Putative Nucleotide-Based Second Messengers in the Archaeal Model Organisms Haloferax volcanii and Sulfolobus acidocaldarius

Research on nucleotide-based second messengers began in 1956 with the discovery of cyclic adenosine monophosphate (3′,5′-cAMP) by Earl Wilbur Sutherland and his co-workers. Since then, a broad variety of different signaling molecules composed of nucleotides has been discovered. These molecules fulfill crucial tasks in the context of intracellular signal transduction. The vast majority of the currently available knowledge about nucleotide-based second messengers originates from model organisms belonging either to the domain of eukaryotes or to the domain of bacteria, while the archaeal domain is significantly underrepresented in the field of nucleotide-based second messenger research. For several well-stablished eukaryotic and/or bacterial nucleotide-based second messengers, it is currently not clear whether these signaling molecules are present in archaea. In order to shed some light on this issue, this study analyzed cell extracts of two major archaeal model organisms, the euryarchaeon Haloferax volcanii and the crenarchaeon Sulfolobus acidocaldarius, using a modern mass spectrometry method to detect a broad variety of currently known nucleotide-based second messengers. The nucleotides 3′,5′-cAMP, cyclic guanosine monophosphate (3′,5′-cGMP), 5′-phosphoadenylyl-3′,5′-adenosine (5′-pApA), diadenosine tetraphosphate (Ap4A) as well as the 2′,3′-cyclic isomers of all four RNA building blocks (2′,3′-cNMPs) were present in both species. In addition, H. volcanii cell extracts also contain cyclic cytosine monophosphate (3′,5′-cCMP), cyclic uridine monophosphate (3′,5′-cUMP) and cyclic diadenosine monophosphate (3′,5′-c-di-AMP). The widely distributed bacterial second messengers cyclic diguanosine monophosphate (3′,5′-c-di-GMP) and guanosine (penta-)/tetraphosphate [(p)ppGpp] could not be detected. In summary, this study gives a comprehensive overview on the presence of a large set of currently established or putative nucleotide-based second messengers in an eury- and a crenarchaeal model organism.


INTRODUCTION
During cellular signal transduction, most external/environmental stimuli do not directly interact with their respective cellular target, but rather cause the intracellular production/release of specific small molecules, which transmit the initial signal and eventually trigger a specific cellular response. In this concept, the initial stimulus is referred to as "first messenger, " while the intracellularly transducing small molecules are called "second messengers" (Newton et al., 2016). Various second messengers have been identified and they can be grouped into four different categories according to their chemical properties: ions, gases and free radicals, lipid-based and nucleotide-based second messengers (Newton et al., 2016). Calcium (Ca 2+ ) is a well-established ionic second messenger, which plays a crucial role in a plethora of eukaryotic signal transduction processes such as the excitability of neural cells, exocytosis, motility, apoptosis, and transcription (Clapham, 2007). A well-characterized example of a gaseous second messenger is nitric oxide (NO). Molecules of this gas are involved in various prokaryotic and eukaryotic signal transduction pathways, such as the regulation of the mammalian nervous system and bacterial quorum sensing and biofilm formation (Zhou and Zhu, 2009;Nisbett and Boon, 2016). Examples of lipid-based second messengers are diacylglycerol or ceramide, which are involved in various eukaryotic signal cascades (Liscovitch and Cantley, 1994;Hilgemann et al., 2018). The most diverse category of second messengers consists of nucleotide-based signaling molecules. These second messengers can either be based on a single (mono-nucleotide-based), two (di-nucleotide-based), or several (oligo-nucleotide-based) nucleotide molecules. Table 1 shows an overview on the majority of currently established or putative nucleotide-based second messengers, including their presence and some examples of their functions in eukaryotes and bacteria.
The information summarized in Table 1 originates from bacteria and eukaryotes. For archaea, only very limited information about the occurrence and the physiological functions of nucleotide-based second messengers is currently available. Until now, only the presence of 3 ,5 -cAMP, 3 ,5 -cdi-AMP and cyclic oligo adenylate (cOA) has been reported in archaea. 3 ,5 -cAMP was identified in the euryarchaea Haloferax volcanii and Methanothermobacter thermoautotrophicus and in the crenarchaea Saccharolobus solfataricus (previously known as Sulfolobus solfataricus) (Leichtling et al., 1986). Additionally, in the euryarchaeon Halobacterium salinarum the levels of 3 ,5 -cAMP were shown to fluctuate during the cell cycle (Baumann et al., 2007). Analogous to its reported function in bacteria, cOA was shown to be involved in type III CRISPR system mediated immunity in the crenarchaeon Sa. solfataricus (Rouillon et al., 2018). In H. volcanii, 3 ,5 -cdi-AMP was shown to be essential and has been implicated in osmoregulation (Braun et al., 2019). Noteworthy, analysis of the presence of the alarmone (p)ppGpp using radioisotopelabeling approaches in a few archaeal species suggested the absence of this signaling nucleotide (Beauclerk et al., 1985;Cimmino et al., 1993;Scoarughi et al., 1995;Cellini et al., 2004). For all other nucleotide-based second messengers there is, to the best of our knowledge, currently no data available whether or not they are produced in archaea. To study the presence of nucleotide-based second messengers in archaea, cell extracts of the euryarchaeal model organism H. volcanii and the crenarchaeal model organism S. acidocaldarius were analyzed using liquid chromatography tandem mass spectrometry (LC−MS/MS) for the presence of representatives of mono-, di-, and oligo-nucleotide-based second messengers. These measurements unveiled that H. volcanii cells contain, besides the already known 3 ,5 -cAMP and 3 ,5 -c-di-AMP, detectable levels of 3 ,5 -cGMP, 3 ,5 -cCMP, 3 ,5 -cUMP, 2 ,3 -cAMP, 2 ,3 -cGMP, 2 ,3 -cCMP, 2 ,3 -cUMP, 5 -pApA and Ap 4 A. Compared to that, S. acidocaldarius cells contained a reduced variety of nucleotides. Besides all four 2 ,3 -cNMPs, only 3 ,5 -cAMP, 3 ,5 -cGMP, Ap 4 A and very minor amounts of 5 -pApA could be detected. The well-established bacterial second messenger 3 ,5 -c-di-GMP, the alarmone (p)ppGpp as well as all three physiologically appearing isomers of cGAMP (2 ,3 -cGAMP, 3 ,3 -cGAMP, and 3 ,2 -cGAMP) could not be detected, suggesting their absence in H. volcanii and S. acidocaldarius when grown under standard laboratory conditions. The same applies for 3 ,5 -cTMP, 3 ,5 -cIMP, 3 ,5 -cXMP, 5 -pGpG and cOA (n = 4; c-tetra-AMP), which were all not present in the examined H. volcanii and S. acidocaldarius cell extracts.
Taken together, the results of this study show that the nucleotide-based second messenger pools of H. volcanii and S. acidocaldarius contain several signaling molecules, whose presence in archaea has not been shown so far. Assuming that many other euryarchaeal and crenarchaeal species make use of similar nucleotide-based second messenger pools, these results offer important leads to further investigate the role and importance of these nucleotides for these organisms.

MATERIALS AND METHODS
Unless stated otherwise all chemicals were purchased from Carl Roth.

Strains and Growth Conditions
H. volcanii strain H26 was grown in selective CA medium (Allers et al., 2004) (0.5 g/L Bacto TM Casamino acids; pH 7.2 adjusted with KOH) modified with an expanded trace element solution (referred to as CAB) (Duggin et al., 2015). Cells were grown at 45 • C in liquid medium while rotating (volumes up to 5 ml) or shaking (volumes > 5 ml).
Since both strains are auxotroph for uracil (H26, pyrE2 and MW001, pyrEF), growth media were supplemented with uracil (Sigma) at a defined concentration (50 µg/ml for H26; 10 µg/ml 1 | Overview on occurrence and exemplary function(s) in eukaryotes and bacteria of all signaling nucleotides addressed in this study.

Nucleotide Extraction From Haloferax volcanii and Sulfolobus acidocaldarius Cells
The extraction of nucleotides from total cell pellets of H. volcanii and S. acidocaldarius cells was performed as described previously (Spangler et al., 2010;Braun et al., 2019 Resuspended pellets were incubated on ice for 15 min followed by a heating step at 95 • C for 10 min. After cooling on ice, the solution was centrifuged at 21,100 × g for 10 min at 4 • C. The resulting supernatant was transferred to a fresh vial. The extraction was repeated two times (three extraction steps in total) with 200 µl fresh extraction solution, omitting the heating step. The supernatants were combined and stored overnight at -20 • C to precipitate proteins. To remove precipitates, the samples were centrifuged again (10 min, 4 • C; 21,100 × g) and the supernatant was transferred to a fresh vial. Final extracts were desiccated using a vacuum concentrator (Eppendorf) at 45 • C.

Nucleotide Extraction From Haloferax volcanii Cell Lysate via Solid Phase Extraction
Nucleotide extracts from total cell pellets of H. volcanii could not be analyzed for their (p)ppGpp content due to certain specific fragmentation patterns of unknown molecular species overlapping with the internal (p)ppGpp standard. Therefore, extraction of (p)ppGpp was performed using a solid phase extraction approach as described previously (Ihara et al., 2015). For solid phase extraction from exponentially growing cultures 15 ml were harvested, for stationary grown cultures 9 ml were harvested. For each sample, an additional 2 ml aliquot of each culture was harvested for the determination of the total protein content (BCA assay). Growth of H26 and time points of sample acquisition were as described above. Cell pellets were snap−frozen in liquid nitrogen and subsequently resuspended in 2 ml ultrapure water on ice. Resuspended pellets were lysed by the addition of formic acid (to a final concentration of 1 M) and incubated for 1 h on ice. The cell lysate was mixed with 2 ml ammonium acetate (pH 4.5) and centrifuged for 5 min at 4 • C and 3,000 × g to remove cell debris. The lysate was further purified on an OASIS Wax cartridge 1 cc (Waters) using centrifugation steps of 1 min at 4,300 × g at 4 • C. The cartridge was equilibrated with 1 ml methanol followed by 1 ml ammonium acetate (pH 4.5), the lysate was loaded in four consecutive loading steps of 1 ml, the cartridge was washed with 1 ml ammonium acetate (pH 4.5) followed by 1 ml methanol and the sample was eluted with 1 ml elution solvent [water (ultrapure)/methanol/ammonium hydroxide (25% (w/v)): 7:2:1 (v/v/v)]. The elution fractions were desiccated using a vacuum concentrator (Eppendorf) at 45 • C.

Quantification of Nucleotides From Cell Extracts by Liquid Chromatography With Tandem Mass Spectrometry
Desiccated nucleotide extracts were resuspended in 200 µl water, centrifuged and diluted 1:2 with the respective standard solution (containing stable isotope labeled nucleotides as well as 100 mg/ml Tenofovir as internal standards) and analyzed by a LC−MS/MS method. Cyclic di-nucleotides were analyzed as described previously (Rao et al., 2010;Bähre and Kaever, 2017). Chromatographic separation was performed by reversed phase chromatography on a C18-column (Nucleodur Pyramid C18 3 µ 50 × 3 mm; Macherey-Nagel; Germany), using water containing 10 mM ammonium acetate and 0.1% acetic acid as eluent A and pure methanol as eluent B, using the following gradient: 0-4 min 0% B, 4-7.3 min 0-10% B, 7.3-8.3 min 10% B, 8.3-11 min 10-30% B, 11-13 min 0% B. The flow rate was 600 µl/min. Mass spectrometric analysis was performed on a tandem mass spectrometer (API4000; MA, United States) performing selected reaction monitoring (SRM). The mass spectrometer was equipped with an electrospray ionization source (ESI) and ionization was performed in positive mode for all analytes.
Cyclic nucleotides, were analyzed as described previously . Chromatographic separation was performed by reversed phase chromatography on a C18-column, using methanol-water [3:97 (v/v)] containing 50 mM ammonium acetate and 0.1% (v/v) acetic acid as eluent A and methanol-water [97:3 (v/v)] containing 50 mM ammonium acetate and 0.1% (v/v) acetic acid as eluent B, using the following gradient: 0-5 min 0-50% B and 5-8 min 0% B. The flow rate was 500 µl/min. Cyclic nucleotides were analyzed on a QTRAP 5500 (Sciex, MA, United States). Ionization was achieved with an ESI in positive mode. In SRM mode 3 ,5 -cNMPs and the 2 ,3 -cNMPs show the same mass transitions due to their high structural similarity. However, these analytes were clearly identified by their different retention times.
Ap 4 A, ppGpp, and pppGpp were analyzed as described previously (Schäfer et al., 2020). Chromatographic separation was performed on a Hypercarb column (30 × 4.6 mm, 5 µm particle size; Thermo Fisher, Scientific MA, United States) using 10 mM ammonium acetate (pH 10) as eluent A and acetonitril as eluent B, using an 8 min gradient from 4 to 60% B. The flow rate was 600 µl/min. All analytes were detected by LC-MS/MS on a QTRAP 5500 (Sciex MA, United States). Ionization of analytes was achieved with an ESI in positive ion mode and SRM was used for analyte detection.
For all used LC-MS/MS methods, the control of the LC and the mass spectrometers as well as data sampling was performed using Analyst software (version 1.7 Sciex, MA, United States). For quantification, calibration curves were created by plotting peak area ratios of the analyte, and the internal standard vs. the nominal concentration of the 10 calibrators. The calibration curve was calculated using quadratic regression and 1/× weighing.
The measured concentration of each nucleotide was normalized for each cell extract sample to the total protein concentration of the respective sample.

Haloferax volcanii and Sulfolobus acidocaldarius Cells Grown Under Standard Conditions Contain Several Mono-Nucleotide-Based (Putative) Second Messengers
We set out to identify nucleotide-based (putative) signaling molecules in the euryarchaeal and crenarchaeal model organisms H. volcanii and S. acidocaldarius. In a first step, we screened for mono-nucleotide-based (putative) second messengers. The presence of 3 ,5 -cAMP in H. volcanii cells has been described 35 years ago (Leichtling et al., 1986). Except for 3 ,5 -cAMP, the presence of other mono-nucleotide-based (putative) second messengers, like other 3 ,5 -cyclic nucleotides (3 ,5 -cNMPs), 2 ,3 -cNMPs or (p)ppGpp, has not been detected in any archaeal species so far. No reports are available on the presence of 3 ,5 -cAMP in S. acidocaldarius, but the closely related species Sa. solfataricus has been shown to produce 3 ,5 -cAMP (Leichtling et al., 1986), implying that this nucleotide is most likely also present in S. acidocaldarius.
H. volcanii strain H26 and S. acidocaldarius strain MW001 were grown under standard laboratory conditions. These cultures were used to obtain cell material from the exponential and stationary growth phases for nucleotide extraction (Supplementary Figure 1). Extraction of nucleotides from the cell extracts followed by LC-MS/MS not only confirmed the presence of 3 ,5 -cAMP in these two species (Figures 1A,B) but also revealed that 3 ,5 -cGMP, 3 ,5 -cCMP and 3 ,5 -cUMP are present in H. volcanii as well ( Figure 1A). These four 3 ,5 -cNMPs showed increased levels in exponentially growing H. volcanii cells compared to stationary cells ( Figure 1A). Similar observations were made for 3 ,5 -cAMP in exponentially growing S. acidocaldarius cells, which contained higher amounts of this cyclic nucleotide compared to stationary growth ( Figure 1B). 3 ,5 -cCMP and 3 ,5 -cUMP were not detected in any S. acidocaldarius sample (Figure 1B), suggesting that these nucleotides, at least under the tested growth conditions, are not synthesized by this crenarchaeon. 3 ,5 -cGMP was the only other 3 ,5 -cNMP detected in S. acidocaldarius ( Figure 1B). However, the levels of 3 ,5 -cGMP were low in both tested conditions. Five of nine replicates from the stationary culture (originating from three biological replicates with three technical replicates each) contained sufficient amounts of 3 ,5 -cGMP for a quantitative analysis. In all other samples from S. acidocaldarius (four replicates from stationary cells and all nine replicates from exponentially growing cells), 3 ,5 -cGMP could be detected as well, but at levels which did not allow for a valid quantification. Therefore no reliable 3 ,5 -cGMP level could be calculated for this crenarchaeon.
With (some) 3 ,5 -cNMPs being present in H. volcanii and S. acidocaldarius the question arises by which enzymes they are produced. Synthesis of 3 ,5 -cAMP in vivo is achieved by adenylate cyclases (ACs) of which six different classes (I-VI) are currently known (Khannpnavar et al., 2020). Very few predicted archaeal ACs fall into class III (Bassler et al., 2018) [cluster of orthologous groups (COG) 2114], while the vast majority belongs to class IV (COG1437), which is characterized by a CYTH (CyaB, thiamine triphosphatase) domain. H. volcanii as well as S. acidocaldarius each contain a single gene encoding for a putative class IV AC (HVO_1648 and Saci_0718). The S. acidocaldarius gene product of Saci_0718 has recently, however, been demonstrated not to function as a cyclase but as a phosphohydrolase of the triphosphate tunnel metalloenzyme (TTM) family (Vogt et al., 2021). An in this context performed systematic sequence similarity network analysis of the CYTH superfamily unveiled that actual class IV ACs only account for a small subgroup, which is entirely of bacterial origin (Vogt et al., 2021). This is in accordance with the observation that a H. volcanii deletion mutant lacking HVO_1648 has unchanged 3 ,5 -cAMP levels (preliminary data). Together, these observations imply that most archaea are likely to synthesize 3 ,5 -cAMP with a yet unknown new class of ACs, which is structurally different from the currently established ones.
Enzymes synthesizing 3 ,5 -cGMP, called guanylate cyclases, have been characterized from bacteria (Marden et al., 2011;An et al., 2013) and eukaryotes (Kang et al., 2019). Protein BLAST searches using the sequences of the guanylate cyclase domains of such enzymes (Marden et al., 2011;An et al., 2013;Kang et al., 2019) against the translated protein databases of species of the genera Haloferax and Sulfolobus yielded no results with significant similarity. This finding is supported by the fact that the COG2114 (adenylate and guanylate cyclase catalytic domain) neither contains a homolog for H. volcanii nor for S. acidocaldarius. All these observations might imply that 3 ,5 -cGMP in H. volcanii and S. acidocaldarius is not generated by a classic guanylate cyclase.
3 ,5 -cCMP and 3 ,5 -cUMP were just recently identified to play an important role as signaling molecules in prokaryotic phage-defense systems (Tal et al., 2021). A recently performed phylogenetic analysis of proteins containing pyrimidine cyclase domains revealed that these types of enzymes are also found in few euryarchaeal species (Tal et al., 2021). However, BLAST searches of these identified putative euryarchaeal pyrimidine cyclases in the proteome of H. volcanii yielded no hit. The same applies when experimentally characterized bacterial pyrimidine cyclases (Tal et al., 2021) were used as template. This observation might suggest the existence of additional and more distantly to the recently discovered pyrimidine cyclases related types of specific cytidylate/uridylate cyclases. Nevertheless, since guanylate and adenylate cyclases were identified or are assumed to be capable of not only producing their respective intrinsic products but also 3 ,5 -cCMP and 3 ,5 -cUMP under certain conditions (Beste et al., 2012;Seifert, 2017), it is possible that the detected 3 ,5 -cCMP and 3 ,5 -cUMP originate from a divergent enzymatic activity of these two types of cyclases.
In addition to 3 ,5 -cAMP, 3 ,5 -cGMP, 3 ,5 -cCMP, and 3 ,5 -cUMP, the cell extracts from both species were also checked for the presence of 3 ,5 -cTMP, 3 ,5 -cIMP, and 3 ,5 -cXMP, however, none of these cyclic nucleotides could be detected ( Table 2). Only very few studies show the presence 3 ,5 -cIMP in biological systems (Newton et al., 1998;Chen et al., 2014), with some of these reports called into question when it comes to the specificity of the detection method used (Seifert, 2015). Therefore, it is difficult to speculate whether the absence of 3 ,5 -cIMP observed in H. volcanii and S. acidocaldarius indicates a general absence of this nucleotide or an absence under the standard conditions used in this study. The absence of 3 ,5 -cTMP and 3 ,5 -cXMP in cell extracts from both model organisms is in line with the fact that these cyclic nucleotides have not been unequivocally identified in any living cell yet.
Additionally to 3 ,5 -cNMPs, H. volcanii and S. acidocaldarius cell extracts were also analyzed for the presence of 2 ,3 -cNMPs. All four examined 2 ,3 -cNMPs, namely 2 ,3 -cAMP, 2 ,3 -cGMP, 2 ,3 -cCMP and 2 ,3 -cUMP, were present in both species in samples from at least one of the two tested growth stages (Figures 2A,B). For H. volcanii cell extracts,  the measured concentrations of 2 ,3 -cAMP, 2 ,3 -cGMP, and 2 ,3 -cCMP (Figure 2A) were much higher than the ones of the corresponding 3 ,5 -isomer ( Figure 1A). Only 2 ,3 -cUMP was present at concentrations similar to 3 ,5 -cUMP. Similar to the 3 ,5 -cNMPs, the concentrations of 2 ,3 -cNMPs in H. volcanii generally increased during exponential growth (Figure 2A). In S. acidocaldarius extracts, 2 ,3 -cNMP levels were in the same range for exponential and stationary growth ( Figure 2B). Similarly, to what was observed for the 2 ,3 -cUMP levels of H. volcanii, levels of this 2 ,3 -cyclic nucleotide in S. acidocaldarius were also considerably lower in comparison to the other three 2 ,3 -cNMPs. Of all stationary samples, only three technical replicates (out of nine in total) contained quantifiable amounts of 2 ,3 -cUMP, whereas samples of exponentially growing S. acidocaldarius cells did not contain any 2 ,3 -cUMP. Production of the detected 2 ,3 -cNMPs in both species has most likely to be linked to the process of RNA-degradation, a major source of 2 ,3 -cNMPs in eukaryotes and bacteria (Thompson et al., 1994;Fontaine et al., 2018), and/or to the activity of homologs of certain RNA cyclases/ligases, which are also known to form 2 ,3 -cyclic phosphates at the 3 -ends of RNAs (Shigematsu et al., 2018). As currently no distinct function as second messenger is ascribed to any 2 ,3 -cNMP it appears likely that they are not used as such in H. volcanii and S. acidocaldarius as well. Still, it also cannot be excluded that they may act in some yet to be discovered signaling network. Next to 3 ,5 -cNMPs and 2 ,3 -cNMPs, the cell extracts were also analyzed for the presence of the alarmone ppGpp and its precursor pppGpp. Analysis of extracts from S. acidocaldarius did not detect any (p)ppGpp at the tested conditions ( Table 2). Since cell extracts of H. volcanii contained substances that interfered with the (p)ppGpp internal standard signal, an alternative solid phase extraction protocol was used (see section "Materials and Methods"). For these extracts the internal (p)ppGpp standard signal was unaffected and the corresponding measurements revealed that neither ppGpp nor pppGpp was present in the H. volcanii samples ( Table 2). These observations are in accordance with former studies examining the occurrence of (p)ppGpp in both species which showed that this alarmone was not produced, even when cells were subjected to stress factors such as starvation (Scoarughi et al., 1995;Cellini et al., 2004). In line with this, a study on the distribution of (p)ppGpp synthetases and hydrolases across the tree of life suggests that H. volcanii and S. acidocaldarius do not contain any of these enzymes (COG0317 contains no hit for both organisms) and that they are in general only very rarely found in archaea (Atkinson et al., 2011). Cyclic Diadenosine Monophosphate, 5 -Phosphoadenylyl-3 ,5 -Adenosine, and Diadenosine Tetraphosphate Are the Only Di-Nucleotide-Based (Putative) Second Messengers Measured in at Least One of the Two Archaeal Model Organisms In a second step, we screened for di-nucleotide-based (putative) second messengers. The only di-nucleotide-based second messenger detected in any archaeon so far is 3 ,5 -c-di-AMP. It was recently shown to be produced in the euryarchaeon H. volcanii by the corresponding di-adenylate cyclase DacZ and osmoregulation had been implicated as a major function of this nucleotide (Braun et al., 2019). Extraction of nucleotides from the cell extracts followed by LC-MS/MS confirmed the presence of 3 ,5 -c-di-AMP during both the exponential and the stationary phase at concentrations similar to what was detected previously (Braun et al., 2019; Figure 3A). In contrast to H. volcanii, S. acidocaldarius cell extracts generated in this study did not contain any 3 ,5 -c-di-AMP. A protein BLAST search using the sequences of an established bacterial (Rosenberg et al., 2015) and an archaeal (Braun et al., 2019) di-adenylate cyclases against the proteome of S. acidocaldarius yielded no significant hits. These observations are consistent with previous bioinformatical analyses showing that proteins containing the 3 ,5 -c-di-AMP synthesizing diadenylate cyclase (DAC)-domain (COG1624) are absent in crenarchaeota, while being frequently found in euryarchaeota (Römling, 2008;Witte et al., 2008;He et al., 2020). This suggests that 3 ,5 -c-di-AMP is likely to be used as second messenger by many euryarchaea, whereas crenarchaea do not seem to utilize this signaling nucleotide.
With 3 ,5 -c-di-AMP being present in H. volcanii, it was not surprising that its (intermediate) degradation product 5 -pApA, was also found in this euryarchaeon ( Figure 3A). The concentration of 5 -pApA was, however, significantly lower than the concentration of 3 ,5 -c-di-AMP. Especially samples from exponentially growing cultures contained only low amounts of 5 -pApA. Although all nine replicates of exponential cultures included 5 -pApA, only two of them contained sufficient amounts for a quantitative analysis. These very low levels of 5 -pApA suggests that most of this linear di-nucleotide in H. volcanii is further degraded to 5 -AMP, the final degradation product of 3 ,5 -c-di-AMP (Commichau et al., 2019). The phosphodiesterases which are degrading 3 ,5 -c-di-AMP and/or 5 -pApA in H. volcanii are currently unknown, as well as a potential function of 5 -pApA as second messenger. Samples from exponentially growing S. acidocaldarius cells did not exhibit any 5 -pApA. Yet, most samples (eight of nine replicates) from stationary S. acidocaldarius cultures included minor amounts of 5 -pApA ( Figure 3B). However, as these amounts were in all samples below the quantification limit no valid average 5 -pApA level could be calculated. As S. acidocaldarius does not contain any 3 ,5 -c-di-AMP these minor amounts of 5 -pApA do certainly not originate from the degradation of 3 ,5 -c-di-AMP. As the genome of S. acidocaldarius has an A + Tcontent of 63% (Chen et al., 2005) it appears possible that the detected molecules of 5 -pApA are intermediate fragments from degraded genomic DNA.
Diadenosine tetraphosphate (Ap 4 A), which was shown in several bacteria to function as an stress induced second messenger (Lee et al., 1983;Pálfi et al., 1991;Kimura et al., 2017;Ferguson et al., 2020), could be detected in cell extracts from both archaeal model organisms (Figures 3A,B). For both species, levels of Ap 4 A were higher during exponential growth. However, samples from exponentially growing H. volcanii cells exhibited a quite broad fluctuation among individual technical replicates and among the biological replicates, with some samples even exhibiting a complete absence of Ap 4 A ( Figure 3A). Intriguingly, Ap 4 A levels in exponentially growing S. acidocaldarius cells were the highest of all putative nucleotide-based second messengers detected in this study ( Figure 3B). Noteworthy, exponential samples from S. acidocaldarius did not show a similar broad fluctuation for their Ap 4 A levels as observed for exponential samples from H. volcanii. The 16-fold difference between Ap 4 A levels in S. acidocaldarius cells during exponential growth and the stationary phase is the largest difference that was observed between the two phases within this study. Whether Ap 4 A is actually used in a second messenger context and what possible biological functions this di-nucleotide could have in H. volcanii and S. acidocaldarius is currently not known. The observed differences between exponential and stationary growth phases as well as the high amounts of Ap 4 A specifically produced by    Table 3 (Exp., exponential; Stat., stationary). exponentially growing S. acidocaldarius cells imply a general physiological relevance of this di-nucleotide. Noteworthy, bioinformatical identification of any Ap 4 A synthesizing enzyme in S. acidocaldarius and H. volcanii is particularly complicated as a broad variety of aminoacyl-tRNA synthetases and also other enzymes like, for example, DNA and RNA ligases, are known to be capable of forming this di-nucleotide (Fraga and Fontes, 2011;Ferguson et al., 2020).
Interestingly, the very well established bacterial second messenger 3 ,5 -c-di-GMP and its (intermediate) degradation product 5 -pGpG were not detected in the cell extracts of H. volcanii and S. acidocaldarius. This suggests that this cyclic di-nucleotide is, unlike to various bacteria, not a key regulatory molecule in these two species. Indeed, a protein BLAST search using the GGDEF-domain, the domain responsible for 3 ,5 -cdi-GMP formation (Paul et al., 2004), of a di-guanylate cyclase from E. coli (Ryjenkov et al., 2005) against the genera Haloferax and Sulfolobus yielded no hits for the genus Sulfolobus and only three hits for the genus Haloferax, of which only one contained the complete GGDEF-motive (hypothetical protein; overall query cover: 40%, percent identity: 38.89%). This is in agreement with previous bioinformatical analyses which showed that not only proteins with a GGDEF-domain are almost completely absent in archaea, but also proteins with all other domains associated with 3 ,5 -c-di-GMP signaling (e.g., EAL-or PilZdomain) (Römling et al., 2013). Nevertheless, an analogous BLAST search against the entire domain of archaea yielded more than 300 hits of putative GGDEF-domain containing enzymes with many of them being predicted to belong to species of the recently discovered Asgard-and DPANN-superphyla. This suggests that 3 ,5 -c-di-GMP might not entirely be absent in the archaeal domain of life. In addition to 3 ,5 -c-di-GMP and 5 -pGpG, cell extracts were also analyzed for the presence of the eukaryotic dinucleotide-based second messengers 2 ,3 -cGAMP and 3 ,2 -cGAMP and their bacterial analog 3 ,3 -cGAMP. None of these three isomers could be detected. The absence of 2 ,3 -cGAMP in H. volcanii and S. acidocaldarius fits with the current idea of 2 ,3 -cGAMP only being present in metazoa (Kranzusch et al., 2015). The recently discovered isomer 3 ,2 -cGAMP was also not detected in any sample. As this isomer also contains an atypical 2-5 phosphodiester linkage it appears very likely that it is also only produced by metazoa. The fact that 3 ,3 -cGAMP is absent in both species might suggest that both do not use any of the prokaryotic anti-phage defense mechanisms, which have been previously linked to bacterial 3 ,3-cGAMP production (Severin et al., 2018;Cohen et al., 2019). This idea is also supported by a complete lack of proteins in the genera of Haloferax and Sulfolobus sharing significant similarities with so far characterized bacterial cyclic GMP-AMP synthases (cGASs) (BLAST search using two characterized bacterial cGASs; Davies et al., 2012;Li et al., 2019). A look at the COGs for both, bacterial (ENOG5028K9C) and metazoan (KOG3963), cGASs unveils that each of them is specific for its respective domain and that they do not root in a common COG which would also include any archaeon.

Cyclic Tetra-AMP Could Not Be Detected in Haloferax volcanii and Sulfolobus acidocaldarius Cell Extracts
Only a few cyclic oligo-nucleotide-based second messengers have been identified. An example of this is c-tetra-AMP (n = 4), which is known to occur, alongside with other isomers of cyclic oligo adenylate (n = 3-6) (cOA), in some crenarchaeal species, such as Sa. solfataricus, a species closely related to S. acidocaldarius (Rouillon et al., 2018). There, cOA was shown to be involved in type III CRISPR system mediated immunity (Kazlauskiene et al., 2017;Niewoehner et al., 2017;Rouillon et al., 2018). No c-tetra-AMP could be detected in any of the samples prepared for this study. As S. acidocaldarius encodes a functional type III CRISPR system that contains a Cas10 subunit (Zink et al., 2021) detection of c-tetra-AMP in extracts from this species could have been expected. However, as the cells in this study were not challenged with any invading virus it is plausible that the observed lack of c-tetra-AMP originates from the type III CRISPR system of S. acidocaldarius being inactive at the tested conditions. H. volcanii and other euryarchaea of the order Halobacteriales contain type I, but lack type III CRISPR systems (Maier et al., 2019), which certainly explains the absence of c-tetra-AMP (and thereby most likely the absence of cOA in general) in H. volcanii.

CONCLUDING REMARK
The results of this study represent the first screening of cell extracts from an euryarchaeal and a crenarchaeal species for a multitude of currently known (and established) nucleotide-based second messengers using a modern and highly sensitive mass spectrometry method. It gives a comprehensive overview on a broad spectrum of (potential) small signaling molecules which are present in the archaeal model organisms H. volcanii and S. acidocaldarius under standard growth and experimental conditions. The function of second messengers includes rapid variations within their levels depending on changing environmental conditions. It thus appears reasonable that the here measured nucleotide levels might considerably change when different cultivation methods/circumstances are used. Even the appearance of a nucleotide species totally absent here (or vice versa) appears then possible. Determining the nucleotide levels in such differentiating growth experiments could help to elucidate the functions of several of the here reported (putative) second messengers in H. volcanii and S. acidocaldarius, and thereby also in archaea in more general.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, upon request.

ACKNOWLEDGMENTS
We thank Annette Garbe, Niklas Rieck and the whole staff of the ZFA Metabolomics for the performance of nucleotide mass spectrometry measurements and their technical assistance. We also thank Chris van der Does for critically reading the manuscript.