J Am Soc Nephrol 12: 1721–1728, 2001 High and Concentrations in Mononuclear Cells of Hemodialyzed Patients

JEROME SAMPOL,*† BERTRAND DUSSOL,†‡ EMMANUEL FENOUILLET,* CHRISTIAN CAPO,§ JEAN-LOUIS MEGE,§ GILLES HALIMI,* GUY BECHIS,* PHILIPPE BRUNET,†‡ HERVE ROCHAT,†‡ YVON BERLAND,†‡ and REGIS GUIEU*¶ *UMR CNRS 6560, Faculte´deMe´decine Secteur Nord; †Service de Ne´phrologie and ‡Centre d’Investigation Clinique, Hoˆpital Sainte Marguerite; §CNRS UPRESA 6020 Unite´des Rickettsies; and ¶Laboratoire de Biochimie, Hoˆpital de la Timone, Marseille, France.

Abstract. Infections are one of the most important complica- clear cells of patients than in healthy volunteers. During HD, tions of hemodialysis (HD). The high concentrations of aden- Ado concentration decreased by 34%, whereas concen- osine (Ado) and of its metabolites during HD may contribute to tration increased by 27%. Before HD, MCADA activity level the dialysis-induced immune deficiency through their known was 2.1-fold lower in patients than in control subjects. After ability to alter lymphocyte function. The influence of HD on HD, MCADA activity increased by nearly 50% but remained Ado metabolism was assessed in mononuclear cells through lower than in control subjects. Ado kinase activity level of the measurement of (1) the concentrations of in patients did not differ from that of control subjects and was mononuclear cells and (2) the activities of mononuclear cell unchanged by HD. The influence of Ado on in vitro mononu- Ado deaminase (MCADA) and Ado kinase, two enzymes clear cell proliferation and interferon-␥ production also was involved in Ado concentration regulation. Nine end-stage renal evaluated. Ado inhibited cell proliferation and interferon-␥ failure hemodialyzed patients (five men and four women; mean production in a dose-dependent manner, and these inhibitions age, 69 Ϯ 10 yr) and eight healthy volunteers (four men and were stronger for patients than for healthy volunteers. The high four women; mean age, 53 Ϯ 19 yr) were included in the study. concentrations of Ado and deoxyadenosine in mononuclear Before HD, Ado, deoxyadenosine, and inosine concentrations cells and the low MCADA activity level likely are involved in were respectively 2.9-, 2.5-, and 2.5-fold higher in mononu- the immune defect of patients who are undergoing HD.

Despite improvements in dialysis and medical therapies, infec- phosphate. The extracellular metabolism of Ado and d-Ado is tions account for a significant proportion (12 to 15%) of deaths mediated by two mechanisms. First, Ado and d-Ado are taken in the chronic dialysis population (1). The immune defect up quickly and efficiently by red blood cells, via an equilibra- observed in patients with end-stage renal disease could explain tive facilitated diffusion system (10) (Figure 1A). Nucleosides partly the frequency of infections. However, hemodialysis also are taken up efficiently by mononuclear cells via an (HD) actually worsens this immunodeficiency because of re- equilibrative facilitated diffusion system and less so by a current predialytic activation of mononuclear cells induced by sodium-dependent concentration system (11). Because of its blood contact with the dialysis membrane and dialysate, which rapid uptake, the plasma half- of Ado is very short (a few in turn results in a deactivation state (2). Many humoral and seconds). Second, Ado and d-Ado are deaminated rapidly into cellular factors likely are implicated in this immune defect (3). inosine and deoxyinosine, respectively, by adenosine deami- Among them, adenosine (Ado) and deoxyadenosine (d-Ado), nase (ADA; Figure 1A). both strong immunosuppressive agents (4–6), might be impli- ADA is found in large amounts particularly in mononuclear cated because plasma concentrations of Ado (7) and metabo- cells (mononuclear cell [MCADA]) lites (8) are high in patients who are undergoing HD. (12,13), where it plays a major role in Ado concentration Ado is released by endothelial cells and by several tissues, regulation in both extracellular (13) and intracellular spaces more particularly during ischemia (9). Intracellular Ado comes (14,15). It also is implicated in T-cell activation via noncova- from the hydrolysis of through a 5'nucleotidase. lent binding to the T-cell antigen CD26 (16). Intracellular Ado d-Ado is formed from the hydrolysis of deoxyadenosine mono- concentration also is regulated strongly by adenosine kinase (AKA) activity (17,18). Indeed, AKA phosphorylates Ado and nucleosides into nucleotides (19). Thus, low MCADA (15) or Received April 21, 2000. Accepted January 26, 2001. AKA activity level (20,21) results in an increased intra- and Correspondence to Dr. Regis Guieu, UMR CNRS 6560, Faculte´deMe´decine extracellular Ado concentration. A normal MCADA activity Secteur Nord, Bd P. Dramard 13015 Marseille, France. Phone: 33-4- 91698843; Fax: 33-4-91657595; E-mail: [email protected] level prevents adenosine accumulation and thus ensures normal 1046-6673/1208-1721 lymphocyte development and function (22). Inherited ADA Journal of the American Society of Nephrology deficiency, which induces high Ado and d-Ado concentrations Copyright © 2001 by the American Society of Nephrology in body fluids, causes severe combined immunodeficiency 1722 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1721–1728, 2001 syndrome (5). The immune system defect occurs because of 30 to 60 yr). Mean hemoglobin level was 15 Ϯ 2.5 g/100 ml; mean the accumulation of toxic metabolites, particularly d- hematocrit was 43 Ϯ 2%; lymphocyte count was 1700 Ϯ 200/␮l, and Ado and deoxyadenosine triphosphates (d-ATP), both of which CD3ϩ cell count was 1105 Ϯ 190. There was no significant differ- ϩ inhibit the reductase activity of T cells (23) ence between CD3 number in patients and in healthy volunteers Ͻ (Figure 1B). Furthermore, lymphocytes are particularly sensi- (ANOVA P 0.05). tive to Ado and d-Ado because of their ability to accumulate d-ATP (24). Reagents ␣ ␤ We showed previously that plasma Ado concentration is Adenosine (crystallized, 99% pure), dipyramidole, , methylene- adenosine-5'-diphosphate, ATP, 9-erythro (2-hydroxy-3-nonyl) ade- increased in patients who are undergoing HD (7). Taking into nine, 6-methylthiopurineriboside (6-MMPR), and dithiothreitol were account the facilitated diffusion system and the concentrative from SIGMA (Saint Quentin Fallavier, France). Deoxycoformycin sodium-dependent transport system of nucleosides across the (dcf) was from Lederle Laboratories (Paris, France). Heparin was cell membrane, we hypothesized that Ado and d-Ado concen- from Sanofi Winthrop (Gentilly, France). Na2-ethylenediaminetet- trations are high in mononuclear cells of patients who are raacetic acid was from SIGMA. Bovine serum albumin was from undergoing HD. Because ADA and AKA are thought to play a Johnson and Johnson Clinical Chemistry (Rochester, NY). Con- major role in the regulation of intracellular concen- canavalin A (Con A) was from SIGMA. The reversed phase chroma- tration, we evaluated their activities in mononuclear cells. tography column (Merck LIChrospher C18, and RP8 250 ϫ 4 mm) Finally, we also evaluated the influence of Ado on the T-cell and other reagents were from Merck (Darmstadt, Germany). proliferation and interferon-␥ (IFN-␥) production in patients who are undergoing HD. We chose IFN-␥ rather than interleu- Blood Samples for Ado, d-Ado, and Inosine Assays kin-2 (IL-2) because IFN-␥ is a good marker of human periph- Sample collection has been described elsewhere (28,29). Briefly, eral blood lymphocytic activity (25) and because IL-2 has been blood collected from the brachial vein (8 ml/sample) was drawn into studied carefully during HD in humans (26,27). vacuum tubes containing a stopping solution (0.2 mM dipyridamole, 4.2 mM Na2-ethylenediaminetetraacetic acid, 5 mM 9-erythro (2- hydroxy-3-nonyl) , 79 mM ␣,␤ methylene-adenosine-5'- Materials and Methods ␮ Patients diphosphate, 1 IU/ml heparin sulfate, 200 g/ml dcf, and 0.9% NaCl), which prevents degradation and uptake of Ado. The samples were For patients and control subjects, the ingestion of coffee or tea was transferred to special Vacutainer tubes (Ficoll-based CPT system; suspended 72 h before samples were taken. Patients who had been Becton Dickinson, Le Pont de Claix, France) and centrifuged at treated with papaverine, dipyridamole, or immunosuppressive agents 500 ϫ g for 30 min. Then 1 ml of interphase cells was pipetted off and during the preceding 6 mo were excluded from the study. All partic- washed three times with 3 ml of the stopping solution before assess- ipants gave their consent according to the Helsinki convention. Nine ment of cell viability by trypan blue dye test exclusion. Granulocyte adults (five men and four women) on maintenance HD, who were contamination and mononuclear cell number were measured (Coulter randomly chosen among 128 patients with end-stage renal failure in Beckman, Fullerton, CA). These techniques resulted in cell prepara- the same center, constituted the group of patients who were undergo- tions that were Ͼ98% viable and that contained Ͻ3% of granulocytes. Ϯ Ϯ ing HD. Their mean age SD was 69 11 yr (range, 49 to 85 yr), Lymphocyte proportion was always more than 95%. Cells were re- Ϯ and their mean duration of maintenance HD was 47 34 mo (range, suspended in stopping solution (6.5 Ϯ 1.5 ϫ 106 cell/ml) and frozen 17 to 132 mo). Causes of end-stage renal failure were diabetic (Ϫ80°C). Samples were submitted to four freeze-thaw cycles (Ϫ80°C, ϭ ϭ nephropathy (n 2), chronic glomerulonephritis (n 1), interstitial ϩ37°C) and centrifuged (2500 ϫ g for 10 min) to obtain clear ϭ ϭ nephritis (n 1), polycystic kidney disease (n 1), nephroangio- supernatant cell extracts. Supernatants were deproteinized by addition ϭ ϭ sclerosis (n 3), and cholesterol embolism (n 1). No patient had of 100 ␮l of perchloric acid (6N) and then centrifuged (1500 ϫ g for anti–human immunodeficiency virus or hepatitis B surface antigen or 10 min). Samples were lyophilized before being chromatographed. hepatitis C virus antibodies. The patients underwent HD three times Ϯ per week for 5 1 h with cellulose diacetate low-flux membranes Sample Preparation for Determining MCADA and (A15, A18; Althin, Miami, FL). The hemodialyzers were never re- used. HD vascular access was either native radial arteriovenous fistula AKA Activities (n ϭ 4) or graft polytetrafluoroethylene fistula (n ϭ 5). Endotoxin- We used the procedure previously described (12) with some mod- free, nonpyrogenic ultrapure bicarbonate dialysate was used. Dialysis ifications. Briefly, samples (8 ml) of whole blood were collected from water was tested every week with Limulus Amoebocyte Lysate assay. each patient and healthy volunteers in special Vacutainer tubes (see ϫ Endotoxin counts were always Ͻ0.01 endotoxin unit/ml. The blood above) and then centrifuged at 500 g for 30 min. A 1-ml sample of Ϯ 6 and dialysate flows were 250 and 500 ml/min, respectively. The mean interphase cells (7.6 1.10 ) was pipetted off and washed four times hemoglobin level was 11 Ϯ 1 g/100 ml, and the mean hematocrit was with 3 ml NaCl 0.9%, to eliminate plasmatic ADA. Aliquots of 1 ml 33 Ϯ 4%. Six patients were receiving recombinant human erythro- were submitted to four freeze-thaw cycles before centrifugation (2500 ϫ poietin (3000 Ϯ 1500 IU/wk). Before HD, mean serum creatinine g for 10 min) to obtain clear supernatant cell extracts. The cell level was 420 Ϯ 80 ␮M, and BUN level was 31 Ϯ 15 mM. After HD, extracts were assayed for ADA and AKA activity levels. mean serum creatinine level was 205 Ϯ 100 ␮M, and BUN level was 10 Ϯ 6 mM. The mean Kt/V was 1.53 Ϯ 0.17 (range, 1.3 to 1.77). Ado, d-Ado, and Inosine Assays Lymphocyte count was 1370 Ϯ 230/␮l. CD3ϩ cell count was 959 Ϯ The technique has been described elsewhere (28,29). Briefly, a 120. Arterial oxygen tension assessed by analyzing blood drawn from Hewlett Packard HP 1100 modular system (Lesullis, France) was the arterial line of the fistula before HD was 95 Ϯ 5 mmHg. used, with a diode array detector. Lyophilized samples (500 ␮l) were ϭ Control subjects were healthy volunteers (n 8; 4 women and 4 mixed with 1 ml of phosphate buffer (NaH2PO4/Na2HPO4 [pH 4]), men) without any medication. Mean age Ϯ SD was 53 Ϯ 19 yr (range, injected in a 1-ml loop, and then eluted with a methanol gradient on J Am Soc Nephrol 12: 1721–1728, 2001 Nucleosides, Mononuclear Cells, and Hemodialysis 1723

a Merck LIChrospher C18 column (0% for 3 min, then 10 to 25% methanol for 15 min). The intra- and interassay coefficients of vari- ations for nucleosides ranged between 1 and 3%. The limit of detec- tion at 254 nm was 1 pmol in 1 ml of plasma matrix injected.

Identification and Quantification Retention times and spectra were compared with those of exoge- nous adenosine and metabolites. Quantifications were made by com- paring areas obtained for samples with areas of known quantities of nucleosides.

MCADA Activity We used the technique previously described (30), with some mod- ifications. Briefly, 750 ␮l of 28 mM Ado was mixed with 125 ␮lof cell extracts and with 125 ␮l of bovine serum albumin 7% in NaCl 0.9%. Aliquots then were incubated for1hat37°C. The reaction was started by adding the substrate and was stopped by cold immersion. The Johnson and Johnson Clinical Chemistry colorimeter test was used to quantify the ammonia concentrations. The intra- and interas- say coefficients of variation ranged between 3 and 5%.

AKA Activity We used the procedure previously described (31), with some mod- ifications. Briefly, 125 ␮l of cell extracts was incubated (37°C for 20 ␮ min) with 875 l of a solution (NaH2PO4/Na2HPO4 [pH 5.5]), sup- plemented with 1 mM 6-MMPR (substrate), 1 mM ATP, 1 mM

MgCl2, and 1 mM dithiothreitol. The reaction was started by adding the substrate and stopped by adding 100 ␮l of perchloric acid (6N). Samples were deproteinized by centrifugation (1500 ϫ g for 10 min), Figure 1. (A) Metabolism of adenosine (Ado), deoxyadenosine (d- and the supernatant was lyophilized before chromatography analysis. Ado), and metabolites. Ado is formed in endothelial cells from the Lyophilized samples were dissolved in 1 ml of phosphate buffer and hydrolysis of nucleotides via a 5' nucleotidase. d-Ado comes from then chromatographed (Merck RP8 column) at 300 nm. The samples DNA breakdown. A part of the Ado and d-Ado formed are rephos- were eluted in a 20 to 65% methanol gradient for 30 min. 6-MMPR- phorylated into nucleotides via an adenosine kinase. Adenosine 5-phosphate was identified by elution time and spectra comparison diphosphate (ADP) and, to a lesser degree, and quantified, as were the nucleosides. The intra- and interassay (ATP) are converted into their deoxy derivatives by a ribonucleotide coefficients of variation for 6-MMPR-5-phosphate ranged between 1 reductase. Most Ado is released in the extracellular spaces where it and 4%. binds to purinergic receptors of target cells, causing smooth muscle relaxation and vasodilation (9). However, Ado half-life in extracellu- Mononuclear Cell Proliferation lar spaces is short because it is taken up quickly by blood cells via an Mononuclear cells of patients or healthy volunteers were obtained equilibrative facilitated diffusion system (10). The part of nucleosides as described above. Cells (4.5 Ϯ 1 ϫ 106/0.5 ml) were cultured in a that have not been taken up is deaminated quickly into inosine or CO incubator at 37°C, for 48 h in 1.5 ml of RPMI 16/40, with 4% of d-inosine via extracellular adenosine deaminase (ADA). The final 2 fetal bovine serum. Cells were preactivated with Con A (1 ng/ml). product of Ado metabolism is uric acid. ADA also is found in large Aliquots obtained from patients or healthy volunteers were separated amount in red blood cells and in mononuclear cells (MCADA), where into four groups: one with dcf (an adenosine deaminase inhibitor), 1 it regulates both intra- and extracellular adenosine concentration. nM/ml; one with dcf ϩ Ado (1 nM/ml); one with dcf (1 nM/ml) ϩ White arrows denote reaction catalyzed by the following: 1, 5'nucle- Ado (3 nM/ml); and one control. Aliquots (50 ␮l) were pipetted off otidase; 2, adenosine kinase (AKA); 3, ; 4, and examined for cell viability (trypan blue dye exclusion test and intracellular ADA (including MCADA); 5, extracellular ADA. (B) count per milliliter with Coulter Epics XL) at time 0, then after 24 and Toxicity of nucleosides and derivatives on lymphocytes. The low 48 h of incubation. ADA level prevents the degradation of Ado and particularly of d-Ado. The accumulation of d-Ado increases the amount of nucleoside phos- phorylated by adenosine kinase, producing high concentrations of IFN-␥ Assay deoxyadenosine triphosphates (d-ATP) in both B and T lymphocytes. IFN-␥ was assayed on cell supernatant after 24 and 48 h of The accumulation of nucleosides and derivatives results in the inhi- activation and culture (concanavalin 1 ng/ml). The IFN-␥ concentra- bition of ribonucleotide reductase and consequently of DNA synthesis tions were measured by use of the quantitative sandwich enzyme (47). Finally, high concentrations of adenosine or analogs induce immunoassay, as recommended by the manufacturer (Immunotech, lymphocyte apoptosis (42,44). d-ATP induces apoptosis by a pathway Marseille, France). This immunoassay uses an immobilized monoclo- involving p53 protein. Indeed, p53 regulates the mRNA of inosine nal antibody specific for human IFN-␥ and a second monoclonal 5'monophosphate dehydrogenase, an enzyme essential for the main- anti–IFN-␥ antibody that is biotinylated. The binding of streptavidin- tenance of GTP/ATP ratio (48). AMP, . peroxidase conjugate to the human complex is followed by the addi- 1724 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1721–1728, 2001 tion of a chromogenic substrate of the peroxidase. The sensitivity of Correlations between Nucleoside Concentrations and the assay was 0.08 IU/ml. The intra- and interassay coefficients of MCADA Activity Levels variation ranged between 5 and 10%. There was an inverse correlation between intramononuclear cell nucleoside concentration and MCADA activity level in Statistical Analyses healthy volunteers (Spearman’s r ϭϪ0.8; P Ͻ 0.05 for both ANOVA one-way analysis was used to compare nucleoside con- Ado and d-Ado). Before HD, there was an inverse correlation centrations and enzyme activity levels between patients and control between Ado or d-Ado concentration and MCADA activity subjects. The Wilcoxon test was used for intragroup comparisons (cell level (r ϭϪ0.73 and r ϭϪ0.55, respectively; P Ͻ 0.05). number and IFN-␥ production as a function of time in the same After HD, the correlations persisted (r ϭϪ0.8 and r ϭϪ0.67; group). The Spearman coefficient of correlation was used for corre- Ͻ lation studies. P Ͼ 0.05 was considered significant. P 0.05). Finally, there was an inverse correlation between the MCADA activity increase during HD and the duration of Results the dialysis treatment (r ϭϪ0.85; P Ͻ 0.005; Figure 3). Intra–Mononuclear Cell Nucleoside Concentrations Before HD, Ado, d-Ado, and inosine concentrations (in Effects of Ado on Mononuclear Cell Number and pmol for 107 cells) were, respectively, 2.9-, 2.5-, and 2.5-fold IFN-␥ Production higher in patients than in control subjects (32.6 Ϯ 10 pmol In control conditions (Figure 4A), cell proliferation at 48 h versus 11 Ϯ 2 pmol; 30 Ϯ 17 pmol versus 12 Ϯ 2 pmol; 48 Ϯ was greater in healthy volunteers (ϩ17% compared with 24 h) 19 pmol versus 19 Ϯ 4.7 pmol; Figure 2A). After HD, Ado than in patients (ϩ12%; P Ͻ 0.05). concentration decreased (34% on average, 32.6 Ϯ 10 versus With dcf alone (Figure 4B), cell proliferation was slightly 21.6 Ϯ 7.9; P Ͻ 0.03). d-Ado concentration did not decrease inhibited after 48 h of incubation but less so in healthy volun- significantly (30 Ϯ 17 versus 26 Ϯ 13), but inosine concen- teers than in patients (P Ͻ 0.05). With dcf and Ado 1 ␮M tration increased significantly by 27% (48 Ϯ 19 versus 61 Ϯ (Figure 4C), cell number at 48 h was lower than that with dcf 12.7; P Ͻ 0.02). alone (Figure 4B). With dcf and Ado (3 ␮M; Figure 4D), cell number decreased significantly at 48 h (P Ͻ 0.05 compared MCADA and AKA Activities with 24 h); however, this decrease was larger in patients Before HD, MCADA activity level was 2.07-fold lower in (Ϫ22%) than in healthy volunteers (Ϫ16%; P Ͻ 0.05). patients than in control subjects (80 Ϯ 18.6 versus 166 Ϯ 33 IFN-␥ concentration in the supernatant of Con A–treated IU; ANOVA P Ͻ 0.001; Figure 2B). After HD, MCADA cells was lower in patients than in healthy volunteers (P Ͻ 0.03 activity level increased significantly (80 Ϯ 18.6 versus 118 Ϯ at 48 h; Figure 5A). With dcf alone (Figure 5A), IFN-␥ 24 IU; P Ͻ 0.01) but remained lower than that in control concentration was slightly inhibited as early as 24 h only in subjects (118 Ϯ 24 versus 166 Ϯ 33 IU; ANOVA P Ͻ 0.005). patients (P Ͻ 0.02 compared with without dcf) and in both Before and after HD, AKA activity level was not significantly patients and healthy volunteers at 48 h (healthy volunteers with different in patients and in healthy volunteers (49 Ϯ 9 versus dcf versus healthy volunteers without dcf, P Ͻ 0.02). With 51 Ϯ 17 and 49 Ϯ 9 versus 47 Ϯ 16 IU, respectively; P Ͼ Ado (1 ␮M) and dcf (Figure 5B), IFN-␥ concentration de- 0.05). creased both in healthy volunteers (mean, Ϫ85% at 24 h and

Figure 2. (A) Ado, d-Ado, and inosine concentrations in mononuclear cells of nine end-stage renal failure patients before (■) and after (1) hemodialysis (HD) and in eight healthy volunteers (controls; ᮀ). Nucleoside concentrations were expressed in pmol for 107 cells. a, ANOVA P Ͻ 0.05 compared with controls; b, Wilcoxon test P Ͻ 0.05 comparing patients before and after HD. (B) MCADA and AKA activities of mononuclear cells from nine end-stage renal failure hemodialyzed patients and eight healthy volunteer. Enzymatic activities are expressed in international units (IU). One unit corresponds to 1 ␮mol/min of substrate transformed. a, ANOVA P Ͻ 0.05 compared with controls; b, Wilcoxon test, P Ͻ 0.05 compared with after HD. J Am Soc Nephrol 12: 1721–1728, 2001 Nucleosides, Mononuclear Cells, and Hemodialysis 1725

Figure 3. Correlation curve between the increase in MCADA activity during the HD session and the duration of the HD expressed in Figure 4. Effects of adenosine and deoxycoformycin (dcf) on mononu- months. Spearman’ s r ϭϪ0.85; P Ͻ 0.005. clear cell proliferation of nine hemodialyzed patients ( ) and eight healthy volunteers (□). dcf was added to prevent quick deamination of Ado into inosine. Cells were activated with concanavalin A (1 ng/ml) and Ϫ95% at 48 h) and in patients (mean, Ϫ77% at 24 h and Ϫ96% were examined for cell viability (trypan blue dye test and count, Coulter 6 at 48 h), compared with dcf alone. With Ado (3 ␮M) and dcf, Epics XL) 24 and 48 h after incubation. Results are expressed for 10 cells at the beginning of the incubation. (A) Spontaneous proliferation; IFN-␥ concentration decreased by 92% at 24 h and by 98% at (B) cells with dcf (1 ␮M); (C) cells with dcf (1 ␮M), and Ado 1 ␮M; and 48 h in patients and by 92% at 24 h and 98% at 48 h in healthy (D) cells with dcf (1 ␮M) and Ado 3 ␮M. a, ANOVA P Ͻ 0.05 volunteers. compared with patients; b, Wilcoxon test P Ͻ 0.05 compared with 24 h; c, ANOVA P Ͻ 0.05 compared with dcf alone at the same time (in Discussion reference to B); d, ANOVA P Ͻ 0.05 compared with spontaneous Abnormal Levels of Ado an Its Metabolites and proliferation (in reference to A). Statistical analysis was performed only Enzymes in Patients Who Were Undergoing HD in the case of an overlap between SD. We found high concentrations of Ado and d-Ado in mono- nuclear cells before HD. These high intracytoplasmic concen- trations are due partly to the low MCADA activity, because sessions (33). This also is suggested by our results that showed AKA activity was normal. Indeed, Ado and d-Ado concentra- increased MCADA activity during HD. However, we did not tions were highly correlated with MCADA activity levels. retain this hypothesis because in previous studies, we demon- Alternatively, the high intracellular nucleoside concentrations strated that undialyzed patients with chronic renal failure had can be explained by the high plasmatic Ado and metabolite normal MCADA activity levels (7) and normal plasmatic Ado concentrations in patients who were undergoing HD (7,8). This concentrations (34). Another hypothesis is that the decrease in hypothesis is supported by the equilibrative facilitated diffu- MCADA activity results from the deactivation of lymphocytes. sion system and the concentrative sodium-dependent nucleo- During HD sessions, lymphocytes are activated by contact with side transports in mononuclear cells. During HD, Ado and the extracorporeal circuit (26,27,35). This activation also is d-Ado concentrations decreased, whereas MCADA activity shown by the predialytic increase in MCADA activity that we levels increased by nearly 50%, but the latter remained lower observed, yet MCADA expression markedly increases when T than in control subjects. The decrease in Ado and d-Ado may cells are activated (36,37). The repeated activation of lympho- be due to the rise in MCADA level, because inosine, the cytes during HD sessions may in turn lead to cell deactivation. product of Ado deamination, increased concomitantly. This Such a phenomenon has been reported for IL-2 receptors, increase in inosine level may participate in the activation of whose number decreases with time during long-term HD treat- lymphocytes that is induced by HD. Indeed, inosine contained ment (27). in dialyzable leukocyte extracts activates lymphocytes (32). Why is MCADA activity low at the basal state in patients Mononuclear Cell Abnormalities who are undergoing HD? One hypothesis is that uremic toxins Many studies have demonstrated the negative impact of Ado depress MCADA activity. This has been suggested by studies and d-Ado on lymphocyte functions. Indeed, T cells exposed to that showed improved erythrocyte ADA activity during HD nucleosides, in a medium with low ADA activity, accumulate 1726 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1721–1728, 2001

Figure 5. IFN-␥ concentration (in IU for 106 cells) in the supernatant of activated lymphocytes (concanavalin A 1 ng/ml) of nine patients (■) and eight healthy volunteers (□). IFN-␥ was assayed for patients and healthy volunteers 24 and 48 h after activation with dcf alone (1 ␮M) or in association with Ado (1 or 3 ␮M) and finally without drug, except concanavalin A. dcf was added to prevent the quick deamination of Ado into inosine. (A) Effects of dcf in comparison with spontaneous proliferation. (B) Effects of Ado (1 or 3 ␮M) in comparison with dcf. A variance analysis (ANOVA, one-way analysis) was used for intergroup comparison. Wilcoxon test was used to compare IFN-␥ production as a function of time. Statistical analysis was performed only in the case of an overlap between the values. A, ANOVA P Ͻ 0.05 compared with healthy volunteers; b, Wilcoxon test P Ͻ 0.05 compared with 24 h.

d-Ado and d-ATP (4,24), which inhibit both ribonucleotide A made IFN-␥ production lower in patients than in healthy reductase and, hence, DNA synthesis (4) (Figure 1B). d-Ado volunteers. Moreover, IFN-␥ production decreased in the also is toxic for ADA-inhibited human peripheral blood lym- presence of Ado but more so for patients than for healthy phocytes (38). d-Ado also has been reported to block RNA volunteers. synthesis (39) and to foster an accumulation of strand breaks in In conclusion, we found high Ado and d-Ado concentra- DNA (40). Moreover, in mixed lymphocyte culture, d-Ado tions and low MCADA activity levels in the mononuclear decreases IL-2 production and inhibits IL-2 receptor expres- cells of patients who were undergoing HD. These may sion (41). d-Ado at 1 to 3 ␮M blocks the transition of the participate in the immune deficiency of these patients. Fur- stimulated lymphocytes from G0 to G1 via the inhibition of ther investigations are needed to determine how HD induced protein phosphorylation (6). Finally, an excess of nucleosides MCADA deficiency. induces apoptosis in human peripheral blood mononuclear cells (42), and adenosine analogs induce apoptosis in normal and neoplastic lymphocytes (43,44). The sensitivity of T cells References 1. Rinehart A, Collins AJ, Keane KJ: Host defenses and infectious to Ado was attributed to their high nucleoside kinase activity complications in maintenance hemodialysis patients. In: Re- (45). placement of Renal Function by Dialysis, 4th Ed., edited by The impact of Ado and d-Ado has never been studied in Jacobs C, Kjellstrand CM, Koch KM, JF Winchester, Dordrecht, lymphocytes from patients who are undergoing HD. We found Kluwer Academic Publishers, 1996, pp 1103–1106 that Ado induced a dose-dependent decrease in cell number 2. Descamps-Latscha B, Herbelin A: Long term dialysis and cellu- and IFN-␥ production. Furthermore, activated lymphocytes lar immunity: A critical survey. Kidney Int 41[Suppl]: S135– from patients, in the absence of any drug except Con A, S142, 1993 proliferated less than those from healthy volunteers and were 3. Descamps-Latscha B: The immune system in end stage renal more sensitive to Ado and/or dcf. Because the intracellular disease. Curr Opin Nephrol Hypertens 2: 883–891, 1993 ADA level is lower in patients, we hypothesized that mono- 4. Ullman B, Gudas LJ, Cohen A, Martin DW Jr: Deoxyadenosine nuclear cells of patients are more sensitive to dcf-induced ADA metabolism and cytotoxicity in cultured mouse T lymphoma cells: A model for immunodeficiency disease. Cell 14: 365–375, decrease and then Ado increase. This hypothesis is supported 1978 by the results of Dong et al. (46) showing that CD26-trans- 5. Hersfield MS, Mitchell BS: Immunodeficiency diseases caused fected Jurkat cells, which express high quantities of MCADA, by adenosine deaminase deficiency and purine nucleosides phos- are more resistant to the inhibitory action of Ado on cell phorylase deficiency. In: The Metabolic and Molecular Bases of proliferation and IL-2 production. Inherited Diseases, 3rd Ed., edited by Scriver CR, Beaudet AL, We also found that the activation of lymphocytes by Con Sly WS, Valle D, New York, McGraw Hill, 1994, pp 1725–1768 J Am Soc Nephrol 12: 1721–1728, 2001 Nucleosides, Mononuclear Cells, and Hemodialysis 1727

6. Sato T, Chan TS : Deoxyadenosine blockade of G0 to G1 nucleosides for T cells. Proc Natl Acad Sci USA 75: 5011–5014, transition in lymphocytes: Possible involvement of protein ki- 1978 nases. J Cell Physiol 166: 288–295, 1996 25. Catalona WJ, Ratliff TL, McCool RE: Gamma-interferon in- 7. Guieu R, Brunet P, Sampol J, Bechis G, Fenouillet E, Mege JL, duced by S. aureus protein A augments natural killing and Capo C, Vitte J, Ibrahim Z, Carrega L, Lerda D, Rochat H, ADCC. Nature 291: 77–79, 1981 Berland Y, Dussol B: Adenosine and hemodialysis in humans. 26. Chatenoud L, Dugas B, Beaurin G, Touam M, Drueke T, J Investig Med 49: 56–67, 2001 Vasquez A, Galanaud P, Bach JF, Delfraissy JF: Presence of 8. Shinzato T, Miwa M, Nakai S, Morita H, Odani H, Inoue I, preactivated T cells in hemodialyzed patients: Their possible role Maeda K: Role of adenosine in dialysis-induced hypotension. in altered immunity. Proc Natl Acad Sci USA 83: 7457–7461, J Am Soc Nephrol 4: 1987–1994, 1994 1986 9. Sollevi A: Cardiovascular effects of adenosine in man: Possible 27. Zaoui P, Green W, Hakim RM: Hemodialysis with cuprophane clinical implications. Prog Neurobiol 27: 319–349, 1986 membrane modulates interleukin 2 receptor expression. Kidney 10. Plagemann PG: Transport and metabolism of adenosine in hu- Int 39: 1020–1026, 1991 man erythrocytes: Effects of transport inhibitors and regulation 28. Guieu R, Sampieri F, Bechis G, Rochat H: Use of HPLC to by phosphate. J Cell Physiol 128: 491–500, 1986 measure circulating adenosine levels in migrainous patients. Clin 11. Roovers KI, Meckling-Gill KA: Characterization of equilibrative Chim Acta 227: 185–194, 1994 and concentrative Naϩ dependent (cif) nucleoside transport in 29. Guieu R, Sampieri F, Bechis G, Halimi G, Dussol B, Berland acute promyelocytic leukemia NB4 cells. J Cell Physiol 166: Y, Sampol J, Rochat H: Development of an HPLC/diode array 593–600, 1996 detector method for the determination of human plasma 12. Lum CT, Sutherland DE, Yasmineh WG, Najarian JS: Peripheral adenosine concentrations. J Liq Chromatogr 22: 1829–1841, blood mononuclear cell adenosine deaminase activity in renal 1999 allograft recipients. J Surg Res 24: 388–395, 1978 30. Giusti G, Galanti B: Colorimetric method. In: Methods of Enzy- 13. Kameoka J, Tanaka T, Nojima Y, Schlossman SF, Morimoto C: matic Analysis, 5th Ed., edited by Bergmeyer HU, Weinheim, Direct association of adenosine deaminase with a T cell activa- Verlag Chemie, 1984, pp 315–323 tion antigen CD26. Science 261: 466–469, 1993 31. Ronca-Testoni S, Lucacchini A: Determination of adenosine kinase activity by high-pressure liquid chromatography. Ital 14. Apasov SG, Sitkovsky MV: The extracellular versus intracellular J Biochem 30: 190–200, 1981 mechanisms of inhibition of TCR-triggered activation in thymo- 32. Komatsu F: Effects of dialyzable leucocyte extracts (DLE) and cytes by adenosine under conditions of inhibited adenosine inosine on stimulated lymphocytes. Bull Tokyo Med Dent Univ deaminase. Int Immunol 11: 179–189, 1999 36: 35–40, 1989 15. Cheng B, Essackjee HC, Ballard HJ: Evidence for control of 33. Severini G: Uremic toxins and adenosine deaminase activity. adenosine metabolism in rat oxidative skeletal muscle by Clin Biochem 27: 273–276, 1994 changes in pH. J Physiol 522: 467–477, 2000 34. Guieu R, Dussol B, Devaux C, Sampol J, Brunet P, Rochat H, 16. Dong RP, Tachibana K, Hegen M, Munakata Y, Cho D, Schloss- Berland YF: Interactions between cyclosporine A and adeno- man SF, Morimoto C: Determination of adenosine deaminase sine in kidney transplant recipients. Kidney Int 53: 200–204, binding domain on CD26 and its immunoregulatory effect on T 1998 cell activation. J Immunol 159: 6070–6076, 1997 35. Donati D, Degiannis D, Combates N, Raskova J, Raska K Jr: 17. Yamada Y, Goto H, Ogasawara N: Purine nucleoside kinases in Effects of hemodialysis on activation of lymphocytes: Analysis human T and B lymphoblasts. Biochim Biophys Acta 761: 34– by an in vitro dialysis model. J Am Soc Nephrol 2: 1490–1497, 40, 1983 1992 18. Spychala J, Mitchell BS, Barankiewicz J: Adenosine metabolism 36. Martin M, Centelles JJ, Huguet J, Echevarne F, Colomer D, during phorbol myristate acetate-mediated induction of HL-60 Vives-Corrons JL, Franco R: Surface expression of adenosine cell differentiation: Changes in expression pattern of adenosine deaminase in mitogen-stimulated lymphocytes. Clin Exp Immu- kinase, adenosine deaminase, and 5' nucleotidase. J Immunol nol 93: 286–291, 1993 158: 4947–4952, 1997 37. Martin M, Huguet J, Centelles JJ, Franco R: Expression of 19. Hurley MC, Lin B, Fox IH: Regulation of deoxyadenosine and ecto-adenosine deaminase and CD26 in human T cells trig- nucleoside analog phosphorylation by human placental adeno- gered by the TCR-CD3 complex. Possible role of adenosine sine kinase. J Biol Chem 260: 1575–1581, 1985 deaminase as costimulatory molecule. J Immunol 155: 4630– 20. Lynch JJ III, Alexander KM, Jarvis MF, Kowaluk EA: Inhibition 4643, 1995 of adenosine kinase during oxygen-glucose deprivation in rat 38. Carson DA, Wasson DB, Lakow E, Kamatani N: Possible met- cortical neuronal cultures. Neurosci Lett 252: 207–210, 1998 abolic basis for the different immunodeficient states associated 21. Golembiowska K, White TD, Sawynok J: Adenosine kinase with genetic deficiencies of adenosine deaminase and purine inhibitors augment release of adenosine from spinal cord slices. nucleoside phosphorylase. Proc Natl Acad Sci USA 79: 3848– Eur J Pharmacol 307: 157–162, 1996 3852, 1982 22. Aldrich MB, Blackburn MR, Kellems RE: The importance of 39. Matsumoto SS, Yu J, Yu AL: The effect of deoxyadenosine plus adenosine deaminase for lymphocyte development and function. deoxycoformycin on replicative and repair synthesis of DNA in Biochem Biophys Res Commun 272: 311–315, 2000 human lymphoblasts and isolated nuclei. J Biol Chem 263: 23. Hirschhorm R: Adenosine deaminase deficiency. In: Immunode- 7153–7158, 1988 ficiencies, 3rd Ed., edited by Rosen FS, Seligmann M, London, 40. Seto S, Carrera CJ, Kubota M, Wasson DB, Carson DA: Mech- Harwood Academic Publishers, 1993, pp 117–131 anism of deoxyadenosine and 2-chlorodeoxyadenosine toxicity 24. Mitchell BS, Mejias E, Daddona PE, Kelley WN: Purinogenic to nondividing human lymphocytes. J Clin Invest 75: 377–383, immunodeficiency diseases: Selective toxicity of deoxyribo- 1985 1728 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1721–1728, 2001

41. Ruers TJ, Buurman WA, van der Linden CJ: 2'Deoxycofor- phocytes by addition of doxorubicin and mitoxantrone. Leuk mycin and deoxyadenosine affect IL2 production and IL2 Lymphoma 36: 559–567, 2000 receptor expression of human T cell. J Immunol 138: 116– 45. Carson DA, Kaye J, Matsumoto S, Seegmiller JE, Thompson L: 122, 1987 Biochemical basis for the enhanced toxicity of deoxyribonucleo- 42. Barbieri D, Abbracchio MP, Salvioli S, Monti D, Cossarizza A, sides toward malignant human T cell lines. Proc Natl Acad Sci Ceruti S, Brambilla R, Cattabeni F, Jacobson KA, Franceschi C: USA 5: 2430–2433, 1979 Apoptosis by 2-chloro-2'-deoxy-adenosine and 2-chloro-adeno- 46. Dong RP, Kameoka J, Hegen M, Tanaka T, Xu Y, Schlossman sine in human peripheral blood mononuclear cells. Neurochem SF, Morimoto C: Characterization of adenosine binding to hu- Int 32: 493–504, 1998 man CD26 on T cells and its biological role in immune response. 43. Meisel H, Gunther S, Martin D, Schlimme E: Apoptosis induced J Immunol 156: 1349–1355, 1996 by modified in human cell culture systems. 47. Newsholmes EA, Leech AR: Pathways for the degradation of FEBS Lett 433: 265–268, 1998 and nucleotides. In: Biochemistry for the 44. Chow KU, Rummel MJ, Weidmann E, Ries J, Jantschke P, Medical Sciences, 5th Ed., edited by Chichester, John Wiley & Boehrer S, Pourebrahim F, Napieralski S, Stein J, Martin H, Sons, 1992, pp 507–508 Hoelzer D, Mitrou PS: Induction of apoptosis by 2-chloro-2'de- 48. Liu Y, Bohn SA, Sherley JL: Inosine 5' monophosphate dehy- oxyadenosine (2-CdA) alone and in combination with othercy- drogenase is a rate determining factor for p53-dependent growth totoxic drugs: Synergistic effects on normal and neoplastic lym- regulation. Mol Biol Cell 9: 15–28, 1998