A Reduction-Based Sensor for Acrolein Conjugates with the Inexpensive Nitrobenzene as an Alternative to Monoclonal Antibody

Acrolein, a highly toxic α, β–unsaturated aldehyde, has been a longstanding key biomarker associated with a range of disorders related to oxidative stresses. One of the most promising methods for detecting acrolein involves the use of antibodies that can recognize the acrolein–lysine conjugate, 3-formyl-3, 4-dehydropiperidines (FDP), within oxidatively stressed cells and tissues from various disease states. We have uncovered here that FDP could reduce nitroarenes in high yields at 100 °C in the presence of excess CaCl2 as a Lewis acid promoter. This unique transformation allowed for the development of a de novo method for detecting levels of FDPs generated from proteins in urine or blood serum samples. Thus we successfully converted a non-fluorescent and inexpensive 4-nitrophthalonitrile probe to the corresponding fluorescent aniline, thereby constituting the concept of fluorescent switching. Its sensitivity level (0.84 nmol/mL) is more than that of ELISA assays (3.13 nmol/mL) and is already equally reliable and reproducible at this early stage of development. More importantly, this method is cost effective and simple to operate, requiring only mixing of samples with a kit solution. Our method thus possesses potential as a future alternative to the more costly and operatively encumbered conventional antibody-based methods.


acrolein
3-formyl-3,4-dehydropiperidine (FDP, 1) "oxidative stress marker" pyridinium (2) vasicine NAD(P)H  This idea could make it possible to detect the acrolein-biomarkers easily without the use of the antibody. Herein, we wish to report a convenient and efficient method for detecting the oxidative stress marker FDP.

Results and Discussion
To commence our investigation, we attempted to use the readily accessible N-Bn-FDP (1a) 55 as a reducing reagent against various substrates (Fig. 2). However, for examples containing carbonyl groups or unsaturated bond activated by an electron-withdrawing group (3a-c), reductions did not take place ( Fig. 2A). On the other hand, we discovered that the nitroarene 3d containing the sulfonyl group could be selectively reduced to the corresponding aniline 4d using 1a (Fig. 2B). While the reduction did not proceed under 60 °C (entry 1) initially with only trace amount of aniline 4d was observed in the trial at 80 °C (entry 2), we found that 100 °C was the most optimal temperature for this reduction (entry 3, 45%). To examine the scope of this reduction, we proceeded with several commercially available nitroarenes as substrates. We found that like 3d activated with an electron withdrawing sulfonyl group, nitrobenzene 3e with an ester group could also reduced in 35% yield (entry 4), and that 4-nitroacetophenone 3f and 4-nitrobenzonitrile 3g were reduced in 50% and 73% yield, respectively (entries 5-6). On the other hand, nitrobenzenes 3h and 3i without an electron-withdrawing group were not reduced at all under the optimized conditions (entries [7][8]. To improve the yield, we explored possibility of Brønsted acid or Lewis acid assisted reduction. Given that the NAD(P)H reduction can be promoted using a metal ion as catalyst 53 , we screened for a suitable acid using nitroarene 3d as a model substrate (Fig. 3). Unfortunately, Brønsted acids such as HCl, H 2 SO 4 , CF 3 COOH, and H 3 BO 3 (entry 2-8) led to decreased yields relative to our initial result (entry 1, 45%). Fortuitously, one equiv of transition metal-based Lewis acidic (e.g. Cu 2+ , In 3+ ) additives led to a remarkable increase of the yields (entries 9-19, up to 76%). We particularly pleased to find that focused on alkaline earth metal-based Lewis acid (Mg 2+ , Ca 2+ ), inspired by NADH reduction catalyzed by metal ion 53 , were equally effective (entries 20-25). With 5 equiv of CaCl 2 being the most effective for this reduction at 89% yield (entry 26). Counter anion dependency on reactivity (among MgCl 2 , MgBr 2 , MgSO 4 , Ca(OH) 2 , CaCO 3 and CaCl 2 ) is most probably due to the solubility of metal species in reaction media; While MgCl 2 and CaCl 2 are soluble in either DMF or DMF/H 2 O, other species are hardly soluble or insoluble (see Fig. S1 in Supplementary Information). We also checked pH of the reaction in the presence of these metal species (Fig. S1). Although additives slightly affected solution pH, the observed reactivity in Fig. 3 could be well explained by the solubility of additives. For detecting FDPs in biological samples, i.e., urine (vide infra), very little amount of samples were diluted with excess H 2 O and then subjected to excess nitrobenzene and metal additives. Hence pH of sample solution should not be significantly concerned for our FDP detection.
To develop this reduction into an effective and simple method for FDP detection, we investigated possible reactions of FDP in biological sample with a non-fluorescent nitroarene probe affording a reaction mixture that contains fluorescence of the converted corresponding aniline. Conceptually, we envisioned that the fluorescence intensity should be proportional to the amount of the aniline, which is related with the amount of FDP at a constant rate via the reduction, and thus, the exact amount of FDP in biological sample could be estimated. Most critically for this endeavor, an adequate nitroarene probe is to detect the level of FDP in the presence of various kinds  of biological materials. We focused on two points for selecting an appropriate nitroarene probe: (a) fluorescence property for switching; and (b) efficient reactivity. With these assessments in mind, we proceeded to screen for nitroarenes that could be reduced to fluorescent anilines. Typical fluorescent anilines are classified electronically as being pull-push. That is, the aromatic ring is substituted with both electro-withdrawing and electron-donating groups such as the combination of nitrile and amino group, which is widely known to fluoresce (Fig. 5A) 56 . It is equally important that excitation and emission wavelengths are sufficiently resolved (non-overlapping), and that the quantum yield is sufficiently high for to an accurate measurement. After screening dozens of candidates including 4g and 4o, we found that 4-nitrophthalonitrile 3p to be the most suitable nitroarene probe. As shown in Fig. 5B, while the reduction of 3p is 78% yield and is complete in 5 h, the overlaid fluorescent spectrum of 3p and 4p suggested a clear switch at emission λ max of 404 nm (Fig. 5C). Moreover, time-dependent measurement reveals a distinct correlation between the fluorescence intensity and the reaction progress (Fig. 5D).
To confirm the reduction-based "fluorescent on" concept for FDP, and to verify chemoselectivity of such concept under biological conditions, we tested robustness of FDP and nitroarene probe in the presence of biologically relevant metals and redox reagents (Fig. 6). Addition of air or hydrogen peroxide afforded no oxidation of FDP   1a (Fig. 6A) and, in the presence of biologically abundant or redox metal species (Mg 2+ , Ca 2+ , Fe 2+ , Cu 2+ ), FDP was recovered quantitatively. For the nitroarene probe, 3p did not react with cystein, cystine, glutathione (GSH) or sodium hydrogen sulfide (NaSH) at 1 μ M, which is supposed as biological concentration 57 (Fig. 6B). Therefore, any biologically relevant metal species and redox agents gave little influence on FDPs and 4-nitrophthalonitrile 3p themselves. Hence undesired consumption of FDPs or background fluorescence increase can be avoided under  the established conditions. Namely, Ca 2+ or Mg 2+ selectively activate the reduction in the presence of both substrates. We do not think that very small amount of metal species existing in biological systems, in comparison to excess CaCl 2 used in this research, can efficiently mediate the reaction, but even if so, this is advantageous for our reduction-based sensor.
To further demonstrate the reliability of this new detection method, we demonstrated biological application using N-lys FDP 1b under established conditions as shown above. We established the precise correlation between  the amount of FDP and the fluorescent intensity using authentic N-lys FDP as the standard (Fig. 7A). The delta values between each fluorescence and control are proportional to that of FDP amount (Fig. 7B). The slope of the linear plot in Fig. 7B indicates fluorescent intensity per FDP unit. According to statistical processing (see calculation in Supplementary Information, Fig. S2), detection limit (LOD) is determined to be 0.84 nmol/mL using this method, which is more sensitive than that of ELISA kit (3.13 nmol/mL).
With these validations in hand, we simplified the method by constructing a kit based on the optimized conditions. The procedure of detection is quite simple with three steps: (a) Mixing the sample and the pre-prepared kit solution containing 1.7 mg of nitroarene probe, and 5.5 mg of CaCl 2 in 50 μ L of DMF-H 2 O; (b) heating at 100 °C for 5 h; and c) measuring the resulting fluorescence (Fig. 8A).
We first evaluated a normal blood serum and prepared serum samples that contain artificially generated FDP via pre-treatment of excess acrolein over 0, 1, 20, and 60 days (Fig. 8B). We would measure the amount of FDP estimated by standard values as calculated in Fig. 7B. The rat blood serum normally has 4.9 ± 0.2 nmol/mL of FDP, which is what we find here. In comparison with this control, 1-day treatment of excess acrolein led to a little increase of FDP (7.7 ± 2.6 nmol/mL), but it the level was 12.7 ± 0.7 nmol/mL in the sample after 20-d treatment.
In addition, a 60-d sample shows further increase of FDP level to 18.7 ± 1.4 nmol/mL. Fluorescent aniline 4p, which was obtained by reduction with FDP in rat blood serum (pretreated with aclorein over 60 days, Fig. 8B), was successfully identified by ESI-MS (Fig. 8B). These results are in good agreement with ELISA assay of antibody recognizing FDP.
We then pursued a more practical experiment by using a real urine sample from a 6-week old mouse. After preparation of 20-fold diluted urine sample and dividing the sample into three sample-lots, we attempted the FDP-detection using both our method in comparison with the ELISA protocol. As shown in Fig. 9A, the diluted urine sample showed a level of FDP of 5.2 ± 0.8 nmol/mL, which is in excellent statistical agreement with the ELISA assay 15 .
Finally, we conducted the detection urine samples from 10 mice (Fig. 9B). Although we conducted these experiment in triplicates (totally 30 lots), the entire set of urine test was carried out smoothly and swiftly using the kit solution within 6 h unlike any medical/laboratory services currently in practice. Our results show that levels of FDP in diluted urines from the 10-mice set are within the 3.4-13.1 nmol/mL range. We have thus succeeded in achieving a sensitive and inexpensive detection of FDP from biological samples extracted from mammalian models through the use of an under explored reaction of FDP.

Conclusion
We have uncovered here N-Bn-3-formyl-3,4-dehydropiperidine, an analog of oxidative stress marker FDP, could reduce nitroarene in high yields using CaCl 2 as a Lewis acid promoter. Based on this unique transformation, we developed a de novo detection method for FDP levels via converting a non-fluorescent nitroarene probe to the corresponding fluorescent aniline, thereby constituting the concept of fluorescent switching. This new method is amenable for actual biological samples; its sensitivity level is comparable to that of ELISA assays and is already equally reliable and reproducible at this early stage of development. More importantly, it is easy to handle, practical to operate, and cost effective. Biological samples containing specific nitrobenzenes, i.e., drugs with strongly electron-withdrawing groups, may react with FDP and affect the analysis. While method should be used with cautions, excess 4-nitrophthalonitrile 3p with CaCl 2 is otherwise preferentially reduced, hence could be applied in most cases for detecting oxidatively stressed diseases. Thus, our method possesses potential as a future alternative to the more costly and operatively encumbered conventional methods. Efforts are underway to further develop this novel detection method.

Experimental Methods
General information. All solvents were of reagent grade. All commercially purchased chemicals were used as received 1 . H and 13 C NMR spectra were obtained from a JEOL RESONANCE AL400 NMR and a JEOL RESONANCE AL300 NMR spectrometer. Signals were internally referenced to solvent residues. High-resolution mass spectral analyses were carried out using micrOTOF-Q III-HC TM (BRUKER). All fluorescent measurement was carried out using JASCO FP6500 spectrofluorometer with 96-well flat-bottomed plates from Corning Inc. Each value of amount is calculated from intensity of authentic standards. All procedures involving experiment animals was approved by the Ethics Committee of RIKEN. The experiments were performed in accordance with the institutional and national guidelines. After stirring for several hours at this temperature, the mixture was concentrated in vacuo to give a crude mixture as sticky gum. The crude residue was monitored in NMR or purified by either preparative TLC or silica gel flash column chromatography with the hexane-EtOAc solvent mixture as the eluting system to give the desired aniline product 4d (19.6 mg, 89% yield).

Sample Preparation and Detection
Procedure. Normal rat serum (100 μ L) purchased from Wako Pure Chemical Industry Ltd. was treated with acrolein 100 μ L (> 100-fold equivalent to serum proteins) for a specific number of days (0, 1, 20, or 60 d). After which, the sample was diluted to 1 mL with distilled water. Fresh urine samples (100 μ L) were supplied from C57BL/6 mouse of RIKEN bio-resource center and diluted 20-fold with distilled water. To a solution of nitroarene probe 3k (1.7 mg, 10.0 μ mol) and CaCl 2 (5.5 mg, 50.0 μ mol) in DMF-H 2 O (50 μ L) was added a given urine sample. After stirring for 5 h at 100 °C, the crude reaction mixture was filtered, and the resulting filtrate was measured by spectrofluorometer at 340 nm/404 nm.
Scientific RepoRts | 6:35872 | DOI: 10.1038/srep35872 ELISA Assay. The measurement was conducted with the Enzyme Linked Immunosorbent Assay (ELISA) kit system (TAKARA, Acrolein-Lysine adduct competitive ELISA kit) 15 and followed by attached instructions. The absorbance at 450 nm was measured using micro plate reader (ImmunoMini NJ1000). Data represent averages of more than twice assay with standard deviations from individual experiments.