Eur. J. Biochem. 209, 759-764 (1992)

c) FEBS 1992

Characterization of tryptophan phosphorescence of aspartate aminotransferase from Escherichia coli ’, James Joseph ONUFFER and Giovanni Battista STRAMRTNI Consiglio Nazionale delle Ricerche, Islituto di Biofisica, Pisa, Italy Department of Molecular and Cell Biology, Division of Biochemistry and Molecular Biology, University of California, Berkeley, USA

Patrizia CIONI I

(Received March 16/July 16, 1992) - EJB 92 0362

The Trp phosphorescence spectrum, intensity and decay kinetics of apo-aspartate aminotransferase, pyridoxamine-5P-aspartate-aminotransferaseand pyridoxal-5P-aspartate aminotransferase were measured over a temperature range 160-273 K. The fine structure of the phosphorescence spectra in low-temperature glasses, with 0-0 vibrational bands centered at 408, 415 and 417 nm, for both apoenzyme and pyridoxamine-5P-enzyme reveals a marked heterogeneity of the chromophore environments. Only for the pyridoxal-5P form of the enzyme is the triplet emission strongly quenched and, in this case, the spectrum displays a unique 0-0 vibrational band centered at 41 5 nm. Concomitant to quenching, there is Trp-sensitized delayed fluorescence of the Schiff base, an indication that quenching of the excited triplet state is due, at least in part, to a process of triplet singlet energy transfer to the ketoenamine tautomer. All three forms of the enzyme are phosphorescent for temperatures up to 273 K. However, across the glass transition temperature the pyridoxal-5P enzyme shows a decrease in lifetime-normalized phosphorescence intensity, a thermal quenching that reduces even further the number of phosphorescing residues at ambient temperature. In fluid solution, the triplet decay is nonexponential and multiple lifetimes stress the heterogeneity in dynamical structure of the chromophores’ sites. For the pyridoxal-5P enzyme, where only one or at most two residues are phosphorescent at 273 K, the nonexponential nature of the decay implies the presence of different conformers of the protein not interconverting in the millisecond time scale.

Aspartate aminotransferase (AATase) is a dimer; each monomer is composed of a large and a small domain and the active sites are shared between the two monomers. The enzyme catalyzes the reversible transamination between the L-dicarboxylic amino acids, Asp and Glu, and the corresponding oxoacids, oxalacetate and 2-oxoglutarate [I]. AATase requires the coenzyme pyridoxal-SP for catalysis, which is converted during the transfer of the amino group, to pyridoxamine-5 P

PI.

According to crystallographic studies on eukaryotic AATase [l], binding of substrates and dicarboxylic inhibitors induces a conformational change modeled as a 13 O rotation of the small domain relative to the large domain, thereby closing the active-site cleft to bulk solvent. Although the structure of AATase has been thoroughly investigated in the crystalline state, little is known of the conformation and, in particular. on the occurrence of open/closed Ftructures in solution, even if it is generally believed that domain closure is required for efficient catalysis [2]. Indirect evidence for the presencc of open and closed conformers in solution has been obtained for chicken mitochondrial AATase using a fluorescence probe [3]. Recent crystallographic studies [4, 51 revealed that the enzyme Correspondence to P Cioni, Consiglio Nazionale delle Ricerche, Istituto di Biolisica, Via S. Lorenzo 26,I-56127 Pisa, Italy

+ 39-50-553501.

FUX:

Abbreviation. AATASE, Aspartatc aminotransferase Escherichia coli. Enzymes. Aspartate aminotransferase (EC 2.6.1.1).

from

from Escherichia coli is very similar in structure to that of chicken cytosolic and mitochondrial isozymes and, it is therefore reasonable to expect that the same structural isomerizations are required for the catalytic function. The fluorescence and phosphorescence properties of Trp residues in proteins have long been utilized to derive information regarding the chemical nature and dynamics of the surrounding protein structure [6,7], the overall hydrodynamic features of the macromolecule as well as the occurrence of conformational changes upon ligand binding [8- 101. In the present investigation, we characterize the phosphorescence properties of the five Trp residues/monomer of AATase from E. coli, both in the apoenzyme and the holoenzyme. The aim is to assess the potential of this intrinsic chromophore as a monitor of possible changes in the conformation of the polypeptide that may occur between oxidized and reduced forms, as well as for intermediates along the reaction pathway. The results show that low-temperature phosphorescence spectra distinguish the Trp residues into three classes and, as opposed to apoenzyme and pyridoxamine-5P forms, the emission is strongly quenched in the pyridoxal-5P-enzyme. All three enzyme forms are phosphorescent at room temperature, and the decay of this emission for the pyridoxal-5P-AATase gives evidence of conformational heterogeneity in solution. MATERlALS AND METHODS Pyridoxal-5P hydrochloride and pyridoxamine-SP hydrochloride were obtained from Sigma. All the other reagents

760 were of analytical grade and double distilled water was used throughout. AATase from E, coli was purified by the method reported in [ll]. The pyridoxamine-5P form of the enzyme was prepared with cysteine sulfinate. Complete saturation by the coenzyme was established from the absorbance ratio A n 0 / A360and A 2 8 0 / A 3 3 for 0 the pyridoxal-5P and pyridoxamine5P forms, respectively. The apo form of the enzyme was prepared by ammonium-sulphate precipitation of the pyridoxamine-SP form at pH 4.9, as described by Wada and Snell (1962) [32]. The absence of coenzyme was tested by the lack of absorption at 330 nm and a residual specific activity which was less than 2 % of that of the holoenzyme. Reconstitution of the apoenzyme with pyridoxal-5P restores 90-100% of the activity. AATase activity was determined as described in [l11. Residual ammonium sulphatc was removed by Sephadex G-25 column chromatography. The concentration of pure protein was estimated from the absorbance at 205 nm using A z o sof 31 [13]. Preparation of Lys Schiff base was carried out in the presence of excess Lys (10 times) so as to avoid the presence of free coenzyme in solution. Prior to luminescence measurements, protein samples were extensively dialyzed (12 h) in 20 mM potassium phosphate, pH 7.4. Luminescence experiments wcrc carried out at a protein concentration of 2 mg/ ml. At this concentration and employing sample cells with 2-mm pathlength, inner-filter artifacts were found to be negligible. Phosphorescence and fluorescence measurements were obtained with a homemade fluorometer/phosphorimeter [14]. The exciting light was provided by a high-pressure 100-W Hg lamp (hBO/W2, Osram). The excitation was selected by a 250-mm grating monochromator (Jobin-Yvon, H25) and the emission was detected with an EM1 9635QB photomultiplier. Luminescence decays in fluid solution were obtained following pulsed excitation by a frequency doubled-flash pumped dye laser (UVSOO M, Candela) with a pulse duration of 1 ps and an energyipulse typically of 1 10 mJ. The decay ofTrp luminescence was monitored at 430 nm by an electronic shutter arrangement permitting the emission to be detected 1 ms after the excitation cut off. The decaying signal was digitalized by an Applescope system (HR-14 RC Electronics), then transferred to an Apple 11 computer for averaging and subsequent exponential-decay analysis by a least-squares method. The analysis of decay curves in terms of the sum of exponential components was carried out by a non-linear least-squaresfitting algorithm implemented by the program Global Analysis (Global Unlimited, LFD University of Illinois, Urbana). To obtain rcproducible phosphorescence data in fluid solution; it is of paramount importance to thoroughly remove all dissolved oxygen. The procedure followed to obtain satisfactory deoxygenation was described in a previous report 181 ~

RESULTS Low-temperature Trp lumincscencc Phosphorescence spectra of apoenzyme, pyridoxal-5Paspartate aminotransferase and pyridoxamine-5P aspartate aminotransferase from E. colr in propylene glycol/20 mM potassium phosphate glass at 160 K are displayed in Fig. l . Upon excitation at 298 nni, the Trp phosphorescence spectrum of apoenzyme and pyridoxamine-5P-enzyme is characterized by three distinct 0-0 vibrational bands centered at 408, 415 and 417 nm. The energy separation reflects a marked

400

125

450 WAVELENG-H

41s

500

(nn)

Fig. 1. Phosphorescence spectra of apo-AATase and pyridoxamine-5PAATase (A) and pyridoxal-SP-AATase (B) in propylene glycol/20 mM potassium phosphate (50:50, by vol.). IZeXLIldllOn. 298 nm; T, 160 K. Protein concentrations were typically 2 mg/ml.

heterogeneity in the chemical environment of the Trp residues. The group that contributes to the blue component (408 nm) has an emission similar to free Trp in the same solvent (407 nm) and could be due to solvent-exposed indole side chains. The bands centered at 415 nm and 41 7 nm represent contributions from buried Trp. As the 0-0 vibronic band in the phosphorescencc spectrum of Trp in non-polar solvents is characterized by peak wavelengths of 41 1 nm and 412 nm, the observed red shift implies a stabilization of the triplet state by aniostropic polar interactions. It should be pointed out that the differences in vibrational structure between apo-AATasc and pyridoxamine-SP-AATase suggests that, as far as the environment of the residue is concerned, there are no major conformational changes in the protein accompanying the binding of reduced cofactor. Environmental heterogeneity of the five Trp residues in the protein is reflected also in their absorption spectrum. Exciting on red edge of the absorption spectrum, at 308 nm, the intensity ofthe 408-nm and 41 5-nm bands decreases markedly relative to the 417-nm component (data not shown). Strong quenching of the phosphorescence emission and a lack of heterogeneity in the spectrum are characteristic of the pyridoxal-5P form of the enzyme (Fig. 1). In this case, we observe a single 0-0 vibronic band centered at 415 nm. Further, an anomalous shoulder in the spectrum of Trp is evident around 470 - 480 nm which, as described later, is due to Trp-sensitized delayed fluorescence of bound pyridoxal5P. Since the 415-nm band coincides with one of the three bands of apoenzyme and the pyridoxalamine-5P form of the enzyme, it would appear that the oxidized enzyme selectively quenches those Trp residues contributing to the 408-nm and 417-nm bands. Phosphorescence intensities after normalization by the fluorescence emission, (P/F), are reported for the three forms of the enzymc at 160 K in Table 1. For pyridoxalamine-5PAATase the value of P/F is the same as that of the apoenzyme and is typical for unperturbed Trp in proteins. In contrast, the pyridoxamine-5P enzyme shows a phosphorescence intensity that is 20% that of the other forms. Thus, while it is known that coenzyme binding quenches Trp fluorescence in both pyridoxal-5P and pyridoxamine-5 P holoprotein [15, 161, the

,

761 Table 1. Relative phosphorescence yiclds and lifetime of apo-AATase, pyridoxarnine-5P-AATase and pyridoxal-SP-A ATdse as a function of temperature in propylene glycoI/tO mM potassium phosphate, pH7.5, (S0:50, by vol.). Concentrations were 2 rng/ml. Aexcitatton, 295; AemlSSLOn, 430 nm.

Enzyme rorm

T

P/F

K Apo-AATase

* 0.02

160 220 250 273

0.10

Pyridoxal-5P- AATase

160 220 250 273

0.02

Pyridoxamine-5P- AATase

160 220 250 273

0.10 & 0.01

0.005

P/F ratio emphasizes that the triplet-state emission is affected exclusively in the Schiff base. A possible reason for this selectivity might be found in the red-shifted absorption spectrum of the oxoenamine tautomer of the Schiff base whose longwavelength absorption band centered at 430 nm provides good overlap with the Trp phosphorescence spectrum for an energy-transfer process. Energy migration among proximal Trp residues in a protein may lead to extensive depolarization of their luminescence. When excited at 300 nm and monitored over the 0-0 vibrational bands, the phosphorescence anisotropy at infinite viscosity, An, was found to be -0.07 0.01 for pyridoxal-5PAATase and -0.08 f 0.01 [or apo and reduced forms of the enzyme. For Trp in rigid matrices, a value of An lower than -0.14 is invariably an indication of singlet-singlet or triplettriplet energy transfer among aromatic residues [17]. The constant value of the phosphorescence anisotropy during the decay of the triplet signal suggests also that the process of energy migration is confined mainly to the fluoresccnce statc. The decay times of the Trp phosphorescence intensity (&tation = 298 nm, l.emisnion = 430 nm) are reported for the three forms of the enzyme at 160 K in Table 1. The decay of apo-AATase and pyridoxal-5P-AATase follows a very nearly exponential law (after subtracting the few percent of a short lived component due to the solvent) with lifetimes of 5.8 f 0.1 s and 5.5 f 0.15 s, respectively, a value typical of unperturbed Trp in proteins that are in glass matrices. For the pyridoxamine-5P-enzyme, the decay is slightly non-exponential and a biexponential fitting yielded a fraction 0.4 0.5 of the emission with a lifetime of 5.0 f 0.1 s, the remainder having a lifetime of 6.6 0.1 s. Altogether, the triplet lifetime of the Trp residues in AATase are little affected by the coenzyme. While this finding could be anticipated for the reduced form, whose emission is similar to the apoprotein, the 80% quenching of phosphorescence associated with the pyridoxal-5P form entails a considerable reduction in z for most of the five chromophores. The unexpectedly long triplet lifetime in the pyridoxal-5P form can be rationalized only if the interaction between the triplet state and coenzyme is negligible between one or at most two indole side-chains and so large with the rest as to make their decay kinetics too rapid for detection. This interpretation is in accord with the absence of two of the three bands in the phosphorescence spectrum.

1.o 0.7 0.3 0.8

0.5 0.046

5.51 2.52 0.92 0.038

1.012 0.323 0.008 0.715 0.241 0.010

1 0.6 0.4 0.1

5.5 1.92 0.48 0.01 8

6.712 4.823 0.833 0.035

5.031 2.902 0.310 0.012

0.05 0.3 0.3 0.8

5.8 3.5 0.45 0.046

5.81 2.32 1.01 0.04

5.8 1.9

Phosphorescence thermal profile

Upon warming a glass matrix into a fluid solution, the phosphorescence lifetime and quantum yield of indole and its derivatives are dramatically reduced by the change in viscosity [18]. One may utilize this remarkable sensitivity of T to the local viscosity about the indole ring, to differentiate Trp residues in a protein according to the dynamic make up of their environments. The decrease in solvent viscosity upon raising the temperature above the glass transition induces a large decrease in the Trp lifetime of AA'Tase and the decay becomes highly multiphasic in all three forms of the protein. These decays were fitted satisfactorily to a biexponential law (x' < 2) and the amplitude and lifetime of the components at selected temperatures are given in Table 1. Although each form of thc enzyme exhibits decay kinetics, the average lifetime, ,z, = c[,zl z2z2, is similar in them up to 250 K. At higher temperatures the pyridoxal-5P form of AATase shows a drastic drop in amplitude of the long-lived component. Selective thermal quenching of one or more Trp residues in a protein entails a decrease in the lifetime-normalized steady-state phosphorescence intensity, P/z, and the residual value of this parameter often gives a fair estimate of the number of Trp that still contribute to the emission at a given temperature [19]. For systems with biexponential decays, P/z can be represented by Pljzl + P2/z2, Pi being the fractional intensity decaying with lifetime T ~ The . behavior of ,ZPi/ziof AATase upon raising the temperature from a glass to 273 K is shown in Fig. 2. Apoenzyme and pyndoxamine-5P forms of the enzyme, inspite of the dramatic decrease in z, do not show any significant change in this ratio across the entire temperature range. The constant value of CPi/ti implies that no component involved in the triplet decay becomes shorter than 1 ms, the detection limit of the appartus, and that even at the highest temperature the measured decay represent the contribution of all five Trp residues. (This finding also confirms that no side chain is solvent exposed as z in aqueous solution is 15 - 20 p).In contrast, the phosphorescence of the pyridoxal-5P-enzyme is largely quenched in fluid solution and at 273 K P/z is reduced to 30%. Although thermal quenching is generally interpreted in terms of a flexible environment about the phosphorescing residues, in the case of pyridoxal5P-AATase, there is also the possibility of temperature-in-

+

762 - "

LO

N

biexponential function. Fitting of the data yielded lifetime 1

__-

--

U

'. \.

.' \

1

\

\

.=

0 150

---.

components of about 40 ms and 10 ms in all three enzymes forms. The pre-exponential terms obtained from the fitting indicate that the short-lived component dominates the emission in the pyridoxal-5P form (CI, = 0.9), as opposed to the pyridoxamine-5P and the apoenzyme where the amplitude of this component is only 50% that of pyridoxal-5P. These results are analogous to those obtained in the organic solvcnt mixture, propylene glycol/20 mM potassium phosphate, except that, for pyridoxamine-SP and apoforms, the amplitude of the long-lived component is 0.8 instead of 0.5. Coenzyme luminescence

In glasses, free pyridoxamine-5P, upon excitation at 298 nm, exhibits a fluorescence band centered at 365 nm and Temperature ( 4 ) a broad phosphorescence emission peaked at 440 nm. Exciting at longer wavelengths, to the red of Trp absorption (330 nm), Fig. 2. Phosphorescence thermal profiles of AATase in apo-AATase(A), the spectra are identical, but, in comparison to that at 298 nm, pyridoxamine-SP-AATasc ( A ) and pyridoxal-SP-AATasc (m) in 50:SO the fluorescence is more intense whereas the phosphorescence (by vol.) propylene glycol/20 rnM potassium phosphate. The excitation wavelength was 295 nm, the emission for decay measurements was is weaker. Under the same conditions, the free Lys-pyridoxalmonitored at 430 nm. The protein concentration was typically 2 mg/ 5P Schiff base (excited at 298 nm) shows two intense fluoresml . cence emission bands with maxima at 365 nm and 483 nm and a single phosphorescence band centered at 455 nm. Like for the reduced form of the cofactor, excitation at 300 nm does not induce any shift in emission spectra but alters their relalive intensity and reduces the phosphorescence emission. The coenzyme luminescence in AATase is evident only with selective excitation, that is 320 nm. The spectra have roughly the same wavelength maxima as those of the free cofactors. While there is no apparent difference in the emission characteristics between free and bound pyridoxamine-5P, the fluorescence intensity ratio F365/F490 for protein-bound pyridoxal-SP has a value of 1.8 as opposed to 10.2 for the free Lys Schiff base. As the two fluorescence bands arise from different tautomers of the Schiff base [20], 3 76 148 221 293 366 this result indicates that when bound to the protein eithcr the equilibrium between tautomers and/or the fluorescence quantum yield strongly favors the oxoenamine form. Although the quantum yield of both fluorescence and -0.5 I , , , , 1 phosphorescence decreases sharply above the glass transition ! 8 IS 22 29 ? I , temperature, quenching effects are most dramatic on the triplet-state emission. Pyridoxamine-SP, free or bound, is no longer detectable phosphorescently above 195 K. At this temperature, the phosphorecence of the free Schiff base is almost Q totally quenched as opposed to that of bound coenzyme which L 1.38 is still detectably by phosphorescence up to 220 K. Q Finally, we consider the anomalous shoulder at 460480 nm (Fig. 1) in the Trp phosphorescence spectrum of o.ni+mrr&-rj-r, I 1 1 I i I ,r pyridoxal-5P-AATase. Two observations indicate that this 1 15 30 41 59 73 emission arises from Trp-sensitized delayed fluorescence of channel number the Schiff base. The difference spectrutn between pyridoxalFig. 3. Phosphorescence decays of apo-AATase (A) and pyridoxal-SP- SP-AATase and apoenzyme (Iucx = 298 nm; data not reportAATase (6) in 20 mM potassium phosphate, pH 7.5 at 0 C. ,Iexcilation ed) shows that the shoulder is resolved into a band centered 295 nm, Asmission, 430 nm; the fitting to a biexponential law yields a at about 490 nm, very similar to the red-fluorcscence band of value of x2 less than 2 in each case. The time interval for each channels the oxidized coenzyme and red shiftcd over 30 nm with respect is 0.14 ms in (A) and 0.51 ms in (B). to the coenzyme phosphorescence band. Further, the cofactor phosphorescence can be excluded since, above 220 K, it is totally quenched (direct excitation) whereas the shoulder in duced quenching by enhanced energy transfer to pyridoxal- the spectrum remains even at higher temperatures 5P, or by other mechanisms, whose dependence on the spatial ( T 2 250 K). The appearance of sensitized delayed fluoresorientation of the chromophorc may be partially relaxed at cence (recently observed also in the pyridoxal-SP-dependent warmer temperature due to a greater protein flexibility. enzyme tryptophan synthase, unpublished results) dcmonIn buffer at 273 K, the phosphorescence decay of the three strates that, one mechanism involved in the phosphorescence forms of the enzyme (Fig. 3) can be adequately fitted to a qucnching of pyridoxal-SP- AATase is triplet-singlet, Forster 200

300

252

I

1

763 type, energy transfer, a process illustrated before in model studies, [21] but never reported to occur in native-protein systems.

DISCUSSION Dimeric B. mli AATase possesses five Trp residues in each subunit, all located within the large domain [22]. Four Trp out of five are in a Rossman-type E/Psupersecondary structure constituting the coenzyme-binding site, and are in close proximity to each other (within 1.9 nm). The fifth residue, Trp319, is placed in an CI helix that, in mitochondria1 AATase, was found to form a rigid core of the macromolecule [22]. All indole side chains are within 1.95 nm of the pyridoxal ring of thc same subunit. While this distance is well within the range of Forster-type Trp to coenzyme energy transfer, the separation from the coenzyme in the adjacent subunit is too large (> 2.68 nm) for any effective interaction. Trp-triplet emission at low temperature has been widely used for probing both the polarity of the local environment of the indole side chain as well as the degree of solvent exposurc. The phosphorescence spectrum of apo-AATase is clearly heterogeneous with three distinct 0-0 vibronic bands indicating that, in terms of this parameters, the five Trp residues in AATase can be broadly grouped into three classes. The number of residues contributing to a particular band is, under conditions of uniform emission yield, simply proportional to the intensity of that band. In AATase, because thc phosphorescence is largely depolarized, we must conclude that inter-Trp energy migration is rather efficient and that, as a consequence, the spectrum is not a true representation of the number of residues associated with each 0-0 vibronic band. Such ambiguity renders the interpretation of phosphorescence data, such as quenching by coenzyme and thermal behaviour, in terms of specific residues a rather difficult task. For example, Trp fluoresccnce is extensively quenched (by approximately 80%) upon pyridoxamine-5P binding to the apoenzyme. The quenching process presumably involves singlet-singlet energy transfer to the coenzyme and therefore preferential quenching would be anticipated for those residues in close proximity or better oriented for it. Churchich [15], from a number of pyridoxamine-5P protein conjugates, estimated an average critical transfer ratio, R,, of about 2.0 nm. According to the crystallographic coordinates of AATase, the separation between the aromatic ring of the coenzyme and the indole ring is 0.3 nm for Trpl40, 1.3 nm for Trp205, 1.6 nm for Trp319,1.7 nmforTrp134 and 1.9 nmfor Trp217. Barring unfavorable orientations (a parameter that is not unambiguously determined unless it is known which is the fluorescence state, L, or Lb), energy transfer would occur from each residue although, on the basis of the distances involved, large differences are anticipated in the efficiency and thus on the extent of quenching. This prediction of selective quenching of fluorescence and consequently of phosphorescence by pyridoxamine-5P is not bornc out expcrimentally as the relative intensity of the 0-0 vibronic bands is similar between pyrimidoxamine-5P-enzyme complex and apoprotein. Since there is no additional quenching at the triplet level by the reduced coenzyme it is possible that preferential quenching of fluorescence be masked by efficient intertryptophan energy migration that observed also in the pyridoxamine-5P complex. Unlike pyridoxamine-SP, pyridoxal-5P affects both fluorescence and phosphorescence emission. The 70% decline in the fluorescence intensity is accompanied by a further 80%

reduction in the phosphorescence intensity. The interaction with the Schiff base displays a clear selectivity for certain Trp residues. This is evident in both the phosphorescence spectrum and the decay kinetics. The former is sharper, with a single 0-0 vibronic band (415 nm) as opposed to three in the apoprotein. The phosphorescence intensity, inspite of extensive quenching by the Schiff base, decays with an unperturbed lifetime of 5.5 s. The implication is that the interaction with pyridoxal-5P is very strong and leads to complete quenching of the majority of the residues, but is practically null, at the triplet level, for the rest. Since the phosphorescence spectrum and decay are homogeneous and the residual intensity is 20% of the total anticipated, it is reasonable to suppose that there is a sole Trp, or at most two, still phosphorescing in the pyridoxal-5P enzyme. We may attempt to identify which is/are the emitting residue(s), by taking advantage of the observation that tripletsensitized delayed fluorescence of pyridoxal-5P confirms that quenching by triplet-singlet (Forster-type) energy transfer is involved in the process. However, since electron transfer is another possible quenching route [23, 241, we stress that until it is shown that quenching is exclusively by energy transfer, this assignment can only be tentative. Based on energy transfer alone, the relative efficiency among Trp residues, assuming a constant overlap, is given by ki2 . rj6/kj2 . ri6, where k is the orientation factor (between transition dipole moments) and r is the distance that separates indole and pyridoxal rings [25]. A quantitative calculation, based on an orientation of the triplet transition dipole moment perpendicular to the indole ring [26], shows that the interaction is weakest with Trp134 (1.7 nm) followed by Trp319 (1.6 nm). The ratio of the rates of energy transfer predicted for these side chains is 0.5. Since, on these grounds, the discrimination between Trp134 and Trp319 is weak, we must conclude that either both are phosphorescent or that other quenching mechanisms are at work. The excited triplet state is strongly affected by changes in the viscosity of its immediate environment and the decrease in the phosphorescence lifetime and quantum yield observed upon warming glassy matrices into fluid solution is a direct measure of the increased flexibility of the protein/solvent structure surrounding the chromophore. When two or more Trp are present in a macromolecule or the latter is not conformationally homogeneous, the thermal-quenching profile of the phosphorescence intensity often provides a means for distinguishing different classes of residues according to the dynamic make up of their local environment 119, 271. The lifetime-normalized phosphorescence intensity is roughly constant ( + l o % ) over 160-273 K for apo and pyridoxamine5P-AATase, but decrease by 70% in the pyridoxal-5P enzyme. This finding indicates that, for the apoenzyme and the pyridoxamine-5P-enzyme, practically all the Trp residues that are emitting at I60 K are still phosphorescent at room tempcrature, whereas pyridoxal-P-AATase, only about 33% of them are detected by phosphorescence. The triplet lifetime at ambient temperature are relatively long (10 - 40 ms) for all three forms of the enzyme (compared to 15 -20 ps for a solvent-exposed residue) and the correlation between T and local viscosity establishes that thesc chromophores are embedded in rather rigid pockets of the polypeptide chain. With both apoenzyme and holoprotein, the triplet decay in fluid solutions cannot be fitted to a monoexponential law. Heterogeneous decay kinetics may arise from distinct flexibility of the sites of Trp residues in the same macromolecule or from the presence of multiple conformers. While, for apoAATase and pyridoxamine-5P-AATase with about five resi-

764 dues emitting a multicomponent decay is easily rationalized in terlns of environmentally different Trp residucs in the same subunit, the heterogeneity in the pyridoxal-5P form, where only one or at most two residues are phosphorescent, necessarily implics conformational differences in the polypeptidc. Indeed, for pyridoxal-5 P-AATase, the phosphorescence at 273 K decays with at least three distinct lifetimes; z1 ,= 10, z2 = 40 and T~ 5 1 ms, the latter representing that portlon of the intensity no longer detectable at ambient temperature (scc thermal profile). Hence, as the number of distinct decay rates is larger than the number of chromophores, we must conclude that at least one of these components represents a different conformer of the macromolecule or subunit not rapidly interconverting during the time of phosphorescence. In principle, other sources of heterogeneity could be represented by incomplete coenzyme saturation and partial dissociation of the dimer. Howcver, while the former was ruled out by absorbance measurements, the small dissociation constant of the dimer [28] and the large protein concentration involved show that the monomer fraction is negligible. Should selective amino acid replacement by site-dirccted mutagenesis establish, besides the correct assignment of emitting residues, that the phosphorescence of the Schiff-base form of the enzyme originates from a single Trp, then the triplet probe could provide a simple means for monitoring the fraction of macromolecules in different conformational states. Indirect evidence for conformational heterogeneity was also obtained in recent solution studies [ 3 ] where the hypothesis of an equilibrium between open and closed states in the unliganded holoenzyme was advanced. The striking similarity in all phosphorescence parameters between apo-AATase and pyndoxamine-SP-AATase suggests that the interaction with the reduccd coenzyme has little influence on the structure of the chromophore environment. From this, one may infer that any conformational equilibrium present in solution is not significantly altered by binding of pyridoxamine-5P. The comparison cannot be extended to the pyridoxal-5P form because the triplet emission is dramatically quenched by bound pyridoxal-5P. In characterizing the luminescence properties of Trp residues in AATase, the present investigation has emphasized the spectral heterogeneity of the emitting centers, the strong and selective interaction of the triplet state of certain residues with pyridoxal-5P and the persistence of long-lived phosphorescence in fluid solution at ambient temperature. Since these spectroscopic parameters are sensitive to both structural and dynamic features of the macromolecule, they can provide, especially if combined with genetic engineering, an intrinsic monitor of the solution structure of the protein and of the changes that may be induced by chemical modification and/ or ligand binding. The authors are grateful to Mr Puntoni Alessandro for his technical assistence and to Ms Neri Claudia for tvning" the manuscrint. < I

REFERENCES 1. Jansonius, J. N. &Vincent, M. G. (1987) in Biological macromol-

ecules and assemblies: active site of enzymes (Jurnak, F. A . & McPherson, A,, cds) vol. 3, pp 187-285, Wiley, New York. 2. Kirsch, J. F., Eichele, G., Ford, G. C., Vincent, 11. G., Jansonius, J. N., Gehring, H. & Christen, P. (1984)J. Mol. Biol. 174,497525. 3. Picot, D., Sandmeier, E.: Thaller, C.; Vincent, M. G.. Christen, P. & Jansonius, J. N. (1991) Eur. J . Biochem. 196, 329- 341. 4. Jaeger, J., Koehler, E., Tucker, P., Sunder, V., Houslcy-Markovie, Z., Fothcringham, I., Edwards, M., Hunter, H., Kirschner, K. & Jansonius, J. N. (1 989) J . Mot. Biol. 209, 499 - 501. 5. Smith, D. L., Ringe, D., Finlayson, W. L. & Kirsch, J. F. (1988) J . Mol. Bid. 191, 301 -302. 6. Saviotti, M. L. & Galley, W. C. (1974) Proc. Nut1 Acad. Sci.U S A 71,4154-4158. 7. Strambini, G. B. (1989)J. Mol. Liq. 42, 155-165. 8. Strambini, G. B. & Gonnelli, M. (1990) Biochemistry 29, 196203. 9. Cioni, P. & Strambini, G. B. (1989) J . Mol. Biol.207, 237-247. 10. Cioni, P., Piras, I,. & Strambini, G. B. (1989) Eur. J . Biochem. 18.5, 573 - 579. 11. Cronin, C. N. & Kirsch, J. F. (1988) Biochemistry 27, 45724579. 12. Wada, H. & Snell, E. E. (1962) J . R i d Chem. 237,127-132. 13. Scopes, R.K. (3974) Anal. Bi~chem.59,277-2x2. 14. Strambini, G. B. (1983) Biophys. J . 43, 127-130. 15. Churchich, J. E. (1965) Biochemistry 4 , 1405- 1410. 16. Arrio-Dupont, M. (1978) Eur. J . Biochem. 91, 369-378. 17. Strambini, G. B. & Galley, W. C. (1976) Nature 260, 554-556. 18. Strambini, G. B. & Gonnelli, M. (1985) Chem. Phys. T,eti. 155. 196 - 201. 19. Strambini, G. B. & Gabellieri, E. (1990) Photochem. Photobid. 5 1, 643 - 648. 20. Kallen, R. G., Korpela, T., Martell, A. E., Matsushina. Y.. Metzlcr, D. E., Morozov, Yuru, V., Ralston, I. M., Savin. F. A., Torchinski. Y. M. & Ueno, H. (1985) in Transaminuses (Christen, P. & Metzler, D. E., eds) pp 38-108, Wiley, New York. 21. Galley, W. C. & Stryer, L. (1969) Biochemistry 8, 1831 -1838. 22. Jansonius, J. N., Eichele, G., Ford, G. C., Pirot, D., Thaller, C. & Vincent, H. G. (1985) in Transaminuses (Christen, P. & Mctzler, D. E., eds) pp 109-231, Wiley, New York. 23. Strambini, G. B. & Gabcllieri, E. (1991) J . Phys. Chem. 95,43474352. 24. Vanderkooi, J. M., Englander, S. W.. Papp, S., Wright, W. W. & Owen, C. S . (1990) Proc. Natl Acad. Sci. USA 87, 5099-5103. 25. Lakowicz, J. R. (1983) in Principles of fluorescence spectroscopy. pp 305 -336, Plenum Press, New York and London. 26. Konev, S . V. (1967) in Fluorescence andphosphorescence o f p r o teins and nucleic acids (Udenfriend, S . , ed.) pp 106- 140, Plenum Press, New York. 27. Domanus, J., Strambini, G. B. & Galley, W. C. (1980) Photochem. Photubiol. 31, 15-21. 28 Herold, M. & Kirschner, K. (1990) Biochemistry 29, 1907- 1913.