Using Affinity To Provide Long-Term Delivery of Antiangiogenic Drugs

Jan 27, 2017 - Antiangiogenic drugs encompass many of the different cancer drugs currently under clinical investigation. One of the drawbacks of ...
0 downloads 0 Views 2MB Size
Subscriber access provided by University of Colorado Boulder

Article

Using affinity to provide long-term delivery of anti-angiogenic drugs in cancer therapy Edgardo Rivera-Delgado, and Horst A. von Recum Mol. Pharmaceutics, Just Accepted Manuscript • DOI: 10.1021/acs.molpharmaceut.6b01109 • Publication Date (Web): 27 Jan 2017 Downloaded from http://pubs.acs.org on January 28, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Molecular Pharmaceutics is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`1

Using affinity to provide long-term delivery of antiangiogenic drugs in cancer therapy

Edgardo Rivera-Delgado and Horst A. von Recum Department of Biomedical Engineering Case Western Reserve University, 10900 Euclid Avenue, Cleveland OH 44106-7207

Corresponding Author: Horst A. von Recum, Department of Biomedical Engineering, Case Western Reserve University, Room 309 Wickenden Building, 10900 Euclid Avenue, Cleveland OH 44106-7207, USA Tel.: +1 216 368 5513; fax: +1 216 368 4969 email: [email protected]

Graphical Abstract

ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 32

`2

Abstract

Anti-angiogenic drugs encompass many of the different cancer drugs currently under clinical investigation. One of the drawbacks of anti-angiogenic therapy, though, is that upon cessation of drug treatment tumors can recur with an accelerated growth rate. In this study we investigate the capacity of using affinity interactions between a polymer made from cyclodextrin and four anti-angiogenic drugs: tranilast, SU5416, 2-methoxyestradiol, and silibinin, with the ultimate goal of creating delivery profiles on the order of anti-angiogenic processes (needing weeks, rather than hours of delivery). In these systems, release rate is dependent on affinity, so using in silico molecular docking studies followed by surface Plasmon resonance we determined that silibinin possesses the highest affinity among the drugs screened. Silibinin also showed a differential binding affinity among various cyclodextrins tested, with a greater affinity towards the larger molecular pocket of γ-cyclodextrin than for βcyclodextrin. Release studies confirmed this affinity to translate into a slower, more sustained release of silibinin. Similarly we found this trend in the release of tranilast. Then using U87 human glioblastoma cells in a ACS Paragon Plus Environment

Page 3 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`3

mouse xenograft model, we showed that affinity-based cyclodextrin polymers loaded with silibinin showed substantially longer release rates than non-affinity control polymers; however, both were capable of inhibiting tumor growth in the timeframe studied. From this work we showed three different, but chemically similar polymers each with a different release rate. Future work is on evaluating longer term tumor models where this longer release rate from affinity delivery systems might have additional advantages over polymers dependent only on diffusion.

Keywords:

anti-angiogenesis,

local

delivery,

cancer,

glioblastoma,

ACS Paragon Plus Environment

affinity,

cyclodextrin

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 32

`4

1. Introduction

In spite of decades of progress in therapies, cancer still remains one of the leading causes of death in developed countries, and is known as the “emperor of all maladies”. One such cancer, glioblastoma multiforme (GBM), is particularly fatal. While GBM cases are rare with less than 10,000 patients diagnosed per year, their prognosis is poor due to nearly universal tumor recurrence.1 Recurrence is further complicated by the side effects arising from the different treatment interventions. The median survival time with the best treatment: surgery combined with radiation therapy and chemotherapy, remains low at 14.6 months.2 While GBM therapy is often systemic, an advantage for studying local drug delivery in this disease is that it remains one of the few cancers with a clinically approved local depot for the delivery of anti-proliferative agents.3,4

Need for local delivery in the brain is partially due to the fact that distribution of therapeutic drugs to the brain following systemic distribution is hindered by both common transport limitations, and also the extra transport obstacle presented by the blood brain barrier. This extra barrier limits the amount and nature of drugs that can be used to treat patients. One well-studied exception is temozolomide; however, although this drug is preferred to treat gliomas, temozolomide has been linked to progression of low-malignancy gliomas into the more malignant GBM.5-7 One way to bypass the blood brain barrier is by the local (in-brain) delivery of chemotherapeutic agents. The primary device available clinically for delivering local therapeutics to brain cancer patients is the implantable wafer Gliadel.8 Gliadel delivers the cytotoxic drug carmustine (or BCNU) from a polyanhydride matrix. Drug release is controlled simply by the diffusion of the drug through the matrix, resulting in a burst of most of the drug at early time points and very little drug remaining to be delivered at later time points.

ACS Paragon Plus Environment

Page 5 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`5

Besides traditional cytotoxic or anti-proliferative cancer drugs, another group of agents under investigation are anti-angiogenic drugs.9 Anti-angiogenic drugs work by preventing the growth of vasculature to the rapidly proliferating tumor. These agents often work through causing endothelial-specific apoptosis which leads to, among other things, sensitization of tumors to standard radiotherapy and chemotherapy.10-13 Of the current investigational drugs, pharmaceutical agents targeting the blood vessel biology represent a significant portion of the drugs under clinical trials.2

A set of four commercially available anti-angiogenic drugs examined in this study to represent the range of drugs that could be delivered locally in the treatment of any solid tumor, with GBM used as a specific cancer model. Tranilast, SU5416, 2-methoxyestradiol(2ME2) and silibinin all have been shown previously to inhibit tumor blood vessel growth. They represent a range of chemistries from hydrophobic to hydrophilic (Figure 1). Of these drugs, silibinin possesses both tumor anti-proliferative properties and anti-angiogenic effects. While silibinin has obvious therapeutic effect, and various mechanisms of action have been implicated, no single mechanism has emerged that fully describes silibinin’s action. Among the putative mechanisms is cyclindependent kinase (CDK) inhibition, through tumor necrosis factor alpha (TNFα). Silibinin shows inhibition of angiogenesis through the downregulation of the vascular endothelial growth factor receptor, Flt-1, when provided in the range of 25-100 µM in vitro.14-16 The drug 2-methoxyestradiol has been shown to inhibit human vascular endothelial cells and other cancer cells. It particularly inhibits endothelial cells with an IC50 of 0.77 µM for proliferation and 1.78 µM for inhibition of migration.17 Attaining these effective drug concentrations in vivo is difficult in part due to the low bioavailability after oral administration of silibinin (70 µM when in its best dosage form: complexed with lecithin/phosphatidylcholine).18,19 2 Another drug, tranilast, has been shown to inhibit the growth of human dermal microvascular endothelial cells tubes in a Matrigel plug assay when provided in the range of 12-100 µg/ml with an IC50 of 136 µM.20 The third ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 32

`6

drug, SU5416, like thalidomide, is derivative of the IMiD class that exerts its action through the Flt-1 receptor. SU5416 shows this effect when provided in the range of 0.1 µM to 100 µM.21 In general drug dosing is crucial since reports of anti-angiogenic therapy have shown mixed success in part due to recurrence and adaption during and upon therapy cessation, some of which could be attributed to incorrect delivery profiles.22,23

Figure 1. Chemical structure of anti-angiogenic agents Four anti-angiogenic small molecule drugs capable of inclusion formation with cyclodextrins were chosen, selecting a broad range of chemistries

ranging from

hydrophobic (2ME2) to

hydrophobic (tranilast, SU5416) to hydrophilic (silibinin).

ACS Paragon Plus Environment

moderately

Page 7 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`7

Affinity-based delivery of anti-angiogenic drugs offer the benefit of tailoring the release of the drug based on the strength of binding of the drug towards a host molecule in the matrix.24,25 We have shown that when this affinity is strong it becomes the rate-limiting step in the drug delivery process, allowing generation of delivery profiles not possible through mechanisms depending on diffusion alone.26 During the burst phase of a diffusiononly system much of the drug released can be wasted if the delivery amount is above the therapeutic threshold, and the amount can even reach the toxic threshold. Additionally with these diffusion-only systems, at later time points they have less drug available, where the amount that is left to be delivered can be ineffective if below the therapeutic threshold. With a drug delivery system using affinity to control the rate-limiting step, the initial burst phase of delivery is greatly reduced, preserving more drug to be delivered at later time points. We have observed months of bioactivity from affinity-delivery in other applications, where chemically similar diffusion only controls failed within days. In the case of the delivery of antibiotics this profile is crucial to ensure complete bacterial killing, and to prevent formation of drug-resistant strains from sub-threshold doses.

Herein we speculate that a similar profile is necessary for anti-angiogenic therapies. Given some reports of recurrence following cessation of anti-angiogenic therapy, approaches that effectively extend the therapeutic window of these anti-angiogenic agents to biologically relevant time scales (on the order of weeks) may allow for a delay in aggressive tumor recurrence. Also using the lower initial drug concentrations possibly through an ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 32

`8

affinity-based system we could delay generation of pathway resistant mechanisms. In future work we predict that when combined with traditional antitumor agents affinity-based delivery of anti-angiogenic drugs may provide a synergistic mean for extending the life of the patients. In this study, we use the affinity-based delivery approach to screen the release of several anti-angiogenic agents. The strongest theoretical binding is initially modeled and then confirmed experimentally by means of surface Plasmon resonance. Finally, the drug with the best affinity-based delivery profile is compared to diffusion-only delivery for its effectiveness at slowing the growth rate of tumors.

2. Experimental

2.1. Materials

Cyclodextrin macromonomers lightly crosslinked with epicholohydrin were used as pre-polymers and purchased from Cyclolab (Budapest, Hungary). Low molecular weight dextran, the chemically similar nonaffinity control pre-polymer was obtained from Polysciences (Warrington, PA). Unless noted otherwise CM5 Chips, HBS-N (a HEPES balanced salt solution pH 7.4) and other surface Plasmon resonance (SPR) reagents were purchased from GE Healthcare (Pittsburgh, PA). Phosphate buffered saline (PBS), ethanolamine, 1-ethyl3-(3-dimethylpropyl)-carbodiimide (EDC), 2-(N-morpholino)ethanesulfonic acid (MES) were from Fisher Scientific(Pittsburgh, PA). The anti-angiogenic drugs SU-5416, silibinin and tranilast, the crosslinking reagent hexamethylene diisocyanate and the SPR reagents n-hydrosuccinimide, 6-amino-β-cyclodextrin and 6-amino-γcyclodextrin were from Sigma-Aldrich (St. Louis, Mo). 2-methoxyestradiol (2ME2) was from Millipore (Bellerica, MA). MTS cell viability assay reagents were from Promega (Madison, WI). Cells were obtained from American Types Culture Collection (ATCC) (Manassas, VA). All other materials were obtained from Fisher Scientific at the highest purity available. ACS Paragon Plus Environment

Page 9 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`9

2.2. Strength of Binding Prediction

Structure data files from all drugs were downloaded from Pubchem, converted to PDBQT format with openBabel and energy minimized within the PyRx software. Each drug was loaded into PyRx as a ligand and βcyclodextrin(β -CD) or γ-cyclodextrins(γ -CD) were used as host. The AutoDock Vina algorithm was used to determine the strength from the resulting interaction.27,28 Affinity results are reported in µM where a lower number represents a stronger predictive binding between the drug and the respective cyclodextrin.

2.3. Surface Plasmon Resonance

In order to determine the binding strength between β-CD and silibinin we performed surface Plasmon resonance (SPR) with a BIACORE X-100 system on CM5 chips. In one channel of the CM5 chip a 10 mM solution of 6-amino-β-cyclodextrin or 6-amino-γ-Cyclodextrin in HBS-N was conjugated onto the CM5 carboxymethylated surface by first activating the surface with an aqueous mixture of 0.4 M EDC and 0.1 M NHS. Subsequently, a 10 mM solution of 6-amino-β-cyclodextrin or 6-amino- γ - cyclodextrin in HBS-N buffer was passed through the chip in the experimental channel. The remaining unconjugated functional groups were deactivated by flowing 3 M ethanolamine over the chip. To distinguish specific affinity interactions from nonspecific interactions, a second control channel was treated identically but without the cyclodextrin containing solution, using the chips underlying carboxylmethyl dextran as the chemically similar non-affinity substrate. Decreasing concentrations of silibinin [1000 µg/ml-62.5 µg/ml] in drug release buffer were passed over the chip with a contact time of 7 minutes. The resulting sensorgram were recorded and analyzed with the BIACORE software. The SPR surface was regenerated with PBS supplemented with 50% (v/v) DMSO which we and others have previously seen to be suitable to dissociate remaining, bound drug. The results are reported ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 32

`10

as the dissociation constant (KD) with the respective Rmax and χ2 for goodness of fit. Values below 10% of the Rmax are considered significant as per BIACORE protocols.

2.4. Polymer Synthesis and Swelling Characterization

A total of 1.25 grams of pre-polymer (either β-cyclodextrin or γ-cyclodextrin macromonomers or low molecular weight dextran (Dex)) were separately dissolved by stirring the pre-polymers into 5 ml of dimethyl sulfoxide (DMSO). Hexamethylene diisocyanate (HDI) totaling 158 µl was added to the polymer mixture for a final ratio of glucose to HDI of 1:0.16. The mixture was then heated in a 70 mm Teflon dish to 70 oC in a bell jar oven for 4 hours. The polymers were washed in 5 steps of 24 hours each; first in DMSO, followed by a wash in equal amounts of DMSO and ultra pure water and finally in ultra pure water three times. The swollen polymers were cut into disks with a 8mm steel punch and left to air dry for 3 days. Polymer disks were then swollen in either DMSO or PBS to determine the swelling ratio in the respective loading and release solutions. Swelling was calculated from the difference between wet and dry weight and normalized by the dry weight volume as in Eq. 1.

.   =

  ℎ −   ℎ   ℎ

2.5. In vitro Drug Loading and Release

Single polymer disks were weighed and placed in vials with differing drug formulations. In order to maximize loading into the devices, solutions of the highest readily soluble amount of each drug were prepared. Specifically, DMSO solutions of 2ME2 (5%), SU5416 (3.5%), silibinin (20%) and tranilast (5%) w/w ratios of drug to polymer content were used to load the polymer disks. Polymer disks were left to load for 2 days, briefly ACS Paragon Plus Environment

Page 11 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`11

rinsed to remove surface adsorbed drug with water, and then air-dried for 3 days. For complete removal of the loading solvent the formulations were then further dried in a vacuum top oven at 50 oC at 0.1 KPa until the weight of the disk remained constant. Four samples were allocated to each drug-polymer condition. For loading determination drug-loaded disks were extracted in 5ml DMSO, and extract evaluated by spectrophotometry (Tecan Safire, Austria) until there was no change in the respective absorbance wavelength after subsequent solvent replacement. To measure release kinetics in a biologically relevant hydrophobic sink, the different drugpolymer formulations disks were placed in a 3 ml release solution consisting of PBS supplemented with 0.25% (v/v) Tween-80 and 0.25% (v/v) DMSO. Samples were shaken in a tabletop incubator at 37 oC with gentle agitation at 100 rpm. Aliquots totaling 1ml of the release solution were taken at different time intervals and replenished with the same amount of fresh release solution. Drug content is obtained by measuring the absorbance of the aliquots from their standard curves. Absorbance of SU5416 was measured at 440 nm, silibinin was measured at 290 nm, and tranilast was measured at 330 nm. Absorption of 2-methoxyestriadol is measured at 203 nm, which had a significant spectral overlap with the release solution, so 2ME2 release was only indirectly determined by bioactivity studies below.

2.6. Scratch Wound Assay

In order to quantify the anti-migratory bioactivity of the loaded drugs being released from our affinity-based polymers we performed a scratch wound assay using C166 endothelial progenitor cells, since progenitor cells are more likely to contribute to vascular remodeling than mature endothelial cells. Based on the modeling and SPR studies, silibinin and 2-methoxyestriadol were predicted to show the greatest difference between affinitybased and non-affinity delivery systems, so we assessed those two drugs to confirm the anti-migratory effect of their released drug. Polymer disks prepared as in the drug release section above were used, except with the addition of briefly rinsing them in ethanol and exposing the polymers to UV light for 10 minutes per side in a ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 32

`12

tissue culture cabinet to sterilized them. Polymers were then placed in 200 µL complete media supplemented with 0.1% (v/v) DMSO. At this concentration DMSO serves as a hydrophobic sink but is below the cytotoxic threshold. Daily release aliquots into this cell medium were taken and replaced with fresh medium to track longitudinal bioactivity. These release medium aliquots were kept frozen at -20 oC until the day of the cell experiments. Frozen samples were determined to have equivalent activity to non-frozen, fresh samples (data not shown). For the cell experiments, C166 endothelial progenitor cells were seeded at 70,000 cells/cm2 into a 96 well plate for 8 hours in 100 µl of serum starvation media containing Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 0.5% (v/v) Fetal Bovine Serum (FBS) and 1% (v/v) Penicillin/Streptomycin. After starvation, cells were scratched perpendicular to the flat surface with a p200 pipette tip. A total of 100 µL from each daily release samples of each polymer condition was used to treat each individual well. Each polymer/drug condition was run in triplicate. Images of the scratched cell monolayers were taken at 10x magnification right after scratching and 24 hours after the initial scratch. Migration was quantified by TSCRATCH software as the percent of the image open after 24 hours divided by the initial wound area as described previously.29 Statistical significance was established by Student’s t-test analysis of the responses as provided by the TSCRATCH software.

2.7. Cytotoxicity Assay

In addition to acting through migration inhibition, anti-angiogenic drugs could act through inhibiting proliferation or even causing cell death. Of particular concern would be in diffusion-only systems where high doses initially can far exceed therapeutic thresholds and reach into toxic thresholds. We tested whether the release was having an anti-proliferative bioactivity (in addition to anti-migratory activity above) by performing a longitudinal MTS assay. Based on both the success of the scratch assay and its mechanism of action, silibinin was the only drug tested. All polymers were prepared as previously described and cell culture handling was as ACS Paragon Plus Environment

Page 13 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`13

in the scratch wound assay methods above. C166 endothelial progenitor cells were loaded into 96 well plates at a concentration of 5,000 cells per well to which the daily release aliquots were added as described above. After 48 hours of incubation media release treatment 20 µL of the MTS solution was applied to each well and absorbance was recorded at 450 nm as per the manufacturer instructions. Results are reported as the average of each day with their respective standard deviations of each polymer group normalized by the value of the no treatment control for each group.

2.8. Glioblastoma Xenograft Model

Final evaluation of these delivery systems was performed using a xenograft tumor model. Specifically human glioblastoma U87 cells were cultured in DMEM supplemented with 2 mM L-glutamine, 100 units/ml penicillin G sodium, 100 µg/ml streptomycin sulfate, 0.25 µg/ml amphotericin B, 1 mM sodium pyruvate and 10% fetal bovine serum prior to use in animal studies. Tumors were induced via subcutaneous injections of 0.5 x 106 U87 cells into the right flanks of 6-8 week old female athymic nude mice. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee (protocol# 2011-0171). Of 10 injected, 7 mice developed comparably sized, palpable tumors. These mice were randomly divided onto 3 groups. The first group had 2 animals implanted with polymer implants without drug, the second group had 2 animals implanted with silibinin-loaded β-CD polymers and the third group had 3 animals implanted with silibinin loaded dextran polymers. Tumor size was tracked by means of caliper measurement of the tumor size along the three different axes from the day of polymer implantation. Tumor volume was estimated based on digital caliper measurement. Tumor shape was assumed to be a hemiellipsoid; volume (V) was calculated by Eq 2.



. 2  =  ℎ

ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 32

`14

where l, w, and h, represent the length, width, and height of the tumor. Animals were sacrificed at the first sign of ulceration, significant body weight loss or any other major physical distress. The results are presented as tumor volume per day after polymer implantation with its respective standard error.

Statistical Analysis

For the scratch assay a t-test was performed as provided by the TSCRATCH software as described above. All other statistical analysis was performed in R. A two-way ANOVA was used to determine the main effects in polymer swelling. For the cumulative release experiments the average of three separate gels in solution is provided with its respective standard deviation. We chose an alpha of 0.05 as a significant threshold for all experiments.

3. Results

Table 1. Predicted dissociation constants based on molecular docking models

Virtual screening of different anti-angiogenic drugs as performed by Autodock Vina. A lower dissociation constant represents a stronger theoretical binding between the drug and the cyclodextrin. Of the screened drugs silibinin shows the strongest theoretical binding for both cyclodextrins. Past studies show strongest differences between affinity-based and non-affinity systems when drugs show binding in the low micromolar range. ACS Paragon Plus Environment

Page 15 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Pharmaceutics

`15

3.1. Anti-angiogenic drugs show a range of affinities to cyclodextrins in molecular modeling.

Out of the panel of anti-angiogenic drugs examined in this study, we observed the highest predicted affinity from silibinin. In general, the predicted affinity of the screened drugs was higher for β-cyclodextrin than for γcyclodextrin (Table 1). Silibinin showed the strongest predicted affinity for both. When examining affinity differences between γ and β cyclodextrins, tranilast showed the biggest difference in the two predicted KD’s. Only SU5416 had comparable predicted affinity for both γ and β cyclodextrins conditions.

3.2. Silibinin affinity to cyclodextrin confirmed using surface Plasmon resonance

Table 2. Dissociation constants determined by surface Plasmon resonance SPR parameters for the interaction between silibinin and the different cyclodextrins. The SPR parameters are the results of a Langmuir 1:1 interaction model fitting between the drug and cyclodextrin. A lower dissociation constant represents a stronger interaction between the drug and the respective cyclodextrin. It is likely that the greater pocket size of γ-cyclodextrin allows for better interaction with silibinin as compared to βcyclodextrin. The Rmax value of γ-cyclodextrin is also significantly higher than for βcyclodextrin which may indicate more inclusion complex formation at equilibrium.

In addition to predicting affinity through modeling, we also directly assessed affinity by surface Plasmon resonance. As predicted above, silibinin displayed the strongest binding affinities, again having a preference towards γ-cyclodextrin over β-cyclodextrin. In this case, the difference was much greater than predicted, with a ACS Paragon Plus Environment

Molecular Pharmaceutics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 32

`16

3 orders of magnitude lower KD towards the γ-cyclodextrin conjugated chip than the β-cyclodextrin chip (Table 2). All reported KD values were within the model confidence interval as evidenced by Chi2 values below 10% of the maximum SPR response.

3.3. Chemically similar, dextran and cyclodextrin polymers show slight differences in swelling.

Figure 2. Swelling of cyclodextrin and dextran polymers depends on solvent identity. Polymers used in these studies were evaluated for their swelling properties in loading (DMSO) and release (PBS) solutions. A difference in swelling among the polymers can contribute to the overall drug-loading process. A two way ANOVA performed on the data distinguished the highest contribution to swelling is provided from the solvent type, followed by the polymer type (p value < 0.01 for both factors). Dextran polymers swell more in both solvents when compared to βcyclodextrin polymers, which partially explains the difference in drug loading behavior between the two polymer types. Dextran polymers swelled slightly more than cyclodextrin polymers in both DMSO and PBS. Both polymer matrices swelled at least twice as much in DMSO than in PBS (Figure 2). A two way ANOVA analysis of the data demonstrated that both the solvent choice and polymer type have a significant contribution in the swelling of the materials with the solvent choice having the greatest effect in swelling (p value