Electrostatic template-assisted deposition of microparticles on electrospun nanofibers: towards microstructured functional biochips for screening applications

Electrostatic Template-Assisted Deposition (ETAD) of microparticles is described as a new process to control the deposition of microparticles by electrospraying onto a substrate. It relies on the construction of an electrostatic template by electrospinning a thin layer of fibers onto a micropatterned collector. Because the fibers cannot release their charges when they are suspended over cavities of the micropatterned collector, an electrostatic template is formed with repulsive and attractive domains. This electrostatic template is then used to guide precisely the particle deposition during the electrospraying step. Microstructured bi-layer composites with a great variety of micropatterns can thus be elaborated with any kind of materials allowing the use of the ETAD process for a wide range of applications. As a proof of concept, the ETAD process was applied for the production of composite scaffolds with poly(e-caprolactone) nanofibers covered by a micropatterned layer of hydroxyapatite. This scaffold was then embedded in a biochip containing 21 wells and used for MG-63 cell proliferation and mineralization studies, showing their possible application in the screening of the scaffold structure for tissue engineering.


Introduction
One of the main goals in cell biology is to nd the optimal in vitro micro-environment to accurately mimic the in vivo one. Indeed, cell behaviors such as proliferation, migration and differentiation deeply depend on their micro-environment, which is dened by different parameters such as the topography, chemistry and stiffness of the substrate and the composition and physiochemical conditions (temperature, pH, O 2 , CO 2 ) of the cell culture medium. 1,2 Among these parameters, the substrate topography is identied as one of the most important impacting the cell fate, especially for stem or bone cells. High throughput screening has been developed in order to test a large number of micro-environment conditions and thus, nd the best topography regarding the intended application. Thanks to microfabrication and microuidics techniques, biochips (i.e. lab on chip devices devoted to biology) have emerged, offering the possibility to test different libraries of molecules or topographies in order to nd the best cellular micro-environment. 2 Furthermore, the high level of repeatability and rapidity provided by the biochips make them very attractive for biologists. 3 Regarding the topography, biochips allow the screening of hundreds different patterns. [4][5][6][7] However, because most of biochips are fabricated by photolithography, etching, or molding techniques, their topography cannot mimic satisfactorily the in vivo extracellular matrix (ECM), which is a nanobrous composite. Therefore, screening a large number of brous structures should be promising, especially to nd the adequate structure of brous scaffolds for tissue engineering, depending on the targeted application.
A simple fabrication technique for the elaboration of submicronic brous scaffolds for tissue engineering is electrospinning. Indeed, electrospun scaffolds have proven to be excellent candidates as substrates for in vivo and in vitro cell growth. 8 Moreover, using this technique, bers can be easily organized in the form of 2D 9-11 or 3D 12-15 microstructured scaffolds, with tailored pore sizes to allow cell colonization and inltration. 16,17 Finally, it has been demonstrated that the topography of the electrospun scaffold has got an important inuence on the cell behavior. [18][19][20] Electrospun membranes have been incorporated in biochips allowing biological experimentations in individual micro-wells having a bottom with non-structured randomly deposited nanobers. [21][22][23] However, to the best of our knowledge, nobody reported the elaboration of biochips allowing cell culture for a wide range of micropatterned nanobrous structures. In this work, we developed a new process, the Electrostatic Template-Assisted Deposition (ETAD), which allows the precise spatial deposition of microparticles onto electrospun nanobers. From this process, we elaborated microstructured biochips to screen the topography of nanobrous scaffolds for bone tissue engineering. For such application, several groups have shown that the addition of hydroxyapatite, the major mineral component of bone, to the nanobrous scaffolds enhance the osteophilic environment for the growth and mineralization of osteoblasts. [24][25][26] We thus elaborated a biochip for which the bottom of each well has its own composite brous microstructure (bars, hexagons, blocks or mazes) made of a layer of electrospun poly(3-caprolactone) (PCL) bers and a layer of electrosprayed hydroxyapatite (HA) microparticles. ETAD process allowed the precise location of the HA particles onto the PCL nanobrous layer. MG-63 osteoblast-like cells were then cultured followed by immunochemistry and alizarin red staining to evaluate the inuence of the micropatterned structures on the mineralization and to demonstrate the potential of application of such biochips.

Electrospinning/electrospraying
A vertical homemade electrospinning set-up was composed by a 18G needle connected to a positive high voltage power supply (Spellmann SL 10) and a rotative collector (rotating in the horizontal plan with a speed of 100 rpm) connected to the ground (high voltage power supply Spellmann SL 10). The polymer solution was pushed through the needle via a silicon feedline connected to a 10 mL syringe installed on a syringe pump.
In a rst step, the PCL solution is electrospun during a time t e (needle-to-collector distance ¼ 13.5 cm, ow rate ¼ 1 mL h À1 , V needle ¼ 25 kV). Then, in a second step, the HA suspension is electrosprayed over the layer of PCL nanobers during 15 min (distance ¼ 13.5 cm, ow rate ¼ 0.62 mL h À1 , V needle ¼ 18 kV).

Fabrication of the biochip collector
The biochip micropatterned collectors were elaborated by using photolithographic processes onto silicon wafers, with a photoresist (SU-8 2050, MicroChem) and a mask aligner (MJB4, SUSS MicroTec). Electron beam evaporator Plassys MEB5505 was then used to deposit a conductive layer (Al-160 nm and Au-40 nm) on the collectors to make them conductive. On each biochip collector, 21 wells were fabricated with 21 different structures. Each well had a diameter of 5 mm and they were spaced from each other by 5 mm. 5 different kinds of structures were studied: hexagon, bar, block, maze and random structure. The characteristic sizes L top and L bot are dened and summarized for each structure (see Fig. 2 and Table 1).

Fabrication of the biochip
The biochip was a homemade device (see Fig. 4b). It was composed of different elements: -A 50 Â 50 mm PMMA plate drilled with 21 holes to form the wells, corresponding to the 21 different electrospun structures.
-A 50 Â 50 mm PDMS seal with a height of 4 mm with 21 holes corresponding to the ones of the PMMA plate.
-A PCL/HA membrane obtained with the biochip collector.
-A 50 Â 50 mm PMMA plate with a height of 5 mm. Before assembling, each element of the biochip was sterilized in ethanol (except the PCL/HA membrane). The four elements were assembled together with four screws (see Fig. 4b). The assembled biochip was then exposed to UV light during 30 minutes to insure good sterilization. The PDMS seal was homemade. A 10/1 (wt/wt) mixture of PDMS-Sylgard Silicone Elastomer and Sylgard Curing Agent 184 (Dow Corning Corp., Seneffe, Belgique) was poured into a Petri dish.
Then, the mixture of PDMS-elastomer and curing agent was cured at 65 C during 1 hour until the polymer became rigid. Aer cooling at room temperature, the PDMS seal was carefully peeled from the Petri dish and perforated with a punch to make 21 holes with a diameter of 5 mm.

SEM characterization
The morphology of the biochips collectors and membranes were observed by scanning electron microscopy (SEM) (VEGA 3 LMV, TESCAN, Czech Republic) with a voltage of 5 kV. Before observation, the samples were sputter-coated with gold using a Q 150RS sputter coater (Quorum Technologies Ltd, United Kingdom).

Raman confocal microspectroscopy
The Raman confocal microscopy mapping were obtained using a Raman spectrometer (LabRam HR by Horiba Jobin-Yvon with a 600 lines per mm grating) coupled to a microscope (Model BX41, Olympus) with xyz mapping stage via optical bers. The excitation of Raman scattering was operated with a green laser at a wave length of 532 nm. The laser beam was focused using the macro-mapping mode with a laser spot of 8 Â 8 mm, steps of 8 mm and with a 10Â microscope lens. A confocal pinhole of 200 mm diameter was placed before the entrance slit to reject the Raman signal from out-of-focus planes. For each spot, a Raman spectrum with a good signal-to-noise ratio was recorded with an integration time of 6 s. The peak at 1106-1131 cm À1 was attributed to PCL whereas the peak at 947-988 cm À1 was attributed to HA signal.

Cell culture and seeding
Human osteoblast cells were obtained from the cell line MG-63 coming from human osteosarcosis. MG-63 osteoblasts were cultured in DMEM® (BioWhittaker® Lonza, France) completed with 10% FBS (Gibco® Life Technologies™, France), 50 U mL À1 penicillin, 50 mg mL À1 streptomycin, 2.5 mg mL À1 amphotericin B (BioWhittaker® Lonza, France), 200 mM sodium pyruvate (Gibco® Life technologies™, France), and 200 mM of glutamine (PAA, France) in 75 cm 2 asks. The cells were incubated at 37 C in a humidied atmosphere of 5% CO 2 and fed every 3 days. When cells reached sub-conuence, they were harvested by trypsin-EDTA treatment and viable cells were counted by trypan blue assay. The MG-63 osteoblasts were then seeded into the biochip 21-well plate at a density of 3000 cells per well and cultured with DMEM®. The seeded biochip were incubated at 37 C in a humidied atmosphere of 5% CO 2 and fed every 3 days. Biochips seeded with MG-63 osteoblasts were cultivated for 21 days in Dulbecco's modied eagle medium (D-MEM®, PAA, Austria) supplemented with 10% fetal bovine serum (FBS), 50 U mL À1 penicillin, 50 mg mL À1 streptomycin, 5 mg mL À1 Fungizone and 1% sodium pyruvate.

Immunocytochemistry staining and observation
MG-63 osteoblasts cells were xed with 4% paraformaldehyde for 1 h at 37 C, permeabilized and saturated with a solution of 0.1% Triton X-100 and 1% BSA diluted in PBS for 1 h. Aer washing with PBS, the biochips were then incubated for 30 min with Alexa Fluor 546-conjugated phalloidin (1/200, Molecular Probes) for F-actin labelling and 5 min with 200 nM DAPI (Sigma) for nuclear staining. Bone growth induction and extracellular matrix synthesis were measured by assaying expression of osteocalcin using polyclonal goat anti-osteocalcin (1/200; Santa Cruz Biotechnology) overnight at 4 C. Aer washing with PBS three times for 10 min, the seeded biochips were incubated with secondary anti-goat antibodies conjugated to Alexa Fluor 488 (1/200, Invitrogen). The biochips were observed under an epiuorescence microscope (LEICA DM 4000 B). Image treatment was carried out with Image J soware. The exposition gain was 3 for the DAPI and osteocalcin images. The exposition time for DAPI was 350 milliseconds and the exposition time for osteocalcin was 920 milliseconds. Osteocalcin expression was observed aer 21 days of cell culture in one biochip device.

Observation of the cell morphology by SEM
Aer 3 days of culture, cells were xed using 2.5% glutaraldehyde diluted in phosphate buffered saline (PBS) and then 1% osmium tetroxide diluted in PBS. The samples were then washed several times with PBS and dehydrated gradually with increasing concentrations of ethanol (25%, 50%, 75%, 90%, and 100%). Finally, the cells were chemically dehydrated with hexamethyldisilazane (HMDS) and observed by SEM as previously described.
2.9. In vitro examination of mineralization using Alizarin red S staining aer 21 days of cell culture Alizarin red S powder was dissolved in distilled water at 2 g for 100 mL. The pH of the Alizarin red S solution was adjusted between 4.1 and 4.3. Aer washing the scaffolds with PBS twice, the cells were xed by adding 75% ethanol in each well of the biochips. The biochips were incubated in the Alizarin red solution for 20 min and then thoroughly rinsed with distilled water. The biochips were then observed under an optical microscope. Image treatment was carried out with Image J soware. Three biochips were incubated with the Alizarin red solution: two biochips seeded with MG-63 cells and one biochip without cells (as a negative control).

Electrostatic template-assisted deposition (ETAD)
It has been shown that when electrospinning is carried out on a collector having protuberances, the suspended portions of nonconducting bers (i.e. which are not in contact with the collector) cannot efficiently release their electric charges 27 whereas the ber portions which are in direct contact with the collector can dissipate the charges. Thus, when using a micropatterned collector with walls and cavities, an electrostatic template showing both repulsive and attractive areas is formed (see Fig. 1a-c). Such electrostatic template has been used to control the deposition of the up-coming electrospun jet and produce 2D 11,20 or even 3D 13,14 patterned scaffolds. Thus, thanks to this electrostatic template effect, it was possible to control very precisely the deposition of electrosprayed hydroxyapatite (HA) microparticles. Indeed bi-layer composite membranes with a layer of PCL electrospun bers covered by a structured layer of electrosprayed hydroxyapatite (HA) microparticles could be elaborated (see Fig. 1d-f). HA microparticles were selectively deposited on the walls of the patterns (i.e. where the PCL ber portions were in direct contact with the collector). This highly selective deposition was observed for honeycomb patterns with an internal diameter as low as 40 mm (Fig. 1e). On the contrary, when the patterned collector was not previously covered by a layer of electrospun nanobers, HA electrosprayed microparticles were deposited both on the pattern walls and in the pattern cavities (see Fig. 1g). This experiment proves without any doubt, that the electrostatic template-assisted deposition (ETAD) of electrosprayed particles is induced by the layer of bers, which is rstly deposited on the patterned collector. Thus, the ETAD process allows the precise spatial deposition of microparticles onto a thin layer of nanobers previously electrospun onto a microstructured collector.
Using the ETAD process, it was possible to produce microstructured composite membranes of a large variety of pattern shapes such as bars, hexagons, blocks or maze (see Fig. 2a-j and Table 1). HA microparticles were precisely deposited on the top of the 21 different patterns of the collector in the case of the presence of a previously deposited thin layer of electrospun PCL bers obtained for an electrospinning time t e ¼ 90 s (Fig. 2f-j). When the time t e is increased to 13 min (corresponding to an average thickness of the PCL brous layer of roughly 80-150 mm), the controlled deposition is still observed but with some differences compared to shorter time t e (see Fig. 2f 0 -j 0 ). Indeed, by increasing t e , the shape of the electrostatic template may change due to ber aggregation or partial loss of the electric contact between the bers and the top of the patterns. Thus, a rst consequence of the increase of t e was that whatever the pattern size L top , ranging from 20 to 120 mm (see Fig. 2b-e and Table 1), the corresponding width L HA of the patterns formed by the deposited HA microparticles was almost constant and ranged between 15 and 20 mm (see Table 1). Therefore, the percentage of surface area covered by the HA microparticles (S HA ) was reduced when electrospraying aer longer electrospinning times. While it was highly dependent on the structures of the patterns at short electrospinning times, ranging from 15 to 90% of the total surface, it became more uniform aer longer times and around 15 to 30% for most of the patterns (Table 1). Regarding the bar and hexagonal patterns, HA particles were located exactly on the top of the patterns for the thinnest ones having an L top of 20 mm, with a low HA surface coverage of 16-20%, not really depending on t e (see Fig. 2f 0 , 2g 0  and 3). However, for patterns with larger width L top , the electric contact couldn't be insured over all the top surface of the patterns leading to bad electric charge release and thus the formation of repulsive areas. For such large L top , the contact was generally insured at the edges of the patterns leading to HA deposition only on these edges. As an example, in the case of Hex 100,120 , aer t e ¼ 13 min of electrospinning, HA was only deposited on the edges of the patterns with S HA ¼ 16 AE 5% (Fig. 2h 0 ) whereas aer a shorter electrospinning time, HA covered all the width L top of the patterns leading to S HA values as high as 74 AE 4% (see Fig. 2h). In the case of block patterns, whereas well-dened HA islands were formed at short t e (S HA ¼ 28 AE 2%, Fig. 2i), HA grids with a pitch corresponding to the distance between the centers of the blocks were observed aer the longest electrospinning time t e ¼ 13 min (S HA ¼ 16 AE 5%, Fig. 2i 0 ). This phenomenon could be explained by the propensity of the PCL bers to self-assemble aer long times of electrospinning 12 leading to the formation of bundles of bers crossing the block gap. Indeed, such bundles, containing small amount of remaining solvent, allow the electric charges dissipation along their axes favoring thus their coverage by HA electrosprayed microparticles resulting in the formation of HA grids.

Biochips for the screening of composite microstructured materials for bone regeneration
Our goal is to demonstrate that the ETAD process can be applied for the elaboration of microstructured functional membranes, which are then implemented in a biochip to study the inuence of the microstructuration on the behavior of MG-63 osteoblast-like cells. Membranes made of a layer of electrospun PCL nanobers covered by a layer of electrosprayed hydroxyapatite microparticles were fabricated using the collector with 21 different micropatterned areas (Fig. 2a and Table 1). Each area of 5 mm in diameter corresponds to the position of a milli-culture well of the biochip. An electrospinning time t e of 13 min was chosen in order to be able to peel off the membrane from the collector and to manipulate it, giving a nal bi-layer scaffold with structures similar to the one shown in Fig. 2f 0 -j 0 . Moreover it has the advantage to ensure smaller variations in HA surface coverage from one pattern to the other, between 15 and 30%, except for the random scaffold which exhibits 85% of HA surface coverage. Aer the electrospinning and electrospraying steps, the structured membrane (Fig. 4a) was peeled off from the collector and integrated in the biochip (Fig. 4b) for cell culture and biological characterization. Under these conditions of elaboration, 15 among the 21 wells of the biochip presented a bottom with a well-controlled structuration and were elaborated in a reproducible way (see Table 1 and Fig. 6).
The biochips were used to screen in parallel the effect of the different brous architectures on the proliferation and activity of osteoblast-like MG-63 cells. The main role of osteoblasts is to produce and secrete organic and inorganic bone extracellular matrix (ECM), called ostoid. 28 The organic bone matrix is constituted of various proteins, such as collagen, osteonectin, osteocalcin, and sialoproteins. The inorganic bone matrix is constituted of calcium-phosphate salts. Osteoconductivity of the various scaffold architectures was investigated by monitoring the proliferation and ECM forming activity of osteoblasts cultured in the biochip wells.
The morphology and spatial organization of the cells were evaluated by SEM observation aer 3 days of cell culture. For the largest patterns such as Hex 20,240 , it is observed that cells preferentially adhered on the pattern walls where HA is located (Fig. 5a). Furthermore, MG63 cells show elongated morphology suggesting a good adhesion with the brous substrate ( Fig. 5b  and c). For the smallest patterns (i.e. when the distance between the HA areas is lower than 100 mm) the cells gathered, covering all the surface of the scaffold and began forming microtissues even aer only 3 days of culture (Fig. S4 in ESI †). The proliferation and localization of MG63 osteoblasts on the micropatterned scaffolds was also observed by staining the cells nuclei with DAPI aer 21 days of culture (see Fig. 6a-e and S1 in ESI †), showing the good proliferation of the cells into the structured wells of the biochip. As previously shown by SEM, a higher density of cells (i.e. stronger blue intensity) is observed where HA is deposited. For hexagonal patterns (Fig. 6a), this result seems to be in contradiction with our previous study 20 for which a preferential location of MG-63 cells inside the nests of the honeycomb was observed. This difference could be explained by the presence of HA microparticles insuring a better affinity with the cells as compared with the hydrophobic surface of the PCL bers. Indeed HA particles have shown to promote adhesion and growth of MG-63 cells. 29  The production of ECM by MG-63 cells was estimated by immunouorescence assays. In contrast with DAPI, important differences are observed in the expression of osteocalcin by the cells as a function of the scaffold microstructure (Fig. 6a 0 -e 0 and S2 in ESI †). It is obvious that some patterns show much higher uorescence intensity than other types of patterns.
Finally, in order to evaluate the mineralization of the scaffolds by the MG-63 osteoblast-like cells, Alizarin red coloration of the membrane aer 21 days of cell culture was performed. In this assay, calcium salts are stained in red and a stronger red intensity indicates the presence of higher amount of calcium salts. Because our membrane already contains a certain amount of hydroxyapatite, a negative control test has been carried out by Alizarin red coloration of a biochip membrane containing hydroxyapatite without cells (Fig. S3 in ESI †). Fig. 6f shows a membrane aer removing from the biochip and coloration by Alizarin red. A signicant difference in Alizarin red intensity is observed, between this membrane, aer cell culture, and an as-produced membrane, thus showing a good mineralization of the scaffolds. However, no signicant  differences in Alizarin red intensity could be distinguished between the different kinds of the micropatterned structures.
These results show that our biochip can be used for cell culture and are compatible with different kinds of biological tests. The membrane can be removed from the device aer the cell culture and can be very easily manipulated for further observations such as microscopy.

Conclusions
In this article we showed a new electrospinning based process, the ETAD process, which allows the precise spatial deposition of microparticles by electrospraying onto a surface thanks to the "electrostatic template" effect obtained by electrospinning of a thin layer of non-conductive bers onto microstructured collectors. This process can be used to prepare bilayered scaffolds with a wide range of 2D architectures and made of two different materials depending on the intended application.
We used this process to elaborate a new kind of biochip, dedicated to the screening of brous composite structures for bone regeneration applications. The biochips contain 21 wells with different micropatterned structures such as hexagon, bar, block, and maze in their bottom. Each well structure was made of electrospun PCL bers covered by a microstructured layer of hydroxyapatite particles. MG-63 osteoblasts-like cells were then successfully cultured into the different wells of the biochip and immunochemistry as well as biochemistry assays were performed as a proof of concept. The biochip can be very easily unbuilt at the end of the cell culture and the scaffold with various microstructures separated from the setup and manipulated for microscope observation or further biological, chemical or mechanical tests.