Immunology Letters 103 (2006) 3338

Immunoglobulins can utilize riboavin (Vitamin B2) to activate the antibody-catalyzed water oxidation pathway
Jorge Nieva b , Lisa Kerwin a , Anita D. Wentworth a , Richard A. Lerner a,c , Paul Wentworth Jr. a,c,
b a Department of Chemistry, The Scripps Research Institute, 10550 N. Torrey Pines Road, La Jolla, CA 92037, United States Department of Molecular and Experimental Medicine, The Scripps Research Institute, 10550 N. Torrey Pines Road, La Jolla, CA 92037, United States c Oxford Glycobiology Institute, Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK

Received 19 November 2005; received in revised form 20 November 2005; accepted 21 November 2005 Available online 15 December 2005

Abstract We have recently discovered a reaction that all antibodies, regardless of source or antigenic specicity can catalyze, that is the reaction between singlet dioxygen (1 O2 * ) and H2 O to generate H2 O2 . We have named this process the antibody-catalyzed water oxidation pathway (ACWOP). As part of our ongoing investigations into the possible biological role of this pathway, we have studied whether isoalloxazine-containing cofactors, that are known to be endogenous photosensitizers via Type-II pathways to generate 1 O2 * , such as riboavin (RF, Vitamin B2) can trigger the ACWOP. Herein we show that regardless of the antigenic specicity or heavy and light chain composition, all antibodies and their fragments are able to intercept the 1 O2 * generated by photo-oxidation of RF in the presence of oxygen (ambient aerobic conditions) to activate the ACWOP. The initial rate of HOOH generation by a panel of murine antibodies ranges from 0.218 to 0.998 M/min. The initial rate of antibody-catalyzed HOOH production is accelerated in D2 O and is quenched in NaN3 , highlighting the key intermediacy of 1 O2 * in the process. Critically, the ACWOP is photo-activated at physiologically relevant concentrations of RF (<50 nM) suggesting that this pathway may be relevant in an in vivo setting. Finally, when activated by RF the ACWOP generates oxidants that accelerate the hemolysis of sheep RBCs hinting at a pathophysiological effect of this RF-induced photo-oxidation pathway. 2005 Elsevier B.V. All rights reserved.
Keywords: Antibody; Riboavin; Catalysis; Oxidation; Hemolysis

1. Introduction One of the most underappreciated facts in immunochemistry is that a certain subclass of antibodies are known to transport isoalloxazine-containing cofactors, such as riboavin (RF), riboavin 5 -monophosphate (FMN) and avin adenine dinucleotide (FAD) in plasma [1,2]. The likely reason that this has escaped further attention is that until recently there was no apparent role for the redox or photochemical properties of avincontaining cofactors in the biological function of immunoglobulins. However, we have recently shown that all antibodies, regardless of source or antigenic specicity, can catalyze the reaction between singlet oxygen (1 O2 * ) and H2 O into hydrogen peroxide (HOOH) [35]. During this process oxidants never

Corresponding author. Tel.: +1 858 784 2576; fax: +1 858 784 2593. E-mail address: paulw@scripps.edu (P. Wentworth Jr.).

before considered to occur in biology, such as dihydrogen trioxide (HOOOH), the hydrotrioxy radical (HOOO ) [6] and ozone (O3 ) [7] are postulated to be formed. We have termed this pathway the antibody-catalyzed water oxidation pathway (ACWOP) [4,8]. Experiments have revealed that, upon activation, this oxidation pathway can be bactericidal to Gram-negative bacteria [7] and we have gained evidence of its occurrence in activated human neutrophils [7,9], during a reversed Arthus inammatory reaction and most recently in human atherosclerotic arteries [10]. When considering the relevance of this pathway in vivo we presume that the 1 O2 * originates from phagocytic cells, such as polymorphonuclear neutrophils, macrophages and eosinophils. However, an alternative but equally well-accepted biological source of 1 O2 * arises from photochemical activation of photosensitizers in the skin, such as occurs in porphyria or in the retinal macular as seen in patients with macular degeneration. Therefore, it was of interest to learn whether RF, itself known to generate 1 O2 * under photochemical activation by visible light

0165-2478/$ see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.imlet.2005.11.020

34

J. Nieva et al. / Immunology Letters 103 (2006) 3338

[11], is capable of activating the water oxidation pathway of antibodies. Herein we show that RF is efcient at activating the water oxidation pathway in antibodies derived from a range of species, and with a mixture of heavy and light chain compositions in a process that is promoted in deuterium oxide (D2 O) and inhibited by sodium azide. The ACWOP can be activated by RF at concentrations of the cofactor that exist under normal conditions in plasma (<50 nM), and upon activation the ROS generated by the antibodies is capable of accelerating hemolysis in vitro. These combined data reinforce the possibility that this pathway may be pathological in vivo. 2. Materials and methods 2.1. Antibodies and fragments

was performed using Prism 4 for Macintosh (Graphpad Software Inc.). 2.3. Photoirradiation of JW38C2 in the presence of D2 O and NaN3 A solution of antibody JW38C2 (murine IgG, 20 M) in PBS (pH 7.4) and RF (60 M) was irradiated on a transilluminator plate (white light, 2.0 mW cm2 Fischer-Biotech transilluminator) at 37 C. Hydrogen peroxide concentration was determined as described above. The assay was also performed in the presence of NaN3 (100 mM) or PBS in D2 O. All experiments were run in at least duplicate and each time point is reported as the mean S.E.M. Data analysis was performed using Prism 4 for Macintosh (Graphpad Software Inc.). 2.4. Hemolysis experiments

Whole antibodies were obtained from PharMingen: S.S.1.2 (mouse IgG2a , ), N.S.8.1.1 (mouse IgG2b , ), N.S.7.1 (mouse IgG3 ). The following murine monoclonal antibodies were obtained in house using hybridoma technology and puried as described previously: JW38C2 (mouse IgG2a , ) [12], DC3 6H10 (mouse IgG2b , ), EP2 19G2 (mouse IgG2b , ) [13], EP2 22B9 (mouse IgG2b , ), EP2 25C10 (mouse IgG2b , ), MT3 38H10 (mouse IgG2b , ), MT3 40C10 (mouse IgG2b , ), MT4 7C5 (mouse IgG2b , ), MT4 7E4 (mouse IgG2b , ), NPN 43C9 (mouse IgG2b , ) [14], PW6 1F5 (mouse IgG2b , ), PW6 2F9 (mouse IgG2b , ), PW6 19A7 (mouse IgG2b , ), PW6 7E9 (mouse IgG2a , ), PW6 8H10 (mouse, IgG). Antibody Fab and Fc fragments were prepared as described previously [3]. AmplexTM Red hydrogen peroxide assay kits (A-12212) were obtained from Molecular Probes. Riboavin and hematoporphyrin IX were obtained from SigmaAldrich and used asreceived. 2.2. Photoirradiation experiments Photosensitizer, riboavin (from a stock in phosphate buffered saline, PBS pH 7.4) or hematoporphyrin IX (from a stock in PBS, pH 7.4), at a range of concentrations was added to a solution of antibody or antibody fragment (20 M in PBS pH 7.4) in a 96-well microtiter plate well (nal reaction volume 200 L). The reaction well was sealed with a glass ball. The protein/sensitizer solution was then irradiated on a transilluminator plate (white light, 2.0 mW cm2 Fischer-Biotech transilluminator) at 37 C. At times throughout the irradiation aliquots were removed and the solution was assayed for HOOH as previously described [3]. In brief, the aliquot (20 L) from the protein solution added into a well of a 96-well microtiter plate (Costar Cat. No. 3603) containing reaction buffer (80 L). Working solution (100 L, 400 M AmplexTM Red reagent and 2 U/mL HRP) was then added and the plate was incubated in the dark for 30 min. The uorescence of the well components was then measured using a Multiwell Plate Reader (EX: 563 nm; EM: 587 nm). The hydrogen peroxide concentration was determined using a standard curve. All experiments were run in at least duplicate and each time point is reported as the mean S.E.M. Data analysis

A 1 mL solution of whole antibody or antibody fragment (20 M) in PBS (pH 7.4), RF (60 M) and sheep red blood cells (RBC, 108 to 109 cells/mL) in a 24-well cell culture plate was irradiated on a transilluminator plate (white light, 2.0 mW cm2 Fischer-Biotech transilluminator) at 37 C. At specic times, aliquots were removed (50 L) and the OD of this solution was measured at 405 nm. Raw OD data were then converted to number of hemolysed cells by interpolation of a standard curve. All experiments were run in at least duplicate and each time point is reported as the mean S.E.M. Data analysis was performed using Prism 4 for Macintosh (Graphpad Software Inc.). 3. Results 3.1. Hydrogen peroxide generation by antibodies during photoirradiation of RF To elucidate whether antibodies are able to utilize the 1 O2 * generated by RF (60 M) upon photoirradiation with visible light to activate the ACWOP we performed irradiation experiments under physiologic conditions (PBS, pH 7.4, 37 C). HOOH production was quantied using the AmplexTM Red assay as previously described [3] (Fig. 1A). For comparison, the HOOH generation by antibodies in the presence of HPIX (60 M) and white light was also measured. The data clearly show that RF is a valid source of 1 O2 * for antibodies and that the photoirradiation of RF in the presence of IgG leads to a time-dependent production in HOOH. The comparison between HPIX and RF reveals that on a mole by mole basis of photosensitizer (60 M) and with the same average light ux (2.0 mW cm2 ) antibody JW38C2 generates HOOH faster with RF (0.646 M/min, r2 = 0.984) than with HPIX (0.381 M/min, r2 = 0.999) (Fig. 1A). The progress curves for HOOH production shown in Fig. 1A are corrected for the initial rate of HOOH generated by irradiation of the sensitizer alone (RF = 0.423 M/min, r2 = 0.998; HPIX = 0.002 M/min, r2 = 0.985). Having established that HOOH is generated by antibody JW38C2 upon photoirradiation with RF it was of interest to explore the scope of this process. Thus, a panel of murine

J. Nieva et al. / Immunology Letters 103 (2006) 3338

35

Fig. 1. Riboavin activates the ACWOP. (A) Progress curves of HOOH production (as measured by the AmplexTM Red method) by murine antibody JW38C2 (20 M) in PBS, pH 7.4 upon photoirradation at 37 C with white light (>400 nm, 2 mW cm2 ) and either RF ( ) or HPIX ( ) (60 M) as the sensitizer. Each point is the mean of at least duplicate measurements S.E.M. and the line of best t are calculated using Prism 4. Please note, no HOOH was generated when these conditions were repeated in the dark. (B) Implicating 1 O2 * . Progress curves of HOOH production (as measured by the AmplexTM Red method) by murine antibody JW38C2 (20 M) upon photoirradation at 37 C with white light (>400 nm, 2 mW cm2 ) of RF (60 M) in either PBS, pH 7.4 ( ) or PBS in D2 O ( ), or PBS, pH 7.4 plus NaN3 (10 mM) ( ). (C) Riboavin at physiologically relevant concentrations. Progress curves of HOOH production (as measured by the AmplexTM Red method) by murine antibody JW38C2 (20 M) in PBS, pH 7.4 upon photoirradation at 37 C with white light (>400 nm, 2 mW cm2 ) with RF at 12.5 nM, 25 nM, 50 nM ( ) and 100 nM ( ). Each point is the mean of at least duplicate measurements S.E.M. (D) Hemolysis by antibody JW38C2. Progress curves of hemolysis of sheep RBCs (3 108 ) as measured by OD 405 nm of aliquots removed during the experiment in PBS, pH 7.4 during photoirradation at 37 C with white light (>400 nm, 2 mW cm2 ). The three groups comprised of RBCs and JW38C2 (20 M) ( ), RBCs and RF (60 M) ( ) and RBCs, RF (60 M) and JW38C2 (20 M) ( ). Please note, no hemolysis was observed when these conditions were repeated in the dark.

monoclonal antibodies, of varying light and heavy chain constitution were tested for their ability to generate HOOH upon exposure to RF (60 M) and white light (2 mW cm2 ) (Fig. 2). The data, summarized in Table 1, reveal that this process is ubiquitous to all antibody compositions. The antibody Fab and Fc fragments also possess the ability to generated HOOH by this process, in-line with our previous observations on the ACWOP in that it occurs with all known immunoglobulin fold proteins thus far tested [4]. The corrected initial rates of HOOH formation varied from 0.218 to 0.998 M/min and are generally higher than the values obtained for when HPIX is the photosensitizer [3], hinting that RF is more efcient than HPIX at the process of 1 O2 * transfer to the IgG for further processing as a substrate in the ACWOP. 3.2. RF-mediated activation of the ACWOP is due to photoproduction of 1 O2 * The intermediacy of 1 O2 * in the RF-photoirradiation activation of the ACWOP was probed by comparing the corrected initial rates for HOOH production by the antibody JW38C2 in PBS, in PBS in D2 O and in PBS in the presence of NaN3 (10 mM). What these experiments show is that the initial rate of HOOH production is fastest in PBS/D2 O and is almost completely

Fig. 2. Initial rate v of all antibodies and fragments tested for RF-induced generation of HOOH. For full experimental details please refer to Section 2.

36

J. Nieva et al. / Immunology Letters 103 (2006) 3338

Table 1 Summary of hydrogen peroxide formation and hemolysis induction by photoirradiation of RF with various antibodies and antibody fragmentsa Entry 1 Clone S.S.1.2 S.S.1.2(Fab) S.S.1.2(Fc) N.S.8.1.1 N.S.8.1.1(Fab) N.S.8.1.1(Fc) N.S.7.1 JW38C2 JW38C2(Fab) JW38C2(Fc) DC36H10 DC36H10(Fab) DC36H10(Fc) EP2 19G2 EP2 22B9 EP2 25C10 MT3 38H10 MT3 43F6 MT3 40C10 MT3 51G9 MT4 7C5 MT4 7E4 NPN 43C9 PW6 1F5 PW6 2F9 PW6 19A7 PW6 7E9 PW6 8D10 PW6 8H10 Source Mouse Isotype 2b , H2 O2 (M/min)b 0.723 0.016 n.d.d n.d. 0.734 0.021 0.762 0.025 n.d. 0.724 0.015 0.998 0.194 n.d. 0.554 0.014 n.d. n.d. 0.549 0.016 n.d. n.d. 0.236 0.040 n.d. 0.430 0.015 0.218 0.023 0.846 0.193 0.431 0.047 0.386 0.016 0.612 0.014 0.678 0.026 n.d. 0.832 0.035 0.671 0.032 Hemolysis (%c ) 79.31 85.61 100.09 103.99 83.97 82.95 44.04 92.56 89.57 100.54 95.17 70.34 94.04 101.16 101.94 90.02 26.97 87.29 103.41 63.29 77.33 93.83 38.33 65.48 74.79 86.99 98.26 84.27 67.28

Mouse

2b ,

3 4

Mouse Mouse

3 , 2a ,

Mouse

2b ,

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
a b

Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse Mouse

2b , 2b , 2b , 2b , 2a , 2b , 2a , 2b , 2b , 2b , 2b , 2b , 2b , 2a , 2b , 2b ,

For in-depth methods please refer to Section 2. Reported as mean S.E.M. of initial rate (analysis of the linear part of the progress curve, r2 > 0.985) measurements (in at least duplicate) corrected for background initial rate of H2 O2 formation by RF alone (0.420 M/min). c Reported as the mean (of at least duplicate measurements) of the theoretical hemolysis maximum (100%) of sheep RBCs after 8 h irradiation with RF (60 M) and antibody or antibody fragment (20 M) corrected for background hemolysis by RF alone (11.4%). d Not determined.

quenched in the presence of NaN3 (Fig. 1B). These data strongly support the role of 1 O2 * in this RF-mediated process. 3.3. The ACWOP is activated at physiologically relevant RF concentrations Riboavin is ubiquitous in body uids and is detectable in plasma at a median concentration of 12 nM with concentrations up to 50 nM detectable in healthy controls. To assess whether the ACWOP could be activated at such low RF concentration, the generation of HOOH by antibody 38C2 was assessed at RF concentrations from 12.5 to 100 nM (Fig. 1C). The rates of HOOH by antibody JW38C2 are easily quantiable; at 50 nM the rate of HOOH formation catalyzed by antibody JW38C2 is 0.003 M/min. Thus, this pathway could well be active in vivo. 3.4. When the ACWOP is activated by RF the ROS generated cause hemolysis To allow a rapid and semi-quantitative assessment of activation of the ACWOP by RF we developed a routine and sen-

sitive assay based around hemolysis of sheep red blood cells (RBCs). Photoirradiation (white light 2.0 mW cm2 ) of RBCs (1019 cells/mL) in the presence of JW38C2 (20 M) and RF (60 M) in PBS, pH 7.4 at 37 C caused hemolysis in a timedependent manner (Fig. 1D). Critically only in the presence of antibody and photoirradiated RF is there signicant hemolysis. Control experiments with photoirradiated RF alone, or with antibody and RF in the dark yielded no signicant hemolysis. The prole of the antibody-mediated hemolysis followed a classic hockey-stick, with a lag-phase of several hours, followed by a rapid rise phase where the vast majority of hemolysis occurs. The length of the lag-phase was critically dependent upon the age and quality of the RBCs used in the assay. Thus, the more aged cells hemolysed with a shorter lag-phase and vice versa. This hockey-stick prole of hemolysis strongly supports the notion that the ROS being generated by the ACWOP is slowly degrading the membrane structure of the RBC, ultimately leading to a catastrophic cell destruction and lysis. Having established the hemolysis assay and shown that one antibody, JW38C2 triggers the hemolysis in a time-dependent manner and whole panel of antibodies and their Fab Fc fragments

J. Nieva et al. / Immunology Letters 103 (2006) 3338

37

were studied in this assay format for their ability to trigger the hemolysis of the sheep RBCs upon photoirradiation with RF. A summary of the data is shown in Table 1. Clearly what the hemolysis experiments show is that, relative to RF alone and regardless of heavy and light chain construction, all antibodies and their fragments generate ROS that contribute to hemolysis. 4. Discussion Since our discovery that antibodies have the intrinsic ability to catalyze the conversion of 1 O2 * into hydrogen peroxide, we have instigated a search for both evidence of its occurrence in complex biological systems and its potential relevance in either immune defense or immunopathology. For this pathway to be active one has to juxtapose immunoglobulins with singlet dioxygen (1 O2 * ). Within a biological setting 1 O2 * can originate from to main sources, activated leukocytes and photosensitization of biological molecules. With this in mind, we have turned out attention to an investigation of antibodies and biological sensitizers such as the isoalloxazine-containing avin cofactors. Riboavin is a water-soluble vitamin that serves as a precursor for avin mononucleotide and FAD [15] and is a well-studied endogenous photosensitizer. Flavin photochemistry is an intensive area for research and was originally promoted because of the proposed involvement of avin excited states in several photobiological and photochemical processes [11]. The RF absorption spectrum in aqueous media exhibits four structureless peaks centered at 446, 475, 265 and 220 nm with high molar extinction coefcients (>104 M1 cm1 ). RF is particularly sensitive to UV and visible light and induces both Type-I (oxygen independent) and critically Type-II (oxygen-dependent) photosensitized oxidation mechanisms. The former involves the formation of free radicals through hydrogen or electron transfer between RF triplet excited state (3 RF* ) and substrates. The Type-II process involves the formation of 1 O2 * by energy transfer from 3 RF* to triplet molecular oxygen (3 O2 ). The photoreactivity of RF has been linked to the lens damage and cataract formation [16] and the deleterious effects of light exposure to nutrient mixtures such as paranteral solutions and infant formulae [17]. Our studies have shown that immunoglobulins and their fragments can intercept the 1 O2 * generated by RF during a Type-II photo-oxidation process to activate the antibody-catalyzed water oxidation pathway (Figs. 1 and 2 and Table 1). The prole of the HOOH generation by antibodies by RF, when compared with HPIX offers an in-sight into the difference between the photochemistry of the two sensitizers. HPIX generates very little superoxide (via Type-I photoxidation) compared to RF. In our experiments, the initial rate v of HOOH formation by HPIX (60 M) was 0.002 M/min as compared to 0.45 M/min by the same concentration of RF. We have shown previously that antibodies can generate >100 equivalents of HOOH without a signicant decrease in the observed rate of HOOH production [4]. Thus, the no-linearity observed in Fig. 1A is presumed to be due to a loss of photosensitizer due to the Type-I photo-oxidation of RF. The intermediacy of 1 O2 * in the HOOH generation by antibodies is always a critical test of whether the ACWOP is

activated or whether this is simply non-specic photoreduction of molecular oxygen by antibody side chains. The rate of HOOH in this RF-photosensitized process is increased in D2 Ocontaining buffers and is quenched in the presence of NaN3 (10 mM) (Fig. 1B). Both of these effects are in-line with the known increase in lifetime of 1 O2 * in deuterated buffers and the quenching of 1 O2 * by azide. The destruction of red blood cells via riboavin photooxidation is a well-studied multifactorial process that is not fully understood [1821]. We decided to utilize this process as a reporter assay for our ACWOP assay and found that all antibodies, regardless of heavy or light chain construction and regardless of whether they were specic for RBC surface markers, accelerate the RF-photoinduced hemolysis (Fig. 1D and Table 1). In most cases the elevation is profound. For comparison, under conditions where the RF-induced hemolysis was 11% of the total sheep RBCs in a samples, the vast majority of Igs had cause between 80 and 100% of the maximal hemolysis. While the relevance of the ACWOP to RF-induced hemolysis in vivo has not at all been investigated, this study hints that Ig, which are present in vivo at >10 mg/mL could contribute in part to the RF-induced hemolysis in the presence of a suitable photoirradiation source. In conclusion, this report shows that the intrinsic catalytic oxidation of water by antibodies can be activated by a known endogenous photosensitizer at concentrations of the sensitizer that are normally observed in vivo. We have shown that this activation is linked to a Type-II photo-oxidation of RF leading to the generation of 1 O2 * that is then intercepted by the Ig. When the ACWOP is activated in the presence of sheep RBCs hemolysis is accelerated relative to RF alone, hinting at a possible pathogenic role for this pathway in vivo. Further studies are being directed at furthering our understanding of this pathway, both in terms of the oxidants it generates, the impact they have in immune defense and their pathological consequences. Acknowledgements The authors would like to thank all members of the Lerner Laboratory, especially Diane Kubitz, for their assistance in antibody production and purication during this project. This work was supported by the Skaggs Institute for Research and the ALSAM foundation. References
[1] Baker H, Frank O, Feingold S, Leevy CM. Vitamin distribution in human plasma proteins. Nature 1967;215:845. [2] Watson CD, Ford HC. High-afnity binding of riboavin and FAD by immunoglobulins from normal human serum. Biochem Int 1988;16:106774. [3] Wentworth AD, Jones LH, Wentworth P, Janda KD, Lerner RA. Antibodies have the intrinsic capacity to destroy antigens. Proc Natl Acad Sci USA 2000;97:109305. [4] Wentworth Jr P, Jones LH, Wentworth AD, Zhu X, Larsen NA, Wilson IA, et al. Antibody catalysis of the oxidation of water. Science 2001;293:18069. [5] Xu X, Muller PR, Goddard III WA. The gas phase reaction of singlet dioxygen with water: a water-catalyzed reaction. Proc Natl Acad Sci USA 2002;99:337681.

38

J. Nieva et al. / Immunology Letters 103 (2006) 3338 [14] Janda KD, Schloeder D, Benkovic SJ, Lerner RA. Induction of an antibody that catalyzes the hydrolysis of an amide bond. Science 1988;241:118891. [15] Cooperman JM, Lopez R. Riboavin. In: Machlin LJ, editor. Handbook of vitamins. New York: Marcel Dekker; 1991. p. 283310. [16] de La Rochette A, Silva E, Birlouez-Aragon I, Mancini M, Edwards AM, Morliere P. Riboavin photodegradation and photosensitizing effects are highly dependent on oxygen and ascorbate concentrations. Photochem Photobiol 2000;72:81520. [17] Lee KH, Jung MY, Kim SY. Effects of ascorbic acid on the light-induced riboavin degradation and color changes in milks. J Agric Food Chem 1998;46:40710. [18] Bereswill S, Fassbinder F, V olzing C, Covacci A, Haas R, Kist M. Hemolytic properties and riboavin synthesis of Helicobacter pylori: cloning and functional characterization of the ribA gene encoding GTPcyclohydrolase II that confers hemolytic activity to Eschericia coli. Med Microbiol Immunol (Berl) 1998;186:17787. [19] Misra RB, Bajpai PK, Joshi PC, Hans RK. An unusual photohaemolytic property of riboavin. Food Chem Toxicol 2001;39:118. [20] Ali L, Naseem L. Hemolyis of human red blood cells by combination of riboavin and aminophylline. Life Sci 2002;70:201322. [21] Ali L, Sakhnini N, Naseem L. Hemolysis of human red blood cells by riboavinCu(II) system: enhancement by azide. Biochemistry (Moscow) 2005;70:10114.

[6] Wentworth Jr P, Wentworth AD, Zhu X, Wilson IA, Janda KD, Eschenmoser A, et al. Evidence for the production of trioxygen species during antibody-catalyzed chemical modications of antigens. Proc Natl Acad Sci USA 2003;100:14903. [7] Wentworth Jr P, McDunn J, Wentworth AD, Takeuchi C, Nieva J, Janda KD, et al. Evidence for antibody-catalyzed ozone formation in bacterial killing and inammation. Science 2002;298:21959. [8] Nieva J, Wentworth P. The antibody-catalyzed water oxidation pathwaya new chemical arm to immune defense? TIBS 2004;29: 2748. [9] Babior BM, Takeuchi C, Ruedi J, Guitierrez A, Wentworth Jr P. Investigating ozone generation by human neutrophils. Proc Natl Acad Sci USA 2003;100:39205. [10] Wentworth Jr P, Nieva J, Takeuchi C, Galve R, Wentworth AD, Dilley RB, et al. Evidence for ozone generation in human atherosclerotic arteries. Science 2003;302:10536. [11] Heelis PF. The photophysical and photochemical properties of avins (isoalloxazines). Chem Soc Rev 1982;11:1551. [12] Wagner J, Lerner RA, Barbas III CF. Efcient aldolase catalytic antibodies that use the enamine mechanism of the natural enzyme. Science 1995;270:1797800. [13] Simeonov A, Matsushita M, Juban EA, Thompson EHZ, Hoffman TZ, Beuscher AE, et al. Blue-uorescent antibodies. Science 2000; 290:30713.