A novel magnet-based scratch method for standardisation of wound-healing assays

A novel magnetic scratch method achieves repeatability, reproducibility and geometric control greater than pipette scratch assays and closely approximating the precision of cell exclusion assays while inducing the cell injury inherently necessary for wound healing assays. The magnetic scratch is affordable, easily implemented and standardisable and thus may contribute toward better comparability of data generated in different studies and laboratories.

www.nature.com/scientificreports www.nature.com/scientificreports/ variables I) standard deviation (SD) and coefficient of variation (CV) of the gap width measured at ten points along the gap, II); straightness of the gap margins (ratio of the measured length of the gap margins to the length of an ideal straight line) and III) gap closure rate (slope of the regression of gap area in mm 2 depending on time) (Suppl. Figs 1 and 2).
Scientific discovery is an iterative process and requires reproducible data, which allow multiple data sets to be combined to generate knowledge. Therefore, the need for scientific community-wide implementation of standardized assays to ensure that the data produced has high intra-study consistency and can be replicated and compared successfully across multiple laboratories is increasingly recognized. However, this can only be achieved with cost-effective, platform-independent, easy-to-use techniques, which can be implemented in any laboratory without technical prerequisites, such as the novel magnetic scratch or cell exclusion assays.
Gap width standard deviation varied significantly (p < 0.001) between methods (Table 1, Suppl. Table 1, Suppl. Fig. 3A). The insert (SD 13.27), followed closely by the magnet (SD 23.69), had the lowest standard deviation of the gap width, and the manual methods had the highest (50 g > 150 g, SD 58.5 resp. 35.05). The coefficient of variation of the gap width was smallest for the magnet with 1.54%, followed by the cell culture insert with 2.9%, the pipette 150 g with 4.39%, and the pipette 50 g with 8.04% (Fig. 3A, Suppl. Fig. 3B). Correspondingly, straightness of the gap margins varied significantly (p < 0.001) between the cell exclusion insert and the other three methods as well as between the magnet and the other three methods (Fig. 3B 3C). Cell exclusion inserts led to the straightest margins (mean deviation from the ideal straight line = 6.9%, CV = 3.61%), the magnet was intermediate (mean deviation from the ideal straight line = 13.2%, CV = 4.06%), and the pipette methods (50 g > 150 g) showed the largest difference (mean deviation from the ideal straight line = 23.9% resp. 21.1%, CV = 4.89% resp. 4.77%) between the measured and an ideal margin (of a perfect straight line) ( Table 1, Suppl. Table 1).
These results highlight the significantly higher geometrical control and precision of the cell culture insert and the magnet in comparison with the manual techniques.
Two cell types (chondrocytes and tenocytes) were included in the validation study of the magnetic scratch to determine whether cell type would, due to differences in cell-cell adhesion, polarity or cytoskeletal organisation, influence gap linearity, gap width SD and CV and gap closure rate. In contrast to chondrocytes, tenocytes have long cell processes and adherens and gap junction-based cell-to-cell contacts forming complex interconnected networks and cell sheets 12,13 . The gap junctions are essential for cell-cell communication and represent a functional network allowing coordination of mechanical and synthetic activity 12,13 . However, cell type showed a significant influence only on gap closure rate (p < 0.001) but not on any indicator of gap geometry, accuracy or precision.
Gap closure rate also differed significantly (p < 0.001) among methods (Fig. 3C, Table 1, Suppl. Table 2, Suppl. Fig. 3D) and was significantly slower (p < 0.001) with the cell exclusion than the injury inducing scratch methods for both cell types as well as overall. This is consistent with previous studies showing differences in cell migration and signalling between cell exclusion and cell injury wound healing assays.
In this study, Fluoresceindiacetat (FDA, C-7521, Sigma-Aldrich) -Propidiumiodid (PI, P4170, Sigma-Aldrich) fluorescent staining confirmed the accumulation of dead cells alongside the scratch gap border for all techniques www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ except the cell culture insert (Fig. 4). The difference in the number of dead cells between all three injury-inducing techniques (magnet, pipette 150 g pipette 50 g) and the cell culture insert was statistically significant (p < 0.001, Suppl. Table 3) at all three time points of analysis (0 h, 8 h and 24 h after wounding) independent of cell type (chondrocytes versus tenocytes, Suppl. Table 4, Suppl. Fig. 4). While the injury free release of contact inhibition is sufficient to induce cell migration in cell exclusion assays 2,6,9,14 , the injury inflicted upon the cells in scratch wound healing models with the associated disruption of cell adhesions and release of intracellular contents and signalling molecules has been shown to alter the microenvironment in the cell culture and to provide a critical input required for cell-cell coordination in the spreading of the wounded monolayer 1,3,4,6,7,9 . Thus, while cell exclusion assays provide repeatability and geometric control, they fail to induce an injury and the insert used in this study correspondingly led to slower gap closure kinetics compared to the three assays that inflicted injury upon the cell layer.
Gap closure rate typically shows a nonlinear decline with decreasing gap size 15 , which we could confirm for all techniques used in this study (Suppl. Fig. 3D). For small gaps this limits the time period in which linear closure rates allow comparative quantification of cell migration to 48 hours. As the scratch assay is predominantly used to study wound healing and cancer cell migration and for corresponding drug discovery screening 7 , the creation of a larger standardized gap, as introduced in this study (mean gap width 1539 μm (magnet) versus 457.6 μm (cell culture insert), Table 1), provides a longer observation period to study, e.g., the tumour cell aggressiveness or the efficacy and cytotoxicity of the screened molecules and compounds.
As damage to the plate surface incurred during scratching may affect migration rates 1,8 , the surface relief of the culture dish was measured using atomic force microscopy (Suppl. Fig. 5). The unscratched surface was characterized by topographical variation within 40 nm in height. Pipette tip scratches with 150-gram pressure increased topographical variation to 160 nm, and the magnetic scratch method to 250 nm, both with grooves parallel to the orientation of the scratch. However, this level of variation in surface topography appears to be no impediment to cell migration since the gap closure rates of the three different injury-inducing scratch methods were comparable and faster than the cell exclusion assay.
In conclusion, the novel magnetic scratch method combines operational ease with precision and geometric control, induces the cell injury inherently necessary for a wound healing assay and provides excellent repeatability, reproducibility and easy standardization across different observers (Table 2). Thus, the proposed method may www.nature.com/scientificreports www.nature.com/scientificreports/ contribute to the generation of data, which are better comparable between different studies and laboratories, and may lead to more efficacy and quality control in research.
Methods production and handling of the magnetic scratching tool. The novel magnetic scratching tool can be handcrafted easily by using standard cell culture well-plates and commercially available neodymium magnets. It can be tailored for different types of well-plates and used to scratch all wells of a plate simultaneously. First, the lids of two well-plates were glued together in a 90° angle along their long sides (Fig. 1) with a commercially available standard super glue, such that one lid serves as bottom plate and one as guiding rail. Then magnetic spheres (neodymium magnets, diameter 5 mm+/−0,1 mm, type K-05-C, Supermagnete, Germany) were glued to the lid serving as bottom plate, which should match the well-plate to be scratched and have the size and shape of the wells engraved as template to facilitate correct placement. The magnetic spheres were glued centred to the farright end of each engraved well with vertically oriented magnetic force (Fig. 1) while using additional magnetic spheres as counter-parts to enable magnetic alignment.
The magnets used as scratching tools were disc-shaped (Supermagnete, S-1.5-0.5-N) with a diameter of 1,5 mm and height of 0,5 mm (+/−0,1 mm tolerance). Our pilot experiments determined four stacked discs to produce the optimal weight and magnetic force to create homogenous scratches over the full length of the wells. Prior to use, the discs were sterilized in 70% ethanol for 10 minutes followed by two washing steps with PBS for 5 minutes each. To position the scraping magnets, the cell culture well-plate was placed on top of the bottom plate offset to the right, such that the left rims of the wells overlapped the right rims of the engraved well shapes with the magnetic spheres on the bottom plate. The scratching magnets (stack of 4 discs) were dropped into the wells above the magnetic spheres using plastic tweezers. Due to the magnetic attraction they fell into place at the left rim of the wells directed toward the magnetic sphere on the bottom plate. When all magnets were in place the well-plate was shifted to the left, using the rail plate to facilitate creation of straight scratches, until it superimposed the bottom plate. Finally, the magnetic discs were removed from all wells using a sterile spherical magnet attached to a metal bar (Suppl. Video).

Validation of the novel scratching tool.
To validate the newly developed scratching tool, it was compared to the current gold standard (manual scratching with a pipette tip) and a cell culture insert for physical exclusion of cells in a defined area (80209, Ibidi TM ). The validation study was carried out in 12-well cell culture plates and performed with equine chondrocytes and tenocytes (3 biological replicates each) by two independent operators to study repeatability and reproducibility of the different scratch assay methods.
The cells had been obtained from three horses, which were euthanized for reasons unrelated to this study. Tissue collection to obtain these cells had been performed according to the "Good Scientific Practice and Ethics in Science and Research" regulation implemented at the University of Veterinary Medicine Vienna. The animal owner's consent to collect and analyse the samples and to publish resulting data was obtained according to the standard procedure and approved by the ethics and animal welfare committee of the University of Veterinary Medicine Vienna. Cells were cultured at routine cell culture conditions using Dulbecco's modified eagle medium (DMEM, LONBE12-707F, Lonza) supplemented with 10% fetal bovine serum (FBS, F7524, Sigma-Aldrich), 1%  www.nature.com/scientificreports www.nature.com/scientificreports/ L-glutamine (K0302, Biochrom), 1% penicillin-streptomycin (P4333-100ML, Sigma Aldrich), and 1% amphotericin (A2612, Biochrom) prior to seeding them onto 12 well-plates for the scratch assays in passage 4-5. Medium was changed twice weekly.
For the three scratching tools (magnet, pipette 50 g and pipette 100 g) 100.000 cells were seeded per well. 24 hours later the confluent cell layers were washed twice with PBS and the wounding of the cell layer was performed: For manual scratching a 1250 µl pipette tip was used applying two different pressures (50 g and 150 g). Application of the correct pressure was monitored by performing the scratching procedure on a precision lab scale. To achieve straight scratches a ruler was used as guiding device.
Scratching with the magnet device was performed as described above. Linear cell culture inserts (80209, Ibidi TM ) for physical exclusion of cells at a defined distance (500 µm +/−100 µm) were used according to the manufacturer´s instructions. In brief, the inserts were placed in the centre of each well. 15.000-20.000 cells were seeded into each of the two insert chambers to achieve confluence the following day. To create the gap, the insert was removed using a sterile forceps.
Independent of the scratching technique the cells were washed twice with PBS following gap creation prior to adding new culture medium.
comparison of the cell free gaps (scratches). The cell free gaps were imaged in phase contrast using the EVOS FL Auto imaging system with a 4x fluorite objective (ThermoFisher Scientific, AMEP4680). To ensure that pictures were always taken at the same position, the "reuse settings" function was applied, which allows resuming of previously used coordinates. Pictures were taken at 0, 4, 8, 12 hours and then every 12 hours until full gap closure (Suppl. Fig. 2). The size of the gaps was measured at all time points using the MRI Wound healing tool (http://dev.mri.cnrs.fr/projects/imagej-macros/wiki/Wound_Healing_Tool) in ImageJ (https://imagej.nih.gov/ ij/, version 2.0.0-rc-43/1.50e). The micrographs taken at 0 hours were additionally analysed for straightness of the wound margins and for homogeneity of the gap width (Suppl. Fig. 1).
Homogeneity of the gap width. Gap homogeneity was evaluated by the mean, standard deviation (SD), and coefficient of variation (CV = SD/mean) of 10 measurements of the gap width performed at equal distances of 128 pixels (Suppl. Fig. 1) using ImageJ (version 2.0.0-rc-43/1.50e, Straight Hand Line Tool, NIH, USA) to measure the distance between the upper and lower gap margins.
Straightness of the gap margins. The margins of the gaps were evaluated relative to an ideal (straight) line, parallel to the horizontal axis of the picture, with a length of 1280 pixel (Suppl. Fig. 1 To distinguish between the overall occurrence of dead cells in the culture and cells, which died in direct consequence to wounding, three regions of interest (ROI) were defined along each scratch margin (lower and upper margin) for each micrograph (Fig. 4): (1) ROI (A) extends 25 µm from the scratch margin into the cell free area of the scratch; (2) ROI (B) extends 25 µm from the scratch margin into the remaining cell monolayer and is the region of main interest as it contains cells which died in direct consequence to scratching; and (3) ROI (C) extends from 25 µm to 50 µm into the remaining cell monolayer. The number of dead cells per ROI was counted manually and the dead cells per ROI of the lower (A1, B1, C1) and upper (A2, B2, C2) scratch margins were added to a total count per ROI. To interpret the influence of the wounding method on cell death and calculate the true amount of "scratch" related cell death per picture (relative number of dead cells in ROI(B)) the following formula was used + − + . Negative values in the raw data resulting from a higher number of dead cells in ROI(C) than ROI(B) were set to zero.
AfM. In order to assess potential damage to the plate surface due to scratching, the surface reliefs of wells scratched with the magnet scratch method and the pipette with 150-gram pressure were measured using atomic force microscopy and compared to a non-scratched well and a positive control (scratch performed using a #3 www.nature.com/scientificreports www.nature.com/scientificreports/ scalpel handle with 50 g pressure). Scratching was performed in wells containing PBS. After scratching, the wells were washed twice with ultrapure water to remove any precipitates of the PBS. The surface morphology was evaluated by Atomic Force Microscopy, using a Nanowizard 4 system from JPK (JPK Instruments AG, Berlin, Germany). The AC mode was used, with PPP-NCHAuD cantilevers (330 kHz nominal resonance frequency, 42 N/m force constant and 125microns length). Scan areas of 8 × 100 µm 2 were imaged at 26 × 256 pixel resolution with the scanning direction perpendicular to the scratch direction. Consecutive scans were performed to achieve a scanned area with a final dimension of 8 × 2000 µm 2 . Three areas per scratch method were imaged. Images were analysed with JPK software and 3D plot using the 3D surface plot in ImageJ.
Statistical analysis. Target variables are: straightness of the gap margins, defined as the length of the edge of the gap divided by the length of an ideal straight line; mean and standard deviation of the gap width measured at ten points along the gap; and the gap area in mm 2 . Gap area was measured at successive time-points and the gap closure rate (slope of the regression of gap area in mm 2 depending on time) was recorded at the period of linear closure between 24 hours (after the initial lag period) and 48 hours (before the gap closed in any of the samples). Descriptive statistics, means, standard deviations and coefficients of variation were calculated for the target variables (straightness of the gap margins, homogeneity of the gap width, gap closure rate). With all these target variables, unifactorial ANOVAs were calculated separately with method as factor for each cell type. Both cell types were also combined in analyses, where only for the target variable "slope" cell-type was entered as an additional factor, since growth of chondrocytes and tenocytes differs. We additionally performed more complicated multifactorial ANOVAs with factors: method, operator, biological replicate, and method*operator, for chondrocytes and tenocytes separately. Both cell types were also combined in analyses, where cell-type was entered as an additional factor. With the more complicated ANOVA models, qualitatively and quantitatively similar results were obtained. ethical approval and informed consent. Tissue collection to obtain the cells used in this study has been performed according to the "Good Scientific Practice and Ethics in Science and Research" regulation implemented at the University of Veterinary Medicine Vienna. The animal owner's consent to collect and analyse the samples and to publish resulting data was obtained according to the standard procedure and approved by the ethics and animal welfare committee of the University of Veterinary Medicine Vienna.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.