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; Weidenbach et al., 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-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).

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., 2022), inhibition by interactions with small proteins (e.g. inhibitory factors in cyanobacteria García-Domínguez et al., 1999; Klähn et al., 2015, Klähn et al., 2018), 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 several of the different regulatory mechanisms for GS activity are reported for one organism. For example, yeast GS (ScGS) is regulated via feedback inhibition by glutamine and additionally 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 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 reversible GS-filament formation to an inactive form of EcGS, often associated with stress responses, 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., 2022).

In M. mazei, a mesophilic methanoarchaeon which is able to fix N₂, regulation of the central N metabolism 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. GlnA₁ is not covalently modified in response to N availability and thus represents an Iα-type-GS (Ehlers et al., 2005). It has been proposed, that GlnA₁ is directly activated under N starvation by the high intracellular concentrations of the metabolite 2-OG (Ehlers et al., 2005). 2-OG represents the internal signal for N limitation, since the internal 2-OG level significantly increases due to missing consumption by GDH under N starvation (M. mazei contains the oxidative TCA part, anabolic). The increased cellular 2-OG concentration has been shown to be directly perceived by GlnA₁, most likely by direct binding resulting in strong activation (Ehlers et al., 2005). Besides, we showed first evidence that 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 showed small effects on the GlnA₁ activity. However, those small effects might be neglectable compared to the strong 2-OG stimulation, particularly taking into account that the indirect GS activity assay shows high deviations in the low activity range. Moreover, initial complex formation analysis by a pull-down approach indicated 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 a potential inhibitory protein interaction with GlnK₁ (Ehlers et al., 2005). Very recently, the first structural analysis of M. mazei GlnA₁ was reported, showing GS complexes with GlnK₁ (Schumacher et al., 2023). 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 that work is entirely missing the effects of 2-OG on GlnA₁ structure and activity, we here aimed to study the regulation of M. mazei GlnA₁ in more detail by evaluating oligomerization and complex formation between GlnA₁, GlnK₁ and sP26 in dependence of 2-OG. This was achieved by employing mass photometry (MP), allowing the measurement of the molecular weight distribution of single complexes in solution, and by high resolution cryo-EM, whilst also performing activity assays.

The strep-tagged purified GlnA₁ was analyzed by size exclusion chromatography (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 (Figure 1—figure supplement 2). 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 of particles 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 nearly exclusively present as dimers with no higher molecular weight complexes present. After the 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) (Figure 1A and 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 allowed to determine the effective concentration of 2-OG for dodecamer assembly to be EC50=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 further verified that no other intermediate oligomeric complexes were detectable during dodecameric assembly (Figure 1A, Figure 1—figure supplement 1A, B). Notably, GlnA₁ did not reach 100% dodecamer assembly after removal and re-addition of 2-OG, although only dodecameric GlnA₁ was used for dialysis (Figure 1—figure supplement 2B, C). We conclude that GlnA₁ is rather unstable in the absence of 2-OG and some of the protein lost its ability to oligomerize after 2-OG was removed by dialysis.

Figure 1 with 4 supplements see all

Download asset Open asset

Furthermore, 2-OG did not only cause dodecamer assembly but also higher enzyme activity. Activity measurements of Strep-GlnA₁ in the presence of increasing 2-OG concentrations showed a strong increase 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). The EC50 for GlnA₁ activity was determined to be 6.3 mM 2-OG (Figure 1D). Thus, we conclude that 2-OG first acts as a trigger for dodecameric assembly of M. mazei GlnA₁ (with an EC50=0.75 mM 2-OG), setting it apart from other bacterial and eukaryotic enzyme variants. Moreover, most likely in addition to the dodecameric assembly, 2-OG is 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 at least 12.5 mM 2-OG is required (EC50=6.3 mM 2-OG).

Previous studies have shown protein interactions between M. mazei GlnA₁ and GlnK₁ as well as GlnK₁ induced effects on GlnA₁ activity (ratio GlnA₁:GlnK₁ 1:1, (Ehlers et al., 2005) and 2:1.4 (Gutt et al., 2021) in activity assays). Consequently, we next tested the effects of GlnK₁ on 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 calculated based on monomer mass) did not result in any dodecamer assembly of GlnA₁ (Figure 2A and B), (ii) no difference in the GlnA₁ dodecameric assembly in the presence of 2-OG was obtained in the presence of purified GlnK₁ (ratio 2:1), (iii) nor was binding of GlnK₁ to GlnA₁ detected by a respective increase in the mass of the higher oligomeric complex (Figure 2B and C). Moreover, the presence of GlnK₁ (ratio 2:1) neither had an influence on the 2-OG affinity (EC50 (- GlnK₁)=1.06 mM 2-OG; EC50 (+ GlnK₁)=1.02 mM 2-OG, EC50 calculated based on the dodecamer/dimer ratio), nor in any ratio on the specific activity of GlnA₁ (Figure 2D and E: exemplarily showing 2:1; Figure 1—figure supplement 1C, 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

Download asset Open asset

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 (Figure 1—figure supplement 2C). 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 (Figure 3, Figure 3—figure supplement 1, Supplementary file 2). Consequently, we achieved a structure demonstrating excellent geometry and density fitting.

Figure 3 with 2 supplements see all

Download asset Open asset

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 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 (Figure 4A, B and 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

Download asset Open asset

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 (Figure 4B and 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 (Figure 4B). Notably, F24 stabilizes the 2-OG via stacking with its phenyl ring (Figure 5). 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.

Figure 5

Download asset Open asset

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 (Figure 5C and E).

Superposition of our structure with the apo-M. mazei 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 (Figure 5A, B and D). R87' and its loop undergo a dramatic flip to coordinate 2-OG and D170 of helix α3 (residues 167–181) (Figure 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 (Figure 5B). This series of structural changes primes the active site for ATP binding by already adopting the side chain conformations that are observed in analog (Met-Sox-P-ADP)-bound structure (transition state) (PDB: 8tfk, Schumacher et al., 2023), thus facilitating nucleotide binding (Figure 5C and 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 (Liaw and Eisenberg, 1994; Figure 5D). In addition, Asp 57' moves deeper into the binding site, facilitating the proton abstraction of NH₄⁺ 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 open active state. This is illustrated by the superposition of the inhibitor-bound, transition-state locked structure (Schumacher et al., 2023; Figure 5C and E).

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., 2024). 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 Figure 5D and 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 (Figure 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 (Figure 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 (Figure 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

Download asset Open asset

In M. mazei, increased 2-OG concentrations act as a 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 different oligomeric states of GlnA₁. We have found mainly 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 (Figure 4B) and functions as a molecular glue between neighboring subunits. The assembly upon 2-OG addition observed using MP appears to be cooperative, fast, and without any detectable intermediate states (Figure 1B and C). Only immediately after thawing a frozen purified GlnA, and in case no additional SEC was performed prior to MP analysis, samples showed additional octameric complexes in MP with low abundancy (Figure 1—figure supplement 3). 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 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 (Figure 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., 2023, clearly demonstrates that 2-OG transfers GlnA₁ into its open active state conformation (Figure 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 (Figure 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 neighboring protomer (Müller et al., 2024). This is comparable to our observed cooperativity of M. mazei dodecamer assembly at low 2-OG concentrations (EC50=0.75 mM, calculated based on the percentage of dodecamer). On the other hand, M. mazei GlnA₁ reaches maximal activity only at much higher 2-OG concentrations (EC50=6.3 mM 2-OG) and likely requires a fully 2-OG-occupied dodecamer for maximal activity. The difference between the two EC50 values strongly points towards the dodecamer assembly being induced by only one 2-OG per hexameric ring, whilst the maximum activity requires one 2-OG in every 2-OG binding site (in agreement with roughly sixfold higher EC50). 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 a 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.

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 validated R66 to be the key residue for this inhibition (Figure 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 (Figure 6B, Figure 7—figure supplement 1).

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). Moreover, 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. However, in the present study, neither an interaction with GlnK₁, nor GlnK₁ effects on GlnA₁ complex formation analyzed by MP, nor an effect of GlnK₁ on GlnA₁ activity was detectable under the conditions used at varying 2-OG concentrations (0.1–12.5 mM) and ratios of GlnK₁ to GlnA₁ (20:1, 2:1, 2:10) (Figure 2). Moreover, the addition of GlnK₁ did not result in a change of the EC50 of 2-OG for the dodecamer GlnA₁ assembly (Figure 2D). Similarly, we could not determine a cryo-EM structure including sP26 despite adding a 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. Taking 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 (Figure 3—figure supplement 2). Such GlnA₁ filaments are also detectable in the cryo-EM images of Schumacher et al., 2023, but were not reported. The filament 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. In yeast for example, GS filamentation was described as mostly depending on stress conditions (Petrovska et al., 2014).

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 dodecameric assembly is novel and described exclusively for M. mazei GlnA₁. Neither in cyanobacteria, enterobacteria or Bacillus has 2-OG been reported as the sole direct activator of the enzyme, 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, which are coordinating 2-OG in M. mazei GlnA₁, in these organisms (Figure 7—figure supplement 1). The cyanobacterial, enterobacterial, and gram-positive GS are assumed to be present in the cell as active dodecamers (Almassy et al., 1986; Bolay et al., 2018; Deuel et al., 1970). These dodecamers are inactivated upon sudden N sufficiency through very different mechanisms all including additional proteins (see Figure 7). In Synechocystis, GS is blocked by small proteins, which are repressed under nitrogen limitation, one in a 2-OG-NtcA mediated way and the other one via a glutamine sensing riboswitch (Bolay et al., 2018; Klähn et al., 2018; Klähn et al., 2015). The enterobacterial GS experiences 2-OG-PII dependent 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; Stadtman, 2001; Travis et al., 2022). Consequently, the GS regulation in M. mazei by a 2-OG triggered assembly is unique across all prokaryotic GS studied so far.

Figure 7 with 1 supplement see all

Download asset Open asset

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.

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) selected 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 (Hanahan, 1983) and BL21 (DE3) pRIL (Supplementary file 1).

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 overnight at 18 °C and 120 rpm. The cells were harvested (6371 × 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 4.135×10⁶ N/m² (Sim-Aminco Spectronic Instruments, Dallas, Texas) followed by centrifugation of the cell lysate for 30 min (13,804 × 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 an 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 hr, 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-purification 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 Scientific, Waltham, Massachusetts).

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. First, a substrate mix containing 257 mM KCl, 143 mM NH₄Cl, 143 mM MgCl₂ (chemicals from Carl Roth GmbH + Co. KG, Karlsruhe, Germany), and 86 mM sodium-glutamate (Sigma-Aldrich, St. Louis, Missouri) was prepared. The assay was performed in a final volume of 1 ml including 350 µl of the substrate mix, 10 µl of lactic dehydrogenase and pyruvate kinase mix (Sigma-Aldrich, St. Louis, Missouri), 50 mM HEPES (final concentration, Carl Roth GmbH +Co. KG, Karlsruhe, Germany), the respective amount of 2-OG (Sigma-Aldrich, St. Louis, Missouri), 1 mM phosphoenolpyruvate (PEP, Sigma-Aldrich, St. Louis, Missouri), 0.42 mM nicotinamide adenine dinucleotide (NADH) and 10 or 20 µg of Strep-GlnA₁. After preincubation at room temperature in a volume of 950 µl for 5 min, the assay mixture was transferred to a cuvette, the time course measurement at 340 nm was started and the enzyme reaction induced by adding 3.6 mM ATP (pH adjusted to 7.0, Roche, Basel, Switzerland) (Figure 1—figure supplement 4). Biological replicates were independent protein expressions and purifications and the assays were performed with four technical replicates per condition, including two concentrations of GnA₁ (2x10 µg and 2×20 µg of Strep-GlnA₁, present in 100 µl were added). Strep-GlnA₁ was stored in E-buffer (described above) or 50 mM HEPES containing 12.5 mM 2-OG which was dialyzed against 50 mM HEPES pH 7.0 using Amicon Ultra cartridges with 30 kDa filters (MilliporeSigma, Burlington, Massachusetts) for the enzyme assays in the absence of 2-OG.

The molecular weight of protein complexes was analyzed 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, 24×60 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 dialyzed against MP buffer using Amicon Ultra cartridges 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. Biological replicates were independent protein expressions and purifications and technical replicates were the repetition of the measurement with the same protein in a fresh mix. 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 EC50-values and creating sigmoidal fitted curves, RStudio (RStudio Team (2020). RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL) was used.

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 the preferred orientation bias, 0.7 mM CHAPSO was added to prevent water-air interface interactions, consequently the concentration of the protein was increased to 6 mg/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 Supplementary file 2.

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 (Figure 3—figure supplement 1).

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 Supplementary file 2.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

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. JMS acknowledges the DFG for an Emmy Noether grant (SCHU 3364/1–1) co-funded 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. We acknowledge financial support by Land Schleswig-Holstein within the funding program Open Access Publikationsfonds.

1. Sent for peer review: March 18, 2024
2. Preprint posted: March 19, 2024
3. Reviewed Preprint version 1: May 28, 2024
4. Reviewed Preprint version 2: February 17, 2025
5. Version of Record published: March 31, 2025

You can cite all versions using the DOI https://doi.org/10.7554/eLife.97484. This DOI represents all versions, and will always resolve to the latest one.

© 2024, Herdering, Reif-Trauttmansdorff et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.