To investigate how CdaR might sense changes in osmolarity and convey these signals to CdaA, to modulate its activity, we took a closer look at the function of the YbbR domains and the membrane localization of CdaR in vivo. As seen in Fig. 2.5, a cdaR mutant is still able to adjust its intracellular c-di-AMP concentration, but shows in general a decrease of c-di-AMP compared to the wild type and the range of adaptation seems to be decreased. A mutant expressing CdaR without its TM domain behaves like the cdaR deletion mutant. Taken together with the topology assays that showed that the TM domain is both, necessary and sufficient for correct membrane localization of the adjacent C-terminus (Fig. 2.3), it can be stated that extracellular localization of the YbbR domains is necessary for proper CdaR-dependent regulation of CdaA activity. This is especially of interest considering the effect of a mutant expressing CdaR without any YbbR domains, meaning only the TM domain. In this mutant we could only measure about a fifth of the c-di-AMP amount compared to the wild type strain. We concluded from this phenotype that expression of the TM domain without the YbbR domains disturbs CdaA dimerization. Considering the before mentioned
three hypotheses that CdaR might (i) sense cell wall alterations, (ii) act as a localizer and anchor for CdaA or (iii) act as a scaffolding protein with cell envelope signal amplifying function, to stabi-lize CdaA self-interaction, model two or three are in good agreement with the observed pheno-types. If CdaR would sense alterations in cell wall composition, the effect of CdaR should be slight conformational rearrangements of CdaR and in consequence of CdaA and thereby specific changes in activity upon YbbR domain interaction with an altered cell wall. A mutant without the sensory YbbR domains should therefore behave like the wild type, except when a signal is perceived. In contrast to that, expression of YbbR-less CdaR impacts CdaA activity per se. In the other two mod-els, where CdaR is important for proper interaction and/or localization of CdaA, the YbbR domains would be a determinant for self-interaction. It has been shown that the YbbR domains are able to interact in the cytosol, if the TM domains are truncated and, furthermore, that CdaA self-interacts, although weaker in the full-length variant compared to mutants lacking TM domains.
Moreover, CdaA and CdaR only interact with each other as full-length proteins, presumably via their TM domains (Rismondo et al., 2016). Hence, it can be hypothesized that CdaA and CdaR self-interactions might be influenced by their membrane localization in a negative way. Considering this, expression of CdaR TM domain that interacts with the TM domains of CdaA could destabilize CdaA self-interaction, if not stabilized by YbbR domain interaction. The phenotype of the YbbR-less CdaR expressing L. monocytogenes strain is in support of this hypothesis (Fig. 2.5). Interest-ingly, a complementation strain shows the opposite phenotype of the cdaR deletion mutant.
Higher c-di-AMP concentrations and a greater change upon osmotic change can be detected, which is in agreement with the proposed model (see Fig. 2.8).
When we compared our experimental findings to what is known about the impact of CdaR on the cellular c-di-AMP concentration in L. monocytogenes, we noticed two interesting differences. On the one hand, effects of cdaR deletion and complementation lead to decreased and increased c-di-AMP level, respectively, which is diametrically opposed to what is known from previous ex-periments and on the other hand c-di-AMP concentrations were in general lower than previously observed (Rismondo et al., 2016). The experimental procedures of these two experiments were distinct by two factors that may be the root for the differences. In Rismondo et al., the bacteria were cultivated in BHI, a very nutrient rich complex medium and c-di-AMP concentrations were measured at an OD600 of 1, the late exponential growth phase (the c-di-AMP concentration in the wild type, which was set to 100% was 99.4 ± 13.4 ng c-di-AMP per mg of protein; Gibhardt, 2015;
Rismondo et al., 2016). In contrast to this set up, we determined c-di-AMP concentrations in LSM, a defined medium and at an OD600 at the early- and mid-exponential phase. It is not known, how c-di-AMP concentration may change upon osmotic stress in dependency of the growth phase, but it is known from c-di-AMP and also other second messenger molecules, c-di-GMP that plays a role in E. coli lifestyle changes from exponential to the stationary growth phase that the growth phase is a determinant for nucleotide second messenger concentrations (Hengge et al., 2015). Similar as for c-di-AMP, the main determinants for the intracellular c-di-GMP concentration are synthesis and degradation by specialized cyclases and PDEs. Interestingly, expression of the PDE PdeH is inhibited and expression of the diguanylate cyclase DgcE increased during the entry into the sta-tionary phase of E. coli and together with other factors the increase in c-di-GMP inhibits motility and facilitates adhesion and matrix production of E. coli in the stationary phase (Hengge et al., 2015; Pesavento et al., 2008). For c-di-AMP, there are studies that demonstrate a cross-talk with the (p)ppGpp regulated stringent response, leading to altered concentrations of c-di-AMP, while the (p)ppGpp concentration depends on the growth phase and nutrient availability (Corrigan et al., 2015). Corrigan and colleagues could show for S. aureus that c-di-AMP levels are indeed rising in late growth phases and that gene expression patterns of cells with a high c-di-AMP
concentration overlap in parts with those of cells undergoing the stringent response. They fur-thermore could show that high c-di-AMP levels activate the RelA enzyme, leading to an increase in (p)ppGpp by an indirect and unknown mechanism (Corrigan et al., 2015). Interestingly, other studies have also demonstrated inhibition of c-di-AMP PDEs by (p)ppGpp, indicating a tight cross-regulation of those two nucleotide second messenger pathways that might play a role in osmo-regulation and will be discussed further below (Corrigan et al., 2015; Huynh et al., 2015; Rao et al., 2010; Whiteley et al., 2015). To conclude, c-di-AMP levels in L. monocytogenes in the early exponential phase might, like in S. aureus, may be lower compared to later growth phases. Fasci-natingly, cellular turgor is, through the accumulation of intracellular osmolytes, which is favored by lower c-di-AMP concentrations, a determinant for cell division in Gram-positive bacteria. In the exponential phase osmolyte import in general has to be controlled tighter, to prevent an excessive increase in turgor that might be detrimental and therefore favors higher c-di-AMP levels (Rojas et al., 2018). The second big difference in the experimental set ups was the medium composition. In a complex medium like BHI, in which a multitude of ionic and peptide osmolytes are present. In contrast to that, the Listeria synthetic medium (LSM) features defined amounts of amino acids, trace elements, vitamins and low amounts of potassium (4.8 mM) and, therefore, compared to a complex medium like BHI or LB a limited, defined amount of low complexity osmotic active sub-stances (Whiteley et al., 2017). The key difference might be the presence of oligopeptides in com-plex media that act as osmolytes in L. monocytogenes and are at least indirectly regulated by the impact of c-di-AMP on the stringent response and thereby expression of oligopeptide permeases (Whiteley et al., 2015; Whiteley et al., 2017). In a minimal medium, lower amounts of c-di-AMP are needed to balance the import of the less abundant osmolytes, like it has been demonstrated for potassium uptake and c-di-AMP levels in B. subtilis (Gundlach et al., 2017).
The different media compositions and or growth phase might also have an impact on the role of CdaR as a regulator. We therefore hypothesized that CdaR can act on CdaA as an inhibitor and activator, meaning it can in general modulate the activity of CdaA, in dependency of different stimuli. Nevertheless, additional experiments in E. coli provided an alternative hypothesis. Using the before mentioned E. coli- and KimA transporter-based c-di-AMP reporter system, we investi-gated impact of CdaR and CdaR variants on CdaA activity. Interestingly, in E. coli, CdaR seems to inhibit CdaA activity if it is membrane localized and at least one YbbR domain is present. In con-trast to the findings in L. monocytogenes, the mutant lacking all YbbR domains did not show a growth phenotype corresponding to low c-di-AMP, but rather showed an intermediate pheno-type, similar to GlmM, meaning it is still able to act in part as an inhibitor on CdaA. This difference of phenotype in L. monocytogenes and E. coli can be a consequence of the differences between the Gram-positive and Gram-negative cell envelope, since the CdaA-CdaR complex would presum-ably be located in the inner membrane of E. coli and the YbbR domains therefore present in the periplasm. Furthermore, different compositions in the lipid bilayer, composition of PG or interac-tions with other proteins could influence CdaA activity and the impact of regulator molecules on its activity (Silhavy et al., 2010). Another difference and is the lack of other proteins of the c-di-AMP signaling pathway in E. coli. Usually this is an advantage for studying DAC activity and the influence of single regulators on cyclases, but in this case we were wondering if other compo-nents of the signaling pathway might have an influence on CdaA activity or the role of CdaR. To elucidate this, we employed a bacterial two hybrid protein-protein interaction assay and indeed found protein-protein interactions between synthesis and degradation machineries (Fig. 2.6). It is therefore possible that CdaA and CdaR might be influences by PdeA and PgpH and vice versa and this potential cross-talk and the resulting local signaling should be the topic of further studies.