e u r o p e a n j o u r n a l o f s o i l b i o l o g y 4 5 ( 2 0 0 9 ) 9 4 – 9 9

ava i lab le a t www.sc iencedi rec t .com

j ourna l homepage : h t tp : / /www.e lsev ier . com/ loca te /e jsob i

Original article

Azospirillum brasilense PII proteins GlnB and GlnZ do notform heterotrimers and GlnB shows a unique trimericuridylylation pattern

Juliana Inaba, Luciano F. Huergo, Ana C. Bonatto, Leda S. Chubatsu, Rose A. Monteiro,M. Berenice Steffens, Giseli Klassen, Liu U. Rigo, Fabio O. Pedrosa, Emanuel M. Souza*

Department of Biochemistry and Molecular Biology, Universidade Federal do Parana, Francisco H. dos Santos s/n,

CP 19046, Curitiba, PR, Brazil

a r t i c l e i n f o

Article history:

Received 17 March 2008

Received in revised form

1 August 2008

Accepted 22 August 2008

Published online 1 October 2008

Keywords:

Azospirillum brasilense

GlnB

GlnZ

PII-like protein,

Nitrogen fixation

* Corresponding author. Tel.: þ55 41 3361 16E-mail address: souzaem@ufpr.br (E.M. S

1164-5563/$ – see front matter ª 2008 Elsevidoi:10.1016/j.ejsobi.2008.08.006

a b s t r a c t

In many organisms, nitrogen metabolism is co-ordinated by a class of highly conserved

proteins from the PII family. In Gram-negative bacteria PII proteins are trimers that can be

covalently modified by uridylylation according to the cellular nitrogen status. Several

prokaryotes have more than one gene that code for PII proteins. In Escherichia coli it was

shown that the two PII proteins (GlnB and GlnK) can form heterotrimers and it was sug-

gested that heterotrimerization of PII proteins could be widespread in Bacteria. The

nitrogen-fixing plant-associative bacteria Azospirillum brasilense code for two PII proteins,

GlnB and GlnZ. The expression of glnB and glnZ genes are induced under nitrogen fixing

conditions and these proteins control both the expression and the activity of the nitroge-

nase enzyme. Here we show that unlike E. coli PII proteins, A. brasilense GlnB and GlnZ, do

not form heterotrimers in vitro. Our data also suggest that A. brasilense GlnB shows a unique

uridylylation pattern.

ª 2008 Elsevier Masson SAS. All rights reserved.

1. Introduction This organism can only fix nitrogen under microaerobic

The associative nitrogen-fixing bacteria Azospirillum brasilense

has attracted considerable attention due to its capacity of

colonize the roots and enhance the growth and yield of several

economically important crops [8,27]. Inoculants of Azospir-

illum spp. have been used in many part of the world for crops

such as rice and maize. The understanding of its nitrogen

metabolism and the fine tuning of its regulatory network is

essential for genetically engineering strains with improved

agronomic efficiency.

67; fax: þ55 41 3266 2042.ouza).er Masson SAS. All rights

conditions and when fixed nitrogen is limiting. Nitrogenase

expression and activity in A. brasilense are controlled by the PII

proteins GlnB and GlnZ [12,13,18,24,25]. In response to

nitrogen levels, these proteins are reversibly modified by the

bifunctional enzyme GlnD. In vitro assays showed that GlnB

and GlnZ are uridylylated by GlnD in the presence of ATP and

2-oxoglutarate. The presence of glutamine inhibits uridylyla-

tion and stimulates deuridylylation of both GlnB and GlnZ [3].

The GlnB protein controls the activity of the transcriptional

activator NtrC through its partner NtrB [13] and is required for

reserved.

e u r o p e a n j o u r n a l o f s o i l b i o l o g y 4 5 ( 2 0 0 9 ) 9 4 – 9 9 95

the activity of NifA, the transcriptional activator of the nif

genes [2]. The GlnZ protein cannot substitute GlnB in these

functions [13]. GlnB and GlnZ also play specific roles in the

control of the nitrogenase activity through the reversible ADP-

ribosylation of dinitrogenase reductase (NifH) [19,24,25].

Under nitrogen fixing conditions GlnZ-UMP3 interacts with

the enzyme DraG (dinitrogenase reductase ADP-ribosyl-

glycohydrolase), which activates nitrogenase by removing the

ADP-ribosyl group from NifH [19]. After an ammonium shock,

the deuridylylated form of GlnB interacts with the enzyme

DraT (dinitrogenase reductase ADP-ribosyl-transferase),

which inactivates nitrogenase by ADP-ribosylation of NifH.

Under these conditions, GlnZ is deuridylylated and the GlnZ–

DraG complex associates with the membrane protein AmtB,

inhibiting nitrogenase activation by DraG [17,18].

The PII family comprises highly conserved signal trans-

duction proteins. These proteins are found in Archaea, Bacteria

and Eukarya and control the activity of transcriptional activa-

tors, key metabolic enzymes and membrane transporters

through direct protein–protein interactions [16,26]. In Proteo-

bacteria, PII proteins undergo reversible uridylylation by the bi-

functional enzyme GlnD [16]. The structural models of some PII

proteins have been determined. The PII proteins from Escher-

ichia coli (GlnB and GlnK) are homotrimers that form a compact

barrel with a central and three lateral clefts [10,30]. Each

subunit extends a flexible region known as the T-loop, which

contains the uridylylation site, a highly conserved tyrosine

residue (Tyr51). Uridylylation of E. coli PII proteins requires ATP

and 2-oxoglutarate and these proteins have a binding site for

each effector [6,22]. The ATP binding sites are located in the

lateral clefts between the PII subunits [30,31]. The 2-oxogluta-

rate binding site of GlnB proteins from E. coli and Herbaspirillum

seropedicae seems to be also in the lateral cleft [9,21]. Recent

data have suggested that PII proteins can also bind ADP and

AMP [20,23,29,32]. The ability of PII proteins to interact and thus

regulate their targets depends on the PII uridylylation status

and the effectors bound.

E. coli PII proteins (GlnB and GlnK) can form heterotrimers

both in vivo and in vitro [15,28]. The formation of heterotrimers

between these proteins seems to promote fine regulation of PII

target activities such as glutamine synthetase [28]. It was

suggested that heterotrimerization could be widespread in

Bacteria. As both PII proteins coded by A. brasilense (GlnB and

GlnZ) are induced under nitrogen limiting conditions [12,17],

we decided to verify if PII heterotrimerization could also occur

in this organism. Here we show that A. brasilense PII proteins,

GlnB and GlnZ, do not form heterotrimers in vitro. The data

also suggest that A. brasilense GlnB trimer is only stable when

completely unmodified or when fully uridylylated.

2. Material and methods

2.1. Expression and purification of A. brasilense GlnB,GlnZ and His-GlnD proteins

The A. brasilense GlnB, GlnZ and His-GlnD (N-terminal 6xHis

tagged version of the GlnD enzyme) proteins were expressed

in E. coli BL 21 from plasmids bearing the T7 promoter and

purified as described previously [3].

2.2. In vitro uridylylation of GlnB and GlnZ andpurification of fully uridylylated forms

The uridylylation reactions were performed as described

previously [3]. The products of the reactions were analyzed by

native gel electrophoresis as described [3] and MALDI-TOF

spectrometry (see below). Samples were collected at 0, 5, 10, 20,

30 and 60 min and the reaction stopped by addition of 1 mM

EDTA and 5 ml of native electrophoresis sample buffer. PII-UMP

was purified after uridylylation by removing GlnD-His with

HisMagnetics beads (Promega Co., Madison, WI) and the system

was dialyzed against 50 mM Tris–HCl pH 7.5, 0.1 M KCl, 20%

glycerol overnight at 4 �C to remove ATP and 2-oxoglutarate.

2.3. Heterotrimer formation of PII

Purified GlnB, GlnB-UMP3, GlnZ and GlnZ-UMP3 proteins were

mixed (final concentration of 0.5 mg/ml) at the indicated ratios

in a buffer containing 100 mM Tris–HCl pH 7.5, 100 mM KCl

and 25 mM MgCl2. When indicated, samples were boiled for

5 min and kept on ice for 5 min. The samples were analyzed by

native gel electrophoresis [3].

2.4. Native polyacrylamide gel electrophoresis (PAGE)

Samples were mixed with 5 ml of loading buffer (62.5 mM Tris–

HCl pH 6.8, 10% glycerol, 0.01% bromophenol blue) and sepa-

rated in a non-denaturing polyacrylamide gel electrophoresis

(10%) as described [14].

2.5. MALDI-TOF analysis

The protein bands separated by native gel electrophoresis

were excised, distained with a solution containing 50% of

acetonitrile (ACN) and ammoniumbicarbonate 25 mM. Samples

were dehydrated in acetonitrile 100%, lyophilized in a Speed-

Vac, and digested with porcine trypsin (10 mg/ml) (Promega Co.,

Madison, WI) overnight at 37 �C. The peptides were extracted

using 30 ml of ACN 50% and trifluoroacetic acid (TFA) 5% for

10 min in an ultrasonic water bath. The supernatant was

collected, lyophilized in a SpeedVac and mixed with 2 ml TFA

0.1% and 1 ml of matrix (a-cyano 4-hydroxycinnamic acid). The

samples were spotted on a MALDI plate (Bruker Daltonics) and

the mass spectra obtained in a MALDI-TOF–MS Autoflex spec-

trometer (Bruker Daltonics, Bremen, Germany) using a positive

reflector mode, accelerating voltage of 20 kV, delay time of

150 ns and acquisition mass-range 800–3200 Da. The spectra

were analyzed by the FlexAnalysis 2.0 software (Bruker Dal-

tonics, Bremen, Germany) and Mascot Program.

3. Results and discussion

3.1. In vitro GlnB uridylylation

The PII proteins are found in solution as trimers. The uridy-

lylation of A. brasilense GlnZ by GlnD (Utase/UR) generates

trimers containing one, two or three subunits linked to UMP

that can be resolved in a non-denaturing gel electrophoresis:

as uridylylation increases the protein mobility due to the

e u r o p e a n j o u r n a l o f s o i l b i o l o g y 4 5 ( 2 0 0 9 ) 9 4 – 9 996

addition of negatively charged groups [3,4]. We observed that,

differently from the A. brasilense GlnZ, only two forms of GlnB

were resolved under native page after an in vitro uridylylation

assay (Fig. 1I). Unmodified GlnB (time zero) produces a single

band and five minutes after starting the reaction, about 30% of

GlnB was modified and migrated faster then unmodified

protein. Ninety percent of uridylylated form was present at

60 min. No intermediate bands corresponding to GlnB-UMP1

and GlnB-UMP2 were observed (Fig. 1I). A similar result was

reported by Araujo et al. (2008) [3]. The bands corresponding to

unmodified GlnB trimer at time 0 and uridylylated GlnB trimer

at 5, 10 and 60 min were then excised from the gel (Fig. 1I),

digested with trypsin and analyzed by Maldi-Tof spectrometry

0 5 10A

B

Reaction

GlnB

GlnB-UMP

m/z

GlnB 48-56

I

II

C

Fig. 1 – (I) Uridylylation of A. brasilense (Ab) GlnB protein. The re

(2 mM), UTP (1 mM), purified GlnB (10 mM) and Ab GlnD (200 nM

analyzed by MALDI-TOF spectrometry. (II) The A–D spectra are

uridylylated and uridylylated forms of the GlnB48–56 peptide con

indicated by arrows. Peak m/z 1287.8950 corresponds to residue

residues 59–72; and peaks 1237.9770 and 1543.8520 to residues

to investigate the presence of hybrid trimers not resolved in

the native gel electrophoresis. A peak with m/z ratio of

1237.977 obtained after digestion of the band corresponding to

the unmodified GlnB trimer at 0 min matched with the frag-

ment containing the residues 48–58 of unmodified GlnB (m/z

1237.6463) that contains the uridylylation site. This peak was

substituted at times 5, 10 and 60 min by a peak with m/z ratio

of 1543.852 that corresponded to the same fragment attached

to a UMP residue (mass difference of 306 Da) (Fig. 1II). These

data confirm previous results [3,19], which showed that GlnB

uridylylation and deuridylylation, in vivo and in vitro, respec-

tively, seem to occur without the formation of hybrid forms

(GlnB-UMP1 and GlnB-UMP2).

20 30 60

D

time (minute)

A

B

C

D

GlnB 48-56

-UMP

GlnB 48-56

-UMP

GlnB 48-56

-UMP

action system contained ATP (100 mM), 2-oxoglutarate

). The bands A–D were excised digested by trypsin and

from excised bands A, B, C and D of panel I. The non-

taining residues 48–56 of GlnB (m/z difference of 306) are

s 91–101 of GlnB; 1365.9790 to residues 73–85; 1646.0780 to

48–56 non-uridylylated and uridylylated, respectively.

e u r o p e a n j o u r n a l o f s o i l b i o l o g y 4 5 ( 2 0 0 9 ) 9 4 – 9 9 97

The uridylylation of each subunit of the trimeric E. coli GlnB

is independent of each other as partially uridylylated trimers

are observed in native PAGE [5]. During the A. brasilense GlnZ

time course uridylylation reaction there is also the formation

of the intermediates, GlnZ-UMP1 and GlnZ-UMP2, which could

be resolved by native gel electrophoresis [3,4]. The results

obtained with the A. brasilense GlnB are different and may be

explained by two possibilities: (1) the A. brasilense GlnB uri-

dylylation is a cooperative process and therefore, once the

first UMP is added to the trimer the addition of the second and

third UMP groups are so fast that we cannot detect the inter-

mediates (GlnB-UMP1 and GlnB-UMP2) under our assay

condition; and (2) the partially uridylylated trimers are not

stable and thus are not readily observed in vitro.

3.2. In vitro heterotrimer formation of A. brasilense PII

proteins

To investigate the formation of GlnB/GlnZ heterotrimers in A.

brasilense the purified GlnB, GlnZ, GlnB-UMP3 and GlnZ-UMP3

were subjected to a denaturation/renaturation reaction as

described in Section 2 and loaded on a native gel electropho-

resis. Studies involving denaturation/renaturation of GlnB

and GlnK proteins were previously reported in E. coli using

urea as denaturing agent [22,28]. The treatment was efficient

to promote hybrid formation, but the renatured protein had

decreased activity [21].

To allow hybrid trimer formation without changing PII

activity we tested a temperature induced denaturation/rena-

turation treatment as described in Section 2. The denatur-

ation/renaturation treatment did not remove the UMP moiety

from GlnB or GlnZ, nor altered the uridylylation profile of both

PII proteins (Fig. 2I and results not shown).

I

Z Z* Z/Z* Z Z* Z/Z*

Room temperature Boiled 5 minutes

II

Boiled 5 minutes

1 2 3 4 5 6

B B* Z Z* B/ZB/B*(1:2)

Fig. 2 – The purified proteins (final concentration 0.5 mg/ml) were

by native gel electrophoresis (10%). When indicated, the sample

electrophoresis. The symbols indicate: B: A. brasilense (Ab) GlnB

At room temperature GlnZ and GlnZ-UMP3 are found as

homogeneous bands as observed by native PAGE (Fig. 2I). When

both proteins (GlnZ and GlnZ-UMP3) were mixed at 1:1 ratio and

kept at room temperature the mono- and di-uridylylated hybrid

trimers were not observed: as expected only the two bands cor-

responding to the homotrimers fully uridylylated and fully

unmodified were detected (Fig. 2I). When GlnZ and GlnZ-UMP3

were mixed (at 1:1 ratio), boiled for 5 min and allowed to renature

on ice for 5 min, new oligomers corresponding to the mono- and

di-uridylylated forms were found in native gel electrophoresis.

These results were expected since the thermic treatment

disrupts the quaternary and tertiary structures of both GlnZ and

GlnZ-UMP3; once the mix is cooled the proteins would renature

randomly leading to the formation of trimers containing 0, 1, 2

and3uridylylated monomers.Theaffinityof interactionbetween

a non-uridylylated GlnZ monomer with a GlnZ-UMP monomer or

with other non-uridylylated GlnZ monomers seems to be similar,

since the most abundant forms of GlnZ observed after thermic

treatment of GlnZ:GlnZ-UMP3 mixture (1:1 ratio) are the mono-

and di-uridylylated forms (Fig. 2II).

In contrast, when purified GlnB and GlnB-UMP3 were mixed,

boiled and allowed to renature, mono- and di-uridylylated

forms were not observed; only two bands corresponding to free

or fully uridylylated GlnB were observed (lanes 3 and 9, Fig. 2II).

When non-uridylylated GlnB and GlnZ (lane 6, Fig. 2II), non-

uridylylated GlnB and GlnZ-UMP-3 (lane 7, Fig. 2II) or non-

uridylylated GlnZ and GlnB-UMP3 (lane 8, Fig. 2II) were mixed

we did not observed formation of any new band, suggesting that

heterotrimerization between these group of proteins did not

occur under our assay condition. Since it is very unlikely that

the thermic treatment did not promote the complete denatur-

ation of these proteins, these results indicate that the A. brasi-

lense GlnB and GlnZ proteins do not form stable heterotrimers in

GlnZ

GlnZ-UMP3

GlnZ-UMP1

GlnZ-UMP2

7 8 9

B/Z* Z/B*

GlnB

GlnB-UMP3

GlnZ

GlnZ-UMP3

B/B*

mixed in a 1:1 (or 1:2, when indicated) ratio and separated

s were boiled 5 min and kept on ice before native gel

; B*: Ab GlnB-UMP3; Z: Ab GlnZ; Z*: Ab GlnZ-UMP3.

e u r o p e a n j o u r n a l o f s o i l b i o l o g y 4 5 ( 2 0 0 9 ) 9 4 – 9 998

vitro, confirming in vivo data [19]. Furthermore our data suggest

that non-uridylylated monomer of GlnB does not form stable

complexes with uridylylated GlnB monomers, thus not allowing

partially uridylylated trimer formation. This could explain why

the mono- and di-uridylylated forms are not observed upon

uridylylation of GlnB in vivo [19] and in vitro ([3] and Fig. 1).

Previous studies in E. coli involving a GlnB D47–53 mutant have

shown that GlnB is found as homotrimer mono-, di- and tri-

uridylylated and the presence of a single T-loop subunit is suffi-

cient to promote PII interaction with their receptors. Hybrid

trimers were able to regulate NtrB activity, promoting NtrC

dephosphorylation, and GS activity by the adenylyltransferase

(ATase) [21]. Van Heeswijk et al. (2000) [28] showed that GlnB

forms heterotrimer with GlnK both in vivo and in vitro, and sug-

gested that heterotrimers GlnK/GlnB are involved in fine tuning

of the regulation of glutamine synthetase by GlnB [28]. The het-

erotrimers had intermediate activity compared with the

homotrimers of GlnB or GlnK. In E. coli both GlnB and GlnK

seem to have several overlapping functions including control of

NtrB [6], ATase [28], AmtB [11]. It was suggested that the distinct

functions of GlnB and GlnK were correlated with the timing of

expression and levels of accumulation of the two proteins [7].

On the other hand, despite the high degree of sequence simi-

larity (81%), the A. brasilense PII proteins GlnB and GlnZ seems to

have evolved to perform very specific functions that usually do

not overlap. Most of the A. brasilense PII targets reported so far

including NifA, NtrB, DraT and DraG are specifically regulated

by only one of the proteins (GlnB or GlnZ) [1,13,18]. The only

target shared by these two proteins seems to be the ammonia

channel AmtB [20], but the effect of either protein on AmtB

activity is not yet known. Thus, the functional segregation

between A. brasilense GlnB and GlnZ seems to exclude a regu-

latory role of GlnB/GlnZ heterotrimer formation.

In conclusion, here we have reported evidence that the A.

brasilense PII proteins GlnB and GlnZ do not form heterotrimers

in vitro and that GlnB shows a unique pattern of trimer uri-

dylylation. Since PII proteins have a central role in the control

of nitrogen metabolism in Bacteria our results may be relevant

for the goal of obtaining new genetically improved strains for

use as inoculant for crop plants. Further structural analysis

will provide the molecular basis of the differences observed

between GlnB and GlnZ trimer formation.

Acknowledgments

We are grateful to Daniela F. Seixas for helping with the

MALDI-TOF analysis, Valter de Baura, Roseli Prado and Julieta

Pie are thanked for their technical support. This work was

supported by CNPq (Instituto do Milenio) and Fundacao

Araucaria.

r e f e r e n c e s

[1] L.M. Araujo, R.A. Monteiro, E.M. Souza, M.B.R. Steffens, L.U.Rigo, F.O. Pedrosa, L.S. Chubatsu, GlnB is specifically requiredfor Azospirillum brasilense NifA activity in Escherichia coli, Res.Microbiol. 155 (2004) 491–495.

[2] F. Arsene, P.A. Kaminski, C. Elmerich, Modulation of NifAactivity by PII in Azospirillum brasilense: evidence fora regulatory role of the NifA N-terminal domain, J. Bacteriol178 (1996) 4830–4838.

[3] L.M. Araujo, L.F. Huergo, A.L. Invitti, C.I. Gimenes, A.C.Bonatto, R.A. Monteiro, E.M. Souza, F.O. Pedrosa, L.S.Chubatsu, Different responses of the GlnB and GlnZ proteinsupon in vitro uridylylation by the Azospirillum brasilense GlnDprotein, Braz. J. Med. Biol. Res. 41 (2008) 289–294.

[4] M.S. Araujo, V.A. Baura, E.M. Souza, E.M. Benelli, L.U. Rigo, M.B. Steffens, F.O. Pedrosa, L.S. Chubatsu, In vitro uridylylationof the Azospirillum brasilense N-signal transducing GlnZprotein, Protein Expr. Purif 33 (2004) 19–24.

[5] M.R. Atkinson, E.S. Kamberov, R.L. Weiss, A.J. Ninfa,Reversible uridylylation of the Escherichia coli PII signaltransduction protein regulates its ability to stimulate thedephosphorylation of the transcription factor nitrogenregulator I (NRI or NtrC), J. Biol. Chem. 269 (1994) 28288–28293.

[6] M.R. Atkinson, A.J. Ninfa, Characterization of the GlnKprotein of Escherichia coli, Mol. Microbiol. 32 (1999) 301–313.

[7] M.R. Atkinson, T.A. Blauwkamp, A.J. Ninfa, Context-dependent functions of the PII and GlnK signaltransduction proteins in Escherichia coli, J. Bacteriol 184(2002) 5364–5375.

[8] Y. Bashan, G. Holguin, L.E. de-Bashan, Azospirillum-plantrelationships: physiological, molecular, agricultural, andenvironmental advances (1997–2003), Can. J. Microbiol. 50(2004) 521–577.

[9] E.M. Benelli, M. Buck, I. Polikarpov, E.M. de Souza, L.M. Cruz,F.O. Pedrosa, Herbaspirillum seropedicae signal transductionprotein PII is structurally similar to the enteric GlnK, Eur.J. Biochem 269 (2002) 3296–3303.

[10] P.D. Carr, E. Cheah, P.M. Suffolk, S.G. Vasudevan, N.E. Dixon,D.L. Ollis, X-ray structure of the signal transduction proteinPII from Escherichia coli at 1.9 A, Acta Crystallogr. sect, D Biol.Crystallogr 52 (1996) 93–104.

[11] G. Coutts, G. Thomas, D. Blakey, M. Merrick, Membranesequestration of the signal transduction protein GlnK by theammonium transporter AmtB, EMBO J 21 (2002) 1–10.

[12] M. de Zamaroczy, A. Paquelin, C. Elmerich, Functionalorganisation of the glnB-glnA cluster of Azospirillum brasilense,J. Bacteriol 175 (1993) 2507–2515.

[13] M. de Zamaroczy, Structural homologues PII and PZ ofAzospirillum brasilense provide intracellular signalling forselective regulation of various nitrogen-dependentfunctions, Mol. Microbiol. 29 (1998) 449–463.

[14] K. Forchhammer, N. Tandeau de Marsac, The PII protein inthe cyanobacterium Synechococcus sp. strain PCC 7942 ismodified by serine phosphorylation and signals the cellularN-status, J. Bacteriol 176 (1994) 84–91.

[15] K. Forchhammer, A. Hedler, H. Strobel, V. Weiss,Heterotrimerization of PII-like signalling proteins:implications for PII-mediated signal transduction systems,Mol. Microbiol. 33 (1999) 338–349.

[16] K. Forchhammer, PII signal transducers: novel functionaland structural insights, Trends Microbiol. 16 (2008) 65–72.

[17] L.F. Huergo, E.M. Souza, M.B. Steffens, M.G. Yates, F.O.Pedrosa, L.S. Chubatsu, Regulation of glnB gene promoterexpression in Azospirillum brasilense by the NtrC protein,FEMS Microbiol. Lett. 223 (2003) 33–40.

[18] L.F. Huergo, L.S. Chubatsu, E.M. Souza, F.O. Pedrosa, M.B.R.Steffens, M. Merrick, Interactions between PII proteins andthe nitrogenase regulatory enzymes DraT and DraG inAzospirillum brasilense, FEBS Lett. 580 (2006) 5232–5236.

[19] L.F. Huergo, E.M. Souza, M.S. Araujo, F.O. Pedrosa, L.S.Chubatsu, M.B.R. Steffens, M. Merrick, ADP-ribosylation ofdinitrogenase reductase in Azospirillum brasilense is regulated

e u r o p e a n j o u r n a l o f s o i l b i o l o g y 4 5 ( 2 0 0 9 ) 9 4 – 9 9 99

by AmtB-dependent membrane sequestration of DraG, Mol.Microbiol. 59 (2006) 326–337.

[20] L.F. Huergo, M. Merrick, F.O. Pedrosa, L.S. Chubatsu, L.M.Araujo, E.M. Souza, Ternary complex formation betweenAmtB, GlnZ and the nitrogenase regulatory enzyme DraGreveals a novel facet of nitrogen regulation in bacteria, Mol.Microbiol. 66 (2007) 1523–1535.

[21] P. Jiang, P. Zucker, A.J. Ninfa, Probing interactions of thehomotrimeric PII signal transduction protein with itsreceptors by use of PII heterotrimers formed in vitro fromwild-type and mutant subunits, J. Bacteriol 179 (1997) 4354–4360.

[22] P. Jiang, J.A. Peliska, A.J. Ninfa, Enzymologicalcharacterization of the signal- transducinguridylyltransferase/uridylyl-removing enzyme (EC 2.7.7.59)of Escherichia coli and its interaction with the PII protein,Biochemistry 37 (1998) 12782–12794.

[23] P. Jiang, A.J. Ninfa, Escherichia coli PII signal transductionprotein controlling nitrogen assimilation acts as a sensor ofadenylate energy charge in vitro, Biochemistry 46 (2007)12979–12996.

[24] G. Klassen,E.M. Souza, M.G. Yates, L.U. Rigo, J. Inaba,F. Pedrosa,Control of nitrogenase reactivation by the GlnZ protein inAzospirillum brasilense, J. Bacteriol 183 (2001) 6710–6713.

[25] G. Klassen, E.M. Souza, M.G. Yates, L.U. Rigo, R.M. Costa, J.Inaba, F.O. Pedrosa, Nitrogenase switch-off by ammoniumions in Azospirillum brasilense requires the GlnB nitrogen

signal transducing protein, Appl. Environ. Microbiol. 71(2005) 5637–5641.

[26] J.A. Leigh, J.A. Dodsworth, Nitrogen regulation in bacteriaand archaea, Annu. Rev. Microbiol. 61 (2007) 349–377.

[27] O. Steenhoudt, J. Vanderleyden, Azospirillum, a free-livingnitrogen-fixing bacterium closely associated with grasses:genetic, biochemical and ecological aspects, FEMS Microbiol.Rev. 24 (2000) 487–506.

[28] W.C. van Heeswijk, D. Wen, P. Clancy, R. Jaggi, D.L. Ollis, H.V.Westerhoff, S.G. Vasudevan, The Escherichia coli signaltransducers PII (GlnB) and GlnK form heterotrimers in vivo:fine tuning the nitrogen signal cascade, Proc. Natl. Acad. Sci.U.S.A. 97 (2000) 3942–3947.

[29] D.M. Wolfe, Y. Zhang, G.P. Roberts, Specificity and regulationof interaction between the PII and AmtB1 proteins inRhodospirillum rubrum, J. Bacteriol 189 (2007) 6861–6869.

[30] Y. Xu, E. Cheah, P.D. Carr, W.C. Van Heeswijk, H.V.Westerhoff, S.G. Vasudevan, D.L. Ollis, GlnK a PII-homologue: structure reveals ATP binding site andindicates how the T-loops may be involved in molecularrecognition, J. Mol. Biol. 282 (1998) 149–165.

[31] Y. Xu, P.D. Carr, T. Huber, S.G. Vasudevan, D.L. Ollis, Thestructure of the PII-ATP complex, Eur. J. Biochem 268 (2001)2028–2037.

[32] O. Yildiz, C. Kalthoff, S. Raunser, W. Kuhlbrandt, Structure ofGlnK1 with bound effectors indicates regulatory mechanismfor ammonia uptake, EMBO J 26 (2007) 589–599.