The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing

Biomaterials xxx (2014) 1e13

Contents lists available at ScienceDirect

Biomaterials journal homepage: www.elsevier.com/locate/biomaterials

The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing Winifred Wing Yiu Yau a, 1, Hongyan Long a, 1, Nils C. Gauthier b, Jerry Kok Yen Chan c, d, e, Sing Yian Chew a, f, * a

Division of Chemical and Biomolecular Engineering, School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459, Singapore Mechanobiology Institute, National University of Singapore, 5A Engineering Drive 1, Singapore 117411, Singapore c Department of Obstetrics and Gynaecology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 119074, Singapore d Department of Reproductive Medicine, KK Women's and Children's Hospital, Singapore 229899, Singapore e Cancer and Stem Cell Biology, Duke-NUS Graduate Medical School, Singapore 169857, Singapore f Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore 308232, Singapore b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 3 July 2014 Accepted 2 October 2014 Available online xxx

While substrate topography influences cell behavior, RNA interference (RNAi) has also emerged as a potent method for understanding and directing cell fate. However, the effects of substrate topography on RNAi remain poorly understood. Here, we report the influence of nanofiber architecture on siRNAmediated gene-silencing in human somatic and stem cells. The respective model cells, human dermal fibroblasts (HDFs) and mesenchymal stem cells (MSCs), were cultured onto aligned or randomly oriented electrospun poly(ε-caprolactone) fibers of different average diameters (300 nm, 700 nm and 1.3 mm). In HDFs, decreasing fiber diameter from 1.3 mm to 300 nm improved Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and Collagen-I silencing efficiencies by ~ 3.8 and ~4.4 folds respectively (p < 0.05) while the effective siRNA uptake pathway was altered from clathrin-dependent endocytosis to macropinocytosis. In MSCs, aligned fibers generated significantly higher level of gene silencing of RE-1 silencing transcription factor (REST) and green fluorescent protein (GFP) (~1.6 and ~1.5 folds respectively, p < 0.05), than randomly-oriented fibers. Aligned fiber topography facilitated functional siRNA uptake through clathrin-mediated endocytosis and membrane fusion. Taken together, our results demonstrated a promising role of three-dimensional fibrous scaffolds in modulating siRNA-mediated gene-silencing and established the critical synergistic role of these substrates in modulating cellular behavior by RNAi. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Electrospinning Mesenchymal stem cells Gene knockdown RNA interference Nanotopography Endocytosis pathway

1. Introduction The manipulation of cell behavior and differentiation remains a major challenge in the field of regenerative medicine. Hence in attempt to steer differentiation in a more specific and delicate manner, alternatives to conventional biochemical cocktails have been explored. In particular, genetic modification of cells by small non-coding RNAs (sncRNAs, e.g. small interfering RNAs, siRNAs) to elicit transient RNA interference (RNAi) has emerged to be a safe

* Corresponding author. Division of Chemical and Biomolecular Engineering, School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459, Singapore. Tel.: þ65 6316 8812; fax: þ65 6794 7553. E-mail address: [email protected] (S.Y. Chew). 1 Both authors contributed equally to this work.

and popular way to induce accelerated cellular differentiation and transdifferentiation [1e6]. Unfortunately, the efficacy of RNAiinduced differentiation has been limited by the delivery and transfection efficiencies of sncRNAs [7]. Although attempts have been made to overcome this limitation by sustained application of sncRNAs from delivery vehicles [8e10] or by modifying the composition of nanoparticle carriers to improve transfection efficiency [11,12], the outcomes remain suboptimal. Therefore, more in-depth understanding and alternative approaches to enhance RNAi are required. One potential method of modifying RNAi outcomes may be to control the architecture of the underlying substrate on which cells are cultured. Besides biochemical cues, topographical features of the extracellular matrix (ECM) also play a significant role in dictating cell fate. In this context, nanofiber constructs have been

http://dx.doi.org/10.1016/j.biomaterials.2014.10.003 0142-9612/© 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

2

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

widely explored due to its biomimicking architecture. Indeed, fiber diameter and orientation can affect the proliferation, apoptosis and differentiation of cells [13e17]. In particular, neural stem cells preferentially differentiated into glial lineage on thin fibers (283 nm) but into neuronal lineage on thicker fibers (749 and 1.452 mm) [13]. Fiber diameter effect on cell behavior has also been established in vivo where nanofibers better facilitated the generation of myelinated axons and myelin sheaths [18], while microfibers enhanced the growth of vascular constructs in electrospun conduits [19]. Additionally, fiber orientation also modulates cell behavior. Aligned fibers could enhance osteogenic differentiation of marrow stromal cells [20] and mesenchymal stem cells [21], as well as neuronal differentiation of neural precursor cells [22] as compared to random fibers. With more extensive research on RNAi- and substrate topography-mediated differentiation, the synergistic effect of these two platforms has been explored [9,10,23]. When applying siRNA to cells cultured on scaffolds, the topography of the scaffold facilitated the extent of differentiation by RNAi [9]. However, the endocytic pathway utilized by cells on scaffolds may be different from that on a two-dimensional surface [24]. Therefore, there is a need to examine the combined effect of ECM topography and gene knockdown. Unfortunately, to date, little is known about the interrelationship of these two factors. The potential of nanotopography in improving endocytosis and enhancing drug and gene delivery has been highlighted in a few studies [24,25], but its effect on gene silencing has yet to be elucidated. In this study, we examined the effect of fiber diameter and orientation on gene silencing efficiency. The nanotopography of the microenvironment was tailor-made by electrospun poly(ε-caprolactone) (PCL) fibers of different average diameters (300 nm, 700 nm and 1.3 mm) and orientation (random and aligned). Using human dermal fibroblasts (HDFs) and human fetal mesenchymal stem cells (MSCs) as model somatic and stem cells, we investigated the effect of substrate topography on siRNA uptake and gene knockdown efficiencies. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was chosen as the knockdown target for both cells since it is a housekeeping gene and has a high endogenous expression level which is not sensitive to substrate topography. Additionally, collagen I and RE1-silencing transcription factor (REST) were chosen as the knockdown targets for HDFs and MSCs respectively in order to investigate the knockdown effect on targets of lower abundance. We hypothesize that gene silencing efficiency may be improved by modifying the topography of the ECM.

and 96-well plate for real-time PCR were bought from Life Technologies. Dulbecco's Modified Eagle Medium (DMEM), DMEM F-12 medium, fetal bovine serum (FBS), and penicillin-streptomycin (p/s) were purchased from Hyclone. Cy5-conjugated oligonucleotide (Cy5-ODN) was obtained from Aitbiotech. PCR reagents, random (dT) primers, M-MLV Reverse Transcriptase were purchased from Promega. RNeasy Mini Kit was purchased from Qiagen. HDFs were purchased from Lonza. Coverslips were purchased from VWR. MSCs and GFP-MSCs were derived from human fetal bone marrow as previously described with ethical approval from institutional research ethics board [26]. SYBR green Supermix was purchased from Bio-rad. 2.2. Scaffold fabrication PCL film was fabricated by spin-coating. PCL was dissolved in TFE to obtain an 8 wt% solution. Thereafter, 80 ml of polymer solution was used and the PCL film was spin-coated onto a round coverslip (diameter 18 mm). For electrospinning of PCL fibers, different parameters were used to generate fibers with different diameters, as indicated in Table 1. 2.3. Evaluation of fiber scaffolds Fiber scaffolds were coated with platinum at 20 mA for 80 s and imaged by scanning electron microscopy (SEM) (JOEL, JSM-6390LA, Japan) with an acceleration voltage of 10 kV. Thereafter, Image J (NIH, USA) was used to measure the average fiber diameter. Over 100 fibers were counted for each sample. 2.4. Cell culture and transfection HDFs were maintained in DMEM supplemented with 10% FBS and 1% p/s. MSCs were maintained in DMEM F-12 supplemented with 10% FBS and 1% p/s. Cells were maintained in a humidified incubator at 37  C with 5% CO2. HDFs and MSCs were seeded at 10,000 and 20,000 cells per cm2 respectively onto PCL film or fibrous scaffolds. Cells were cultured for 3 days in 24-well plates and stained with the LIVE/ DEAD® Cell Viability Assay kit according to manufacturer's protocol. The elongation factor and cell area of live cells were measured by Image J. Elongation factor was computed by the equation: (Cell length/cell width) e 1. The cell area was the surface area of live cells spreading on the fibers or film. Over 100 cells per sample were calculated for statistical analysis. All siRNA transfection was carried out in 1% FBS with TKO as the transfection agent. Cells were seeded for 24 h in 12-well plates before transfection. HDFs were transfected with GAPDH or Collagen I siRNA for 48 h at a volume ratio of 1:3 (siRNA:TKO) following manufacturer's protocol. MSCs were transfected with GAPDH or REST siRNA for 48 h at a volume ratio of 1:1. Thereafter, cells were harvested for realtime PCR for evaluation of knockdown efficiency. Cells transfected with random strands of siRNA were taken as a negative control (siNEG). For GFP silencing, lentiviral transduced MSCs over-expressing GFP were transfected with GFP siRNA for 48 h [27,28]. To quantify the extent of GFP silencing in GFP-MSCs, GFP signal per cell was measured using Image J. The signal intensity relative to GFP-MSCs transfected with siNEG was calculated. A lower relative signal intensity denotes a higher knockdown efficiency and vice versa. GFP transfection efficiency was computed by the equation: (1-signal intensity after siGFP transfection/signal intensity after siNEG transfection) 100%. The particle size of siRNA/TKO complexes freshly prepared or incubated in DMEM for 48 h were measured by Zetasizer (Malvern Instrument, UK). Three samples were measured for each group. 2.5. Cy5-ODN uptake assay

2. Materials and methods 2.1. Materials PCL (Mw ¼ 80,000) and 2,2,2-tetrafluoroethanol (TEF, 99%), chlorpromazine, filipin, amiloride, and DMSO (99.9%) were obtained from Sigma Aldrich. The LIVE/ DEAD® Cell Viability Assay kit (calcein-AM and ethidium homodimer-1), Alexa Fluor® 488 phalloidin, Molecular Probes transferrin and dextran, Ambion In Vivo siRNA controls (siNEG), GAPDH siRNA, type I Collagen siRNA, REST siRNA, GFP siRNA

HDFs or MSCs were seeded on film or fibrous scaffolds for 24 h before they were transfected with Cy5-ODN/TKO complexes (using the same volume ratio as siRNA) for 48 h. After transfection, the cells were solubilized with RLT lysis buffer (Qiagen RNeasy Mini Kit) and fluorescence intensity of the lysate was measured by Infinite F200 microplate reader (TECAN). Two hundred micro liters of lysate was added to each well for intensity measurement. A standard curve was generated with known concentrations of Cy5-ODN. The amount of Cy5-ODN retained by the cells was then derived from the standard curve and further normalized by the total amount of RNA

Table 1 Parameters for electrospinning of PCL fibers.

Concentration (wt%) Solvent Needle Flowing rate (mL/h) Rotator speed (rpm) Voltage (kV) Distance (cm) Fiber diameter (nm) *

300 nm Random

300 nm Aligned

700 nm Random

700 nm Aligned

1.3 mm random

1.3 mm aligned

12 TFE:PBS (2) ¼ 1:4 22 G 0.7 300 15 15 317 ± 33.7

12 TFE:PBS (1) ¼ 1:4 22 G 0.7 2300 15 15 306 ± 55.6

12 TFE:PBS (1) ¼ 1:4 30 G 0.3 300 10 12 712 ± 82.3*,#

12 TFE 30 G 0.5 2000 7 12 709 ± 93.7*,#

12 TFE 21 G 0.5 300 7 12 1339 ± 99.3*

12 TFE 30 G 1 2000 10 12 1297 ± 144*

: p < 0.05 with respect to 300 nm fibers; #: p < 0.05 with respect to 1300 nm fibers.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13 2.8. Dextran and transferrin uptake assay

Table 2 Primer sequences for real-time PCR. Primers

Sequence (50 -30 )

Product length (bp)

GAPDH

Forward: ATCAGCAATGCCTCCTGCAC Reverse: TGGCATGGACTGTGGTCATG Forward: AGATGAAGATGTAGGCCGGGTGA Reverse: GCCAAATCCTTTCTCTCCTGTAGC Forward: CAATGCTGCCCTTTCTGCTCCTTT Reverse: ATTGCCTTTGATTGCTGGGCAGAC Forward: AAGTGCAGAGAAACAGGCCAAAGC Reverse: TTCAGGTGTGCCATGTAGTGGTCA

103

CycB Col1A1 REST

3

103 125 138

HDFs or MSCs were seeded on PCL film and fibers 24 h prior to transferrin or dextran uptake assay. For transferrin uptake, cells were pre-incubated for 1 h in the presence of transfection complex and 20 ml/ml transferrin. After that, cells were allowed to grow in fresh medium for 10, 30 and 60 min. For dextran uptake, cells were pre-incubated for 1 h in the presence of transfection complex. After that, 1 mg/ ml dextran was added to the cells for 10 min. The cells were either harvested immediately or allowed to grow for another 20 or 50 min in fresh medium before they were harvested and examined by confocal microscopy using high magnification lens (100). More than 20 cells were imaged for each sample and the trend was consistent across three experimental repeats. 2.9. Statistical analysis

in the sample. All experimental groups were performed in triplicates and the experiment was repeated for three times.

Statistical analysis was performed using SPSS software. One-way ANOVA and the Tukey post-hoc test were used when samples had equal variances. Otherwise, the ManneWhitney U test was used.

2.6. Drug treatment

3. Results

The protocol for drug treatment was adapted from literature [29,30]. The concentrations of drugs were further optimized by LIVE/DEAD® Cell Viability Assay to ensure cell viability during the course of experiment. Briefly, 24 h after seeding, HDFs were pre-incubated in 1% FBS at 37  C for 1 h with DMSO (control), 10 mg/ml filipin, 0.5 mg/ml chlorpromazine, 1 mM amiloride or at 4  C. Thereafter, cells were transfected with Cy5-ODN, collagen I or REST siRNA in the presence of inhibitors at the same temperature for 4 h. To evaluate gene uptake, cells transfected with Cy5ODN were fixed with 10% formalin, stained with DAPI and subjected to confocal imaging (LSM710, Zeiss) under the same setting for all the samples. The intensity of Cy5 signal per cell was measured from the confocal images by Image J. The intensity values were normalized to cells treated with DMSO on film, and the value of this sample group was set as 100%. Over 100 cells were measured for each sample. To evaluate silencing efficiency, the transfection complex containing collagen I or REST siRNA was replaced by fresh medium for a further incubation of 44 h at 37  C. After that, cells were harvested for RNA extraction and real-time PCR analysis.

2.7. RNA extraction and real-time PCR Total RNA was extracted from cells (pooled from 3 to 4 scaffolds per sample) using RNeasy mini kit following manufacturer's protocol. First strand cDNA was synthesized using random (dT) primers and M-MLV Reverse Transcriptase. Realtime PCR was performed with StepOnePlus real-time PCR system (Applied Biosystems) using the comparative Ct method after ensuring that all primers had similar amplification efficiencies. The primers for real-time PCR are listed in Table 2. Cyclophilin B (CycB) was the internal control for all real-time PCR results. The results of mRNA expression were normalized to the respective scrambled siRNA control. Three independent repeats were conducted for each experiment.

3.1. Fiber orientation affected morphology of HDFs Six PCL fibrous scaffolds with significantly different average diameters (~300 nm vs. ~700 nm vs. ~1.3 mm, p < 0.05, Table 1) and orientation (random and aligned) were fabricated by electrospinning to provide different topographical cues to modulate cellular response (Fig. 1). To clearly elucidate topographical effect on cellular response, rid of surface chemistry effects, PCL film was fabricated by spin-coating and was included as a control to show the classical cellular behavior on two-dimensional culture. The average thickness of the fiber scaffolds was ~224 mm, with no significant difference between each group (Supp Fig. 6). HDFs cultured on different PCL fibers exhibited different morphology in terms of elongation factor (Fig. 2). In general, HDFs were more elongated when cultured on aligned fibers (Fig. 2B). Moreover, the elongation factor of HDFs appeared to increase with fiber diameter. Cells cultured on 700 nm and 1.3 mm fibers were significantly more elongated than those cultured on 300 nm fibers (~6.5 for 700 nm and 1.3 mm fibers vs ~4 for 300 nm fibers, p < 0.05). On the other hand, fiber diameter did not seem to affect the spreading area of HDFs, although they were much smaller when compared to cells cultured on PCL films (Fig. 2C).

Fig. 1. Representative SEM images of PCL fibers and film. Scale bar ¼ 2 mm.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

4

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

3.2. Fiber diameter affected silencing efficiency of GAPDH and collagen I in HDFs Fig. 3A to D show the effect of fiber diameter and orientation on gene silencing in HDFs. In order to eliminate the effect of fiber diameter on the constitutive expression of the two genes, the values were normalized to the mRNA level on the same substrate but transfected with scrambled siRNA. Therefore, a normalized relative mRNA expression level of 0% represents 100% knockdown efficiency and vice versa. Interestingly, regardless of the silencing target, HDFs cultured on film and 300 nm fibers generally showed higher knockdown efficiencies than 700 nm and 1.3 mm fibers (p < 0.05). In addition, such enhancement appeared most prominent at low concentrations of siRNA. Specifically, when cells were transfected with 2 nM of siRNA, a maximum increase of ~3.8 fold was observed for GAPDH (23.9% on 300 nm fibers vs. 6.3% on 1.3 mm fibers, p < 0.05, aligned fibers). For collagen I, the largest difference was ~4.4 fold (44.8% for 300 nm fibers vs. 10.2% for 1.3 mm fibers, p < 0.05, random fibers). This trend was consistent on random and aligned fibers, implying that fiber diameter effect was greater than

fiber orientation effect in terms of facilitating gene silencing in HDFs. Taken together, regardless of the target gene or fiber alignment, 300 nm fibers better facilitated gene knockdown in HDFs.

3.3. SiRNA uptake and retention was positively related to gene silencing efficiency Fig. 3E shows the relative amount of total Cy5-ODN retained in HDFs cultured on fibers of different average diameters. Interestingly, 300 nm fibers significantly facilitated Cy5-ODN uptake and retention in HDFs when compared to 700 nm and 1.3 mm fibers. This was in line with the earlier observation that 300 nm fibers facilitated better siRNA directed gene silencing over the larger fibers. In addition, the particle size of siRNA/TKO complexes after 48 h incubation in DMEM was 278 ± 54 nm, which remained similar as the size of the freshly prepared complexes (269 ± 26 nm). The result suggested that fiber diameter could affect the amount of gene uptake and retention and thereby modulate silencing efficiency.

Fig. 2. Effect of fiber diameter and orientation on morphology of HDFs. (A) Representative fluorescence images showing HDFs stained with Calcein-AM (green). Scale bar ¼ 50 mm. (B) Elongation factor and (C) cell area of HDFs cultured on fibers with different diameter and orientation. Results show mean ± SD from three independent experiments with 9 scaffolds (n ¼ 3). More than 200 cells were analyzed for each group. Statistical analysis was done using ANOVA and ManneWhitney U test, for (B) and (C) respectively. *: p < 0.05 compared to Film. #: p < 0.05 compared to random fibers of the same fiber diameter. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

5

Fig. 3. HDFs cultured on 300 nm fibers showed higher knockdown efficiency. Real-time PCR analysis showing the percentage of knockdown for GAPDH mRNA on (A) random fibers and (B) aligned fibers, as well as knockdown for collagen I mRNA on (C) random fibers and (D) aligned fibers. Results show mean ± SD from 8 different scaffolds (n ¼ 8). Statistical analysis was done using ManneWhitney U test. *: p < 0.05 compared to Film. #: p < 0.05 compared to 1300 nm þ: p < 0.05 compared to 700 nm. (E) HDFs cultured on 300 nm fibers showed a higher amount of Cy5-ODN uptake relative to total RNA. Result shows mean ± SD from 12 different scaffolds (n ¼ 12). Statistical analysis was done using ANOVA. *: p < 0.05.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

6

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

3.4. Effective gene silencing on all fibers was facilitated by clathrinmediated endocytosis To investigate the importance of clathrin-mediated endocytic pathway for internalization of siRNA, HDFs were pre-incubated with chlorpromazine, an inhibitor of this pathway before transfection. Chlorpromazine significantly reduced the amount of gene uptake for all the samples except 1.3 mm fibers, which gave a low amount of gene uptake even in the control group that was treated with carrier (DMSO) only (Fig. 4A; Supp. Fig. 1A and B). Nevertheless, chlorpromazine significantly reduced the gene silencing efficiency of collagen I in cells cultured on fibers (Fig. 4B). To further confirm this observation, transferrin, a marker for clathrinmediated endocytosis, was added to the cells during transfection and its colocalization with Cy5-ODN was examined. As indicated in Supp. Fig. 2, colocalization of Cy5-ODN was detected in all sample groups. Altogether, the results suggested the involvement of clathrin-mediated endocytic pathway in effective gene silencing in HDFs on fiber scaffolds under our experimental conditions. 3.5. Effective gene silencing on thinner fibers and film was facilitated by macropinocytosis In addition to clathrin-mediated endocytosis, macropinocytosis was also responsible for effective silencing in HDFs. When amiloride was added to block macropinocytosis, HDFs cultured on fibers were able to take up a similar amount of siRNA as the DMSO control group (Fig. 4C; Supp. Fig. 1A and C). However, the percentage of gene knockdown was greatly reduced especially on 300 nm fibers, 700 nm fibers and film (Fig. 4D). This suggested that a low amount of siRNA taken up via macropinocytosis could result in extensive gene silencing, particularly on nanofibers. We observed the colocalization of dextran and Cy5-ODN in cells cultured on 300 nm, 700 nm fibers and film (Supp. Fig. 3), confirming that Cy5-ODN uptake was dependent on macropinocytosis. 3.6. Cholesterol depletion only affected gene silencing on 700 nm fibers and film Cholesterol depletion by filipin treatment reduced both gene uptake and silencing efficiency on film (Fig. 4E and F; Supp. Fig. 1 A and D). On 300 nm fibers, a significant 25% reduction in gene uptake only slightly affected knockdown. The contrary was observed on 700 nm fibers, where insignificant reduction in Cy5-ODN uptake led to 90% reduction in silencing efficiency, suggesting that effective gene silencing on 700 nm fibers was cholesterol dependent. Taken together, cholesterol-dependent endocytic pathway was partially responsible for functional gene uptake in HDFs cultured on film and 700 nm fibers. 3.7. Energy depletion reduced effective gene silencing on thinner fibers and film Gene uptake and silencing in all the sample groups were energy dependent (Fig. 4G and H; Supp. Fig. 1 A and E). When transfection was carried out at 4  C, silencing efficiency of 300 nm, 700 nm fibers and film were significantly reduced by more than 65%. Energy depletion also slightly reduced gene silencing on 1.3 mm fibers although the difference was not significant. 3.8. Fiber orientation affected morphology of MSCs Next, we extended our study to stem cells, with the use of MSCs as the cell type. MSCs adopted a more elongated phenotype when cultured on aligned fibers, as was the case with HDFs (Fig. 5). In

addition, elongation factor for MSCs increased with increasing fiber diameter (~9 for 1.3 mm aligned fibers vs ~6 for 300 nm fibers) (Fig. 5B). On the contrary, MSCs maintained a similar cell area regardless of underlying substrate topography (Fig. 5C). 3.9. Aligned fibers favored gene knockdown in MSCs In contrast to HDFs, gene silencing in MSCs was less affected by fiber diameter (Fig. 6A). Only a small difference in silencing efficiency was recorded between 300 nm and 1.3 mm fibers. Instead, fiber orientation seemed to have a more profound effect in altering the silencing efficiency of genes that exist in lower abundance (REST and GFP vs. GAPDH as indicated in Fig. 6 A to D and Supp Fig. 1 respectively). As shown in Fig. 6A, MSCs seeded on film and aligned fibers exhibited higher silencing efficiencies than random fibers when REST was targeted. The trend was consistently observed across different concentrations of siRNA (2, 10, 20 and 50 nM) (Fig. 6B). Similarly, silencing of GFP was more robust in MSCs cultured on aligned fibers (Fig. 6C and D ~ 80% reduction on aligned fibers vs. 50% on random fibers). In contrast, no difference in gene silencing was observed in response to fiber orientation when GAPDH was targeted (Supp. Fig. 4). 3.10. Aligned fibers facilitated gene uptake in MSCs Fig. 6E shows the amount of Cy5-ODN internalized by MSCs on random vs. aligned fibers. Consistent with gene silencing outcomes, aligned fibers induced a higher level of gene uptake and retention in MSCs when compared to random fibers. The enhancement was observed at different siRNA concentrations (10, 20 and 50 nM), and was most obvious at 50 nM (p < 0.05). 3.11. Effective silencing on aligned fibers was facilitated by clathrinmediated pathway On random fibers and film, inhibiting siRNA uptake via clathrinmediated pathway using chlorpromazine did not affect gene knockdown efficiency (Fig. 7A and B; Supp. Fig. 5A and B). In contrast, chlorpromazine significantly decreased gene knockdown on aligned fibers (p < 0.05). These observations were further supported by transferrin uptake assay, which showed more obvious uptake via clathrin-mediated pathway on aligned fibers (Fig. 8) than random fibers and film (Supp. Fig. 2), Taken together, clathrinmediated pathway was mainly responsible for functional siRNA uptake in MSCs on aligned fibers. 3.12. Effective silencing on film was mediated by macropinocytosis Blockage of macropinocytosis by amiloride slightly reduced Cy5-ODN uptake and completely abolished gene knockdown in MSCs that were cultured on films but not fibers (Fig. 7C and D; Supp. Fig. 5A and C). In line with gene knockdown experiments, Cy5-ODN colocalized with dextran while entering cells that were cultured on films only (Fig. 9 and Supp. Fig. 3). Taken together, macropinocytosis was a prominent pathway that led to gene silencing when MSCs were cultured on a flat two-dimensional substrate. 3.13. Gene uptake by MSCs is cholesterol and energy dependent Gene silencing in MSCs was found to be sensitive to cholesterol and energy depletion. As shown in Fig. 7E and F and Supp. Fig. 5A and D, filipin, an inhibitor for caveolae and lipid-raft dependent endocytic pathways, significantly reduced the amount of gene uptake and silencing efficiency in all the sample groups. Similarly,

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

7

Fig. 4. Pathway inhibition reduced Cy5-ODN uptake and collagen I knockdown in HDFs. Relative intensity of Cy5-ODN in HDFs treated with (A) chlorpromazine, (C) amiloride, (E) filipin or (G) transfected at 4  C. Results show mean ± SD from three independent experiments with 12 scaffolds (n ¼ 3). Knockdown efficiency of collagen I in HDFs in the presence of (B) chlorpromazine, (D) amiloride, (F) filipin or (H) at 4  C. Result shows mean ± SD from three independent experiments with 10 scaffolds (n ¼ 3). Statistical analysis was done using ANOVA. *: p < 0.05. **: p < 0.01. ***: p < 0.001.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

8

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

Fig. 5. Effect of fiber diameter and orientation on morphology of MSCs. (A) Representative fluorescence images showing MSCs stained with Calcein-AM (green). Scale bar ¼ 50 mm. (B) Elongation factor and (C) cell area of MSCs cultured on fibers with different diameter and orientation. Results show mean ± SD from three independent experiments with 9 scaffolds (n ¼ 3). More than 200 cells were analyzed for each group. Statistical analysis was done using ANOVA. *: p < 0.05 compared to Film. #: p < 0.05 compared to random fibers of the same fiber diameter. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

when transfection was carried out at 4  C, gene uptake and silencing were significantly decreased (Fig. 7G and H; Supp. Fig. 5A and E). Interestingly, 20% of silencing efficiency was still retained in MSCs cultured on aligned fibers, suggesting that a portion of functional gene uptake was energy independent. 4. Discussion Substrate topography [20e22] and gene silencing [3e6] have emerged as powerful signals to direct cell fate and tissue regeneration. We have shown that silencing of collagen I expression in fibroblasts by nanofibers endowed with siRNA reduced the formation of fibrous capsule around the scaffold, hence facilitating host-implant integration in vivo [31]. Meanwhile, knockdown of REST in stem cells facilitated neuronal differentiation [3,9]. Coupled with nanofiber topography, we demonstrated that the synergistic effects further promoted neuronal commitment [9]. Therefore, these two gene knockdown targets were adopted in this study. Here, in attempt to identify factors that can enhance RNAi, we investigated the effect of substrate topography on siRNA silencing

efficiency. Our results showed that substrate topography not only influenced the extent of gene silencing, but also altered the pathways responsible for functional siRNA delivery. Substrate topography was found to modulate the level of gene uptake and knockdown. In HDFs, fiber diameter played an important role in manipulating gene silencing. The knockdown efficiencies of either GAPDH or collagen I on 300 nm fibers were generally higher than that on 700 nm or 1.3 mm fibers at all siRNA concentrations tested. Unlike HDFs, fiber diameter effect was not as prominent as fiber orientation effect on effective gene silencing in MSCs. MSCs cultured on aligned fibers and film exhibited better knockdown than randomly-oriented fibers. Although aligned fibers did not show enhanced knockdown for GAPDH, it is possible that the high abundance of GAPDH could have masked the substrate topography effect [32,33]. Nevertheless, fiber orientation effect on gene silencing was observed for REST and GFP knockdown, and therefore is not gene-specific. Not surprisingly, the amount of gene uptake was in line with knockdown efficiency. Regardless of cell type, cells that retained more siRNA showed better knockdown. Specifically, for HDFs,

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

9

Fig. 6. Effects of fiber orientation and siRNA concentration on knockdown efficiency for REST and GFP. (A) MSCs cultured on aligned fibers showed a higher knockdown efficiency for REST. MSCs were transfected with 10 nM siRNA for 48 h. Result shows mean ± SD from 6 different scaffolds (n ¼ 6). (B) Aligned fibers enhanced knockdown efficiency for REST at different concentrations of siRNA. Result shows mean ± SD from 6 different scaffolds (n ¼ 6). (C) Confocal images showing GFP-overexpressed MSCs transfected with scrambled or GFP siRNA. Nuclei were counterstained with DAPI (blue). Scale bar ¼ 100 mm. (D) Knockdown efficiency for GFP as represented by the percentage decrease of intensity level of fluorescence signal after silencing. Results show mean ± SD from three independent experiments with 9 scaffolds (n ¼ 3). *: p < 0.05 compared to Film. #: p < 0.05 compared to random fibers of the same fiber diameter. (E) MSCs cultured on aligned fibers showed a higher amount of Cy5-ODN uptake relative to total RNA. Result shows mean ± SD from 9 different scaffolds (n ¼ 9). Statistical analysis was done using ANOVA. *: p < 0.05. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

300 nm fibers resulted in the highest retention of siRNA. On the other hand, MSCs cultured on aligned fibers retained more siRNA than random fibers. Since the average particle size of siRNA/TKO complexes remained fairly stable within the 48 h study duration (269 ± 26 nm for freshly prepared complexes vs. 278 ± 54 nm after

48 h incubation in DMEM), these observations strongly suggested that substrate topography could potentially become a biomedical tool to modulate the efficiency of gene uptake and knockdown. Recently, more evidence have suggested that functional siRNA uptake depended on the endocytic pathway via which the siRNA

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

10

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

Fig. 7. Pathway inhibition reduced Cy5-ODN uptake and REST knockdown in MSCs. Relative intensity of Cy5-ODN in MSCs treated with (A) chlorpromazine, (C) amiloride, (E) filipin or (G) transfected at 4  C. Results show mean ± SD from three independent experiments with 9 scaffolds (n ¼ 3). Knockdown efficiency of REST in MSCs in the presence of (B) chlorpromazine, (D) amiloride, (F) filipin or (H) at 4  C. Result shows mean ± SD from three independent experiments with 9 scaffolds (n ¼ 3). Statistical analysis was done using ANOVA. *: p < 0.05.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

11

Fig. 8. Enhanced colocalization of Cy5-ODN with transferrin in MSCs cultured on aligned fibers. Representative images showing MSCs cultured on aligned fibers. MSCs were preincubated with Cy5-ODN and transferrin at 4  C for 1 h. Cellular uptake were then initiated by incubating at 37  C for 10 min, 30 min, 60 min. Colocalization between red and green signal were shown in white (white arrows). Scale bar ¼ 10 mm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

was internalized [16,23]. Very often, only a small portion of internalized siRNA contributed to effective gene silencing [34]. Therefore, we further investigated the pathways responsible for siRNA uptake with respect to substrate topography. Tables 3 and 4 summarize the effect of substrate topography on the utilization of each pathway by HDFs and MSCs respectively. Clathrin-mediated endocytosis has shown to be a very efficient pathway for gene uptake [23,29,35]. Consistently, our results showed that inhibition of this pathway caused significant reductions in cellular uptake of siRNA. Both HDFs and MSCs cultured on fibers utilized this pathway for internalization of siRNA. One exception was HDFs cultured on 1.3 mm fibers, where the siRNA uptake in the DMSO control group was very low to begin with. Upon pathway inhibition, only a slight reduction in siRNA uptake was observed. Nevertheless, the reduction in siRNA uptake was enough to cause a significant disruption in gene knockdown on all scaffolds, suggesting that clathrin-mediated endocytosis was responsible for delivery of functional siRNA into HDFs. Although this pathway is very efficient, it may direct the siRNA to lysosomes for degradation without releasing it to the RNAi machinery in the cytoplasm [29,36]. It seemed to be the case in MSCs. On both film and fibers, inhibition of clathrin-mediated pathway significantly reduced the amount of siRNA uptake. However, on film and random fibers, the corresponding gene knockdown efficiency was not significantly affected, suggesting that the internalized siRNA was not functional. In other words, clathrin-mediated pathway was only responsible for functional siRNA internalization on aligned fibers, but not on random fibers. This signifies the importance of substrate topography on siRNA-mediated gene silencing. By manipulating the alignment of the underlying substrate, MSCs can

be directed to use a different pathway for functional siRNA internalization. Beside clathrin-mediated endocytosis, macropinocytosis has also been reported to facilitate functional transfection of cationic lipids into Chinese hamster ovary cells and primary foreskin fibroblasts [37,38]. Indeed, our results for both cell types showed that in conventional two-dimensional culture, macropinocytosis was responsible for efficient siRNA uptake and silencing. Although the inhibition of macropinocytosis only slightly reduced siRNA uptake, gene silencing was greatly abolished. Interestingly, on fibers, there seemed to be a fiber diameter effect on functional siRNA delivery by macropinocytosis in somatic cells. Specifically, HDFs cultured on thinner fibers (300 nm and 700 nm) behaved like those on film, where negligible reduction in siRNA uptake following inhibition of macropinocytosis led to a significant drop in knockdown efficiency. However, on 1.3 mm fibers, the reduction in gene knockdown was insignificant. In fact, HDFs cultured on 1.3 mm fibers seemed to utilize clathrin-mediated endocytosis as their sole pathway for functional siRNA delivery on fibers since the inhibition of all other endocytic pathways by cholesterol and energy depletion did not significantly affect gene knockdown efficiency. This may explain their low silencing efficiency even in DMSO control group. Taken together, HDFs appeared to utilize different pathways for functional siRNA delivery with fiber diameter variation. Therefore, by altering fiber diameter, the pathway for effective gene silencing could potentially be manipulated to enhance gene knockdown efficiency. A number of endocytic pathways, including macropinocytosis, caveolae-mediated pathway and lipid-raft mediated pathway are cholesterol dependent [39]. Therefore, the effect of cholesterol depletion on effective gene silencing was also investigated.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

12

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

Fig. 9. More extensive colocalization of Cy5-ODN with dextran in MSCs cultured on Film. Representative images showing MSCs cultured on Film. MSCs were pre-incubated with Cy5-ODN at 4  C for 1 h. They were then pulse-labeled with dextran (green) for 10 min (10 min) at 37  C following a 20 min (30 min) and 50 min (60 min) chase. Nuclei were counterstained with DAPI (blue). Colocalization between red and green signal was shown in white (white arrow). Scale bar ¼ 10 mm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Interestingly, only HDFs on 700 nm fibers and film were sensitive to cholesterol depletion and showed impaired gene knockdown. Cholesterol-dependent uptake of siRNA on 300 nm fibers, however, did not contribute significantly to gene silencing. Therefore, future work should focus on cholesterol-dependent pathways in order to elucidate the exact roles of each pathway on functional siRNA delivery. More quantitative studies on the relative importance of each endocytic pathway would also be desirable. Specific blockage of each individual pathway by siRNA would be a useful supplement to our approach using pharmacological inhibitors. More recently, Lu et al. introduced membrane fusion as a newly defined pathway for functional gene silencing [29]. This cholesterol-dependent and energy-independent pathway accounted for the internalization of functional siRNA by cationic lipid complex delivery in monkey kidney epithelial cells. In fact, our results suggest that this pathway might have contributed to effective gene silencing in MSCs. Specifically, functional siRNA delivery into MSCs was cholesterol-dependent, since depletion of

cholesterol on both fibers and film significantly reduced gene knockdown. Moreover, gene knockdown was energy independent on aligned fibers. When transfection was carried out at 4  C, 20% knockdown was still observed on aligned fibers but not random fibers or film. To visualize the uptake, MSCs were pulse-labeled with siRNA at 4  C followed by a chase for up to 1 h. Interestingly, on aligned fibers, siRNA was still retained on the membrane surface after 1 h, suggesting that these membrane bound siRNA could have contributed to effective silencing. Therefore, aligned fibers appeared to have facilitated functional siRNA delivery by membrane fusion. Taken together, fiber alignment and diameter play important roles in siRNA-mediated gene silencing. In particular, in adult somatic cell, siRNA uptake mechanism was altered from macropinocytosis to clathrin-mediated endocytosis by increasing fiber diameter. In the case of stem cells, fiber orientation was the determining factor altering gene uptake pathway from macropinocytosis to clathrin-mediated endocytosis. This is an

Table 3 Extent of involvement of uptake pathways for functional gene knockdown in HDFs.

Table 4 Extent of involvement of uptake pathways for functional gene knockdown in MSCs.

Clathrin-mediated Macropinocytosis Cholesterol dependent Energy dependent

300 nm

700 nm

1.3 mm

þ þþ

þ þþ þþþ þþþ

þþþþ

þþ

Random fibers

Film (∞) þþ þþ þþ

Legend: þ: >1.5 times reduction in gene silencing efficiency with pathway inhibition. þþ: 2 to 5 times reduction in gene silencing efficiency with pathway inhibition. þþþ: >5 times reduction in gene silencing efficiency with pathway inhibition. þþþþ: >10 times reduction in gene silencing efficiency with pathway inhibition.

Clathrin-mediated Macropinocytosis Cholesterol dependent Energy dependent

Aligned fibers

Film (∞)

þ þþþ þþ

þþ þ

þþþþ þþ þþþþ

Legend: þ: >1.5 times reduction in gene silencing efficiency with pathway inhibition. þþ: 2 to 5 times reduction in gene silencing efficiency with pathway inhibition. þþþ: >5 times reduction in gene silencing efficiency with pathway inhibition. þþþþ: >10 times reduction in gene silencing efficiency with pathway inhibition.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003

W.W.Y. Yau et al. / Biomaterials xxx (2014) 1e13

encouraging fact because besides providing physical cues to alter cell phenotype, scaffolds may now play a more active role in improving gene knockdown efficiency. These findings would help researchers optimize their platform for RNAi-directed cell fate manipulation by maximizing the efficiency of gene knockdown, after some fine-tuning according to the target cell type and the transfection agent used. In recent years, approaches to improving knockdown efficiency have been largely inclined to the development of better transfection agents, such as siRNA-incorporated nanoparticles. Our study offers an alternative method by preconditioning the siRNA recipients. By simply altering the size and orientation of the underlying substrate, the cells could become more susceptible to RNAi signals. Nevertheless, it is to be noted that this study has been focusing on the upstream events of RNAi. Future work should therefore look into topographical effect on downstream events such as functionality of the RISC complex. 5. Conclusion In conclusion, our results highlighted the potential to manipulate the efficiency of gene knockdown in human somatic and stem cells by altering substrate architecture, such as fiber diameter and orientation. When using cationic transfection reagents, somatic cells and stem cells behaved differently according to the underlying substrate. Our results suggested that alteration in the amount of gene uptake and the pathway for functional siRNA delivery may be possible reasons. Given that substrate topography can modulate gene silencing, it would also be a promising tool to improve the effectiveness of scaffold-based RNAi-induced cell fate modification. Acknowledgments Partial funding support by The MechanoBiology Institute, Singapore (R714-013-007-271) and MOE AcRF Tier 1 (RG75/10) are acknowledged. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.biomaterials.2014.10.003. References [1] Yau WWY, Rujitanaroj PO, Lam L, Chew SY. Directing stem cell fate by controlled RNA interference. Biomaterials 2012;33(9):2608e28. [2] Yao YC, Wang CM, Varshney RR, Wang DA. Antisense makes sense in engineered regenerative medicine. Pharm Res Dord 2009;26(2):263e75. [3] Low WC, Yau WWY, Stanton LW, Marcy G, Goh E, Chew SY. Directing neuronal differentiation of primary neural progenitor cells by gene knockdown approach. DNA Cell Biol 2012;31(7):1148e60. [4] Levy O, Ruvinov E, Reem T, Granot Y, Cohen S. Highly efficient osteogenic differentiation of human mesenchymal stem cells by eradication of STAT3 signaling. Int J Biochem Cell Biol 2010;42(11):1823e30. [5] Xue YC, Ouyang KF, Huang J, Zhou Y, Ouyang H, Li HR, et al. Direct conversion of fibroblasts to neurons by reprogramming PTB-regulated microRNA circuits. Cell 2013;152(1e2):82e96. [6] Jayawardena TM, Egemnazarov B, Finch EA, Zhang LN, Payne JA, Pandya K, et al. MicroRNA-mediated in vitro and in vivo direct reprogramming of cardiac fibroblasts to cardiomyocytes. Circ Res 2012;110(11). 1465-þ. [7] Dominska M, Dykxhoorn DM. Breaking down the barriers: siRNA delivery and endosome escape. J Cell Sci 2010;123(8):1183e9. [8] Lobovkina T, Jacobson GB, Gonzalez-Gonzalez E, Hickerson RP, Leake D, Kaspar RL, et al. Vivo sustained release of siRNA from solid lipid nanoparticles. Acs Nano 2011;5(12):9977e83. [9] Low WC, Rujitanaroj PO, Lee DK, Messersmith PB, Stanton LW, Goh E, et al. Nanofibrous scaffold-mediated REST knockdown to enhance neuronal differentiation of stem cells. Biomaterials 2013;34(14):3581e90. [10] Nelson CE, Gupta MK, Adolph EJ, Shannon JM, Guelcher SA, Duvall CL. Sustained local delivery of siRNA from an injectable scaffold. Biomaterials 2012;33(4):1154e61.

13

[11] Sahay G, Querbes W, Alabi C, Eltoukhy A, Sarkar S, Zurenko C, et al. Efficiency of siRNA delivery by lipid nanoparticles is limited by endocytic recycling. Nat Biotechnol 2013;31(7). 653-U119. [12] Lin PJC, Tam YYC, Hafez I, Sandhu A, Chen S, Ciufolini MA, et al. Influence of cationic lipid composition on uptake and intracellular processing of lipid nanoparticle formulations of siRNA. Nanomed Nanotechnol 2013;9(2): 233e46. [13] Christopherson GT, Song H, Mao HQ. The influence of fiber diameter of electrospun substrates on neural stem cell differentiation and proliferation. Biomaterials 2009;30(4):556e64. [14] Zhu JX, Cai Q, Zhang X, Hu XQ, Li L, Wang WP, et al. Biological characteristics of mesenchymal stem cells grown on different topographical nanofibrous poly-L-lactide meshes. J Biomed Nanotechnol 2013;9(10):1757e67. [15] Erisken C, Zhang X, Moffat KL, Levine WN, Lu HH. Scaffold fiber diameter regulates human tendon fibroblast growth and differentiation. Tissue Eng Pt A 2013;19(3e4):519e28. [16] Chen M, Patra PK, Warner SB, Bhowmick S. Role of fiber diameter in adhesion and proliferation of NIH 3T3 fibroblast on electrospun polycaprolactone scaffolds. Tissue Eng 2007;13(3):579e87. [17] Badami AS, Kreke MR, Thompson MS, Riffle JS, Goldstein AS. Effect of fiber diameter on spreading, proliferation, and differentiation of osteoblastic cells on electrospun poly(lactic acid) substrates. Biomaterials 2006;27(4):596e606. [18] Jiang X, Mi R, Hoke A, Chew SY. Nanofibrous nerve conduit-enhanced peripheral nerve regeneration. J Tissue Eng Regen Med 2012. [19] Bashur CA, Eagleton MJ, Ramamurthi A. Impact of electrospun conduit fiber diameter and enclosing pouch pore size on vascular constructs grown within rat peritoneal cavities. Tissue Eng Pt A 2013;19(7e8):809e23. [20] Ma JY, He XZ, Jabbari E. Osteogenic differentiation of marrow stromal cells on random and aligned electrospun poly(L-lactide) nanofibers. Ann Biomed Eng 2011;39(1):14e25. [21] Chang JC, Fujita S, Tonami H, Kato K, Iwata H, Hsu SH. Cell orientation and regulation of cell-cell communication in human mesenchymal stem cells on different patterns of electrospun fibers. Biomed Mater 2013;8(5):055002. [22] Mahairaki V, Lim SH, Christopherson GT, Xu LY, Nasonkin I, Yu C, et al. Nanofiber matrices promote the neuronal differentiation of human embryonic stem cell-derived neural precursors in vitro. Tissue Eng Pt A 2011;17(5e6): 855e63. [23] Liu X, Ma L, Gao C. RNAi functionalized scaffold for scarless skin regeneration. Organogenesis 2013;9(2):76e8. [24] Teo BK, Goh SH, Kustandi TS, Loh WW, Low HY, Yim EK. The effect of micro and nanotopography on endocytosis in drug and gene delivery systems. Biomaterials 2011;32(36):9866e75. [25] Adler AF, Speidel AT, Christoforou N, Kolind K, Foss M, Leong KW. Highthroughput screening of microscale pitted substrate topographies for enhanced nonviral transfection efficiency in primary human fibroblasts. Biomaterials 2011;32(14):3611e9. [26] Chan J, O'Donoghue K, De la Fuente J, Roberts IA, Kumar S, Morgan JE, et al. Human fetal mesenchymal stem cells as vehicles for gene delivery. Stem Cells 2005;23(1):93e102. [27] Chong MS, Chan J. Lentiviral vector transduction of fetal mesenchymal stem cells. Methods Mol Biol 2010;614:135e47. [28] Chan J, Kumar S, Fisk NM. First trimester embryo-fetoscopic and ultrasoundguided fetal blood sampling for ex vivo viral transduction of cultured human fetal mesenchymal stem cells. Hum reprod 2008;23(11):2427e37. [29] Lu JJ, Langer R, Chen JZ. A novel mechanism is involved in cationic lipidmediated functional siRNA delivery. Mol Pharmacol 2009;6(3):763e71. [30] Hsu CYM, Uludag H. Cellular uptake pathways of lipid-modified cationic polymers in gene delivery to primary cells. Biomaterials 2012;33(31): 7834e48. [31] Rujitanaroj PO, Jao B, Yang J, Wang F, Anderson JM, Wang J, et al. Controlling fibrous capsule formation through long-term down-regulation of collagen type I (COL1A1) expression by nanofiber-mediated siRNA gene silencing. Acta Biomater 2013;9(1):4513e24. [32] Barber RD, Harmer DW, Coleman RA, Clark BJ. GAPDH as a housekeeping gene: analysis of GAPDH mRNA expression in a panel of 72 human tissues. Physiol Genomics 2005;21(3):389e95. [33] Lee MN, Ha SH, Kim J, Koh A, Lee CS, Kim JH, et al. Glycolytic flux signals to mTOR through glyceraldehyde-3-phosphate dehydrogenase-mediated regulation of Rheb. Mol Cell Biol 2009;29(14):3991e4001. [34] Gilleron J, Querbes W, Zeigerer A, Borodovsky A, Marsico G, Schubert U, et al. Image-based analysis of lipid nanoparticleemediated siRNA delivery, intracellular trafficking and endosomal escape. Nat Biotechnol 2013;31(7):638e46. [35] Hsu SH, Ho TT, Tseng TC. Nanoparticle uptake and gene transfer efficiency for MSCs on chitosan and chitosan-hyaluronan substrates. Biomaterials 2012;33(14):3639e50. [36] Rejman J, Bragonzi A, Conese M. Role of clathrin- and caveolae-mediated endocytosis in gene transfer mediated by lipo- and polyplexes. Mol Ther 2005;12(3):468e74. [37] Zhang XX, Allen PG, Grinstaff M. Macropinocytosis is the major pathway responsible for DNA transfection in CHO cells by a charge-reversal amphiphile. Mol Pharmacol 2011;8(3):758e66. [38] Hufnagel H, Hakim P, Lima A, Hollfelder F. Fluid phase endocytosis contributes to transfection of DNA by PEI-25. Mol Ther 2009;17(8):1411e7. [39] Howes MT, Mayor S, Parton RG. Molecules, mechanisms, and cellular roles of clathrin-independent endocytosis. Curr Opin cell Biol 2010;22(4):519e27.

Please cite this article in press as: Yau WWY, et al., The effects of nanofiber diameter and orientation on siRNA uptake and gene silencing, Biomaterials (2014), http://dx.doi.org/10.1016/j.biomaterials.2014.10.003