Morphology regulation of nanowire forests for tumour-cell behaviours

Three-dimensional nanowire arrays based on different biomaterials have emerged as promising platforms for the studies of cell behaviours over the years. Fabrication of nanosurfaces with complex geometries and controllable morphologies is still a challenge. In this work, nanowire forests (NWFs), which exhibited various aspect ratios, were fabricated using a reactive ion etching technique. Especially, by regulating bombardment periods of oxygen and argon plasma, the aspect ratio of nanowires in NWFs could be decreased from 29:1 to 7:1. Based on superhydrophilicity of the NWFs, the interaction between the NWFs with various aspect ratios and HeLa cells (human cervical carcinoma cell line) after different culturing periods was systematically investigated. NWFs with a low aspect ratio provide appropriate nanosubstrates for the cells, enabling them to grow on and adhere to the surface. NWFs with a high aspect ratio exhibited low mechanical strength under initial cell adhesion, which further hindered cell growth and migration as the incubation time increased. These ﬁndings provide essential guidance in designing nanostructure geometries for more efﬁcient capture of circulating tumour cells and manipulation of cytomorphology in future.


INTRODUCTION
For years, cancer has been the greatest danger to human life and health [1,2]. As a valuable biomarker of cancer liquid biopsy, the detection and analysis of circulating tumour cells (CTCs) offer the possibility for clinical diagnosis, prognosis, and treatment [3][4][5]. With the rapid development of microelectro-mechanical systems and bionanotechnology, surfacefunctionalised and nanostructured substrates have emerged as promising strategies for CTC capture [6][7][8]. Nevertheless, most related researches were mainly focused on EpCAM-positive CTCs (e.g. A549, MCF-7, HepG2) in previous studies [9,10], and it is also necessary to develop a method for capturing EpCAM-negative CTCs (e.g. HeLa cells (human cervical carcinoma cell line)) to further develop the traditional negative selection strategy. Numerous researches have demonstrated that nanoscaled cellular surface components (e.g. microvilli [11], filopodia [12]) could affect cell adhesion [13], migration [14], and other behaviours [15]; besides, surface nanotopography and chemical composition on substrates were believed to be able to manipulate these components [16][17][18]. Compared with planar interfaces, CTCs incline to attach to nanostructured surfaces similar to the extracellular matrix (ECM) [19]; therefore, it would be interesting to mimic nanostructures in ECM and assess interactions between CTCs and these nanostructures to optimise the substrate topographical design for CTC isolation [20,21]. Owing to high surface-to-volume ratios and tunable surface energies, nanowire or nanopillar arrays are able to interact with massive cells in a high-throughput and highspatiotemporal-resolution manner [22]. Qi et al. fabricated silicon nanowire (SiNW) arrays of different densities and high aspect ratios with self-assembly nanoelectrochemistry [23]. The adhesion and spreading behaviours of HepG2 and LX-2 were characterised on the SiNW arrays. The results showed that SiNW arrays could not only enhance the cell-substrate adhesion force but also restrict cell spreading. Further, in order to explore how cell morphologies influenced the cell capture efficiency, Kim et al. prepared SiNWs and quartz nanopillars (QNPs) by metal-assisted chemical etching and nanosphere lithography, respectively [24]. The experimental results demonstrated that the nanoscale topography could guide the formation of cellular needle-like filopodia, which dramatically improved the adhesion between cells and nanostructured substrates. In terms of cell capture yield, the SiNW arrays were four times (91.8%) higher than the planar glass substrate (24.1%). Furthermore, cell behaviours (e.g. proliferation, adhesion, migration) were recently found to be strongly dependent on the shape, size, density, and especially aspect ratio of vertical nanostructures; in this respect, the interaction between HeLa cells and different nanostructures is worth to be explored. On the other hand, chemical compositions of the nanostructures, which affect the wettability of the surfaces would further affect cell behaviours. For instance, hydrophilic or superhydrophilic surfaces were adopted to improve cell adhesion, whereas hydrophobic or superhydrophobic surfaces were used for anti-adhesion and anti-growth of cells [25]. In view of this, hydrophilic nanostructures with different aspect ratios are of importance for the investigation of HeLa cells. So far, diversiform nanostructures have been prepared by advanced methods, including metal-assisted chemical etching [26], ultrafast laser irradiation [27], nanoimprint lithography [28], chemical vapour deposition [29], and so forth; however, most of these approaches rely on high-end equipment, and some are with serial features, and thus are complex and timeconsuming. With these methods, it is usually difficult to fabricate nanostructures over large areas on certain substrates in need, which, as a result, has a restriction on cellular behaviour investigation.
The mechanism of interaction between tumour cells and polymer nanowires deserves to be further explored. In this work, a substrate-independent facile technique based on reactive ion etching (RIE) was developed to prepare superhydrophilic nanowire forests (NWFs). The feature of NWFs could be precisely defined according to their aspect ratios. On such NWFs, HeLa cells were cultured, and then the influence of the NWFs on filopodia and lamellipodia morphologies, regulation of cell adhesion, and cell growth were systematically studied. It is expected that these systematic observations could serve as a guide for improving nanostructure design and cell capture efficiency.

Nanostructure fabrication
It is known that numerous studies have attempted to generate polymer nanowires on various substrates by using a simple oxygen (O 2 ) plasma etching [30][31][32][33]. However, it should be noted that etching mask layers were usually introduced artificially [30][31][32]. In this work, the carbon-based polymer materials were allowed directly to etch by O 2 plasma in RIE mode without using any artificial self-masking layer. Besides, the employment of argon (Ar) plasma enabled further regulation of the aspect ratio of the NWFs. In our previous work, nanowires with various morphologies have been obtained from polymers after plasma bombardment [34,35]. A variety of factors, including different plasma sources, gas flowing rates, radio frequency power, and etching time were proven to have an impact on the size and morphology of the nanowires. Taking all these factors into consideration, a controllable fabrication method for NWFs was developed.
The process for preparing NWFs with various aspect ratios is depicted in Figure 1. Firstly, a polyimide (PI) layer was spin-coated on a Si substrate (Figure 1(a)). Then, the PI layer, which had a thickness of 5.61 µm, was subjected to RIE (RIE-150, Beijing Zhongke Tailong Electronic Technology Co., Ltd., China) involving O 2 plasma to form NWFs with a high aspect ratio (HNWFs) in Figure 1(b). During the process of plasma etching, the radio frequency power was set at 200 W, the chamber pressure was 5 mTorr, and the flowing rates for O 2 and Ar were 50 and 20 sccm, respectively. Under these conditions, the treatment time of O 2 plasma lasted for 40 min. As shown in Figure 2(a), there are linear relationships between the etching time of O 2 plasma and the height of the NWFs, as well as the pristine PI layer. Only when the etching time was increased to 40 min, the PI layer was completely etched to form nanowires. Due to the recombination of neutral particles and reaction fragments, HNWFs could be polymerised further to NWFs with a low aspect ratio (LNWFs). The relationship between the etching time of Ar plasma and the aspect ratio of the NWFs is shown in Figure 2(b). The aspect ratio of NWFs decreased obviously with the increase of etching time of Ar plasma. Otherwise, Si-NWF composite nanostructures could be obtained when the etching time of Ar plasma was longer than 80 min. In order to acquire LNWFs, the final treatment time of O 2 and Ar plasma was 40 and 80 min, respectively. It should be noted that such a technique is not dependent on the material of substrates. Figure 3 primarily describes the morphological characteristics of the prepared HNWFs and LNWFs using scanning electron microscopy (SEM) images. The HNWFs are relatively higher and denser, while LNWFs are relatively lower and sparser (Figures 3(a) and (c)). It can be clearly observed that the nanowires In order to utilise specific values to further characterise NWFs with different aspect ratios, the average height and diameter of the nanowires were calculated in Figure 4. The size of nanowires in the HNWFs ranging from 50 to 500 nm with an average diameter of 145 nm was observed, and the HNWFs were 4.25 µm in height. Meanwhile, nanowire tips in HNWFs tended to form aggregates or clusters; consequently, the diame-ters of nanowires in the LNWFs were enlarged, ranging from 200 to 1200 nm after treatment of Ar plasma. As shown in this figure, the average diameter of individual nanowires in the LNWFs was 424 nm, and the height was reduced to 3.12 µm. Finally, the aspect ratios of the HNWFs and LNWFs were calculated to be 29:1 and 7:1, respectively.

Surface wettability
As was stated, surface wettability has a considerable impact on cell adhesion. The wettability of a planar Si, the HNWFs and the

FIGURE 4 Diameter distributions of the HNWFs and LNWFs
LNWFs were characterised by measuring static contact angles (CAs) of deionised water through a sessile drop test. As shown in Figure 5, CA of the planar Si is 72.5 • . In contrast, the surfaces with HNWFs and LNWFs are superhydrophilic (CA < 5 • ) because of the highly porous nanostructures.

Cell proliferation and viability
To investigate the biocompatibility of the different NWFs, first of all, the proliferation of HeLa cells (American Type Culture Collection number: CCL-2) on their surfaces was detected  compared with the planar Si and the HNWFs, the OD value of the LNWFs is obviously higher. It is well known that OD values reflect the cytotoxicity, which is usually affected by surface hydrophobicity and nanoscale topography of biomaterials [36]. Previous studies have shown that cells proliferate significantly on hydrophilic surfaces with low aspect ratio nanostructures [25]. In this respect, the surface nanostructures of LNWFs are conducive to promote the growth and proliferation of HeLa cells.
To further assess the viability of HeLa cells on the NWFs, cells adhered to different substrates were stained using a LIVE/DEAD viability/cytotoxicity kit (ThermoFisher, USA). The number of live (depicted as green) and dead cells (depicted as red) per unit area was counted under the observation of a fluorescence microscope. The viabilities of HeLa cells after different incubation time are shown in Figure 7. The cell viability of the HNWFs was lower than planar Si at the initial 4 h. It is supposed that HeLa cells can adhere to tips of the nanowires in HNWFs via the formation of focal adhesions [37,38]. Therefore, the HNWFs with lower mechanical strength hinder the deformation of the cell membrane under the numerous anchoring points. As the incubation time increased, the cell viability on LNWFs maintained at a high level, from 80% @4 h to 93% @24 and 48 h. While that on the HNWFs at 4 h was only 22%, even when the culturing time was prolonged to 48 h, the viability was only lifted to 57%, which was far lower than the viability on the LNWFs surface. In addition, as the planar Si surface could not provide a friendly environment for cell adhesion, the cell viability reduced from an initial 79% @4 h to 44% @48 h. Based on these results, we can say HeLa cells prefer to grow on surfaces of the LNWFs.
As shown in Figure 8, there were fewer dead cells on LNWFs after incubation for 48 h, and the shape of the cells was fusiform. However, a large number of dead cells were observed on the surface of HNWFs, and the cell shapes maintained were relatively spherical. This further suggested that there must be some special reason for cell death, which needs to be further analysed by SEM.

Cell adhesion and spreading
The diverse morphologies of HeLa cells that adhered to various substrates after cultivation for 4, 24, and 48 h are shown in Figure 9. The cells on the planar Si surface exhibited a small number of short filopodia (Figure 9(a)). As shown in Figure 9(b), there was a growing tendency in the number and length of filopodia at 24 h. However, the performance of cell states still was poor in the absence of physical stimulation.
With an increase of the incubation period to 48 h, the cells still maintained a relatively hemispherical shape (Figure 9(c)). This indicates the cells have an inferior growth state because the planar Si surface could not provide a good support for cell adhesion. Figure 9(d) shows the interaction between HeLa cells and the HNWFs at 4 h. Due to the relatively high cell adhesion force, the HNWFs were uprooted from the substrate. As a result, the immobilised cells on the HNWFs restricted themselves to a much smaller area and were not well-cultured on this substrate. When the nanowires were placed in the cell culture medium for 24 h, all the nanowires lied down onto the substrate, which further hindered cell adhesion and migration (Figures 9(e) and (f)). In contrast, the cells on the surface with the LNWFs exhibited a flat and polygonal shape as shown in Figures 9(g) to (i). In the initial stage, the cells had a large number of filopodia stretched out to form mechanical anchors on the rough surfaces ( Figure 9(g-2). The lamellipodia at the edge of the cells gradually began to emerge with the increase of the cultivation time (Figures 9(h-2) and (i-2)). However, the  To further illustrate cell adhesion characteristics, the spreading areas of HeLa cells on different substrates were also investigated, and the results are shown in Figure 10. At incubation time of 48 h, the average spreading areas on substrates of planar Si (130.25 ± 56.26 µm 2 ) and HNWFs (165.12 ± 23.18 µm 2 ) were much smaller than that on LNWFs (242.33 ± 9.57 µm 2 ). The cell shapes on both planar Si and HNWFs were more circular, compared with that on the LNWFs, indicating that the cells on the LNWFs were well spread and freely cultured. Simi-lar to the previously reported results, the nanostructures on the substrate could increase the cellular contact areas and provide more connection sites for cell adhesion [39].

CONCLUSION
Superhydrophilic HNWFs and LNWFs were fabricated using a simple RIE technique. As were explored and compared, the LNWFs promoted the formation of filopodia and lamellipodia. Also, stronger cell adhesion and larger cell spreading areas were more likely to appear on the LNWFs. In addition, due to enhanced mechanical anchoring, the cells markedly proliferated on the LNWFs. Further, the LNWFs on different substrates and their simple fabrication method could be a promising approach for controlling the adhesion of cancer cells, which might be integrated into sensing devices for CTC detections.