Regulat ion of Glutamine Synthetase in the Diazotroph Rhodospi r i l lum rubrum

Anders Jonsson

Regulation of Glutamine Synthetase in the Diazotroph

Rhodospirillum rubrum

Anders Jonsson

Stockholm University

Doctoral thesis

©Anders Jonsson, Stockholm 2007

ISBN 978-91-7155-495-6, pp 1-63

Printed in Sweden by Universitetsservice AB, Stockholm 2007

Distributor: Department of Biochemistry and Biophysics

Tillägnad min Mamma

..jag har väntat så länge på just den här dan, och de e skönt när den

äntligen kommer..

List of Publications

This thesis is based on the following publications, referred to by their Roman numerals:

I. Jonsson, A. and Nordlund, S. 2007. In vitro studies of the uridylylation of the three PII protein paralogs from Rhodospirillum rubrum: the transferase activity of R. ru-

brum GlnD is regulated by α-ketoglutarate and divalent cations but not by glutamine. J Bacteriol 189:3471-78.

II. Jonsson

#, A., Teixeira#, P. F. and Nordlund, S. 2007. The

activity of adenylyltransferase in Rhodospirillum rubrum is only affected by α-ketoglutarate and unmodified PII pro-teins, but not by glutamine, in vitro. FEBS J 274:2449-60.

III. Jonsson

#, A., Teixeira#, P. F. and Nordlund, S. 2007. A

novel peroxiredoxin activity is located within the C-terminal end of Rhodospirillum rubrum adenylyltrans-ferase. J Bacteriol (Under revision).

IV. Jonsson

#, A., Teixeira#, P. F. and Nordlund, S. 2007. Re-

duced activity of glutamine synthetase in Rhodospirillum

rubrum mutants lacking adenylyltransferase, GlnE. Manu-script.

# the authors have contributed equally to the publication

Contents

Introduction .....................................................................................................1 Diazotrophic bacteria ....................................................................................................2 Rhodospirillum rubrum ..................................................................................................2 Nitrogenase...................................................................................................................3 Post-translational and genetic regulation of nitrogenase ...............................................4 The master nif gene activator NifA ................................................................................5

Covalent modification of glutamine synthetase (GS)....................................10 Structure and mechanistic aspects of GS ...................................................................11 Feedback inhibition .....................................................................................................12 Oxidative modification of GS.......................................................................................12 Other ways of GS regulation .......................................................................................13

GlnD and the PII proteins monocycle ...........................................................15 GlnD activity................................................................................................................15 Uridylylation/deuridylylation of R. rubrum PII proteins .................................................17 Other GlnD studies......................................................................................................18 PII proteins..................................................................................................................19 R. rubrum PII proteins .................................................................................................20 PII protein structures and specific point mutations ......................................................21 Other PII protein 3D structures....................................................................................22 GlnK-AmtB complexes ................................................................................................23 Ammonia transporters (AmtB).....................................................................................24

Adenylylation/deadenylylation of GS ............................................................25 The bifunctional enzyme GlnE ....................................................................................27 GlnE in R. rubrum .......................................................................................................28 Integration of signals ...................................................................................................30 Is our in vitro data relevant for in vivo studies?............................................................31 Peroxiredoxin activity of GlnE (Paper III).....................................................................35 glnE mutation (Paper IV).............................................................................................38 Concluding remarks ....................................................................................................39

En enkel förklaring till vad jag gjort ...............................................................43

Acknowledgements .......................................................................................44

References....................................................................................................46

Abbreviations

σ Sigma factor of RNA polymerase NTP Nucleotide triphosphate Å Ångström (10-10 m) Kd Dissociation constant pKa Acid dissociation constant CTAB Hexadecyltrimethylammonium bromide TCA Tricarboxylic acid cycle Prxs Peroxiredoxins SDS-PAGE Sodiumdodecylsulfatepolyacrylamidegelelectrophoresis

1

Introduction

Why is nitrogen important? If we take a close look at compounds that are essential to life i.e. DNA,

RNA, proteins, cofactors, there is a nitrogen atom or several nitrogen atoms within these essential compounds. Of course, there are other important ele-ments, but nitrogen is definitely one of the most crucial for sustaining life. Nitrogen atoms in the air exist as dinitrogen (N2) comprising around 80 % of the atmosphere. Dinitrogen is an inert gas and its unavailable to most living organisms. However there are organisms that can convert atmospheric nitro-gen to a more accessible form namely ammonia. Rhodospirillum rubrum is one of these organisms, the so called diazotrophs. The diazotrophs contain an enzyme that catalyzes the conversion of the inert dinitrogen to ammonia, namely the nitrogenase complex. Ammonia, the end product of the nitro-genase reaction, is a substrate for the enzyme glutamine synthetase (GS) that catalyzes the committed step in the assimilation of ammonia resulting in the production of the amino acids, glutamine and glutamate. Glutamine is in-volved as nitrogen donor in the biosynthesis of purines, pyrimidines and many other amino acids. Glutamate is the donor of amino groups in most of the reactions catalyzed by transaminases producing other amino acids (164). The pathway for ammonia assimilation represents around 15 % of the cells ATP requirement, so the regulation of GS is very important in terms of wast-ing or not wasting energy (108, 164). The GS enzyme is regulated by three different mechanisms; feedback inhibition, transcriptional regulation, and covalent modification by an AMP group. This work focuses on the covalent regulation of GS and detailed studies of adenylyltransferase, the bifunctional enzyme catalyzing the reversible modification of GS.

2

Diazotrophic bacteria

The ability to fix nitrogen is restricted to prokaryotic organisms, both free-living and symbiotic. The genera Rhizobiaceae are symbiotic diazotrophs, which can form close symbiotic associations with the roots of specific plants, namely the legumes. The interplay between the two has been reported to produce up to 50 % of the biological nitrogen fixation (34). The associa-tion starts with complex signaling between the plant root and the rhizobia followed by entrance into the plant root via an infection thread. Within the plant, the bacteria turns into a nitrogen fixing organelle, the bacteroid, and the plant cell simultaneously undergoes morphological changes that finally lead to the so called root nodule. The nodule is a nitrogen fixing organ that both players benefit from, the rhizobia obtain energy from the plant and the plant acquires nitrogen produced by the nitrogenase complex synthesized within the bacterium (62).

Other symbiotic diazotrophic bacteria are the so called endophytic and as-sociative that can form symbiosis with, for example, grasses (Azoarcus sp) or sugar cane (Gluconacetobacter diazotrophicus). Although their coloniza-tion is different from rhizobia, the alliance is again beneficial to both organ-isms.

The other group of diazotrophs is the free-living diazotrophs that can be divided according to the preferred growth conditions such as photosynthetic, aerobic or anaerobic, and they are extremely diverse with respect to the pro-karyotic world, from archaea to cyanobacteria, (there are also symbiotic cyanobacteria). The two most studied free-living diazotrophs are Klebsiella

pneumoniae and Azotobacter vinelandii, more specifically, for detailed ge-netic regulation, and the 3D structure of nitrogenase respectively.

Rhodobacter capsulatus and R. rubrum are the most studied within the photosynthetic group of diazotrophic bacteria.

Rhodospirillum rubrum

R. rubrum was first described in 1887 as “Spirillum rubrum” by E. Esmarch (49) and the ability to fix nitrogen was reported 1949 by Gest and Kamen (66). R. rubrum is a spiral shaped bacterium that can carry out anoxygenic photosynthesis. Photosynthesis is a life essential process where sunlight is captured and transformed into usable energy. Water is the electron donor in this process in plants, and bacteria can use alternative carbon sources or sul-phur compounds as electron bank. For R. rubrum, malate is an excellent carbon source, when they grow anaerobically in the light (photoheterotro-phic), but growth is promoted in an array of different growth conditions (17, 37). In R. rubrum, the photosystem is cyclic, in which no products are

3

formed per se, but protons are translocated to the periplasm, generating a proton motive force that is the driving force for the ATP synthase.

R. rubrum has been studied in detail because it possesses the ability to form so called chromatophores under photosynthetic conditions. The chro-matophores are invaginations of the inner membrane, and upon cell break-age, these membrane invaginations form inside-out vesicles that can easily be used for bioenergetic studies. Furthermore R. rubrum can be regarded as a model organism with respect to post-translational regulation of nitrogenase. It was the first diazotrophic bacterium where this phenomenon was reported (128, 148). In addition, R. rubrum encodes two nitrogenase systems: iron only nitrogenase, and the more studied molybdenum containing nitrogenase.

Nitrogenase

The overall nitrogenase reaction outlined below shows that six electrons are required for reduction of one dinitrogen into two ammonia and two electrons for the production of one hydrogen. For each electron used, two molecules of ATP are hydrolyzed.

In order to fix nitrogen, electrons are required, and the direct electron donor to dinitrogenase reductase is usually a ferredoxin or a flavodoxin (ferredoxin N in R. rubrum (40)). The reduction pathway to either of them depends on the physiology of the organism, but in R. rubrum this pathway has to a large extent been elucidated by Edgren et al. (41, 42), the so called fix-system. Briefly, in their proposed model, the fix-system consists of 4 different pro-teins, encoded by fixABCX. FixAB is reduced by a soluble cofactor and de-livers electrons to FixC. FixC is presumably located at the membrane deliv-ering one electron to FixX and the other one to ubiquinone in the chromato-phore membrane, which is reoxidized by reversed electron flow driven by the proton motive force (12). FixX then reduces ferredoxin, the final electron donor to nitrogenase. It should be noted that the pyruvate-ferredoxin oxi-doreductase encoded by nifJ in R. rubrum can also support nitrogenase activ-ity, although to a lesser extent compared to the fix-system under normal photoheterotrophic growth (42, 117).

The actual nitrogenase complex consists of dinitrogenase reductase or the Fe-protein (encoded by nifH), and dinitrogenase (encoded by nifDK) or the MoFe-protein. Dinitrogenase reductase is a homodimeric protein that con-tains a single [4Fe-4S] cluster that is extremely oxygen sensitive, a half-life around 30 seconds for dinitrogenase reductase in air has been reported (39, 181). The other protein in the complex, dinitrogenase is a α2β2-heterotetrameric protein with two heterodimeric halves. Each half contains a P-cluster and a FeMo-co factor. The P-cluster can be visualized as two [4Fe-

4

4S] clusters sharing a sulphur atom. The FeMo-co cluster is much more complex in structure and it is believed to be the site of reduction of dinitro-gen. With all these clusters within the nitrogenase complex there is a path-way for one electron at a time. It starts with reduction of dinitrogen reduc-tase with electrons from ferredoxins or flavodoxins. When Mg-ATP binds to dinitrogenase reductase a conformational change is induced within the pro-tein leading to a more exposed [4Fe-4S] cluster. This conformation is much more prone for docking into dinitrogenase and thereby facilitates the elec-tron transfer to the P-cluster on dinitrogenase, and from which it is trans-ferred to the FeMo-cofactor. A schematic representation of nitrogenase and its electron donors are outlined in Figure 1.

Figure 1. Schematic representation of the nitrogenase system and the nitrogenase reaction. The electron donors to nitrogenase are included in the picture.

Post-translational and genetic regulation of nitrogenase

As shown above, the nitrogenase reaction requires substantial amounts of ATP, so not surprisingly nitrogen fixation is a tightly regulated process. In all diazotrophs studied, nitrogen fixation is transcriptionally regulated and in some, e.g., R. rubrum nitrogenase is also regulated post-translationally by

5

reversible covalent modification. Two enzymes are involved in the post-translational regulation of nitrogenase: dinitrogenase reductase activating glycohydrolase (DRAG), and dinitrogenase reductase ADP-ribosyl trans-ferase (DRAT). Their function can be described by the so called “switch-off” effect (149). When growing R. rubrum diazotrophically and exposing the cells to NH4

+, nitrogenase activity declines. Depending on the concentra-tion of NH4

+, and/or the optical density of the culture, nitrogenase recovery to the initial activity will depend on the time it takes for the NH4

+ added, to be metabolized (174). The reason for the decrease in nitrogenase activity is the covalent modification on Arg101 on one of the subunits of dinitrogenase reductase, which is catalyzed by DRAT (123-126, 221). The modification is an attachment of an ADP-ribose from NAD+ (159, 160). The ADP-ribose moiety is detached when the “switch off” effector is removed or consumed, in a reaction catalyzed by DRAG (127, 147, 148). Other “switch-off” effec-tors reported are, i.e., glutamine, asparagine, NAD+ and exposure to darkness (99, 143), consequently even centrifugation of the cells will induce a “switch-off” effect. The actual signal transduction from the “switch-off” effector to DRAT or DRAG has not been established to date, but several models have been suggested, among them the involvement of the regulatory PII proteins as a signal link is one.

The genes encoding DRAT and DRAG are the draT and the draG genes respectively, which, interestingly are not regulated in response to nitrogen availability. Downstream draG is another gene, draB, that probably is co-transcribed together with draT and draG (54). Its function is unknown, but sequence analysis shows similarities to glutaredoxin-dependent arsenate reductase that converts arsenate to arsenite for disposal. The draTGB operon is localized around 400 bp upstream from the nif HDK genes but in the op-posite direction.

The master nif gene activator NifA

The nifHDK genes and other nif genes involved in maturation and processing of the nitrogenase complex and the metalloclusters are usually in different operons depending on species. The common theme for the nif genes is upregulated during nitrogen fixing conditions and downregulated when the cell is exposed to oxygen, i.e., the two signals are antagonistic. The master transcriptional activator for the nif genes is NifA that belongs to the EBPs-enhancer-binding proteins, a unique set of proteins that interact with σ54 RNA polymerase. The binding sites for EBPs are usually located around 100 bp upstream of the transcription initiation site with the conserved nucleotide sequence TGT-N10-ACA (136). NifA from most organisms usually contains three domains: the GAF domain, the central AAA+ domain, and the HTH-type DNA binding domain. The amino-terminal regulatory GAF domain is

6

the signaling part of NifA that binds different metabolites, very diverse though. For example, in A. vinelandii but not in K. pneumonia, the GAF domain binds α-ketoglutarate, a signaling molecule in diazotrophs (120, 121, 133). In Azospirillum brasilense, the GAF domain seems to bind GlnB, a PII protein (6). The central or AAA+ domain is responsible for σ54 RNA poly-merase interaction, or more precisely open complex formation, displaying ATPase activity (50). The carboxyl-terminal domain of NifA contains a HTH-type DNA binding domain that is required for recognition of the up-stream activator sequences (139, 140). The loop formation required for tran-scriptional activation by EBPs is outlined in Figure 2.

Figure 2. Transcriptional activation by EBPs. A- Showing RNA-polymerase with bound σ54 and the Enhancer Binding Protein (EBP) bound to the upstream activating sequence (UAS) located around 100 bp upstream of the transcription initiation site. The -12 and -24 boxes constitute the RNA polymerase binding site relative to the transcriptional start site. B- The required loop formation for transcription and NTP hydrolysis for the open complex formation is illustrated. The integration host factor (IHF) that in some cases promotes the loop formation is also included in the picture. Adapted from (33).

The activity of NifA is regulated in a variety of ways with respect to fixed nitrogen. In K. pneumoniae for example, under nitrogen excess conditions, NifL inhibits NifA activity by direct protein-protein interaction that can be relieved under nitrogen limited conditions by the PII regulatory protein GlnK or the modified form of GlnK, GlnK-UMP, as illustrated in Figure 3A (the covalent modification of PII proteins will be further discussed in the next chapter). In other words the low nitrogen signal for K. pneumoniae

7

would be the GlnK regulatory protein. This makes sense since under low nitrogen conditions expression of glnK (in an operon with amtB) and the operon nifL-nifA is activated, as outlined in Figure 3A, by the phosphory-lated transcriptional activator NtrC-P (more about the NtrB/C system later in this section).

Figure 3. NifA activation in K. pneumoniae (A) and A. vinelandii (B) adapted from (33).

As outlined in Figure 3B, in A. vinelandii α-ketoglutarate binds to NifA and thereby prevents protein-protein interaction of NifA and NifL, mimick-ing the GlnK role in K. neumoniae (83, 120). Interestingly, the role of GlnK in A. vinelandii seems to be tethering together NifL and NifA in a ternary complex leading to inactivation of NifA under nitrogen excess conditions (119). In general the low nitrogen signal in A. vinelandii seems to be high α-ketoglutarate. This is logical since under nitrogen limited conditions the α-ketoglutarate level is high (166). Also, glnK-amtB and nifL-nifA expression in A. vinelandii is not upregulated during nitrogen limiting conditions, differ-ing from K. pneumoniae (13, 36). As mentioned before oxygen needs to be sensed within these systems, and this is accomplished by NifL (76, 176, 184). NifL contains an N-terminal PAS domain that detects signals through an associated cofactor, in this case FAD (76, 176).

In diazotrophs that lack the redox sensitive flavoprotein NifL, NifA activ-ity is regulated differently. NifA in this case contains an additional linker

8

region of up to 100 amino acids between the AAA+ domain and the DNA binding C-terminal domain. This interdomain linker comprises a Cys-X4-Cys motif in addition to two Cys residues at the end of the AAA+ domain (51). In Bradyrhizobium japonicum all these cysteines and a metal ion are required for NifA activity in vivo (52). This could indicate a model for redox control of NifA, giving the four cysteines a coordination role of the metal ion. Sequence comparison between the free-living diazotrophs Az. brasi-

lense, Herbaspirillum seropedicae, R. rubrum, together with the symbiotic organisms B. japonicum and Sinorhizobium meliloti, do indeed show the presence of all four cysteine residues. Interestingly, in the above mentioned free-living diazotrophs, NifA seems to require uridylylated GlnB for activa-tion (6, 187, 217), which is not the case for the symbiotic organisms. The activity is probably regulated through interaction between the modified GlnB protein and the GAF domain of NifA.

In some diazotrophs NifA is also transcriptionally regulated by the classi-cal two-component system NtrB/C, where NtrB is a histidine kinase, con-taining two domains: the N-terminal sensor domain and the C-terminal kinase domain. NtrC, an EBP-α54 dependent protein, resembles NifA in do-main structure except in the N-terminal, where NtrC instead of the GAF domain has the so called REC domain (33, 129, 146, 171). The REC domain is a receiver domain that gets phosphorylated by histidine kinases. In K.

pneumoniae the level of phosphorylated NtrC, NtrC-P, controls expression of the glnK-amtB and nifL-nifA operons, which is high under nitrogen limita-tion and low under nitrogen excess conditions as illustrated in Figure 3A (134). The level of NtrC-P is controlled by another PII protein in K. pneu-

moniae (and in Escherichia coli) namely GlnB, which under high nitrogen conditions is deuridylylated and thereby can form a binary complex together with NtrB that reduces the ability of catalyzing phosphorylation of NtrC (33, 156). Under low nitrogen conditions, GlnB is uridylylated and does not in-teract with NtrB, which then catalyzes phosphorylation of NtrC leading to high expression of NtrC controlled operons.

The NtrB/C system in E. coli has been extensively studied in Ninfa`s laboratory showing specific features very similar to K. pneumoniae NtrB/C, except that other genes/operons relevant for E. coli under low nitrogen con-ditions are activated by phosphorylated NtrC (7, 9, 10, 18, 88, 90, 96, 97, 157, 158, 186). Among the genes upregulated in E. coli under low nitrogen conditions are glnA (encoding GS) and the glnK-amtB operon. Interestingly, the glnK-amtB and nifL-nifA operons in A. vinelandii are not subjected to NtrC dependent activation (as depicted in Figure 3B), and furthermore GlnK is the only PII regulatory protein present (135). Figure 4 shows a schematic picture of the NtrB/C system.

9

Figure 4. Regulation of the two component system NtrB/C. The transcriptional activator NtrC (NtrC-P) is active under low nitrogen conditions when NtrB acts as a kinase since the uridylylated GlnB cannot interact with NtrB. Under high nitrogen conditions NtrC is dephosphorylated by the phosphatase activity of NtrB in complex with non-uridylylated GlnB. The signals for modification/demodification of GlnB by GlnD are included in the picture and are discussed in the GlnD section.

In the α-proteobacteria R. capsulatus (encoding 2 NifA proteins), the global redox responsive RegB/RegA two-component system is responsible for nifA2 expression in combination with NtrC (48). In the symbiotic bacte-rium S. meliloti, nifA expression is induced by the FixL/FixJ system (33). In R. rubrum and Az. brasilense, NifA seems to be constitutively expressed since in a R. rubrum ntrB/C mutant, nitrogenase is expressed (112, 212). As mentioned before the requirement for NifA activity though, is the uridyly-lated form of GlnB, GlnB-UMP.

Nitrogen control in cyanobacteria and expression of nitrogen fixing genes (nifHDK) require the transcriptional regulator NtcA, which belongs to the CAP (catabolite gene activator or cyclic AMP receptor protein) family (75, 142). Unlike the situation in several proteobacteria, a cyanobacterial PII protein is not involved in NtcA function and NtrB/C homologues have not been detected.

The above exemplified regulatory mechanisms are just a scratch on the surface. There are several other regulatory cascades not discussed. A reason for this diversity in regulation could be a very fast evolution of the regula-tory networks, giving the organisms a better ability to cope with the envi-ronment (33).

10

Covalent modification of glutamine synthetase (GS)

Glutamine is synthesized in a reaction catalyzed by GS that requires ATP, glutamate and NH4

+. The glutamine produced can be further assimilated together with α-ketoglutarate and NADPH into two molecules of glutamate in a reaction catalyzed by GOGAT, (glutamate synthase) (164). The two reactions are outlined below:

The ammonium ions required for the reaction can be obtained in different ways: from the environment by diffusion through the cell membrane, or by the ammonium transporting protein, AmtB. During nitrogen fixation ammo-nium are provided by the nitrogenase reaction. The GS/GOGAT pathway is dominant in low nitrogen conditions, and in nitrogen rich conditions another enzyme contributes to ammonia assimilation, namely glutamate dehydro-genase (GDH) (137), with its reaction outlined below:

GDH is not a very efficient enzyme during low nitrogen conditions due to the relatively high Km for ammonia (around 1 mM).

Another reaction catalyzed by GS is the γ-glutamyl-transferase reaction as outlined below:

The γ-glutamylhydroxamate formed can easily be measured spectroscopi-cally (170), therefore the reaction which is non-physiological is commonly used for GS activity measurements.

11

Structure and mechanistic aspects of GS

The structure of GS has been determined initially to 3.5 Å resolution (2) and all the way to 2.5 Å (67, 68, 114, 152, 210). Bacterial GS is a dodecamer with identical subunits, divided into two hexameric rings superimposed on each other. In eukaryotes GS has an octameric structure (72). Hydrogen bonding between the hexameric rings keeps the dodecameric structure to-gether (2). Additional “adhesion” between the subunits is contributed by the C-terminal helix, called the “helical thong” that is inserted into a hydropho-bic hole in one of the subunits in the other hexameric ring (45). The active sites can be described as a bifunnel in which ATP and glutamate arrive at opposite ends, and the center of the bifunnel contains two divalent cation sites, n1 and n2, where either magnesium or manganese binds. The dimen-sion of the bifunnel is around 30 Å wide at the opening, 15 Å wide in the middle and 45 Å deep. Solving the enzymatic GS mechanism involved crys-tal structures of enzyme-ligand complexes from (69, 113) showing the in-termediate states. Among the players involved is Asp50 from an adjacent subunit that is involved in ATP interaction, the Glu327 flap that is required for closing the active site and stabilization of the transition state, the n1 and n2 divalent cation sites, and other amino acids mentioned in the figure text. The mechanism is illustrated in Figure 5.

Figure 5. GS mechanism a schematic view. When ATP enters on the top of the bifunnel, its γ-phosphate binds to the n2 ion site. This results in a movement of Asp50 toward the ammonium entering site, and Arg359 moves toward the site where the γ-carboxylate group of glutamate binds increasing the affinity for gluta-mate and ammonium ions. When the γ-carboxylate of glutamate (glutamate enters from the bottom of the bifunnel) binds to the n1 ion and above the Glu327 flap, the active site is now closed and no water can enter. The γ-phosphate of ATP is then transferred to the γ-carboxylate of glutamate, thereby forming γ-glutamyl phosphate. The ammonium ion enters at the middle of the bifunnel and binds at the negatively charged pocket formed by Glu327, Asp50, Tyr179, Glu212 and Ser53 (from an adjacent subunit). Now, Asp50 deprotonates the ammonium ion, forming ammonia that in turn attacks the δ-carbon of the γ-glutamyl phosphate intermediate, releasing

12

the phosphate group. A salt bridge is formed between Glu327 and the tetrahedral adduct, a proton is accepted by Glu327 from the adduct, thereby neutralizing the salt bridge and forming glutamine. In the last step the Glu327 flap opens and glutamine is released. Adapted from (45).

Feedback inhibition

GS is also regulated by feedback inhibition by some of the end products from glutamine metabolism: serine, alanine, glycine, AMP, CTP, trypto-phan, histidine, carbamoyl phosphate, glucosamine-6-phosphate and NAD+ (203, 204). Crystal structures of GS in complex with alanine, serine and glycine showed that these amino acids bind to the glutamate site (115, 116), and AMP to the ATP site. It seems likely, by kinetic analysis, that GS activ-ity by feedback regulation is a competitive inhibition. But homotropic regu-lation within the active sites exists also (45).

Oxidative modification of GS

Oxidative modification of GS is another way of regulation that has been reported (109, 111). One way of protection against this phenomenon could be simply surface exposed methionines. In a report by Levine et al. they examined E. coli GS exposed to H2O2 (110). They concluded that surface exposed methionines were oxidized (forming methionine sulfoxide), which was not the case for the more interior or buried methionines. Altogether 8 out of 16 methionine residues were oxidized, and there was little effect on GS catalytic activity. Interestingly, the oxidized methionines were highly distributed at the entrance of the active site suggesting a locally protective antioxidant function. In order to reduce methionine sulfoxide back to me-thionine another enzyme needs to take action, namely methionine sulfoxide reductase. This enzyme is annotated in the R. rubrum genome which might indicate that this antioxidant system also exists in R. rubrum, also R. rubrum GS contains even more methionines compared to E. coli GS.

Another way of examining GS susceptibility to oxygen radicals is with the so called MCO system (metal catalyzed oxidation) where hydroxyl radi-cals are formed that react with specific amino acids of the enzyme thereby impairing its activity (103). Generation of hydroxyl radicals and oxygen radicals are outlined below:

13

In equation (1) Fe2+ is oxidized by molecular oxygen and a superoxide radi-cal anion is formed, that can undergo a dismutation reaction in equation (2) to form H2O2. In equation (3), H2O2 can react with Fe2+ and generate a hy-droxylradical via the Fenton reaction.

Often reducing equivalents such as ascorbate or DTT are used in vitro to recycle the oxidized metal ion (103). When the metal ion bound to an en-zyme is involved in these oxidation-reduction reactions, the locally gener-ated oxygen species may react at specific sites within the enzyme impairing its activity (109). Enzymes prone to this specific oxidation are, among oth-ers, GS and adenylyltransferase (GlnE), from E. coli (103). Interestingly, from a physiological standpoint, GS can also be protected against oxidative damage (MCO-system) by protector enzymes or antioxidants (102). More about this in the peroxiredoxin section.

Other ways of GS regulation

GS has also been reported to be ribosylated under different growth condi-tions, but the relevance of this modification has not been elucidated (199). However, in R. rubrum, ribosylation has been proposed to participate in fine tuning of GS activity or being involved in coordination of nitrogenase and GS regulation (199). Curiously no decrease in activity of GS by ribosylation of R. rubrum GS was reported, which however was the case for E. coli GS. In E. coli, the ribosylation of GS is not physiological, since the ribosylation was carried out with an NAD:arginine ADP-ribosyltransferase from turkey erythrocytes (141). The site for E. coli GS ribosylation is Arg172 located in the central channel of the dodecamer, thereby perhaps inducing a disruption in the normal subunit alignment leading to loss in GS catalytic activity. Other organisms with reported GS ribosylation are S. meliloti (122), where a unique rhizobial GSIII was ribosylated and losing activity. Many rhizobia differ from enteric bacteria by having more than one GS. GSI resembles the single GS found in other bacteria, and GSII is similar to eukaryotic GS not very common in bacteria (30). ADP-ribosylation has also been reported in Streptomyces griseus (153) and in the cyanobacterium Synechocystis sp. strain PCC 6803 (183). In a later report, GS from the same cyanobacterial strain was found to be inactivated by two polypeptides (IF7 and IF17). The peptides were expressed and bound to GS under high ammonium conditions, thereby lowering GS activity in an in vitro system. Interestingly, glutamine and arginine are the most abundant amino acids present in nitrogen rich con-ditions, also IF7 and IF17 contain around 25 % glutamine and arginine with their genes (gifA and gifB) induced in nitrogen rich conditions. This could, according to the authors, indicate that the particular amino acid composition of both factors may be considered a kind of product feedback mechanism regulating the ammonium assimilating pathway in cyanobacteria (64). Re-

14

cently the ammonium inactivated GS was shown to be reactivated in vivo by a mechanism involving proteolytic removal of IF7 and IF17 (63).

Another interesting binding partner for GS is the regulatory protein GlnK1. This binary complex is formed in the archaea Methanosarcina mazei under low α-ketoglutarate conditions, resulting in low GS activity. Surpris-ingly, when the α-ketoglutarate level goes up, GS activity also increases even though the binary complex is still intact. In addition, α-ketoglutarate can itself bind to GS and stimulate GS activity (44).

The most common form of post-translational regulation of GS in proteo-bacteria is by adenylylation/deadenylylation that is covered in a later chap-ter, but first the fascinating GlnD/PII protein monocycle, and its characteris-tics is discussed.

15

GlnD and the PII proteins monocycle

The regulation of GS modification/demodification in E. coli depends on the correct uridylylation status of the PII protein GlnB. Uridylylation and de-uridylylation of PII proteins are catalyzed by the bifunctional enzyme UTase/UR or GlnD (hereafter called GlnD). The activity of GlnD needs to be explained together with the PII protein. Therefore selected characteristics of the PII proteins necessary to understand the GlnD function is included in this section, and more about the trimeric PII protein in the next section.

GlnD activity

GlnD enzyme activity was first described by Adler et al. (1) and later puri-fied by Francis et al. and Garcia et al. (61, 65). A more detailed characteriza-tion of GlnD has been done by Kamberov et al. (96), and more extended studies were performed by Jiang et al. (89). They showed that the uridylyla-tion status of GlnB (one of the PII proteins present in E. coli) is regulated by the small molecule effectors glutamine for nitrogen and α-ketoglutarate for carbon. The glutamine effect is excerted by GlnD that catalyzes the de-uridylylation of GlnB when glutamine is high and the reverse reaction when glutamine is low. Glutamine also inhibits the uridylylation reaction, (which is concentration independent of the carbon signal α-ketoglutarate) by lower-ing the rate of catalysis. In an accompanying paper in which the steady-state levels of GlnB modification were examined, they concluded that the extent of GlnB uridylylation quickly reached a level that was characteristic of the glutamine concentration and again, independent of the α-ketoglutarate con-centration when the α-ketoglutarate level was varied within its reported physiological range 0.1-0.9 mM (145, 179).

The carbon signal α-ketoglutarate regulates the activity of GlnB by direct binding to the trimeric protein synergistically with ATP. At physiological concentrations of ATP, all ATP sites are saturated. Under these conditions, the binding of α-ketoglutarate displays negative cooperativity, with a Kd of around 5 µM, for the first molecule, well below the in vivo concentration of α-ketoglutarate (0.1-0.9 mM) (145, 179). The second and third α-ketoglutarate molecules have reported Kd values at the high end of the physiological α-ketoglutarate range. By using the techniques of equilibrium dialysis and/or ultrafiltration together with 14C labeled ATP and α-

16

ketoglutarate they showed that unmodified GlnB binds three molecules of ATP and one α-ketoglutarate molecule, and uridylylated GlnB binds three molecules of each (89). Also modified GlnB reduces the negative coopera-tivity in α-ketoglutarate binding (89, 145). Interestingly, their studies showed that ATP and α-ketoglutarate (even in the upper physiological range) stimu-lates both uridylylation and deuridylylation by binding to GlnB and GlnB-UMP, suggesting that only glutamine is the primary regulator of the uridyly-lation status of GlnB. Furthermore, when they characterized the GlnD activ-ity they suggested Mg2+ to be the relevant metal ion cofactor for both reac-tions.

The other PII protein in E. coli, GlnK, has also undergone crucial exami-nation by Ninfa and his coworkers in many respects (7, 8, 11, 14, 15). Its ability to be uridylylated is very similar to GlnB but deuridylylation of GlnK-UMP is slower than for GlnB-UMP in vitro (10). In another report GlnK-UMP was suggested to be faster deuridylylated than GlnB-UMP in vivo (11).

The uridylylation/deuridylylation reactions in E. coli of a PII protein by GlnD are outlined in Figure 6.

Figure 6. The role of E. coli GlnD in the covalent modification of GlnB and GlnK. Under high glutamine conditions GlnB and GlnK are deuridylylated and at low glutamine conditions, uridylylated. The figure is based on data using Mg2+ in the assay.

The uridylylation status of PII proteins determines the specific interacting partner and in addition, α-ketoglutarate and/or ATP are also involved in the PII proteins quest for a partner, an issue further discussed in the PII protein section.

17

Uridylylation/deuridylylation of R. rubrum PII proteins

The results concerning uridylylation/deuridylylation of the three PII proteins in R. rubrum (GlnB, GlnK and GlnJ) do indicate differences between R.

rubrum and E. coli PII proteins (Paper I). In the uridylylation reaction GlnB was more effectively uridylylated together with Mg2+ compared to Mn2+. This was not the case for GlnJ and GlnK for which Mn2+ was much more effective. The relevance of differences with respect to divalent ion used in the assay is not clear; the effect could be either on the PII proteins or on GlnD itself. There are Mn2+ dependent enzymes in R. rubrum, e.g., GS and DRAG, also in vivo GlnJ has been shown to be uridylylated, indicating suf-ficient Mn2+ concentrations (216). Unpublished results from our laboratory also show that GlnJ is dissociated from a membrane fraction in a Mn2+/α-ketoglutarate/ATP dependent fashion, but not when Mn2+ was replaced with Mg2+. Recently, a report confirmed the above results, but did not include any studies on differences in divalent cations (200).

Glutamine did not exert any inhibitory effect similar to the E. coli system even though we also tried with R. rubrum cell extract to exclude the possibil-ity that we might have lost a component necessary for the inhibitory effect. To rule out the possibility that the effect was on the R. rubrum PII proteins we also carried out the assay with E. coli GlnK showing the same result. We also studied the ability of mutated variants of the PII proteins (Q39E) to be uridylylated. A Q39E variant of E. coli GlnB has been shown to have re-duced ATP binding and very defective α-ketoglutarate binding properties. In addition the variant is uridylylated with Mn2+ but not with Mg2+ (92). Two of the R. rubrum variants, GlnB Q39E and GlnK Q39E were not uridylylated even though we used Mn2+ in the assay, but the GlnJ Q39E variant was uridylylated albeit to a much lesser extent than native GlnJ.

Deuridylylation of GlnB-UMP and GlnJ-UMP was stimulated by gluta-mine, irrespective of the α-ketoglutarate and ATP concentration, but Mn2+ was definitely required. In E. coli ATP and α-ketoglutarate are required for deuridylylation when Mg2+ is used in the assay, but not when Mn2+ is pre-sent.

We also showed that deuridylylation of GlnB-UMP is inhibited by GlnB like the E. coli system, but UTP did not inhibit deuridylylation. In addition to this, we have showed that R. rubrum GlnD has a higher affinity for GlnB than for GlnB-UMP since deuridylylation occurred only when GlnB was omitted from the reaction, this might indicate the same active site for GlnB and GlnB-UMP.

18

The active site in E. coli GlnD has been thoroughly studied by specific point mutations followed by purification of the enzyme (145). Certain point mutations in the conserved nucleotidyl transferase domain (NT) resulted in loss or reduced uridylylation activity (D107N), and a D105N mutation re-sulted in complete loss of both activities. This aspartate (D105N) is believed to chelate the Mg2+ ion involved in both activities. The UMP removing activ-ity of GlnD was almost abolished in a G98A mutation. Together these results point toward a single active site on the peptide. When comparing the primary sequences for R. rubrum and E. coli GlnD, they show very high similarity in the NT region. All the above amino acids mutated in E. coli GlnD are pre-sent in R. rubrum. In addition the R. rubrum and E. coli GlnD sequences contain two ACT domains (the name ACT comes from the first letter of the three first proteins containing the ACT domain (70)), close to the C-terminal of GlnD (22, 78). ACT domains are involved in sensing amino acid levels that in the case for GlnD would be glutamine, but unfortunately no bio-chemical experiments have been performed highlighting the function of the ACT domains in GlnD.

Experiments in Roberts laboratory concerning R. rubrum GlnD activity in vivo showed that different levels of GlnD activity are required for post-translational regulation of nitrogenase (DRAT/DRAG activity), nif gene expression (NifA activity), and NtrB/NtrC function (glnJ expression) (216). Interestingly, a R. rubrum glnD mutant was not lethal as reported for many other organisms, although the growth rate was severely affected with ammo-nium as the sole nitrogen source. According to their results, the slow growth could be due to low GS activity since in a glnD mutant PII proteins cannot be uridylylated, leading to inactivation of GS although not all GS gets ade-nylylated as shown by Western Blot (216). The reason for this result is fur-ther discussed in the GlnE in R. rubrum section (Paper II). Another pheno-type was that the glnD mutant could not fix nitrogen, which could again be explained by the absence of uridylylated GlnB.

Other GlnD studies

Detailed studies at the molecular level of GlnD together with its substrates GlnB and GlnK, have only been done for E. coli (10, 89, 96), but the PII protein from Az. brasilense (GlnZ) and GlnB from R. rubrum have also been shown to be substrates for E. coli GlnD (4, 220). Our report concerning GlnD (Paper I) and a recent published report characterizing GlnD in H.

seropedicae are not as detailed as the E. coli GlnD information, but they provide some insight into the variation of GlnD function (16). Most studies regarding GlnD have been on the physiological level and/or by mutational approaches (25, 26, 28, 43, 154, 163, 172, 190).

19

Very little is known about the transcriptional regulation of glnD, but the E. coli gene is monocistronic and expressed at very low levels (96, 101, 150, 192). The R. rubrum gene is located in the same fashion as in S. meliloti and Rhizobium leguminosarum bv. viciae with the mvi gene downstream of glnD encoding a putative virulence factor (172, 175). The gene organization in R.

rubrum might indicate that the genes are transcribed as an operon (around 100 bp in between the genes).

To date, the glnD gene has not been identified in the gram-positive bacte-rium Bacillus subtilis, the archaebacteria (Methanobacterium thermoauto-

trophicum and Methanococcus jannaschii) or cyanobacteria (e.g., Synecho-

cystis). Even though the glnD gene has not been identified in cyanobacteria, the cyanobacterial PII protein GlnB can still be modified, although not by uridylylation. Instead, the covalent modification is phosphorylation at Ser49 (59, 60). The modification status of GlnB in cyanobacteria seems to be modulated by a kinase/phosphatase system with the respective activity sepa-rated on two different peptides (82). The phosphatase activity has been iden-tified and characterized (81, 105), but the kinase has not so far (55). Notably, there is no glutamine effect on the PII protein modification level in cyano-bacteria compared with E. coli. Cyanobacterial GlnB does indeed bind ATP and α-ketoglutarate synergistically, but at low α-ketoglutarate concentra-tions, cyanobacterial GlnB binds a single ATP and a single α-ketoglutarate. At high α-ketoglutarate concentrations it binds three α-ketoglutarate and three ATP molecules. It seems that cyanobacterial GlnB displays negative cooperativity of ATP binding that is relieved by α-ketoglutarate binding. When saturating ATP concentrations were used, the cyanobacterial GlnB protein showed the same characteristics as the E. coli GlnB, i.e. negative cooperativity towards α-ketoglutarate (57, 59, 82).

PII proteins

The regulatory protein PII was first identified 1969 as the second peak on a gelfiltration column, hence its name (182). This highly conserved protein family has since been found to play significant roles in both nitrogen me-tabolism and carbon sensing, present in eukaryota (plants and eukaryotic algae), archaea, and bacteria. The E. coli glnB gene was first sequenced by Son and Rhee who confirmed earlier results showing that the site of uridyly-lation indeed is Tyr51 (167, 185). By mutational studies of glnB and glnD a homologue to glnB was identified 1995, designated glnK (191, 193). The GlnK protein is 67 % identical to GlnB at the primary sequence level with the same conserved site of uridylylation as GlnB, Tyr51.

PII proteins can be divided into three different families by adapting the criteria of Arcondeguy et al., namely GlnB, GlnK and NifI (5). A monocis-tronic operon or a polycistronic operon together with glnA or nadE are the

20

characteristic of the GlnB family of PII proteins, where glnA is the gene encoding GS and nadE an ammonia dependent NAD synthetase. GlnBs are mainly found in proteobacteria and cyanobacteria and the GlnB primary sequence is characterized by a lysine as residue 3 and either of the acidic residues (glutamate, aspartate) at position 5. The glnK family of PII proteins is transcribed in an operon together with the amtB gene probably indicating a functional relationship between GlnK and AmtB. More about this issue later in the thesis.

Usually with some exceptions, GlnK have hydrophobic residues at posi-tion 3 and 5. The last PII family, NifI is encoded by the nifH-linked genes found in the diazothrophic methanogens. Their function does not seem to resemble that of GlnB and GlnK, instead they show similarities to the DRAT/DRAG system present in a variety of bacteria. For example in the methanogenic archaea Methanococcus maripaludis, nitrogenase activity is regulated by a post-translational mechanism involving associa-tion/deassociation of the regulatory proteins NifI1 and NifI2 (35), two genes with same name always occuring in pairs. The nifI genes are not restricted to archaea, they have also been identified in other organisms.

R. rubrum PII proteins

As mentioned previously, R. rubrum contains three PII proteins. Firstly the glnB gene was identified and sequenced with the glnA gene located 135 bp downstream of glnB (94). Further studies showed that the glnB-glnA tran-script is probably processed giving an additional regulatory level to both GlnB and GS (21). The involvement of GlnB in regulation of GS activity was also examined with R. rubrum cell-free extracts grown under different conditions, together with overexpressed his-tagged GlnB and GlnB-UMP showing characteristics different from the E. coli system (deadenylylation of GS), which in part could be attributed to the use of cell-free extracts (93). These results are discussed further in the GlnE section.

A more in vivo based analysis of GlnB has also been performed showing that GlnB is modified in response to N status of the cell (95). The results from this report confirmed that uridylylation is the covalent modification of a PII protein, and in addition the results show that an uridylylated PII protein is required for nitrogen fixation. The addition of a so called “switch-off” effector leads to an unmodified PII-protein (addition of NH4Cl, NAD+, glutamine, and partially by glutamate). No effect by addition of oxygen or exposing the sample to darkness was reported. At that time only one PII protein was known to be present in R. rubrum, giving the possibility that the Western Blots performed showing the above results could be influenced by the two other PII proteins that we today know are present in R. rubrum.

21

The glnJ and glnK genes linked together with amtB1 and amtB2 respec-tively were identified by Zhang et al. (214). They showed that either GlnB or GlnJ is required for proper DRAT/DRAG regulation.

Previous reports from the same group also showed that any of the three R.

rubrum PII proteins can take part in the covalent regulation of GS in vivo (214, 215). Another report from the same group demonstrated that GlnJ is expressed only under nitrogen low conditions (218), which also has been verified by RT-PCR in our laboratory (unpublished results Wang et al.).

PII protein structures and specific point mutations

Structures of E. coli GlnB, GlnB-ATP, GlnK and GlnK-ATP have been solved between 1.9-2.7 Å (19, 208, 209). All have a common βαβ double motif together with three different loops. The proposed T-loop is between amino acids Gly37-Phe55 with a Tyr51 at the apex of the loop being the site of uridylylation. Mutational studies within this loop, especially of GlnB, have been performed giving plentiful results with respect to structure-interaction characteristics (92). For example removal of the apex of the T-loop (∆47-53) prevents uridylylation, but also alters the interaction with all three, at that time known receptors for GlnB (GlnE, GlnD and NtrB). Inter-estingly, binding of ATP and α-ketoglutarate was not affected, implying the important role of the T-loop in interaction with GlnB targets (92). Two other mutations, G41A and A49P, eliminated the interaction with GlnD and NtrB respectively, but the interaction with the other targets was unaffected (92). Other mutations reported in the same paper, not located in the T-loop, are T29M and G89A. These mutations seem to affect the overall structure of GlnB and certainly no target receptor interaction could be detected. Another mutation is Q39E, which has lower interaction ability for α-ketoglutarate, GlnD and NtrB, but could still interact with GlnE and bind ATP.

We have also constructed this mutation in all PII proteins present in R.

rubrum followed by purification and examining their effect on GlnD and GlnE activity (Paper I and II respectively). We have also examined the effect of a Y51F mutation, showing in vitro that the site of uridylylation is indeed Y51 for all of the R. rubrum PII proteins (Paper I).

E. coli GlnB also contains the so called B-loop, stretching from amino acid Gln82 to Asp88, separating the second α-helix from the fourth β-sheet, and the third loop (C-loop) is located at the C-terminal of GlnB. The T-loop in both GlnB and GlnK has been very difficult to resolve, but for GlnB this was possible by lattice contact with another molecule, and in the case for GlnK, when the T-loop was constrained. One of the differences between GlnK and GlnB lies in the T-loop where residues 47-49 in GlnK form a 310 helix, which is not the case for GlnB. Highly conserved residues for both GlnK and GlnB (Gly27, Thr29, Gly35, Lys58, Gly87, Gly89, Lys90, Arg101

22

and Arg103) do map an ATP binding site that is formed by the B-loop of one subunit together with the C-loop, and sequences of either end of the T-loop from an adjacent subunit. All of the above amino acids involved in ATP binding are also present in all three R. rubrum PII proteins (GlnB, GlnJ and GlnK).

Other PII protein 3D structures

Other structures solved to date are Thermus thermophilus GlnK and GlnK-ATP 1.85 Å (173), Thermotoga maritima PII-like protein-ADP 2.5 Å (178), and the gram negative β-proteobacterium Neisseria meningitidis 1.85 Å (144), where the authors suggest a α-ketoglutarate binding site located in the conserved regions of (Gly37, Arg38, Gln39, Lys40, Thr83, Gly84, Gly89, Lys90 and Arg101). In the diazotroph H. seropedicae GlnK 2.1 Å structure paper (130), they reported an interesting observation. In non-enteric di-azotrophs, the C-terminal consensus is XGXDAX and in enterics EDDAAI (amino acids 106-112). This could lead to a conformational change in the C-loop, a bend at position G108, thereby changing the characteristics of the PII protein. The non-enteric diazotrophs compared are Rh. capsulatus, Rhodo-

bacter sphaeroides, B. japonicum, Az. brasilense, R. leguminosarium,

Azorhizobium caulinodans, and R. rubrum GlnB. The other two R. rubrum PII proteins (GlnJ and GlnK) do not match the consensus. The enteric bacte-ria E. coli and K. pneumoniae were used for the comparison.

In the cyanobacteria Synechococcus sp. PCC 7942 and Synechocystis sp. PCC 6803, the core of their GlnB (2 Å) is similar to the E. coli protein with differences in the T-loop and with a C-loop being able to take on multiple conformations (207). As mentioned before, cyanobacterial PII proteins are phosphorylated at Ser49, but heterologous expression of Synechococcus sp. PCC 7942 GlnB in E. coli results in uridylylation of this protein, albeit with low efficiency (57, 58). Cyanobacterial PII has a conserved Gln57, which in all proteobacteria and actinobacteria is a proline, therefore this residue has been proposed to be involved in recognition events typical for cyanobacteria.

In the model plant organism Arabidopsis thaliana a PII protein has been resolved (1.9 Å), with peptide segments in the N and C-terminals highly conserved in plant PII proteins (20). The N-terminal is supposed to be inter-acting with other proteins and the C-terminal, when ATP binds in, induces a conformational change of the protein. The protein has been proposed to sense the carbon status by α-ketoglutarate and energy by ATP, with the same conserved residues as enterics for binding of ATP. No covalent modification of the protein has been reported. Unique and highly conserved residues for plants are Trp22, Arg37 and Asp65.

Lastly the archaeae M. jannaschii GlnK1 has been crystallized, not only without ligand, but also in complex with Mg-ATP and Mg-ATP-α-

23

ketoglutarate, to 2.1, 1.2 and 1.6 Å resolution respectively (211). The latter two crystals have given a unique insight to the structure-function relation of PII proteins in general. Unliganded GlnK1 was resolved to 2.1 Å with all three T-loops in the extended flexible conformation, and some crystals also contained ADP. The three T-loops have a positive surface potential charge mainly attributed by three arginines at position 45, 47 and 49. In the crystal structure of GlnK1-Mg-ATP, all nucleotide binding pockets contained ATP in complex with Mg2+. The Mg-ATP ligand turns the T-loops into a compact conformation creating a salt bridge between residue Glu44 and Lys58, and the overall surface charge becomes less positive. In the crystal structure, GlnK1-Mg-ATP-α-ketoglutarate, all monomers bind ATP but only one monomer binds Mg2+ plus a single molecule of α-ketoglutarate. In the monomer containing Mg-ATP and α-ketoglutarate, the T-loop is in the com-pact conformation whereas the other two T-loops are in the extended form as in the ATP free form. This certainly shows that Mg2+ is required to fix the T-loop in its compact state, and that ATP is necessary, but not sufficient to create this major conformational change. The fixation of the T-loop creates a binding site for α-ketoglutarate, changing the surface potential to strongly negative. A network of hydrogen bonds between α-ketoglutarate and I52, V53 and D54 in the T-loop are formed, together with π stacking interaction with the aromatic ring of Y51.

GlnK-AmtB complexes

In M. jannaschii the interaction of GlnK1 and AmtB was also examined both by modeling the binary complex using AmtB from Archaeoglobus fulgidus and in vitro using both components from M. jannaschii (211). In the model-ing system unliganded GlnK1 with its extended positive T-loop can form a binary complex together with the more negatively charged cytoplasmic face of AmtB. Upon binding of Mg-ATP and even later α-ketoglutarate, the T-loop goes from extended to compact, and concomitantly a more negative surface charge of GlnK1 is displayed leading to dissociation of GlnK1 from AmtB.

Interestingly, in silico experiments carried out by Andrade et al. (3) using A. fulgidus AmtB and E. coli GlnB as docking partners showed a positive charge on the cytoplasmic surface of AmtB and a negative surface charge of E. coli GlnB in contrast to the above model by Yildiz et al. where the charges are reversed for the two partners. The latter group claims the interac-tion is due to surface complementarity and the other group charge comple-mentarity. Yildiz et al. also combined M. jannaschii GlnK1 and AmtB in vitro and they could show by different experiments that the binary complex is stable in the absence of Mg-ATP. Also Mg-ATP inhibits complex forma-tion even further if α-ketoglutarate is also included.

24

Ammonia transporters (AmtB)

It has been shown that each AmtB monomer consists of a central hydropho-bic channel surrounded by a right handed helical bundle of 11 transmem-brane α-helices, and like GlnK, AmtB is a trimeric protein (3, 100, 219). Structures with and without ammonia or methyl ammonia show a vestibule that recruits NH4

+/NH3. A long hydrophobic channel lowers the pKa of NH4+

and thereby transfer is allowed. Within the channel two conserved his resi-dues favor NH3 conductance. Structures of E. coli GlnK in complex with AmtB have been resolved by Gruswitz et al. and Conroy et al. to 1.96 and 2.5 Å resolution respectively (27, 71). In short their papers showed that the tip of the extended T-loop of GlnK (Arg47) protrudes into the AmtB channel and thereby blocks ammonia transport and the T-loop turns into the extended conformation. The crystal structures showed ADP instead of ATP binding in the GlnK monomers. They also showed that uridylylated GlnK does not interact with AmtB and uridylylation of Y51 was not possible when GlnK is in the complex. In other words, uridylylation is not the driving force for GlnK dissociation from AmtB. Instead, as shown by Durand et al. (38), complex dissociation required α-ketoglutarate, ATP and Mg2+ different to the situation in M. jannaschii where only Mg-ATP induces dissociation of the complex (211). Significant dissociation of the complex in E. coli could also be obtained with Mg2+ alone.

In addition to the preliminary results mentioned before in the uridylyla-tion/deuridylylation of R. rubrum PII proteins section, purified R. rubrum GlnJ together with washed membrane fractions as a source of R. rubrum AmtB show that the dissociation requires α-ketoglutarate, ATP and Mn2+ instead of Mg2+. When performing the same experiment with the GlnJQ39 variant no dissociation could be detected. Other organisms in which seques-tration of GlnK to AmtB has been identified are A. vinelandii (29), B. sub-

tilis (32), Corynebacterium glutamicum (189) and Az. brasilense (79). In S. elongatus and in A. thaliana the PII protein has been shown to

interact with the key enzyme in arginine biosynthesis N-acetyl-L-glutamate kinase (20, 73) where an unmodified PII protein binds to (NAG) kinase and stimulates its activity.

In addition to the just mentioned interactions, one can appreciate the complexity of PII proteins by adding NtrB, GlnD, GlnE, NifA, DRAT/DRAG and GS to the repertoire of targets for PII proteins. Further-more PII proteins can also form heterotrimers in vivo, not only in vitro, which makes the complexity of PII proteins even more interesting (58, 194).

25

Adenylylation/deadenylylation of GS

The pioneering work regarding GS regulation was initiated in the late 1960s, mainly in the Stadmann`s (reviewed in (188)) and Holzer`s laboratories fol-lowed by more detailed studies by Ninfa and coworkers. GlnE was one of the first signal transduction enzymes identified, and it provided the first ex-ample of reversible nucleotidylylation as a signal (104, 206). Adenylylation of GS occurs under high nitrogen conditions and deadenylylation under low nitrogen conditions, reactions catalyzed by the bifunctional enzyme ade-nylyltransferase/adenylylremoving, (ATase/AR) enzyme, from now denoted GlnE.

An increase in the number of subunits modified concomitantly decreases GS activity, and up to twelve subunits can independently and additively be modified (168). The site of adenylylation, Tyr397, is located in the adenyly-lation loop just outside the bottom entrance of the bifunnel. Several kinetical reports concerning the GS/GlnE monocycle in E. coli has been published (10, 87, 90, 91).

The adenylylation activity of GlnE is synergistically activated by GlnB and glutamine. Either effector alone can partially activate GlnE but the com-bination is the most effective. The activation of adenylylation shows a bi-phasic curve with respect to increasing concentrations of α-ketoglutarate with a peak value around 10 µM α-ketoglutarate, which is much below the second Kd value (~150 µM) for α-ketoglutarate but above the first Kd value (~5 µM) (87, 90). This Kd value might indicate the optimal α-ketoglutarate concentration for GlnB converting the GlnE enzyme into a form that binds glutamine more avidly. The fact that GlnB can, as mentioned in the PII pro-tein section, synergistically bind ATP and α-ketoglutarate, and GlnB with three ATP (saturated) and one α-ketoglutarate (unsaturated) molecule bound, gives the most efficient adenylylation reflects the above reported optimal Kd value (90). Under physiological conditions the α-ketoglutarate concentration has been reported to vary from 0.1 to 0.9 mM in cells (145, 179). At these concentrations the main role of α-ketoglutarate in the regulation of GS ade-nylylation state would be to inhibit the adenylylation reaction since GlnB is saturated with α-ketoglutarate.

The α-ketoglutarate does not regulate the binding of GlnB or GlnB-UMP to the enzyme, the effect is rather after GlnB and GlnB-UMP is already bound to the enzyme (87). GlnB-UMP inhibits the adenylylation reaction

26

perhaps by reducing the synergy effect of GlnB and glutamine, or by com-peting with glutamine for the enzyme (87).

The deadenylylation activity of GlnE requires Pi and is favored by high concentrations of α-ketoglutarate corresponding to saturated GlnB-UMP (90). GlnB and glutamine weakly inhibit this reaction, probably again, by the glutamine and GlnB-UMP competition for the enzyme suggesting that GlnE exclusively binds either GlnB-UMP or glutamine at distinct sites (91). It is notable that GlnB and GlnB-UMP also appear to act from distinct sites. Fig-ure 7 summarizes the effectors and proteins/enzymes involved in the cova-lent regulation of GS in E.coli.

Figure 7. Schematic description of adenylylation/deadenylylation of E. coli GS under high and low glutamine (gln)/α-ketoglutarate (α-KG). The uridylylation status of PII proteins is determined solely by the enzyme GlnD and the effector molecule glutamine. GlnE in complex with an uridylylated or deuridylylated PII protein cata-lyzes adenylylation/deadenylylation of GS. Adenylylation occurs mainly under high glutamine but low α-ketoglutarate and deadenylylation under the reverse conditions.

27

The bifunctional enzyme GlnE

GlnE contains two copies of a conserved nucleotide sequence with two char-acteristic/conserved aspartates that coordinate two Mg2+ ions. The N-terminal domain is responsible for deadenylylation and the C-terminal do-main for adenylylation of GS. A model for GlnE was proposed by Jaggi et al. (85) based on overexpression and purification of the respective domains, with further details added in a later publication (23). In their GlnE model, in addition to the two activity domains a third domain, a regulatory domain is located between the N and C terminal domains. GlnB and GlnB-UMP are suggested to bind into this regulatory domain that also contains two Q-linkers. Q-linkers are around 15-25 amino acids long with mostly glutamine, arginine, glutamate, serine and proline residues present and every 4 to 5 residue is hydrophobic (205). Like the name implies (apart from being glutamine rich) Q-linkers do connect domains that tether structurally distinct but interacting domains usually in a wide range of two component bacterial regulatory systems. When GlnB or GlnB-UMP bind into the regulatory do-main, the active sites will become more accessible due to a shift in the rela-tive positions of the N and C-terminal domains. The relative position shift by binding of GlnB or GlnB-UMP in the regulatory domain might be caused by a conformational change between a helical region in the first Q-linker and the third helix in the regulatory domain, which otherwise would be closer since both are amphipathical helices. Glutamine probably binds into the C-terminal adenylylation domain (23).

Recently two reports concerning GlnE were published regarding struc-ture-function analysis and kinetic characterization of regulation by GlnB, GlnB-UMP, glutamine and α-ketoglutarate (87, 91). Much of the very im-pressive data was obtained by purified truncated versions of GlnE similar to the above report by Jaggi et al., but many more experimental ap-proaches/designs were used to verify their results.

In short, their results suggest multiple sites for GlnB/GlnB-UMP within GlnE. The two domains separately bind GlnB and GlnB-UMP poorly, but significantly better when linked to the central region of GlnE. Glutamine together with the purified C-terminal domain stimulated adenylylation of GS suggesting a site for glutamine in the C-terminal. The binding of glutamine increased the affinity for GlnB, and interestingly certain point mutations within the central region of GlnE dramatically impaired the binding of glutamine even though the glutamine binding site is proposed to be in the C-terminal. Truncated versions of GlnE missing the C-terminal did not have any adenylylation or deadenylylation activity suggesting a role for the C-

28

terminal in both activities. Elimination or certain point mutation within the central region of GlnE abolished deadenylylation activity, and the adenylyla-tion activity could not be activated by GlnB, even though the GlnE variant still could bind GlnB/GlnB-UMP. This could mean that communication between the GlnB/GlnB-UMP sites and the adenylylation/deadenylylation sites are destroyed, meaning that the central domain plays a very important role in intramolecular communication. A model for GlnE regulation and conformation is presented in Figure 8.

Figure 8. A model for GlnE in E. coli showing the two activity domains and binding sites for GlnB, GlnB-UMP and glutamine. In A, GlnE is in the inactive form in the absence of effectors. In B, the deadenylylation activity is stimulated by GlnB-UMP with a site for GS-AMP on the N-terminal domain (AR). In C, adenylylation of GS is stimulated by the synergistic effect of GlnB and glutamine (gln) with the active site for GS in the C-terminal domain. Adapted from (91).

GlnE in R. rubrum

Previous experiments in R. rubrum examining GlnE activity from cell-free extracts grown under different conditions indicated characteristics differing from the E. coli system. Neither GlnB-UMP nor α-ketoglutarate activated the deadenylylation of GS (93). Also a partial purification of GlnE indicated a Mw 70 kDa, which is considerably smaller than the E. coli GlnE (198). Many reports have shown modification and demodification of GS from R.

rubrum with samples taken from different growth conditions or with addition of different molecules, but no reports with purified enzymes except with the addition of purified GlnB and GlnB-UMP into cell-free extracts (93, 215, 216, 218, 220). Due to the difference in GlnE activity compared to E.

coli and the molecular mass, we decided to purify wild-type GlnE from R.

rubrum. GlnE adenylylation activity was recovered from around 250 litres of cells after several column steps. Unfortunately, even though the activity was

29

recovered, the enzyme GlnE could not be identified by Edman sequencing, but during the process we were able to develop several GlnE assays. In 2002, the genome sequence draft for R. rubrum was released and thereby we could clone GlnE into an overexpression vector with six histidines in the C-terminal.

With all components purified, our results from Paper II show that all PII proteins from R. rubrum can activate the adenylylation activity of GlnE, in agreement with previous studies (214, 215). Interestingly, adenylylation of GS was not stimulated by glutamine, and instead of ATP, ADP could be used as a substrate for GlnE. A high concentration of α-ketoglutarate inhib-ited adenylylation like the E. coli system. The highest adenylylation activity was achieved with no α-ketoglutarate present whereas in E. coli a biphasic curve for the α-ketoglutarate effect has been reported (87, 90). GlnB-UMP also inhibited adenylylation like in E. coli.

When examining GlnE activity with PII-Q39E R. rubrum variants, we could conclude that the α-ketoglutarate effect was indeed on the PII proteins, not on GlnE. The deadenylylation activity of R. rubrum GlnE required only Pi, and deadenylylation was neither stimulated by GlnB-UMP nor by α-ketoglutarate. These results are clearly different from the E. coli system in which GlnB-UMP and α-ketoglutarate are absolutely required for deadenyly-lation (90). The fact that deadenylylation in R. rubrum does not require uridylylated PII proteins can be explained in terms of a R. rubrum glnD mu-tant. In this mutant, small amounts of unmodified GS were shown to be pre-sent, even in cells grown in nitrogen sufficient conditions (216). The most likely explanation for this phenotype is the fact that deadenylylation of GS in R. rubrum does not require the modified forms of PII proteins and accord-ingly deadenylylation is the default activity of GlnE.

When examining the reversibility of GlnE together with native PII pro-teins and PII-Q39E variants, the results not only verified the reversibility of GlnE, the results again showed that the α-ketoglutarate effect is on the PII proteins. Also, as opposed to the deadenylylation experiments, α-ketoglutarate was required for the reversibility experiments.

From our results it seems that GlnB and GlnB-UMP are competing for the enzyme and thereby provide regulation, since only GlnB is promoting the adenylylation activity. The fact that GlnB-UMP does not stimulate deade-nylylation is intriguing.

Possibly during a transition from nitrogen rich to nitrogen poor conditions (nitrogen fixing conditions) when GlnB becomes uridylylated, an increase in the level of GlnB-UMP prevents GlnB from binding GlnE. This will prevent further adenylylation of GS and thereby GS is active. Also, under nitrogen poor conditions, the α-ketoglutarate level is high and according to our results this leads to inhibition of adenylylation and stimulation of deadenylylation. The level of α-ketoglutarate when growing cells under nitrogen poor condi-tions, was higher which is in agreement with the above results (Paper II).

30

Integration of signals

For a bacterium to maintain a steady state control of both consumption and uptake of nutrients, a balance needs to be set between catabolism and anabo-lism. A sudden specific change, or perturbation, in the environment needs to be adjusted within the cell for survival and many molecular mechanisms are present in bacteria for coping with these changes, that otherwise would be fatal or lead to decreased cell growth. One very important anabolic pathway is the previosly mentioned GS/GOGAT pathway. In E. coli, this pathway has at least three signaling molecules involved in nitrogen metabolism, gluta-mine, interacts with both GlnD, GlnE and is also a substrate for GOGAT. The other substrate for GOGAT, α-ketoglutarate acts as a signal by its inter-action with the PII proteins thereby changing the affinity for PII protein tar-gets. PII proteins also bind ATP (or ADP in some crystals), which seems to be required for its function. Glutamate, a product of GOGAT and a substrate for GS does not seem to have a signaling function within the E. coli system, which also probably holds true for NADPH and NH4

+. In R. rubrum, the system is clearly different from the E. coli situation, for

example glutamine only stimulates deuridylylation of GlnB-UMP and GlnJ-UMP (possibly also GlnK-UMP). No inhibition by glutamine of uridylyla-tion of R. rubrum GlnB, GlnJ, GlnK and E. coli GlnK (Paper I). Neither activity (adenylylation/deadenylylation) of GlnE in R. rubrum was affected by glutamine (Paper II). The α-ketoglutarate effect seems to be on only the PII proteins in R. rubrum, and this metabolite is required for uridylylation, with different concentrations depending on the PII protein investigated and which divalent cation is used. Deuridylylation of GlnB-UMP and GlnJ-UMP seems to occur irrespective of the α-ketoglutarate concentration, but only if Mn2+ is present. High α-ketoglutarate probably results in three α-ketoglutarate molecules bound to the PII protein and thereby the PII protein-GlnE association is disrupted and adenylylation of GS cannot proceed. The association is probably also prevented by uridylylation of PII proteins that is stimulated by an increase in the α-ketoglutarate concentration. In Figure 9 our in vitro findings with purified components involved in GS regulation in R. rubrum are summarized. The divalent cation influences on GlnD and/or the effect of ATP on PII proteins are not included in the figure.

31

Figure 9. The two monocycles involved in GS and PII (illustrated by GlnB) proteins covalent modification in R. rubrum. See text for details.

Is our in vitro data relevant for in vivo studies?

The modification/demodification cycles of GS and PII proteins observed in vivo can actually be explained by integrating our in vitro data. When R. ru-

brum grows diazotrophically, ATP is produced from photosynthesis and GS is active, assimilating the ammonium ions from nitrogenase. GlnB is in the uridylylated form preventing interaction with GlnE and activating NifA for nif gene expression. The role of GlnJ-UMP during nitrogen fixation condi-tions has not been fully elucidated yet in R. rubrum. When cells are exposed to ammonium, GS will probably very rapidly convert the ammonium ions into glutamine leading to a peak in the glutamine concentration (98). A high concentration of glutamine leads to deuridylylation of PII proteins that stimulates adenylylation of GS. Unmodified GlnJ is presumably located to the membrane blocking the ammonia transporter (AmtB). α-ketoglutarate plays a very important role as a signal in our model for GS regulation and previous experiments have shown that during low nitrogen conditions (nitro-gen fixing conditions for R. rubrum) the α-ketoglutarate level is high and in nitrogen rich conditions the level is low Paper II (145, 179, 213). When the ammonia is consumed the α-ketoglutarate level increases back to the initial level. Again, according to our in vitro results, under these conditions GS is unmodified. An increase in the α-ketoglutarate concentration also leads to

32

uridylylation of the PII proteins and uridylylated GlnB (GlnB-UMP) cannot promote adenylylation of GS, but GlnB-UMP interacts with NifA for nif gene activation. Unpublished results from our laboratory show that an in-crease in the α-ketoglutarate concentration makes GlnJ dissociates from the membrane, and in vivo this is perfectly sound since the ammonium trans-porter should be opened for its substrate during low nitrogen conditions.

Figure 10 exemplifies the two modification/demodification cycles in vivo of GS and GlnB by Western Blot, together with GS in vivo activity meas-ured by the γ-glutamyl transferase assay using CTAB to permeabilize the cells.

Figure 10. Effect on GS and PII modification by ammonium ions. At time 0 min-utes, 150 µM NH4Cl was added to the culture, and at the indicated time points sam-ples were withdrawn from the cell culture. In A, the γ-glutamyl transferase assay in the presence of CTAB was used to monitor GS inactivation in both wild-type and a glnE mutant (UR 1660). In B, Western Blots of GS and GlnB (from wild-type) were performed showing the different modification states upon exposure and consumption of NH4Cl.

Our in vitro results show that GlnE activity is independent of the gluta-

mine, which instead acts at the level of GlnD deuridylylation activity of GlnB-UMP and GlnJ-UMP leading to a specific ratio of modi-fied/unmodified PII protein. We believe that the regulation of GlnE activity in R. rubrum is accomplished only by α-ketoglutarate on the PII proteins and the ratio of modified/unmodified PII proteins by the indirect effect of gluta-mine on GlnD.

33

When considering the GS/GOGAT pathway as a metabolic hub integrat-ing carbon, nitrogen and energy status in the cell, the flux through this path-way would reflect the level of α-ketoglutarate in the cell and thereby the adenylylation/deadenylylation level of GS. Under nitrogen fixing conditions, GS is in the active form (93, 214), consuming the low amount of ammonium present leading to a slower flux through the pathway and concomitantly an increase in α-ketoglutarate concentration. The α-ketoglutarate would then bind to the PII proteins and thereby prevent interaction with GlnE, under these conditions the default deadenylylation activity of GlnE is predominant. After a shift to nitrogen-rich conditions the flux through the GS/GOGAT pathway is increased and the α-ketoglutarate level decreases. Under these conditions the adenylylation activity of GlnE predominates and GS has been reported to be in the inactive state (adenylylated) (93, 214). The lack of glutamine effect on GlnE could be explained by a previous report showing that the glutamine concentration in R. rubrum does not vary depending on nitrogen source. Instead the glutamate pool is varied substantially depending on nitrogen availability and it can be expected that differences in the gluta-mate pool are reflected in variations in the α-ketoglutarate concentration (98).

Is it possible to explain the differences between E. coli and R. rubrum with respect to GS regulation? Probably not, but certainly differences with respect to growth and protein set up are involved. For example the α-ketoglutarate permease in E. coli (encoded by kgtP) is not present in R. ru-

brum. We have complemented R. rubrum wild-type in trans with this gene to examine the effect of this permease (unpublished data). No phenotype was observed with respect to α-ketoglutarate influence on GS regulation, but the strain was able to grow with α-ketoglutarate as sole carbon source. The level of α-ketoglutarate was nearly the same as in wild-type meaning that the steady state level was not affected by the permease. The same type of ex-periments were conducted with a α-ketoglutarate permease free Synechococ-

cus strain bearing the kgtP gene showing impaired growth when exposed to α-ketoglutarate (196).

Interestingly, cyanobacteria do not have a complete tricarboxylic acid cy-cle (TCA) (lack of α-ketoglutarate dehydrogenase) which leads to produc-tion of α-ketoglutarate that is not further catabolized in the TCA. The only biosynthetic role of α-ketoglutarate in cyanobacteria would be as a precursor of glutamate synthesis which requires ammonium assimilation. It is notable that both E. coli and R. rubrum do have the TCA enzyme α-ketoglutarate dehydrogenase and thereby a complete TCA. In conclusion cyanobacteria do not have the α-ketoglutarate permease or α-ketoglutarate dehydrogenase, R.

rubrum has only the α-ketoglutarate dehydrogenase, and E. coli has both enzymes, which might explain the variations for GS regulation between the species.

In terms of attempting to manipulate the intracellular α-ketoglutarate con-centration, overexpressing the enzyme PEP (phosphoenolpyruvate) car-

34

boxykinase, a key enzyme in gluconeogenesis that diverts carbon from the TCA cycle in E. coli has been performed (145). The phenotype when over-expressing the enzyme resulted in a lower level of α-ketoglutarate, leading to, under nitrogen starvation conditions, a gentle change in Ntr gene tran-scription and GS deadenylylation. They concluded that a high intracellular concentration of α-ketoglutarate is required for a complete response to nitro-gen starvation (145).

Recently it has been showed that the phosphoenolpyruvate-dependent phosphotransferase system (PTS) is involved in nitrogen regulation. The PTS system is composed of enzyme I (EI) and histidine phosphocarrier pro-tein (HPr), the glucose specific EIIAglc protein, and the glucose permease referred to as EIIBCglc. Phosphoryl groups are transferred (relayed between the enzymes) from phosphoenolpyruvate to the EIIB part of the permease and to the incoming glucose. Dephosphorylated EIIAglc signals glucose suf-ficiency and phosphorylated EIIAglc the reverse. In an experiment, E. coli grew together with a poor carbon source and glutamine as a nitrogen source, the uridylylation status of GlnB (unuridylylated) indicated nitrogen suffi-cient conditions. On the other hand, when growing the cells with glucose and glutamine, GlnB was uridylylated signaling nitrogen depletion (131). This could be explained by glucose inhibiting the uptake of other alternative car-bon sources, the so called catabolite repression, as described for the lac ZYA operon (138).

When the catabolite repression is relieved (growth using poor carbon sources, i.e., glycerol) glutamine uptake is increased leading to deuridylyla-tion of GlnB, adenylylation of GS, and inhibition of NtrC dependent gene expression. This could be explained by the CRP-cAMP complex activating the expression of the glnHPQ operon, encoding a glutamine transporter (24, 197), thereby raising the intracellular glutamine response (56, 132).

In addition to the carbon PTS, E. coli also contains a nitrogen PTS (161, 162). This system has been proposed also to be involved in carbon/nitrogen metabolism, and it is probably not taking part in the sugar transport. Instead, it seems to be exclusively involved in signal transduction. The system is composed of an EI protein encoded by ptsP, an HPr homologue (Npr) en-coded by ptsO, and an EIIANtr encoded by ptsN (165). Interestingly, two of the genes, ptsN and ptsO are found within the same operon as the rpoN gene encoding σ54, the alternative sigma factor for RNA polymerase that controls several operons involved in nitrogen metabolism, the same gene organiza-tion is present in R. rubrum. The gene rpoN itself is not per se a nitrogen regulated gene (80, 137), but the association together with ptsN and ptsO might suggest some involvement with nitrogen regulation.

35

Peroxiredoxin activity of GlnE (Paper III)

When comparing the nucleotide sequence of R. rubrum glnE to E. coli glnE, it is around 0.4 kilobases longer than reported for E. coli. The primary se-quence is 143 amino acids longer at the C-terminal end. This part of R. ru-

brum GlnE shows very high similarity to an E. coli bacterioferritin comigra-tory protein (BCP), which belongs to the peroxiredoxin family (Prxs) (86). Peroxiredoxins are low efficiency peroxidases using alternative reductants when reducing H2O2. The only redox active residues are cysteines and no prosthetic group is present in the peroxiredoxin group. Enzymatically, mem-bers of the peroxiredoxin family can be divided into three subgroups, 1-Cys Prxs, typical 2-Cys Prxs and the atypical 2-Cys Prxs. The peroxidase reac-tion is composed of two steps centered around a redox active cysteine called the peroxidatic cysteine, generally near residue 50 (46, 47). All three per-oxiredoxin classes appear to have a common first step in which the peroxi-datic cysteine attacks the peroxide substrate and gets oxidized to a cysteine sulfenic acid (Cys-SOH). The second step of the peroxidase reaction, the resolution of the cysteine sulfenic acid, distinguishes the three peroxiredoxin classes as described in (202).

The typical and atypical 2-Cys Prxs have two redox active cysteines, the peroxidatic, and another cysteine near residue 170, called the resolving cys-teine. Typical 2-Cys Prxs are obligate homodimers (and higher oligomeric states have been reported) containing two active sites (201). In the second step of the peroxidase reaction for typical 2-Cys Prxs, the peroxidatic cys-teine sulfenic acid from one subunit is attacked by the resolving cysteine from the other subunit, which results in the formation of a stable intersubunit disulfide bond. The atypical 2-Cys Prxs have the same mechanism as typical 2-Cys Prxs but they are functionally monomeric resulting in an intramolecu-lar disulfide bond instead (31, 180). To recycle the disulfide, both appear to use thioredoxin (or other disulfide reductases) as electron donor, completing the catalytic cycle (202). Only one active cysteine is conserved in the last class of Prxs, the 1-Cys Prxs, where the peroxidatic cysteine sulfenic acid is probably reduced by a thiol containing electron donor, for example glu-tathione or glutaredoxin (53, 77, 107, 155).

The reported BCP from E. coli seems to be a monomeric peroxiredoxin even under oxidized conditions indicating a member of the 1-Cys Prxs or the atypical 2-Cys Prxs class. Replacing the three different cysteines present in the primary sequence of E. coli BCP showed that Cys45 is the peroxidatic cysteine and the other two cysteines present were not important for activity, indicating the 1-Cys Prxs class. Interestingly, recycling of BCP required thioredoxin, a characteristic for the typical and atypical 2-Cys Prxs. In a later report the sequence surrounding the peroxidatic cysteine was shown to be FVCP for typical/atypical 2-Cys Prxs and PVCT for 1-Cys Prxs (169). The

36

surrounding sequence in E. coli BCP is PGCT. Summing up, it seems to be complex to classify the E. coli BCP.

To analyze the C-terminal part of R. rubrum GlnE (from now referred to as BCP), we cloned the corresponding sequence using a start codon in frame with GlnE into an overexpressing vector. By using a non-reducing SDS-PAGE system, we could detect homodimeric formation of BCP indicating that R. rubrum BCP is a typical 2-Cys Prxs. We also used non-reducing SDS-PAGE to analyze the oligomeric state of overexpressed native GlnE but the results were inconclusive. To further characterize the peroxiredoxin ac-tivity of BCP we replaced the two cysteine residues indicated in Figure 11 with serine and analyzed the ability of the native, single and double mutant BCP to protect GS with the MCO system described in the oxidative modifi-cation of GS section. By replacing the first cysteine residue, BCP activity was abolished, but for the other cysteine mutation some activity was re-tained. The variant with both point mutations showed no protection activity. These results are similar to the E. coli BCP characterization (86). In Figure 11 the primary sequences of E. coli and R. rubrum BCP are compared using Clustal W.

Figure 11. The primary sequence similarities between E. coli and R. rubrum BCP using Clustal W. The arrows indicate the two cysteines replaced by serine in R. rubrum BCP.

When using purified GlnE activity (see GlnE in R. rubrum section), gel-filtration results indicated a GlnE dimer around 220 kDa (unpublished re-sults). This might be in accordance with our non-reducing SDS-PAGE data indicating that R. rubrum BCP is a homodimer. The GlnE activity was puri-fied under oxidizing conditions that might trigger disulfide bonds to be formed between the subunits. We also showed that native GlnE possesses BCP activity in a concentration dependent manner and by removing the BCP part from GlnE (GlnE∆BCP) no BCP activity could be detected.

When initially analyzing BCP activity we used an ion exchange flow through fraction as a source of GS activity, but with this fraction we were

37

not able to detect any decrease in GS activity that would be the case if GS were not protected. This indicated that the fraction did possess some kind of radical protection mechanism. We later learned that we needed to use pure R. rubrum GS to detect the protection. The lowering of GS activity when exposed to the MCO system, has been shown to be due to weakening of the interactions between the subunits and/or non-enzymatic cleavage of GS (170). We detected some degradation products of GS when exposed to the MCO system but most of the protein remained intact even though around 90 % of GS activity was lost. To understand these results we analyzed the oli-gomeric state of GS using native gels, and the results indicated that GS ac-tivity was low possibly due to disassembly of the dodecameric structure when exposed to the MCO system.

It is certainly very tempting to speculate in terms of a specific protection activity of GlnE since it interacts with GS, or that the BCP domain provides a self protective mechanism for GlnE. To test the latter we exposed pure GlnE or GlnE without the BCP part (GlnE∆BCP) to the MCO system and stopped the reaction with EDTA, followed by analyzing the GlnE activity as adenylylation of GS. No change in GlnE activity could be detected for any of the proteins, indicating no self protecting role within GlnE.

In order to determine if GlnE had a specific protection mechanism for GS, a glnE mutant was generated. By growing R. rubrum aerobically we specu-lated that radicals should be formed and thereby the protection mechanism for GS should come into effect, leading to impaired growth rate in the mu-tant. When comparing the growth rate between the glnE mutant and wild-type, no difference was obtained, indicating that BCP is not essential. Proba-bly the lack of a specific oxygen sensitive phenotype in the glnE mutant suggests that the BCP part of GlnE is just one amongst many other protec-tion mechanisms e.g. glutathione peroxidase, other peroxiredoxins, catalase and superoxide dismutase. Genes for all these enzymes have been identified in the R. rubrum genome. This result could also explain the lack of decrease in GS activity when we used an ion exchange flow through fraction as a source of GS activity as mentioned above. The fraction probably contained other protecting enzymes.

An E. coli BCP mutant grew slower than wild-type in an aerobic envi-ronment and was also hypersensitive towards H2O2 (86). In Helicobacter

pylori, a BCP mutant phenotype was very weak in respect to oxidative stress, explained by the overlapping function of AhpC (Alkyl hydroperoxide reductase C), another much more abundant peroxiredoxin in H. pylori (195).

Why do peroxiredoxins exist when there are so many other antioxidants present in the cell? Possibly these enzymes have different specificities to-ward different peroxides, they could also be expressed at different levels within the cell or at different stages of cell growth. These questions are just beginning to be answered.

38

glnE mutation (Paper IV)

When further characterizing the R. rubrum glnE mutant, we expected GS activity to be very high as a phenotype, which would perhaps influence cell growth. Surprisingly, high GS activity was not detected even though the adenylylation/deadenylylation activities were abolished, and the mutant grew as wild-type. We also complemented the mutant in trans with glnE and glnE∆BCP, showing that adenylylation and deadenylylation could be de-tected. Complementing the mutant with BCP in trans did not restore any activity. To understand the observed lower GS activity we measured the specific activity of GS and performed Western Blots to examine the level of GS in each strain. The overall specific activity and level of GS were lower in all strains, with some individual differences between the strains.

At this point we decided to sequence glnA, the gene encoding GS, since the above results did not provide any answers. It turned out that all the glnA genes did have a point mutation except in the UR 1660 strain. This strain was not generated in our laboratory and the mutant was not designed exactly the same way as our glnE mutant strains (213). The reason for lower specific activity in UR 1660 could be due to lower amount of the GS protein. This means that a suppressor mutation could be in either the ntrB/C operon since expression of glnA is regulated through that system, or a mutation in the glnBA promoter. In the other glnE mutant strains the lower GS specific ac-tivity could be explained by previous studies of GS catalytic mechanism (45) briefly discussed in the GS section of this thesis. One really interesting resi-due is Glu328 that was mutated into a Lysine residue. The so called flap residue Glu328 has the ability to close the active site during transition state and thereby prohibit hydrolysis of the intermediate, γ-glutamyl phosphate. We cannot totally exclude that the difference in specific activity of GS is due to the assay conditions, i.e. that we used the γ-glutamyl hydroxamate assay. Arg344 (mutated to Ser) has been reported to be important in coordination of ADP, and for providing additional contacts for inter-subunit stabilization (45). Another suppressor mutation in one of the glnE mutant strains was Pro183 (mutated to Arg) which is close to the ammonium binding pocket and consequently likely to be important for the catalytic mechanism. The other suppressor mutations were located in regions that are conserved be-tween the R. rubrum and Salmonella typhimurium proteins but not taking part in the catalytic process.

What is the relevance for the cell to lower GS activity? Experiments done in S. typhimurium, showed that when overexpressing GS in a glnE mutant background, a severe growth defect was identified (106), but if GS was not overexpressed cell growth was normal, similar to our results. They con-cluded that the reason for the growth defect was depletion of the glutamate pool and not excess amounts of glutamine which would be synthesized by a very active GS. This report also postulated that the reason for adenylylation

39

of GS is mainly for protection of the glutamate pool, not depletion of the ATP pool as proposed by (177). A report from Zhang et al., showed that a ∆glnBglnJ double or ∆glnBglnJglnK triple mutant in R. rubrum showed a severe growth defect on a selection of different media, that was suppressed by mutations mapping to genes involved in GS regulation (213). Among the suppressor mutations they reported, are genes encoding ntrB/C, amtB and within glnA. They also generated a glnE mutant (UR 1660) and overex-pressed GS in that strain, giving the same phenotype as for S. typhimurium, a severe growth defect. We also used this strain in our experiments mentioned above (UR 1660). Their conclusion for the growth defect in the PII-mutants was due to increased GS activity that resulted in a 1000 fold increase in glutamine concentration and a 5 fold drop in glutamate concentration, and the suppressor mutations would be to prevent to high GS activity. Our find-ing that all glnE mutations studied had suppressor mutations goes well with the above argument: too high GS activity is deleterious to cell growth. Ex-actly what causes the toxicity with high GS activity is not clear. It is proba-bly related to an increase in the glutamine concentration and/or depletion of the glutamate/α-ketoglutarate or ATP pools. In R. rubrum the glutamine concentration is basically kept at a constant level in different growth condi-tions (98). To keep the glutamine concentration at a constant level, GS needs to be properly regulated and this would not be the situation in a ∆glnE, ∆glnBglnJ or ∆glnBglnJglnK mutant.

Concluding remarks

In a number of ways, GS regulation in R. rubrum is different from the E. coli

system that has been regarded as the paradigm for GS regulation. We do believe that the effector glutamine acts on GlnD and α-ketoglutarate on GlnE and the PII proteins. The uridylylation status or the ratio of modi-fied/unmodified PII proteins is set by the concentrations of both glutamine and α-ketoglutarate. The adenylylation status of GS is set by the α-ketoglutarate concentration and the ratio of modified/unmodified PII protein. Divalent cations do have an effect on uridylylation/deuridylylation of PII proteins but the physiological relevance of this phenomenon needs to be further investigated.

Another novel action of GlnE, the peroxiredoxin activity, has been dis-covered and partially characterized together with a putative explanation for the lowering of GS activity upon exposure to the MCO system.

When generating glnE mutants, we concluded that suppressor mutations will be produced when trying to manipulate enzymes/proteins involved in the regulation of GS, in accordance with the proposal from Zhang et al., when examining the R. rubrum ∆glnBglnJ or ∆glnBglnJglnK mutant (213). The exact cause for this occurrence is hard to distinguish but it is tempting to

40

speculate in terms of a too high intracellular level of glutamine, since the level of glutamine in R. rubrum has been reported to be constant independent of nitrogen availability that in this case could indicate an abnormal situation for R. rubrum (98). On the other hand, a low level of glutamate might also be the cause for suppressor mutation but in R. rubrum, the glutamate concen-tration has been shown to be varied with respect to nitrogen availability (98), maybe low glutamate is easier for the cells to cope with compared with high glutamine. Other metabolic candidates causing the suppressor mutations could be α-ketoglutarate and/or ATP. Altogether the suppressor mutations generated in the ∆glnE, ∆glnBglnJ or ∆glnBglnJglnK mutants emphasize the importance of a functional GS regulation in R. rubrum and need to be further investigated.

The ATP/ADP ratio as a signal in GS regulation has largely been ignored in this thesis due to previous results showing that under physiological condi-tions the ATP/ADP ratio is not changed, so the role as a signaling switch might be questionable (118, 151). Recently though, it has been suggested that the ATP/ADP ratio could act as a signaling switch indicating the intra-cellular energy situation (200). It was shown that ADP competes with ATP in the GlnJ-AmtB1 complex and thereby inhibits the dissociation between GlnJ and AmtB1. It might be a matter of interpreting published results, but their proposal is certainly interesting. In addition their results also showed that exposure to darkness promoted deuridylylation of GlnJ and perhaps also GlnB.

Another putative signal in R. rubrum would be the ratio between NADH/NAD+, which is high when the cells are exposed to light and low under darkness (84). NAD+ and darkness are “switch off” effectors for nitro-genase. Previous in vivo experiments have shown that a R. rubrum PII pro-tein is deuridylylated when exposed to NAD+ under nitrogen fixing condi-tions (95). We have not been able to verify this in vitro with purified compo-nents, indicative of a component missing in the assay.

Recently the 3D structure of M. jannaschii GlnK was released, showing for the first time the binding sites for α-ketoglutarate, ATP and Mg2+ (211). ATP and Mg2+ need to bind into GlnK before α-ketoglutarate thereby form-ing a compact structure of the T-loop. ATP and Mg2+ can alone form the compact structure that otherwise is an extended structure of the T-loop. An-other interesting finding from their crystal structures is the salt bridge formed in the compact state of GlnK between Glu44 and Lys58.

When interpreting the results from M. jannaschii with respect to R. ru-

brum PII protein configurations, our in vitro uridylylation results show that if only ATP or α-ketoglutarate is present in the uridylylation assay together with a divalent cation no uridylylation is observed, all three molecules need to be present. In other words, not only the T-loop has to be in the compact formation, also α-ketoglutarate has to be bound to the PII protein for uridyly-lation to occur or for GlnD interaction. Interestingly, for deuridylylation to

41

proceed, only Mn2+ and glutamine need to be mixed together with GlnB/GlnJ-UMP and GlnD, ATP and/or α-ketoglutarate can be omitted from the assay. The conformation (extended/compact) of the T-loop in this situa-tion is not known since the PII proteins are in the uridylylated state.

In our adenylylation assay ATP is the substrate (stabilized by Mg2+), which probably induces a compact T-loop conformation of the PII protein, but for interaction with GlnE, α-ketoglutarate is not necessary as the case is for GlnD.

When examining the membrane association of GlnJ in R. rubrum, it can be postulated that the T-loop is in the extended conformation, and upon dis-sociation, when ATP, Mg2+, Mn2+ and α-ketoglutarate is added, GlnJ might adopt the compact conformation of the T-loop (200) (unpublished results). The relevant divalent cation for GlnJ dissociation in our experiments is defi-nitely Mn2+, which is further supported by our in vitro results concerning uridylylation of GlnJ (Paper I). Our results indicate different characteristics from other reported GlnK-AmtB complex dissociations where only Mg2+ has been used (38, 211), except in B. subtilis were the release of GlnK from the membrane was shown to occur with either Mg2+ or Mn2+ (74).

Future attractive experiments would be to investigate different GlnJ vari-ants, for example, one with a substitution of Arg47, at the tip of the T-loop. How would that affect AmtB1 interaction? Another interesting feature of the T-loop, the salt bridge between Glu44 and Lys58 formed in the compact configuration would be interesting to analyze by site-directed mutagenesis. The effect of different amino acid substitutions on the ability of GlnJ to in-teract not only with AmtB1, but also with GlnE and GlnD would provide further knowledge about the T-loop within PII proteins.

Other informative experiments would for example be cross-linking ex-periments with the PII proteins and GlnE to elucidate if there are multiple sites for PII/PII-UMP as for E. coli GlnE and under which conditions the complex is formed. Another interesting experiment could be site-directed mutagenesis of GlnD to further investigate the glutamine site, and if there also might be a α-ketoglutarate site. The role of NAD+ or the ratio between NADH/NAD+ needs to be further examined with respect to PII protein modi-fication.

The different roles of the three PII proteins in R. rubrum need to be fur-ther clarified. Under nitrogen rich conditions mostly unmodified GlnB is present in the cell, probably indicating its involvement in adenylylation of GS. Under nitrogen poor conditions expression of glnJ is induced by the NtrB/C system, and post-translationally both GlnJ and GlnB are modified. Proposed roles for GlnB-UMP under these conditions would be activation of NifA and competition for GlnE with non-uridylylated GlnB to inhibit ade-nylylation of GS. The role of GlnJ-UMP under nitrogen poor conditions has not been fully elucidated yet.

42

When a nitrogen poor culture is exposed to ammonium ions (“switch off”), GlnB/GlnJ-UMP are deuridylylated and GlnB interacts again with GlnE, for adenylylation of GS. Under these conditions GlnJ probably blocks the ammonium transporter (AmtB) since non-uridylylated GlnJ has been shown to interact with AmtB. To date the role for GlnK has not been eluci-dated.

43

En enkel förklaring till vad jag gjort

Kväve är essentiellt för liv eftersom många livsnödvändiga molekyler inehåller kväve, t.ex. DNA, RNA, och proteiner. Kväve finns i luften och utgör nästan 80 % av atmosfären. Vi kan endast tillgodogöra oss kvävet i atmosfären om det omvandlas till ammonium och endast ett fåtal bakterier, bl.a. den jag studerat, Rhodospirillum rubrum har den egenskapen. Denna process kallas för kvävefixering och utförs av ett protein som kallas för nitrogenas. När nitrogenaset som finns inne i cellen producerat ammonium byggs det in i aminosyran glutamin som används för att bilda andra kväveföreningar. Det protein som bildar glutamin kallas glutaminsyntetas. Glutaminsyntetas kan stängas av om det redan finns för mycket glutamin i cellen, eller sättas på om det finns för lite glutamin. Min forskning har varit koncentrerad till regleringen av detta protein.

Jag har undersökt regleringen genom att isolera alla proteiner involverade i denna process. Detta har varit en utmaning med tanke på att det finns flera tusen olika proteiner i en bakteriecell och att varje protein även finns i olika mängder i cellen. När alla proteiner blivit isolerade har dessa använts för att undersöka vad som händer med glutaminsyntetaset.

Ett av de isolerade proteinerna (GlnE) har jag även muterat. Detta innebär att man tar bort proteinet eller förändrar dess egenskaper i bakteriecellen. Mutering kan liknas vid att ta bort delar från en bilmotor om man tänker sig bilmotorn som en bakteriecell, för att sedan undersöka om den startar. Beroende på vilken del man tar bort kommer den att fungera lite olika. Resultaten från dessa studier visar att glutaminsyntetasets karaktär förändras när man muterar GlnE vilket är mycket viktig information för vidare studier av R. rubrum och även av andra organismer.

Vidare visar resultaten av min forskning att regleringen av glutaminsyntetas inte är som i model organismen E. coli. Istället har resultaten påvisat flera viktiga skillnader gällande signaler för av och på mekanismen för glutaminsyntetas i R. rubrum. Då forskningsområdet angående reglering av glutaminsyntetas är väldigt stort så blir resultaten från min forskning mycket användbara. Den har även påvisat en helt ny aktivitet i ett av de isolerade proteinerna som aldrig tidigare blivit beskrivet.

44

Acknowledgements

Finally this challenge is done! It has been frustrating at times, but also very rewarding. Apart from learning science, I have also learned a lot about my-self maybe that has been the most rewarding experience of all. The first person I would like to thank is my supervisor Stefan N, not only for giving me a salary, but also for being very patient. It has been a pleasure talking science with you, but more importantly, your interest in discussing football made every day much more enjoyable. I will miss you. I also would like to thank Tomas E, for your ability and patients in respect to teaching me a lot about molecular biology techniques. I must say that the last sprint has been extremely long though….

Pedro, my lab-companion and co-author, your contribution to my/our pro-ject has been invaluable! How can you know so much science and still be so humble?

Helen, thank you for always remembering my birthday, but more impor-tantly, take good care of Tomas E. Fang Sun! To former and present mem-bers of the group, Anders L, Tiago, Agneta, and the girls in my office, Hanna and Kajsa, it has been a pleasure having you around and good luck with your own work.

To everyone at the department who has helped me in my research, or in other areas, thank you!

Jag vill även tacka Sven och Lilly Lawskis, Ekströms, Smitts, och Helge Ax:son Johnsons stipendiefonder för stålarna!

45

Den lilla släkten jag har; Olle (styvbror), Ola (styvpappa), Wenche (Olles fru), Christopher (Wenche o Olles son), Ingrid (moster), Ans (morbror), Jenny (kusin), Pernilla (kusin), Gunilla (Louise mamma), och Hans (mammas snubbe, alias ”crockodile shoes”). Tack alla för att ni finns. Mina vänner: Larsson, Nisse, Björn, Pudas, Bjälken, TJ, Nille, Asplund, Foppa, Jakob, Hagge, Dalle, Grane, Swante, Frazze, Hellquist, Goran, och Gileus. Ni alla definierar vänskap för mig på olika sätt. Till mamma, den starkaste och mest omtänksamma! Jag är stolt över att vara din son. Louise, nu är plugget slut och nya utmaningar väntar…ILY!

46

References

1. Adler, S. P., D. Purich, and E. R. Stadtman. 1975. Cascade control of Escherichia coli glutamine synthetase. Properties of the PII regulatory protein and the uridylyltransferase-uridylyl-removing enzyme. J Biol Chem 250:6264-72.

2. Almassy, R. J., C. A. Janson, R. Hamlin, N. H. Xuong, and D. Eisenberg. 1986. Novel subunit-subunit interactions in the structure of glutamine synthetase. Nature 323:304-9.

3. Andrade, S. L., A. Dickmanns, R. Ficner, and O. Einsle. 2005. Crystal structure of the archaeal ammonium transporter Amt-1 from Archaeoglobus fulgidus. Proc Natl Acad Sci U S A 102:14994-9.

4. Araujo, M. S., V. A. Baura, E. M. Souza, E. M. Benelli, L. U. Rigo, M. B. Steffens, F. O. Pedrosa, and L. S. Chubatsu. 2004. In vitro uridylylation of the Azospirillum brasilense N-signal transducing GlnZ protein. Protein Expr Purif 33:19-24.

5. Arcondeguy, T., R. Jack, and M. Merrick. 2001. P(II) signal transduction proteins, pivotal players in microbial nitrogen control. Microbiol Mol Biol Rev 65:80-105.

6. Arsene, F., P. A. Kaminski, and C. Elmerich. 1996. Modulation of NifA activity by PII in Azospirillum brasilense: evidence for a regulatory role of the NifA N-terminal domain. J Bacteriol 178:4830-8.

7. Atkinson, M. R., T. A. Blauwkamp, V. Bondarenko, V. Studitsky, and A. J. Ninfa. 2002. Activation of the glnA, glnK, and nac promoters as Escherichia coli undergoes the transition from nitrogen excess growth to nitrogen starvation. J Bacteriol 184:5358-63.

8. Atkinson, M. R., T. A. Blauwkamp, and A. J. Ninfa. 2002. Context-dependent functions of the PII and GlnK signal transduction proteins in Escherichia coli. J Bacteriol 184:5364-75.

9. Atkinson, M. R., E. S. Kamberov, R. L. Weiss, and A. J. Ninfa. 1994. Reversible uridylylation of the Escherichia coli PII signal transduction protein regulates its ability to stimulate the dephosphorylation of the transcription factor nitrogen regulator I (NRI or NtrC). J Biol Chem 269:28288-93.

10. Atkinson, M. R., and A. J. Ninfa. 1999. Characterization of the GlnK protein of Escherichia coli. Mol Microbiol 32:301-13.

11. Atkinson, M. R., and A. J. Ninfa. 1998. Role of the GlnK signal transduction protein in the regulation of nitrogen assimilation in Escherichia coli. Mol Microbiol 29:431-47.

47

12. Bauer, C. C., L. Scappino, and R. Haselkorn. 1993. Growth of the cyanobacterium Anabaena on molecular nitrogen: NifJ is required when iron is limited. Proc Natl Acad Sci U S A 90:8812-6.

13. Blanco, G., M. Drummond, P. Woodley, and C. Kennedy. 1993. Sequence and molecular analysis of the nifL gene of Azotobacter vinelandii. Mol Microbiol 9:869-79.

14. Blauwkamp, T. A., and A. J. Ninfa. 2003. Antagonism of PII signalling by the AmtB protein of Escherichia coli. Mol Microbiol 48:1017-28.

15. Blauwkamp, T. A., and A. J. Ninfa. 2002. Physiological role of the GlnK signal transduction protein of Escherichia coli: survival of nitrogen starvation. Mol Microbiol 46:203-14.

16. Bonatto, A. C., G. H. Couto, E. M. Souza, L. M. Araujo, F. O. Pedrosa, L. Noindorf, and E. M. Benelli. 2007. Purification and characterization of the bifunctional uridylyltransferase and the signal transducing proteins GlnB and GlnK from Herbaspirillum seropedicae. Protein Expr Purif. doi:10.1016/j.pep.2007.04.012.

17. Buchanan, B. B., and D. I. Arnon. 1990. A reverse KREBS cycle in photosynthesis: consensus at last. Photosynth Res 24:47-53.

18. Bueno, R., G. Pahel, and B. Magasanik. 1985. Role of glnB and glnD gene products in regulation of the glnALG operon of Escherichia coli. J Bacteriol 164:816-22.

19. Cheah, E., P. D. Carr, P. M. Suffolk, S. G. Vasudevan, N. E. Dixon, and D. L. Ollis. 1994. Structure of the Escherichia coli signal transducing protein PII. Structure 2:981-90.

20. Chen, Y. M., T. S. Ferrar, E. M. Lohmeier-Vogel, N. Morrice, Y. Mizuno, B. Berenger, K. K. Ng, D. G. Muench, and G. B. Moorhead. 2006. The PII signal transduction protein of Arabidopsis thaliana forms an arginine-regulated complex with plastid N-acetyl glutamate kinase. J Biol Chem 281:5726-33.

21. Cheng, J., M. Johansson, and S. Nordlund. 1999. Expression of P(II) and glutamine synthetase is regulated by P(II), the ntrBC products, and processing of the glnBA mRNA in Rhodospirillum rubrum. J Bacteriol 181:6530-4.

22. Chipman, D. M., and B. Shaanan. 2001. The ACT domain family. Curr Opin Struct Biol 11:694-700.

23. Clancy, P., Y. Xu, W. C. van Heeswijk, S. G. Vasudevan, and D. L. Ollis. 2007. The domains carrying the opposing activities in adenylyltransferase are separated by a central regulatory domain. FEBS J 274:2865-77.

24. Claverie-Martin, F., and B. Magasanik. 1991. Role of integration host factor in the regulation of the glnHp2 promoter of Escherichia coli. Proc Natl Acad Sci U S A 88:1631-5.

25. Colnaghi, R., P. Rudnick, L. He, A. Green, D. Yan, E. Larson, and C. Kennedy. 2001. Lethality of glnD null mutations in Azotobacter vinelandii is suppressible by prevention of glutamine synthetase adenylylation. Microbiology 147:1267-76.

48

26. Connelly, H. M., D. A. Pelletier, T. Y. Lu, P. K. Lankford, and R. L. Hettich. 2006. Characterization of pII family (GlnK1, GlnK2, and GlnB) protein uridylylation in response to nitrogen availability for Rhodopseudomonas palustris. Anal Biochem 357:93-104.

27. Conroy, M. J., A. Durand, D. Lupo, X. D. Li, P. A. Bullough, F. K. Winkler, and M. Merrick. 2007. The crystal structure of the Escherichia coli AmtB-GlnK complex reveals how GlnK regulates the ammonia channel. Proc Natl Acad Sci U S A 104:1213-8.

28. Contreras, A., M. Drummond, A. Bali, G. Blanco, E. Garcia, G. Bush, C. Kennedy, and M. Merrick. 1991. The product of the nitrogen fixation regulatory gene nfrX of Azotobacter vinelandii is functionally and structurally homologous to the uridylyltransferase encoded by glnD in enteric bacteria. J Bacteriol 173:7741-49.

29. Coutts, G., G. Thomas, D. Blakey, and M. Merrick. 2002. Membrane sequestration of the signal transduction protein GlnK by the ammonium transporter AmtB. EMBO J 21:536-45.

30. Darrow, R. A., and R. R. Knotts. 1977. Two forms of glutamine synthetase in free-living root-nodule bacteria. Biochem Biophys Res Commun 78:554-9.

31. Declercq, J. P., C. Evrard, A. Clippe, D. V. Stricht, A. Bernard, and B. Knoops. 2001. Crystal structure of human peroxiredoxin 5, a novel type of mammalian peroxiredoxin at 1.5 A resolution. J Mol Biol 311:751-9.

32. Detsch, C., and J. Stulke. 2003. Ammonium utilization in Bacillus subtilis: transport and regulatory functions of NrgA and NrgB. Microbiology 149:3289-97.

33. Dixon, R., and D. Kahn. 2004. Genetic regulation of biological nitrogen fixation. Nat Rev Microbiol 2:621-31.

34. Dixon, R. O. D., and C. T. Wheeler. 1986. Nitrogen fixation in plants. Chapman and Hall, New York.

35. Dodsworth, J. A., N. C. Cady, and J. A. Leigh. 2005. 2-Oxoglutarate and the PII homologues NifI1 and NifI2 regulate nitrogenase activity in cell extracts of Methanococcus maripaludis. Mol Microbiol 56:1527-38.

36. Drummond, M., J. Clements, M. Merrick, and R. Dixon. 1983. Positive control and autogenous regulation of the nifLA promoter in Klebsiella pneumoniae. Nature 301:302-7.

37. Dubbs, J. M., and F. R. Tabita. 2004. Regulators of nonsulfur purple phototrophic bacteria and the interactive control of CO2 assimilation, nitrogen fixation, hydrogen metabolism and energy generation. FEMS Microbiol Rev 28:353-76.

38. Durand, A., and M. Merrick. 2006. In vitro analysis of the Escherichia coli AmtB-GlnK complex reveals a stoichiometric interaction and sensitivity to ATP and 2-oxoglutarate. J Biol Chem 281:29558-67.

39. Eady, R. R., B. E. Smith, K. A. Cook, and J. R. Postgate. 1972. Nitrogenase of Klebsiella pneumoniae. Purification and properties of

49

the component proteins. Biochem J 128:655-75. 40. Edgren, T., and S. Nordlund. 2005. Electron transport to

nitrogenase in Rhodospirillum rubrum: identification of a new fdxN gene encoding the primary electron donor to nitrogenase. FEMS Microbiol Lett 245:345-51.

41. Edgren, T., and S. Nordlund. 2004. The fixABCX genes in Rhodospirillum rubrum encode a putative membrane complex participating in electron transfer to nitrogenase. J Bacteriol 186:2052-60.

42. Edgren, T., and S. Nordlund. 2006. Two pathways of electron transport to nitrogenase in Rhodospirillum rubrum: the major pathway is dependent on the fix gene products. FEMS Microbiol Lett 260:30-5.

43. Edwards, R., and M. Merrick. 1995. The role of uridylyltransferase in the control of Klebsiella pneumoniae nif gene regulation. Mol Gen Genet 247:189-98.

44. Ehlers, C., K. Weidenbach, K. Veit, K. Forchhammer, and R. A. Schmitz. 2005. Unique mechanistic features of post-translational regulation of glutamine synthetase activity in Methanosarcina mazei strain Go1 in response to nitrogen availability. Mol Microbiol 55:1841-54.

45. Eisenberg, D., H. S. Gill, G. M. Pfluegl, and S. H. Rotstein. 2000. Structure-function relationships of glutamine synthetases. Biochim Biophys Acta 1477:122-45.

46. Ellis, H. R., and L. B. Poole. 1997. Novel application of 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole to identify cysteine sulfenic acid in the AhpC component of alkyl hydroperoxide reductase. Biochemistry 36:15013-8.

47. Ellis, H. R., and L. B. Poole. 1997. Roles for the two cysteine residues of AhpC in catalysis of peroxide reduction by alkyl hydroperoxide reductase from Salmonella typhimurium. Biochemistry 36:13349-56.

48. Elsen, S., W. Dischert, A. Colbeau, and C. E. Bauer. 2000. Expression of uptake hydrogenase and molybdenum nitrogenase in Rhodobacter capsulatus is coregulated by the RegB-RegA two-component regulatory system. J Bacteriol 182:2831-7.

49. Esmarch, E. 1887. Uber die Reinkultur eines Spirillum. Zentralbl. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt. I Orig. 1:225-30.

50. Eydmann, T., E. Soderback, T. Jones, S. Hill, S. Austin, and R. Dixon. 1995. Transcriptional activation of the nitrogenase promoter in vitro: adenosine nucleotides are required for inhibition of NIFA activity by NIFL. J Bacteriol 177:1186-95.

51. Fischer, H. M. 1994. Genetic regulation of nitrogen fixation in rhizobia. Microbiol Rev 58:352-86.

52. Fischer, H. M., T. Bruderer, and H. Hennecke. 1988. Essential and non-essential domains in the Bradyrhizobium japonicum NifA protein: identification of indispensable cysteine residues potentially

50

involved in redox reactivity and/or metal binding. Nucleic Acids Res 16:2207-24.

53. Fisher, A. B., C. Dodia, Y. Manevich, J. W. Chen, and S. I. Feinstein. 1999. Phospholipid hydroperoxides are substrates for non-selenium glutathione peroxidase. J Biol Chem 274:21326-34.

54. Fitzmaurice, W. P., L. L. Saari, R. G. Lowery, P. W. Ludden, and G. P. Roberts. 1989. Genes coding for the reversible ADP-ribosylation system of dinitrogenase reductase from Rhodospirillum rubrum. Mol Gen Genet 218:340-7.

55. Forchhammer, K. 2004. Global carbon/nitrogen control by PII signal transduction in cyanobacteria: from signals to targets. FEMS Microbiol Rev 28:319-33.

56. Forchhammer, K. 2007. Glutamine signalling in bacteria. Front Biosci 12:358-70.

57. Forchhammer, K., and A. Hedler. 1997. Phosphoprotein PII from cyanobacteria--analysis of functional conservation with the PII signal-transduction protein from Escherichia coli. Eur J Biochem 244:869-75.

58. Forchhammer, K., A. Hedler, H. Strobel, and V. Weiss. 1999. Heterotrimerization of PII-like signalling proteins: implications for PII-mediated signal transduction systems. Mol Microbiol 33:338-49.

59. Forchhammer, K., and N. Tandeau de Marsac. 1995. Phosphorylation of the PII protein (glnB gene product) in the cyanobacterium Synechococcus sp. strain PCC 7942: analysis of in vitro kinase activity. J Bacteriol 177:5812-7.

60. Forchhammer, K., and N. Tandeau de Marsac. 1994. The PII protein in the cyanobacterium Synechococcus sp. strain PCC 7942 is modified by serine phosphorylation and signals the cellular N-status. J Bacteriol 176:84-91.

61. Francis, S. H., and E. G. Engleman. 1978. Cascade control of E. coli glutamine synthetase. I. Studies on the uridylyl transferase and uridylyl removing enzyme(s) from E. coli. Arch Biochem Biophys 191:590-601.

62. Gage, D. J. 2004. Infection and invasion of roots by symbiotic, nitrogen-fixing rhizobia during nodulation of temperate legumes. Microbiol Mol Biol Rev 68:280-300.

63. Galmozzi, C. V., M. J. Fernandez-Avila, J. C. Reyes, F. J. Florencio, and M. I. Muro-Pastor. 2007. The ammonium-inactivated cyanobacterial glutamine synthetase I is reactivated in vivo by a mechanism involving proteolytic removal of its inactivating factors. Mol Microbiol 65:166-79.

64. Garcia-Dominguez, M., J. C. Reyes, and F. J. Florencio. 1999. Glutamine synthetase inactivation by protein-protein interaction. Proc Natl Acad Sci U S A 96:7161-6.

65. Garcia, E., and S. G. Rhee. 1983. Cascade control of Escherichia coli glutamine synthetase. Purification and properties of PII uridylyltransferase and uridylyl-removing enzyme. J Biol Chem

51

258:2246-53. 66. Gest, H., and M. D. Kamen. 1949. Studies on the metabolism of

photosynthetic bacteria IV. Photochemical Production of Molecular Hydrogen by Growing Cultures of Photosynthetic Bacteria. J Bacteriol 58:239-45.

67. Gill, H. S., and D. Eisenberg. 2001. The crystal structure of phosphinothricin in the active site of glutamine synthetase illuminates the mechanism of enzymatic inhibition. Biochemistry 40:1903-12.

68. Gill, H. S., G. M. Pfluegl, and D. Eisenberg. 2002. Multicopy crystallographic refinement of a relaxed glutamine synthetase from Mycobacterium tuberculosis highlights flexible loops in the enzymatic mechanism and its regulation. Biochemistry 41:9863-72.

69. Gill, H. S., G. M. Pfluegl, and D. Eisenberg. 1999. Preliminary crystallographic studies on glutamine synthetase from Mycobacterium tuberculosis. Acta Crystallogr D Biol Crystallogr 55:865-8.

70. Grant, G. A. 2006. The ACT domain: a small molecule binding domain and its role as a common regulatory element. J Biol Chem 281:33825-9.

71. Gruswitz, F., J. O'Connell, 3rd, and R. M. Stroud. 2007. Inhibitory complex of the transmembrane ammonia channel, AmtB, and the cytosolic regulatory protein, GlnK, at 1.96 A. Proc Natl Acad Sci U S A 104:42-7.

72. Haschemeyer, R. H. 1968. Electron microscopy of enzymes. Trans N Y Acad Sci 30:875-91.

73. Heinrich, A., M. Maheswaran, U. Ruppert, and K. Forchhammer. 2004. The Synechococcus elongatus PII signal transduction protein controls arginine synthesis by complex formation with N-acetyl-L-glutamate kinase. Mol Microbiol 52:1303-14.

74. Heinrich, A., K. Woyda, K. Brauburger, G. Meiss, C. Detsch, J. Stulke, and K. Forchhammer. 2006. Interaction of the membrane-bound GlnK-AmtB complex with the master regulator of nitrogen metabolism TnrA in Bacillus subtilis. J Biol Chem 281:34909-17.

75. Herrero, A., A. M. Muro-Pastor, and E. Flores. 2001. Nitrogen control in cyanobacteria. J Bacteriol 183:411-25.

76. Hill, S., S. Austin, T. Eydmann, T. Jones, and R. Dixon. 1996. Azotobacter vinelandii NIFL is a flavoprotein that modulates transcriptional activation of nitrogen-fixation genes via a redox-sensitive switch. Proc Natl Acad Sci U S A 93:2143-8.

77. Hofmann, B., H. J. Hecht, and L. Flohe. 2002. Peroxiredoxins. Biol Chem 383:347-64.

78. Hsieh, M. H., and H. M. Goodman. 2002. Molecular characterization of a novel gene family encoding ACT domain repeat proteins in Arabidopsis. Plant Physiol 130:1797-806.

79. Huergo, L. F., L. S. Chubatsu, E. M. Souza, F. O. Pedrosa, M. B.

52

Steffens, and M. Merrick. 2006. Interactions between PII proteins and the nitrogenase regulatory enzymes DraT and DraG in Azospirillum brasilense. FEBS Lett 580:5232-6.

80. Hunt, T. P., and B. Magasanik. 1985. Transcription of glnA by purified Escherichia coli components: core RNA polymerase and the products of glnF, glnG, and glnL. Proc Natl Acad Sci U S A 82:8453-7.

81. Irmler, A., and K. Forchhammer. 2001. A PP2C-type phosphatase dephosphorylates the PII signaling protein in the cyanobacterium Synechocystis PCC 6803. Proc Natl Acad Sci U S A 98:12978-83.

82. Irmler, A., S. Sanner, H. Dierks, and K. Forchhammer. 1997. Dephosphorylation of the phosphoprotein P(II) in Synechococcus PCC 7942: identification of an ATP and 2-oxoglutarate-regulated phosphatase activity. Mol Microbiol 26:81-90.

83. Jack, R., M. De Zamaroczy, and M. Merrick. 1999. The signal transduction protein GlnK is required for NifL-dependent nitrogen control of nif gene expression in Klebsiella pneumoniae. J Bacteriol 181:1156-62.

84. Jackson, J. B., and A. R. Crofts. 1968. Energy-linked reduction of nicotinamide adenine dinucleotides in cells of Rhodospirillum rubrum. Biochem Biophys Res Commun 32:908-15.

85. Jaggi, R., W. C. van Heeswijk, H. V. Westerhoff, D. L. Ollis, and S. G. Vasudevan. 1997. The two opposing activities of adenylyl transferase reside in distinct homologous domains, with intramolecular signal transduction. EMBO J 16:5562-71.

86. Jeong, W., M. K. Cha, and I. H. Kim. 2000. Thioredoxin-dependent hydroperoxide peroxidase activity of bacterioferritin comigratory protein (BCP) as a new member of the thiol-specific antioxidant protein (TSA)/Alkyl hydroperoxide peroxidase C (AhpC) family. J Biol Chem 275:2924-30.

87. Jiang, P., A. E. Mayo, and A. J. Ninfa. 2007. Escherichia coli glutamine synthetase adenylyltransferase (ATase, EC 2.7.7.49): kinetic characterization of regulation by PII, PII-UMP, glutamine, and alpha-ketoglutarate. Biochemistry 46:4133-46.

88. Jiang, P., and A. J. Ninfa. 1999. Regulation of autophosphorylation of Escherichia coli nitrogen regulator II by the PII signal transduction protein. J Bacteriol 181:1906-11.

89. Jiang, P., J. A. Peliska, and A. J. Ninfa. 1998. Enzymological characterization of the signal-transducing uridylyltransferase/uridylyl-removing enzyme (EC 2.7.7.59) of Escherichia coli and its interaction with the PII protein. Biochemistry 37:12782-94.

90. Jiang, P., J. A. Peliska, and A. J. Ninfa. 1998. The regulation of Escherichia coli glutamine synthetase revisited: role of 2-ketoglutarate in the regulation of glutamine synthetase adenylylation state. Biochemistry 37:12802-10.

91. Jiang, P., A. A. Pioszak, and A. J. Ninfa. 2007. Structure-function

53

analysis of glutamine synthetase adenylyltransferase (ATase, EC 2.7.7.49) of Escherichia coli. Biochemistry 46:4117-32.

92. Jiang, P., P. Zucker, M. R. Atkinson, E. S. Kamberov, W. Tirasophon, P. Chandran, B. R. Schefke, and A. J. Ninfa. 1997. Structure/function analysis of the PII signal transduction protein of Escherichia coli: genetic separation of interactions with protein receptors. J Bacteriol 179:4342-53.

93. Johansson, M., and S. Nordlund. 1999. Purification of P(II) and P(II)-UMP and in vitro studies of regulation of glutamine synthetase in Rhodospirillum rubrum. J Bacteriol 181:6524-29.

94. Johansson, M., and S. Nordlund. 1996. Transcription of the glnB and glnA genes in the photosynthetic bacterium Rhodospirillum rubrum. Microbiology 142 1265-72.

95. Johansson, M., and S. Nordlund. 1997. Uridylylation of the P(II) protein in the photosynthetic bacterium Rhodospirillum rubrum. J Bacteriol 179:4190-4.

96. Kamberov, E. S., M. R. Atkinson, J. Feng, P. Chandran, and A. J. Ninfa. 1994. Sensory components controlling bacterial nitrogen assimilation. Cell Mol Biol Res 40:175-91.

97. Kamberov, E. S., M. R. Atkinson, and A. J. Ninfa. 1995. The Escherichia coli PII signal transduction protein is activated upon binding 2-ketoglutarate and ATP. J Biol Chem 270:17797-807.

98. Kanemoto, R. H., and P. W. Ludden. 1987. Amino acid concentrations in Rhodospirillum rubrum during expression and switch-off of nitrogenase activity. J Bacteriol 169:3035-43.

99. Kanemoto, R. H., and P. W. Ludden. 1984. Effect of ammonia, darkness, and phenazine methosulfate on whole-cell nitrogenase activity and Fe protein modification in Rhodospirillum rubrum. J Bacteriol 158:713-20.

100. Khademi, S., J. O'Connell, 3rd, J. Remis, Y. Robles-Colmenares, L. J. Miercke, and R. M. Stroud. 2004. Mechanism of ammonia transport by Amt/MEP/Rh: structure of AmtB at 1.35 A. Science 305:1587-94.

101. Kim, I. H., S. J. Kwak, J. Kang, and S. C. Park. 1998. Transcriptional control of the glnD gene is not dependent on nitrogen availability in Escherichia coli. Mol Cells 8:483-90.

102. Kim, K., I. H. Kim, K. Y. Lee, S. G. Rhee, and E. R. Stadtman. 1988. The isolation and purification of a specific "protector" protein which inhibits enzyme inactivation by a thiol/Fe(III)/O2 mixed-function oxidation system. J Biol Chem 263:4704-11.

103. Kim, K., S. G. Rhee, and E. R. Stadtman. 1985. Nonenzymatic cleavage of proteins by reactive oxygen species generated by dithiothreitol and iron. J Biol Chem 260:15394-7.

104. Kingdon, H. S., B. M. Shapiro, and E. R. Stadtman. 1967. Regulation of glutamine synthetase. 8. ATP: glutamine synthetase adenylyltransferase, an enzyme that catalyzes alterations in the regulatory properties of glutamine synthetase. Proc Natl Acad Sci U

54

S A 58:1703-10. 105. Kloft, N., G. Rasch, and K. Forchhammer. 2005. Protein

phosphatase PphA from Synechocystis sp. PCC 6803: the physiological framework of PII-P dephosphorylation. Microbiology 151:1275-83.

106. Kustu, S., J. Hirschman, D. Burton, J. Jelesko, and J. C. Meeks. 1984. Covalent modification of bacterial glutamine synthetase: physiological significance. Mol Gen Genet 197:309-17.

107. Lee, S. P., Y. S. Hwang, Y. J. Kim, K. S. Kwon, H. J. Kim, K. Kim, and H. Z. Chae. 2001. Cyclophilin a binds to peroxiredoxins and activates its peroxidase activity. J Biol Chem 276:29826-32.

108. Lengeler, J. W., G. Drews, and H. G. Schlegel. 1999. Biology of the Prokaryotes. Blackwell Science, Oxford UK.

109. Levine, R. L. 1983. Oxidative modification of glutamine synthetase. I. Inactivation is due to loss of one histidine residue. J Biol Chem 258:11823-7.

110. Levine, R. L., L. Mosoni, B. S. Berlett, and E. R. Stadtman. 1996. Methionine residues as endogenous antioxidants in proteins. Proc Natl Acad Sci U S A 93:15036-40.

111. Levine, R. L., C. N. Oliver, R. M. Fulks, and E. R. Stadtman. 1981. Turnover of bacterial glutamine synthetase: oxidative inactivation precedes proteolysis. Proc Natl Acad Sci U S A 78:2120-4.

112. Liang, Y. Y., F. Arsene, and C. Elmerich. 1993. Characterization of the ntrBC genes of Azospirillum brasilense Sp7: their involvement in the regulation of nitrogenase synthesis and activity. Mol Gen Genet 240:188-96.

113. Liaw, S. H., and D. Eisenberg. 1994. Structural model for the reaction mechanism of glutamine synthetase, based on five crystal structures of enzyme-substrate complexes. Biochemistry 33:675-81.

114. Liaw, S. H., G. Jun, and D. Eisenberg. 1993. Extending the diffraction limit of protein crystals: the example of glutamine synthetase from Salmonella typhimurium in the presence of its cofactor ATP. Protein Sci 2:470-1.

115. Liaw, S. H., G. Jun, and D. Eisenberg. 1994. Interactions of nucleotides with fully unadenylylated glutamine synthetase from Salmonella typhimurium. Biochemistry 33:11184-8.

116. Liaw, S. H., C. Pan, and D. Eisenberg. 1993. Feedback inhibition of fully unadenylylated glutamine synthetase from Salmonella typhimurium by glycine, alanine, and serine. Proc Natl Acad Sci U S A 90:4996-5000.

117. Lindblad, A., J. Jansson, E. Brostedt, M. Johansson, U. Hellman, and S. Nordlund. 1996. Identification and sequence of a nifJ-like gene in Rhodospirillum rubrum: partial characterization of a mutant unaffected in nitrogen fixation. Mol Microbiol 20:559-68.

118. Lindblad, A., and S. Nordlund. 1997. Electron transport to nitrogenase in Rhodospirillum rubrum: Role of energization of the

55

chromatophore membrane. Photosynth Res 53:23-8. 119. Little, R., V. Colombo, A. Leech, and R. Dixon. 2002. Direct

interaction of the NifL regulatory protein with the GlnK signal transducer enables the Azotobacter vinelandii NifL-NifA regulatory system to respond to conditions replete for nitrogen. J Biol Chem 277:15472-81.

120. Little, R., and R. Dixon. 2003. The amino-terminal GAF domain of Azotobacter vinelandii NifA binds 2-oxoglutarate to resist inhibition by NifL under nitrogen-limiting conditions. J Biol Chem 278:28711-8.

121. Little, R., F. Reyes-Ramirez, Y. Zhang, W. C. van Heeswijk, and R. Dixon. 2000. Signal transduction to the Azotobacter vinelandii NIFL-NIFA regulatory system is influenced directly by interaction with 2-oxoglutarate and the PII regulatory protein. EMBO J 19:6041-50.

122. Liu, Y., and M. L. Kahn. 1995. ADP-ribosylation of Rhizobium meliloti glutamine synthetase III in vivo. J Biol Chem 270:1624-8.

123. Lowery, R. G., C. L. Chang, L. C. Davis, M. C. McKenna, P. J. Stephens, and P. W. Ludden. 1989. Substitution of histidine for arginine-101 of dinitrogenase reductase disrupts electron transfer to dinitrogenase. Biochemistry 28:1206-12.

124. Lowery, R. G., and P. W. Ludden. 1989. Effect of nucleotides on the activity of dinitrogenase reductase ADP-ribosyltransferase from Rhodospirillum rubrum. Biochemistry 28:4956-61.

125. Lowery, R. G., and P. W. Ludden. 1988. Purification and properties of dinitrogenase reductase ADP-ribosyltransferase from the photosynthetic bacterium Rhodospirillum rubrum. J Biol Chem 263:16714-9.

126. Lowery, R. G., L. L. Saari, and P. W. Ludden. 1986. Reversible regulation of the nitrogenase iron protein from Rhodospirillum rubrum by ADP-ribosylation in vitro. J Bacteriol 166:513-8.

127. Ludden, P. W., and R. H. Burris. 1976. Activating factor for the iron protein of nitrogenase from Rhodospirillum rubrum. Science 194:424-6.

128. Ludden, P. W., and R. H. Burris. 1979. Removal of an adenine-like molecule during activation of dinitrogenase reductase from Rhodospirillum rubrum. Proc Natl Acad Sci U S A 76:6201-5.

129. MacFarlane, S. A., and M. Merrick. 1985. The nucleotide sequence of the nitrogen regulation gene ntrB and the glnA-ntrBC intergenic region of Klebsiella pneumoniae. Nucleic Acids Res 13:7591-606.

130. Machado Benelli, E., M. Buck, I. Polikarpov, E. Maltempi de Souza, L. M. Cruz, and F. O. Pedrosa. 2002. Herbaspirillum seropedicae signal transduction protein PII is structurally similar to the enteric GlnK. Eur J Biochem 269:3296-303.

131. Maheswaran, M., and K. Forchhammer. 2003. Carbon-source-dependent nitrogen regulation in Escherichia coli is mediated

56

through glutamine-dependent GlnB signalling. Microbiology 149:2163-72.

132. Mao, X. J., Y. X. Huo, M. Buck, A. Kolb, and Y. P. Wang. 2007. Interplay between CRP-cAMP and PII-Ntr systems forms novel regulatory network between carbon metabolism and nitrogen assimilation in Escherichia coli. Nucleic Acids Res 35:1432-40.

133. Martinez-Argudo, I., R. Little, and R. Dixon. 2004. Role of the amino-terminal GAF domain of the NifA activator in controlling the response to the antiactivator protein NifL. Mol Microbiol 52:1731-44.

134. Martinez-Argudo, I., R. Little, N. Shearer, P. Johnson, and R. Dixon. 2004. The NifL-NifA System: a multidomain transcriptional regulatory complex that integrates environmental signals. J Bacteriol 186:601-10.

135. Meletzus, D., P. Rudnick, N. Doetsch, A. Green, and C. Kennedy. 1998. Characterization of the glnK-amtB operon of Azotobacter vinelandii. J Bacteriol 180:3260-4.

136. Merrick, M. J. 1992. Regulation of nitrogen fixation genes in free-living and symbiotic bacteria, p. 835-876. In G. Stacey and R. H. Burris (ed.), Biological nitrogen fixation. Chapman and Hall, New york.

137. Merrick, M. J., and R. A. Edwards. 1995. Nitrogen control in bacteria. Microbiol. Rev. 59:604-22.

138. Monod, J. 1947. The phenomenon of enzymatic adaptation. Growth 11:223-89.

139. Morett, E., and M. Buck. 1988. NifA-dependent in vivo protection demonstrates that the upstream activator sequence of nif promoters is a protein binding site. Proc Natl Acad Sci U S A 85:9401-5.

140. Morett, E., W. Cannon, and M. Buck. 1988. The DNA-binding domain of the transcriptional activator protein NifA resides in its carboxy terminus, recognises the upstream activator sequences of nif promoters and can be separated from the positive control function of NifA. Nucleic Acids Res 16:11469-88.

141. Moss, J., S. J. Stanley, and R. L. Levine. 1990. Inactivation of bacterial glutamine synthetase by ADP-ribosylation. J Biol Chem 265:21056-60.

142. Muro-Pastor, A. M., A. Valladares, E. Flores, and A. Herrero. 2002. Mutual dependence of the expression of the cell differentiation regulatory protein HetR and the global nitrogen regulator NtcA during heterocyst development. Mol Microbiol 44:1377-85.

143. Neilson, A. H., and S. Nordlund. 1975. Regulation of nitrogenase synthesis in intact cells of Rhodospirillum rubrum: inactivation of nitrogen fixation by ammonia, L-glutamine and L-asparagine. J Gen Microbiol 91:53-62.

144. Nichols, C. E., S. Sainsbury, N. S. Berrow, D. Alderton, N. J. Saunders, D. K. Stammers, and R. J. Owens. 2006. Structure of

57

the PII signal transduction protein of Neisseria meningitidis at 1.85 A resolution. Acta Crystallograph Sect F Struct Biol Cryst Commun 62:494-7.

145. Ninfa, A. J., P. Jiang, M. R. Atkinson, and J. A. Peliska. 2000. Integration of antagonistic signals in the regulation of nitrogen assimilation in Escherichia coli. Curr Top Cell Regul 36:31-75.

146. Nohaile, M., D. Kern, D. Wemmer, K. Stedman, and S. Kustu. 1997. Structural and functional analyses of activating amino acid substitutions in the receiver domain of NtrC: evidence for an activating surface. J Mol Biol 273:299-316.

147. Nordlund, S., U. Eriksson, and H. Baltscheffsky. 1977. Necessity of a membrane component for nitrogenase activity in Rhodospirillum rubrum. Biochim Biophys Acta 462:187-95.

148. Nordlund, S., U. Eriksson, and H. Baltscheffsky. 1978. Properties of the nitrogenase system from a photosynthetic bacterium, Rhodospirillum rubrum. Biochim Biophys Acta 504:248-54.

149. Nordlund, S., and P. W. Ludden. 2004. Post-translational regulation of nitrogenase in photosynthetic bacteria, p. 176-196. In W. Klipp, B. Masephol, J. R. Gallon, and W. E. Newton (ed.), Genetics and regulation of nitrogen fixation in free-living bacteria, vol. 2. Kluwer Academic Publishers, The Netherlands.

150. O'Connell, K. P., S. J. Raffel, B. J. Saville, and J. Handelsman. 1998. Mutants of Rhizobium tropici strain CIAT899 that do not induce chlorosis in plants. Microbiology 144 ( Pt 9):2607-17.

151. Paul, T. D., and P. W. Ludden. 1984. Adenine nucleotide levels in Rhodospirillum rubrum during switch-off of whole-cell nitrogenase activity. Biochem J 224:961-9.

152. Pellegrini, M., N. Gronbech-Jensen, J. A. Kelly, G. M. Pfluegl, and T. O. Yeates. 1997. Highly constrained multiple-copy refinement of protein crystal structures. Proteins 29:426-32.

153. Penyige, A., A. Kalmanczhelyi, A. Sipos, J. C. Ensign, and G. Barabas. 1994. Modification of glutamine synthetase in Streptomyces griseus by ADP-ribosylation and adenylylation. Biochem Biophys Res Commun 204:598-605.

154. Perlova, O., R. Nawroth, E. M. Zellermann, and D. Meletzus. 2002. Isolation and characterization of the glnD gene of Gluconacetobacter diazotrophicus, encoding a putative uridylyltransferase/uridylyl-removing enzyme. Gene 297:159-68.

155. Peshenko, I. V., and H. Shichi. 2001. Oxidation of active center cysteine of bovine 1-Cys peroxiredoxin to the cysteine sulfenic acid form by peroxide and peroxynitrite. Free Radic Biol Med 31:292-303.

156. Pioszak, A. A., P. Jiang, and A. J. Ninfa. 2000. The Escherichia coli PII signal transduction protein regulates the activities of the two-component system transmitter protein NRII by direct interaction with the kinase domain of the transmitter module. Biochemistry 39:13450-61.

58

157. Pioszak, A. A., and A. J. Ninfa. 2003. Genetic and biochemical analysis of phosphatase activity of Escherichia coli NRII (NtrB) and its regulation by the PII signal transduction protein. J Bacteriol 185:1299-315.

158. Pioszak, A. A., and A. J. Ninfa. 2004. Mutations altering the N-terminal receiver domain of NRI (NtrC) That prevent dephosphorylation by the NRII-PII complex in Escherichia coli. J Bacteriol 186:5730-40.

159. Pope, M. R., S. A. Murrell, and P. W. Ludden. 1985. Covalent modification of the iron protein of nitrogenase from Rhodospirillum rubrum by adenosine diphosphoribosylation of a specific arginine residue. Proc Natl Acad Sci U S A 82:3173-7.

160. Pope, M. R., S. A. Murrell, and P. W. Ludden. 1985. Purification and properties of the heat-released nucleotide-modifying group from the inactive iron protein of nitrogenase from Rhodospirillum rubrum. Biochemistry 24:2374-80.

161. Powell, B. S., D. L. Court, T. Inada, Y. Nakamura, V. Michotey, X. Cui, A. Reizer, M. H. Saier, Jr., and J. Reizer. 1995. Novel proteins of the phosphotransferase system encoded within the rpoN operon of Escherichia coli. Enzyme IIANtr affects growth on organic nitrogen and the conditional lethality of an erats mutant. J Biol Chem 270:4822-39.

162. Rabus, R., J. Reizer, I. Paulsen, and M. H. Saier, Jr. 1999. Enzyme I(Ntr) from Escherichia coli. A novel enzyme of the phosphoenolpyruvate-dependent phosphotransferase system exhibiting strict specificity for its phosphoryl acceptor, NPr. J Biol Chem 274:26185-91.

163. Read, R., C. A. Pashley, D. Smith, and T. Parish. 2007. The role of GlnD in ammonia assimilation in Mycobacterium tuberculosis. Tuberculosis (Edinb) 87:384-90.

164. Reitzer, L. 2003. Nitrogen assimilation and global regulation in Escherichia coli. Annu Rev Microbiol 57:155-76.

165. Reizer, J., A. Reizer, M. H. Saier, Jr., and G. R. Jacobson. 1992. A proposed link between nitrogen and carbon metabolism involving protein phosphorylation in bacteria. Protein Sci 1:722-6.

166. Reyes-Ramirez, F., R. Little, and R. Dixon. 2001. Role of Escherichia coli nitrogen regulatory genes in the nitrogen response of the Azotobacter vinelandii NifL-NifA complex. J Bacteriol 183:3076-82.

167. Rhee, S. G. 1984. 5'-Nucleotidyl-O-tyrosine bond in glutamine synthetase. Methods Enzymol 107:183-200.

168. Rhee, S. G., W. G. Bang, J. H. Koo, K. H. Min, and S. C. Park. 1988. Regulation of glutamine synthetase activity and its biosynthesis in Escherichia coli: Mediation by three cycles of covalent modification. Springer-Verlag, Berlin.

169. Rhee, S. G., H. Z. Chae, and K. Kim. 2005. Peroxiredoxins: a historical overview and speculative preview of novel mechanisms

59

and emerging concepts in cell signaling. Free Radic Biol Med 38:1543-52.

170. Rhee, S. G., P. B. Chock, and E. R. Stadtman. 1989. Regulation of Escherichia coli glutamine synthetase. Adv Enzymol Relat Areas Mol Biol 62:37-92.

171. Ronson, C. W., B. T. Nixon, and F. M. Ausubel. 1987. Conserved domains in bacterial regulatory proteins that respond to environmental stimuli. Cell 49:579-81.

172. Rudnick, P. A., T. Arcondeguy, C. K. Kennedy, and D. Kahn. 2001. glnD and mviN are genes of an essential operon in Sinorhizobium meliloti. J Bacteriol 183:2682-85.

173. Sakai, H., H. Wang, C. Takemoto-Hori, T. Kaminishi, H. Yamaguchi, Y. Kamewari, T. Terada, S. Kuramitsu, M. Shirouzu, and S. Yokoyama. 2005. Crystal structures of the signal transducing protein GlnK from Thermus thermophilus HB8. J Struct Biol 149:99-110.

174. Schick, H. J. 1971. Substrate and light dependent fixation of molecular nitrogen in Rhodospirillum rubrum. Arch Mikrobiol 75:89-101.

175. Schluter, A., M. Nohlen, M. Kramer, R. Defez, and U. B. Priefer. 2000. The Rhizobium leguminosarum bv. viciae glnD gene, encoding a uridylyltransferase/uridylyl-removing enzyme, is expressed in the root nodule but is not essential for nitrogen fixation. Microbiology 146 2987-96.

176. Schmitz, R. A. 1997. NifL of Klebsiella pneumoniae carries an N-terminally bound FAD cofactor, which is not directly required for the inhibitory function of NifL. FEMS Microbiol Lett 157:313-8.

177. Schutt, H., and H. Holzer. 1972. Biological function of the ammonia-induced inactivation of glutamine synthetase in Escherichia coli. Eur J Biochem 26:68-72.

178. Schwarzenbacher, R., F. von Delft, P. Abdubek, E. Ambing, T. Biorac, L. S. Brinen, J. M. Canaves, J. Cambell, H. J. Chiu, X. Dai, A. M. Deacon, M. DiDonato, M. A. Elsliger, S. Eshagi, R. Floyd, A. Godzik, C. Grittini, S. K. Grzechnik, E. Hampton, L. Jaroszewski, C. Karlak, H. E. Klock, E. Koesema, J. S. Kovarik, A. Kreusch, P. Kuhn, S. A. Lesley, I. Levin, D. McMullan, T. M. McPhillips, M. D. Miller, A. Morse, K. Moy, J. Ouyang, R. Page, K. Quijano, A. Robb, G. Spraggon, R. C. Stevens, H. van den Bedem, J. Velasquez, J. Vincent, X. Wang, B. West, G. Wolf, Q. Xu, K. O. Hodgson, J. Wooley, and I. A. Wilson. 2004. Crystal structure of a putative PII-like signaling protein (TM0021) from Thermotoga maritima at 2.5 A resolution. Proteins 54:810-3.

179. Senior, P. J. 1975. Regulation of nitrogen metabolism in Escherichia coli and Klebsiella aerogenes: studies with the continuous-culture technique. J Bacteriol 123:407-18.

180. Seo, M. S., S. W. Kang, K. Kim, I. C. Baines, T. H. Lee, and S. G. Rhee. 2000. Identification of a new type of mammalian

60

peroxiredoxin that forms an intramolecular disulfide as a reaction intermediate. J Biol Chem 275:20346-54.

181. Shah, V. K., and W. J. Brill. 1973. Nitrogenase. IV. Simple method of purification to homogeneity of nitrogenase components from Azotobacter vinelandii. Biochim Biophys Acta 305:445-54.

182. Shapiro, B. M. 1969. The glutamine synthetase deadenylylating enzyme system from Escherichia coli. Resolution into two components, specific nucleotide stimulation, and cofactor requirements. Biochemistry 8:659-70.

183. Silman, N. J., N. G. Carr, and N. H. Mann. 1995. ADP-ribosylation of glutamine synthetase in the cyanobacterium Synechocystis sp. strain PCC 6803. J Bacteriol 177:3527-33.

184. Soderback, E., F. Reyes-Ramirez, T. Eydmann, S. Austin, S. Hill, and R. Dixon. 1998. The redox- and fixed nitrogen-responsive regulatory protein NIFL from Azotobacter vinelandii comprises discrete flavin and nucleotide-binding domains. Mol Microbiol 28:179-92.

185. Son, H. S., and S. G. Rhee. 1987. Cascade control of Escherichia coli glutamine synthetase. Purification and properties of PII protein and nucleotide sequence of its structural gene. J Biol Chem 262:8690-5.

186. Song, Y., D. Peisach, A. A. Pioszak, Z. Xu, and A. J. Ninfa. 2004. Crystal structure of the C-terminal domain of the two-component system transmitter protein nitrogen regulator II (NRII; NtrB), regulator of nitrogen assimilation in Escherichia coli. Biochemistry 43:6670-8.

187. Souza, E. M., F. O. Pedrosa, M. Drummond, L. U. Rigo, and M. G. Yates. 1999. Control of Herbaspirillum seropedicae NifA activity by ammonium ions and oxygen. J Bacteriol 181:681-4.

188. Stadtman, E. R. 2001. The story of glutamine synthetase regulation. J Biol Chem 276:44357-64.

189. Strosser, J., A. Ludke, S. Schaffer, R. Kramer, and A. Burkovski. 2004. Regulation of GlnK activity: modification, membrane sequestration and proteolysis as regulatory principles in the network of nitrogen control in Corynebacterium glutamicum. Mol Microbiol 54:132-47.

190. Van Dommelen, A., V. Keijers, E. Somers, and J. Vanderleyden. 2002. Cloning and characterisation of the Azospirillum brasilense glnD gene and analysis of a glnD mutant. Mol Genet Genomics 266:813-20.

191. van Heeswijk, W. C., S. Hoving, D. Molenaar, B. Stegeman, D. Kahn, and H. V. Westerhoff. 1996. An alternative PII protein in the regulation of glutamine synthetase in Escherichia coli. Mol Microbiol 21:133-46.

192. van Heeswijk, W. C., M. Rabenberg, H. V. Westerhoff, and D. Kahn. 1993. The genes of the glutamine synthetase adenylylation cascade are not regulated by nitrogen in Escherichia coli. Mol

61

Microbiol 9:443-57. 193. van Heeswijk, W. C., B. Stegeman, S. Hoving, D. Molenaar, D.

Kahn, and H. V. Westerhoff. 1995. An additional PII in Escherichia coli: a new regulatory protein in the glutamine synthetase cascade. FEMS Microbiol Lett 132:153-7.

194. van Heeswijk, W. C., D. Wen, P. Clancy, R. Jaggi, D. L. Ollis, H. V. Westerhoff, and S. G. Vasudevan. 2000. The Escherichia coli signal transducers PII (GlnB) and GlnK form heterotrimers in vivo: fine tuning the nitrogen signal cascade. Proc Natl Acad Sci U S A 97:3942-7.

195. Wang, G., A. A. Olczak, J. P. Walton, and R. J. Maier. 2005. Contribution of the Helicobacter pylori thiol peroxidase bacterioferritin comigratory protein to oxidative stress resistance and host colonization. Infect Immun 73:378-84.

196. Vazquez-Bermudez, M. F., A. Herrero, and E. Flores. 2000. Uptake of 2-oxoglutarate in Synechococcus strains transformed with the Escherichia coli kgtP gene. J Bacteriol 182:211-5.

197. Willis, R. C., K. K. Iwata, and C. E. Furlong. 1975. Regulation of Glutamine Transport in Escherichia coli. J Bacteriol 122:1032-7.

198. Woehle, D. L. 1992. PhD thesis. Dept of Biochemistry. University of Wisconsin, Madison.

199. Woehle, D. L., B. A. Lueddecke, and P. W. Ludden. 1990. ATP-dependent and NAD-dependent modification of glutamine synthetase from Rhodospirillum rubrum in vitro. J Biol Chem 265:13741-9.

200. Wolfe, D. M., Y. Zhang, and G. P. Roberts. 2007. Specificity and Regulation of Interaction between the PII and AmtB1 Proteins in Rhodospirillum rubrum. J Bacteriol. doi:10.1128/JB.00759-07.

201. Wood, Z. A., L. B. Poole, R. R. Hantgan, and P. A. Karplus. 2002. Dimers to doughnuts: redox-sensitive oligomerization of 2-cysteine peroxiredoxins. Biochemistry 41:5493-504.

202. Wood, Z. A., E. Schroder, J. Robin Harris, and L. B. Poole. 2003. Structure, mechanism and regulation of peroxiredoxins. Trends Biochem Sci 28:32-40.

203. Woolfolk, C. A., and E. R. Stadtman. 1964. Cumulative feedback inhibition in the multiple end-product regulation of glutamine synthetase activity in Escherichia coli. Biochem Biophys Res Commun 17:313-9.

204. Woolfolk, C. A., and E. R. Stadtman. 1967. Regulation of glutamine synthetase. III. Cumulative feedback inhibition of glutamine synthetase from Escherichia coli. Arch Biochem Biophys 118:736-55.

205. Wootton, J. C., and M. H. Drummond. 1989. The Q-linker: a class of interdomain sequences found in bacterial multidomain regulatory proteins. Protein Eng 2:535-43.

206. Wulff, K., D. Mecke, and H. Holzer. 1967. Mechanism of the enzymatic inactivation of glutamine synthetase from E. coli.

62

Biochem Biophys Res Commun 28:740-5. 207. Xu, Y., P. D. Carr, P. Clancy, M. Garcia-Dominguez, K.

Forchhammer, F. Florencio, S. G. Vasudevan, N. Tandeau de Marsac, and D. L. Ollis. 2003. The structures of the PII proteins from the cyanobacteria Synechococcus sp. PCC 7942 and Synechocystis sp. PCC 6803. Acta Crystallogr D Biol Crystallogr 59:2183-90.

208. Xu, Y., P. D. Carr, T. Huber, S. G. Vasudevan, and D. L. Ollis. 2001. The structure of the PII-ATP complex. Eur J Biochem 268:2028-37.

209. Xu, Y., E. Cheah, P. D. Carr, W. C. van Heeswijk, H. V. Westerhoff, S. G. Vasudevan, and D. L. Ollis. 1998. GlnK, a PII-homologue: structure reveals ATP binding site and indicates how the T-loops may be involved in molecular recognition. J Mol Biol 282:149-65.

210. Yamashita, M. M., R. J. Almassy, C. A. Janson, D. Cascio, and D. Eisenberg. 1989. Refined atomic model of glutamine synthetase at 3.5 A resolution. J Biol Chem 264:17681-90.

211. Yildiz, O., C. Kalthoff, S. Raunser, and W. Kuhlbrandt. 2007. Structure of GlnK1 with bound effectors indicates regulatory mechanism for ammonia uptake. EMBO J 26:589-99.

212. Zhang, Y., A. D. Cummings, R. H. Burris, P. W. Ludden, and G. P. Roberts. 1995. Effect of an ntrBC mutation on the posttranslational regulation of nitrogenase activity in Rhodospirillum rubrum. J Bacteriol 177:5322-6.

213. Zhang, Y., E. L. Pohlmann, M. C. Conrad, and G. P. Roberts. 2006. The poor growth of Rhodospirillum rubrum mutants lacking PII proteins is due to an excess of glutamine synthetase activity. Mol Microbiol 61:497-510.

214. Zhang, Y., E. L. Pohlmann, P. W. Ludden, and G. P. Roberts. 2001. Functional characterization of three GlnB homologs in the photosynthetic bacterium Rhodospirillum rubrum: roles in sensing ammonium and energy status. J Bacteriol 183:6159-68.

215. Zhang, Y., E. L. Pohlmann, P. W. Ludden, and G. P. Roberts. 2000. Mutagenesis and functional characterization of the glnB, glnA, and nifA genes from the photosynthetic bacterium Rhodospirillum rubrum. J Bacteriol 182:983-92.

216. Zhang, Y., E. L. Pohlmann, and G. P. Roberts. 2005. GlnD is essential for NifA activation, NtrB/NtrC-regulated gene expression, and posttranslational regulation of nitrogenase activity in the photosynthetic, nitrogen-fixing bacterium Rhodospirillum rubrum. J Bacteriol 187:1254-65.

217. Zhang, Y., E. L. Pohlmann, and G. P. Roberts. 2004. Identification of critical residues in GlnB for its activation of NifA activity in the photosynthetic bacterium Rhodospirillum rubrum. Proc Natl Acad Sci U S A 101:2782-7.

218. Zhang, Y., D. M. Wolfe, E. L. Pohlmann, M. C. Conrad, and G.

63

P. Roberts. 2006. Effect of AmtB homologues on the post-translational regulation of nitrogenase activity in response to ammonium and energy signals in Rhodospirillum rubrum. Microbiology 152:2075-89.

219. Zheng, L., D. Kostrewa, S. Berneche, F. K. Winkler, and X. D. Li. 2004. The mechanism of ammonia transport based on the crystal structure of AmtB of Escherichia coli. Proc Natl Acad Sci U S A 101:17090-5.

220. Zhu, Y., M. C. Conrad, Y. Zhang, and G. P. Roberts. 2006. Identification of Rhodospirillum rubrum GlnB variants that are altered in their ability to interact with different targets in response to nitrogen status signals. J Bacteriol 188:1866-74.

221. Zumft, W. G., and F. Castillo. 1978. Regulatory properties of the nitrogenase from Rhodopseudomonas palustris. Arch Microbiol 117:53-60.