1), increases slightly with Q size but remains low, within 6-7 kcal/mol, for all acrylamide derivatives, which is characteristic of the quenching moiety. Conclusions

Figure 4. RNase T1 phosphorescence quenching rate constant (b) of acrylamide derivatives as a function of the molecular weight. kq(BA) has been divided by 2. Also included is the theoretical rate constant (9) for HA, TA, and SA calculated by eq 3, assuming an increase in rp equal to an increment in molecular radius of 0.36, 1.0, and 1.37 Å for HA, TA, and SA, respectively. All values refer to 25 °C.

charged derivatives, the local protein electric field may affect the distance distribution of Q. Quenching of RNase T1 phosphorescence by HA, TA, and SA was carried out in acetate buffer, pH 5.5, at 25 and -5 °C, across the 0-100 mM concentration range. To attenuate electrostatic interactions, with negatively charged SA the ionic strength was maintained large at 0.3 M (0.2 M KCl plus 0.1 M acetate). With TA, whose stock supply contains 7% KCl, the ionic strength was kept constant at 0.12 M by adjusting the buffer concentration. All through, the addition of Q led to a progressive shortening of τ that resulted in linear Stern-Volmer plots, as expected for dynamic quenching processes. The magnitude of kq derived from the slope is reported in Table 1, and the value obtained at 25 °C is displayed in Figure 4 as a function of the molecular weight. In comparison with acrylamide, kq decreased 1.7-fold with TA and 2.7-fold with SA, whereas with HA the difference is of the order of the experimental error. The above Qs are formed by the acrylamide moiety plus a bulking group, which is presumed to be inert with respect to

This study demonstrated that acrylamide quenching of superficially buried Trp residues of RNase T1 and parvalbumin is due to long-range interactions with the quencher in the solvent. Its rate is insensitive to an increase, up to 100-fold, in solvent viscosity, implying that external quenching of buried Trp residues in proteins fully complies with the rapid diffusion limit regime. The good agreement between experimental and theoretical rates indicates that eq 3 can be confidently applied to predict the contribution of external quenching in proteins in general, thus providing a tool for distinguishing quenching pathways that do not bear relevance to protein dynamics. References and Notes (1) Persson, E.; Halle, B. J. Am. Chem. Soc. 2008, 130, 1774. (2) Eppler, R. K.; Hudson, E. P.; Chase, S. D.; Dordick, J. S.; Reimer, J. A.; Clark, D. S. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 15672. (3) Henzler-Wildman, K.; Kern, D. Nature 2007, 450, 964. (4) Hammes-Schiffer, S.; Benkovic, S. J. Annu. ReV. Biochem. 2006, 75, 519. (5) Kale, S.; Ulas, G.; Song, J.; Brudvig, G. W.; Furey, W.; Jordan, F. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 1158. (6) Marlow, M. S.; Dogan, J.; Frederick, K. K.; Valentine, K. G.; Wand, A. J. Nat. Chem. Biol. , 6, 352. (7) Wand, A. J. Nat. Struct. Biol. 2001, 8, 926. (8) Bouvignies, G.; Bernado, P.; Meier, S.; Cho, K.; Grzesiek, S.; Bruschweiler, R.; Blackledge, M. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 13885. (9) Lakowicz, J. R.; Weber, G. Biochemistry 1973, 12, 4171. (10) Saviotti, M. L.; Galley, W. C. Proc. Natl. Acad. Sci. U.S.A. 1974, 71, 4154. (11) Strambini, G. B. Biophys. J. 1987, 52, 23. (12) Strambini, G. B.; Gonnelli, M. Biochemistry 2009, 48, 7482. (13) Vanderkooi, J. M. Topics in Fluorescence Spectroscopy. In Biochemical Applications; Lakowicz, J. R., Ed.; Plenum: New York, 1991; Vol. 3, p 113. (14) Cioni, P.; Strambini, G. B. J. Am. Chem. Soc. 1998, 120, 11749. (15) Calhoun, D. B.; Englander, S. W.; Wright, W. W.; Vanderkooi, J. M. Biochemistry 1988, 27, 8466. (16) Calhoun, D. B.; Vanderkooi, J. M.; Englander, S. W. Biochemistry 1983, 22, 1533.

Protein Phosphorescence Quenching (17) Vanderkooi, J. M.; Englander, S. W.; Papp, S.; Wright, W. W.; Owen, C. S. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 5099. (18) Calhoun, D. B.; Vanderkooi, J. M.; Woodrow, G. V., III; Englander, S. W. Biochemistry 1983, 22, 1526. (19) Strambini, G. B.; Kerwin, B. A.; Mason, B. D.; Gonnelli, M. Photochem. Photobiol. 2004, 80, 462. (20) Owen, C. S.; Vanderkooi, J. M. Comments Mol. Cell. Biophys. 1991, 7, 235. (21) Wright, W. W.; Owen, C. S.; Vanderkooi, J. M. Biochemistry 1992, 31, 6538. (22) Eftink, M. R. Topics in Fluorescence Spectroscopy. In Principles; Lakowicz, J. R., Ed.; Plenum: New York, 1991; Vol. 2, p 53.

J. Phys. Chem. B, Vol. 114, No. 29, 2010 9697 (23) Strambini, G. B.; Gonnelli, M. J. Phys. Chem. B 2010, 114, 1089. (24) Saltiel, J.; Atwater, B. W. Spin-Statistical Factors in DiffusionControlled Reactions; Wiley & Sons: New York, 1988; Vol. 14. (25) Gohda, K.; Oka, K.; Tomita, K.; Hakoshima, T. J. Biol. Chem. 1994, 269, 17531. (26) Swain, A. L.; Kretsinger, R. H.; Amma, E. L. J. Biol. Chem. 1989, 264, 16620. (27) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 265. (28) Gonnelli, M.; Strambini, G. B. Biochim. Biophys. Acta 2009, 1794, 569.

JP103615Y