One-pot synthesis of reactive-oxygen species (ROS)-self-immolative polyoxalate prodrug nanoparticles for hormone dependent cancer therapy with minimized side effects

A new reactive oxygen species (ROS)-sensitive, self-immolative biodegradable polyoxalate prodrug based on the anticancer chemotherapeutic hormone analog diethylstilbestrol was synthesized via one-pot step-growth polymerization. The nanoparticles prepared from this prodrug undergo self-immolative degradation releasing the chemotherapeutic drug in ROS-rich environments, e.g. , in cancer cells. This new ROS self-immolative polyprodrug backbone eliminates the need for a linker between polymer chain and drug, resulting in a more speci ﬁ c drug release and minimized toxic side e ﬀ ects to non-ROS-producing cells as proven by in vitro experiments. The strategy enables re-utilization of a successful chemotherapeutic agent that has been clinically under-utilized due to dose-related side e ﬀ ects.


H NMR measurements of the polyoxalate prodrug degradation
To 10.0 mg of the polymer, 0.6 mL of DMF-d7 was added and the NMR tube was incubated at 37 °C. Afterwards, deuterated PBS buffer solution (pH 7.4, 160 µL) was added ( Figure 1A, top). Finally, 40 µL of H 2 O 2 was added to make a 1 mM solution, and the experiments proceeded for 24 up to 72 h ( Figure 1A middle and bottom). For the 1 H NMR measurements of the polymer incubated for longer times (e.g., 72h) the incubation was performed outside of the magnet (NMR tube Heat&Mix Apparatus, Labio a.s., CZ).

Size exclusion chromatography (SEC) analysis
The weight-average molecular weight (M w ), number-average molecular weight (M n ), and the respective dispersity Đ = (M w /M n ) were obtained by size exclusion chromatography (SEC) analysis. The SEC of the isolated polyprodrugs was performed at 25 °C with two PLgel MIXED-C columns (300 × 7.5 mm, SDV gel with particle S2 size 5 μm; Polymer Laboratories, USA) and with UV (UVD 305; Watrex, Czech Republic) and RI (RI-101; Shodex, Japan) detectors. N,N-Dimethylformamide (Sigma-Aldrich, Czech Republic) with LiBr (0.01 % v/v) was used as a mobile phase at a flow rate of 1 mLmin -1 . The molecular weight values were calculated using Clarity software (Dataapex, Czech Republic). Calibration with PMMA standards was used.

Scattering techniques
The dynamic light scattering (DLS) measurements were performed using an ALV CGE laser goniometer consisting of a 22 mW HeNe linear polarized laser operating at a wavelength  = 632.8 nm, an ALV 6010 correlator, and a pair of avalanche photodiodes operating in the pseudo cross-correlation mode. The samples were loaded into 10 mm diameter glass cells and maintained at 25  1 °C. The data were collected using the ALV Correlator Control software and the counting time was 30 s. The measured intensity correlation functions were analyzed using the algorithm REPES 1 resulting in the distributions of relaxation times shown in 2 ( ) equal area representation as . The mean relaxation time or relaxation frequency is related to the diffusion coefficient of the nanoparticles as , where is the scattering vector with n ( ) with the Boltzmann constant, the absolute temperature and the viscosity of the solvent. The dispersity of the nanoparticles was accessed by using cumulant analysis 2 of the correlation functions measured at 90° as: where is the amplitude of the correlation function. The parameter is known as the second cumulant and it  2 was used to compute the dispersity of the samples .

Preparation of the polyoxalate prodrug nanoparticles
A predetermined amount of PDEB1 polyprodrug (12.5 mg) was dissolved in acetonitrile (10 mL) and precipitated into a water solution (20 mL) containing 2 times the amount of Tween-80® (FDA-approved) stabilizing agent (25 mg). The solvent was evaporated and the polyprodrug NPs were concentrated to the desired volume (5 mL). In the case of rubrene loaded NPs, rubrene 2.5 wt% (per polymer wt) was added and dissolved in acetonitrile. PLA NPs (control) were prepared as PDEB1 aforementioned.

Diethylstilbestrol (DEB) amount in the polyoxalate prodrugs
The total amount of the chemotherapeutic DEB in the polyprodrug backbone was measured by high performance liquid chromatography (HPLC, Shimadzu, Japan) with a reverse-phase column Chromolith Performance RP-18e (100 x 4.6 mm), eluent water-acetonitrile with acetonitrile gradient 0-100 vol%, flow rate

Toxicity mechanism
DEB gets oxidized to a DEB-quihinone which binds covalently to the DNA, which is a DNA damaging mechanism that occurs in any cell type without distinction, 3 and is held responsible for the carcenogenicity of long-term In a concentration-dependent manner, after 24h incubation DEB caused changes in the chromosomal number (at 20 µM), disturbed the microtubuli network (at 50 µM) and raised toxicity (100 µM). 4 In all tested cell lines in this study, the PDEB1 polyprodrug NPs caused, next to the specific toxic effects described below, a general decrease in cell viability and growth (see Figure S7).

Prostate cancer cells
The PDEB1 polyprodrug was tested in two model cell lines for hormone therapy-resistant (PC3) and therapyresponsive (LNCaP) prostate cancer. The DEB-induced M-phase arrest, aneuploidy and apoptosis induction in a concentration range of 30 µM in PC3 cells, which was described by Robertson, et al. 5 They further reported that DEB equally triggered cell cycle arrest and apoptotic death in LNCaP and PC3 cells, via an estrogen receptorindependent mechanism. The microscopy studies of PC3 showed that, despite being resistant to hormone therapy, they were notably affected by the PDEB1 polyprodrug NPs. The cycle arrest and nuclear fragmentation occurred as described in the literature. 5,6,7 Similar effects were observed for the LNCaP cells, except that LNCaP, unlike PC3, also displayed an impairment of the cytoskeleton because the cells detached in vast numbers and forming lose clumps. This was demonstrated using a light microscope, where LNCaP cells were visualized without washing to avoid losing detached cells. Note that these samples (see Figure S7) contained Tween 80 and DMSO at maximal 5.3 µM resp. 0.01% concentration, i.e., the concentrations of both chemicals were below the toxicity threshold. Moreover, the overall concentration of Tween 80 was identical in the free DEB and the PDEB1 NPs samples, which effectively excludes Tween 80 as the reason for detachment. This cell detachment is therefore an impairment that will add to the long-term toxicity of PDEB1 NPs for LNCaP. Beyond the detachment, PDEB1-treated LNCaP also showed the same toxic effects as PC3, i.e., significant M-phase arrest S7 ( Figure S7). It is important to note that the cell death by mitotic arrest-induced apoptosis has a timeframe of several days. 5 Fluorescence imaging of the clustered detached LNCaP cells confirmed that after 3 days incubation with PDEB1 NPs, the detached LNCaP were still viable (propidium iodide staining negative, see Figure S8). Due to restrictions concerning cell density, the AlamarBlue TM assay was not incubated for longer than 5 days. But it is expected that, due to the slow nature of apoptosis induction via cell cycle arrest, the final toxicity of PDEB1 NPs on LNCaP and PC3 cells is even higher than what we were able to document in the range of 5 days.

Breast cancer cells
The apoptosis induction through estrogen-receptors (specifically ERα type I) is described as major mechanism of DEB apoptosis induction in breast cancer cells. 6 But it is generally acknowledged that DEB acts via several pathways to induce toxicity, and the DEB binding to other estrogen-related receptors in breast cancer cells is under discussion. 7 The fact that ERα type I acts in a genomic pathway and triggers apoptosis swiftly, after 24 h and not via cell cycle arrest, likely contributed to the high efficiency of PDEB1 versus MCF7. 6 On the other hand, the high numbers of G2/M arrested cells suggests the involvement of at least one more pathway -that induces apoptosis more slowly, in the range of several days. The MCF7 cells from all tested cell lines showed the most severe toxicity, where the vast majority of cells were damaged, accumulated in the mitotic phase or displaying nuclear aneuploidy (see Figure S7). Similar to the prostate cancer cells, the slow nature of cell cycle arrestinduced cell death indicates that the final toxicity of PDEB1 on MCF7 may be higher than can be measured after 5 days. In MCF7 the difference between DEB toxicity and polyprodrug toxicity was striking, and higher than in any of the other tested cell lines. The MCF7 cells were highly sensitive to the polyprodrug PDEB1 NPs.

Non-cancerous cells
In stark contrast to the MCF7 cells, in the non-cancerous HF cells the toxicity of PDEB1 was mostly visible in the decreased cell numbers, morphology change into a less elongated shape and fragmentation aneuploidy of the nucleus in several cells (DEB-induced aneuploidy in fibroblasts as reported by Ochi. 4 The general cell death (number of dead cells) was less excessive than in MCF7 and an increase of the G2/M cell population like in the cancer cell samples was not observed in the fibroblast samples exposed to PDEB1.
We can conclude that HF fibroblasts were in general less affected by the PDEB1 polyprodrug NPs in comparison with the used cancer cells. HF cells displayed decreased cell growth, shape changes and nuclear aneuploidy and fragmentation. The cancer cells displayed all this as well, but in addition also significant G2/M arrest, S8 detachment (LNCaP) and general higher cell damage. PDEB1 lowered the required dose in LNCaP and also in therapy-resistant PC3 cells, and it was particularly effective in the MCF7 breast cancer model system. Finally, while being highly efficient against cancer cells, towards non-cancerous cells the polyprodrug NPs was barely more toxic than the gold standard of biocompatible polymers polylactide, PLA (see Figure S10).

Chemiluminescence of rubrene in PDEB1 polyoxalate prodrug NPs in cells
The oxalate bonds are oxidized by H 2 O 2 to produce 1,2-dioxetanedione (or other high-energy intermediates which interact with an appropriate fluorophore to form an activated complex, the mechanism still not fully elucidated). After decomposition of the complex along with CO 2 release, the excited fluorophore decays to the ground state with fluorescent emission (Scheme S1). The MCF7 cells were selected to demonstrate the degradation of PDEB1 by chemiluminescence. The H 2 O 2 -triggered cleavage of the oxalate groups in the prodrug backbone includes an intermediate step, which in the presence of rubrene will cause a proportionate chemiluminescence emission of the dye molecule (see Scheme S1). The appearance of rubrene luminescence is therefore a direct sign that the degradation of PDEB1 NPs is taking place.  Scheme S1. Proposed mechanism for fluorescence from oxalate based polymers. [,8,9]