Integrin activation and internalization on soft ECM as a mechanism of induction of stem cell differentiation by ECM elasticity

Abstract

The mechanism by which ECM elasticity induces lineage specification of stem cells has not been clearly understood. Integrins are well-documented mechanosensors that are positioned at the beginning of the sensing pathway. By using an antibody specifically recognizing the active conformation of β1 integrin, we observed that β1 integrin activation in bone marrow mesenchymal stem cells (BMMSCs) was induced by soft substrate to a significantly greater degree than by stiff substrate. In contrast, however, the level of cell surface integrin on soft substrate was significantly lower than that on stiff substrate. Soft substrate markedly enhanced the internalization of integrin, and this internalization was mediated mainly through caveolae/raft-dependent endocytosis. The inhibition of integrin internalization blocked the neural lineage specification of BMMSCs on soft substrate. Furthermore, soft substrate also repressed the bone morphogenetic protein (BMP)/Smad pathway at least partially through integrin-regulated BMP receptor endocytosis. A theoretical analysis based on atomic force microscopy (AFM) data indicated that integrin-ligand complexes are more easily ruptured on soft substrate; this outcome may contribute to the enhancement of integrin internalization on soft substrate. Taken together, our results suggest that ECM elasticity affects integrin activity and trafficking to modulate integrin BMP receptor internalization, thus contributing to stem cell lineage specification.

Fig. 1.

β1 Integrin activation in BMMSCs significantly increases on soft substrate. (A) Activated and total β1 integrin levels in BMMSCs 2 h after seeding on stiff or soft substrate analyzed by Western blotting. GAPDH was used to normalize for equal loading. Gels are representative of five experiments. (B and C) Statistical analysis of results in A. P values are for differences in β1 integrin levels between stiff and soft substrates (mean ± SEM; n = 5). (D) Immunocytochemical staining of activated and total β1 integrin levels in BMMSCs 2 h after seeding on stiff or soft substrate. In photos marked “activated integrin” and “total integrin,” solid arrowheads indicate β1 integrin on the cell surface and open triangles indicate β1 integrin in cytosol. Microphotographs are representative of six experiments. (Scale bar: 20 μm.) (E and F) Fluorescence intensity of β1 integrin quantified using ImageJ software. Statistical analysis of results in D is shown. P values are for differences in β1 integrin levels between stiff and soft substrates (mean ± SEM; n = 6).

Fig. 2.

Substrate stiffness affects integrin subcellular localization in BMMSCs. (A) Flow cytometry dot plots of BMMSCs freshly harvested and analyzed for cell surface expression of activated β1 integrin. Data are representative of three experiments. (B) Statistical analysis of results in A. P values are for differences in activated β1 integrin levels between stiff and soft substrates (mean ± SEM; n = 3). (C) Typical immunocytochemical images of activated β1 integrin expression in BMMSCs cultured on stiff or soft substrate. (Scale bar: 10 μm.) (D and E) Activated β1 integrin (D) and total integrin (E) on the cell surface analyzed by biotin labeling and capture ELISA. P values are for differences in β1 integrin levels between stiff and soft substrates (mean ± SEM; n = 3).

Fig. 3.

β1 integrin internalization is enhanced on soft substrate. Internalization of total β1 integrin (A) and activated β1 integrin (B) in BMMSCs on stiff or soft substrate in the absence or presence of 0.6 μM PMQ (inhibitor of receptor recycling) was determined for the times indicated by biotin labeling and capture ELISA (mean ± SEM; n = 4).

Fig. 4.

Soft substrate enhances integrin internalization via caveolae/raft-dependent endocytosis. (A) BMMSCs on soft substrate were stained with anti-activated β1 integrin and either anti-clathrin or anti-caveolin-1 antibodies, followed by FITC-conjugated anti-mouse and TRITC-conjugated anti-rabbit and anti-goat secondary antibodies. Data are representative of 12 experiments. (Scale bar: 10 μm.) BMMSCs on stiff or soft substrate were pretreated with 200 μM MDC, 10 mM MBCD, or medium alone for 1 h, and internalization of total β1 integrin (B) and activated β1 integrin (C) for 15 min was determined by biotin labeling and capture ELISA. P values are for differences in internalized β1 integrin levels between stiff and soft substrates in each group (mean ± SEM; n = 4). BMMSCs on stiff or soft substrate were pretreated with 10 mM MBCD or medium alone for 1 h, and total β1 integrin (D) and activated β1 integrin (E) on the cell surface were analyzed by biotin labeling and capture ELISA. P values are for differences in β1 integrin levels on the cell surface between stiff and soft substrates in each group or MBCD+ vs. MBCD− on soft substrate (mean ± SEM; n = 3). BMMSCs transfected with CAV-1 siRNA or control RNA were cultured on stiff or soft substrate, and internalization of total β1 integrin (F) and activated β1 integrin (G) for 15 min was determined by biotin labeling and capture ELISA. P values are for differences in internalized β1 integrin levels on the cell surface between stiff and soft substrates in each group (mean ± SEM; n = 3).

Fig. 5.

Caveolin-1–mediated internalization may contribute to BMMSC fate determination via substrate stiffness. (A) BMMSCs were cultured on a plastic six-well plate or soft substrate for 7 d in the presence or absence of 10 mM MBCD, followed by determinations of Nestin, MAP2, and NFL expression by immunocytochemical staining. Data are representative of four experiments. (Scale bar: 30 μm.) (B–D) Fluorescence intensities of Nestin, MAP2, and NFL were quantified using ImageJ software. Statistical analysis of results in A is shown. P values are for differences between MBCD+ vs. MBCD− on soft substrate (mean ± SEM; n = 4).

Fig. 6.

Soft substrate represses the BMP–Smad pathway probably through caveolin-dependent endocytosis of BMPRIA. (A) BMMSCs on soft substrate were stained with anti-activated β1 integrin and anti-BMPRIA antibodies, followed by FITC-conjugated anti-mouse and TRITC-conjugated anti-rabbit secondary antibodies. Data are representative of 8 experiments. (Scale bar: 10 μm.) (B) Phosphorylated and total Smad1, -5, -8 levels in BMMSCs 2 h after seeding on stiff or soft substrate were analyzed by Western blotting. Gels are representative of 6 experiments. (C) Statistical analysis of results in B. P values are for differences in ratios of phosphorylated/total Smad1, -5, -8 between stiff and soft substrates (mean ± SEM; n = 6). (D) BMPRIA on the cell surface was analyzed by biotin labeling and capture ELISA. P values are for differences in BMPRIA levels between stiff and soft substrates (mean ± SEM; n = 3). (E) BMMSCs on stiff or soft substrate were pretreated with 10 mM MBCD or medium alone for 1 h, and internalization of BMPRIA for 15 min was determined by biotin labeling and capture ELISA. P values are for differences in internalized BMPRIA levels between stiff and soft substrates in each group (mean ± SEM; n = 4). (F) BMMSCs transfected with β1 integrin siRNA or control RNA were cultured on soft substrate, and internalization of BMPRIA for 15 min was determined by biotin labeling and capture ELISA. P values are for differences in internalized BMPRIA levels on the cell surface between β1 integrin siRNA- and control RNA-transfected cells (mean ± SEM; n = 3). (G) After biotin labeling of cell surface proteins in BMMSCs on the soft substrate, 25 μg/mL of β1 integrin blocking antibody or isotype control IgG was added, and internalization of BMPRIA for 15 min was determined by biotin labeling and capture ELISA. P values are for differences in internalized BMPRIA levels between β1 integrin blocking antibody and isotype control IgG-treated cells (mean ± SEM; n = 3). (H) Phosphorylated and total Smad1, -5, -8 levels in BMMSCs 2 h after seeding on stiff or soft substrate in the presence of 25 μg/mL β1 integrin blocking antibody or isotype control IgG were analyzed by Western blotting. Gels are representative of 3 experiments. (I) Statistical analysis of results in B. P values are for differences in the ratios of phosphorylated/total Smad1, -5, -8 between stiff and soft substrates in each group (mean ± SEM; n = 3).