The HPLC method for analyzing cellular nucleotides after their extraction from animal cells or tissues is widely used for quantification of adenine nucleotides. In the course of such a study for which a series of samples of the sponge T. muricata were used we found that this sponge contains considerable amounts of diadenylates with 2′,5′-phosphodiester linkage. These diadenylates were present in the perchloric acid extract in various 5′-phosphorylated (tri-, di-, mono-) and unphosphorylated (“core”) forms, the proportions of which depended on the particular sample (Figure 1). In some samples minor amounts of 2′,5′-triadenylates were also present (Figure 2). Such data indicate that the oligoadenylates, initially synthesized as 5′-triphosphate derivatives resulting from OAS activity exhibited by the sponge were afterwards dephosphorylated to a different degree. However, the mechanisms that regulate phosphatase activities in particular sponges are not known yet.

Treatment of the extract with alkaline phosphatase simplified the subsequent analysis as all compounds were separated in their “core” forms (Figure 2). Quantification of adenylates revealed that their amounts varied by an order of magnitude between different samples of T. muricata (see Experimental section, 3.1): the amount of 2′,5′-linked diadenylates (A2′p5′A) was 2–50 pmol/mg wet weight, whereas that of adenosine (reflecting the total amount of adenine nucleotides ATP, ADP and AMP) varied from 2 to 80 pmol/mg wet weight.

Surprisingly, in many samples 2-5A was present in quantities that considerably exceeded those of the total adenine nucleotide pool (Figure 2). Such a variation between the amounts of adenylates obviously shows differences in the physiological status of sponges resulting i) from cultivation effects and ii) from variability between different specimens. In accordance with the data published previously [37], elevated levels of 2-5A evidently correspond to stress conditions for the sponge (data not shown).

A peak with retention time close to adenosine and UV spectrum (λ_max = 256 nm) different from oligoadenylates regularly accompanied high levels of A2′p5′A. Its intensity increased after dephosphorylation of the extract and we presupposed the presence of some hetero-oligomer, which might also have a 2′,5′-linkage. Therefore, the chromatographic fraction containing the unknown compound (peak 1, Figure 2) was collected and subjected to respective chemical and enzymatic treatments (see Figure 3) to establish the structure of the compound. Since alkaline hydrolysis of this compound produced equimolar amounts of the products [(3′-AMP + 2′-AMP) and guanosine)], and digestion with snake venom phosphodiesterase (PDE) produced equimolar amounts of 5′-GMP and adenosine, the unknown compound has to be a (A, G) dimer. Resistance to the treatment with ribonuclease T2 indicated that the dimer has a 2′,5′-internucleotide linkage instead of the usual 3′,5′-linkage.

As a result of these treatments, the unknown compound was identified as adenylyl(2′-5′)guanosine (A2′p5′G) (Figure 4, 1). Subsequent mass-spectrometric analysis (m/z 611.2) confirmed the identity of the product.

The mean content of A2′p5′G in different samples was about a magnitude lower than that of A2′p5′A′ A positive correlation observed between the amounts of A2′p5′A and A2′p5′G in different samples of T. muricata (Figure 5) indicated that formation of the (A, G) dimer was directly related to the activity of OAS in a particular sponge sample. Such a finding inspired us to look more carefully at minor components present in different samples in order to search for additional oligonucleotides with 2′,5′-linkage.

The area of a peak next to NAD⁺ (peak 2, Figure 2) correlated positively with the A2′p5′A peak as well (not shown), but differently from A2′p5′A, A2′p5′G and NAD⁺ peak, its area did not increase by phosphatase treatment of the extract. The UV spectrum of that peak was very similar to NAD⁺. Such a behavior (including the observed retention time) would be expected for Nir-5′pp5′-Ado-2′p5′-Ado (NAD2′p5′A), a compound formed from NAD⁺ and ATP by mammalian OAS [15]. MALDI-MS analysis of this compound in positive ion mode was in agreement with NADpA (m/z 993.1; 992 Da for neutral molecule). However, since NAD⁺ molecule has two ribose moieties with free 3′ and 2′ OH groups, 5′-AMP (from the substrate ATP) might bind, alternatively, to nicotinamide ribose moiety. The negative ion mode MALDI-MS analysis proved to be useful in solving this problem. The analysis of NAD⁺ in negative ion mode yielded the strongest signals with m/z 540 (the ionized molecule lacking nicotinamide) and 426 (corresponds to the fragment of ADP), whereas the peak of expected mass (m/z 662) was low. Appearance of NAD⁺ in negative ion mode MALDI-MS analysis as fragments with m/z 540 > 426 has been reported, while the signal with m/z 662 has not been detected at all [41]. Similarly to NAD⁺, NADpA in negative ion mode yielded the strongest signals with m/z 869.1 and 755.1, together with the low intensity signal that corresponded to the expected mass with m/z 991.1. On the basis of the fragments formed in both cases we can conclude that the adenylate additional to NAD⁺ in NADpA was bound to adenine nucleotide moiety of the NAD⁺ molecule. Since NADpA isolated from the sponge extract was resistant to RNase T2 treatment, the identified compound was Nir-5′pp5′-Ado-2′p5′-Ado (Figure 4, 2).

NAD⁺ has been reported to be a preferred acceptor molecule for the OAS in human (HeLa) cell extracts, however, the respective 2′-adenylated derivatives of NAD⁺ were not detected in the extracts synthesizing 2-5A [42].

Another small peak in the samples with higher A2′p5′A content was found whose area increased after treatment of the extract with alkaline phosphatase (peak 3, Figure 2, inset). As a result of the respective treatments (Figure 3) and mass-spectrometric analysis, two compounds in the collected chromatographic fraction were identified: one of them was adenylyl(2′–5′)uridine (A2′p5′U) (Figures 4, 3) (m/z 572.1) and the other was adenylyl(2′–5′)cytidine (Figure 4, 4) (A2′p5′C) (m/z 571.1) (Figure 4). From these two compounds A2′p5′U was prevalent, its molar amount was about two orders of magnitudes lower than that of A2′p5′A.

Thus, dimers containing all four ribonucleotides in the form of A2′p5′N (N is A, G, U or C) were found in dephosphorylated extracts of T. muricata, the molar content of these compounds decreasing in the order: A2′p5′A > A2′p5′G > A2′p5′U > A2′p5′C. Although the exact meaning of the synthesis of such 2′,5′-linked oligomers for the sponge physiology is not known, decrease in cellular nucleotide content, especially that of ATP, would have negative impact on total metabolic activity and could indicate stress conditions for the sponge.

To expand the knowledge about the natural occurrence of 2′,5′-linked heteronucleotides in sponges, the detectable levels of these metabolites could be found, guided by the example of T. muricata, in sponges with high 2-5A content. The occurrence of 2-5A has been shown only for G. cydonium, using radioimmunoassay method for establishing its concentration [29]. The content of 2-5A in G. cydonium that was determined in the present study was up to 5 pmol per mg wet weight, being in the same range with values found earlier. Two other sponges that had previously demonstrated high OAS activities [39] were studied for their 2-5A content. In C. reniformis, the amount of 2-5A was about 2 pmol per mg wet weight, whereas in C. nucula 2-5A was accumulated in higher amounts (14 and 3 pmol per mg wet weight for dimer and trimer, respectively) (Figure 6). Therefore, C. nucula was chosen for comparative studies of the occurrence of 2′,5′-linked heteronucleotides.

The presence of A2′p5′G (peak 1, Figure 6) was easily detected in dephosphorylated extracts of C. nucula and also confirmed as described above. Similarly to T. muricata, the amount of A2′p5′G (which also increased by phosphatase treatment) in the extract of C. nucula was about a magnitude lower than that of A2′p5′A.

To establish other possible 2′,5′-linked oligomers in the extract of C. nucula, chromatographic fractions (18 fractions, á 0.5 mL) in the region between guanosine and (2′–5′)ApApA peaks (Figure 6) were collected and analyzed. The following hetero-dimers (migrating together with some other components of the fraction) were identified in the analyzed region: peak 3 (23.5–24 min)–A2′p5′U (Figure 4, 3) (m/z 572.1) and A2′p5′C (Figure 4, 4) (m/z 571.1), together with traces of adenosine and A2′p5′A; peak 4 (17.5–18min)–guanylyl(2′–5′)uridine (G2′p5′U) (Figure 4, 5) (m/z 588.1) as a minor component migrating together with dGuo; peak 5 (19–19.5 min)–guanylyl(2′-5′)adenosine (G2′p5′A) (Figure 4, 6) (m/z 611.1) as a minor component migrating together with an unknown compound (UV_max 261 nm). The molar content of 2′,5′-linked dimers in the extract that was calculated by peak areas and from the data of alkaline hydrolysis of the collected fractions decreased in the order: A2′p5′A > A2′p5′U > A2′p5′G > A2′p5′C > G2′p5′U ≥ G2′p5′A.

In comparison with T. muricata, two additional compounds with the formula G2′p5′N (N is A or U) were found in the extract of C. nucula. In order to synthesize such hetero-oligomers, GTP has to be bound at acceptor site and ATP at donor site of the OAS. Since ATP is a prevalent ribonucleotide in cells (see also Figure 6 for the content of respective nucleosides), the affinity of GTP towards OAS from C. nucula should be considerable.

To date, no data are available in the literature about the ability of a sponge OAS to catalyze the formation of co-oligomers of ATP and another ribonucleotide. Therefore, to verify the hypothesis of in vivo synthesis of the 2′,5′-heteronucleotides in sponges, in vitro reactions were performed with NAD⁺ and four ribonucleotides, used alone or in mixtures with ATP.

The extracts with specific activities of 138 pmol ATP/min per μg protein and 1.64 nmol ATP/min per μg protein were used in the assays for T. muricata and C. nucula, respectively. NAD⁺ alone remained intact in the extract, however, other ribonucleotides which were used alone as substrates instead of ATP were easily converted to their respective 2′,5′-linked oligomers. In the case of GTP, the specific activity was 43 pmol/min per μg protein for T. muricata and 2.34 nmol/min per μg protein for C. nucula. Unexpectedly, pyrimidine nucleotides UTP or CTP were utilized quite efficiently to produce the respective 2′,5′-linked oligomers. The calculated specific activity for T. muricata was 54 pmol UTP/min per μg protein or 38 pmol CTP/min per μg protein. For C. nucula, the specific activities relative to pyrimidine nucleotides were 69 pmol UTP/min per μg protein and 13 pmol CTP/min per μg protein.

These data show that the specific activity of the ribonucleotides towards OAS decreases in the order: ATP > GTP ≅ UTP ≅ CTP or GTP ≅ ATP ≫ UTP > CTP for T. muricata and C. nucula, respectively.

The product profiles formed from the ribonucleotides were studied under the assay conditions where most of the given substrate was utilized for oligomerization. In these conditions, 2′,5′-linked oligomers up to pentamers (ATP), tetramers (GTP) and trimers (UTP and CTP) were formed (Table 1).

The relative amounts of the formed oligomers are presented in Table 2. As it can be seen, purine nucleotides were better substrates for chain elongation than pyrimidine nucleotides. The quantitative data of oligomerization of CTP by the C. nucula extract are not included in Table 2 due to a deaminating activity which was present in the extract. This resulted in a quite extensive conversion of CTP to UTP (with a specific activity of 22.6 pmol/min per μg protein) and in the production of a mixture of 2′,5′-linked products consisting of C and U homo-oligomers as well as of (C, U) heterooligomers.

Thus, as illustrated by the example of T. muricata and C. nucula, the substrate specificity of OAS depends on the particular sponge species. The inter-species comparison of known sponge OAS protein sequences refers to a very low similarity between these proteins, most probably due to their independent evolution after the divergence from their common ancestor [36]. Although the OAS genes from T. muricata and C. nucula have not been sequenced, different affinities of ribonucleotides towards OAS might be explained by differences in the enzyme structures of the respective sponges.

To study the formation of hetero-oligomers, ATP was combined with GTP, UTP, CTP or NAD⁺ in roughly equimolar concentrations. Since G2′p5′U was identified after phosphatase treatment of perchloric acid extract of C. nucula, the combination of GTP and UTP was used additionally in case of this sponge. Taking into account the relative ability of a sponge OAS to catalyze the oligomerization of these substrates when used alone (see above), the nature and multiplicity of the products formed by one or another of the sponges in the experiments with the mentioned mixtures is not surprising (Table 3).

The results demonstrate that adenylate was preferred in 5′-terminal position in hetero-oligomers, and the yield of dimeric product A2′p5′N was generally the highest (Figure 7A, C, D). As an exception, there was only a slight preference for ATP in 5′-terminal position when the ATP + GTP mixture was added to the C. nucula extract (Figure 7 B). In the reaction with GTP + UTP, GTP was an efficient acceptor molecule while UTP terminated the chain (Table 3). In the case of T. muricata, where the difference between OAS specific activities towards purine and pyrimidine nucleotides was smaller than that observed for C. nucula, pyrimidine nucleotides could also serve (though with a smaller efficiency) as acceptor molecules for oligomerization of ATP in the presence of UTP or CTP (Table 3, Figure 7C).

Thus, these data provide a clear proof for the ability of OAS to catalyze in vivo the production of 2′,5′-heteronucleotides. The results show that besides ATP, other nucleotides are efficiently used for the formation of 2′,5′-linked oligomers under in vitro conditions. Especially, the usage of pyrimidine nucleotides as acceptor molecules for the oligomerization refers to the substrate specificity of the sponge OAS that is remarkably different from that of vertebrate OASs.

It has previously been proposed that OAS proteins have evolved from a common ancestor of archaeal CCA-adding enzymes and poly(A) polymerases [24,25]. The OAS proteins show the highest similarity with the class I CCA-adding enzymes, which not only catalyze the synthesis of the CCAsequence to the 3′-tRNA end, but are also capable of poly(C) synthesis [43]. Interestingly, OASs from T. muricata and C. nucula are capable of catalyzing the synthesis of 2′,5′-linked homo-oligomers not only from ATP, but also from GTP, UTP and CTP. These findings indicate that OASs from sponges have broader nucleotide specificity than that known for vertebrate OASs. Further studies of these enzymes from sponges allow for better characterization of their nucleotidyltransferase properties.

Moreover, research in this field may help to understand the function of OAS proteins in sponges, the most basal multicellular animals. Taking into consideration that several OAS genes are expressed in a sponge and that, on the other hand, OAS protein sequences in different sponge species differ considerably from one another [36], the functions of OASs in sponges may also be diversified. Though at present suggestions about the role(s) of OASs in sponges are speculative, one can consider a possible role of these enzymes in the control of cell growth, which may be regulated by different signal pathways depending on the inducing factor. Another intriguing aspect is the participation of OASs in immune responses of sponges against bacteria or viruses as also proposed earlier [32,34,36]. The elucidation of the pathways involved in these host-defence reactions would give us valuable information about the innate immune system of invertebrates. In its turn, thorough studies of OASs from sponges could enlighten some aspects of the suggested multifunctionality of OAS proteins in vertebrates.

The marine sponges Thenea muricata (Demospongiae) were collected on a soft bottom sediment off Korsfjord (close to Bergen, Norway) at 300 m depths (60° 07′ 44 N, 004° 49′ 9 E). The sponges were either fixed directly in liquid nitrogen or maintained in tanks with running sea water for establishing a laboratory-based culture and then fixed. The total number of analyzed samples (corresponding to sponge specimens) was 64. The samples were kindly provided by Hans Tore Rapp from Bergen University. The marine sponges Chondrilla nucula, Geodia cydonium and Chondrosia reniformis (Demospongiae) collected from the Tyrrhenian Sea (Napoli Bay) were kindly provided by Dr. S. de Rosa (CNR, Napoli, Italy). The specimens of G. cydonium collected from the Adriatic Sea (Rovinj, Croatia) and C. reniformis collected from Cala Montgo, L’Escala (Spain) were gifts from Prof. Dr. W. E. G. Müller (University of Mainz, Germany) and Dr. M.-L. Schläppy (MPI for Marine Microbiology, Bremen, Germany), respectively. The animals were cut into pieces, frozen in liquid nitrogen and kept at −70 °C until their further use.

The sponge tissue was disrupted by means of TissueLyser (Qiagen, Retsch GmbH, Germany). Grinding jars were pre-chilled in liquid nitrogen and the frozen samples were pulverized in a highspeed regime (at 30 Hz) for 1 min.

The reaction mixtures containing 20 mM Tris-HCl, pH 8, 50 mM KCl, 5 mM MgCl₂, the sponge protein extract (final protein concentration from 0.005 to 0.1 mg/mL) and the substrate(s) (ATP, GTP, UTP, CTP, NAD⁺) were incubated at 37 °C for required periods of time (5 min–20 h). The reaction was stopped by heating (4 min, 95 °C). Substrates were used in the following concentrations: 0.86 mM ATP, 0.67 mM GTP, 0.76 mM UTP, 0.78 mM CTP and 0.97 mM NAD⁺.

For the determination of specific activity where the appearance of the firstly synthesized dimer was registered, short reaction times (5–15 min) and low sponge protein concentrations (0.05 to 0.005 mg protein/mL) were suitable. To study the total pool of oligomeric products formed in different reactions, the protein concentrations were 0.1 mg/mL and incubation times were longer. Overnight incubations (17–20 h) for all substrates were suitable for T. muricata. For C. nucula, 1 h incubations were used with all substrates or their combinations, except for pyrimidine nucleotides UTP or CTP when used alone. In the latter case, the incubation time was 18 h.

Products of the reactions were analyzed either directly or after dephosphorylation of the reaction mixture (see 3.6).

The analysis was performed with an Agilent 1100 HPLC instrument (Agilent Technologies, USA) on a C₁₈ reverse-phase column (Phenomenex Luna C18(2), 250 × 4.6 mm, 5 μm; pre-colum SecurityGuard C18, 4 × 3 mm). Eluent A was 50 mM ammonium phosphate pH 7.0 and eluent B was 50% methanol in water. The products were separated at 30 °C in a linear gradient of eluent B (0–40%, 20 min); the column was equilibriated with eluent A before the next injection (10 min) (gradient 1). Alternatively, the column was thermostated at 20 °C and the linear gradient of eluent B (0–60%, 30 min) was used (gradient 2). The absorption was measured at 260 nm. The adenylates and other products were identified by comparing their retention times with those of authentic compounds and by their UV spectra (using previously created UV spectra library). The retention times of the respective compounds with 2′,5′- and 3′,5′-linkage differed substantially, the latter having longer retention in the column′ For example, by applying gradient 1 the retention time of G2′p5′U was 16 min, whereas that of G3′p5′U (Sigma) was 20 min; the retention time of A2′p5′U was 20′9 min versus 23 min for A3′p5′U (Sigma).

The products were quantified by integrating the corresponding peak areas (software of Agilent Technologies). Taking into account the molar absorption coefficients of the compounds at 260 nm, molar concentrations were calculated. Molar absorption coefficients for 2′,5′-adenylates have been given previously [45]. The following molar absorption coefficients were used for other oligomers: G2′p5′G–2.11 × 10⁴, G2′p5′G2′p5′G–3.16 × 10⁴, G2′p5′G2′p5′G2′p5′G–4.21 × 10⁴, U2′p5′U–1.8 × 10⁴, U2′p5′U2′p5′U–2.7 × 10⁴, C2′p5′C–1.36 × 10⁴, C2′p5′C2′p5′C–2.04 × 10⁴.

The composition of oligoribonucleotides was determined by analysis of the core oligomers after removal of the 5′-phosphate groups by alkaline phosphatase treatment. The dephosphorylation reaction was performed by adding the appropriate amounts of 10 × reaction buffer (100 mM MgCl₂, 200 mM Tris-HCl, pH 8) and shrimp alkaline phosphatase (SAP, USB Corporation) at final concentration of 0.02–0.04 U/μL and incubating the mixture at 37 °C for 1 h; thereafter the enzyme was inactivated (5 min, 95 °C).

The collected HPLC fractions were directly subjected to mass-spectrometric analysis. The analysis was carried out with a Bruker Daltonics Autoflex II TOF/TOF instrument in a reflector negative mode. THAP (2,4,6-trihydroxyacetophenone) matrix solution was used as described previously [39]. [THAP-H⁺], [ApA-H⁺], [ApApA-H⁺] and [ApApApA-H⁺] with m/z of 167, 595.14, 924.1 and 1253.15, respectively, were used as external calibrants. DHB (2,5-dihydroxybenzoic acid) matrix solution was used for registration of masses in the region with m/z from 547 to 549. The mass signals of NAD⁺ and NAD2′p5′A were registered also in a reflector positive mode, using THAP as a matrix solution. The discrepancies between registered and theoretical monoisotope masses were in the range of 0.1 Da.