Figure 1.
Overall Scheme of Real-Time Adaptive Cell Culture System.
Data are acquired using phase-contrast time-lapse microscopy and sent to a server for image processing where the confluency is calculated and predicted. 4 hours prior to reaching a predefined threshold for confluency (0.5 confluence), an email and text message is sent to alert the user to prepare for subculture. When the cells have achieved the threshold for confluency, another email and text message is sent to remind the user. The cells are subcultured and the total cell number is counted. If the target cell number has been achieved, the experiment is terminated. Otherwise, the cells are replated at low density for further cell expansion and the process is repeated until the target cell number is achieved.
Figure 2.
Calculation and Evaluation of Confluency.
(A) Process of computer-generated confluency. (i) Original image. (ii) Image is inverted during the initial part of the restoration process to remove halo and shade-off artifacts. (iii) After restoration. (iv) Basic thresholding is applied to obtain a confluency mask. (v) The confluency mask is dilated to capture cellular processes that are hard to discern from background. (vi) The computer-generated confluency mask (green) is overlaid on top of the original image. (B) Evaluation of computer-generated confluency versus human-generated confluency. (i) Human-generated cell tracing. (ii) The cell tracing is digitized and filled in to generate a confluency mask. (iii) The computer-generated confluency mask (green) is overlaid on top of the human-generated confluency mask (red) with overlapping regions (yellow, true positive) against the background (black, true negative). Green-only regions (false positive). Red-only regions (false negative). (iv) The computer- (green) and human-generated (red) confluency masks are overlaid on top of the original image with overlapping regions highlighted (yellow).
Figure 3.
Remote Monitoring of Confluency and Predictive Modeling via the Internet.
A screen capture of the graphic user interface from the real-time adaptive cell culture system illustrating the calculated confluency level and predictions (top) and individual fields of view (bottom).
Figure 4.
(A) Human-directed subculture. Confluency prediction was based on the human operator's previous cell culture experience. The graphs show archived time- lapse image data from the human-directed subculture that was processed with the confluency prediction model for the purpose of comparison with the computer-directed subculture. (B) Computer-directed subculture. The predefined threshold for confluency (blue line). The calculated confluency (red line). The predicted confluency (green line). 1 frame is equivalent to 5 minutes. (C) Variance in confluency prediction and actual confluency measurement in computer-directed subculture. Variance (grey line). Trendline (black line).
Figure 5.
4 hours prior to reaching the predefined threshold for confluency, an email and text message is sent to alert the human user to prepare for subculture. Once the threshold for confluency is reached, a reminder email is sent.
Figure 6.
Total Theoretical Cell Yield Achieved from Human- and Computer-directed Subcultures.
(A) Human-directed subculture. (B) Computer-directed subculture. Both human- and computer-directed subcultures required 3 serial passages to achieve a theoretical cell yield of 50 million C2C12 cells without exceeding 0.5 confluency.
Table 1.
Confluence Measurements from Computer-Directed Subcultures.
Table 2.
Average Cell Yield per 35 mm Petri Dish for Human- and Computer-directed Subcultures.
Figure 7.
ALP Staining of Human- and Computer-directed Subcultures.
(A) ALP-stained plates. BMP-2-treated cells stain positive (blue) for the osteogenic marker, ALP. (B) Quantification of ALP intensity (fold difference compared to control). This shows that cells expanded from both human- and computer-directed subcultures were still responsive to BMP-2-induced ALP expression, indicating that cells were differentiating towards an osteoblast fate.
Figure 8.
Myogenin and Pax7 Staining of Human- and Computer-directed Subcultures.
After 96 h in myogenic conditions, cells that have fused and differentiated towards a myocyte fate (elongated cells containing multiple nuclei) stain positive for the myogenic marker Myogenin (green). Undifferentiated cells stain positive for the stemness marker, Pax7 (red). This shows that cells expanded from both human- and computer-directed subcultures were capable of myocyte differentiation while undifferentiated cells retained their stemness marker.