Targeting a fixed percentage of granulocyte differentiation using experiments designed via nonlinear model predictive control

Previous efforts to control cellular differentiation have been largely experimental. Although some mathematical models for this process exist, rarely has a quantitative approach been employed to design experiments that predictably direct the cell fate. As an initial step towards this aim, a control strategy for sustaining a desired constant level of differentiated human promyelocytic leukemia (HL60) cells is proposed. Nonlinear model predictive control is applied using a model of HL60 cell differentiation which reflects the dominant observable cell states along the known granulocyte/monocyte differentiation pathway. After implementing changes identified by a preliminary experimental trial, the experimental realization of the controller successfully achieved the target differentiation level and demonstrated the need for a quantitative approach to experiment design.