Báo cáo khoa học: The domains carrying the opposing activities in adenylyltransferase are separated by a central regulatory domain ppt

In this study, we further map the domain junctions of adenylyltransferase on the basis of solubility and enzymatic analysis of truncation constructs, and show for the first time that aden

adenylyltransferase are separated by a central regulatory domain

Paula Clancy1, Yibin Xu1,2, Wally C van Heeswijk1,3, Subhash G Vasudevan1,4and David L Ollis5

1 Department of Biochemistry and Molecular Biology, James Cook University, Townsville, Australia

2 Structural Biology Division, The Walter and Eliza Hall Institute of Medical Research, Parkville, Australia

3 Department of Molecular Cell Physiology, Faculty of Earth and Life Sciences, Vrije Universiteit, Amsterdam, the Netherlands

4 Dengue Unit, Novartis Institute for Tropical Diseases, Singapore

5 Research School of Chemistry, Australian National University, Canberra, Australia

In Escherichia coli, adenylyltransferase (AT) catalyzes

the adenylylation (Scheme 1) and deadenylylation

(Scheme 2) of glutamine synthetase (GS) according to

the reactions:

GSþ ATP ! GS-AMP þ PPi

Scheme 1.

GS-AMPþ Pi! GS þ ADP

Scheme 2.

In its unmodified form, GS catalyzes ATP-depen-dent ammonia incorporation into glutamate, forming glutamine (Gln), and in so doing drives the uptake of ammonia by enteric bacteria [1–4] The adenylylated form of GS has little activity The two activities of AT are mechanistically distinct, but are functionally the reverse of each other, and must be carefully controlled

by the organism in order to prevent futile cycling of ATP Growth will occur in a low-ammonia environ-ment if GS is active, with the deadenylylation activity

Keywords

adenyltransferase; intramolecular signaling;

monoclonal antibody; regulatory domain

Correspondence

Y Xu, Structural Biology Division, The

Walter and Eliza Hall Institute of Medical

Research, Parkville, Victoria, 3050, Australia

Fax: +61 3 9347 0852

Tel: +61 3 9345 2305

E-mail: xu@wehi.edu.au

(Received 6 February 2007, revised 1 April

2007, accepted 4 April 2007)

doi:10.1111/j.1742-4658.2007.05820.x

Adenylyltransferase is a bifunctional enzyme that controls the enzymatic activity of dodecameric glutamine synthetase in Escherichia coli by rever-sible adenylylation and deadenylylation Previous studies showed that the two similar but chemically distinct reactions are carried out by separate domains within adenylyltransferase The N-terminal domain carries the deadenylylation activity, and the C-terminal domain carries the adenylyla-tion activity [Jaggi R, van Heeswijk WC, Westerhoff HV, Ollis DL & Vasudevan SG (1997) EMBO J 16, 5562–5571] In this study, we further map the domain junctions of adenylyltransferase on the basis of solubility and enzymatic analysis of truncation constructs, and show for the first time that adenylyltransferase has three domains: the two activity domains and

a central, probably regulatory (R), domain connected by interdomain Q-linkers (N-Q1-R-Q2-C) The various constructs, which have the oppo-sing domain and or central domain removed, all retain their activity in the absence of their respective nitrogen status indicator, i.e PII or PII-UMP

A panel of mAbs to adenylyltransferase was used to demonstrate that the cellular nitrogen status indicators, PII and PII-UMP, probably bind in the central regulatory domain to stimulate the adenylylation and deadenylyla-tion reacdeadenylyla-tions, respectively In the light of these results, intramolecular sign-aling within adenylyltransferase is discussed

Abbreviations

AT, adenylyltransferase; BPM, b-polymerase motif; GS, glutamine synthetase; Gln, glutamine; a-KG, a-ketoglutarate; R domain, regulatory domain; UT, uridylyltransferase.

of AT being switched on and the adenylylation activity

switched off Conversely, in a high-ammonia

environ-ment, unnecessary ATP consumption can be reduced

by turning down GS activity, with progressive

adenyly-lation It is completely switched off when all 12

sub-units are converted to the inactive GS-AMP form In

this situation, the adenylylation activity of AT is

switched on, and the deadenylylation activity must be

switched off The two antagonistic activities of AT are

profoundly influenced by the signal transduction

protein PII, and also by the small molecule effectors

a-ketoglutarate (a-KG) and Gln PII is the nitrogen

status indicator of the cell: PII-UMP implies low

nitro-gen, and unmodified PII implies nitrogen excess The

modulation and demodulation of PII is carried out by

uridylyltransferase (UT) In a low-nitrogen

environ-ment, the uridylylation activity of UT is stimulated

and PII is converted to PII-UMP [5,6] PII-UMP binds

to AT as a complex with ATP and a-KG and inhibits

its adenylylation activity, at the same time stimulating

deadenylylation activity [7,8] In a high-nitrogen

envi-ronment, the supply of a-KG is depleted and the Gln

concentration increases In this case, the

uridylyl-removing activity of UT causes PII-UMP to be

con-verted to PII [5] The unmodified PII and Gln bind to

AT, with a consequent reduction in deadenylylation

activity and activation of the adenylylation activity

[7,8] In addition to the regulation of GS activity by

PII, the latter also regulates coordinately the

transcrip-tion of the gene encoding GS by a two-component

sys-tem [8,9], which is not discussed here It has been

shown that the two activities of AT reside on separate

domains [10] In the previous study, two constructs of

AT were made: AT-N consisted of residues 1–423

(AT-N:1–423) and was found to have the

deadenylyla-tion activity AT-C consisted of residues 425–946

(AT-C:425–946) and contained the adenylylation

activ-ity Both constructs contained the signature sequence

that is also found in the active site of rat DNA

poly-merase b [11] Construct numbering is based on

whe-ther it is C-terminal (AT-C) or N-terminal (AT-N),

and where the starting and finishing residues fall in the

polypeptide chain (Fig 1)

The present extended study on AT domains [10] was

based on three observations which suggested that the

domain boundaries of AT-N and AT-C were not

defined by the earlier truncations First, AT-N:1–423

was poorly soluble, and only a limited amount of

pro-tein could be isolated for enzymatic characterization

Extension of this construct by just 17 amino acids to

AT-N:1–440 produced a completely soluble domain

with deadenylylation activity [12] The structure of this

domain was determined by X-ray crystallography [13]

Second, the completely soluble AT-C:425–946 trunca-tion construct was highly susceptible to proteolysis in the N-terminal region, such that about seven amino acids were readily cleaved off during purification Third, another putative Q-linker sequence (Q2) [14] was noted between residues 607 and 627 in addition

to the previously noted Q-linker 1 from residues 441

to 462 [10] Q-linkers are linker sequences ( 15–25 residues long) that tether structurally distinct but inter-acting domains in a wide range of prokaryotic two-component regulatory and sensory proteins such as NTRB⁄ NTRC and NIFA ⁄ NIFL Individual Q-linker sequences are not strongly conserved, and they have a low probability of having an a or b secondary struc-ture They are rich in Gln (and hence Q-linkers), Arg, Glu, Ser and Pro residues, with a hydrophobic residue such as Leu, Ala, Ile or Val every four or five residues Q-linkers flank highly conserved and system-specific N-terminal and C-terminal domains in these types of proteins [10] Together, these observations suggested that the opposing activities of the two domains may be separated by a third central domain, and that complete

AT can be represented as N-Q1-R-Q2-C The solu-bility of truncated domains is widely regarded as an indicator that domain boundaries have been correctly chosen [15] Overhanging amino acid stretches that are not part of the domain or missing stabilizing end resi-dues (when the truncation construct is not the full length of the domain) hamper correct protein folding during overexpression, leading to aggregation and reduced solubility [16] Accordingly, a series of N-ter-minal and C-terN-ter-minal truncations of AT have been produced (Fig 1), guided by secondary structure pre-diction (predictprotein) [17,18], to define the domain boundaries of AT and gain a better insight into the intramolecular signal transduction mechanism of the protein

In order to understand the structure, function and regulation of AT, mouse mAbs were also produced and used to analyze the actions of AT [19] Using the truncated AT constructs, the mAb-binding sites are defined, and the presence of a central regulatory domain is demonstrated

Results

AT has a central regulatory domain

On the basis of studies that showed solubility of a truncation construct of a water-soluble multidomain protein is a good indicator of correct folding and domain boundaries [15,16], AT-RQ2:463–627 (central

R domain + Q2), AT-C:607–946 (Q2 + C-terminal

domain) and AT-DR:1-440- - -628-946 (N-terminal

domain+ C-terminal domain) truncations (Fig 1)

were expressed from plasmids bearing the

correspond-ing section of the gene and subjected to a rapid

solu-bility test In addition to the previously noted

Q-linker from residues 441 to 462, the presence of the

second Q-linker from residues 607 to 627 provided

the positions for the three truncations

The three truncation constructs had bands of similar

intensity for both whole cell extract and the cell-free

lysate in western blot, demonstrating that the expressed constructs were soluble (data not shown) This result implies that the N-terminal domain (AT-N:1–440) [12,13], R domain (AT-RQ2:463–627), C-terminal domain (AT-C:607–946) and a construct that was formed from the N-terminal and C-terminal domains (AT-DR:1-440- - -628-946) consisted of stable domains

The deadenylylation activity of the N-terminal domain (AT-N:1–440) has already been reported [12]

Fig 1 Schematic representation of the truncation constructs of AT Truncations of the AT protein (946 residues long) were designated AT-N

or AT-C, depending on their location in the linear polypeptide chain, and their starting and finishing residues (e.g AT-N:1–440 refers to the N-terminal 440 residues of AT) Also indicated on the diagram are the positions of the two predicted b-polymerase motifs (BPM1 and BPM2) [11], the two Q-linkers (Q1 and Q2) [14], and the amino acid sequence and predicted secondary structure of the truncation region of the pro-tein between residues 421 and 600 [17,18] H (helix) and L (loop) The solubility of the constructs is shown in parentheses: Soluble (S), partly soluble (PS), insoluble (IS), thermal induction at 37 C (– T), and isopropyl thio-b- D -galactoside induction at low temperature, i.e 18 C (– I).

This construct had deadenylylation activity that was

independent of both the small effector molecule a-KG

and the effector protein PII-UMP, even when they

were present in molar excess In comparison with the

entire protein, the construct was 1000 times less

act-ive (data not shown)

The AT-N:1–423 construct was previously reported

to be as active as wild-type AT and regulated in the

same way [10] This discrepancy in the activities of

these two N-terminal constructs probably arose from

the degree of purity of the two protein preparations

The AT-N:1–423 construct was not very soluble, and

the protein preparation was only partly purified [10],

whereas the AT-N:1–440 construct was fully soluble,

and the protein preparation was extremely pure,

allow-ing structural determination from protein crystals [13]

It is quite possible that endogenous AT and other

fac-tors not removed from the protein preparation

contri-buted to the activity reported for AT-N:1–423

It was previously demonstrated that the

adenylyla-tion activity of the AT-C:425–946 truncaadenylyla-tion construct

was independent of PII [10] even when PII was present

in molar excess The various adenylylation activity

lev-els of AT, AT-C:432–946, AT-C:551–946, AT-C:607–

946 and AT-DR:1-440- - -628-946 are shown in Table 1

The AT-C:642–946 truncation construct had no

ade-nylylation activity (data not shown) All the C-terminal

truncation constructs were used in the assay at 0.6 lm

rather than 0.025 lm (entire AT) to give similar

activity levels as those of entire AT The ATDR:1–440

-628–946 truncation construct expressed poorly and

was not purified To avoid interference from the poten-tially more active endogenous AT (the expression strain was not glnE–), this construct was used at

 0.2 lm in the assay

Intact AT needs both PII and Gln to stimulate full adenylylation activity Removal of either effector cau-ses a drop in activity If PII is omitted from the ade-nylylation assay, there is a 70% drop in activity, and

if Gln is omitted from the assay, there is a  60% drop in activity Omission of both PII and Gln virtu-ally abolishes activity By contrast, the adenylylation activity for each of the C-terminal truncation con-structs and AT:DR:1-440- - -628-946 is independent of PII, as their activity level is the same whether the PII effector protein is present or not (Table 1)

The previously reported adenylylation domain, AT-C:425–946 ( 60 kDa), has now been redefined by the truncation construct AT-C:607–946 ( 39 kDa) All the C-terminal truncation constructs were dependent on Gln for full activity, because the removal

of Gln from the assay resulted in a drop in adenylyla-tion activity ranging from  50% to 85% When Gln was omitted from the assay, further removal of PII still had no effect on the activity of the truncated C-terminal proteins (Table 1)

These results suggest that the opposing and central domains inhibit activity by some form of stearic hind-rance, and that binding by either effector protein alle-viates this inhibition and encourages their respective activity A schematic presentation of these results is shown in Fig 2

Table 1 Role of the R domain in regulation of adenylylation activity Adenylylation assays using AT and C domain truncation constructs These assays show the changes in activity of ATwt(purified), AT-C:432–946 (purified), AT-C:551–946 (purified), AT-C:607–946 (cell lysate) and AT-DR:1–440- - -628–946 (cell lysate) under various conditions Activity was assessed by determining the adenylylation state of GS by measuring the production of c-glutamyl hydroxamate with various combinations of effector molecules present in the assay Standard assay conditions were used (50 n M GS, 25 n M AT ⁄  0.6 l M construct, 25 n M PII, 1 m M Gln) All the C-terminal truncation constructs were used in the assay at 0.6 l M rather than 0.025 l M (AT wt ), to give similar activity levels to that of the whole AT protein The AT-C:607–946 and AT-DR:1–440- - -628–946 construct preparations were partly purified, so their concentrations were being determined approximately from bands in western blots The first 5 min were fitted with a linear regression using Microsoft Excel The R 2 coefficients for these curves are usually > 0.9 The initial rates for all the proteins are expressed as a proportion of their standard activity, and the number following in paren-theses is the relative activities expressed as the rate per l M All the truncation constructs have a similar molar activity, which is approxi-mately 10-fold less than that of wild-type AT.

Condition

Protein

AT wt

(N-Q1-R-Q2-C) (0.025 l M )

AT-C:432–946 [N(10)-Q1-R-Q2-C]

(0.6 l M )

AT-C:551–946 [R(55)-Q2-C]

(0.6 l M )

AT-C:607–946 (Q2-C) (0.6 l M )

AT-DR:1–440- - -628–946 (N-C)

(0.2 l M )

Epitope mapping of AT mAbs using truncation

constructs

mAbs to full-length AT were produced during the

course of this study using established protocols [20]

Initially, a panel of 10 mAbs was screened in ELISA

using the overlapping constructs AT-N:1–548 and

AT-C:425–946 (data not shown) Five mAbs were

chosen for further characterization The AT mAb

6B5 was chosen because it only bound to AT-N:1–

548 and was therefore denoted the N domain mAb

The two mAbs 5A7 and 39G11 bound in the

overlap-ping region of AT-N:1–548 and AT-C:425–946, and

were therefore denoted the R domain mAbs The two

mAbs 6A3 and 27D7 only bound to AT-C:425–946,

and were therefore denoted the C domain mAbs Cell

lysates from all of the AT truncation constructs,

and complete AT, were separated by 12% SDS⁄

PAGE and immunoblotted with purified N + C

polyclonal mix, 6B5, 5A7, 6A3, 27D7 or crude 39G11

(Fig 3)

The N domain mAb 6B5 binds somewhere in the first 423 residues of the protein, as the mAb can detect all the N-terminal truncation constructs tested and intact AT, but not the C-terminal constructs starting

at AT-C:432–946 (Fig 3) A further truncation con-struct, AT-N:1–311, was also detectable using mAb 6B5 (data not shown) Therefore, this mAb binds somewhere in the first 311 residues of AT

On the other hand, the two C domain mAbs, 6A3 and 27D7, showed the opposite pattern, where all of the C-terminal constructs were detected and none of the N-terminal truncations starting with AT-N:1–548 were detected (Fig 3) Therefore, these mAbs bind in the last 305 residues of the protein, i.e the region from residues 642 to 946 A further truncation construct, AT-C:712–946, was not detectable by the two C domain mAbs (data not shown) Therefore, these two mAbs bind between residues 642 and 711, which is the adenylylation catalytic site

Both of the R domain mAbs detected the AT-RQ2: 463–627 truncation construct mAb 39G11

Adenylylation

AT+PII

“open adenylylation conformation”

AT

“closed conformation”

AT-C:432-946

AT-ΔR:1-440 -628-946

AT-C:551-946 AT-C:607-946

N N

R

C

R R

R

N

Deadenylylation

AT

“closed conformation”

C R

N

AT+PII-UMP

“open deadenylylation conformation”

C R

N

AT-N:1-440

N

UMP UMP UMP

Fig 2 Schematic representation of the different truncations of AT in adenylylation and deadenylylation The activity results from the adenyly-lation and deadenylyadenyly-lation assays are summarized in this diagram The adenylyadenyly-lation active site is shown in white, and is accessible to GS in all the conformations except the uncomplexed ‘closed’ conformation, and the deadenylylation active site, shown in gray, is accessible to GS-AMP in all conformations except the uncomplexed ‘closed’ conformation Uncomplexed AT has a ‘closed’ conformation and has minimal activity in either assay Removal of the N or R domains gives rise to polypeptides with similar adenylylation activity to that of PII-complexed

AT, and removal of the R + C domain gives rise to a polypeptide that has activity independent of PII-UMP in deadenylylation Addition of PII

to the adenylylation assay or PII-UMP to the deadenylylation assay causes a shift in the position of the N domain relative to the C domain, and AT adopts the ‘open’ conformation The complexed AT is then capable of adenylylating GS or deadenylylating GS-AMP, depending on the other effectors present in the assay.

does not detect AT-N:1–423 to AT-N:1–467, on

the N-terminal side, but does detect AT-N:1–501

and AT-N:1–548 (residues 468–501) On the

C-ter-minal side, mAb 39G11 detects AT-C:466–946 and

does not detect from AT-C:508–946 onwards

(resi-dues 466–507) (Fig 3) Therefore, this mAb binds

somewhere in the region between residues 468 and 501

mAb 5A7 does not detect AT-N:1–423 to AT-N: 1–501, on the N-terminal side, but does detect AT-N: 1–548 (residues 502–548), and on the C-terminal side it has the same binding profile as mAb 39G11

B

G

6B5 (N-terminal)

(R domain)

D 5A7 (R domain)

E 6A3 (C-terminal)

F 27D7 (C-terminal)

A Polyclonal N+C mix

T 4

48kD 50kD

54kD58kD 64kD

18kD

108kD

59kD 55kD 50kD 46kD 39kD 35kD

Deadenylylation domain Central R Adenylylation domain

domain

5A7

(residues 466–507) (Fig 3) Therefore, this mAb binds

somewhere in the region between residues 502 and

507

Mapping PII and PII-UMP-binding sites using AT

mAbs

The adenylylation reaction requires the PII protein

and Gln as the allosteric effectors, whereas the

deadenylylation reaction requires the uridylylated

form of the effector protein PII-UMP and a-KG as

the allosteric effectors In order to determine the

effects of the various mAbs on the enzymatic

activit-ies of AT, purified N domain mAb, 6B5, was added

to either the adenylylation or deadenylylation assay

at a molar ratio of 10 : 1, and was found to have no

effect on either of the activities of AT (data not

shown) This suggests that mAb 6B5 binds at a site

(within residues 1–311) that does not influence

activ-ity directly or indirectly by blocking cofactor binding

or any related conformational changes Similarly, the

two C domain mAbs, 6A3 and 27D7, also had no

impact on deadenylylation under these conditions,

and only partially inhibited adenylylation, by 73%

and 52%, respectively These mAbs also partially

inhibited the adenylylation activity of all the

PII-independent C-terminal truncation constructs (data

not shown), suggesting that these mAbs, which bind

in the catalytic site (residues 642–711), are affecting

the interaction with GS

These results are biologically interesting because

they show that blocking one activity does not

con-sequently influence the opposing activity This

obser-vation demonstrates the necessity to regulate both

opposing activities in a coordinated manner, as

other-wise the activity of GS is not regulated properly

In addition, these results suggest that mAbs 6A3

and 27D7, whose binding site overlaps b-polymerase

motif BPM2, do not bind to BPM1 Apparently,

BPM1, which has a homologous amino acid sequence

to BPM2 [11], is not antigenically similar to BPM2

Inhibition of PII binding in adenylylation

by R domain mAbs Similarly, the two R domain mAbs, 39G11 and 5A7, were used in adenylylation assays with AT and AT-C:432–946 to investigate the impact of these mAbs

on adenylylation activity (Fig 4) Intact AT and the PII-independent truncated construct, AT-C:432–946 (R-Q2-C), were chosen for these assays because they were shown to bind the two R domain mAbs (Fig 4) and were fully soluble

Preincubation of the two R domain mAbs, 5A7 and 39G11, in the adenylylation assay mix with intact AT resulted in a reduction of adenylylation activity (Fig 4), but neither of these mAbs had an impact on the adenylylation activity of the AT-C:432–946 trunca-tion construct (Fig 4) This result implies that the ade-nylylation activity of AT is probably inhibited by mAbs 5A7 and 39G11 via inhibition of a signaling event To corroborate this finding, PII was omitted from the adenylylation assay, and the results show that whereas omission of PII had no impact on AT-C:432–

946, intact AT was inhibited to the same level as when the mAbs 5A7 or 39G11 were present with PII (Fig 4) This result implies that the binding of mAbs 5A7 or 39G11 prevents the PII binding that is neces-sary to fully stimulate adenylylation in intact AT

Inhibition of PII-UMP binding in deadenylylation

by R domain mAbs Likewise, the two R domain mAbs, 39G11 and 5A7, were tested in the deadenylylation assay with AT, in order to investigate the impact of these mAbs on deadenylylation activity Interestingly, both 5A7 and 39G11 (data not shown) completely eliminated the deadenylylation activity of intact AT (Fig 4)

Omission of PII-UMP from the deadenylylation assay also completely eliminated the activity in intact

AT, but not in AT-N:1–440, which has been shown to

be PII-UMP independent [12] In order to show that

Fig 3 Monoclonal antibody-binding regions of the AT protein (A) Western blot analysis of 12% SDS ⁄ PAGE gel of whole cell extracts for the various truncation constructs using a purified mix of AT-N:1–548 and AT-C:425–946 polyclonal antibody for detection The bands indica-ting the appropriate induced polypeptides are marked with arrows The mAbs were screened against all the truncation constructs, but only the truncation constructs that bound to the mAbs are presented: (B) purified 6B5; (C) crude 39G11; (D) purified 5A7; (E) purified 6A3; (F) purified 27D7 The N domain mAb 6B5 also detected a truncation construct comprising the first 311 amino acids of the protein (data not shown), so this mAb binds somewhere in the first 311 residues of the protein The two R domain mAbs bind in the N-terminal region of this domain, with mAb 39G11 binding in the region between residues 468 and 501, and mAb 5A7 binding in the region between residues 502 and 507 From this panel of constructs, the two C domain mAbs, 6A3 and 27D7, appear to bind between residues 642 and 946, i.e the last

305 residues of the AT protein However, they do not detect a smaller C-terminal truncation construct, AT-C:712–946 (data not shown), so they are actually binding somewhere in the adenylylation catalytic site (G) Monoclonal antibody binding regions within AT Also shown in the diagram are the b-polymerase motifs (BPM1 and BPM2) [11] and the two Q-linkers (Q1 and Q2) [14].

the inhibition of deadenylylation is via the prevention

of PII-UMP binding, the assay was slightly modified

so that no a-KG was added and twice as much

GS-AMP and PII-UMP protein was added to the assay Under these in vitro conditions, PII-UMP was the only effector responsible for the stimulated deade-nylylation activity In this modified condition, both 5A7 and 39G11 completely inhibited the deadenylyla-tion activity of intact AT (Table 2)

Discussion Previous work demonstrated that AT had two domains with catalytic activity at either end of the pro-tein [10,11] Examination of the propro-tein sequence sug-gested that there were two Q-linkers flanking a central region that separated the protein into three domains (N-Q1-R-Q2-C), in contrast to the previous suggestion

of a two-domain protein [10]

The crystallization of AT-N:1–440 demonstrated that the N-terminal region of AT before the first Q-lin-ker is a biologically relevant, complete domain con-taining the deadenylylation active site [13] Assay data obtained using various truncation constructs indicated that the central domain acted as a regulatory domain (see later)

Indirect evidence from equivalent assay results with the entire AT protein where the R domain mAbs have been added or the effector protein omitted suggest that both the R domain mAbs are blocking the bind-ing of PII or PII-UMP This means the two effector proteins are probably binding somewhere in or near the R domain antibody-binding region between resi-dues 466 and 507 in the N-terminal region of the cen-tral R domain Whether the two forms of PII are binding at exactly the same site or not cannot be

Shift in AT activity with addition of either R domain mAb or removal of PII

G- ) ni

A

No shift in activity of

AT-C:432-946 with addition of

either R domain mAb or

removal of PII

G- ) ni

B

Shift in AT activity with addition of either R domain mAb or removal of PII-UMP

G- ) ni

C

-1

14

-1

14

0

27

Fig 4 Inhibition of activity in AT and truncation constructs by

R domain mAbs These assays show the changes in activity of (A)

AT and (B) the C-terminal truncation construct AT-C:432–946 in

adenylylation with R domain mAbs 5A7 and 39G11 present [no

AT ⁄ AT-C:432–946 + PII + Gln (dark blue), AT⁄ AT-C:432–946 +

PII + Gln (pink), AT ⁄ AT-C:432–946 - PII + Gln (red), AT ⁄ AT-C:432–

946 + PII + Gln + 5A7 (green), AT ⁄ AT-C:432–946 + PII + Gln +

39G11 (blue)] and (C) AT in deadenylylation with R domain

mAb 5A7 present [no AT + PII-UMP + a-KG (dark blue),

AT + UMP + a-KG (pink), AT + UMP + a-KG (red), AT +

PII-UMP + a-KG + 5A7 (green)] Standard assay conditions (50 n M GS ⁄

GS-AMP, 25 n M AT ⁄  0.6 l M construct, 25 n M PII ⁄ PII-UMP, 1 m M

Gln ⁄ 20 m M a-KG) were used, and the mAbs were preincubated

with AT ⁄ AT-C:432–946 (1 : 1) for 30 min at room temperature All

assays were performed in duplicate and with ATwtas a reference.

Error bars have not been shown on the curves, as they hinder

visual inspection The standard error range for all the curves is

generally < 0.4.

Table 2 Inhibition of PII-UMP binding by R domain mAbs 5A7 and 39G11 This table shows the initial deadenylylation activity of AT stimulated only by PII-UMP, in the presence and absence of the

R domain mAbs 5A7 and 39G11 The activity was determined using initial rate assays, which measured the production of c-gluta-myl hydroxamate (c-GH) by GS-AMP The curves were fitted with a linear regression using Microsoft Excel, and the resulting rates are shown here The R2coefficient for each curve is shown in paren-theses.

Condition

Deadenylylation rate (nmol of c-GH produced per well per min per min)

GS-AMP + AT + PII-UMP + a-KG 3.12 (1.00)

2 · GS-AMP + AT + 2· PII-UMP-a-KG 0.54 (1.00)

2 · GS-AMP + AT + 2 · PII-UMP-a-KG + 5A7

0.00

2 · GS-AMP + AT + 2 · PII-UMP-a-KG + 39G11

0.00

ascertained from these data The fact that the two

effector proteins are probably binding in the R

do-main further supports the notion that it has a

regula-tory role The GlnK paralog and its uridylylated form

were also used in assays with and without R domain

mAbs in the same way as PII and PII-UMP, and

showed the same activity inhibition patterns (data not

shown), suggesting they also bind somewhere in this

region

When the deadenylylation domain and⁄ or R domain

are removed from AT, the resulting polypeptides

become independent of PII in adenylylation These

results suggest that the R domain regulates

adenylyla-tion activity by interacting with the N domain, so that

its position relative to the C domain blocks the

ade-nylylation capacity of AT (‘closed conformation’ in

Fig 2)

Although all of these truncation constructs had

activity independent of PII binding, they were reliant

on Gln for full adenylylation activity, which suggests

that the binding site for Gln is within the C domain,

rather than the R domain Although direct binding of

Gln was not demonstrated, the fact that its removal

from the assay reduced adenylylation activity for all

the C-terminal constructs shows it definitely binds to

the C-terminal domain of AT

Similarly, deadenylylation activity is also

indepen-dent of PII-UMP when the R and C domains are

removed Therefore, in the ‘closed conformation’

(Fig 2), deadenylylation activity is also blocked A

similar phenomenon is seen in the enzyme activities

present in the N-terminal domain of

aspartokinase-homoserine dehydrogenase I [21] Removal of either of

the activity domains resulted in a decrease in the

regu-lation of the activity of the remaining domain

The binding of PII or PII-UMP somewhere within

the N-terminal region of the R domain may alter the

resting state conformation that exists between the two

opposing domains, resulting in an ‘open conformation’

(Fig 2), which allows adenylylation or deadenylylation

to proceed, depending on the effector molecules

pre-sent This model does not provide any insights into

how uridylylation of the PII effector protein causes

deadenylylation activity to be favored over

adenyly-lation activity when the AT protein is in its ‘open

conformation’

AT is approximately 1000 times more active in

deadenylylation than the PII-UMP-independent N

domain polypeptide and 10 times more active in

ade-nylylation than the PII-independent C domain

poly-peptides This suggests that when either effector

protein binds to entire AT, the respective active site⁄

domain adopts a more suitable conformation for the

appropriate reaction, allowing it to proceed more effi-ciently than in the truncation constructs

The signal of PII or PII-UMP binding is transmitted

to the activity domains of AT, presumably by con-formational changes in the domains and⁄ or Q-linkers

In Fig 5A, a speculative mechanism based on secon-dary structure prediction is suggested for the allevi-ation of stearic hindrance by the opposing domain in the ‘open conformation’ model (Fig 2) On the basis

of sequence and truncation analysis, it appears that the Q1-linker contains an amphipathic helix (residues 448–461) with a hydrophobic face, and the potential PII⁄ PII-UMP binding region contains three helices, the third of which is also amphipathic with a hydro-phobic face (residues 498–516) (Fig 5B) This fits with the observation that AT-C:466–946 (R-Q2-C) is far less soluble than AT-C:432–946 [N(10)-Q1-R-Q2-C] (data not shown)

It is possible that the hydrophobic face of the amphi-pathic helix in the Q1-linker interacts with the hydro-phobic face of the third helix in the N-terminal region

of the R domain (Fig 5C) and the binding of either effector protein disrupts the interaction, so the N and

R domains are separated, allowing AT to adopt the

‘open conformation’ The further changes that occur in the protein so that adenylylation is favored over deade-nylylation and vice versa, depending on the effec-tor protein present, can only be determined by crystallization of AT complexed to the PII and PII-UMP proteins

To conclude, in this work we have refined the domain structure of the bifunctional AT enzyme by providing compelling evidence for the presence of a central regulatory domain flanked by the two activity domains Specific mAbs that bind AT in the R domain probably block the binding of the effector proteins PII, GlnK, PII-UMP and GlnK-UMP, supporting the con-cept that the central domain plays a regulatory role

Experimental procedures Bacterial strains, media and growth conditions

All the E coli strains (primer sequences are available on request) were grown in LB medium supplemented, when appropriate, with ampicillin (100 lgÆL)1), chloramphenicol (25 lgÆL)1) [BL21(DE3)RecA] or ammonium chloride (0.5% w⁄ v for expression of adenylylated GS in DH5a) Bacterial strains containing pND707-derived vectors (thermoinducible k promoter) [22] were cultured at 37C, and induction was carried out by rapid shift to 42C at

A595 0.5–0.6 with further culture for 2 h Bacterial strains containing the T7-based expression plasmid pETDW2

(derived from pETMCS1)-derived vectors were cultured

continuously at 37C, and protein overexpression was

induced with isopropyl thio-b-d-galactoside (0.4 mm at A595

0.5–0.6) at 18C, with further culture for 2 h This vector

was used for expression of truncation constructs, which had

poor solubility when expressed at 37C in the

thermoin-ducible pND707 vector

DNA manipulations

Standard DNA manipulations were carried out essentially

as described previously [23] Oligonucleotides (sequences are available on request) used for PCR amplification and nucleotide sequence determination were from AusPep Pty Ltd (Parkville, Australia) DNA sequencing was carried out

Fig 5 Analysis of the Q1-linker and PII ⁄ PII-UMP-binding region (A) Structural prediction for the Q1-linker and PII ⁄ PII-UMP-binding region covering residues 441–520 H, helix; C, coil; E, sheet ( EXPASY : APSSP ) (B) Top view representation of the predicted helical region in Q1 and the third helix of the R domain ( EXPASY : HELIXWHEEL ) The respective amino acids and their relative positions in the helix are indicated on the helical wheel The hydrophobic residues [32] are italicized and highlighted in red, and the hydrophilic ⁄ polar residues are in normal text (C) Schematic representation of the Q1-linker and PII ⁄ PII-UMP-binding region of the R domain The hydrophobic side of the a-helix in Q1 (shown

in red) is possibly associated with the hydrophobic face in the third helix in the N-terminal region of the R domain (shown in red) Binding of PII or PII-UMP in this region (shown in blue) may disrupt the interaction between the two hydrophobic faces, causing them to separate, and consequently relieving the stearic hindrance between the two opposing domains.