Five substrates were selected from our library and fabricated according to our previous protocol (Wang et al., 2016b) (Figure 1A). Briefly, cSAPs #1 and #2 was composed of SiO₂ with 5 μm diameter and polystyrene (PS) with 200 or 400 nm diameter, #3, #4, and #5 was composed of SiO₂ with 2 μm diameter and PS with 65 nm diameter, carboxy-PS (PSC) with 50 or 100 nm diameter, and the cSAPs were fabricated within the 24-well tissue culture plates (TCPS, Falcon). TCPS was the flat control group in this study.

Scanning electron microscopy (SEM, ZEISS SUPRA ® 55, Carl Zeiss, Germany) and water contact angle (WCA) was used to characterize the surface structures and wettability of these cSAPs using an automated contact angle measurement device (PT-705B, Precise Test, China) at room temperature. Other detailed characterizations have been done previously (Wang et al., 2016b).

H1 hESCs and human induced pluripotent stem cells (hiPSCs) were routinely cultured in Matrigel (BD, United State) coated 60 mm Petri dishes with mTeSR1 medium (Stem Cells, United State), and passaged using ReleSR (Stem Cells, United State) with a ratio of 1:3 every other day. E14 mESCs and mouse induced pluripotent stem cells (miPSCs) were routinely maintained on gelatin (Yeasen, China) coated 6-well tissue culture plates in mouse ESC medium (DMEM-knockout, GIBCO, United State) containing 20% FBS, 1% NEAA, 1% l-Glutamine, 1% penicillin, and streptomycin (P/S, GIBCO, United State), 0.15 mM 1-thioglycerol (Sigma, United State), 1000 U/mL leukemia inhibitory factor (Millipore), and two chemical inhibitors (2i), 3 μM CHIR99021 and 1 μM PD0325901 (Stemgent, United State). mESCs and miPSCs were passaged using trypsin-EDTA (Invitrogen, United State) with a ratio of 1:6 every other day. All the medium was changed daily.

All cells used in this study were passaged three times on cSAPs prepared in 24-well tissue culture plate or control surfaces coated by Matrigel or Gelatin for further research. The experimental group was named cSAP #X (X = 1, 2, 3, 4, 5) and TCPS (as a flat control). Cell inoculatiion density of mESCs and miPSCs in experimental group was 2  ×  10⁴ cells/cm². Cell density of hESCs and hiPSCs was adjusted according to the colony quantity. Cell culture in experimental group was same to the routinely cultured cells, mESCs and miPSCs was passaged at a ratio of 1:6, hESCs and hiPSCs was passaged at a ratio of 1:3. Medium was changed every day. Cell morphology was captured every day by an inverted microscope (Olympus-IX71).

Total RNA was extracted from cells with the MiniBEST universal RAN extraction Kit (Takara, Japan). cDNA was made with 1 mg of total RNA based on the manufacturer’s protocol using the PrimeScript^TM RT Master Mix (Takara, Japan), and subsequent real-time qPCR was carried out in triplicates, using the 2 × RealStar Green Mixture (GENE STAR, China) on Light Cycler 96 System cording to manufacturer’s instructions (Roche, United State). Amplifications were performed using the following conditions: 95°Cfor 2 min, followed by 40 cycles of 95°C for 15 s, 60°C for 15 s, and 72°C for 30 s. Gene expression was normalized to Gapdh, and then the TCPS control. The sequences of all primers used were listed in Supplementary Table S1.

hESCs were co-cultured with mouse stromal cell line AGM-S3 to induce their differentiation into hematopoietic cells, as reported before (Chen et al., 2017). hESCs were cut into small squares of 1.5 × 10³ cells, and then plated onto the irradiated AGM-S3 cells and cultured in the hPSC maintenance medium [high glucose Dulbecco’s modified Eagle’s medium, DMEM, F-12 nutrient mixture, 20% knockout serum replacement (KSR), 1% l-glutamine, 1% non-essential amino acid solution (NEAA; Gibco), and 5 ng/ml basic FGF (b-FGF; Wako)] for 3 days (Chen et al., 2017). The medium was changed to the hematopoiesis-inducing medium (Iscove’s Modified Dulbecco’s Medium (IMDM) containing 10% fetal bovine serum (FBS; Hyclone), 1% NEAA (Gibco), 60 ng/ml ascorbic acid (Sigma), and 20 ng/ml vascular endothelial growth factor (VEGF; Wako)), referred as day 0 (D0) (Chen et al., 2017; Zhou et al., 2019). The culture medium was replaced every day. The co-cultures were dissociated by 0.05–0.25% trypsin-EDTA (Invitrogen) at D8 or D14 for further analysis. This experiment was run at least three times in parallel.

Flow cytometry experiments were performed according to a previous report (Chen et al., 2017). Briefly, the co-cultured cells at D8 or D14 were dissociated with 0.25% trypsin-EDTA solution. The cell suspension was obtained by filtration through a 70-μm nylon mesh. Before immunostaining, cells were blocked by rabbit serum at 4°C for 30 min. Cells were stained with the anti-CD34/CD43 antibodies (at D8 and D14), and anti-CD34/CD45 or anti-GPA/CD71 antibodies (at D14) at 4°C for 30 min (Supplementary Table S2). Flow cytometric analysis was performed using the FACS Canto II system (BD Biosciences). All FACS data were analyzed using FlowJo 10 software.

According to the previous studies, the hematopoietic colony-forming assay of co-cultured cells was performed (Chen et al., 2017; Sun et al., 2020). Briefly, co-cultured cells were dissociated by 0.25% trypsin/EDTA into single cells at day 14.5 × 10⁴ cells were suspended in 80% MethoCult H4230 (Stem cells) containing 100 ng/ml stem cell factor (SCF), 100 ng/ml interleukin-6 (IL-6), 10 ng/ml interleukin-3 (IL-3), 10 ng/ml Fms-related tyrosine kinase three ligand (FL), 10 ng/ml thrombopoietin (TPO), 10 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF), and 4 units/ml erythropoietin (EPO), and then incubated in 5% CO₂ at 37°C for another 14 days. The number of colony-forming unit erythroid (CFU-E) colonies was counted at 7–10 days, while the number of burst-forming unit-erythroid (BFU-E), colony-forming unit-mixed (CFU-Mix), and colony-forming unit-granulocyte/macrophage (CFU-GM) colonies were counted at 12–14 days. This experiment was run at least three times in parallel.

H1 hESCs cultured on cSAPs and flat control (TCPS) for two passages (4 days/passage) were dissociated with 0.5 mM EDTA for RNAseq. Total RNA of each 1 × 10⁶ cells was extracted by 1 ml TRIzol (Life Technologies) and purified according to the standard instruction. RNA Sequence analysis was performed by BGI Tech (Shenzhen, China). The analysis of gene function was performed with the multi-group data mining system of Dr. Tom (http://report.bgi.com). Gene changes on cSAP #5 compare to TCPS over 2-fold were defined as significant.

Cells were cultured in multi-well plates were fixed with 4% paraformaldehyde (PFA) for 30 min at 4°C, and washed three times with wash buffer (5% FBS in PBS). They were then incubated with membrane permeation reagent (PBS containing 0.3% Triton-100 and 5% FBS) at 4°C for 30 min, stained overnight at 4°C with anti-OCT4/SOX2/SSEA4 antibodies (mouse anti-human), washed three times with PBS containing 5% skim milk, and then incubated for 30 min at room temperature with secondary antibodies (FITC-conjugated secondary Ab, goat anti-mouse) (Supplementary Table S2). Nuclei were labeled with DAPI. After washing three times with PBS, the sample was imaged under a fluorescence microscope.

All data are presented as means ± SD; statistical analyses were performed by GraphPad Prism 8 using the Student’s t-test, one-way ANOVA with Dunnett post hoc test, and two-way ANOVA with Tukey’s post hoc test. p < 0.05 was considered significant.

cSAPs were fabricated by mixing different particles together and depositing them on the tissue culture plates (TCPS); after evaporation, particles were distributed on the surface according to the principle of self-assembly (Figure 1A). The surface topography of cSAPs was measured by SEM and showed that the large particles and small particles were orderly distributed on the surfaces (Figure 1B). Due to the differences of the surface on chemestry and topography, surface wettability showed that water contact angle (WCA) of cSAP #1 and #2 (SiO₂ = 5 μm; PS = 200 and 400 nm) was more hydrophobic (85.8 ± 3.5 and 96.8 ± 4.6 degree), than the cSAP #3, #4, and #5 (SiO2 = 2 μm; PS = 65 nm, PSC = 50 and 100 nm), and the WCA of cSAP #3, #4, and #5 was 31.8 ± 1.1, 32.3 ± 3.2, and 25.8 ± 2.4 degree, respectively (Figure 1C).

The colony morphology of miPSCs on cSAPs was significantly different from that on the TCPS control (Figure 2A). miPSCs colonies on cSAPs were more 3D-like, especially on the cSAP #1 and #2. The PSC colonies were dome-like morphology on the cSAP #4 and #5, neither 3D spheroids nor 2D-like colonies. miPSCs cultured on cSAPs without LIF (Leukemia Inhibitory Factor) had a higher percentage of Oct4-GFP positive cells after 7 days than TCPS without and even with LIF, indicating that cSAPs could maintain the pluripotency of PSCs (Figure 2B and Supplementary Figure S1A). Percentage of Oct4-GFP positive cells was more than 90% on cSAPs, except the cSAP #3, which was similar to the cells cultured within LIF (i.e., ∼80%).

Gene expression analysis showed that pluripotent markers (i.e., Oct4, Nanog, and Sox2) of miPSCs were similar (i.e., Oct4 and Nanog, fold changes < 1.5) or higher (i.e., Sox2, fold changes > 1.5) on the cSAPs compared to the TCPS control without LIF (Figure 2C). Gene expression of miPSCs was also analyzed under LIF conditions. The results showed that Sox2 expression was significantly higher on cSAPs than TCPS, while Oct4 and Nanog were similar between surfaces (Figure 2D). The high expression of Sox2 may indicate neural stem and progenitor cells in some cSAPs groups (Ellis et al., 2004). The effects of cSAPs on miPSCs and mESCs were different because three genes of mESCs expressed similarity between surfaces (fold changes < 1.5, Figure 2C and Supplementary Figure S1B).

Gene expression of mesoderm markers of miPSCs was higher on cSAPs than the TCPS control without LIF (Supplementary Figure S1C). On average, the expression of mT, mSnail2, and mFoxa2 on cSAPs, except cSAP #5, was significantly higher than the TCPS control (fold changes > 2, Supplementary Figure S1C).

hiPSCs and hESCs also showed different colony morphology compared with TCPS and slight differences between cSAPs, similar to miPSCs on cSAPs showing a more 3D-like colony morphology (Figures 3A,B). For hESCs can not grow well without Matrigel, all surfaces used in culturing hESCs were precoated with Matrigel. FACS analysis (Figure 4 and Supplementary Figure S2) and the hematopoietic colony form assay (Figure 5) showed that only H1 hESCs cultured on cSAP#4 and #5 had better hematopoietic differentiation efficiency than the TCPS control when co-cultured with AGM-S3, so only #4 and #5 were chosen for further analysis of hESCs pluripothency. Besides, immunostaining of pluripotent markers, i.e., OCT4, SOX2, and SSEA4, showed that passaged hESCs were in high pluripotency on cSAPs, i.e., cSAP #4 and cSAP #5, and TCPS (Figure 3C).

Cell number of hESC-HCs from cSAP-derived hESCs was higher than the TCPS control after 8 and 14 days (Figure 4 and Supplementary Figure S2). cSAPs, especially cSAP #5, strongly increased the cell populations of CD34⁺CD43⁻ (∼2-folds), CD34⁻CD43⁺ (∼4-folds), and CD34⁺CD43⁺ cells (∼5-folds; hematopoietic progenitors) at D8 (Figure 4B and Supplementary Figure S2A). Also, cell populations of CD34⁺CD43⁻ (∼1.5-folds), CD34⁻CD43⁺ (∼1.5-folds), and CD34⁺CD43⁺ (∼3-folds), CD34⁺CD45⁻ (∼2-folds), CD34⁻CD45⁺ (∼5-folds on cSAP #5), CD34⁺CD45⁺ (∼3-folds), and CD71 + GPA + cells (∼3-folds; erythroid progenitors) at D14 (Figure 4C and Supplementary Figure S2B).

The morphology of hESC-HCs on cSAPs was better compared to the TCPS control (Figure 5A). The colony assay proved that the hematopoiesis potentials of hESCs were significantly enhanced on cSAPs, especially #5. The numbers of CFU-GM, CFU-E, BFU-E, and CFU-MIX generated from cSAP-derived hESC-HCs were approx. 3-fold, 2-fold, 2-fold, and 7.5-fold higher than the TCPS control. Colony number represents the hematopoiesis potentials of H1 hESCs derived from different culture conditions to produce granulocyte/macrophage progenitors (CFU-GM), erythroid progenitors (CFU-E and BFU-E), and hematopoietic progenitors (CFU-MIX) after D14 of co-cultures (Figure 5B). The erythroid cells were confirmed by May-Grunwald-Giemsa (MGG) staining (Figure 5C). After co-culture differentiation and hematopoietic colony form culture, the absolute numbers of four kinds of colonies generated by H1 hESCs cultured on cSAPs#4 and #5 increased significantly compared with TCPS, especially cSAPs#5. Percentage of cell subtypes showed that H1 hESCs cultured on cSAPs# 5 were prone to generate the erythro-myeloid progenitor (EMP) in hematopoietic differentiation. Therefore, the cSAP #5-derived hESCs had significantly higher hematopoietic potentials to produce hematopoietic progenitors (i.e., CD34⁺CD45⁺ cells) and erythroid progenitors (i.e., GPA + CD71⁺ cells) compared to the TCPS control (Figure 5D ).

RNA-seq was used to analyze the change of cellular transcriptional profiling of hESCs after culturing on cSAP #5 (Figure 4A). Results showed that 300 genes (e.g., EGR3, MT1M, EGR1, DDX60, FOS, and MX1) were up-regulated significantly (32% of total differential expression genes), while the 627 genes (e.g., NPPB, GALR1, SIX2, FOXG1, NBL1, RASGRF2, TNNT2, FGF8, NOTUM, TEK, TF, CXCL14, TNC, and CAV1) were down-regulated significantly (68% of total differential expression genes) on cSAP #5 compared to the TCPS control (Figures 6A,B).

According to RNAseq analysis, the genes regulated on cSAPs were enriched in mineral absorption, longevity regulating, Toll-like receptor signaling, HIF-1a signaling, Notch signaling that were up-regulated, and the focal adhesion, TGF-β signaling, PI3K-Akt signaling, and MAPK signaling that were down-regulated (Figure 6B). According to the RNA-seq results we found that expression of CAV1 and CXCL14 was down-regulated (Figure 6B), and was consistent with qPCR verification (Figure 6C). CXCL14 is found concentrated in the focal adhesions, the down-regulated expression of CXCL14 (∼2.79 fold) on cSAP #5 resulted in lower expression of PTK2 (∼1.61 fold) and VCL (∼2.83 fold) (Figure 6C). The dot-plot also showed focal adhesion was enriched in down-regulated DEGs (Figure 6B). CAV1 can function as a scaffolding protein contribute to the activation of the MAP kinase pathway (Mineo et al., 1996; Engelman et al., 1998). In this study, the down-regulated MAPK pathway in Figure 6B may be the result of lower CAV1 (∼3.90 fold) expression on cSAP #5 (Figure 6C). Lower expression of CAV1 can promote hematopoietic differentiation of hESCs when co-cultured with stromal cells under hematopoietic induction condition (Choi et al., 2012). Therefore, cSAP-cultured hESCs having higher blood cell differentiation ability could be a consequence of down-regulation of focal adhesion and MAPK signaling (Figure 6D).