Efficient Synthesis of Acylated, Dialkyl α-Hydroxy-Benzylphosphonates and Their Anticancer Activity

An efficient method applying acyl chlorides as reagents was developed for the acylation of the hindered hydroxy group of dialkyl α-hydroxy-benzylphosphonates. The procedure did not require any catalyst. A few acylations were also performed with the SC-enantiomer of dimethyl α-hydroxy-benzylphosphonate, and the optical purity was retained. A part of the acyloxyphosphonates was tested against eight tumor cell lines of different tissue origin at c = 50 μM concentration. The compounds elicited moderate cytostatic effect against breast, skin, prostate, colon, and lung carcinomas; a melanoma cell line; and against Kaposi’s sarcoma cell lines. Then, dose-dependent cytotoxicity was assayed, and benzoylation of the α-hydroxy group was identified as a moiety that increases anticancer cytotoxicity across all cell lines. Surprisingly, a few analogues were more toxic to multidrug resistant cancer cell lines, thus evading P-glycoprotein mediated drug extrusion.


Introduction
The most important synthesis of α-hydroxyphosphonates is the Pudovik reaction of oxo compounds (aldehydes and ketones) with dialkyl phosphites [1]. Different kinds of base and acid catalysts were described. Base catalyst may be triethylamine [2], TEA/MgCl 2 [3], barium hydroxide [4,5], or potassium phosphate [6]. Microwave assistance was also useful during the syntheses [7]. A number of solvent-free methods were elaborated for the Pudovik reaction itself; however, the work-up and preparation (extraction, chromatography, recrystallization) requested a considerable quantity of solvent [8][9][10]. One author of this paper together with coworkers developed a green procedure for the synthesis of α-hydroxybenzylphosphonates. The method comprised the reaction of dialkyl phosphites with benzaldehyde derivatives in a minimum quantity of acetone, in the presence of 5% of triethylamine. After a few hours' reflux, the adduct precipitated on cooling [11].
Beyond their potential bioactivity, the α-hydroxyphosphonates may be important intermediates in a series of syntheses [1]. The most important reactions are alkylations [1], acylations [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27], phosphorylations [28], substitutions [1], rearrangements [1], and dealkylations [1]. As regards acylations, a series of α-(aryloxyacetoxy)-alkylphosphonate derivatives [12][13][14][15] was prepared mainly by acylation of different α-hydroxyphosphonates with α-hydroxyphosphonates with carboxylic acid chlorides, including aryloxy-butyryloxy or -valeroxy [16] and heterocyclic derivatives [17,18]. A part of the products prepared were described as herbicidal agents. Among the more complex examples, modification of αhydroxyphosphonates with acetyl chloride was also described, but in a complicated manner (see below) [19]. Carboxylic acid anhydrides [20][21][22][23] and the acids themselves [24][25][26] were also used in the acylation of hydroxyphosphonates. Among the carboxylic acids, benzoic acid and propionic acid were also applied [25,26]. Comparing the literature examples on the acylation of diethyl α-hydroxy-benzylphosphonates (Table 1), the experiences may be summarized as follows. Using acetyl-chloride in the presence of an irondoped, single-walled carbon nanotube catalyst at 90 °C without any solvent, the acetyloxy-benzylphosphonate was obtained with a yield of 87% (Table 1, entry 1) [19]. The need for the special catalyst is a disadvantage. Acetic acid anhydride was applied in a solvent-free microwave (MW)-assisted acylation. Although the outcome was practically quantitative (98%), and the reaction time was only 3 min [22], the use of a kitchen MW equipment (and hence, the lack of the temperature range) does not allow a reproduction ( Table 1, entry 2). Another method involving Ac2O utilized trichlorotitanium trifluoromethanesulfonate as the catalyst at 26 °C in a solvent-free manner affording the product an 89% yield (Table 1, entry 3) [21]. A similar acylation was performed with the anhydride of benzoic acid. In this case, there was a need for 80 °C and for a longer reaction time ( Table 1, entry 4) [21]. In both cases, the use of the special catalyst means a disadvantage. Last but not least, α-hydroxy-benzylphosphonate was acylated with propionic acid and benzoic acid under the conditions of the Mitsunobu reaction in boiling acetonitrile, furnishing the corresponding esters in 75/80% yields (Table 1, entries 5 and 6) [26]. Obviously, this is an elegant way of acylation. It is noteworthy that a tosyloxybenzylphosphonate was also described [27]. The phosphorylation of α-hydroxyphosphonates [28] was elaborated by us. In this paper, we wished to describe a simple and efficient (robust) acylation of α-hydroxyphosphonates. It was also our plan to prepare optically active acylated hydroxyphosphonates and to test the cytotoxic activity of the acyloxy derivatives against different cancer cell cultures.  [28] a No exact temperature was provided due to the use of a kitchen oven. b Was applied in equivalent quantity as a reagent.
In this paper, we wished to describe a simple and efficient (robust) acylation of α-hydroxyphosphonates. It was also our plan to prepare optically active acylated hydroxyphosphonates and to test the cytotoxic activity of the acyloxy derivatives against different cancer cell cultures.

The Acylation of Racemic Diethyl and Dimethyl α-Hydroxy-Benzylphosphonates
It was shown that acylated α-hydroxyphosphonates were synthesized in different ways applying acyl chlorides, carboxylic acids, and anhydrides under diverse conditions including special catalysts, microwave irradiation, or 4,4 -azopyridine as the reagent of the Mitsunobu reaction. The temperature range embraced 26-90 • C, and the yields fell in the range of 75-90% [19,21,22,26]. As pointed out, none of these methods were too attractive. We wished to use acyl chlorides under simple conditions.
The starting diethyl and dimethyl α-hydroxy-arylmethylphosphonates (1a-d and 3a) were prepared as described earlier [11], or by the extension of this method (see Experimental). Derivatives 1c and 1d were new.
In the first round, racemic diethyl α-hydroxy-arylmethylphosphonates 1a-d were reacted with 3 equivalents of acetyl chloride in toluene in the presence of 1.1 equivalents of triethylamine. The role of the base was to bind the HCl liberated during acylation. The acylation of unsubstituted 1a at 25 • C in a closed vessel required a reaction time of 24 h. However, in the other instances with 4-Cl, 4-CF 3 , and 3,4,5-triF electron-withdrawing substituents in the phenyl ring (1b, 1c, and 1d, respectively), there was need for a higher temperature of 50-60 • C. Then, the diethyl hydroxy-benzylphosphonate (1a) was acylated with 1.5 equivalents of butyryl chloride and benzoyl chloride, as described above. After purification by column chromatography, the acyloxyphosphonates (2a-f) were obtained with yields of 69-97% (Table 2).

The Acylation of Racemic Diethyl and Dimethyl α-Hydroxy-Benzylphosphonates
It was shown that acylated α-hydroxyphosphonates were synthesized in different ways applying acyl chlorides, carboxylic acids, and anhydrides under diverse conditions including special catalysts, microwave irradiation, or 4,4′-azopyridine as the reagent of the Mitsunobu reaction. The temperature range embraced 26-90 °C, and the yields fell in the range of 75-90% [19,21,22,26]. As pointed out, none of these methods were too attractive. We wished to use acyl chlorides under simple conditions.
The starting diethyl and dimethyl α-hydroxy-arylmethylphosphonates (1a-d and 3a) were prepared as described earlier [11], or by the extension of this method (see Experimental). Derivatives 1c and 1d were new.
In the first round, racemic diethyl α-hydroxy-arylmethylphosphonates 1a-d were reacted with 3 equivalents of acetyl chloride in toluene in the presence of 1.1 equivalents of triethylamine. The role of the base was to bind the HCl liberated during acylation. The acylation of unsubstituted 1a at 25 °C in a closed vessel required a reaction time of 24 h. However, in the other instances with 4-Cl, 4-CF3, and 3,4,5-triF electron-withdrawing substituents in the phenyl ring (1b, 1c, and 1d, respectively), there was need for a higher temperature of 50-60 °C. Then, the diethyl hydroxy-benzylphosphonate (1a) was acylated with 1.5 equivalents of butyryl chloride and benzoyl chloride, as described above. After purification by column chromatography, the acyloxyphosphonates (2a-f) were obtained with yields of 69-97% (Table 2). Then, racemic dimethyl α-hydroxy-benzylphosphonate (3a) was subjected to acylation by reaction with 1.1 equivalents of valeryl-, propionyl-, or benzoyl chloride or 3 equivalents of acetyl chloride in a few combinations. These acylations required more forcing conditions owing to the lower reactivity of the dimethyl esters compared with the diethyl ones (e.g., 3a vs. 1a). The C4-, C2-, and C1-acyloxyphosphonates (4a-d) were prepared in 45-87% yields ( Then, racemic dimethyl α-hydroxy-benzylphosphonate (3a) was subjected to acylation by reaction with 1.1 equivalents of valeryl-, propionyl-, or benzoyl chloride or 3 equivalents of acetyl chloride in a few combinations. These acylations required more forcing conditions owing to the lower reactivity of the dimethyl esters compared with the diethyl ones (e.g., 3a vs. 1a). The C 4 -, C 2 -, and C 1 -acyloxyphosphonates (4a-d) were prepared in 45-87% yields ( Table 3).  All products (2a-f and 4a-d) were characterized by 31 P, 13 C, and 1 H NMR as well as HRMS. From among the ten acyloxyphosphonates, five (2c,d and 4a-c) were new compounds. 31 P, 13  All products (2a-f and 4a-d) were characterized by 31 P, 13 C, and 1 H NMR as well as HRMS. From among the ten acyloxyphosphonates, five (2c,d and 4a-c) were new compounds. 31 P, 13 C and 1 H NMR spectra of the products can be found in the Supplementary Materials section.
The method developed applies acid chlorides in smaller or larger excess to acylate the sterically hindered α-hydroxy group of arylmethylphosphonates (1 and 3). Contrary to earlier cases described in the literature, as there was no need for catalysts, and as the yields were mostly high, our method may be regarded a robust one.

The Acylation of the S-Enantiomer of Dimethyl α-Hydroxy-Benzylphosphonate
The resolution of dimethyl α-hydroxy-benzylphosphonate (3a) was performed according to an earlier procedure [29]. The optically active S-hydroxyphosphonate (3a*) with an ee of 84% was also subjected to O-acylation with acetyl-, propionyl-, and benzoyl chloride to afford hydroxyphosphonates 4b*, 4c*, 4d*, respectively (Table 4). According to chiral HPLC, no racemization occurred; hence, the optical activity was preserved during the acylations. See Figure 1 (Aa,Ba,Ca). For clarity, the chromatograms of the corresponding racemates are also shown ( Figure 1 (Ab,Bb,Cb)). Retention of the configuration is the consequence of the fact that the O-acylation does not affect the chirality center. The products were characterized by 31 P NMR chemical shifts, as well as by specific rotations. All products (2a-f and 4a-d) were characterized by 31 P, 13 C, and 1 H NMR as well as HRMS. From among the ten acyloxyphosphonates, five (2c,d and 4a-c) were new compounds. 31 P, 13 C and 1 H NMR spectra of the products can be found in the Supplementary Materials section.
The method developed applies acid chlorides in smaller or larger excess to acylate the sterically hindered α-hydroxy group of arylmethylphosphonates (1 and 3). Contrary to earlier cases described in the literature, as there was no need for catalysts, and as the yields were mostly high, our method may be regarded a robust one.

The Acylation of the S-Enantiomer of Dimethyl α-Hydroxy-Benzylphosphonate
The resolution of dimethyl α-hydroxy-benzylphosphonate (3a) was performed according to an earlier procedure [29]. The optically active S-hydroxyphosphonate (3a*) with an ee of 84% was also subjected to O-acylation with acetyl-, propionyl-, and benzoyl chloride to afford hydroxyphosphonates 4b*, 4c*, 4d*, respectively (Table 4). According to chiral HPLC, no racemization occurred; hence, the optical activity was preserved during the acylations. See Figure 1 (Aa,Ba,Ca). For clarity, the chromatograms of the corresponding racemates are also shown ( Figure 1 (Ab,Bb,Cb)). Retention of the configuration is the consequence of the fact that the O-acylation does not affect the chirality center. The products were characterized by 31 P NMR chemical shifts, as well as by specific rotations.
The cytostatic effect of the hydroxyphosphonate derivatives was dependent on the cell type; A431 human skin carcinoma and Ebc-1 lung carcinoma proved to be the most sensitive, as several compounds elicited a cytostatic effect against these cell lines, while proliferation of PC-3 prostatic carcinoma and Kaposi's sarcoma were the least affected by the species studied. The most pronounced cytostatic effect was induced by acylated hydroxyphosphonate 4c against A2058 human melanoma cells (55.4 ± 1.5%). Derivatives 1a, 2a, and 4d showed significant antitumor effect against A431 cells (40.8 ± 4.8%, 30.5 ± 2.6, and 36.1 ± 0.6%, respectively). Hydroxyphosphonate 1a and compound 4d elicited an antiproliferative effect against MDA-MB-231 cells as well (29.7 ± 4,7%, and 36.1 ± 2.3%, respectively). In case of Ebc-1, hydroxyphosphonates 3a and 2d proved to be the most effective (36.8 ± 1.9% and 33.0 ± 2.4%, respectively. In case of HT-29 and KS cells, a Molecules 2022, 27, 2067 5 of 16 moderate (<30%) cytostatic effect could be observed. Against HT-29 cells, 1d, 4c, 4b*, and 4c* were the most effective compounds, whereas 2c and 2d proved to be moderately antiproliferative. The cytostatic activity of the members of the hydroxyphosphonate family is summarized in Table 5 and Figure 2.

Determination of IC 50 Values
Based on the results shown in Table 5, we decided to measure cytotoxicity to quantify the effect of different substituents. To have a broader view, we chose 7 cell lines of different origin. To investigate and compare the effect of 3,4,5-trifluorination and 3-trifluoromethylation of the benzene ring, and alkylation/arylation of the α-OH group, we assayed the cytotoxicity of phosphonate derivatives 1a, 1c,d, 2a, and 2e,f using the PrestoBlue viability reagent. Up to 500 µM, only compound 2f was toxic enough to obtain IC 50 values. (It is noted that hydroxyphosphonates 1c,d were not tested against 143/B.) The toxicity of 2f was robust throughout the cell panel; IC 50 values ranged from 234 µM against 143/B osteosarcoma line to 363 µM against CAKI-1 renal carcinoma line, (Table 6) highlighting the contribution of an extra benzene ring to toxicity. Interestingly, when we compared A431 and its ABCB1-expressing multidrug resistant (MDR) derivative, A431-B1, species 2f was slightly (but not significantly) more toxic to the resistant line. To obtain a more detailed structure-activity relationship, we tested the compounds investigated in the cytostatic screen (see Table 6) against Mes-Sa mCherry (Mes-Sa mCh) and Mes-Sa/Dx5 eGFP (Dx5 eGFP) cells to establish basic SAR for α-hydroxyphosphonates as in an earlier study [38]. The results are shown in Figure 3 and Table 7.
Another important observation is that the enantiomers with S-configuration (4b*, 4c*, and 4d*) gave different results compared with their racemic form: none of them killed at least 50% of Mes-Sa mCh or Dx5 eGFP cells up to 500 µM. Thus, the S-form seems to be inactive; therefore, most probably the R-form is the biologically active entity in the cytotoxicity tests. To clarify this observation, further experiments will be carried out in due course.  . IC50 values of the tested compounds against Mes-Sa mCh (Mes) and Mes-Sa/Dx5 eGFP (Dx5). Compounds that did not trigger at least 50% growth inhibition at 500 μM were considered nontoxic; these are shown above the dashed line at 500 μM.
Another important observation is that the enantiomers with S-configuration (4b*, 4c*, and 4d*) gave different results compared with their racemic form: none of them killed at least 50% of Mes-Sa mCh or Dx5 eGFP cells up to 500 μM. Thus, the S-form seems to be inactive; therefore, most probably the R-form is the biologically active entity in the cytotoxicity tests. To clarify this observation, further experiments will be carried out in due course.
The Dx5 eGFP cell line, which is the MDR derivative of Mes-Sa, was more susceptible to the tested compounds. Compound 2f showed the highest 4.6-fold selectivity to Dx5 eGFP. Due to P-glycoprotein (P-gp) overexpression, Dx5 eGFP is resistant to doxorubicin and other P-gp substrate chemotherapeutics, while, at the same time, hypersensitive to so-called MDR-selective agents, such as NSC57969, which depletes intracellular iron through P-gp, triggering cell death [39][40][41]). To assess if P-gp plays a role in the observed hypersensitivity of Dx5 eGFP cells against the tested analogues, we repeated the experiments in the presence of tariquidar-a P-gp inhibitor (Table 7). One can see a trend in decreasing selectivity that was due to the increased toxicity trend against Mes-Sa mCh and the decreased toxicity trend against Dx5 eGFP. This was a bit unexpected, as mostly, in the presence of tariquidar, the toxicity of compounds against Mes-Sa mCh cells do not change . Phosphonates 1a, 1c,d, 2a, 2c-f, 4c, and 4c*,d* were not toxic to A431 and A431-B1, only species 2f (Table 6) and 4d (data not shown) were active with a selectivity of 1.07 and 1.2, respectively. Based on these results, we conclude that increased susceptibility of MDR cells is not conveyed by P-gp. Nevertheless, all of the analogues proved to be equally effective against MDR cells, suggesting that they can evade P-gp-mediated drug resistance in cancer. IC 50 values of the tested compounds against Mes-Sa mCh (Mes) and Mes-Sa/Dx5 eGFP (Dx5). Compounds that did not trigger at least 50% growth inhibition at 500 µM were considered nontoxic; these are shown above the dashed line at 500 µM. Table 7. Cytotoxicity expressed as IC 50 values and standard deviation (SD) of phosphonate analogues. TQ: presence of 1 µM tariquidar. nt: no IC 50 value was detected up to 500 µM. p < 0.05, *; p < 0.01,**. The Dx5 eGFP cell line, which is the MDR derivative of Mes-Sa, was more susceptible to the tested compounds. Compound 2f showed the highest 4.6-fold selectivity to Dx5 eGFP. Due to P-glycoprotein (P-gp) overexpression, Dx5 eGFP is resistant to doxorubicin and other P-gp substrate chemotherapeutics, while, at the same time, hypersensitive to so-called MDR-selective agents, such as NSC57969, which depletes intracellular iron through P-gp, triggering cell death [39][40][41]). To assess if P-gp plays a role in the observed hypersensitivity of Dx5 eGFP cells against the tested analogues, we repeated the experiments in the presence of tariquidar-a P-gp inhibitor (Table 7). One can see a trend in decreasing selectivity that was due to the increased toxicity trend against Mes-Sa mCh and the decreased toxicity trend against Dx5 eGFP. This was a bit unexpected, as mostly, in the presence of tariquidar, the toxicity of compounds against Mes-Sa mCh cells do not change . Phosphonates 1a, 1c,d,  2a, 2c-f, 4c, and 4c*,d* were not toxic to A431 and A431-B1, only species 2f (Table 6) and 4d (data not shown) were active with a selectivity of 1.07 and 1.2, respectively. Based on these results, we conclude that increased susceptibility of MDR cells is not conveyed by P-gp. Nevertheless, all of the analogues proved to be equally effective against MDR cells, suggesting that they can evade P-gp-mediated drug resistance in cancer.

General
The 31 P, 13

Cell Lines and Culture Conditions-In Vitro Cytostasis Assays
In vitro cytostatic effect of the compounds was studied on MDA-MB-231 human breast adenocarcinoma [30], A431 human epidermoid carcinoma [31], PC-3 human prostate adenocarcinoma [32], Ebc-1 human lung squamous cell carcinoma [33], MDA-MB-453 human metastatic epithelial breast carcinoma [34], A2058 human melanoma [35], HT-29 human colorectal carcinoma [36], and Kaposi's sarcoma (KS) [37] cells. MDA-MB 435 cell line was a generous gift of Dr. Angels Fabra, Hospital of Duran e Reynalds, Barcelona, Spain, 1995 and obtained from Dr. Janet E. Price [36]. The other cell lines were generous gifts of Dr. József Tóvári (Department of Experimental Pharmacology, National Institute of Oncology, Budapest, Hungary). MDA-MB-231, MDA-MB-453, PC-3, Ebc-1, and A431 cells were cultured in DMEM medium (Lonza, Basel, Switzerland) supplemented with 10% FBS (EuroClone, Pero, Italy), 2 mM L-glutamine (BioSera, Nuaille, France), penicillinstreptomycin antibiotics mixture (50 IU/mL and 50 µg/mL, respectively), 1 mM sodium pyruvate (both obtained from Lonza, Basel, Switzerland), and 1% nonessential amino acid mixture (BioSera, Nuaille, France). KS, A2058, and HT-29 cells were cultured in RPMI-1640 medium (Lonza, Basel, Switzerland) supplemented with 10% FBS (EuroClone, Pero, Italy), 2 mM L-glutamine (EuroClone, Pero, Italy), and penicillin-streptomycin antibiotics mixture (50 IU/mL and 50 µg/mL, respectively) (Lonza, Basel, Switzerland). The cultures were maintained at 37 • C in a humidified atmosphere with 5% CO 2 . The cells were grown to confluency and then divided into 96-well tissue culture plates (Sarstedt, Nümbrecht, Germany) with the initial cell number of 5.0 × 10 3 cells/well. After 24 h incubation at 37 • C, the cells were treated with the compounds in 200 µL final volume containing 1.0 v/v% DMSO (Merck, Darmstadt, Germany) at 50 µM concentration overnight, whereas control cells were treated with serum-free medium only, or with DMSO (c = 1.0 v/v%) at the same conditions. After incubation, the cells were washed twice with serum-free medium. Subsequently, the cells were cultured for additional 72 h in 10% serum containing medium at 37 • C; then, the MTT (Merck, Darmstadt, Germany) solution (at c = 0.37 mg/mL final concentration) was added to each well. The respiratory chain [45] and other electron transport systems [46] reduce MTT, and thereby form non-water-soluble violet formazan crystals within the cell [47]. The amount of these crystals may be determined by spectrophotometry and serves as an estimate for the number of mitochondria, and hence, the number of living cells in the well [48]. After 3 h of incubation with MTT, the cells were centrifuged with 2000 rpm for 5 min and then the supernatant was removed. The obtained formazan crystals were dissolved in DMSO (100 µL) and the optical density (OD) of the samples was measured at λ = 540 nm and 620 nm, respectively, using ELISA Reader (iEMS Reader, Labsystems, Vantaa, Finland). OD 620 values were subtracted from OD 540 values. The percent of cytostasis was calculated with the following equation: Cytostatic effect (%) = [1 − (OD treated / OD control )] × 100 where values OD treated and OD control correspond to the optical densities of the treated and the control wells, respectively. In each case, two independent experiments were carried out with four parallel measurements. Statistical analysis of data was performed using Student's t test at the 95% confidence level.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not relevant.