Collagen type II modification by hypochlorite (cid:1)

Oxidation of proteins is a common phenomenon in the inflammatory process mediated by highly reactive agents such as hypochlorite (HOCl/OCl – ) produced by activated neutrophils. For instance, in rheumatoid arthritis hypochlorite plays an important role in joint destruction. One of the major targets for HOCl/OCl – is collagen type II (CII) — the primary cartilage protein. In our study, HOCl/OCl – mediated collagen II modifications were tested using various methods: circular dichroism (CD), HPLC, ELISA, dynamic light scattering (DLS), fluorimetry and spectrophotometry. It was shown that hypochlorite action causes deamination with consecutive carbonyl group formation and transformation of tyrosine residues to dichlorotyrosine. Moreover, it was shown that ammonium chloramine (NH 2 Cl) formed in the reaction mixture reacts with CII. However, in this case the yield of carbonyl groups and dichlorotyrosine is lower than that observed for HOCl/OCl – by 50%. CD data revealed that collagen II exists as a random coil in the samples and that chlorination is followed by CII fragmentation. In the range of low HOCl/OCl – concentrations (up to 1 mM) 10–90 kDa peptides are predominant whereas massive production of shorter peptides

Protein oxidation by reactive oxygen species (ROS) plays an important role in the pathomechanism of inflammatory diseases. ROS are produced by neutrophils, the main cells of inflammed synovial fluid in rheumatoid arthritis (RA). They are responsible for cartilage damage and joint erosion observed in this disease. The most toxic ROS generated by activated neutrophils is hypochlorite (HOCl/ OCl -) -the product of the myeloperoxidase/Cl -/hydrogen peroxide system (Weiss et al., 1982). Collagen type II (CII) is a primary protein component of cartilage and destruction of this protein caused by hypochlorite and by collagenase is one of the principal pathological reactions observed in RA (Edwards & Hallett, 1997).
It has been demonstrated (Weiss et al., 1983) that HOCl/OClcan react rapidly with amino acids and proteins to form low molecular and protein N-mono(di)chloramines. Davies et al. (1993) have shown that hypochlorite action leads to intensive fragmentation of tritiated acid soluble bovine collagen type II and that some of the amino acid-derived chloramines (N-chloroleucine or N-chloroalanine) increase the proteolytic susceptibility of N-chloramine modified collagen. At present, however, the knowledge of the early stages of HOCl/OClmediated collagen oxidation is poor. In our study we used various methods to monitor hypochlorite-mediated changes in CII aminoacid residues as well as changes in the structure of this protein.
(2 and 3). Collagen samples (1 mg/ml) in 0.2 M phosphate buffer, pH 7.4, with 0.2 M NaCl were incubated with 2 mM ammonium chloramine or with 1, 3 or 5 mM taurine chloramine for 2 h at 37 o C. The progress of reaction (2) was monitored by the decrease of absorption at 253 nm and the resulting NH 2 Cl concentrations were calculated as above. Estimation of dichlorotyrosine. Dichlorotyrosine concentration was calculated from difference spectra of chlorinated collagen (2 mg/ml) using e 306nm = 4.7 mM -1´c m -1 (Rudie et al., 1980). The reference cuvette contained native collagen solution (2 mg/ml). To reduce turbidity the samples were incubated with a 10 mg/ml solution of pronase E for 30 min prior to measurements.
Estimation of tyrosine and bityrosine. The fraction (%) of intact tyrosine moieties in collagen samples (2 mg/ml) was calculated from the relative fluorescence intensity (RFI) for emission at 307-309 nm (excitation at 275 nm). Bityrosine fluorescence was excited at 310 nm and monitored for emission at 410 nm.
Assay of carbonyl groups. Carbonyl groups were determined as dinitrophenylhydrazones according to Oliver (1987). Assays were carried out in 0.1 ml samples of native and chlorinated collagen (0.2 mg) and the concentration of protein dinitrophenylhydrazones was calculated using absorption coefficient e 368nm = 21 mM -1´c m -1 .
Assay of ammonium. Ammonium, a HOCl-mediated deamination product, was assayed using Nessler reagent. Fifty microliters of collagen (0.1 mg) sample was added to the assay buffer containing Nessler stock solution diluted with 0.12 M KOH (1:4, v/v). Absorption at 430 nm was measured after 30 min incubation at 37 o C and ammonium concentration was evaluated from a standard curve obtained for 5-100 mM NH 4 H 2 PO 4 solutions. HPLC assays.
(1) Assays were performed using an apparatus (Waters Millipore Comp., U.S.A.) equipped with a mBondpack C-18 column (3.9´300 nm) and the buffer set as follows: (A) 0.1% TFA in water, (B) 0.07% TFA in 80% acetonitrile in water. Gradient was formed from 0 to 100% of B within 15 min. Twenty micrograms of unchlorinated or HOCl-modified collagen was loaded onto the column.
(2) Assays were performed using a Shimadzu SCL 10A apparatus equipped with a Supelcosil TM LC-318 C-18 column (4´250 mm) and the buffer set as follows: (A) 0.1% TFA in water, (B) 0.1% TFA in acetonitrile. Gradient was formed from 0 to 100% of B within 30 min. One hundred microliters of sample was filtered through a Milipore YM 10 kDa membrane and loaded on to the column.
Electrophoresis. SDS/PAGE was performed in a MiniProtean II apparatus (BioRad, U.S.A.) using Tris/Tricine protein gels, 3.25 m g of total protein was loaded into each well. The gels were stained with Coomassie Blue R-250 or silver stained.
Competitive ELISA. To determine collagen epitope modification by chlorination competitive ELISA was used. Briefly, microtiter plates (Corning, U.S.A.) were coated with 5 mg/ml of CII. The antigens tested (CII modified by HOCl/OClor taurine monochloramine) at 0.5-50 mg/ml were mixed with an anti-CII monoclonal antibody (12.5 mg) and then added and incubated with a fixed competitive antigen for 1 h at room temperature and washed 3´to remove nonadsorbed immune complexes. The plates were then incubated with biotinylated goat anti-mouse antibody for 45 min at room temperature. Horseradish peroxidase (HRP) conjugated streptavidin diluted 1:1000 in 1% BSA/PBS was added and the plates were incubated for 45 min at room temperature. Then o-phenylenediamine dihydrochloride (OPD) was used as a substrate (5 mg of OPD in 10 ml of phosphate/citrate buffer pH 5.0) and incubated with 40 ml of 30% H 2 O 2 for 30 min at room temperature. The reaction was stopped with 3 M H 2 SO 4 . Absorbance was measured at 492 nm. As a control we used AgAb complex formation without preincubation with free test antigen (A 490 = 1.25).
Spectrophotometric measurements were carried out with a U-2000 spectrophotometer Vol. 50 Collagen type II modification by hypochlorite (Hitachi, Japan) and fluorescence spectra were recorded with an F-2000 fluorescence spectrophotometer (Hitachi, Japan). Circular dichroism spectra (CD) of 2.2 mM collagen solutions were measured with a J710 spectropolarimeter (Jasco, Japan). Dynamic light scattering measurements were performed with an MS200 (831.6 nm) laser unit (DynaPro, England) for collagen solutions of 1.5 mg/ml in 0.2 M phosphate buffer with 0.2 M NaCl. Samples of collagen unchlorinated and chlorinated with 1, 3 or 5 mM HOCl were filtered (100 nm) prior to the measurements. Radii (nm) were evaluated on the assumption that collagen exists in solution as a randomly coiled structure. Calculations were performed using an updated version of Dynamics Software (DynaPro, England).

RESULTS
It is widely known that protein chlorination results in modification of reactive amino-acid side chains containing such functional groups as: -SH, -SS-, RSR, -NH 2 , OH (in Tyr) or indole (in Trp) (Hampton et al., 1998). Chlorination of collagen II with HOCl/OClleads to the conversion of tyrosine to dichlorotyrosine as it was observed for other proteins (Kettle, 1996;Olszowski et al., 1996). As shown in Table 1 dichlorotyrosine formation could be eas-ily traced as absorbance increase at 305-308 nm in the difference spectra of chlorinated CII. Since CII does not contain tryptophan moieties it was possible to record the weak tyrosine fluorescence at 305 nm. This emission diminished with increasing HOCl/OClconcentration and could be treated as a measure of the content of intact tyrosine moieties in CII (Table 1). The dichlorotyrosine formation in CII was dose dependent and correlated with the loss of tyrosine moieties in the samples. Moreover, a shorter chlorination time (30 min) at the same range of HOCl/OClconcentrations (0.1-4 mM) produced a similar content (up to 40 mM) of dichlorotyrosine in 22 mM collagen (not shown). On the other hand, the 5 mM taurine monochloramine-mediated collagen chlorination did not lead to dichlorotyrosine formation although the more reactive 5 mM NH 2 Cl did produce dichlorotyrosine. The yield of the latter process was, however, lower (50%) than that of corresponding 5 mM HOCl/OClsolution (not shown).
Additionally bityrosine formation was detected in CII samples treated with 5 mM HOCl/OCl -. However, the bityrosine fluorescence at 410 nm was too weak to permit evaluation of its concentration. Since CII does not contain other (e.g. tryptophan) moieties which could serve as markers of oxidation, the dichlorotyrosine content seems to correlate with the degree of CII modification. Data represent averages of three measurements and refer to 22 mM CII concentration (calculated for 90 kDa CII species). *The evaluated radii represent the most abundant fraction (70-90%) of total content. Artificial radii over 100 nm were out of measurement range. All samples showed an additional artificial radius of 0.2 nm in the assay buffer control.
The HOCl/OClmediated chlorination of lysine e-amino groups in collagen converted them to aldehydes with concomitant deamination. Ammonium evolved in the reaction mixture was quickly converted with an excess of HOCl to mono-and dichloramines which in turn could chlorinate the protein. Ammonium chloramine-mediated CII chlorination was confirmed spectrophotometrically. Our results suggest that up to 15% of the initial amount of ammonium monochloramine reacted with CII during a 2 h incubation (27 ammonium monochloramines/collagen, mol/ mol). Ammonium and carbonyl group contents in HOCl/OCltreated CII are shown in Table 1. Whereas the carbonyl group content gradually increased with HOCl/OClconcentration to reach 45 mM (approx. 2 carbonyl groups for a 90 kDa fragment), the ammonium concentration was the highest for relatively low (1 mM) HOCl/OClconcentration ( Table 1). The latter result suggests that intensive deamination prevails up to 1 mM hypochlorite concentration, whereas at higher HOCl/OClconcentration the deamination is followed by the reaction that transforms the ammonium produced to chloramines (NH 2 Cl and NHCl 2 ). Circular dichroism spectra (CD) revealed changes in the content of secondary structures in CII samples. As shown in Fig. 1, an increase of the HOCl/OClconcentration decreased the ellipticity in the range of 190-240 nm (n-p*). All spectra correspond to random coil structure and suggest an absence of triple helix in all samples. Thus, the lower q corresponds to a higher content of degraded uncoiled a(II) collagen chains. Since the spectra showed the same position (220 nm) of q maximum it could be concluded that no other secondary collagen structures formed during chlorination. These results correlate with the concomitant decrease of the content of high molecular CII fraction -calculated from the peak height for 10.9 min retention time (Table 2). In addition, the presence of low molecular mass peptides in collagen samples filtered through a 10 kDa membrane was confirmed. The HPLC profiles shown in Fig. 2 suggest that only the highest HOCl/OClconcentration let to a massive formation of low molecular mass (< 10 kDa) peptides.
The antigen-antibody reaction between chlorinated collagen and a monoclonal antibody specific to native collagen was used to determine epitope deterioration caused by HOCl/OCland taurine monochloramine. HOCl/OClin a dose dependent manner affected the formation of immune complexes as shown in Fig. 3. This indicates that HOCl/OCl -, used at concentrations higher than 1 mM, seriously changed the epitope structure of CII, which did not compete with native CII for the specific antibody. In contrast to HOCl/OCl -, taurine monochloramine did not affect the ability of collagen to react with the antibody specific for the native protein (Fig. 4). SDS profiles (data not shown) indicated that the molar mass of unchlorinated CII was Vol. 50 Collagen type II modification by hypochlorite 475 90-95 kDa and that the HOCl/OCl --mediated fragmentation of CII could be monitored even within low HOCl/OClconcentration range (up to 1 mM). Samples of chlorinated CII gave one main band whose thickness diminished proportionally to the chlorination level both on Commassie and silver stained gels. Although no additional low molecular bands were found, dichlorotyrosine and carbonyls were detected in fractions (molecular mass <10 kDa) derived from CII samples chlorinated with 3 and 5 mM HOCl/OCl -(not shown). The latter result suggests the presence of low molecular mass oxidatively modified peptides in the samples. The dynamic light scattering method (DLS) was used to monitor changes in CII aggregates in water. The DLS data shown in Table 1 suggest that HOCl/OClchlorination decreased the radius of existing CII structures. Unchlorinated CII micelles had the radius of 476 S. Olszowski and others 2003 Peptide bonds (%)** 100 78.6 32.5 17.5 *CD data for 2.2 mM CII; **HPLC data derived from CII peak height (10.9 min retention time). For other details see Methods -HPLC procedure (1).

Figure 2. HPLC profiles of CII derived peptide ( < 10 kDa) fraction of unchlorinated CII (A) and CII chlorinated with 5 mM HOCl/OCl -(B).
Fractions of CII chlorinated with 1 and 3 mM HOCl/ OCl showed the same profile as profile A. 30 nm which corresponds to a structure of molecular mass 900-9000 kDa, much higher (10-100 times) than the mass of a single collagen molecule (90 kDa). Although the calculated mass depends strongly on the assumptions made (globular proteins or pollulans) the hypochlorite mediated relative decrease of the radius is unquestionable. Samples containing chlorinated collagen showed a radius corresponding to the molecular mass 150-200 kDa.

DISCUSSION
In this study we tested the ability of HOCl/ OCl -, a product of activated neutrophils, to modify the structure of collagen type II -the major protein present in joint cartilage. The present data suggest that hypochlorite modifies amino-acid side chains, secondary structure and type of aggregate. Assuming the molecular mass of CII to be approx. 90 kDa one could estimate the dichlorotyrosine and aldehyde content per one collagen molecule. The numbers of dichlorotyrosine moieties as well as of aldehyde groups vary from 0.25 to 2 per collagen molecule in collagen II samples modified with 1-5 mM HOCl/OCl -. Unlike the stable dichlorotyrosine moieties, aldehydes are easily oxidized with an excess of HOCl/OCl -. Therefore, due to further oxidation, carbonyl groups were not detected in collagen samples chlorinated with more concentrated HOCl/ OCl -(> 5 mM).
At low hypochlorite concentrations (up to 1 mM) ammonium content seems to be a better marker of lysine group oxidation since carbonyl formation is concomitant with deamination. These results are also supported by the fact that free amino group level in CII samples was not elevated as shown by the 2,4,6-trinitrobenzenesulfonic method (not shown) and are consistent with the suggestion that HOCl/OCl --modified amino groups are unavailable for reaction with fluorescamine (Davies et al., 1987). At higher HOCl/OClconcentrations (3 and 5 mM) ammonium level was decreased which could be interpreted in terms of the ammonium chloramine formation in the presence of HOCl/OCl - (Table 1). All these facts taken together suggest that collagen is chlorinated with both HOCl/OCland ammonium chloramine and that the higher the initial HOCl/OClconcentration, the higher the ammonium monochloramine content in the reaction mixture.
As suggested by the HPLC, SDS/PAGE and CD data, hypochlorite chlorination was followed by CII fragmentation even at low hypochlorite concentrations (less than 1 mM). Unlike in the case of albumin (Olszowski et al., 1996) and fibronectin (Vissers & Winterbourn, 1991), for collagen this tendency seems to prevail over aggregation. The degree of CII fragmentation evaluated from CD measurements was very similar to that calculated from experiments with tritium labelled type II bovine articular collagen (Davies et al., 1993). The data from competitive ELISA showed that HOCl/OClat low concentrations (up to 1 mM) did not affect the formation of immune complexes between chlorinated CII and anti-collagen specific antibodies. This may suggest that the structure of the epitopes was not affected. Nevertheless, lysine and tyrosine modifications in CII molecules were observed. On the other hand, HOCl/OClat higher concentrations (35 mM) significantly diminished Vol. 50 Collagen type II modification by hypochlorite 477 Bars represent unchlorinated CII (o) and CII chlorinated with 1 (n) or 3 (n) mM taurine monochloramine. Other details as for Fig. 3.
the binding of the antibody due to the fragmentation of CII. In contrast to HOCl/OCl -, taurine monochloramine did not affect CII binding to the antibody. Therefore, these results suggest that taurine may prevent collagen structure deterioration in vivo, since taurine monochloramine is formed from hypochlorite at the inflammation site (Weiss et al., 1982). Our results demonstrating that chlorination was followed by fragmentation, it leads to the conclusion that the structures found in chlorinated CII samples (DLS) are aggregates of chlorinated low molecular CII-derived peptides. However, since physiological hypochlorite or chloramine concentrations are lower (mM) than those used in our model system, only slight (< 10%) collagen degradation could be expected in the synovial fluid. Thus it could be suggested that in the inflammatory process the radius of CII micelles decreases approx. 3-5 times. The components of these modified micelles are probably weakly bound peptides (10-90 kDa) containing carbonyl groups and exhibiting more anionic character than intact collagen (Olszowski et al., 1996). The dramatic change in the size and composition of CII aggregates may influence the properties of the synovial fluid. At present, the biological role of this process is unknown, although, as it has been observed for fibronectin (Olszowski et al., 1994), it is possible that this modification may influence PMN activation in the synovial fluid.
Finally, our data confirm previous observations that chlorination of a protein by HOCl/OClmay alter its properties, including structure, susceptibility to proteolysis, immunogenity and pathogenecity (Marcinkiewicz et al., 1991;1992;Kwaoeny-Krochin et al., 2002). However, the effect of HOCl/OClon the protein properties will be different, depending on the exposure of target amino acids in the chlorinated protein (e.g., ovalbumin vs. collagen).