Non-viral Gene Delivery of HIF-1α Promotes Angiogenesis in Human Adipose-Derived Stem Cells

Abstract

Stable and mature vascular formation is a current challenge in engineering functional tissues. Transient, non-viral gene delivery presents a unique platform for delivering genetic information to cells for tissue engineering purposes and to restore blood flow to ischemic tissue. The formation of new blood vessels can be induced by upregulation of hypoxia-inducible factor-1α (HIF-1), among other factors. We hypothesized that biodegradable polymers could be used to efficiently deliver the HIF-1α gene to human adipose-derived stromal/stem cells (hASCs) and that this treatment could recruit an existing endogenous endothelial cell population to induce angiogenesis in a 3D cell construct in vitro. In this study, end-modified poly(β-amino ester) (PBAE) nanocomplexes were first optimized for transfection of hASCs and a new biodegradable polymer with increased hydrophobicity and secondary amine structures, N’-(3-aminopropyl)-N,N-dimethylpropane-1,3-diamine end-modified poly(1,4-butanediol diacrylate-co-4-amino-1-butanol), was found to be most effective. Optimal PBAE nanocomplexes had a hydrodynamic diameter of approximately 140 nm and had a zeta potential of 30 mV. The PBAE polymer self-assembled with HIF-1α plasmid DNA and treatment of hASCs with these nanocomplexes induced 3D vascularization. Cells transfected with this polymer-DNA complex were found to have 10⁶-fold upregulation HIF-1α expression, an approximately 2-fold increase in secreted VEGF, and caused the formation of vessel tubules compared to an untransfected control. These gene therapy biomaterials may be useful for regenerative medicine.

Graphical Abstract

4. Introduction

The field of therapeutic cellular reprogramming is rapidly expanding in response to advances in genetic and biological technologies ranging from CRISPR to the Human Genome Project [1]. With this influx of research and technology, there arises a demand for safe, reproducible, and scalable cellular gene delivery technologies. Human adipose-derived stromal/stem cells (hASCs) offer potential as a sustainable source for cellular reprogramming due to their accessibility, ease of culture, multipotency, low immunogenicity, and ability to home to sites of inflammation in vivo [2–4]. Because of this, reprogrammed hASCs are being used for a multitude of applications including tissue engineering, cancer immunotherapy, drug delivery, and regenerative medicine [5–9]. In the field of tissue engineering, hASCs have demonstrated the ability to differentiate into multiple cell lineages and are able to be transplanted across immune haplotypes with low immunogenicity [6,9]. Additionally, these cells can be modified genetically to promote certain cell lineages or combat barriers to tissue regeneration, such as hypoxia [10–12].

Oxygen tension is known to have a major effect on proliferation, differentiation, cell fate, and secretion of vascular cytokines [13,14]. Together, these processes can dictate outcomes of tissue regeneration following cell implantation. The HIF-1α transcription factor is the major biochemical molecular oxygen sensor and has downstream effects on cytokine secretion and genetic activation [15,16]. At physiologic normoxia (referring to normal tissue oxygenation), HIF-1α is hydroxylated by prolyl hydroxylase (PHD), an oxygen-dependent process, and degraded in the cytoplasm. In hypoxia, which refers to levels lower than the homeostatic oxygenation, hydroxylation of HIF-1α by PHD is inhibited by the low levels of oxygen [17]. The intact HIF-1α translocates to the nucleus and interacts with transcription factors to promote genes involved in survival, metabolism, and growth factor secretion [17]. Of particular interest to the field of tissue engineering is the secretion of vascular endothelial growth factor (VEGF), a key mediator of angiogenesis that is induced by HIF-1α, into surrounding tissues, which can further improve regeneration [18]. In this study, we utilize gene delivery to increase the expression of HIF-1α in hASCs to promote the formation of vascular structures.

As the fields of gene delivery and cellular therapy become more established, a need arises for low risk, effective gene delivery technologies for cellular therapies. Typically, gene delivery is divided broadly into two categories: viral and non-viral [19–21]. Viral gene delivery uses biological viruses modified for safety in order to transfer genes to cells, while non-viral methods deliver DNA plasmids to the cell using specialized materials. Though viral delivery is often associated with increased efficiency and prolonged gene expression, there are significant concerns about its safety and efficacy in the clinic due to high risk of genotoxicity, limited DNA packaging capacity, complex production processes, cytotoxicity, immunogenicity, and mutagenicity [21–23]. Non-viral gene delivery vectors, while usually less efficient, provide a safer alternative as well as greater ease of scale up for future manufacturing and clinical use [21–23]. In contrast to viruses, which have evolved for efficient gene delivery, non-viral vectors must be designed to overcome delivery barriers through their material properties [24]. Within the field of non-viral gene delivery, there is a large range of engineered materials with properties that aim to address these remaining challenges of cell internalization, cytotoxicity, and delivery efficiency. Biodegradable cationic polymers that use electrostatic interactions to couple with negatively charged DNA and create polymer-DNA polyplexes have shown potential to be a safe, controlled, and nontoxic alternative to viral gene delivery [24–26]. Among cationic polymers, poly(β-amino ester)s (PBAE)s with hydroxyl side chains and primary amine end-groups have demonstrated their ability to effectively deliver DNA to a variety of cell types as well as their biodegradability and low cytotoxicity [2,27,28].

In addition to careful selection of materials, it is important to consider the biology of the delivery target when attempting non-viral gene transfection in vitro. Transfection efficiency and cytotoxicity depend not only on the delivery materials but also on biological factors such as permeability of the membrane, frequency of endocytosis, ease of endosomal escape, possibility of lysosome degradation, and extent of nuclear uptake within the cell [22,29,30]. We have developed a high-throughput polymer library screening approach based on synthetic PBAEs with differing side chain, backbone, and end-cap combinations [2,23,28]. With this method we have been able to synthesize hundreds of polymers that have demonstrated improved efficiency and reduced cytotoxicity across many cell types compared to commercial transfection reagents such as Invitrogen Lipofectamine™ (Thermo Fisher) [2,23]. We have also developed a freeze-drying method to store PBAE-DNA polyplexes long-term for increased practicality of future clinical use, and we showed that PBAE polymer transfections can retain hASC functionality better than viral methods [2,23]. Building on this emerging nanobiotechnology approach, we aimed here to improve transfection efficacy to hASCs, reduce cytotoxicity, and produce a functional, clinically relevant outcome by synthesizing a novel PBAE polymer to deliver HIF-1α to hASCs and to promote the formation of vascular structures in vitro.

Previous studies using non-viral gene delivery for angiogenesis have included non-viral delivery of VEGF using cationic polymers and incorporation of HIF-1α DNA by electroporation and poly-L-lysine (PLL) polymers [31–33]. These studies have shown limited efficacy, and the use of PBAE polymers improves transfection compared to PLL and electroporation and can also enhance safety due to biodegradability and lower cytotoxicity. Likewise, the delivery of HIF-1α is advantageous compared to VEGF due to its ability to regulate multiple downstream effectors of angiogenesis, including VEGF among other key factors [16]. The investigation of a new monomer endcap structure allows for the synthesis of PBAE polymers that have not been previously reported and the creation of enhanced nanocomplexes. The polymer formulation (44E49) focused on in this study offers increased hydrophobicity and an additional secondary amine group when compared to leading previous PBAE formulations and was hypothesized to improve non-viral gene delivery to hASCs.

5. Materials and Methods

5.1. Experimental Design

A diagram of the experimental design is shown in Figure 1. Poly(β-amino ester) (PBAE)s were synthesized by combinatorial chemistry of different backbone monomers, side chains, and end-caps, then complexed with a green fluorescent protein (GFP) DNA plasmid. These polymer-DNA polyplexes were used to transfect human adipose-derived stem cells (hASCs), and the polymer with the highest transfection yield was selected as the optimal polymer for hASC delivery moving forward.

Figure 1: Experimental design.

hASCs were transfected non-virally in monolayer with HIF-1α polyplexes; transfected cells were encapsulated in a fibrin gel, and the formation of vascular structures was observed over time.

hASCs were then transfected with a HIF-1α plasmid and encapsulated in fibrin gels. Samples were collected at days 1, 2, 4, and 7 after transfection for RT-PCR, ELISA, and whole mount immunostaining. Whole mount immunostaining was performed using a nuclear stain and a stain for CD31, an endothelial cell marker, to determine any changes in vascular morphogenesis following HIF-1α transfection.

5.2. Polymer Synthesis

Poly(β-amino ester)s were synthesized from combinations of backbone, side chain, and end-cap monomers using a two-step process that has been described previously [34] and is shown in Figure 2A–B. Monomers were purchased from Alfa Aesar (B4, S3, S4, S5, E7), Sigma Aldrich (E6), Monomer-Polymer and Dajac Labs (B5), and EMD Millipore (E49). Each polymer was synthesized in a neat solution from one backbone monomer (B4 or B5) mixed with a side chain monomer (S3, S4, or S5). Molar ratios of backbone to side-chain polymers were either 1.1:1 or 1.2:1. The reaction mixture was then stirred at 90°C for 24 hours in the dark. Then, the resulting acrylate-terminated base polymers were dissolved in tetrahydrofuran (THF), mixed with a solution of end-cap in THF at a final concentration of 100 mg/mL base polymer and 0.2 M end-cap, and stirred at room temperature for 1 hr at 600 rpm. The final PBAEs were then purified by precipitation into anhydrous diethyl ether, centrifuged at 3200 rcf for 5 min at 4°C to isolate the polymer, and washed twice with diethyl ether. The resulting polymers were then dried under vacuum for 48 hr, dissolved in anhydrous dimethyl sulfoxide (DMSO) at 100 mg/mL, and stored with desiccant at −20°C in small aliquots to minimize freeze-thawing. The nomenclature of each PBAE follows the numbering of the monomers; for example, the polymer synthesized from B4, S4, and E49, N’-(3-aminopropyl)-N,N-dimethylpropane-1,3-diamine end-modified poly(1,4-butanediol diacrylate-co-4-amino-1-butanol), is referred to as “44E49.”

Figure 2: Two-step PBAE synthesis reaction.

A) Backbone (B4, B5), side chain (S3, S4, S5), and end-cap (E6, E7, E49) monomers are used to generate PBAEs with distinct properties. The structure of representative polymer 44E49 is shown. B) Molecular weight and polydispersity of polymer 44E49 determined by GPC

5.3. Polymer Characterization

HA-HIF1alpha P402A/P564A-pcDNA3 plasmid was a gift from William Kaelin (Addgene plasmid # 18955; http://n2t.net/addgene:18955 ; RRID:Addgene_18955) [35]. Nanoparticle size distribution and mean nanoparticle size of resulting complexes were measured using nanoparticle tracking analysis (NTA, Nanosight NS300, Malvern Panalytical, Malvern, UK). Nanocomplexes were made in 25 mM NaAc as described above, then diluted 1:5 (v/v) in 0.1x PBS. For stability measurements, the z-average mean diameter of the diluted nanocomplexes was measured by dynamic light scattering (DLS), and the zeta potential was measured by electrophoretic mobility (Zetasizer NanoZS, Malvern, UK). Data are presented as mean +/− standard deviation of three independent samples. Nanocomplexes were made in 25 mM NaAc as described above, then diluted 1:5 (v/v) in 1xPBS to a final concentration of 10 μg/mL DNA (200 μg/mL PBAE). Diluted nanocomplexess were incubated at 37C with agitation. At predetermined time points, nanocomplexes were diluted 1:1 in water, and the z-average mean diameter of the diluted Nanocomplexes was measured by DLS and the zeta potential by electrophoretic mobility. To determine polymer/DNA complex efficiency, nanocomplexes were made in 25 mM NaAc as described above, then diluted 1:5 (v/v) in 1xPBS or NaAc to a final concentration of 10 μg/mL DNA (200 μg/mL PBAE). Samples were mixed with 30% glycerol as a loading buffer at a 1:5 ratio (v/v) of loading buffer to nanocomplexess, then loaded into a 1% agarose gel (UltraPureTM Agarose, ThermoFisher Scientific) with 1 μg/mL ethidium bromide. Each well contained 125 ng DNA. The gel was run for 40 min under 100 V, then visualized by UV exposure. To determine the timecourse of DNA release, s were made in 25 mM NaAc as described above, then diluted 1:5 (v/v) in 1xPBS or 1xPBS with 10% FBS to a final concentration of 10 μg/mL DNA (200 μg/mL PBAE). Diluted nanocomplexes were incubated at 37C with agitation. At predetermined time points, nanocomplexes were mixed with sucrose to a final concentration of 30 mg/mL sucrose (9.4 μg/mL DNA) and frozen at −80C. Once all time points were taken, frozen nanocomplex aliquots were thawed and mixed with 30% glycerol as a loading buffer at a 1:5 ratio (v/v) of loading buffer to NANOCOMPLEXs. Samples were then loaded into a 1% agarose gel (UltraPureTM Agarose, ThermoFisher Scientific) with 1 μg/mL ethidium bromide, with 117.5 ng DNA (2.35 μg PBAE) loaded per well. The gel was run for 30 min under 80 V, then visualized by UV exposure. To confirm spherical polymer-DNA complex morphology and size, complexes were negative-stained with uranyl acetate prior to imaging, and images were acquired on a Philips/FEI BioTwin CM120 transmission electron microscope (TEM).

5.4. hASC Isolation and Culture

Human subcutaneous adipose tissue was obtained in the form of lipoaspirate from a female Caucasian donor undergoing elective surgery, with written informed consent, and with Institutional Review Board approval. ASCs were isolated as previously described [36,37] and their characteristics have been reported previously [38]. Briefly, tissue was digested with collagenase type 1 (1 mg/mL; Worthington Biochemical Corp.) to isolate the stromal vascular fraction of cells. Cells were filtered through a 100μm filter. Cells were plated onto tissue culture plastic and were termed “passage 0 ASC.” Cells were passaged with trypsin when they reached 80-90% confluence, termed “passage 1 ASCs”. Upon reaching 80-90% confluence, these cells were lifted with trypsin and used as “passage 2 ASCs” for all experiments with ASCs groups. Growth medium consisted of: high glucose DMEM (Gibco) with 10% fetal bovine serum (FBS; Atlanta Biologicals), 1% penicillin/streptomycin (Gibco), and 1 ng/mL basic fibroblast growth factor (FGF-2; PeproTech).

Isolated passage 3 hASCs were seeded into 96-well plates at a density of 10⁴ cells/well in monolayer in 100 μL/well hASC growth media [high-glucose DMEM with pyruvate (Gibco 11995) with 1% penicillin/streptomycin, 10% fetal bovine serum, and 1 ng/mL basic fibroblast growth factor (bFGF)]. After transfection (described below), cells were incubated in monolayer at 37°C for 3-24 hours, trypsinized with 0.25% trypsin/EDTA (Thermo Fisher Scientific), and resuspended in 12 mg/mL fibrinogen (from bovine plasma, Sigma Aldrich, St. Louis, MO) in PBS at a cell density of 4,000 cells/μL. Fibrin gels 4 mm in diameter were then created by adding 3 μL of a 10 U/mL thrombin solution to 12 μL of the fibrinogen/cell solution and immediately mixing and pipetting into a PDMS mold in a 24-well plate. The media was changed from hASC growth media to endothelial growth media (EBM-2 + 6% FBS + 1% penicillin/streptomycin, Lonza Group, Basel, Switzerland).

5.5. In Vitro Optimization of hASC Transfection

hASCs were seeded into 96-well plates at a density of 10⁴ cells/well in monolayer and cultured in growth media overnight. Immediately prior to transfection, media was changed to 100 μL/well of serum-free high-glucose DMEM with pyruvate. PBAEs and GFP DNA (pEGFP-N1, Elim Biopharmaceuticals, Hayward, CA) were both diluted in 25 mM sodium acetate buffer (pH 5; NaAc) and mixed at polymer-DNA ratios of 30, 60, and 90 (w/w). The Polymer-DNA nanocomplexes were allowed to form for 10 minutes. Cells were then transfected by adding 20 μL nanocomplexes to each well at a final dose of 600 ng DNA/well and incubated for 2 hours at 37°C and 5% CO₂. After 2 hours, the media was changed to original serum-containing growth media. Cell viability was calculated using an MTS assay (Cell Titer 96, Promega Corporation, Madison, WI) 24 hr after transfection and was normalized to the viability of untreated cells. Transfection was measured using flow cytometry (Accuri C6, BD Biosciences with a Hypercyt high-throughput autosampler from Intellicyt, Ann Arbor, MI) 48 hours after transfection. Transfection efficiency is defined as the percent of cells that were GFP⁺. An example of cells transfected with GFP using polymer 44E49 is shown in Figure 3A–B. Normalized GFP fluorescence was calculated from the geometric mean fluorescence intensity of the cells. Transfection yield was calculated by multiplying (cell viability) x (transfection efficiency) x (normalized GFP fluorescence). After selection of the polymer with the highest transfection yield, differing transfection weight percentages and dosages of the polymer-DNA complexes were applied to hASCs, and transfection yield was optimized. The optimized polymer (44E49 polymer formed at a 20 w/w ration polymer/DNA and used at a 1x dose of 600 ng per well) was then compared to a variety of formulations and doses of Lipofectamine™ 2000 and 3000 (Thermo Fisher Scientific) (see Figure S1). After the polymer dosing and polymer-DNA w/w ratio were optimized with GFP, the polymer was complexed with the corresponding amount of HIF-1α DNA and used for subsequent experiments.

Figure 3: Optimization of hASC transfection with PBAE polyplexes.

Micrographs show A) phase contrast + GFP and B) GFP channel 24 hours post-transfection with polymer 44E49 at 20 w/w. Error bars represent the mean ± mean standard error. Scale bars represent 100 μm. C) Normalized cell viability was measured by MTS assay 24hr post-transfection. D) Transfection efficiency was measured as percent of cells transfected and E) normalized GFP fluorescence as a measure of transfected signal intensity. F) Transfection yield was calculated from viability and transfection. Statistical significance determined by one-way ANOVA tests and Dunnett’s multiple comparison post-tests. Post tests were performed between means of each group at each w/w and lipofectamine. Polymer compositions with higher means and significant difference from Lipofectamine are marked for significance. G) transfection yield of the polymer 44E49 at varying doses and polymer-DNA weight ratio.

5.6. Evaluation of Vascular Morphogenesis

5.6.1. Quantitative RT-PCR

Quantitative RT-PCR was performed on hASCs transfected with HIF-1α and encapsulated in fibrin gels at days 1, 2, 4, and 7 post-transfection. RNA was extracted by adding 10x gel volume of TRIzol (Thermo Fisher Scientific), homogenizing with a sterile pestle, and following the manufacturer’s instructions. cDNA was synthesized using an iScript cDNA synthesis kit (BioRad) and a thermocycler (BioRad MyCycler Thermo Cycler). RT-PCR (Applied Biosystems® StepOne™ Real-Time PCR System, Thermo Fisher Scientific) was performed on HIF-1α, VEGF, and GAPDH as a housekeeping gene for reference. HIF-1α (5’-TGCTCATCAGTTGCCACTTC-3’, 3’-CTTCACCGTTGACTACTCGT-5’), VEGF-A (5’-GCCTTGCCTTGCTGCTCTA-3’, 3’-ATCTCGTCGTTCCGTTCCG-5’), and GAPDH (5’-CACCCACTCCTCCACCTTTGA-3’, 3’-AGTTTCCACCTCCTCACCCAC-5’) primers were purchased from OriGene (Rockville, MD).

5.6.2. Assessment of VEGF Production

For n = 4 samples, gels containing non-virally transfected hASCs were incubated with 0.5 mL of endothelial growth media in a 24-well plate, and VEGF secreted into the media was measured 1, 2, 4, and 7 days after transfection with HIF-1α using a Human VEGF₁₆₅ Standard TMB ELISA kit (PeproTech, Rocky Hill, NJ).

5.6.3. Immunocytochemistry

Samples for whole-mount immunostaining were collected 1, 2, 4, and 7 days post-transfection. Samples were fixed in 4% paraformaldehyde (PFA) for 3 hr at 4°C. Samples were then washed with PBS and blocked with 10% serum in 0.1% Tween/PBS (Polyethylene glycol sorbitan monolaurate, Polyoxyethylenesorbitan monolaurate, Sigma-Aldrich, St. Louis, MO) solution for 3 hr. Samples were stained with 1:100 endothelial cell marker CD31 (PECAM-1, Clone M-20, Santa Cruz Biotechnology, Dallas, TX) in 0.1% Tween/PBS overnight and washed 3x with 0.1% Tween/PBS for 1hr each. Alexafluor 488 (Jackson ImmunoResearch) secondary antibody was added at 1:100 in 0.1% Tween/PBS and incubated at 4°C overnight. Samples were then washed 2x for 1hr with 0.1% Tween/PBS. Samples were washed a final time for 1hr with 1:2000 nuclear stain 4′,6-diamidino-2-phenylindole (DAPI, Invitrogen, Carlsbad, CA) in 0.1% Tween/PBS. Images were taken on a Zeiss LSM800 GaSP confocal microscope at 10x magnification with constant slice thickness and gain settings. Three images were taken for each gel in each group. Each group had 3-4 gels. For gels without vessels, or with too few vessels for 3 images in each gel to contain vessels, images without vessels were included in quantification measurements. Measurements of tubule density, tubule length, tubule width and number of tubule junctions were acquired by image thresholding and manual, double-blinded ImageJ (NIH) measurement.

5.7. Statistical Analysis

All experiments were performed with a minimum of n = 3 replicates. Error bars in all figures indicate standard error of the mean. GraphPad Prism was used for all statistical analysis. 2 sample Student’s t-tests were used to measure differences between two groups and one-way ANOVA with Dunnett’s post-tests were used to measure differences when there were multiple comparisons. Two-way ANOVA was used to assess significant differences between groups over multiple time points using Tukey’s post-tests. In all cases, α = 0.05 were used to assess statistical significance (* indicates p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001).

6. Results

6.1. In Vitro Optimization of hASC Transfection

Transfection optimization results are shown in Figure 3. A select group of PBAEs were chosen for the library screen due to past performance with similar cell types [2]. Acceptable cell viability for further investigation was determined to be ≥60% cell viability after 24 hours. Top-performing polymers from the polymer library screen in Figure 3 are 44E6 (60 w/w) at 1× dose, 44E49 (30 w/w) at 1x dose, and 53E6 (30 w/w) at 1x dose. All three polymers had cell viabilities >60% (Figure 3C) and transfection efficiencies >40% (Figure 3D). 44E49 significantly outperformed 44E6 and 53E6 in normalized GFP fluorescence (Figure 3E), a measure of transfected signal intensity, leading to the highest transfection yield overall (Figure 3F). We then moved forward with polymer 44E49 (Figure 2A) and optimized transfection yield by varying polymer-DNA weight ratios and polymer-DNA dosage (Figure 3G). The highest transfection yield was a 1× dose at 20 w/w (Figure 3G).

For comparison to a leading commercially-available standard, we tested a range of Lipofectamine™ products and mixtures (Supplementary Figure S1). Lipofectamine™ 3000 performed the highest in all areas and was included in Figure 3. Lipofectamine™ 3000 had transfection efficiency <20%, viability approximately 80%, and the highest transfection intensity of all groups. However, due to its low transfection efficiency in these hard-to-transfect cells, its overall transfection yield was approximately 3 times lower than that of 44E49, with an unpaired t-test indicating statistical significance between the two groups (p<0.001).

6.2. Nanoparticle Characterization

The synthesis procedure, general structure, and molecular weight of selected polymer 44E49 are shown in Figure 2A–B. Nanoparticle tracking analysis (NTA) reveals the polymer-DNA polyplex hydrodynamic diameter distribution from approximately 80-220 nm with mean hydrodynamic diameter of 140 ± 20 nm (Figure 4A). Polymer-DNA polyplex zeta potential was 30 ± 2 mV in 1:1 (v/v) of 0.1x PBS (Figure 4B). Transmission electron microscopy of dried nanoparticles shows spherical morphology and slightly smaller size than the hydrated nanoparticles (Figure 4F).

Figure 4: Nanoparticle characterization.

A) The size distribution for polymer 44E49-HIF1α complexes (20 w/w) was measured by NTA, with mean hydrodynamic diameter of 140 ± 20 nm. B) The mean particle zeta potential is 30 ± 2 mV. C) in vitro DNA release of polymer-DNA complexes in PBS and (D) serum. E) DNA gel retardation assay shows polymer-DNA complexes at varied w/w ratio. F) TEM shows dried 44E49-HIF1α polymer-DNA complexes of comparable size to hydrated complexes by NTA. Scale bar represents 100 nm. G) Stability of 44E49-HIF1α polymer-DNA complex zeta potential (H) and 44E49-HIF1α polymer-DNA complex hydrodynamic diameter.

Figure 4C and Figure 4D show stability and in vitro DNA release of 44E49 polymer-DNA complexes (20 w/w) in PBS and serum respectively. There is no detectable DNA release in PBS over at least 14 hrs, with some release detectable after 24hrs due to polymer hydrolysis, as shown in Figure 4C, but DNA release in serum begins at 8 hours and increases over a 24 hour period as shown in Figure 4D. When nanocomplexes are evaluated over a range of weight ratios (5 w/w – 90 w/w), there is no visible separation of polymer and DNA under any of the conditions, indicating full complexation in all w/w groups (Figure 4E). The polymer-DNA complex hydrodynamic particle size (Figure 4G) and zeta potential (Figure 4H) both remain relatively constant over 4 hrs. At later time points, there is a gradual decrease in zeta potential and increase in particles size, indicating aggregation.

6.3. Evaluation of Vascular Morphogenesis

Figure 5A shows RT-PCR results for both HIF-1α and downstream VEGF mRNA expression normalized to a non-transfected control and housekeeping gene GADPH 1, 2, 4, and 7 days post-transfection. The left axis corresponds to HIF-1α expression and the right axis corresponds to VEGF expression. Both genes were significantly upregulated from the control with peak expression 2 days after transfection. Peak expression of HIF-1α was 10⁶-fold over the control, and peak expression of VEGF was 6-fold over the control. Expression decreases after 2 days. Figure 5B shows the concentration of secreted VEGF per single gel in 500 μL of media. Samples were collected 1, 2, 4, and 7 days after transfection. Two-way ANOVA tests comparing the transfected group to the non-transfected group show that the groups are statistically significantly different (p<0.0001). Secretion was observed to plateau between 4-7 days, with peak VEGF secretion by the HIF-1α-transfected group roughly double that of the control group (470 ± 40pg/mL and 230 ± 40 pg/mL, respectively).

Figure 5: HIF-1α and VEGF overexpression after transfection with HIF-1α DNA.

A) RT-PCR shows that hASCs transfected with PBAE 44E49-HIF-1α complexes (20 w/w) have elevated levels of both HIF-1α and VEGF mRNA 1, 2, 4, and 7 days post-transfection. B) ELISA quantification shows that HIF-1α transfected cells in gels secrete more VEGF than control cells at 1, 2, 4, and 7 days post-transfection. Statistical significance determined by two-way ANOVA tests and Tukey’s post-tests comparing the transfected group to the non-transfected group at each time point (*** indicates p<0.001).

Whole-mount immunostaining in Figure 6A shows vessel development and morphology 1, 2, 4, and 7 days post-transfection. Morphology measurements shown in Figure 6B include average number of tubules/image field, average tubule length, average vessel width, and average number of junctions/tubule. Non-transfected groups did not display any quantifiable tubule formation. For the experimental group, the number of tubules per field increased linearly over time, the average tubule length remained approximately constant, the vessel width plateaued at roughly 10-15 μm, and the number of junctions/tubule increased sharply after 2 days.

Figure 6: HIF-1α transfection of hASCs enhances vessel formation.

A) Representative images of whole-mount immunostaining samples collected 1, 2, 4, and 7 days post-transfection with PBAE 44E49-HIF-1α complexes. DAPI channel showing blue for nuclear stain and green channel for endothelial marker CD31. Scale bars represent 100 μm B) Quantification of vessel number and morphology measurements. Statistical significance determined by Two-way ANOVA with Tukey’s post-test comparing the transfected group to the non-transfected group at each time point (** indicates p<0.01; *** indicates p<0.001; **** indicates p<0.0001). Tests for significance were also performed using nonparametric t-tests between the transfected group at day 7 vs. transfected group at day 1 (indicated by the comparison bars).

7. Discussion

7.1. In Vitro Optimization of hASC Transfection

Genetic modification of human mesenchymal stem cells is traditionally mediated through viral based gene transduction, with major motivations for this route being delivery efficiency as well as length of delivered gene expression [4]. However, viral techniques are limited therapeutically due to safety, manufacturing, and mutagenicity concerns. Non-viral techniques typically offer lower efficiencies but greater ease of scale-up and greater safety. For this reason, we aimed to use a polymer library approach to screen for an optimal non-viral delivery vehicle that can efficiently transfect human adipose-derived stem cells (hASCs) and direct their function to form vasculature.

Mesenchymal stem cells are significantly heterogeneous, and transfection efficiencies are largely impacted by uptake pathways that vary among cell types. For mesenchymal stem cells, general endocytosis and membrane disruption have been indicated as the main mechanisms of cellular uptake of non-viral gene delivery vehicles [39]. Actin-mediated and clathrin-mediated cellular uptake pathways have also been indicated in MSC cellular uptake of positively charged nanoparticles regardless of the magnitude of particle surface charge [40]. Clathrin-mediated cellular uptake has been reported to correlate to transfection efficacy with PBAE polyplexes and exhibits a size and shape selection preference for spherical particles 100-150 nm in diameter [41,42], which matches the 44E49 nanoparticulates used in this study for hASC transfection. In comparing the most effective polymer (44E49) to less effective polymers for transfection of hASCs (such as 44E6 and 44E7 ), it is striking that the only difference is a small change of just a few atoms to the amine end group, beyond the shared aminopropyl linkage and the common poly(1,4-butanediol diacrylate-co-4-amino-1-butanol) backbone that composes the large majority of all 3 polymers.

After cell uptake, the encapsulated DNA must escape the endosome or lysosome, reach the nucleus, and be transcribed. In PBAE-based polyplexes, the highly positive surface charge is thought to initiate endosomal escape through the proton sponge effect, although this may be highly variable based on the structures of the PBAE amine end-caps [34]. The end-cap structures of the leading polymers from our optimization study (44E6, 53E6, and 44E49) are shown in Figure 2. Differences in the end-cap structures E6 and E49 include an additional secondary amine and increased hydrophobicity in E49 compared to E6. Previously, a link was suggested between the number of secondary end-cap amines and intracellular buffering capacity, leading to higher transfection [34]. This may be an additional explanation for the increased transfection yield of polymer 44E49 compared to 44E6 and 53E6. Most likely, it is a combination of factors rather than any single intracellular delivery mechanism that contributes to the increased yield of polymer 44E49.

One of the current standards of non-viral transfection is Lipofectamine™, a lipid-based transfection agent. When used in parallel with the developed PBAEs, Lipofectamine™ reagents transfected only a small percentage of hASCs, leading to a transfection yield 3 times lower than that of 44E49, at best. This may be due to vast differences in uptake mechanisms and endosomal escape mechanisms for a lipid-based delivery system compared to that of a cationic polymer system. For comparison to general MSC transduction methods, lentiviral MSC transduction has been reported with 80% efficiency and ≥80% viability, compared to the PBAE 44E49 transfection efficiency of approximately 50% and viability of approximately 75% [43,44].

Similar PBAE formulations have evaluated for biodegradability in the literature and the half-life of ester bond degradation in physiological conditions was determined to be ~4.6 hours with full degradation in approximately 1 day [45]. The stability of the 44E49 particles is ~4 hours and DNA begins to be released starting at 8 hours (in serum) as shown in Figure 4. PBAEs are degraded by hydrolyzable ester bonds in physiological conditions and the hydrophilic small molecule monomer breakdown products are small enough to be cleared out by the kidneys (<7 nm) following an in vivo application. Thus, these non-viral nanomaterials are promising for therapeutic use due to their potential efficacy and safety.

7.2. Evaluation of Vascular Morphogenesis

Hypoxia has been shown to be a reliable inducer of vascular morphogenesis in hASCs [9]. However, there are logistical limitations to the creation of physically hypoxic environments both in vivo and in the clinic. Non-viral gene delivery of HIF-1α may offer a more feasible, safe, and easily scalable alternative to induce angiogenesis for tissue engineering and regenerative medicine therapies. To do so, it is necessary to demonstrate efficacy of non-viral gene delivery similar to or greater than that of physical hypoxia. Previously we reported robust vessel tubule formation and genetic upregulation of the HIF-1α pathway of hASCs in 2D culture treated with physical hypoxia compared to normoxia-treated groups [9]. Here, we show that hASCs in 3D culture transfected with HIF-1α showed similar VEGF upregulation to that seen in 2D hypoxic culture (approximately 6-fold higher mRNA levels than the control) and vascular morphogenesis comparable to vessel structures reported in 2D hypoxia.

Likewise, there have been previous efforts at inducing overexpression of HIF-1α in MSCs through viral transduction [3,46]. As a comparison, Lampert et al. and Li et al. used adenovirus to transduce human MSCs and rabbit hASCs to overexpress HIF-1α and found prolonged upregulation of angiogenic genes HGF, VEGF, bFGF, PDGFR-α, and PDGRF-β [3,47]. Lampert et al. reported VEGF expression approximately 1.5-fold higher than that of the control, and Li et al. reported VEGF expression approximately 4.5-fold higher than that in the control [47] . Both reported no changes in the differentiation capacity of HIF-1α-overexpressing MSCs aside from a slight decrease in BMP-2 expression, which could impact osteogenic potential [3,46,47].

Similarly, Yang et al. used PBAE polymer’s to transfect bone marrow-derived hMSCs with VEGF plasmid, and reported a ~2-fold increase in VEGF protein secretion (day 4-9) when compared to untransfected controls in vitro and this expression level was found to be sufficient to improve ischemic limb salvage in vivo [48]. This is an interesting comparison to the current results with PBAE 44E49/HIF-1α polyplexes, as HIF-1α is a gene upstream in the VEGF secretion pathway, and yet results are comparable for VEGF expression (~2-fold increase in VEGF protein secretion with polymer 44E49 in adipose-derived stromal/stem cells over untransfected controls at day 7). There are further advantages to transfection with HIF-1α, as it upregulates several important downstream angiogenic genes in addition to upregulation of VEGF.

A potential challenge of the PBAE non-viral transfection system is the transient expression of the delivered gene. As shown in Figure 5, mRNA expression peaks 2 days after transfection and rapidly decreases by 7 days. Additionally, the HIF-1α transcription factor is not stable under normoxia and degrades rapidly [49]. On the one hand, these kinetics may be favorable. In a study by Silva et al., endothelial cells were dosed with soluble VEGF and the concentration and frequency of dosing were varied over a 5-day period [50]. Vessel formation was found to be most robust after a high initial dose followed by continuous low dosing of VEGF over the 5-day period. These results are consistent with the behavior of VEGF secretion shown in Figure 5B and the pattern of vessel development in Figure 6. Upregulation of HIF has also been linked to cancer pathogenesis [51] and a transient expression profile may be beneficial from a safety perspective as a promising way to blunt harmful effects of HIF. Figure 5A shows HIF expression peaks on day 2 and continues to decrease over the 7-day period. It is likely that expression continues to decrease after, as the cells are in the presence of normoxia, which degrades HIF transcription factors. On the other hand, if, due to a lack of significant physical hypoxia, degradation of HIF-1α may be occurring more quickly than ideal for a tissue engineering or regenerative medicine application, other aspects of this system could be further engineered. For example, Fu et al. report a genetic polymorphism in the HIF-1α reporter for certain prostate cancers that allows for stabilization under normoxia [49]. It may be possible to increase duration and efficacy of transfection by changing the plasmid sequence to this mutated HIF-1α. Overall, transient gene expression of HIF-1α may be ideal for the purposes of angiogenesis due to the successful formation of vascular structures within 7 days, as shown in Figure 6, while also limiting overexpression of HIF-1α to prevent potential cancer progression [52].

8. Conclusions

Using the novel 44E49 PBAE as a non-viral delivery system to hASCs, it is possible to initiate vascular tubule formation in vitro in response to upregulation of HIF-1α. The morphology and gene expression of these transfected structures are comparable to that of physical hypoxia. Likewise, downstream VEGF upregulation appears to be equitable with current reports of MSCs transduced virally with HIF-1α. We demonstrate the potential of PBAEs to deliver HIF-1α to adipose-derived stromal/stem cells for tissue engineering and regenerative medicine vascularization therapies.

2. Statement of Significance.

Not only is the formation of stable vasculature a challenge for engineering human tissues in vitro, but it is also of valuable interest to clinical applications such as peripheral artery disease. Previous studies using HIF-1α to induce vascular formation have been limited by the necessity of hypoxic chambers. It would be advantageous to simulate endogenous responses to hypoxia without the need for physical hypoxia. In this study, 3D vascular formation was shown to be inducible through non-viral gene delivery of HIF-1α with new polymeric nanocomplexes. A biodegradable polymer N’-(3-aminopropyl)-N,N-dimethylpropane-1,3-diamine end-modified poly(1,4-butanediol diacrylate-co-4-amino-1-butanol) demonstrates improved transfection of human adipose-derived stem cells. This nanobiotechnology could be a promising strategy for the creation of vasculature for tissue engineering and clinical applications.