¹H-NMR chemical shifts are influenced by the π-electron density of the attached carbon. Ha represents the proton trans to the substituent, and Hb the proton cis to that (Figure 1). There was a good correlation between δHa and δC_β for monomers [18]. The ¹H- and ¹³C-NMR chemical shift data for nine (meth)acrylates taken from the literature [18] and their biological activity (hemolytic activity, in vivo toxicity), physicochemical parameters and theoretical parameters, also taken from the literature [1,3,15], are shown in Tables 1, 2 and 3, respectively. We investigated QSARs between 1/H₅₀ and log P or NMR chemical shifts (δHa, δC_β). The QSAR for 9 monomers (MA, EA, nPA, nBA, IBA, MMA, EMA, nPMA, and nBMA) yielded good results for log P (r² = 0.95) (QSAR 1), whereas there was no QSAR for δHa. Similarly, a good linear QSAR between log 1/H₅₀ and heat of formation (HF) for the same data set was obtained (r² = 0.86) (QSAR 2). Since there was a good QSPR between log P and HF for monomers (r² = 0.86), the entropic and enthalpic factors could affect the overall log P value. Dillingham et al. [3] previously reported a good QSAR between the hemolytic activity and log P for (meth)acrylates, and also a relationship between log P and molar refraction (MR) (r² = 0.70) or molecular volume (Vm) (r² = 0.92). We investigated a relation between Vm and HF (Table 3) for 9 (meth)acrylates and it was found that the good linear relationship was obtained at r² = 0.97; as HF increased, Vm declined. Putz et al. [16] described Hansch physico-chemical parameters as follows: (1) hydrophobicity (log P), corresponding to trans-cellular membrane diffusion and with translation motion of the molecules; (2) polarizability, accounts for the dipole perturbation and ionic interaction; and (3) optimal total energy (Etot), which contains steric information about the molecule’s 3D structure since it is given by the equilibrium conformation. We previously investigated a relationship between log P and van der Waals (VDW) area, dipole moment (μ) or HOMO energy for seven methacrylates monomers that were calculated using the PM3 method and it was found that two molecular parameters, VDW area and HOMO energy, particularly the former, contributed significantly to the variation of log P values [15]. Thus, log 1/H₅₀ could be related to HF (QSAR 2).

On the other hand, a linear relationship between log 1/H₅₀ and δC_β was obtained for separations of acrylates and methacrylates, particularly the latter (r² = 0.87). For acrylates, there was a good linear QSAR between log 1/H₅₀ and HF (r² = 0.96) or log P (r² = 0.95), but not δC_β (r² = 0.46). Thus, the hemolytic mechanism may differ between acrylates and methacrylates.

Acrylates and methacrylates can be classified together as a single group of unspecifically reactive chemicals [21], or as two groups: the acrylates as electrophiles and the methacrylates as ester narcotics [22]. In general, QSARs for acrylates and methacrylates are based on the assumption that their monomers with the same functional group (e.g., R: the alcohol moiety in Figure 1) have the same mode of action. However, the classes of these chemicals may not be identical to each other because the metabolic activity or conjugation with GSH differs between acrylates and methacrylates; a difference in GSH reactivity (k_GSH) has been observed between readily reactive acrylates and more slowly reactive methacrylates [5,11,20]. We previously investigated the relationship between k_GSH values using reported data and δC_β values for 12 acrylates and methacrylates and it was found that a good relationship was obtained [19]. The unsaturated β-carbon atom in (meth)acrylate molecules is the most probably site of attack in the Michael addition [11]. As shown in QSPR 1, in this work there was a significant relationship between GSH rate constants and δC_β for (meth)acrylates. McCarthy et al. [20] investigated depression of the erythrocyte GSH by acrylates and methacrylates in vitro, and found that alkylation of the erythrocyte membrane may be result from interaction between erythrocytes and the reactive acrylates, MA and EA, and also that there may be a process that can reduce the effective intracellular acrylate concentration, consequently leading to a decrease of cellular GSH depression in the erythrocyte system. Koleva et al. [23] reported that α,β-unsaturated carbonyl compounds such as acrylates are common environmental pollutants that are able to interact with proteins, enzymes, and DNA through various mechanisms. A common mechanism of action (Michael-type addition) may not be responsible for the hemolytic activity of acrylates because there was no QSAR between log 1/H₅₀ and δC_β for these monomers. Therefore we investigate the hemolytic mechanism of acrylates and methacrylates using DPPC liposomes as a model of erythrocyte membranes. The results are described in Section 2.3.

Next, we investigated the QSARs between ipLD₅₀ and δHa or δC_β for (meth)acrylates, and the results are shown in Table 4. A significant linear QSAR was obtained for δHa or δC_β (QSARs 3 and 4; in both cases, r² = 0.78). An increase of δHa or δC_β enhanced the in vivo toxicity. We found good QSPRs between δC_β and the χ-, η- or ω-term (in three cases, r² = 0.99). As expected, there were also good QSARs between ipLD₅₀ and the χ-, η- or ω- term for methacrylates (QSARs 7, 8 and 9; r² = 0.7 − 0.8). Furthermore, there was a parabolic, not linear, QSAR (QSAR 6) for log P for (meth)acrylates. This finding was similar to that reported previously [1]. The QSAR in terms of both δC_β and log P yielded a better result (QSAR 5, r² = 0.94).

Lawrence et al. [1] also previously reported a good QSAR between ipLD₅₀ (mice) and the σ charge on the carbonyl carbon, Q^σ(C) for (meth)acrylates using the Hansch model; Q^σ(C) was associated with high toxicity [9]. They obtained Q^σ(C), using the method of del Re [24] employing parameters that reproduce dipole moments. We also investigated the QSPR between δC_β and Q^σ(C) for selected (meth)acrylates and obtained a good linear QSPR (r² = 0.99). As shown for QSAR 7, the σ charge of monomer molecules was considered to play a role in the toxicity, since ester hydrolysis and nucleophilic attack are affected by the σ charge [1,11].

According to the QSARs 3 and 4, (meth)acrylates with a large δHa (δC_β) value should have potent toxicity. Talalay et al. [25] previously reported that MA, acrylonitrile and acrolein were more highly active inducers of QR (NAD(P)H: (quinone-acceptor) oxidoreductase) in Hepa 1clc7 cells in comparison to MMA. MMA and acrylamide are inactive intracellular inducers of QR. The relationship between QR reactivity, GSH reactivity or in vivo toxicity and the NMR chemical shifts of the β-carbon for these active acrylates are summarized in Table 5.

The δHa (δC_β) (ppm) declined in the order acrolein > acrylonitrile > MA > acrylamide > MMA. MA, acrylamide and acrolein, which are potent QR inducers, showed a large NMR chemical shift value of the β-carbon, compared to that of acrylamide; these compounds are known to be major intracellular reducing agents, and scavengers of reactive oxygen species (ROS) generated in various cellular processes [29]. Electrophilic xenobiotics such as vinyl monomers become conjugated to GSH and decrease its level within the cell [5,30]. When cellular GSH is exhausted, unscavenged ROS accumulate in cells, thus exerting toxic effects [31]. Ishikawa et al. [32] previously showed that MMA upregulates the expression of genes encoding phase II enzymes such as glutathione S-transferase and quinone oxidoreductase (NAD(P)H) in L929 cells. However, from the present result based on the NMR chemical shifts, it was assumed that such reactivity of MMA would be markedly lower than that of active acrylates such as acrolein, acrylonitrile and MA. Acrolein [33] and acrylonitrile [34] are well-known carcinogens. McCarthy et al. [20] reported that the carcinogenetic mechanism of acrylates may be related to alkylation of protein thiols involved in tumor promotion. Oxidative stress caused by these compounds, and the resulting oxidative damage, induce apoptosis and are involved in carcinogenesis. We predicted the GSH reactivity (k_app) under cell-free conditions, and found that k_app declined in the order acrolein > acrylonitrile > MA > acrylamide > MMA. This strong decrease of GSH in hepatocytes for acrolein and acrylonitrile has been reported previously [35], and is supported by the predicted k_app data for these monomers. Also, the predicted ipLD₅₀ value declined in the order acrolein > acrylonitrile > MA > acrylamide > MMA. Both acrolein and acrylonitrile were most toxic, as supported by reported oral-LD₅₀ data. The induction of QR activity and GSH reactivity for active acrylates was dependent on the NMR chemical shifts of the β-carbon. On the other hand, although acrylamide was an inactive QR inducer and its predicted toxicity was relatively low, it showed toxicity under experimental conditions (Table 5). Biotransformation of acrylamide is thought to occur through glutathione conjugation and decarboxylation, with the formation of toxic glycinamide [36].

Induction of phase II enzymes and elevation of the glutathione level are well known to be involved in the toxic and carcinogenic effects of electrophiles and reactive forms of oxygen.

The following chemicals and reagents were obtained from the indicated sources. Methyl acrylate (MA), ethyl acrylate (EA), methyl methacrylate (MMA), ethyl methacrylate (EMA) and n-butyl acrylate (nBA)(Tokyo Chemical Industry Co., Ltd., Tokyo, Japan); l-α-dipalmitoylphosphatidylcholine (DPPC)(Sigma Chemical Co., USA); deuterium oxide, 3-(trimethylsilyl)propionic acid sodium salt (TMSPA)(Merck, Darmstadt, Germany).

The ¹H- and ¹³C-NMR chemical shift data for various monomers in chloroform-d (CDCl₃) were taken from the literature [18,19]. Briefly, the chemical shifts of the indicated monomers were measured in CDCl₃ at 35 °C at 125 and/or 500 MHz, respectively, using tetramethylsilane (TMS) as an internal standard.

Preparation of liposomes: Briefly, the method of DPPC liposome preparation was as follows: DPPC was accurately weighed and dissolved in chloroform. The solution was evaporated to a dry thin film on the bottom of a round-bottom test tube and left under vacuum for 30 min. Deuterium oxide (D₂O, Merck, Darmstadt, Germany) in 0.01 M phosphate buffer at pD 7.0 was then added, and sonication was performed under a nitrogen atmosphere for 10 min at 60 °C. After incubation for 20 min at room temperature, unilamelar DPPC liposomes were prepared by centrifugation at 20,000× g for 15 min. ¹H-NMR spectra were measured at 25 °C on a JEOL (Tokyo, Japan) JNM-GX 270 or ALPHA 500 instrument at a resolution of 0.01 ppm and 0.0013 ppm, respectively. An NMR sample tube with a coaxial capillary was used. The coaxial capillary with monomers was inserted into an NMR sample tube with the liposomes. Then, NMR spectra were measured at 25 °C and 50 °C, respectively. The external standard was TMSPA [17,38].

An aliquot sample (20 μL) of DPPC and the indicated concentration of monomer in 0.01 M phosphate buffer solution at pH 7.0 was placed into a DSC specimen container. The specimen was allowed to equilibrate for 14 h at 5 °C, then the specimen was scanned in a sealed calorimetric container on a DSC-Rigaku calorimeter (Rigaku Denki Co. Ltd., Tokyo, Japan) at a heating rate of 5 °C min⁻¹ with a range setting of 0.5 mcal s⁻¹. The instrument was calibrated with indium as a standard. The enthalpy (ΔH) was calculated from the area under the DSC curve [17,38,39].

The concentration eliciting 50% (H₅₀) hemolysis of rabbit erythrocytes for (meth)acrylates was taken from the literature [3].

LD₅₀ (50% lethal dose) data for intraperitoneal injection of mice with acrylate and methacrylate monomers were taken from the literature [3]. Briefly, male albino ICR mice weighing 25 ± 5 g were used as the test animals, and the LD₅₀ dose for each compound was calculated in terms of 7-day mortality.

Parameters η, χ and ω were calculated using Equations 1–3, respectively (Table 4). HOMO, LUMO, and heats of formation were taken from our reported studies [14,15]. Briefly, calculations of heats of formation were performed using the PM3/CONFLEX method. To obtain fine geometry details in the present study, initial geometry optimization was first performed using CONFLEX5 (Conflex, Tokyo, Japan). Thereafter, calculations using the PM3 method in the MOPAC 2000 program were carried out on a Tektronix CAChe workstation (Fujitsu Ltd., Japan).