In this study, a perfusion-based microfluidic system was designed to enhance the production of differentiated hepatocytes from MSCs. The design showed substantial advantage in promoting hepatogenesis of MSCs. Our design has two major advantages over other similar devices which use negative pressure to seal the open cover, such as the hybrid microfluidic–vacuum system [54, 55]. First, the use of LDW techniques reduced the fabrication time of our microfluidic system from days to hours compared with the conventional micro-electromechanical techniques [48]. Also, the fabrication of a large culture chamber is made easy with the help of LDW techniques. Second, unlike other previous designs [54, 55], cells are not pressed against the microfluidic device at the edges of culture chamber during assembly due to a close match between the cell culture region and the culture chamber in our system; this design prevents cell damage during assembly. In addition, the microfluidic device can be reversibly bonded and unbonded to the culture substrate, which is useful for further immunocytochemistry or manipulation of cell colonies. For example, unwanted colonies can be removed easily without affecting the healthy ones during maintenance. This device may also be suitable for the differentiation of ESCs or induced pluripotent stem cells (iPSCs).

Besides a larger cell culture surface area, a geometry that allows uniform flow distribution within the culture chamber is also a featured characteristic which confers several advantages for hepatic differentiation of MSCs. The culture chamber of our microfluidic device (culture area ≥ 400 mm2) was significantly larger than that previously reported (culture area = 7 mm2) [20]. Incorporating this device into an automated system, scale-up production for cell therapy is possible. Moreover, the flow field distribution of the culture medium inside the chamber is more uniform because of the quadrilateral geometry of the culture chamber, because the flow field of a rectangular chamber creates a more uniform flow (Fig. 1c, d). On the contrary, uneven flow occurs in a circular chamber (Additional file 1: Figure S5). Hence, a more homogeneous flow distribution within the culture chamber can be achieved in our device, which offers a favorable environment for hepatic differentiation of MSCs.

The microfluidic system, which mimics the shear flow microenvironment, theoretically confers advantages over the static culture system. It supplies a continuous flow of fresh medium and facilitates constant removal of metabolic wastes, which provides a constant microenvironment to maintain cell viability and function over a longer culture period. On the contrary, there are inherent limitations in the traditional in-vitro static culture systems, such as nutrient supply and metabolic end-product accumulation, including release of toxins that negatively influences the cell growth [56]. The utilization of our bio-micofluidic device with a favorable environment efficiently facilitates and accelerates hepatic differentiation of MSCs. In addition, a previous study has demonstrated that the microenvironment is essential for the stemness properties of stem cells [57]. The cellular microenvironments of MSCs in the microfluidic device and the culture dish were similar, and so was the cell viability of MSCs.

The microenvironment in the culture chamber is the key for efficient hepatic differentiation of MSCs in this study. The continuous flow not only constantly supplies nutrients for hepatic differentiation but also removes the unhealthy cells that loosely attached onto the culture chamber. As a consequence, hepatocyte-like cells with homogeneous and favorable functional activities are obtained. A previous study indicates that removal of cell-secreted factors suppresses cell growth and differentiation [58]. However, this action did not seem to suppress cell growth and differentiation in our study. This might be due to the use of a two-step hepatic differentiation protocol which offers necessary growth factors to support hepatogenesis [9]. These factors can be maintained at a relatively stable and higher concentration in the microfluidic system compared with the static culture. A higher expression of hepatic marker genes in the microfluidic device group may be attributable to the mechanical microenvironment as well as more homogeneous cell distribution in the device.

Shear stress plays a crucial role in regulating proliferation, differentiation, and cellular morphology of MSCs [5961]. The shear stress generated in our microfluidic device was much lower (0.00142 Dyne/cm2) than those previously reported to regulate the proliferation and osteogenic differentiation of MSCs (0.3–2.7 Dyne/cm2) [5961]. It did not affect the lineage commitment of MSCs because there were no significant difference between the microfluidic system and static culture in surface immunophenotype after 3 days of maintenance (Fig. 4). This means that shear stress was not a key factor for stemness properties of MSCs in our study. Moreover, a previous study indicates that low shear stress could further stimulate maturation signals through sensory systems on polarized hepatic cells [62]. Indeed, higher expression of hepatic marker genes was observed in the microfluidic device and the mechanical signals might have contributed to functional maturation of hepatocyte-like cells.

Based on our results, MSCs in the microfluidic device after 3 weeks of hepatic induction expressed a high level of AFP and HNF4. Because AFP is specifically expressed in hepatocyte progenitors, it is therefore one of the indicators for hepatic differentiation [63, 64]. The high level of AFP after 3 weeks of hepatic differentiation means that use of the microfluidic device can accelerate the hepatic differentiation. The results indicated that both hepatocyte progenitors (AFP high expression) and matured hepatocytes (HNF4 high expression) in the microfluidic device are greater than those in the culture dish, due to the optimal differentiation conditions. Since previous studies have indicated that hepatic progenitor cells (or fetal liver cells) possess greater regeneration potential than fully matured hepatocytes in experimental liver failure models [49, 65, 66], the cells differentiated in the microfluidic device should be suitable for cell therapy compared with those generated from 2D static culture. The hepatic differentiation from MSCs in the microfluidic device is therefore a more efficient method than that in the culture dish.