Proton nuclear magnetic resonance studies on glutaminebinding protein from Escherichia coli

J. Mol. Biol.

(1989) 210. 849-857

Proton Nuclear Magnetic Resonance Studies on Glutaminebinding Protein from Escherichia coli Formation

of Intermolecular

and Intramolecular Ligand Binding

Qichang Shent, Virgil Simplaceanu,

Hydrogen

Bonds upon

Patricia F. Cottam and Chien HoI

Department of Biological Sciences, Carnegie Mellon University 4400 Fifth Avenue, Pittsburgh, PA 15213, U.S.A. (Received 26 October 1988, and in revised form 11 August 1989) Proton nuclear magnetic resonance studies have revealed several structural and dynamic3 properties of the glutamine-binding protein of Escherichia coli. When this protein binds L-glutamine, six low-field, exchangeable proton resonances appear in the region from +55 to + 10 parts per million downfield from water (or + 10.2 to + 14.7 parts per million downfield from the methyl proton resonance of 2,2-dimethyl-2-silapentane-5-sulfonate). This suggests that the binding of L-glutamine induces specific conformational changes in the protein molecule, involving the formation of intermolecular and intramolecular hydrogen bonds between the glutamine-binding protein and L-glutamine, and within the protein molecule. The oxygen atom of the y-carbonyl group of L-glutamine is likely to be involved in the formation of an intermolecular hydrogen bond between t’he ligand and the binding protein. We have shown that at least one phenylalanine and one methyl-containing residue are spatially close to this intermolecular hydrogen-bonded proton. The intermolecular and intramolecular hydrogen-bonded protons of the ligand-protein complex undergo solvent exchange. The local conformations around these intermolecular and intramolecular hydrogen bonds are quite stable when subjected to pH and temperature variations. From these results, the utility of proton nuclear magnetic resonance spectroscopy for investigating such binding proteins has been shown, and a picture of the ligand-binding process can be drawn.

proposed. According to this model, which is based mainly on biochemical and genetic evidence, a ligand-binding protein plays two essential roles in the two-step transport process: first, recognition, i.e. binding of its specific ligand with high affinity; second, ligand translocation across the cytoplasmic membrane by interaction between the binding protein-ligand complex and the corresponding membrane-associated protein components. The translocation process is believed to require metabolic energy. For a discussion on this subject, refer to Ames (1986). In Escherichia coli, the active transport of L-glutamine is carried out primarily by a single transport system, composed of a glutamine-binding protein (GlnBP§) and two membrane-associated 0 Abbreviations used: GlnBP, glutamine-binding

1. Introduction A variety of compounds and ions, including amino acids, peptides, sugars, anions and vitamins, can be transported into Gram-negative bacteria against a concentration gradient by a particular class of active transport systems. Each member of this class of transport systems is characterized by an indispensable periplasmic binding protein, and several cytoplasmic membrane-bound protein components. Since osmotic shock reduces the activity of these transport systems, they are referred to as osmotic shock-sensitive transport systems (for reviews, see Ames, 1986; Furlong, 1987). A model for ligand transport by the osmotic shock-sensitive transport systems has been

t Present address: Department of Pediatrics, Medical Center, University of Massachusetts, Worcester. MA 01605, U.S.A. 1 Author to whom all correspondence should be addressed. 0022-2836/89/240f!.4949

$03.00/0

protein; n.m.r., nuclear magnetic 2,2-dimethyl-2-silapentane-5-sulfonate: Overhauser effect, p.p.m., parts per billion.

849

0

resonance; DSS, Eu‘OE, nuclear per million: p.p.b..

1989 Academic

parts

Press Limited

850

Q.

Shen

proteins (Weiner & Heppel, 1971; Nohno et al., 1986). The most studied component of the system is GlnBP. It is a monomeric protein with a molecular weight of 24,935. A large number of lysine and arginine residues make it a basic protein with a p1 value of 86. Of the naturally occurring amino acids, it binds only L-Gln, forming a 1 : 1 molar ratio complex. The ligand dissociation constant (Kn) is about 5 x lo-’ M at 5°C and pH 7.2. Upon binding of L-Gin, GlnBP undergoes a ligand-induced conformational change as revealed by fluorescence spectroscopy (Weiner & Heppel, 197 1 ), proton nuclear magnetic resonance (n.m.r.) spectroscopy (Kreishman et al., 1973), and microcalorimetry (Marty et aE., 1979). There are two L-Gln analogs, y-glutamylhydrazide and y-glutamylhydroximate, that’ bind competitively to the protein (Weiner & Heppel, 1971). The dependence of the L-Gln transport on GlnBP has been demonstrated by in vitro reconstitution of the transport in a membrane vesicle system (Hunt & Hong, 1983). All these features make the glutamine transport system of E. coli a tempting model for the study of the relationship among the structure, dynamics and function of periplasmic binding proteins. The formation and the properties of intramolecular and intermolecular hydrogen bonds between GlnBP and L-Gln as investigated by ‘H-n.m.r. spectroscopy are reported here. n.m.r. offers a powerful means for obtaining structural and dynamic information on proteins under near in vivo conditions (Jardetzky & Roberts, 1981; Wagner, 1983; Wiithrich, 1986).

2. Materials and Methods (a)

Materials

AntibioticswerepurchasedfromSigma.nL-[cr2,3,4.j.6-‘H6Jphenylalanine nr.-[~,2,4,5,6,7-~H,1([2HlPhe), and m-[cq2,3,5,6-‘H,]tyrosine tryptophan (]‘H]Trp), ([2H]Tyr) were prepared as described by Matthews et al. (1977). i,-[3,4-3H(N)]glutamine was purchased from DuPont NEN. L-Glutamine-15N1(amide-15N) was purchased from MSD Isotopes. All other chemicals and reagents were purchased from commercial sources and used without further purification.

et al Plasmids p
GlnBP was purified from cultures of E. coli GLNP, and BK9MDG (r- m- thihsdS metC A-GlnP endB), which were gifts from Drs L. A. Heppel and J.-S. Hong, respectively. AT2471 (tyrA4 thi reEA1 1- spoT1: Taylor & Trotter. 1967) and KA197 (pheA97 thi-I relA1 1- spoTI: Hoekstra et al.. 1974) were supplied by Dr Barbara J. Bachmann, EGSC, New Haven, CT. [‘H]Tyrand [‘HIPhe-labeled GlnBP were purified from AT2471 or KAI97 following transformation of these auxotrophs with plasmids pGPl-2 and pJW133. [‘H]Trp-labeled GlnBP was obtained from CLGTR trp : : Tn10, which was derived from GLNP, by transduction with a lysate of PiCM (Rosner, 1972) grown on NK5145 trp : : TnlO. After selection for tetracycline resistance and tryptophan auxotrophy, this strain was transformed with plasmids pGPl-2 and pJWlO7, NK5145 was a gift from Dr N. Kleckner.

cell growth

Minimal medium composed of 7 g of K,HPO,, 3 g of KHIPO,, 1 g of (NH&S04, @l g of MgSO, .7H,O, and JO g sodium succinatejiter of water was used for the growth of GLNP, cells. Cells of AT2471 and KA197 were grown in minimal medium containing 30 mg kanamycin and ampicillin/l and enriched with either 1% (w/v) tyrosine assay medium and 200 ,mc[‘H]Tgr or 1 */o (w/v) phenylalanine assay medium and 200 PM-[‘HIPhe. For growing CLGTR, the minimal medium was supplemented with 1% tryptophan assay medium or 1% (w/v) Casamino acids with @4% (w/v) glucose, and 100 ,uM-[‘H]Trp. in the presence of 30 mg kanamycin and ampicillin/l, and 15 mg tetracycline hydrochloride/l, L broth (Miller, 1972). with the addition of 30 mg of kanamycin and ampicillin/l. was used for growing cells of BKSMDG. GLNP, cells were grown with aeration in minimal medium at 37°C to the late exponential phase. Strains carrying the coupled plasmids were grown with aeration in their respective media at 32”C, until the A,,, of the cultures reached about 2. Exclusive synthesis of GlnBP was induced following a temperature shift to 42 “C and the addition of rifampicin as outlined by Tabor & Richardson (1985). (d) Puri$cation

of G1nRP

GlnBP was released from cells by osmotic shock (Neu & Heppel, 1965), or by chloroform-shock (Ames et ad., 1984). GlnBP was precipitated from shock fluid by the addition of ammonium sulfat,e to saturation. The protein pellet was resuspended and dialyzed against 10 mlw-Tris HCl (pH 7*6), containing 0.1 mM-K,EDTA. The suspension was loaded onto a DE52 anion exchange column equilibrated with the same buffer. The pass-through fraction was collected, concentrated by ultrafiltration, and chromatographed on a Sephadex G75 column equilibrated with 100 mr+potassium phosphate (pH 7.2). GlnBP was purified to homogeneity, as shown by SDS/acrylamide gel electrophoresis. (e)

(b) Strains and plasmids

and

Assay

of protein

concentration

Protein concentrations were determined by the Lowry procedure (Lowry et al.. 1951) using cryst,alline bovine serum albumin as the protein standard. (f)

Determination,

of glutamine-bin,din.g

activity

The K, value for the GlnBP-Gln complex was determined by equilibrium dialysis as described by Level (1972) with minor modifications. Radioactivity was counted on ir . :rckman liquid scintillation counter (model LS-7000). (2)

n.nb.r.

sample

preparation

1‘I~i,it;, I ($lnBI’ preparations were concentrated by ’ ii *fitration. To prepare GlnBP samples in 2H20r the protein samples were diluted and recormentrated 3 to 4

L-Glutamine-induced

times with 100 mu-potassium phosphate in ‘Hz0 at pH 7.2. When H,O was required as solvent for the GlnBP samples, ‘H,O was added to 5% (v/v) to provide an internal lock. If not specifically indicated, the concentration of GlnBP was 2 mM, and the buffer used was 100 mM-potassium phosphate (pH 7.2). (h) n.m.r. measurements ‘H-n.m.r. measurements were carried out at 29”C, unless otherwise stated, on a Bruker WM-300 or AM-300 n.m.r. spectrometer operating in the Fourier transform mode at 300 MHz for ‘H measurements, using a 18bit digitizer. The measurements generally used 90” pulses, a relaxation delay of 1 s, a spectral width of 8 kHz, and 16,384 data file size. The ‘H chemical shifts are reported relative to the resonance of H,O (or ‘H*HO when using ‘HZ0 as a solvent) taken as 0 parts per million (p.p.m.), which, in turn, is 4.75 p.p.m. downfield from the methyl proton resonance of 2.2-dimethyl-2-silapentane-5-sulfonate (DSS) at 29°C. For measurement of the temperature dependence, the chemical shift of water at each temperature was measured by using a sample of buffer with a trace of DSS added. The spectra were then referenced indirectly to DSS using the water proton resonance as a secondary standard. The positive sign in the chemical shift indicates that the resonance is downfield from the reference signal and the negative sign indicates that the resonance is upfield from that of the reference. The accuracy of the chemical shift measurement is estimated as +O.Ol p.p.m. ‘H-n.m.r. spectra of unlabeled or ‘H-labeled GlnBP in HZ0 were obtained by using the jump-and-return pulse sequence (Plateau et al., 1983) with a 100~PCS delay between the two 90” pulses. ‘H-‘H nuclear Overhauser effect (NOE) measurements were performed by accumulating the difference between blocks of free induction decays with on and off-resonance pre-irradiation of a chosen resonance (truncated driven NOE; Keller & Wiithrich, 1981). The irradiation time was

GlnBP/Gln

q

5: 15

GlnBP/

200 ms, in order to achieve selectivity and to minimize the effect of spin diffusion. Typically, a total of 8000 to 16,000 scans were collected. The non-selective spin-lattice relaxation time (t,) values of the 6 low-field exchangeable proton resonances were measured by inverting the whole spectrum, including water. A composite inversion pulse 9o”(X)240”( Y)9O”(X) (Freeman et al., 1980) was used in order to compensate for the effects of resonance offsets and spatial inhomogeneity of the radiofrequency field, and the jump-and-return pulse sequence of Plateau et al. (1983) was used to measure the magnetization recovered after delays of variable duration. t, values were obtained by fitting the data to the equation Mz(t)=A,+A, (l-e-@‘). For determination of the selective t, values, each lowfield exchangeable proton resonance was selectively inverted (incompletely) by pre-irradiation with a decoupler pulse of power of 20 L (44 Hz) and a duration of 14 ms. After a variable delay of (PO1 to 04 s, the recovered magnetization was measured and the data were processed as described above.

3. Results (a) Binding activity of GlnBP Purified GlnBP has a dissociation constant (Ko) of 5 x lo-’ at 5°C upon binding of L-Gin, which is comparable to the results reported by Weiner & Heppel (1971). (b) Effects of L-Gln and

its analogs on the ‘H-n.m.r. spectra of GlnBP in H,O

In order to investigate the conformational changes induced by the binding of the ligand, ‘H-n.m.r. spectra of GlnBP in H,O with increasing amounts of r,-Gln have been obtained (Fig. 1). Ligand-induced spectral changes can be observed,

“”

GlnBP/Gln=5.5

GlnBP/Gln

851

Changesin GlnBP

Conformational

,,

= 5: 3

-

Gln = 5.1

9.97

7.84 7.74

6.87

6.44

5.69

6.44 A

GlnBP I

t11.0

I

+10.0

I

t9.0

I

t8.0

I

I

t7.0

t6.0

p.p.m. from Hz0 Figure 1. Effects of L-glutamine on the low-field region of the 300 MHz 100 mw-potassium phosphate (pH 7.2) in H,O and at 29°C.

I

+5.0 ‘H-n.m.r.

spectrum

I

t4.0 of 2 mM-GlnBP

in

852

Q. Shen et al.

GlnBP/Y-Glutamylhydroxlmote

=l

GlnBP/Y-Glutamylhydrazlde:l

5

10

9 72

7 81 7.78

n1D3

I

I

t 11.0

/

t10.0

6.34

7 91 7.715

/

t 9.0

I

t8.0

/

t7.0

1

t6.0

+5.0

I

t4.0

p.p.nl. frsm Hz3 y-Glutamylhydroxlmate OHHHO II I HOW-C-C-C-C-C-OH / t. H H

I

r-Glufamylhydraz~de 0 HHH II / , I H2N-N-C-C-C-C-C-OH

II

1

NH,

0 /I

H H NH2

i- Glutom~ne 0 H ii W 0 II I I I :I H-N-C-C-C-C-C-OH / III 1, i-, 1, NH,

Figure 2. Effects of I,-glutamine analogson the low-field regionof the 300MHz ‘H-n.m.r. spectrumof :! mM-(AnBI’ in 100mM-potassiumphosphate(pH 7.2) in H,O and at 29°C.

especially in the low-field proton resonance region, where six broad. well-resolved resonances appear between + 50 and + 1WI p.p.m, downfield from H,O (or +9.7 and + l-t.7 p.p.m. downfield from I>SS). These resonances increase in intensity until the molar ratio of r,-Gln/GlnBP reaches 1 : 1. while their positions do not shift during the course of the titration. The ligand-induced sp&ral changes that can be seen in other regions of the proton spectrum are also complete at a 1 : 1 molar ratio of rAln/GlnBP. Thus, our results confirm the findings of previous biochemical and ‘H-n.m.r. studies suggesting that there is only one l&and-binding sit,e in GlnBP for I,-Gin (lYciner & Heppel. 1971: Krrishman ef nl.. 1973). A series of r,-Gin analogs were examined for their effect on t,he ‘H-n.m.r. spectrum of GlnRP in H,O. Neither I,-Glu nor L-Asn causes any observable specbtral change. The two known competit,ive binding inhibitors. y-glutamylhydroximate and y-glutamylhydrazide, do induce spectral changes similar to those induced by r,-Gin, including t’he appearance of the six low-field resonances (Fig. 2), although some of these resonancesa,re shifted with respect to those in the presenceof r,-Gln. Tn the presenceof a tenfold excess of y-glutamylhydrazidr. the addition of an

equal-molar amount of I,-Gln to GlnHP results in a spectrum that is essentially ident’ical with that of t,he GlnBP--Gin complex (result,s not shown). These results demonstrate that the simultaneous appearance of t)he six low-field resonancesis a consequence of a specific int,eraction between GlnBP and its ligand: and that GlnRP exhibits the highest affinitv towards its naturally occurring liga,nd. r,-Gln.

(c) Solvent-exchangeable

nature cg the low-field rpsomznnces

The six low-field proton resonancsesare due, t,o solvent-exchangeable protons, since t,hey vanish in the ‘H-n.m.r. spectrum of the GlnBP-Gin complex dissolved in 2H20. As t’hese resonances can be observed in t,he ‘H.-n.m.r. spectrum taken in H,O. the solvent) excha,nge rate must be slow 011 the chemical shift time-scale. On t,he basis of the most upfield resonance (+57 p.p.m. from H,(j). the exchange rate should be less than 1680 s ‘. An e&mate of the exchange rates for these protons (*an be derived from their spin-lattice relaxation time (tl) values. summarized in Table 1. Tt can br seen that the selective t, values arc in the range 20 t)o 80 ms. while the no~xwlwtivr f, valrws ~IW

Table 1 Span-lattice relaxation times (tl) of the six low-$eld, exchangeable proton resonancecs determined by selective saturation recovery and non-selective inversion recowry Krs011ancr (p.p.111. H,O at WC’)

from

Selectivef, (ms) Non-selective f, (s)

+ 569 19 (k.5)

1.2

+e1Ci 84 (k18) 1 .o

proton

+I237 49 (k21) 12

+ 7.74 40 (&%5) 1.3

+ 7+4-i 4% (&13) 14

+ 9% 43 (*lY) I .:3

L-Glutamine-induced

Changes

Conformational

in

a53

GlnBP

6.87

I t

I

I

IO.0

+ 8.0

1

I

I

t6.0 p.p.m.

from

I

t 4.0

1

I

I

t2.0

I

1

-1-O

-2.0

I

-3.0 p.p.m.

Hz0

1

1

-4.0

-5.0

from

I

-6.0

-7.0

H20

Figure 3. ‘HP’H NOE difference spectrum of 10 mivr-GlnBP plus 200 mM-L-glutamine in 100 mM-potassium phosphate (pH 7.2) in H,O and at 29°C. with irradiation at the +6.87 p.p.m. resonance for 200 ms. substantially longer. being between 1 and 1.5 s: close t’o the t, value of H,O. The marked difference between t,he two sets oft, values for these low-field prot,on resonances is another indication that these protons undergo a moderately fast exchange with H,O. The inverse of their select’ive t, value can be regarded as a rough estimate of the upper limit of the exchange rate. The selective t, of 50 ms for the resonance at +6.87 p.p.m. gives 20 5-l as an upper limit of the solvent exchange rate, which is more stringent than the upper limit imposed by t>he chemical shift time-scale. The non-selective t, values impose a lower limit for t’he exchange rate which must be larger than, say, 0.5 s-r, significantly faster than the inverse of the non-selective tl value. The upper and lower limits of the exchange rat’es for the ot’her exchangeable proton resonances should be of similar magnit.ude. since they have similar t, values. (d) ‘H-‘H resonances

XOE studies of the low-$eld proton of thr GlnRP-L-Gln complex

‘H-‘H SOE measurements of GlnBP with a 20fold excess of L-Gin have been performed in order to reveal the spatial relat,ionship between the six lowfield proton resonances and other protons from ClnBP and/or L-Gin. Irradiation of the resonance at +6*87 p.p.m. from H,O causes an intensity decrease in the resonances arising from the y and p protons of free r,-Gln. in addition to a decrease of the intensit)irs of several aromatic and .aliphatic proton anti one ring-current-shifted proton resonances, resonance at -4.62 p.p.m. from H,O (Fig. 3). The saturation transfer to the y and /l proton resonances increases when the irradiation time is increased from 40 t)o 200 ms. In a parallel experiment, the intensit’ies of the 7 and fl peaks in the difference

spectrum are found to increase when t,he molar ratio of L-Gln/GlnBP is increased from 1 : 1 to 9 : 1. This indicates that the observed saturation transfer is indeed to the y and fl proton resonances of the free L-Gln via the dissociation of the GlnBP-Gln complex. Irradiation of each of the other five low-field proton resonances was found to cause negative NOES, but only on resonances from GlnBP (results not shown). Thus, these five resonances arise solely from the GlnBP molecule and are not located near the ligand-binding sit’e. (e)

Experiments

with,

(amkde-

1 i~V)-Cln

Transferred NOE measurements show that the proton resonating at +6.87 p.p.m. from H,O is close to the y and fl protons of the bound r,-Gln. A y-amide proton of L-Gln seems most likely to be involved in an intermolecular hydrogen bond with GlnBP. The possibility that this resonance might arise from any of t’he two y-amide protSons has been ruled out by the observation tha)t the addition of (amide-15N)-labeled r,-Gln to GlnBP does not lead to a splitting of the resonance at +6.87 p.p.m. from H,O. Because of the scalar coupling to the 15N nucleus, the resonances from the amide protons in the r5N-labeled r,-Gln would each split int’o a doublet, with a .JISN~,n coupling constant of about 90 Hz (Blomberg et al., 1976). This has indeed been observed in the spectrum of free ‘“N-labeled r,-Gln. (f)

Experiments

with

partially

deuterated GlnHP

Homonuclear ‘H-‘H KOE difference spe&a of [‘H]Tyr-, [‘HIPhe-, and [‘H]Trp-labeled Gin BPS in t,he presence of I>-Gin have been compared with that

854

Q. Shen et al.

I

/

8.5

I

I

7.5

I

I

6.5

I

I

5.5

p.p.m.

h

I

4.5

from

/

I

3.5

I

I,

2.5

8.5

1.5

1

I

7.5.

I

I

I1

6.5

I

5.5 p.p.m.

I

4.5 from

5.5

/

!

I

3.5

I

(

2.5

1

,

I.5

4.5

from

3.5

2.5

I.5

H,O

6.87 4

id)

/

6.5 p.p.m.

(Cl

8.5

7.5

Hz0

75

8.5

I

p.p.m.

H,O

I

I

5.5

6.5

/

4.5

from

/

/

i

3.5

1

2.5

I

I.5

Hz0

Figure 4. ‘H-‘H ISOE difference spectra of 5 mw-unlabeledand partially deuterated GlnlVPs plus 15mlvr-L-glutamine in 100 mw-potassium phosphate (pH 7.2) in H,O and at 29°C: (a) unlabeled GlnBP with irradiation at the +687 p.p.m. resonance: (b) [‘HIPhe-labeled GlnBP with irradiation at the +687 p.p.m. resonance; (c) [‘HI; Tyr-labeled GlnBP with irradiation

at the

6.87 p.p.m.

resonance;

(d) [‘H]Trp-labeled

GlnBP

with

irradiation

at the

+687

p.p.m.

disappear from the difference spectrum of [‘HIPhe-labeled GlnBP. Hence, these resonances can be assigned to the aromatic protons from at least one Phe residue. The origin of the resonances at + 3.05 and + 2.57 p.p.m. from H,O in the NOE difference spectrum of the unlabeled GlnBP is not known. They could be amides or the C-2 and/or C-4

of unlabeled GlnBP. As shown in Figure 4, when the resonance at +6*87 p.p.m. from H,O is irradiated, the NOE difference spectra of ]‘H]Tyrand [*H]Trp-labeled GlnBP are identical with that of the unlabeled GlnBP, but three resonances at + 182, + 2.23 and + 2-45 p.p.m. from H,O present in the difference spectrum of unlabeled GlnBP

Table 2 Temperature coefieients of the low-field, exchangeableproton resonancesof the Gln-GlnBP complex at pH 7.2 Resonance from H,O + + f + +

9.97 7.84 7.74 687 6.33

f6.16 +569

(p.p.m at 29°C)

Resonance (p.p.m. from DSS) + + + + +

Temperature coefficient (p.p.b./“C)

1472 12.59 12.49 1162 11.08

-44 - 2.0 -69 -97 -043

+ 1091 + 1064

- 3.8 - 13.0

t This conclusion is baaed on the observed 95 Hz) in a uniformly “N-labeled GlnBP unpublished results).

splitting (N.

resonance.

Tjandra,

Nature

of hydrogen

bond

Intramolecular Intramolecular Intramolecular Intermolecular (to an amide)i Intramolecular (to the indole NH of Trp220) intramolecular (to an amide)? Intramolecular (to an amide)t

of this resonance (with a JIsN ,” value of about V. Simplaceanu, P. F. Cottam & C. Ho,

L-Glutamine-induced

pH 5.4

Conformational

Changes

855

in GlnBP

A

pti 7.2

pH 8.0

L....

11.0

-----IL-

1

10.0

9.0

a.0 ppm

7.0 from

6.0

-L

5.0

4.0

H,O

Figure 5. Effects of pH on the low-field proton resonances of 2 m&&l&P plus 6 mr+L-glutamine in 100 mM-potassium phosphate in H,O at 29°C.

protons of the sole histidyl residue in the protein molecule. A ring-current-shifted proton resonance at -4.62 p.p.m. from H,O in the NOE difference spectrum of unlabeled GlnBP-Gln complex might arise from a methyl proton close to an aromatic ring, possibly a Phe residue (Fig. 3). (g) Eflectx

of pH and exchangeable

temperature on the low-jield proton resonances

The effects of’ pH on the six low-field, exchangeable proton resonances are shown in Figure 5. No significant change in the positions or linewidths of these resonances, in particular the resonance at +6.87 p.p.m. from H,O, has been detected over the pH range from 5.4 to 8.0 at 29°C. Spectra taken at pH 7.2 and at temperatures ranging from 3 to 48°C are shown in Figure 6. Small shifts in the position of the various resonances can be seen as a function of temperature. The temperature coefficients, summarized in Table 2, are quite small, varying from -96 to -6.9 part,s per billion (p.p.b.)/“C, with the exception of the resonances at +6*87 and +5.69 p.p.m. from H,C which have temperature coefficients of -9.7 and -13.Op.p.b./“C, respectively. In the ‘H-n.m.r. spectra of a uniformly 15N-labeled GlnBP, we have observed a doublet with a JlsNmIH value of about 95 Hz for each of these two resonances (N. Tjandra, V. Simplaceanu, P. F. Cottam & C. Ho, unpublished results). These results suggest that the resonances at +687 and 5.69 p.p.m. arise from hydrogen bonds of amide protons (see Table 2).

4. Discussion Our ‘H-n.m.r. studies on GlnBP show that ligand-induced conformational changes occur in GlnBP when it binds either its natural ligand, L-Gln, or the two Gin competitive inhibitors. Spectral changes are observable in the ring-currentshifted proton resonance region as well as in the aromatic proton region (results not shown), but the most striking effect is the appearance of six low-

f ,‘., ‘,j

\.i

3Oc -c--c I

I

16.0

I

15.0

I

14.0

I

13.0

p.p.m.

I

12.0

from

I

il.0

IO.0

DDS

I

I

I

!

I

I

I

II.0

IO.0

9.0

8.0

7.0

6.0

5.0

p.p.m.

from

Hz0

at 25OC

Figure 6. Effects of temperature on the low-field exchangeable proton resonances of 4 mn&lnBP plus 20 mM-L-glutamine in 100 mM-potassium phosphate (pH 7.2) in H,O.

field, exchangeable proton resonances in the region from +55 to +lOp.p.m. from H,O at 29°C (or +10.2 to 14.7 p.p.m. from DSS), as shown in Figure 1. They are observable only when H,O is the solvent and they occur in the hydrogen-bonded proton resonance region (Mason, 1987). It is well known that hydrogen bonding can induce a lowfield chemical shift of up to 10 p.p.m. (Emsley et al., 1967; Becker, 1969; Emsley, 1980). The chemical shifts of these low-field resonances change slightly

&. Shen

when the temperature is varied from 3 to 47”C, as shown in Figure 6. The temperature coefficients are within the limits found by Deslauriers & Smit’h (1980) for hydrogen-bonded protons in model peptides. except for the resonance at +569p.p.m. from H,O whose temperature coefficient (Table 2) is somewhat larger (-13 p.p.b./“C). On the basis of these observations, we conclude that t,he six lowfield, exchangeable proton resonances are due to hydrogen-bonded protons. The solvent exchange behavior of hydrogenbonded protons has been studied extensively (Englander et al., 1972: Wagner: 1983). It is achieved by breaking and reforming the hydrogen bonds. For an exposed, freely accessible hydrogenbonded proton at t’he surface of the protein, the solvent exchange rate is, in general, too fast, for the prot’on resonance to be observable. However, for an internal hydrogen-bonded proton, the solvent accessibility will be controlled by the fluctuation rat,e of the molecular conformation and thus, the exchange ra,te should be slower. Tn particular, we believe that for the case of the GlnBP-Gln complex, the exchange rat’e would be primarily the rate of the ligand on/off process, which modulates the protein conformation. From the ligand-titration experiment illustrated in Figure 1. it is apparent that the resonance at’ + 6.44 p.p.m. from H,O in the ligand-free GlnBP shifts upfield t,o + 6.33 p.p.m. from H,O when r,-Gln is bound, and both free and bound proteins co-exist in slow exchange until the molar ratio reaches 1 : 1 and only the bound form is present. As the magnit’ude of the shift is 0.11 p.p.m. or 33 Hz at 300 MHz, the dissociation rate should be significantly less than 33 Hz, i.e. the lifetime of the ligand-protein complex should be longer than 30 ms. The resonance at +6.33 p.p.m. from Hz0 exchanges very slowly; it can be observed for several hours in *H,O at room temperature. It also has an almost zero temperature coefficient (Table 2), so that it must belong to a proton buried inside the prot,ein, with restricted solvent accessibility. This resonance has been assigned to the indole NH of the Trp220 by site-specific mutagenesis (Shen, 1988; Shen et al., 1989). Of the low-field resonances, only the resonance at +687 p.p,m. from H,O yields a saturation transfer to the ligand resonances (Fig. 3), especially to the fi and y protons of L-Gln. It is int,eresting to note that’ there appears to be no shift of the L-Gln resonances upon binding, as suggested by the saturation transfer experiement performed with a 1 : 1 molar ratio of protein to ligand. In this case, there is essentially no free ligand, as the KD value is 5 x lop7 M. The saturation transfer to the free glutamine molecule occurs in two steps: a negative NOE is generated from the resonance at +6+37 p.p.m. from H,O to the B and y protons of the bound glutamine (and, to some extent, through spin diffusion to the a proton); the glutamine molecule is then released and another glutamine molecule is bound in its place. The intensities of the aliphatic

et al. proton resonances of t hc free I,-(:ln rnolec~ulr arc’ thus reduced. The fat+ t’hat, the saturation transfer is most effective for the fi and ;’ protons indicates that the resonancae at +6%7 p.p.m. from H,O is closer in space t’o t,he 1’ end of the bound glutaminca than to the C( end. Since there is no splitt,ing of the resonance at +6+%7 p.p.m. from H,O when using I%-y-amide-Gin, the only possibility is that this resonance is a hydrogen bond to the y-carbony of the glut,amine. Further support is found in Figure 2. which shows that when eit~her of the two glutamine analogs: y-glutamylhydrazide or y-glutamylhvtlroximate, is added in place of I,-Gln. all six new resortantes appear. but are slight,ly shifted in clomparison with t’he ClnKP-(Zln complex. Tn cont,rxst with the ot,her five resonances. the resonance at + 6.87 p.p.m. from H20 is shifted appreciably t,o hi@r field (i.e. from +6%7 to t6.60. or t,o +6.34 p,p,m.. rcspcc+ ively). This is an indication of a weaker hvdrogerr bond (Emsley. 1980) and correlat’es well with t,hr smaller afinity of the protein for these ligands. Ax the resonance at +6.87 p.p,m. from H,O is sensitive to substitutions at the y- amide. it, is most likely an int,ermolecular hydrogen bond bet,ween the binding protein and t.hr y-carbonyl of the bound ligand. Though the involvement of hydrogen bonds in the binding events between periplasmic binding proteins and their ligands has been claimed in several cases, to the best of our knowledge there arc no reports of a, direct observation of such bonds in aqueous solution. except, as recently report’ed by our laboratory (Kuckel it al.. 1989). Our ‘H-n.m.r. studies show that the formation of intermolecular and int,ramolecular hydrogen bonds a.lso occurs when I,-hist,idine binds to the histidine-binding protein J from Salmonella typhimurium. It has been inferred from X-ray diffraction stlldies that cxtlensive intermolecula,r and intrabonds are formed between molecular hydrogen I,-arabinose and arabinose-binding protein from E. coli (Quiocho 8 Vyas, 1984), and between the unsolvated sulfate ion and it,s binding protein from R. typhkurium (Pflugrat,h & Quiocho, 1985) in the crystalline state. It appears t,hat the formation of intermolecular and intramolecular hydrogen bonds may have a general significance for the specific binding between periplasmic binding proteins and their respective ligands. Since hydrogen bonds are directional, especially t,hose involving NH protons (Baker & Hubbard. 1984), it can be envisaged that. the intermolecular hydrogen bonds confer geometrical specificity on t)he binding sites in t~hr binding proteins. therefore ensuring the correct fit) for the ligands in t)heir binding sites (perhaps involving a lock-and-key mechanism). The formation of intramolecular hydrogen bonds reflects specific conformational changes in the protein molecule and may stabilize certain conformations of t,he ligand-bound binding proteins required for the subsequent, ligand translocation process. In conclusion. GlnBP binds strongly and with high specificity its natural ligand, L-Gln. Upon binding, the protein undergoes conformat,ional

L-Glutamine-induced

Coqformational

changes as evidenced by alterations in the ringcurrent-shifted and aromatic-shifted proton resonance regions as well as the formation of several hydrogen bonds detectable by ‘H-n.m.r. spectroscopy as six resolved, low-field-shifted, solventexchangeable proton resonances. One of them, the resonance at + 637 p.p.m. from H,O, is an intermolecular hydrogen bond, while the other five appear to be intramolecular hydrogen bonds that stabilize the conformation of the GlnBP molecule in the ligand-bound form. Since the appearance of the six low-field proton resonances is little affected by changing the pH and temperature, we believe that these exchangeable proton resonances arise from intermolecular and intramolecular hydrogenbonded protons which are internal and their local environment quite stable. Earlier biochemical studies have shown that for a comparable pH range, and an even wider temperature range, the L-Gln binding activity of GlnBP as measured by the KD does not change significantly (Weiner & Heppel, 1971). In light of our ‘H-n.m.r. results, this functional stability can now be understood as due to an additional structural stability in GlnBP upon binding of L-glutamine. We thank Dr Jen-Shiang Hong for sending us strain BKSMDG with the plasmids pGPl-2, pJWlO7 and pJW133; and Dr Leon A. Heppel for sending us strain GlnP, needed for this work. We also thank Dr Susan R. Dowd and Dr Hoai-Thu N. Truong for helpful discussions. This research is supported by research grants from the National Science Foundation (DMB-8816384) and the National Institut,es of Health (GM-26874). Preliminary results were presented at the 29th Annual Meeting of the Biophysical Society, 24 to 28 February 1985, Baltimore, MD, U.S.A. This work is taken from the Ph.D. thesis of Q.S.. Carnegie Mellon University, August 1988, Pittsburgh. PA, U.S.A.

Resolution Nuclear Magnetic Resonance Spectroscopy, vol. 1, pp. 534-551, Pergamon Press, Oxford. Englander, S. W., Downer, N. W. & Teitelbaum, H. (1972). Annu. Rev. Biochem. 41, 903-925. Freeman, R., Kempsell, S. P. & Levitt. M. H. (1980). J. Magn.

Ames, G. F.-L. (1986). Annu. Rev. Biochem. 55, 397425. Ames, G. F.-L., Prody, C. & Kustu. S. (1984). J. Bacterial. 160, 1181-1183. Baker. E. N. & Hubbard, R. E. (1984). Progr. Biophys. Mol. Biol. 44, 97-179. Becker, E. D. (1969). High Resolution n.m.r.: Theory and Chemical Applications. pp. 232-235, Academic Press, New York. Blomberg, F., Maurer, W. & Riiterjans, H. (1976). Proc. Nat. Acad. Sci., U.S.A. 73, 1409-1413. Buckel, S., Cottam, P. F., Simplaceanu, V. & Ho, C. (1989). J. Mol. Biol. 208, 477-489. Deslauriers, R. & Smith, T. C. P. (1980). Biol. Magn. Reson.

2, 243-344.

Emsley, J. (1980). Chem. Sot. Rev. 9, 91. Emsley. ,J. W.. Feeney. ,J. & Sutcliffe, L. H. (1967). High

Reson.

38, 453479.

Furlong, C. E. (1987). In Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F. C., Ingraham, J. L., Low. K. 13.. Magasanik. B., Schaechter, M. B. & Cmbarger, H. E. American Society of eds), vol. 1, pp. 768-796, Microbiology. Washington, D.C. Hoekstra, W. P. M., Storm, P. K. & Zuiderweg, E. M. (1974). Mutat Res. 23, 319-326. Hunt, A. G. & Hong, J.-S. (1983). Biochemistry, 22. 851-854. Jardetzky. 0. & Roberts, G. C. K. (1981). n.m.r. in Molecular Biology, Academic Press, New York. Keller, R. M. & Wiithrich, K. (1981). Biol. Magn. Reson. 3, l-52. Kreishman, G. P., Robertson, D. E. & Ho. (‘. (1973). Biochem. Biophys. Res. Commun,. 53, 18-23. Lever. ,J. E. (1972). Anal. Biochem. 50. 73-83. Lowry, 0. H., Rosenbrough, N. J., Farr, A. I,. & Randall, R. J. (1951). J. Biol. Chem. 193, 265-275. Marty, B., Gaudin. C., Ragot, M. & Belaich. ,J. (1979). Biochem. Biophys. Res. Commun. 86, 1118-l 125. Mason, J. (1987). Editor of Multinuclear n.m.r., pp. 173174, Plenum Press, New York and London. Matthews, R., Matthews, K. & Opella, 8. cJ. (1977). Biochim. Biophys. Acta, 497, l-13. Miller, D. M., Olson, J. S., Pflugrath, J. W. & Quiocho, F. A. (1983). J. Biol. Chem. 258, 1366513672. Miller, J. H. (1972). In Experiments in Molecular Genetics, pp. 352-355, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Neu, H. C. & Heppel. L. A. (1965). J. Biol. (Them. 240, 3685-3692.

Nohno,

T., Saito, T. & Hong,

Genet.

205,

Pflugrath,

J.

(London),

Plateau, Reson.

References

857

Changes in GlnBP

J.-S. (1986).

Mol.

Gen.

26&269.

W.

& Quiocho,

314,

257-260.

P., Dumas,

F.

C. & Gueron.

A.

(1985).

Nature

M. (1983). d. Magn.

54, 4653. F. A. & Vyas. N.

Quiocho, K. (1984). Nature (London), 310, 381-386. Rosner, J. L. (1972). Virology, 49, 679-687. Shen, Q. (1988). Ph.D. thesis, Carnegie Mellon ITniversity. Pittsburgh, PA. Shen, Q., Simplaceanu, V., Cottam, P. F., Wu. J.-L., Hong. J.-S. & Ho, C. (1989). J. Mol. Riol. 210, 854-867. Tabor, S. & Richardson, C. C. (1985). Proc. Nat. Acad. Sci., U.S.A. 82, 1074-1078. Taylor, A. L. & Trotter, C. D. (1967). Hacteriol. Rev. 31, 332-353. Wagner. G. (1983). Quart. Rev. Biophys. 16. l-5i.

Weiner, J. H. & Heppel, L. A. (1971)..I.

Biol.

Chem.

246,

6933-6941. Willis, R. C. & Seegmiller, J. E. (1976). Anal. Riochem. 72, 6&77. Wiithrich, K. (1986). In n.m.r. in Protein and Nucleic Acids, pp. 37-39. Wiley Interscience. New York.

Edited by B. W. Matthews