Advanced Oxidation Protein Products as Novel Mediators of Inflammation and Monocyte Activation in Chronic Renal Failure^{1, 2}

Uremic patients suffer from a dysregulation of the immune system, characterized by the paradoxical coexistence of a clinically and biologically evident state of immunodeficiency and activation of immunocompetent cells, including T and cells, and of monocytes (reviewed in 1 . Substantial evidence accumulated in recent years has implicated activated monocyte-derived proinflammatory cytokines in such immunologic disorders (2, 3, 4). The presence of a chronic inflammatory state has also been widely documented in end-stage renal disease patients receiving maintenance hemodialysis (recently reviewed in 5 . It is commonly attributed to the constantly renewed activation of circulating neutrophils and monocytes following blood passage through dialysis circuits and subsequent generation of activated complement components due to contact of plasma with bioincompatible membranes and/or transfer of endotoxins from the dialyzate to the blood compartment. This conjunction leads to 1) a massive generation of reactive oxygen species (ROS),⁴ e.g., O₂⁻, H₂O₂, ¹O₂, and OH⁻, and chlorinated oxidants such as HOCl by activated neutrophils; and 2) an increased production of cytokines, e.g., IL-1β and TNF-α, by activated monocytes. Recent studies suggest that both the oxidative stress induced by ROS and the proinflammatory effects of these cytokines are reinforced by profound defects in antioxidant (6, 7) and anticytokine systems (3, 4) and largely contribute to β₂m amyloid arthropathy (8, 9) and accelerated atherosclerosis (10, 11), which remain leading causes of morbidity and mortality in dialysis patients (12).

The exquisite vulnerability of proteins to ROS is now well documented (13, 14, 15, 16). Oxidation of amino acid residues such as tyrosine, leading to the formation of dityrosine, protein aggregation, cross-linking, and fragmentation, is an example of ROS-mediated protein damage in vitro. In contrast, evidence for the presence of such oxidatively damaged proteins in vivo and their possible clinical significance was still lacking until recently (15, 17). Indeed, in the search for whether such protein oxidative damage could reflect the dialysis-associated oxidative stress, we were able to isolate and characterize dityrosine-containing protein cross-linking products in the plasma of dialysis patients, which we designated advanced oxidation protein products (AOPP) (17).

The contribution of uremia per se to the chronic inflammatory state has been suggested, and consistent evidence has been afforded that both monocyte activation and a defect in antioxidant systems occur early in the course of chronic renal failure and gradually increase with its progression to end-stage renal disease (4, 7). Interest has focused on the role of “uremic toxins” generated during the course of chronic renal failure, some of which having known effects on neutrophil and monocyte functions (18, 19). Among these, growing efforts are being devoted to the potential toxicity of advanced glycation end products (AGE) (19, 20, 21, 22, 23, 24).

The formation of AGEs has been widely documented during diabetes and aging and has been held to be responsible for tissue degradation. The presence of increased plasma levels of AGEs has been observed in dialysis patients, independently of diabetes, and it is likely that, like ROS, AGEs contribute to β₂m deposits (21, 22). Interestingly, the hypothesis that oxidative stress is implicated in AGE formation might also be relevant to the uremic toxicity syndrome (24, 25). More recently, we reported that the well-characterized AGE-pentosidine accumulates with progression of chronic renal failure in close relationship with neopterin, a well-characterized monocyte activation marker (26).

In the search for whether AOPP could act as mediators in the monocyte-mediated inflammatory process associated with chronic renal failure and participate in the monocyte activation state, we determined 1) in vivo, AOPP levels in the plasma of uremic patients with varying degrees of renal failure in relationship with immunologic markers; and 2) in vitro, the potency of AOPP to induce the monocyte respiratory burst.

One hundred and sixty-two nondialyzed chronic renal failure patients at various stages of renal failure were enrolled in the study after giving informed consent. They included mild renal failure defined by creatinine clearance (Ccr) of 41 to 80 ml/min (n = 73), moderate renal failure with Ccr of 20 to 40 ml/min (n = 53), and advanced renal failure with Ccr <20 ml/min (n = 36). Primary renal diseases are indicated in Table I. Patients suffering from diabetes mellitus, systemic lupus erythematosus, malignant tumors, or acute infection or receiving immunosuppressive therapy at the time of blood sampling were excluded from the study. Controls consisted of 31 healthy adults recruited among blood donors from our blood transfusion center.

Venous blood (5–10 ml) was collected in standard sterile polystyrene vacuum tubes, with 5 mM EDTA. After centrifugation (600 × g for 10 min), the plasma was stored in 500-μl aliquots at −70°C until use. Assays were conducted on duplicate samples thawed once.

To investigate the effect of AOPP on monocyte respiratory burst, monocytes were isolated from blood (20 ml) of normal volunteers by a two-step procedure involving erythrocyte sedimentation on dextran followed by leukocyte-rich plasma layer sedimentation on Nycoprep (1.068 solution; Nycomed Pharma, Oslo, Norway) according to the supplier’s instructions.

AOPP were determined in the plasma using the semiautomated method previously devised in our laboratory (17). Briefly, AOPP were measured by spectrophotometry on a microplate reader (model MR 5000, Dynatech, Paris, France) and were calibrated with chloramine-T (Sigma, St. Louis, MO) solutions that in the presence of potassium iodide absorb at 340 nm (27). In test wells, 200 μl of plasma diluted 1/5 in PBS was placed on a 96-well microtiter plate (Becton Dickinson Labware, Lincoln Park, NJ), and 20 μl of acetic acid was added. In standard wells, 10 μl of 1.16 M potassium iodide (Sigma) was added to 200 μl of chloramine-T solution (0–100 μmol/liter) followed by 20 μl of acetic acid. The absorbance of the reaction mixture is immediately read at 340 nm on the microplate reader against a blank containing 200 μl of PBS, 10 μl of potassium iodide, and 20 μl of acetic acid. The chloramine-T absorbance at 340 nm being linear within the range of 0 to 100 μmol/liter, AOPP concentrations were expressed as micromoles per liter of chloramine-T equivalents.

Human serum albumin (HSA; type V; Sigma) was exposed to HOCl (Fluka, Buchs, Switzerland) as described previously (17). Briefly, HOCl stock solution (100 mM) was freshly prepared in PBS, and the concentration was measured by spectrophotometry using a molar extinction coefficient of 350 M⁻¹ cm⁻¹ at 290 nm at pH 12. Various concentrations of oxidants of HOCl were added to HSA at the indicated HSA/HOCl molar ratio. The AOPP-HSA preparation was incubated for 30 min at room temperature and then dialyzed overnight against PBS and tested for AOPP content. AOPP-HSA were also prepared by exposing purified HSA (100 mg/ml) to chloramine-T or hydrogen peroxide (H₂O₂) at the indicated concentrations and was dialyzed overnight against PBS. Dityrosine synthesis was adapted from the Anderson method (28). Briefly, 270 mg of tyrosine was dissolved in 250 ml of 0.2 M borate buffer, pH 9.5, in the presence of 6 mM H₂O₂. The reaction was started by the addition of 6.3 mg of horseradish peroxidase and performed for 18 h at 37°C. Then the mixture was concentrated almost to dryness in a rotatory evaporator under vacuum at 35°C. The brown powder was suspended in water acidified by concentrated HCl. A precipitated was removed by filtration through a glass filter funnel (G4; Millipore, Bedford, MA). Further separation from unreacted l-tyrosine and impurities was performed by chromatography on a cation exchanger. The brown solution was then applied to a fibrous cellulose phosphate (Sigma; 50–100 μm) column (1.5, 25 cm) equilibrated with 0.2 N acetic acid. After washing, the elution was performed with 0.5 M NaCl in 0.2 N acetic acid. Fractions of 3 ml were collected and evaluated after dilution in 0.1 N NaOH spectrophotometrically at 280 and 315 nm. Fluorescence detection (l_ex = 320 nm, l_em = 410 nm) was also monitored. Fractions exhibiting dityrosine fluorescence were pooled and concentrated by lyophilization. The concentrate was solubilized in distilled water, filtrated, then loaded on a Dowex 50W-X8 (Bio-Rad) column previously soaked in 1 M HCl. The column was extensively washed with water to remove all NaCl and acetic acid. The dityrosine was subsequently eluted with 2 M ammonium hydroxide. Fluorescent fractions were concentrated in a rotatory evaporator under vacuum. The final fraction was reprecipitated several times in methanol-ether and stored under ether. Finally, a slightly yellow dityrosine was obtained (yield, 12%) and characterized by UV spectra, fluorescence emission, and magnetic resonance spectroscopy (29, 30).

Spectroscopic analyses on HSA, AOPP-HSA, and purified dityrosine were performed on a Kontron SF 25 spectrophotometer; UV and visible spectra were recorded in PBS at pH 7.5 at room temperature. Absorption and emission spectra were recorded in denaturing 6 M urea buffer at pH 7.5. Protein-bound dityrosine production was assayed in plasma or HSA samples by fluorescence measurements after dilution of the sample in 20 mM phosphate buffer, pH 7.5, in the presence of 7 M urea (10). After a 30-min incubation, the fluorescence emission spectra of dityrosine was recorded from 550 to 350 nm following excitation at 320 nm using a Kontron SF 25 spectrophotometer (Kontron, Zurich, Switzerland) and was measured at its maximum at 410 nm. The assay was calibrated by means of external standardization using a calibration curve generated in the same urea medium with authentic dityrosine. Its concentration was monitored spectrophotometrically at 315 nm, E = 5 mM⁻¹ cm⁻¹ at pH 7.5 (29, 30).

Carbonyl residues were determined as previously described (31) using dinitrophenylhydrazine. Briefly, samples were submitted to 10 mM dinitrophenylhydrazine in 2.5 M HCl for 1 h, followed by deproteinization with 20% TCA. The pellet was washed three times in ethanol/ethyl acetate and solubilized in guanidine 6 M. The carbonyl concentration was measured by spectrophotometry at an OD of 370 nm with ε₃₇₀ = 22 mM⁻¹ cm⁻¹. The protein concentration was determined in parallel using OD at 280 nm in reference to BSA.

The AGE-HSA used in vitro was prepared by incubating HSA (type V; Sigma; 50 mg/ml) with 500 mM glucose in PBS for 65 days at 37°C under sterile conditions. The AGE-pentosidine, as a marker of nonenzymatic glycation of proteins, was measured using a modification of the method described by Odetti et al. (32). Briefly, plasma proteins or AGE-HSA were precipitated on TCA. The pellets were hydrolyzed in 2 ml of 6 N HCl and dissolved in 250 μl 0.01 M heptafluorobutyric acid (Sigma). The equivalent of 4 mg of plasma protein was injected into an HPLC system (Waters Division of Millipore, Marlborough, MA). A 25- × 0.46-cm C₁₈ Vydac type 218TP (10 μm) column was used (Separations Group, Hesperia, CA). HPLC was programmed with a linear gradient of 10 to 17% acetonitrile from 0 to 35 min. Pentosidine was eluted at approximately 30 min as monitored by fluorescence excitation at 335 nm and emission at 385 nm. Pentosidine prepared according to the method of Sell and Monnier (33) was used as standard, and results were given in picomoles per milligram of protein.

Thiobarbituric-reacting substances, including malondialdehyde (MDA), were determined using a commercially available kit according to the supplier’s instructions (Sobioda, Grenoble, France). Briefly, tetraethoxypropane was used as standard; each molecule of tetraethoxypropane hydrolyzes to yield one molecule of MDA under assay conditions. One hundred microliters of plasma was mixed with a mixture of thiobarbituric and perchloric acid and boiled for 1 h, then butanol (Carlo Erba, Milan, Italy) was added. Tubes were vortexed and centrifuged to extract MDA. Fluorometric measurements (excitation at 532 nm and emission at 553 nm) were performed on supernatants, using a Kontron SFM 25.

The level of selenium-dependent plasma GSH-Px was determined according to the method of Anderson (34) and expressed as micromoles of NADPH oxidized per milliliter.

Commercially available kits were used for measuring plasma levels of neopterin (monocyte activation marker; RIA, Behring Diagnostic, Rueil-Malmaison, France), IL-1R antagonist (IL-1Ra; ELISA Quantikine, R&D Systems, Minneapolis, MN), soluble CD23 (B cell activation marker; ELISA, BioSource Europe, Fleurus Belgium), and TNF-α (ELISA, BioSource). Plasma levels of TNF soluble receptors, TNF-sR55 and TNF-sR75, were determined as described previously (35) with specific mAbs provided by Dr. H Gallati (F. Hoffmann-La Roche, Basel, Switzerland). The plasma level of soluble CD25 (T cell activation marker) was determined with a kit (EIA, Cobas Core, sIL-2R, Hoffmann-La Roche) provided by M. E. Nobile (Roche, Neuilly, France).

The capacity of AOPP-HSA or AGE-HSA to activate the monocyte respiratory burst was measured by chemiluminescence, using dimethylbiacridinium (lucigenin) as the chemoluminigenic substrate. In this system the reductive dioxygenation of lucigenin that yields luminescence is strictly NADPH-oxidase dependent (36). One hundred microliters of monocyte suspension (2 × 10⁵/ml) was automatically injected into tubes containing 100 μl of HBSS (basal activity), the tested preparations at the indicated concentrations (native HSA, AOPP-HSA or AGE-HSA, or human AB serum-opsonized zymosan (2 × 10⁹ yeast cell wall units/ml), or PMA (Sigma). Chemiluminescence production was measured in duplicate in a luminometer (LB953 Berthold, Wildbad, Germany), and luminescence intensity was expressed in counts per minute.

The data were analyzed using standard statistical methods (Statistica Software, Tulsa, OK). Differences between means were evaluated using Student’s paired or unpaired t test where appropriate or ANOVA for comparing more than two groups. Relationships between variables were tested using simple (Pearson’s r correlation coefficient) or multiple linear regression analysis as indicated. All values are reported as the mean ± SEM. Statistical significance was set at p < 0.05.

The mean plasma level of AOPP was significantly higher in chronic renal failure patients than in healthy control subjects (52 ± 2 vs 29 ± 5 μmol/l; p < 0.001). AOPP concentrations increased over a nearly threefold range from the incipient to the advanced stage of chronic renal failure (p < 0.001; Table II). An inverse relationship between AOPP levels and creatinine clearance was seen (r = −0.46, p < 0.001; Fig. 1). However, no significant effect of the underlying nephropathy on circulating levels of AOPP was identified (Table III).

The mean plasma level of MDA, the lipid peroxidation product, was significantly higher at the incipient stage of chronic uremia than in controls, but remained stable during the progression of renal failure (Table II). Thus, no relationship was found between AOPP and MDA levels (r = 0.16, p = NS). In contrast, the plasma level of GSH-Px, a major antioxidant enzymatic system, decreased significantly with the progression of renal failure, reaching half its initial level at an advanced stage of chronic renal failure (Table II; p < 0.001). An inverse relationship was found between AOPP and GSH-Px levels (r = −0.34, p < 0.001). However, this correlation was not significant after these two parameters were adjusted for creatinine clearance (r = 0.16, p = NS).

A study of the relationships among AOPP, AGEs, and cell activation markers was performed in a representative group of 56 patients, equally distributed across the range of chronic renal failure (Table IV). Plasma AGE-pentosidine levels gradually increased over a fourfold range with progression from early to advanced chronic renal failure (p < 0.001), and a close relationship was observed between AOPP and AGE-pentosidine levels (r = 0.52, p < 0.001; Fig. 2). This correlation remained significant after these two parameters were adjusted for creatinine clearance (r = 0.42, p < 0.01).

Plasma concentrations of activation markers of T cells (soluble CD25) and B cells (soluble CD23) were significantly higher in chronic renal failure patients than in controls and gradually increased with the progression of renal failure, reaching levels in the advanced stage twice that of the early stage. Likewise, the mean plasma level of the monocyte activation marker, neopterin increased strikingly with loss of renal function, reaching values 30-fold higher in advanced chronic renal failure (Table III). While no relationship was found between plasma AOPP and soluble CD25 or soluble CD23 levels, a highly significant correlation was observed between neopterin and AOPP (r = 0.55, p < 0.0001) or AGE-pentosidine levels (r = 0.84, p < 0.0001; Fig. 3). Both correlations remained significant by multiple regression analysis after adjustment of AOPP and AGE-pentosidine values for creatinine clearance (r = 0.32, p = 0.007 and r = 0.67, p < 0.0001, respectively).

IL-1β was only rarely present in chronic renal failure patients, mostly in those with advanced disease (data not shown). In contrast, IL-1Ra plasma levels were higher in chronic renal failure patients than in controls (p < 0.01; Table IV), and weak correlations were observed between IL-1Ra and AOPP levels (r = 0.30, p < 0.01) or AGE-pentosidine (r = 0.26, p = 0.05); these correlations were not significant by multiple regression analysis when IL-1Ra and AOPP values were adjusted for creatinine clearance.

In all chronic renal failure patients TNF-α was detected and increased with deterioration of renal function (Table IV). Significant correlations were observed between TNF-α and AOPP (r = 0.36, p = 0.004) or AGE-pentosidine levels (r = 0.50, p = 0.0001; Fig. 3). Likewise, both TNF-sR55 and TNFsR75 levels increased with the progression of chronic renal failure (Table IV), and significant correlations were observed with the levels of AOPP (r = 0.41, p = 0.002 and r = 0.55, p = 0.0001, respectively) or AGE-pentosidine (r = 0.46, p = 0.0004 and r = 0.60, p = 0.0001, respectively; Fig. 4). However, when all values were adjusted for creatinine clearance, r correlation coefficients remained significant only with TNF-sR75. As previous studies, including ours (3, 4), have shown that the final determination of a cytokine’s biologic activity is best reflected by the ratio of the cytokine to its inhibitor, we determined the molar ratios TNF-sR55/TNF-α and TNF-sR75/TNF-α and performed regression analysis between these ratios and AOPP or pentosidine values. For AOPP, the r correlation coefficients were 0.15 (p = 0.07) and 0.07 (p = NS) with TNF-sR55/TNF-α and TNF-sR75/TNF-α, respectively, and for pentosidine the r values were 0.06 (p = NS) and 0.18 (p = NS), respectively.

We first determined the optimal conditions for HSA oxidation as assessed by AOPP, dityrosine, and carbonyl concentrations and with respect to the nature and concentration of the oxidant. As shown in Figure 5, treatment of HSA with HOCl or chloramines induced a dose-dependent increase in AOPP, dityrosine, and protein carbonyl concentrations, whereas such an increase was not obtained with H₂O₂. The ineffectiveness of H₂O₂ remained consistent at 1- or 10-M concentrations (data not shown).

Since chlorinated oxidants appeared to be the most efficient oxidants in inducing AOPP-HSA, we further studied spectroscopic characteristics of HOCl-induced AOPP-HSA compared with those of native HSA and purified dityrosine. Data obtained from UV-visible spectrophotometry are presented in Figure 6. They illustrate the relative absorbance contribution of native HSA and AOPP-HSA in the 200 to 450 nm range. Below 250 nm, the spectra showed typical absorbance of the peptide bonds in both native and AOPP-HSA. Significant differences occurred in the 250 to 300 nm region typical of aromatic residues. In native HSA, this peak at 280 nm was sharp and well individualized, whereas in AOPP-HSA, it was considerably altered, demonstrating a modification in the content or characteristics of tyrosine and/or tryptophan residues induced by chlorinated oxidants. In addition, above 300 nm, the presence of an absorbance shoulder at 315 nm appeared to be a feature of HSA treated with chlorinated oxidants, as shown in AOPP-HSA. Interestingly, in the same pH conditions, purified dityrosine showed a peak at the same wavelength, at 315 nm. Of note, no band was detectable in native HSA.

To further investigate the relationships between AOPP-HSA and dityrosine spectral characteristics, fluorescence spectrometry was used. AOPP and dityrosine were diluted in 20 mM phosphate buffer at pH 7.5 in the presence of 6 M urea. According to fluorescence characteristics of dityrosine, which shows emission spectrum with a peak at 410 nm after excitation at 320 nm, fluorescence spectra of both AOPP-HSA and native HSA were recorded under these conditions (Fig. 6,B). Both purified dityrosine and AOPP-HSA presented the same emission fluorescence with a maximum at 410 nm. Flat spectra were observed for both HSA and control urea buffer. Excitation spectra recorded at a fixed wavelength of 410 nm (Fig. 2 B), indicated an overlapping between dityrosine and AOPP-HSA within the 320 to 350 nm, whereas the contribution of native HSA was close to that of the control urea buffer. One interesting feature in native HSA is the presence of the peak typical of aromatic residues around 280 nm (as previously mentioned for absorbance spectra), which is not visible in either AOPP-HSA or dityrosine. Taken together, these spectral characteristics clearly demonstrate the presence of dityrosine in AOPP-HSA.

As shown in Figure 7, HOCl-induced AOPP-HSA triggered a respiratory burst in isolated human monocytes, as measured by lucigenin-amplified chemiluminescence. The chemiluminescence production increased with the HSA/HOCl molar ratio, reaching a maximum at 1:60. AOPP-HSA at a HSA/HOCl molar ratio of 1:120 induced a lower chemiluminescence production, and this was observed as a significant increase in both AOPP and dityrosine and a moderate increase in carbonyl concentrations compared with AOPP-HSA at a HSA/HOCl molar ratio of 1:60. In contrast, no oxidative burst was triggered by native HSA or by purified dityrosine at the concentration range of 1 to 100 μM, which is the same as that found in the optimal concentration of AOPP for inducing respiratory burst.

AOPP-HSA preparation at the HSA/HOCl molar ratio 1:60 induced a chemiluminescence response similar to that obtained with opsonized zymosan but significantly lower than that obtained with PMA (p < 0.05). AGE-HSA used at the same final concentration of 2 mg/ml induced a significant rise of chemiluminescence compared with native HSA. However, chemiluminescence production induced by AGE-HSA preparation was significantly lower than that induced by AOPP-HSA (Fig. 8). Interestingly, this AGE-HSA preparation, adjusted to a 20 mg/ml protein concentration, contained AOPP (55 μmol/l); this concentration was similar to that found in AOPP-HSA obtained after HSA treatment with HOCl using a 1:60 molar ratio. In contrast, the latter AOPP-HSA, which induced the maximal chemiluminescence production, contained the same amount of AGE-pentosidine as native HSA, thus ruling out the contribution of AGEs in AOPP-induced monocyte respiratory burst.

The present study provides several lines of evidence to suggest that AOPP that we previously isolated and biochemically characterized in plasma from hemodialyzed patients (17), act as mediators of the monocyte activation state associated with chronic uremia.

First, plasma levels of AOPP are significantly elevated in uremic patients compared with those in healthy individuals. The increase starts at an early stage of chronic renal failure and gradually rises with the progression of renal failure, as emphasized by the highly significant inverse relationship between plasma concentrations of AOPP and the glomerular filtration rate. Second, and more importantly, the accumulation of AOPP during the course of chronic renal failure is not only related to the decrease in kidney excretory function, but is also closely associated with several immunoinflammatory markers. However, whether the level of AOPP is related to the rate of progression of renal failure needs further investigation by a sequential follow-up of AOPP levels in patients with the same type of nephropathy but with distinct rates of decline in glomerular filtration.

The present finding that AOPP accumulation coexists with decreased GSH-Px level, while the plasma concentration of malondialdehyde remains stable, supports the contention that AOPP are more accurate markers of oxidative stress than lipid peroxidation products (17). The occurrence of oxidative stress in chronic renal failure patients whether on dialysis or not has been suggested in studies showing an imbalance between oxidant and antioxidant systems and their cofactors (6, 7). Recent reports have stressed the role of such an oxidative stress in the accelerated atherosclerosis process associated with end-stage renal disease (12).

To determine the role of AOPP in uremia-associated immune dysregulation, we analyzed their relationships with cell activation markers, with emphasis on monocytes as a potential source of oxidants and proinflammatory cytokines. While no relationship was found between AOPP and T or B cell activation markers, a close correlation was observed between AOPP and neopterin, the monocyte activation marker. This selective relationship between AOPP and monocyte activation was further established with positive correlations between AOPP and TNF-α and its soluble receptors, and, to a lesser degree, with IL-1Ra, although these correlations tended to be of only borderline significance when values were corrected for creatinine clearance. These latter findings suggested that the relationship was not related to cytokine or cytokine inhibitor biologic activities, and this was further evidenced by the absence of correlation between AOPP and the TNF-sR55/TNF-α or TNF75/TNF-α molar ratios, which may better reflect the biologic activity of TNF-α (3, 4).

Of note, in another state of profound immune dysregulation, in HIV patients, we observed very high plasma levels of AOPP. In this clinical condition, characterized by a pronounced oxidative stress in the absence of renal failure, AOPP was an exquisite marker of oxidative stress correlating tightly with the degree of monocyte activation (37).

Another important aspect of the present study was to further investigate the relationship between AOPP and advanced glycosylation proteins as assessed by AGE-pentosidine. We and other investigators have previously documented that chronic renal failure is associated with increased AGE formation independent of diabetes or aging (24, 25, 26). In the present study, the close correlation observed between plasma AGE-pentosidine and AOPP demonstrates that this relationship already exists in uremic patients not yet undergoing dialysis and further suggests that AGE and AOPP may share common mechanisms of formation and/or common biologic activities in vivo. Moreover, several studies pointed to the involvement of oxidative pathways in the formation of AGE, notably in patients with end-stage renal failure (24, 38).

Our previous observations demonstrated that, in vivo, AOPP result from oxidant-induced protein cross-linking and correlate with dityrosine levels. To determine the mechanisms by which AOPP could be involved in monocyte activation, the potential proinflammatory effects of AOPP-HSA produced in vitro by exposing HSA to chlorinated oxidants were studied. As evidenced in the case of carbonyl and dityrosine levels in oxidant-treated HSA, chlorinated oxidants appear very efficient in inducing AOPP. Spectroscopy studies reveal that AOPP-HSA could be identified as follows: in UV-visible spectra, by perturbation of the aromatic residue band associated with a large band at 320 nm, the latter property being used for routine quantification of AOPP; in fluorescence spectra by the loss of tryptophan and tyrosine emission and the appearance of the dityrosine contribution clearly visualized in the excitation spectrum. Therefore, the spectral measurement of AOPP, which is maximum at 320 nm, is the result of an overlapping between chromophores, among which dityrosine is identified as a predominant component.

We then demonstrated that AOPP-HSA, but not purified dityrosine, possesses the ability to activate isolated monocytes in vitro, emphasizing three remarkable features of the induced respiratory burst: 1) its induction is dependent on the chlorinated nature of the oxidant, i.e., HOCl and chloramine, but not H₂O₂; 2) its intensity is directly related to the level of protein oxidation as defined by the molar ratio of protein to oxidant and ascertained by dityrosine and carbonyl measurements; and 3) it is related to the protein cross-linking structure, since purified dityrosine alone has no effect on monocytes.

In the past, studies on the interactions between proteins and oxidants have focused on the structural changes induced by oxidants generated by water pulse radiolysis, including superoxide and hydroxyl radicals (13, 14). Such studies have demonstrated that structural modifications of proteins (selective loss of an amino acid, fragmentation, or aggregation) were highly dependent on the nature of the oxidant. Since AOPP formation is optimal with chlorinated oxidants, it is interesting to speculate that its formation in vivo might result from enzymatic activity of phagocyte-derived MPO (39). Interestingly, although the best-characterized product of MPO is HOCl, MPO-derived enzymatic activity can also generate dityrosine (16), a compound that correlates with AOPP levels, and aldhehyde compounds, which may exert a potent biologic effect at the site of inflammation (40). MPO can also convert l-serine to glycolaldehyde, which mediates the formation of carboxymethylysine, described as a byproduct of protein glycation (41). The pathophysiologic relevance of MPO-derived chlorinated oxidant is also illustrated by its role in the formation of oxidatively modified low density lipoproteins (42).

Another important aspect of the present study was to further investigate the relationship between AOPP and AGEs. Both AOPP-HSA and AGE-HSA triggered monocyte NADPH oxidase activation, leading to superoxide anion production. Of note, while AOPP-HSA did not contain AGE-pentosidine, significant concentrations of AOPP were found in AGE-HSA preparations. Taken together, these data suggest that AGE-proteins contain some structural motif that results from an oxidative process and support the hypothesis that AGE-mediated biologic activities might depend on their level of oxidation (43). The interaction between AGEs and macrophages is now well established. Macrophages may internalize AGE-modified proteins via a specific receptor, or RAGE (21, 44), which is also expressed by endothelial cells (45). Interestingly, the macrophage RAGE can be up-regulated by TNF-α (21, 46). How macrophages process AOPP-HSA remains elusive, but taking into account a possible structural resemblance between AGE and AOPP, the involvement of a similar receptor-mediated process is plausible.

In conclusion, AOPP may represent a novel class of proinflammatory mediators acting as a mediator of oxidative stress and monocyte respiratory burst. The monocyte is thus, at the same time, the elective cellular target of AOPP and a potential source of oxidants inducing AOPP.

We are grateful to Dr. P. Soubiran (Laboratoire d’Analyses Médicales Spécialisé en Immunologie, Nice, France) for measurement of plasma levels of soluble CD25, and to Dr. M. Thévenin, Ph.D. (Laboratoire de Biochimie A, Necker Hospital, Paris, France), for determination of plasma GSH-Px activity. We thank Françoise Tresset for skillful technical assistance, Mathilde Labrunie, M.S., for statistical analysis, and Dr. M. Webb M.D. (St. Thomas Hospital, London, U.K.), for critical advise on the manuscript.