Esters of Bendamustine Are by Far More Potent Cytotoxic Agents than the Parent Compound against Human Sarcoma and Carcinoma Cells

The alkylating agent bendamustine is approved for the treatment of hematopoietic malignancies such as non-Hodgkin lymphoma, chronic lymphocytic leukemia and multiple myeloma. As preliminary data on recently disclosed bendamustine esters suggested increased cytotoxicity, we investigated representative derivatives in more detail. Especially basic esters, which are positively charged under physiological conditions, were in the crystal violet and the MTT assay up to approximately 100 times more effective than bendamustine, paralleled by a higher fraction of early apoptotic cancer cells and increased expression of p53. Analytical studies performed with bendamustine and representative esters revealed pronounced cellular accumulation of the derivatives compared to the parent compound. In particular, the pyrrolidinoethyl ester showed a high enrichment in tumor cells and inhibition of OCT1- and OCT3-mediated transport processes, suggesting organic cation transporters to be involved. However, this hypothesis was not supported by the differential expression of OCT1 (SLC22A1) and OCT3 (SLC22A3), comparing a panel of human cancer cells. Bendamustine esters proved to be considerably more potent cytotoxic agents than the parent compound against a broad panel of human cancer cell types, including hematologic and solid malignancies (e.g. malignant melanoma, colorectal carcinoma and lung cancer), which are resistant to bendamustine. Interestingly, spontaneously immortalized human keratinocytes, as a model of “normal” cells, were by far less sensitive than tumor cells against the most potent bendamustine esters.

Exploring the properties of derivatives of bendamustine (1), esters of bendamustine (2-7, Fig 1) [31] comprising basic moieties (4-7) were prepared as potential prodrugs with higher solubility compared to simple esters such as compounds 2 and 3, which were mainly prepared as synthetic intermediates [32,33]. Very recently, we reported on the stability of the nitrogen mustard and the ester moieties in compounds 2-7 against hydrolysis and enzymatic cleavage in buffer, in the presence of porcine butyrylcholine esterase as well as in human and murine plasma [34]. The moderately basic morpholinoethyl ester (4) proved to be of particular interest with respect to both solubility and stability [34]. Preliminary data suggested considerably increased cytotoxicity of the esters compared to the parent compound 1, the basic compounds being of particular interest [31]. Based on the assumption that higher antiproliferative activity may result from increased cellular accumulation, additional mechanisms of action or both, we compared compounds 2-7 with bendamustine (1) regarding cytotoxicity against a panel of human cancer cell types, representing hematologic and solid malignancies. Additionally, the induction of p53 expression and apoptosis, cellular enrichment and the involvement of the organic cation transporters OCT1 and OCT3 were investigated.
Genetically modified HEK293 cells. Transfected human embryonic kidney cells (HEK293) were cultured in EMEM, supplemented with penicillin/streptomycin (100 U/mL / 100 μg/mL) and 250 μg/mL hygromycin B. The cells were cultured in a water-saturated atmosphere with 5% CO 2 at 37°C and passaged once to twice a week. Adherently growing cells were treated with trypsin/EDTA (0.5 mg/mL / 0.22 mg/mL) (PAA, Pasching, Austria) and washed with medium prior to transfer into new culture flasks. Cells growing in suspension were passaged after mechanical separation of cell agglomerates.
In addition to long-term exposure, IC 50 values were determined after an incubation period of 96 hours, following the same procedure as described above (4 instead of 8 replicates). As suggested by the National Cancer Institute [41] the corrected T/C values (T/C corr ) were plotted against the logarithm of the concentrations, and the IC 50 values were calculated using Prism 5.01 (GraphPad Software, La Jolla, CA, USA) according to the "log (inhibitor) vs. normalized response-variable slope" equation.

Detection of Apoptosis (Annexin V/Propidium Iodide Assay)
Apoptosis was determined by incubating proliferating Jurkat cells with medium containing 1, 2, 4 or 5 at a concentration of 10 μM or 0.1% of DMSO (untreated control). After different periods of incubation (6, 24, 48 hours), samples were analyzed using the Annexin V-FITC apoptosis detection kit l (BD Biosciences, Heidelberg, Germany) according to the manufacturer's protocol using 10 6 cells/mL. Cells were analyzed using a FACSCalibur flow cytometer (BD Biosciences, Heidelberg, Germany). The compensation was performed for each experiment with annexin V-FITC (530/30 BP filter) and propidium iodide (585/42 BP filter), respectively. At least 1 Á 10 4 events were registered per sample and debris as well as cell aggregates were excluded by forward (FSC) versus side scatter (SSC) gates. Raw data were analyzed using FlowJo V10 software (Treestar Inc., Ashland, OR, USA).

Detection of p53 Expression by Immunoblotting
The expression of the tumor suppressor p53 by NCI-H460 and HT-29 cells was determined after incubating the cells with compounds 1, 2, 4 and 5 at different concentrations for 24 hours. The cells (from a 10-cm culture dish, 70% confluency) were washed twice with PBS and harvested by scraping after addition of ice-cold buffer A (10 mM HEPES pH = 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, protease inhibitor mix (Sigma-Aldrich)). After adding Nonidet P-40 (NP-40) (Sigma, Taufkirchen, Germany) to a final concentration of 1%, the cell suspensions were vortexed and subsequently centrifuged at 13000 g and 4°C for 30 seconds. The pellets were re-suspended in buffer B (buffer A + 400 mM NaCl, 1% NP-40) and gently agitated using a Sarmix M2000 (Sarstedt, Nümbrecht, Germany) at 4°C for 15 min, followed by centrifugation (13000 g, 4°C, 5 min; Microfuge; Eppendorf, Hamburg, Germany). The concentration of soluble protein was determined according to Bradford using the Bio-Rad Protein Assay (Bio-Rad Laboratories, Munich, Germany).

Quantification of Cell-Associated Bendamustine and Derivatives
Accumulation of bendamustine and selected bendamustine esters by HT-29 and NCI-H460 cells was determined by HPLC. For this purpose, 1.5 Á 10 6 cells/well were seeded into 6-well plates (Sarstedt, Nümbrecht, Germany). After two days of cultivation, the cells were incubated in PBS containing 30 μM of 1, 2, 4 or 5 and 200 μM of umbelliferone as internal standard at 37°C for 10 minutes. After removing the medium, the cells were washed three times with PBS. Untreated cells were detached with trypsin/EDTA and counted. The treated cells were harvested by adding 200 μL of ice-cold perchloric acid (1 M) and scraping. Subsequently, the samples were vortexed and sonicated (Branson 3200 ultrasonic cleaner; Branson, Danbury, USA) for 10 minutes. After 5 minutes of centrifugation (13000 g, 4°C), the supernatants were filtered (0.2 μm Phenex; Phenomenex, Aschaffenburg, Germany) and directly analyzed by HPLC with fluorescence detection according to a recently reported validated procedure [34]. The normalization of the measured concentration to the cell count allowed for a calculation of the cell-associated amount of the respective test compound. Additionally, the ratio of cell-associated substance compared to the applied concentration (30 μM) was calculated based on an average cell volume of 3 pL.

Flow Cytometric Determination of OCT1 and OCT3 Activity
The previously described fluorescent substrate of organic cation transporters, 4-(4-dimethylaminostyryl)-N-methylpyridinium (ASP + ) [42][43][44][45], was used to determine the function of OCT1 and OCT3 by flow cytometry in the absence and the presence of bendamustine and selected bendamustine esters. HEK293 cells expressing the transporter of interest were trypsinized and washed twice with PBS, prior to re-suspension of 0.5 Á 10 6 cells in 500 μL of PBS containing 2% (v/v) FCS. The cells were incubated with ASP + (100 μM stock solution in PBS containing 10% DMSO) for 5 minutes, allowing for cellular accumulation, and subsequently analyzed with a FACSCalibur flow cytometer. ASP + was excited at 488 nm and the fluorescence was measured using a 530/30 nm and a 585/42 band-pass filter. At least 1 Á 10 4 single cells were analyzed by appropriate FSC/SSC gating.
For the determination of K m , cell suspensions of HEK-OCT1 and HEK-OCT3 cells as well as HEK-Co cells as control for unspecific uptake of ASP + were incubated with ASP + at increasing concentrations at 37°C for 5 minutes. Subsequently, two volumes of an ice-cold 200 μM solution of the standard inhibitor of OCT1 and OCT3, tetrapentyl ammonium (TPA) [46][47][48], in PBS/FCS were added to one volume of cell suspension. The mixture was immediately analyzed. The mean fluorescence intensities (MFI) were calculated using FlowJo V10, and the difference between total (HEK-OCT1 or HEK-OCT3 cells) and unspecific (HEK-Co cells) uptake was plotted as specific uptake against the concentration of ASP + . The linearity of ASP + uptake (1 μM) at 37°C was assessed between 30 seconds and 7 minutes (HEK-OCT1) or 9 minutes (HEK-OCT3), respectively. K m was calculated according to the Michaelis-Menten equation using GraphPad Prism 5.01.
The inhibition of ASP + uptake was measured to determine the IC 50 values of TPA and compounds 1, 2, 4 and 5. The cells were pre-incubated with the respective test compound at different concentrations at room temperature for 10 minutes. Afterwards, ASP + was added to a final concentration of 1 μM, and the samples were incubated at 37°C in the dark for 5 minutes. Subsequently, the cells were washed, re-suspended in ice-cold PBS/FCS and stored on ice in the dark until measurement.
The normalized mean fluorescence intensities of the OCT expressing cells, set to 100% in the absence of an inhibitor, were calculated by subtracting the mean fluorescence intensity of unspecific ASP + uptake into HEK-Co control cells. The normalized fluorescence intensities in the presence of an inhibitor were plotted against the logarithm of the concentrations of the test compounds to calculate IC 50 values. The "log (inhibitor) vs. normalized response-variable slope" equation of GraphPad Prism 5.01 was used for this purpose.

Imaging of Cellular ASP + Uptake by Confocal Laser Scanning Microscopy
HEK-Co, HEK-OCT1 and HEK-OCT3 cells were seeded into 8-well μ-Slides (Ibidi, Munich, Germany) (2 Á 10 4 cells/well) and allowed to attach for 48 hours. Prior to the staining procedure, the medium was replaced by PBS and the DNA probe Draq5 (5 μM) (Biostatus, Shepshed, UK) was added to each well. To monitor the inhibition of ASP + uptake, 200 μM of TPA or 15 μM of compound 5, respectively, were added and incubated at 37°C for 5 minutes. Subsequently, ASP + was added to reach a concentration of 1 μM, and the cells were incubated for 5 minutes in the dark. Imaging was performed with an Axiovert 200 M confocal microscope coupled to a Zeiss LSM 510 scanning device (Carl Zeiss, Oberkochen, Germany) using a Plan-Apochromat 63x/1.40 oil immersion objective. Draq5 was excited at 633 nm and fluorescence was detected using a 650 nm long-pass filter, ASP + was excited at 488 nm and detected using a 565/35 nm band-pass filter.

Determination of OCT1 and OCT3 Expression by Various Cancer Cells
The expression of SLC22A1 and SLC22A3 mRNA in different cancer cell types was analyzed by quantitative RT-PCR using the Roche LightCycler system. Total RNA was isolated using the NucleoSpin RNA Purification Kit (Macherey-Nagel, Dueren, Germany) according to the manufacturer's instructions. One μg of total RNA was reversely transcribed using the iScript Kit (Biorad, Munich, Germany) according to the manufacturer´s instructions. SLC22A1, SLC22A3 and β-actin mRNA levels were determined using the LightCycler System and the FastStart DNA Master SYBR Green I Kit (both from Roche, Mannheim, Germany). The primer pair for the amplification of the SLC22A1 cDNA fragment was oOCT1-RT.for (CTGCCTGGTGAATGCT GAGC) and oOCT1-RT.rev (ACATCTCTCTCAGGTGCCCG), for the SLC22A3 cDNA fragment oOCT3-RT.for (CAAGCAATATAGTGGCAGGGG) and oOCT3-RT.rev (CCTCAAAGGTGAGA GCGGGA) and for the β-actin fragment oActin.for (TGACGGGGTCACCACACACTGTGTGCC CATCTA) and oActin.rev (CTAGAAGCATTTGCGGTGGACGATGGAGGG). PCR was performed according to the manufacturer´s instructions with 0.5 μM of the respective sense and antisense primers, 4 mM MgCl 2 and 1-fold FastStart DNA Master SYBR green I mix in a total volume of 20 μL including 1 μL of the synthesized sscDNA. Cycling conditions were as follows: 10 min denaturation at 94°C, followed by 45 cycles of 10 s denaturation at 94°C, 15 s primer annealing at 64°C and 30 s of elongation at 72°C. The amount of β-actin, SLC22A1 and SLC22A3 cDNAs were determined using a serial plasmid dilution (pOCT1.31; from 10 6 to 10 4 fg) as amplification standard. The β-actin concentration, calculated in relation to the standard curve, was set to 100% and the respective SLC22A1 and SLC22A3 mRNA values are given as a percentage of βactin amplification.

Cytotoxicity of Bendamustine and Derivatives
The cytotoxicity of compounds 1-7 against tumor cells was determined both as an end point and kinetically. Additionally, the toxicity of 1, 2, 4 and 5 was determined in kinetic assays at spontaneously immortalized human keratinocytes (HaCaT), [49], as a model for "normal" cells. IC 50 values of compounds 1-7 (Table 1 and S2 Fig) were calculated after 96 hours of incubation and the cytotoxic drug effect was measured over a period of 5 days (Fig 2 and S3-S13 Figs). In case of the crystal violet assay, the kinetic approach allows the distinction between cytotoxic, cytostatic and cytocidal drug effects [40].
In contrast to the parent compound bendamustine, the derivatives 2-7 exhibited considerably higher potencies up to factors > 100 both, against the cancer cells investigated. With respect to the cytotoxic effects, compounds 2-7 fall into two groups: the alkyl esters (2, 3) and the mofetil ester 4 on one hand, and the basic heterocyclic esters 5-7 on the other hand. Compared to bendamustine, compounds 2-4 showed a 10-to 30-fold increase in potency, and 5-7 were 60-to 120-fold more potent than 1.
The results of the kinetic toxicity assays on HaCaT cells (Fig 2, S13 Fig) compared to the IC 50 values (Table 1) and the data from kinetic cytotoxicity assays on cancer cells (Fig 2 and  S3-S13 Figs) revealed a preferential toxicity against tumor cells, suggesting a more favorable "therapeutic index", in particular in case of the esters 4 and 5.
Data on the in vitro cytotoxicity of bendamustine are scarce. In the literature, for myeloma cells IC 50 values around 100 μM or even higher are reported [22,23]. A very recent study on several hematologic malignancies revealed IC 50 values between approximately 10 μM and 250 μM. Especially mantle cell lymphoma, Burkitt's lymphoma and T-cell acute lymphoblastic leukemia derived cell lines were relatively sensitive to bendamustine treatment [50]. The IC 50 value (approximately 50 μM) reported by Hiraoka et al. [50] for Jurkat cells is in good agreement with our data. Published plasma levels of 1 after intravenous administration (C max = 6 μg/mL (% 17 μM) [51]; C max = 11 μg/mL (% 31 μM) [52]) suggest that the chemotherapy with bendamustine must be considered ineffective in case of tumor entities showing IC 50 values in the two-to three-digit micromolar range in vitro. In this context the up to 100-fold antiproliferative activity of the bendamustine esters, in particular 5-7, suggest both, higher efficacy in case of malignancies for which the parent compound is approved and a possible extension of the scope of indications.

Induction of Apoptosis and p53 Expression by Compounds 1, 2, 4 and 5
Bendamustine was reported to trigger apoptosis [22]. In search for an explanation of the higher antiproliferative activity of the bendamustine esters compared to the parent compound, we determined early and late stage of apoptosis by flow cytometry (annexin V/propidium iodide staining) after treatment of Jurkat cells with 10 μM of compounds 1, 2, 4 or 5 for 6, 24 and 48 hours (Fig 3; S14 Fig). As becomes obvious from Fig 3, annexin V + /PI + cells, defined as secondary necrotic cells, amounted to approximately 10% of the total cell population, most probably resulting from sample preparation. Only a fraction of approximately 1% of the cell population was Annexin V -/PI + and considered necrotic. As expected from the results of the MTT assay, at a concentration of 10 μM, bendamustine had no effect compared to the control cells, regardless of the period of incubation. After 6 hours, the fraction of early apoptotic cells was less than 1% except for compound 5 (~2.5%; cf. Supporting Information, S14 Fig). Compounds 2 and 4 showed induction of apoptosis in around 10% of the cells after 24 hours of incubation. After 48 hours of incubation, 15% of the cells were in an early apoptotic state and a considerable fraction (20-25%) was secondary necrotic. In agreement with the results from the chemosensitivity assays, compound 5 was more potent, exhibiting a more rapid onset of action and a higher maximal response compared to 1, 2 and 4 regarding induction of apoptosis. Approximately 50% of the cells were either early apoptotic (20%) or secondary necrotic (30%) after 24 hours. Two days after treatment, only around 15% of the cells were viable, whereas the majority of the cells was secondary necrotic (70%). Bendamustine was reported to induce the expression of p53 [21,23,50]. As the derivatives of 1 were much more potent against large cell lung and colorectal cancer cells, we investigated the induction of p53 expression in NCI-H460 and HT-29 cells after 24 hours of incubation with compounds 1, 2, 4 and 5 (Fig 4). Especially in NCI-H460 cells, p53 expression was significantly induced by the treatment with 10 μM of compounds 2, 4 and 5, while 5 led to the most pronounced effect. Bendamustine also weakly induced the expression of p53, although at a tenfold higher concentration (100 μM). Interestingly, p53 was only detectable in nuclear extracts of NCI-H460 cells after treatment, whereas HT-29 cells exhibited a high constitutive expression of the tumor suppressor protein. Nevertheless, the treatment with compounds 1, 2, 4 and 5 increased the expression of p53 in HT-29 cells but, similar to NCI-H460 cells, compounds 2, 4 and 5 caused a higher expression level than 1. Another notable observation was a slight induction of p53 expression in HT-29 cells after treatment with 10 μM of 1, since proliferation assays revealed no toxic effect at this concentration. It is known from the literature that concentrations of ! 80 μM of bendamustine are required to produce a significant induction of p53 expression in various tumor cell types [22]. Both, the induction of p53 and apoptosis, correlated with the cytotoxic potency of bendamustine and derivatives.

Cellular Accumulation of Bendamustine and Derivatives
Higher cellular uptake of the neutral (2, 3) and basic (4-7) bendamustine esters compared to the parent compound could account for the increased potency in terms of antiproliferative activity, induction of apoptosis and p53 expression. Therefore, we performed HPLC analyses to determine the amount of cell-associated test compounds 1, 2, 4 and 5 at a concentration of 30 μM after 10 minutes of incubation. NCI-H460 and HT-29 cells were selected as examples of solid tumors, which surprisingly proved to be sensitive against bendamustine esters (Fig 5).  The recently reported validated HPLC method was applied [34]. To prevent hydrolysis of the nitrogen mustard group during sample preparation and analysis, cell lysis and deproteination were performed under acidic conditions (1 M perchloric acid, sonication). The esters were proven to be stable over the incubation period of 10 minutes in the presence of cells (S15 Fig). The cellular association of the internal standard umbelliferone did not significantly differ between both cell lines (p > 0.05). The amounts of cell-associated bendamustine (1) were extremely low (HT-29: 0.07 ± 0.011 nmol/10 6 cells; NCI-H460: 0.03 ± 0.001 nmol/10 6 cells). The ratio of cell-associated to the applied concentration of the test compound (accumulation factor) was 1:3 for NCI-H460 cells and 2:3 for HT-29 cells, indicating an incomplete uptake. In contrast, compounds 2, 4 and 5 revealed considerably higher cell-associated amounts, with 5 reaching the highest cellular concentrations. A pronounced cellular enrichment of 2, 4 and 5 was particularly observed in HT-29 cells. Compounds 2 (2.05 ± 0.39 nmol/10 6 cells) and 4 (1.59 ± 0.23 nmol/10 6 cells) revealed a 20-to 30-fold, and 5 (5.01 ± 0.34 nmol/10 6 cells) an approximately 70-fold higher cellular enrichment than bendamustine. Qualitatively, the cellular accumulation of the bendamustine derivatives was comparable in HT-29 and NCI-H460 cells, though at a lower level (factor of approximately three) in case of the latter. The amounts of cell-associated 2, 4 and 5 were 0.28 ± 0.06 nmol/10 6 cells, 0.59 ± 0.04 nmol/10 6 cells and 1.89 ± 0.14 nmol/10 6 cells, respectively.
The levels of cellular enrichment correlate very well with the antiproliferative activities, underlining a crucial role of the ester moiety depending on the chemical nature, covering a neutral group (2) or substructures with different degree of basicity (4 and 5). In particular, the contribution of the pyrrolidino group in 5, which is positively charged under assay conditions, becomes obvious from the increased cellular accumulation which is paralleled by the antiproliferative activity, the induction of apoptosis and p53 expression. The high cellular accumulation of 5 and the comparable effects of 5-7 concerning the toxicity suggest an important role of the basic substituent and the positive charge for the cellular association and thus for the toxicity. Apart from that, the significant difference between HT-29 and NCI-H460 cells might result from a different extent of diffusion and transporter-mediated uptake. Therefore, a possible contribution of organic cation transporters (OCT) was taken into account [53]. To test this hypothesis, functional studies on recombinant OCT1 and OCT3, expressed in HEK293 cells, were performed. Additionally, the expression of the respective transporters by the tumor cell types selected for cytotoxicity studies was investigated.

Effect of Bendamustine Derivatives on the Activities of OCT1 and OCT3
ASP + uptake by OCT1 and OCT3 expressing HEK293 cells Confocal laser scanning microscopy revealed specific uptake of the fluorescent substrate ASP + by organic cation transporter expressing HEK293 cells (Fig 6). OCT1 activity was lower than that of OCT3. The activities of both transporters were inhibited by compound 5 and the reference inhibitor TPA (Fig 6).
Determination of the affinities of OCT1 and OCT3 to ASP + as substrate. Flow cytometry was applied to determine K m values (Fig 7). The kinetics of ASP + -uptake and thus the increase in fluorescence were linear for both transporters (Fig 7A), allowing the determination of initial velocities (v 0 ) at different concentrations (Fig 7B). The affinity to ASP + was not significantly different for OCT1 (K m = 3.1 ± 0.1 μM) and OCT3 (K m = 2.7 ± 0.1 μM). By contrast, the maximal uptake rate for ASP + was significantly higher for OCT3 [v max = 5321 ± 89 MFI/(10 4 cells Á min)] than for OCT1 [v max = 2808 ± 78 MFI/(10 4 cells Á min)] (p < 0.001).
Expression of OCT1 (SLC22A1) and OCT3 (SLC22A3) by cancer cells. Previously, OCT1-mediated uptake of cytostatics in tumor cells was reported for chronic myeloid leukemia [56], chronic lymphocytic leukemia [57], and colon carcinoma [54], whereas OCT3 expression was associated with colorectal [58] and renal cancer [59]. We applied RT-PCR for the examination of OCT1 and OCT3 expression in all used cancer cell lines (Fig 9, S16 Fig). Whereas OCT1 was weakly expressed in SK-ES1 and SK-MEL3 cells (in both cell lines below 0.3%, related to β-actin expression) OCT3 expression was high in HT-29 cells (0.52%, proportional to β-actin expression), and clear bands were detectable upon analysis of Capan-1 (0.05%, proportional to β-actin expression) and LNCaP cells (0.14%, proportional to β-actin expression). The differential expression (Fig 9, S16 Fig) does not support the hypothesis that the chemosensivities of the investigated cancer cell types (Table 1) are primarily determined by OCTmediated uptake of the bendamustine esters.

Conclusion
Although the investigated compounds are sufficiently stable to act as antitumor agents on their own, it cannot be precluded that the esters are only prodrugs, allowing for increased intracellular accumulation of bendamustine. Having the N-Lost moiety in common, alkylating property is a characteristic feature of both, the parent compound and the derivatives. In concert with the   increase in apoptotic processes and elevated p53 expression, the data may be interpreted as a hint to a dual mechanism of action, in particular in case of the basic bendamustine esters.
Supporting Information S1 Fig