Tunable electrospun scaffolds of polyacrylonitrile loaded with carbon nanotubes: from synthesis to biological applications

Growing cells in a biomimetic environment is critical for tissue engineering as well as for studying the cell biology underlying disease mechanisms. To this aim a range of 3D matrices have been developed, from hydrogels to decellularized matrices. They need to mimic the extracellular matrix to ensure the optimal growth and function of cells. Electrospinning has gained in popularity due to its capacity to individually tune chemistry and mechanical properties and as such influence cell attachment, differentiation or maturation. Polyacrylonitrile (PAN) derived electrospun fibres scaffolds have shown exciting potential due to reports of mechanical tunability and biocompatibility. Building on previous work we fabricate here a range of PAN fibre scaffolds with different concentrations of carbon nanotubes. We characterize them in‐depth in respect to their structure, surface chemistry and mechanical properties, using scanning electron microscopy, image processing, ultramicrotomic transmission electron microscopy, x‐ray nanotomography, infrared spectroscopy, atomic force microscopy and nanoindentation. Together the data demonstrate this approach to enable finetuning the mechanical properties, while keeping the structure and chemistry unaltered and hence offering ideal properties for comparative studies of the cellular mechanobiology. Finally, we confirm the biocompatibility of the scaffolds using primary rat cardiomyocytes, vascular smooth muscle (A7r5) and myoblast (C2C12) cell lines.


Introduction
Cell fate is driven by a wide range of biological signals of the microenvironment. [1]The composition and stiffness of the tissue microenvironment changes during development and disease.Mechanical properties and composition of the extracellular matrices (ECMs) are integrated into tissues and influence multiple cell behaviours, including differentiation, migration and proliferation, tissue homeostasis, remodeling and development.Therefore, developing in vitro models, permitting independent modulations of physical and chemical properties is essential to identify parameters that are influencing cell behavior.To recapitulate the complexity of the microenvironment, multiple in vitro models have been designed.Hydrogels, [2] made of hydrated polymers, share a lot of commonalities with the ECM [3] and are extensively employed for cell culture.However, commonly used strategies have either low biocompatibility, require reticulation after cell embedding, are incorporating chemicals impacting cell phenotype, are not suitable for long term cell culture, or present difficulties to perform cellular and molecular analysis.Moreover, chemical and physical parameters are difficult to control independently, adding to the many limitations of hydrogels. [4,5]Sponges and ceramic scaffolds are suitable for bone tissue regeneration, but their stiffness and porosity cannot be easily tuned. [6]Organoids and other 3D cell do not allow fine control of ECM properties, suffer from poor oxygen and nutrient transport, necrosis and limited maturation; hence impairing their reliability for many applications.As an alternative approach, fibrous scaffolds appear to be a relevant scaffold type for 3D cell culture, compatible with live imaging and most biological studies. [7,8]ibrous scaffolds are commonly obtained from decellularized tissues, electrospinning, 3D printing, or thermally induced phase separation.Because decellularized tissue suffers from heterogeneity and poor reproducibility, [9] this option is often considered less favorable as choice for 3D models.On the other hand, nanofibers obtained by electrospinning are increasing in popularity.During electrospinning, a polymer solution is submitted to a high electric field. [10]Temperature, humidity, polymer choice, viscosity and concentration influence the resulting fiber properties.The fibrillar and interconnected porous architecture of electrospun fiber networks is closely mimicking fibrous tissue, ECM and capillary morphology. [5,11]lso, fibers can be designed to incorporate biological stimuli such as proteins. [12]Although electrospinning can use both synthetic and natural polymers, synthetic polymers are frequently preferred, since their physical and chemical properties can be more easily controlled and tuned.They are low cost and easy to process, suited to a wide span of application and adaptable to industrial technology. [13]They offer better resistance to enzymatic degradation, and to immunogenic reactions. [14]Many synthetic polymers have already been used for electrospinning, such as polyvinyl acrylate (PVA), [15] polycaprolactone (PCL), [16] polystyrene (PS), polyvinyl chloride (PVC), poly (lactic acid) (PLA), or PVDF. [10,12,17,18]However, so far, their usefulness for cell biological studies has been limited by poor biocompatibility.Electrospun fibers scaffolds have often been associated with poor proliferation, increased cytotoxicity due to post-processing and/or chemical residues, and unphysiological mechanical properties. [5]Those limitations can be tackled by using electrospinning of Polyacrylonitrile (PAN) fibres.PAN is a thermoplastic polymer, soluble in polar solvents, with high inter and intramolecular cohesion energy and tunable mechanical properties.It is often used for the fabrication of carbon nanofibres, due to its high carbon rate (56 %), its mechanical flexibility, chemical stability (after heat treatment) and good moldability. [19]PAN can further be enriched with multi-walled carbon nanotubes in order to increase the control of graphitization process and the resulting fibers have been envisioned as supports for biological applications. [20]However, the graphitized nature of the resulting material can represent a limitation for its biocompatibility -even if differentiation of iPSC was successfully achieved. [21]In a recent study, our group assessed the relevance of PAN/MWCNT based scaffolds to study the behavior of glioma cancer cells versus scaffold's mechanical properties. [22]he satisfying biocompatibility of our scaffold is likely related to weakly graphitic nature of the fibers since they endure a mild temperature treatment before cell exposure.However, many questions remain open on the mechanisms explaining how the formulation influences the properties of the final material, and on the potential of these scaffolds to grow other cell models.In the present article, we characterize in detail the structure, alignment, chemical and mechanical properties of the scaffold, as well as the biocompatibility with three different cell types.The scaffolds open interesting perspectives as model material for the growth of many cell types both for drug screening applications and fundamental studies in tissue engineering.

Functional groups at the surface of the fibres
To generate a range of fibre networks with different mechanical and morphological properties we electrospun PAN networks with aligned, or non-aligned fibres and different concentrations of MWCNTs (0 %, 0.0015 %, 0.00625 % and 0.05 % w/w MWCNT).To characterize the fibre mats, we first assessed the surface chemistry of scaffolds with and without prior heat-treatment by performing vibrational spectroscopy analysis.Raman spectroscopy, particularly suited for graphitic materials, did not provide significant insight as the graphitization of our scaffold is very limited (Figure SI 1 of supplementary information).
We preferred using FTIR (Fourier Transformed InfraRed) spectroscopy (Figure 1A).Before heat treatment, the spectrum of nanofibres shows characteristic vibrations at 1094 cm À 1 for secondary alcohol, 1452 cm À 1 for alkane group, at 1667 cm À 1 for trisubstituted alkene, 2242 cm À 1 for C�N and 2940 cm À 1 for CH alkane.After heat treatment, the spectrum shows a weak band at 807 cm À 1 attributed to the C=CH aromatic group and triazine ring.Nanofibres contain phenol OH functional groups as shown by the 1300 cm À 1 band.The strong band at 1400 cm À 1 is characteristic of CH alkane group and the strong 1600 cm À 1 band witnesses the presence of C=C cyclic alkane.Stabilized PAN contains CN as shown by 2210 and 2242 cm À 1 bands.The strong band at 2345 cm À 1 is attributed to C=O groups.There are also NH groups (2941 cm À 1 ).Finally, the broad weak bands from 3000 to 3600 cm À 1 correspond to aromatic NH2, to NH and OH stretching, to C=NH and CH ring.The comparison of the spectra before and after the heat treatment, including the attenuation of the CN bands and the enhancement of the C=C bands, shows that PAN has undergone aromatization and dehydrogenation, which is in good agreement with the color change of the samples from white to yellow/brown.The functional groups on the surface enable the formation of hydrogen bond between nanofibres and cell surface, which makes the matrix biocompatible and favorable for cell attach- ment.The surface chemistry with varied functional groups enables coating of the nanofibres with biomolecules such as extracellular matrix (ECM) proteins (e. g. laminin or fibronectin) to better mimic the cells microenvironment.Neither the addition of carbon nanotubes (in the here used concentrations), nor the thermal treatment had noticeable impact on the surface chemistry of the scaffolds (Figure 1B).

Morphology and inner structures
Next, we investigated the morphologic characteristics of the fibres using scanning electron microscopy (SEM).Because the rotation speed of the drum collector impacts the alignment of nanofibres, we changed the speed from 2000 rpm for aligned (A) nanofibres (Figure 2A) to 20 rpm, for non-aligned (NA) nanofibres (Figure 2B).The distribution of fibre orientation and coherency were quantified using ImageJ and the Orientation J plugin (Figure 3C-D).As expected, the orientation distribution was narrower and showed a single peak for aligned fibres and was unaltered by the addition of MWCNT (Figure 2C-D).
We further analysed the structural homogeneity of each scaffold after thermal treatment.A fully quantitative characterization of nanometric fibres on a macroscopic sample measuring several tens of cm 2 is of course barely possible.We, however, tried to provide a coarse picture of the structural homogeneity on a limited number of regions randomly chosen and give the results in Figure 3.For this fibre, diameters were analysed with the ImageJ plugin DiameterJ, using SEM images of different regions of the aligned nanofibres scaffolds (Figure 3A, produced by electrospinning a solution of 10 % w/w polyacrylonitrile in N, N-Dimethylformamide with an applied voltage of 20 kV, a flow rate of 2.4 ml/h and a rotating speed of 2 000).The analysis of fibre diameter from five distinct parts (~1 cm 2 , containing 60-150 visible fibres) of the same scaffold suggested a high level of homogeneity with 680 � 30 nm fibre diameter (Figure 3B).The pore areas ranged from 0.42 μm 2 to 489 μm 2 with an average pore area of 27.23 μm 2 , but no significant difference between different areas was observed (Figure 3C).We further performed SEM analyses of the inner structure of a 600 μm-thick scaffold after delamination.This confirmed homogeneity was not limited to the surface, but both the diameter and the pore area were conserved from the surface to the inside of the scaffold (Figure 3F-3G).
We tested the effect of MWCNTs on structural parameters, such as the diameter, mesh hole area and the orientation.SEM of both aligned and non-aligned (NA) scaffolds bearing different MWCNT concentrations (0 for no MWCNT, + for 0.0015 %, + + for 0.00625 % and + + + for 0.05 % MWCNT) suggested that none of the structural parameters were affected by the addition of MWCNTs (Figure 4 A-D).This indicated the potential to use the MWCNTs to tune mechanical properties, while keeping the structure of the scaffolds constant.We would like to underline the fact that this automatized processing of SEM images constitutes a robust semi-quantitative approach for the characterization of fibrous scaffolds that could be applied to other types of scaffolds.
To better understand the impact of MWCNT on the PAN fibres, we next analysed their organisation within the fibres using X-ray nanotomography and transmission electron microscopy (TEM) (Figure 5A).To this aim, we used MWCNT of the same dimensions (mean diameter external of 25 nm, 400 μm of  length), bearing iron-based nanoparticles in their inner channel, allowing us to discriminate the MWCNT from the PAN fibres.Using X-ray nanotomography of PAN fibres with 0.05 % (w/w) concentration of MWCNT we detected bright regions bearing electron rich elements -corresponding to Fe-based compounds present in the inner core of the fibres (Figure 5BÀ C).Similar cross-sectional TEM images showed MWCNT in the center of the fibres and along their length for the 0.05 % MWCNT fibres (Figure 5D-E-F), which were absent in the control PAN fibres without MWCNTs (Figure 5G-H-I).Among the 195 sections of fibres, 19 were showing MWCNTs, in line with the low mass content of MWCNT in the fibres and the homogeneous dispersion of the nanotube in the polymer solution.These results constitute the first direct evidence of the internal localization of carbon nanotubes inside ex-PAN fibres, which was an experimental challenge considering the very low electronic contrast between the fibres and carbon nanotubes.They also confirm that the mechanical effects produced by MWCNT incorporation are not related to MWCNTs lying at the surface of the fibres.

Mechanical properties
Previously, addition of MWCNT to electrospun fibres was shown to modulate the elastic modulus of the scaffolds. [22]To get further insight into the mechanical properties of the individual fibres and fibre-networks, depending on the alignment and MWCNTs concentration, we performed atomic force microscopy (AFM) and nanoindentation experiments, respectively.We first characterized the mechanical properties of the scaffolds using AFM with a silicon nitride cantilever functionalized with silica beads (Figure 6A).We found an increase in Young's modulus with the concentration of MWCNTs: ~115.7 kPa (without MWCNTs), ~102 kPa (with 0.0015 %MWCNTs), ~165.9 kPa (with 0.00625 %MWCNTs) and ~257,7 kPa (with 0.05 %MWCNTs) (Figure 6B).These values are larger than the ones we reported in ref. [22].Due to the nanofibers properties, those high values are probably due to a strong attractive force that can be seen in the curves (Figure 6C).
High loading force values (from 1 nN) led to a stiffness overestimation due to the influence of the substrate as reported previously.To determine the macro-scale mechanical properties of the scaffolds, we performed nanoindentation in liquid phase with a 0.45 N/m probe.In this context, the measured Young's moduli were close to what was reported previously, [18] whereby the Young's Modulus of the nanofibers increased with the concentration of MWCNTs (2.923 kPa without MWCNTs, 12.34 kPa with 0.0015 %MWCNTs and 63.71 kPa with 0.00625 % MWCNTs).No significative differences in stiffness appeared between aligned and non-aligned fibres measured without MWCNTs: 2.923 kPa for the aligned nanofibers and 1.303 kPa for the non-aligned (Figure 7).Those stiffnesses appear relevant in physiological and pathological context, therefore suitable for cell biology studies in a pertinent mechanical environment.

Modification of the morphology of the scaffolds
Increasing pore size is known to increase cellular infiltration through the scaffolds. [23,24]To modulate the pore size we decided to hang loads on the fibres attached perpendicularly to the direction of alignment of fibres during heat treatment.Pore size increased significantly and proportional to the loaded weight (Figure 8FÀ G).When we attached loads along the fibre's direction, the alignment increased, and pore size decreased (Figure 8).We also noticed a waviness of the loaded fibres.Curvotaxis is known to impact the shape of the cell, the nucleus and the cytoskeleton, which again affect gene expression, focal adhesion organisation or cell migration. [25]To characterize the morphology of fibres after loads were attached, we analysed the waviness, as done in ref. [26].Here waviness is the ratio of two distances (Lc/Lo), whereby Lc is the length of the fibre following the wave and Lo is the length of the straight path between the beginning and the end of the fibre (Figure 8H, measured with ImageJ on SEM images).By increasing load weight, we found a decrease in the ratio of Lc/Lo, indicating an increase in waviness (Figure 8I).Therefore, applying a load to the scaffolds during the thermal stabilization step allowed controlling the topography and mean mesh size of the scaffolds.This opens exciting perspective towards the growth of cells of larger dimensions, and/or developing in vivo in concave environments.

Biocompatibility
PAN scaffolds are not soluble in cell culture medium, and not biodegradable in contact with cells (Cornu et al., 2019).Biocompatibility is essential in the design of scaffolds for cell culture and electrospun fibres scaffolds have been described with poor proliferation, cytotoxicity due to chemical residues, and non-biologically relevant mechanical properties.Therefore, we next analysed cytotoxicity of our scaffold, using live & dead assays for three cell types, for which alignment is described as significant in vivo: A7r5 cells (vascular smooth muscle cells), C2 C12 (myoblasts cells that are precursor of contractile skeletal muscle cells), and neonatal rat cardiomyocytes (NRC).The cells were seeded on scaffolds containing different amounts of MWCNT and on glass coverslips (CS) as the control.For the NRC, the scaffolds were functionalized with fibronectin or laminin, while fibronectin was used for C2C12 and A7r5 cells.Live and dead cells were quantified after 5 days of culture.All cell types spread and interacted with all the compositions of scaffolds (Figure 9; microscopy maps are exemplified for A7r5 cells grown on MWCNT-free scaffolds in Fig. 9B, and other maps can be found in Supplementary Information SI2).For all scaffolds with different concentrations of MWCNTs and alignments, we found a minimum of ~80-85 % of living cells after 5 days (Figure 9A, C, D) implying very satisfying biocompatibility of the scaffolds, [27] whereby cells were found throughout the whole fiber mat (data not shown).To summarize, with these new evidences of the biocompatible character of three cell lines belonging to muscular tissues, we can now assess that the PAN/MWNCT electrospun scaffolds present a strong interest for the in vitro study of a variety of cells ranging from brain to muscular tissues (including cardiac tissues).

Experimental 3D Nanofibre scaffolds
Polyacrylonitrile nanofibres were produced by electrospinning a solution of 10 % w/w polyacrylonitrile (Sigma Aldrich, 181315, Mw 150,000) in N, N-Dimethylformamide (DMF, Sigma Aldrich, D4551, > 99 %, molecular biology grade).The rotating drum module of the  electrospinning set-up (IEM Electrospinner) was placed 15 cm away from the needle.A voltage of 20 kV and a flow rate of 2.4 ml/h were applied.To obtain aligned fibres, the rotating speed of the drum was 2 000 rpm, and 20 rpm to obtain non-aligned fibres.Multi-walled carbon nanotubes (MWCNT) were purchased from the Nanocyl company.These MWCNT are widely used for research as they have been extensively characterized and their structural and chemical properties were published in detail. [28]Their main features involve a diameter of 9.5 nm, a length of 1.5 μm, a 90 % purity, amounts of catalyst residues of 9 wt.% of Al2O3, and of 1 wt.% of total Fe).They were added in the electrospun solution (0 %, 0.0015 %, 0.00625 % and 0.05 % w/w MWCNT) to tune the scaffolds mechanical properties.To localize MWCNTs in the fibres, other MWCNT of same diameter and length, but containing Fe (MWCNT + 5 %w/w Fe, CEA) were produced according to an aerosol assisted Chemical Vapor Deposition CVD technique [29,30] and used for TEM, and X-ray tomography experiments.TEM analysis shows that Febased particles are encapsulated in carbon nanotube core.Raman spectroscopy gave an I D /I G ratio of 0.4.After overnight drying at room temperature, the nanofibres were put in a chamber furnace for heat treatment: a ramp of 120 °C/h to reach 250 °C was applied, followed by 2 h at 250 °C under air.Before biological use, the nanofibre matrices were then cut and sterilized in autoclave.

Modification of pore area
To modify the structure of the scaffold, and more precisely the pore area (i.e., the mesh of the net on top view, related to the pore volume), fibres were hung with loads from 0 to 56 g during heat treatment according to the scheme depicted in Figure 10.To increase pore area, loads were hung perpendicular to the alignment of fibres, whereas to decrease the pore area, loads were hung parallel to the alignment.

FTIR
ATR-FTIR (Attenuated Total Reflection FTIR) spectra were recorded on a Thermo Nicolet Nexus FTIR (Thermo Fisher) with a diamond crystal from 4000 cm À 1 to 400 cm À 1 .OMNIC software is used to record FTIR spectra.The infrared spectrum of the background is measured before acquisition of the sample spectrum, at 32 scans and a resolution of 4 cm À 1 .

Scanning Electron Microscopy
Scanning Electronic Microscopy (SEM) analysis was performed on a Hitachi S4800.Three scanning electron micrographs were captured of each sample mounted onto SEM stubs with carbon tape.Samples were imaged at five hundred times, with an acceleration voltage of 15KV.Prior to SEM analysis, samples were coated with 10 nm of gold using the SC7620 metallizer (Quorum Technologies).

Tomographic Transmission Electron Microscopy
Samples were embedded in LR (London Resin) white resin (a hydrophilic acrylic resin) and cut with a Leica UC7 ultramicrotome.Ultrathin sections (70 nm) were collected on three hundred mesh carbon coated copper grids.TEM analyses were performed on a JEOL 2200FS microscope operated at 200 kV.This microscope is equipped with a field emission gun (FEG) and an in-column Omega-type energy filter.Images were acquired on a CCD Gatan UltraScan 4000 (4kx4k) camera.
DiameterJ [31] The DiameterJ algorithm (plugin of the ImageJ software) was used to analyse 8-bit SEM images of 500 dpi resolution, according to the procedure summarized in Figure 11.The first step of the procedure consists in segmenting the image into black (background) and white (fibres) colors using different algorithms.The segmented image named "T6" and obtained from the thresholding technique was selected for analysis.The second step of DiameterJ, applied on this image, gives some averaged structural information such as the so-called super pixel (mean fibre diameter), pore area (determined by counting the number of black pixels).Then, Orientation J, an Image J plugin included in Diameter J, gives an orientation histogram.We fitted it with a Gaussian function of the following equation.Here μ is the mean value and σ is the standard deviation.We have thought of how close the data are close together.The more aligned the fibres are, the smaller the value of σ will be.Using Orientation J directly on the SEM image, we can obtain a value called coherency.Briefly, after having chosen and applied a Gaussian analysis window equals to the mean fibre diameter, the Fourier gradient structure tensor is evaluated for each pixel in the image over a Gaussian window. [32]The orientation histogram built is a weighted histogram where the weight is the coherency.The coherency gives us the number of fibres aligned with the principal orientation found.

X-ray nanotomography
A fibre of PAN without MWCNT and one made from the solution of PAN with 0.05 % MWCNT (containing iron-based nanoparticles) have been analysed by X-ray nanotomography.This experiment was performed at the European Synchrotron Radiation Facility (ESRF, Grenoble, France).The isolated fibre is stuck with UV glue on a quartz capillary of 100 μm diameter (Figure 12).Scans are done at 17.5 keV with 2505 projections on 360 degrees in holotomography   with a resolution of 25 nm pixels.To have a 3D reconstitution of the fibres and be able to see nanotubes, the stack of images is analysed with "volume viewer", an Image J plugin.

Atomic Force Microscopy (AFM)
AFM measurements were done using a Dimension Fastcan ICON PT (Brucker, CBMN, Bordeaux, FRANCE).Calibration of cantilevers was done on sapphire before each experiment.Samples were placed on a wafer of silicon that was stuck on a microscope slide and measured, in liquid phase, using triangular Silicon Nitride cantilevers (MLCT-F, Brucker) functionalized with silica beads (diameter 7.75 μm).The nominal spring constant was 0.5 N/m and the resonance frequency 20 kHz.Matrices of force/displacement curves were measured (5 μm, 8*8, ramp size between 500 nm and 1 μm, approach and withdrawal velocities 1 μm/s, force setpoint 3 or 5 nN).The indentation curve is divided in small intervals and fitted with Hertz model to obtain apparent Young's modulus.The force curves were then analysed with the software Gwyddion.The length and height of the linear part of the approach curve is reported in Excel.Then, the Young modulus is obtained with the formula: ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi tip radius p ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi Where tip radius = 7.75 μm and Poisson's = 0.5.

Nanoindentation
Nanoindentation measurements were performed using a Chiaro nanoindenter system (Optics11, Amsterdam) mounted on a Leica DMI8 epifluorescence microscope.The scaffold was immersed in PBS and the probe was lowered on it.Distinct parts of the scaffold were studied.The tip radius of the probe was 22 μm and its stiffness was 0.45 N/m.The Hertzian contact mechanics model was used to obtain the Young's modulus according to this equation: where R is the radius of the indenter tip, h the indentation depth, P the applied force.E eff is the effective Young's modulus obtained according to this equation: where Poisson's ratio υ is set to 0.5. [33]ll culture for live & dead assay Fibres were sterilized by autoclave and added in 24-well plates.After Plasma treatment (100 % air, 5 min), scaffolds were immerged in 300 μl of fibronectin or laminin at 1μg/ml in PBS under sterile condition for 1 hour at 37 °C and washed twice with PBS.The A7R5 and C2 C12 cells were cultured on the nanofibers in DMEM with 10 % FBS, penicillin and streptavidin and NRC cells on 77 % DMEM, 18 % M199, 2 % horse serum, 2 % GlutaMAX, 1 % P/S.After five days in cell culture, live & dead assay was performed thanks to the Viability/cytotoxicity Kit for mammalian cells (L3224, Thermo Fischer).Cells were incubated at 20 °C for 30 min in the staining solution and imaged using an optical Leica microscope.

Statistical analysis
For the morphological characterization, the measurements were performed on 4 different samples for each condition and on at least 50 fibres or pores per sample.The values are expressed as mean values over all measurements + /-standard error of the mean (SEM).
For the mechanical characterization, the measurements were performed on 2 different samples for each condition and on at least 3 different localizations per sample.The values are expressed as mean values over all measurements + /-SEM.P-values were calculated from one-way ANOVA with Dunnett correction for multiple comparisons using Prism GraphPad software.For the live and dead assay, the experiments were carried out in biological triplicates.The values are expressed as mean values over all measurements + /-SEM.

Conclusions
Based on the encouraging results reported in the recent literature on the growth of neural cells on 3D scaffolds made of thermally treated PAN electrospun fibres, we have presented a multi-technique analysis of the structure of such scaffolds both at the supra-fibre and the intra-fibre scale, and first biocompatibility assays on cell lines belonging to muscular tissues.Using an automatized SEM image analysis routine, we managed to depict the homogeneity of the fibre diameter and of the mesh size distribution all over the scaffolds in a semi-quantitative way.More importantly, we evidenced that the incorporation of MWCNT in the formulation of the scaffold did not have significant impact on the structure of the fibrous scaffold.We have managed to image the MWCNT in the inner core of the fibres, ensuring that their mechanical effect on the Young modulus of the scaffolds is truly a core effect and is not linked to the presence of MWNT at the surface of the fibres.Therefore, these results demonstrate that we are able to propose scaffolds presenting a range of different elastic properties without impacting any other topologic parameters of the fibre scaffold.We even showed that specific loading of scaffolds during thermal stabilization allowed tuning the curvature of the fibres, offering another control on the size and geometry of the hosting sites for cell growth.Altogether, we believe that the aspresented 3D scaffolds made of PAN/MWCNT electrospun fibres represent a precious tool for the in vitro study of a large range of cell types, e. g. cells from cardiac tissues as demonstrated by the compatibility assays presented in this article.

Figure 1 .
Figure 1.FTIR is used to characterize the surface chemistry of the scaffold (A) FTIR spectrum and structure of PAN nanofibres before (blue) and after (red) thermal stabilization with a proposed structure (inset) from (Bailey et al. 1970).(B) FTIR spectrum and structure of PAN nanofibres with 0.05 %MWCNT.

Figure 2 .
Figure 2. Electrospinning was used for fabrication of aligned and nonaligned fibres network.SEM images of PAN nanofibres scaffolds obtained by electrospinning from a solution of PAN in DMF with a drum rotation speed of 2000 rpm (A) and 20 rpm (B).Distribution of alignement direction of fibres in PAN membranes made of aligned (C) and non-aligned (D) fibres.

Figure 3 .
Figure 3. PAN nanofibres (A) obtained by electrospinning after thermal stabilization with the letters corresponding to the different locations of the scaffold that was cut to be analyzed by SEM (B) Mean fibre diameter measured in the area denoted a to e. (C) Pore distribution of aligned scaffold measured in the area denoted a to e. SEM images of aligned PAN nanofibres scaffold at the surface (D) and inside (E).Comparison of the diameter (F) and pore area (G) of the surface and the internal part of the scaffold.

Figure 4 .
Figure 4. To determine if the addition of MWCNTs had an impact on the fibremorphology, an analysis of SEM images of the different scaffolds was made.Comparison of fibres mean diameter (A) and pore area distribution (B) for non-aligned (na) or aligned (a) fibres with different amount of MWCNT.(C) Standard deviation of the orientation calculated from the orientation histogram given by Diameter J plugin.(D) Coherency of the different membrane obtained with Orientation J applied on SEM images.Here 'a' stands for aligned while 'na' stands for non-aligned and 0 stands for 0 % MWCNT, + for 0.0015 %MWCNT, + + for 0.00625 %MWCNT and + + + for 0.05 % MWCNT.

Figure 5 .
Figure 5.To better understand the impact of MWCNTs, we needed to understand their organisation within the scaffold.TEM analysis and X-Ray nanotomography were performed to determine this organisation.A. TEM image of MWCNT containing 5 % Fe particles (shown by the red arrow).B-C.X-Ray nanotomography of a PAN electrospun fibre containing MWCNT.TEM images of ultramicrotome cut of aligned fibres with 0.05 % MWCNT (DÀ E-F) or without (G-H-I).

Figure 6 .
Figure 6.AFM measures fibres' stiffness.A. Image obtained by AFM with PFT-QNM mode to show the points where the measurements were made (white crosses).B. Young's Modulus of scaffolds obtained by AFM using a silicon nitride cantilever functionalised with silica beads in liquid phase.C. A force-curve showing the strong interference between the tip and the scaffold for AFM measurements.

Figure 7 .
Figure 7. Young's modulus is determined to characterize scaffold's stiffness.Young modulus of aligned or non-aligned fibres without MWCNT and of aligned fibres with 0.0015 % and 0.00625 % MWCNT, obtained by nanoindentation.

Figure 8 .
Figure 8.To increase pore area, a load was hung on scaffolds during heat treatment.SEM images of fibres hung without load (A), with load of 29 g (B) and 56 g (D) perpendicular to fibres alignment, with loads of 29 g (C) and 56 g (E) parallel to fibres alignment, after heat treatment.F. Pore size areas (μm 2 ) measured with Diameter J. G. Curve of pore size area according to weight of loads hung perpendicular to alignment of fibres.H. SEM image of fibre after heat treatment, hung with a load of 29 g perpendicular to fibres alignment.Lo and Lc are drawn here and measured with Image J. I. Values of the waviness parameter according to the weight load used and the orientation of traction.

Figure 9 .
Figure 9. Live & dead assays are performed to check the biocompatibility of the scaffold.This assay (N = 3) shows no toxicity neither for A7R5 (A) nor for C2 C12 (C) nor for NRC (D).(B) A7R5 cells on aligned nanofibres without MWCNT.The nucleus is stained in blue, the phalloidin in green and YAP in red. mÞ

Figure 10 .
Figure 10.To increase cell infiltration, the increase of pore is necessary.Here is a scheme of the modification of pore area process.The rectangle illustrates the scaffold and the draws represents the main direction of The fibres.F Ĥ illustrates the direction of the traction which is perpendicular to the fibres alignment on the left and parallel on the right.Figure 11.To characterize the morphology of the fibres, SEM images were analysed with Diameter J plugin.Here is a functional scheme of the plugin.

Figure 11 .
To characterize the morphology of the fibres, SEM images were analysed with Diameter J plugin.Here is a functional scheme of the plugin.

Figure 12 .
Figure 12.Optical microscopy image oft he X-ray nanotomography experimental assembly used to characterize MWCNT organisation within the.