Báo cáo khoa học: Allosteric monofunctional aspartate kinases from Arabidopsis pptx

Indeed, activity measurements carried out on protein extracts from various plants revealed that Lys-sensitive AK activity is inhibited in a synergistic manner by Keywords aspartate kinas

Gilles Curien, Mathieu Laurencin, Myle`ne Robert-Genthon and Renaud Dumas

Laboratoire de Physiologie cellulaire Ve´ge´tale (PCV-DRDC), CEA-CNRS-INRA-Universite´ Joseph Fourier, Grenoble, France

The essential amino acids Lys, Met and Thr and the

methylating agent S-adenosyl-l-methionine (SAM) are

derived in plant and bacteria from Asp The first step

of this branched metabolic pathway consists of the

activation of Asp to aspartyl phosphate in the presence

of ATP This reaction is catalyzed by aspartate kinase

(AK; EC 2.7.2.4) The number of AK isoforms varies

greatly among different organisms (from one in yeast,

to at least five in plants) A fascinating aspect of AK

is the existence of very different allosteric control patterns, depending on the source organisms and the isoforms considered

Plant monofunctional AK activity is inhibited by Lys, as reported for bacterial AKs, but displays an additional feature that is specific to the plant enzyme Indeed, activity measurements carried out on protein extracts from various plants revealed that Lys-sensitive

AK activity is inhibited in a synergistic manner by

Keywords

aspartate kinase; lysine;

S-adenosyl-L -methionine; slow inhibition, synergy

Correspondence

G Curien, Laboratoire de Physiologie

cellulaire Ve´ge´tale (PCV-DRDC), 17 avenue

des Martyrs, 38054 Grenoble, France

Fax ⁄ Tel: +33 4 38 78 50 91

E-mail: gcurien@cea.fr

(Received 21 September 2006, revised 31

October 2006, accepted 6 November 2006)

doi:10.1111/j.1742-4658.2006.05573.x

Plant monofunctional aspartate kinase is unique among all aspartate kinases, showing synergistic inhibition by lysine and S-adenosyl-l-methionine (SAM) The Arabidopsis genome contains three genes for monofunctional aspartate kinases We show that aspartate kinase 2 and aspartate kinase 3 are inhibited only by lysine, and that aspartate kinase 1 is inhibited in a synergistic manner by lysine and SAM In the absence of SAM, aspartate kinase 1 displayed low apparent affinity for lysine compared to aspartate kinase 2 and aspartate kinase 3 In the presence of SAM, the apparent affin-ity of aspartate kinase 1 for lysine increased considerably, with K0.5values for lysine inhibition similar to those of aspartate kinase 2 and aspartate kinase 3 For all three enzymes, the inhibition resulted from an increase in the apparent Kmvalues for the substrates ATP and aspartate The mechan-ism of aspartate kinase 1 synergistic inhibition was characterized Inhibition

by lysine alone was fast, whereas synergistic inhibition by lysine plus SAM was very slow SAM by itself had no effect on the enzyme activity, in accord-ance with equilibrium binding analyses indicating that SAM binding to aspartate kinase 1 requires prior binding of lysine The three-dimensional structure of the aspartate kinase 1–Lys–SAM complex has been solved [Mas-Droux C, Curien G, Robert-Genthon M, Laurencin M, Ferrer JL & Dumas R (2006) Plant Cell 18, 1681–1692] Taken together, the data suggest that, upon binding to the inactive aspartate kinase 1–Lys complex, SAM promotes a slow conformational transition leading to formation of a stable aspartate kinase 1–Lys–SAM complex The increase in aspartate kinase 1 apparent affinity for lysine in the presence of SAM thus results from the displacement of the unfavorable equilibrium between aspartate kinase 1 and aspartate kinase 1–Lys towards the inactive form

Abbreviations

AK, aspartate kinase; SAM, S-adenosyl- L -methionine.

SAM and Lys [1,2] The mechanism of plant AK

synergistic inhibition by Lys and SAM has never been

characterized, and the exact function of this

plant-spe-cific control is not clear The publication of the

Ara-bidopsisgenome complicated the matter further Three

genes potentially coding for monofunctional AK

enzymes exist in this plant (At5g13280 for AK1;

At5g14060for AK2; At3g02020 for AK3) The

corres-ponding proteins have never been characterized at a

biochemical level It is thus still unclear whether

mono-functional AKs are all synergistically inhibited by Lys

and SAM In addition, we recently demonstrated that

plant bifunctional AK–homoserine dehydrogenases

from Arabidopsis thaliana (isoforms I and II) are

acti-vated by various amino acids [3] Whether

monofunc-tional AKs from plants are sensitive to these activators

is unknown In order to answer these questions and to

characterize the mechanism of the plant-specific

syner-gistic inhibition of AK by Lys plus SAM, the three

Arabidopsis cDNAs potentially coding for

monofunc-tional AK enzymes were cloned, and the

correspond-ing proteins were overproduced in Escherichia coli,

purified to homogeneity and characterized This work

allowed us to show that only one AK (AK1) is

inhib-ited in a synergistic manner by Lys and SAM, and to

characterize in detail the nature of the inhibition of the

plant AKs by Lys AK1 was cocrystallized in the

pres-ence of Lys and SAM, and the structure of the

com-plex was solved in this laboratory [4], representing the

first structure of an AK The present kinetic analyses

complement this structural analysis

Results

cDNA cloning, overproduction of the

recombinant enzymes in E coli and purification

The predicted amino acid sequences of AK1, AK2 and

AK3 contain putative N-terminal plastid-targeting

sequences, in accordance with the chloroplast

localiza-tion of AK activity [5] In order to obtain recombinant

enzymes in sufficient quantities for biochemical

analy-ses, cDNAs encoding the mature enzymes (with the

putative transit peptides removed) were cloned into

bacterial overexpression plasmids The sequences of

the PCR-cloned cDNAs matched the published

sequence [6]

All proteins were expressed in soluble forms in

E coli BL21 codon (+) On the first anion exchange

column, AK3 was eluted at a much lower ionic strength

(25 mS) than AK2 (40 mS) or AK1 (50 mS) A second

purification step consisted of fractionation on a

Q-Sepharose column for AK1 and gel filtration for

AK2 and AK3 Ten to fifty micrograms of highly puri-fied proteins (Fig 1) were obtained per liter of culture

On the denaturing gel documented in Fig 1, the pro-teins migrated in agreement with their predicted molecular masses (AK1, 55.9 kDa; AK2, 53.2 kDa; and AK3, 55.1 kDa) All proteins were stable when stored at ) 80 C in their storage buffer AK3 proved

to be highly unstable when stored at room temperature, losing 95% of its activity in 24 h During this period of time, AK1 and AK2 retained all their activity

Kinetic parameters The three enzymes displayed hyperbolic kinetics with both ATP and Asp (not shown) The AK1 catalytic constant (kcat) was two-fold higher than the AK2 kcat and three-fold higher than the AK3 kcat(Table 1) The AK3 Km for ATP was about two-fold and three-fold lower than observed for AK2 and AK1, respectively The AK3 Km for Asp was about two-fold lower than observed for AK1 and AK2 (Table 1) In agreement with a random mechanism for AK [7], the apparent Km

Fig 1 Protein purification Proteins were separated on a 10% poly-acrylamide (w ⁄ v) slab gel under denaturing conditions and stained with Coomassie brilliant blue R-250 Lane 1: molecular mass mark-ers Lanes 2, 4 and 5: soluble proteins of the E coli extract contain-ing AK1, AK2 and AK3, respectively Lanes 3, 5 and 6: purified AK1, AK2 and AK3, respectively (1 lg).

Table 1 Kinetic parameters Activities were measured in 50 m M Hepes (pH 8.0), 150 m M KCl, 20 m M MgCl 2 and 200 l M NADPH with 100 n M AK and 1 l M aspartate semialdehyde dehydrogenase

at 30 C.

KmATP(l M ) KmAsp(l M ) kcat(s)1)

for one substrate was independent of the concentration

of the other substrate in the absence of inhibitor

Regulatory properties

Regulatory properties were subsequently examined in

the presence of physiologic concentrations of Asp

(1 mm) and ATP (2 mm) The results shown in Fig 2

show that the enzymes were inhibited in a sigmoidal

manner by increasing concentrations of Lys AK1

dis-played a much higher K0.5 value for Lys inhibition

(570 ± 20 lm; Fig 2A) compared to AK2 (K0.5¼ 10.2 ± 0.7 lm; Fig 2B) and AK3 (K0.5¼ 7.4 ± 0.4 lm; Fig 2C) For the three enzymes, no inhibition

by SAM could be detected in the absence of Lys Interestingly, in the presence of Lys, AK1 but not AK2 or AK3 was inhibited by SAM (Fig 3) As shown in Fig 3A, for AK1 the K0.5value for Lys inhi-bition decreased dramatically from 570 lm in the absence of SAM to 4.2 ± 0.2 lm in the presence of a saturating concentration of SAM (300 lm) Thus, at saturation with SAM, AK1 displayed a K0.5value for Lys similar to that of the SAM-insensitive AK2 and AK3 enzymes

The response curves of AK1 to SAM in the presence

of Lys are shown in Fig 3B,C In the presence of Lys, AK1 activity decreased in a sigmoidal manner as a function of SAM concentration Increasing the Lys concentration decreased the K0.5 values for SAM (15 lm in the presence of 100 lm Lys)

Nature of the inhibition

In order to determine the origin of the inhibition, AK activities were measured in the presence of variable concentrations of ATP and Asp for different concen-trations of Lys (as well as in the presence of Lys plus SAM for AK1) The results are shown in Fig 4 for AK1 and in Fig 5 for AK2 (qualitatively identical results were obtained with AK3; not shown) In the presence of Lys, the apparent Kmvalues for both ATP and Asp increased Note that the increase in the apparent Km values for the substrates in the presence

of the inhibitors was more pronounced at low concen-trations of the second substrate Thus, whereas in the absence of inhibitor, the apparent Kmfor one substrate was independent of the concentration of the other sub-strate (empty squares for AK1; Fig 4A,B) and empty circles for AK2; Fig 5A,B), a dependence was observed in the presence of the inhibitor Increasing the second substrate concentration decreased the Km effect of Lys This effect is expected when competitive inhibition occurs with a two-substrate enzyme The same behavior was observed with the SAM-sensitive AK1 enzyme in the presence of Lys plus SAM The two-substrate nature of AKs necessitates caution in the examination of the inhibitor effect on catalytic constant At subsaturating (i.e limiting) fixed concen-tration of ATP (or Asp) and variable concenconcen-trations

of Asp (or ATP), the maximal v⁄ [E]0 value is an apparent catalytic constant Increasing the inhibitor concentration increases Km for both the fixed and the variable substrates, thus leading to a decrease in apparent maximal velocity In order to check whether

0

0.2

0.4

0.6

0.8

1

[Lysine], µM

A

AK1

0

0.2

0.4

0.6

0.8

1

[Lysine], µM

0

0.2

0.4

0.6

0.8

1

[Lysine], µM

Fig 2 Inhibition of Arabidopsis monofunctional AK isoenzymes by

Lys AK activities were measured in buffer D in the presence of

1 m M Asp, 2 m M ATP and variable concentrations of Lys Activities

were normalized to unity in the absence of inhibitors Curves are

the best fit obtained by nonlinear regression analysis of the

experi-mental data using a Hill equation (A) AK1 (j), (B) AK2 (s) and (C)

AK3 (n) activities versus Lys concentration Values for AK1 are:

K 0.5 ¼ 570 ± 10 l M , n H ¼ 2.4 ± 0.1 Values for AK2 are: K 0.5 ¼

12.5 ± 0.8 l M , n H ¼ 1.3 ± 0.1 Values for AK3 are: K 0.5 ¼

7.4 ± 0.2 l M , nH¼ 2.6 (± 0.2).

the true catalytic constant was affected by Lys,

extra-polated apparent catalytic constants obtained by

curve-fitting in bisubstrate variation experiments were

replotted as a function of the fixed substrate

concen-tration (ATP or Asp) As shown in Figs 4C,D,G,H and 5C,D, extrapolated maximal v⁄ [E]0values (i.e true catalytic constants) were about 25 s)1 for AK1 and

14 s)1 for AK2 These values are similar, within the limits of experimental error, to the true catalytic con-stant of the uninhibited enzymes (Table 1) Thus, inhi-bition of plant AKs results only from a modification

of the apparent Km for the two substrates ATP and Asp

Equilibrium binding experiments Kinetic experiments indicated that SAM alone had no effect on AK1 activity (Fig 3B), even after preincuba-tion of the enzyme with SAM However, this did not exclude SAM binding to the free enzyme without hav-ing any effect on the enzyme activity in the absence of Lys In order to determine whether SAM binds to the free enzyme, equilibrium binding experiments were car-ried out with S-adenosyl-l-[methyl-3H]methionine As shown in Fig 6, bound radioactivity was undetectable

in the absence of Lys On the contrary, in the presence

of 1 mm Lys, bound radioactivity was detected Lys alone was able to inhibit AK1 (Fig 2A), indicating that it can bind to the enzyme in the absence of SAM Thus, kinetics and equilibrium binding experiments show that SAM and Lys binding to AK1 is sequential, with Lys binding preceding SAM binding

Slow-inhibition kinetics AK1 inhibition by Lys alone was fast Indeed, a steady-state rate was achieved during the mixing time (less than 6 s; Fig 7A), indicating rapid equilibration

of AK1, and AK1–Lys and AK1–substrate complexes

In marked contrast, when AK1 activity was measured

in the simultaneous presence of SAM and Lys, a delay was observed in attainment of the steady state When the reaction was initiated with the enzyme (Fig 7B), rates decreased with time until the steady state was reached When, instead, the enzyme was preincubated for 5 min with the two inhibitors and diluted with the inhibitors in the reaction mix, progressive reactivation

of the enzyme was observed, indicating that the inhibi-tion was reversible (not shown) Slow inhibiinhibi-tion of plant AK has never been described before, probably because the low abundance of the enzyme in extracts from plants required long incubation times [1,2] Three mechanisms have been proposed for the analysis of slow-binding inhibition [8,9] Mechanism A assumes that the formation of an enzyme–inhibitor (EI) com-plex is a single, slow step Mechanism B assumes the rapid formation of an EI complex that then undergoes

0

0.2

0.4

0.6

0.8

1

[Lys], µM

A

0 µM

20 µM

300 µM [SAM]

0

0.2

0.4

0.6

0.8

1

[SAM], µM

B

10 µM Lys

0

0.2

0.4

0.6

0.8

1

[SAM], µM

C

100 µM Lys

Fig 3 Synergistic inhibition of AK1 by Lys and SAM AK1 activity

was measured in buffer D in the presence of 1 m M Asp, 2 m M ATP

and variable concentrations of Lys and SAM (A) AK1 activity versus

Lys concentration for three different concentrations of SAM (0, 20,

300 l M ) Curves are the best fit obtained by nonlinear regression

analysis of the experimental data using a Hill equation K 0.5 values

for Lys in the presence of 0, 20 and 300 l M SAM are 570 ± 10 l M ,

82 ± 2 l M , and 4.5 ± 0.5 l M , respectively Hill numbers (n H ) were

2.4 ± 0.2, 2.3 ± 0.1, and 2 ± 0.3, respectively (B) AK1 activity

ver-sus SAM concentration in the absence and the presence of 10 l M

Lys The curve in the presence of 10 l M Lys is the best fit obtained

by nonlinear regression analysis of the experimental data using a

Hill equation The K0.5value for SAM in the presence of 10 l M Lys

is 131 ± 6 l M The Hill number in the presence of 10 l M Lys is

1.75 ± 0.1 (C) AK1 activity versus SAM concentration in the

pres-ence of 100 l M Lys The K0.5value is 15 ± 0.3 l M The Hill number

is 2 ± 0.05.

0 1 2 3 4 5

pap of rA TP

[Asp], mM

A

0 6 12 18 24 30

1- )

[Asp], mM

C

0

1 2 3 4

[Asp], mM

E

[Lys]=3 µM [SAM]=400 µM

0 6 12 18 24 30

pap s(

1- )

[Lys]= 3 µM [SAM]= 400 µM

G

2 3 4 5 6 7

pa f oA rs

[ATP], mM

B

0

0 6 12 18 24 30

pap s(

1- )

[ATP], mM

D

0 2 4 6

p of rA sp

[ATP], mM

[Lys ]= 3 µM [SAM] = 400 µM

F

0

0 6 12 18 24 30

pap s(

1- )

[ATP], mM

[Lys]= 3 µM [SAM] = 400 µM

H

Fig 4 Nature of the inhibition of AK1 by Lys or Lys plus SAM Bisubstrate variation experiments were carried out with AK1 in the absence of inhibitor (h), in the presence of Lys alone (500 l M ) (j) or in the presence of Lys (3 l M ) plus SAM (400 l M ) ( ) Hyperbolic curves (not shown) obtained for a fixed concentration of one substrate and variable concentrations of the other substrate were fitted with Michaelis–Menten equations to calculate apparent kinetic parameters (Kappm and kappcat) (A) AK1 apparent K m for Asp versus ATP concentration in the absence (h) and presence (j) of 500 l M Lys (B) AK1 apparent Km for ATP versus Asp concentration in the absence (h) and presence (j) of 500 l M Lys (C) AK1 apparent kcat values (extrapolated from bisubstrate variation experiments) versus Asp concentration for 500 l M Lys (D) AK1 extrapolated apparent k cat values versus ATP concentration in the presence of 500 l M Lys The extrapolated maximal apparent kcatvalue is similar (within the limits of experimental error) to the true kcatvalue of AK1, indicated

by a dotted line (Table 1) (E, F, G, H) Same as (A), (B), (C) (D), respectively, for bisubstrate variation experiments carried out with AK1 and in the presence of Lys plus SAM.

a slow and favorable isomerization to an EI* complex.

In the mechanism C, isomerization precedes inhibitor

binding It is possible to distinguish between these

mechanisms by analysis of the relationship between the

exponential decay constant (kobs) of the progress curve

and the inhibitor concentration [8,9] A linear

relation-ship is observed for mechanism A, and hyperbolic

rela-tionships for mechanisms B and C: however, kobs

increases with inhibitor concentration when

mechan-ism B applies, and decreases when mechanism C

applies

Concerning AK1, kinetic results indicate that Lys

alone inhibits the enzyme (although with a low

apparent affinity) and that the inhibition is fast Also,

equilibrium binding analyses indicate that SAM

bind-ing follows Lys bindbind-ing Slow inhibition in the

pres-ence of SAM results either from slow binding of

SAM to the AK1–Lys complex (mechanism A), or

from a slow conformational transition of the AK1–

Lys–SAM complex induced by SAM (mechanism B),

or finally, from the binding of SAM to an isomer of

the enzyme–Lys complex in slow equilibrium with

another isomer (mechanism C) In order to distin-guish between the three possible mechanisms, progress curves were obtained with AK1 for 100 lm Lys (i.e under conditions where Lys alone was only margin-ally inhibitory) and for different concentrations of SAM The kobs values were obtained by nonlinear least-square fitting of the progress curves using Eqn (1):

At¼ At0 vs t þðvs viÞ  ð1  e

k obs tÞ

kobs

ð1Þ where Atis the absorbance at time t, At0is the absorb-ance at t0, viis the initial velocity of the reaction, vs is the steady-state velocity of the reaction, and kobsis an exponential decay constant

The results shown in Fig 7C reveal a hyperbolic relationship between kobs and SAM concentration The kobs value increased when SAM concentration increased The results are thus consistent with mechan-ism B, i.e a mechanmechan-ism in which slow inhibition is due

to slow isomerization following SAM binding to an enzyme–Lys complex

0 1 2 3

K m

[Asp], mM

A

0 5 10

15

20

k act

pap s(

[Asp], mM

C

4

0 2 4 6 8

K m

p ofr A sp

B

0 5 10 15 20

D

Fig 5 Nature of the inhibition of AK2 by Lys (A) AK2 apparent K m for Asp versus ATP concentration in the absence (s) or in the presence (d) of 25 l M Lys (B) AK2 apparent Kmfor ATP versus Asp concentration in the absence (s) or in the presence (d) of 25 l M Lys (C) AK2 extrapolated apparent kcatvalues (calculated from bisubstrate variation experiments) versus Asp concentration in the presence of 25 l M Lys The maximal value is similar to the true AK2 k cat value, symbolized by a dotted line (Table 1) (D) AK2 extrapolated apparent k cat values (cal-culated from bisubstrate variation experiments) versus ATP concentration in the presence of 25 l M Lys The maximal value is similar to the true AK2 kcatvalue (Table 1).

AK2 and AK3 also displayed slow inhibition

kinet-ics by Lys (not shown) In the absence of structural

data for these enzymes, the results cannot yet be

inter-preted in terms of a mechanistic model

Allosteric control of monofunctional AKs

is highly specific

No additional inhibition of the three AKs was

observed upon addition of Thr or Leu, in the presence

of 5 lm Lys for AK3 and AK2, and in the presence of

100 lm Lys and 20 lm SAM for AK1 This contrasts

with E coli AKIII, which proved to be inhibited

synergistically by Lys and Leu [10], and Bacillus

polymyxa monofunctional AK, which is inhibited in a

concerted manner by Lys and Thr [11] In addition,

the activators of Thr-sensitive bifunctional

AK–homo-serine dehydrogenases from plants (Ala, Cys, Ile, Leu,

Ser and Val [3]) had no effect on monofunctional

A thaliana AK activities, either in the absence or in

the presence of the inhibitors The other amino acids

tested (Met, Gln, Asn, Glu, Arg) had no effect on the

enzyme activities at 2.5 mm either in the presence or

the absence of the inhibitor Lys or Lys plus SAM (for

AK1)

Discussion

The present article describes in detail the kinetic and

regulatory properties of the three Lys-sensitive

mono-functional AKs from Arabidopsis We show for the

first time that all three enzymes are inhibited by Lys,

but only one isoform (AK1) is inhibited synergistically

by Lys and SAM In vitro kinetic measurements

indi-cate that all three enzymes are efficiently inhibited by

physiologic concentrations of Lys (80 lm) Indeed, the

K0.5values are 80 lm, 10 lm and 7 lm, for AK1, AK2 and AK3, respectively, for activity assayed in the pres-ence of physiologic concentrations of Asp, ATP and SAM (for AK1) AK2 and AK3 would be more strongly inhibited by Lys than AK1 under these condi-tions AK1 activity is also highly sensitive to changes

in SAM concentrations in the physiologic context of the leaf (K0.5for SAM in the presence of 80 lm Lys is close to the physiologic concentration of SAM, 20 lm

0

0.25

0.5

0.75

1

[SAM], µM

Fig 6 Equilibrium binding of S-adenosyl- L -[methyl- 3 H] methionine to

AK1 in the absence (h) and in the presence (j) of 1 m M Lys The

curve in the presence of 1 m M Lys is the best fit obtained by

nonlin-ear regression analysis of the experimental data using the equation

of a hyperbol A Kdvalue of 5.7 ± 0.7 l M could be calculated.

0.9 1

0 100 200 300 400

time (s)

A

[Lys] = 500 µ M

No SAM

0.6 0.7 0.8 0.9 1

time (s)

[SAM] (µ M ) 400 200 100 50 [Lys]= 100 µ M B

0

0 0.015 0.03 0.045 0.06

0 100 200 300 400 500

[SAM], µ M

[Lysine]= 100 µ M C

Fig 7 Slow inhibition of AK1 in the presence of Lys plus SAM (A) Progress curves were obtained in the presence of Lys and in the absence of SAM, with the reaction initiated with AK1 (B) Progress curves were obtained in the presence of Lys and SAM in the reac-tion media Reacreac-tions were initiated with AK1 free of inhibitors (C) Observable rate constant of AK1 versus SAM concentration Pro-gress curves were obtained as indicated in (B) for 100 l M Lys and different concentrations of SAM For each curve, the exponential decay constant was obtained by curve-fitting using Eqn (1) Experi-ments were repeated twice for a given SAM concentration Data points were fitted with the equation derived in [6] for mechanism B (two-step process).

[12]) It is clear from our results that AKs are not

sat-urated by the substrates in vivo Thus, control of plant

AK activities by modification of the enzyme Kmvalues

for the substrates is an efficient control mechanism

Although the Thr-sensitive bifunctional AKs [3] and

Lys-sensitive monofunctional AKs from Arabidopsis

have in common a control of their activity via

modifi-cation of the Km values for both substrates, a striking

difference is the absence of activation of

monofunc-tional AKs by small amino acids (Ala, Cys, Ile, Leu,

Ser, Val) Interestingly, monofunctional AKs display

low Km values for Asp and ATP (1–2 mm in the

absence of inhibitors) compared to bifunctional AKs

(5–15 mm in the absence of the activators) The

activa-tors of bifunctional AKs reduce the Kmvalues for both

ATP and Asp to values similar to those measured here

for monofunctional AKs, suggesting that under

physi-ologic conditions, all five AKs display similar kinetic

efficiencies We proposed that Ala, because of its

abundance in the chloroplast, was the physiologic

acti-vator of bifunctional AKs A hypothetical functional

advantage of this allosteric interaction could be a

feed-forward control, coupling the Asp-derived amino acid

pathway to nitrogen and carbon metabolism

Accord-ing to this hypothesis, one would expect to also

observe allosteric activation of Lys-sensitive AKs Its

absence suggests that the effect of activation of

Thr-sensitive bifunctional AKI and AKII might be to

increase their sensitivity to Thr inhibition rather than

to provide coupling with carbon and nitrogen

metabo-lism This might also explain why the activation of

AKI and AKII need not be highly specific for the

acti-vator [3]

A survey of the expression pattern of the AK genes

using the Arabidopsis microarray database

Genevesti-gator [13] indicated the presence of AK1 and AK3

mRNAs in all the examined organs and tissues at

sim-ilar levels Specific expression of the AK3 gene in

vas-cular tissues has been reported [14] As the phenotype

of a knockout mutant of AK3 [14] is indistinguishable

from that of a wild-type Arabidopsis, other AKs (more

probably the closely related AK2) can replace AK3, at

least under controlled growth conditions No data

could be found for the AK2 gene in the Arabidopsis

microarray database Genevestigator However,

nor-thern blot analyses [15] suggested that the gene

enco-ding AK2 is expressed in all tissues of Arabidopsis

Unless specific control of translation takes place

in vivo, these results suggest that the three AKs are

coexpressed in leaf chloroplasts Together with our

kinetic results, they suggest that a fraction of the flux

controlled by Lys is insensitive to SAM (i.e the flux

generated by the activity of AK2 and AK3)

Mechanism of inhibition of AK1 by Lys and SAM The mechanism of the synergistic inhibition of AK1 by Lys and SAM was analyzed according to the recent three-dimensional structure of the AK1–Lys–SAM complex in this laboratory [4] AK1 displayed a K0.5 value for Lys inhibition in the absence of SAM about 50-fold higher than that observed for the SAM-insen-sitive AKs This might be due to a much higher affin-ity of AK1 for the substrates compared to AK2 and AK3 That is, much more Lys might be required to displace more strongly bound substrates However, all three AKs display similar Km values for the substrates (Table 1) Thus, the high AK1 K0.5value for Lys inhi-bition in the absence of SAM (Fig 2A) is a conse-quence of the enzyme’s lower affinity for Lys compared to AK2 and AK3

Equilibrium binding analyses carried out with AK1 indicated sequential binding of Lys and then SAM to AK1 (Fig 6) In the crystal structure of AK1 in com-plex with Lys and SAM, the SAM-binding site in the regulatory domain of the enzyme is formed in part by a loop that also participates in the Lys-binding site This suggests that the SAM-binding site is not already pre-sent on the enzyme and requires prior binding of Lys

In the presence of SAM, the apparent affinity of AK1 for Lys was much higher than in its absence and was similar to that of the SAM-insensitive AK2 and AK3 This increase in apparent affinity in the presence

of SAM does not result from a direct molecule-to-molecule interaction between SAM and Lys, but is mediated by the protein Indeed, the Lys- and SAM-binding sites are in close proximity in the crystal struc-ture, and two adjacent amino acids of the polypeptide chain, S371 and V372, interact with Lys and with Lys and SAM, respectively [4] The increase in apparent affinity of AK1 for Lys in the presence of SAM results from a slow conformational rearrangement of the pro-tein that is induced by SAM, as indicated by the hyperbolic relationship between kobsand SAM concen-tration (Fig 7C)

The data can be used to propose a model for the inhibition of AK1 by Lys and SAM (Scheme 1) In Scheme 1, E represents the enzyme in the active con-formation ES represents the enzyme–substrate (ATP plus Asp) complex In this model, upon Lys binding in the regulatory domain, the enzyme adopts a novel con-formation (E*) that is unable to bind the substrates (Lys alone is inhibitory; Fig 2A) The transition E–Lys fi E*–Lys is fast (Fig 7A), but the equilib-rium is not favorable That is, a high concentration of Lys is required to shift the whole enzyme population

to the inhibited form In addition, the E*–Lys form

has acquired the ability to bind SAM (Fig 6)

Follow-ing the formation of the encounter complex with SAM

(SAM–E*–Lys), a slow conformational transition

induced by SAM occurs (Fig 7C) As shown in the

protein structure [4], Lys is trapped inside the protein

in the E** state The ribose and adenine moieties of

SAM are also deeply buried in the protein, but the

Met moiety of SAM is exposed to the solvent Most

probably, the dissociation of the SAM–E**–Lys

com-plex occurs when the protein comcom-plex is in the SAM–

E*–Lys conformational state The release of the

coin-hibitors is sequential, with Lys release following SAM

release In this model, the reinforcement of Lys

inhibi-tion by SAM would result from the displacement of

the unfavorable equilibrium between E–Lys and E*–

Lys, owing to the formation of an enzyme form

stabil-ized by SAM

The sigmoidal shape of the Lys inhibition curves

(Fig 2A) is in accordance with the identification of

two equivalent Lys-binding sites at the interface

formed by two regulatory domains in the protein

dimer [4] Residues from both monomers contribute to

the formation of a Lys-binding site, thus providing

strong coupling between both subunits In the crystal

structure, the number of interactions between dimers is

low (eight hydrogen bonds, 2.4% of each subunit

area), in agreement with native gel electrophoresis

results showing that the enzyme behaves

predomin-antly as a dimeric enzyme (95%) in equilibrium with a

tetramer However, kinetic experiments indicate that

the Hill numbers for AK1 are close to or slightly

higher than 2 This may indicate that a fraction of the

enzyme population forms tetramers in solution under

the conditions of the kinetic experiments

Binding experiments indicated hyperbolic saturation

curves for SAM in the presence of 1 mm Lys (Fig 6)

This suggests that there are no cooperative homotropic

interactions between the two SAM molecules in the

protein dimer under these conditions This is supported

by the three-dimensional structure The SAM-binding site of a monomer is entirely formed by amino acids from that monomer, with no obvious physical inter-actions between this site and the other subunit Kinetic experiments showed sigmoidal SAM saturation curves (Fig 3B,C) This may indicate that long-range interac-tions occur in the enzyme between SAM sites However,

as SAM binding follows Lys binding, the cooperative homotropic interaction observed for SAM under these conditions may be only apparent and result from homo-tropic interactions between Lys-binding sites

As discussed by Mas-Droux et al [4], differences in amino acid sequences were observed in AK2 and AK3 compared to AK1 at the level of the SAM-binding site W392SR394 amino acids are involved in SAM binding in AK1 The tryptophan residue is not found

in AK2, and the loop is longer in AK3 These differ-ences could explain the absence of SAM effects on AK2 and AK3

Comparison of plant and bacterial Lys-sensitive AKs

In addition to the specific control of AK1 by Lys plus SAM, Arabidopsis AK allosteric control displayed dif-ferences compared to the E coli Lys-sensitive AKIII enzyme First, E coli AKIII is inhibited in synergy by Lys and Leu [10], but no effect of Leu could be observed with the A thaliana monofunctional AKs In addition, the three A thaliana AKs display slow-inhi-bition kinetics, a feature that has never been reported for the bacterial AKIII enzyme Finally, inhibition of

A thaliana enzymes results exclusively from a modifi-cation of the apparent Kmfor the substrates ATP and Asp Three studies examined the inhibition pattern of

E coli AKIII by Lys [16–18] All report competitive inhibition with respect to ATP (as for the plant enzyme; this work) but noncompetitive inhibition with respect to Asp [16,17] However, in these studies, the effect of Lys on E coli AKIII was tested with high concentration of the second substrate ATP In these conditions, the Km effect may have been minimized (see Figs 4A,B and 5A,B for A thaliana AK) Lower concentrations of ATP were used by Wampler & West-head [18], and the apparent Kmfor Asp increased from 1.6 mm in the absence of Lys to 5 mm at 560 lm Lys, suggesting a competitive component in the inhibition

by Lys In the same study, the authors reported a modification of maximal velocity in the presence of Lys for a fixed concentration of ATP and variable con-centrations of Asp However, as the ATP concentra-tion was fixed, they probably observed a decrease in apparent maximal velocity (a consequence of a

E

s L -E

s L

-*

E

s L

-* E -M

w o l S

Scheme 1.

decrease in the apparent affinity for the second

sub-strate ATP used at a fixed concentration) rather than a

decrease in the true maximal velocity (true kcat) If this

is correct, then plant AK and bacterial AK inhibition

mechanisms may be similar

The publication of the AK1–Lys–SAM complex

structure was followed by the release of the

Methanococ-cus jannaschii AK structure in complex with Mg-ADP

and Asp [19] (Protein Data Bank entry 2HMF) and two

structures of the E coli AKIII Lys-sensitive enzyme

(AKIII–Lys–Asp and AKIII–Asp–ADP complexes;

Protein Data Bank entries 2J0X and 2J0W, respectively)

[20] This offers the possibility of examining the

inhibi-tion mechanisms of the plant and the bacterial enzymes

in the light of structural data The three-dimensional

structure of AK1 in complex with Lys and SAM [4]

revealed that the conformation of the ATP-binding site

in this complex was unsuitable for nucleotide binding

In E coli AKIII cocrystallized with Lys, the

ATP-bind-ing site was also in a conformation that prevented the

accommodation of ATP This is consistent with

compet-itive inhibition with respect to ATP observed both with

plant AKs and with E coli AKIII [16–18]

The active site arginine side chain (R198 in M

jann-aschii AK and R202 in E coli AKIII) was shown to

be responsible for a bidendate interaction with the Asp

substrate a-carboxyl group [19] The side chain of the

corresponding Arg residue (R230) in the A thaliana

AK1–Lys–SAM complex is farther away (7 A˚) and

forms interactions with the SAM-binding site,

suggest-ing that the bindsuggest-ing of the inhibitors removes an

inter-action that stabilizes Asp in its binding site This is in

agreement with the observed increase in Kmvalues for

Asp in the presence of the inhibitor(s) In E coli

AKI-II cocrystallized with Lys, an Asp molecule was

pre-sent in the active site (one per dimer) [20] In this

complex, the active site Arg residue (R202) is

posi-tioned more than 2 A˚ further from the Asp substrate

molecule than in the AKIII–Asp–ADP complex This

suggests that the interaction between the AKIII active

site and the Asp substrate is weaker in the inhibited

complex This is in agreement with an increase in the

apparent Kmfor Asp observed in the presence of Lys,

as reported by Wampler & Westhead [18] Structural

data thus suggest that the plant and the E coli

enzymes are controlled by similar mechanisms

The E coli Lys-sensitive AK has been considered by

Monod et al [21] and others [10,22] to be a V-type

allosteric enzyme Both plant and E coli AKs indeed

display features of allosteric V systems, as substrate

saturation curves are hyperbolic in the absence and in

the presence of the effector However, in the model

proposed by Monod et al for ‘V systems’, the

alloster-ic effector modifies kcatand the substrate has the same affinity for the two states of the enzyme (the R and T states) It would be somewhat misleading to consider plant AKs as V-type allosteric enzymes, as Lys modi-fies exclusively the apparent Kmfor the substrates The denomination ‘V system’ and the associated model may not be appropriate for this enzyme

Inhibition and synergistic inhibition of AKs Synergistic control by Lys and SAM is specific to the plant enzyme, but other AKs also display synergistic inhibition Lys-sensitive AKIII from E coli is inhibited

in synergy by Lys and Leu [22] AKIII from Bacillus subtilis[23,24] and AK from

Rhodopseudomon-as[25,26], display synergistic inhibition by Lys and Thr Synergistic inhibition requires the existence in these enzymes of two sites, one for each coinhibitor The reg-ulatory domain of A thaliana AK1 as well as that of

E coli AKIII is formed of two ACT domains [27] In AK1 from Arabidopsis, the two coinhibitors Lys and SAM bind in one of the two ACT domains (ACT1) [4]

In the E coli AKIII structure, the Lys molecule was also found in ACT1 [20] The enzyme structure in the presence of Lys plus Leu is still unknown Thus we do not know whether the coinhibitors in the other synergis-tically inhibited AKs bind in a position similar to where the SAM molecule binds in AK1 or whether another site (in ACT2, for example) is involved in the binding

of the coinhibitor According to the first hypothesis, AK1 may provide an explanatory model for the other AKs that are synergistically inhibited

Experimental procedures

Chemicals

Amino acids were obtained from Sigma-Aldrich (St Quentin Fallavier, France) SAM was purified as previously described [28]

Bacterial strains

Escherichia coli strain DH10B was used for cloning, and

Darmstadt, Germany) was used for recombinant protein production

Construction of the plasmids

The cDNA sequences corresponding to the predicted mature proteins were amplified by PCR using an A thali-ana cDNA library [29] The 5¢ and 3¢ oligonucleotides