Evidence for Chemical and Cellular Reactivities of the Formaldehyde

ARTICLE pubs.acs.org/crt

Evidence for Chemical and Cellular Reactivities of the Formaldehyde Releaser Bronopol, Independent of Formaldehyde Release Mustapha Kireche,† Jean-Luc Peiffer,‡ Diane Antonios,§ Isabelle Fabre,‡ Elena Gim
enez-Arnau,† Marc Pallardy,§ Jean-Pierre Lepoittevin,† and Jean-Claude Ourlin*,‡ †

Laboratoire de Dermatochimie, Institut de Chimie de Strasbourg, CNRS and Universit
e de Strasbourg (UMR 7177), 4 rue Blaise Pascal, 67081 Strasbourg, France ‡ AFSSAPS, Unit
e BCM/DLC, 635 rue de la garenne, 34740 Vendargues, France § Universud, INSERM UMR 996, Faculty of Pharmacy, 5 rue JB Cl
ement, 92290 Ch^atenay-Malabry, France ABSTRACT: Formaldehyde and formaldehyde releasers are widely used preservatives and represent an important group of skin sensitizers. Formaldehyde is very often suspected to be the sensitizing agent of formaldehyde-releasers; however, many reported clinical cases of contact allergy to these molecules such as bronopol (2-bromo-2nitropropane-1,3-diol) indicate negative skin reactions to formaldehyde suggesting a more complex mechanism. The aim of this study was to compare the chemical reactivity and biological activity of formaldehyde with those of two formaldehyde releasers: 2-bromo-2nitropropane-1,3-diol and 1,3-dimethylol-5,5-dimethylhydantoin. A key step in the sensitization to chemicals is the formation of the hapten
protein antigenic complex via covalent binding between the chemical sensitizer and amino acids in proteins. The chemical reactivity of the three compounds was thus addressed using 13C NMR analysis of adduct formation upon incubation with a set of nucleophilic amino acids. The biological activity was measured in two in vitro models based on dendritic cells and a monocytic cell line (CD34-DC and THP-1 model) through monitoring of a panel of biomarkers. The results obtained show that 2-bromo-2nitropropane-1,3-diol produces low amount of free formaldehyde in physiological buffers but that its degradation generates various molecules including 2-bromoethanol. In addition, 2-bromo-2-nitropropane-1,3-diol also generates adducts with amino acids, not observed with formaldehyde alone, that could be explained by the reactivity of 2-bromoethanol. In parallel, in a cellular approach using the human monocytic THP-1 cell line, 2-bromo-2-nitropropane-1,3-diol activates THP-1 cells at concentrations that are not correlated to simple formaldehyde release. This observation is confirmed in the more physiological model CD34-DC. Moreover, in the THP-1 model, the expression profiles of several biomarkers are specific to 2-bromo-2-nitropropane-1,3-diol. Finally, the use in the cellular model of the pure degradation products identified by NMR reveals the reactivity of bromonitromethane. In contrast, 1,3dimethylol-5,5-dimethylhydantoin presents chemical and biological reactivities similar to those of formaldehyde. Taken together, these data suggest that 2-bromo-2-nitropropane-1,3-diol is an atypical formaldehyde releaser, releasing low amounts of formaldehyde at physiological conditions but producing multiple degradation products among which 2-bromoethanol and bromonitromethane are potential candidates for explaining the specific allergic reactions to 2-bromo-2-nitropropane-1,3-diol.

’ INTRODUCTION Allergic contact dermatitis (ACD) is a delayed-type hypersensitivity induced by multiple skin contacts with environmental substances. Skin chemical sensitizers compose a wide family of structurally unrelated low molecular weight compounds.1 However, they share two common features: hydrophobicity and electrophilicity. Electrophilic properties of chemical allergens are at the basis of their reactivity against nucleophilic groups leading to protein-haptenization and immunogenicity of the protein
hapten complex. Hydrophobicity favors the penetration of chemical sensitizers in the skin. Apart from its suspected carcinogenic potential, formaldehyde is a well-known sensitizer in humans, commonly included in the top 20 list of most prevalent chemical allergens and therefore included in r 2011 American Chemical Society

the standard battery test. Formaldehyde is classified as a strong sensitizer in the local lymph node assay (LLNA), a regulatory in vivo assay on mice for sensitization prediction. As emphasized above, formaldehyde shares a strong electrophilic property with most chemical allergens. Its high reactivity makes formaldehyde suitable for use as a biocide or a preservative agent (e.g., in cosmetics) as well as a fixing agent in biological preparations.2 Because of its toxic properties, formaldehyde is subject to regulation. In Europe, free formaldehyde must not exceed 0.2% in cosmetics except in products for oral hygiene (maximal concentration: 0.1%). Free formaldehyde is not authorized in Received: June 22, 2011 Published: October 28, 2011 2115

dx.doi.org/10.1021/tx2002542 | Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology

ARTICLE

aerosol cosmetics. Moreover, when the concentration of free formaldehyde exceeds 0.05% in finished cosmetic products, a specific label of “contains formaldehyde” must be added.3 The side effects of formaldehyde have triggered the development and use of the so-called formaldehyde
releasers (FR) which are small organic molecules capable of releasing one or several molecules of formaldehyde. Because of the slow and progressive release of formaldehyde, these molecules are widely used as a surrogate of formaldehyde itself. Many FR are available on the market such as 1,3-dimethylol
5,5-dimethylhydantoin (1, DMDMH), imidazolidinyl urea (IZU), diazolidinyl urea (DZU), 2-bromo-2-nitropropane-1,3-diol, also called bronopol (2, BNP), 5-bromo-5-nitro-1,3-dioxane (BND), quaternium-15 (Q-15), and methenamine (MTN) (for reviews, see refs 4
6). Unfortunately, contact sensitization to FR is also well documented, and their use in cosmetic products is also under restrictions in Europe.3 In particular, clinical evidence and epidemiological studies of contact sensitization to FR suggest that prevalence to formaldehyde allergy is about 2
3% in sensitized European populations, and prevalence to DMDMH, IZU, DZU, BNP, and Q-15 varies around 0.5
2% among allergic patients depending on the studies.4 High percentage of cross-reactivity between formaldehyde and FR (e.g., DMDMH) is a hallmark of this group of allergens.7 Preferential cross-reactivity is also observed between structurally related molecules such as IZU and DZU. In contrast, certain FR exhibit low cross-reactivity such as BNP for which often 2/3 of BNP positive patients do not cross-react with formaldehyde. A recent study from Lundov et al. confirms such observations.8 Indeed, the absence of cross-reactivity in many patients suggests that allergy to FR, and BNP in particular, is probably more complex than the simple release of formaldehyde from FR chemicals. In order to test this hypothesis, we have compared BNP and DMDMH for their respective formaldehyde-releasing potential, chemical reactivity, and in vitro biological activity. DMDMH was selected as a “real” formaldehyde-releaser, based on literature data. As such, it was expected in our assays to present reactivities and biological activities similar to those of formaldehyde alone. First, both 13C-labeled compounds were analyzed for their reactivity against a set of nucleophilic amino acids by 13C NMR compared to that of formaldehyde alone, and the respective reactive degradation products were identified. Then, in vitro biological activities were assessed and also compared to those of formaldehyde alone. Because only BNP showed biological activities unrelated to formaldehyde release, purified BNP degradation products were also tested in the biological assay. Taken together, our data suggest that the BNP sensitization mechanism is not due to the release of formaldehyde alone and propose candidate molecules to explain BNP specific allergic contact dermatitis.

using a reported procedure.9 The same methodology was used for the synthesis of 1,3-[13C]-dimethylol-5,5-dimethylhydantoin ([13C]-1), this time employing dimethyl hydantoin and [13C]-formaldehyde as reagents. BNP (2) was purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). 2-Bromo-2-nitropropane-1,3-[13C]-diol ([13C]-bronopol) ([13C]-2) was synthesized according to a reported procedure for the synthesis of 2 and employing bromonitromethane and [13C]-formaldehyde as reagents.10 Sodium phosphate monobasic and sodium phosphate dibasic, for the preparation of buffers, were purchased from Merck (Darmstadt, Germany). t-Butanol, used as a NMR internal standard, was acquired from Acros Organics (Illkirch, France). GIBCO cell culture medium RPMI 1640, with GlutaMAX I and 10% fetal bovine serum, were purchased from Invitrogen (Cergy Pontoise, France). N-Acetylated amino acids were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France) and were used as delivered. All other chemicals, including bromonitromethane (BNMT), 2-bromoethanol (BRE), and tris-(hydroxymethyl)nitromethane (THNM) used in cell culture experiments, and solvents were also purchased from Sigma-Aldrich.

’ MATERIALS AND METHODS

Reference Compounds: Reaction of [13C]-Formaldehyde with Nucleophilic Amino Acids. To [13C]-formaldehyde (1 μL,

Caution: Skin contact with formaldehyde, DMDMH, and BNP must be avoided. Since these are sensitizing substances, they must be handled carefully. Chemicals. Formaldehyde (37% in aqueous solution, stabilized with 10% methanol) was purchased from Merck (Darmstadt, Germany). For cell culture experiments where methanol is undesired, formaldehyde (prepared from paraformaldehyde, 20% solution in distilled water under sealed vials) was purchased from Electron Microscopy Sciences (EMS, Hatfield, PA). [13C]-Formaldehyde (20% in aqueous solution) and deuterated solvents were obtained from Euriso-Top (Saint Aubin, France). DMDMH (1) was synthesized from dimethyl hydantoin and formaldehyde

Measurement of Formaldehyde Release from DMDMH (1) and BNP (2) by 13C NMR. A calibration curve was at first established with 26 solutions of formaldehyde of known concentrations (from 0 to 0.37 M). The NMR samples were prepared as follows: x μL aqueous solution of formaldehyde at 37% (0 < x < 15 μL), (450
x ) μL of phosphate buffer (0.13 M at pH 7.8) and 150 μL of a solution of t-butanol in deuterated water (0.1 M) as internal standard. Using these proportions, the final buffer solution in the NMR tube was 0.1 M at pH 7.4. One-dimensional 13C NMR spectra of each solution were carried out on a Bruker Avance spectrometer at 75 MHz. Optimal NMR parameters for the acquisition of the 13C signal of formaldehyde were previously adjusted on a Bruker Avance spectrometer at 125 MHz, especially to calculate the relaxation time T1 (6 s). The time between two impulsions was set to 30 s (D1 = 5  T1) and the number of scans to 200. For each of the 26 13C NMR spectra registered, the area of the corresponding formaldehyde signal at 82.0 ppm (hydrated form) was compared to that of the t-butanol signal at 30.3 ppm (internal standard technique) using NMRnotebook software (version 2.0, NMRtec, Illkirch Graffenstaden, France). The calibration curve was obtained by correlating the calculated area of the formaldehyde 13C NMR signal with the known formaldehyde concentration of each solution. Quantification of formaldehyde released by DMDMH (1) and BNP (2), in a phosphate buffer solution (0.1 M at pH 7.4) or in the cell culture medium (RPMI 1640, with GlutaMAX I and 10% fetal bovine serum), was carried out by extrapolation of the area of the 13C NMR signal of released formaldehyde into the calibration curve. For each formaldehyde releaser, three solutions of different concentration were analyzed after 24 h. NMR samples were prepared as follows: x mg of 1 or 2 (0 < x < 40 mg), 450 μL of phosphate buffer (0.13 M at pH 7.8) or 450 μL of cell culture medium, and 150 μL of a solution of t-butanol in deuterated water (0.1 M). The final buffer solution in the NMR tube was 0.1 M at pH 7.4. Onedimensional 13C NMR spectra of each solution were carried out on a Bruker Avance spectrometer at 75 MHz. 7 μmol), dissolved in deuterated water (150 μL), was added a phosphate buffer solution (450 μL, 0.1 M at pH 7.4) containing the amino acid in its N-acetylated form (10 equiv). The final solution (600 μL) was filtered into a NMR tube, and the reaction was followed by 13C NMR. The amino acids tested were N-Ac-Cys, N-Ac-Lys, N-Ac-His, N-Ac-Arg, N-Ac-Trp, N-Ac-Asn, N-Ac-Ser, N-Ac-Tyr, N-Ac-Thr, N-Ac-Met, and N-Ac-Gln.

Reaction of 1,3-[13C]-Dimethylol-5,5-dimethylhydantoin ([ C]-1) and 2-Bromo-2-nitropropane-1,3-[13C]-diol ([13C]-2) with Nucleophilic Amino Acids. As a general procedure, into a 13

solution of [13C]-1 (1.3 μg, 7 μmol) or [13C]-2 (1.4 μg, 7 μmol) in 2116

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology deuterated water (150 μL) was added a phosphate buffer solution (450 μL, 0.1 M at pH 7.4) or the cell culture medium (450 μL), containing the amino acid in its N-acetylated form (10 equiv). The final solution (600 μL) was filtered into a NMR tube. The reaction was followed by 13C NMR. The amino acids tested were N-Ac-Cys, N-Ac-Lys, N-Ac-His, N-Ac-Arg, N-AcTrp (only in the phosphate buffer as it is not soluble in the cell culture medium), N-Ac-Asn, N-Ac-Ser, N-Ac-Tyr, N-Ac-Thr, N-Ac-Met, and N-AcGln. Structural Characterization of the Adducts. The reactivity of [13C]-1 and [13C]-2 with the amino acids tested was followed by onedimensional 13C NMR carried out on a Bruker Avance 300 spectrometer at 75 MHz. In order to elucidate the structures of the products formed after 24 h of reaction, two-dimensional heteronuclear [1H
13C]-NMR experiments were carried out. 1H and 13C NMR data were obtained by heteronuclear single-quantum correlation (HSQC) and heteronuclear multiple-bond correlation (HMBC) experiments, recorded on a Bruker Avance 400 (1H, 400 MHz; 13C, 100 MHz) spectrometer. One and twodimensional spectra were treated using NMRnotebook software (version 2.0, NMRtec, Illkirch Graffenstaden, France). Chemical shifts (δ) are reported in ppm with respect to TMS, acetonitrile being the internal standard (δ 1H = 2.06, δ 13C = 119.7). Structures of the products were assigned using a combination of HSQC and HMBC data, and by comparing the measured chemical shifts with those calculated by ACD/ CNMR and ACD/HNMR Predictor (version 6.0, ACD/Laboratories, Toronto, Canada) and Office Chem Draw Ultra (version 11.0, CambridgeSoft, Cambridge, Massachusetts, United States of America) software.

Generation of Dendritic Cells from Human Cord Blood CD34+ Progenitor Cells. Human umbilical cord blood samples were obtained from Biopredic International (Rennes, France) after full-term delivery from women who were clearly informed about the aim of the study and gave their informed consent. Mononuclear cells were isolated by density centrifugation on a Ficoll gradient (Lymphocyte Separation Medium LSM 1077, PAA, Les Mureaux, France). Cells bearing the CD34+ antigen were isolated from the mononuclear fraction through magnetic positive selection using MiniMacs separation columns (Miltenyi Biotec, Bergish Glabash, Germany) and anti-CD34+ antibodies coated on magnetic beads (Direct CD34 Progenitor Cell Isolation Kit, Miltenyi Biotec, Bergish Glabash, Germany). After purification, the isolated cells were 80
95% CD34+ cells. CD34+ cells were adjusted to the concentration of 1.5  105 cells/ mL and cultured at 37 °C in a humidified 5% CO2 atmosphere in RPMI 1640 Glutamax I medium (Gibco, Invitrogen, Paisley, UK), 10% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin, and 1 mM sodium pyruvate (Gibco Invitrogen), and supplemented with 200 U/mL GMCSF (Abcys SA, Paris, France), 50 U/mL rhTNF-α (R&D Systems, Lille, France), 50 ng/mL Stem Cell Factor (SCF, Abcys SA), and 50 ng/ mL Flt3 Ligand (Flt3-L, Peprotech, Tebu, Le Perray-en Yvelines, France). From day 4 to day 7, cells were diluted 1:2 each day by adding complete RPMI medium. At day 4, the added volume was supplemented with GM-CSF (200 U/mL), rhTNF-α (50 U/mL), and IL-4 (1000 U/mL). At day 7, CD34-DC were treated for 24 h with different concentrations of bronopol (12.5, 25, and 50 μM) or 150 μM of formaldehyde. Cell viability after treatment with the different molecules was never less than 70% as assessed by propidium iodine (10 μg/mL) (Invitrogen, Eugene, OR) THP-1 Cell Culture and Chemical Treatment. THP-1 cells (ATCC ref: TIB-202, American Tissue Culture Collection, Manassas, Virginia, USA) were cultured and maintained at a cell density between 2  105 cells/mL and 1  106 cells/mL in RPMI 1640 with Glutamax I complemented with 10% FCS under classical cell culture protocols and facilities. Frozen stocks were regularly thawed to avoid high passages subcultures. For induction studies, cells were grown for at least 24 h and collected, counted, and seeded in fresh medium at a density of 8  105 cells/mL. Chemicals, extemporaneously solubilized at a 1000 the final concentration in saline or DMSO, were added to fresh medium at

ARTICLE

2 the final concentration with an equal volume of cell suspension. The final concentration of cells and chemicals was therefore 4  105 cells/ mL and 1, respectively. To reduce the cytotoxicity, cells were incubated for 2 h with the tested compounds, washed, and incubated with fresh medium for an additional 4 h (for 6 h samples) or 22 h (for 24 h samples). Cell viability after treatment with the different molecules was never less than 60% as assessed by 7-AAD (7-amino-actinomycin D (Beckman Coulter, Villepinte, France)) staining. Formaldehyde, BRE, BNP, and DMDMH were used dissolved in saline. THNM and BNMT were used dissolved in DMSO. As indicated above, the final concentration of DMSO never exceeded 0.1% in the cell culture assay. At this concentration, there was no difference observed between saline versus DMSO-treated cells whatever the biomaker analyzed. Flow Cytometry. For the CD34-DC model, cell staining was performed using the following mAbs: PE conjugated anti-CD86 (BT7, Diaclone, Besanc-on, France), PE conjugated anti-CD83 (HB15e, BD Biosciences), FITC conjugated anti-HLA-DR (Immu-357, Immunotech), PE conjugated anti-CD40 (MAB89, Immunotech), and PE conjugated anti-CD54 (84H10 Immunotech). Appropriate nonspecific isotype-matched antibodies were used at the same concentrations as controls. Cell labeling procedures were identical to the one used for THP-1. Results were then analyzed using the CellQuest Software (Becton Dickinson, San Jose, CA) based on a collection of 1  104 cells with a FACScalibur flow cytometer (Becton Dickinson). For the THP-1 model, cells were centrifuged and washed with PBS complemented with 1% FCS and 0.1% sodium azide. For labeling, 2.5  105 cells were incubated for 30 min on ice in the presence of 5 μL of the antibody. CD86-PE and the corresponding isotype control were from BD Biosciences (San Jose, CA). 7-AAD was added to each tube (5 μL) for viability determination. Cytofluorometry acquisitions were performed on a FC500 cytometer (Beckman Coulter), and data analysis was performed only on 7-AAD negative cells (viable cells) using CXP software. For both models, results were determined using RFI (relative fluorescence intensity) calculated with the following formula or cMFI (corrected MFI): RFI = (MFI spe
MFI iso) treated/(MFI spe
MFI iso) vehicle, where MFI is the total mean fluorescence intensity of samples labeled with isotype (iso) or antigen-specific antibody (spe).

Messenger RNA Expression Analysis Using Real-Time PCR. After the indicated incubation times (6 and 24 h), 3 to 4  106 cells were collected per sample, and cells were lysed at 4 °C (Cytoplasmic RNA Preparation Protocol, Qiagen, Courtaboeuf France). RNA extraction was performed using a column-based RNA extraction kit (RNeasy Kit, Qiagen, Courtaboeuf France). Total RNA content was measured at 230, 260, and 280 nm by spectrometry for quantification and quality assessment. Then, 1 μg of tRNA was reverse-transcribed using the Superscript II reversetranscription kit (Invitrogen, Paisley, UK). 1/50th of each reverse-transcription was then used for real-time PCR analysis. Real-time PCR was performed on a LightCycler apparatus using the Fast Start DNA Master plus SybrGreen kit (Roche, Mannheim Germany). Expression of GAPDH, HMOX1, IL-8, HSPA6, and IL-1β mRNAS was monitored on each sample run in duplicate. The primer sequences were as follows (forward and reverse primers, respectively): GAPDH: 50 ACTGGCGCTGCCAAGGCTGT30 and 50 GCCCCAGCGTCAAAGGTGGA30 . HMOX1: 50 GCGACAGTTGCTGTAGGGCTTT30 and 50 ATGTGCTTTTCGTTGGGGAAGA30 . IL-8: 50 TGGCTCTCTTGGCAGCCTTC30 and 50 TTCTGTGTTGGCGCAGTGTG30 . HSPA6: 50 TGGAGGCCCATGTCTTCCAT30 and 50 TCTGCCAGCTGGTTGTGCTC30 . IL-1β: 50 AGGGCTGGCAGAAAGGGAAC30 and 50 GGGATTGGCCCTGAAAGGAG30 Of note, GAPDH was used as the control house-keeping gene. Expression of this gene was not affected significantly by all of the 2117

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology Chart 1. Chemical Structures of 1,3-Dimethylol-5,5-dimethylhydantoin (DMDMH) (1), 1,3-[13C]-Dimethylol-5, 5-dimethylhydantoin ([13C]-1), 2-Bromo-2-nitropropane1,3-diol (BNP) (2) and 2-Bromo-2-nitropropane-1,3-[13C]diol ([13C]-Bronopol) ([13C]-2)a

ARTICLE

Table 1. Quantitative Release of Formaldehyde from 1 and 2 initial

concentration of free concentration of free

concentration HCHO in phosphate chemicals of chemical (M)a buffer (M)b

HCHO in culture medium (M)b

HCHOc

0.049

0.049

0.046

HCHO

0.098

0.098

0.075

HCHO

0.146

0.146

0.127

1

0.096

0.060

0.041

1

0.193

0.092

0.046

1 2

0.289 0.091

0.121 NDd

0.053 ND

2

0.182

ND

ND

2

0.273

0.017

ND

a

a

* indicates a 13C labeled position.

Concentration of chemical at time 0 in the phosphate buffer (0.1 M, pH 7.4) or in the cell culture medium RPMI 1640, with GlutaMAX I and 10% fetal bovine serum. b Concentration of free formaldehyde (HCHO) in the solution 24 h after the preparation of the samples. c Control experiment. d Not detected. 13

Figure 1. Calibration curve: correlation of the calculated area of the formaldehyde (HCHO) 13C NMR signal with the concentration of formaldehyde in the 26 solutions or standards. treatments. Therefore, the results are expressed as fold factor calculated by comparing the Cq (Quantification Cycle) values obtained from treated and untreated (vehicle) samples and corrected for GAPDH expression. IL-8 Protein Measurement by ELISA. After 24 h of incubation with chemical compounds, cell culture medium supernatants were collected after centrifugation of the cells. Cell culture supernatants (100 μL) were analyzed in duplicate using the Ready-Set-Go Human IL-8 ELISA kit (EBiosciences, San Diego, CA, USA). The detection limit of the assay is 8 pg/mL. Nontreated samples presented undetectable levels of IL-8 production. Data are expressed as pg/mL of IL-8 protein by interpolation of the calibration curve and correction from noise signal. Statistical Analysis. All experiments were repeated at least three times, with data expressed as mean ( standard deviation (SD). Significance of data were determined using the Student’s t test. Data presenting a p-values of e0.05 were considered as statistically significant.

’ RESULTS Measurement of Formaldehyde Release from DMDMH (1) and BNP (2) by 13C NMR. To measure quantitatively the release

of formaldehyde from DMDMH (1) and BNP (2) (Chart 1),

C NMR was used as a noninvasive technique that did not affect the medium or the equilibrium of formaldehyde release. The concentration of free formaldehyde present in solutions of 1 and 2 was measured by interpolating the area of the 13C NMR signal of detected formaldehyde into a previously established calibration curve (Figure 1). In Table 1 are shown, for each FR, the amounts of released formaldehyde in the case of three solutions of different concentration of each releaser either in a phosphate buffer solution at physiological pH (0.1 M at pH 7.4) or in the cell culture medium (measurements carried out after 24 h). Control of sample pH values was important, as it is known that the release of formaldehyde from different FR can be dependent on the pH. Bronopol (2), for example, is known to release a lot of formaldehyde in alkaline buffer compared with that of acidic buffer.11 A control experiment performed with solutions containing formaldehyde exclusively pointed out that in the culture medium formaldehyde was slightly consumed, probably due to its reaction with some of the ingredients. The cell culture medium is actually constituted by amino acids, peptides such as glutathione, and proteins potentially reactive toward formaldehyde. DMDMH (1) released simply formaldehyde in the phosphate buffer in a concentration dependent manner. In the cell culture medium, the amount of free formaldehyde was quite low compared to that quantified in phosphate buffer. This could be anticipated from the control experiments as released formaldehyde could react with the culture medium ingredients. However, it could be easily seen when comparing results of the control experiments with those of compound 1 that the decrease of the formaldehyde concentration in the culture medium was much higher when it was released from 1. In addition to the probable reaction with the constituents of the medium, one might also hypothesize that the release of formaldehyde from compound 1 was slower in the culture medium, compared to the release in phosphate buffer. In opposition to compound 1, it was not possible to detect the release of formaldehyde from BNP (2) at physiological pH after 24 h in phosphate buffer or in the cell culture medium. Extra measures after 7 and 40 days indicated the presence of formaldehyde at 0.02 and 0.104 M, respectively, for the 0.273 M solution of 2 in phosphate buffer (results not shown). These data indicated that 2 can release formaldehyde

2118

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology but very slowly and not significantly in the time course of the in vitro tests performed. Reaction of 1,3-[13C]-Dimethylol-5,5-dimethylhydantoin [13C]-1 with Nucleophilic Amino Acids. Compounds and adducts observed after 24 h of reaction between [13C]-1 and nucleophilic amino acids are shown in Figure 2. At 24 h, adducts were only observed in the reactions with N-Ac-Cys, N-Ac-Lys, N-Ac-His, and N-Ac-Arg. Structures of these adducts have already been reported in previous studies.12 Differences between reactivity in both the phosphate buffer and the cell culture media are

Figure 2. Compounds and adducts revealed from the reaction of [13C]1 with nucleophilic amino acids and characteristic 1H and 13C NMR data (* indicates a 13C labeled position).

ARTICLE

described here, together with completed NMR data description, allowing the elucidation of structures. Reaction with N-Ac-Cys, in both media afforded an important new signal at 65.8 ppm, correlated via 1J(C, H) coupling (HSQC experiments) to protons at 4.62 ppm and long-range correlated (HMBC experiments) with protons at 3.13 ppm. In parallel, only the [13C]-1 signal at 60.5 ppm disappeared almost completely. The correlation 65.8/4.62 ppm was characteristic of a methylene group in a thiohydroxymethyl structure. The δ of 3.13 ppm was typical of protons of the β-methylene group of N-Ac-Cys. These δ were in agreement with compound [13C]-3 resulting from the reaction of the thiol group of N-Ac-Cys on the electrophilic carbon atom of released formaldehyde, whose precursor was most likely the [13C]-1 hydroxymethyl group at 60.5 ppm. The chemical structure of [13C]-3 was confirmed by comparison of the δ with those of a reference compound made from the reaction of [13C]-formaldehyde with N-Ac-Cys. In the case of the reaction with N-Ac-Lys, formaldehyde released at 82.0 ppm was observed in both media. However, it seemed to have no further reactivity with the amino acid itself. In addition, a signal at 52.6 ppm appeared, being much more resolved in the experiments conducted in the culture medium. It was correlated via 1J(C, H) coupling to protons at 4.29 ppm, and no long-range correlations were identified. The couple 52.6/4.29 ppm corresponded well with [13C]-4. However, doubt remains about this structure. Indeed, [13C]-1 could follow a nucleophilic attack by the ε-NH2 group of N-Ac-Lys on both hydroxymethyl groups. Nevertheless, the observed loss of intensity of the [13C]-1 signal at 62.5 ppm (Figure 3) suggested that this attack should occur on the hydroxymethyl group located in between both [13C]-1 carbonyl chemical functions. In the presence of N-Ac-His and N-Ac-Arg in phosphate buffer, a fast release of formaldehyde was noticed, reactive toward the amino acids to afford adducts [13C]-5 and [13C]-6, respectively. The new 13C NMR signal at 70.5 ppm correlated via 1J(C, H) coupling to protons at 5.30 ppm for the N-Ac-His reaction and at 65.4 ppm equally correlated with protons at 4.63 ppm for the N-Ac-Arg reaction. Chemical structures were corroborated by

Figure 3. Reaction of [13C]-1 with N-Ac-Lys. One-dimensional 13C NMR spectrum after 24 h of reaction in the cell culture medium. 2119

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology Scheme 1. Degradation Products of [13C]-Bronopol ([13C]-2) and Important 1H and 13C NMR Dataa

a

* indicates a 13C labeled position.

comparison of the δ with those of reference compounds made from the reaction of [13C]-formaldehyde with both amino acids. The same results were observed when the reactions were performed in the cell culture medium. Reaction of 2-Bromo-2-nitropropane-1,3-[13C]-diol ([13C]Bronopol) ([13C]-2) with Nucleophilic Amino Acids. As previously shown,12 [13C]-2 underwent degradation in aqueous media giving several products (Scheme 1). Some of these degradation products could be characterized by 13C NMR. [13C]-2 produced 2-bromo-2-nitro-[13C]-ethanol [13C]-7 by releasing one unit of formaldehyde. Then, [13C]-7 could lose the nitro group forming 2-bromo-[13C]-ethanol [13C]-8 or lose still another molecule of formaldehyde to obtain bromonitromethane.13 However, as the resulting bromonitromethane did not contain any 13C label, it could not be identified in the 13C NMR experiments carried out in this work. Finally, [13C]tris-(hydroxymethyl)nitromethane [13C]-9 was the most important compound detected. It is essential to stress here that many of these degradation products were potentially reactive toward the nucleophilic amino acids tested. This was reflected in the diverse adducts formed after 24 h of reaction with [13C]-2 and shown in Figure 4. Interesting reactivity was observed with N-Ac-Cys, N-Ac-Lys, N-Ac-His, and N-Ac-Trp.12 The reactivity of [13C]-2 with N-Ac-Cys showed differences in phosphate buffer and in the culture medium. In phosphate buffer, a strong and diverse reactivity was noticed. The previously described adduct [13C]-3 resulting from the reaction of N-AcCys with released formaldehyde was observed. The new NMR signals at 49.1/4.30 ppm indicated that, rapidly, [13C]-3 followed an intramolecular cyclization to afford [13C]-10. This kind of cyclizations has been already described in the literature for the Nα nucleophilic attack on the hydroxymethyl group.14 However, these two products were minor compared to other adducts formed. The most important 13C NMR signal was at 59.8 ppm and corresponded to [13C]-9. The correlation at 65.4/ 4.28 ppm confirmed the presence of [13C]-7 in the reaction mixture, indicating the degradation of [13C]-2. [13C]-8, another degradation product, was not detected in the spectra. However, signals at 61.8/3.90 ppm corresponded well to a methylene group included in a hydroxyethylthio chemical function, fitting with the structure of [13C]-11. This result could indicate that [13C]-8 was well formed but reacted very quickly with N-Ac-Cys forming [13C]-11 by nucleophilic attack of the thiol group on the carbon atom of [13C]-8 bearing bromine (Scheme 2). A new

ARTICLE

peak at 30.2 ppm appeared, correlated by 1J(C, H) coupling to a proton at 2.92 ppm. It was long-range correlated with protons of the β-methylene group of N-Ac-Cys at 2.94/2.78 ppm and with its Cα proton at 4.82 ppm. Similarly, a peak at 33.6 ppm appeared, correlated by 1J(C, H) coupling to a proton at 3.04 ppm. It was also long-range correlated with protons of the β-methylene group of N-Ac-Cys at 2.94/2.78 ppm and with its Cα proton at 4.80 ppm. These NMR data corresponded well with adducts [13C]-12 and [13C]-13. It is well known that ethylthio derivatives can undergo cyclization and form ethylenesulfonium ion intermediates.15 [13C]-11 could form such an intermediate that after nucleophilic attack by hydroxyl or bromine ion could afford [13C]-12 and [13C]-13, respectively (Scheme 2). This mechanism of going through a sulfonium ion intermediate was confirmed when the experiments were carried out in the cell culture medium. In this case also adduct [13C]-14 was observed with characteristic δ at 33.3/3.02 ppm (Figure 5). Signals of [13C]-13 and [13C]-14 were exactly of the same intensity showing that nucleophilic attack by bromine ion on the ethylene sulfonium intermediate occurred equally on both electrophilic positions, whereas [13C]-11 and [13C]-12 were not of the same intensity as [13C]-11 was obtained from nucleophilic attack of a hydroxyl ion on the ethylene sulfonium intermediate and also from direct attack of N-Ac-Cys on [13C]-8. Interestingly, in the cell culture medium, adducts resulting from the reactivity with released formaldehyde were not observed but only those resulting from the reactivity with [13C]-8. Finally, signals corresponding to the dimer of N-Ac-Cys were also detected indicating the involvement of oxidoreduction processes in the reaction mixture. Some studies reported in the literature suggest that the formation of N-Ac-cystine can be favored by the formation in the reaction mixture of radical anion intermediates derived from bronopol.16 Globally, the same kind of reactivity was observed with N-AcLys. In phosphate buffer degradation products [13C]-7, [13C]-8, and [13C]-9 could be seen in the spectra. Two adducts resulting from the reaction of the ε-NH2 group of N-Ac-Lys with released formaldehyde appeared at 71.5/4.47 ppm, [13C]-15, and at 73.1/ 3.94, 3.14 ppm, [13C]-16. Their structures were elucidated by comparison with reference compounds obtained from the reaction of N-Ac-Lys with [13C]-formaldehyde. But again, these two adducts were minor compared to those obtained by the reaction of the ε-NH2 group of N-Ac-Lys with [13C]-8. According to the computer predictions, the δ of 54.3/3.58 ppm, 58.3/3.87 ppm, and 51.3/3.24 ppm matched well with adducts [13C]-17, [13C]18, and [13C]-19, respectively (Figure 4). Looking at their chemical structures and from a mechanistic point of view, it was not surprising that the ε-NH2 group of N-Ac-Lys could react with the electrophilic sites of [13C]-8, as described for N-Ac-Cys. The formed adducts could then follow an intramolecular cyclization by a second nucleophilic attack of the amino group forming an electrophilic intermediate of the aziridinium kind this time.17 Nucleophilic attack on this intermediate by bromine or hydroxyl anions could explain the production of the other adducts detected. In the cell culture medium, the release of formaldehyde was slightly observed, but no adducts resulting from its interaction with N-AcLys were noticed. Only degradation products of [13C]-bronopol and adducts [13C]-17, [13C]-18, and [13C]-19 were identified. With N-Ac-His and N-Ac-Trp, except for [13C]-7 and [13C]-9, we only saw little signals for adducts resulting from the reaction of the lateral chain of the amino acid with released formaldehyde, at 70.5/5.30 ppm in the case of N-Ac-His ([13C]-5) and at 68.6/ 5.48 ppm in the case of of N-Ac-Trp ([13C]-20). 2120

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology

ARTICLE

Figure 4. Compounds and adducts from the reaction of [13C]-2 with nucleophilic amino acids and characteristic 1H and 13C NMR data (* indicates a 13 C labeled position).

Activation of THP-1 Cells in Vitro by Formaldehyde, DMDMH (1) and BNP (2). The THP-1 cell line has been

proposed by several authors as a surrogate to dendritic cells for the in vitro identification of chemical sensitizers. The capacity of this cell line to detect chemical sensitizers was recently evaluated by Ashikaga et al.18 We compared the capacity of formaldehyde and FRs to activate THP-1 cells. Classically, THP-1 activation is measured using CD86 surface expression at low toxic concentrations of tested chemicals. Recently, the monitoring of IL-8 secretion in cell culture supernatants has also been proposed as a complementary biomarker.19 In Figure 6, the results obtained for two concentrations of formaldehyde and DMDMH (1) and three concentrations of BNP (2), selected to induce similar low toxicity, are presented (Figure 6A). In these conditions, all three compounds induced a concentration-dependent overexpression of CD86 above the threshold of 1.5 defined by Ashikaga et al. (Figure 6B). Moreover, a similar response of the cells was observed when the secretion of IL-8 in the cell culture medium was assessed (Figure 6C). Interestingly, DMDMH induced cellular activation at concentrations similar to those of formaldehyde. In contrast, BNP induced cellular activation at concentrations 3 to 4 times lower compared to that of

formaldehyde suggesting that THP-1 activation was not due to only the formaldehyde release of BNP. As one molecule of BNP can release a maximum of two molecules of formaldehyde, the activation potential of BNP was not explained by formaldehyde release. Phenotypical Changes by Formaldehyde and BNP (2) in CD34-DC. In CD34-DC, several contact sensitizers such as NiSO4, CoCl2, 1-chloro-2,4-dinitrobenzene (DNCB), cinnamaldehyde, and 1,4-phenylenediamine (pPD) have been previously described to induce different expression of CD86, CD83, HLADR, and CD40.20
22 In this work, we focused our investigations on CD34-DC marker surface expression in response to formaldehyde and BNP (2). CD34-DCs were differentiated for 7 days using GM-CSF, TNF-α, SCF, and Flt3-L. IL-4 was added at day 4 to reduce the number of residual CD14+ cells (data not shown). At day 7, CD34-DCs were incubated for an additional 24 h period in the presence or in the absence of the chemicals, and the expressions of CD86, CD40, CD54, HLA-DR, and CD83 were evaluated. The highest chemical concentration used was selected based on cell viability above 70%. Expression of each surface marker was expressed as the percentage of positive cells and the fold induction of the RFI. 2121

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology Results showed that the expression of CD86 was increased in CD34-DC treated with formaldehyde. BNP induced a concentration dependent expression of CD86 (Table 2). Both the RFI and the percentage of positive cells were modified; however, only RFI reached statistical significance. Moreover, BNP and formaldehyde were able to induce a slight upregulation of CD40 at their highest concentration. Concerning CD54, CD83, and HLA-DR, neither BNP nor formaldehyde induced a notable effect on the expression of these markers. Indeed, these data confirmed in a more physiological model the original observations made on THP-1 and excluded a specific mechanism related to the cell line. However, because of the limited availability of CD34+ progenitor cells, the following experiments were performed only on THP-1 cells. Gene Expression Profiles induced by Formaldehyde, DMDMH (1) and BNP (2) in THP-1 Cells. To compare the biological effects of the three selected molecules, the expression of several genes, currently monitored in our laboratory as Scheme 2. Hypothetical Mechanism for the Formation of Adducts [13C]-11, [13C]-12, [13C]-13, and [13C]-14 via a Sulfonium Intermediate

ARTICLE

potential biomarkers of sensitization, was assessed in the THP-1 model. Our hypothesis was based on the assumption that if BNP and DMDMH mechanisms followed formaldehyde release, their biological effects on THP-1 should be similar to those induced by formaldehyde alone. Four genes (hmox-1, Il-8, hspa6, and Il-1β) were examined after 6 or 24 h of incubation with the tested compounds. As shown in Figure 7, BNP induced an early and specific response with HMOX1, IL-8, and HSPA6 (Figure 7A
C). At this time point, formaldehyde and DMDMH did not induced gene expression. After 24 h, the IL-1β gene was induced similarly by all three molecules, confirming that the responses observed at 6 h were specific to BNP. Taken together, these results suggested that BNP exerted at low concentration biological effects that were distinct from those induced by formaldehyde alone. In contrast, DMDMH induced an activation profile undistinguishable from that of formaldehyde alone. Activation of THP-1 Cells in Vitro by Degradation Products of BNP (2). We stated the hypothesis that the biological effects of BNP were the consequence of the production of its degradation products. As described in the results relative to 13C NMR data and in Scheme 1, three degradation products have been identified: 2-bromoethanol (BRE), 2-bromo-2-nitro-ethanol, and tris(hydroxymethyl)nitromethane (THNM). A fourth degradation product, bromonitromethane (BNMT), could not be observed in this study but was observed by others.13 Three of these molecules were acquired commercially and tested for their capacity to activate the THP-1 cell line. As shown in Figure 8, BRE did not induce any toxicity or CD86 overexpression even at high concentrations. THNM showed a moderate activation potential at concentrations over 500 μM, which was not compatible with those of its parent compound BNP. In contrast, BNMT was capable of inducing a strong overexpression of CD86 at very low concentrations (10
15 μM). These data suggested that only BNMT could mimic the biological reactivity of BNP in such a cellular model. BNMT Induction of a Gene Expression Profile Similar to That of BNP (2) in the THP-1 Model. As only BNMT showed a potent biological activity, we then hypothesized that if BNMT

Figure 5. Reaction of [13C]-2 with N-Ac-Cys. One-dimensional 13C NMR spectrum after 24 h of reaction in the cell culture medium. 2122

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology

ARTICLE

Figure 6. Activation of THP-1 cells by formaldehyde, BNP, and DMDMH. THP-1 cells were incubated for 24 h with tested compounds as described in Materials and Methods. Cells were collected and analyzed by flow cytometry for cell mortality (A) and CD86 expression (B). In parallel, cell culture medium was collected and analyzed by ELISA for IL-8 production (C). For all graphs, data are expressed as the mean ( standard deviation calculated from at least three independent experiments. Calculations of data are described in Materials and Methods. Nontreated samples exhibited basal cell mortality values (3.28 ( 0.06) and undetectable expression levels of IL-8 protein. Of note, in graph B (CD86) the shaded bar indicates the threshold of specific activation as defined in ref 18. Concentrations are in micromolar. *, p < 0.05; **, p < 0.01.

was responsible for the biological activities of BNP, it should induce a similar gene expression profile compared to that of its parent compound in our model. As shown in Table 3, at a concentration where all tested chemicals induced a similar CD86 protein overexpression, only BNMT and BNP induced significantly the same three genes (hmox-1, IL-8, and HSPA6), whereas formaldehyde and DMDMH did not. Moreover, the amplitude of the regulation was similar for both BNMT and BNP. These data also suggested that BNMT and BNP share similar biological activities, making BNMT a potential reactive metabolite of BNP.

’ DISCUSSION The aim of this study was to compare the chemical and biological reactivity of two unrelated formaldehyde
releasers: DMDMH (1)

and BNP (2). The rationale for this work was that based on clinical observations, allergic reactions to BNP (2) are probably not explained by formaldehyde release only. For that purpose, the reactivity of 1,3-[13C]-dimethylol-5,5-dimethylhydantoin ([13C]-1) and 2-bromo-2-nitropropane-1,3-[13C]-diol ([13C]-bronopol) ([13C]-2) with nucleophilic amino acids was studied. [13C]-1 was chosen as a “real” formaldehyde releaser, as shown in the measurement of formaldehyde release by 13C NMR.23 As such, it was expected that its reactivity with nucleophilic amino acids was going to be dominated by the reactivity of the released formaldehyde itself, which is a strong electrophile. 13 C NMR follow-up of the reactivity after 24 h of reaction confirmed this hypothesis when experiments were conducted in phosphate buffer at physiological pH and also in the cell culture medium used in this study for cellular assays. N-Ac-Cys, 2123

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology

ARTICLE

Table 2. Phenotypical Changes Induced by Formaldehyde and BNP in CD34-DCa CD86 RFI

CD40 %

RFI

CD54 %

RFI

HLA-DR %

RFI

CD83 %

RFI

%

control

1(0

10.4 ( 4

1(0

72.8 ( 20

1(0

74.4 ( 27.4

1(0

88.9 ( 10.2

1(0

HCHO (150 μM)

1.5 ( 0.6b

27.6 ( 23.2

1.3 ( 0.7

73.7 ( 16.2

0.8 ( 0.2

70.9 ( 27.1

0.9 ( 0.4

86.5 ( 6.9

1.1 ( 0.2

7.1 ( 8.8 8.1 ( 4.4

BNP (12.5 μM)

1.1 ( 0.1

13.9 ( 8.8

1 ( 0.1

73.2 ( 23.6

1.1 ( 0.2

72.6 ( 31.3

1.1 ( 0.1

89.8 ( 8.5

1 ( 0.1

5.6 ( 5.9

BNP (25 μM)

1.1 + 0.3

19.3 ( 14

1 ( 0.1

71.4 ( 21

0.9 ( 0

71.7 ( 30.3

1.3 ( 0.1

89.4 ( 9.1

1 ( 0.1

6.5 ( 5.6

BNP (50 μM)

1.3 ( 0.2b

23.9 ( 11

1.3 ( 0.4

73.9 ( 17.1

1.1 ( 0.3

75.2 ( 23.1

1 ( 0.2

88.2 ( 7.9

1 ( 0.2

5.8 ( 3.6

a

CD34-DCs were generated from hematopoietic progenitor cells obtained from human cord blood in the presence of GM-CSF, TNF-a, SCF, Flt-3L, and IL-4 for 7 days as described in the Materials and Methods section. Cells were then washed and stimulated or not (control) with BNP or formaldehyde (HCHO) for 24 h. Expression of each surface marker was expressed as the percentage of positive cells and the RFI. Results are shown as the mean ( SD of 3 independent experiments. b p < 0.05.

N-Ac-His, and N-Ac-Arg were reactive toward formaldehyde released from [13C]-1 forming adducts containing on the lateral chain of the amino acid a hydroxymethyl unit. Only the reaction with N-Ac-Lys gave an adduct ([13C]-4) resulting from the direct reaction of the ε-NH2 group of the lateral chain with a hydroxymethyl group of [13C]-1. However, this adduct was obtained in negligible amount. In conclusion, both on formaldehyde release measurements and the adduct formation study, [13C]-1 properties were undistinguishable from formaldehyde itself. On the contrary, 13C NMR experiments to measure the release of formaldehyde from 2-bromo-2-nitropropane-1,3-[13C]-diol ([13C]-bronopol) ([13C]-2) clearly were not able to evidence free formaldehyde after 24 h in aqueous solutions. Moreover, the 13 C NMR follow-up of the reactivity of [13C]-2 with several nucleophilic amino acids produced results suggesting a more complex degradation scheme. Various degradation products have been evidenced which can be divided in two groups: formaldehyde and nonformaldehyde. The formaldehyde release by BNP is suggested by (i) the detection of very low but detectable amounts of free formaldehyde in phosphate buffer at high BNP concentrations; (ii) the appearance of degradation products such as 2-bromo-2-nitro-[13C]-ethanol ([13C]-7) and [13C]tris-(hydroxymethyl)nitromethane ([13C]-9) only explained by an initial formaldehyde release (see Scheme 1); and (iii) the detection of formaldehyde specific adducts when [13C]-2 was incubated with nucleophilic amino acids in phosphate buffer. Interestingly, the absence of free formaldehyde observed in aqueous solutions of BNP can be retrospectively explained by the immediate reactivity of released formaldehyde with BNP itself leading to the major degradation product tris(hydroxymethyl)nitromethane. Nonformaldehyde reactive degradation products have also been evidenced by (i) the identification of three compounds generated in aqueous solution in the absence of nucleophilic amino acids (2-bromo-2-nitro-[13C]ethanol ([13C]-7), 2-bromo-[13C]-ethanol ([13C]-8), and [13C]-tris-(hydroxymethyl)nitromethane ([13C]-9)); and (ii) the detection of BNP specific adducts in phosphate buffer or cell culture medium only explained by the reactivity of its degradation products. Of note, one of these degradation products, 2-bromo-[13C]-ethanol ([13C]-8), was particularly reactive toward amino acids such as N-Ac-Cys and N-Ac-Lys. Many of the characterized adducts were issued from the reaction of [13C]-8 with the lateral chains containing a thiol or an amino group. Moreover, the abundance of BNP specific adducts was much higher than formaldehyde specific adducts suggesting that free formaldehyde is not predominant over the various

degradation products derived from BNP. Finally, it is noteworthy to observe that differences in adduct formation between phosphate buffer and culture medium for BNP but not DMDMH suggest an equilibrium between the different degradation pathways, which is affected by external conditions. As potential consequences, one could make the assumption that in vivo degradation of BNP may vary between individuals. Depending on skin properties (pH, microflora, metabolism, ...), each individual may generate variable amounts of BNP metabolites. For example, an individual generating high formaldehyde from BNP may present cross-reacting symptoms of ACD, while another individual may generate more alternative metabolites (such as the ones identified here) and present little cross-reactivity between BNP and formaldehyde in patch-testing. The biological effects of the three molecules tested in vitro were in agreement with the observations made at the physicochemical level. Two cellular models were used in this study. The first was the CD34-DC model based on the in vitro generation of DC-like cells through differentiation of cord blood derived CD34+ progenitors. The phenotypic changes observed in CD34-DC in the presence of chemical sensitizers have already been described by our laboratory19
21 as well as by others (as reviewed in ref 24). The second model was based on the THP-1 cell line. This model has been developed as a potential regulatory method for the prediction of the sensitization potential of chemical. This method with two others is presently under prevalidation phase at ECVAM (European Center for the Validation of Alternative Methods). In this approach, the overexpression of CD86 or CD54 above a threshold defined as 1.5 for CD86 and 2 for CD54 is considered as a specific response to a chemical sensitizer.18 Because, formaldehyde, BNP, and DMDMH induced CD86 well above this threshold, we did not monitor CD54 but included IL-8, a novel promising biomarker, measured in a supernatant using an ELISA method.19 Indeed, the combination of these two complementary biomarkers allowed us to define a range of concentrations activating the cells for each tested compounds. In these conditions, BNP induced a cellular activation monitored with CD86 expression and IL-8 production at concentrations well below the one required for formaldehyde alone or DMDMH to induce a similar response. These results strongly suggested that the observed BNP effects are not due solely to formaldehyde release. These phenotypical modifications observations were confirmed using the CD34-DC model, excluding the fact that the observed effects were limited to the THP-1 cell line. To explore the specific biological effects of BNP, we measured the expression of a set of genetic biomarkers by real-time PCR. At the 2124

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology

ARTICLE

Figure 7. Gene expression profiles induced by formaldehyde, BNP, and DMDMH in THP-1 cells. THP-1 cells were incubated for 6 h (A
C) or 24 h (D) with tested compounds. Then, cells were collected, and mRNAs were extracted, reverse-transcribed, and analyzed by Q-PCR as indicated in Materials and Methods. Early expression genes such as HMOX1 (A), IL-8 (B), and HSPA6 (C) were monitored from 6 h samples. Late expression gene such as IL-1β (D) was monitored on 24 h samples. For all graphs, data are expressed as fold factor by comparison with nontreated samples and corrected from the expression of the house-keeping gene GAPDH. Data are presented as the mean ( standard deviation calculated from at least three independent experiments. Concentrations are in micromolar. *, p < 0.05; **, p < 0.01.

same concentrations used to activate the cells, BNP induced specifically a set of early genes (hmox1, Il-8, and hspa6). In

contrast, formaldehyde and DMDMH induced only late expression genes such as Il-1β. The strong differences observed 2125

dx.doi.org/10.1021/tx2002542 |Chem. Res. Toxicol. 2011, 24, 2115–2128

Chemical Research in Toxicology

ARTICLE

Figure 8. Activation of THP-1 cells by degradation products of BNP. THP-1 cells were incubated for 24 h with BNP or its degradation products: bromonitromethane (BNMT), 2-bromoethanol (BRE), and tris-(hydroxymethyl)nitromethane (THNM) as described in Materials and Methods. Cells were collected and analyzed by flow cytometry for cell mortality (A) and CD86 expression (B). For all graphs, data are expressed as the mean ( standard deviation calculated from at least three independent experiments. Calculations of data are described in Materials and Methods. Nontreated samples exhibited basal cell mortality values (3.28 ( 0.06). Of note, in graph B (CD86) the shaded bar indicates the threshold of specific activation as defined in ref 18. Concentrations are in micromolar. *. p < 0.05; **, p < 0.01.

Table 3. Comparison of Gene Expression Profile Induced by BNMT, BNP, and DMDMHa BIOMARKERS

CD86 (FCM)

HMOX-1 (Q-PCR)

IL-8 (Q-PCR)

HSPA6 (Q-PCR)

b

DNCB

7.95 ( 2.78

186 ( 115.5

6.44 ( 0.84

151.6 ( 107.04

HCHO

3.36 ( 1.71

1.7 ( 0.23

1.26 ( 0.23

0.92 ( 0.32

DMDMH

3.24 ( 0.65

1.2 ( 0.21

0.9 ( 0.29

0.77 ( 0.25

BNP

3.55 ( 1.03

18 ( 4.10

10.23 ( 5.35

652.70 ( 152.37

BNMT

2.68 ( 0.52

13.76 ( 3.56

10.04 ( 4.04

792.83 ( 186.83

a

THP-1 cells were treated with various compounds and analyzed for CD86 expression (FCM) and gene expression (Q-PCR) as described in the Materials and Methods section. Data are expressed as the mean of RFI (FCM) or fold factor (Q-PCR) ( SD derived from 3 independent experiments. Data in bold font exhibited a p-value