Abstract: Objective To observe the influence of the stiffness of three-dimensionally bioprinted extracellular matrix analogue on the differentiation of bone marrow mesenchymal stem cells (BMSCs) into skin appendage cells. Methods (1) Sodium alginate of 1 g and 4 g gelatin, 3 g sodium alginate and 8 g gelatin were mixed respectively, and the two mixtures were dissolved in 100 mL ultra-pure water respectively to prepare two sodium alginate-gelatin composite hydrogels, named 1A4G hydrogel and 3A8G hydrogel, which were used in the subsequent experiments. The morphology of the two hydrogels at room temperature, after condensation for 15-30 min at 4 ℃ (the same condensation condition below), after condensation and cross-linking with 25 g/L calcium chloride solution (the same cross-linking condition below), and after condensation and three-dimensional printing with a three-dimensional bioprinter (the same three-dimensional printer below) and cross-linking were observed respectively. Young′s modulus (stiffness) of the two kinds of hydrogels was measured by Young′s modulus tester after condensation and cross-linking (n=3). Two kinds of hydrogels were cross-linked and freeze-dried, and their pore structure was observed by scanning electron microscope. Two hydrogels were cross-linked and freeze-dried, and the porosity was detected by anhydrous ethanol replacement method (n=3). (2) BMSCs were isolated from femur and tibia of 20 C57BL/6 mice (no limitation with sex, born 7 days) and cultured, and the second passage of cells was used for further test. The BMSCs single cell suspension (1.0×10⁷ /mL) was mixed with 1A4G hydrogel and 3A8G hydrogel respectively at 1∶9 volume ratio to prepare BMSCs-loaded 1A4G hydrogel and BMSCs-loaded 3A8G hydrogel for three-dimensional printing. One construct was printed with 1 mL cell-loaded hydrogel (the same dosage for printing below). Mesenchymal stem cells (MSCs) specific medium was added after cross-linking, and the printed constructs were divided into 1A4G group and 3A8G group according to the hydrogel. One construct of each group cultured for 7 days was tested with live/dead kit to count the live cells and dead cells in 50-fold field of view. Nine printed constructs from each of the two groups were taken, and BMSCs of nine wells (1.0×10⁶ per well) cultured with 2 mL MSCs specific medium were set as two-dimensional culture group. After 1, 3, 5 day (s) of culture, three printed constructs from 1A4G group and 3A8G group respectively and three wells of cells from two-dimensional culture group were taken to detect the absorbance value in culture medium by cell counting kit 8, denoting the cell proliferation activity. (3) BMSCs-loaded 1A4G hydrogel and BMSCs-loaded 3A8G hydrogel of 10 mL respectively were prepared as in experiment (2), which were respectively mixed with 0.5 mL plantar dermis homogenate extracted from 10 C57BL/6 mice of 1 day postnatal with unknown sex, then three-dimensionally printed, cross-linked, cultured with MSCs specific medium for 3 days and then changed to sweat gland specific medium. The printed constructs were divided into 1A4G group and 3A8G group according to their hydrogel. After 7 days of culture with sweat gland specific medium, the expressions of epithelial cell surface markers cytokeratin-5 (CK5) and CK14, sweat gland cell surface markers CK18 and Na⁺ /K⁺ -ATPase (NKA), and hair follicle cell surface markers CK17 and alkaline phosphatase (ALP) at protein level in cells of printed constructs in the two groups were detected by immunofluorescence method. The expressions of CK5, CK14, CK18, NKA (detecting ATP1a1), CK17, and ALP at mRNA level in cells of printed constructs in the two groups were detected with real-time fluorescent quantitative reverse transcription polymerase chain reaction (n=3). Data were statistically analyzed with independent sample t test, Fisher′s exact probability test, analysis of variance for factorial design, and Bonferroni method. Results (1) Compared with that of 3A8G hydrogel, 1A4G hydrogel had lower viscosity and better fluidity at room temperature. Both kinds of hydrogels were gel-like after condensation, based on which, the shape of cross-linked hydrogels was uniform and regular, with three-dimensional printing and cross-linking made hrdrogels forming solid crisscross cylindrical constructs. The Young′s modulus of 1A4G hydrogel was (52±6) kPa, which was obviously lower than (218±5) kPa of 3A8G hydrogel (t=40.470, P<0.01). The pore structure of the two hydrogels was similar, with all the cross-sections showing porous network structure. The porosity of the two hydrogels was similar (t=0.930, P>0.05). (2) The distribution of live/dead cells between 1A4G group and 3A8G group was similar after 7 days of culture (P>0.05), most of which were live cells. The absorbance value in culture medium of printed constructs among 1A4G group, 3A8G group, and two-dimensional culture group didn′t show statistically significant differences after 1, 3, 5 day (s) of culture (P>0.05). Compared with that after 1 day of culture within each group, the absorbance value in culture medium of printed constructs in 1A4G group and 3A8G group was significantly increased after 3 and 5 days of culture (P<0.05 or P<0.01), and the absorbance value in culture medium of cells in two-dimensional culture group was significantly increased after 5 days of culture (P<0.01). Compared with that after 3 days of culture within each group, the absorbance value in culture medium of printed constructs in 1A4G group and 3A8G group and that of cells in two-dimensional culture group was significantly increased after 5 days of culture (P<0.01). (3) After 7 days of culture with sweat gland specific medium, the CK5, CK14, CK18, NKA, CK17, and ALP were positively expressed at protein level in cells of printed constructs in the two groups. After 7 days of culture with sweat gland specific medium, the expressions of CK5, CK14, CK18, and NKA at mRNA level in cells of printed constructs were close between the two groups (t=0.362, 0.807, 0.223, 1.356, P>0.05); the expressions of CK17 and ALP at mRNA level in cells of printed constructs in 3A8G group were 1.96±0.21 and 55.57±11.49, respectively, which were significantly higher than 1.05±0.42 and 2.01±0.27 in 1A4G group (t=3.333, 8.074, P<0.05 or P<0.01). Conclusions BMSCs cultured three-dimensionally in 1A4G and 3A8G hydrogels tend to differentiate into sweat gland cells, but the BMSCs cultured three-dimensionally in 3A8G hydrogel show a stronger tendency to differentiate into hair follicle cells than the cells cultured in 1A4G hydrogel. It suggests that relatively high stiffness of three-dimensionally bioprinted extracellular matrix analogue facilitates not only differentiation of BMSCs into sweat gland cells, but also their differentiation into hair follicle cells.