Mimic of photocycle by a protein folding reaction in photoactive yellow protein.

The blue light receptor photoactive yellow protein (PYP) displays rhodopsin-like photochemistry based on the trans to cis photoisomerization of its p-coumaric acid chromophore. Here, we report that protein refolding from the acid-denatured state of PYP mimics the last photocycle transition in PYP. This implies a direct link between transient protein unfolding and photosensory signal transduction. We utilize this link to study general issues in protein folding. Chromophore trans to cis photoisomerization in the acid-denatured state strongly decelerates refolding, and converts the pH dependence of the barrier for refolding from linear to nonlinear. We propose transition state movement to explain this phenomenon. The cis chromophore significantly stabilizes the acid-denatured state, but acidification of PYP results in the accumulation of the acid-denatured state containing a trans chromophore. This provides a clear example of kinetic control in a protein unfolding reaction. These results demonstrate the power of PYP as a light-triggered model system to study protein folding.

Signal transduction in biology starts with the activation of receptor proteins. In the case of photochemical signal transduction, signals are generated by photoreceptors containing lightabsorbing chromophores like retinal. Photoactive yellow protein (PYP) 1 is a water-soluble blue-light receptor in Ectothiorhodospira halophila and related eubacteria (1,2). E. halophila displays negative phototaxis toward blue light, and PYP is considered to be the photoreceptor for this response (3). The chromophore of PYP is p-coumaric acid (pCA) (4,5), commonly found in plants as a metabolite derived from tyrosine. Receptor stimulation by light triggers a photocycle in which the anionic pCA in the active site of PYP first is isomerized from trans to cis, and subsequently is protonated (6 -12). The absorbance spectrum of the pCA is shifted from max ϭ 446 nm for the initial pG trans state of PYP (see Table I for a summary of the PYP states studied here) to max ϭ 355 nm for the longest-lived photocycle intermediate pB (1,13). The pB state is considered to be the signal state for negative phototaxis because of its life time and its resemblance to the S 373 intermediate of the archaebacterial sensory rhodopsin I (1,13,14). A number of results indicate that a large structural change occurs during the formation of the pB state, resulting in the partial unfolding of PYP (12,(15)(16)(17)(18)(19)(20)(21).
To explore the relationship between protein folding in PYP and signaling state formation during the photocycle, we have studied the kinetics of protein folding in PYP. Recently, we demonstrated that refolding of PYP from the fully denatured state containing a cis chromophore involves the pB photocycle intermediate as an on-pathway folding intermediate (19). Thus, the folding pathway in PYP can be studied both by conventional stopped-flow techniques and by photoexcitation.
Here we extend this novel approach for studying protein folding to the transitions between native PYP and its acid-denatured state pB trans (also called pB dark ) in rapid mixing pH jump experiments.

EXPERIMENTAL PROCEDURES
Absorbance and Fluorescence Spectroscopy-PYP was overexpressed in Escherichia coli and purified as described previously (9). PYP was used at a concentration of 8 M for absorbance measurements and 2 M for fluorescence spectroscopy. Steady state equilibrium measurements of the PYP absorbance spectrum at different pH values and temperatures were performed using a Cary 300 UV-visible spectrophotometer (Varian) equipped with a Cary 1 ϫ 1 Peltier element. The kinetics of the last photocycle step were recorded at 446 and 340 nm at different pH values (in the same buffer as the stopped-flow experiments; see below) after 20 s illumination with broadband blue actinic light using the Cary 300 UV-visible spectrophotometer (Varian) or a Hewlett-Packard 8453 diode array spectrophotometer. Intrinsic aromatic and chromophore fluorescence from PYP was measured using a PerkinElmer Life Sciences LS50B spectrometer both in steady state experiments at pH 2.0 and in kinetic experiments on the photocycle at pH 3.9. Fluorescence excitation was at 280 nm for intrinsic aromatic fluorescence, 350 nm for chromophore fluorescence from the acid-denatured state, and 440 nm for chromophore fluorescence from the pG trans state. The fluorescence emission spectrum of the pB photocycle intermediate at pH 3.9 was reconstructed from each kinetic trace at different wavelengths by fitting the data as a monoexponential decay, and extrapolation to t ϭ 0, i.e. immediately after the actinic illumination.
Stopped-flow Absorbance Spectroscopy-Stopped-flow absorbance spectroscopy for pH jump experiments on PYP unfolding and refolding was performed using an SX-18MV stopped-flow spectrophotometer (Applied Photophysics) with a volumetric ratio of 1:1. For unfolding transitions to low pH, yielding pB trans , PYP in 10 mM potassium phosphate (pH 7.3) and 50 mM KCl was mixed rapidly with a buffer solution of 50 mM citrate, 50 mM KCl, and different concentrations of HCl or KOH. For refolding transitions, PYP was first denatured at pH 2.4 or 2.0 in 10 mM potassium phosphate, 25 mM citrate, and 10 or 20 mM HCl, and subsequently refolded by rapid mixing with solutions containing different concentrations of KOH, to yield final pH values from pH 3.2 to 7.6. To investigate the effect of the pCA isomerization state on the refolding process, acid-denatured PYP at pH 2.0 in 10 mM potassium phosphate, 25 mM citrate, and 20 mM HCl was exposed to UV-A light from a Cuda I-150 light source for 4 min to photoisomerize the pCA from trans to cis, yielding pB cis . This PYP solution was mixed with solutions containing different concentrations of KOH. Stopped-flow absorbance traces at 445 and 340 nm were recorded at least four times, averaged, and analyzed using the SX-18MV software. The refolding jump experiments from pB cis to a final pH value of 7.6 were also performed at different tem- peratures to allow the thermodynamic analysis of this folding transition.
Data Analysis-The equilibrium absorbance data on the acid denaturation of PYP were described as a two-state transition using Equation 1, where the equilibrium constant K D/N is defined by 10 n(pK Ϫ pH) ; Abs is the observed absorbance at 446 or 340 nm; and a n , b n , and a u , b u are the parameters for the sloping base lines to describe the pH dependence of native and acid-denatured PYP. pK and n are the pH at the transition midpoint and the number of protons taken up in the unfolding transition, respectively.
The rate constants for the unfolding and refolding reactions observed by stopped flow absorbance spectroscopy were analyzed as follows. First, the data were fit as a monoexponential process. The resulting rate constants (k obs ) in the protein folding reactions are the sum of the protein refolding rate (k f ) and unfolding rate (k u ) (k obs ϭ k u ϩ k f ), with the ratio of k f to k u determined by the equilibrium constant K D/N (K D/N ϭ k u /k f ). Combining these two equations, k u and k f can be obtained from K D/N and k obs . The resulting equations link the equilibrium and kinetic data, providing a quantitative test of the two-state nature of the transition.
The pH dependence of the kinetics of protein unfolding and refolding is caused by the uptake or release of protons upon reaching the transition state. In general, the pH dependence of the refolding kinetics can be described by the following equation, where a change in pH of ␦pH causes a change in the folding rate ␦logk f (22,23).
R is the gas constant, T is the absolute temperature, ⌬G ‡-D is the activation free energy for reaching the transition state from the denatured state, and ⌬Q ‡-D is the number of moles of protons involved upon reaching the transition state. Two different phenomenological descriptions of the pH dependence of ␦logk f /␦pH were used: (i) a linear dependence, in which the ⌬Q ‡-D does not depend on pH (Equation 5); and (ii) a phenomenological description, in which deviations from linearity are described by an additional quadratic term (Equation 6).
The temperature dependence of the rate constants k for folding and photocycle transitions was described using activation changes in enthalpy, entropy, and heat capacity ⌬H # , ⌬S # , and ⌬C p # at an arbitrarily chosen reference temperature of 298 K (Equation 7).
T is the temperature in degrees Kelvin, h is Plank's constant, and k B is Boltzmann's constant.
Equilibrium thermal denaturation curves for pB trans and pB cis at pH 2.0 were analyzed using Equation 8 as an equilibrium between the native and the unfolded state and two sloping base lines. The temperature dependence for the free energy for unfolding ⌬G u (T) was described as Equation 9, where T m is the temperature at the transition midpoint.

RESULTS AND DISCUSSION
Effect of the Chromophore Isomerization State on Refolding Kinetics-At low pH the initial pG trans state of PYP is converted to the blue-shifted, partially unfolded 2 pB trans state with an apparent pK a of 2.8 (24,25) (Fig. 1C). This acid denaturation process involves the protonation of the pCA chromophore, resulting in the blue-shift of the absorbance spectrum of PYP from 446 to 350 nm. We studied the kinetics of the transitions between pG trans and pB trans as determined in pH jump experiments using stopped-flow absorbance spectroscopy. Because of the large difference in absorbance maximum, these transitions can be sensitively monitored at both 350 and 445 nm. At 25°C the transitions between pG trans and pB trans were beyond the time resolution (2 ms) of the rapid mixing device (Fig. 1A) in the pH range 2.0 to 7.5. By performing the pH jump experiments at 5°C, the kinetics were slowed down sufficiently to allow the accurate determination of the rate constant for the transitions between pG trans and pB trans in a wide pH range (Fig. 1B, filled circles). At all pH values the signals could be described as a monoexponential transition. The logarithm of the refolding rate (k f ) and unfolding rate (k u ) show a linear pH dependence, with slopes of 0. 30   The isomerization state of the pCA chromophore is indicated as a superscript.

FIG. 1. Mimic of the PYP photocycle by a pH jump.
A, the refolding kinetics from the acid-denatured state of PYP at pH 2.0 to the native state at pH 4.7 monitored at 445 nm using stopped-flow absorbance spectroscopy at 25°C. Kinetic traces 1 and 2 are from acid-denatured PYP, which contains trans pCA (trace 1) and cis pCA (trace 2), respectively. Open circles depict the absorbance changes at 445 nm during the last photocycle transition at pH 4.7. The initial absorbance value at t ϭ 0 for traces 1 and 2 was 0, showing that most of the initial rapid phase occurred within the mixing dead time. B, pH dependence of refolding and unfolding rates, and the last photocycle step at 5°C and 25°C. Filled circles are rate constants for refolding and unfolding at 5°C of PYP containing trans pCA observed in pH jump experiments.
Open circles and open squares are rate constants for pH jump refolding of PYP containing cis pCA at 25 and 5°C, respectively. Filled diamonds and filled triangles are photocycle last step rate constants at 25 and 5°C at the indicated pH values, respectively. The arrow indicates the rate constant for the last photocycle step at pH 7.5 reported previously using laser spectroscopy (13). C, equilibrium pH titration of PYP monitored at 446 nm (filled squares) and 350 nm (open triangles). and ⌬Q ‡-N (see Equation 4) are independent of pH in the range 1.7 to 7.5. The rates of unfolding and refolding were found to be identical at pH 2.8, which is the midpoint transition in equilibrium pH titrations (Fig. 1C). Equilibrium titrations at 5°C were essentially identical to those performed at 25°C (25), with a pK a of 2.8 and an n value of 1.15 (see Equation 1). Thus, the sum of the slopes of the pH dependence observed for the unfolding and refolding transitions is equal to number of protons taken up during the transition as found in the equilibrium data. The monoexponential behavior and good correspondence between equilibrium and kinetic data (using Equations 2 and 3) indicate that the transition between pG trans to pB trans can be described as a two-state process.
To directly study the effect of the isomerization state of the pCA chromophore on the kinetics of refolding from the aciddenatured state to the native pG trans state, pB trans was illuminated with UV-A light at pH 2.0 to generate pB cis . The formation of pB cis could be monitored by a small but characteristic red-shift in the absorbance spectrum of the acid-denatured state (data not shown), as the absorbance maximum of pB trans is at 350 nm, and that of pB cis at 355 nm (25). Both species are fully stable at low pH. The formation of pB cis had a strong effect on the protein refolding kinetics; strongly biexponential behavior was observed. The very fast phase of ϳ50% could be attributed to refolding from pB trans . The remaining 50% of the signal displayed dramatically decelerated folding kinetics, caused by refolding from pB cis (Fig. 1A). Apparently, the UV-A illumination resulted in the formation of a stable mixture of equally populated pB trans and pB cis states. The pH dependence of the kinetics for refolding from pB cis to pG trans was determined, revealing that isomerization of pCA to its cis isomer decelerated the kinetics of refolding by 3-5 orders of magnitude (Fig.  1B, open circles at 25°C and open squares at 5°C). The deceleration of refolding is attributed to the energy barrier caused by the cis to trans isomerization of the pCA that needs to occur for refolding from pB cis . For free pCA the energy barrier for thermal isomerization has been estimated to be 125 kJ/mol (see Ref. 18). This experiment demonstrates how the energy surface for protein refolding can be experimentally modified by the introduction of a specific chemical process into the energy landscape for folding.
The pH dependence of the logarithm of refolding rates for pB cis significantly deviates from linear behavior both at 25 and 5°C (Fig. 1B). This reveals that, for pB cis , the value of ⌬Q ‡-D is pH-dependent, in contrast to the situation found for pB trans . The change in ⌬Q ‡-D for pB cis refolding as a function of pH was estimated by analysis of the data using a phenomenological quadratic equation (Equation 6). This analysis revealed that the number of protons ⌬Q ‡-D released by pB cis upon reaching the transition state decreases from ϳ1 to ϳ0 in the pH range from 3.2 to 7.5. The interpretation of these observations is discussed below.
The deceleration of PYP refolding by isomerization of the pCA to the cis conformation is analogous to the effect of isomerization of the imide bond of Pro residues (26). Because such Pro isomerization events occur with a significant energy barrier (85 kJ/mol), they can be rate-limiting for protein folding, as we found for pCA isomerization during PYP folding. However, the pCA isomerization effect exhibits two attractive features not found for Pro isomerization; (i) the extent of pCA isomerization in unfolded PYP is under direct experimental control (photoisomerization), and (ii) pCA isomerization is a central part of the functional cycle of PYP.
Mimic of the Last Photocycle Transition by pH Jump Refolding-We directly tested the relationship between the protein refolding reaction from pB cis to pG trans , as observed by stopped-flow absorbance spectroscopy, and the pB to pG trans photocycle transition, as observed after photoexcitation of PYP. To this end, the kinetics of the last PYP photocycle transition were measured by time-resolved absorbance spectroscopy as a function of pH. These measurements showed that the photocycle kinetics are exactly mimicked by those found for the protein refolding process at all pH values in the range 3.2-7.5 (Fig. 1B, closed diamonds at 25°C and closed triangles at 5°C). This observation indicates a high level of structural similarity between the equilibrium denatured state pB cis and the kinetic photocycle intermediate pB.
To further test the identity of the pB cis to pG trans protein refolding process and the pB to pG trans photocycle transition, the temperature dependence values of the kinetics of both transitions were compared. The temperature dependence of the kinetics observed in rapid mixing experiments was found to be essentially identical to that reported previously (15) for the pB to pG trans photocycle transition (Fig. 2). Both reveal a strongly curved temperature dependence that can be accurately described (Equation 7) by the occurrence of an activation change in heat capacity (Table II). These results demonstrate that the pB to pG trans photocycle transition is a protein refolding process from an acid-denatured state to the native state.
The strong deceleration of the folding kinetics by the cis chromophore described above has an important biological consequence. The pB state is considered to be the signal generating intermediate in the PYP photocycle. Only the pB state containing a cis chromophore has a physiologically relevant lifetime of hundreds of milliseconds. The lifetime observed for pB containing a trans chromophore is less than 1 ms under physiological conditions, which would be very inefficient for photochemical signal transduction.
Comparison of the Structure of Kinetic and Equilibrium pB Species-To further investigate the structure of pB trans , pB cis , pB, and pG trans , we studied and compared the aromatic fluorescence signals from these four species. Excitation of pG trans at 280 nm (pH 7.0) elicits a fluorescence emission peak at 330 nm (Fig. 3A, spectrum 1), consistent with the 5 Tyr residues and single Trp side chain in PYP. In the pB trans state at pH 2.0, the amplitude of this peak is reduced by ϳ39% and the emission maximum is shifted from 330 to 350 nm (Fig. 3A, spectrum  2). In addition to this aromatic fluorescence band, the spectra of pG trans and pB trans both display a second emission band of lower intensity, at 495 and 435 nm, respectively. These bands are caused by fluorescence emission from the pCA chromophore. The pCA in the pG trans state is known to have an emission maximum at 495 nm (27) (Fig. 3C, spectrum 13). To confirm that the emission band at 435 nm is caused by pCA, the   FIG. 2. Comparison of the temperature dependence of PYP refolding from pB cis and of the pB to pG trans photocycle transition. Temperature dependence of the rate constants for refolding for PYP containing cis pCA from pH 2.0 to 7.6 (filled triangles). The photocycle kinetics at different temperatures at pH 7.5 were reported previously (15) (filled circles). We attribute the small difference between the two curves to minor differences in experimental condition, such as buffer composition.
pH Jump Refolding Mimics the Photocycle of PYP emission spectrum of this species was measured upon excitation of the chromophore at 350 nm, and was indeed found to be at 435 nm (Fig. 3C, spectrum 9). This reveals fluorescence resonance energy transfer (FRET) from the aromatic amino acids to the pCA chromophore. The FRET pCA emission band in pB trans is a factor of 7.5 stronger that that in pG trans , presumably because of the increased spectral overlap between the emission from the aromatic amino acids and the absorbance of the pCA chromophore in pB trans .
The fluorescence emission spectra of pB cis were obtained by performing fluorescence spectroscopy on a mixture containing 50% pB trans and 50% pB cis , obtained as described above (Fig. 3, A (spectrum 3) and C (spectrum 10)). The data were then corrected for the contribution of the 50% pB trans to yield the pure spectrum of pB cis at pH 2.0 (Fig. 3, A (spectrum 4) for aromatic fluorescence and C (spectrum 11) for chromophore fluorescence). The intensity of fluorescence emission from the aromatic amino acids is increased a factor of 2.0 by the trans to cis isomerization of the pCA in the acid-denatured state of PYP, whereas the FRET emission band at 435 nm is completely lost in pB cis . The chromophore fluorescence from the pCA in pB cis is strongly quenched: fluorescence excitation at 350 nm does not yield an emission band at 435 nm, in contrast with pB trans (Fig. 3C, spectrum 11). In summary, the pB cis state exhibits fluorescence properties quite different from pB trans : (i) increased fluorescence of aromatic residues, (ii) strong quenching of pCA fluorescence, and, possibly because of this, (iii) no emission band by FRET from aromatic side chains to the pCA chromophore. This indicates that pCA trans to cis photoisomerization results in structural changes in the aciddenatured state of PYP.
The three fluorescence characteristics that clearly distinguish pB trans from pB cis were used to compare the structure of the kinetic pB photocycle intermediate with that of the equilibrium species pB cis . The fluorescence emission properties of the pB photocycle intermediate were reconstructed from kinetic measurements at pH 3.9 during the pB to pG trans photocycle transition, observed after switching off actinic illumination. Under the conditions used, essentially all PYP was converted to pB, and changes in fluorescence emission during relaxation to pG were determined. Both the aromatic (Fig. 3B, closed circles) and chromophore (Fig. 3C, closed circles, spectrum 12) fluorescence properties of the kinetic photocycle species pB are highly similar to those of the equilibrium aciddenatured species pB cis , indicating that these two states of PYP have a very similar structure.
Stabilization of the Acid-denatured State by Chromophore trans to cis Photoisomerization-To further investigate the effect of pCA isomerization on the acid-denatured state of PYP, the thermal stability of pB trans and pB cis were determined. Thermal denaturation of pB trans to the fully unfolded state resulted in a shift in absorbance maximum from 350 to 338 nm, and an increase in extinction coefficient at the absorbance maximum by 23% (Fig. 4). This shift occurred in a single, cooperative unfolding transition with a midpoint temperature of 42°C.
Conversion of 50% of the pB trans population to pB cis by UV-A illumination of the sample had a marked effect on the thermal denaturation curve. In this case, two thermal transitions were observed: one at 42°C, corresponding to pB trans , and a second at 72°C, corresponding to the thermal denaturation of pB cis (Fig. 4, closed squares). These data show that the trans to cis isomerization of the pCA chromophore significantly increases the thermal stability of the acid-denatured state (Table III). Quantitative analysis (using Equations 8 and 9) of the denaturation curve of pB cis was complicated by the fact that, even at 100°C, the thermal denaturation of this state was not complete, but indicates that pCA isomerization stabilizes the aciddenatured state of PYP by ϳ16 kJ/mol.
The native conformation of PYP has a strong energetic preference for the trans chromophore. Our results show that for the acid-denatured state the opposite is true: the cis conformation is energetically favored. However, when the pG state is acidified, it is partially unfolded to pB trans , not pB cis . This demonstrates that acid denaturation of PYP is under kinetic control (Fig. 4B); because of the higher activation barrier for the formation of pB cis , the acid-denatured state containing trans pCA is accumulated at low pH. Kinetic control was previously proposed by the Agard group (28,29) for protein refolding. Here, we report that, upon acidification, native PYP is converted to the local free energy minimum of the pB trans state, even though the pB cis state has a significantly lower free energy.
The pH Dependence of Protein Folding Kinetics in PYP: Transition State Movement-The results on the pH dependence of the kinetics for the formation of pG trans from the three blue-shifted states studied here result in the following challenge: which model can describe (i) the pH-independent noninteger slopes in the presence of trans-pCA and (ii) the transition from linear to nonlinear pH dependence upon chromophore isomerization?
This question is corroborated by a re-analysis that we performed of the published pH dependence of the pB to pG photocycle transition of a set of seven PYP mutants (30 -34) on a log-log scale, as we did for wild-type PYP (Fig. 5). Two groups of mutant PYPs could be classified according to the pH-dependent profile of refolding kinetics. The E46Q, M100A, R52A and T50V mutants show a nonlinear pH dependence, as does the wild type (Fig. 5A). The E46D, E46A, and Y42F mutants exhibit a linear pH dependence in the pH ranges studied (Fig.  5B). Apparently, not only chromophore photoisomerization, but also these side-directed mutations shift the pH dependence of the barrier for the pB to pG transition from nonlinear to linear. In addition, the pH dependence of the pB to pG transition in the E46D, E46A, and Y42F mutants is characterized by fractional ⌬Q ‡ values (see Fig. 5B).
We examined if the position of Tyr 42 , Glu 46 , Thr 50 , Arg 52 , and Met 100 in the x-ray structure of PYP (35) provides further insights into the pH dependence of the photocycle kinetics. All of these residues contribute to the first shell of atoms surrounding the pCA chromophore (see Fig. 2 of Ref. 12). The phenolic group of Tyr 42 and the acidic side chain of Glu 46 are directly hydrogen-bonded to the phenolate oxygen of the pCA chromophore, whereas the side chain of Thr 50 is hydrogen-bonded to that of Tyr 42 . The guanidino group of Arg 52 and the sulfur atom of Met 100 are placed immediately adjacent to the pCA. Interestingly, only mutants that disrupt the active site hydrogen bonds between the chromophore and Glu 46 and Tyr 42 result in a linear pH dependence.
Previously, the pH dependence of the kinetics of the pB to pG trans photocycle transition was described using equations developed for the pH dependence of enzyme catalysis, in which the pH dependence of the reaction is caused by equilibrium changes in the protonation state of groups with specific pK a values (30 -34). In this description a log-linear pH dependence pH Jump Refolding Mimics the Photocycle of PYP of the kinetics is interpreted as the titration of a functionally important group with a pK a outside the studied pH range. This results in a slope of 1 for low pH values and Ϫ1 for high pH values. Fractional values for the slope are difficult to understand based on such a model. Thus, the fractional values of ⌬Q ‡ , observed both for the folding kinetics of pB trans of native PYP and for the photocycle kinetics of the E46D, E46A, and Y42F mutants, provide an important clue to the origin of the pH dependence of the kinetics. Because we found that the pB cis to pG trans transition is a protein refolding event, we investigated an alternative interpretation, based on the analysis developed to describe the dependence of the kinetics of protein folding on denaturant concentration (chevron analysis). We propose that the fractional ⌬Q ‡ values for the transitions between pG trans and pB trans can be described in analogy with the analysis of a denaturant chevron plot for a two-state transition, with m ‡ values for unfolding and refolding that are independent of denaturant concentration. The effect of dena-turant concentration on the kinetic of folding provides information on solvent-exposed surface area upon transition state formation. Analogously, the effect of pH reveals changes in charge ⌬Q ‡ that are required to reach the folding transition state and provides a measure of the electrostatic interactions in the transition state (22,23,36,37). In this proposal the transfer of protons between the solvent and the protein occurs only partially during the formation of the transition state for pB trans refolding. The pH independence of the ⌬Q ‡ values for pB trans then indicates that the position of the transition state does not depend on pH, with ϳ25% (0.30/(0.30 ϩ 0.85)) of the native electrostatic interactions formed in the transition state for refolding. The most straight forward candidate of the major contribution to the observed ⌬Q ‡ values is the pCA chromophore, because the deprotonation of this buried group is an essential step in the formation of the native state.
The curved pH dependence for the kinetics of the pB to pG trans transition correspond to a curved chevron plot, which has been found for a range of proteins. Such curved denaturant chevron plots can be caused by two different phenomena: (i) the involvement of a folding intermediate (38,39) or (ii) movement of the transition state in a two-state transition (40). We propose a transition state movement to explain the nonlinear pH chevron plot of pB cis . First, this proposal is in line with the absence of indications for intermediates in the pB to pG photocycle transition or pH jump experiments. Second, this model can describe the observed curved pH dependence of the kinetics: a pH-induced transition state movement involves a change in the apparent ⌬Q ‡-D . This change in ⌬Q ‡-D for refolding from pB cis may reflect the titration behavior of a specific group or may be caused by the pH-induced shift from one barrier to a second barrier. We prefer the latter interpretation, based on the following reasoning. For refolding from pB trans , the chromophore deprotonation event is likely to result in the linear pH dependence of refolding. The refolding process from pB cis involves an additional process: thermal pCA re-isomerization. This is expected to give rise to an additional energy barrier on the folding landscape of pB cis . In this proposal, a transition state movement occurs from a pH-dependent pCA deprotonation barrier to a pH-independent pCA isomerization barrier (Fig. 6B). The pCA deprotonation then results in a log-linear pH dependence, as observed for pB trans , whereas the kinetics of pCA isomerization do not depend on pH. A shift between these two barriers results in the observed curved pH dependence. In the case of the mutants with a perturbed hydrogen bond network around the pCA chromophore, the pCA deprotonation process may well be strongly decelerated. This would prevent the isomerization step from becoming the dominant bar- . The pure pB cis spectrum (spectrum 4) was calculated from the spectra of pB trans (spectrum 2) and 50% mixture of pB trans and pB cis (spectrum 3) at pH 2. The spectrum of the pB photocycle intermediate (spectrum 8) at pH 4 was reconstructed from each kinetic trace at different wavelengths by taking the values at t ϭ 0. Native PYP fluorescence spectra (spectra 1 and 5) at pH 7 were recorded for direct comparison in the two independent experiments (A and B). Spectra 6 and 7 are the pG spectrum reconstructed from each kinetic trace, and continuously scanned at pH 4, respectively. C, chromophore fluorescence excited at 350 nm (spectrum 9 -12) or 440 nm (spectrum 13). The pure pB cis spectrum (spectrum 11) was calculated from pB trans (spectrum 9) and a 50%/50% mixture of pB trans and pB cis (spectrum 10). pCA fluorescence spectra of the pB photocycle intermediate at pH 4 were also reconstructed from each kinetic trace at different wavelengths (spectrum 12). Spectrum 13 is a pG fluorescence at pH 4.0.
FIG. 4. Kinetic control for protein unfolding to the acid-denatured state of PYP exerted by chromophore isomerization. A, thermal denaturation of pB trans (filled circles) and a pB trans /pB cis mixture (filled squares) at pH 2.0 monitored by absorbance at 337 nm. B, schematic energy surface of PYP at low pH.

TABLE III
Thermal stability of pB trans and pB cis at pH 2.0, 298 K The data for pB cis thermal denaturation (see Fig. 4A) did not show a plateau, complicating their analysis. The data were analyzed using two sloping baselines for the native and denatured state (a), and with one sloping baseline for the native state and a fixed value for the thermally denatured state at 95°C (b). pH Jump Refolding Mimics the Photocycle of PYP rier in the studied pH range, resulting in the observed pHindependent slope. An attractive aspect of this analysis of the pH dependence of the PYP photocycle kinetics, is that it offers a tool to disentangle the multiple molecular events that occur during the pB to pG photocycle transition.
Conclusions-A striking feature of the results reported here is that the kinetics of protein refolding for the acid-denatured state of PYP containing cis pCA is exactly same as that of the photocycle step from pB to pG trans over a wide pH (Fig. 1B) and temperature (Fig. 2) range. Therefore, the final step in the PYP photocycle corresponds to a protein refolding reaction, implying a direct link between transient protein unfolding and photosensory signal transduction (Fig. 6A).
We demonstrate that the isomerization state of the pCA chromophore has a number of significant effects on the aciddenatured state of PYP: (i) the kinetics of refolding to pG trans are decelerated by 3-5 orders of magnitude for cis pCA; (ii) the fluorescence properties of both the pCA chromophore and the aromatic amino acids are altered, indicating differences in tertiary structure; and (iii) the stability of the acid-denatured state are stabilized by ϳ16 kJ/mol by the presence of cis pCA. The stabilizing effect of the cis chromophore of 16 kJ/mol is likely to contribute significantly to the reduced refolding rate for pB cis . From the finding that both the aromatic fluorescence and the stability of the acid-denatured state are significantly affected by the isomerization of the pCA chromophore, it can be concluded that protein-chromophore interactions play an important role in the acid-denatured state.
Our results demonstrate how PYP provides as a strong model system to study protein folding, in which the folding reactions can be initiated not only by jumps in pH or denaturant concentration but also by pCA photoexcitation. We have used this opportunity to study three general issues in protein folding. (i) We performed a comparison of partially folded species observed in equilibrium and kinetic experiments, indicating a clear resemblance between the equilibrium species pB cis and the kinetic intermediate pB. (ii) We provided a clear example of kinetic versus thermodynamic control in protein unfolding. (iii) We investigated the origin of linear and nonlinear pH dependence of the barrier for protein folding. We formulate a general model that can provide a quantitative explanation for transitions that exhibit a pH-independent factional value of ⌬Q ‡ , which has also been reported for bacteriorhodopsin (41). We propose a transition state movement by a change in the highest activation barrier from pCA deprotonation to pCA isomerization to explain the nonlinear pH dependence of the refolding kinetics of pB cis (Fig.  6B). This proposal provides a specific example of the structural basis for such a transition state movement and must be considered as an alternative for interpretations based on the involvement of the pK a of specific groups when nonlinear pH chevron plots are observed.
Acknowledgments-We thank Philippe Cluzel, Marvin Makinen, Joe Piccirilli, Tobin Sosnick, and Aihua Xie for valuable comments and discussions, and thank Applied Photophysics for excellent technical support. A, mimic of the PYP photocycle by a protein folding reaction. B, a pH-induced transition state movement by a change in the highest activation barrier from a pH-dependent chromophore deprotonation energy barrier (TS1) to a pH independent chromophore isomerization barrier (TS2). The order in which the two barriers occur is not known.