Introduction

Nitrogen is one of the key elements in life and it is essentially required in form of ammonium for biomolecules such as proteins or nucleic acids. Two major pathways of ammonium assimilation in bacteria and archaea are known. Under nitrogen (N) sufficiency glutamate dehydrogenase (GDH) is active and generates glutamate from oxoglutarate and ammonium (reviewed in van Heeswijk et al., 2013). However, under N limitation low ammonium concentrations lead to inactive GDH as a result of its low ammonium affinity, whereas the expression of glutamine synthetase (GS) is strongly induced in response to N limitation (Bolay et al., 2018; Gunka and Commichau, 2012; Stadtman, 2001). Consequently, under low ammonium conditions GS together with glutamate synthase (GOGAT) are responsible for ammonium assimilation via the GS/GOGAT pathway, one of the major intersections in central carbon and N metabolism. Accordingly, GS present across all domains of life plays a central role in cellular N assimilation under low N availability. The enzyme, its structure and regulation has been investigated in detail in different organisms for more than half a century (e.g. Dos Santos Moreira et al., 2019; Stadtman, 2001; Woolfolk and Stadtman, 1967).

Most of the GS are grouped into three major classes based on their monomeric size and oligomerization properties (overview in Dos Santos Moreira et al., 2019). GSI and GSIII, both found in bacteria and archaea mostly form dodecamers, whereas GSII found in Eukaryotes form decamers of smaller subunits (Dos Santos Moreira et al., 2019; He et al., 2009; Valentine et al., 1968; van Rooyen et al., 2011). The GSI class can be further grouped into Iα-type GS and Iβ-type GS based on their amino acid sequence and respective molecular mechanisms of activity regulation. Iß-type GS contain a conserved adenylylation site (Tyr397 residue near the active site) that allows for covalent modification of Iβ-type GS and leads to inactivation of the enzyme (Brown et al., 1994; Magasanik, 1993; Shapiro and Stadtman, 1970), whereas Iα-type GS are not covalently modified and mainly show feedback inhibition by end products of the glutamine metabolism including glutamine (Fisher, 1999; Gunka and Commichau, 2012).

GS regulation on transcriptional level

Since in contrast to GDH, GS catalyzed generation of glutamine requires ATP, most organisms strictly regulate the expression of GS in response to the nitrogen availability on the transcriptional level. In gram negative bacteria mainly transcriptional activation of the coding gene (glnA) under low nitrogen availability occurs via a transcriptional activator (e.g. NtrC in Escherichia coli (Jiang et al., 1998)). For several gram positive bacteria however, the mechanism of regulation is a de-repression of glnA transcription under N limitation, which has also been shown for methanoarchaea (Cohen-Kupiec et al., 1999; Fedorova et al., 2013; Fisher, 1999; Fisher and Wray, 2008; Hauf et al., 2016; Weidenbach et al., 2010, 2008). Whereas in gram positives the signal perception is complex and often also involves protein interactions of GS with transcriptional regulators (reviewed in Gunka and Commichau, 2012), signal perception and transduction in methanoarchaea occurs directly via the small effector molecule 2-oxoglutarate (2-OG), which increases under N limitation. It has been shown that binding of 2-OG to the global N repressor protein NrpR significantly changes the repressor conformation resulting in dissociation from its respective operator (Lie et al., 2007; Weidenbach et al., 2010; Wisedchaisri et al., 2010). In addition to expression regulation the activity of GS is also strictly regulated in all organisms in response to changing N availabilities, however the underlying molecular mechanism(s) of inhibition significantly differ for the various GS classes and in various organisms (Reitzer, 2003)

Regulation of GS activity: highly diverse and often complex in various organisms

An extensive repertoire of cellular control mechanisms regulating GS activity in response to N availability has been observed in different organisms. Inhibitory mechanisms in response to an N upshift range from feedback inhibition by e.g. glutamine or other end products of the glutamine metabolism (e.g. E. coli (Stadtman, 2004), Bacillus subtilis (Deuel et al., 1970), yeast (Legrain et al., 1982)), proteolytic degradation (yeast, (Legrain et al., 1982), covalent modification by adenylylation of the 1ß-type GS subunits (e.g. enterobacteriaceae), thiol-based GS regulation (e.g. in soybean nodules (Masalkar and Roberts, 2015)), inhibition by regulatory proteins (e.g. in gram positive bacteria (Travis et al., 2022a)), inhibition by interactions with small proteins (e.g. inhibitory factors in Cyanobacteria (García-Domínguez et al., 1999; Klähn et al., 2018, 2015)), to directly effecting the activity through the presence or absence of the small metabolite 2-OG, which has been shown for the first time for Methanosarcina mazei (Ehlers et al., 2005). Moreover, often complex regulations for GS activity including several of the different regulatory mechanisms are reported for one organism. For example, yeast GS (ScGS) is regulated via feedback inhibition by glutamine and is susceptible to proteolytic degradation under N starvation. It was also found to assemble into nanotubes (He et al., 2009) and under advanced cellular starvation into inactive filaments (Petrovska et al., 2014). In E. coli, the activity of the Iß-type GS (EcGS) is controlled by cumulative feedback inhibition and covalent modification (reviewed in Reitzer, 2003). It has been shown that each of the 12 subunits can be modified by adenylylation (Tyr397) resulting in an inactivation of the respective subunit (Stadtman, 1990). Moreover, the adenlylylation of single subunits in the complex in addition makes the other subunits more susceptible to cumulative feedback inhibition by various substances (Stadtman, 1990). These substances either bind the glutamine-binding pocket or have an allosteric binding site (Liaw et al., 1993; Woolfolk and Stadtman, 1967). The dodecameric structure of EcGS has been known for a long time (Almassy et al., 1986; Yamashita et al., 1989). However, when artificially exposed to divalent cations (Mn²⁺, Co²⁺) it randomly aggregates and produces long hexagonal tubes (paracrystalline aggregates) (Valentine et al., 1968). The detailed structural information on the mechanisms of this GS-filament formation to an inactive form of EcGS, often associated with stress responses, as well as the reversion into individual active dodecamers has only recently been described by cryo-electron microscopy (cryo-EM) analysis (Huang et al., 2022). The B. subtilis GS has been shown to be feedback regulated. In addition, binding of the transcriptional repressor GlnR to the feedback inhibited complex not only activates the transcription repression function of GlnR (Fisher and Wray, 2008) but also stabilizes the inactive GS conformation potentially changing from a dodecamer into a tetradecameric structure (Travis et al., 2022a).

In M. mazei, a mesophilic methanoarchaeon, which is able to fix N₂, regulation of the central N metabolisms has been studied extensively on the transcriptional and post-transcriptional level (Jäger et al., 2009; Prasse and Schmitz, 2018; Veit et al., 2005). A central role of 2-OG for the perception of changes in N availabilities has been proposed as has been demonstrated for cyanobacteria (Forchhammer, 1999; Herrero et al., 2001). The activity of M. mazei GS encoded by glnA₁, is regulated by several different mechanisms. GS /GlnA₁) is not covalently modified in response to N availability and thus represents a Iß-type-GS (Ehlers et al., 2005). It has been demonstrated to get directly activated under N starvation due to the high intracellular concentrations of the metabolite 2-OG which directly induces GlnA₁ activity (Ehlers et al., 2005). 2-OG represents the internal signal for N limitation since under N starvation the internal 2-OG level significantly increases due to missing consumption by GDH (M. mazei contains the oxidative TCA part, anabolic). The increased cellular 2-OG concentration has been shown to be directly perceived by the GlnA₁ most likely by direct binding resulting in strong activation (Ehlers et al., 2005). Besides, we showed first evidence that in addition two small proteins interact with M. mazei GlnA₁, the PII-like protein GlnK₁ and small protein sP26 comprising 23 amino acids (Ehlers et al., 2005; Gutt et al., 2021). The presence and potential interaction of both small proteins show small effects on the GlnA₁ activity, however compared to the strong 2-OG stimulation only to a very low extend, which might be neglectable and due to the indirect GS activity assay. Moreover, based on initial complex formation analysis using size exclusion chromatography (SEC) and pull-down approaches first indications were obtained that in the absence of 2-OG the GlnA₁/GlnK₁ complexes are more stable than in the presence of high 2-OG. This led to the conclusion that due to the shift to N sufficiency after a period of N limitation, GlnA₁ activity is reduced due to the lower 2-OG concentration but also due to the inhibitory protein interaction with GlnK₁ (Ehlers et al., 2005). Very recently the first structural analysis of M. mazei GlnA₁ was reported, showing first GS complexes with GlnK₁ (Schumacher et al. 23). Based on their findings Schumacher et al. propose a regulation of GlnA₁ activity by oligomeric modulation, with GlnK₁ stabilizing the dodecameric structure and the formation of GlnA₁ active sites. Since this proposed model is entirely missing the effects of 2-OG on GlnA₁ activity, we here aimed to study the obtained effects regulating M. mazei GlnA₁ activity in more detail by evaluating oligomerization and complex formation between GlnA₁, GlnK₁ and sP26 in dependence of 2-OG employing mass photometry (MP) allowing molecular weight distribution of single complexes in solution and by high resolution cryo-EM.

Results

2-OG is responsible for GlnA₁-dodecamer formation in M. mazei

The strep-tagged purified GlnA₁ was analyzed by SEC in the presence of 12.5 mM 2-OG demonstrating that GS is exclusively present in a dodecameric structure, no other oligomers were detectable (suppl. Fig. S1). To investigate the effects of 2-OG on M. mazei GlnA₁ in more detail, we employed MP, a method that allows to measure the molecular weight distribution in solution. Strep-tagged purified GlnA₁ (after SEC) was dialyzed into a 2-OG free HEPES buffer (see Materials and Methods) and subsequently analyzed by MP, demonstrating that in the presence of low 2-OG concentrations (0.1 mM) all of the M. mazei GlnA₁ was exclusively present as monomers/dimers with no higher molecular weight complexes present. After addition of 12.5 mM 2-OG, the size distribution shifted towards a higher molecular weight complex of 630-700 kDa (calculated based on the measured dimer-size in each measurement; expected molecular weight of dodecamer: 634 kDa) (Fig. 1A, B). This molecular weight corresponds to a fully assembled dodecamer species, the same oligomeric structure that is adapted in GS from other prokaryotes. Using 2-OG concentrations varying between 0.1 and 12.5 mM, complex analysis showed that up to 62 % of all particles were assembled in a dodecamer. This further allowed to determine the binding affinity of GlnA₁ to 2-OG to be K_D = 0.75 ± 0.01 mM 2-OG (based on two biological replicates, calculated with the percentage of dodecamer) as described in Materials and Methods, and verified that no other intermediate oligomeric complexes were detectable during dodecameric assembly (Fig. 1A, C, suppl. Fig. S2A, B). Activity measurements of Strep-GlnA₁ in the presence of increasing 2-OG concentrations showed a strong increase of the activity from 0.0 U/mg in the absence of 2-OG up to 7.8 ± 1.7 U/mg in the presence of 12.5 mM 2-OG (six independent protein purifications, Fig. 1D). Thus, we conclude that 2-OG acts as a trigger for dodecameric assembly of M. mazei GlnA₁, setting it apart from other bacterial and eukaryotic enzyme variants. Moreover, most likely in addition to the dodecameric assembly, 2-OG is additionally required for a further 2-OG induced conformational switch of the active site, since saturated GlnA₁ activities are not reached in the presence of 5 mM 2-OG, when most of the GlnA₁ is in a dodecameric structure. For full activity, the presence of 12.5 mM 2-OG is required.

Figure 1:

GlnK₁ has no detectable effects on GS dodecamer assembly or activity under the tested conditions

Previous studies have shown protein interactions between M. mazei GlnA₁ and GlnK₁ as well as GlnK₁ induced effects on GlnA₁ activity. Consequently, we next tested the effects of GlnK₁ presence on the GlnA₁ oligomerization in the presence of 2-OG. Performing the MP analysis under the tested conditions as before but in the presence of purified GlnK₁ demonstrated that (i) in the absence of 2-OG varying ratios between GlnA₁ and GlnK₁ (20:1, 2:1, 2:10 based on monomers) did not result in any dodecamer assembly of GlnA₁ (Fig. 2A, B), (ii) no difference in the GlnA₁ dodecameric assembly in the presence of 2-OG was obtained in the presence of purified GlnK₁ (2:1), (iii) nor was binding of GlnK₁ to GlnA₁ detected by a respective increase in the mass of the higher oligomeric complex (Fig. 2B, C). Moreover, the presence of GlnK₁ (2:1) neither had an influence on the 2-OG affinity (K_D (-GlnK₁) = 1.06 mM 2-OG; K_D (+ GlnK₁) = 1.02 mM 2-OG, K_D calculated based on the dodecamer/dimer ratio), nor in any ratio on the specific activity of GlnA₁ (Fig. 2 D, C: exemplarily showing 2:1; suppl. Fig. S2C, D). Consequently, we conclude that under the conditions tested using purified proteins, GlnA₁ dodecamer assembly occurs independently of GlnK₁ and no binding of GlnK₁ to the dodecameric GlnA₁ occurs. However, we cannot exclude that cellular components/metabolites not present in these experiments are crucial for a GlnA₁-GlnK₁ interaction.

Figure 2:

Structural basis of oligomer formation by 2-OG

To now unravel the structural mechanism underlying M. mazei GlnA₁ activation by 2-OG, we employed cryo-EM and single-particle analysis. Treating freshly purified Strep-GlnA₁ with 12.5 mM 2-OG, effectively shifted the equilibrium towards fully assembled homo-oligomers as depicted in the MP experiments. In the micrographs, fully assembled ring-shaped particles are visible. However, initial attempts to obtain a 3D reconstruction were hindered by the pronounced preferred orientation of particles within the ice, a challenge which has been overcome by introducing low concentrations of CHAPSO (0.7 mM). In our final dataset, all particles exhibited well-distributed oligomers in diverse orientations. Leveraging this dataset, we aligned the particles to a 2.39 Å resolution structure, revealing well resolved side chains that facilitated seamless model building (Fig. 3, suppl. Fig. S3, suppl. Tab. S2). Consequently, we achieved a structure demonstrating excellent geometry and density fitting.

Figure 3:

The detailed structural analysis uncovered that GlnA₁ assembles into a dodecamer characterized by stacked hexamer rings. A single GlnA₁ protomer is composed of 15 β-strands and 15 α-helices and is split in into a larger C-domain and an N-domain by helix α3. The dodecameric arrangement is achieved through two distinct interfaces, the hexamer interfaces and inter-hexamer interfaces.

Hexamer interfaces are situated between subunits within each ring, while inter-hexamer interfaces occur between subunits derived from adjacent rings (Fig. 4A, B, C). The structures are highly similar to Gram-positive bacterial GS structures (PDB: 4lnn, Murray et al., 2013),with root mean squared deviations (rmsds) of 0.5–1.0 Å.

Figure 4:

A closer inspection of the density reveals the density for the bound 2-OG at an allosteric site localized at the interface between two GlnA₁ protomers in vicinity of the GlnA₁ catalytic site (Fig. 4B, D). Several residues are contributing to its binding. R172’ and S189’ coordinate the γ-Carboxy-group. Additionally, two tightly bound water molecules are detectable in the binding site. One is interacting with the γ-Carboxy group while being stabilized by another water that is coordinated by S38 and R26. Latter arginine is coordinating the α-Keto-group and, together with R87 and R173’, the α-Carboxy group of 2-OG (Fig. 4B). Notably, F24 stabilizes the 2-OG via stacking with its phenyl ring. This binding contribution from two GlnA₁ protomers at the intersubunit junction enhances activation by boosting readiness and the rate of full complex assembly. It operates akin to molecular glue that facilitate the observed cooperative assembly.

A comparison with the substrate-bound GlnA₁ structure (PDB: 8tfk, Schumacher et al. 2023) revealed that the catalytically important residues in M. mazei are the aspartic acid (D57) that abstracts the proton from ammonium and the catalytic glutamic acid, Glu307. The active site of M. mazei GlnA₁ is formed at the interface between two subunits in the hexamer and formed by five key catalytic elements surrounding the active site: the E flap (residues 303–310), the Y loop (residues 369–377), the N loop (residues 235–247), the Y* loop (residues 152–161) and the D50’ loop (residues 56-71). The latter one is the only one that originates from adjacent neighboring protomer (Fig 5C, E).

Figure 5:

Superposition of our structure with the apo-M. mazei X-ray structure (PDB: 8tfb, Schumacher et al., 2023) reveals that 2-OG binding also triggers further movements that lead to structural changes in the substrate binding pocket (Fig. 5A, B, D). R87’ and its loop undergo a dramatic flip to coordinate 2-OG and D170 of helix α3 (residues 167-181) (Fig. 5A). This, combined with the action of other coordinating residues, initiates a motion that is propagated through the entire protein. Notably, helix α3 shifts forward, causing F184 to flip over and facilitate a T-shaped aromatic interaction with F202. The resulting pull on F202 causes F204 to flip, allowing π-stacking with the purine moiety of ATP (Fig. 5B). This series of structural changes primes the active site for ATP binding by already adopting the side chain conformations that are observed in analogue (Met-Sox-P-ADP)-bound structure (transition state) (PDB: 8tfk, Schumacher et al., 2023), thus facilitating nucleotide binding (Fig. 5C, E).

Additionally, the D50’ loop adopts a position similar to the transition state in a catalytic competent conformation. This involved a remodeling of the loop, leading to the positioning of key catalytic residues in a catalytic competent configuration. Compared to the apo structure (Schumacher et al., 2023), R66 flips out of the catalytic pocket, now accommodating R319 which participates in phosphoryl transfer catalysis (Fig. 5D). In addition, Asp 57’ moves deeper into the binding site, facilitating the proton abstraction of NH ⁺ and preparing for its attack on the phosphorylated glutamate. Similar to the ATP/ADP binding site, these catalytic elements are primed to ideally stabilize the tetrahedral transition state. This is illustrated by the superposition of the inhibitor-bound, transition-state locked structure (Schumacher et al., 2023) (Fig. 5C, E).

Feedback inhibition by glutamine does not affect the dodecameric M. mazei GlnA₁ structure

For bacteria it is known, that GS can be feedback inhibited. Very recently, the first feedback inhibition of an archaeal GS by glutamine has been reported for Methermicoccus shengliensis GS (Müller et al., 2023). The specific arginine residue identified to be relevant for the feedback inhibition is R66. Consequently, we generated the respective M. mazei GlnA₁ mutant protein changing the conserved arginine to alanine (R66A) (see also Fig. 5D, E) and compared the purified strep-tagged mutant protein with the wildtype (wt) protein. In the presence of 12.5 mM 2-OG, the mutant protein showed the same specific activity as obtained for the wt. However, when supplementing 5 mM glutamine, exclusively the wt was strongly feedback inhibited, whereas the R66A mutant protein was not significantly affected (Fig. 6A). In B. subtilis, R62 is responsible for feedback inhibition. The superposition of the apo-BsGS structure (PDB: 4lnn, Murray et al., 2013) with our 2-OG-bound GlnA₁ reveals a similar positioning of the respective M. mazei R66 (Fig. 6B) indicating a similar mechanism. Moreover, we can rule out an effect on the oligomeric structure of GlnA₁ by MP analysis, clearly showing that glutamine does not induce disassembly of the dodecameric wt GlnA₁ (Fig. 6C). Instead, this effect can be explained with the role of R66 being an important residue to bind to glutamine in the product state of the enzyme.

Figure 6:

Discussion

2-OG is crucially required for M. mazei GS assembly to an active dodecamer and induces the conformational state towards an active open state

In M. mazei increased 2-OG concentrations act as central N starvation signal (Ehlers et al., 2005). Here we demonstrated the importance of 2-OG as the major regulator of M. mazei GlnA₁ activity by using independent methods, MP and cryo-EM, to detect and structurally characterize single complexes with high resolution and quantify the different oligomeric complexes. We have found mono- and dimeric GlnA₁ (apo GlnA₁) to be inactive and crucially require 2-OG to form an active dodecameric complex. Moreover, this dodecameric conformation is the only active state of GlnA₁. In the first step, 2-OG assembles the dodecamer by binding at the interface of two subunits (Fig. 4B) and functions as a molecular glue between neighbouring subunits. The assembly upon 2-OG addition observed using MP appears to be cooperative, fast and without any detectable intermediate states (Fig. 1B, C). Only immediately after thawing a frozen purified GlnA₁ preparation and in case that no additional SEC was performed prior to MP analysis, samples showed additional octameric complexes in MP with low abundancy (suppl. Fig. S4). However, octameric complexes were never observed in cryo-EM or detected by SEC analysis of frozen purified GlnA₁ samples. Consequently, octamers are most likely broken or disassembled GlnA₁-dodecamers or dead-ends in assembly with no physiological function, rather than an incomplete dodecamer during assembly. Thus, our findings are contrary to the assembly model proposed by Schuhmacher et al. (Schumacher et al., 2023).

As a second step of activation, the allosteric binding of 2-OG causes a series of conformational changes in GlnA₁ protomers, which prime the active site for the transition state and hence catalysis of the enzyme. This conformational change of the ATP-binding pocket of the dodecameric GlnA₁ upon 2-OG binding goes hand in hand with the observed increased activity at higher 2-OG concentrations (Fig. 1). Comparing our 2-OG-bound GlnA₁ dodecameric structure and the dodecameric M. mazei GlnA₁ transition state (PDB: 8tfk) and apo structures (PDB: 8ftb) reported by Schumacher et al. (Schumacher et al., 2023), clearly demonstrates that 2-OG transfers GlnA₁ into its open transition state conformation (Fig. 5). The conformation of our 2-OG-bound dodecamer resembled the transition state conformation (ADP-Met-Sox-bound complex) reported by Schumacher et al., even though in our case no ATP was added (Fig. 5E). A reconfiguration of the active site upon 2-OG-binding has also been reported for GS in Methanothermococcus thermolithotrophicus (Müller et al., 2024). In this report, which does not delineate dodecamer assembly at all, it was demonstrated that binding of 2-OG in one protomer-protomer interface of a dodecameric GS causes a cooperative domino effect in the hexameric ring of M. thermolithotrophicus GS (Müller et al., 2024). A 2-OG bound protomer undergoes a conformational change and thereby induces the same shift in its neighbouring protomer (Müller et al., 2024). This is comparable to our observed cooperativity of M. mazei dodecamer assembly at low 2-OG concentrations (K_D = 0.75 mM, percentage of dodecamer). On the other hand, M. mazei GlnA₁ reaches maximal activity only at much higher 2-OG concentrations and likely requires a fully 2-OG-occupied dodecamer for maximal activity. The here obtained high activities by 2-OG saturation (up to 9 U/mg) in comparison with previously described M. mazei GlnA₁ activities in the absence of 2-OG in the significantly lower range (mU/mg) (Gutt et al., 2021; Schumacher et al., 2023) support our conclusion that 2-OG is substantial for the GlnA₁ active state.

GlnA₁ activity is further regulated by feedback inhibition, small proteins and possibly filament formation

M. mazei GlnA₁ belongs to the group of Iα-type GS, which are known to be feedback inhibited. We confirmed a strong feedback inhibition by a genetic approach and found R66 to be the key residue for this inhibition (Fig. 6) as suggested in Müller et al. 2024. The mechanism of feedback inhibition has been described in detail for B. subtilis GS (Murray et al., 2013). There, R62 plays the central role by binding glutamine and inducing a well ordered inactive structure at the substrate-binding pocket upon glutamine-binding (Murray et al., 2013). The homologous M. mazei R66 likely conveys a similar way of inhibition to B. subtilis GS (Fig. 6B, alignment in suppl. Fig. S5).

Further regulations by the two small proteins sP26 and the PII-like protein GlnK₁ have previously been reported for M. mazei (Ehlers et al., 2005; Gutt et al., 2021; Schumacher et al., 2023). However, in the present study neither an interaction with GlnK₁, nor GlnK₁ effects on GlnA₁ complex formation analysed by MP, nor an effect of GlnK₁ on GlnA₁ activity was detectable under the conditions used at varying 2-OG concentrations (0.1 to 12.5 mM) and ratios of GlnK₁ to GlnA₁ (20:1, 2:1, 2:10) (Fig. 2). Moreover, the addition of GlnK₁ did not result in a change of the K_D for 2-OG for the dodecamer GlnA₁ assembly (Fig. 2D). In previous reports, GlnK₁ was shown to interact with GlnA₁ in vivo after a nitrogen upshift by pull-down approaches (Ehlers et al., 2005), pointing towards an inhibitory function of GlnK₁ under shifting conditions from N limitation to N sufficiency. Similarly, we could not determine a cryo-EM structure including sP26 despite adding large excess of the small protein either obtained by co-expression or by addition of a synthetic peptide. Because these attempts were unsuccessful, we speculate that yet unknown cellular factor(s) might be required for an interaction of GlnA₁ with both small proteins, GlnK₁ and sP26, which however is difficult to simulate under in vitro conditions with purified proteins. Taken this into account, we speculate about a potential function of the two small proteins beyond GlnA₁ inactivation or activation. Since the GlnA₁ reaction is coupled to the GOGAT reaction (GS/GOGAT system) and the products of the two reactions replenish the substrates for one other, it is tempting to speculate that GlnA₁ and GOGAT experience metabolic coupling by sP26 and/or GlnK₁ e.g. by being involved in recruiting or separating GOGAT from GlnA₁.

Finally, higher oligomeric states of GS enzymes have been known for a long time for organisms like yeast and E. coli (He et al., 2009; Huang et al., 2022; Petrovska et al., 2014; Valentine et al., 1968). Interestingly, dependent on the ice thickness and on higher concentrated areas of the grids, we could also observe filament-like structures of M. mazei GlnA₁ in cryo-EM and resolved their structure at a resolution of 6.9 Å (suppl. Fig. S6). Such GlnA₁ filaments are also detectable in the cryo-EM images of Schumacher et al. 2023 but were not reported. Their interface is much alike the previously reported E.Coli GS filament structures (Huang et al., 2022). The physiological relevance of filamentation in M. mazei however remains unresolved and raises the question, whether an additional rapid modulation of GlnA₁ activity through higher oligomeric states exists, as described e.g. for yeast GS most depending on stress conditions (Petrovska et al., 2014).

M. mazei GlnA₁ shows unique properties

Overall, we have confirmed 2-OG to be the central activator of GlnA₁ in M. mazei, which assembles the active dodecamer and induces a conformational switch towards an active open state. Though 2-OG has previously been reported as an on-switch for (methano)archaeal GS activity (Ehlers et al., 2005; Müller et al., 2024; Pedro-Roig et al., 2013), the 2-OG-triggered assembly is novel and described exclusively for M. mazei GlnA₁. In this respect, the way of GS regulation in M. mazei is unique across all prokaryotic GS studied so far. Neither in cyanobacteria, enterobacteria or Bacillus is 2-OG a direct activator, nor is complex (dis-)assembly a mode of regulating GS activity in any other of these model organisms. This is further supported by the absence of up to three of those four arginines - coordinating 2-OG in M. mazei GlnA₁ - in these organisms (suppl. Fig. S5). The cyanobacterial, enterobacterial and gram positive GS are present in the cell as active dodecamers (Almassy et al., 1986; Bolay et al., 2018; Deuel et al., 1970). However, these dodecamers are inactivated upon sudden N sufficiency through very different mechanisms: Synechocystis GS is blocked by small proteins (inhibitory factors), the enterobacterial GS experiences gradual adenylylation of subunits which abolishes the enzyme activity and B. subtilis GS is feedback inhibited by glutamine and further inhibited by binding of the transcription factor GlnR (Almassy et al., 1986; Bolay et al., 2018; Klähn et al., 2018, 2015; Stadtman, 2001; Travis et al., 2022b) (see Fig. 7).

Figure 7:

The direct 2-OG activation and glutamine feedback inhibition of M. mazei GS are two fast, reversible and very direct ways of reacting towards the changing N status of the cell. We propose that the direct activation through 2-OG without any required additional protein as it is the case for all other regulations, is a more simple and direct regulation of GS. Due to the evolutionary placement of methanoarchaea and haloarchaea, where a direct 2-OG regulation has been found exclusively, this may represent an ancient regulation.

Materials and methods

Strains and plasmids

For heterologous expression and purification of Strep-tagged GlnA₁ (MM_0964), the plasmid pRS1841 was constructed. The glnA₁-sequence along with the sP26-sequence (including start-codon: ATG) were codon-optimized for Escherichia coli expression and commercially synthesized by Eurofins Genomics on the same plasmid (pRS1728) (Ebersberg, Germany). Polymerase chain reaction (PCR) was performed using pRS1728 as template and the primers (Eurofins Genomics, Ebersberg, Germany) GlnAopt_NdeI_for (5’TTTCATATGGTTCAGATGAAAAAATG3’) and GlnA1opt_BamHI_rev (5’TTTGGATCCTTACAGCATGCTCAGATAACGG3’). The resulting GlnA₁_opt PCR-product and vector pRS375 were restricted with NdeI and BamHI (NEB, Schwalbach, Germany); the resulting pRS375 vector fragment and the GlnA₁ fragment were ligated resulting in pRS1841. For heterologous expression of Strep-GlnA₁, pRS1841 was transformed in E. coli BL21 (DE3) cells (Thermo Fisher Scientific, Waltham, Massachusetts) following the method of Inoue (Inoue et al., 1990). For generating the Arg66Ala-mutant, a site-directed mutagenesis was performed. pRS1841 was PCR-amplified using primers sdm_GlnA_R66A_for (5’ATTGAAGAAAGCGATATGAAACTGGCGC3’) and sdm_GlnA_R66A_rev (5’CGCGGTAAAGCCCTGAATGCTGCTACC3’) by Phusion High-Fidelity polymerase (Thermo Fisher Scientific, Waltham, Massachusetts) followed by religation resulting in plasmid pRS1951. For heterologous expression, pRS1951 was transformed into E. coli BL21 (DE3).

In order to co-express sP26 along with Strep-GlnA₁, the construct pRS1863 was generated. pRS1728 with the codon-optimized sP26-sequence and pET21a (Novagen, Darmstadt, Germany) were restricted with NdeI and NotI and the resulting untagged sP26_opt was ligated into the pET21a backbone yielding pRS1863. pRS1863 was co-transformed with pRS1841 into E. coli BL21 (DE3) cells (Thermo Fisher Scientific, Waltham, Massachusetts) selecting for both Kanamycin (pRS1841 derived) and Ampicillin (pRS1863 derived) resistance.

The plasmid pRS1672 was constructed for producing untagged GlnK₁. The GlnK₁ gene was PCR-amplified using primers GlnK1_MM0732.for (5’ATGGTTGGCTATGAAATACGTAATTG3’) and GlnK1_MM0732.rev (5’TCAAATTGCCTCAGGTCCG3’) and cloned into pETSUMO by using the Champion™ pET SUMO Expression System (Thermo Fisher Scientific, Waltham, Massachusetts) according to the manufacturer’s protocol. pRS1672 was then transformed into E. coli DH5α and BL21 (DE3) pRIL (suppl. Tab. S1).

Heterologous expression and protein purification: Strep-GlnA₁ and GlnK₁

Heterologous expression of Strep-GlnA₁-variants (pRS1841 and pRS1951) and Strep-GlnA₁-sP26-coexpression (pRS1841 + pRS1863) were performed in 1 l Luria Bertani medium (LB, Carl Roth GmbH + Co. KG, Karlsruhe, Germany). E. coli BL21 (DE3) containing pRS1841, pRS1841 and pRS1863 or pRS1951 was grown to an optical turbidity at 600 nm (T₆₀₀) of 0.6 - 0.8, induced with 25 µM isopropylβ-d-1-thiogalactopyranoside (IPTG, Carl Roth GmbH + Co. KG, Karlsruhe, Germany) and further incubated over night at 18 °C and 120 rpm. The cells were harvested (6,371 x g, 20 min, 4 °C) and resuspended in 6 ml W-buffer (100 mM TRIS/HCl, 150 mM NaCl, 2.5 mM EDTA, (chemicals from Carl Roth GmbH + Co. KG, Karlsruhe, Germany), 12.5 mM 2-oxoglutarate (2-OG, Sigma-Aldrich, St. Louis, Missouri), pH 8.0). After the addition of DNase I (Sigma-Aldrich, St. Louis, Missouri), cell disruption was performed twice using a French Pressure Cell at 4.135 x 10⁶ N/m² (Sim-Aminco Spectronic Instruments, Dallas, Texas) followed by centrifugation of the cell lysate for (30 min (13,804 x g, 4 °C). The supernatant was incubated with 1 ml equilibrated (W-buffer) Strep-Tactin sepharose matrix (IBA, Gottingen, Germany) at 4°C for 1 h at 20 rpm. Strep-tagged GlnA₁ was eluted from the gravity flow column by adding E-buffer (W-buffer + 2.5 mM desthiobiotine (IBA, Gottingen, Germany)). Strep-GlnA₁ was always purified and stored in the presence of 12.5 mM 2-OG, either in E-buffer or 50 mM HEPES, pH 7.0 at 4 °C for a few days or with 5 % glycerol at −80 °C (chemicals from Carl Roth GmbH + Co. KG, Karlsruhe, Germany).

His₆-SUMO-GlnK₁ was expressed similarly using E. coli BL21 (DE3) pRIL + pRS1672. Expression was induced with 100 µM IPTG, incubated at 37 °C, 180 rpm for 2 h and harvested. The pellet was resuspended in phosphate buffer (50 mM phosphate, 300 mM NaCl, pH 8 (chemicals from Carl Roth GmbH + Co. KG, Karlsruhe, Germany)) and the cell extract was prepared as described above. His-tag-affinity chromatography-purifcation was performed with a Ni-NTA agarose (Qiagen, Hilden, Germany) gravity flow column, the protein was purified by stepwise-elution with 100 and 250 mM imidazole (SERVA, Heidelberg, Deutschland) in phosphate buffer. SUMO-protease (Thermo Fisher Scientific, Waltham, Massachusetts) was used according to the manufacturer’s protocol to cleave the His₆-SUMO-GlnK₁ and obtain untagged GlnK₁ by passing through the Ni-NTA-column after the cleavage. Elution fractions of protein purifications were analyzed on 12 % SDS-PAGE gels and the protein concentrations were determined by Bradford (Bio-Rad Laboratories, Hercules, California) or Qubit protein assay (Thermo Fisher Sceintific, Waltham, Massachusetts).

Determination of glutamine synthetase activity

The glutamine synthetase activity was determined by performing a coupled optical assay (Shapiro and Stadtman, 1970). The assay was performed as described in Gutt et al. 2021 with modifications. Modifcations included the use of 50 mM HEPES, the adjustment of ATP-pH to 7.0 and the use of 5 mM glutamine (Sigma-Aldrich, St. Louis, Missouri) in some assays. The assays were performed with four technical replicates per condition including two concentrations of GnA₁ (2 x 10 µg and 2 x 20 µg of Strep-GlnA₁). Strep-GlnA₁ was stored in E-buffer or 50 mM HEPES containing 12.5 mM 2-OG which was dialysed against 50 mM HEPES pH 7 using Amicon® Ultra catridges with 30 kDa filters (MilliporeSigma, Burlington, Massachusetts) for the enzyme assays in the absence of 2-OG.

Mass photometry

The molecular weight of protein complexes was analysed by mass photometry (MP) using a Refeyn twoMP mass photometer with the AcquireMP software (Refeyn Ltd., Oxford, UK). All measurements were performed in 50 mM HEPES, 150 mM NaCl pH 7.0 (MP-buffer, chemicals from Carl Roth GmbH + Co. KG, Karlsruhe, Germany) on 1.5 H, 24 x 60 mm microscope coverslips with Culture Well Reusable Gaskets (GRACE BIO-LABS, Bend, Oregon). Strep-GlnA₁ and untagged GlnK₁ were prepared as described above. Prior to MP experiments, a size exclusion chromatography (SEC) was performed with GlnA₁ in the presence of 12.5 mM 2-OG on a Superose™ 6 Increase 10/300 GL column (Cytiva, Marlborough, Massachusetts) with a flow rate of 0.5 ml/min. Only the dodecameric fraction was used for MP experiments and dialysed against MP buffer using Amicon® Ultra catridges with 30 kDa filters (MilliporeSigma, Burlington, Massachusetts) beforehand. The Gel Filtration HMW Calibration Kit (Cytiva, Marlborough, Massachusetts) was used as a standard in SEC. 75 – 200 nM monomeric Strep-GlnA₁ were used in the MP measurements, GlnK₁ was added accordingly in the desired ratio calculated based on monomers. The analysis of the acquired data was performed with the DiscoverMP software by applying a pre-measured standard (Refeyn Ltd., Oxford, UK). Counts were visualized in mass histograms as relative counts, which were calculated for the Gaussian fits of the measured peaks. For the determination of K_D-values and creating sigmoidal fitted curves, RStudio (RStudio Team (2020). RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL) was used.

Cryo-electron sample preparation and Data collection

Purified GS at a concentration of 1.5 mg/mL was rapidly applied to glow-discharged Quantifoil grids, blotted with force 4 for 3.5 s, and vitrified by directly plunging in liquid ethane (cooled by liquid nitrogen) using Vitrobot Mark IV (Thermo Fisher Scientific, Waltham, Massachusetts) at 100% humidity and 4 °C. To overcome prefererred orientation bias, 0.7 mM CHAPSO was added to prevent water-air interface interactions, consequently the concentration of the protein was increased to 6mg/ml. We added purified commercially synthesized sP26 (Davids Biotechnologie, Regensburg, Germany) to all samples, but the peptide did not stably bind under the observed conditions. Data was acquired with EPU in EER-format on an FEI Titan Krios G4 (Cryo-EM Platform, Helmholtz Munich) equipped with a Falcon IVi detector (Thermo Fisher Scientific, Waltham, Massachusetts) with a total electron dose of ∼55 electrons per Ų and a pixel size of 0.76 Å. Micrographs were recorded in a defocus range of −0.25 to −2.0 μm. For details see suppl. Table S2.

Cryo EM - Image processing, classification and refinement

All data was processed using Cryosparc (Punjani et al., 2017). Micrographs were processed on the fly (motion correction, CTF estimation). Using blob picker, 878,308 particles were picked, 2D-classified and used for ab initio reconstruction. Iterative rounds of ab initio and heterogenous refinement were used to clean the particle stacks. The final refinements yielded models with an estimated resolution of 2.39 Å sets at the 0.143 cutoff (suppl. Fig. S3).

An initial model was generated from the protein sequences using alphaFold (Jumper et al., 2021), and thereupon fitted as rigid bodies into the density using UCSF Chimera (Pettersen et al., 2021). The model was manually rebuilt using Coot (Emsley et al., 2010). The final model was subjected to real-space refinements in PHENIX (Liebschner et al., 2019). Illustrations of the models were prepared using UCSF ChimeraX (Pettersen et al., 2021). The structure is accessible under PDB: 8s59. For details see suppl. Table S2.

Acknowledgements

We thank the members of our laboratories for useful discussions on the experiments, as well as Claudia Kiessling for technical assistance. This work was supported by the German Research Council (DFG) priority program (SPP) 2002 ‘Small proteins in Prokaryotes, an unexplored world’ [Schm1052/20-2]. We acknowledge the contribution of the CryoEM Facility of the Philipps University of Marburg. J.M.S. acknowledges the DFG for an Emmy Noether grant (SCHU 3364/1-1) cofunded by the European Union (ERC, TwoCO2One, 101075992). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them. We thank Sandra Schuller for useful discussions and help in preparing manuscript figures. GKAH was supported by the Max Planck Society.

Figure supplements

Figure S1:

Figure S2:

Figure S3:

Figure S4:

Figure S5:

Figure S6:

Supplementary tables

Table S1:

Table S2: