Does DarA interact with the glutamate synthase?

Currently, the most promising target for an interaction with DarA is the glutamate synthase GltAB (GOGAT). Two independently acquired mutations in the transcriptional regulatorgltC were identified but surprisingly no mutation in the gltAB operon itself. It is not entirely clear why this is the case but this might be due to the small sample size of the analyzed suppressors.

GltAB was identified in our bioinformatic analysis where we searched for putative interaction partners of DarA that fit rational criteria. As a reminder, these criteria were conservation among related species, cytosolic localization, promising structure and less abundance in relation to the putative regulator DarA. In B. subtilis, glutamate is synthesized by the glutamate synthase GltAB in a reductive, NADPH-dependent reaction out of glutamine and the TCA cycle intermediate 2-oxoglutarate (Gunka and Commichau, 2012). Subsequently, glutamate and acetyl coenzyme A (acetyl-CoA) are used as a starting point for the arginine biosynthesis (Cuninet al., 1986). A stimulating interaction of DarA with GltAB would increase the available

glutamate which then can be fed into the arginine biosynthesis.

An interaction is highly likely from a structural point of view. Inspired by our experimental and bioinformatic results parallel work by Richts (2018) nicely showed how perfectly the structures of DarA and GltAB fit together. Structural compatibility was visualized by molecular docking of theB. subtilis DarA trimer to the GltAB homologue from Azospirillum brasilense in its enzymatically active, dodecameric form. The DarA homotrimer exhibits a tripod-like tertiary structure formed by the unusually long B-loops that protrude from the protein core. The tips are negatively charged, 40 Å apart and nicely fit into three positively charged depressions that are formed by the big subunit GltA. These are located in an 60 Å central cavity of the GltAB holoenzyme. In this model two DarA trimers would sandwich the dodecameric GltAB holoenzyme by binding into two cavities formed by two GltA trimers (Richts, 2018). This proposed mechanism resembles the stimulating interaction of PII with the NAGK inS. elongatus (Llácer et al., 2007). Availability of K⁺ is reported by alterations in the c-di-AMP pool which can be integrated by DarA since binding of the ligand leads to changes in the orientations of the B-loops (Gundlach et al., 2015a). Consequently, depending on the binding state of DarA a putative interaction with GltA(B) could easily be either facilitated or diminished. Nevertheless, the structural indications should be viewed critically.

The Azospirillum GOGAT structure was only resolved at a resolution of 9.5 Å and was acquired by a combination of three-dimensional cryoelectron microscopy, small angle x-ray scattering and bioinformatic modeling with a structural similar dehydrogenase as a template (Cottevieille et al., 2008).

The transcriptional control of the GOGAT has been studied extensively in the past but, to our knowledge, post-translational control of a bacterial glutamate synthase has not been reported so far (Gunka and Commichau, 2012). However, post-translational regulation of an eukaryotic GOGAT has been reported recently for the model plant Arabidopsis thaliana (Takabayashi et al., 2016). The bacterial GOGAT found in E. coli orB. subtilis is a

NADPH-dependent type of glutamate synthase (NADPH-GOGAT). In addition, two other types exist in nature which either use NADH (NADH-GOGAT) or reduced ferredoxin (Fd-GOGAT) as the electron donor. The NADH-GOGAT can be found in yeast, lower eukaryotes and in the nonphotosynthetic tissues of plants, while the Fd-GOGAT is found in cyanobacteria, algae and in the photosynthetic tissue of plants (Suzuki and Knaff, 2005). ACR11, an ACT-domain-containing protein (acronym for aspartate kinase, chorismate mutase, and TyrA),

directly interacts with the Fd-GOGAT inA. thaliana. ACR11 seems to stabilize the enzyme complex which is reflected by reduced Fd-GOGAT activity in anacr11 mutant (Takabayashi et al., 2016). Interestingly, ACR11 also increases the activity of the glutamine synthetase by direct binding and thus seems to balance the GS/GOGAT activity (Osanaiet al., 2017).

Considering the structural and mechanistic similarities between the GOGAT-types and the significant sequence similarity (>40 %) of the Fd-GOGAT to the big subunit GltA of the NADPH-GOGAT, GltAB might also be regulated on a post-translational level (Cottevieille et al., 2008; Suzuki and Knaff, 2005). Although,gltABexpression is slightly reduced at low K⁺ concentrations, GS/GOGAT activity is not influenced by K⁺ (Krüger, unpublished; Measures, 1975). Consequently, it would be reasonable for DarA to sense K⁺ limitation depending on its c-di-AMP binding status and to adjust the activity of GltAB accordingly. Our results indicate this but unambiguous proof for an interaction of DarA with GltAB is still pending.

Up to now, several different experiments were not able to verify an interaction of DarA with GltAB. However, most of them were either not entirely functional or the negative results have to be interpreted carefully. Previous GOGAT activity assays with differentB. subtilis cell-free crude extracts were inconclusive (Richts, 2018). However, the extracts were not from aktrAB kimA ahrC deletion background which might explain the negative results since the interaction seems unnecessary under these conditions and DarA is still bound to c-di-AMP which likely inhibits an interaction (see Section 4.3.2). On the one hand, the GOGAT activity assay with purified proteins also suggests that c-di-AMP-un-/bound DarA does not influence the activity of GltABin vitro (see Figure 3.14). On the other hand, thein vitro assay surely does not reflect the native cellular conditions like substrate and cofactor concentrations and unknown factors might even be missing. Furthermore, the assay relies on the oxidation of the cofactor NADPH which occurs at the NADPH site of the small subunit GltB. Consequently, binding of DarA to the big subunit GltA might not hinder the oxidation of NADPH but only affect the reaction at the glutaminase site or the synthase site, where glutamine or 2-oxoglutarate are converted to glutamate, respectively. Both are located in GltA (Agnelli et al., 2005; Vanoniet al., 1996). The pull-down assay was also not entirely functional since DarA did bind unspecifically to the matrix. A BACTH assay with DarA and GltA conducted by Fülleborn (2018) was not conclusive. However, the other subunit GltB was not present for complex formation resulting in only little self-interaction of GltA. The most promising experimental evidence is the SPINE from Fülleborn (2018) since he was able to reproducibly detect the big subunit GltA in a cross-linked SPINE elution fraction. However, this cannot be taken as a definite proof since suitable controls were missing. A perfect control for this does not exist since a strain lacking DarA is not able to grow under this condition and a strain expressing an untagged DarA leads to partial unspecific binding of DarA to the matrix. The SPINE fractions showed a lot of background which might indicate a false positive binding of GltA. This could be excluded by cultivation of the strain with high amounts of K⁺ and using this as a methodological control.

Since GltAB is the most tempting target candidate for DarA, additional experimental setups should be pursued. For example, Förster resonance energy transfer (FRET) can be used as anin vitroor as anin vivoapproach. For this purpose, DarA and GltA have to be genetically fused to derivatives of the cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) which in pair function as a donor and acceptor fluorophor, respectively (Miyawakiet al., 1997).

FRET has already been successfully used to show an interaction of the PIIprotein GlnB with the NAGK from S. elongatus (Lüddecke and Forchhammer, 2013). In addition, ITC with the purified proteins can be used to detect a possible interactionin vitro without relying on NADPH oxidation. Purification of functional GltAB was demonstrated in this work starting with simultaneous overexpression of His6-taggedgltB and untaggedgltA using a two-plasmid approach with subsequent co-purification (see Figure 3.12). For illustration, the localization of GltAB within glutamate metabolism as well as the connection of glutamate to the TCA cycle and to the arginine biosynthesis are depicted in Figure 4.1.

Acetyl-CoA

Figure 4.1: Connection of central carbon and glutamate metabolism and overview of arginine biosynthesis. (A) Localization of GltAB within glutamate metabolism and connection to the TCA cycle.

(B)Arginine biosynthesis. Red arrows indicate glutamate dependent reactions. In both representations not all enzymes, steps, substrates and products are shown to ensure a good overview.