High complexity of Glutamine synthetase regulation in Methanosarcina mazei: Small protein 26 interacts and enhances glutamine synthetase activity

Small ORF (sORF)‐encoded small proteins have been overlooked for a long time due to challenges in prediction and distinguishing between coding‐ and noncoding‐predicted sORFs and in their biochemical detection and characterization. We report on the first biochemical and functional characterization of a small protein (sP26) in the archaeal model organism Methanosarcina mazei, comprising 23 amino acids. The corresponding encoding leaderless mRNA (spRNA26) is highly conserved on nucleotide level as well as on the coded amino acids within numerous Methanosarcina strains strongly arguing for a cellular function of the small protein. spRNA26 level is significantly enhanced under nitrogen limitation, but also under oxygen and salt stress conditions. Using heterologously expressed and purified sP26 in independent biochemical approaches [pull‐down by affinity chromatography followed by MS analysis, reverse pull‐down, microscale thermophoresis, size‐exclusion chromatography, and nuclear magnetic resonance spectroscopy (NMR) analysis], we observed that sP26 interacts and forms complexes with M. mazei glutamine synthetase (GlnA1) with high affinity (app. KD = 0.76 µm± 0.29 µm). Moreover, seven amino acids were identified by NMR analysis to directly interact with GlnA1. Upon interaction with sP26, GlnA1 activity is significantly stimulated, independently and in addition to the known activation by the metabolite 2‐oxoglutarate (2‐OG). Besides, strong interaction of sP26 with the PII‐like protein GlnK1 was demonstrated (app. KD = 2.9 µm ± 0.9 µm). On the basis of these findings, we propose that in addition to 2‐OG, sP26 enhances GlnA1 activity under nitrogen limitation most likely by stabilizing the dodecameric structure of GlnA1.

Small ORF (sORF)-encoded small proteins have been overlooked for a long time due to challenges in prediction and distinguishing between coding-and noncoding-predicted sORFs and in their biochemical detection and characterization. We report on the first biochemical and functional characterization of a small protein (sP26) in the archaeal model organism Methanosarcina mazei, comprising 23 amino acids. The corresponding encoding leaderless mRNA (spRNA26) is highly conserved on nucleotide level as well as on the coded amino acids within numerous Methanosarcina strains strongly arguing for a cellular function of the small protein. spRNA26 level is significantly enhanced under nitrogen limitation, but also under oxygen and salt stress conditions. Using heterologously expressed and purified sP26 in independent biochemical approaches [pull-down by affinity chromatography followed by MS analysis, reverse pull-down, microscale thermophoresis, size-exclusion chromatography, and nuclear magnetic resonance spectroscopy (NMR) analysis], we observed that sP26 interacts and forms complexes with M. mazei glutamine synthetase (GlnA 1 ) with high affinity (app. K D = 0.76 µM AE 0.29 µM). Moreover, seven amino acids were identified by NMR analysis to directly interact with GlnA 1 . Upon interaction with sP26, GlnA 1 activity is significantly stimulated, independently and in addition to the known activation by the metabolite 2-oxoglutarate (2-OG). Besides, strong interaction of sP26 with the PII-like protein GlnK 1 was demonstrated (app. K D = 2.9 µM AE 0.9 µM). On the basis of these findings, we propose that in addition to 2-OG, sP26 enhances GlnA 1 activity under nitrogen limitation most likely by stabilizing the dodecameric structure of GlnA 1 .

Introduction
Modern genomic and transcriptomic technologies combined with systematic genome-wide approaches have uncovered an unexpected genome complexity in prokaryotes. Besides genes encoding larger proteins and genes of noncoding RNAs (ncRNAs), global approaches have over the past decade discovered a wealth of hidden small genes containing short open reading frames (sORFs) in many prokaryotic genomes [1][2][3]. These sORFs often encode proteins smaller than 50 amino acids (aa) in length and have been typically missed in genome annotations by automated gene predictions due to too strict assumptions and traditionally considering only standard proteins in the automated gene annotation tools [4,5]. This led to neglecting the existence of an additional layer of complexity represented by small proteins. Besides, small proteins have been difficult to detect biochemically due to technical limitation. Though, nowadays new technologies are emerging, which enable their global profiling in genome-wide approaches, and systematic approaches for global identification are used, for example, by sophisticated bioinformatic predictions, peptidomics, ribosome profiling, and combinations of those [5][6][7][8][9][10][11][12][13][14][15]. Due to their small size, the small proteins are frequently predicted to modify the activity of larger proteins or complexes via physical interactions or interact with the membranes. However, information regarding specific physiological role(s) of verified small proteins is lacking for the majority of confirmed small proteins. Only a fraction of small proteins experimentally identified in diverse prokaryotes have been functionally characterized. This characterization demonstrated that they can play important roles in different functional scenarios and have a broad range of function from cell division, signal transduction, modeling membrane protein recruitment, or (membrane) protein activity to modulation or being part of a larger mostly membrane integrated protein complex (reviewed by Ref. [2,16,17]). In archaea, only a few small proteins have been reported, the majority of which is regulated in response to specific stress. However, for most of them verified functional analysis is scarce or missing [7,8,[18][19][20][21][22][23][24].
Methanosarcina mazei strain G€ o1 belongs to the methylotrophic methanogens of the order Methanosarcinales, which have the most versatile substrate spectrum within the methanogenic archaea and significantly contribute to the production of the greenhouse gas [25]. M. mazei is able to fix molecular nitrogen under nitrogen (N) limitation. The regulation of the N metabolism particularly of the nitrogen fixation as well as the glutamine synthetase is well studied on the transcriptional and post-transcriptional levels [26][27][28][29][30][31]. Overall regulation occurs in response to the N availability, where N limitation is in general perceived internally by sensing the internal 2-oxoglutarate (2-OG) pool, which increases under N limitation due to reduced consumption by the ammonium-dependent glutamate dehydrogenase and increasing glutamine synthetase/glutamate synthase pathway (GS/GOGAT) for glutamate synthesis [32]. We recently showed that post-transcriptional regulation by small RNA 154 , originally identified in a global RNA sequencing (RNAseq) approach [27], plays a central role in N regulation of several components of the N-cycle including glutamine synthetase [28]. Besides, RNAseq and term-seq approaches as well as unpublished RNAseq data sets identified not only high numbers of small noncoding RNAs but also numerous small mRNAs in M. mazei containing putative sORFs [27,33]. Aiming to elucidate whether the predicted sORFs identified are translated in vivo, the full cytosolic proteome of M. mazei was examined using reversed-phase LC-MS MS approaches (bottom-up and top-down strategies) as described in Ref [7,8,19,23]. In these comprehensive studies, overall 68 small proteins were detected and experimentally validated with high or mid confidence. In this present study, the small protein 26 (sP26), identified with low confidence in our previous study but verified by a recent ribosome-profiling analysis (R. A. Schmitz & M. Gutt, unpublished data), was selected aiming to identify its physiological role, due to the fact that the respective encoding mRNA (spRNA26) is upregulated under N starvation [27]. Moreover, the genomic location of spRNA26 downstream of the operon encoding the acetyl-CoA-decarbonylase/synthase (ACS) complex [27] suggested ACS, which itself is upregulated in response to N starvation on the post-transcriptional level [26], as an attractive potential target for sP26. Very recently, nuclear magnetic resonance spectroscopy (NMR) spectroscopy analysis demonstrated that chemically synthesized sP26 is unstructured, whereas bioinformatic tools predict that unstructured sP26 can potentially fold into a structure upon complex formation with a target [21]. Here, we demonstrate by several independent biochemical approaches that sP26 interacts with glutamine synthetase (GlnA 1 ) and enhances its activity in M. mazei. Glutamine synthetase, the central component of the N-cycle, is known to be strictly regulated in response to the cellular nitrogen status by various molecular mechanisms in other Prokaryotes. They range from feedback inhibition, post-translational modifications, and allosteric inhibitors to interactions with other proteins [34,35].

Results
Aiming to gain insight into the physiological role of small protein sP26 in M. mazei, we addressed our hypothesis that sP26 is directly or indirectly involved in N regulation. First we examined, whether the archaeal small protein interacts with other M. mazei proteins and verified the identified interacting partner glutamine synthetase using various approaches.
spRNA26 and the corresponding small protein are highly conserved in Methanosarcina strains and upregulated under N limitation in M. mazei spRNA26 was identified using a differential RNAseq (dRNA-seq) approach and shown to be expressed under N limitation, whereas under N sufficiency no transcript was detectable [27]. It represents an~70 nucleotide (nt) long leaderless RNA encoding a small protein of 23 amino acids. A tryptic peptide from this protein was identified in the cell extraction of cells grown under N starvation conditions via a 2DLC-MS bottom-up proteomic analysis (annotated spectra of the peptide Fig. S1) [8]. The sp26-derived tryptic peptide was identified with a lower level of stringency than applied in previously reported publications [7,8,19], with a protein-level false discovery rate of 5%, and an MS2 fragment ion tolerance of 0.05 Da, however, can be considered as unambiguously identified. This is further supported by verifying translation in ribosome-profiling approaches (Schmitz and Gutt, unpublished data). spRNA26 is located within the 316-bp intergenic region between MM2083 encoding orotidine 5 0 -monophosphate decarboxylase and the operon encoding the acetyl-CoA-decarbonylase/synthase (ACS) complex. Its promoter (TATA box and BRE box) was identified between nt 26 and 40 upstream of the transcriptional start site (+1) based on the distance to the start site and their consensus sequence in M. mazei (Fig. 1A). Two further alternative BRE boxes can be predicted (indicated with a dashed line). The repressor-binding motif of the general transcriptional regulator nitrogen regulator protein R (NrpR) [29] was not detected. However, the repetitive elements identified upstream of the BRE box (indicated with gray boxes) might represent a weak binding site for the transcriptional activator nitrogen regulator protein A which has been identified in the upstream promoter region of the nif operon [31]. Multiple alignments of the respective spRNA26 and homologs in other Methanosarcina species showed that the sORF experimentally verified in M. mazei is highly conserved with regard to amino acid as well as the nucleotide sequence (Fig. 1B,C). Further, high conservation of the promoter and the 5' upstream region was identified among the M. mazei strains of which four are missing the BRE box identified for M. mazei G€ o1 and thus might use one of the alternative BRE boxes (Fig. 1B). All homologs are predicted to start with valine, indicating that a noncanonical start codon is used for translation initiation (GTG). Interestingly, at the third position of the small protein three Methanosarcina barkeri strains contain methionine (ATG) in contrast to all other strains encoding valine (GTG) at the third position. This might indicate the presence of an alternative translation start. Unfortunately, the peptide detected by the LC-MS MS analysis starts with the 5th position (lysine) of the predicted sORF (see Fig. 1C, The similarity is shown in black-gray-white boxes (black symbols 100% similarity); additionally, the identity is shown in a gray bar and a nucleotide logo above the nucleotide alignment. (C) Amino acid alignment of sP26 homologs in Methanosarcina species. The protein sequence was generated by blasting the nucleotide sequence, creating every possible translation frame, and aligning every possible sequence to sp26 using ClustalW and further evaluated by hand. The similarity is shown in black-gray-white boxes (black symbols 100% similarity), and the identity is shown in a gray bar and a nucleotide logo above the alignment; *, indicates the peptide identified by LC-MS/ MS; -, indicates a stop. All alignments were generated using Geneious Prime 2020.1.2 (https://www.geneious.com) and sequence data from NCBI with following Accession numbers: M. acetivorans indicated with a box in the sequence logo), thus cannot be used to identify the native start of the small protein.
In the M. thermophila strains and in M. falvescens, apparently a deletion of nt 33 occurred causing the generation of a translational stop codon subsequently resulting in a very short version of sP26 (Fig. 1C). Overall, the high conservation of the amino acid as well as the nucleotide sequence argues for a potential physiological role of sP26. Northern blot and Illumina RNAseq analysis confirmed the reported transcriptional start site (+1) and upregulation of spRNA26 under N limitation (~1.5 to 2.5 fold) as well as under oxygen and salt stress (Fig. 2). In agreement with the RNAseq data, spRNA26 appears to be processed into three shorter fragments (65, 63, and 61 nt) in varying amounts depending on the growth phase and stress conditions ( Fig. 2A,C).
To study potential effects of the small protein sP26 in vivo, the respective sORF was additionally expressed in M. mazei from a plasmid under the control of the constitutive promoter pmcrB with the ribosome-binding site of mcrB (pRS1242, plasmid map, see Fig. S2d). Growth analysis under N sufficient and N limiting conditions showed no significant phenotype; growth rates comparable to the control containing the empty pWM321 vector were obtained independent of additional sP26 synthesis (µ (+N) = 0.080 h À1 vs. 0.086 h À1 , µ (ÀN) = 0.082 h À1 ). Cell extract preparations and subsequent western blot analysis using antibodies raised against synthetic sP26 demonstrated that chromosomally expressed sP26 is induced under N starvation. The majority is in the soluble fraction independently from the N availability but in a higher oligomeric state or bound to a protein with higher molecular mass (Fig. 3). sP26 is interacting with glutamine synthetase In order to identify cell proteins directly interacting with sP26, we studied potential complex formation between sP26 N-terminally fused to a SUMO-tag (Sumo-His 6 -sP26) and M. mazei cell extract proteins by pull-down experiments using affinity chromatography on Ni-NTA agarose for detecting complexes. His 6 -SUMO-His 6 -sP26 was heterologously expressed in E. coli (pRS1229, see Fig. S2c) and purified to an apparent homogeneity of 98% by Ni-NTA affinity chromatography as described in Methods. Purified His 6 -SUMO-His 6 -sP26 (1 mg) was incubated for 1 h at 4°C with~27.5 mg cell extract protein of M. mazei cells grown under N limitation. Subsequently, the protein mixture was applied to Ni-NTA agarose. After washing the chromatography material to remove all cell extract proteins, unspecifically binding to His 6 -SUMO-His 6 -sP26 or the Ni-NTA agarose, His 6 -SUMO-His 6 -sP26, and potential specifically interacting proteins were eluted in the presence of 250 mM imidazole. The respective elution fractions were analyzed by denaturating SDS/PAGE and subsequent silver staining. As exemplarily shown in Fig. 4A, only one distinct additional protein band corresponding to an~42 K D protein was detected in at least three independent biological replicates. The elution fractions were concentrated and reanalyzed on an SDS/PAGE stained with Coomassie blue (Fig. 4A, panel 2). The additional protein band was analyzed, and the presence of glutamine synthetase (GlnA 1 ) was demonstrated in two independent biological experiments by LC-MS/MS analysis (see Materials and methods). Unspecific binding of M. mazei proteins to the SUMO-fusion protein or the affinity chromatography material was excluded by using purified SUMO-fusion protein and M. mazei cell extract and loading M. mazei cell extract to the affinity chromatography material followed by elution (see Fig. 4A panel 3). To confirm complex formation with GlnA 1 , a reverse pulldown was performed. Strep-tagged GlnA 1 (pRS375) and SUMO-His 6 -sP26 (pRS1229) were individually heterologously expressed in E. coli. Strep-GlnA 1 was purified to homogeneity by affinity chromatography using Strep-TactinÒ sepharoseÒ as described in Materials and methods. 1 mg purified Strep-GlnA 1 was incubated with the respective E. coli cell extract with SUMO-His 6 -sP26 expressed (~50 mg) for 30 min at 4°C followed by affinity chromatography using Strep-TactinÒ sepharoseÒ. Strep-GlnA 1 eluted from the column mainly in elution fraction 2, subsequent western blot analysis using antibodies directed against the His 6 -tag confirmed the co-elution of SUMO-His 6 -sP26 in the Strep-GlnA 1 elution fractions (Fig. 4B).
To obtain further experimental evidence for the interaction between sP26 and GlnA 1 , microscale thermophoresis (MST) analysis was performed and the dissociation constant (K D ) determined. GlnA 1 was purified as N-terminal (His) 6 -fusion protein as described by   [32]. sP26 was chemically synthesized (ssP26) and nt-647-NHS labeled as described in Methods. Interactions between the labeled ssP26 (20 nM, based on the monomeric MW) and purified His 6 -GlnA 1 in the range of 0.13 nM to 4.15 µM (calculated based on the monomeric MW) were analyzed by MST. Significant binding was observed in two independent biological experiments, each representing completely independent protein expressions and purifications and containing up to three technical replicates. One biological experiment is depicted in Fig. 5A; the K D of 75 AE 40 nM was calculated based on integrating at least three technical replicates (panel 1), and panel 2 shows the respective calculation with one technical replicate. To verify the interaction, His 6 -sP26 was expressed in E. coli (pRS1245, see Fig. S2e) and purified by fractionated ammonium precipitation followed by affinity chromatography on HisPur TM cobalt superflow agarose, and size exclusion using 10 kDa filters as described in detail in Methods resulting in an~95% purified protein fraction. When using purified and nt-647-NHS-labeled GlnA 1 (20 nM, monomeric) and purified His 6 -sP26 in the range of 4 nM to 140 µM, a K D of 54 µM AE 18 µM (monomeric) was calculated based on several technical replicates ( Fig. 5B, panel 1), which was validated in a second independent biological replicate with three technical replicates. The specific binding with high affinity was further verified using purified untagged sP26 generated from purified SUMO-sP26 by removing the SUMO-fusion protein obtaining an even higher affinity for the taggless sP26 (see Fig. 5C, depicting one of three biological replicates). The fact that the K D of the untagged sP26 version is two orders of magnitude lower indicates that the His-tag is hampering binding.
We performed NMR analysis of purified 15 N-labeled-untagged sP26 in the presence and absence of the target protein GlnA 1 and 2-OG (~12 mM). As depicted in the 2D 1 H 15 N HSQC spectrum (Fig. 6) showing signals of NH-sites in sP26 in the absence of the target, 18 out of 20 expected amide signals of sP26 are visible (shown in black); the spectrum is in complete agreement with the respective one of chemically synthesized sP26 [21]. In the presence of GlnA 1 (shown in red), seven amino acids of sP26 cannot be detected. They built the interaction surface, and their signals vanish either due to faster relaxation or due to chemical exchange-induced line broadening, further confirming specific binding of sP26 to GlnA 1 and allowing to identify the amino acids of sP26 that directly interact with GlnA 1 to be K5, S10, M11, M15, S16, S17, and E18. In addition, backbone amide signals of L7, M12, and K21 do not completely vanish but become significantly weaker upon addition of GlnA 1 .  as well as His 6 -sP26 and untagged sP26 were purified by affinity chromatography after heterologous expression in E. coli. GlnA 1 activity was determined by using the coupled optical test assay described by Shapiro [36], which couples the consumption of ATP by the conversion of ammonium and glutamate to glutamine catalyzed by glutamine synthetase to the oxidation of nicotinamide adenine dinucleotide (reduced) (NADH) by lactate dehydrogenase. In the course of optimizing the test assay, we observed that the buffer system is crucial and significantly effects the specific activity of glutamine synthetase. Using the originally reported MOPS buffer, which was used in the previous report on GlnA 1 activity [32], significantly lower specific activities were observed in comparison with other buffer systems. N-(2-hydroxyethyl)piperazine-N 0 -(2ethanesulfonic acid) (HEPES) buffer was proven to show the highest specific activities of GlnA 1 in the absence but also in the presence of the metabolite 2-OG, which has been demonstrated to significantly enhance GlnA 1 activity [32]. Due to the higher specific activity, we consequently exclusively used the HEPES buffer-based assay in the following experiments and purified GlnA 1 in HEPES buffer. In the presence of increasing amounts of purified His 6 -sP26 (1.8-18 µM) (Fig. 7A), which were pre-incubated with 0.95 µM His 6 -GlnA 1 (calculated based on the monomeric MW) for 5 min at RT in the test assay before starting by supplementing ATP, glutamine synthetase activity was significantly enhanced. The positive effect of His 6 -sP26 on GlnA 1 activity was confirmed by at least five further independent biological replicates using independent protein purifications. Further, the presence of contaminating adenosintriphosphatase activities in purified His 6 -sP26 fractions that could interfere with the test assay was excluded. In general, stimulation of glutamine synthetase activity up to ninefold was obtained by His 6 -sP26 depending on the respective protein purification (its quality condition) and consequential saturation level (see summarizing Table S1). One representative set resulting in 2.1-fold stimulation is exemplarily shown in Fig. 7B. Elucidating the effects of His 6 -sP26 on glutamine synthetase activity in the presence of the known activator 2-OG (5 mM) clearly demonstrated that the obtained activating effect of sP26 is independent of the activation by 2-OG and stimulates the activity in addition to the activation due to 2-OG (see Fig. 7C, showing one exemplarily chosen biological activity set). Several independent biological replicates (independent protein purifications) confirmed the additional stimulation of GlnA 1 activity by His 6 -sP26 in the presence of 2-OG up to fivefold (Table S1). For one fraction of purified His 6 -sP26, we additionally tested the positive effects in the presence of 5 mM 2-OG until achieving apparent saturation by His 6 -sP26, indicating that~10 µM His 6 -sP26 is saturating when using 0.95 µM GlnA 1 in the test assay (calculations based on GlnA 1 monomers) (Fig. 7F).

5358
The  however, showed a tenfold-enhanced activity compared with the His 6 -tagged version of the protein (Fig. 7D, E), indicating that the tag is hampering the activity or the binding to GlnA 1 . To address the predicted higher oligomeric state of GlnA 1 in the presence of 2-OG [32], we analyzed the conformation and oligomeric state of purified GlnA 1 by structural integrity analysis using the Tycho system (NanoTemper, Munich, Germany) as described in Methods. The analysis showed two different transformation events first the unfolding of the monomer at 63°C (black line) in the absence of 2-OG and additionally in the presence of the metabolite the dissociation of the oligomer at 48°C (light gray line) (Fig. 8A). The addition of increasing concentrations of 2-OG resulted in increasing amounts of the oligomeric conformation of GlnA 1 and complementary decreasing amounts of the monomeric fraction ( Fig. 8A: 5 mM 2-OG light gray line; Fig. 8B: 5, 7, 10, and 15 mM 2-OG from light gray to dark gray). Besides, binding of 2-OG generally resulted in higher stability of both conformations indicated by the significant shift to higher denaturation temperatures (e.g., shifting from 60 to 70°C for the monomer). Evaluating different independent purifications, this variation in the oligomeric fraction of GlnA 1 might explain variability in activity and the range of stimulation observed by sP26 (Table S1). At present however, it is unclear whether the reason for this variability is based on biological or technical variables in the preparation.
sP26 is interacting with GlnK 1 and stimulates GlnA 1 activity in addition to activation by GlnK 1 Previous studies have shown that the PII-like protein GlnK 1 is modulating the GlnA 1 activity in M. mazei and GlnA 1 inhibition was predicted due to upshifts in N availability [32]. Consequently, we included GlnK 1 in our studies evaluating sP26 effects on GlnA 1 activity. Interaction studies between purified proteins by MST analysis clearly showed and verified the predicted specific interaction between GlnA 1 and GlnK 1 . Using His 6 -GlnA 1 and nt-647-NHS-labeled His 6 -GlnK 1 , a K D of 1.5 µM AE 0.7 µM was calculated (always based on the monomeric MW) and verified in additional independent biological experiments including a reverse labeling experiment (nt-647-NHS-labeled His 6 -GlnA 1 and His 6 -GlnK 1 , K D = 17 µM AE 3 µM) (Fig. 9D,E). For all those experiments, the presence of 5 mM 2-OG did not change the binding affinity (K D values, data not shown). Interactions between the second PII-like protein in M. mazei (GlnK 2 ) and GlnA 1 could not be detected by MST analysis, strongly arguing for GlnA 1specific interaction with GlnK 1 .
Elucidating potential interactions between GlnK 1 and sP26 showed in two independent biological experiments that nt-647-NHS-labeled His 6 -GlnK 1 besides interacting with GlnA 1 similarly interacts with His 6 -sP26 with high affinities and as well independent of 2-OG (see Fig. 9B, K D = 2.5 µM AE 1.6 µM). The interaction was further confirmed with untagged sP26 ( Fig. 9C, K D = 2.9 AE 0.9 µM, representing one out of three biological experiments), as well as by an independent reverse labeling experiment analyzing the interaction between His 6 -GlnK 1 and nt-647-NHS-labeledsynthesized sP26 (Fig. 9A, K D = 51 nM AE 25 nM). Interaction between sP26 and GlnK 2 was not detected.
Overall, these findings strongly argue that GlnK 1 modulation has to be taken into further account when analyzing the sP26 effects on GlnA 1 activity. Consequently, we included GlnK 1 in evaluating the enhancing effects of sP26 on GlnA 1 activity. Aiming to first validate the reported inhibitory effects of GlnK 1 on GlnA 1 activity in the new buffer system, we reanalyzed the GlnK 1 effect in the modified activity assay. Using the originally MOPS-based system, we confirmed the inhibitory effects of GlnK 1 on GlnA 1 activity; however, changing the buffer system (to NaH 2 PO 4 or HEPES) resulted in positive effects of GlnK 1 on GlnA 1 activity independent of the presence of 2-OG (data not shown). Using the HEPES-based test assay, we obtained strong evidence in two biological independent replicates (each including three technical replicates) that the stimulating effect of sP26 on glutamine synthetase activity is not only independent of 2-OG but also independent and in addition to the stimulating effect of GlnK 1 (Fig. 9F, depicting one of two biological replicates). In the presence of 2-OG, the additive effects of GlnK 1 and sP26 were not anymore detectable most likely due to limiting concentrations of NADH and ATP in the test assay (saturation of the activity) (Fig. 9G).
In order to evaluate potential complexes formed between the three proteins, purified His 6 -GlnK 1 (77.7 µg), His 6 -GlnA 1 (168.5 µg), and His 6 -sP26 (24 µg) were pre-incubated at RT for 5 min and analyzed by size-exclusion chromatography (SEC) in the absence and presence of 5 mM 2-OG in the buffer as described in Methods (ENrich 650 column; Bio-Rad, Feldkirchen, Germany). Analysis of the respective SEC fractions was performed via LC-MS/MS as described in Methods. The results summarized in Fig. 10 clearly show that the addition of 2-OG facilitated the identification of both sP26 and GlnK 1 in the highest molecular weight (MW) fraction (> 600 kDa) besides GlnA 1 , providing further evidence toward the close association of these three proteins in the presence of 2-OG.

Discussion
The recently recognized existence of previously overlooked hidden small proteins in bacterial and archaeal genomes became an emerging research field. Here, we report on the physiological role of the first small archaeal protein in M. mazei. Concordantly we obtained strong evidence that sP26 interacts specifically and with high affinity with the glutamine synthetase (GlnA 1 ) in M. mazei the key enzyme of ammonium assimilation under N limiting growth conditions. Interaction with sP26 directly effects and enhances the activity of GlnA 1 independently and in addition to the inducing metabolite 2-OG and the modulating PII-like nitrogen sensory protein GlnK 1 .
Ammonium assimilation under N starvation via the GS/GOGAT pathway is one of the major intersections in central N metabolism (see Fig. 11, inset). Consequently, in all organisms synthesis and activity of glutamine synthetase (GS) is strictly regulated by N availability [34,37,38]. The GS in prokaryotes and eukaryotes can be classified in three different families, GSI, GS II, and GSIII, all of which are representing large homo-oligomeric complexes with 8, 10, or 12 monomers. The decameric GSII is present in eukaryotes, whereas GSI comprising two subtypes, GSI-a and GSI-b, and GSIII are found in bacteria and archaea mostly forming dodecamers [35]. GSI are typically feedback inhibited by the end products glutamine and AMP. In addition and in contrast to GSI-a, GS of subtype GSI-b are further inhibited in response to N sufficiency by a specific covalent modification of a tyrosine residue near the active side (adenylation by an adenylyltransferase) [35,39,40]. In M. mazei, the essential GS, GlnA 1 , does not exhibit an adenylylation site nor have homologues of adenylyltransferase been identified in the M. mazei genome [39,41] excluding posttranslational regulation of M. mazei GS in response to N by covalent modification [35,39,40]. Accordingly, GlnA 1 represents a GS of the GSI-a subdivision. The gene encoding the essential GS in M. mazei (glnA 1 ) is under the direct transcriptional control of the global nitrogen repressor NrpR and thus exclusively transcribed under N limitation [27,29]. In addition, the expression of glnA 1 is post-transcriptionally regulated by sRNA 154 , which significantly stabilizes glnA 1 transcripts under N limitation [28]. In respect to posttranslational N-dependent regulation of GlnA 1 activity, we have shown in the past that GlnA 1 activity is directly stimulated by the cellular metabolite 2-OG [32]. Due to severe reduction in glutamate dehydrogenase activity under N limitation, the cellular concentrations of 2-OG increase drastically; consequently, the cellular 2-OG concentration is considered to reflect the internal signal for N starvation [32]. Perceiving the signal for N starvation as a result of increasing intracellular 2-OG concentrations is also proposed for other autotrophically growing microorganisms as reported, for example, for cyanobacteria [34,42,43], and has been demonstrated for the post-translational modulation of nitrogenase activity by PII-like proteins (NifJ 1 and NifJ 2 ) in response to N availability (switch on, switch off mechanism by direct protein interaction) in Methanococcus maripaludis [44,45]. As known for other methanoarchaea, the cellular 2-OG concentration in M. mazei also mediates the N status to the transcriptional regulatory machinery. Under N limitation and resulting high internal 2-OG concentration, 2-OG binds to the global repressor NrpR, significantly lowering the binding affinity of NrpR to its respective operator. Consequently, NrpR leaves the operator and transcription can be initiated by RNA polymerase [29]. This further emphasizes the central role of 2-OG in sensing and transmitting the internal N status. In addition, a PII-like protein, GlnK 1 , allows   (20 nM) and untagged sp26 varied between 0.107 nM and 23 µM; estimated K D (sP26): 2.9 µM AE 0.9 µM (n = 3). (D) Purified and labeled His 6 -GlnK 1 * was applied to purified His 6 -GlnA 1 whose concentration varied between 1.7 nM and 55.5 µM (calculated based on the monomeric MW). Calculated K D (His 6 -GlnA 1 ): 1.5 µM AE 0.7 µM (with n = 2, error bars represent the standard deviation of 1 biological experiment with 2 technical replicates). (E) Purified labeled His 6 -GlnA 1 * (20 nM) was analyzed with purified His 6 -GlnK 1 with a concentration range from 6 nM to 105 µM. Calculated K D (His 6 -GlnK 1 ): 17 µM AE 3 µM (n = 3, error bars represent the standard deviation of 1 biological experiment with three technical replicates). In general (A, B, C, D, E), out of several biological replicates, one respective biological replicate is exemplarily shown in panel 1-based on two or three technical replicates. (F, G) Glutamine synthetase activity of His 6 -GlnA 1 (0.95 µM) determined as described in Fig. 7 pre-incubated at RT for 5 min, with His 6 -sP26 (2.9 µM), His 6 -GlnK 1 (0.65 µM), and His 6 -sP26 (2.9 µM) plus His 6 -GlnK 1 (0.65 µM) in the absence (F) or presence of 5 mM 2-OG (G) (generally, concentrations calculated based on the monomeric MW). Depicted are exemplarily data from one representative purification batch of His 6 -GlnA 1 , His 6 -sP26, and His 6 -GlnK 1 (standard deviation of three technical replicates). The interaction was confirmed in a second biological replicate (including three technical replicates) (Table S1). fine-tuning control of the glutamine synthetase activity under changing N availabilities and was predicted to inhibit GlnA 1 activity due to an ammonium upshift after a period of N limitation [32].
Realizing that a wealth of small proteins exists in M. mazei identified by systematic global screens for translated small proteins [7,8,19], now unraveled that GlnA 1 regulation in M. mazei is even more complex and argues for a mechanistically novel post-translational regulation by a small protein. Using different approaches, we obtained conclusive experimental evidence that the small protein sP26 (molecular mass 2.7 kDa), which is upregulated under N limitation, interacts with GlnA 1 in M. mazei and simultaneously increases its enzyme activity independently and in addition to the stimulating metabolite 2-OG and modulating GlnK 1 . (a) Pull-down experiments demonstrated that purified M. mazei SUMO-His 6 -sP26 directly interacts with GlnA 1 in cell extracts, which was verified by a reverse pull-down, demonstrating coelution of SUMO-His 6 -sP26 when purifying Strep-GlnA 1 (Fig. 4). (b) High binding affinity between sP26 and GlnA 1 was observed using MST analysis (including reverse labeling, see Fig. 5). This specific binding was not depending or affected by 2-OG. (c) NMR analysis of sP26 in the presence and absence of GlnA 1 identified seven amino acids specifically interacting with GlnA 1 (Fig. 6). (d) sP26 significantly enhances glutamine synthetase activity (Fig. 7). The EC50 for sP26 in respect to GlnA 1 activity enhancement was calculated to be 0.43 µM, which is in the same range as the K D determined for the untagged sP26. (e) This activity stimulation was independent and additive to the 2-OG activation and GlnK 1 modulation (Fig. 9F). (vi) First SEC analysis provided evidence toward close association of GlnA 1 , GlnK 1 , and sP26 in the presence of 2-OG (Fig. 10) indicating a quaternary structure.
Interestingly, in cyanobacteria, where 2-OG is also reflecting the internal N status, small proteins modulating GS activity have been reported several years ago [46,47]. However, these proteins are in the range of 7 and 17 kDa and have been demonstrated to inhibit GS activity by protein-protein interaction due to an ammonium upshift after N starvation [46]. Those socalled inactivating factors (IF7 and IF17) are expressed under N sufficiency combining transcriptional regulation but also post-transcriptional regulation by regulatory RNAs, an antisense RNA (NsiR4) in case of IF7 [48] and a recently discovered glutamine-binding riboswitch in case of IF17 [49]. As already observed by   [32], the PII protein GlnK 1 is able to modify GlnA 1 activity. Direct protein interaction has been exclusively shown by pull-down approaches without quantification. Here, we succeeded to demonstrate by MST analysis that GlnK 1 binds to GlnA 1 with high affinity and further excluded that the interaction is dependent on 2-OG. In contrast however, to the previous report [32] we obtained strong evidence that GlnK 1 is activating GlnA 1 activity when the enzyme activity is determined in a HEPES buffer-based activity assay (Fig. 9F,G). The observation of contradictory effects of GlnK 1 on GlnA 1 activity depending on the buffer systems suggests that the buffer system is highly influencing/effecting GlnA 1 activity. We can only speculate that the different buffer systems, salt, and/or differences in the pK S might affect the interaction between GlnA 1 and GlnK 1 , and/or the oligomeric conformation of GlnA 1 . However, since in all buffers tested except MOPS buffer used in the 2005 [32] GlnA 1 activity was enhanced by GlnK 1 presence, we conclude that GlnK 1 is indeed activating GlnA 1 activity. We clearly demonstrated that sP26 also binds to GlnK 1 with high affinity and in a 2-OG independent manner (Fig. 9A-C). Thus, in future it will be important to elucidate the structural changes in GlnA 1 in the presence of GlnK 1 and sP26 (e.g., by NMR analysis), as well as the quaternary complex formation between all three proteins in the presence and absence of 2-OG in more detail and determine the respective ratio as well as the respective oligomeric conformation of GlnA 1 .

Hypothetical model for post-translational regulation of GlnA 1 by sP26
Since strong interactions between sP26 and GlnA 1 have been observed, and it is highly unlikely that sP26 displays a catalytic activity for, for example, covalent modification of GlnA 1 , we propose that the obtained sP26-dependent induction of GlnA 1 activity is most likely due to effects on GlnA 1 oligomeric structure or stability. Moreover, sP26 might in addition directly or indirectly effect GlnA 1 interaction with GOGAT. Potential effects on the GlnA 1 /GOGAT interaction by sP26 are particularly attractive, since we have detected peptides derived from both large subunits of GOGAT in several sP26 pull-down experiments in the LC-MS/MS analysis.
Based on our findings, we hypothesize the following model depicted in Fig. 11. In the presence of 2-OG, GlnA 1 nearly exclusively expressed under N starvation forms higher oligomeric structures (dodecamers) resulting in GlnA 1 activation. However, the dodecamers are not very stable. Only in the additional presence and upon interaction with sP26, which is induced under N starvation, GlnA 1 dodecamers change into a more stable (tighter) complex and consequently induce GlnA 1 activity in addition to 2-OG. GlnK 1 only expressed under N limitation binds to GlnA 1 most likely mediated by interacting with sp26 resulting in further stabilization of the dodecamer. In vivo, sP26 is always present under N limitation; however, in vitro GlnK 1 was shown to bind to GlnA 1 and stimulate its activity also in the absence of sP26. The first complex analysis by size-exclusion chromatography indicates quaternary complexes of all three proteins are formed only in the presence of 2-OG. Thus, in future studies the quaternary complexes as well as the hypothesized change in the dodecameric structure (most likely two hexameric ring structure face to face) have to be proven, for example, by NMR analysis and in combination with hydrogen-deuterium exchange mass spectrometry (HDX-MS) to identify the interaction sites of all proteins at the protein surfaces. Moreover, since sP26 is highly conserved in the Methanosarcinas, it is attractive to speculate that for other methanoarchaeal glutamine synthetases, modulations by the sP26 homologs might occur.

Strains and plasmids
All plasmids used have been constructed as follows, confirmed by sequencing and are summarized in Table S2. pRS1242 was generated for overproducing sP26 N-terminally fused to a (His) 6 -tag (His 6 -sP26) in M. mazei. A DNA fragment including the respective gene was commercially synthesized (Eurofins Scientific, Nantes, Luxemburg) as follows: The gene was placed under the control of the promoter and the ribosome-binding site of mcrB [50], followed by an additional ATG as new translation start and sORF26 with an N-terminal His 6 -Tag. Downstream of the sORF26 sequence the methanoarchaeal transcriptional terminator was added. Further, a flanking SacI restriction site (5 0 ) and a KpnI restriction site (3 0 ) were added resulting in a 379-bp fragment. The synthesized DNA fragment was inserted into pEX (Eurofins Scientific), the corresponding plasmid designated pRS1209 (Fig. S2a). Using SacI and KpnI (NEB, Schwalbach, Germany), the respective 373-bp fragment was generated and ligated into the SacI and KpnI-restricted pWM321 [51] resulting in plasmid pRS1242 (Fig. S2d).
pRS1228 was constructed by amplifying sORF26 from pRS1209 using primers sORF26_1 rev (AAAAAATTAAG-GATAATTTCCGTGCCTC) and sORF26_1 for (GTG CCTGTGATGAAAAATCTGGCTG) and TA cloning the respective fragment into pET-SUMO (Invitrogen) vector. In contrast to pRS1229, the respective construct pRS228 provides no additional His 6 -tag between SUMO and sP26 (Fig. S2d).
pRS375 was constructed by amplifying glnA 1 from pRS196 [32] using a forward primer replacing the N-terminal (His) 6 -tag by a Streptag (Mm GlnA H/S for 5 0 -CCATGGGCTGGAGCCACCCGCAGTTCGAAAAAAG CAGCGGCCTGGTGCCGCGC-3 0 ) and adding a NcoI restriction site and the reverse primer (Mm GlnA H/S rev 5 0 -CCGCCGCGCCCGCCGCCCGCCGCAAGCTTGTC-GACGGAGCTCGAATTCGGATCC-3 0 ) adding a BamHI restriction site. The obtained 1.4 kbp PCR product was cloned into pRS196 using NcoI and BamHI restriction sites. The resulting plasmid was designated pRS375.

Growth analysis of M. mazei
All M. mazei strains were grown at 37°C in 50 mL or 1-L enclosed serum bottles under anaerobic conditions, with the gaseous phase containing N 2 and CO 2 (80/20) (Air Liquide, Paris, France). The minimal medium was supplemented with 150 mM methanol and 40 mM acetate as carbon and energy sources, as described in Veit et al. (2006) [54]. For growth under N limitation, the N 2 from the gas atmosphere served as the sole N source, and for N sufficiency, the medium was supplemented with 10 mM NH 4 Cl. For different stress conditions, the cultures were grown under various conditions, at 30°C, at 42°C, under oxygen stress (20 mL sterile air was added to the gas atmosphere at a turbidity of 0.25), and under salt stress (additional 500 mM NaCl). Growth was generally monitored by determining the turbidity of the cultures at 600 nm.
The Illumina sequencing data were obtained from M. mazei cultures grown under N starvation and N sufficiency (15 mM) which were harvested in the exponential growth phase. RNA isolation was performed as described above, and RNAseq was performed as described by Prasse et al. (2017) [28].

Localization of sP26 in M. mazei cells
Methanosarcina mazei was grown anaerobically in complex medium under N sufficiency or limitation, as described in Ehlers et al 2005 [53]. The optical turbidity was measured at 600 nm with spectrophotometer DU 640 (Beckmann Coulter, Krefeld, Germany). At mid-exponential growth phase, cells were harvested at 2831 g and 4°C for 30 min, resuspended in 50 mM Tris (pH 6.8), and disrupted using a dismembrator (3 min at 1600 r.p.m.). Following centrifugation for 30 min at 15 700 g and 4°C, the supernatant (soluble fraction) and pellet (insoluble fraction) were used for western blot analysis using antibodies directed against the synthetic sP26.
Purification of SUMO-sP26 fusion proteins and sP26, and co-elution analysis SUMO-His 6 -sP26 was purified from 1 L culture of E. coli BL21 (DE3) carrying the plasmid pRS1229 grown in LB at 37°C. At a turbidity of 0.6 at 600 nm, the culture was supplemented with 100 µM IPTG and further incubated for 2 h at 37°C with rigorous shaking. After harvesting (4000 g at 4°C for 30 min) and resuspending in buffer A (50 mM NaH 2 PO 4 , 300 mM NaCl, pH 8), the cells were disrupted using a French pressure cell two times (at 4.135 9 10 6 NÁm À2 ) followed by centrifugation at 13 865 g at 4°C for 30 min. The cytosolic supernatant was incubated and stirred with 1 mL Ni-NTA agarose (Qiagen, Hilden, Germany) at 4°C for 1 h. After pouring the slurry in an empty column, it was washed twice with 8 mL buffer A containing 20 mM imidazole. The SUMO-His 6 -sP26 protein was eluted with 5x 1 mL buffer A containing 100, 250, and 500 mM imidazole. Elution fractions containing the SUMO-His 6 -sP26 protein (250 mM imidazole) were combined and dialyzed overnight against buffer A. SUMO-sP26 was purified from 1 L culture of E. coli BL21 (DE3) carrying the plasmid pRS1228 as described above. The SUMO domain from the fusion protein was removed by SUMO protease (Thermo Fisher Scientific TM , Waltham, USA, 10 U per 20 µg fusion protein) using buffer B (50 mM HEPES, 300 mM NaCl, pH 7.5) and the remaining tag-less sP26 separated from the SUMO-tag as well as the protease by Co-NTA chromatography (Thermo Fisher Scientific TM , Waltham, USA).
For co-elution experiments, 1 mg of purified SUMO-His 6 -sP26 protein was incubated with 500 µL M. mazei crude cell extract grown under N limitation (27.5 mg) and immobilized to 0.5 mL Ni-NTA agarose for 1 h at 4°C. All nonbinding proteins were washed off with 5 9 1 mL buffer A and the SUMO-His 6 -sP26 and potentially interacting proteins eluted with 4 9 1 mL buffer A supplemented with 250 mM imidazole. The elution fractions were analyzed with Coomassie-and silver-stained SDS/PAGE and Tris/Tricine/PAGE (see Gel electrophoretic separation). The respective M. mazei crude cell extract was generated as follows: 1 L M. mazei culture grown exponentially under N limitation was harvested at 4000 g and 4°C for 30 min and the cell pellet resuspended in 500 µL 50 mM Tris/HCl pH 6.8 supplemented with DNaseI (Thermo Fisher Scientific TM , Waltham). Cells were twice disrupted by GinoGrinder (SPEX CertiPrep, Metuchen, NJ, USA) on ice at 1300 strokes for 3 min each, followed by centrifugation at 13 865 g and 4°C for 30 min to separate cell debris.

Identification of co-eluting proteins by LC-MSMS analysis
Coomassie blue (R-250)-stained gel bands were excised and destained. The individual bands were reduced with dithiothreitol (10 mM, 56°C, 1 h) and subsequently alkylated in the presence of iodoacetamide (50 mM, room temperature, 30 min). Enzymatic digestion of was performed overnight at 37°C by the addition of sequencing grade trypsin (Promega, Madison, WI, USA) (100 ng per sample) in 100 µL of ammonium bicarbonate (ABC) buffer (100 mM pH 7.4). The peptides were extracted via one wash with 60% acetonitrile with 0.1% trifluoroacetic acid and a second wash with 100% acetonitrile. The peptides were then dried via vacuum evaporation prior to LC-MS analysis. For LC-MS analysis, the samples were suspended in UHPLC loading buffer (3% acetonitrile + 0.1% trifluoroacetic acid).
Chromatographic separation was performed on a Dionex U3000 UHPLC system (Thermo Fisher Scientific, Darmstadt, Germany) equipped with an Acclaim PepMap 100 column (3 lm particle size, 75 lm 9 150 mm) and µ-precolumn (300 lm 9 5 mm) coupled online to LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific). The eluents used were as follows: eluent A: 0.05% formic acid (FA) and eluent B: 80% ACN + 0.05% FA. The separation was performed over a programmed 90-min run. Initial chromatographic conditions were 5% B for 5 min followed by an increase to 10% B over 1 min, subsequently a linear gradient from 10% to 40% B over 54 min, a 5-min increase to 95% B, and 10 min at 95% B. Following this, an interrun equilibration of the column was achieved by 15 min at 5% B. A constant flow rate of 300 nLÁmin À1 was used, and 8 lL of sample was injected per run. Data acquisition on the LTQ Orbitrap Velos mass spectrometer utilized CID activation (NCE 35). A full scan MS acquisition was performed (resolution 60 000) scan range 300-1500 m/z, maximum IT 100 ms. Subsequent data-dependent MS/MS (resolution 7500, minimum intensity 500), of the top 10 most intense ions, single charged, and undetermined charged state ions were excluded, and dynamic exclusion was enabled (90-s duration, repeat count 2 in 30 s); internal lockmass was enabled on 445.12003 m/z.
MS data files were searched against a FASTA database containing the full M. mazei proteome (accessed from Uni-Prot 2016.03.16) plus predicted sORF encoded proteins [27,33], and the cRAP list of commonly occurring laboratory contaminants (version 1.0, 2012.01.01). The searches were performed using the Proteome Discoverer software package (version 1.4.0.288) using the SequestHT search algorithm. A tryptic search was performed (precursor tolerance 10 p.p.m., fragment tolerance 0.04 Da, missed cleavages 2); Variable protein modification: oxidation of methionine and fixed modification: cysteine carbamidomethylation. Strict parsimony criteria were applied with a target FDR of 0.01 (1%) applied at peptide level. In addition, proteins required two high-confident peptides to be considered as identified. harvested, resuspended in 6 mL W-buffer (100 mM Tris, pH 8.0; 150 mM NaCl; 1 mM EDTA), and disrupted using a French pressure cell at 4135 x 10 6 NÁm À2 , followed by centrifugation at 20 000 g for 20 min. The cell-free crude extract was incubated with 1 mL W-buffer equilibrated Strep-TactinÒ sepharoseÒ matrix (IBA, G€ ottingen, Germany) incubated for 30 min at 4°C under slightly swivel. The column was washed with 10 mL W-buffer. Strep-GlnA 1 was eluted in the presence of 2.5 mM D-Desthiobiotin (IBA) followed by buffer exchange using an Amicon Centrifugal Filter (30 KDa) (Merck) according to the manufacturer's instructions. To prepare cell-free crude extract containing SUMO-His 6 -sP26, E. coli BL21 pRIL / pRS1229 was grown in 1 L LB at 37°C to a turbidity at 600 nm of 0.6 and the expression of SUMO-His 6 -sP26 was induced by adding 100 µM IPTG, followed by further incubation for 2 h. Cells were harvested, and cell-free extract was prepared as described above. Approximately 50 mg of crude extract containing SUMO-His 6 -sP26 was mixed with 1 mg purified Strep-GlnA 1 and incubated with 1 mL Strep-TactinÒ sepharoseÒ matrix (IBA) for 30 min at 4°C under slightly swivel. Nonbinding proteins were subsequently washed from the column with 10 mL W-buffer. Strep-GlnA 1 and potential interacting proteins were eluted in the presence of 2.5 mM D-Desthiobiotin (IBA). Aliquots of wash and elution fractions of the reverse cochromatography were separated by SDS/PAGE according to Laemmli (1970), and the elution fractions were further analyzed by western blot analysis as described in Weidenbach et al. [30] using commercial antibodies directed against the His-tag following the instructions of the manufacture (Qiagen).
Purification of His 6 -GlnK 1 , His 6 -GlnA 1 , and His 6 -sP26 One litre of the respective E. coli BL21 / pRIL strains carrying pRS203 (His 6 -GlnK 1 ), pRS196 (His 6 -GlnK 1 ), or pRS1245 (His 6 -sP26) was grown in LB medium at 37°C under rigorous shaking to a turbidity of 0.7 at 600 nm. The cultures were cooled down to 18°C followed by supplementing with 10 µM IPTG and further incubation at 18°C for~18 h. The cells were harvested at 4000 g and 4°C for 20 min and resuspended in 20 mL buffer B (50 mM HEPES, 300 mM NaCl, pH 7.5). Cytosolic extracts were generated as described above and the His 6 -tagged proteins purified by metal affinity chromatography and gravity flow using 0.5 mL HisPur TM cobalt superflow agarose (Thermo Fisher Scientific TM , Waltham) according to manufacturer's protocol, eluting proteins at 50 mM imidazole. The imidazole of the elution fraction was removed until the concentration of imidazole was below 1 mM using Ami-conÒ centrifugal filters (0.5 mL size, GlnA 1 : 30 kDa and GlnK 1 : 10 kDa) (Merck KGaA). His 6 -sP26 was purified as described above with the following additional steps: The generated cytosolic extract was fractionated by (NH 4 ) 2 SO 4 precipitation prior the metal affinity chromatography. His 6 -sP26 precipitated at~40% to 45% ammonium sulfate. The cytosolic extract was supplemented with 30% (NH 4 ) 2 SO 4 (v/v) and incubated under stirring for 1 h at 8°C. After centrifugation at 13.865 x g and 4°C for 45 min, the respective supernatant was supplemented with additional 15% ammonium sulfate and processed as described. The generated second pellet (precipitated His 6 -sP26) was resuspended in 20 mL Buffer B and dialyzed overnight against buffer B. The protein was purified using HisPur TM cobalt superflow agarose (0.5 mL) (Thermo Fisher Scientific TM , Waltham). Finally, the His 6 -sP26 protein was separated from contaminating larger proteins using size-exclusion 10 kDa AmiconÒ centrifugal filters (0.5 mL size) and the buffer was exchanged to buffer B without imidazole using 3 kDa AmiconÒ centrifugal filters (0.5 mL size) (both from Merck KGaA).
Due to the lack of aromatic amino acids within the small protein His 6 -sP26, the Pierce TM BCA protein assay kit (Thermo Fisher Scientific TM , Waltham) was used to determine the concentration of all proteins and was performed according to manufacturer's protocol. Absorbance was measured at 562 nm by the spectrophotometer Ultrospec 2100 UV/VIS (GE Healthcare TM , Chicago, IL, USA). The concentration was calculated by using a BSA standard (50-2000 µgÁmL À1 ).
Interaction and affinity analysis by microscale thermophoresis (MST) using Monolith NT.115 labeling of 0.95 or 0.87 was achieved, respectively, and for analysis, 20 nM labeled protein, supplemented with 0.05% Tween 20 and 1 mgÁmL À1 BSA, was applied to nM to µM concentrations of nonlabeled putative interaction partners. The measurements were performed using standard treated capillaries, an excitation power of 20%, high MST-power, and MST-on of 15 s. In general, the proteins were incubated at RT for 5 min prior to loading the capillaries. K D values were calculated using the Nanotemper tool MO. Affinity Analysis. Data of three independent pipetted measurements were analyzed (MO. Affinity Analysis software version 2.3, NanoTemper Technology).

Determination of glutamine synthetase activity
Glutamine synthetase (GS) activity was determined by using the coupled optical test assay described by Shapiro [36], which couples the consumption of ATP by the conversion of ammonium and glutamate to glutamine catalyzed by glutamine synthetase to the oxidation of NADH by lactate dehydrogenase. Conformation analysis using Tycho NT.6 His 6 -GlnA 1 was analyzed using Tycho NT.6 (NanoTemper) by applying a standard capillary (10 µL) with 1.5 mgÁmL À1 enzyme in buffer B. Thermal unfolding profiles were recorded within a temperature gradient between 35 and 95°C. In case the assays were performed with the addition of 2-OG (2.5, 5, 7.5, 10 and 12.5 mM), it was supplemented, respectively, 5 min before start.

Complex analysis using size-exclusion chromatography
All proteins were heterologous expressed in E. coli and purified as described above: 168.75 µg His 6 -GlnA 1 , 77.5 µg His 6 -GlnK 1 , and 24 µg His 6 -sP26 in different combinations (His 6 -GlnA 1 ; His 6 -GlnA 1 + His 6 -sP26; His 6 -GlnA 1 + His 6 -GlnK 1 ; His 6 -GlnA 1 + His 6 -GlnK 1 His 6 -sP26) were incubated together at RT for 5 min in the presence and absence of 5 mM 2-OG prior to loading on the analytical ENrich 650 column (Bio-Rad Laboratories, Inc., Hercules, USA). The column was equilibrated with 50 mM Tris/HCl and 150 mM NaCl and supplemented with 5 mM 2-OG when the complexes were pre-incubated in the presence of 5 mM 2-OG and the applied protein isocratic eluted with the respective buffer with a flow rate of 1 mLÁmin À1 . One millilitre fractions were collected and concentrated by TCA precipitation (15% v/v). The pellets were resuspended in 50 mM ammonium acetate for LC-MS/MS analysis.

Proteomic analysis of SEC fractions
SEC fractions were precipitated before being suspended in 100 mM ammonium acetate. Aliquots (10 µL) from each fraction were made up to a volume of 60 µL with 100 mM ABC buffer (pH 7.4). The proteins were reduced with dithiothreitol (10 mM, 56°C, 1 h) and subsequently alkylated in the presence of chloroacetamide (50 mM, room temperature, 30 min). Enzymatic digestion was performed overnight at 37°C by the addition of sequencing grade trypsin (Promega) (100 ng per sample) in 100 µL of ABC buffer. The samples were acidified via the addition of 10 µL of 10% TFA to stop the digestion before being dried down under vacuum (Concentrator plus; Eppendorf, Hamburg, Germany). The dried peptides were stored at À20°C prior to LC-MS analysis. On the day of LC-MS analysis, the samples were suspended in 15 µL of UHPLC loading buffer (3% acetonitrile + 0.1% trifluoroacetic acid). Chromatographic separation was performed on a Dionex U3000 UHPLC system (Thermo Fisher Scientific, Darmstadt) equipped with an Acclaim PepMap 100 column (2 lm particle size, 75 lm 9 500 mm) and µ-precolumn (300 lm 9 5 mm) coupled online to Fusion Lumos mass spectrometer (Thermo Fisher scientific). The eluents used were as follows: eluent A: 0.05% formic acid (FA) and eluent B: 80% ACN + 0.05% FA. The separation was performed over a programmed 60-min run. Initial chromatographic conditions were 4% B for 3 min followed by linear gradients from 4% to 50% B over 30 min, a 1min increase to 90% B, and 10 min at 90% B. Following this, an inter-run equilibration of the column was achieved by 15 min at 4% B. A constant flow rate of 300 nLÁmin À1 was used, and 1 lL of sample was injected per run. Data acquisition on the Fusion Lumos mass spectrometer utilized HCD activation (NCE 30). A full scan MS acquisition was performed (resolution 120 000) scan range 300-1500 m/z, AGC target 4e5, maximum IT 50 ms. In a subsequent data-dependent MS/MS (resolution 30 000, minimum intensity 5e4, maximum IT 100 ms) of the most intense ions for the following 3 s; the MIPS mode was activated for peptides, single charged and > 7+ charged peptides were excluded and dynamic exclusion was activated (60-s duration). The internal barrier mass was activated at 445.12003 m/z.
MS data files were searched against a combined set of fasta databases containing the full M. mazei proteome plus all predicted sORF encoded proteins (PTK_92x_Mmazei_Full_Plus_SEP_190425.fasta), and the His-tag protein sequences provided for sp26 and GlnA 1 (up26HisTagged + GlnAHis.fasta). In addition, protein sequences for the expression system, E. coli D3, and the cRAP list of common laboratory contaminants were included for all searches (PTK_proteome_Ecoli.fasta and cRAP47.fasta). The searches were performed using the Proteome Discoverer software package (version 2.2.0.388) using the SequestHT search algorithm. A semi-tryptic search was performed (Precursor tolerance 10 p.p.m., fragment tolerance 0.02 Da, missed cleavages 2) with variable modifications; methionine oxidation and protein N-terminal acetylation, fixed modification; cysteine carbamidomethylation.
Strict parsimony criteria have been applied with a target FDR of 0.01 (1%) applied at the PSM, peptide, and protein level. In addition, proteins required two high-confident peptides to be considered identified, or in the case of sP26, single peptides were manually validated to assure quality of the identification.

Protein production for NMR analyses
Plasmid pRS1228 carrying the SUMO-tagged sP26 sequence was transformed into E. coli BL21 cells and heterologously expressed. The necessary 15 N-labeling scheme for NMR spectroscopy was performed by using M9 minimal medium containing 15 N-enriched NH 4 Cl (Cambridge Isotope Laboratories, Cambridge, MA, USA). The transformed cells were incubated at 37°C until a turbidity at 600 nm of 0.6 was reached and protein expression was subsequently induced by addition of 1 mM IPTG. The cell cultures were further incubated at 18°C overnight. The cells were harvested by centrifugation (5000 r.p.m., 15 min, 4°C) and resuspended in purification buffer (50 mM HEPES pH 7, 300 mM NaCl) supplemented with one EDTA-free protease inhibitor tablet (cOmplete TM , Roche, Germany).
Protein purification was accomplished by tandem Co-NTA chromatography. Cleavage of the SUMO-tag was performed with SUMO protease at 20°C overnight. Purity of the protein was confirmed by SDS/PAGE and mass spectrometry (MALDI).

NMR spectroscopic analysis
NMR samples were prepared in NMR buffer containing 50 mM HEPES pH 7, 300 mM NaCl, 1 mM DSS, and 10% D 2 O. Spectra were recorded on a 600 MHz Bruker spectrometer at 298 K. Referencing was done by setting the DSS signal to 0.00 p.p.m.; referencing of the indirect dimension was calculated according to the 1 H referencing [58]. Spectra were processed and analyzed using TopSpin 3.5pl7 (Bruker Biospin, Ettlingen, Germany).

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig, S1. Annotated spectra for the proteotypic peptide. Fig. S2. Schematic plasmid maps of expression vectors. Table S1. GlnA 1 activity determined as described in Methods using purified proteins of independent purifications. Table S2. Strains and plasmids used.