Potassium and c-di-AMP amounts in a darA mutant

The cytosolic localization of DarA under several conditions strongly suggested that DarA does not regulate a (potassium) transporter by direct binding. Nevertheless, supplementation with higher amounts of KCl counteracted the inhibitory effect of DarA on growth in MSSM medium with ammonium as the nitrogen source and low amounts of potassium. It might be that DarA integrates the c-di-AMP signal to affect potassium homeostasis in another way under this condition, which had escaped detection before. Since c-di-AMP levels correspond to the potassium concentration to adjust import of K⁺ accordingly we also wondered whether DarA affects the levels of c-di-AMP (Gundlach, 2017; Gundlach et al., 2017b). It should be noted that direct modulation of the cellular c-di-AMP homeostasis is not common for c-di-AMP targets. However, the c-di-AMP-binding TrkH family protein CabP was recently described to alter c-di-AMP levels in Streptococcus pneumoniae in a K⁺ independent manner (Zarrella et al., 2018).

To investigate the above hypotheses, the intracellular amounts of K⁺ and c-di-AMP were determined from cells grown in the minimal medium where absence of DarA is beneficial for growth. For theses purposes, the B. subtilis wild type (168) and the darAdeletion mutant (GP1712) were grown as described for the metabolite determination in MSSM medium with ammonium and 0.1 mmKCl. The intracellular K⁺ amounts were determined by inductively coupled plasma optical emission spectrometry (ICP-OES) and normalized to the culture volume and OD600. c-di-AMP was extracted from aliquots of the same cultures, quantified using high performance liquid chromatography-coupled tandem mass spectrometry (HPLC-MS/MS) and normalized to the total protein amount determined with the Bradford method.

As shown in Figure 3.8, deletion ofdarAhad no impact on the intracellular amounts of K⁺ or c-di-AMP.

Figure 3.8: Deletion ofdarAhas no impact on c-di-AMP and potassium homeostasis.TheB. sub-tilis wild type (WT, 168) and the darA deletion mutant (GP1712) were grown in MSSM medium with ammonium and low (0.1 mm) amounts of potassium. Cells were harvested at an OD600 of 0.5. The intracellular amounts of potassium were determined by ICP-OES in the laboratory of Dr. Dietrich Hertel (Göttingen) and were normalized to the culture volume and OD600. The amounts of extracted, intracellular c-di-AMP were determined by HPLC-MS/MS in the laboratory of Prof. Dr. Volkhardt Kaever (Hannover) and were normalized to the total cellular protein amount. Data points represent independent biological replicates, bars indicate the calculated means (n= 3).

Apparently, DarA does not integrate the c-di-AMP signal to affect K⁺ homeostasis, at least not under the condition where DarA inhibits growth (MSSM medium with ammonium and 0.1 mmKCl). Furthermore, DarA does not affect c-di-AMP homeostasis under the same condition. The results show that the inhibitory effect of DarA on growth in MSSM medium with ammonium and 0.1 mm KCl is not caused by a modulation of potassium or c-di-AMP homeostasis by DarA. In summary, several studies did not elucidate the reason for this phenotype. The most likely cytosolic interaction partner of (most likely c-di-AMP-bound) DarA still escaped detection despite pursuing a multitude of different approaches.

3.5 Rational bioinformatic search of interaction partners

The c-di-AMP-bound DarA most likely interacts with a cytosolic target in MSSM medium with ammonium and low amounts of potassium, which inhibits growth in liquid medium.

Despite this specific phenotype and numerous experiments, the interaction partner still escaped detection. We reasoned that a rational bioinformatic search could provide an unbiased overview of proteins that are in theory suitable for an interaction with DarA.

For this purpose, we analyzed the proteome of B. subtilis to identify putative targets.

Several rational requirements were set and proteins not matching them were discarded. First, a putative interaction partner is most likely conserved among c-di-AMP-producing firmicutes also expressing darAor homologues. Consequently, the Bacillus subtilis 168 proteome was compared with the proteome of the related, well studied firmicutesListeria monocytogenes EGD-e and Staphylococcus aureus subspecies aureus NCTC 8325. Second, regulation by binding of DarA only makes sense if the interaction partner is less abundant than DarA.

Third, DarA is a trimeric, PII-like protein. As a consequence, only proteins which structurally resemble known PIItargets are highly suggestive of an interaction. These are foremost trimeric, hexameric or dodecameric. Fourth, our results strongly indicate that DarA is solely located in the cytosol and so should be the interaction partner. A list of the rational interaction partners can be found in the Appendix (see Table 6.1). It should be noted that 509 proteins were excluded since protein quantification data were missing, but a separate list was kept (see Appendix: Table 6.2). Missing protein quantification data might be attributed to membrane proteins, proteins that are not expressed, only expressed under very specific conditions or the protein amount of these is extremely low in general. In either case, suitability as an interaction partner of DarA is arguably questionable. We identified 160 proteins that were conserved, less abundant compared to DarA, and cytosolic or with unknown location. Structural data were not available for all of them but promising candidates were identified. Only eleven proteins matched all criteria (see Table 3.2).

Interestingly, again enzymes and proteins related to glutamate/arginine metabolism were found. Most notable hits were the dodecameric glutamate synthase GltAB and the hexameric N-acetyl-l-glutamate kinase (NAGK) ArgB. As mentioned before, the NAGK is regulated by

a PIIprotein in the cyanobacteriumS. elongatusand catalyzes the second, rate limiting step in arginine biosynthesis (Burillo et al., 2004; Maheswaran et al., 2004). Furthermore, molecular docking of DarA to GltAB suggested that this enzyme is a promising interaction partner of DarA since both protein structures fit together perfectly (Richts, 2018). The other hits were not as suggestive of an interaction. The glutamate dehydrogenase GudB is not a suitable target since the enzyme is not functional in the laboratory strain 168 (Belitsky and Sonenshein, 1998). It should be noted that B. subtilis encodes a second glutamate dehydrogenase, RocG, but quantification data were missing for this protein. HprK, MecA, Rho and YojN fulfilled our rational criteria. However, analysis of their tertiary structure and symmetry suggested that an interaction with DarA is not very likely. The structures of YhfE, YtaG and YtoP might be suitable for an interaction with DarA, however, all three proteins are only poorly characterized and we have no additional experimental indications (as in the case for GltAB or ArgB) that suggest an interaction. Taken together, ArgB and GltAB were identified as promising interaction partners of DarA. However, it remained elusive why neither of the two was identified in prior experiments, despite linking DarA to glutamate metabolism.

Table 3.2: Proteins that fulfilled our criteria for an interaction with DarA.

Name Description Localization Structure^a

ArgB N-Acetylglutamate 5-phosphotransferase Cytoplasm 6-hom GltA Glutamate synthase (large subunit) Cytoplasm 12-het GltB Glutamate synthase (small subunit) Cytoplasm 12-het GudB Glutamate dehydrogenase, trigger enzyme − 6-hom HprK PtsH-HPr kinase/ phosphorylase − 6-hom

MecA Adaptor protein − 6-hom

Rho Transcriptional termination protein − 6-hom

YhfE Similar to glucanase − 12-hom

YojN Similar to nitric-oxide reductase − 6-hom

YtaG Dephospho-CoA kinase − 3-hom

YtoP Similar to glutamyl aminopeptidase − 12-hom

a hom: homomer, het: heteromer.