Conversion of RNA aptamer into modified DNA aptamers provides for prolonged stability and enhanced antitumor activity

Abstract

Aptamers, synthetic single-strand oligonucleotides with a similar function to antibodies, are promising as therapeutics because of their minimal side effects. However, the stability and bioavailability of the aptamers pose a challenge. We developed aptamers converted from RNA aptamer to modified DNA aptamers targeting phospho-AXL with improved stability and bioavailability.

Based on the comparative analysis of a library of 17 converted modified DNA aptamers, we selected aptamer candidates, GLB-G25 and GLB-A04 that exhibited highest bioavailability, stability and robust anti-tumor effect in vitro experiments.

Backbone modifications such as thiophosphate or dithiophosphate and a covalent modification of the 5’-end of the aptamer with polyethylene glycol optimized the pharmacokinetic properties, improved the stability of the aptamers in vivo by reducing nuclease hydrolysis and renal clearance, and achieved high and sustained inhibition of AXL at a very low dose.

Treatment with these modified aptamers in ovarian cancer orthotopic mouse models significantly reduced tumor growth and the number of metastases. This effective silencing of the phospho-AXL target thus demonstrated that aptamer specificity and bioavailability can be improved by chemical modification of existing aptamers for phospho-AXL. These results lay the foundation for translation of these aptamer candidates and companion biomarkers to the clinic.

Introduction

Aptamers are synthetic single-strand DNA or RNA oligonucleotides that act as synthetic antibodies to bind and regulate the activity of proteins on the basis of their tertiary structural interactions ^{1, 2}. Aptamers are designed and selected through repeated rounds of selection called Systematic Evolution of Ligands by Exponential Enrichment (SELEX) ^{3, 4}, involving separation and repetition steps to gradual increase the selection pressure and resulting in final sequences that highly specific binding to their target (proteins, nucleic acids, small molecules, or cells) through non-covalent interactions (electrostatic, hydrophobic, and three-dimensional (3D) structural interactions) ⁵. In essence, aptamers can be considered “synthetic antibodies” that can act like monoclonal antibodies and bind to the target with high specificity and affinity; unlike antibodies, however, aptamers show little or no immunogenic effects ⁵. Aptamers are easy to manufacture and they can be chemically modified and/or conjugated with other molecules to increase their specificity and effectiveness. For these reasons, aptamers have been used as tools for several therapeutic approaches such as biosensing, drug delivery, and theranostics ¹.

Some limitations to the use of aptamers as therapeutic agents include rapid degradation by nucleases in body fluids and tissues ⁶, removal by renal filtration ⁶, bioavailability ⁷, and cross-reactivity ⁸. Because most of the limitations are based on the structural stability of the aptamer and aptamer-target complex, which in turn is governed by the overall charge and reacting groups on the aptamer, chemical modifications of either the nucleotides or the backbone (thus altering the binding characteristics) could overcome the limitations and increase the stability, specificity, bioavailability and functional reactivity of the aptamer.

AXL is a receptor tyrosine kinase that belongs with its homologs Tyro3 and Mer in the TAM (Tyro3-AXL-Mer) receptor kinase subfamily ⁹. The AXL receptor functions as a sensor for extracellular ligands. Activation mechanisms involve the stimulation of the extracellular domain of receptors with bridging protein ligands. The main ligand of AXL is growth arrest-specific protein 6 (Gas6), a γ-carboxylate protein that binds with high affinity to the AXL receptor, inducing the dimerization and auto-phosphorylation of tyrosine residues ¹⁰. This leads to recruitment, phosphorylation, and activation of several downstream proteins involved in the regulation of survival, growth, differentiation, adhesion, proliferation, and motility ¹¹. AXL is ubiquitously expressed and has been identified in a wide range of organs and cells, including monocytes, macrophages, and endothelial cells in the heart, skeletal muscles, liver, and kidneys ¹². Overexpression or increased activity of Gas6/AXL interaction has been reported in many human cancers, such as prostate ¹³, esophageal ¹⁴, thyroid ¹⁵, breast ¹⁶, lung ¹⁷, liver ¹⁸, and ovarian ¹⁹ cancer and glioblastoma ²⁰, and it has been associated with poor prognosis ^{21, 22}.

We have reported that AXL overexpression in ovarian cancer, the most lethal gynecological malignancy, is associated with reduced overall survival ²³. Silencing of AXL would lead to the overall decrease in the ovarian cancer tumor growth and an increase in survival. Therefore, there is a need for novel strategies for silencing and blocking AXL signaling pathway. A wide range of AXL inhibitors has been described in the literature, including promiscuous small molecule kinase inhibitors ²⁴ or monoclonal antibodies ²¹. Most of the studies to date use RNA aptamers for targeting. We here report the synthesis and testing of converted 2’-fluoro pyrimidine (2’-F Py), dithiophosphate modified DNA aptamers targeting AXL receptor, using an RNA aptamer sequence, named GL21.T as starting point, to demonstrate the effect of our modifications on the increase in the therapeutic potential ²². The rationale to convert the RNA aptamer to DNA aptamer was to improve the stability of the RNA aptamer. Unmodified DNA aptamers by themselves are much more stable than RNA aptamers due to the removal of the 2’-hydroxyl group present in RNA and replacement with a 2’-proton. Once one begins substituting the 2’-position with a fluorine atom, three of the four DNA and RNA bases are in fact identical (2’-F-dA = 2’-F-A; 2’-F-dC = 2’F-C; 2’-F-dG = 2’-F-G). The fourth base, 2’-F-dU, is a hybrid that can be thought of as a 2’-F-substituted U or as a U-substituted 2’-F-dT. The F-substitution provides the flexibility and H-bonding properties of an RNA sugar while also providing the increased stability of a DNA sugar. Considering these facts our root aptamer is then a DNA/RNA hybrid. At the outset of this project, we had in mind to develop first aptamers with monothioate substitutions and later aptamers with selectively placed dithioates. At that time, off-the-shelf reagents were only available for creating dithioate DNA bases (not RNA). Thus leading us to use selectively dithioated dA and dG DNA bases. Later, monothioated RNA reagents came on the market, and have since disappeared again. Monothioated substitutions can be incorporated easily into either RNA or DNA using a thio-oxidation step during oligo synthesis, so that 2’-F-monothioated oligos of either RNA or DNA can be easily synthesized.

Moreover, we demonstrate here that of all the backbone modifications such as thiophosphate or dithiophosphate, in which non-bridging oxygens are substituted with sulfur atoms, increase the aptamer relative affinity to basic groups in proteins and overcome degradation by nucleases ²⁵ while enhance the specificity of the aptamers. Polyethylene glycol (PEG) was shown to improve the stability of the aptamers in vivo by reducing the nuclease hydrolysis and renal clearance ²⁶. However, studies have shown that using PEGylated liposomes the presence of anti-PEG antibodies could induce clearance of subsequent doses of PEGylated substances ²⁷. It is worth noting that the anti-PEG antibodies could lead to decrease clinical effectiveness. Thus, it may be necessary to test patients for PEG antibodies prior to treatment with PEGylated substances ^{2, 27}.

Results

GLB-G25 and GLB-A04 best reduce the expression of p-AXL among the 17 dithiophosphate modified DNA aptamers.

We redesigned converted modified DNA aptamers to increase and surpass the binding, bioavailability and stability of the existing aptamer by introducing four different modifications. We converted an RNA aptamer into DNA aptamers. First, we modified by adding 2’-F Py to GLB-B0 that represents the starting sequence from which we generated the library (which was composed of 17 different modified DNA aptamers). Monothiophosphate groups were added in seven positions (1, 3, 10, 17, 25, 31, and 33) of the GLB-B0 sequence. To improve the stability in serum, bioavailability, and pharmacokinetic properties of the converted DNA aptamers, we made further modifications consisting of dithiophosphate groups on the phosphate groups of adenine or guanine bases in specific positions of the sequence and PEG on the 5’ end (Figure 1A). We selected two DNA aptamers, GLB-G25 and GLB-A04, from among the library of 17 dithiophosphate DNA aptamers on the basis of their biological activity against ovarian cancer cell lines.

Figure 1: Aptamer selection is based on the inhibition of p-AXL.

(A) Representative table of the 17 different modifications. Dithiophosphate modification is represented by the number 6 for adenine and 8 for guanine. (B) Unstimulated expression of total AXL in ovarian cancer cell lines. β-actin was used as an endogenous control. (C) AXL mRNA expression of in OC cells by analysis of Cancer Cell Line Encyclopedia (CCLE) database. CCLE converted raw Affymetrix CEL files to a single value for each probeset using robust multi-array average (RMA) and quantile normalization. (D) Aptamer selection was based on the inhibition of p-AXL. GLB-G25 and GLB-A04 were selected among the modified DNA aptamers on the basis of p-AXL reduction, as determined by Western blot analysis. Gas6-stimulated (200 ng/mL) SKOV3.ip1 cells were treated with the 17 modified aptamers for 3 hours at a concentration of 0.4 μM. A scramble aptamer was used as a negative control. β-actin was used as a loading control. Data are presented as means ± standard deviation of three independent experiments. *P < 0.05; NS= No statistical significance compared with the scramble aptamer.

The expression of the AXL receptor in ovarian cancer cell lines was examined by Western blot analysis. HeyA8, SKOV3.ip, and OVCAR5 cell lines showed increased AXL expression, whereas A2780 and FUOV1 showed reduced levels (Figure 1 B).

We also analyzed AXL mRNA expression in ovarian cancer cell lines by the Cancer Cell Line Encyclopedia (CCLE) database. HeyA8, OVCAR8, ES2, and SKOV3.ip1 cell lines highly expressed AXL in among the other OC cells (Figure 1 C). Based on these data, we selected SKOV3-ip1 and OVCAR5 for further in vitro and in vivo experiments.

To select the best sequences among the 17 2’-F Py dithiophosphate modified DNA aptamers, we performed a screen for phospho-AXL (p-AXL) down-modulation in the presence of Gas6, the major ligand of the AXL receptor, in ovarian cancer cell lines (SKOV3.ip1, Figure 1 D; HeyA8, Figure S1). We observed that some of the sequences (G03, G11, G21, A22, and A27) did not show any effect on p-AXL expression in the SKOV3.ip1 ovarian cancer cell line as compared with the scramble sequence (Figure 1 D left panel). p-AXL expression was consistently reduced with GLB-G25 (45%) and GLB-A04 (62%), indicating that these were the best sequences among the 17 modified sequences. From analysis of three independent experiments (Figure 1 D right panel), we found other potential aptamer candidates (A17, A08 and A24) that we will explore in the future as potential alternatives to the selected ones.

Characterization and comparison of predicted three-dimensional structures of converted modified DNA aptamers.

To predict the secondary structures in our molecular modeling, we used the MFold server ²⁸ which showed that the aptamers fold into hairpin-like structures with four regions: one head/loop region, one stem region, and two single-stranded ends (Figure 2 A). Next, we generated 3D models of RNA and DNA aptamers based on the predicted secondary structures using the MFold server (Figure 2 B). The position of the 2’-F Py modification along the sequence is represented by green dots for GL21.T, GLD-1, GLB-G25, and GLB-A04. The dithiophosphate groups are highlighted with a red rectangle and monothiophosphate is represented with yellow dots. These models suggest that the translation and modifications (2’-F Py and dithio-) had no effect on the secondary and tertiary structures of GLB-G25 and GLB-A04 aptamers and they maintained similar 3D structures as GL21.T and GLD-1, first generation monothiophosphate modified anti-AXL aptamer ²³.

Figure 2: 2’-fluoro pyrimidine (2’-F Py) dithiophosphate modifications do not have any effect on the two- and three-dimensional structure of aptamers; tightly bind the extracellular domain of AXL receptor, down-regulate p-AXL compared withGL21.T and GLD-1 and have antitumor activity in vivo in a SKOV3.ip1 ovarian cancer mouse model.

(A) Predicted two-dimensional structures of GL21.T (left) and GLB-A04 (right) according to MFold software. (B) Three-dimensional structure of GL21.T (RNA sequence), GLD-1 (converted monothiophosphate DNA sequence), GLB-G25, and GLB-A04 (converted dithiophosphate DNA sequences) according to MFold software. Green dots represent 2’-F Py modification along the sequence and yellow dots are the thiophosphate and dithiophosphate modifications. (C) GLB-G25 and GLB-A04 binding curve obtained by MicroScale Thermophoresis (MTS), scramble aptamer used as negative control. Error bars represent the standard deviation of three independent experiments in (two technical repeats each). (D) Gas 6-stimulated (200 ng/mL) SKOV3.ip1 cells were treated with 0.4 μM of GL21.T, DNA unmodified aptamer, GLD-1, GLB-G25, and GLB-A04. GLB-G25 and GLB-A04 strongly reduced p-AXL expression at 0.4 μM concentration compared with GL21.T, DNA unmodified and GLD-1 aptamers. A scramble aptamer was used as a negative control (β-actin was used as a loading control). (E) Scheme of treatment for SKOV3.ip1 orthotopic animal models: 7 days after cell injection, mice were treated twice per week with 1600 pmol [0.6 mg kg-1 intravenous (i.v.)] aptamers for 5 weeks. (F) Mouse body weight analysis for the SKOV3.ip1 orthotopic animal model. No significant differences were observed between the groups, indicating no toxicity. (G) Tumor weight. (H) Numbers of nodules. Data are presented as box and whisker plots, whiskers extend from minimum to maximum values. ****P < 0.0001,**P < 0.001; *P < 0.05 (n = 10 mice per group).

By microscale thermophoresis, we analyzed the binding affinity of GLB-G25 and GLB-A04 aptamers for the extracellular domain of AXL receptor. Our data clearly showed that GLB-G25 and GLB-A04 bind to AXL protein with a KD of 239 ± 43 nM and 592 ± 92 nM respectively, whereas the scrambled aptamer is not binding to AXL protein (Figure 2 C). Therefore, we determined whether GLB-G25 and GLB-A04 aptamers have a better ability to reduce the level of the p-AXL activity compared with GL21.T, DNA unmodified and GLD-1. In these experiments, we treated SKOV3ip.1 with 0.4 μM of aptamers following Gas6 stimulation. We observed that GLB-G25 inhibited Gas6-induced p-AXL expression by 39% and GLB-A04 by 37%. These results are consistent with those previously obtained with GL21.T and GLD-1 aptamers (Figure 2 D). Next, we assessed whether GLB-G25 and GLB-A04 aptamers have a better antitumor activity in orthotopic SKOV3.ip1 animal model (n = 10 for each group in both animal models). SKOV3ip.1 milion cells were injected intraperitoneally. GL21.T, DNA unmodified, GLD-1, GLB-G25 PEG and GLB-A04 PEG were injected intravenously twice per week (Figure 2E). We observed a significant therapeutic effect. GLB-G25 PEG and GLB-A04 PEG significantly reduced tumor weight of 82% reduction (Figure 2G) and number of tumor nodules (Figure 2H) compared to scramble-PEG. However, no effect on body weight was observed (Figure 2F), indicating that GLB-G25 PEG and GLB-A04 PEG were not toxic.

GLB-G25 and GLB-A04 bind the extracellular domain of AXL in the centroid position by stronger interaction than GL21.T and GLD-1.

Previous studies showed that GL21.T has an affinity for the extracellular domain of the AXL receptor ²². This extracellular domain (ECD) is composed of two immunoglobulin-like domains (Ig1 and Ig2) and two fibronectin type 3-like domains (FNIII). Several studies have reported that Ig-like domains interact with DNA ^29–32. However, no data have been reported for FNIII domains. On the basis of this data, we hypothesized that aptamers bind to Ig-like domains of the AXL receptor.

We docked model structures GL21.T, GLD-1, GLB-G25, GLB-A04 and scramble aptamers to a homology model of the extracellular domain of the AXL receptor (dimer) using Autodock Vina software (Figure 3 A and S 4). We observed no differences in the poses related to the interactions between the head and stem region of the aptamers with the centroid of the AXL extracellular domain (Figure 3 B). We used the same procedure used for GL21.T, GLD-1, GLB-G25, and GLB-A04 to dock the scramble aptamer with AXL extracellular domain. Among all the possible poses, we selected the highest scored pose for the analysis. Based on docking, we predicted the affinity of the aptamers to the AXL receptor; GLB-G25 showed a predicted affinity of −13.62 kcal/mol and GLB-A04 showed a predicted affinity of −13.52 kcal/mol, similar to the predicted affinities of GL21.T (−12.25 kcal/mol) and GLD-1 (−11.96 kcal/mol). This suggests that 2’-F Py and dithiophosphate modifications do not interfere with the AXL interaction. These results suggest that the stem region of the aptamers forms major interactions with AB loops of the Ig1 domains of the AXL receptor. This observation is consistent with interactions of DNA with Ig-like domains of the transcription factor, in which major interactions are observed between the DNA major groove and loops of Ig-like domain ^{29, 31, 32}.

Figure 3: GL21.T, GLD-1, GLB-G25, and GLB-A04 bind to the AXL receptor with a centroid position and interactions between GLB-G25 and GLB-A04 aptamers and the AXL receptor are stronger than for interactions between GL21.T and GLD-1 aptamers and the AXL receptor.

(A) Rigid docking analysis three-dimensional models of GL21.T, GLD-1, GLB-G25, and GLB-A04 aptamers and the extracellular domain of the AXL receptor (dimer). (B) GLB-G25 and GLB-A04 have stronger interactions with the AXL receptor than GL21.T and GLD-1, as indicated by additional hydrogen bonds (see Table 1).

In Table 1, we report major interacting residues and bases for GL21.T, GLD-1, GLB-G25, GLB-A04 and the scramble aptamer. We observed that GLB-G25 and GLB-A04 engage in additional hydrogen bonds that could explain the stronger binding and hence the higher inhibition of AXL compared with GL21.T and GLD-1. GLB-G25 (represented by green in Figure 3 B) and GLB-A04 (represented by orange) interacted more strongly with the Ig-like domain of the AXL receptor than did GL21.T and GLD-1. Adenine 33 and thymine 19, interacting with arginine 103 on different monomers (free energy scoring: −19.78 and −38.13 kcal/mol for GLB-G25 and −16.99 and −38.13 kcal/mol for GLB-A04), are the key bases leading to the strong affinity between GLB-G25 and GLB-A04 and the AXL receptor (Figure 3 B).

Table 1: Interacting Residues (Ionic Pairs and Hydrogen Bonds) between AXL Receptor and GL21.T, GLD-1, GLB- G25, GLB-A04 and Scrambled Aptamer^a.

Sequence name	Free energy scoring, (kcal/mol)	Distance, (Å)	AXL interacting residues	Aptamer interacting residues
GL21.T	−14.939	3.076	Arg48	G11
	−28.728	2.937	Arg48	G28
	−15.743	3.097	Arg48	G29
	−35.167	2.932	Arg103	A33
GLD-1	−14.16	3.397	Arg48	G28
	−11.171	3.328	Arg48	G29
	−32.933	3.06	Arg103	A33
GLB-G25	−15.074	3.374	Arg48	G28
	−12.959	3.237	Arg48	G29
	−19.778	3.084	Arg103	T19
	−38.128	3.063	Arg103	A33
GLB-A04	−15.082	3.373	Arg48	G28
	−12.959	3.238	Arg48	G29
	−16.99	3.168	Arg103	T19
	−38.127	3.063	Arg103	A33
Scramble	−11.16	3.17	Arg185	T17
	−4.4	3.01	Arg16	C3
	−3.5	2.82	Gln76	C3
	−1.6	2.9	Aspi15	T17

^aGLB-G25 and GLB-A04 aptamers present stronger interactions with the Ig-like domain of the AXL receptor than do GL21.T and GLD-1 aptamers, supporting the idea that dithiophosphate modifications enhance the binding affinity. The interaction energy is weak for scrambled aptamer.

However, the scramble aptamer interacts with different residues (Arg185, Arg16, Gln76 and Asp115 interacting with thymine 17 and cytosine 3) and weak interaction energy (free energy score −11.16, −4.4, −3.5 and −1.6 kcal/mol) than GL21.T, GLD-1, GLB-G25 and GLB-A04. The predicted affinity for scramble aptamer seems much weaker (−3.1717 kcal/mol) than the GL21.T, GLD-1, GLB-G25, and GLB-A04 aptamers as well (S 4).

GLB-G25 and GLB-A04 reduce phosho-AXL expression, invasion, and migration in ovarian cancer cell lines.

To determine whether GLB-G25 and GLB-A04 can reduce the expression of the active form of AXL, we treated SKOV3.ip1 and OVCAR5 cells with GLB-G25 and GLB-A04 following Gas6 stimulation by immunoblotting analysis. Our data showed that GLB-G25 and GLB-A04, compared with the scramble aptamer, strongly reduced the expression of p-AXL by about 50% (Figure 4 A), however no effect was shown on total AXL expression in both cell lines. No significant effect was observed with GLB-G25 and GLB-A04 aptamers in cell proliferation in SKOV3.ip1 and OVCAR5 as compared with the scramble aptamer (Supplementary Figure 5 A). It has been reported that AXL receptor is involved in cancer cell metastasis ³³ to elucidate the effects of GLB-G25 and GLB-A04 on the role of AXL downstream signaling in cancer cell metastasis, we analyzed invasion and migration in SKOV3-ip1 and OVCAR5 cells. Treatment with GLB-G25 and GLB-A04 significantly reduced invasion (42% in SKOV3.ip1 and 73% in OVCAR5 for GLB-G25 and around 80% for GLB-A04 in both cell lines), as shown in Figure 4 B. GLB-G25 and GLB-A04 significantly inhibited cell migration in both cell lines (Figure 4 C). No significant effect was observed with GLB-G25 and GLB-A04 aptamers in migration in the AXL negative cell line FUOV1 as compared with the scramble aptamer (Supplementary Figure 5 B). Therefore, we analyzed the migration marker p-FAK by immunoblotting and we found that after treatment with GLB-G25 and GLB-A04 significantly reduce levels of phosphorylated focal adhesion kinase (p-FAK) in both cell lines.

Figure 4: GLB-G25 and GLB-A04 reduce p-AXL expression, migration, and invasion of ovarian cancer cells.

(A) Gas6-stimulated (200 ng/mL) SKOV3.ip1 (left) and OVCAR5 (right) cell lines were treated with GLB-G25, and GLB-A04 (400 nM). GLB-G25 and GLB-A04 strongly reduced p-AXL expression. Scramble aptamer was used as a negative control (β-actin was used as a loading control). (B) Expression of p-FAK after treatment with GLB-G25 and GLB-A04. SKOV3.ip1 and OVCAR5 cell lines were seeded followed by treatment with aptamers (400 nM) for 24 hours and analyzed by Western blot using β-actin as a loading control. (C) Effect of aptamers on ovarian cancer invasion. Imaging analysis showed a marked reduction of invasiveness in both cell lines (left), confirmed by the number of invading cells (****P= 0.0001 for both aptamers) compared with the scramble aptamer (right). (D) Effect of aptamers on ovarian cancer migration. Images analyzed showed a significant reduction of migration in both cell lines (left), confirmed by the number of migrated cells (*P < 0.05, **P < 0.001 and ****P < 0.0001) for both aptamers compared with the scramble aptamer used as a negative control (right).

2’-F Py, dithiophosphate, and PEG modifications enhance stability in human serum, pharmacokinetic profile, and bioavailability in vivo.

Stability, bioavailability, and pharmacokinetic profile are keys for the therapeutic applications of drugs. To determine if our modifications could improve aptamer stability, we incubated GLB-G25 and GLB-A04 (32 pmol) in 95% human serum at 37°C for a period ranging from 1 hour to 7 days. Evaluation by denaturing PAGE showed that GLB-G25 and GLB-A04 were stable for up to 72 hours (Figure 5 A).

Figure 5: Dithiophosphate modifications affect aptamer stability in human serum, aptamer pharmacokinetic profile, and aptamer penetration in vivo.

(A) Aptamer stability was studied for 7 days to determine stability in human serum. GLB-G25 and GLB-A04 were stable for up to 72 hours in human serum, far exceeding the previously reported stability values of 8 hours for GL21.T and 24 hours for GLD-1. (B-D) In vivo pharmacokinetic profile for Cy5-labeled GLB-G25 PEG, Cy5-labeled GLB-A04, and Cy5-labeled GLB-A04 PEG. PEGylated aptamers were retained (bioavailability) at 10 μg/mL after 500 minutes (n = 3 mice per group). (E) Ex vivo tumor, kidney, spleen, liver, lung, heart, and brain penetration for Cy7.5-labeled GLB-G25 PEG and Cy7.5-labeled GLB-A04 PEG compared with Cy7.5-labeled scramble aptamers. Results are based on the scale shown and values are expressed by average radiance (p/s/cm²/sr; n = 3 mice per group).

Owing to their small size, aptamers are easily eliminated by renal clearance and degraded by nucleases ³⁴. Aptamer modifications and conjugations with molecules that can increase the aptamer size may affect circulation, tissue accumulation, metabolism, and renal clearance ³⁵. Given this information, we evaluated the pharmacokinetic profile of GLB-G25 and GLB-A04 with and without PEG modification in vivo after intravenous administration of a single bolus dose. PEGylation increased the size of the aptamers and reduced time of elimination (t_1/2) (~253.8 minutes for GLB-G25 PEG; ~305.9 minutes for GLB-A04 PEG compared with ~112.9 minutes for GLB-A04), renal clearance (CL) [~0.000578 μg∙(μg/ml)⁻¹ for GLB-G25 PEG; ~0.002377 μg∙(μg/ml)⁻¹ for GLB-A04 PEG compared with ~ 0.0146 μg∙(μg/ml)⁻¹ for GLB-A04], and volume distribution (V) [~ 0.0393 μg∙(μg/ml)⁻¹ for GLB-G25 PEG; no data were reported for GLB-G25; ~0.272 μg∙(μg/ml)⁻¹ for GLB-A04 PEG compared with ~0.547 μg∙(μg/ml)⁻¹ for GLB-A04; (Figure 5 B-D)].

To determine whether GLB-G25 and GLB-A04 were targeted in the tumors, we intravenously (via tail vein) injected Cy7.5-labeled GLB-G25 and GLB-A04 into mice bearing SKOV3.ip1 ovarian cancer cells and obtained images 2 hours after injection. Next, we studied the tissue and tumor distribution of GLB-G25 PEG and GLB-A04 PEG in an orthotopic model of ovarian cancer. Two million SKOV3.ip1 ovarian cancer cells were injected into the peritoneal cavity of mice (three mice per group). Two weeks later, tumor-bearing mice were injected intravenously with Cy7.5-labeled GLB-G25 PEG and GLB-A04 PEG. Mice were euthanized and tissues were dissected 2 hours after aptamer injection. Images were obtained and average radiance was analyzed (Figure 5 E). Cy7.5 GLB-G25 PEG and Cy7.5 GLB-A04 PEG significantly targeted the tumor (Average Radiance 3.081×10⁷ p/s/cm²/sr for Cy7.5 GLB-G25 PEG and 3.866×10⁷ p/s/cm²/sr for Cy7.5 GLB-A04 PEG) compared with Cy7.5 scramble PEG (Average Radiance 1.02483×10⁷ p/s/cm²/sr and 1.00353×10⁷ p/s/cm²/sr respectively) used as a negative control.

2’-F Py, dithiophosphate, and PEG modification of GLB-G25 and GLB-A04 inhibit ovarian cancer tumor growth and metastasis in vivo.

Next, we assessed antitumor activity in two different orthotopic SKOV3.ip1 and OVAR5 animal models (n = 10 for each group in both animal models). Ovarian cancer cells were injected intraperitoneally. GLB-G25 PEG and GLB-A04 PEG were injected intravenously twice per week, alone or in combination with paclitaxel injected intraperitoneally once per week (S2 A).

We observed a significant therapeutic effect in both in vivo models. GLB-G25 PEG and GLB-A04 PEG alone or in combination with paclitaxel significantly reduced tumor weight (in mice bearing SKOV3.ip1: 74% reduction for GLB-G25 PEG and 91% reduction for GLB-A04 PEG; in mice bearing OVCAR5: 68% reduction for GLB-G25 PEG and 84% reduction for GLB-A04 PEG) and number of tumor nodules (in mice bearing SKOV3.ip1: 68.5% reduction for GLB-G25 PEG and 78.5% reduction for GLB-A04 PEG (Figure 6 A and B); in mice bearing OVCAR5: 64% reduction for GLB-G25 PEG and 80% reduction for GLB-A04 PEG; S2 B and C). However, no effect on body weight was observed (S2 B and S3 A), indicating that GLB-G25 PEG and GLB-A04 PEG were not toxic.

Figure 6: GLB-G25 PEG and GLB-A04 PEG, alone and in combination with paclitaxel, have antitumor activity in vivo in a SKOV3.ip1 ovarian cancer mouse model.

(A) SKOV3.ip1 tumor-bearing mice treated with GLB-G25 PEG and GLB-A04 PEG, alone or in combination with paclitaxel, exhibited lower tumor weights (B), fewer tumor nodules (C), Treatment with GLB-G25 PEG and GLBA04 PEG in vivo induced significant down-modulation of p-AXL in SKOV3.ip1 tumors compared with scramble aptamer PEG. GLB-G25 PEG and GLB-A04 PEG in combination with paclitaxel reduced p-AXL expression compared with paclitaxel alone or in combination with scramble aptamer PEG. β-actin was used as a loading control. (D) Prediction of tumor weight in different ovarian cancer orthotopic mouse models under six experimental conditions using a mathematical model. Tumor growth rates (r) were first estimated for SKOV3ip.1 cell line(black), and then their tumor growth inhibition (α) and death rates (β) were estimated under treatment with the two aptamers GLB-G25 PEG (light green) and GLB-A04 PEG (dark green), paclitaxel chemotherapy (red), and the combination therapies GLB-G25 PEG + paclitaxel (light blue) and GLB-A04 PEG + paclitaxel (dark blue). Estimated parameter values are shown in the figure insets, which may be used to obtain model-predicted tumor weight change over time under each treatment condition. Treatment with GLB-G25 PEG and GLBA04 PEG in vivo induced (E) lower Ki67 index (G) and lower microvessel density (CD31-positive staining) than tumor-bearing mice treated with scramble aptamer. Scramble aptamer PEG plus paclitaxel was used as a negative control. Quantification of proliferation (F) and angiogenesis (H) in vivo is also shown. Data are presented as means ± standard deviation. ****P < 0.0001; **P < 0.001 (n = 10 mice per group).

Next, we assessed whether GLB-25 PEG and GLB-A04 PEG could reduce p-AXL expression in tumor tissue from mice bearing SKOV3.ip1 ovarian cancer cells. Our findings demonstrated that GLB-G25 PEG and GLB-A04 PEG alone or in combination with paclitaxel significantly reduced p-AXL expression (Figure 6 C).

Mathematical model results are shown in Figure 6 D and S3 C; estimated parameter values are shown in the figure insets. The model was then used to predict drug treatment outcome in vivo, indicated by tumor weight change over time. We observe that with or without treatment, the tumor increases monotonically over the entire treatment course, but the various treatments show suppressing effects on the growth of both OVCAR5 (S3 C) and SKOV3ip.1 tumors (Figure 6 D).

Comparing the individual treatments with their combination counterparts, in the case of OVCAR5 (Figure. S3 C), one combination form (GLB-G25 PEG + paclitaxel) results in a larger growth inhibition rate (αGLB-G25.pac > αGLB-G25) and a larger death rate (βGLB-G25.pac > βGLB-G25); while the other combination (GLB-A04 PEG + paclitaxel) results in a smaller growth inhibition rate (αGLB-A04.pac < αGLB-A04) and a smaller death rate (βGLB-A04.pac < βGLB-A04), but still shows a better outcome than any of the individual treatments (because of the synergistic effects of aptamers plus chemotherapy). In the case of SKOV3.ip1 (Figure 6 D), the growth inhibition rates and death rates obtained from both combination cases (i.e., GLB-G25 PEG + paclitaxel and GLB-A04 PEG + paclitaxel) are smaller than those values obtained from the corresponding individual treatments; and only one combination (GLB-G25 PEG + paclitaxel) shows a better outcome over the individual treatments. Interestingly, with respect to SKOV3.ip1, one individual treatment, i.e., the one using GLB-A04 PEG alone, shows the most favorable outcome among all treatment options. This indicates that the two aptamers, GLB-G25 PEG and GLB-A04 PEG, may have different functional interactions with the tumor. Comparing both mouse models, while the tumor growth rate for SKOV3 is slightly higher than that for OVCAR5, both aptamers (GLB-G25 PEG and GLB-A04 PEG), whether used alone or in combination with chemotherapy, seem to exhibit greater effects on suppressing SKOV3ip.1 than OVCAR5. Cell proliferation, assessed by Ki67 expression, was also decrease significantly in mice treated with GLB-G25 PEG and GLB-A04 PEG (P<.0001) alone or in combination with Paclitaxel (P<.0001) compared with the group that received scramble PEG aptamer alone or scramble PEG aptamer in combination with paclitaxel in SKOV3.ip1 (Figure 6 E-F) and OVCAR5 (S3 E-F). Moreover, we observed that the treatment with GLB-G25 PEG and GLB-A04 PEG alone and in combination with paclitaxel significantly reduce the number of pro-angiogenic CD31-positive cells in SKOV3.ip1 (Figure 6 G-H) and OVCAR5 (S3 G-H) in vivo models compared with scramble PEG aptamer alone or in combination with Paclitaxel.

Discussion

The modified aptamers, reported here, represent one of the few aptamers converted from RNA into DNA. In terms of aptamer development and therapeutic application, DNA aptamers have several advantages compared their RNA counterparts. DNA is more chemically and biologically stable, easier to synthesize and present a longer shelf life than RNA ³⁶. Our results demonstrate that converted DNA aptamers modified to 2’-F Py, dithiophosphate, and PEG significantly improve the pharmacokinetic and biological properties of these new molecules compared with the previous anti-AXL aptamers, GL21.T and GLD-1. Our data clearly show that the antitumor effects of dithiophosphate PEGylated DNA aptamers combined with paclitaxel highlight the potential impact of this novel anti-AXL DNA aptamer on the dosing of paclitaxel. Whereby, the use of the anti-AXL DNA aptamer will lead to lower requirements of paclitaxel.

Molecular targeted therapy is becoming a promising integral part of cancer treatment along with conventional surgery, radiotherapy, and chemotherapy ³⁷. Epithelial ovarian cancer is the most lethal gynecologic malignancy, and effective systemic therapeutic approaches beyond traditional surgery and chemotherapy are clearly required to improve patient outcomes ³⁸. Recent advances in our understanding of the underlying molecular aberrations in epithelial ovarian cancer and in the development of new drugs for epithelial ovarian cancer are providing unique opportunities for precision medicine in the management of this disease ³⁹.

Nucleic acid aptamers represent excellent potential therapeutic tools to specifically recognize and inhibit oncogene functions ⁴⁰. This new class of molecular ligands presents a wide range of advantages compared with other ligands such as monoclonal antibodies or peptides⁴¹. Aptamers are poorly immunogenic, easy to manufacture, and modified chemically with low cost and high effectiveness⁴¹. However, two factors are critical for aptamer development in clinical applications: susceptibility to exo and endo nucleases and renal excretion ⁴⁰. Due to this fact in this work, we have converted an RNA aptamer into DNA aptamer with a better features, more stable and higher activity and bioavailability.

Dithiophosphate backbone modifications provide resistance against nucleases and enhance the binding properties for aptamers, especially compared with monothiophosphate aptamers ⁴². Our findings showed that similar to GL21.T and GLD-1, GLB-G25 and GLB-A04 preserved 3D conformation. GLB-G25 and GLB-A04, the AXL receptor–targeted aptamers, have better binding to their target than GL21.T and GLD-1, via adenine 33/arginine 103 and thymine 19/arginine 103 interactions that are not present in GL21.T and GLD-1 aptamers targeting AXL. These next-generation aptamers improve the biological activity of aptamers by inducing a 50% reduction of p-AXL at low concentrations. Therefore, PEG conjugation increases the aptamer size and represents a strategy to overcome rapid renal excretion and increase nuclease resistance ⁴².

PEGylated GLB-G25 and GLB-A04 showed prolonged stability compared with GL21.T and GLD-1 and dramatically improved the pharmacokinetic properties in vivo compared with aptamers without PEG. We observed that PEGylated aptamers had a more prolonged retention time, better resistance to nucleases, better inhibition of tumor growth, and reduced doses and administration frequency compared with GL21.T and GLD-1 aptamers (see Supplementary Table 1 and Supplementary Table 2). Therefore, next-generation 2’-F Py dithiophosphate aptamers have a basal potential application for any target and can be used in the development of chimeras conjugating with noncoding RNA or nanoparticles. PEGylated dithiophosphate modified aptamers could be used not just as target inhibitors but also as a vehicle to specifically deliver other therapeutic moieties.

Materials and methods

Aptamer synthesis

Single dithio-substituted versions of GLB-B0 at positions A04 and G25, as well their scramble sequences were synthesized on an Expedite 8909 Oligo Synthesizer (Applied Biosystems, Foster City, CA) using standard phosphoramidite chemistry ⁴³. DNA synthesis reagents were purchased from Glen Research Inc. (Sterling, VA). Fluoro-dC and fluoro-dU phosphoramidites were purchased from both Glen Research and Sigma-Aldrich (St. Louis, MO). Sulfhydryl, amine, and Cy5 modification were introduced at the 5’ end using thiol modifier C6SS, 5’ Amino-Modifier C6, and Cy5 phosphoramidites (Glen Research). Following synthesis and 5’ modification, thioaptamers were deprotected in 30% ammonium hydroxide for 24 hours at room temperature. They were dried overnight on a lyophilizer with trace amounts of Tris base to protect the DMT group. Dried pellet was re-solubilized and thioaptamers were purified by reverse phase chromatography on a Hamilton PRP-1 column using a 100mM triethylamine acetate buffer (pH 8.4) and acetonitrile gradient. Fractions collected were analyzed by PAGE and EtBr staining for integrity of DNA. Fractions containing the desired length product were pooled and dried. Thioaptamer concentrations were determined using extinction coefficients calculated from OligoCalc ⁴⁴. Before each treatment, aptamers were subjected to a short denaturation–renaturation step (5 minutes at 85°C, 3 minutes on ice, 10 minutes at 37°C).

Aptamer PEGylation

For conjugation of thioaptamers with 10-kd PEG, approximately 100 nmole of HPLC-purified C6SS thiol-modified thioaptamer was suspended in 500 μl of phosphate-buffered saline (PBS) with pH 7.4 and reduced with 20 μM DTT for 1 hour. DTT was removed by filtration through 3-kd spin filters. Reduced thioaptamers were incubated with 3-fold molar excess of monofunctionalized 10K PEG-maleimide (Creative PEG Works, Durham, NC, USA) for 3 hours. After the reaction, the mixture was filtered to 10-kd cutoff spin filters to remove unreacted PEG-maleimide. The efficiency of PEGylation was monitored by mobility shift on polyacrylamide gel.

Aptamer double conjugation

For double conjugation of thioaptamers with 10K PEG and Cy7.5, a set of thioaptamers was also synthesized and purified; both thiol modifier C6SS and a different amino modifier, Amino-Modifier C6 dT, were used. Amino-Modifier C6 dT was in the penultimate position and served to conjugate to Cy7.5 dye using standard NHS chemistry before PEGylation. The conjugation reaction was carried out in borate buffer (pH 8.5) with 10-fold molar excess of dye in the presence of 10% DMSO to keep the dye solubilized. After 2 hours, the reaction was quenched with 10 μl of 1M Tris (pH 8.0) and the unreacted dye was removed by 3-kDa cutoff spin filters. The dye conjugation was monitored by PAGE and fluorescence imaging.

Cell lines

SKOV3-ip1, HeyA8, and A2780-PAR cells were maintained in RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum and 0.1% gentamicin sulfate (Gemini Bioproducts, Calabasas, CA, USA). OVCAR-5 and FUOV1 cells were maintained in Dulbecco modified Eagle–F12 medium (Corning Cellgro, NY, USA) supplemented with 10% heat-inactivated fetal bovine serum and 0.1% gentamicin sulfate (Gemini Bioproducts, Calabasas, CA, USA). All cell lines were maintained in 5% CO₂ and 95% air at 37°C. All cell lines were obtained from the MD Anderson Characterized Cell Line Core Facility, which supplies authenticated cell lines and is routinely screened to confirm the absence of Mycoplasma using a MycoAlert mycoplasma detection kit (Lonza Rockland, ME, USA) as described by the manufacturer. All in vitro experiments were conducted with 60–80% confluent cultures.

Immunoblotting

Whole cell lysates were prepared from cultured cells by subjecting them to ice-cold lysis buffer supplemented by protease and phosphatase inhibitor cocktails (Sigma-Aldrich) in RIPA buffer. Proteins were isolated and then quantified with Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Total protein samples (30 μg) were subjected to electrophoresis on 7.5%, 10%, and 4% to 15%–gradient sodium dodecyl sulfate polyacrylamide gels (Bio-Rad, Hercules, CA, USA) and then each was electrophoretically transferred onto a polyvinylidene difluoride membrane (Millipore, Darmstadt, Germany). Membranes were blocked with nonfat dry milk, washed, and probed overnight at 4°C with primary anti-p-AXL (1:1000; Cell Signaling, Danvers, MA, USA), anti-AXL antibody (1:1000 dilution; R&D System, Minneapolis, MN, USA), p-FAK (1:1000 dilution; BD Biosciences, San Jose, CA, USA), FAK (1:1000 dilution; BD Biosciences, San Jose, CA, USA), and anti-β-actin antibody as a loading control (1:10000 dilution; Sigma-Aldrich), followed by horseradish peroxidase–conjugated secondary rabbit or mouse antibody (Cell Signaling). Bound antibodies were visualized using an enhanced chemiluminescent horseradish peroxidase antibody detection kit (GE Healthcare, Wauwatosa, WI, USA).

Structural modeling

The secondary structures of the aptamers were predicted using the MFold server (23). On the basis of the predicted secondary structure, we divided aptamer sequences into four regions, one head/loop region, one stem region, and two single-stranded ends. Each region was modeled using the DNA model builder server (http://knotts.byu.edu/dna_model_builder.html). Resulting models were imported into Molecular Operating Environment (MOE) software and connected manually to generate single-stranded DNA structure. For DNA aptamers, 3’-dithio-dG and 3’-dithio-dA modifications were done in GLB-A04 and GLB-G25, respectively. For the GL21.T RNA aptamer, a hydroxyl group was added to the sugar moiety and thymine was converted to uracil manually. Partial charges were added and energy minimizations were carried out using the Amber10: EHT force field. A homology model of extracellular domains of AXL (Ig₁ and Ig₂ domains) was built using the Tyro3 structure (PDB ID: 1RHF) as a template in MOE using default settings.

Molecular docking

Molecular docking experiments were performed using Autodock Vina ⁴⁵. Modeled aptamers of GL21.T, GLB-G25, and GLB-A04 were prepared for docking using the prepare_ligand4.py script. All torsion angle remarks were deleted from the PDBQT file and TORSDOF was set to zero. The receptor model was prepared for docking with default settings the using prepare_receptor4.py script. The concave surface of the receptor structure was used to generate a grid for docking, and the exhaustiveness setting was set to 8. Ten docked poses were generated for each aptamer. Results were analyzed in MOE.

Microscale thermophoresis (MTS)

Microscale thermophoresis (MST) binding experiments were carried out with 5 nM Cy5-labeled aptamers (GLB-G25, GLB-A04) and 100 nM Cy3 labelled scrambled control aptamer in binding buffer (1x PBS pH 7.4, 0.1% Triton X 100) with a range of concentrations of AXL protein (Creative Biomart, Shirley, NY) from 0,118 to 3781 nM at 40% MST power, 30% LED power (Cy5 aptamers) or 80% LED power (Cy3 aptamer) in standard capillaries on a Monolith NT.115 pico device (Cy5 aptamers) or Monolith NT115 (Cy3 aptamer) at 25°C (NanoTemper Technologies, Munich, Germany). Data was analyzed using MO. Affinity Analysis software (version 2.3, NanoTemper Technologies) at the standard MST-on time of 1.5 s. Data fits possessing amplitudes >5 units combined with Signal to Noise levels >5 units were defined as binding events. Error bars represent the standard deviation of two independent experiments in (two technical repeats each).

In order to calculate fraction bound, the ΔFnorm value of each point is divided by the amplitude of the fitted curve, resulting in values from 0 to 1 (0 = unbound, 1 = bound), and processed using the Kaleidagraph software (4.1) and fitted using the KD fit formula derived from the law of mass action. Error bars represent the standard deviation of three independent experiments in (two technical repeats each).

Aptamers comparison by in vivo tumor growth inhibition

Orthotopic tumor model was established by implantation of 10⁶ SKOV3.ip1 ovarian cancer cells via in nude mice. Nude mice bearing SKOV3.ip1 were randomly divided into seven groups (10 mice/group) and treated with scramble aptamer PEG (1600 pmol or 0.6 mg kg-1), GL21.T (1600 pmol or 0.6 mg kg-1), DNA unmodified aptamer (1600 pmol or 0.6 mg kg-1), GLD-1 (1600 pmol or 0.6 mg kg-1), GLB-G25 PEG (1600 pmol or 0.6 mg kg-1), GLB-A04 PEG (1600 pmol or 0.6 mg kg-1). GL21.T, DNA unmodified aptamer, GLD-1, GLB-G25 PEG, GLB-A04 PEG, and scramble aptamer PEG were administered intravenously biweekly. Animals were sacrificed after 5 weeks of treatment, and tumors were removed. Tumor weight and number and location of tumor nodules were recorded. Tumor tissue was snap-frozen for lysate preparation.

Stability in human serum

To determine the stability in human serum (Sigma-Aldrich), we incubated GLB-G25 and GLB-A04 aptamers at a concentration of 4 μM in 95% human serum at different times (1, 3, 6, 12, 24, 48, 72, and 168 hours). At each time point, 32 pmol of aptamers was withdrawn and incubated for 2 hours at 37°C with proteinase K solution (Millipore) to remove serum proteins that interfere with electrophoretic migration. Following treatment with proteinase K, 1x TBE and 10x urea RNA loading dye (Invitrogen, Carlsbad, CA, USA) were added to the samples. Each sample was heated to 95°C for 5 minutes and then stored at −80°C. At all-time points, samples and controls (aptamer alone and human serum alone) were loaded into an 8M urea 15% acrylamide gel and separated by electrophoresis. The urea acrylamide gel was stained with a 1:10,000 dilution of SYBR Gold (Life Technologies, Carlsbad, CA, USA) in 1x TBE for 30 minutes and visualized by ultraviolet exposure.

Invasion assays and wound healing

Cell invasiveness was assessed using transwell chamber assay. Transwell chambers (Greiner Bio One, Kremsmünster, Austria) were coated with matrigel (Corning, Cellgro) containing extracellular matrix proteins. SKOV3.ip1 and OVCAR5 cells treated with GLB-G25, GLB-A04, or scramble aptamer were suspended in serum-free medium and added into the upper matrigel-coated chambers (5.0×10⁵ cells/per chamber). Complete medium containing 10% fetal bovine serum was added to the lower chambers as a chemo-attractant. The cells were incubated at 37°C in 5% CO₂ for 24 hours. After incubation, the cells in the upper chamber were removed with cotton swabs. Cells that invaded the lower chambers were fixed and stained using the Hema3 staining set (Fisher HealthCare Protocol). Cells in five random fields were counted using ImageJ 1.48v software. Experiments were repeated three times.

Cell migration was measured by a wound healing assay. SKOV3.ip1 and OVCAR5 cells (2.0×10⁵ cells/per well) were plated onto six-well plates before treatment with GLB-G25, GLB-A04, or scramble aptamer and then incubated at 37°C until they reached 100% confluence to form a monolayer. Each cell monolayer was carefully scratched using a p200 pipet tip, and then cellular debris was removed by washing with Hank’s balanced salt solution (Gibco, Carlsbad, CA, USA). The cells were treated with GLB-G25, GLB-A04, or scramble aptamer. Images (magnification 10 x) were captured at 0 and 36–48 hours (depending on the cell line) after scratching using a phase-contrast Nikon eclipse TE2000-U microscope. The rate of migration was measured by quantifying the total distance that the cells migrated from the edge of the scratch toward the center of the scratch. The obtained values were expressed as open area. Experiments were repeated three times.

Pharmacokinetic and ex vivo bioavailability in tumor bearing mice

Following a single intravenous bolus administration of Cy5-labeled GLB-G25, Cy5-labeled GLB-G25 PEG, Cy5-labeled GLB-A04, or Cy5-labeled GLB-A04 PEG (1600 pmol in 100 μl) into 8-week-old nude mice (three mice per group), blood was collected via abdominal vena cava or cardiac puncture at 5, 15, and 30 minutes and 1, 2, 4, 8, 24, 48, and 72 hours after the injection. The fluorescence intensity of each plasma sample (10 μl) was measured by the Tecan M1000 plate reader (Packard, Downers Grove, IL) at 550/570 nm (excitation/emission wavelengths) to determine the pharmacokinetic parameters of Cy5-GLB-G25, Cy5-GLB-G25 PEG, GLB-A04, and GLB-A04 PEG. Plasma was also collected from saline-injected mice and used as a baseline. The fluorescence intensity of each sample was interpolated to a standard curve.

To investigate bioavailability, we intravenously injected Cy7.5-labeled GLB-G25, Cy7.5-labeled GLB-A04, and Cy7.5-labeled scramble aptamers (1600 pmol/mouse or 0.6 mg kg⁻¹) into mice bearing SKOV3.ip3 orthotopic ovarian cancer tumors (n = 3). The bioavailability of the aptamers was assessed 2 hours after injection using an IVIS-spectrum imaging system. Animals were euthanized and tissue from the tumor and major organs such as the kidney, spleen, liver, lung, heart, and brain were collected, and tissues of equal size and shape were deposited onto a black plate (which absorbed light and reduced background and crosstalk) for measurement of fluorescence intensities. Images were captured and visualized using a Xenogen IVIS Spectrum imaging system (Caliper Life Sciences, Waltham, MA, USA) with a 788-nm excitation wavelength filter and 808-nm emission wavelength filter for Cy7.5.

Orthotopic tumor implantation and drug treatment

Orthotopic models of ovarian cancer were developed as described previously ⁴⁶. SKOV3.ip1 and OVCAR5 ovarian cancer cells were harvested using trypsin-EDTA, neutralized with fetal bovine serum-containing medium, washed, and re-suspended in appropriate numbers in Hank’s balanced salt solution prior to injection. To assess the therapeutic activity of GLB-G25 PEG, GLB-A04 PEG, and scramble aptamer PEG, we loaded aptamers alone and in combination with paclitaxel in the capillary. Nude mice bearing SKOV3.ip or OVCAR5 tumors were randomly divided into eight groups (10 mice/group) and treated with paclitaxel (35 μg) only, scramble aptamer PEG (1600 pmol or 0.6 mg kg⁻¹), scramble aptamer PEG + paclitaxel (1600 pmol or 0.6 mg kg⁻¹ + 35 μg paclitaxel) GLB-G25 PEG (1600 pmol or 0.6 mg kg⁻¹), GLB-G25 PEG + paclitaxel (1600 pmol or 0.6 mg kg⁻¹ + 35 μg paclitaxel), GLB-A04 PEG (1600 pmol or 0.6 mg kg⁻¹), or GLB-A04 PEG + paclitaxel (1600 pmol or 0.6 mg kg⁻¹ + 35 μg paclitaxel). Paclitaxel was administered intraperitoneally once weekly, and GLB-G25 PEG, GLB-A04 PEG, and scramble aptamer PEG were administered intravenously biweekly. Animals were killed after 5 weeks of treatment, and tumors were removed and processed for immunohistochemical analyses. Tumor weight and number and location of tumor nodules were recorded. Tumor tissue was fixed in formalin for paraffin embedding, frozen in optimal cutting temperature medium for preparation of frozen slides, or snap-frozen for lysate preparation.

Immunohistochemistry

Immunohistochemical analysis for Ki67 was performed on 4-μm formalin-fixed, paraffin-embedded epithelial cancer sections. Slides were deparaffinized and dehydrated, then subjected to antigen retrieval using 1x Diva Decloaker (BioCare Medical, Pacheco, CA) under a steamer. Endogenous peroxidases were blocked with 3% hydrogen peroxide in methanol followed by washes with PBS. Nonspecific binding was blocked with 5% normal horse serum and 1% normal goat serum in PBS. Samples were incubated with primary antibody against Ki67 (1:200; Neomarkers, Portsmouth, NH) overnight at 4°C, followed incubation with goat anti-rabbit horseradish peroxidase secondary antibody (Jackson Immuno Research Laboratories, West Grove, PA, USA) diluted in blocking solution. Immunohistochemical analyses for CD31 were performed on 8-μm sections of fresh frozen cancer specimens embedded in optimal cutting temperature medium. Slides were fixed with cold acetone and acetone:chloroform and rehydrated with PBS. Nonspecific binding was blocked with 5% normal horse serum and 1% normal goat serum in PBS. Samples were incubated with the primary antibody, rat anti-mouse CD31 (1:200; BD Pharmingen, San Diego, CA, USA), overnight at 4°C. After washing, samples were incubated with peroxidase-conjugated goat anti-rat secondary antibody (Jackson ImmunoResearch Laboratories). The slides were incubated with 3,3’-diaminobenzidine (Sigma-Aldrich) at room temperature, counterstained with hematoxylin for 15 seconds, and mounted on a slide to be analyzed on a bright-field microscope (magnification 20x).

Mathematical modeling of cancer treatment

Building on our prior work on mathematical modeling of cancer treatment and biophysical modeling of drug delivery ^47–53, we developed a model for predicting tumor response to drug treatment based on aptamers, alone or in combination with chemotherapy:

$$\frac{dW}{dt}=\left(r_i-{\alpha }_i\right)·\left(1-\frac{W}{K_{\text{w}}}\right)·W-{\beta }_i·W$$

where $W$ is the tumor weight, $i$ represents the treatment group (control or drug treatment), $r_i$ is the tumor growth rate, ${\alpha }_i$ is the growth inhibition rate due to aptamers, ${\beta }_i$ is the tumor death rate due to chemotherapy, and K_W is a constant that represents the carrying capacity of the tumor system. In our analysis, for each of the ovarian cancer orthotopic mouse models (OVCAR5 and SKOV3ip.1), KW was set to the largest tumor weight measured in their corresponding experiments.

For each orthotopic mouse model, we first fit the model to control group data (i.e., in the absence of treatment where both α = 0 and β = 0 at all times) to determine the tumor growth rate (r). We then fit the model to the drug treatment data for (i) aptamers alone (GLB-G25 PEG and GLB-A04 PEG), (ii) chemotherapy alone (paclitaxel), and (iii) their combined form in order to determine the growth inhibition rates (α) and tumor death rates (β), first independently and then when delivered in combination.

Statistical analysis

Box and whisker plots along with statistical analyses were generated. We used the nonparametric test Man Whitney to assess the effect of different treatments on tumor growth and number of nodules. Two references were considered: untreated GL21.T, GLD-1, GLB-G25 PEG and GLB-A04 PEG, respectively scrambled PEG.

Analysis of variance ANOVA (for multiple group comparison) was used to compare independent samples from different groups. All statistical tests were performed using GraphPad Prism 7.0. All data are presented as mean ± standard deviation. P<0.05 was considered statistically significant.

Supplementary Material

Supporting Information

Materials, methods, and data regarding: Supplementary Figure 1 shows aptamer selection in HeyA8 cell line. Supplementary Figure 2 shows the schematic representation of the treatment for in two orthotopic ovarian cancer cell models (SKOV3.ip1 and OVCAR5). Supplementary Figure 3 shows orthotopic tumor implantation, drug treatment; immunohistochemistry and mathematical modeling of cancer treatment in OVCAR5 in vivo animal model. Supplementary Figure 4 shows that the structural modeling and molecular docking for scramble aptamer. Supplementary Figure 5 shows cell viability of GLB-G25 and GLB-A04 in SKOV3ip.1 and OVCAR5 ovarian cancer cell lines wound healing assay performed using AXL negative ovarian cancer cell line (FUOV1).

Abbreviations:

SELEX

Systematic Evolution of Ligands by Exponential Enrichment

3D

Three-dimensional

TAM

Tyro3-AXL-Mer

Gas6

Growth arrest-specific protein 6

2’-F Py

2’-Fluoro pyrimidine

PEG

Polyethylene glycol

CCLE

Cancer Cell Line Encyclopedia

p-AXL

phospho-AXL

ECD

Extracellular domain

Ig

Immunoglobulin-like domains

FNIII

Fibronectin type 3-like domains

t_1/2

indicates time of elimination

V

Volume distribution

CL

Renal clearance

MOE

Molecular Operating Environment

denaturing PAGE

denaturing Polyacrylamide gel electrophoresis

EtBr

Ethidium bromide

MTS

Microscale thermophoresis