MA

JGa

b

c

d

a

ARRAA

KNGM

I

A2spr

aiGE

((m

h0

Microbiological Research 171 (2015) 65–72

Contents lists available at ScienceDirect

Microbiological Research

j ourna l h om epage: www.elsev ier .com/ locate /micres

utational analysis of GlnB residues critical for NifA activation inzospirillum brasilense

uliana Inabaa,∗, Jeremy Thorntonb, Luciano Fernandes Huergoc, Rose Adele Monteiroc,iseli Klassend, Fábio de Oliveira Pedrosac, Mike Merrickb, Emanuel Maltempi de Souzac

Department of Chemistry, Universidade Estadual de Ponta Grossa, Av. Gal. Carlos Cavalcanti, 4748, CEP 84030-900 Ponta Grossa, PR, BrazilDepartment of Molecular Microbiology, John Innes Centre, Norwich NR4 7UH, United KingdomDepartment of Biochemistry and Molecular Biology, Universidade Federal do Paraná, CP 19046, CEP 81531-990 Curitiba, PR, BrazilDepartment of Basic Pathology, Universidade Federal do Paraná, CP 19046, CEP 81531-990 Curitiba, PR, Brazil

r t i c l e i n f o

rticle history:eceived 18 November 2014eceived in revised form 8 December 2014ccepted 14 December 2014vailable online 31 December 2014

eywords:ifA activationlnB activityutagenesis

a b s t r a c t

PII proteins are signal transduction that sense cellular nitrogen status and relay this signals to othertargets. Azospirillum brasilense is a nitrogen fixing bacterium, which associates with grasses and cerealspromoting beneficial effects on plant growth and crop yields. A. brasilense contains two PII encodinggenes, named glnB and glnZ.

In this paper, glnB was mutagenised in order to identify amino acid residues involved in GlnB signaling.Two variants were obtained by random mutagenesis, GlnBL13P and GlnBV100A and a site directedmutant, GlnBY51F, was obtained. Their ability to complement nitrogenase activity of glnB mutant strainsof A. brasilense were determined. The variant proteins were also overexpressed in Escherichia coli, purifiedand characterized biochemically. None of the GlnB variant forms was able to restore nitrogenase activityin glnB mutant strains of A. brasilense LFH3 and 7628. The purified GlnBY51F and GlnBL13P proteins could

not be uridylylated by GlnD, whereas GlnBV100A was uridylylated but at only 20% of the rate for wildtype GlnB. Biochemical and computational analyses suggest that residue Leu13, located in the � helix1 of GlnB, is important to maintain GlnB trimeric structure and function. The substitution V100A led toa lower affinity for ATP binding. Together the results suggest that NifA activation requires uridylylatedGlnB bound to ATP.

ntroduction

PII proteins compose a family of signal transducers found inrchaea, Bacteria and plants (Forchhammer 2004; Huergo et al.012). Proteins of this family sense nitrogen, carbon and energytatus, integrating them and relaying these signals to a variety ofrotein targets, especially proteins involved in the general nitrogenegulatory (Ntr) system (Arcondeguy et al. 2001).

PII proteins have a highly conserved structure with ∼112 aminocid residues, and many organisms have two or more genes encod-

ng PII proteins. In Escherichia coli there are two genes encodinglnB and GlnK (Cheah et al. 1994; Van Heeswijk et al. 1996).. coli GlnB is a homotrimer, each monomer of which contains two

∗ Corresponding author. Tel.: +55 42 32203731.E-mail addresses: jinaba@uepg.br (J. Inaba), eremy.thornton@jic.ac.uk

J. Thornton), huergo@ufpr.br (L.F. Huergo), rose.adele@terra.com.brR.A. Monteiro), giseli@ufpr.br (G. Klassen), fpedrosa@ufpr.br (F.d.O. Pedrosa),

ike.merrick@jic.ac.uk (M. Merrick), souzaem@ufpr.br (E.M. de Souza).

ttp://dx.doi.org/10.1016/j.micres.2014.12.005944-5013/© 2014 Elsevier GmbH. All rights reserved.

© 2014 Elsevier GmbH. All rights reserved.

�-helices and six �-strands connected by three conserved motifs:the T-loop, between residues 37–55; the B loop, between residues82–88; and the C loop between residues 102–105 (Cheah et al.1994). The T and Bloops of one subunit form a lateral cleft with theC-loop of another (Xu et al. 1998). The T loop has a highly conservedtyrosine residue (Tyr51) which is covalently modified dependingon the cellular nitrogen status by the GlnD protein, a bifunctionalenzyme with uridylyltransferase/uridylyl-removing activities thatincorporate an UMP residue at Tyr51 of PII under low nitrogen con-ditions. The UMP group is removed by GlnD acting as an uridylylremoving enzyme when the cellular levels of glutamine increase(Atkinson and Ninfa 1999; Son and Rhee 1987; Van Heeswijk et al.1996).

Azospirillum brasilense is a free living nitrogen fixing bacterium,which associates with several grasses and cereals promoting ben-eficial effects on plant growth and crop productivity (Steenhoudt

and Vanderleyden 2000; Okon and Itzigsohn, 1995). A. brasilenseencodes two PII genes, glnB and glnZ (De Zamaroczy et al. 1990,1996). Despite the high similarity (82%) of these proteins at theamino acid level they cannot form hybrid trimers, as reported for

6 ical Re

ERZtacKpbKnZt

isesboadgfNeNsemngb

cradTve

M

B

arN(wDa((2(aL(

R

b

6 J. Inaba et al. / Microbiolog

. coli GlnB and GlnK (Inaba et al. 2009; Van Heeswijk et al. 2000).ather they are involved in different regulatory processes (Deamaroczy 1998). A. brasilense GlnZ regulates ammonium influxhrough the ammonium transporter AmtB (Huergo et al. 2006) andlso participates in the post-translational control of nitrogenase byontrolling the activity of the enzyme DraG (Huergo et al. 2007;lassen et al. 2001). On the other hand, A. brasilense GlnB partici-ates in the reversible NH4

+ dependent inactivation of nitrogenasey controlling the activity of the enzyme DraT (Huergo et al. 2007;lassen et al. 2005) and GlnB is also necessary for activation of theif− specific transcriptional activator, NifA (Arsene et al. 1996; Deamaroczy et al. 1993). This latter activity of GlnB is the focus ofhis paper.

A. brasilense nifA is fully expressed under low oxygen levels andn the absence of ammonium ions and its repression involves aynergistic effect between oxygen and ammonium (Fadel-Pichetht al. 1999). The NifA protein contains three domains; an N terminalensory domain, a central activator domain and a C terminal DNAinding domain and it can sense both the cellular ammonium andxygen status. The N terminal domain of NifA is not essential for itsctivity and this domain exerts an inhibitory effect on NifA activityepending on the cellular nitrogen status (Arsene et al. 1996). Nitro-en sensing is mediated by GlnB acting through this domain sinceull length NifA is inactive in the absence of GlnB, but an N truncatedifA requires neither GlnB nor low fixed nitrogen for activity (Liangt al. 1992; De Zamaroczy et al. 1993). The GlnB protein binds to the

terminal domain of NifA (Chen et al. 2005; Sotomaior et al. 2012),uggesting a model for NifA activity control involving PII (Arsenet al. 1996). Furthermore, an A. brasilense glnD mutant had a Nifinus phenotype (Dommelen et al. 2002). In this mutant, GlnB is

ot covalently modified as cellular nitrogen levels decrease, sug-esting that GlnB-UMP is the form responsible for activating NifA,ut the exact mechanism of control is not yet known.

To further characterize GlnB residues necessary for NifA activityontrol we have produced two glnB point mutants obtained by aandom mutagenesis strategy, coding for GlnBL13P and GlnBV100And a GlnBY51F mutant obtained by site directed mutagenesis andetermined the activity of the encoded proteins in vitro and in vivo.he point mutations altered the uridylylation profile of the GlnBariants and their ability to activate NifA, and consequently nifHxpression, in A. brasilense.

aterials and methods

acterial strains and growth conditions

The bacterial strains of A. brasilense used in this work were: FP2, Nif+ spontaneous streptomycin (Smr) and nalidixic acid (Nalr)esistant mutant of Sp7, ATCC 29145 wild type, 7628 [glnB::kmalr Kmr (kanamycin) Nif−] (De Zamaroczy et al. 1996) and LFH3

Nalr, �glnB Nif−) (Huergo et al. 2006). The E. coli strains usedere S17.1 (RP4-2 Tc::Mu Km::Tn7 Tra recA) (R. Simon 1983),H10B (Smr, F′ [proAB+lacZ�M15]) (Green and Sambrook 2012)nd BL21 (�DE3) E. coli B F−dcm ompT hsdS(rB−mB−) gal �(DE3)Stratagene). A. brasilense strains were grown in NFbHP mediumMachado et al. 1991) with the required antibiotics containing0 mM NH4Cl (high ammonium conditions), or 5 mM glutamatenitrogenase derepressing conditions). The cultures were grown in

shaker at 30 ◦C and 120 rpm. The E. coli strains were cultured inuria–Bertani medium (LB) at 37 ◦C with the relevant antibioticsGreen and Sambrook 2012).

andom and site directed mutagenesis of glnB gene of A. brasilense

To obtain random mutants of the glnB gene protein a PCR-ased random mutagenesis using manganese and reduced dNTP

search 171 (2015) 65–72

concentrations was performed as described by Lin-Goerke et al.(1997). A XbaI/HindIII glnB fragment of pMLA–MLV1 plasmid(Araújo et al. 2004) was subcloned into pTZ18R (Mead et al.1986), generating pJI8, which expresses glnB-His6, under placpromoter. This plasmid was used as a template in the mutagenicPCR reaction using the Universal and Reverse primers, the productswere digested with KpnI/HindIII enzymes and cloned into pTZ18Rvector. Transformants were randomly chosen for sequencing andtwo clones with random substitutions were obtained: glnBL13P,and glnBV100A, named pJI6 and pJI9, respectively.

The site directed glnBY51F mutant was constructed by PCR usingspecific mutagenic primers pglnBY51F 5′-G GAA GTC GAC AAA CTCCAC C-3′. The substituted thymine to adenine in this position isshown in underlined and it is also contained a SalI restriction siteshowed in bold. The primer contained a mutated codon leading tothe change of tyrosine residue 51 (Y51) to phenylalanine (F51). ThepJI8 plasmid (0.5–1.0 ng) was used as template in a reaction sys-tem containing: PCR buffer (10 mmol/l Tris–HCl pH 8.3, 50 mmol/lKCl), 1.5 mmol/l MgCl2, 0.2 mmol/l dNTP, 10 pmol of reverse and themutagenic primer and 1.5 U of Taq DNA polymerase. The amplifi-cation reaction consisted of one cycle 95 ◦C – 5 min, followed by 30cycles of 94 ◦C – 30 s, 57 ◦C – 30 s and 72 ◦C – 2:30 min and a singlecycle of 72 ◦C – 10 min. The product of this reaction was digestedwith KpnI/SalI enzymes. A second amplification reaction was per-formed by using pJI8 plasmid as template in a system similar to thatabove using 10 pmol of forward and reverse primers, which flankthe insert of pTZ18R vector, to amplify the glnB gene of A. brasilense.The product of this reaction was digested with SalI/SacI. The vec-tor pUC18 digested with KpnI/SacI were linked to the product ofthe first amplification reaction digested with KpnI/SalI fragmentand the SalI/SacI product of the second amplification reaction. Theproduct of this ligation was transformed into E. coli DH10B andthe bacteria spread onto LA plates supplemented with ampicillin(250 �g/ml) and X-gal (0.03 mg/ml). Plasmid DNA from transfor-mant colonies was isolated and sequenced to confirm nucleotideexchange. A clone containing the desired substitution was selectedand named pJIB51.

To purify native and mutant variants of the GlnB protein, theNdeI/BamHI fragments from pJI6, pJI9 and pJIB51 were isolatedand cloned into pET29a (Novagen) yielding plasmids were namedpJI2.6, pJI2.9 and pJIB52, respectively, which express GlnB and itsvariants from the native translation start without the N terminalhistidine tag. To evaluate GlnB function in vivo, the glnB gene and itsvariants were isolated from the plasmids pJI8, pJI6, pJI9 and pJIB52by enzymatic digestion (XbaI/HindIII) and cloned into the broad-host range vector pLAFR3.18 (Machado et al. 1995), yielding pJI8.2,pJI10, pJI11 and pJIB53, respectively. These plasmids were trans-formed into E. coli S17-1, then conjugated into A. brasilense wildtype strain (FP2) and into the glnB mutants strains 7628 (glnB::km)and LFH3 (�glnB) as described (Pedrosa and Yates 1984).

Nitrogenase activity

A. brasilense was grown in NFbHP medium supplemented with5 mM glutamate for 18 h, at 120 rpm. The nitrogenase activitywas measured by the acetylene reduction method (Pedrosa andYates 1984). The protein content was determined as described byBradford (1976), after cell lysis with 0.2 M NaOH.

Western blot analysis

For GlnB and NifH immunodetection in cell-free extracts of A.

brasilense, proteins were separated by a 12.5% (GlnB) or 10% (NifH)SDS-PAGE and transferred to an ECL nitrocellulose membrane (GEhealthcare) and probed with anti-A. brasilense GlnB (1/5000) andanti H. seropedicae NifH (1/5000) (Huergo et al. 2006). The second

J. Inaba et al. / Microbiological Research 171 (2015) 65–72 67

Fig. 1. Alignment of the amino acid sequences of constructed mutants and native GlnB. Alignment of the amino acid sequences of the variant forms of A. brasilense GlnBL13P(AbGlnBL13P), GlnBY51F (AbGlnBY51F), GlnBV100A (AbGlnBV100A) and the GlnB and GlnZ (or GlnK) proteins of A. brasilense (Ab), E. coli (Ec), H. seropedicae (Hs) and R.r ets, 31w rved

s

abwct

E

p[BL0bTaaa3HAafds

I

1tG26on

N

TaFp

brasilense strain (FP2) showed a nitrogenase activity of 34.4 nmolof ethylene mg protein−1 min−1 under nitrogen fixing conditions.The glnB mutant 7628 had no nitrogenase activity, but when

Table 1Effect of GlnB, GlnBL13P, GlnBV100A and GlnBY51F on nitrogenase activity inA. brasilense. Nitrogenase activities are given in nmol of ethylene min−1 mg ofprotein−1. The results represent the means and standard deviations of three inde-pendent experiments.

Strain Plasmid Nitrogenaseactivity

FP2 34.4 ± 2.97628(glnB::Tn5) 07628(glnB::Tn5) pJI8.2(A. brasilense glnB6His) 14.0 ± 1.157628(glnB::Tn5) pJI10(A. brasilense glnBL13-P6His) 07628(glnB:Tn5) pJI11(A. brasilense glnBV100-A6His) 0LFH3(�glnB) 0

ubrum (Rr). The substituted residues are identified by a box. The � helix, the � sheere aligned using the CLUSTAL W program (Thompson et al. 1994) and the conse

imilarity and dots represent the low similarity amino acids.

ntibody used was a horseradish peroxidase conjugate anti rab-it immunoglobulin (GE healthcare) (1:5000 dilution). The signalsere detected by using the ECL development reagent (GE health-

are) and the image was captured by an UVP Imaging System usinghe Labwork4 program (UVP, Inc., Upland, CA, USA).

xpression and purification of native and mutant forms of GlnB

To overexpress and purify native and mutant forms of GlnB,lasmids pLH25 (containing A. brasilense glnB cloned into pET29a(+)16], pJI2.6, pJI2.9 or pJI53 were introduced into the E. coli strainL21(�DE3) and cultured at 20 ◦C in flasks containing 800 ml ofB medium supplemented with 50 �g/ml Km, 30 mM NH4Cl and.5 mM IPTG for 18 h. Cells were kept on ice for 30 min, harvestedy centrifugation and re-suspended in a buffer containing 50 mMris–HCl, pH 7.5, 50 mM KCl and 20% glycerol. Cells were broken in

French press cell (at 2000 psi) for three times and heat treatedt 80 ◦C for 3 min. An EDTA free protease inhibitor cocktail wasdded and cell debris removed by centrifugation at 30,000 × g for0 min, at 4 ◦C. The soluble proteins fractions were applied in ai-trap Heparin column (GE Healthcare) equilibrated with buffer

(50 mM Tris–HCl pH 7.5, 50 mM KCl). Proteins were eluted by linear gradient of KCl 0.1–1.0 M. Dialyzed samples were keptrozen at −80 ◦C before SDS–PAGE. Protein concentrations wereetermined by the Bradford assay using bovine serum albumin astandard.

n vitro uridylylation of GlnB

Uridylylation assays were performed in a reaction containing0 �M of purified GlnB, GlnBL13P, GlnBV100A or GlnBY51F pro-eins, 0.1 mM ATP, 1 mM UTP, 2 mM 2-oxoglutarate, and 0.1 �MlnD in a buffer containing 100 mM Tris–HCl pH 7.5, 100 mM KCl,5 mM MgCl2. Samples were collected at time 0, 5, 10, 20, 30 and0 min and the reaction stopped by adding 1 mM EDTA and 5 �lf electrophoresis sample buffer. The products were analyzed byative gel electrophoresis.

on-denaturing protein gel electrophoresis

Samples were mixed with 5 �l of loading buffer (62.5 mM

ris–HCl pH 6.8, 10% glycerol, 0.01% bromophenol blue) and sep-rated in a non-denaturing gel electrophoresis as described byorchhammer and Tandeau de Marsac (1994). Electrophoresis waserformed at 4 ◦C and 100 V.

0 helix and the T, C and B loops are presented above the sequences. The sequencesregions are shaded. Asterisks represent the identical amino acids, colons the high

MALDI-TOF analysis

Samples of PII uridylylation reaction were collected at time 0,10, 30 and 60 min and mixed with a saturated sinapinic acid solu-tion at a ratio of 1:9 (sample:acid). The samples were spotted on aMALDI plate (Bruker Daltonics) and the mass spectra obtained ona MALDI-TOF Autoflex mass spectrometer (Bruker Daltonics, Bre-men, Germany) using positive linear mode, accelerating voltage of20 kV, delay time of 330 ns and acquisition mass-range 5–20 kDa.The spectra were analyzed by the FlexAnalysis 2.0 software (BrukerDaltonics, Bremen, Germany).

Results

The A. brasilense GlnB variants GlnBV100A, GlnBL13P andGlnBY51F prevent expression of nitrogenase

Two strategies to obtain point mutations of A. brasilense glnBwere used. First, a random mutagenesis strategy and secondly asite directed mutagenesis to change the uridylylation site of GlnB.This allowed obtaining three GlnB variants (GlnBL13P, GlnBY51Fand GlnBV100A), which have amino acid changes located in �1helix (GlnBL13P), the T loop (GlnBY51F) and C loop (GlnBV100A)of GlnB (Fig. 1). The ability of these variants to activate NifAin vivo was evaluated by measuring nitrogenase activity in A.brasilense strains expressing a 6His tagged version of GlnB andits derivatives from the plac promoter (Table 1). The wild type A.

LFH3(�glnB) pJI8.2 (A. brasilense glnB6His) 47.1 ± 9.4LFH3(�glnB) pJI10(A. brasilense glnBL13-P6His) 0LFH3(�glnB) pJI11(A. brasilense glnBV100-A6His) 0LFH3(�glnB) pJIB53 (A. brasilense glnBY51F6His) 0

68 J. Inaba et al. / Microbiological Research 171 (2015) 65–72

Fig. 2. Expression of GlnB6His, GlnBY51F6His and NifH in A. brasilense. Five micro-grams of crude extracts from cultures grown under nitrogen fixing conditions wereseparated by SDS PAGE and analyzed by Western blot using polyclonal anti PIIand NifH antibodies. Lanes: 1. purified GlnB6His (50 ng); 2. FP2 strain; 3. LFH3 (glnBma

pteGbngscpaawaon

rnLntwm1Hna

G

scsfp(bt

osaAwVd

Fig. 3. Native PAGE profile of purified native and mutant forms of GlnB. Five micro-

utant); 4. LFH3 pJI5 (GlnB6His, plac); 5. LFH3 pJIB53 (GlnBY51F6His, plac). Arrows, – wild type GlnB; b – GlnB6His.

lasmid pJI8.2, expressing A. brasilense GlnB6His, was introducedhe Nif+ phenotype of this mutant was restored (14 nmol ofthylene mg protein−1 min−1). In contrast, the expression oflnBL13P6His, GlnBV100A6His and GlnBY51F6His in strain 7628,y plasmids pJI10, pJI11 and pJIB53 respectively, did not restoreitrogenase activity. None of these plasmids affected A. brasilenserowth. In A. brasilense the glnA gene, which encodes glutamineynthetase (GS), is located immediately downstream from glnB andonsequently the glnB insertional mutation in strain 7628 has aolar effect on glnA expression (De Zamaroczy 1998). This mutationlso indirectly affects GlnZ deuridylylation (Huergo et al. 2006)nd probably affects the uridylylation status of GlnB. Consequentlye also analyzed nitrogenase activity in strain LFH3 which carries

n in frame glnB deletion (Huergo et al. 2006). In LFH3, as in 7628,nly the strain expressing wild type GlnB6His was able to restoreitrogenase activity (Table 1), confirming the previous results.

To evaluate whether GlnBY51F was acting on the transcriptionalegulation of nif genes via NifA, or on posttranslational control ofitrogenase via DraT/DraG system, the A. brasilense glnB mutantFH3, expressing either GlnB6His or GlnBY51F6His was grown underitrogen fixing conditions and tested for the presence of NifH pro-ein by western blotting using anti NifH antibody (Fig. 2). NifHas detected in wild type A. brasilense FP2, but not in the glnButant strain (LFH3) as previously shown (De Zamaroczy et al.

993; Huergo et al. 2006) and not in the strain expressing GlnBY51F.ence, GlnBY51F is unable to substitute wild type GlnB to activateifH expression, suggesting that it has its effect at the level of NifActivity.

lnBL13P affects the quaternary structure of the protein

In order to further characterize the effects of the amino acidubstitutions, each the GlnB variant proteins was purified andompared with the wild type. To evaluate effects on the nativetructure of GlnB, non-denaturating gel electrophoresis was per-ormed. GlnB, GlnBV100A and GlnBY51F had the same migrationattern, showing a single band corresponding to trimeric GlnBFig. 3, lanes 1–3). In contrast GlnBL13P showed at least two majorands (Fig. 3, lane 4), suggesting that Leu13 is an important residueo maintain GlnB quaternary structure.

Residue Leu13 is a conserved residue in PII proteins and analysisf structural models of the GlnB variants (Fig. 4) shows that the sub-titution L13P, leads to disruption of � helix 1 with the formation of

loop in this region, creating additional hydrogen bonds (betweensp67-Ile112, Arg101-Thr104, Gly108-Asp110 and Gly109-Ile112)hen compared to the GlnB wild type. The substitutions Y51F and100A caused fewer predicted alterations in the structure (Fig. 4),espite affecting GlnB activity.

grams of A. brasilense GlnB (lane 1), GlnBY51F (lane 2), GlnBV100A (lane 3) andGlnBL13P (lane 4) were subjected to native-page electrophoresis (10%) and stainedwith Coomassie blue.

GlnBV100A, GlnBL13P and GlnBY51F are affected in uridylylation

To evaluate the influence of the substitutions on PII uridylyla-tion, native and variants of GlnB were purified and uridylylationreactions were performed. Wild type GlnB was 30% uridylylated5 min after initiation of the reaction and reached about 90% uridy-lylation after 60 min (Fig. 5). In contrast, GlnBV100A was only 3%modified after 5 min and only 23% was found as GlnBV100A-UMPafter 60 min. When analyzing the GlnBL13P profile on SDS-PAGEno modification was observed in a time course experiment. Toconfirm the GlnB modification results, MALDI-TOF analyses wereperformed. At time zero, unmodified GlnB was detected for thewild-type and all three variants (peak at m/z of 12,355) (Fig. 6A–D).After 60 min of reaction, the modified form (m/z of 12,674, whichcorresponds to the monomer linked to UMP) was observed only forGlnB and GlnBV100A but with lower intensity in the latter protein(Fig. 6A and D). The results confirm that GlnBV100A has a loweruridylylation rate compared to wild type GlnB and that neitherGlnBL13P nor GlnBY51F are uridylylated.

GlnB variants V100A and Y51F are defective in ATP binding

The ability of PII to bind ATP is well established and theC-terminal region of PII proteins is involved in ATP binding. Inter-action with this effector, together with 2-oxoglutarate, controlsPII interaction with target proteins such as NtrC and glutaminesynthetase (Jiang and Ninfa 2007). A number of structures of PIIproteins with ATP bound have been solved, as for A. brasilense GlnZ(Truan et al. 2014) and those for E. coli GlnB and GlnK. Whilstthe details of the interactions vary slightly, residues from the Bloop (Gly87 and Gly89) and from the adjacent C loop (particu-larly Arg101 and Arg103) are major contacts (Xu et al. 1998, 2001).Hence, alteration of residue Val100, as in GlnBV100A, might beexpected to affect ATP binding.

To analyze whether ATP binding was affected by the singleamino acid substitutions, ITC experiments were carried out inwhich the GlnB, GlnBY51F and GlnBV100A proteins were titrated

with ATP in the presence of MgCl2 and 2-oxoglutarate (Fig. 7). Thebinding equilibrium of ATP to GlnB proteins can be evaluated by theheat changes caused when the ligand attaches to its partner. Theresults show that GlnBY51F and GlnBV100A had a lower affinity for

J. Inaba et al. / Microbiological Research 171 (2015) 65–72 69

Fig. 4. Comparison between structure models of GlnB and GlnB mutant forms of A. bra3Djigsaw program. T loop of GlnB (a) and GlnBY51F (b); �1 helix of GlnB (c) and GlnBL13are indicated by arrows.

Fig. 5. Uridylylation of GlnB, GlnBV100A, GlnBY51F and GlnBL13P proteins. Thereaction was stopped at time 0 (lane 1), 5 (lane 2), 10 (lane 3), 20 (lane 4), 30 (lane 5)and 60 min (lane 6), loaded on non-denaturing polyacrylamide gels and stained withCoomassie blue. The unmodified (a) and modified PII-UMP (b) forms are indicatedby arrows. The percentage of modified forms in relation of the total amount of theprotein (b%) was determined by densitometry with the exception of GlnBL13P (nd.– not determined).

silense. GlnB wild type and variant structure models were constructed using theP (d); C loop of GlnB (e) and GlnBV100A (f). The altered amino acids of each variant

ATP when compared to wild type GlnB and in fact, this effect wasmuch more drastic in GlnBV100A supporting that V100A substitu-tion affected ATP binding.

Discussion

The highly conserved structure of PII suggests that particularresidues are likely to play key roles in its multiple functions. PIIproteins of different organisms have been shown to bind differ-ent targets such as NtrB (Jiang and Ninfa 2009), AmtB in E. coli, A.brasilense and R. rubrum (Huergo et al. 2006; Conroy et al. 2007;Wolfe et al. 2007), DraT, DraG, and NifA in Rhodospirillum rubrumand A. brasilense (Arsene et al. 1996; Chen et al. 2005; Huergo et al.2006; Rajendran et al. 2011; Sotomaior et al. 2012) and PipX (Llaceret al. 2010; Zhao et al. 2010) and NAGK in the cyanobacterium Syne-choccocus elongates (Llacer et al. 2007). Because of the broad rangeof targets it is possible that different parts of GlnB are responsi-ble for interaction with different proteins. Indeed, the interactionof AmtB with PII occurs by T loop contact (Conroy et al. 2007)whereas in the DraG-GlnZ complex the T loop does not partici-pate in protein–protein contacts (Rajendran et al. 2011). Moreover,the T loop can interact with target proteins in different conforma-tions. For example, in the PII-PipX complex, one PII trimer bindsto three monomers of PipX through an extended conformationof the T loops (Llacer et al. 2010; Zhao et al. 2010). The aminoacid substitutions studied in the work reported here are locatedin highly conserved regions of PII proteins. The Leu13 and Tyr51residues are almost invariant and residue 100 is nearly alwayseither valine or isoleucine (Fig. 1). The substitution L13P is pre-

dicted to disrupt � helix1 and our data indicate that this affects thetrimeric structure of GlnB. This variant was also unable to be uridy-lylated by GlnD, suggesting that the trimeric structure of GlnB isnecessary for GlnD interaction and uridylylation. The Tyr51 residue

70 J. Inaba et al. / Microbiological Research 171 (2015) 65–72

Fig. 6. MALDI-TOF analysis of the uridylylation products of GlnB, GlnBL13P, GlnBY51F and GlnBV100A. MALDI-TOF analysis of the uridylylation products of GlnB (a), GlnBL13P( n. The1 e rati

i(isoiePiTa

b), GlnBY51F (c) and GlnBV100A (d) collected at times 0, 30 and 60 min of reactio2,350 and 12,670 respectively. The mass spectra were obtained using a signal/nois

n the T loop is the site of uridylylation by GlnD in many PII proteinsHuergo et al. 2012). In the present study, we have shown that Y51Fs unable to be uridylylated despite having a conserved trimerictructure and ATP binding ability, confirming that Y51F is the sitef UMP modification in A. brasilense GlnB. Although residue V100s located in the lateral cleft between two monomers (Rajendrant al. 2011; Xu et al. 2001) substitution by alanine did not alter the

II trimeric structure but did alter ATP binding ability. This residues usually a hydrophobic amino acid such as valine or isoleucine.hus, it is possible that the hydrophobic side chain of thesemino acids is important to maintain ATP binding. The binding of

unmodified and modified forms of GlnB correspond to the peaks with m/z abouto of 30.

effectors, either ATP and 2-oxoglutarate or ADP, within the lateralcleft of the PII trimer has a significant effect on the conformation ofthe protein and is therefore likely to influence the ability of the pro-tein to be uridylylated by GlnD (Truan et al. 2010). The significantlyreduced ATP binding ability of the V100A variant is consequentlyquite likely to influence the conformation of the PII trimer, therebyaccounting for its significantly impaired uridylylation properties.

In A. brasilense, NifA is active only in the presence of GlnB,since in a glnB mutant NifA is present but unable to activate nifHexpression (Arsene et al. 1996). When the N terminal domain ofNifA is deleted GlnB is no longer needed to activate NifA and, as a

J. Inaba et al. / Microbiological Re

Fig. 7. ATP titration of GlnB, GlnBV100A and GlnBY51F proteins. Calorimetric outputfrom binding experiments of ATP to GlnB (black), GlnBY51F (blue), and GlnBV100A(red). The experiments were carried out in a microcalorimeter VP-ITC (Microcal).The purified and dialyzed GlnB (107.8 �M), GlnBY51F (67.4 �M) and GlnBV100A(136.6 �M) were titrated by ATP (3.8, 5.1 and 10.3 mM, respectively). The molarrit

csoe

isfi(unliGipiGa

osG

atio refers to ligand concentration/protein concentration in the cell reaction. (Fornterpretation of the references to color in this figure legend, the reader is referredo the web version of this article.)

onsequence, NifA activity is not influenced by the cellular nitrogentatus. Hence, GlnB is necessary for nitrogen control and activationf NifA through interaction with the N terminal domain (Arsenet al. 1996,Chen et al. 2005).

Our characterization of three GlnB variants offers further insightnto the role of GlnB in regulating NifA activity in A. brasilense. It washown previously that an A. brasilense glnD mutant was unable tox nitrogen suggesting that GlnB-UMP is essential for NifA activityDommelen et al. 2002). Our experiments showed that GlnBY51F isnable to be uridylylated and fails to complement a glnB mutant foritrogenase activity, consistent with the concept that the uridyly-

ated form of GlnB is essential for NifA activity, and potentially fornteraction of GlnB with the N terminal domain of NifA. Data forlnBV100A also support this hypothesis. This variant is impaired

n ATP binding leading to a potential conformational change thatrevents optimal uridylylation and, like GlnBY51F, it is impaired

n nitrogenase activity and by implication in NifA activity. Finally,lnBL13P does not form trimers, is again impaired in uridylylationnd hence is not capable of activating NifA.

In summary, our characterization of a number of variantsf A. brasilense GlnB containing single amino acid substitutionsuggest that ATP binding and uridylylation are important forlnB-dependent activation of NifA. Furthermore, the conserved

search 171 (2015) 65–72 71

residue Leu13 is important to maintain the trimeric structure ofGlnB.

Author’s contributions

JI participated in the study design, carried out the experimen-tal studies, data analysis and was responsible for the manuscriptwriting and submission of the final manuscript. JT participatedin the experimental studies of protein purification and ITC. LFH,RAM, FOP participated in study design and assisted with data anal-ysis. MM contributed to study design, data analysis and supervisedmanuscript writing. GK and EMS conceived the study, partici-pated in study design, data analysis, supervised and supervisedmanuscript writing. All authors read and approved the manuscript.

Acknowledgements

This research was supported by CAPES, National Institute ofScience and Technology on Biological Nitrogen Fixation (INCT-FBN/CNPq-MCT), Fundac ão Araucária, CNPq, and Instituto doMilênio/CNPq. We thank Valter Baura, Roseli Prado and Julieta Piefor technical assistance and Luíza Araújo for the anti GlnB poly-clonal antibodies.

References

Araújo LM, Monteiro RA, de Souza EM, Steffens MBR, Rigo LU, Pedrosa FO, et al. GlnBis specifically required for Azospirillum brasilense NifA activity in Escherichia coli.Res Microbiol 2004;155(6):491–5.

Arcondeguy T, Jack R, Merrick M. PII signal transduction proteins, pivotal players inmicrobial nitrogen control. Microbiol Mol Biol Rev 2001;65(1):80–105.

Arsene F, Kaminski PA, Elmerich C. Modulation of NifA activity by PII in Azospiril-lum brasilense: evidence for a regulatory role of the NifA N-terminal domain. JBacteriol 1996;178(16):4830–8.

Atkinson MR, Ninfa AJ. Characterization of the GlnK protein of Escherichia coli. MolMicrobiol 1999;32(2):301–13.

Bradford MM. A rapid and sensitive method for the quantification of microgramsquantities of protein utilization the principle of protein-dye binding. AnalBiochem 1976;72:248–54.

Cheah E, Carr PD, Suffolk PM, Vasudevan SG, Dixon NE, Ollis DL. Structure ofthe Escherichia coli signal transducing protein PII. Struct Lond Engl 19931994;2(10):981–90.

Chen S, Liu L, Zhou X, Elmerich C, Li J-L. Functional analysis of the GAF domain ofNifA in Azospirillum brasilense: effects of Tyr→Phe mutations on NifA and itsinteraction with GlnB. Mol Genet Genomics 2005;273(5):415–22.

Conroy MJ, Durand A, Lupo D, Li X-D, Bullough PA, Winkler FK, et al. The crystalstructure of the Escherichia coli AmtB-GlnK complex reveals how GlnK regulatesthe ammonia channel. Proc Natl Acad Sci U S A 2007;104(4):1213–8.

Dommelen AV, Keijers V, Somers E, Vanderleyden J. Cloning and characterisation ofthe Azospirillum brasilense glnD gene and analysis of a glnD mutant. Mol GenetGenomics 2002;266(5):813–20.

Fadel-Picheth CM, de Souza EM, Rigo LU, Funayama S, Yates MG, Pedrosa FO. Regula-tion of Azospirillum brasilense nifA gene expression by ammonium and oxygen.FEMS Microbiol Lett 1999;179(2):281–8.

Forchhammer K. Global carbon/nitrogen control by PII signal transduction incyanobacteria: from signals to targets. FEMS Microbiol Rev 2004;28(3):319–33.

Forchhammer K, Tandeau de Marsac N. The PII protein in the cyanobacterium Syne-chococcus sp. strain PCC 7942 is modified by serine phosphorylation and signalsthe cellular N-status. J Bacteriol 1994;176(1):84–91.

Lin-Goerke JL, Robbins DJ, Burczak JD. PCR-based random mutagenesis using man-ganese and reduced dNTP concentration. Biotechniques 1997;23(3):409–12.

Green MR, Sambrook J. Molecular cloning: a laboratory manual. Cold Spring Harbor,NY: Cold Spring Harbor Laboratory Press; 2012.

Van Heeswijk WC, Hoving S, Molenaar D, Stegeman B, Kahn D, WesterhoffHV. An alternative PII protein in the regulation of glutamine synthetase inEscherichia coli. Mol Microbiol 1996;21(1):133–46.

Van Heeswijk WC, Wen D, Clancy P, Jaggi R, Ollis DL, Westerhoff HV, et al. TheEscherichia coli signal transducers PII (GlnB) and GlnK form heterotrimersin vivo: fine tuning the nitrogen signal cascade. Proc Natl Acad Sci U S A2000;97(8):3942–7.

Huergo LF, Chubatsu LS, de Souza EM, Pedrosa FO, Steffens MBR, Merrick M. Inter-actions between PII proteins and the nitrogenase regulatory enzymes DraT and

DraG in Azospirillum brasilense. FEBS Lett 2006;580(22):5232–6.

Huergo LF, Merrick M, Pedrosa FO, Chubatsu LS, Araujo LM, de Souza EM. Ternarycomplex formation between AmtB, GlnZ and the nitrogenase regulatory enzymeDraG reveals a novel facet of nitrogen regulation in bacteria. Mol Microbiol2007;66(6):1523–35.

7 ical Re

H

I

J

J

K

K

L

L

L

M

M

M

O

P

R

2 J. Inaba et al. / Microbiolog

uergo LF, Pedrosa FO, Muller-Santos M, Chubatsu LS, Monteiro RA, Merrick M, et al.PII signal transduction proteins: pivotal players in post-translational control ofnitrogenase activity. Microbiology 2012;158(1):176–90.

naba J, Huergo LF, Bonatto AC, Chubatsu LS, Monteiro RA, Steffens MB, et al. Azospir-illum brasilense PII proteins GlnB and GlnZ do not form heterotrimers and GlnBshows a unique trimeric uridylylation pattern. Eur J Soil Biol 2009;45(1):94–9.

iang P, Ninfa AJ. Escherichia coli PII signal transduction protein controlling nitrogenassimilation acts as a sensor of adenylate energy charge in vitro. Biochemistry(Mosc) 2007;46(45):12979–96.

iang P, Ninfa AJ. Sensation and signaling of �-ketoglutarate and adenylylate energycharge by the Escherichia coli PII signal transduction protein require coopera-tion of the three ligand-binding sites within the PII trimer. Biochemistry (Mosc)2009;48(48):11522–31.

lassen G, de Souza EM, Yates MG, Rigo LU, Inaba J, de Oliveira Pedrosa F. Control ofnitrogenase reactivation by the GlnZ protein in Azospirillum brasilense. J Bacteriol2001;183(22):6710–3.

lassen G, de Souza EM, Yates MG, Rigo LU, Costa RM, Inaba J, et al. Nitrogenaseswitch-off by ammonium ions in Azospirillum brasilense requires the GlnB nitro-gen signal-transducing protein. Appl Environ Microbiol 2005;71(9):5637–41.

iang YY, de Zamaroczy M, Arsene F, Paquelin A, Elmerich C. Regulation of nitrogenfixation in Azospirillum brasilense Sp7: involvement of nifA, glnA and glnB geneproducts. FEMS Microbiol Lett 1992;100(1–3):113–9.

lacer JL, Contreras A, Forchhammer K, Marco-Marin C, Gil-Ortiz F, Maldonado R,et al. The crystal structure of the complex of PII and acetylglutamate kinasereveals how PII controls the storage of nitrogen as arginine. Proc Natl Acad SciU S A 2007;104(45):17644–9.

lacer JL, Espinosa J, Castells MA, Contreras A, Forchhammer K, Rubio V. Structuralbasis for the regulation of NtcA-dependent transcription by proteins PipX andPII. Proc Natl Acad Sci U S A 2010;107(35):15397–402.

achado HB, Funayama S, Rigo LU, Pedrosa FO. Excretion of ammonium byAzospirillum brasilense mutants resistant to ethylenediamine. Can J Microbiol1991;37(7):549–53.

achado HB, Yates MG, Funayama S, Rigo LU, Steffens MB, de Souza EM, et al. ThentrBC genes of Azospirillum brasilense are part of a nifR3-like-ntrB-ntrC operonand are negatively regulated. Can J Microbiol 1995;41(8):674–84.

ead DA, Szczesna-Skorupa E, Kemper B. Single-stranded D.N.A. blue T7 promoterplasmids: a versatile tandem promoter system for cloning and protein engineer-ing. Protein Eng 1986;1(1):67–74.

kon Y, Itzigsoh R. The development of Azospirillum as a commercial inoculant forimproving crop yields. Biotechnol Adv 1995;13(3):415–24.

edrosa FO, Yates Mg. Regulation of nitrogen fixation (nif) genes of Azospiril-

lum brasilense by nifA and ntr (gln) type gene products. FEMS Microbiol Lett1984;23(1):95–101.

ajendran C, Gerhardt ECM, Bjelic S, Gasperina A, Scarduelli M, Pedrosa FO, et al.Crystal structure of the GlnZ-DraG complex reveals a different form of PII-targetinteraction. Proc Natl Acad Sci U S A 2011;108(47):18972–6.

search 171 (2015) 65–72

R. Simon UP. A broad host range mobilization system for in vivo genetic engi-neering: transposon mutagenesis in gram negative bacteria. Nat Biotechnol1983;1(9):784–91.

Son HS, Rhee SG. Cascade control of Escherichia coli glutamine synthetase. Purifica-tion and properties of PII protein and nucleotide sequence of its structural gene.J Biol Chem 1987;262(18):8690–5.

Sotomaior P, Araujo LM, Nishikawa CY, Huergo LF, Monteiro RA, Pedrosa FO, et al.Effect of ATP and 2-oxoglutarate on the in vitro interaction between the NifAGAF domain and the GlnB protein of Azospirillum brasilense. Braz J Med Biol Res2012;45(12):1135–40.

Steenhoudt O, Vanderleyden J. Azospirillum, a free-living nitrogen-fixing bacteriumclosely associated with grasses: genetic, biochemical and ecological aspects.FEMS Microbiol Rev 2000;24(4):487–506.

Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W: improving the sensitivityof progressive multiple sequence alignment through sequence weighting,position-specific gap penalties and weight matrix choice. Nucleic Acids Res1994;22(22):4673–80.

Truan D, Bjelic S, Li X-D, Winkler FK. Structure and thermodynamics of effectormolecule binding to the nitrogen signal transduction PII protein GlnZ fromAzospirillum brasilense. J Mol Biol 2014;426(15):2783–99.

Truan D, Huergo LF, Chubatsu LS, Merrick M, Li X-D, Winkler FK. A newPII protein structure identifies the 2-oxoglutarate binding site. J Mol Biol2010;400(3):531–9.

Wolfe DM, Zhang Y, Roberts GP, Specificity. Regulation of interaction between thePII and AmtB1 proteins in Rhodospirillum rubrum. J Bacteriol 2007;189(19):6861–9.

Xu Y, Carr PD, Huber T, Vasudevan SG, Ollis DL. The structure of the PII–ATP complex.Eur J Biochem 2001;268(7):2028–37.

Xu Y, Cheah E, Carr PD, van Heeswijk WC, Westerhoff HV, Vasudevan SG, et al.GlnK, a PII-homologue: structure reveals ATP binding site and indicates howthe T-loops may be involved in molecular recognition. J Mol Biol 1998;282(1):149–65.

De Zamaroczy M. Structural homologues PII and Pz of Azospirillum brasilense provideintracellular signalling for selective regulation of various nitrogen-dependentfunctions. Mol Microbiol 1998;29(2):449–63.

De Zamaroczy M, Delorme F, Elmerich C. Characterization of three differentnitrogen-regulated promoter regions for the expression of glnB and glnA inAzospirillum brasilense. Mol Genet Genomics 1990;224(3):421–30.

De Zamaroczy M, Paquelin A, Elmerich C. Functional organization of the glnB-glnAcluster of Azospirillum brasilense. J Bacteriol 1993;175(9):2507–15.

De Zamaroczy M, Paquelin A, Peltre G, Forchhammer K, Elmerich C. Coexistence of

two structurally similar but functionally different PII proteins in Azospirillumbrasilense. J Bacteriol 1996;178(14):4143–9.

Zhao M-X, Jiang Y-L, Xu B-Y, Chen Y, Zhang C-C, Zhou C-Z. Crystal structure of thecyanobacterial signal transduction protein PII in complex with PipX. J Mol Biol2010;402(3):552–9.