156
Biochimica et Biophysica Acta, 1160(1992) 156-162 © 1992Elsevier SciencePublishers B.V. All rights reserved 0167-4838/92/$05.00
BBAPRO 34318
Multinuclear NMR studies of the trp-repressor Jeremy N.S. Evans
a,b D e n n i s
N . A r v i d s o n c, R o b e r t P. G u n s a l u s a n d M a r y F. R o b e r t s d
c
a Department of Biochemistry / Biophysics, Washington State University, Pullman, WA (USA), b Department of Chemistry, Washington State University, Pullman, WA (USA), c Department of Microbiology, University of California, Los Angeles, CA (USA) and d Department of Chemistry, Boston College, Chestnut Hill, MA (USA)
(Received 18 March 1992)
Key words: Tryptophan-repressor;Protein-DNA complex;NMR; Protein-ligandcomplex The binding of the corepressor, L-tryptophan, to the Escherichia coli trp-aporepressor in solution has been examined by 13C-and ~9F-NMR spectroscopy. The binding of a number of tryptophan analogues have been studied by equilibrium dialysis. Evidence is presented that support the crystallographic studies (Schevitz, R.W., Otwinowski, Z., Joachimiak, A., Lawson, C.L. and Sigler, P.B. (1985) Nature 317, 782-786) that Val-58 is within the ring currents of the bound tryptophan and also close in space to the indole 5'-position, on the basis of heteronuclear 19F{1H}-NOE experiments. The tryptophan carboxylate is in hydrogen-bonding distance to a highly positively charged residue, probably Arg-54 and this bond strengthens on formation of the trp-repressor-DNA complex.
Introduction
The trp-repressor is involved in the transcriptional regulation of four operons in Escherichia coli which encode for enzymes in the L-tryptophan biosynthetic pathway, namely the t r p E D B A , trpR and a r o H operons and a m t r permease for L-tryptophan uptake from the medium [1-5]. Binding of the co-repressor, L-tryptophan, to the aporepressor results in the formation of the active aporepressor-corepressor complex, which subsequently binds tightly to the operators for each operon to repress their transcription by RNA polymerase. The trp-repressor is a small protein composed of two identical subunits (M r 12 356) [3]. Each dimer has two co-repressor binding sites that are identical and independent [6-8]. Furthermore, each L-tryptophan molecule is in proximity to amino-acid residues from both subunits of the aporepressor dimer [9,10]. Work from Sigler's laboratory [8,9] has refined the X-ray crystal structure of the trp-holorepressor and the indole propionic acid pseudorepressor complex [11] to 1.65 A. The structure suggests that the corepressor is bound rigidly in a hydrophobic binding pocket, with Correspondence to: J. Evans, Department of Biochemistry/ Biophysics, Washington State University,Pullman, WA, USA. Abbreviations: LW, linewidth; NOE, nuclear Overhauser effect; DQF, double quantum filtered; COSY, correlated spectroscopy.
the orientation of the indole ring well-defined (and in opposite senses in the two different complexes). However, it does not follow that the same holds in solution. Preliminary NMR studies in solution [6,12,13] suggest that there are differences between the two complexes and the assignment of intermolecular NOEs between repressor and co-repressor certainly imply that there is little dynamical motion in the co-repressor indole ring, but the orientation differs from that described by the crystal structure. More recent NMR work by Lane [14], by Jardetsky's laboratory [15,16] and by Roberts' laboratory [17] suggests that there is a significant change in the dynamics of the D- and E-helices of the protein in the presence of tryptophan. In the X-ray crystal structure of the trp-repressor-DNA complex proposed by Sigler and co-workers [18], it is suggested that the principal protein-base contacts are mediated through structured water. This structure, which has already made it into the textbooks of biochemistry [19], has been suggested to represent a non-specific complex which results from the crystallization conditions [20,21]. Indeed, it has been suggested that the trp-repressor may bind to the operator in a tandem fashion [22,23], as has been suggested for the metJ-repressor [24,25]. These criticisms have been addressed by Sigler's laboratory [26], in which it is argued that the crystalline protein-DNA complex does not change lattice form at the higher salt concentrations normally used for specific protein-DNA complex formation (up to 200 mM
157 NaCI) and the structure is consistent with the mutagenesis studies of Bass et al. [27]. In this paper, we examine the binding of tryptophan to the aporepressor from E. coli by solution-state 13C-and WF-NMR and the effect of addition of non-specific plasmid DNA. Materials and Methods
Chemicals. All chemicals were purchased from Sigma or Aldrich and were of the highest grade obtainable. L-Tryptophan was used without further purification. 4'-, 5'- and 6'-DL-fluorotryptophan were subjected to treatment with a Chelex column to remove paramagnetic impurities. Other tryptophan analogues were purified by HPLC by known methods [28]. [1-13C]-DLtryptophan was obtained from MSD Isotopes (Canada). [3-13C]-DL-tryptophan was a generous gift from Dr. N.E. Mackenzie (University of Arizona). Labelled tryptophans were purified further using Dowex 1 if required. [4',6',7'- ~H 3]-5'-DL-fluorotryptophan was synthesised from 5'-DL-fluorotryptophan using known methods [29]. Growth of cells. The trp-aporepressor-overproducing strain used in this work was E. coli LE392 containing the multiple copy trpR ÷ plasmid pRPG47 [3,5]. This plasmid, a pBR322 derivative, contains the trpR structural gene downstream from the E. coli lacUV5 promoter and ribosome binding site. Cell cultures were grown (at UCLA) with aeration to late logarithmic phase in a New Brunswick 200 1 fermenter in a medium containing 0.75% Bacto-Tryptone, 0.5% yeast extract, 0.5% lactose, 0.5% NaCl. The cells were harvested by centrifugation in a Sharples continuous centrifuge. Cells were stored for periods up to one year at -70°C. Protein purification. E. coli trp-aporepressor was purified by known methods [30] at UCLA. Purified protein was then shipped in 20% glycerol buffer (20 mM potassium phosphate buffer (pH 7.6)) on solid CO 2 to either MIT, Oxford, or WSU. Sample preparation. For NMR experiments, the aporepressor was concentrated, diluted and concentrated with 10 mM potassium phosphate buffer in 100% DaO, 100 mM KC1, 0.5 mM EDTA ((pH 7.6), except in Fig. 3 where the pH was 8.2), using a 3 ml Amicon stirred ultrafiltration cell and PM-10 membrane (M r cut-off 10 kDa) under about 75 psi of gaseous nitrogen pressure. The final volume of concentrated protein was usually 0.4 ml. Determination of analogue dissociation constants. This was carried out using a standard equilibrium dialysis competition assay [31], using L-[2-14C] tryptophan (0.2 /~M, final concentration of 53.5 mCi/mmol) and analogue at a variety of concentrations. NMR Spectroscopy. High-field Fourier transform (FT) NMR studies on the trp-repressor were performed on either a Bruker AM-250 (5.85 T, 250 MHz
1H, at Oxford), a Bruker AM-400 WB (9.39 T, 400 MHz XH, at Tufts University), a Bruker AM-500 (11.75 T, 500 MHz 1H, at Oxford) or a Varian VXR-500S (11.75 T, 500 MHz 1H at WSU) NMR spectrometer. Deuterium was used in all cases for locking the field. Chemical shifts were referenced externally to samples of similar dielectric constant: 13C-NMR spectra to dioxan in D20 buffer (6 c = 67.4 ppm); 19F-NMR spectra to trifluoroacetic acid in D20 (t~F 0 ppm). Sample temperature was maintained with the spectrometer variable temperature control unit, using boil-off liquid nitrogen. All samples containing trp-repressor were maintained at 298 K. Heteronuclear NMR was carried out with single frequency probes. Proton decoupling was achieved by gated broad band irradiation, using the Waltz-16 composite pulse sequence to minimize dielectric heating. Spectra were obtained with 32K data points and the FIDs Fourier transformed with line broadening (5 Hz for 19F or 10 Hz for ~3C). Results
Binding of tryptophan analogues As a prelude to understanding the interactions between tryptophan and trp-aporepressor, we have examined the ability of trp-aporepressor to bind a number of L-tryptophan analogues, some of which are known to affect trio gene regulation in vivo. The equilibrium inhibition constant (K i) for each analogue was obtained using equilibrium dialysis and the results are shown in Table I. 13C-NMR studies The binding of [1-13C]-DL-tryptophan to the aporepressor and the binding of [1-13C]-oL-trp-repressor to non-specific plasmid DNA has been examined by 13CNMR and the results are shown in Fig. 1. Fig. la shows the 100.6 MHz 13C-NMR spectrum of the trpaporepressor at 0.6 mM + 1.1 equivalent of [1-13C]-DL tryptophan. The free (principally o-tryptophan - see the K~ given in Table I) tryptophan resonance occurs at 177.3 ppm (pH 8.0) and titrates normally over the pH range 8.0-10.0 (data not shown). The bound tryptophan resonance occurs at 175.8 ppm and does not titrate over the same pH range. The original spectrum is regained on adjusting back to pH 8.0. The addition of pBR322 plasmid DNA (2 mg, final concentration 1.4 /zM), which contains non-specific sequences, resulted in an additional resonance at 174.8 ppm (Fig. lb). This resonance was assigned to the ternary complex between tryptophan, aporepressor and DNA, did not titrate over the range 7.6-10.6, like the binary complex at 175.8 ppm. The resonance due to free tryptophan titrated normally. In order to confirm that repressor-DNA complexes form under these conditions, the K d for non-specific binding of trp-repres-
158 TABLE 1 Inhibition constants (K i) for ligand binding to trp-aporepressor Ligand tested
Ki (txM)
none (aporepressor alone) L-tryptophan tryptamine indole a tryptophol a-methyl-L-tryptophan ~ 7-aza-L-tryptophan b indole-3-propionic acid 3-,6-indole-acrylic acid L-tryptophanamide 5-methyl-L-tryptophan b 4-fluoro-L-tryplophan b 5-fluoro-L-tryptophan b 6-methyl-L-tryptophan b 7-methyl-L-tryptophan b L-tryptophan methyl ester t.-tryptophan ethyl ester v-abrine o-tryptophan L-phenylalanine L-tyrosine
60 190 60 110 > 1000 > 500 38 6 325 19 110 65 170 173 140 214 280 10130 > 12000 > 31000
(K) (J) (I)
(rl)
a Greater than 95% pure after storage for one week in binding assay buffer as determined by reverse phase HPLC (purity and stability of other analogues was not analyzed). b t~Lmixture used for assays, contribution of o-form is assumed to be negligible.
(F) (E)
[ 1 - ~ 3 C ] Trp
+
repressor
(c) I
(B) (A) (
A
)
~
/
~
t 80
~ 70
PPM --T
r
i
T-'l
'l
;
,
|BO
t
,
I
I
l
~
t
;
,
170
O0M
Fig. 1. 100.4 MHz 1H-decoupled t3C-NMR spectra of trp-aporepressor (0.6 raM) in 10% DzO buffer (see Materials and Methods). (A), plus (1-13C)-~L-tryptophan (1.1 equiv.); (B), sample (a) plus 2 mg plasmid pBR322 DNA.
Fig. 2. 100.4 MHz JH-decoupled [3C-NMR spectra of trp-aporepressot (1.6 raM) in 10% D20 buffer (see Materials and Methods) with progressive additions of (1-13C)-DL-tryptophan at (A), 0 equiv, trp; (B), 0.25 equiv, trp; (C), 0.5 equiv, tro; (D), 0.75 equiv, trp; (E), 0.9 equiv, trp and sample (E) followed by successive additions of plasmid pBR322 DNA at (F), 0.5 equiv. DNA; (G), 1.0 equiv. DNA; (H), 2.0 equiv, DNA and (I), DNA alone; (J), DNA plus 0.9 equiv, trp and (K), free trp alone.
159
19F-NMR studies
sor to pBR322 was determined to be 400/~M (in 200 mM KC1 at 298 K). This implies that about 60% of the holorepressor will be bound to DNA, which is consistent with the relative intensities of the resonances. When the same experiments were repeated with protein at the higher concentration of 1.6 mM and at pH 7.6, the results were somewhat different, as shown in Fig. 2. At 0.5 equiv, of tryptophan, two resonances are visible, the bound at 176.0 ppm and the free at 175.0 ppm. The difference in the position of the free tryptophan resonance compared with Fig. 1 can be attributed to the difference in pH of the two experiments (note however that the bound resonances do not shift). Addition of increasing amounts of tryptophan results in an increase in intensity of only the resonance assigned to free tryptophan. The addition of pBR322 plasmid DNA (at a higher concentration than in the experiment shown in Fig. 1) resulted in considerable broadening of the resonances, but with further addition of DNA, two resonances are obtained. These can be assigned to the free tryptophan at 175.0 ppm and the ternary complex at 174.5 ppm. Since none of the binary trp-repressor complex resonance remains, presumably all the binary complex is bound to DNA. Further addition of DNA results in significant broadening, including the free tryptophan resonance, consistent with non-specific binding of tryptophan to DNA alone.
The binding of 4'- and 6'-DL-fluorotryptophan to the aporepressor at 0.4 mM has been examined and is shown in Fig. 3. Both the 19F-NMR resonances from 4'- and 6'-DL-fluorotryptophan, when added at 0.9 equiv, to aporepressor, were dramatically broadened on binding .to the protein (LW[4'-free] 19 Hz, LW[4'bound] 84 Hz; LW[6'-free] 22 Hz, LW[6'-bound] 103 Hz), without any significant change in chemical shift (Fig. 3a,b). If the K d values for the fluorotryptophans are comparable to tryptophan, then under these conditions, the o-form is free, while about 70% of the L-form is bound. For the study of the binding of 5'-fluorotryptophan, we synthesised [4',6',7'-2H3]-5 'OL-fluorotryptophan and the spectrum was noticeably different (Fig. 4) from the first two cases. Three chemically shifted resonances are visible in Fig. 4 (lower spectrum) - the small broad resonance at lower field is a minor impurity in the synthetic fluorotryptophan and the two upfield resonances are broad and slightly shifted with respect to the free fluorotryptophan (LW about 150 Hz for 0.15 ppm shifted peak and LW about 300 Hz for 1.5 ppm shifted peak). The narrower peak presumably represents the superposition of free 5'-0fluorotryptophan and free 5'-L-fluorotryptophan (in slow exchange with the bound state). The broader peak represents bound 5'-L-fluorotryptophan (in slow exchange with the free state). The upfield change in
6-F-Trp
4-F-Trp
(]3)
(A)
t
2 7 Hz
*repressor
"'"'l'"'"'""'"'""i'"'"'"'""'""l
120
I I0
PPM
.
I00
.
.
.
'"'"'"'"'""T'""'"'"'"'"I'"'""
120
! I0
PPM
Fig. 3. 376 MHz IH-decoupled 19F-NMR spectra of trp-aporepressor (0.6 mM) plus (A) 6'-or (B) 4'-fluoro-DL-tryptophan (0.9 equiv.) (lower) and free 6'- or 4'-fluoro-oL-tryptophan (upper).
160
Irradiate at:
.~
G F
i C ~..g
D .~:=--~ "A
I
i
,
,
'
'
I•0
,
I
,
,
~
O 0 PPM
I
i
,
I
I
I
,
i
"
\
C B
,
-I•0
Fig. 4. 235 MHz ]9F{1H}-heteronuclear 1-D NOE difference NMR spectra of trp-aporepressor (0.6 mM) plus (4',6',7'-2H3)-5'-DL-flu orotryptophan (0.9 equiv.). The 19F chemical shift scale is arbitrarily referenced to inorganic fluoride at 0 ppm. Off-resonance spectrum (lower) with Ill-irradiations (for 500 ms) on-resonance at shown positions (A, 0.58; B, 0.66; C, 0.70; D, 1.22; E, 4.32; F, 6.90; G, 7.08 ppm) in 250 MHz l H-spectrum (sideways).
chemical shift (relative to 5'-DL-fluorotryptophan in the absence of protein) is consistent with the interaction of the 5'-position of the indole ring with an amino-acid side-chain methyl group• In order to investigate this latter possibility further, we studied the interaction of [4',6',7'-2H3]-5'-u-fluoro tryptophan with aporepressor by 1-D heteronuclear Overhauser effect spectroscopy (Fig. 4, upper spectra)• Irradiation of selected resonances in the 1H-spectrum gave rise to heteronuclear NOEs to the bound 19F-resonance, principally in the aliphatic region of the spectrum. This experiment was performed at lower field strength than the experiment shown in Fig. 3, so that
the free and bound resonances are better resolved (the T2 relaxation times are longer at the lower field, leading to narrower lines). The effect of binding 5'-Lfluorotryptophan on the conformation of the trp-repressor was compared with that of binding tryptophan at equal protein concentrations by use of the DQFCOSY experiment. It was found that while the chemical shifts of the cross-peaks arising from the trp-holorepressor did not alter significantly, their intensities were reduced markedly in the complex with 5'-L-fluorotryptophan. These data (not shown) indicated that 5'-L-flUorotryptophan had a greater effect on repressor dynamics than did tryptophan, but did not not appear to
161 perturb the conformation of the holorepressor unduly. This result is not surprising considering that fluorine is to a good approximation isosteric with hydrogen.
Discussion The equilibrium dissociation (inhibition) data for the tryptophan analogues suggest that (i), the indole ring and the a-carboxylate are responsible for corepressor affinity of binding to the trp-aporepressor, while the a-amino group contributes a negative effect towards binding (see indole, indole-3-propionic acid and tryptamine in Table I); (ii), a methyl group substituted for a hydrogen atom at the a-carbon (a-methyl-L-tryptophan) caused considerably weakened binding as did substitution of a nitrogen atom for carbon at position seven of the indole ring (%aza-L-tryptophan) and (iii), the weakened binding of L-abrine is probably due to the loss of a hydrogen bond donor compared to L-tryptophan. We note that our data are largely consistent with that of Marmorstein et al. [8]. The 13C-NMR results are consistent with a strong ionic interaction between the tryptophan carboxylate and a basic group on the protein. The upfield shift of the bound resonance implies that electron density being removed from the carboxylate carbon, as would occur in a strong ionic interaction. The failure of the carboxylate resonance to titrate in the range where the amino group of tryptophan should deprotonate also suggests a strong ionic interaction for the amino group. From Sigler's structure [9] it is clear that Arg-54 forms a bridge with the carboxylate of tryptophan. Such a group is undoubtedly the strong positive charge which gives rise to the upfield shift of the bound resonance of [1-13C]trp-repressor complex. Furthermore, Arg-54 clearly perturbs the pK a of the bound tryptophan carboxylate, so that it does not titrate over the range investigated. The observation of an additional, further upfield shifted resonance upon addition of DNA suggests that the binding of DNA strengthens the ionic interaction for the tryptophan carboxylate, since the ternary complex is shifted further upfield from the binary complex. This is consistent with the findings of Sigler et al. [9]. The ~9F-NMR results suggest that while 4'- and 6'-fluoro-oL-tryptophan do bind to the trp-aporepressor, the local environment does not change significantly. In contrast, 5'-fluoro-DL-tryptophan binds tightly, which is consistent with the equilibrium dialysis measurements and the upfield shift is consistent with a Van der Waals interaction of the 5'-position of the indole ring with an amino acid side-chain methyl group, such as Val-58 or I1e-57. The heteronuclear NOE experiments attempt to identify which amino-acid residue might be inducing the shift. The largest NOE was observed when ~H-irradiation was carried out at
0.58 ppm (position A in Fig. 4), which is consistent with the position of the upfield shifted Val-58 y-methyl resonance (data not shown). The NOE observed when ~H-irradiation was carried out at 1.22 ppm (position D in Fig. 4) is consistent with the y'-methyl resonance of Val-58. Inspection of the X-ray crystal structure of the trp-aporepressor-corepressor complex [9] reveals that Val-58 is positioned for a favourable hydrophobic interaction with a fluorine atom, or a methyl group in the case of 5'-methyl-L-tryptophan, at the 5'-position of the indole ring. These data are also consistent with the results of Squires and co-workers, who found that 5'-methyl-L-tryptophan mediates repression in vitro at a lower concentration than L-tryptophan itself while 6-methyl-tryptophan is required at higher concentration [32]. Our conclusions are, therefore, that Val-58 is within the ring currents of the bound tryptophan and close in space to the indole 5'-position, on the basis of heteronuclear 19F{1H}-NOE experiments. Also, the tryptophan carboxylate is in hydrogen-bonding distance to a highly positively charged residue, probably Arg-54 and this bond strengthens on formation of the trp-repressor-DNA complex.
Acknowledgements We would like to thank Bill Bachovchin (Tufts University) for use of his NMR spectrometer, Tom Gerig (UC Santa Barbara) and the late Andy Derome (Oxford) for help with the heteronuclear 19F{1H}-NOE experiments, Neil MacKenzie (University of Arizona) for the gift of [3-13C]tryptophan. This work was supported by the Research Corporation Trust UK and WSU Grant-in-Aid (JNSE), National Institutes of Health Grant GM 29456 (RPG) and the WSU NMR Center is supported by NIH grant RR 0631401, NSF grant CHE 9115282 and Battelle Pacific Northwest Laboratories Contract No. 12-097718-A-L2.
References 1 Yanofsky, C. (1971) J. Am. Med. Ass. 218, 1026-1035. 2 Bennett, G.N., Schweinguter, M.E., Brown, K.D., Squires, C. and Yanofsky, C. (1976) Proc. Natl. Acad. Sci. USA 73, 2351-2355. 3 Gunsalus, R.P. and Yanofsky, C. (1980) Proc. Natl. Acad. Sci. USA 77, 7117-7121. 4 Zurawski, G., Gunsalus, R.P., Brown, K.D. and Yanofsky, C. (1981) J. Mol. Biol. 145, 47-73. 5 Heatwole, V.M. and Somerville, R.L. (1991) J. Bacteriol. 173, 3601-3604. 6 Arvidson, D.N., Bruce, C. and Gunsalus, R.P. (1986) J. Biol. Chem. 261,238-243 (1986). 7 Lane, A.N. (1986) Eur. J. Biochem. 157, 405-413. 8 Marmorstein, R.Q., Joachimiak, A., Sprinzl, M. and Sigler, P.B. (1987) J. Biol. Chem. 262, 4922-4927. 9 Schevitz, R.W., Otwinowski, Z., Joachimiak, A., Lawson, C.L. and Sigler, P.B. (1985) Nature 317, 782-786.
162 10 Lawson, C.L., Zhang, R.-G., Schevitz, R.W., Otwinowski, Z., Joachimiak, A. and Sigler, P.B. (1988) Proteins 3, 18-31. 11 Lawson, C.L. and Sigler, P.B. (1988) Nature 333, 869-871. 12 Lane, A.N. and Jardetsky, O. (1985) Eur. J. Biochem. 152, 395-404. 13 Lane, A.N. and Jardetsky, O. (1985) Eur. J. Biochem. 152, 405-409. 14 Lane, A.N. (1989) Eur. J. Biochem. 182, 95-104. 15 Arrowsmith, C.H., Pachter, R., Altman, R.B., Iyer, S~B. and Jardetsky, O. (1990) Biochemistry 29, 6332-6341. 16 Arrowsmith, C.H., Czaplicki, J., Iyer, S.B. and Jardetsky, O. (1991) J. Am. Chem. Soc. 113, 4020-4022. 17 Hyde, E.I., Ramesh, V., Frederick, R. and Roberts, G.C.K. (1991) Eur. J. Biochem. 201,569-579. 18 Otwinowski, Z., Schevitz, R.W., Zhang, R.-G., Lawson, C.L., Joachimiak, A., Marmorstein, R.Q., Luisi, B. and Sigler, P.B. (1988) Nature 335, 321-329. 19 Voet, D. and Voet, J.G. (1990) Biochemistry, p. 873, Wiley, New York. 20 Matthews, B.W. (1988) Nature 335, 294-295. 21 Brennan, R.G. and Matthews, B.W. (1989) Trends Biochem. Sci. 14, 286-290. 22 Kumamoto, A.A., Miller, W.G. and Gunsalus, R.P. (1987) Genes Dev. 1,556-564.
23 Staake, D., Walter, B., Kisters-Woike, B., Von WilckenBergmann, B. and Miiller-Hill, B. (1990) EMBO J. 9, 1963-1967 and corrigendum p. 3023. 24 Rafferty, J.B., Somers, W.S., Saint-Girons, I. and Phillips, S.E.V. (1989) Nature 341,705-710. 25 Phillips, S.E.V., Manfield, I., Parsons, I., Davison, B.E., Rafferty, J.B., Somers, W.S., Magarita, D., Cohen, G.N., Saint-Girons, I. and Stockley, P.G. (1989) Nature 341, 711-715. 26 Luisi, B. and Sigler, P.B. (1990) Biochim. Biophys. Acta 1048, 113-126. 27 Bass, S., Sorrells, V. and Youderian, P. (1988) Science 242, 240-245. 28 Krstulovic, A.N. and Powell, A.M. (1979) J. Chromatogr. 171, 345-356. 29 Bak, B., Led, J.J. and Pederson, E.J. (1969) Acta. Chem. Scand. 23, 3051. 30 Arvidson, D.N., Kumamoto, A.A. and Gunsalus, R.P. (1987) in Protein Purification: Micro to Macro. UCLA Symposia on Molecular and Cellular Biology, New Series, Vol. 68 (Burgess, R., ed.), pp. 401-407, Alan R. Liss, New York. 31 Collins, C.M. and Collier, R.J. (1984) J. Biol. Chem. 259, 1515915162. 32 Squires, C.L., Lee, F.D. and Yanofsky, C. (1975) J. Mol. Biol. 92, 93-111.
Biochimica et Biophysica Acta, 1160(1992) 156-162 © 1992Elsevier SciencePublishers B.V. All rights reserved 0167-4838/92/$05.00
BBAPRO 34318
Multinuclear NMR studies of the trp-repressor Jeremy N.S. Evans
a,b D e n n i s
N . A r v i d s o n c, R o b e r t P. G u n s a l u s a n d M a r y F. R o b e r t s d
c
a Department of Biochemistry / Biophysics, Washington State University, Pullman, WA (USA), b Department of Chemistry, Washington State University, Pullman, WA (USA), c Department of Microbiology, University of California, Los Angeles, CA (USA) and d Department of Chemistry, Boston College, Chestnut Hill, MA (USA)
(Received 18 March 1992)
Key words: Tryptophan-repressor;Protein-DNA complex;NMR; Protein-ligandcomplex The binding of the corepressor, L-tryptophan, to the Escherichia coli trp-aporepressor in solution has been examined by 13C-and ~9F-NMR spectroscopy. The binding of a number of tryptophan analogues have been studied by equilibrium dialysis. Evidence is presented that support the crystallographic studies (Schevitz, R.W., Otwinowski, Z., Joachimiak, A., Lawson, C.L. and Sigler, P.B. (1985) Nature 317, 782-786) that Val-58 is within the ring currents of the bound tryptophan and also close in space to the indole 5'-position, on the basis of heteronuclear 19F{1H}-NOE experiments. The tryptophan carboxylate is in hydrogen-bonding distance to a highly positively charged residue, probably Arg-54 and this bond strengthens on formation of the trp-repressor-DNA complex.
Introduction
The trp-repressor is involved in the transcriptional regulation of four operons in Escherichia coli which encode for enzymes in the L-tryptophan biosynthetic pathway, namely the t r p E D B A , trpR and a r o H operons and a m t r permease for L-tryptophan uptake from the medium [1-5]. Binding of the co-repressor, L-tryptophan, to the aporepressor results in the formation of the active aporepressor-corepressor complex, which subsequently binds tightly to the operators for each operon to repress their transcription by RNA polymerase. The trp-repressor is a small protein composed of two identical subunits (M r 12 356) [3]. Each dimer has two co-repressor binding sites that are identical and independent [6-8]. Furthermore, each L-tryptophan molecule is in proximity to amino-acid residues from both subunits of the aporepressor dimer [9,10]. Work from Sigler's laboratory [8,9] has refined the X-ray crystal structure of the trp-holorepressor and the indole propionic acid pseudorepressor complex [11] to 1.65 A. The structure suggests that the corepressor is bound rigidly in a hydrophobic binding pocket, with Correspondence to: J. Evans, Department of Biochemistry/ Biophysics, Washington State University,Pullman, WA, USA. Abbreviations: LW, linewidth; NOE, nuclear Overhauser effect; DQF, double quantum filtered; COSY, correlated spectroscopy.
the orientation of the indole ring well-defined (and in opposite senses in the two different complexes). However, it does not follow that the same holds in solution. Preliminary NMR studies in solution [6,12,13] suggest that there are differences between the two complexes and the assignment of intermolecular NOEs between repressor and co-repressor certainly imply that there is little dynamical motion in the co-repressor indole ring, but the orientation differs from that described by the crystal structure. More recent NMR work by Lane [14], by Jardetsky's laboratory [15,16] and by Roberts' laboratory [17] suggests that there is a significant change in the dynamics of the D- and E-helices of the protein in the presence of tryptophan. In the X-ray crystal structure of the trp-repressor-DNA complex proposed by Sigler and co-workers [18], it is suggested that the principal protein-base contacts are mediated through structured water. This structure, which has already made it into the textbooks of biochemistry [19], has been suggested to represent a non-specific complex which results from the crystallization conditions [20,21]. Indeed, it has been suggested that the trp-repressor may bind to the operator in a tandem fashion [22,23], as has been suggested for the metJ-repressor [24,25]. These criticisms have been addressed by Sigler's laboratory [26], in which it is argued that the crystalline protein-DNA complex does not change lattice form at the higher salt concentrations normally used for specific protein-DNA complex formation (up to 200 mM
157 NaCI) and the structure is consistent with the mutagenesis studies of Bass et al. [27]. In this paper, we examine the binding of tryptophan to the aporepressor from E. coli by solution-state 13C-and WF-NMR and the effect of addition of non-specific plasmid DNA. Materials and Methods
Chemicals. All chemicals were purchased from Sigma or Aldrich and were of the highest grade obtainable. L-Tryptophan was used without further purification. 4'-, 5'- and 6'-DL-fluorotryptophan were subjected to treatment with a Chelex column to remove paramagnetic impurities. Other tryptophan analogues were purified by HPLC by known methods [28]. [1-13C]-DLtryptophan was obtained from MSD Isotopes (Canada). [3-13C]-DL-tryptophan was a generous gift from Dr. N.E. Mackenzie (University of Arizona). Labelled tryptophans were purified further using Dowex 1 if required. [4',6',7'- ~H 3]-5'-DL-fluorotryptophan was synthesised from 5'-DL-fluorotryptophan using known methods [29]. Growth of cells. The trp-aporepressor-overproducing strain used in this work was E. coli LE392 containing the multiple copy trpR ÷ plasmid pRPG47 [3,5]. This plasmid, a pBR322 derivative, contains the trpR structural gene downstream from the E. coli lacUV5 promoter and ribosome binding site. Cell cultures were grown (at UCLA) with aeration to late logarithmic phase in a New Brunswick 200 1 fermenter in a medium containing 0.75% Bacto-Tryptone, 0.5% yeast extract, 0.5% lactose, 0.5% NaCl. The cells were harvested by centrifugation in a Sharples continuous centrifuge. Cells were stored for periods up to one year at -70°C. Protein purification. E. coli trp-aporepressor was purified by known methods [30] at UCLA. Purified protein was then shipped in 20% glycerol buffer (20 mM potassium phosphate buffer (pH 7.6)) on solid CO 2 to either MIT, Oxford, or WSU. Sample preparation. For NMR experiments, the aporepressor was concentrated, diluted and concentrated with 10 mM potassium phosphate buffer in 100% DaO, 100 mM KC1, 0.5 mM EDTA ((pH 7.6), except in Fig. 3 where the pH was 8.2), using a 3 ml Amicon stirred ultrafiltration cell and PM-10 membrane (M r cut-off 10 kDa) under about 75 psi of gaseous nitrogen pressure. The final volume of concentrated protein was usually 0.4 ml. Determination of analogue dissociation constants. This was carried out using a standard equilibrium dialysis competition assay [31], using L-[2-14C] tryptophan (0.2 /~M, final concentration of 53.5 mCi/mmol) and analogue at a variety of concentrations. NMR Spectroscopy. High-field Fourier transform (FT) NMR studies on the trp-repressor were performed on either a Bruker AM-250 (5.85 T, 250 MHz
1H, at Oxford), a Bruker AM-400 WB (9.39 T, 400 MHz XH, at Tufts University), a Bruker AM-500 (11.75 T, 500 MHz 1H, at Oxford) or a Varian VXR-500S (11.75 T, 500 MHz 1H at WSU) NMR spectrometer. Deuterium was used in all cases for locking the field. Chemical shifts were referenced externally to samples of similar dielectric constant: 13C-NMR spectra to dioxan in D20 buffer (6 c = 67.4 ppm); 19F-NMR spectra to trifluoroacetic acid in D20 (t~F 0 ppm). Sample temperature was maintained with the spectrometer variable temperature control unit, using boil-off liquid nitrogen. All samples containing trp-repressor were maintained at 298 K. Heteronuclear NMR was carried out with single frequency probes. Proton decoupling was achieved by gated broad band irradiation, using the Waltz-16 composite pulse sequence to minimize dielectric heating. Spectra were obtained with 32K data points and the FIDs Fourier transformed with line broadening (5 Hz for 19F or 10 Hz for ~3C). Results
Binding of tryptophan analogues As a prelude to understanding the interactions between tryptophan and trp-aporepressor, we have examined the ability of trp-aporepressor to bind a number of L-tryptophan analogues, some of which are known to affect trio gene regulation in vivo. The equilibrium inhibition constant (K i) for each analogue was obtained using equilibrium dialysis and the results are shown in Table I. 13C-NMR studies The binding of [1-13C]-DL-tryptophan to the aporepressor and the binding of [1-13C]-oL-trp-repressor to non-specific plasmid DNA has been examined by 13CNMR and the results are shown in Fig. 1. Fig. la shows the 100.6 MHz 13C-NMR spectrum of the trpaporepressor at 0.6 mM + 1.1 equivalent of [1-13C]-DL tryptophan. The free (principally o-tryptophan - see the K~ given in Table I) tryptophan resonance occurs at 177.3 ppm (pH 8.0) and titrates normally over the pH range 8.0-10.0 (data not shown). The bound tryptophan resonance occurs at 175.8 ppm and does not titrate over the same pH range. The original spectrum is regained on adjusting back to pH 8.0. The addition of pBR322 plasmid DNA (2 mg, final concentration 1.4 /zM), which contains non-specific sequences, resulted in an additional resonance at 174.8 ppm (Fig. lb). This resonance was assigned to the ternary complex between tryptophan, aporepressor and DNA, did not titrate over the range 7.6-10.6, like the binary complex at 175.8 ppm. The resonance due to free tryptophan titrated normally. In order to confirm that repressor-DNA complexes form under these conditions, the K d for non-specific binding of trp-repres-
158 TABLE 1 Inhibition constants (K i) for ligand binding to trp-aporepressor Ligand tested
Ki (txM)
none (aporepressor alone) L-tryptophan tryptamine indole a tryptophol a-methyl-L-tryptophan ~ 7-aza-L-tryptophan b indole-3-propionic acid 3-,6-indole-acrylic acid L-tryptophanamide 5-methyl-L-tryptophan b 4-fluoro-L-tryplophan b 5-fluoro-L-tryptophan b 6-methyl-L-tryptophan b 7-methyl-L-tryptophan b L-tryptophan methyl ester t.-tryptophan ethyl ester v-abrine o-tryptophan L-phenylalanine L-tyrosine
60 190 60 110 > 1000 > 500 38 6 325 19 110 65 170 173 140 214 280 10130 > 12000 > 31000
(K) (J) (I)
(rl)
a Greater than 95% pure after storage for one week in binding assay buffer as determined by reverse phase HPLC (purity and stability of other analogues was not analyzed). b t~Lmixture used for assays, contribution of o-form is assumed to be negligible.
(F) (E)
[ 1 - ~ 3 C ] Trp
+
repressor
(c) I
(B) (A) (
A
)
~
/
~
t 80
~ 70
PPM --T
r
i
T-'l
'l
;
,
|BO
t
,
I
I
l
~
t
;
,
170
O0M
Fig. 1. 100.4 MHz 1H-decoupled t3C-NMR spectra of trp-aporepressor (0.6 raM) in 10% DzO buffer (see Materials and Methods). (A), plus (1-13C)-~L-tryptophan (1.1 equiv.); (B), sample (a) plus 2 mg plasmid pBR322 DNA.
Fig. 2. 100.4 MHz JH-decoupled [3C-NMR spectra of trp-aporepressot (1.6 raM) in 10% D20 buffer (see Materials and Methods) with progressive additions of (1-13C)-DL-tryptophan at (A), 0 equiv, trp; (B), 0.25 equiv, trp; (C), 0.5 equiv, tro; (D), 0.75 equiv, trp; (E), 0.9 equiv, trp and sample (E) followed by successive additions of plasmid pBR322 DNA at (F), 0.5 equiv. DNA; (G), 1.0 equiv. DNA; (H), 2.0 equiv, DNA and (I), DNA alone; (J), DNA plus 0.9 equiv, trp and (K), free trp alone.
159
19F-NMR studies
sor to pBR322 was determined to be 400/~M (in 200 mM KC1 at 298 K). This implies that about 60% of the holorepressor will be bound to DNA, which is consistent with the relative intensities of the resonances. When the same experiments were repeated with protein at the higher concentration of 1.6 mM and at pH 7.6, the results were somewhat different, as shown in Fig. 2. At 0.5 equiv, of tryptophan, two resonances are visible, the bound at 176.0 ppm and the free at 175.0 ppm. The difference in the position of the free tryptophan resonance compared with Fig. 1 can be attributed to the difference in pH of the two experiments (note however that the bound resonances do not shift). Addition of increasing amounts of tryptophan results in an increase in intensity of only the resonance assigned to free tryptophan. The addition of pBR322 plasmid DNA (at a higher concentration than in the experiment shown in Fig. 1) resulted in considerable broadening of the resonances, but with further addition of DNA, two resonances are obtained. These can be assigned to the free tryptophan at 175.0 ppm and the ternary complex at 174.5 ppm. Since none of the binary trp-repressor complex resonance remains, presumably all the binary complex is bound to DNA. Further addition of DNA results in significant broadening, including the free tryptophan resonance, consistent with non-specific binding of tryptophan to DNA alone.
The binding of 4'- and 6'-DL-fluorotryptophan to the aporepressor at 0.4 mM has been examined and is shown in Fig. 3. Both the 19F-NMR resonances from 4'- and 6'-DL-fluorotryptophan, when added at 0.9 equiv, to aporepressor, were dramatically broadened on binding .to the protein (LW[4'-free] 19 Hz, LW[4'bound] 84 Hz; LW[6'-free] 22 Hz, LW[6'-bound] 103 Hz), without any significant change in chemical shift (Fig. 3a,b). If the K d values for the fluorotryptophans are comparable to tryptophan, then under these conditions, the o-form is free, while about 70% of the L-form is bound. For the study of the binding of 5'-fluorotryptophan, we synthesised [4',6',7'-2H3]-5 'OL-fluorotryptophan and the spectrum was noticeably different (Fig. 4) from the first two cases. Three chemically shifted resonances are visible in Fig. 4 (lower spectrum) - the small broad resonance at lower field is a minor impurity in the synthetic fluorotryptophan and the two upfield resonances are broad and slightly shifted with respect to the free fluorotryptophan (LW about 150 Hz for 0.15 ppm shifted peak and LW about 300 Hz for 1.5 ppm shifted peak). The narrower peak presumably represents the superposition of free 5'-0fluorotryptophan and free 5'-L-fluorotryptophan (in slow exchange with the bound state). The broader peak represents bound 5'-L-fluorotryptophan (in slow exchange with the free state). The upfield change in
6-F-Trp
4-F-Trp
(]3)
(A)
t
2 7 Hz
*repressor
"'"'l'"'"'""'"'""i'"'"'"'""'""l
120
I I0
PPM
.
I00
.
.
.
'"'"'"'"'""T'""'"'"'"'"I'"'""
120
! I0
PPM
Fig. 3. 376 MHz IH-decoupled 19F-NMR spectra of trp-aporepressor (0.6 mM) plus (A) 6'-or (B) 4'-fluoro-DL-tryptophan (0.9 equiv.) (lower) and free 6'- or 4'-fluoro-oL-tryptophan (upper).
160
Irradiate at:
.~
G F
i C ~..g
D .~:=--~ "A
I
i
,
,
'
'
I•0
,
I
,
,
~
O 0 PPM
I
i
,
I
I
I
,
i
"
\
C B
,
-I•0
Fig. 4. 235 MHz ]9F{1H}-heteronuclear 1-D NOE difference NMR spectra of trp-aporepressor (0.6 mM) plus (4',6',7'-2H3)-5'-DL-flu orotryptophan (0.9 equiv.). The 19F chemical shift scale is arbitrarily referenced to inorganic fluoride at 0 ppm. Off-resonance spectrum (lower) with Ill-irradiations (for 500 ms) on-resonance at shown positions (A, 0.58; B, 0.66; C, 0.70; D, 1.22; E, 4.32; F, 6.90; G, 7.08 ppm) in 250 MHz l H-spectrum (sideways).
chemical shift (relative to 5'-DL-fluorotryptophan in the absence of protein) is consistent with the interaction of the 5'-position of the indole ring with an amino-acid side-chain methyl group• In order to investigate this latter possibility further, we studied the interaction of [4',6',7'-2H3]-5'-u-fluoro tryptophan with aporepressor by 1-D heteronuclear Overhauser effect spectroscopy (Fig. 4, upper spectra)• Irradiation of selected resonances in the 1H-spectrum gave rise to heteronuclear NOEs to the bound 19F-resonance, principally in the aliphatic region of the spectrum. This experiment was performed at lower field strength than the experiment shown in Fig. 3, so that
the free and bound resonances are better resolved (the T2 relaxation times are longer at the lower field, leading to narrower lines). The effect of binding 5'-Lfluorotryptophan on the conformation of the trp-repressor was compared with that of binding tryptophan at equal protein concentrations by use of the DQFCOSY experiment. It was found that while the chemical shifts of the cross-peaks arising from the trp-holorepressor did not alter significantly, their intensities were reduced markedly in the complex with 5'-L-fluorotryptophan. These data (not shown) indicated that 5'-L-flUorotryptophan had a greater effect on repressor dynamics than did tryptophan, but did not not appear to
161 perturb the conformation of the holorepressor unduly. This result is not surprising considering that fluorine is to a good approximation isosteric with hydrogen.
Discussion The equilibrium dissociation (inhibition) data for the tryptophan analogues suggest that (i), the indole ring and the a-carboxylate are responsible for corepressor affinity of binding to the trp-aporepressor, while the a-amino group contributes a negative effect towards binding (see indole, indole-3-propionic acid and tryptamine in Table I); (ii), a methyl group substituted for a hydrogen atom at the a-carbon (a-methyl-L-tryptophan) caused considerably weakened binding as did substitution of a nitrogen atom for carbon at position seven of the indole ring (%aza-L-tryptophan) and (iii), the weakened binding of L-abrine is probably due to the loss of a hydrogen bond donor compared to L-tryptophan. We note that our data are largely consistent with that of Marmorstein et al. [8]. The 13C-NMR results are consistent with a strong ionic interaction between the tryptophan carboxylate and a basic group on the protein. The upfield shift of the bound resonance implies that electron density being removed from the carboxylate carbon, as would occur in a strong ionic interaction. The failure of the carboxylate resonance to titrate in the range where the amino group of tryptophan should deprotonate also suggests a strong ionic interaction for the amino group. From Sigler's structure [9] it is clear that Arg-54 forms a bridge with the carboxylate of tryptophan. Such a group is undoubtedly the strong positive charge which gives rise to the upfield shift of the bound resonance of [1-13C]trp-repressor complex. Furthermore, Arg-54 clearly perturbs the pK a of the bound tryptophan carboxylate, so that it does not titrate over the range investigated. The observation of an additional, further upfield shifted resonance upon addition of DNA suggests that the binding of DNA strengthens the ionic interaction for the tryptophan carboxylate, since the ternary complex is shifted further upfield from the binary complex. This is consistent with the findings of Sigler et al. [9]. The ~9F-NMR results suggest that while 4'- and 6'-fluoro-oL-tryptophan do bind to the trp-aporepressor, the local environment does not change significantly. In contrast, 5'-fluoro-DL-tryptophan binds tightly, which is consistent with the equilibrium dialysis measurements and the upfield shift is consistent with a Van der Waals interaction of the 5'-position of the indole ring with an amino acid side-chain methyl group, such as Val-58 or I1e-57. The heteronuclear NOE experiments attempt to identify which amino-acid residue might be inducing the shift. The largest NOE was observed when ~H-irradiation was carried out at
0.58 ppm (position A in Fig. 4), which is consistent with the position of the upfield shifted Val-58 y-methyl resonance (data not shown). The NOE observed when ~H-irradiation was carried out at 1.22 ppm (position D in Fig. 4) is consistent with the y'-methyl resonance of Val-58. Inspection of the X-ray crystal structure of the trp-aporepressor-corepressor complex [9] reveals that Val-58 is positioned for a favourable hydrophobic interaction with a fluorine atom, or a methyl group in the case of 5'-methyl-L-tryptophan, at the 5'-position of the indole ring. These data are also consistent with the results of Squires and co-workers, who found that 5'-methyl-L-tryptophan mediates repression in vitro at a lower concentration than L-tryptophan itself while 6-methyl-tryptophan is required at higher concentration [32]. Our conclusions are, therefore, that Val-58 is within the ring currents of the bound tryptophan and close in space to the indole 5'-position, on the basis of heteronuclear 19F{1H}-NOE experiments. Also, the tryptophan carboxylate is in hydrogen-bonding distance to a highly positively charged residue, probably Arg-54 and this bond strengthens on formation of the trp-repressor-DNA complex.
Acknowledgements We would like to thank Bill Bachovchin (Tufts University) for use of his NMR spectrometer, Tom Gerig (UC Santa Barbara) and the late Andy Derome (Oxford) for help with the heteronuclear 19F{1H}-NOE experiments, Neil MacKenzie (University of Arizona) for the gift of [3-13C]tryptophan. This work was supported by the Research Corporation Trust UK and WSU Grant-in-Aid (JNSE), National Institutes of Health Grant GM 29456 (RPG) and the WSU NMR Center is supported by NIH grant RR 0631401, NSF grant CHE 9115282 and Battelle Pacific Northwest Laboratories Contract No. 12-097718-A-L2.
References 1 Yanofsky, C. (1971) J. Am. Med. Ass. 218, 1026-1035. 2 Bennett, G.N., Schweinguter, M.E., Brown, K.D., Squires, C. and Yanofsky, C. (1976) Proc. Natl. Acad. Sci. USA 73, 2351-2355. 3 Gunsalus, R.P. and Yanofsky, C. (1980) Proc. Natl. Acad. Sci. USA 77, 7117-7121. 4 Zurawski, G., Gunsalus, R.P., Brown, K.D. and Yanofsky, C. (1981) J. Mol. Biol. 145, 47-73. 5 Heatwole, V.M. and Somerville, R.L. (1991) J. Bacteriol. 173, 3601-3604. 6 Arvidson, D.N., Bruce, C. and Gunsalus, R.P. (1986) J. Biol. Chem. 261,238-243 (1986). 7 Lane, A.N. (1986) Eur. J. Biochem. 157, 405-413. 8 Marmorstein, R.Q., Joachimiak, A., Sprinzl, M. and Sigler, P.B. (1987) J. Biol. Chem. 262, 4922-4927. 9 Schevitz, R.W., Otwinowski, Z., Joachimiak, A., Lawson, C.L. and Sigler, P.B. (1985) Nature 317, 782-786.
162 10 Lawson, C.L., Zhang, R.-G., Schevitz, R.W., Otwinowski, Z., Joachimiak, A. and Sigler, P.B. (1988) Proteins 3, 18-31. 11 Lawson, C.L. and Sigler, P.B. (1988) Nature 333, 869-871. 12 Lane, A.N. and Jardetsky, O. (1985) Eur. J. Biochem. 152, 395-404. 13 Lane, A.N. and Jardetsky, O. (1985) Eur. J. Biochem. 152, 405-409. 14 Lane, A.N. (1989) Eur. J. Biochem. 182, 95-104. 15 Arrowsmith, C.H., Pachter, R., Altman, R.B., Iyer, S~B. and Jardetsky, O. (1990) Biochemistry 29, 6332-6341. 16 Arrowsmith, C.H., Czaplicki, J., Iyer, S.B. and Jardetsky, O. (1991) J. Am. Chem. Soc. 113, 4020-4022. 17 Hyde, E.I., Ramesh, V., Frederick, R. and Roberts, G.C.K. (1991) Eur. J. Biochem. 201,569-579. 18 Otwinowski, Z., Schevitz, R.W., Zhang, R.-G., Lawson, C.L., Joachimiak, A., Marmorstein, R.Q., Luisi, B. and Sigler, P.B. (1988) Nature 335, 321-329. 19 Voet, D. and Voet, J.G. (1990) Biochemistry, p. 873, Wiley, New York. 20 Matthews, B.W. (1988) Nature 335, 294-295. 21 Brennan, R.G. and Matthews, B.W. (1989) Trends Biochem. Sci. 14, 286-290. 22 Kumamoto, A.A., Miller, W.G. and Gunsalus, R.P. (1987) Genes Dev. 1,556-564.
23 Staake, D., Walter, B., Kisters-Woike, B., Von WilckenBergmann, B. and Miiller-Hill, B. (1990) EMBO J. 9, 1963-1967 and corrigendum p. 3023. 24 Rafferty, J.B., Somers, W.S., Saint-Girons, I. and Phillips, S.E.V. (1989) Nature 341,705-710. 25 Phillips, S.E.V., Manfield, I., Parsons, I., Davison, B.E., Rafferty, J.B., Somers, W.S., Magarita, D., Cohen, G.N., Saint-Girons, I. and Stockley, P.G. (1989) Nature 341, 711-715. 26 Luisi, B. and Sigler, P.B. (1990) Biochim. Biophys. Acta 1048, 113-126. 27 Bass, S., Sorrells, V. and Youderian, P. (1988) Science 242, 240-245. 28 Krstulovic, A.N. and Powell, A.M. (1979) J. Chromatogr. 171, 345-356. 29 Bak, B., Led, J.J. and Pederson, E.J. (1969) Acta. Chem. Scand. 23, 3051. 30 Arvidson, D.N., Kumamoto, A.A. and Gunsalus, R.P. (1987) in Protein Purification: Micro to Macro. UCLA Symposia on Molecular and Cellular Biology, New Series, Vol. 68 (Burgess, R., ed.), pp. 401-407, Alan R. Liss, New York. 31 Collins, C.M. and Collier, R.J. (1984) J. Biol. Chem. 259, 1515915162. 32 Squires, C.L., Lee, F.D. and Yanofsky, C. (1975) J. Mol. Biol. 92, 93-111.
