A Novel 96well-formatted Micro-gap Plate Enabling Drug Response Profiling on Primary Tumour Samples

Drug-based treatments are the most widely used interventions for cancer management. Personalized drug response profiling remains inherently challenging with low cell count harvested from tumour sample. We present a 96well-formatted microfluidic plate with built-in micro-gap that preserves up to 99.2% of cells during multiple assay/wash operation and only 9,000 cells needed for a single reagent test (i.e. 1,000 cells per test spot x 3 selected concentration x triplication), enabling drug screening and compatibility with conventional automated workstations. Results with MCF7 and MDA-MB-231 cell lines showed that no statistical significance was found in dose-response between the device and conventional 96-well plate control. Primary tumour samples from breast cancer patients tested in the device also showed good IC50 prediction. With drug screening of primary cancer cells must consider a wide range of scenarios, e.g. suspended/attached cell types and rare/abundant cell availability, the device enables high throughput screening even for suspended cells with low cell count since the signature microfluidic cell-trapping feature ensures cell preservation in a multiple solution exchange protocol.


Dilution procedures for uniform cell seeding in the microfluidic device
A uniform cell seeding should be obtained in this assay since the total amount of cells from each tumour is extremely few. In addition, any shift on the quantity of cells among MGP units could considerably affect the outcome of tumour response assay since cell-cell contact would produce more extra-cellular matrix for higher drug resistance 5,6 . To minimize the variation on cell number between MGP units, a two-step dilution operation was applied in the preparation of cell suspension to MGP. The complete operation of preparation was 1. Measure the density of original cell suspension. 2. Determine the times should the cell suspension be diluted (Ex. N times). 3. Dilute the cell suspension into √N times and then the expected density would be obtained after two repeats. 4. After the dilution, density was measured again to determine the corresponding volume of cell suspension to be introduced.

Estimation of cell loss
Cell loss was minimized by removing solution through surrounding microchannels instead of directly through culture wells. To confirm cells were successfully conserved during the operation, quantity of cells before and after solution exchange was compared. A fluorescent dye (Hoechst, H33342) was used to stain cells for better identification. After cells were seeded overnight, fluorescent dye was then introduced to identify the initial number of cells. The optimal cycles of solution exchange were conducted as stated in following section. After completed solution exchange, images were captured to quantify number of cells in each MGP unit. Difference in quantity of cells was then compared and efficiency of this operation was shown in rate of cells remained.
Rate of cells remained = (Initial number-Final number)/(Initial number) X 100% hours respectively for confirmation of long-term stability of concentration. Fluorescent images were captured by MetaMorph to compare the concentration before and after incubation. Captured images were analyzed by imageJ software for intensity. After the operation stated above, the least repeats of solution exchange was obtained.

Repeats for sufficient solution exchange
Supplementary Figure S3 showed the concentration of fluorescent dye (FITC) in the culture well of a MGP unit after every repeat of solution exchange. According to the results, we need to conduct at least 2 repeats for the sufficient solution exchange on MGP, it will take around 30 seconds for one well unit. In addition, concentration of solution in MGP units was relatively stable during 24 hours of incubation after 2 repeats of solution exchange comparing to 1 repeat of that.

The correlation between manual and auto quantification
The cell quantification in the research was all conducted by a commercialized software, MetaMorph. The software allows cell quantification by setting the minimal and maximal diameter of single stained objects, minimal intensity of objects comparing to local back ground and the minimal and maximal stained area. By applying the auto quantification, quantity of stained cells can be easily and rapidly obtained. In order to apply this tool for quantification of viability assessment in this research, the stained condition should be optimized and strictly followed. On the other hand, while finding the optimal staining condition on the MGP, the optimal settings in MetoMorph were also conducted for accurate cell quantification.
Before applying the quantification method, we should examine the accuracy of auto quantification and also build a correlation between manual count and auto count. The images of fluorescent stained cells were randomly chosen by random number. Three views of each well were chosen to compare the result of manual count and auto count.
Manual count was conducted firstly before auto count. Then, auto quantification was conducted by MetaMorph. The results of manual count would not affect the outcome of auto quantification since the number of cells was concluded after processing the quantification in software. Before processing quantification, one could only change the settings in MetaMorph for optimal cell identification. Once the optimal cell identification was found, the auto quantification would then be processed. So, the two methods of quantification would not affect to each other.
Supplementary Figure S4 showed the result of correlation between auto quantification and manual quantification. The number of counted cells was normalized against that of manual count which could directly find the difference between the two methods.
Result showed that the weighting factor of auto quantification was 1.0322, which further indicated that the method was averagely 3.2% more than manual quantification. In addition, no statistical significance between the two methods was found which proved that auto quantification could provide reliable results of quantification in this research.

Morphology of MCF7 and MDA-MB-231 cells during treatment
The morphology of MCF7 cells during 24 hour cisplatin and docetaxel treatment was shown in Supplementary Figure.

Comparisons between the microfluidic culture and conventional cultures
Comparisons with conventional cultures were presented in Supplementary Table S2. Microscale culture in the MGP is slightly different from macro culture in 96-well plate in volume density, surface area to volume ratio. In addition, the shifts on concentration of solution would not be an issue since the volume of working solution was relatively high than microcultures. According to the results conducted on MGP and 96-well control, the response profile of cells treated by different anticancer drugs showed slightly different with each other since the culture environment was different.
However, by applying statistical analysis, the results showed no statistical significance between the platforms which proves that the microculture platform is a viable tool for anticancer drug assay.