PROTEINS: Structure, Function, and Genetics 40:112–125 (2000) Aromatic Interactions in Homeodomains Contribute to the Low Quantum Yield of a Conserved, Buried Tryptophan Vikas Nanda* and Ludwig Brand Department of Biology, Johns Hopkins University, Baltimore, Maryland ABSTRACT Trp 48, a conserved, buried residue commonly found in the hydrophobic core of homeodomains, has an unusually low fluorescence quantum yield. Chemical denaturation of Drosophila homeodomains Engrailed and Antennapedia(C39S) result in a four-fold increase in quantum yield, while unfolding of Ultrabithorax causes a twenty-fold enhancement. Global analysis of time-resolved fluorescence decay monitored at multiple emission wavelengths reveals sub-nanosecond lifetime components which dominate the overall intensity. Based on structure and sequence analysis of several homeodomains, we deduce that quenching is due to a transient, excited-state NH . . . hydrogen bond involving Trp 48 and a conserved aromatic residue at position 8. Additionally, both time-resolved fluorescence of indole-benzene mixtures and an electrostatic model of the proposed tryptophan-aromatic interaction substantiate different aspects of this mechanism. A survey of the Protein Data Bank reveals many proteins with tryptophanaromatic pairs where the indole nitrogen participates in a NH . . . hydrogen bond with the ring of another aromatic residue. Chemical denaturation of one protein found in this survey, human fibronectin type III module 10, causes an enhancement of the fluorescence quantum yield. This unique interaction has implications for many other systems and may be useful for studying larger, multi-tryptophan containing proteins. Proteins 2000;40:112–125. © 2000 Wiley-Liss, Inc. Key words: NH . . . hydrogen bond; fluorescence; protein unfolding; fibronectin type III; quenching INTRODUCTION Homeotic proteins are transcription factors that control body patterning of an organism during development. They contain a ⬃60 amino acid long element, the homeodomain (HD), which confers DNA binding specificity. One of the striking features of this class of proteins is the extensive sequence and structural homology found among diverse, distantly related members. Structures of Drosophila HDs such as Antp,1 Ubx,2 Eng,3 the yeast HDs Mata1 and Mat␣2,4 the murine Msx-1,5 and several other members of the HD class all show a similar helix-turn-helix fold. All of these proteins have a conserved tryptophan on the DNA recognition helix (helix III).6,7 For many of these HDs, this tryptophan is the only one present, making interpretation © 2000 WILEY-LISS, INC. of its fluorescence considerably more straightforward. As a result, we have an unique opportunity to look at several similar environments around the buried tryptophan and potentially make structurally accurate interpretations of the fluorescence data. The conserved Trp 48 of HDs from a variety of organims is characterized by a low fluorescence quantum yield (⌽F). The single tryptophan RNA-binding Bicoid from Drosophila shows a twenty fold enhancement in signal upon unfolding by chemical denaturation.8 Mat␣2 from yeast has two tryptophans, both on the recognition helix: one buried in the globular core and the other exposed to solvent. Heating Mat␣2 to the melting temperature of 56°C causes a doubling of intensity and a red-shift in the peak emission wavelength (Carra and Privalov, unpublished data). Chemical denaturation of both wild-type Msx-1 from mouse and mutant version of Msx-1, where a large number of nonconsensus residues are changed to alanine, causes a significant signal enhancement and a red-shift of the emission spectrum of the single tryptophan.9 The human HD Pbx-1 which belongs to the TALE HD class10 contains a single tryptophan with an extremely low quantum yield and a peak emission of 310 nm.11 Herein, we demonstrate a similar phenomenon in three Drosophila HDs: Ubx, Eng, and Antp(C39S), where chemical denaturation results in an enhancement of the ⌽F and a red-shift of peak emission wavelength.57 Direct structural data is available for all of these proteins except Bicoid and Msx-1, although a homology-based structure has been constructed for Msx-1.5 One may attribute an intensity enhancement upon unfolding to specific quenching interactions with polar side chains or backbone groups in the folded protein.12–14 A closer examination of the side-chain environment around Trp 48 in these structures may suggest a mechanism for the low ⌽F in HDs. The peptide backbone has been shown to quench tryptophan in aqueous solution, but it is not known whether similar effects are observed inside a protein.12 Studies of Abbreviations: HD, homeodomain; Ubx, Ultrabithorax; Eng, Engrailed; Antp, Antennapedia; FnIII, fibronectin type III; ⌽F, fluorescence quantum yield; DAS, decay-associated spectra; TRES, timeresolved emission spectra; PDB, protein data bank; GnHCl, guanidine hydrochloride. Grant sponsor: National Science Foundation; Grant number: MCB9810812. *Correspondence to: Vikas Nanda, Department of Biology, Johns Hopkins University, Baltimore, Maryland 21218. E-mail: email@example.com Received 12 October 1999; Accepted 18 February 2000 113 FLUORESCENCE OF HOMEODOMAINS specific quenching mechanisms of amino acid functional groups have shown that lysine and tyrosine quench by excited-state proton transfer while glutamine, asparagine, glutamic acid, aspartic acid, cysteine and histidine quench by excited-state electron transfer.13 Disulfides in the proximity of tryptophan can be a strong quencher.15 However, these same studies show that most aliphatic residues and phenylalanine have negligible effects on tryptophan emission at physiological pH. Aromatic and aliphatic residues, including tryptophan, extensively populate the hydrophobic core of most globular proteins. Due to the absence of quenching residues, buried tryptophans generally have a high ⌽F. The steady state and time-resolved fluorescence of several Drosophila homeodomains in both the native and denatured state are analyzed and compared for spectroscopic insight into the nature of the side-chain environment around Trp 48. The unusual fluorescence behavior of the HDs is attributed to a NH . . . hydrogen bond (also known as an atypical or unconventional hydrogen bond) exciplex between the indole nitrogen of Trp 48 and the center of the aromatic ring of residue 8. This is supplemented with measurements done on indole-benzene mixtures and electrostatic calculations of a tryptophanphenylalanine model interaction. Finally, we survey the PDB for other proteins with similar interactions in order to uncover systems where tryptophan may be quenched due to the formation of such a complex. MATERIALS AND METHODS Materials Ubx HD plasmid was obtained from Tom Tulius and protein was purified according to the protocol outlined in Ekker et al.16 Eng protein was obtained from Neil Clarke and Antp(C39S) was obtained from Peter Privalov. Ubx and Eng were stored in phosphate buffer (50 mM K2HPO4/ KH2PO4 pH 7.5, 250 mM NaCl, 1 mM dithiothreitol) and Antp(C39S) was stored in glycine buffer (20 mM glycineHCl pH 5.0, 100 mM NaCl). Human fibronectin Type III module 10 was provided by Kenneth Ingham. Steady State Spectroscopy All steady-state excitation and emission data was measured using a SLM-48000 spectrofluorometer. The excitation and emission slit widths were 4 nm. For emission spectra, the emission polarizer was vertical and the excitation polarizer was set to the magic angle. Emission spectra were corrected for instrumental sensitivity using correction factors obtained with a rhodamine quantum counter. Temperature in the cuvette was maintained at 25°C using a Neslab temperature controlled water bath. To remove scattering aggregates, protein was clarified with a low protein binding 0.22 micron filter (Millipore Corp., Bedford MA) prior to measurements. Protein concentrations were kept at ⬃10 M and low light intensities were used to reduce effects of photobleaching.17 All buffers were measured independently and found to be free of contaminating fluorescence. Time-Resolved Fluorescence Spectroscopy Time-correlated single photon counting measurements were performed with a picosecond dye laser pumped by a mode-locked yttrium/aluminum-garnet laser (Spectra Physics 3000 series) as the light source. Dye laser output was frequency doubled to 289 or 296 nm to excite tryptophan. Neutral density filters were placed in the excitation path to attenuate the excitation beam and reduce photobleaching. Emission slits were 1 nm and polarizers were set to magic angle. The fluorescence emission was monitored at a series of wavelengths between 295 and 450 nm in 5 nm intervals using a single-photon-counting apparatus with time-resolution of ⬃60 ps. A Ludox scattering solution was used to measure the instrument response. The series of decay curves at multiple emission wavelengths were globally fit using the method of least squares to four discrete exponentials18 with the pre-exponential factors varying as a function of wavelength (Eq. 1). I is intensity, em is emission wavelength, t is time, ␣ is the pre-exponential factor or amplitude and is the lifetime. I共t, em兲⫽ 冘 ␣i共em兲exp共⫺t/i兲 (1) i Wavelength-averaged amplitude components for each lifetime were calculated using the single wavelength amplitudes from the global fit (Eq. 2): 具␣i典⫽ 冘 ␣i共em兲/关 em 冘冘 i ␣i共em兲兴 (2) em The amplitude weighted average lifetimes therefore is simply (Eq. 3): 具典⫽ 冘 具␣i典i (3) i Intensity weighted DAS were generated by plotting the relative contribution of each lifetime component to the emission intensity at each of the wavelengths monitored (Eq. 4).19 –21 Ii共em兲⫽␣i共em兲i (4) TRES were generated by fitting the series of emission intensities monitored at successive wavelengths at fixed time intervals after excitation. Model Studies of Indole in Benzene Steady state and time-resolved measurements of 50 M indole (Aldrich Chemical Co., Milwaukee, WI, 99⫹% pure) solutions in cyclohexane (Sigma Chemical Company, St. Louis, MO) were measured as a function of added benzene (Sigma). Benzene and cyclohexane were measured under identical instrument settings to test for background fluorescence contamination, which was not found. We selected 289 nm as the excitation wavelength because it excites into a major absorption band of indole, and because absorbance of benzene at 289 is small enough that additions of benzene to indole in cyclohexane did not cause significant inner filter artifacts. Emission was monitored at 300 and 310 nm. The bimolecular quenching constant Ksv for the 114 V. NANDA AND L. BRAND benzene-indole interaction was determined by fitting the single exponential fit lifetime data to the Stern-Volmer equation (Eq. 5).22 0 is the lifetime at zero benzene concentration. 0/⫽Ksv 关benzene兴⫹1 (5) 1 NR⫽cenar⫺NTrp (9c) ⫽acos关共W䡠A兲/共兩W兩兩A兩兲兴 (10a) 2 1 2 1 W⫽共CTrp ⫺C␦Trp 兲⫻共CTrp ⫺NTrp 兲 (10b) 1 1 ⫺C␥ar兲⫻共C␦ar ⫺Car兲 A⫽共C␦ar (10c) Time-resolved fluorescence decay of indole dissolved in 100% benzene excited at 289 nm was monitored at multiple emission wavelengths between 290 and 400 nm in 5 nm intervals. This data was globally analyzed to generate DAS and TRES. Interactions were recorded if the distance between the indole nitrogen and the center of the ring was less than or equal to 6.00 Å. The survey includes all structures deposited in the PDB prior to June, 1999. Hydrogen Bond Energetics RESULTS HD Steady State Fluorescence The potential functions used for modeling the NH . . . hydrogen bond are the sum of the van der Waals interaction and the electrostatic interaction between the NH and the atoms of the aromatic group.23 The van der Waals interaction is modeled by a 12-6 Lennard-Jones function (Eq. 6). UvdW⫽ε关共r0/r兲12⫺2共r0/r兲6兴 (6) Here, r is the distance between the two atoms, r0 is the distance at the minimum energy of the van der Waals potential and ε is an energy parameter that depends on atom type. The pairwise parameter values used are: r0(HH) ⫽ 2.852 Å, ε(HH) ⫽ 0.038 kcal/mol, r0(CH) ⫽ 3.469 Å, ε(CH) ⫽ 0.038 kcal/mol, r0(NH) ⫽ 3.299 Å, ε(NH) ⫽ 0.125 kcal/mol, r0(NC) ⫽ 4.014 Å, ε(NC) ⫽ 0.125 kcal/mol. The electrostatic interaction is given by a Coulombic term (Eq. 7). Uel⫽332䡠qiqj/r (7) Partial charges for the indole were taken from semiempirical calculations on 3-methylindole in the ground and excited-state.24 Charge assignments for the Phe side chain were taken from studies of Phe–water interactions.25 The total energy is the sum of all pairwise electrostatic and van der Waals interactions. Structural Survey of the PDB The C program AROMATIC was designed to examine PDB26 files for interactions between the indole nitrogen and the center of an aromatic residue (only Tyr or Phe). Figure 1 shows the three geometric constraints considered. The following are the calculations used to obtain d, and (Eqs. 8 –10 vectors in bold, ‘ar’ subscript denotes vector for the aromatic acceptor): 1 d⫽兩NTrp ⫺cenar兩 (8a) cenar⫽共C␥ar⫹Car兲/2 (8b) ⫽acos关共PN䡠NR兲/共兩PN兩兩NR兩兲兴 (9a) 1 PN⫽P⫺NTrp (9bi) 2 1 ⫹C␦Trp 兲/2 P⫽共CTrp (9bii) 共pseudoatom in same direction as Hε1兲 Two single tryptophan-containing HDs, Eng, and Ubx, show a significant enhancement in intensity upon loss of native structure. Ubx has a twenty-fold increase in 8.0 M urea and a 20 nm red-shift in peak emission wavelength from 335 nm to 355 nm (Fig. 2A). Eng shows a four-fold enhancement and a similar red-shift upon unfolding (Fig. 2B). Antp(C39S), which contains two tryptophans, shows a four-fold increase and a 5 nm red-shift with GnHCl unfolding (Fig. 2C). HD Time-Resolved Fluorescence Table I summarizes the global analysis of Ubx, Eng, and Antp(C39S) fluorescence decay. Tryptophan has a strong absorption band at 289 nm, making this generally a good excitation wavelength to use. Additionally, measurements are done at 295 nm in order to minimize tyrosine fluorescence contamination. The spectroscopic contributions of tyrosine are small, and do not significantly affect the lifetime distribution. In the native state, Ubx fluorescence decay is dominated by two sub-nanosecond lifetime components which account for 82–90% of the excited-state fluorophore population (which is proportional to the preexponential factor, ␣, if the radiative decay rate of tryptophan is assumed to be constant for all species, generally a fair assumption for proteins).27 The two longer lifetimes account for only 10 –18% of the total population, but contribute significantly to the intensity (which is proportional to the product of ␣ and ). Unfolding of Ubx alters the lifetime distributions dramatically such that the dominant lifetime is ⬃3.5 ns. Similar behavior is observed for both Eng and Antp(C39S). Intensity-weighted DAS of Eng in the native and denatured state show how the relative contributions of the different lifetime components change as a function of wavelength (Fig. 3A). In the native protein, the two sub-nanosecond lifetimes have blue-shifted associated spectra around 330 nm, while the longer lifetimes have peak emission around 350 nm. In the denatured protein, none of the lifetime components have a peak emission shorter than 350 nm, consistent with the signature of a solventexposed tryptophan in the denatured state. It is obvious that the two shortest lifetimes account for a significant fraction of the overall emission intensity in native Eng, while the 3.20 ns lifetime accounts for the majority of the fluorescence intensity in unfolded protein. The DAS of Ubx FLUORESCENCE OF HOMEODOMAINS 115 Fig. 1. Orientation parameters for Trparomatic interactions: Three geometric constraints used to characterize the interaction between tryptophan and adjacent aromatic side chains. (A) d, distance, between the indole nitrogen and the geometric center of the aromatic ring. (B) , angle of projection, between the direction of the indole nitrogen to its proton and the indole nitrogen to the ring center. (C) , interplanar dihedral angle, between the planes of tryptophan and another aromatic residue. (data not shown) have similar lifetime distributions as Eng for both the native and denatured state. Antp(C39S), with a second exposed tryptophan at position 56, yields different DAS (Fig. 3B). Both native and unfolded protein DAS are dominated by longer lifetime species. Still, sub-nanosecond lifetimes contribute more to native state fluorescence than to unfolded protein fluorescence. Time-resolved emission spectra (TRES) are effectively snapshots of the emission spectrum at various times after excitation. TRES that change over time are indicative of relaxation of the environment during the lifetime of the excited-state, or of discrete excited-state heterogeneity, or of both phenomenon simultaneously. The TRES of Eng (Fig. 4) are representative of what we also see in Antp(C39S) and Ubx. The native state TRES show a significant time dependence for peak emission wavelength. It is difficult to conclude whether this is due to relaxation or to heterogeneity because the native protein has a very low ⌽. Even a minute fraction of unfolded protein could make TRES evolve towards longer wavelengths. The TRES of denatured Eng show little change in spectral shape with time, which is consistent with a highly exposed residue, where water relaxes around tryptophan in times faster than the resolution of the lifetime instrument. Time-Resolved Fluorescence of Indole-Benzene Mixtures It has been demonstrated in the chemical literature that indole fluorescence is quenched in the presence of benzene.28 Measuring the emission decay kinetics of indole in cyclohexane as a function of benzene concentration allows us to determine whether quenching is due to ground or excited-state interactions. Previous studies describe a biexponential decay in the presence of benzene, reporting that benzene emission adds an additional component to the signal.29 We find that the decay of indole is monoexponential up to 10.0 M benzene in cyclohexane with excellent 2 values. Despite this discrepancy, the results from the two studies are comparable. Our data gives a Ksv of 1.32 and shows that an exciplex is formed between the indole nitrogen and the benzene ring.29 It should be noted that the time-resolved emission decay of indole in 11.22 M (100%) benzene is not monoexponential, and in fact requires four lifetimes to fit the data. This may be due to the presence of weaker van der Waals interactions of indole with a second benzene that only occurs at very high 116 V. NANDA AND L. BRAND Fig. 2. Steady state fluorescence emission spectra of several HDs: Relative emission spectra (left) and normalized spectra (right) included for (A) Ubx in storage buffer and storage buffer ⫹ 8 M urea, (B) Eng in storage buffer and storage buffer ⫹ 5.4 M GnHCl, (C) Antp(C39S) in storage buffer and storage buffer ⫹ 5.4 M GnHCl. N denotes native protein spectra and D denotes spectra of protein in denaturant. Excitation wavelength is 295 nm for all of these measurements. concentrations.30,31 There are some similarities of the indole-benzene DAS and TRES with those of HDs (Fig. 5). The dominant emission component in the DAS is a 0.47 ns lifetime. The TRES evolve during the lifetime of the excited-state with a broad spectral shift, although the abruptness of the shift is more indicative of heterogeneity than relaxation. Electrostatic Model of the NH . . . Interaction Although recent experimental studies have directly observed the covalent character of the hydrogen bond in proteins,32,33 the attractive forces that induce a hydrogen bond to form can be well described by electrostatic modeling of the interaction. Generally, carbon is not a good hydrogen bond acceptor. However, in an aromatic ring, the relative geometry of the six partially negative carbons forms a composite favorable charge that can participate in a hydrogen bond with a proton donor such as nitrogen. An atomic charge approximation which ignores the presence of delocalized electron clouds and treats benzene as an array of C-H dipoles has been successfully applied to modeling the NH-benzene interaction, 23 and phenylalanine-phenylalanine interactions.34 Upon excitation of tryptophan, the dipole moment increases substantially and changes direction, which is reflected by a change in the partial charge distribution on the atoms of indole.24 Quantum chemical calculations of the ground and first singlet excited-state show that indole nitrogen goes from a partial negative charge to a partial positive charge upon excitation. If tryptophan is involved in a NH . . . interaction, making the nitrogen positive should strengthen the electrostatic interaction with the carbons of benzene. Figure 6 shows the energy of a NH . . . hydrogen bond where the charges on the indole are those assigned for 3-methylindole in the ground (S0) 117 FLUORESCENCE OF HOMEODOMAINS TABLE I. Four Exponential Global Fit to Native and Denatured Homeodomains at Two Excitation Wavelengths† Native Denatured 289 nm Ubx 0.11 0.68 2.49 6.01 295 nm ␣ 0.69 0.21 0.08 0.02 0.18 0.80 2.72 6.89 具典 ⫽ 0.54 2 ⫽ 1.261 289 nm ␣ 0.55 0.27 0.16 0.02 0.38 1.45 3.67 9.16 具典 ⫽ 0.89 2 ⫽ 1.292 295 nm ␣ 0.20 0.32 0.47 0.01 0.28 1.26 3.57 7.56 具典 ⫽ 2.35 2 ⫽ 1.053 ␣ 0.16 0.30 0.52 0.02 具典 ⫽ 2.43 2 ⫽ 1.089 Eng 289 nm 0.15 0.48 1.95 4.78 具典 ⫽ 0.62 2 ⫽ 1.042 Antp(C39S) 289 nm 0.14 0.84 2.52 4.63 具典 ⫽ 1.66 2 ⫽ 1.042 295 nm ␣ 0.52 0.36 0.09 0.04 0.17 0.49 2.14 5.07 289 nm ␣ 0.53 0.37 0.07 0.03 0.28 1.16 3.00 4.28 具典 ⫽ 0.57 2 ⫽ 1.030 0.09 0.91 2.83 5.64 0.28 1.16 3.20 4.91 具典 ⫽ 2.33 2 ⫽ 0.997 295 nm ␣ 0.25 0.27 0.39 0.09 295 nm ␣ 0.15 0.27 0.43 0.16 具典 ⫽ 2.26 2 ⫽ 1.007 289 nm ␣ 0.44 0.23 0.31 0.03 具典 ⫽ 1.30 2 ⫽ 1.012 0.29 1.25 3.02 4.61 ␣ 0.15 0.28 0.50 0.06 295 nm ␣ 0.14 0.28 0.47 0.10 具典 ⫽ 2.27 2 ⫽ 0.998 0.07 0.68 2.30 4.04 ␣ 0.15 0.21 0.38 0.25 具典 ⫽ 2.03 2 ⫽ 1.047 Lifetimes are globally fixed across all emission wavelengths. ␣s are summed over all wavelengths and normalized. 具典 is ␣ weighted: ¥ ␣ii/¥ ␣i 䡠 2 values are for the global analysis. † and first singlet excited-state (1La).24 The energy of the hydrogen bond becomes more favorable in the excitedstate.35 It’s also interesting to note that the equilibrium distance from the nitrogen to the center of the aromatic ring decreases from 3.78 Å in the S0 state to 3.65 Å in the 1 La state. DISCUSSION The low quantum yield of Trp 48 is a prevalent phenomenon in HDs. The challenge of this study has been to develop a model of quenching that is consistent with past and present observations in the context of known structures. The steady-state fluorescence spectra of the three HDs included in this study and other work done on Mat␣2, Msx-1, and Bicoid show a red-shift upon unfolding consistent with the transport of Trp 48 from a highly apolar environment into a polar solvent. The large intensity increase is observed in each of the HDs we have surveyed, suggesting that some common native structure interaction quenches Trp 48. The magnitude of the intensity change varies for different HDs. In summary, Bicoid8 and Ubx show the largest change (⬃20 fold), while Msx-19 and Eng show a 4 –5 fold enhancement. Antp(C39S) shows a similar change, and the increase attributed to Trp 48 is believed to be the dominant one. Trp 56 of Antp(C39S) is more exposed and should only increase slightly in the presence of denaturant.* The doubling of intensity observed during Mat␣2 melting may understate the enhancement of Trp 48 for two reasons: first, Mat␣2 also has two tryptophans in similar environments to Antp so the change in Trp 48 may be partially masked by Trp 56, and second, tryptophan fluorescence decreases with increasing temperature, so presumably, chemical denaturation of Mat␣2 would show an even greater change in signal intensity. By looking at the details of the side-chain environment in many of these homeodomains, we can attempt to infer what type of interactions are responsible for reducing the Trp 48 ⌽F. Structural Analysis of HDs The crystal and NMR structures of Antp, Eng, Ubx, and Mat␣2 are scrutinized for details on the side-chain environment of Trp 48. Any heavy atoms within 6.00 Å of the tryptophan Cε2 as determined using the NEIGHBORS function of Insight 95.036 are considered for possible interaction with the indole fluorophore (Fig. 7). Due to the high degree of sequence and structural homology between HDs, residues 8, 13, 16, 17, 49, and 52 are always seen in contact with Trp 48. Which of these interactions could give rise to the characteristic low ⌽F of HDs? Most contacts *Increasing [urea] or [GnHCl] generally causes a small enhancement and no detectable spectral shift when unaccompanied by a structural change. 118 V. NANDA AND L. BRAND Fig. 3. Intensity weighted DAS of HDs excited at 295 nm using four global lifetimes: (A) Native (left) and denatured (right) Eng. Lifetime values in Table I. (B) Native (left) and denatured (right) Antp(C39S). Lifetime values in Table I. Fig. 4. TRES of Eng excited at 289 nm: in the native state (left) and denatured state (right). Emission spectra at six time points are shown: 0, 0.1, 0.3, 1.0, 3.0, and 10.0 ns after excitation. Spectra evolve towards longer wavelengths with increasing time. observed are with aromatic ring groups, aliphatic sidechains, or the alkyl chains of polar residues. The exception is in the crystal structure of Eng where several terminal amines are found in proximity to Trp 48. N1 and N2 of Arg 5 and the N of Lys 52 and Lys 55 are potentially very good quenchers of tryptophan fluorescence. Additionally, bound water found in the HD/DNA interface may also reduce the Trp 48 ⌽F. However, neither bound water nor polar side-chain amine interactions are consistently found for all the structures investigated and therefore would not provide a general explanation of quenching in HDs. Additionally, quenching by solvent is not sufficient to account for the magnitude of the quantum yield reduction.37 We therefore suggest that the primary mechanism of quenching is by interactions with the aromatic side-chains of adjacent phenylalanines or tyrosines. The four aromatic residues in the core of Eng (Phe 8, Phe 20, Trp 48, and Phe 49) pack in a “herringbone” FLUORESCENCE OF HOMEODOMAINS 119 Fig. 5. Indole in 100% benzene excited at 289 nm using four global lifetimes: (A) Intensity weighted DAS: 1 ⫽ 0.47 ns, 2 ⫽ 2.98 ns, 3 ⫽ 1.67 ns, 4 ⫽ 11.51 ns. (B) TRES: Emission spectra at six time points are shown: 0, 0.1, 0.3, 1.0, 3.0, and 10.0 ns after excitation. Spectra evolve towards longer wavelengths with increasing time. Fig. 6. Energy of a Trp-Phe NH . . . hydrogen bond: Interaction energy as a function of the distance between the indole nitrogen and the center of the aromatic ring. conformation.3 Of these, residue 8 is either a phenylalanine or a tyrosine for the HDs included in this study, and is aromatic in 80% of known HDs.6 Residue 49 is always phenylalanine except in the case of Mat␣2, where it is a valine. Residue 20 is always phenylalanine, but does not make contact with Trp 48. Tyrosine has been found to quench tryptophan through excited-state proton transfer,13 but the same study finds that phenylalanine generally does not affect tryptophan ⌽ F . If a universal mechanism for the ⌽ F of HDs does in fact exist, it presumably involves residue 8, which is always in contact with Trp 48 and is always aromatic. Additionally, this mechanism must explain the interchangeability of phenylalanine and tyrosine at position 8. Evidence for just such a mechanism consistent with both fluorescence and structural data comes from another type of protein which also has a low ⌽ F . Steady state measurements of the immunophilin FKBP12 find Trp 59 has a ⌽F of 0.014,38 which is a factor of ten lower than free tryptophan.39 Urea-induced unfolding of human FK506 binding protein shows a 15-fold enhancement.40 NMR characterization of the FKBP59 immunophilin, which is homologous to FKBP12, demonstrates the presence of a hydrogen bond between the indole nitrogen of Trp 89 and the center of the aromatic ring of Phe 129.41 The two rings are oriented perpendicularly, with the indole nitrogen proton pointing directly towards the center of the phenylalanine ring. Time-resolved fluorescence measurements on FKBP12 further suggests that the presence of this hydrogen bond correlates with the weak fluorescence of Trp 59.42 We suggest that a similar interaction between Trp 48 and residue 8 may be responsible for the low ⌽F observed in HDs. The hypothesis that fluorescence is quenched by an atypical hydrogen bond from Trp 48 to residue 8 is predominantly based on elimination of other known mechanisms of tryptophan quenching. Assuming that the mechanism is a general one for the entire class of proteins, driven by a common interaction, we can eliminate quenching by polar residues, cysteine, histidine, water, or backbone amides. However, if a tryptophanaromatic pair is indeed responsible for Trp 48 low ⌽ F , then why do X-Ray and NMR-derived structures not show clear evidence of an NH . . . interaction? The majority of structures show Trp 48 in an orientation where the C␦ 1 is pointing to the ring instead of the nitrogen. It could be argued that most of the structures available are DNA/protein co-crystals, which influences the tryptophan orientation. This would imply that adding specific DNA to a HD should cause an increase in intensity similar to protein unfolding, but we have not been able to observe this. Additionally, structures of protein in the absence of DNA have moderately improved tryptophan geometries for hydrogen bond formation, but these still diverge significantly from tryptophan–aromatic interactions such as those seen in immunophilins. Therefore, DNA binding-induced structural changes do not seem to be an issue. One possibility is that excitation of Trp 48 facilitates an excited-state rearrangement, allowing hydrogen bond for- Fig. 7. Side chain environment around Trp 48: (A) Eng–1ENH, (B) Antp/DNA–9ANT, (C) Ubx/Exd/DNA–1B8I and (D) Mat␣2/Mata1/DNA–1AKH. Trp 48 and residue 8 (Trp 179 and Phe 136 in 1AKH) are highlighted. Individual crosshatches represent coordinates of bound water oxygen atoms. Graphics generated using Insight 95.0.34 FLUORESCENCE OF HOMEODOMAINS 121 Fig. 8. Excited state rearrangement model: This cartoon depicts the rotation believed to be required for formation of the hydrogen bond in HDs. Through graphical analysis of available structures, a 70 – 80° rotation in the direction shown is sufficient to bring the indole nitrogen in proximity to the aromatic acceptor. The rotation along the C␣-C␤ bond may either be a result of picosecond fluctuations in structure or directed electrostatically driven motion of the side chain in the indole plane as a result of its change in dipole moment. mation and quenching. An in-plane rotation of the indole side chain would allow the necessary interactions to occur (Fig. 8). There are clues in the fluorescence behavior that substantiate this mechanism. tophans undergoing an excited-state rotation and hydrogen bond formation as indicated in the cartoon (Fig. 8). There are two possible reasons for an excited-state rearrangement to occur. It is known that local side-chain fluctuations occur on the picosecond time-scale.43 It may be that Trp 48 fluctuates rotationally in its plane during the lifetime of the excited-state. If one of these motions brings the indole NH close enough to the aromatic acceptor, our electrostatic model suggests that there will be a strong attraction between the residues resulting in hydrogen bond formation and subsequent de-excitation. The second mechanism postulated is an induced excited-state rearrangement based on the change in tryptophan dipole moment upon excitation. The dipole moment of the indole ring changes magnitude from 2.1 to 5 D upon excitation and changes in direction.44 This change in dipole moment may induce torque on the indole side chain due to the local electric field which is relaxed by the in-plane rotation. Such a directed process would make quenching more efficient and potentially account for the low ⌽F observed in HDs. This would also explain TRES of HDs in the native state, where the rotational relaxation of transition energy (E 䡠 ⌬) is reflected in a red-shifting emission maximum with time. The indole-benzene complex behaves as a pure dynamic quenching system where both 具典 and ⌽F decrease six fold in the presence of benzene. The low Ksv of benzene quenching of indole suggests that the efficiency of this interaction is low. This argues strongly for quenching caused only by collisions where both indole and benzene need to be in a particular relative orientation. All other things being equal, quenchers with spherical symmetry such as iodide would therefore be relatively more efficient since there would be fewer orientation constraints on the quenching interaction. The bimodal nature of the indole- Excited State Rearrangement and Quenching The time-resolved fluorescence behavior of HDs brings up a number of issues. One notable discrepancy is that for Ubx and Antp(C39S): native ⌽native ⫽ denatured ⌽denatured (11) This is often indicative of a mixture of static and dynamic quenching mechanisms.22 If the ratios are equal, the mechanism can be considered purely dynamic, and if native ⫽ denatured then the mechanism can be described as purely static. However, the decay of both native and denatured homeodomains are multi-exponential, making it difficult to assign meaningful values for native and denatured. One may summarize the lifetime parameters by describing each state with an average lifetime 具典, but this can be misleading. The intensity weighted 具典 will depend strongly on tiny fractional contributions from longer lifetimes. A longer lifetime from a minute fraction of unfolded or misfolded protein in a native state ensemble could easily increase the 具典 significantly. The ␣ weighted 具典 will depend strongly on the resolution of the instrument, and the fitted pre-exponential factors for short lifetimes may be inaccurate. Assuming the time resolution of our instrument is adequate to resolve amplitudes for the short lifetime components, it suggests that Trp 48 is involved in both ground and excited-state quenching processes. The ground-state component may be a fraction of tryptophan already forming an atypical hydrogen bond with residue 8 and the excited-state component would represent those tryp- 122 V. NANDA AND L. BRAND TABLE II. Several Proteins with Structural Evidence for NH . . . Hydrogen Bonds† PDB id Protein Structure type Residues dÅ ° ° 1lqi 1a88 1axi 1xtc 1epj 1fgp 1iln 1lis 1mfn 1pls 1waj 3cp4 1rlr 1vrt 1igt 1pcr 1oya 2puc 1luc 2fke 1hom 9ant 1b8i 1akh 1enh 3hdd insecticidal ␣ scorpion toxin chloroperoxidase L HGH and receptor cholera toxin epidermal growth factor phage coat protein interleukin-2 red abalone lysin mouse fibronectin pleckstrin homology domain bacteriophage DNA polymerase cytochrome P450CAM ribonucleotide reductase R1 HIV-1 reverse transcriptase immunoglobulin photosynthetic reaction center old yellow enzyme purine repressor PurR w/DNA bacterial luciferase human FKBP12 Antp Antp/DNA Ubx/Exd/DNA Mat␣2/Mata1/DNA Eng Eng/DNA NMR/Md11 Xray/1.9 Å Xray/2.1 Å Xray/2.4 Å NMR/Md 4 NMR/Md 3 theoretical Xray/1.9 Å NMR/Md 3 NMR/Md 3 Xray/2.8 Å Xray/2.3 Å Xray/2.5 Å Xray/2.2 Å Xray/2.8 Å Xray/2.7 Å Xray/2.0 Å Xray/2.6 Å Xray/1.5 Å Xray/1.7 Å NMR/Md 4 Xray/2.4 Å Xray/2.4 Å Xray/2.5 Å Xray/2.1 Å Xray/2.8 Å W39-F18 W183-F165 W157-Y174 W88-Y12 W49-Y37 W37-Y32 W121-F117 W62-Y133 W22-Y32 W21-F62 W205-F113 W374-F381 W309-F368 W229-Y183 W47-Y35 W66-Y177 W116-Y196 W147-Y126 W194-Y42 W59-F99 W48-Y8 W48-Y8 W48-Y8 W179-F136 W48-F8 W48-F8 5.86 3.10 3.46 5.53 3.79 4.59 3.14 3.68 3.98 3.61 5.76 3.81 3.43 3.55 4.06 3.79 3.46 2.99 5.65 3.34 4.75 5.65 5.69 5.38 5.21 5.21 22 18 12 15 10 18 10 8 18 18 16 20 20 16 9 14 18 18 18 8 105 85 87 90 85 95 128 106 110 88 88 93 90 94 101 106 105 86 84 95 72 102 83 84 71 90 100 34 21 34 113 88 † X-ray crystal structures include the reported resolution under Structure Type. NMR structures are referred to by specific model number. Geometric parameters listed for four homeodomains. Information for both free and DNA bound forms included for Antp (9ANT, 1HOM) and Eng (3HDD, 1ENH). benzene TRES may be best described as a mixture of states: one being the atypical hydrogen bond, and the other a less specific, stacked van der Waals complex. MassAnalyzed Pulsed Field Threshold Ionization Spectroscopy measurements of 1:1 indole-benzene clusters in the gas phase show that the atypical hydrogen bond is the primary conformation.31 As an aside, it is interesting to note that the lifetime distributions of unfolded Antp(C39S) and Eng are different are very different (Fig. 3), indicating that the environments around Trp 48 and Trp 56 in the Antp(C39S) random coil are dissimilar. In Antp(C39S), Trp 48 is flanked by Ile 47 and Phe 49 while Trp 56 is between Lys 55 and Lys 57. In Eng, Trp 48 is flanked by Ile 47 and Phe 49. If one assumes the denatured state has no persistent structure, this observation suggests that the difference in denatured protein DAS signatures is due to possible electrostatic effects from the presence of the polar lysine side chains adjacent to Trp 56 of Antp(C39S). Surveying the PDB With the evidence mounting for an aromatic mediated tryptophan quenching in HDs, the question arises as to the generality of this phenomenon. By surveying the PDB for proteins with tryptophan-aromatic interactions, we compile a list of candidates that may potentially have a low tryptophan ⌽F. Three geometric parameters: distance (d), angle of projection () and dihedral angle () for interact- ing pairs of tryptophan and phenylalanine or tyrosine are measured for structures in the PDB (Fig. 1). Potentially interesting interactions would also include tryptophanaromatic histidine or tryptophan-tryptophan complexes, but these two cases are excluded in the present study. Only structures with orientations within 20° of ⫽ 90° and ⫽ 0° are kept. Table II lists several of the proteins obtained from this survey. In addition to the immunoglobulin structure, 1IGT mentioned in Table II, over twenty other structures of FAB fragments bound to various epitopes are also culled from the PDB. A large number of immunophilin mutants, homologues, protein-ligand complexes, etc., are also found in this survey. 2FKE is included in Table II as a representative of the interaction geometry in immunophilins. Several of these proteins have a low ⌽F and clearly show an enhancement upon denaturation. The same geometric parameters are listed for several HD structures. As previously mentioned, there is no direct observation of the hydrogen bond in the HD structures, which is reflected in the values for d, , and . IL-2 has a tryptophan-phenylalanine pair involved in a helical i, i ⫹ 4 interaction buried in the hydrophobic core. The only potential quenching group is a cysteine which is greater than 6.0 Å away from Trp 121.45 As Table II shows, the distance and orientation parameters are ideal for formation of an NH . . . hydrogen bond between Trp 121 and Phe 117. A crystal structure of human IL-2 (3INK) shows that the cysteine and a nearby histidine are not in 123 FLUORESCENCE OF HOMEODOMAINS Fig. 9. Denaturation of human FnIII-10 in GnHCl: Intensity change (solid line, crosses) is ⬃4 fold. The peak shift represented as a function of I350nm/I320nm (dashed line, open circles). bonds have roles in both protein structure and ligand binding interactions.50 –52 Mutational studies on binding site residues of the CAP/DNA complex show that introducing an aromatic residue which acts as a H-bond acceptor can contribute ⬃0.5 to 1 kcal/mol to binding.53 If the energy of the NH . . . hydrogen bond is near ⫺2.81 kcal/mol, as the physical model suggests, this would be significant enough to contribute to the energy of protein stabilization which is on the order of 5 to 10 kcal/mol.54 It is a more favorable interaction than typical nonbonding aromatic-aromatic interactions which are on the order of 1–2 kcal/mol.55 The omnipresence of specific tryptophans in FnIII and HDs and the conserved adjacent aromatics suggest that these edge-face interactions are significant to stability. It is interesting to note that in a sequence alignment of FnIII domains, the otherwise conserved Trp-aromatic interaction is absent in interleukin-7 receptor, but it is replaced by a histidine-phenylalanine pair,56 an interaction which has also been implicated in forming NH . . . hydrogen bonds in proteins.50 One might speculate that this interaction is conserved not only because of the energetic contribution, but also due to the specificity of packing that such an interaction entails. Orienting bulky aromatics in such a fashion may provide a scaffold for the remaining core. CONCLUSIONS the correct orientation, or in the vicinity to quench Trp 121. Chemical denaturation studies performed on wildtype human IL-2 shows a 50% increase in signal intensity upon unfolding and a large red-shift of the emission wavelength peak from 324 nm to ⬃350 nm indicating a removal of Trp 121 from a hydrophobic, but quenching environment.46 In the same study, a mutant with an additional exposed tryptophan also shows an enhancement and a red-shift of emission upon denaturation. Human FnIII Module 10 has a single tryptophan and an extremely low ⌽F in the native state. Addition of GnHCl to native human FnIII module 10 shows a four-fold enhancement and a significant red-shift in emission (Fig. 9). The emission spectrum of the native protein is dominated by tyrosine contributions. The same effect has been noted in studies of human FnIII Module 1.47 There is little structural evidence for other mechanisms of quenching. It is plausible that the atypical hydrogen bond between Trp 22 and Phe 32 in FnIII modules is responsible for tryptophan quenching. This mechanism is consistent with sequence conservation of interacting residues which would give a general description of the low ⌽F in fibronectins. Stability and Conservation The presence of stabilizing edge-face aromatic interactions may be the reason that Trp 22 is highly conserved in FnIII type folds and why Trp 48 is conserved in HDs. It is known from experimental studies and physical models that the energetics of the NH . . . bond is approximately half the energy of standard hydrogen bonds found in proteins.23,48 Generally, these hydrogen bonds lose in competition to more favorable side chain interactions.49 Still, numerous examples exist where NH . . . hydrogen The unusual fluorescence of Trp 48 in HDs cannot be satisfactorily explained by conventional quenching mechanisms. Structure and sequence analysis indicates that interactions of Trp 48 with adjacent aromatics may result in a low ⌽F. The mechanism presented at our current level of understanding is a transient NH . . . hydrogen bond between Trp 48 and the aromatic ring of residue 8 caused by an excited-state in plane rotation of the indole side chain. Several other proteins with direct structural evidence of an NH . . . hydrogen bond have a low ⌽F. A practical application of this phenomenon to many of the proteins found in the PDB survey is the potential for using tryptophan-aromatic pairs to study systems with multiple tryptophans. As was observed in the case of Antp(C39S), the sensitivity of total tryptophan emission to conformation is dominated by the enhancement observed in Trp 48. Even with proteins with a large number of tryptophans, a conformational change which can effectively switch a tryptophan “on or off” may cause an interpretable change in fluorescence. With the accessibility and advantages of using fluorescence to study conformation and interactions, knowledge that an atypical hydrogen bond involving tryptophan occurs in a particular system can be very useful. ACKNOWLEDGMENTS This work was only possible with the valuable input and expertise and the materials provided by several people. We thank Richard Frazee and Tom Tulius for providing the Ubx plasmid and teaching us the purification, Neil Clarke for the generous gift of Eng protein and valuable discussions on HD structure, Peter Privalov for providing Antp protein and communicating valuable information on Mat␣2 124 V. NANDA AND L. BRAND fluorescence, Zhenlan Li for additional help with the Antp system, Ken Ingham for the generous gift of human fibronectin, George Rose, Thomas Woolf, Bertrand GarciaMoreno, Hirsh Nanda, and Alan Grossfield for valuable discussions on working with the PDB and modeling the tryptophan-phenylalanine interaction, Dima Toptygin for training on the fluorescence instrumentation and valuable discussion on data analysis, Andy Russo and Michael Rodgers for valuable discussions on pertinent biochemical techniques and protein purification. REFERENCES 1. Qian Y, Billeter M, Otting G, Müller M, Gehring W, Wüthrich K. The structure of the Antennapedia homeodomain determined by NMR spectroscopy in solution: Comparison with prokaryotic repressors. Cell 1989;59:573–580. 2. Passner JM, Ryoo HD, Shen L, Mann RS, Aggarwal AK. Structure of a DNA-bound Ultrabithorax-Extradenticle homeodomain complex. Nature 1999;397:714 –719. 3. Clarke ND, Kissinger CR, Desjarlais J, Gilliland GL, Pabo CO. Structural studies of the engrailed homeodomain. Protein Sci 1994;3:1779 –1787. 4. Li T, Stark MR, Johnson AD, Wolberger C. Crystal structure of the MATa1/MAT␣2 homeodomain heterodimer bound to DNA. Science 1995;270:262–269. 5. Li H, Tejero R, Monleon D, et al. Homology modeling using simulated annealing of restrained molecular dynamics and conformational search calculations with CONGEN: application in predicting the three-dimensional structure of murine homeodomain MSX-1. Protein Sci 1997;6:956 –970. 6. Laughon A. DNA binding specificity of homeodomains. Biochemistry 1991;30:11358 –11367. 7. Gehring WJ, Affolter M, Bürglin T. Homeodomain proteins. Annu Rev Biochem 1994;63:487–526. 8. Subramaniam V, Rivera-Pomar R, Jovin TM. Su-Pos447: the conserved tryptophan in the recognition helix of the bicoid homeodomain is a sensitive probe of structure and folding. Biophys J 1999;76:A108. 9. Shang Z, Issac VE, Li H, et al. Design of a “minimAL” homeodomain: the N-terminal arm modulates DNA binding affinity and stabilizes homeodomain structure. Proc Natl Acad Sci 1994;94: 8373– 8377. 10. Sanchez M, Jennings PA, Murre C. Conformational changes induced in Hoxb-8/Pbx-1 heterodimers in solution and upon interaction with specific DNA. Mol Cell Bio 1997;17:5369 –5376. 11. Bürglin TR. Analysis of TALE superclass homeobox genes (MEIS, PBC, KNOX, Irogquois, TGIF) reveals a novel domain conserved between plants and animals. Nucl Acids Res 1997;25:4173– 4180. 12. Chen Y, Liu B, Yu H-T, Barkley MD. The peptide bond quenches indole fluorescence. J Am Chem Soc 1996;118:9271–9278. 13. Chen Y, Barkley MD. Towards understanding tryptophan fluorescence in proteins. Biochemistry 1998;37:9976 –9982. 14. Harris DL, Hudson BS. Photophysics of tryptophan in bacteriophage T4 lysozyme. Biochemistry 1990;29:5276 –5285. 15. Cowgill RW. Fluorescence and the structure of proteins. XVIII. Spatial requirements for quenching by disulfide groups. Biochim Biophys Acta 1970;207:556 –559. 16. Ekker SC, von Kessler DP, Beachy PA. Differential DNA sequence recognition is a determinant of specificity in homeotic gene action. EMBO J 1992;11:4059 – 4072. 17. Toptygin D, Rodgers ME, Brand L. Tu-Pos490: Non-exponential fluorescence decay of photobleached indole and tryptophan. Biophys J 1999;76:A361. 18. Grinvald A, Steinberg IZ. On the analysis of fluorescence decay kinetics by the method of least-squares. Anal Biochem 1974;59: 583–598. 19. Donzel B, Gauduchon P, Wahl P. Study of the conformation in the excited-state of two tryptophanyl diketopiperazines. J Am Chem Soc 1974;96:801– 808. 20. Knutson JR, Walbridge DG, Brand L. Decay-associated fluorescence spectra and the heterogeneous emission of alcohol dehydrogenase. Biochemistry 1982;21:4671– 4679. 21. Ross JBA, Schmidt CJ, Brand L. Time-resolved fluorescence of the two tryptophans in horse liver alcohol dehydrogenase. Biochemistry 1981;20:4369 – 4377. 22. Lakowicz JR. Principles of fluorescence spectroscopy. New York: Plenum Press; 1983. 23. Levitt M, Perutz MF. Aromatic rings act as hydrogen bond acceptors. J Mol Biol 1988;201:751–754. 24. Muiño PL, Harris D, Berryhill J, Hudson B, Callis PR. Simulation of solvent dynamics effects on the fluorescence of 3-methylindole in water. SPIE—Time-resolved laser spectroscopy in biochemistry III 1992;1640:240 –251. 25. Bolis G, Clementi E. Analytical potentials from “ab initio” computations for the interaction between biomolecules. 3. Reliability and transferability of the pair potentials. J Am Chem Soc 1977;99: 5550 –5557. 26. Bernstein FC, Koetzle TF, Williams GJB, et al. The protein data bank: a computer-based archival file for macromolecular structure. J Mol Biol 1977;112:535–542. 27. Strickler SJ, Berg RA. Relationship between absorption intensity and fluorescence lifetime of molecules. J Chem Phys 1962;37:814 – 822. 28. Van Duuren BL. Solvent effects in the fluorescence of indole and substituted indoles. J Org Chem 1961;26:2954 –2960. 29. Suwaiyan A, Klein KA. Picosecond study of solute-solvent interaction of the excited-state of indole. Chem Phys Lett 1989;159:244 – 250. 30. Hager J, Ivanco M, Smith MA, Wallace SC. Two-color threshold photoionization spectroscopy of jet-cooled indole clusters. Chem Phys 1986;105:397– 416. 31. Braun JE, Grebner ThL, Neusser HJ. van der Waals versus hydrogen-bonding in complexes of indole with argon, water and benzene by mass-analyzed pulsed field threshold ionization. J Phys Chem 1998;102:3273–3278. 32. Isaacs ED, Shukla A, Platzman PM, Hamann DR, Barbiellini B, Tulk CA. Covalency of the hydrogen bond in ice: a direct x-ray measurement. Phys Rev Lett 1999;82:600 – 603. 33. Wang Y-X, Jacob J, Cordier F, Wingfield P, Stahl SJ, Lee-Huang S, Torchia D, Grzesiek S, Bax A. Measurements of 3hJNC⬘ connectivities across hydrogen bonds in a 30 kDa protein. J Biomol NMR 1999;14:181–184. 34. Hunter CA, Singh J, Thornton JM. - interactions: the geometry and energetics of phenylalanine-phenylalanine interactions in proteins. J Mol Biol 1991;218:837– 846. 35. Pimentel GC. Hydrogen bonding and electronic transitions: the role of the Franck-Condon principle. J Am Chem Soc 1957;79:3323– 3326. 36. Insight95.0. San Diego: MSI/Biosym; 1995. 37. Gallay J, Vekshin N, Sopkova J, Vincent M. Evidences for picosecond excited state reactions of indole derivatives in alcoholic solvents, reverse micelles and proteins. SPIE—time-resolved laser spectroscopy in biochemistry III 1994;2137:390 –399. 38. Silva ND, Prendergast FG. Tryptophan dynamics of the FK506 binding protein: time-resolved fluorescence and simulations. Biophys J 1996;70:1122–1137. 39. Chen RF. Fluorescence quantum yields of tryptophan and tyrosine. Anal Lett 1967;1:35– 42. 40. Egan DA, Logan TM, Liang H, Mtayoshi E, Fesik SW, Holzman TF. Equilibrium denaturation of recombinant Human FK Binding Protein in Urea. Biochemistry 1993;32:1920 –1927. 41. Craescu CT, Rouvière N, Popescu A, Cerpolini E, Lebeau M-C, Baulieu E-E, Mispelter J. Three-dimensional structure of the immunophilin-like domain of FKBP59 in solution. Biochemistry 1996;34:11045–11052. 42. Rouvière N, Vincent M, Craescu CT, Gallay J. Immunosupressor binding to the immunophilin FKBP59 affects the local structure dynamics of a surface ␤-strand: time-resolved fluorescence study. Biochemistry 1997;36:7339 –7352. 43. McCammon JA, Wolynes PG, Karplus M. Picosecond dynamics of tyrosine side chains in proteins. Biochemistry 1979;18:927–942. 44. Pierce DW, Boxer SG. Stark effect spectroscopy of tryptophan. Biophys J 1995;68:1583–1591. 45. Dryden D, Weir MP. Evidence for an acid-induced molten-globule state in interleukin-2; a fluorescence and circular dichroism study. Biochim Biophys Acta 1991;1078:94 –100. 46. Weir MP, Chaplin MA, Wallace DM, Dykes CW, Hobden AN. Structure-activity relationships of recombinant human interleukin 2. Biochemistry 1988;27:6883– 6892. FLUORESCENCE OF HOMEODOMAINS 47. Litvinovich SV, Novokhatny VV, Brew SA, Ingham KC. Reversible unfolding of an isolated heparin and DNA binding fragment, the first type III module from fibronectin. Biochim Biophys Acta 1992;1119:57– 62. 48. Knee JL, Khundkar LR, Zewail AH. Picosecond photofragment spectroscopy. III. Vibrational predissociation of van der Waal’s clusters. J Chem Phys 1987;87:115–127. 49. Mitchell JBO, Nandi CL, McDonald IK, Thornton JM, Price SL. Amino/aromatic interactions in proteins: is the evidence stacked against hydrogen bonding? J Mol Biol 1994;239:315–331. 50. Armstrong KM, Fairman R, Baldwin RL. The (i, i ⫹ 4) Phe-His interaction studied in a alanine-based ␣-helix. J Mol Biol 1993;230: 284 –291. 51. Perutz MF. The role of aromatic rings as hydrogen-bond acceptors in molecular recognition. Phi Trans R Soc A 1993;345:105–112. 125 52. Burley SK, Petsko GA. Amino-aromatic interactions in proteins. FEBS Lett 1986;203:139 –143. 53. Parkinson G, Gunasekera A, Vojtechovsky J, Zhang X, Kunkel TA, Berman H, Ebright RH. Aromatic hydrogen bond in sequence specific DNA recognition. Nat Struct Biol 1996;3:837– 841. 54. Pace NC. Conformational stability of globular proteins. Trends Biochem Sci 1990;15:14 –17. 55. Burley SK, Petsko GA. Aromatic-aromatic interaction: a mechanism of protein structure stabilization. Science 1985;229:23–28. 56. Bazan JF. Structural design and molecular evolution of a cytokine receptor subfamily. Proc Natl Acad Sci 1990;87:6934 – 6938. 57. Nanda V, Brand L. WPos277 Fluorescence characterization of homeodomains: Quenching of a buried tryptophan by aromatic hydrogen bonding associations. Biophys J 1999;76:A448.