Polyphosphate as a donor of high-energy phosphate for the synthesis of ADP and ATP

All processes in living systems that are able to maintain a steady-state in which the entropy within the system is lower than that of its non-living surrounding, as well as reactions in non-living systems, always proceed with an increase in entropy (Sanchez, 2011). The transformations of energy among all of its forms, including entropy and free energy at finite temperature, follow the laws of thermodynamics (Ge and Qian, 2010). Energy transformation processes have two arms: (1) energy conservation and (2) energy dissipation. Both forms are required to drive metabolism in living systems. The energy is usually lost as heat. The free energy (ΔG; Gibbs free energy) gained or lost during a reaction can be calculated as ΔG=ΔH−(T×ΔS), where H is enthalpy, T is temperature and S is entropy. According to Lippman (1941), free energy that is produced in a living system during catabolism is stored in energy-rich compounds, for example, ATP. By that, free energy can be subsequently used to drive endergonic reactions that constitute and maintain anabolic reactions. The breakdown of ‘foodstuffs’ results in the formation of a special kind of chemical energy before it is converted into other forms of energy, for example, mechanical work (muscle), osmotic work (secreting glands) or heat (fat tissue). These mutual relationships imply that during enzymatic hydrolysis of high-energy phosphate bond(s) in ATP [e.g. by ATPase(s)], biochemically useful energy is released; part of this energy can be converted into heat.

In the extracellular space of metazoan systems ATP is not the major metabolite containing ‘energy-rich’ bonds (high-energy phosphate bonds/acid anhydride linkages), but is instead inorganic polyphosphate (polyP) (reviewed in Kornberg et al., 1999; Schröder and Müller, 1999; Kulaev et al., 2004). ATP exists within cells at comparably high concentrations (3–10 mM). It can be released into the extracellular space where it acts as signaling molecule at a level of ∼10 nM; there it has only a short half-life of ∼15 min. In contrast, in the extracellular space, polyP is stored in relatively larger amounts, especially in vesicles in blood platelets (at a concentration of ∼130 mM), as well as in a free state in the blood (∼3 μM) (reviewed in Wang et al., 2016). Importantly, polyP is metabolically hydrolyzed by an alkaline phosphatase (ALP) enzyme (Lorenz and Schröder, 2001). In line with the Lippman concept of free energy, we have to accept that this energy is channeled at first into anabolic reactions, through formation of chemical bonds that usually requires energy. In the extracellular space, energy is required, among other reasons, for contraction/relaxation processes, for example, during biomechanical processes in the cartilage (Muiznieks and Keeley, 2013) or other coiling/recoiling transformations (Rauscher and Pomès, 2012). In the extracellular space, it is also assumed that energy is consumed during enzymatic and non-enzymatic processes associated with blood clotting or the reactions of the complement system (Nesargikar et al., 2012). Finally, production of entropy and dissipation of free energy associated with spontaneous relaxation processes (i.e. self-organization) in the extracellular space also require energy-producing biochemical reactions.

It is well documented that, in bacteria, inorganic polyP is synthesized by the polyphosphate kinase-1 (PPK1) which uses ATP as a phosphodonor (reviewed in Brown and Kornberg, 2008). Intracellularly, within bacteria, polyP acts as an energy store. The energy-rich phosphates in polyP are considered to be used for the synthesis of nucleotides. In bacteria, during polyP degradation, the polyphosphate-AMP-phosphotransferase generates ADP, the endo-polyphosphatase PPN forms polyP molecules with shorter chain lengths and the exo-polyphosphatase PPX releases free inorganic phosphate. It is also interesting that the unicellular eukaryote Dictyostelium discoideum requires linear inorganic polyP, with its high-energy phosphoanhydride bonds, for complete realization of its developmental program (Livermore et al., 2016). In addition, recent results also indicate that the energy level in the extracellular microenvironment decisively controls the proliferation capacity of mammalian cells (Loo et al., 2015). During this regulation loop, the liver cells secrete a kinase, a creatine kinase, that acts extracellularly to generate phosphocreatinine that directly fuels growth of cells.

ATP-consuming kinases are crucially involved in the control of certain metabolic events, for example, phosphorylation of secreted proteins controlling biomineralization, not only intracellularly, but also extracellularly (Tagliabracci et al., 2012). Furthermore, extracellular ATP acts as an excitatory transmitter during synaptic transmission by interacting, for example, with purinergic receptors, ecto-ATPases or through phosphorylation via ecto-protein kinases. In addition, ATP has been shown to be involved in the modulation of cell–cell spreading of Ca²⁺ signals between mast cells, and it has been suggested that it is involved in the cell–cell contacts during activation of T-lymphocytes (see: Redegeld et al., 1997). Hence, we have to search for a potential pathway that allows the generation of ATP from polyP. The first candidate enzymes involved in such transferase reactions would be those that can act as an adenylate kinase (AK; EC 2.7.4.3). This phosphotransferase activity catalyzes the interconversion of adenine nucleotides and, by that, plays an important role in cellular energy homeostasis. There are seven AK enzyme isoforms in mammals (reviewed in Dzeja and Terzic, 2009): the muscle AK1 and AK2, which are the cytosolic and mitochondrial isoforms of the inner and outer membranes; AK3, which is a a GTP:AMP phosphotransferase of intramitochondrial AMP; AK4 and AK5, which are located in the mitochondrial matrix of neuronal and pancreas cells; AK6, which has been identified in the cell nucleus and suggested to have a role in energy provision there; and finally AK7, which occurs in epithelial cells. The most likely candidate for performing a phosphotransferase reaction is AK1, which is secreted into the extracellular space (Choo et al., 2008). This enzyme, most likely the isoform AK1β, has also been suspected to mediate the synthesis of ATP, which acts as a crucial mediator of chemosensory transduction (Quillen et al., 2006). It has been demonstrated that AK1 secretion is required for the accumulation of extracellular ATP, for example, in myotubes (Choo et al., 2008).

Some evidence exists that ADP is also used as energy source in the extracellular space, for example, for the extracellular chaperon clusterin. This glycosylated secretory heterodimeric protein of ∼75 kDa has been proposed to be secreted from the cells in response to apoptotic signals (see Trougakos, 2013) and is considered to be a unique regulator of extracellular proteostasis. Even though clusterin comprises a putative dinucleotide-binding structure (Tsuruta et al., 1990), any effect of ATP on the action of this chaperone remains unknown (Poon et al., 2000). Since the binding domain, found in clusterin, has the same sequence as that found in ADP-binding proteins, a thorough study of other energy donors might be advisable (Wierenga and Hol, 1983).

The existence of extracellular matrix vesicles of 20 nm to 400 nm globular particles has been extensively demonstrated (Ciancaglini et al., 2006; Golub, 2009). The function of these organelles has been determined to some extent. It is striking that the matrix vesicles are enriched with the tissue-nonspecific ALP (known as ALPL) and other nucleotide-converting enzymes, for example, a nucleotide pyrophosphatase/phosphodiesterase (Terkeltaub, 2006). Since their initial observation (Ali, 1976), the potential role of the matrix vesicles for processes in mineralization of (calcified) cartilage, bone and dentin has been substantiated (Golub, 2009). In the present study, evidence is given that matrix vesicles are abundantly formed after exposure of the cells to polyP.

Based on our recent contribution providing first evidence of an increased formation/release of extracellular ATP in the presence of inorganic polyP (Müller et al., 2015), we continue to use the osteosarcoma cell line SaOS-2 (Pautke et al., 2004) for a more detailed elucidation of the role of polyP as an extracellular energy source for the formation of ADP and ATP. In order to dissect the transfer of the metabolic energy, conserved in the high-energy acid anhydride bonds of polyP, to adenosine nucleotides, we applied the natural inhibitor of the AK enzyme(s), A(5′)P5(5′)A (Yan and Tsai, 1999). It is established that ATP is released by bone cells via vesicular exocytosis (Orriss et al., 2013). This process can be inhibited by N-ethylmaleimide (NEM) (Akopova et al., 2012; Sivaramakrishnan and Fountain, 2015) and brefeldin A (Yewdell and Bennink, 1989; Brandao-Burch et al., 2012). In turn, we applied these two inhibitors and determined that, even in their presence, the ATP level in the medium increased. It is interesting to mention is that A(5′)P5(5′)A exists in large quantities in blood platelets (Davies et al., 1995), those cells which are richest in polyP (reviewed in Morrissey et al., 2012). By performing this analysis, we could demonstrate that the formation of ADP in the extracellular space is independent of the synthesis of ATP via the AK. In conclusion, the data in the present study provide further evidence that it is polyP that substantially contributes, via its acid anhydride bonds, to phosphorylation events in the extracellular space.

Na-polyP was added to the SaOS-2 cell medium, together with extra Ca²⁺ in order to keep the Ca²⁺ concentration in the medium at 2 mM [denoted Na-polyP (Ca²⁺)], causes a dose-dependent increase in cell viability, as measured with a colorimetric XTT assay. After an incubation period of 3 days, Na-polyP at a concentration of 10 µg/ml and 30 µg/ml caused a significant increase of cell growth from 0.69±0.08 optical density at 500 nm (OD) units OD to 1.08±0.11 and 1.35±0.14 OD units, respectively. Addition of the mineralization activation cocktail (MAC), composed of β-glycerophosphate, ascorbic acid and dexamethasone (Wiens et al., 2010b) increased the basal level of cell growth to 1.45±0.15 OD units; in the presence of 10 µg/ml and 30 µg/ml Na-polyP a further significant increase to 2.13±0.23 OD units and 2.33±0.27 OD units, respectively, was found. If, in addition, these cell assays were supplemented with the A(5′)P5(5′)A inhibitor (50 µM) no significant change of cell growth was recorded (Fig. 1).

Cells incubated for 3 days on microscope slides in the presence of the MAC and either in the absence or the presence of 30 µg/ml of Na-polyP were subjected to immunostaining using antibodies against AK1 (Fig. 2). In addition, the cells were stained with DAPI to identify the individual nuclei in the cells. The samples grown in the absence of Na-polyP showed a lower cell density on the slides (Fig. 2A,B), compared to those seen on samples from Na-polyP-treated cultures (Fig. 2D,E). With respect to the intensities of the staining for the AK1 antigen on and within the cells, no striking differences could be seen (Fig. 2A versus D). The fluorescence intensities in both assays, i.e. without and with polyP, were calculated by using ImageJ and found to be very similar (for the control as well as for the polyP-exposed cells). It is noticeable that, besides the more homogenous staining of the inside of the cells, some grains with a diameter of ∼500 nm are visible (Fig. 2C,F). Those grains are of a similar size to the extracellular matrix vesicles that are released from cells treated with polyP (see below); those vesicles originate from intracellular precursors via apical membrane microvilli (Thouverey et al., 2009).

The SaOS-2 cells remained unexposed to polyP or were incubated with 30 µg/ml of Na-polyP for 3 d in the presence of the MAC. Then the cells were taken, sliced and reacted either with the anti-ALP (herein used to refer to ALPL) or AK1 antibodies; the immunogold complexes were visualized with gold-labeled antibodies, as described in the Materials and Methods. In the control series for the ALP and the AK1 reaction, it was established that the slices that did not react with the gold-labeled secondary antibodies did not show any electron-dense particles on the surface of the cells or within the cells. Examples are given for ALP (Fig. 3A–C). The number of grains within a 0.2 µm×0.2 µm membrane area was <1.

If the cells were incubated in the absence of polyP and then reacted with anti-ALP antibody followed by gold/silver labeling (Fig. 3D–F), electron-dense gold/silver immune complexes could be seen rarely, and when present, they were found in vesicles (Fig. 3E,F). In contrast, if the cells were incubated with Na-polyP the immune–gold complexes were present in both vesicles and on the surface of the cells (Fig. 3G–I). Those cells present 3.2 grains within a cell membrane area of 0.2 µm×0.2 µm.

The distribution pattern of AK1 (Fig. 4) in SaOS-2 cells is similar to that seen with antibodies against ALP. After incubation of the cells in the absence of polyP the gold/silver antigen–antibody grains are distributed (almost) exclusively in vesicles (Fig. 4A,B). The number of grains associated with the cell membrane is <1 per 0.2 µm×0.2 µm membrane area. This is very much in contrast to what is seen in cells treated with Na-polyP. Under these conditions, the grains are densely packed within vesicles that also exist on the surface of the cells (Fig. 4E,F) with 2.8 grains per 0.2 µm×0.2 µm membrane area.

In a previous study (Müller et al., 2015), we reported on the change of the morphology of the cells from a more even cell surface in the absence of polyP, to a pronounced lobulated surface after incubation with polyP. This shaping was studied in more depth here by using the high-resolution transmission electron microscopy (TEM) technique (Fig. 4C,D,G,H). While the cells grown in the absence of polyP, also have an undulated cell surface, the pattern of protrusion formation is strongly intensified in polyP-treated SaOS-2 cells. It is striking that, very frequently, spherical vesicles accumulate in the nearest vicinity of the cells that appear to be released by the cells. The size of these vesicles measures between 250 and 450 nm, dimensions that are characteristic for matrix vesicles, and particularly vesicles that have been implicated in bone mineral deposition (Shapiro et al., 2015). Therefore, we operationally term these structures matrix vesicles (Fig. 4G,H).

We next tested whether gene expression of AK1, the most likely candidate for the cell surface-associated AK, depended upon the presence of polyP in SaOS-2 cells (Fig. 5). To check whether the gene is inducible, the cultures were incubated under standard serum (5%) conditions and under serum-starved (2.5%) culture conditions. The expression of the gene was determined by performing quantitative real-time PCR (qRT-PCR). The data show that the cells respond to a lowering of the serum concentration in the culture medium to 2.5% with a significant upregulation of the steady-state-expression of AK1 from 0.24±0.03 expression units, correlated to the expression of the reference gene GAPDH, to 0.51±0.06 units. In contrast, the level of AK1 expression remained unchanged within the concentration range of 3 to 30 µg/ml of polyP (Fig. 5); a level of around 0.25 expression units was measured.

Under defined culture conditions (10⁶ cells per ml) and 1 ml of incubation medium, the change in the extracellular ADP concentration was determined with a colorimetric/fluorometric assay system as described in the Materials and Methods. The initial ADP level at 30 min after inoculation, was low with ∼2.5 pmoles per 10⁶ cells for the controls (Fig. 6A); this level remained unchanged when the inhibitor A(5′)P5(5′)A (50 µM) was added. However, if the cultures were supplemented with 30 µg/ml of Na-polyP (Ca²⁺), a significant increase in the ADP level to 4.1±0.5 pmoles (per 10⁶ cells) was measured in the absence of A(5′)P5(5′)A and 5.3±0.6 pmoles in the presence of A(5′)P5(5′)A, with the difference between the two values obtained from polyP-treated cultures, minus and plus A(5′)P5(5′)A, being significant. During a longer incubation period, of 60 min and 180 min, the differences between the four assays became even more pronounced. Focusing on the 60 min incubation measurements, the values increased drastically and reached, in the polyP experiments, levels of 16.9±1.3 pmoles [minus A(5′)P5(5′)A] and 21.3±1.4 pmoles [plus A(5′)P5(5′)A]. Again, the amount of ADP in the polyP-treated cultures was significantly higher when A(5′)P5(5′)A was present. From these data, we conclude that increasing ADP is formed extracellularly in polyP-treated cultures; the increase is highest in cultures that additionally contain A(5′)P5(5′)A.

A parallel series of experiments was performed to determine the change in the extracellular ATP concentration in the culture medium for SaOS-2 cells in the presence of polyP. The ATP level was quantified by application of the luciferin-luciferase assay. In the absence of polyP, the level of ATP was low, with ∼125 pmoles (per 10⁶ cells) formed during the incubation period of 30 min. This level increased, again in the absence of polyP, to ∼220 pmoles (10⁶ cells) after 60 min and to ∼200 pmoles (10⁶ cells) after 180 min (Fig. 6B).

The ATP level in cultures that were exposed to polyP (30 µg/ml) were substantially higher (Fig. 6B). In the presence of A(5′)P5(5′)A, the level increased even after a 30 min incubation period to 631±48 pmoles and after 60 min to 713±61 pmoles (10⁶ cells). Subsequently after 180 min, the concentration slightly decreased. In the presence of A(5′)P5(5′)A, there was less ATP, at 319±22 pmoles after 30 min, 285±26 pmoles after 60 min (10⁶ cells) and to 328±31 pmoles after 180 min (10⁶ cells). These data indicate that, again, the ATP level is higher in polyP-supplemented cultures; however, addition of A(5′)P5(5′)A significantly reduces the ATP concentration.

Two inhibitors of the vesicular exocytosis, NEM and brefeldin A, were applied to clarify whether the change in the extracellular ATP is connected with an effect of polyP on the ATP release from SaOS-2 cells. The two inhibitors were added to the cultures at a concentration of 100 µM and the incubation was performed for 30 min, 60 min and 180 min in the absence of the AK inhibitor A(5′)P5(5′)A (Fig. 7). In the absence of polyP, the ATP level in the medium is low at 128.2±17.1 pmoles (10⁶ cells) (after an incubation period of 30 min); however, it increases significantly to 214.4±38.3 pmoles (10⁶ cells) and 185.3±31.3 pmoles (10⁶ cells) after 60 and 180 min, respectively (Fig. 7). The two inhibitors NEM and brefeldin A reduce significantly the ATP levels by more than 50%. In contrast, in the assays with cells incubated with polyP, the two inhibitors do not significantly alter the ATP level in the medium; in the presence of polyP, the ATP level is high at >500 pmoles (10⁶ cells).

The present study adds additional experimental evidence for an enzyme-driven generation of the energy-rich nucleotides ADP and ATP in the extracellular space. The data can be grouped into three blocks: first, the translocation of the key enzyme players involved in the generation of the metabolic fuel in the extracellular space, ALP (the ALPL form) and AK1, onto the cell surface in response to polyP exposure; second, the formation of matrix vesicles during the activation of the cells with this inorganic polymer; and third, the accumulation of ADP and ATP in the environment of the cells after incubation with polyP. For the studies presented here, the bone-cell-related osteoblastic osteosarcoma cell line SaOS-2 has been used. This selection was made since these cells show not only a high hydroxyapatite-mineralizing capacity but also express a high level of ALP (Murray et al., 1987). Furthermore, SaOS-2 cells strongly release ATP in the extracellular space (Buckley et al., 2003), a nucleotide that supports the hydroxyapatite-mineralizing capacity of cells (Orriss et al., 2009, 2013).

ALP is known to be an enzyme that is present in different isoforms within cells or bound to the cell membrane and even in a free state in the extracellular space (reviewed in: Dzeja and Terzic, 2009). The reaction catalyzed by ALP is irreversible, at least in eukaryotic systems (Kim and Wyckoff, 1991). Based on these data, it has been proposed that both intracellular and extracellular AKs play important roles during processes involved in nucleotide energetic signaling, for example, regulating actin assembly and disassembly during cell movement and chemotaxis. In our recent studies, we have focused on the generation of the energy-rich nucleotides by exo-enzymes that is required to initiate and maintain extracellular energy-consuming processes, like chaperone functions and extracellular nucleotide signaling (see Introduction). It is striking and compelling that both ALP, as we have recently shown (Müller et al., 2015), and the AK translocate to the cell membrane in response to exposure to polyP (see scheme in Fig. 8). This finding is a first compelling hint that ALP, as a processive exo-polyphosphatase (Lorenz and Schröder, 2001), catalyzes the stepwise hydrolysis of the energy-rich acid anhydride linkages of polyP. As outlined in the Introduction, it is thermodynamically persuasive and necessary to predict that the free energy liberated during the ALP reaction (ΔG negative) is partially converted into biochemically usable energy, meaning the formation of covalent bonds. The most likely candidates which can act as acceptors for the storage of the metabolic energy (energy-rich phosphates) are adenosine, AMP and ADP, which are nucleotides that are present in the extracellular space (Wink et al., 2003). The increase in the extracellular ADP level in response to polyP exposure strongly suggests that such a pathway exists. The underlying enzyme(s) are not yet known. Neither a polyphosphate:AMP phosphotransferase nor a polyphosphate:ADP phosphotransferase activity, as known from bacteria (Ishige and Noguchi, 2000, 2001; Resnick and Zehnder, 2000), that might act in concert with the AK have been detected in higher eukaryotic systems. On the other hand, a reversible formation of ADP from AMP and short-chain polyP (tripolyphosphate) by muscle adenylate kinase has been reported (Lieberman, 1956), even though the tripolyphosphate is only a weak phosphate donor for the vertebrate enzyme (Sanders et al., 1989).

The second enzyme, which is similarly translocated during polyP treatment is the AK. In contrast to the ALPL gene, which can be induced by polyP treatment (Müller et al., 2011), the AK1 gene, which we focused on as the most likely isoform (see Dzeja and Terzic, 2009), undergoes no gene induction after polyP treatment. In other systems, expression of the AK1 gene has been described as being promoted by inducible stimuli (Dzeja et al., 2011). Here, it is shown that AK1 is induced under serum starvation. The AK1 enzyme reversibly converts ADP into ATP and AMP. While polyP has no effect on gene induction in the cell system used here, this polymer causes an increase in intracellular grains that have sizes typical for matrix vesicles (250–450 nm; Shapiro et al., 2015) or acidocalcisomes (100–200 nm; Docampo et al., 2005).

Matrix vesicles are extracellular globular structures that are, besides other enzymes, also filled with ALP (reviewed in Golub, 2009). Those vesicles originate from intracellular vesicles that are released from the cells after activation (Maeda et al., 2002). Until now AKs have not been described to exist in matrix vesicles, while other enzymes, catalyzing the reversible metabolic formation of energy-rich phosphoanhydride bonds, for example, ecto-nucleotide pyrophosphatase/phosphodiesterase, are present in the vesicles (reviewed in Yegutkin, 2008). Since large numbers of matrix vesicles, or matrix vesicle-related structures, accumulate around polyP-exposed cells it is fascinating to accept, as a working hypothesis, that the ALP and AK enzymes are two of the main enzymes that are released from the interior of the cells to the exterior space via matrix vesicles.

In order to dissect the initial formation of AMP/ADP nucleotides, following hydrolysis of polyP by ALP, to the ATP generation step mediated by the AK, the cells were incubated with the AK inhibitor A(5′)P5(5′)A (Melnick et al., 1979). In the presence of A(5′)P5(5′)A, the ADP pool in the extracellular space of the cells increases several-fold. This boost is also seen at a reduced level in the absence of this inhibitor. This change in the nucleotide level is fairly rapid, from 60 min to 180 min, suggesting that the enzymes catalyzing the formation of these nucleotides from the precursor(s) are already present in the medium or are rapidly released by the cells, perhaps via the matrix vesicles. A more detailed elucidation of the enzymatic steps involved in this/these reactions(s) is in progress. Most interesting is the observation that the ATP level in the extracellular space strongly increases in response to polyP treatment. This amplification can be attributed to the function of the AK, since the ATP level is reduced to a control level if the inhibitor A(5′)P5(5′)A is present (Fig. 8). In ongoing studies, we determined that Na-polyP species with an average chain of 20 to 40 phosphate units are most effective at increasing the ATP level in the extracellular medium of SaOS-2 cells (our unpublished observations). Polymers/oligomers of polyP with shorter phosphate units, such as those with a chain length of six units [like Na-hexametaphosphate (Na₆O₁₈P₆)], were almost ineffective.

Previous studies have revealed that ATP is exported to the extracellular space via the vesicular exocytosis pathway (Orriss et al., 2009, 2013). This efficient export can be blocked by the specific inhibitors NEM and brefeldin A; these inhibitors are non-toxic to the cells. In the present study, we performed a series of experiments to clarify whether the alteration of the ATP level measured in the extracellular medium in response to polyP is modified by these two inhibitors. As expected, and in line with the published data (Orriss et al., 2013), it was quantified that the ATP level drastically (by more than 50%) drops in the assays without polyP during the 30 to 180 min incubation period in the presence of NEM and brefeldin A. However, the ATP level in the medium remains constant in assays containing polyP, irrespective of the presence of these two inhibitors. This finding suggests that the hypothesized polyP-dependent ATP-generating enzyme system(s) either could activate the ATP exocytosis system or counterbalance the ATP reduction. This aspect needs to be clarified in the future.

The data collected in the present study show that there is polyP-driven formation of ADP and ATP in the extracellular space (Fig. 7). The results are in accordance with the finding that a larger amount of ADP is formed in extracellular space if the cells are incubated with polyP. This effect is especially obvious if the initial reaction, ADP formation, is uncoupled from the second ATP-generating AK system. This isolated initial process surely involves the hydrolysis of the high-energy phosphoric anhydride bonds by ALP. The catabolic reaction, hydrolysis of the anhydride bonds, also requires a coupled anabolic enzyme that catalyzes the formation of ADP from the less energy-rich nucleoside/nucleotide precursor under consumption of metabolic energy released during the ALP reaction. ADP is very likely an energy donor for a series of reactions or functional proteins in the cytoplasm, for example, for clusterin. Since, in the absence of the A(5′)P5(5′)A, the ALP-mediated hydrolysis and the proposed subsequent metabolic energy-conserving reaction (ADP formation) are coupled with the AK reaction, a strong elevation of the extracellular ATP pool is seen. This nucleotide then drives the ecto-kinase reactions as well as the transmission signaling originating from the purinergic receptor(s) (Cekic and Linden, 2016). In this context, it is interesting to note that polyP amplifies inflammatory signaling in human endothelial cells through interaction with the P2Y purinoceptor 1, resulting in a release of Ca²⁺ from internal stores (Dinarvand et al., 2014). The same effect was determined for rat neurons and astrocytes; there, the inorganic polymer interacts with the P2Y₁ purinergic receptor, again mobilizing intracellular Ca²⁺ (Holmström et al., 2013).

Sodium polyphosphate (Na-polyP of an average chain of 40 phosphate units; M_r 4000) was obtained from Chemische Fabrik Budenheim (Budenheim, Germany). P¹,P⁵-di(adenosine-5′) pentaphosphate (pentasodium salt; A(5′)P5(5′)A; #D 4022) was from Sigma, Taufkirchen, Germany.

Human osteogenic sarcoma cells (SaOS-2 cells; Fogh et al., 1977; purchased from Sigma 89050205) were cultured in McCoy's medium (containing 1 mM CaCl₂) with 5% heat-inactivated fetal calf serum (FCS), 2 mM l-glutamine and gentamicin (50 μg/ml) in 25 cm² flasks or in six-well plates (Sigma-Greiner) in a humidified incubator at 37°C (Wiens et al., 2010a). The cells were seeded at a density of 2×10⁴ cells per 3 ml well in 24-multi-well plates (Sigma) and cultured for 3 days in McCoy's medium with 5% FCS. Increasing concentrations of Na-polyP were added to the cultures together with CaCl₂ [Na-polyP (Ca²⁺)]; in order to prevent depletion of Ca²⁺ in the medium (Müller et al., 2011), the concentration of Ca²⁺ in the medium was adjusted to 2 mM. Where indicated the cells were pre-incubated for 3 days in the presence of the mineralization activation cocktail (MAC; 5 mM β-glycerophosphate, 50 mM ascorbic acid and 10 nM dexamethasone to induce biomineralization), as described previously (Wiens et al., 2010b). Then they were transferred into 24-multi-well plates and continued to be incubated. In order to inhibit AK, 40 µM A(5′)P5(5′)A was added to the cells when indicated. In the studies to determine the extracellular ATP and ADP levels after 30 min, 60 min or 180 min at a cell concentration of 10⁶/ml, the cells were pre-incubated with the AK inhibitor A(5′)P5(5′)A at a concentration of 50 µM for 1 h (Quillen et al., 2006). If not mentioned otherwise, the cells were routinely incubated with 30 µg/ml of Na-polyP (Ca²⁺). The effect of the two inhibitors of the vesicular exocytosis, N-ethylmaleimide (NEM; no. E3876 Sigma) and of brefeldin A (no. B7651 Sigma), was determined at a concentration of 100 µM. Incubation periods of 30 min, 60 min and 180 min had been selected.

Growth/viability of the cells was determined by the colorimetric method based on the tetrazolium salt XTT (Cell Proliferation Kit II; Roche, Mannheim, Germany) (Mori et al., 2007). The OD₅₀₀ was determined and background values were subtracted.

Polyclonal antibodies against AK1 [AK1 antibody (H-90): sc-28785; raised in rabbits; Santa Cruz Biotechnology, Dallas, TX] were used at a dilution of 1:2000. The cells were fixed in 70% ethanol (−20°C; 10 min) and washed in 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 30 min. Then, the samples were treated with 3% bovine serum albumin (BSA; Sigma) and – after washing with PBS – incubated with secondary antibodies (goat anti-rabbit-IgG labeled with Cy3; Dianova, Hamburg; Germany). The cell nuclei were stained with DAPI [2-(4-amidinophenyl)-1H-indole-6-carboxamidine; Sigma]. Finally, the slices were inspected with an Olympus AHBT3 microscope under immunofluorescence light with excitation light wavelengths suitable for either Cy3 or DAPI. ImageJ (v1.48, NIH) was used applied to relatively quantify the fluorescence intensity in the cells (plug-in available at https://imagej.nih.gov/ij/plugins/lut-editor.html): cell area and mean fluorescence were measured together with several adjacent background readings. Subsequently, the total corrected cellular fluorescence (TCCF) was calculated. This TCCF was then equalized against the mean TCCF of the controls (absence of Na-polyP). The results of the fluorescence originating from cells exposed to polyP were obtained in the same field and expressed as a fold increase over the controls (McCloy et al., 2014).

Transmission electron microscopic (TEM) observations of the immune-labeled samples were performed with a Tecnai 12 microscope (FEI Electron Optics, Eindhoven; Netherlands) as described previously (Kokkinopoulou et al., 2014; Müller et al., 2015). In brief, the cells were fixed in paraformaldehyde and glutaraldehyde, embedded in agarose and then in LR-white resin (no. 62661; Sigma-Aldrich). Then ultrathin sections (80 nm) were cut (Microsystems, Wetzlar; Germany) which were stained with antibodies (polyclonal AK1 antibodies no. A4107, Sigma-Aldrich) and incubated with a secondary anti-rabbit antibody coupled to 10-nm gold particles (1:50 diluted with water, cat. ab39597 Abcam, Cambridge, UK). In a second series of experiments, the samples were reacted with a monoclonal IgG antibody against human ALP (1:2000, produced in mouse, no. ab33, Abcam). Immunogold labeling was performed with a secondary anti-mouse-IgG antibody coupled to 10-nm gold particles. The samples were enhanced with silver (Danscher, 1981), and contrasted with uranyl acetate and lead citrate.

Semi-quantitative analysis of the labeling density of the gold-silver grains on the cell membrane was performed by counting of defined areas for the respective cell membranes. Images of cells were taken with a SEM under the same conditions. Then, membrane areas measuring 0.2 µm×0.2 µm were selected randomly, and counted. A total of 20 cells were examined; 10–20 membrane areas were counted per cell and the average number of particles is given.

High-resolution TEM was performed with a Tecnai F20 electron microscope (FEI, Gräfelfing; Germany).

The SaOS-2 cells were incubated for 5 days in the absence of the MAC in 48-well plates (initial density of 1×10⁵ cells/well) at different polyP (as a Ca²⁺ salt) concentrations. After terminating the reaction, total RNA was extracted and freed from possible DNA contamination by using DNase. Then first-strand cDNA was synthesized by using the M-MLV RT (Promega, Mannheim; Germany). Finally, qRT-PCR was performed with ∼5 µg of total RNA in reaction mixtures of 40 µl using the iCycler (Bio-Rad, Hercules; CA) and 1:10 serial dilutions in triplicate. The reactions were run using the temperature cycles described previously (Wiens et al., 2010b). The expression levels of AK1 were correlated to the reference gene GAPDH, essentially as described previously (Wiens et al., 2010b).

The following primer pairs were used. For AK1 expression, the sequence of Homo sapiens was taken (cDNA clone MGC:1808 IMAGE:2988248, complete cds; accession number GenBank BC001116.2): fwd, 5′-GACGCCCTAAAGTAGCAACG-3′ (nt₆₄₂ to nt₆₆₁) and rev, 5'-GTGCTCAGCTGTCCATGAAA-3′ (nt₇₈₆ to nt₈₀₅). As the reference, the expression of the human GAPDH gene was used (NM_002046.3): fwd, 5′-CCGTCTAGAAAAACCTGCC-3′ (nt₈₄₅ to nt₈₆₃) and rev, 5′-GCCAAATTCGTTGTCATACC-3′ (nt₁₀₅₉ to nt₁₀₇₈; 215 bp).

The determination of the extracellular ATP level was performed as described before (Ahmad et al., 2004; Orriss et al., 2009, 2013; Müller et al., 2015). Immediately before the determination of the ATP level, the cells were transferred into medium without serum and continued to be incubated for an additional period of 30 min, 60 min or 180 min at a cell concentration of 10⁶ cells/ml. Then 1 ml of the conditioned medium was collected and transferred into chilled polypropylene tubes (#Z334006 Sigma). After centrifugation (12,000 g) for 5 min, aliquots (100 µl) from the supernatant were used in a luciferin-luciferase assay (no. LL-100-1, Kinshiro, Toyo Ink; Japan) carried out as described previously (Müller et al., 2015). An ATP standard curve with seven points was generated with known amounts of ATP. The standard curve ranged around the ATP level present in the experimental system. The ATP concentration determined is given as pmol/10⁶ cells (1 ml aliquot of medium). The extracellular ADP concentration was determined by use of an ADP colorimetric/fluorometric assay kit (ab83359; Abcam, Cambridge, UK), following the protocol described previously (Mayeur et al., 2013). The ATP and ADP levels were measured in five parallel assays.

After verification that the respective values follow a standard normal Gaussian distribution and that the variances of the respective groups are equal, the results were statistically evaluated using the independent two-sample Student's t-test (Petrie and Watson, 2013).

We thank Ms Maren Müller (Physics at Interfaces) and Mr G. Glaßer (Electron Microscopy), Max Planck Institute for Polymer Research, Mainz (Germany) for continuous support and excellent guidance.