Introduction
BCL-2, a member of the antiapoptotic protein family, plays a key and central role in preventing cell death, a defining characteristic of malignant cells [
1]. BCL-2 confers an antideath phenotype and its overexpression contributes to the genesis of hematopoietic and lymphatic cancers [
2‐
6]. Aberrant BCL-2 expression is driven by
t(14;18) chromosomal rearrangement of the BCL-2 gene in many follicular (FL) and diffuse large B cell (DLBCL) lymphomas [
3‐
6]. In chronic lymphocytic leukemia, impaired degradation of BCL-2 mRNA is linked to nucleolin and/or microRNA expression, resulting in continuous production of BCL-2 protein and the subsequent survival of leukemic cells [
7,
8]. In cancers of the breast, skin, prostate, sarcomas, and lung, BCL-2 is implicated in the development of chemo-resistance [
9]. Given its biological importance, BCL-2 is a desirable target for therapeutic development. Numerous approaches have been reported to block or modulate the production of BCL-2 at the DNA level (e.g., retinoids and histone deacetylase inhibitors), at the RNA level (targeted antisense oligonucleotides, siRNA or miRNA), or the protein level (e.g., pan inhibitors of BH3 family members of BCL-2, or recently the specific BCL-2 inhibitor, ABT-199) reviewed in [
10].
There is an emerging understanding that CpG islands surround mammalian cell promoter regions and that these non-methylated regions contribute to gene regulation [
11,
12]. The recognition that these genomic regions are DNAse I-hypersensitive enabled the discovery of cis-regulatory elements that act as transcription factors, enhancers, silencers, repressors, or control regions, which regulate gene expression [
13‐
15]. Additionally, higher-order secondary structures (quadruplexes, cruciforms, or I-motifs) that surround the promoter regions of oncogenes may also serve as cis-regulatory domains to modulate transcription [
16,
17]. It has been demonstrated that exposing cells to short DNA sequences containing these motifs reduced mRNA and protein levels [
18,
19]. Others have recognized the importance of regulatory regions specific for BCL-2 [
20‐
23]. Young and Korsmeyer demonstrated that a series of 20 base deletions between the P1 and P2 promoter of BCL-2 decreased transcription. Miyashita et al. reported that p53-dependent regions upstream of the BCL-2 gene act as negative regulatory elements, and Duan et al. showed long-range regulatory effects on BCL-2 transcription by enhancers in the IgH 3’ region. The observations support the hypotheses that these regulatory regions may be favorable therapeutic targets for hybridization due to the accessible chromatin state during oncogene up-regulation and transcription.
We describe a novel approach to blocking transcription termed DNA interference (DNAi
®). DNAi therapeutic candidates are a new class of nucleic acid-based drugs. They are single-stranded sequences of unmodified phosphodiester DNA having lengths of 20–34 bases. These sequences are designed to be complementary to non-coding, non-transcribed regions of genomic DNA upstream of gene transcription start sites. The hybridization of the DNA-interfering oligonucleotide to its targeted region results in gene modulation with phenotypic changes and a modulation of mRNA and protein levels. While DNAi against BCL-2 is described in this paper, DNAi oligonucleotides may be designed to also target other regions of the genome to modulate genes. We describe here the development of PNT2258, containing PNT100, a 24-base single-stranded DNA oligonucleotide specific to BCL-2 encapsulated in amphoteric liposomes. The liposomes protect the oligonucleotides from nuclease degradation, facilitate cellular uptake, and enable endosomal escape to the nucleus, where the biological effects of the oligonucleotides occur [
24‐
26]. PNT2258 is currently undergoing clinical development against BCL-2-driven malignancies [
27].
Materials and methods
PNT100 and control DNAi oligonucleotides
Oligonucleotides were produced by multi-step solid-phase organic synthesis involving on-column cleavage from solid support, base de-protection, followed by ion exchange (IEX) purification, ultrafiltration/diafiltration, concentration, and freeze-drying. The PNT100 and controls include: PNT100, 5′-CACGCACGCGCATCCCCGCCCGTG-3′; methylated PNT100, 5′-CAXGCAXGXGCATCCCXGCCXGTG-3′, where X represents a methylated cytosine base; scrambled control, 5′-CGGCGTGCACCCCACCCACGCCGT-3′; reverse complement control, 5′-CACGGGCGGGGATGCGCGTGCGTG-3′, which is the reverse sequence of PNT100 and is 100 % homologous with the coding strand; mismatched control, 5′-CACGCACGCGCATCCTTGCCCGTG-3′ or 5′-CACGCACGCGCATCCTTGCCCATG-3′; randomer, 5′-NNNNNNNNNNNNNNNNNNNNNNNN-3′, where N represents a wobble; PNT100cy 5′-CACGCGCGCGCATCCCCGCCCGTG-3′. Oligonucleotides were purchased from TriLink, Dow, Sigma, or NITTO DENKO Avecia.
Lipids and transfection agents
NeoPhectin AT containing the cationic cardiolipin 1, 3-Bis-(1,2-bis-tetradecyloxy-propyl-3-dimethylethoxyammoniumbromide)-propane 2-ol was purchased from NeoPharm (Waukegan, IL). 1-Palmitoyl-2-oleoyl-sn-glycero-3 phosphocholine (POPC) and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) were purchased from Lipoid GmbH or Avanti Polar Lipids. Cholesteryl hemisuccinate (CHEMS) and cholesteryl-4-[[2-(4-morpholinyl)ethyl]amino]-4-oxobutanoate (MOCHOL) were produced by Merck & Cie (Shaffhuasen, Switzerland).
Preparation of encapsulated oligonucleotides and PNT2258
During preliminary screening, oligonucleotides were encapsulated in a variety of liposome compositions, including the cationic NeoPhectin AT system and into various amphoteric liposomes denoted as SMARTICLES
® (from Novosom AG, now Marina Biotech). While NeoPhectin spontaneously forms liposomes upon mixing with oligonucleotides, the other lipids that generate the amphoteric liposomes were mixed in ethanol and combined with an acidified (pH 4) aqueous solution of each oligonucleotide tested. The mixing process with the latter results in encapsulation in a 30 % ethanol suspension followed by dilution in an excess of phosphate-buffered saline with simultaneous neutralization to pH 7.5 [
26,
28,
29]. For PNT2258, a cross-flow injection technique was utilized to enable continuous mixing of the lipid ethanolic solution with aqueous PNT100 to encapsulate PNT100 in liposomes followed by an immediate pH shift to neutralize the mixture. Ethanol was removed by dialysis or diafiltration to exchange saline with sucrose (phosphate-buffered sucrose), followed by sterile filtration and filling into glass vials.
Characterization of the physicochemical properties of formulation solutions and dosing solutions of oligonucleotides
OD260-derived concentrations, representing total oligonucleotide content, were used to compare encapsulated oligonucleotide dosing solutions used for animal studies with PNT100, control oligonucleotides, and prototype PNT2258 formulations. Standard curves with each oligonucleotide were used along with spectral scans to confirm the OD260 contribution resulted from the oligonucleotides and not the liposome/lipid nanoparticles. The percent unencapsulated oligonucleotide was determined by OD260 following ultrafiltration using Centrisart 100 K cutoff membranes. Where indicated, the PNT2258 concentrations were reported as PNT100 content after correcting for purity as determined by ion exchange (IEX) chromatography. Reverse-phase HPLC was used to quantify lipids in the nanoparticles. Particle diameters and zeta potentials were determined by dynamic light scattering using a Malvern Nano ZS (Malvern, PA).
Cell culture and treatment with DNAi
Cell lines were maintained in suspension or monolayer cultures in media supplemented with 10 % fetal bovine serum (FBS). Breast (MDA-MB-231, BT-474, and T47D), melanoma (A375 and M14), prostate PC-3, lymphoma (Daudi-Burkitt’s, SU-DHL-6, and Pfeiffer), and mouse mammary NMuMG lines were purchased from ATCC®. WSU-DLCL2 and WSU-FSCCL lines were purchased from DSMZ GmbH (Germany) or provided by Ramzi Mohammad of Wayne State University. The cell lines were authenticated by the providers and were maintained under the recommended conditions for propagation and experimental use.
Unless otherwise noted, cell exposure studies utilized methylated oligonucleotides without formulation. For adherent lines, cells were seeded in 6- or 24- well plates or T-25 flasks (Corning Life Sciences) at 2.0 × 105 cells per flask in 5 mL media. One day after passage, the medium was replaced with fresh media containing the test oligonucleotides. Cultures were incubated at 37 °C in a humidified atmosphere of 5 % CO2. Cells were washed with 1× PBS and incubated with 0.25–1 % trypsin and 0.02 % EDTA to disperse the cells. The number of living and dead cells was assessed following 0.1 % trypan blue exposure, with the percentage of inhibition reported as a percent of live cells present in saline-treated controls. The effects of treatment on proliferation were assessed using MTT assay with 2,500 cells per well seeded in 96-well plates. Suspended cells were grown in 24-well plates and treated as described above. For PC-3 exposure studies, oligonucleotides were formulated with NeoPhectin. Typically, cells were exposed for 30 min or 6 h, washed, then the effects of treatment on proliferation were assess by CellTiter-Glo® (Promega) 48 h post-treatment or BCL-2 and GAPDH expression was measured by qPCR.
Pharmacokinetics, pharmacology, tissue distribution, and xenograft studies
Pharmacokinetic studies and xenograft animal studies
PNT2258 and its prototypes were evaluated as single agents or in combination in four different human tumor xenograft model systems. Two of the models were non-Hodgkin’s lymphoma models (WSU-DLCL2 (a diffuse large B-cell) and Daudi-Burkitt’s). The other xenograft models included A375 melanoma and the PC-3 hormone refractory prostate carcinoma. For WSU-DLCL2, female ICR SCID mice (Taconic) or female C.B-17 SCID mice were implanted subcutaneously with donor WSU-DLCL2 xenograft fragments in their flanks. A parallel set of WSU-DLCL2 xenograft mice were also used for tissue distribution studies. For Daudi-Burkitt’s, female C.B-17 SCID mice were implanted subcutaneously with 1 × 107 Daudi cells in their flanks. A parallel set of Daudi-Burkitt’s xenograft mice were also used for pharmacokinetic and tissue distribution studies. For A375, female nu/nu mice were implanted subcutaneously in the flank with 1 mm3 fragments. For PC-3, male SCID/NCr mice were subcutaneously implanted with 5 × 106 PC-3 cells. All xenograft tumor models were conducted through contract or research collaborations at MPI Research (PC-3), South Texas Accelerated Research Therapeutics (A375 and WSU-DLCL2), Piedmont Research (Daudi and A375), Karmanos Cancer Center (WSU-DLCL2), and Van Andel Research Institute (PC-3). BALB/C mice were purchased from Charles River Labs and used for pharmacokinetic studies. Test samples were provided in a blinded manner. All protocols and procedures employed in this work were approved by each testing centers’ Institutional Animal Care and Use Committee (IACUC).
Animal monitoring, tumor measurement, and data calculation
Upon attaining tumor volumes of 100–200 mm
3, animals were randomized into treatment groups. Thereafter, clinical signs, tumor measurements, and body weights were recorded 3–5 times per week. Tumor volume was calculated using the formula (
l ×
w
2)/2, where
l and
w are the length and width of the tumor, respectively. Animals were euthanized when tumor sizes reached 1,000 mm
3, 2,000 mm
3 or at approximately 60 days depending on the research organizations’ approved protocols. Efficacy endpoints, including time to tumor endpoint, tumor growth delay, and net log
10 cell kill, were calculated as follows using previously described methods [
30]. Gross cell kill was calculated using the following formula: [T-C (days)]/(3.32 * Td), where T-C is the tumor growth delay and Td is the tumor volume doubling time (days). Individual tumor volumes which decreased to <50 % of their volumes at treatment initiation for three consecutive measurements were considered partial regressions (PR). Individual tumor volumes that were not measurable for three consecutive measurements were considered complete regressions (CR). Complete regressions persisting until the end of the study were considered tumor-free survivors (TFS). Data and statistics were analyzed using Prism 5.0 (GraphPad; San Diego, CA) and Microsoft Excel.
Plasma measurement of PNT2258 by hybridization–ligation and plasma immune markers
Whole blood was collected in K2EDTA-coated tubes, placed on ice, centrifuged to obtain plasma and stored at −80 °C until analyses. Samples were treated with 10 % (v/v) Tween-20 detergent and heated to 90 °C to liberate PNT100 from PNT2258 then diluted fourfold with a template probe (complementary and specific to the entire sequence of PNT100) labeled with biotin on its 3′-end and a 9-mer overhang to the opposing end. The solution was incubated at 37 °C for 1 h in NeutrAvidin-coated plates, prior to the addition of a mixture containing a digoxigenin-label signal probe which ligates the 3′ terminus of PNT100 with the 5′end of the ligation probe. Unbound ligation probe was washed away prior to antibody (targeting digoxigenin) addition and conjugation to alkaline phosphatase. AttoPhos® substrate was added and the reaction terminated with EDTA solution prior to fluorescent signal measurement (excitation: 435 nm; emission: 555 nm). The lower limit of quantitation (LLOQ) was 3 ng/mL using PNT2258 as a standard. Multiplex immunoassays of mouse plasma obtained from WSU-DLCL2-tumored animals 8 h post-PNT2258 dose were assayed in triplicate per the Affymetrix Procarta Mouse 37-plex kit protocols (Fremont, CA) and visualized using a Luminex 100 IS System (Luminex Corporation, Austin, TX). Analyte concentrations were calculated from the standard curves using Bio-Plex Manager 4.1.1 (Bio-Rad Laboratories, Hercules, CA). Statistical analysis was done using Student t statistic; P values <0.05 were considered significant.
Pharmacodynamic sampling of tumors and PNT2258 tissue levels
Tumors and organs were collected, snap frozen, weighed, and stored until analyses. Tissue levels of PNT2258 were assessed by two independent labs, Charles River Labs and Helix Diagnostics, using the hybridization–ligation assay described above for plasma analyses (LLOQ: 50 μg/g of tissue) or by capillary gel electrophoretic detection (LLOQ: 5 ng/g of tissue). PNT2258 levels in xenograft tumors were measured through direct hybridization with capture and extender probes that recognize only PNT100 amidst the total RNA extract (LLOQ 300,000 copies of PNT100). Tumor homogenates were prepared by pulverizing tumors under liquid nitrogen, followed by homogenization in 900 μL of homogenizing solution (Affymetrix) supplemented with 9 μL of proteinase K (50 mg/mL). The homogenates were incubated at 65 °C for 30 min, then clarified by centrifugation, and stored at −70 °C until analyses. Tumors excised from animals treated with PNT100 formulated with NeoPhectin, PNT100R formulated with NeoPhectin, NeoPhectin alone, or sucrose (vehicle control) were pooled into groups, dissected mechanically into single cell suspensions, and subjected to protein analysis by Western blot using antibodies (Santa Cruz Biotechnology) along with total protein quantitation by BCA assay (Pierce).
Discussion
Oligonucleotide candidates modulating gene expression may be targeted at the level of either RNA or genomic DNA. Approaches to modulate genomic DNA include triplex, quadruplex oligonucleotides, methylated forms of DNA or RNA, or mismatched or single-base modifications of the DNA or RNA oligonucleotides. These methodologies hybridize to transcription start sites to interfere with transcription, cause DNA cleavage, homologous recombination or act to stimulate inherent DNA repair mechanisms to correct the mismatches triggered by the hybrid double-stranded DNA/DNA or chimeric RNA/DNA interactions [
38,
39]. To our knowledge, none of these DNA-targeted approaches have progressed into the clinic. In contrast, single- or double-stranded RNA-targeted oligonucleotide approaches have progressed to the point of providing clinical success reviewed in [
40] and [
41]. These include antisense or siRNA that cause mRNA cleavage and disruption of the RNA translational machinery, RNA modulation agents that correct gene defects by exon skipping, or microRNA miRs and antimiRs agents that regulate the expression of multiple pathways by replacing absent sequences or antagonizing sequences, respectively. A common feature among these approaches is the use of chemical modifications (e.g., phosphothioate, 2’-modified nucleic acids including MOE modifications and cET, conformationally restricted nucleic acids bases including LNA and inverted bases) to enhance activity, pharmacokinetics, and pharmacodynamics [reviewed in [
42] ]. In addition, in some cases, the active RNA sequences are not 100 % complementary to their target mRNA.
The 24-base BCL-2-targeted oligonucleotide, PNT100 represents a new class of DNA therapeutics being developed that is distinct from other nucleic acid approaches. PNT100 is an unmodified phosphodiester DNA sequence that is 100 % complementary to its homologous sequence of genomic DNA. Our data demonstrate equivalent antitumor activity of mePNT100 and unmethylated PNT100, but no antitumor effects with control scrambled, randomer, or reverse complement sequences. Moreover, in cellular studies, PNT100’s antiproliferative activity was not affected by a methyltransferase inhibitor, suggesting that methylation is not required for activity. Virtually identical immune response to PNT2258 and its scrambled control in xenografted animals suggests that PNT100’s sequence specificity and not immunogenicity drives BCL-2 modulation and antitumor activity. It is recognized that there are limitations to truly defining control sequences for PNT100. As such, there is the possibility that interactions between the liposomes and specific oligonucleotide sequences may be responsible for the differential effects seen. However, if these effects exist, they should be dose dependent and should have been observed at doses of 10 mg/kg encapsulated oligonucleotide, a dose level that shows antitumor activity for PNT2258. Therefore, by demonstrating, the specificity of PNT100’s pharmacodynamic activity using two independent liposomal systems compared to empty liposomes, and no effect with encapsulated control sequences suggests this is unlikely.
We propose that using unmodified sequences complementary only to a target gene region may offer advantages. The sequence of PNT100 does not possess toll-like receptor (TLR) immunostimulatory CpG motifs, a property that precludes the need to use chemical modifications to mask immune recognition. Further, the observation that PNT100 cannot be measured unless the liposomes are disrupted suggests fully encapsulated PNT100 may also contribute to immune avoidance. Indeed, no significant changes in immune-stimulatory cytokines or clinical signs of anaphylaxis observed following dosing of PNT2258 in patients with advanced solid tumors lends support to this approach [
27]. Clinically, dose-dependent BCL-2-targeted effects were observed including reductions in lymphocyte counts [
27]. PNT2258 was well tolerated in the clinic with doses up to 150 mg/m
2 (equivalent to ~4 mg/kg).
The lipid composition used for PNT2258 with its amphoteric pH-tunable nature has several distinct features. First, the overall particle charge is anionic at blood pH which prevents aggregation with blood components and eliminates the need for PEGylation. Second, the composition is sterol rich with CHEMS and MOCHOL, providing the cholesterol backbone to anchor the pH responsive headgroup compositions. Third, ether linkages often used to anchor polyethylene glycol (PEG) or pH responsive headgroups are not used. Finally, the amphoteric nature enables a transition of charge from cationic to net neutral to anionic or vice versa depending on the microenvironment [
24,
26]. For example, during transient acidic ethanolic conditions used during formulation, there is an overall positive charge to the mixture enabling the efficient encapsulation of the negatively charged PNT100. A shift to pH 7.5 renders an overall anionic charge to the particle, repelling unencapsulated oligonucleotide, which is removed during ultrafiltration. The PNT2258 drug product typically has a zeta potential of −40 mV. The anionic surface characteristics and unique lipid composition enables stable encapsulation of the oligonucleotide, without surface PNT100, thereby obviating the need for PEG spacers to prevent aggregation or immune recognition. Additionally, the anionic nature likely alters opsonin adhesion or activation (e.g., complement factors) compared with cationic carriers [
43]. It may also attract exchangeable apolipoproteins such as apoE to facilitate cellular uptake through lipoprotein and/or other receptor-mediated uptake pathways [44 and references therein]. The novel lipid MOCHOL has been shown to have a pKa of approximately 6.5 [
26], which is within the range identified for facilitating endosomal escape [
44]. We believe the unique features of the lipids and the ratios of the liposomal system permits PNT2258, under the appropriate pHs and ionic environments to facilitate endosomal escape, a key feature enabling PNT100 access to nuclei.
PNT2258 exhibits good systemic exposure following intravenous administration and demonstrates antitumor activity against xenografted tumors. Our working hypothesis centered on the assumption that BCL-2 expression is generally low in non-cancerous cells and entry of PNT100 into non-cancerous tissue would not interfere with normal homeostasis. Similarly, in tumor types driven by BCL-2 transcription, PNT2258 should demonstrate good single-agent activity (compare WSU-DLCL2 with Daudi-Burkitt’s), but should potentiate the combination drug in tumor types with high BCL-2 expression if BCL-2 resistance is implicated in disease resistance (e.g., Daudi, PC-3, and A375). The key challenge was to achieve sufficient systemic exposure while avoiding dose-limiting lipid toxicity. Therefore, an overriding goal was to ensure that sufficient lipid doses (>100 mg/kg lipid, equivalent to 4 mg/kg PNT2258) could be safely administered to animals to overcome the reticuloendothelial system (RES) clearance mechanisms and enable sufficient exposure to be efficacious without causing observable toxicities. Lipid particle numbers are proportional to the nanoparticle diameters [
45] and both these parameters greatly influence circulation lifetimes [
46], hepatic uptake [
47], and importantly, tolerability and access to extrahepatic tissues in the absence of targeting ligands [
48,
49]. Moreover, the pharmacokinetics and tissue access of liposomes are also influenced by surface charge and recognition which are a function of particle size and composition [
46]. Liver fenestrae and sinusoids across species represent physical barriers of ~100 nm such that nanoparticles with an average diameter size range of ~130 nm will be primarily removed by resident macrophages and will not readily access hepatocytes, which represent the population of liver cells to which lipid toxicity is attributed [
47]. As a result, the liver is generally resilient and can tolerate lipid doses of 100 mg/kg or greater, depending on the rate of delivery, as evidenced by the safety and tolerability of parenteral nutrition containing daily lipid doses of up to 60 g (administered at 2.5 g/h). These factors were taken into account during the development of PNT2258 to ensure that a broad therapeutic window could be identified. This is evidenced by the relatively flat dose response of PNT2258 above 10 mg/kg (EC50 estimated to be between 3 and 10 mg/kg across the models) and tolerability with daily doses of 30 mg/kg (or 750 mg/kg lipid). We speculate that the unique mixture of lipids (that lack ether linkages or PEG) and particle size of the amphoteric liposomes contributes to the therapeutic activity of PNT2258 in several ways. These include imparting stability [
24,
26], surface characteristics which prolong circulation times in blood, and permit metabolic breakdown to enable repeated dosing. Toxicology findings in preclinical studies with PNT2258 showed dose-dependent toxicities that were reversible, and attributable to high doses of lipid (>300 mg/kg) rather than PNT100 [
27].
PNT2258 shows broad activity against a variety of tumor types with robust single-agent activity in DLBCL where BCL-2 transcription drives the genesis and survival of the tumors. Antitumor activity and long-term survival in combination with docetaxel or rituximab in chemo-resistant tumors were also demonstrated. BCL-2 overexpression is linked to lower overall survival and adversely influences progression-free survival in subtypes of chemotherapy naïve NHL patients [
50]. After front-line R-CHOP therapy, 30–50 % of these patients fail to respond, with BCL-2 expression remaining high in patients with germinal center DLBCL, suggesting the need for BCL-2-targeted interventions. High-risk patients with revised International Prognostic Index (R-IPI) scores of 3–5 or those with double-hit (BCL-2/MYC positive) phenotypes demonstrate even worse prognosis with short survival timelines [
51]. Moreover, there is a high correlation between BCL-2 expression and the presence of the t(14;18) translocation. Our preclinical data show that PNT2258 demonstrates good single-agent activity and show additive effect with rituximab in DLBCL where the t(14;18) rearrangement exists (WSU-DLCL2) and synergistic activity against Daudi-Burkitt’s, suggesting that targeting BCL-2 can potentiate other therapies. Early clinical data indicate antitumor effect in patients whose tumors may be BCL-2 dependent [
27]. Similarly, combination with docetaxel results in an additive effect in prostate and melanoma models. These results support the rationale for combining PNT2258 with approved agents such as docetaxel and rituximab to potentiate their cytotoxic activity in tumor types where BCL-2 plays a role in resistance.