Sugar Transport by the Bacterial Phosphotransferase System

2 downloads 0 Views 837KB Size Report
cies are functionally different and that their intercon- version may regulate sugar transport via the PTS. The. C-terminal Cys of Escherichia coli E1 was reacted ...
Vol. 269, No. 32, Issue of August 12, pp. 20263-20269,1994 Printed in U.S.A.

THEJOURNAL OF B I O ~ I CCHEMISTRY AL 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Sugar Transport by the Bacterial Phosphotransferase System CHARACTERIZATION OF THE ESCHERICHIA COLI ENZYME I MONOMERDIMER EQUILIBRIUM BY FLUORESCENCE ANISOTROPY* (Received for publication, January 6, 1994, and in revised form, April 12, 1994)

Franqoise ChauvinS, Ludwig Brand§, and Saul Rosemann From the McCollum-Pratt Institute, the Department Baltimore, Maryland 21218

of

Biology, The Johns Hopkins University,

Enzyme I (EI), the first proteinof the bacterial phos- sugar uptake for about 1.5 min, followed by essentially no uptake for about 2 min, followed by uptake ata rate that the cells photransferase system (PTS), exists in a monomer/dimer (M/D) equilibrium. We have proposed that the two spe- can handlemetabolically (6). ( b ) Transport of the non-metabocies are functionally different and that their intercon- lizable glucose analogue, methyl a-glucoside, by whole cells of version may regulate sugar transport via thePTS. The Salmonella typhimurium (7) and Staphylococcus aureus (8)exC-terminal Cys of Escherichia coli E1 was reacted with hibits similar kinetics. A brief, initial rapid rate of uptake is pyrene maleimide (Han,M. K., Roseman, S., and Brand, followed by a continuous decline in the rate until theconcenL. (1990) J. Biol. Chem. 265,1985-1995), and the pyrene tration of internalsugar-phosphatereaches a steady-state conjugate used to characterize the M/D equilibrium by level, about 107-foldbelow the calculated equilibrium state. ( c ) fluorescenceanisotropy.Theproperties of unlabeled When methyl a-glucoside is taken upby S. typhimurium memand pyrene-labeledE1 are indistinguishable.Values for brane vesicles, a system withfewer variables, similar data are the apparent association constant,K&, and the steady- obtained, and thereis no obvious explanation for the results(9). state anisotropy of the monomer and the dimer were The PTS is an ensemble of proteins (four in the caseof gluobtained under a variety of conditions. K& increases 23- cose transport) that transfer the phosphoryl group from PEP to fold, from 0.45 x lo6 to 10.7 x 10' M", as the temperature the sugar, and in addition, regulate other important and diincreases from 6 to 30 "C; the association appears to be K&, verse physiological processes. We have suggested (1) that the entropically driven. Under all conditions tested, the f o r phospho-E1 is 6 1 2 - f o l d less than for dephospho-EI. PTS inEscherichia coli and S. typhimurium may itself be regulated by Enzyme I, the first protein in the phosphotransfer Forphospho-EI,PEPand Mg2' inducea240-foldincrease of K& when both ligands are present. Based on sequence. This enzyme is cytoplasmic, with monomer molecuthese data, E1 was preincubated under conditions that lar weights of 63.5 and 63.2 for the E. coli and S. typhimurium change KL,,and theinitial activitiesof the different spe- proteins, respectively (10, 11).The protein reversibly dimerizes cies were determined at37 "C in a PTS sugar phospho- (12-17), and it has been suggested but notproven that only the rylation assay with PEP as the phosphoryl donor. The dimer is competent for accepting the phosphoryl group from initial rate depends on M/D the ratio; itis maximal when PEP. The phosphomonomer may be the catalytic unit that E1 is 100% dimer, and zero when E1 is 100% monomer. In transfers this group t o HPr, the next protein in thesequence. the latter case, the rate gradually increases in the assay Enzyme I has also been shown to interact with other cellular mixture.Theresultshaveimportantimplicationsfor proteins, acetate kinase (18), and an ATP-dependent kinase how the PTS regulates sugar transport and other phys-that is regulatedby NAD+/NADH (19). iological phenomena. Thus, themonomer/dimer transition of Enzyme I may playa central role in regulating a wide variety of processes, from sugar transport to the transcription of certain catabolic operons The net reaction for the uptake of sugars and sugar anato chemotaxis toward PTS sugar substrates.Physiological syslogues via the bacterialphosphoeno1pyruvate:glycosephospho- tems exist with the potential for testing thevalidity of hypothtransferase system as is follows. eses on the importance of the monomer/dimer transition. For example, naturalbacterialmembrane vesicles accumulate Sugar,, + PEP, e sugar-Pi,+ pyruvate, methyl a-glucoside 6-phosphate, but not when the membranes REACTION 1 are isolated from mutants deleted in the structural genes that The K& for this reaction is - los, and,as discussed by Weigel et encode Enzyme I or other PTS proteins. However, if the missing al. (1) and in recent reviews (2-5), the enormous thermodyhomogeneous PTS proteinsareinsertedintothe deficient PTS be vesicles, they become fully capable of taking up and phosphonamic driving force provided by PEP requires that the rigorously controlled. A few examples follow. ( a )When glucose rylating the sugars(20). The role of the monomer/dimer tranis added to starvedStreptococcus lactis cells, there isa burst of sition could be determined ina system of this typewith a well characterized spectroscopic method for making the measure* The costs of publication of this article were defrayed in part by the ments. payment of page charges. This article must therefore be hereby marked The monomer of Enzyme I contains a C-terminal Cys residue "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to 3 additional Cys residues, all located in the C-terminal and indicate this fact. domain (about half) of the molecule (10, 11).The C-terminal $ Supported by National Science Foundation Grant DIR 8721059. 8 Supported by National Institutes of Health Grant GM11632. Cys reacts more rapidly with-SH reagents than do the other3 ll Supported by National Institutes of Health Grant GM38759. Cys residues (151, and thisdifference in reactivity permits speThe abbreviations used are: PTS, the phosphoeno1pyruvate:glycose phosphotransferase system; PEP, phosphoenolpyruvate; HPr,histidine- cific labeling of the C-terminal-SH group with reagentssuch as 5,5'-dithiobis-(2-nitroben- pyrene maleimide. containingphosphocarrierprotein;DTNB, zoic acid); DTT, dithiothreitol;EI, Enzyme I; M/D, monomer/dimer. In the presentpaper, we show that E.coli Enzyme I and the

20263

The MonomerlDimer Equilibrium

20264

derivatizedenzymebehaveidentically underall conditions tested (activity in sugarphosphorylation, monomer/dimer transition, etc.). Emission anisotropy of the pyrene-labeled enzyme is shown to be suffkiently sensitive to estimate the monomer and dimer concentrations in equilibrium mixtures. The equilibrium constant was therefore determined under a variety of conditions relevant to those used for sugar phosphorylation assays in vitro and in vesicles. This method also allowed us to determine the kinetics of the transition, as shown in the accompanying article (21). MATERIALSANDMETHODS DTNB was purchased from Sigma, pyrene maleimide from Molecular Probes. HPr was a gift from P. Coyle, and I1IGicwas from Dr. Norman Meadow. Membranes from the deletion mutant S. typhimuriurn SB2950 were prepared as described (22). Enzyme I was purified from an overproducing strain of E. coli (lo), according to a procedure described previously (22). The protein concentration was determined by the Lowry method (23), using bovine serum albumin as a standard,and by DTNB titrations (4 cysteinedmonomer), using the methods of Ellman (24) and Habeeb (25) as described by Han et al. (15). In all cases, the "dimer fraction" ( f D ) and "monomer fraction" (f,) relate to the proportion ofprotornersinvolved in the species, so t h a t f, + f, = 1. Enzymatic activity was determined as described by Weigel et al. (22), in the presence of the other PTS proteins, where Enzyme I was rate-limiting. The effect of the phosphorylation state of Enzyme I on the equilibrium constant could be determined because phospho-Enzyme I is stable for a t least 48 h in theabsence of the phosphorylating ligands MgZ' and PEP a t 0 "C.The enzyme was phosphorylated by incubation at 37 "C in the presence of 2 mM PEP and 5 l ~ MgZ' l ~ for 30 min. The dephosphorylated form of Enzyme I was obtained by dialysis overnight at 4 "C in the presence of 10 nm pyruvate. The extent of completion of the phosphorylation and dephosphorylationreactions was checked bythe lactatedehydrogenaseassay described by Waygood etal. (27); it was at least 95% in all cases. Derivatization of Enzyme I-The protein was specificallyderivatized at the C-terminal cysteine with pyrene maleimide. The labeled enzyme was chromatographed on ACA44 filtration gel at room temperature to separate the singly reacted form, which is able to dimerize, from the multiply reacted molecule, which is unable to dimerize (15). The sepabuffer, 100 mM KC1,5 mM MgCI,, l mM EDTA, ration was in 100 mM 2.5 mM DTT, pH 6.5. Unless stated otherwise, the conditions for labeling, isolating, and characterizing the enzyme wereas described (15,26). The stoichiometry of labeling was determined by DTNB titration of the remaining free thiols and yielded a value of 1 (+ 0.1) pyrene moiety/ subunit. The enzymatic activity of the labeled enzyme was indistinguishable from that of the unlabeled enzyme. The minimal buffer used throughout this study was 10 mM KP,, 100 mM KCI, 1mM EDTA, 0.2mM DTT,pH 6.5 (buffer B).In most ofthe experiments, the following ligands were alsopresent at the indicated concentrations:5 l ~ MgCl, l ~ and 2 m M PEP. The complete buffer was called bufferA. The solutions were preincubated for at least 45 min beforethe fluorescence measurements, in order to allow the system to reach equilibrium. Fluorescence Measurements-Steady-state fluorescence data were collected with a SLM 8000 photon counting spectrophotometer at controlled temperature. The absorbance was less than 0.1 at the wavelength of excitation to avoidinner filter effects and multiple excitations. The spectra were obtained under magic angle conditions to avoid polarization artifacts. The steady-state anisotropy, rae,is defined as shown by Equation 1, where I , and I, are, respectively, the parallel and perpendicular components of the polarized fluorescence light and G is a correction factor for the different instrumentresponse to the two polarized components.

KP,

r, =

I,, - G.I, I,, + 2.G.IH

(Eq. 1)

The fluorescence anisotropy was recordedin the T arrangement (simultaneous recording of both parallel and horizontal polarized emitted light). Pyrene was exciteda t 337 nm(8-nm slits). The fluorescence light was recorded in the photon counting mode after passage through two matched color filters (CS 5-59, half-maximum transmittance at 380 and 480 nm). The count rate was always higher than 10,000 countds; 61 readings of I , and IHwere used to calculate a given anisotropy value. The G factor was determined using either pyrene maleimide coupledto 2-mercaptoethanol in methanol as a fast rotating probe or with the

of

Enzyme I of the PTS

0.1 3

0.12

>-

R

0

sz

0.11

4

0.10

0.09

1

5

1 00

10

[Ell

I

1000

(pdml)

FIG.1. Determination of K:, by steady-state anisotropy. The fluorescence anisotropy of pyrene was measured with mixtures of pyrene-labeled EnzymeI and unlabeled Enzyme I a t 6 "C. For concentrations ranging from 6 to 10 pg/ml, only pyrene-labeled Enzyme I was present. At higher concentrations, 10 pglml pyrene-labeled Enzyme I was mixedwith unlabeled protein to give the indicated final concentration. Phosphorylated Enzyme I was in buffer A. The solutions were equilibrated at 6 "C for at least 1 h before determining the anisotropy. protein solution, excited with horizontally polarized light. Using either determination, G = 0.3.The anisotropy of the 2-mercaptoethanoladduct was 0 + 0.005. In the binding isotherm experiments, the G factor was determined for each data point. Typically, it did not vary outside the range 0.299-0.305. RESULTS

Determination of the Equilibrium Constant of the Monomer1 Dimer Dunsition by Steady-state EmissionAnisotropy-While it is desirable to obtain equilibrium data over as wide a concentration range as possible, fluorescence measurements are most conveniently done over a limited concentration range. At very high concentrations there are errors due to inner filter effects, and atvery low concentrations the signal tonoise ratio decreases. In orderto avoid these problems, the pyrene-labeled Enzyme I was used at the optimum concentration range for fluorescence measurements, while the desired change in enzyme concentration was achieved by addition of unlabeled enzyme. In order to use this protocol, it wasnecessary to establish that the equilibrium behavior of the pyrene-labeled Enzyme I was identical to that of unlabeled Enzyme 1. Evidence that this is indeed the case is presented later in thispaper. Fig. 1gives fluorescence anisotropy as a function of increasing concentration of phospho-Enzyme I in buffer A. Except for the temperature, 6 "C, these are theconditions used t o obtain optimal Enzyme I activity in the sugarphosphorylation assay. Pyrene-labeled EnzymeI concentration was variedfrom 6 to 10 pg/ml, a range that gives reliablesignals. In samples with higher concentrations of Enzyme I, 10 pg/ml pyrene-labeled E1 were present and unlabeled Enzyme I was added to the final concentration. The fluorescence anisotropy was measured in the protein concentration range of 6 pg/ml to 1 mg/ml after equilibration for at least 1 h at 6 "C; the emission anisotropy remained constant after 60 min. The datashown in Fig. 1were used to calculate the valuesof the equilibrium constant, K&, and of the steady-state fluorescence anisotropy values of the monomer and thedimer, rMand r-,, respectively. As shown below, the monomer-dimer ratio does not modify the fluorescence intensity properties of the pyrene derivative. Therefore, the contributions to the observed fluorescence anisotropy, r,,, of the monomer and dimer species are given by their respective molar fractions, fM and f-,,as shown by Equation 2

of Enzyme I of the PTS

The MonomerlDimer Equilibrium = (fM.rM) + (fD'rD)

2) (Eq.

If K:q is the apparentequilibrium constant for dimer formation, then K:, = [Dl/[M12

(Eq. 3)

TABLEI Fluorescence intensity and emission anisotropyof mixtures of unlabeled and pyrene-labeled EnzymeI Mixtures of unlabeled and pyrene-labeled Enzyme I in theindicated proportions wereprepared so that the total Enzyme I concentration was constant a t 36 pg/ml. The mixtureswere incubated a t 23 "C in buffer A for 2 h priorto the measurement. The expected dimer content was 80%.

The total Enzyme I concentration (calculated as protomer) is shown below. [EII, = [MI + 2.[Dl

0 = 2.Kq.[MI2+ [MI - [EII, fM

-1 + fM

=

=(Eq. [Ml/[EIl,

vl +

0.11 0.17 0.33 0.50 1.0

(Eq. 5) 6)

8.K&.[EIIt

4.Kq.[EIl,

Fluorescence intensity

Fraction of pyrenelabeled Enzyme I

(Eq. 4) 90.7

20265

'.OO

1

(Eq. 7)

31.7 48.8 136.7 257.4

Emission anisotropy

0.131 0.129 0.128 0.126 0.128

1 I

I'

The values of K&, r,, and rDwere obtained with theprogram NONLIN, a gift from ProfessorMichael Johnson (Universityof Virginia). The fitting function used is given by Equation 8.

0.80

-

/\

1 ,

.--0.60 c

i\

-

v)

I

J

I

!

When phospho-Enzyme I is at 6 "C and pH 6.5 in thepresence of 5 mM MgC1, and 2 m~ PEP, the values of K&, r,, and r, are 3.3 x lo6M-I, 0.084, and 0.131, respectively. The 67% confidence intervals are 1.3 x lo6 to 5.7 x lo6 for K&, 0.074-0.090 for r,, and 0.129-0.132 for rD. An idea of the reliability of these values can beobtained by I comparing the uncertaintyof the anisotropy determination, the 360 400 440 480 520 width of the 67% confidence intervals, and their reproducibilWavelength (nm) ity; a given anisotropy value was determinedfrom the average FIG.2. Fluorescence emission spectrum of pyrene-labeled Enof 61 datapoints giving a typical standard deviation of 0.0035. zyme I. The emission spectrum was recorded a t 6 "C for samples conWhen the entire experiment (raSas a function of Enzyme I taining pyrene-labeled Enzyme I only. The concentrations of Enzyme I of K& ob- were 4.1 pg/ml (continuous line) and 1.85 mg/ml (dashed line). The concentration) was repeated three times, the values tained by the analyses were 3.3 x lo6, 2.0 x lo6, and 4.0 x lo6 dimer content was 23 and 93%, respectively. The excitation wavelength was 337 nm, and bandwidthswere 8 nm for the excitation and 4 nm for M-I, respectively, giving a n experimental standard deviation of the emission. No pyrene excimer was formed in the concentrated pro1.0 x lo6 M-I, less than thehalf-width of the 67% confidence tein sample, asevidenced by the lack of any broad peak in the470-500 interval (1.5-2 x lo6 "I). The values of r, were 0.131, 0.130, nm region. and 0.135, respectively, giving a n experimental standard deviation only slightly greater than the half-width of the 67% dimer was changed employing only pyrene-labeled enzyme. confidence interval. For r,, the values were 0.084, 0.091, and Fig. 3 shows that fluorescence intensity is directly proportional 0.083. to the probe concentration, both at 23 "C (A) and 6 "C ( B ) ,and K& are nearly in- is independent of the dimercontent (f, = 0.60.96 at 23 "C, and The derived fitted parameters rM, rD, and dependent of the valuesof the initial guesses used in the fitting f, = 0.35-0.8 at 6 "C). program; the initiallyguessed K& can vary from 1 x lo4 to 1 x Thus, the fluorescence intensity of pyrene conjugated to the lo9 M-I. The guessed values of the anisotropy, r,, can be as low C-terminal cysteine of Enzyme Idoes not change as the enzyme as 0.05 and r, as high as 0.3for K&between 1x lo5and 1x lo8 associates, and Equation 2 is validated. "I. In view of the simplicity, rapidity, and potential of the emisThe Influence of Pyrene Labeling on the Propertiesof Enzyme sion anisotropy method, it wasapplied to studies of the effects Z-In order to determine the effect of pyrene labeling on the of temperature, ligands, state of phosphorylation, and pH on monomer/dimer equilibrium, it was shown that when the total K&. enzyme concentration was constant, the steady-stateemission Effect of Emperature-Table I1 gives K& and the emission anisotropy of pyrene was independentof the fraction of labeled anisotropy of the monomer and the dimer of dephospho-EnEnzyme I; data are given in Table I for a ratio [pyrene-labeled zyme I (without ligands) at four temperatures. Since the visEIl/[EIl, varying between 0.1 and 1.0 at a total Enzyme I con- cosity decreases with increased temperature,r, and r, should centration of 36 pg/ml, at 23 "C (expected dimer content: 80%). likewise decrease. The expected results were obtained in the Thus, labeled and unlabeled subunits form dimers with equal experimental temperature range, 6-30"C; rMchanged from affinity. 0.104 t o 0.081 and rDfrom 0.166 to 0.119. K& increased by As noted above, it was essential to determine thevalidity of 23-fold in this temperature interval. The valuesof K& versus the simplifying assumption made in Equation 2. The fluores- temperature permitted calculation of the free energies of the cence of pyrene was therefore studied as a function of the transition andof the changes in enthalpy and entropy (Fig. 4). monomer/dimer content. The emission spectrum of pyrene-la- The correlation coefficient was 0.995. AH and A S are both posibeled Enzyme Idoes not change upon dimerization,as shown in tive, 21kcal.mo1" and 98cal.mol".K", respectively. These valFig. 2. Specifically, no excimer band at 470-490 nm is seenfor ues agreeboth in theirsign and range with the values obtained the dimer. In addition, the fluorescence intensity of pyrene- by Kukuruzinskaetal. with S. typhimurium (13) and by labeled Enzyme I was measured as the fraction of monomer/ Neyroz et al. (14) with E. coli Enzyme I.

6

The MonomerlDimer Equilibrium

20266

of Enzyme

I

of

the PTS

Me,

FIG.5 . Effects of phosphorylation, and pH on the monomer/dimer transition at 23 “C. Steady-state anisotropy was [WRME-EIl (pdml) measured as a function of Enzyme I concentration: 0,dephosphorylated FIG.3. Pyrene fluorescence intensity is not modified by dimer-enzyme; A, phosphorylated enzyme. In each experiment, the total pyization of Enzyme I. The total steady-state intensity of pyrene fluo- rene-labeled enzyme was constant (from 6 to 12 pg/ml), and unlabeled rescence ( I , = I, + 2 x G x I”) is shown as a function of pyrene-labeled protein was added to the indicated concentrations. The data were colEnzyme I concentration. Only pyrene-labeled Enzyme I was used. Flu- lected and analyzed as described under “Results”to obtain K&, r,, and orescence intensity is directly proportional to pyrene concentration at rD.The two latter values were used to calculate f D , the dimer fractions 23 “C (panel A ) and at 6 “C (panel B ) ; the fraction of dimer, fd, varied shown in panels A , B , and C , where f, = [rss- rM]/[rD - rMl.A, buffer B. from 0.6 to 0.96 and from 0.35 to 0.8, respectively. B , same as A + 5 mM MgC1,. C , same as B , except pH = 7.5. D , total fluorescence intensity, It, corresponding to the anisotropy values in C . I , TABLEI1 is independent of the monomer/dimer fraction. This result is typical of Temperature dependence of the monomerldimer equilibrium all conditions studied. Dephospho-EnzymeI was in buffer B. At each temperature, the values of Ke’, the association equilibrium constant, rMand rD,the anisoAn example of the method used to test thevalidity of using tropy of tKe monomer and dimer species, respectively, were determined Equation 8 is given in Fig. 5. If the quantum yield and the using at least 15 different Enzyme I concentrations. The 67%confidence lifetime of pyrene do not vary with dimer formation, the total intervals are given in parentheses below each value. 0

50

100

150

200

250

300

fluorescence intensity, It,is expected t o be constant when the quantity of labeled protein is maintained constant throughout the totalprotein concentration range. Fig. 5 0 shows that at pH “C “1 7.5, the total fluorescence intensity does, in fact, remain con6 0.166 (0.260.69) (0.103-0.106) (0.153-0.177) stant. This was checked for each condition studied and the 0.093 10 1.2 0.135 results were comparable to that shown. Thus, fluorescence is (0.129-0.141) (0.7-1.7) (0.092-0.094) independent of the monomer/dimer state of EnzymeI and 0.134 23 5.9 0.0975 Equation 8 applies. (0.131-0.137) (3.68.9) (0.096-0.099) 0.081 0.119 The effects of phosphorylation state and ligand presence on 30 (0.116-0.120) (6-18) (0.077-0.084) K&at 23 “C are shown in Fig. 5.The conditions chosen for these experiments are relevantto those used in the sugar phosphorylation assay in vitro and in vesicles. In these experiments, the anisotropy values were collected as shown in Fig. 1and the values of KAq, r,, and rDwere obtained as described above. The dimer fraction is shown for clarity. Indeed, ligand presence also affects the values of rM and rD (see below). In each case, the cr dimer fraction was calculated from the observed anisotropy ( f D Q Y = [rBa - rM]/[rD - r M ] )The . line is the theoretical dimer fraction, as given by the valueof KLq obtained from analysis of the data. -00, These experiments were done in theabsence of PEP. However, there was less than5% hydrolysis of the phosphoprotein over the time course of the experiments, as checked by the lactate dehydrogenase assay. The results in Fig. 5 may be summarized asfollows. (i)K&for 3.6 3.2 the dephosphoenzyme was 6-12-fold greater thanfor phosphoEnzyme I.(ii) Mg2‘ increased K& 7-8-fold for both phospho- and dephosphoenzymes. (iii) Therewere only small changes in KLq FIG.4. Effect of temperature on the monomer/dimer transifor both the phospho- and dephosphoenzymes as the pH was tion. The data in Table I1 are plotted. Dephospho-Enzyme I was in (iv) The largest buffer B. The estimated values of AH and A S for the transitionare given varied from 6.5 to 7.5 in thepresence of under “Results.” effect on KLq was observed (Fig. 7) when 2 m~ PEP was added to the phosphoenzyme in thepresence of M e . PEP increased KLq 30-fold. That this ligand promotes dimer formation sugEffects ofPhosphorylation, Ligands, and pH on K&-Before be one or more PEP binding sites in the determining KLqunder different conditions, it was important to gests that there must show that the fitting function, Equation 8, applied as these phosphoprotein. The PEPeffect was also observed at 6 “C, but neither 2 mM PEP nor 4 mM pyruvate had any effect on the perturbations were introduced into the system. Temperature

&X

‘M

TD

Me.

20267

The MonomerlDimer Equilibriumof Enzyme I of the PTS TABLE I11 Effects of ligands, phosphorylation, and pH on Kjq and emission anisotropy of the monomer and dimer ut 23 "C Incubationswere in buffer B at the indicated pH and inthe presence of the ligands as shown. The values of K&, rM,and rDwere determined using at least 12 different EnzymeI concentrations. The67%confidence intervals are given in parenthesesbelow each value of the equilibrium constant. State of Enzyme I"

pH

Dephospho.

6.5

0

Dephospho.

6.5

Dephospho.

rD

gqX 10.~

0.097

0.134

5

0.093

0.124

7.5

5

0.095

0.130

Phospho.

6.5

0

0.101

0.154

Phospho.

6.5

5

0.105

0.135

Phospho.

6.5

0.078

0.140

5.9 (3.6-8.9) 44 (15-100) 31 (14-59) 0.95 (0.54-1.4) 8 (613.8) 240

Phospho.

7.5

0.118

0.145

[MgCIJ

TM

mM

5

+ 2 mM PEP 5

M"

(0-1000) 2.7 (1.4-4.4)

Dephospho., dephosphorylated; phospho., phosphorylated.

2

0

4

6

8

10

TIME (MIN)

FIG.6. Enzymatic activity as a function of the initial stateof phospho-Enzyme I. Theactivity of the enzymewasassayed by measuring the rate of [14C]methylglucosidephosphorylation as described under "Materials and Methods," with Enzyme I rate-limiting, but at the same concentration in each assay mixture. The monomer/ dimer levels of the two samples were altered by preincubating eachat the following concentration and temperature: A, 37 "C, 1.5 mg/ml, 15 min (all dimer at time zero);0, 1"C, 7 pg/ml, 4 h (all monomer at time zero). Preincubations were in buffer A in the presence of 0.1 mg/ml bovine serum albumin.

dephosphoprotein in the absence of Mg2' (data not shown). I t 1 .o seems likely that the dephosphoprotein only forms a PEP binding site afterit complexes with Me. 0.8 Table I11 summarizes the values of KLq and the emission anisotropy of the monomer and dimer forms under these dif/ A I ferent conditions. The spread in the rMand rDis significantly 0.6 I wider than the67% confidence intervals: (a)The anisotropyof the phosphorylated protein tends to be higher than thatof its 0.4 dephosphorylated counterpart, ( b ) magnesium tends to cause ( c )a n increase in pH causes the the anisotropy to decrease, and 0.2 anisotropy of the phosphoprotein to increase. The change in anisotropy suggeststhat the ligands notonly shift KAq,but also cause changes inconformation of the mono0.0 1 10 100 1 000 mer and dimer, These changes are important in interpreting the kinetics and mechanismof dimerization. Effect of Equilibrium on Activityof E n z y m e I-We have suggested that themonomer/dimer transition plays an important role in regulating the PTS (1, 3, 4, 12, 13, 26). This idea is supported by the results inFigs. 6 and 7. In Fig. 6, the kinetics of sugar phosphate formation by the complete PTS system was measured at two initial states of Enzyme I, one where it was it was dimer. In almost entirely monomer and the other where each case thelevel of total Enzyme I in the assay mixture was the same (1.25 pg/ml) and rate-limiting. Phospho-Enzyme I in the presence of Mgz+and PEP was dissociated to virtually100% monomer by preincubation at low temperature andlow concentration (1"C, 7 pg/ml, lower curue) and associated to virtually 100% dimer (upper curue) by preincubation of a 1.5 mg/ml 0.0 I I 0 2 4 6 8 10 solution at 37 "C. Each incubation mixture was then diluted into the enzymatic assay mixture to 1.25 pg/ml at 37 "C and [PEP1 (mM) sugar phosphate formation was followed. The mixture containFIG.7. Activity as a function of the initial monomer/dimer ing thecold sample was brought to 37 "C within a few seconds. state of phospho-Enzyme I. The temperature of the preincubations An initial velocity of 0.32 pmol sugar-phosphate4min x pg was maintained constantat 23 "C. A, anisotropy data as a function of phospho-Enzyme I concentration were collected and analyzed as deEnzyme I) was obtained when phospho-Enzyme I was initially scribed in Fig. 5. The buffer was:10 m~ KP,, 100 m~ KCl, 5 m~ MgCl,, in the dimer form (upper curve). By sharp contrast, the rate 1 mM EDTA, 0.2 mM Dm, pH 6.5. 0,no PEP; A, 2 mM PEP.Data was zero when phospho-EnzymeI wasinitially monomeric collection was at 23 "C. B , samples were preincubated with the indi(lower curve). After 8-10 min at 37 "C in the presence of all cated concentrations of PEP as described for panel A, and aliquots (as in Fig. 6). components of the PTS, the rate of the sample initially contain-assayed for activityby the sugar phosphorylation method The quantity of [l4CImethyl glucoside 6-phosphate formed 1inmin at ing only monomer approaches that of the sample containing 37 "C is presented. dimer. In the accompanying paper (21), we show that dimer formation takes minutes. formation of the monomer. The lag that is observed in Fig. 6 It was possible that low temperatures have two simultane- could conceivably result from the conformational change as the ous effects: dissociation of the dimer anda change in thecon- monomer is shifted from 1"C t o 37 "C, rather than because of

lA

20268

The MonomerlDimer Equilibriumof Enzyme I of the PTS

the time required for dimerization. To test this idea, the temperature was maintained at 23 "C for all samples, but the monomer/dimer composition was shiftedby including different quantities of PEP in the preincubation. As noted above (Table III), 2 mM PEP increasesK& of the phosphoprotein by 30-fold. The results areshown in Fig. 7. The data on the MiD transition are shown in Fig. 7A. This figure was generated in a manner similar t o that indicated in regard to Fig. 5. Fluorescence anisotropy valueswere collected as indicated for Fig. 1 as a functionof phospho-E1 concentration. This data was fitted by non-linear least squares analysis, and the values of K&, rM,and rD were obtained as described previously. These values are given in Table 111. For each datapoint, the dimer fraction was calculated from the observed anisotropy ( f , = [rss - rM]/[rDrM]).These values are indicated in the figure as a function of the Enzyme I Concentration. The solid curve represents the dimer fraction simulated at all protein concentration based on the value of K&. As indicated in Fig. 7A, the addition of 2 mM PEP causes thevalue of K& of the phosphoprotein to increase by 30-fold. At all concentrations of protein, there is a higher dimer fraction in thepresence of PEP. Due to thefact that the monomer concentration is rather low at all protein concentrations in the presence of PEP, the quality of the data is notas good in the presence of PEP (triangles) as in its absence (circles)and theconfidence interval for the data set obtained in the presence of PEP wasvery wide (Table 111). The data regarding the enzymatic activity are shown in Fig. 7B. At the concentration of protein used in the preincubation (7 pg/ml), there is 4 times more dimer in the sample containing 2 mM PEP than the sample without PEP (panel A). PhosphoEnzyme I preincubated with PEP displayed about 3 times more (initial) activity inthePTSsugar phosphorylation assay (panel B). Therefore, the ratio of monomer/dimer at the start of the reaction is the determiningfactor in the initialactivity of Enzyme I in sugar phosphorylation via the PTS. DISCUSSION

Two of the major questions concerning Enzyme I are as fol-

lows. What is themechanism of dimerization? Is the monomer/ dimer transitionphysiologically significant? The DNA sequences o f p t s l from E. coli and S. typhimurium show that the Enzyme I monomers are 63.5 and 63.2 kDa, respectively (10,ll). TheS. typhimurium monomer was shown to dimerize by sedimentation equilibrium and analytical gel chromatography (12,13); the association was found to be highly temperature-dependent, and K& was determined for a few selected conditions. High sensitivity differential scanning calorimetry and protease studies showed that theS. typhimurium monomer contains two structural domains, each consisting of about half of the molecule (11).The N-terminal domain, containing the active site His residue, is the compact proteaseresistant stable structural unit in the monomer, whereas the C-terminal domain, which contains the 4 Cys and 2 Trp residues, isa highly flexible structural unitrequired for interaction with PEP.'

Some experiments have also been conducted with E. coli Enzyme I. The intrinsic tryptophanfluorescence of the protein was studied withnanosecond time-resolved techniques (14,281, and the results were consistent with a monomer/dimer transition. In addition (E), theC-terminal Cys residue of the E. coli enzyme was specifically labeled with pyrene maleimide; the labeled protein was found to dimerize and was fully active in the PTS sugar phosphorylation assay. The pyrene-labeled protein maybecome important for studying both the mechanism of dimerization and the behavior of Enzyme I in complex systems. However, this requires that the monomer and dimerexhibit different spectroscopic properties, and that the labeled proteins behave identically and indistinguishably from the native monomer and dimer. The present studies were designed to answer these questions using fluorescence anisotropy of E. coli Enzyme I. By determining anisotropy and total fluorescence with the pure pyrene-labeled protein and with mixtures of labeled and unlabeled proteins over a 100-200-fold concentration range, we found the following. ( a ) The labeled monomer and dimer exhibit different, measurable anisotropy values (rMand rD, respectively). ( b ) The labeled and unlabeled proteins behave identically and interchangeably in all respects. ( c )The anisotropy of a mixture of the two species is defined by a simple relationship (Equation2). From this result, themole fractions of monomer and dimer in an equilibrium mixture can bedetermined by its anisotropy, which in turn defines K& under those conditions. ( d ) Similar experiments were conducted to determinethe effects of temperature, phosphorylation, ligands ( M e , PEP, pyruvate), and pH on K&, rM,and r p While a pH change from 6.5 to 7.5 had a minimal effect, the K& increased 23-fold as the temperaturewas increased from 6 to 30 "C (dephosphoprotein, no ligands). The following ratios of K& were obtained at 23 "C: dephospho/phospho-EI, 6-10; either species, +Mg2+/-Mg2', 7-8. The largest effect was observed with PEP and M e . K&,for the phosphoprotein increased 240-fold in the presence of the two ligands. In early experimentson the enzymatic activity of Enzyme I (27), a lag period was observed when the protein was equilibrated at 0 "C prior to assay but notwhen it was preincubated at the assay temperature. These results led to a n hypothesis (4) on how the enzyme functions, i.e. the dimer accepts the phosphoryl group fromPEP, while the phosphomonomer transfers it to HPr.3 Since fluorescence anisotropy can determine the monomer/ dimer content of an Enzyme I solution, it was possible to a t least test the first part of the hypothesis. The equilibrium was perturbed intwo different ways. ( a )A solution containingonly monomer was generatedby preincubating Enzyme I a t 1"C at low concentration, while 100% dimer wasformed at 37 "C, high concentration. ( 6 ) Since PEP strongly promotes dimer formation of the phosphomonomer, the temperature was maintained constant (23 "C), but aliquots of Enzyme I were preincubated

domain is apparently required for dimerization and is highly flexible, the sequence data suggest that folding and interaction of two monomers to give the dimer follows a well defined path, possibly involving more The S. typhimurium and E.coli proteins substitute for each otherin than one conformer. In preliminary experiments, dimers were genersugar phosphorylation in vitro and in vivo and cross-react immunologi- ated using two kinds of monomers, each derivatized with a different cally. In addition, they behave virtually identically when studied by fluorophore at their C termini. The probes were chosen so that energy scanning microcalorimetry and proteasetreatment (Dr. C. LiCalsi, un- transfer can occur. There was no detectable energy transfer,indicating to one anotherin the dimer (onthe order published data). The functional and structural homology between these that the C termini are not close proteins is surprising since they differby 16 amino acid residues in of 75-100 A). The N-terminal half of EnzymeI contains the active site His residue their primary sequences. By contrast, there are many Enzyme I mutants that are inactive or defective and involve single amino acid sub- and accepts the phosphoryl group from phospho-HPr but neither from stitutions. The ptsZ DNA sequences show that 13 of the 16 amino acid PEP nor phospho-EnzymeI (11). There is no evidence that this half of can dimerize. Thus, it appears likely that the differences between the S. typhimurium and E. coli wild type proteins the molecule are inthe N-terminal domain, whilethe C-terminalhalf of the molecule phosphomonomer can reversibly phosphorylateHPr. We have not ruled is highly conserved,withonlythree changes. Since the C-terminal out the possibility that phosphodimer can also phosphorylate HPr.

The Monomer lDimer Equilibrium of Enzyme I of the PTS with various concentrations of PEP. All samples were assayed in the complete sugar phosphorylation system at 37 "C. The results provide strong evidence that themonomer is inactive in accepting the phosphoryl group from PEP. Finally, changes in the anisotropy suggest that Enzyme I monomer and dimer undergo significant conformational changes upon phosphorylation and in the presence of some of the ligands. These results agree with those obtained by measuring the rates of reaction of the 4 Cys residues with DTNB (15). The rapid reaction of the C-terminal Cys-575, was not affected by phosphorylation or the presence of ligands, whereas there was a marked change in the reaction rates of the 3 internal Cys residues, all located in the C-terminal domain, and different ligands gave different rates. It is likely that the monomer/dimer transition is not a simple two-state process, but involves intermediate conformers of monomer or dimer or both. Results published in the following paper show that this approach allows the study of the kinetics of the monomer/dimer transition. Acknowledgments-We thank Dr. Dmitri Toptygin for helpfuldiscussions. We thank Professor Michael Johnson for making his program NONLIN available to us and providing aid in its use. We are especially grateful to Pat Coyle for gifts of proteins and the chromatographic data described in the text. REFERENCES 1. Weigel, N., Kukuruzinska, M. A,, Nakazawa, A,,Waygood, E. B., and Roseman, S. (1982) J. Biol. Chem. 257, 14477-14491 2. Postma, P. W.,and Lengeler, J. W. (1985) Microbiol. Reu. 49, 232-269 3. Roseman, S., and Meadow, N.D. (1990) J. Biol. Chem. 265,2993-2996

20269

4. Meadow, N. D., Fox, D. K., and Roseman, S. (1990)Annu. Reu. Biochem. 59, 497642 5. Postma, P. W., Lengeler, J. W., and Jacobson, G . R. (1993) Microbiol. Reu. 57, 543-594 6. Mason, P. W., Carbone, D. P., Cushman, R. A., and Waggoner, A. S. (1981) J. Biol. Chem. 266, 1861-1866 7. Stock, J. B., Waygood, E. B., Meadow, N. D., Postma, P.W., and Roseman, S. (1982)J. Biol. Chem. 257, 14543-14552 8. Simoni, R. D., and Roseman, S. (1973) J. Biol. Chem. 248,966-979 9. Liu, K. D.,and Roseman, S. (1983)Proc.NatlAcad. Sci. U.S.A. 80,7142-7145 10. Saffen, D.W., Presper, K. A,, Doering, T. L., and Roseman, S. (1987) J. Biol. Chem. 262, 16241-16253 11. LiCalsi, C., Crocenzi, T., Freire, E., andRoseman, S. (1991)J. Biol. Chem. 266, 19519-19527 12. Kukuruzinska, M.A., Harrington, W. E, andRoseman,S. (1982)J.Biol. Chem. 257,14470-14476 13. Kukuruzinska, M. A,, Turner, B. W., Ackers, G. K., and Roseman, S. (1984) J. Biol. Chem. 259,11679-11681 14. Neyroz, P., Brand, L., and Roseman,S. (1987)J. Biol. Chem. 262,15900-15907 15. Han, M. K., Roseman, S., and Brand, L. (1990) J. Biol. Chem. 265,1985-1995 16. Saier, M. H., Schmidt, M. R., and Lin, P. (1980) J. Biol. Chem. 255,85794584 17. Misset, O.,Brouwer, M., and Robillard, G. T.(1980) Biochemistry 19,883490 18. Fox, D.K., and Roseman, S. (1986) J. Biol. Chem. 261,1349%13503 19. Dannelly, H. K., and Roseman, S. (1992) Proc. Natl. Acad. Sci. U.S. A. 89. 11274-11276 20. Beneski, D.A., Misko, T. P., andRoseman, S. (1982) J. Biol.Chem. 257, 14565-14575 21. Chauvin, F., Brand, L., and Roseman, S. (1994) J. Biol. Chem. 269, 2 0 2 7 s 20274 22. Weigel, N., Waygood, E. B., Kukuruzinska, M. A,, Nakazawa, A,, and Roseman, S. (1982) J. Biol. Chem. 267, 14461-14469 23. Lowry, 0.H., Rosebrough, N. J., Farr,A. L., and Randall, R. J. (1951) J . Biol. Chem. 193,26&275 24. Ellman, G. L. (1959)Arch. Biochem. Biophys. 82, 7&77 25. Habeeb, A. F. S. A. (1972) Methods Enzymol. 26, 457464 26. Han, M. K., Knutson, J. R., Roseman, S., and Brand, L. (1990) J. Biol. Chem. 265, 1996-2003 27. Waygood, E. B., Meadow, N. D., and Roseman, S. (1979)AnaL Biochem. 95, 293-304 28. Chauvin, F., lbptygin, D., Roseman, S., and Brand, L. (1992) Biophys. Chem. 44,163-173