Synthesis of Novel Biologically Active Proflavine Ureas Designed on the Basis of Predicted Entropy Changes

A novel series of proflavine ureas, derivatives 11a–11i, were synthesized on the basis of molecular modeling design studies. The structure of the novel ureas was obtained from the pharmacological model, the parameters of which were determined from studies of the structure-activity relationship of previously prepared proflavine ureas bearing n-alkyl chains. The lipophilicity (LogP) and the changes in the standard entropy (ΔS°) of the urea models, the input parameters of the pharmacological model, were determined using quantum mechanics and cheminformatics. The anticancer activity of the synthesized derivatives was evaluated against NCI-60 human cancer cell lines. The urea derivatives azepyl 11b, phenyl 11c and phenylethyl 11f displayed the highest levels of anticancer activity, although the results were only a slight improvement over the hexyl urea, derivative 11j, which was reported in a previous publication. Several of the novel urea derivatives displayed GI50 values against the HCT-116 cancer cell line, which suggest the cytostatic effect of the compounds azepyl 11b–0.44 μM, phenyl 11c–0.23 μM, phenylethyl 11f–0.35 μM and hexyl 11j–0.36 μM. In contrast, the novel urea derivatives 11b, 11c and 11f exhibited levels of cytotoxicity three orders of magnitude lower than that of hexyl urea 11j or amsacrine.


Introduction
Previous studies have shown that compounds based on an acridine chromophore possess a wide range of interesting biological activities [1]. The main type of biological effect exhibited by acridine derivatives is that of antiproliferative activity [2]. The main interaction mode of acridines with biological macromolecules is an intercalation reaction with nucleic acids where an electron-deficient acridine chromophore inserts between adjacent base pairs [3,4]. This process affects cell division on the molecular and enzymatic level, leading to the suppression of further cell proliferation. The enzymatic inhibitory effect of acridine-based compounds takes the form of specific interactions with topoisomerases I/II, whereby the ability to stabilize G-rich sequences of telomeric DNA in the form of a G-quadruplex indirectly inhibits the activity of the enzyme telomerase [5].
Genomic DNA is densely packed and undergoes superhelical strain during the process of cell division. Two principal enzymes are involved in this process: topoisomerases I In this study, we present a pharmacological model based on the entropy changes in the urea derivatives 11j-11n accompanying their DNA binding process. This pharmacological model could be used as a tool in the design of new urea-based intercalators with potential anticancer activity. Telomerase is a special type of DNA polymerase which maintains an elongation of a telomeric DNA shortening in tumor cells after each cell division cycle. Telomerase is an RNA reverse transcriptase that synthesizes telomeric DNA from an RNA template, thereby leading to the uncontrollable proliferation of malign cells [6]. Previous studies have shown that substituted acridines such as compounds 4 and 5 exhibit the ability to interact with the telomeric DNA, resulting in interesting cytostatic activities ( Figure 1) [8,9].
Recently, the total synthesis of inubosine B6, an acridine alkaloid that initiates neuroregeneration, has been reported ( Figure 1) [10]. Antimicrobial properties are another important category of biological activity displayed by acridine derivatives [11,12]. Proflavine 7 is the example of a typical antibacterial and antifungal agent, while quinacrine 8 exhibits antimalarial properties (Figure 1) [13].
In this study, we present a pharmacological model based on the entropy changes in the urea derivatives 11j-11n accompanying their DNA binding process. This pharmacological model could be used as a tool in the design of new urea-based intercalators with potential anticancer activity.

Results and Discussion
The stability of the intercalation between ctDNA and ureas 11j-11n ( Figure 1) is highly dependent upon the length of the urea alkyl chains. Previous studies have observed that the enlargement of alkyl chains led to the reduced K B value of the binding complex, and vice versa (Figure 2A) [14].  The binding constants KB × 10 5 of the ctDNA-urea 11j-11n intercalation complexes are given as a molar concentration. A value of the change in the standard entropy ΔS° of urea derivatives 11j-11n given in cal.mol −1 . K −1 was obtained using the PM7 method. Several carbon atoms define the number of carbons in the alkyl chain of the derivative. 11j-n-hexyl, 11k-n-pentyl, 11l-n-butyl, 11m-n-propyl, 11n-ethyl.
In order to predict the strength of the binding between the ligand and DNA, MM-GBSA (PBSA) calculations should be performed for specific intercalation complexes followed by a normal mode analysis to reveal the thermodynamic parameters of the binding process; more specifically, the binding free energy change and its enthalpic and entropic components [15]. Entropy changes are the consequence of the loss of degrees of . Relationship of the binding constant K B and the change in the standard entropy ∆S • . The binding constants K B × 10 5 of the ctDNA-urea 11j-11n intercalation complexes are given as a molar concentration. A value of the change in the standard entropy ∆S • of urea derivatives 11j-11n given in cal.mol −1 . K −1 was obtained using the PM7 method. Several carbon atoms define the number of carbons in the alkyl chain of the derivative. 11j-n-hexyl, 11k-n-pentyl, 11l-n-butyl, 11m-n-propyl, 11n-ethyl.
This finding prompted us to determine the relationship of the K B values and the ∆S • values for the urea derivatives 11j-11n ( Figure 2B). The term ∆S • reflects the structural features of the chains of specific urea derivatives, as is shown in Figure 2, and expresses the gap in the values of the standard entropy, S • , obtained at the temperatures of 300 K and of 340 K using the PM7 method (Table 1). These findings will be discussed further within the framework of the urea-based molecular model design study. In order to predict the strength of the binding between the ligand and DNA, MM-GBSA (PBSA) calculations should be performed for specific intercalation complexes followed by a normal mode analysis to reveal the thermodynamic parameters of the binding process; more specifically, the binding free energy change and its enthalpic and entropic components [15]. Entropy changes are the consequence of the loss of degrees of freedom and enthalpy changes arise from conformation discrepancies when an intercalation complex is formed [6].
Furthermore, it is also necessary to consider the appropriate modes by which the ligand binds to DNA in order to obtain accurate results for the required calculations. However, this task is complicated by the complexity of the intermolecular interactions involved in the binding process. As a result, our study simplifies this process by employing an approach based on the following considerations.
We hypothesized that only the chains of ureas 11j-11n undergo structural changes during the formation of the complex. This idea arose from our expectation that the flexible alkyl chains would adapt to the macromolecule during the binding process rather than vice versa. We further hypothesized that a DNA lattice could undergo similar changes during the formation of complexes with urea derivatives 11j-11n. Given such hypotheses, we could expect that the enthalpy changes in intercalation reactions across the entire series of urea derivatives 11j-11n would not differ substantially; therefore, ∆∆H rxn =~0.
On such a basis, it is possible to suggest that the stability of the binding complex depends substantially on the entropy changes in derivatives 11j-11n which accompany an intercalation reaction. This can be rationalized using the above-mentioned relationship of K B values versus the ∆S • values ( Figure 2B). In terms of structure, the ∆S • value reflects the structural features of the chains of derivatives 11j-11n. By extending this relationship as a prediction tool, the ∆S • value could predict how strongly the urea model can bind to DNA.
Unexpectedly, the cytotoxicity (IC 50 ) of ureas 11j-11n was not found to increase with the increasing of the binding constant, K B , a finding which contradicts the hypothesis that DNA is the typical pharmacological target of such substances ( Figure 3). A positive correlation was found for the relationship of cytotoxicity (IC 50 ) and lipophilicity (LogP), which thereby reflects the fact that the ability of specific urea derivatives to penetrate the cell membrane is of equal importance in the anticancer activity as its ability to bind to DNA ( Figure 4A). Analogously, the prediction of the lipophilicity of a given urea model may offer a useful estimate of its anticancer potential. The task which we addressed was, therefore, how to utilize the two opposing physicochemical properties of logP and K B to predict the biological activity of a specific urea model. freedom and enthalpy changes arise from conformation discrepancies when an intercalation complex is formed [6]. Furthermore, it is also necessary to consider the appropriate modes by which the ligand binds to DNA in order to obtain accurate results for the required calculations. However, this task is complicated by the complexity of the intermolecular interactions involved in the binding process. As a result, our study simplifies this process by employing an approach based on the following considerations.
We hypothesized that only the chains of ureas 11j-11n undergo structural changes during the formation of the complex. This idea arose from our expectation that the flexible alkyl chains would adapt to the macromolecule during the binding process rather than vice versa. We further hypothesized that a DNA lattice could undergo similar changes during the formation of complexes with urea derivatives 11j-11n. Given such hypotheses, we could expect that the enthalpy changes in intercalation reactions across the entire series of urea derivatives 11j-11n would not differ substantially; therefore, ΔΔHrxn = ~0.
On such a basis, it is possible to suggest that the stability of the binding complex depends substantially on the entropy changes in derivatives 11j-11n which accompany an intercalation reaction. This can be rationalized using the above-mentioned relationship of KB values versus the ΔS° values ( Figure 2B). In terms of structure, the ΔS° value reflects the structural features of the chains of derivatives 11j-11n. By extending this relationship as a prediction tool, the ΔS° value could predict how strongly the urea model can bind to DNA.
Unexpectedly, the cytotoxicity (IC50) of ureas 11j-11n was not found to increase with the increasing of the binding constant, KB, a finding which contradicts the hypothesis that DNA is the typical pharmacological target of such substances ( Figure 3). A positive correlation was found for the relationship of cytotoxicity (IC50) and lipophilicity (LogP), which thereby reflects the fact that the ability of specific urea derivatives to penetrate the cell membrane is of equal importance in the anticancer activity as its ability to bind to DNA ( Figure 4A). Analogously, the prediction of the lipophilicity of a given urea model may offer a useful estimate of its anticancer potential. The task which we addressed was, therefore, how to utilize the two opposing physicochemical properties of logP and KB to predict the biological activity of a specific urea model.   Bearing in mind the facts outlined above, it is clear that it would not be possible to design a urea-based model that exhibits both strong DNA binding capabilities and high levels of lipophilicity. One solution, therefore, would be to find a compromise structure, a molecule with sufficient lipophilicity to penetrate a cell membrane but which could still be able to form a stable complex with DNA. For this purpose, a pharmacological model was developed which utilized the relationship of IC50 versus LogP and the relationship of KB versus ΔS° as prediction tools, both of which were discussed above. According to this model, only a determination of the value of logP and the value of ΔS° would be required to determine the binding capacity and anticancer activity for a given urea model.
The pharmacological model is based on a scoring system that determines a specific range of score values which urea derivative models must achieve before they would be considered for synthesis. The score value is the sum of the logP value and the ΔS° value of the urea model and the acceptable range for the derivative models was derived from the score values for the urea derivatives hexyl 11l and butyl 11n, compounds which display a reasonable range of biological activity ( Figure 4).
In order to identify potential new urea-based intercalators with increased biological activity, quantum mechanics and cheminformatics were used to obtain the logP and ΔS° values for specific urea models. Urea models I-XXIII were then modelled using ChemSketch software [16] (Figure 5). Bearing in mind the facts outlined above, it is clear that it would not be possible to design a urea-based model that exhibits both strong DNA binding capabilities and high levels of lipophilicity. One solution, therefore, would be to find a compromise structure, a molecule with sufficient lipophilicity to penetrate a cell membrane but which could still be able to form a stable complex with DNA. For this purpose, a pharmacological model was developed which utilized the relationship of IC 50 versus LogP and the relationship of K B versus ∆S • as prediction tools, both of which were discussed above. According to this model, only a determination of the value of logP and the value of ∆S • would be required to determine the binding capacity and anticancer activity for a given urea model.
The pharmacological model is based on a scoring system that determines a specific range of score values which urea derivative models must achieve before they would be considered for synthesis. The score value is the sum of the logP value and the ∆S • value of the urea model and the acceptable range for the derivative models was derived from the score values for the urea derivatives hexyl 11l and butyl 11n, compounds which display a reasonable range of biological activity (Figure 4).
In order to identify potential new urea-based intercalators with increased biological activity, quantum mechanics and cheminformatics were used to obtain the logP and ∆S • values for specific urea models. Urea models I-XXIII were then modelled using ChemSketch software [16] (Figure 5).
The optimization of models I-XXIII were performed using the PM7 method and a post processing normal mode analysis provided the values of standard entropy, S • ( Table 2). The calculations were performed using the MOPAC2016 software package, while the logP values for the model structures were obtained using Molinspiration online software [17,18]. The logP and ∆S • values were summed together in order to provide the score values of urea models I-XXIII (Table 2). Specific models would be considered as the leading structures if their score value fell within the range defined by the values of 20.1 and 24.7. Only models which featured six carbon atoms in their chains were synthesized: models VIII-XIV, XVII and XVIII ( Figure 5). The optimization of models I-XXIII were performed using the PM7 method and a post processing normal mode analysis provided the values of standard entropy, S° ( Table 2). The calculations were performed using the MOPAC2016 software package, while the logP values for the model structures were obtained using Molinspiration online software [17,18]. The logP and ΔS° values were summed together in order to provide the score values of urea models I-XXIII (Table 2). Specific models would be considered as the leading structures if their score value fell within the range defined by the values of 20.1 and 24.7. Only models which featured six carbon atoms in their chains were synthesized: models VIII-XIV, XVII and XVIII ( Figure 5).   Models VIII-XIV, XVII and XVIII were used as the basis for the synthesis of urea derivatives 11a-11i, which were prepared following our previously published procedure (Scheme 1). The synthesis of 3,6-diisothiocyanatoacridine (9) was improved substantially, with the yield of the reaction increasing from the previous 53% to 83%. Isothiocyanate 9 was purified through crystallization from toluene and was subsequently used in the synthesis of the related thioureas 10a-10i which were isolated as light yellow solids. The next step of the process was the preparation of the final urea derivatives 11a-11i through Models VIII-XIV, XVII and XVIII were used as the basis for the synthesis of urea derivatives 11a-11i, which were prepared following our previously published procedure (Scheme 1). The synthesis of 3,6-diisothiocyanatoacridine (9) was improved substantially, with the yield of the reaction increasing from the previous 53% to 83%. Isothiocyanate 9 was purified through crystallization from toluene and was subsequently used in the synthesis of the related thioureas 10a-10i which were isolated as light yellow solids. The next step of the process was the preparation of the final urea derivatives 11a-11i through the reaction of the related thioureas with mesitylnitrile oxide in methanol. The synthesized derivatives were then used in the subsequent biological and biochemical experiments.

Scheme 1. Synthesis of ureas 11a-11i.
Derivatives 11a-11i were subjected to in vitro screening on cancer cell lines conducted by the Developmental Therapeutic Program of the National Cancer Institute (NCI). A one-dose screen of each derivative against NCI-60 panels consisting of sixty human cancer cell lines was performed in order to identify the derivatives which exhibited the highest biological activity. On the basis of these results, ureas 11b, 11c and 11f were selected for further evaluation of their growth inhibition activity. From the panel of sixty human cancer cell lines, a single cell line per tumor type was selected where the derivatives displayed the highest overall levels of cytostatic activity (Table 3). Urea derivative 11j, the parent structure in the preceding molecular design study, was used as a standard to evaluate the pharmacological model presented herein.
The results presented in Table 3 show clearly that ureas 11b, 11c, 11f and 11j did not display higher levels of cytostatic activity than doxorubicin. However, the ureas did exhibit promising anticancer activities that exceeded that of amsacrine, especially in the case of the HCT-166, RXF393 and NCI-H522 cancer cell lines (Table 3, Table S2). Against the selected cancer cell lines, urea derivatives 11b, 11c, 11f and 11j also displayed superior levels of biological activity than both fluorouracil and cisplatin.

Scheme 1. Synthesis of ureas 11a-11i.
Derivatives 11a-11i were subjected to in vitro screening on cancer cell lines conducted by the Developmental Therapeutic Program of the National Cancer Institute (NCI). A one-dose screen of each derivative against NCI-60 panels consisting of sixty human cancer cell lines was performed in order to identify the derivatives which exhibited the highest biological activity. On the basis of these results, ureas 11b, 11c and 11f were selected for further evaluation of their growth inhibition activity. From the panel of sixty human cancer cell lines, a single cell line per tumor type was selected where the derivatives displayed the highest overall levels of cytostatic activity (Table 3). Urea derivative 11j, the parent structure in the preceding molecular design study, was used as a standard to evaluate the pharmacological model presented herein. The results presented in Table 3 show clearly that ureas 11b, 11c, 11f and 11j did not display higher levels of cytostatic activity than doxorubicin. However, the ureas did exhibit promising anticancer activities that exceeded that of amsacrine, especially in the case of the HCT-166, RXF393 and NCI-H522 cancer cell lines (Table 3, Table S2). Against the selected cancer cell lines, urea derivatives 11b, 11c, 11f and 11j also displayed superior levels of biological activity than both fluorouracil and cisplatin.
In order to provide an overall view of the cytostatic activity of the urea derivatives, the GI 50 values of each derivative for all cancer cell lines are listed in a decimal logarithmic form in the supplementary materials (Tables S1-S4). In summary, ureas 11b, 11c and 11f were found to be the most cytostatic against leukemia cell lines and moderate activity was also shown against colon cell lines (Table 4, Table S2). In order to determine the toxicity (IC 50 ) of the novel derivatives, the biological activity of ureas 11b, 11c, 11f and 11j were performed against immortalized foreskin fibroblasts ( Table 5). The IC 50 values obtained in the studies show that azepyl 11b and phenylethyl 11f urea exhibited levels of cytotoxicity three orders of magnitude lower than their cytostatic activity. The highest level of cytotoxicity was displayed by hexyl urea 11j, a level even greater than that of amsacrine after 48 h and 96 h, whereas phenyl urea 11c was found to be more cytostatic. Figure 6 expresses the IC 50 values in a decimal logarithm scale in order to allow a better comparison of the cytotoxicity of the proflavine ureas and amsacrine.

General Remarks
All chemicals and reagents were of reagent grade. NMR spectra were measured on Varian VNMRS NMR and Varian Mercury Plus 400 FT NMR spectrometers at room

General Remarks
All chemicals and reagents were of reagent grade. NMR spectra were measured on Varian VNMRS NMR and Varian Mercury Plus 400 FT NMR spectrometers at room temperature (600, 400 MHz). δ values were referenced on a residual solvent signal as follows: for 1 H-DMSO-d 6 (2.49 ppm) and for 13 C-DMSO-d 6 (39.5 ppm). Reactions were monitored with thin-layer chromatography (TLC) using Silufol plates with detection at 254 nm. Preparative column chromatography was performed using Aluminum oxide Merck 90 neutral (grain size 200 nm). High-resolution mass spectra (HRMS) were recorded on a micrOTOF-QII quadrupole-time-of-flight mass spectrometer (Bruker Daltonics) with an electrospray ionization source. Optical rotations were determined using a P-2000 Jasco polarimeter. Melting points were recorded on a Koffler hot block and are presented uncorrected.

Synthesis of 3,6-Diisothiocyanatoacridine (9)
Proflavine hemisulfate hydrate (7) (300 mg, 1.43 mmol), thiophogene (0.22 mL, 2.87 mmol) and chloroform (30 mL) were poured into a flask to which 5 mL of the stock solution of sodium carbonate (0.45 g in 25 mL H 2 O) per 5 min was added while the mixture was shaken intensively. A solution of sodium bicarbonate (1.50 g in 15 mL H 2 O) was then added and the reaction mixture was stirred for 20 min. In turn, an organic layer with the soluted product 9 was separated and dried over CaCl 2 . The crude product 9 was purified by flash chromatography on neutral aluminum oxide with chloroform as an eluent. The purified isotiocynate 9 was then crystalized from toluene (50 mL) with the addition of activated coal (100 mg) at 80 • C. All physicochemical properties were in accordance with the previously published data [14].

General Protocol for the Synthesis of Thioureas 10a-10k
Amine (2.0 mmol) was added to a heterogeneous solution of 3,6-diisothiocyanatoacridine (9) (60 mg, 0.20 mmol) in methanol (1.50 mL) [14]. The reaction mixture was stirred vigorously and the course of the reaction was monitored using TLC in which chloroform was used as an eluent. When the TLC result was negative in the presence of reactant 9, diethylether was added and the resultant mixture was stirred for an additional two hours to maximize the precipitation of the product. The heterogeneous mixture was then filtered off and the crude product was washed with ethylacetate then with n-hexane in a funnel. The final product was crystalized from the DMF-methanol mixture.

Screening of Anticancer Activity-NCI-60 Panels
The anticancer activity of proflavine ureas 11a-11j was tested against NCI-60 panels consisting of sixty human cancer cell lines. Derivatives 11a-11j were initially at a single dose of 10 µM to identify the individual derivatives with the highest level of growth inhibition. Urea derivatives 11b, 11c, 11f and 11j were found to have exhibited significant growth inhibition in the One-Dose Screen, and these derivatives were further evaluated against NCI-60 panels at five concentration levels to determine the growth inhibition concentration, GI 50 . The GI 50 values for doxorubicin, fluorouracil and cisplatin were downloaded from the NCI data repository (https://dtp.cancer.gov/databases_tools/data_search.htm).

Normal Mode Analysis
To build urea models I-XXIII, 3D Viewer (ver. 11.01) and ChemSketch (ver. 11.02) software as components of the ACD Labs software package were used [16]. Molecular design studies were performed with Gabedit [19] and MOPAC2016 software using the PM7 method [17]. The optimization of models I-XXIII was controlled by the commands: LET; DDMIN = 0; GNORM = 0.1; AUX; EPS = 78; EF; XYZ. The post-processing normal modes analysis was controlled using the keywords: FORCE; THERMO; PRECISE; EPS = 78.

Calculation of ∆S • Value
The change in the standard entropy, ∆S • , of specific urea models was calculated using the value of the standard entropy, S • , at a temperature of 300 and 340 degrees of Kelvin. The values of the standard entropy are listed in the Mopac output file.

Calculation of LogP Value
LogP calculations were performed using the "Calculation of Molecular Properties and Prediction of Bioactivity" module within the options of the Molinspiration web tool in order to obtain the logP value of individual urea models [18].

Calculation of Score Value
The score value of individual urea models was obtained as the sum of the LogP value and the ∆S • value.

Conclusions
Building on our previous studies on the urea derivatives 11j-11n, the pharmacological model was developed in order to facilitate the design of novel urea-based proflavine intercalators which could act as cytostatic substances. The model using a scoring system that utilized the lipophilicity and the change in the standard entropy for individual structures.
The physicochemical values of logP and ∆S • were obtained through a theoretical approach. In order for a given urea model to be synthesized for in vitro screening, it was necessary to achieve a score value which fell within the score range defined by the structure-activity relationship for ureas 11j-11n. Using this scoring system, urea models VIII-XIV, XVII and XVIII were selected as the leading models for synthesis from which all had at least six carbon atoms in their chains.
An improved protocol was applied in the preparation of the urea derivatives in which the yield of the synthesis of 3,6-diisotiocyanatoacridine (9) was enhanced substantially, from the previous yield of 53% to 83%.
Urea derivatives 11a-11j were subjected to further investigation through in vitro screening against NCI-60 human cancer cell lines in order to determine their anticancer activity. The strongest cytostatic effect was exhibited by azepyl 11b, phenyl 11c and phenylethyl urea 11f, all of which inhibited the growth of cancer cell lines even at submicromolar concentrations, while ureas 11a, 11d, 11e, 11g, 11h and 11i displayed no effect whatsoever. In comparison to hexyl urea 11j, the parent structure in the preceding molecular design study, the cytostatic activity of ureas 11b, 11c and 11f was found to have improved only slightly, but this moderate improvement is in line with the results of the scoring system based on opposing physicochemical properties. In contrast, the cytotoxicity of ureas 11b, 11c and 11f was more than three orders of magnitude lower than the cytotoxicity of urea 11j. In summary, this is the main outcome of the study, especially compared to hexylurea 11j, which showed almost no selectivity for tumor cells over non-cancer cells.
In conclusion, a series of novel proflavine ureas, derivatives 11a-11i, were synthesized based on molecular modeling design studies. The structure of the novel ureas was proposed on the basis of the pharmacological model. The lipophilicity (LogP) and the changes in the standard entropy (∆S • ), as the input parameters of the pharmacological model, were obtained using quantum mechanics and cheminformatics. The pharmacological model is based on a scoring system, in which urea models are required to achieve within a specific range in order to be considered suitable for synthesis. The anticancer activity of the synthesized derivatives was evaluated against NCI-60 human cancer cell lines. The strongest anticancer activity was exhibited by ureas azepyl 11b, phenyl 11c and phenylethyl 11f, although the levels of activity were only a slight improvement on those recorded for hexyl urea 11j. The GI 50 values recorded for the derivatives against HCT-116 cancer cell line give an indication of the cytostatic effect of the ligands azepyl 11b-0.44 µM, phenyl 11c-0.23 µM, phenylethyl 11f-0.35 µM and hexyl 11j-0.36 µM. In contrast, the novel ureas 11b, 11c and 11f exhibited levels of cytotoxicity three orders of magnitude lower than that of hexyl urea 11j.