The FASEB Journal express article10.1096/fj.02-1136fje. Published online May 20, 2003.

Soluble purine-converting enzymes circulate in human blood

and regulate extracellular ATP level via counteracting

pyrophosphatase and phosphotransfer reactions

Gennady G. Yegutkin, Sergei S. Samburski, and Sirpa Jalkanen

MediCity Research Laboratory and Department of Medical Microbiology, Turku University, and

National Public Health Institute, FIN-20520 Turku, Finland

Corresponding author: Gennady G. Yegutkin, MediCity, Tykistökatu 6A, FIN-20520 Turku,

Finland. E-mail: gennady.yegutkin@utu.fi

ABSTRACT

Extracellular ATP and other purines play a crucial role in the vasculature, and their turnover is

selectively governed by a network of ectoenzymes expressed both on endothelial and

hematopoietic cells. By studying the whole pattern of purine metabolism in human serum, we

revealed the existence of soluble enzymes capable of both inactivating and transphosphorylating

circulating purines. Evidence for this was obtained by using independent assays, including

chromatographic analyses with 3H-labeled and unlabeled nucleotides and adenosine, direct

transfer of γ-terminal phosphate from [γ-32

P]ATP to NDP/AMP, and bioluminescent

measurement of ATP metabolism. Based on substrate-specificity and competitive studies, we

identified three purine-inactivating enzymes in human serum, nucleotide pyrophosphatase (EC

3.6.1.9), 5′-nucleotidase (EC 3.1.3.5), and adenosine deaminase (EC 3.5.4.4), whereas an

opposite ATP-generating pathway is represented by adenylate kinase (EC 2.7.4.3) and NDP

kinase (EC 2.7.4.6). Comparative kinetic analysis revealed that the Vmax values for soluble

nucleotide kinases significantly exceed those of counteracting nucleotidases, whereas the

apparent Km values for serum enzymes were fairly comparable and varied within a range of 40�

70 µmol/l. Identification of soluble enzymes contributing, along with membrane-bound

ectoenzymes, to the active cycling between circulating ATP and other purines provides a novel

insight into the regulatory mechanisms of purine homeostasis in the blood.

Key words: purine metabolism • human serum • soluble nucleotide kinases • adenosine

E xtracellular ATP and other purines mediate diverse signaling events in several tissues,

including the vascular system. ATP/ADP are released into the bloodstream from the

vascular endothelium, smooth muscles and circulating blood cells through the opening of

channel-like pathways, via vesicular exocytosis, or leakage upon cells lysis (1). Once released,

they influence vascular tone and platelet aggregation and mediate several inflammatory

responses of lymphocytes, granulocytes, and macrophages (1�4). Most of these effects are

mediated through G protein-coupled P2Y receptors as well as via �cytolytic� P2X7 and other

P2X receptors with intrinsic pore-forming activity (5). Subsequent to signal transduction,

circulating ATP/ADP are rapidly inactivated to AMP by the enzymes of E-type NTP-

diphosphohydrolase family (E-NTPDase; previous names ecto-ATPDase, CD39) and further to

adenosine via ecto-5′-nucleotidase/CD73 reaction (6�8). These two ecto-nucleotide hydrolases

are ubiquitously coexpressed among the endothelial and hematopoietic cells and are considered

to be the major regulators of the duration and magnitude of purinergic signaling in the blood (2,

9). General schemes of nucleotide inactivation have also included a role for alkaline

phosphatases, phosphodiesterases, and the members of the ecto-nucleotide pyrophosphatase (E-

NPP) family, otherwise known as plasma cell membrane glycoprotein-1 (PC-1), autotaxin, B10

(10, 11). The resulting adenosine interacts with its own G protein-coupled receptors (5) and

usually serves as a powerful antiinflammatory agent with effects that are opposite of ATP/ADP

(12�14). Subsequently, the fate of adenosine is defined by its uptake into the cells via

equilibrative or Na+-dependent nucleoside transporters (15); however, an alternative possibility

of its extracellular deamination by ecto-adenosine deaminase anchored to the lymphocyte surface

via CD26/dipeptidyl peptidase IV has been recently proposed (16, 17). Along with purine-

inactivating pathways, backward nucleotide resynthesis via sequential ecto-nucleotide kinase-

mediated phosphotransfer reactions AMP → ADP → ATP has been described recently by us (8,

17) and others (18�20).

In contrast to traditional paradigms that focus on cell-associated mechanisms of extracellular

purine turnover, the existence of soluble enzymes in the blood and their contribution to the

metabolism of circulating purines have received little attention. Although human plasma was

shown to contain the major purine-degrading enzymes, including E-NPPs (21�23), E-NTPDases

(24, 25), adenosine deaminase (26), alkaline phosphatases, and 5′-nucleotidase (27), these

enzymic activities were primarily considered in the context of their clinical applications as

potential diagnostic tools in various (patho)physiological conditions. These studies aimed to

evaluate the whole pattern of purine metabolism in human serum, and the results presented

below demonstrate the existence of an extensive network of soluble enzymes coordinately

mediating both degradation and transphosphorylation of circulating purines.

MATERIALS AND METHODS

Serum preparation

Blood was taken from healthy volunteers by venipuncture and was allowed to clot at room

temperature for 30 min. Blood samples were centrifuged for 15 min at 2000g followed by

freezing of the serum at �70°C. Serum was also obtained from White New Zealand rabbits and

from FVB/n mice by collecting the blood from the heart immediately after the animals were

killed.

Thin layer chromatography (TLC) analysis of the metabolism of 3H-labeled nucleotides

and adenosine

Nucleotide-converting activities were assayed with [2,8-3H]ATP (specific activity 19 Ci/mmol;

Sigma, St. Louis, MO), [2,8-3H]ADP (25.3 Ci/mmol; Sigma), and [2-

3H]AMP (18.6 Ci/mmol;

Amersham, Little Chalfont, UK) as appropriate substrates. Serum (5�7 µl) was incubated at

37°C in a final volume of 60 µl RPMI-1640 medium containing 5 mmol/l β-glycerophosphate,

the indicated concentrations of the substrate with tracer 3H-nucleotide (∼4 × 10

5 dpm), and γ-

phosphate-donating ATP/NTP (in case of nucleotide kinase studies). Incubation times were

chosen to ensure the linearity of the reaction with time, so that the amount of the converted

nucleotides did not exceed 15% of the initially introduced substrate. Aliquots of the mixture

were applied to Alugram SIL G/UV254 sheets (Macherey-Nagel, Duren, Germany) and separated

by TLC with isobutanol/isoamyl alcohol/2-ethoxyethanol/ammonia/H20 (9:6:18:9:15) (Fig. 1A). 3H-labeled substrates and their derivatives were visualized in UV light and quantified by

scintillation β-counting, as described earlier (28). Likewise, for measurement of adenosine

deaminase, serum was incubated with [2-3H]adenosine (24 Ci/mmol; Amersham) followed by

separation of 3H-nucleosides with isobutanol/ethyl acetate/methanol/ammonia (7:4:3:4) (Fig.

1B).

High-pressure liquid chromatography (HPLC) analysis of nucleotide metabolism

Serum (20 µl) was incubated for 30 min at 37°C with 50 µM NDP/NTP in the final volume of

300 µl RPMI-1640 medium followed by immediate extraction of exogenous nucleotides with

200µl of 0.6 mol/l perchloric acid. The obtained supernatant (450 µl) was neutralized with ~50

µl 4N KOH, clarified by centrifugation, and injected onto a 4-µm, 150 × 4.6 mm, Synergi

Hydro-RP 80A column protected by reversed-phase C18 guard cartridges (all from Phenomenex,

Torrance, CA). Buffer A was 0.1 mol/l KH2PO4 (pH 6.0) supplemented with 1 mmol/l ion

pairing reagent tetrabutylammonium dihydrogen phosphate (Fluka Chemica, Buchs,

Switzerland), and buffer B was a 15% (v/v) solution of methanol in buffer A. The amount of

buffer B was changed linearly between the following time points: 0 min, 0%; 1 min, 25%; 14

min, 35%; 18 min, 100%; 25 min, 100%; and 30 min, 0%. The reequilibration time was 10 min

with buffer A, resulting in a total cycling time of 40 min between injections. The flow rate was 1

ml/min, and peaks were monitored by absorption at 258 nm. Individual standard curves were

constructed with the use of known concentrations of nucleotide and nucleoside standards injected

vs. the observed chromatographic peak areas (Fig. 1C).

ATP quantification by luciferin-luciferase assay

Serum (15 µl) was incubated at room temperature in a final volume of 200 µl RPMI-1640

containing 1 mmol/l β-glycerophosphate and various NDP/NTP. Subsamples of the medium

were mixed with EDTA at a final concentration of 2 mmol/l, and ATP concentration was assayed

as described previously (29).

Transphosphorylation of [γ-32

P]ATP with AMP/NDP

Human serum (10 µl) was incubated for 30 min at 37°C in a final volume of 30 µl RPMI-1640

containing 1 mmol/l β-glycerophosphate, 1 µmol/l ATP with tracer [γ-32

P]ATP (3000 Ci/mmol;

Perkin Elmer, Boston, MA), and 50 µmol/l AMP/NDP, unless otherwise stated. In some studies,

mouse and rabbit serum (10 µl) and human leukemic T-cell lymphoblast line Jurkat (5 × 104

cells; ATCC) were used as enzyme sources instead of human serum. Aliquots of the reaction

mixture were spotted onto Polygram CEL-300 PEI sheets (Macherey-Nagel) and separated by

TLC with 0.75 mol/l KH2PO4 (pH 3.5).

Data analysis

Pooled data are presented as mean ±SE, where n represents the number of experiments performed

with separate serum samples in duplicates. Statistical comparisons were made using Student�s t

test, and P<0.05 were considered significant. Data from kinetic experiments were subjected to

computer analyses, using the Michaelis-Menten equation to determine the Km and Vmax values

(GraphPad Prism, version 3.03, San Diego, CA).

RESULTS

Catabolism of 3H-labeled purines in human serum

In the initial time-course experiments, human serum was incubated with 20 µmol/l [3H]ATP

followed by TLC separation of 3H-labeled ATP and its metabolites. Progressive decay of

[3H]ATP concentration was accompanied by accumulation of [

3H]adenosine in the assay

medium, and there was a little formation of [3H]AMP as an intermediate metabolite (Fig. 2A).

Use of a higher, saturating, concentration of [3H]ATP in additional quantitative and competitive

studies confirmed the presence of ATP-hydrolyzing activity in human serum that can be

abolished by the chelating agent EDTA (Fig. 2B). [3H]ATP hydrolysis in the serum remained

insensitive to inhibitors of ion-exchanging (ouabain) and intracellular (sodium azide, ortho-

vanadate) ATPases but was partially inhibited by trypanocidal agent suramin concurrently

serving as antagonist of P2 receptors and inhibitor of E-type ATPases (Fig. 2B). Serum was also

capable of hydrolyzing [3H]AMP, and this nucleotidase reaction can be prevented with the

specific ecto-5′-nucleotidase inhibitor α,β-methylene-ADP (APCP) but not with suramin (Fig.

2C). Together, these findings suggest the presence of at least two nucleotide-hydrolyzing

activities in human serum that display strong cation dependence, coordinately degrade ATP via

AMP to adenosine, and are differently modulated by inhibitors.

Although adenosine was the principle metabolite formed in the course of [3H]ATP hydrolysis,

slight generation of 3H-labeled inosine and hypoxanthine suggests the presence of other soluble

purine-inactivating enzymes, adenosine deaminase and probably purine nucleoside

phosphorylase, in human serum (Fig. 2A). For direct quantification of adenosine deaminase

activity, we used [3H]adenosine as initial substrate and particular solvent mixture that enabled

better TLC separation of 3H-labeled adenosine and its nucleoside metabolites (see Materials and

Methods and Fig. 1). Indeed, human serum successfully deaminates [3H]adenosine, and this

catalytic reaction was inhibited by ∼95% in the presence of adenosine deaminase inhibitor

EHNA (Fig. 2C).

Transphosphorylation of 3H-labeled and unlabeled nucleotides in human serum

To test the alternative possibility of backward ATP resynthesis in cell-free serum, we determined

the conversion of 3H-labeled AMP/ADP into high-energy

3H-phosphoryls in the presence of

various NTP. Unlabeled NTP activated [3H]AMP phosphorylation with the following rank order

of γ-phosphate-donating potency: ATP >> UTP > ITP ≥ GTP (Fig. 3A). Use of [3H]ADP as

another phosphate acceptor revealed even more efficient [3H]ADP conversion into [

3H]ATP and

this phosphoryl transfer can be equally activated by both ATP and other NTP (Fig. 3B).

Metabolism of both 3H-nucleotides was significantly inhibited by EDTA, whereas the specific

adenylate kinase inhibitor Ap5A prevented ATP-mediated phosphorylation of only [3H]AMP but

not [3H]ADP. These data suggest the existence of two different phosphorylating activities in

human serum capable of mediating sequential ATP/NTP-induced AMP conversion via ADP into

ATP and possessing hallmark characteristics of adenylate kinase and NDP kinase.

Kinetic analysis of soluble purine-converting enzymes

In the context of a complex mixture of enzymes sharing similarities in substrate specificity, the

role of individual components primarily depends on the enzyme kinetics and local substrate

concentrations. For kinetic analysis of 5′-nucleotidase and adenosine deaminase, serum was

incubated with [3H]AMP and [

3H]adenosine as corresponding substrates and Fig. 4A displays the

enzyme velocities plotted vs. the increasing substrate concentrations. Concentration-response

curves for nucleotide kinases were generated under conditions that approximated first-order rates

of reaction, that is, γ-phosphate-donating ATP was maintained at a nearly saturated level,

whereas the concentration of phosphate acceptor, [3H]AMP or [

3H]ADP, was varied (Fig. 4B).

All enzyme assays were conducted under the same initial rate conditions, and the kinetic

parameters were calculated using the Michaelis-Menten equation. Except for adenylate kinase

with relatively low substrate affinity, the apparent Km values for other serum enzymes, 5′-

nucleotidase, adenosine deaminase, NDP kinase (Fig. 4C), and ATPase (30), are fairly

comparable and vary within a range of 40�70 µmol/l. However, the Vmax values for adenylate

kinase and especially NDP kinase significantly exceed those of purine-inactivating enzymes

(Fig. 4C), suggesting the predominance of phosphotransfer reactions under simultaneous

appearance of ATP and NDP/AMP in the external milieu.

HPLC analysis of nucleotide metabolism in human serum

Nucleotide-converting activities were also evaluated by using HPLC. Nucleotides and

nucleosides were identified by construction of the individual standard curves (see Fig. 1C) as

well as by internal injections of authentic standards. The major peak corresponding to

hypoxanthine was clearly detected in the case of control nontreated serum (Fig. 5A), and these

findings are consistent with earlier established �hypoxanthine pool� freely circulating in human

blood (31). Addition of exogenous ADP to the serum caused its simultaneous conversion into

AMP and ATP presumably via reverse adenylate kinase reaction (Fig. 5B), whereas GTP alone

did not induce significant shifts in serum purine metabolism (Fig. 5C). Combined serum

treatment with ADP plus GTP dramatically activated NDP kinase-mediated phosphotransfer

reactions through the following scheme: ADP + GTP ↔ ATP + GDP (Fig. 5D).

Bioluminescent determination of ATP metabolism in human serum

Phosphotransfer reactions in human serum were independently measured by using a luciferin-

luciferase assay. Serum incubation with 4.5 µmol/l ADP caused detectable ATP formation (Fig.

6A; inset), and these data fit well with the above radio-TLC measurements of [3H]ADP

phosphorylation into [3H]ATP (see Fig. 3B; first bar) and with HPLC analysis of ADP

metabolism (see Fig. 5B). Together, these findings can be reasonably explained by reverse

soluble adenylate kinase reaction mediating transphosphorylation of two ADP molecules into

ATP and AMP. However, the most pronounced ATP synthesis was observed under combined

serum exposure to ADP and various NTP (Fig. 6A) confirming the necessity of a γ-phosphate-

donating NTP for efficient ADP conversion into ATP via soluble NDP kinase reaction. Vice

versa, serum incubation with ATP alone induced its modest hydrolysis, whereas UDP, IDP, and

to a lesser extent, GDP accelerated ATP metabolism via activation of phosphotransfer reactions

(Fig. 6B).

Metabolism and transphosphorylation of [γ-32

P]ATP

Serum was incubated with [γ-32

P]ATP, and Figure 7 illustrates the major interconversion

pathways of this nucleotide. Incubation of human serum with 1�50 µmol/l [γ-32

P]ATP caused

progressive nucleotide hydrolysis with concomitant formation of 32

PPi (lanes 2, 8, 9), and this

pyrophosphatase reaction can be prevented by EDTA (lane 10). Likewise, rabbit serum also

predominantly degrades [γ-32

P]ATP via NPP reactions (lane 13), whereas incubation of [γ-32

P]ATP with mouse serum caused the appearance of both 32

PPi and 32

Pi (lane 12). The revealed

patterns of [γ-32

P]ATP degradation in human and rabbit serum show clear-cut differences from

the E-NTPDase-mediated sequential cleavage of 32

Pi from [γ-32

P]ATP found in the experiments

with Jurkat T lymphocytes (lane 11) and other lymphoid and endothelial cells (8, 17). In addition

to identification of soluble NPP as the major ATP-degrading enzyme in human serum, use of [γ-32

P]ATP allowed us to trace transfers of the γ-terminal 32

P-phosphate in the course of soluble

phosphotransfer reactions. Addition of [γ-32

P]ATP together with unlabeled UDP, GDP, or ADP

to the human serum activated NDP kinase reaction with the formation of corresponding [γ-32

P]NTP (lanes 3�5). Serum incubation with [γ-32

P]ATP plus AMP caused their

transphosphorylation into [β-32

P]ADP (lane 6), and this adenylate kinase reaction can be

prevented by Ap5A (lane 7). Notably, Ap5A concurrently diminished NPP-mediated [γ-32

P]ATP

hydrolysis to 32

PPi (lane 7), which may be reasonably explained by the ability of NPPs to

metabolize diadenosine polyphosphates as an alternative substrates (11).

Assessment of reliability of the used approaches

Studies on purine metabolism in serum or plasma must be performed and interpreted with certain

caution because even slight hemolysis during blood sampling would induce the appearance of

artificially high levels of ATP and other nucleotides. However, concentration of ATP in the

analyzed serum samples was <10−8

mol/l as measured by highly sensitive luciferin-luciferase

assay. In addition, HPLC analysis did not reveal any traces of other NTP/NDP even at the

highest sensitivity range (detection limit ∼5�10 pmol per injection). Use of a large excess of β-

glycerophosphate as an alternative phosphorylated substrate in all enzyme assays also allowed us

to exclude the potential contribution of nonspecific phosphatases in the measured activities.

Furthermore, additional ultracentrifugation of the serum (60 min at 100,000g) did not

significantly affect the activities of soluble enzymes, thus excluding the possible contribution of

membrane-bound ectoenzymes that could be associated with fine cell fragments and not

precipitated during routine preparation of the blood samples.

DISCUSSION

By using radio-TLC and HPLC assays as the most sensitive and versatile approaches for

screening the metabolism of extracellular purines, we evaluated the whole pattern of purine

metabolism in human serum and further identified the major soluble activities involved in these

interconversion pathways. The elements of the purine-inactivating cascade comprise at least

three separate enzymes displaying ATP-pyrophosphohydrolyzing, AMP-hydrolyzing, and

adenosine-deaminating activities, whereas an opposite ATP-regenerating pathway is mediated

via sequential adenylate kinase and NDP kinase reactions (Fig. 8).

An unambiguous distinction between the members of the E-NTPDase and the E-NPP family is

complicated by their coexpression to a variable extent among the mammalian tissues, sharing

similarities in substrate specificity (10, 11), similar pattern of inhibition by some P2 antagonists

such as suramin and pyridoxalphosphate azophenyl disulphonic acid (PPADS), and insensitivity

to the classic inhibitors of intracellular ATPases (32�34). The E-NPPs are particularly abundant

in fibroblasts, chondrocytes, and osteoblasts, where they regulate soft tissue calcification and

bone mineralization (35), whereas E-NTPDases dominate in the vascular endothelium and are

usually considered to be the major regulators of blood fluidity and platelet activation (2, 4).

Soluble NPPs can also be liberated via proteolysis of the parent membrane-bound molecules and

are primarily represented in plasma by 127-kDa isozyme NPP1 (otherwise known as PC-1) (35,

36). Serum PC-1 levels have been shown to be elevated in patients with degenerative arthritis

and scleroderma (23) and decreased in idiopathic infantile arterial calcification (21). Soluble

ATP-hydrolyzing activities have also been described in human plasma (24, 25, 30).

Present TLC studies with [γ-32

P]ATP as substrate clearly demonstrate that practically all ATP in

human (and rabbit) serum is directly converted into AMP, presumably via NPP1/PC-1-mediated

pyrophosphatase reaction. The patterns of ATP hydrolysis in mouse (see Fig. 7) and rat (30)

serum samples suggest the existence of other nucleotidases in rodent blood. These enzymes may

include other members of the NPP family such as NPP2/autotaxin, which can also be secreted to

the bloodstream, sequentially hydrolyze both ATP and ADP, and is even capable of cleaving PPi

to two Pi (11, 35). Alternatively, we do not exclude the coexistence of two independent, NPP-

like and NTPDase-like, activities in mouse serum. Furthermore, recent data on the secretion of

soluble E-NTPDases from the nerve terminals (37) and from vascular endothelial cells (29)

suggest the possibility of transient appearance of soluble NTPDases in human blood under

certain (patho)physiological conditions and their contribution to local metabolism of

extracellular ATP.

Circulating AMP can be degraded to adenosine and further to inosine by serum 5′-nucleotidase

and adenosine deaminase, respectively. The 5′-nucleotidase/CD73 is an ubiquitously expressed

ectoenzyme attached to the membrane matrix via GPI anchor (6, 38) and, under various stimuli,

it can be released into external milieu from nerve terminals (37), endothelial cells (29), and

lymphocytes (39). The major source of soluble 5′-nucleotidase in the blood is thought to be the

plasma membranes of hepatocytes, and serum 5′-nucleotidase is increased in certain

hepatobiliary, hematological, immunological, and neoplastic disorders (27). Another purine-

inactivating enzyme, adenosine deaminase, is widely expressed in intestine, thymus, and other

lymphoid and nonlymphoid tissues (13, 40). Serum also contains soluble adenosine deaminase

that seems to originate from lymphocytes or the monocyte/macrophage lineage, and its activity is

changed in various diseases in which the immune status is changed, such as rheumatoid arthritis

and tuberculosis (26). Although the existence of purine-converting enzymes has been described

earlier in human plasma, here for the first time we performed comparative and kinetic analyses

of these soluble enzymes.

Understanding the impact of serum enzymes on the purine homeostasis within the larger

framework of membrane-associated ecto-nucleotidases is another important issue. Endothelial E-

NTPDases, in concert with ecto-5′-nucleotidase, are thought to be the major terminators of the

proinflammatory and prothrombotic effects of ATP/ADP in the vasculature (9). However,

endothelial metabolism prevails in the microcirculation, where the area of endothelial surface

exposed to unit volume of blood is very high, whereas within large vessels, the capacity to

metabolize circulating purines resides approximately equally at the luminal surface of the vessel

wall and in the flowing blood (24). Soluble nucleotidases can also potentially offset the reduced

scavenging effectiveness of membrane-bound ecto-nucleotidases under certain inflammatory

conditions or traumatic shock where the integrity of a blood vessel wall is compromised. In this

context, it is pertinent to note that the recombinant soluble form of human NTPDase/CD39 was

capable of potently inhibiting thrombosis and tissue injury even when given to mice 3 h after

stroke (41), and it is currently considered a promising aspirin-insensitive antithrombotic drug (2,

4). Importantly, contrary to the endothelial and lymphocyte E-NTPDases, serum NPPs directly

convert ATP into AMP and PPi, thereby bypassing the generation of a principal platelet-

recruiting agent ADP. In addition, because PPi is one of the major physiological inhibitors of

calcification, preventing basic calcium phosphate (hydroxyapatite) crystal deposition in bone and

cartilage (35), serum NPPs may concurrently serve as powerful physiological inhibitors of

mineralization in vivo.

Another important novelty of this work is the demonstration of the hitherto unrecognized ATP-

generating pathway in human serum that is sequentially mediated via phosphotransfer reactions.

Although both adenylate kinase and NDP kinase have been traditionally viewed as intracellular

regulators of endogenous nucleotide pools, recent studies demonstrated the expression of these

ectoenzymes on the surface of endothelial cells (8, 19), lymphocytes (17), and other cell types

(18, 20). Interestingly, a certain proportion of ecto-nucleotide kinases can be secreted from

airway submucosal glands in response to cholinergic stimuli (20). The identification of soluble

NDP kinase capable of rapidly (within minutes) and indifferently transphosphorylating

micromolar concentrations of nucleoside di- and triphosphates may represent an auxiliary and

currently unappreciated route regulating the pattern of nucleotide receptor stimulation and/or

desensitization in the blood. Particularly, in the settings of thrombus formation, where ADP and

other nucleotides are massively released from dense granules of activated platelets and injured

endothelium (2, 4), rapid interconversion of ADP/NTP by serum NDP kinase could prevent

recruitment of additional platelets via a directional switch between purine- and pyrimidine-

selective receptors. Serum adenylate kinase may also contribute to the clearance of circulating

ADP through its reverse transphosphorylation into ATP and AMP.

In conclusion, the results presented here demonstrate the coexistence of two opposite, ATP-

generating and ATP-consuming, pathways in cell-free human serum, and these findings may

significantly add to our understanding of the regulation of the duration and magnitude of

purinergic signaling in the blood. In addition, there is increasing interest in the therapeutic

potential of purinergic compounds, and these strategies are likely to include search for selective

agonists and antagonists for P1 and P2 receptor subtypes (42), regulation of extracellular purine

metabolism by using antiinflammatory drugs (14), and enzyme therapies with recombinant

soluble E-NTPDases (4, 41). Therefore, our findings may open up further research to assess the

potential diagnostic applications of purine-converting enzymes in clinical biochemistry.

ACKNOWLEDGMENTS

This work was supported by grants from the Finnish Academy and the Sigrid Juselius

Foundation.

REFERENCES

1. Kunapuli, S. P., Daniel, J. L. (1998) P2 receptor subtypes in the cardiovascular system.

Biochem. J. 336, 513�523

2. Robson, S., Sevigny, J., Imai, M., Guckelberger, O., Enjyoji, K. (2000)

Thromboregulatory potential of endothelial CD39/nucleoside triphosphate diphosphohydrolase:

modulation of purinergic signalling in platelets. Emerging Therapeutic Targets 4, 155�171

3. Di Virgilio, F., Chiozzi, P., Ferrari, D., Falzoni, S., Sanz, J. M., Morelli, A., Torboli, M.,

Bolognesi, G., Baricordi, O. R. (2001) Nucleotide receptors: an emerging family of regulatory

molecules in blood cells. Blood 97, 587�600

4. Marcus, A., Broekman, M., Drosopoulos, J., Pinsky, D., Islam, N., Gayle, R., III,

Maliszewski, C. (2001) Thromboregulation by endothelial cells: significance for occlusive

vascular diseases. Arterioscler. Thromb. Vasc. Biol. 21, 178�182

5. Ralevic, V., Burnstock, G. (1998) Receptors for purines and pyrimidines. Pharmacol.

Rev. 50, 413�492

6. Zimmermann, H. (1992) 5′-Nucleotidase: molecular structure and functional aspects.

Biochem. J. 285, 345�365

7. Meghji, P., Pearson, J. D., Slakey, L. L. (1995) Kinetics of extracellular ATP hydrolysis

by microvascular endothelial cells from rat heart. Biochem. J. 308, 725�731

8. Yegutkin, G. G., Henttinen, T., Jalkanen, S. (2001) Extracellular ATP formation on

vascular endothelial cells is mediated by ecto-nucleotide kinase activities via phosphotransfer

reactions. FASEB J. 15, 251�260

9. Zimmermann, H. (1999) Nucleotides and cd39: principal modulatory players in

hemostasis and thrombosis. Nat. Med. 5, 987�988

10. Zimmermann, H. (2000) Extracellular metabolism of ATP and other nucleotides. Naunyn

Schmiedebergs Arch. Pharmacol. 362, 299�309

11. Goding, J. W. (2000) Ecto-enzymes: physiology meets pathology. J. Leukoc. Biol. 67,

285�311

12. Cronstein, B. N. (1994) Adenosine, an endogenous anti-inflammatory agent. J. Appl.

Physiol. 76, 5�13

13. Spychala, J. (2000) Tumor-promoting functions of adenosine. Pharmacol. Ther. 87, 161�

173

14. Chan, E. S., Cronstein, B. N. (2002) Molecular action of methotrexate in inflammatory

diseases. Arthritis Res 4, 266�273

15. Griffith, D. A., Jarvis, S. M. (1996) Nucleoside and nucleobase transport systems of

mammalian cells. Biochim. Biophys. Acta 1286, 153�181

16. Franco, R., Valenzuela, A., Lluis, C., Blanco, J. (1998) Enzymatic and extraenzymatic

role of ecto-adenosine deaminase in lymphocytes. Immunol. Rev. 161, 27�42

17. Yegutkin, G. G., Henttinen, T., Samburski, S. S., Spychala, J., Jalkanen, S. (2002) The

evidence for two opposite, ATP-generating and ATP-consuming, extracellular pathways on

endothelial and lymphoid cells. Biochem. J. 367, 121�128

18. Lazarowski, E. R., Homolya, L., Boucher, R. C., Harden, T. K. (1997) Identification of

an ecto-nucleoside diphosphokinase and its contribution to interconversion of P2 receptor

agonists. J. Biol. Chem. 272, 20402�20407

19. Buxton, I. L., Kaiser, R. A., Oxhorn, B. C., Cheek, D. J. (2001) Evidence supporting the

Nucleotide Axis Hypothesis: ATP release and metabolism by coronary endothelium. Am. J.

Physiol. Heart Circ. Physiol. 281, H1657�H1666

20. Donaldson, S. H., Picher, M., Boucher, R. C. (2002) Secreted and cell-associated

adenylate kinase and nucleoside diphosphokinase contribute to extracellular nucleotide

metabolism on human airway surfaces. Am. J. Respir. Cell Mol. Biol. 26, 209�215

21. Rutsch, F., Vaingankar, S., Johnson, K., Goldfine, I., Maddux, B., Schauerte, P., Kalhoff,

H., Sano, K., Boisvert, W. A., Superti-Furga, A., et al. (2001) PC-1 nucleoside triphosphate

pyrophosphohydrolase deficiency in idiopathic infantile arterial calcification. Am. J. Pathol. 158,

543�554

22. Park, W., Masuda, I., Cardenal-Escarcena, A., Palmer, D. W., McCarty, D. J. (1996)

Generation of inorganic pyrophosphate from extracellular adenosine triphosphate by human

serum and plasma. J. Rheumatol. 23, 1233�1236

23. Cardenal, A., Masuda, I., Ono, W., Haas, A. L., Ryan, L. M., Trotter, D., McCarty, D. J.

(1998) Serum nucleotide pyrophosphohydrolase activity; elevated levels in osteoarthritis,

calcium pyrophosphate crystal deposition disease, scleroderma, and fibromyalgia. J. Rheumatol.

25, 2175�2180

24. Coade, S. B., Pearson, J. D. (1989) Metabolism of adenine nucleotides in human blood.

Circ. Res. 65, 531�537

25. Beukers, M. W., Pirovano, I. M., van Weert, A., Kerkhof, C. J., IJzerman, A.P., Soudijn,

W. (1993) Characterization of ecto-ATPase on human blood cells. A physiological role in

platelet aggregation? Biochem. Pharmacol. 46, 1959�1966

26. Ungerer, J. P., Oosthuizen, H. M., Bissbort, S. H., Vermaak, W. J. (1992) Serum

adenosine deaminase: isoenzymes and diagnostic application. Clin. Chem. 38, 1322�1326

27. Sunderman, F. W., Jr. (1990) The clinical biochemistry of 5′-nucleotidase. Ann. Clin.

Lab. Sci. 20, 123�139

28. Yegutkin, G. G., Burnstock, G. (1998) Steady-state binding of [3H]ATP to rat liver

plasma membranes and competition by various purinergic agonists and antagonists. Biochim.

Biophys. Acta 1373, 227�236

29. Yegutkin, G., Bodin, P., Burnstock, G. (2000) Effect of shear stress on the release of

soluble ecto-enzymes ATPase and 5′-nucleotidase along with endogenous ATP from vascular

endothelial cells. Br. J. Pharmacol. 129, 921�926

30. Yegutkin, G. G. (1997) Kinetic analysis of enzymatic hydrolysis of ATP in human and

rat blood serum. Biochemistry (Mosc) 62, 619�622

31. Simmonds, R. J., Harkness, R. A. (1981) High-performance liquid chromatographic

methods for base and nucleoside analysis in extracellular fluids and in cells. J. Chromatogr. 226,

369�381

32. Crack, B. E., Beukers, M. W., McKechnie, K. C., Ijzerman, A. P., Leff, P. (1994)

Pharmacological analysis of ecto-ATPase inhibition: evidence for combined enzyme inhibition

and receptor antagonism in P2X-purinoceptor ligands. Br. J. Pharmacol. 113, 1432�1438

33. Grobben, B., Claes, P., Roymans, D., Esmans, E. L., Van Onckelen, H., Slegers, H.

(2000) Ecto-nucleotide pyrophosphatase modulates the purinoceptor-mediated signal

transduction and is inhibited by purinoceptor antagonists. Br. J. Pharmacol. 130, 139�145

34. Yegutkin, G. G., Burnstock, G. (2000) Inhibitory effects of some purinergic agents on

ecto-ATPase activity and pattern of stepwise ATP hydrolysis in rat liver plasma membranes.

Biochim. Biophys. Acta 1466, 234�244

35. Terkeltaub, R. A. (2001) Inorganic pyrophosphate generation and disposition in

pathophysiology. Am. J. Physiol. Cell Physiol. 281, C1�C11

36. Cardenal, A., Masuda, I., Haas, A. L., Ono, W., McCarty, D. J. (1996) Identification of a

nucleotide pyrophosphohydrolase from articular tissues in human serum. Arthritis Rheum. 39,

252�256

37. Todorov, L. D., Mihaylova-Todorova, S., Westfall, T. D., Sneddon, P., Kennedy, C.,

Bjur, R. A., Westfall, D. P. (1997) Neuronal release of soluble nucleotidases and their role in

neurotransmitter inactivation. Nature 387, 76�79

38. Resta, R., Yamashita, Y., Thompson, L. F. (1998) Ecto-enzyme and signaling functions

of lymphocyte CD73. Immunol. Rev. 161, 95�109

39. Airas, L., Niemela, J., Salmi, M., Puurunen, T., Smith, D. J., Jalkanen, S. (1997)

Differential regulation and function of CD73, a glycosyl-phosphatidylinositol-linked 70-kD

adhesion molecule, on lymphocytes and endothelial cells. J. Cell Biol. 136, 421�431

40. Aldrich, M. B., Blackburn, M. R., Kellems, R. E. (2000) The importance of adenosine

deaminase for lymphocyte development and function. Biochem. Biophys. Res. Commun. 272,

311�315

41. Pinsky, D. J., Broekman, M. J., Peschon, J. J., Stocking, K. L., Fujita, T., Ramasamy, R.,

Connolly, E. S., Jr., Huang, J., Kiss, S., Zhang, Y., et al. (2002) Elucidation of the

thromboregulatory role of CD39/ectoapyrase in the ischemic brain. J. Clin. Invest. 109, 1031�

1040

42. Burnstock, G. (2002) Potential therapeutic targets in the rapidly expanding field of

purinergic signalling. Clin Med 2, 45�53

Received January 3, 2003; accepted March 31, 2003.

Fig. 1

Figure 1. Chromatographic analyses of the nucleotides and nucleosides. A, B) TLC images of the purine standards (4–6 nmol/spot) separated by using two different solvent mixtures (see Materials and Methods) and visualized under reflective UV light with an imaging analyzer FluorChem 8000 (Alpha Innotech, San Leandro, CA). C) HPLC chromatogram of nucleotide/nucleoside standards. The amount of each compound injected was between 0.2 and 0.3 nmol. Ado, adenosine; Hyp, hypoxanthine; Ino, inosine; Xanth, xanthine.

Fig. 2

Figure 2. Catabolism of adenine nucleotides and adenosine by human serum. A) [3H]ATP (20 µmol/l) was incubated with 30 µl serum in the starting volume of 120 µl RPMI-1640, and aliquots of the medium were collected and assayed by TLC for the relative amounts of [3H]ATP and its 3H metabolites. Serum was incubated for 60 min with 300 µmol/l [3H]ATP (B) as well as [3H]AMP and [3H]adenosine (C). Serum was also pretreated with suramin (sur; 200 µmol/l), EDTA (2.5 mmol/l), ortho-vanadate (100 µmol/l), ouabain (oub; 200 µmol/l), sodium azide (NaN3; 3 mmol/l), APCP (25 µmol/l), and EHNA (8 µmol/l) for 20 min before addition of 3H substrate. Ordinate shows the enzyme activities expressed as nmol 3H nucleotides/adenosine degraded by 1 ml serum per hour (mean ±SE; n=4–8). *P<0.05 as compared with the corresponding enzyme activities of nontreated serum.

Fig. 3

Figure 3. Nucleotide kinase activities in human serum. Serum was incubated for 60 min with 300 µmol/l [3H]AMP (A) or [3H]ADP (B) in the absence and presence of 1 mmol/l unlabeled NTP. In some experiments, serum was pretreated for 20 min with Ap5A (200 µmol/l) or EDTA (2.5 mmol/l) before addition of substrates. Conversion of 3H-labeled AMP/ADP into high-energy metabolites is expressed on the ordinate as nmol 3H nucleotides phosphorylated by 1 ml serum per hour (mean ±SE; n=3–4).

Fig. 4

Figure 4. Kinetic analysis of purine-converting enzymes in human serum. A) The activities of 5'-nucleotidase (5-NT) and adenosine deaminase (ADA) were measured by serum incubation for 60 min with the increasing concentrations of [3H]AMP and [3H]adenosine as respective substrates. Ordinate shows the rates of 3H substrate breakdown. B) Adenylate kinase (AdK) and NDP kinase (NDPK) activities were determined by serum incubation for 60 min with 1 mmol/l ATP and the increasing concentrations of [3H]AMP and [3H]ADP, respectively. Ordinate shows the rates of 3H nucleotides phosphorylation. C) The kinetic parameters Vmax and Km were calculated from the saturation curves and expressed as mean ±SE for two independent experiments.

Fig. 5

Figure 5. HPLC analysis of nucleotide metabolism in human serum. Serum was incubated for 30 min in the absence (A) and presence of ADP (B), GTP (C), and ADP plus GTP (D). Nucleotide concentrations were 50 µmol/l. Exogenous nucleotide substrates and their metabolites were separated by reverse-phase HPLC and are indicated in the chromatograms shown.

Fig. 6

Figure 6. Bioluminescent determination of ATP concentration. A) Serum was incubated with 4.5 µmol/l ADP in the absence and presence of 100 µmol/l NTP. Aliquots of the medium were periodically collected and assayed for ATP. The inset illustrates on a larger scale the time course of ADP conversion into ATP in the absence of NTP. B) Serum was also incubated with 4.5 µmol/l ATP and 100 µmol/l of certain NDP followed by measurement of the rate of ATP decay. The graphs show the mean data of three independent experiments performed with different serum samples.

Fig. 7

Figure 7. TLC analysis of the transfer of γ-phosphate from [γ-32P]ATP. Human serum (HS) was incubated for 30 min with 1 µmol/l ATP plus tracer [γ-32P]ATP in the presence of the indicated concentration of certain nucleotide as second substrate. Serum was also pretreated with EDTA (2 mmol/l) and Ap5A (30 µmol/l) for 10 min before addition of substrates. [γ-32P]ATP was also incubated with mouse serum (MS), rabbit serum (RS), and Jurkat T-cells (Jur). The arrows indicate the positions of inorganic phosphorus (Pi), pyrophosphate (PPi), and nucleotide standards. The first lane is a blank showing the radiochemical purity of [γ-32P]ATP after a 30-min incubation in the absence of serum.

Fig. 8

Figure 8. Purine metabolism in human serum. The potential exchange activities of circulating nucleotides in cell-free human serum as well as their conversion into adenosine (Ado) and inosine/hypoxanthine (Ino/Hyp) are shown. The following soluble enzymic activities have been identified: nucleotide pyrophosphatase (NPP; EC 3.6.1.9); 5'-nucleotidase (5-NT; EC 3.1.3.5); adenosine deaminase (ADA; EC 3.5.4.4); adenylate kinase (AdK; EC 2.7.4.3); and NDP kinase (NDPK; EC 2.7.4.6).