ZO-1 boosts the in vitro self-renewal of pre-haematopoietic stem cells from OCT4-reprogrammed human hair follicle mesenchymal stem cells through cytoskeleton remodeling

Abstract

Background

The challenge of expanding haematopoietic stem/progenitor cells (HSPCs) in vitro has limited their clinical application. Human hair follicle mesenchymal stem cells (hHFMSCs) can be reprogrammed to generate intermediate stem cells by transducing OCT4 (hHFMSCs^OCT4) and pre-inducing with FLT3LG/SCF, and differentiated into erythrocytes. These intermediate cells exhibit gene expression patterns similar to pre-HSCs, making them promising for artificial haematopoiesis. However, further investigation is required to elucidate the in vitro proliferation ability and mechanism underlying the self-renewal of pre-HSCs derived from hHFMSCs.

Methods

hHFMSCs^OCT4 were pre-treated with FLT3LG and SCF cytokines, followed by characterization and isolation of the floating cell subsets for erythroid differentiation through stimulation with hematopoietic cytokines and nutritional factors. Cell adhesion was assessed through disassociation and adhesion assays. OCT4 expression levels were measured using immunofluorescence staining, RT-qPCR, and Western blotting. RNA sequencing and Gene Ontology (GO) enrichment analysis were then conducted to identify proliferation-related biological processes. Proliferative capacity was evaluated through CCK-8, colony formation assays, Ki67 index, and cell cycle analysis. Cytoskeleton was observed through Wright‒Giemsa, Coomassie brilliant blue, and phalloidin staining. Expression of adherens junction (AJ) core members was confirmed through RT‒qPCR, Western blotting, and immunofluorescence staining before and after ZO-1 knockdown. A regulatory network was constructed to determine relationships among cytoskeleton, proliferation, and the AJ pathway. Student’s t tests (GraphPad Prism 8.0.2) were used for group comparisons. The results were considered significant at P < 0.05.

Results

Pre-treatment of hHFMSCs^OCT4 with FLT3LG and SCF leads to the emergence of floating cell subsets exhibiting small, globoid morphology, suspended above adherent cells, forming colonies, and displaying minimal expression of CD45. Excessive OCT4 expression weakens adhesion in floating hHFMSCs^OCT4. Floating cells moderately enhanced proliferation and undergo cytoskeleton remodelling, with increased contraction and aggregation of F-actin near the nucleus. The upregulation of ZO-1 could impact the expressions of F-actin, E-cadherin, and β-catenin genes, as well as the nuclear positioning of β-catenin, leading to variations in the cytoskeleton and cell cycle. Finally, a regulatory network revealed that the AJ pathway cored with ZO-1 critically bridges cytoskeletal remodelling and haematopoiesis-related proliferation in a β-catenin-dependent manner.

Conclusions

ZO-1 improved the self-renewal of pre-HSCs from OCT4-overexpressing hHFMSCs by remodeling the cytoskeleton via the ZO-1-regulated AJ pathway, suggesting floating hHFMSCs^OCT4 as the promising seed cells for artificial hematopoiesis.

Introduction

The transplantation of artificial blood cells derived from reprogrammed somatic cells is a promising technique for treating various haematologic diseases and refractory anaemia. However, the challenge of expanding haematopoietic stem/progenitor cells (HSPCs) in vitro has limited the clinical application of this technology. Recently, mouse or human functional HSPCs were shown to be effectively expanded in vitro for at least one month using a specially designed 3a-medium [1, 2]. This discovery offers valuable methods and culture systems for supporting the expansion of artificial hematopoietic stem cells (HSCs). However, exploring additional viable strategies is imperative. One such strategy worth investigating is the identification of intermediate stem cells that exhibit pluripotency between induced pluripotent stem cells (iPSCs) and HSPCs. This process imitates embryonic haematopoiesis, during which hematopoietic endothelial cells (HECs) generate a subset of pre-HSCs with dual angiogenic and hematopoietic capabilities, prior to the development of the mature HSCs [3, 4]. Ideally, these intermediates can be maintained in a self-renewal medium and subsequently induced to differentiate into specific cell types through the application of haematopoietic cytokines.

In our earlier research, we discovered that human hair follicle mesenchymal stem cells (hHFMSCs) derived from the bulge and papilla of hair follicles can be induced into iPSCs by the introduction of four pluripotent factors (Oct4, Sox2, C-Myc, and Klf4) [5]. Additionally, we found that introducing only OCT4 can reprogram hHFMSCs into intermediate cells (hHFMSCs^OCT4), and these reprogrammed cells can be used to generate mature erythrocytes directly by stimulating them with specific cytokines [6]. Furthermore, we observed an intriguing phenomenon in which pre-stimulation with low concentrations of FLT3LG and SCF cytokines resulted in the generation of suspended cell subsets capable of in vitro passage and expansion, as well as delaying differentiation, and gene expression profiles analogous to those observed in pre-HSCs were also identified [7]. Upon sequential stimulation by a cascade of hematopoietic factors, the suspended cells underwent prompt differentiation into erythroid cells [6]. As a result, this process eliminates the need for sorting HSCs and maintaining their status. Significantly, a unique subset of suspended cells displaying distinct cellular morphology characteristics was identified, characterized by a small volume, round shape, and low adhesion, and subsequently designated as floating cells derived from hHFMSC^OCT4 (floating hHFMSCs^OCT4) [6, 7]. Nevertheless, the correlation between alterations in morphology and adhesion of reprogrammed cells and their capacity for self-renewal remained ambiguous.

OCT4, also known as the POU domain transcription factor or POU5F1, plays a crucial role in maintaining the self-renewal and pluripotency of embryonic stem cells (ESCs) [8, 9]. Additionally, studies have shown that OCT4 not only promotes the G1/S transition but also enhances the self-renewal of hHFMSCs by inhibiting the gene expression of p21 [10, 11]. In a previous study, we found that OCT4-induced reprogramming of hHFMSCs resulted in changes in the expression of cytoskeleton- and adhesion-related genes, leading to the loss of hair follicle and skin development potential. Interestingly, the reprogrammed hHFMSCs showed upregulation of pluripotent and haematopoietic genes, with enrichment of Gene Ontology (GO) terms related to haematopoietic lineage differentiation [7]. Based on these findings, we hypothesize that cytoskeleton genes and adhesion molecules may play a role in regulating the dedifferentiation and self-renewal of floating hHFMSCs^OCT4.

Prior research showed that ZO-1(also known as TJP1) is mainly located on adherens junctions (AJs) in nonepithelial cells [12]. Our previous studies found a notable rise in ZO-1 expression in floating hHFMSCs^OCT4. Additionally, the expression of the cytoskeleton gene ACTN2 and the AJ molecule E-cadherin significantly changed [7]. Currently, AJs are believed to regulate cell proliferation via the nuclear transport of β-catenin [13]. This study will examine how cytoskeletal remodelling and AJ signal transduction coordinate in the self-renewal of floating hHFMSCs^OCT4. To explore this mechanism, floating cell subsets will be isolated from adherent cell subsets and subjected to cellular and molecular biology experiments, as well as transcriptome sequencing. The anticipated outcomes of this research are expected to offer both experimental and theoretical support for the in vitro hematopoietic application of novel seed cells with expansion capabilities.

Materials and methods

Cell isolation and culture

hHFMSCs were isolated and identified as described in our previous work [5, 6]. hHFMSCs were cultured in H-DMEM/F12 (Gibco) supplemented with 10% FBS (Gibco), 10 ng/ml bFGF (Acro Biosystems), and 100 U/ml penicillin‒streptomycin (Solarbio). The hHFMSCs were frozen in a cryopreservation solution composed of 50% H-DMEM/F12 medium, 40% FBS, and 10% dimethyl sulfoxide and stored in liquid nitrogen at passages 0–2. The cells were thawed and expanded for experimentation at passages 3–5. hHFMSCs, floating hHFMSCs^OCT4, and adherent hHFMSCs^OCT4 were maintained in hHFMSCs medium on Matrigel (Corning)-coated culture plates.

Lentivirus transduction and pre-induced with haematopoietic cytokines

hHFMSCs were transduced with the lentiviral vector pLV-EF1-OCT4-IRES-EGFP as previously described [6]. In brief, hHFMSCs seeded onto Matrigel-coated plates were infected with lentivirus expressing OCT4 in the presence of polybrene and expanded for 7 days in hHFMSC medium. The transduced hHFMSCs (hHFMSCs^OCT4) were cultured in pre-induced medium with 100 ng/ml FLT3LG, 100 ng/ml SCF, 10% FBS, and 100 U/ml penicillin-streptomycin for at least 14 days. The medium was then switched to hHFMSC medium for continued expansion. After 12 days of expansion, floating cells were separated from the adherent cells, and the plate was rinsed twice with PBS to remove any remaining floating cells, as verified by microscopy. The floating cells were then centrifuged at 800 rpm for 5 min, then seeded on Matrigel-coated plates. Once 70-80% confluent, the cells were detached with 0.25% trypsin-EDTA and passaged for further culture.

Haematopoietic differentiation and erythroid precursors expansion

Further hematopoietic differentiation was conducted by culturing floating hHFMSC^OCT4 in hematopoietic media supplemented with cytokines. The cells were plated in 0.5% methylcellulose on ultra-low adhesion plates at a density of 1 × 10⁴ cells /well. The floating cells were then cultured in hematopoiesis medium (StemSpan SFEM Serum-Free Medium from Stem cell technologies) supplemented with 0.5% methylcellulose, 10% knockout serum (Gibco), and various cytokines including BMP4, VEGF, bFGF, SCF, Flt3, IL3, IL6, G-CSF, IGF-II, EPO, and TPO (R&D Systems), and 100U/mL penicillin-streptomycin for 14 days [4]. The haematopoietic precursors derived from suspension cells were subsequently treated with TrypLE (Gibco) and cultured in erythroid cell expansion medium (StemSpan SFEM Serum-Free Medium), supplemented with 0.5% methylcellulose, 10% knockout serum, 100U/mL penicillin-streptomycin, 3 U/mL EPO, 100 ng/mL SCF, 20 ng/mL IL3, and various nutritional factors known to support erythrocyte maturation, such as myoinositol (Sigma-Aldrich), folic acid (Sigma-Aldrich), vitamin B12 (Sigma-Aldrich), monothioglycerol (Sigma-Aldrich), for a period of 7 days. Half-medium was changed every two days.

Dissociation assay

The floating subset was subcultured and expanded without trypsin, using only PBS rinsing to achieve a single-cell suspension, indicating low cell-to-cell adhesion in this subpopulation. A cell dissociation assay was performed to observe cell‒cell adhesion in both subsets, as previously described [7]. Cells were seeded on Matrigel-coated 24-well plates at 2 × 10⁵ cells/well and cultured overnight to confluence. After treatment with dispase II (2.4 U/mL) and 0.1% crystal violet staining, the percentage of single cells was calculated. Each sample was tested in triplicate, with three counts per well.

Cell adhesion assay

As culture density increased, floating cells grew in suspension above the layers, moving with the shaking of the plate, which indicated decreased cell-substrate adhesion. A cell adhesion assay was conducted to assess adhesion to the extracellular matrix, following previously established methods [7]. The cells were plated in 24-well plates at 2 × 10⁵ cells/well and cultured until hHFMSCs adhered to the plates and fully expanded. The plates were washed with digestive enzyme-free PBS, and the remaining cells were stained with 0.1% crystal violet. The percentage of remaining cells relative to the total number of cells was proportional to the cell-extracellular matrix adhesion. Each sample was tested three times, and three counts were taken per well.

Immunofluorescence (IF)

To assess the hematopoietic potential of floating subsets derived from pre-induced hHFMSCs^OCT4, IF was conducted to detect the expression of CD45 protein in viable colonies, which emerged after 12 days for expansion of pre-induced hHFMSCs^OCT4. The culture medium of colonies was replaced with PBS, and fixed with 4% paraformaldehyde (Solarbio) for 3 min, followed by rinsing in PBS buffer for 3 times. Colonies were then blocked with 10% FBS for 1 h at room temperature and incubated with mouse anti-human PE-CD45 (1:100 dilution, BD Pharmingen) or PBS (as a negative control) for 1 h at 37℃ in dark. Colonies were then washed with PBS buffer for 3 times. Alexa Fluor-555 goat anti-mouse IgG (1:200 dilution, Cell Signaling Technology) was used as secondary antibody and colonies were visualized with a fluorescence microscope (Olympus). Peripheral blood cells were used as a positive control, and nuclei were counterstained with 5 µg/mL DAPI (Solarbio) for 2 min in the dark.

We examined the factors behind floating subset suspension and self-renewal regulation by assessing OCT4 expression and β-catenin nuclear localization using IF. Cells were seeded on glass slides in 24-well plates at 1 × 10⁴ cells/well, fixed with 4% paraformaldehyde for 20 min, and permeabilized with 0.1% Triton X-100 for 20 min at room temperature. Nonspecific binding was blocked with 1% BSA/PBS. Cells were incubated overnight at 4°C with primary antibodies: mouse anti-human OCT3/4 (ready-to-use, Maxim), mouse anti-human β-catenin (1:150, Maxim), or PBS (control). Alexa Fluor 594-conjugated goat anti-mouse IgG (1:200) was then added and incubated for 1 h at room temperature. Nuclei were stained with 5 µg/ml DAPI for 5 min in the dark. Fluorescence microscopy was used for visualization, and ImageJ software measured the average immunofluorescence intensity for each cell group.

RT‒qPCR

RT-qPCR was employed to analyze the mRNA expression levels of the genes OCT4, ZO-1, ACTN2, E-Cadherin, β-catenin, and Cyclin D1 in two distinct subsets. Total RNA was isolated using TRIzol Reagent (Sparkjade, China). cDNA was synthesized with the Prime Script RT reagent Kit (+ gDNA Eraser) and then subjected to qPCR using TB Green^® Premix Ex Taq™ II (TaKaRa). The gene mRNA levels were determined using 50 ng of cDNA on an Applied Biosystems 7500 real-time PCR system, and the samples were normalized to GAPDH with the autoset baseline. The relative expression was calculated as 2^−ΔΔCt. Each gene was detected in three replicates for different cell groups. The primer sequences are provided in Table 1.

Table 1 Primers for RT‒qPCR

Western blot analysis

Western blot analysis was used to examine OCT4, ZO-1, and β-catenin protein levels in two subsets. Cells were harvested, lysed in ice-cold RIPA buffer (Solarbio)with 1% PMSF(Solarbio), and centrifuged at 12,000×g for 30 min at 4 °C to remove debris. Protein concentration was measured using a BCA Protein Assay Kit (Solarbio). Equal amounts of proteins were subjected to sodium dodecyl sulphate‒polyacrylamide gel electrophoresis (SDS‒PAGE), after which the proteins were transferred to polyvinylidene fluoride membranes (Millipore). The membrane was incubated with 5% non-fat milk and then incubated with primary antibodies at 4 °C overnight. Then, the membrane was incubated with an HRP-conjugated secondary antibody at room temperature for 2 h. The membranes were finally stained with an enhanced chemiluminescence (ECL) Western blot system (Millipore) (n = 1). The optical density values were measured three times at varying exposure times. Table 2 lists the optimal antibody dilutions for Western blot analysis.

Table 2 Optimal dilutions of antibodies used for Western blot/IF analysis

Wright-Giemsa staining

We observed cell morphology of haematopoietic precursors derived from floating hHFMSCs^OCT4 using Wright‒Giemsa staining. Approximately 1,000 ~ 2,000 cells were washed twice in cold PBS with 2% FBS and diluted in 400𝜇L of cold PBS plus 1% FBS. Samples were loaded into wells of Thermo Scientific Cytospin 4 (Thermo Fisher Scientific) and spun at 800 rpm for 5 min. Slides were fixed with either methanol/acetic acid (v/v 3:1) for 3 min or precooled acetone for 10 min and dried for 30 min. They were stained with Wright-Giemsa dye for 5 min, soaked in PBS for 10 min, and quickly washed in distilled water.

The cell morphology and ratio of nuclear to cytoplasm of two subsets were observed using Wright‒Giemsa staining. Cells were seeded on glass slides in 24-well plates at 1 × 10⁴ cells/well and fixed with methanol for 10 min. The cells were stained with Wright-Giemsa dye for 20 min, soaked in PBS for 10 min and quickly washed in distilled water. Each cell group was stained in triplicate. Photos were taken in three random fields of vision and then analysed with Image J software. The data was expressed as mean ± standard deviation, and the comparisons between the two groups were performed with independent sample t-tests.

Coomassie brilliant blue staining

The arrangement of cytoskeletal microfilaments was observed suing Coomassie brilliant blue staining. Cells were seeded on glass slides in 24-well plates at 1 × 10⁴ cells/well and treated with 0.1% Triton X-100 for 20 min. Then, the cells were fixed with 3% glutaraldehyde for 15 min and stained with 0.2% Coomassie Brilliant Blue R250 (Beyotime) for 40 min. Each cell group was stained in triplicate.

Phalloidin staining

Actin molecules make up cytoskeletal microfilaments, which have various shapes, with F-actin being a common type. The expression and localization of F-actin were analysed using fluorescent dye-labelled phalloidin, a specific ligand for F-actin. Cells were seeded on glass slides in 24-well plates at 1 × 10⁴ cells/well and fixed in 4% paraformaldehyde for 20 min. Then, cells were treated with 0.1% Triton X-100 for 30 min, followed by a 20-minute incubation with Alexa Fluor^® 555 Phalloidin (Beyotime). Nuclei were counterstained with 5 µg/ml DAPI for 5 min in the dark at room temperature. Each cell group was stained in triplicate.

Cell counting Kit-8 (CCK-8) assay

The CCK-8 assay was employed to assess the cell proliferation activity of two distinct subsets. Cells were seeded in Matrigel-coated 96-well plates at a density of 1000 cells/well and monitored for 7 days at 37 °C in a 5% CO₂ atmosphere with one change of fresh medium on day 4. CCK-8 working solution was added to each well, and the cells were incubated for 2 h. The absorbance at 450 nm was measured. Each cell group was counted thrice, and the numerical data were presented as means ± standard deviations.

Colony formation and intercellular distance assay

The self-renewal ability of two cell subsets was evaluated using a 2D colony formation assay. Cells were seeded in Matrigel-coated 6-well plates at 20 cells/cm² and grown for 5–7 days. They were then washed with PBS, fixed with 4% paraformaldehyde for 15 min, and stained with 0.1% crystal violet for 30 min. Colonies were categorized as large (> 50 cells) or small (20–50 cells) and counted. Clone formation rate was calculated as (stained clones/seeded cells) × 100%. Each group was tested three times, with data shown as mean ± standard deviation. An independent sample t-test compared the two groups.

To study cell-to-matrix and cell-to-cell interactions within colonies, cell-cell distances were measured. Three colonies were randomly selected from each cell group, and three visual fields were chosen per clone. The nearest distance between cytoplasmic edges in each field was measured and analyzed using GraphPad Prism (8.0.2).

Immunocytochemistry (ICC)

ICC staining of Ki67 was employed to assess the cell proliferation in two distinct subpopulations. Cells were seeded on glass slides in 24-well plates at 1 × 10⁴ cells/well, fixed with 4% paraformaldehyde (Solarbio) for 20 min, and subsequently permeabilized with 0.1% Triton X-100 (Solarbio) for 20 min at room temperature. ICC staining was performed using a standard immunoperoxidase staining procedure (mouse anti-human Ki67 antibody, 1:150 Maxim) and PBS (as a negative control) at 4°C overnight. Haematoxylin was used as a counterstain. Each cell group was stained in triplicate.

DAPI intensity and cell cycle analysis

Cells were seeded on glass slides in 24-well plates at 1 × 10⁴ cells/well, fixed with 4% paraformaldehyde for 20 min, treated with 0.1% Triton X-100 for 30 min, and stained with 5 µg/ml DAPI for 5 min in the dark. Fluorescence microscopy was then used to capture images of the stained nuclei. Cell cycle staging data processing followed literature methods with slight modifications [14]. We quantified DNA content in 35 images, comprising 10 hHFMSCs, 13 adherent hHFMSCs^OCT4, and 12 floating hHFMSCs^OCT4, to obtain cell cycle profiles for the G1, S, and G2/M phases. Post-ZO-1 knockdown cell cycle staging was assessed in 10 images, including 5 negative control (NC) and 5 si-ZO-1 treated samples. The CellProfiler 2.1 pipeline was employed to segment DAPI-stained nuclei, exclude border cells, and measure single-cell DAPI intensity. An R 3.1.0 script reads this data, exports a histogram of combined DAPI strength per group, and calculates cell cycle stage percentages, saving them in a .txt file. GraphPad Prism 8.0.2 then generates a cell cycle percentage plot.

RNA sequencing and GO enrichment analysis

To explore the potential role of differences in adhesion in regulating self-renewal, we used transcriptome sequencing to analyse gene expression differences in floating cell subsets versus adherent cells. The expression profiles of mRNAs from hHFMSCs, adherent hHFMSCs^OCT4, and floating hHFMSCs^OCT4 were determined by next-generation sequencing (NGS) as previously described [5]. The transcriptome sequencing and analysis were conducted by OE biotech Co., Ltd. (Shanghai, China). Raw data (raw reads) were processed using Trimmomatic. The reads containing ploy-N and the low quality reads were removed to obtain the clean reads. Then the clean reads were mapped to reference genome using hisat2. FPKM and read counts value of each transcript (protein_coding) was calculated using bowtie2 and eXpress. differentially expressed genes (DEGs) were identified using the DESeq (2012) functions estimateSizeFactors and nbinomTest. P value < 0.05 and fold Change (FC) > 2 or FC < 0.5 was set as the threshold for significantly differential expression. Hierarchical cluster analysis of DEGs was performed to explore transcripts expression pattern. GO enrichment and KEGG pathway enrichment analysis of DEGs were respectively performed using R based on the hypergeometric distribution. GO enrichment analysis of the DEGs was carried out via functional annotation in the DAVID database (https://david.ncifcrf.gov/), and the P value and false discovery rate (FDR) were ultimately obtained. The enrichment scores can be calculated by:

$$Enrichment{\text{ }}scores = {\raise0.7ex\hbox{${\frac{m}{n}}$} \!\mathord{\left/{\vphantom {{\frac{m}{n}} {\frac{M}{N}}}}\right.\kern-\nulldelimiterspace}\!\lower0.7ex\hbox{${\frac{M}{N}}$}}$$

Where, N is the number of genes with GO annotation among all genes. n is the number of genes with GO annotation among differentially expressed genes in N; M is the number of genes annotated with a particular GO term among all genes; m is the number of differentially expressed genes annotated as a particular GO term.

Significant genes were visualized by the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database (http://string-db.org/), and the network was constructed using Cytoscape software (https://cytoscape.org/). The filtered clean reads were mapped to the reference genome database (accession number: GCF_000001405.38) by Hisat2, v2.2.1.0 (http://ccb.jhu.edu/software/hisat2/index.shtml). For each group of cells, three independent biological replicates were sequenced.

RNA interference (RNAi)

To verify the role of ZO-1 in the expression of AJ core protein members, we subsequently knocked down the ZO-1 gene in floating hHFMSCs^OCT4 by RNAi technology. The siRNA sequences used for ZO-1 in our experiments were designed and generated by RiboBio. Cells were seeded on glass slides in 24-well plates at a density of 1 × 10⁵-5 × 10⁵ cells/well and treated with ribo FECT™ CP compound (RiboBio) at 37 °C in a 5% CO₂ atmosphere when the cell density reached 30-50%. Floating hHFMSCs^OCT4 were transfected with siRNA or NC for 24–96 h to determine the optimal silencing conditions. The expression of Cy3 was observed via fluorescence microscopy. The knockdown of ZO-1 in floating cells was replicated twice. The plasmid vector information used for siRNA transfection is provided in Table 3.

Table 3 The plasmid vector information for siRNA transfection

Statistical analysis

All the numerical data from the disassociation assay, cell adhesion assay, RT‒qPCR, Western blot, IF, and ICC are presented as the means ± standard deviations. Comparisons between two groups were performed with Student’s t tests (GraphPad Prism 8.0.2). The results were considered significant at P < 0.05. DEGs from RNA sequencing were identified using the DESeq (2012) functions estimateSizeFactors and nbinomTest. P value < 0.05 and fold Change (FC) > 2 or FC < 0.5 was set as the threshold for significantly differential expression.

Results

Emergence of floating cell subsets from OCT4-transduced hHFMSCs

The OCT4 gene was introduced into hHFMSCs as described previously [6]. To enhance the hematopoietic capacity of hHFMSCs^OCT4, we conducted a pre-induction experiment before erythroid differentiation. hHFMSCs^OCT4 were stimulated with FLT3LG and SCF, followed by expansion in hHFMSC medium (Fig. 1a). Extended OCT4 transduction and pre-induction time led to the emergence of a new subset of floating cells (floating hHFMSCs^OCT4), characterized by their small size, round shape, and ability to grow in suspension. FLT3LG and SCF significantly increased the quantity of these floating cells (Fig. 1b).

Fig. 1

FLT3LG and SCF caused floating cells to appear from hHFMSCs^OCT4. (a) Schema showing the transformation of floating cells derived from hHFMSCs^OCT4 into erythroblasts. (b) Cell morphology changes and floating cells observed under immunofluorescence microscope, with a floating cell indicated by a yellow arrow and a floating cell colony outlined by a dashed circle. (c) Phase contrast microscope used to observe floating cells’ morphology. Floating cell indicated by yellow arrow. (d) Collecting and re-cultivating suspended cells. (e and f). Floating cell colonies were selected to identify CD45 expression, with green indicating EGFR, red indicating PE-CD45, and blue indicating Hoechst. Peripheral blood cells served as a positive control

Floating cell morphology with small volume, short pseudopodia, and spheroid shape was observed under phase contrast microscope (Fig. 1c). Colonies were formed during expansion of pre-induced cells on Matrigel-coated plates (Fig. 1d). CD45⁺ colony was identified by picking and re-culturing on Matrigel-coated plates, with one colony out of 32 containing CD45⁺ cells (Fig. 1e, f).

Floating cells differentiation into mature erythrocytes

Due to the limited number of CD45⁺ cells, we isolated all identifiable floating cells based on their shape and adhesion properties, without specifically sorting for CD45 positivity. These cells underwent erythroid differentiation when exposed to haematopoietic cytokines. After 7–14 days of stimulation, the floating cells became rounder and produced burst forming unit-erythroid (BFU-E) and colony forming unit-erythrocyte (CFU-E). The remaining adherent cells maintained irregular polygon or fusiform shapes and did not form haematopoietic-like clusters (Fig. 2a).

Fig. 2

Floating hHFMSCs^OCT4 differentiated into erythrocytes. (a) Erythroblast differentiation and cell morphology change in response to hematopoietic cytokines. Expansion D7 refers to erythroid cells expanded for 7 days in erythroid cell medium. (b) Erythroblast growth and maturation in erythroid expansion medium with nutrients. Green arrow: promyelocyte and mesmyelocyte. Orange arrow: metarubricyte. Red arrows: enucleated erythrocyte. (c) Erythrocytes were enucleated in a culture medium without cytokines and the enucleation efficiency was measured

BFU-E and CFU-E were cultured in erythrocyte expansion medium and enriched through attached culture overnight. Primitive erythroblasts (green arrows) and orthochromatic erythroblasts (orange arrows) were observed under Wright-Giemsa staining, with evident cytoplasmic hemoglobinization in orthochromatic erythroblasts. Some erythrocytes denucleated to form reticulocytes (red arrows) (Fig. 2b). After 3 days of enucleation culture, the enucleation rate reached 85% (Fig. 2c).

Significant upregulation of OCT4 resulted in the emergence of floating subsets

Floating hHFMSCs^OCT4 were separated and their adhesion properties were tested. Dissociation assays showed that a higher percentage of single cells came from floating hHFMSCs^OCT4 (87%) compared to adherent hHFMSCs^OCT4 (48%) after treatment with dispase II. Both floating hHFMSCs^OCT4 and adherent hHFMSCs^OCT4 had higher percentages of single cells than hHFMSCs (11%) (Fig. S1a), indicating reduced cell-cell adhesion in the floating subsets. Subsequently, cell-matrix adhesion was assessed via adhesion assays. The percentage of cells remaining attached to Matrigel after rinsing with PBS was lower for floating hHFMSCs^OCT4 (9.2%) than for adherent hHFMSCs^OCT4 (32.5%), and both percentages were lower than that for hHFMSCs (75.8%) (Fig. S1b).

To determine why floating hHFMSCs^OCT4 were suspended, we analyzed OCT4 expression levels in both groups. IF staining showed that OCT4 was mainly found in the nuclei of transduced cells, with a higher percentage of positive cells in the floating group (91%) compared to the adherent group (37%) (Fig. 3a and b). RT-qPCR revealed a 14.166-fold increase in OCT4 mRNA in floating hHFMSCs^OCT4 compared to adherent cells (Fig. 3c). Western blot analysis (Fig. 3d) also showed higher OCT4 protein levels in the floating group. These findings suggest that the heightened overexpression of OCT4 may weaken adhesion and contribute to the appearance of floating hHFMSCs^OCT4 after pre-induction with FLT3LG and SCF.

Fig. 3

The difference in OCT4 overexpression between adherent hHFMSCs^OCT4 and floating hHFMSCs^OCT4. (a) IF analysis of OCT4 expression and localization in hHFMSCs, adherent hHFMSCs^OCT4, and floating hHFMSCs^OCT4. NC, negative control. Red represents OCT4, blue represents DAPI, and green represents EGFP, (n = 3). (b) and (c) Histograms were generated to validate the expression of OCT4 by RT‒qPCR (n = 3) and (d) Western blots (n = 1), respectively. Full-length blots/gels are presented in Supplementary Figure S3. **P˂0.01, ***P˂0.001, ****P˂0.0001

Proliferative capacity of floating hHFMSCs^OCT4 moderately enhanced

Higher OCT4 gene expression is believed to lead to reprogrammed cells that can proliferate well [15]. RNA-seq was used to study the role of cell adhesion in OCT4’s regulation of self-renewal. First, principal components were analyzed to compare sample similarity within groups and diversity between groups. The results showed that cells within each group were closer together, indicating better repetitiveness, while distances between groups were significantly greater (Fig. S2a). This suggests that OCT4 induced hHFMSCs to generate new cell populations, with cells of different adhesion potentially having different gene transcripts.

Next, DAVID database was used to cluster upregulated genes based on their GO terms, and several biological processes were identified in the floating hHFMSCs^OCT4, including re-entry into the mitotic cell cycle, cell proliferation, regulation of cell growth, positive regulation of the cell cycle process, and cell proliferation of the inner cell mass. Unexpectedly, the genes associated with cell cycle arrest, negative regulation of the mitotic cell cycle, and negative regulation of the cell cycle G1/S phase transition were upregulated (Fig. S2b). Furthermore, when comparing adherent hHFMSCs^OCT4 with hHFMSCs, we found the enriched biological process GO terms of the upregulated genes were primarily linked to the promotion of cell proliferation. These terms included regulation of the mitotic cell cycle, regulation of cell growth, regulation of the cell cycle G1/S phase transition, positive regulation of the cell cycle process, and regulation of stem cell proliferation (Fig. S2b).

We assessed cell growth curves of two subgroups using a CCK-8 kit. The results indicated that hHFMSCs had the lowest growth, adherent hHFMSCs^OCT4 exhibited the highest proliferation, while floating hHFMSCs^OCT4 showed moderate growth, particularly between days 3 and 7 (Fig. 4a). Colony formation efficiency was assessed, showing large colony (≥ 50 cells) formation rates of 19.79%, 14.76%, and 5.21% for adherent hHFMSCs^OCT4, floating hHFMSCs^OCT4, and hHFMSCs, respectively. Small colony (< 50 cells) rates were 10.4%, 5.2%, and 1.6% for the same groups (Fig. 4b). It indicates that large and small colonies of adherent cells formed faster than floating cells, possibly due to their growth pattern on Matrigel. However, floating cells showed a higher colony formation rate compared to hHFMSCs, suggesting improved proliferation ability even in adherent growth conditions. The analysis of cell-cell distances within colonies reveals that adherent subpopulations exhibit significantly reduced intercellular spacing compared to stem cell colonies, with floating subpopulations demonstrating the highest degree of compactness. These findings imply stronger cell-to-matrix or cell-to-cell interactions in floating cell colonies, which may enhance molecular signal transduction (Fig. S3a).

Fig. 4

The difference in self-renewal and proliferative ability between the adherent hHFMSCs^OCT4 and floating hHFMSCs^OCT4. (a) Cell proliferation curve of hHFMSCs, adherent hHFMSCs^OCT4 and floating hHFMSCs^OCT4 obtained from the CCK-8 assay (n = 3). (b) Colony formation assay of hHFMSCs, adherent hHFMSCs^OCT4, and floating hHFMSCs^OCT4; the enlarged views show the difference between the three cell clones (n = 3). The histogram is the rate of colony formation for each cell. (c) ICC of proliferation-associated protein Ki67 expression in three cell groups, and Ki67 protein was localized in the nucleus (n = 3). *P˂0.05, **P˂0.01, ****P˂0.0001. (d) Discrete cutoffs for distinct cell cycle phases are visually selected to calculate the percentages of cells in G1, S, and G2/M phases using R software

Additionally, we examined the proliferation index via Ki67 protein expression and performed ICC, which revealed that the percentage of cells positive for Ki67 in floating hHFMSCs^OCT4 (89.8%) and adherent hHFMSCs^OCT4 (86.03%) was significantly greater than that in hHFMSCs (55.37%) (Fig. 4c).

To clarify cell cycle differences among the three groups, we analyzed DNA content via DAPI staining and compared the percentages of cells in G1, G2, and S phases. As shown in Fig. S3b and Fig. 4d, adherent cells had more cells in the S phase (19.16%) compared to hHFMSCs (6.98%), with similar G2/M phase numbers, indicating more cells left G1 for DNA synthesis. Moreover, floating cells had the highest percentage in the G2/M phase (54.19%), suggesting the most active cell division and proliferation.

The cytoskeleton of floating hHFMSCs^OCT4 was remodelled

The cytoskeleton regulates cell adhesion and junctions, impacting cellular signal transduction during differentiation and development [16, 17]. We examined cell morphology and cytoskeleton using Wright-Giemsa staining, finding spindle-shaped hHFMSCs, polygonal adherent cells, and round floating hHFMSCs^OCT4. Additionally, the diameters of the floating hHFMSCs^OCT4 were significantly reduced by 7.78-fold and 3.45-fold compared to those of the hHFMSCs and adherent subsets, respectively. Furthermore, compared with those in the hHFMSCs and the adherent subsets, the nuclear-to-plasma ratio in the floating hHFMSCs^OCT4 subgroup exhibited significant increases (1.38-fold and 2.96-fold, respectively) (Fig. 5a).

Fig. 5

Remodelling of the cytoskeleton in floating hHFMSCs^OCT4. (a) The cytoplasm and nucleus were stained with Wright‒Giemsa stain in three groups of cells (n = 3). (b) Cytoskeletal microfilaments in the three groups were stained with Coomassie brilliant blue (n = 3). (c) F-actin morphology and distribution were detected by phalloidin staining (n = 3). These images were all taken from randomly selected fields. *P˂0.05, **P˂0.01

Subsequently, Coomassie brilliant blue staining revealed that the cytoskeletal microfilaments of the hHFMSCs exhibited a relatively parallel arrangement and were distributed in filamentous structures along the longitudinal axis of the cells. Conversely, in adherent hHFMSCs^OCT4, the microfilaments were observed to coalesce, forming irregular networks that aggregated towards the nucleus. Notably, in the floating hHFMSCs^OCT4, the microfilaments further clustered into irregular networks and even formed numerous small nodules in the cytoplasm adjacent to the nucleus (Fig. 5b).

Cytoskeletal microfilaments, composed of actin molecules, exhibit diverse morphologies, with F-actin being a prominent variant [18, 19]. F-actin expression and localization were analyzed using fluorescent dye-labeled phalloidin, a specific ligand for F-actin. The staining results demonstrated the presence of F-actin in both the cell membrane and the cytoplasm. Specifically, in hHFMSCs, F-actin exhibited elongation and alignment in parallel bundles, whereas in adherent hHFMSCs^OCT4, F-actin exhibited a grid-like pattern with intertwining branches that were primarily localized around the nucleus. However, in the floating hHFMSCs^OCT4, F-actin exhibited increased contraction and aggregation near the nucleus, resulting in the formation of a clot (Fig. 5c). These findings indicate that OCT4-transduction induced changes in arrangement of cytoskeleton, and the cytoskeleton of floating cell was reassembled conspicuously, implying a novo cell type generated.

Cell adhesion and cytoskeleton control self-renewal of floating hHFMSCs^OCT4 through the AJ pathway

KEGG analysis was used to study the self-renewal mechanisms in floating hHFMSCs^OCT4. The AJ pathway was found to be downregulated in floating cells compared to hHFMSCs, and the DEGs in the AJ pathway were annotated using the KEGG database (Fig. 4a). In the AJ pathway, TJ proteins like ZO-1 control actin polymerization and AJ formation. Adhesion strength changes can regulate Wnt, MAPK, and TGF-β signaling to control cell growth and differentiation.

Subsequently, AJ genes were detected using RT-qPCR, showing increased expressions of ZO-1 and ACTN2 (actin) and decreased expression of E-cadherin in floating cells compared to adherent cells. β-catenin and Cyclin D1 were significantly upregulated (Fig. 6b). This suggests AJ’s role in regulating cytoskeletal dynamics, cell adhesion, and the cell cycle. Additionally, the nuclear localization of β-catenin was observed through IF analysis, which revealed the presence of the β-catenin protein in the nuclei of both adherent hHFMSCs^OCT4 and floating hHFMSCs^OCT4. However, the expression level of β-catenin was notably elevated in floating hHFMSCs^OCT4 (P = 0.0194), suggesting the potential activation of downstream genes, including Cyclin D1 (Fig. 6c and d).

Fig. 6

Expression of AJ core genes and target genes in floating hHFMSCs^OCT4. (a) RT‒qPCR analysis of the expression of ZO-1, ACTN2, E-cadherin, β-catenin, and Cyclin D1 (n = 3). (b) IF staining images showing the location of the β-catenin protein in hHFMSCs, adherent hHFMSCs^OCT4, and floating hHFMSCs^OCT4. (c) Fluorescence intensity of the merged colour in the nucleus was measured in three groups using ImageJ. *P˂0.05, **P˂0.01, ***P˂0.001

ZO-1 is essential for rebuilding the cytoskeleton and starting β-catenin signaling

To confirm ZO-1’s role in the AJ pathway, we used RNAi technology to knockdown the ZO-1 gene in floating hHFMSCs^OCT4. The efficiency of siRNA transfection was assessed using immunofluorescence microscopy and the most effective siRNA sequences and timing were determined through RT-qPCR (Fig. S3a and b). Following knockdown, both mRNA and protein levels of ZO-1 in floating hHFMSCs^OCT4 were significantly reduced (Fig. 7a). Then, a notable decrease in mRNA levels of ACTN2 and β-catenin was observed, with no changes in E-cadherin or Cyclin D1 gene expression (Fig. 7b). Protein expression of β-catenin also significantly decreased by 2.45 folds (Fig. 7c).

Fig. 7

Changes in AJ gene expression and F-actin arrangement in floating hHFMSCs^OCT4 after ZO-1 knockdown. (a) RT‒qPCR and Western blot showing the expression of ZO-1 after RNAi (n = 2). Full-length blots/gels are presented in Supplementary Figure S4. (b) RT‒qPCR and detection of the mRNA expression of ACTN2, E-cadherin, β-catenin, and Cyclin D1 (n = 3). The error bars represent the standard deviations of measurements in three separate sample runs. (c) Western blots showing the protein expression levels of β-catenin (n = 1). Full-length blots/gels are presented in Supplementary Figure S5, n = 1. (d) Phalloidin staining showing F-actin morphology and distribution (n = 3). (e and f) Fluorescence microscopy (Olympus) captured DAPI-stained cell images, and CellProfiler software detected nuclei and calculated their integrated intensity. R software then used visually determined cutoffs to calculate the percentages of cells in G1, S, and G2/M phases. NC, negative control. si-ZO-1, ZO-1-knockdown. *P˂0.05, **P˂0.01, ***P˂0.001, ****P˂0.0001

Moreover, we conducted phalloidin staining to determine the expression, structure, and distribution of F-actin following the knockdown of the ZO-1 gene. These findings revealed a decrease in F-actin expression in floating hHFMSCs^OCT4 after ZO-1 knockdown. Additionally, the cytoskeletal microfilaments exhibited relative extension and a relative decrease in the number of micronodules and vertical branches. Nevertheless, there was a relative increase in the number of microfilaments parallel to the cell membrane (Fig. 7d). These findings suggested that the expression of ZO-1 can affect the expression and branching of F-actin through cooperation with AJ molecules.

To assess the impact of ZO-1 knockout on the floating subpopulation’s cell cycle, we compared the percentages of cells in G1, S, and G2 phases between si-ZO-1 and control NC. As illustrated in Fig. 7e and f, si-ZO-1 cells showed reduced S phase (2.11%) and G2 phase (16.32%) percentages compared to NC cells (11.11% and 27.77%). These findings indicate that ZO-1 knockdown significantly disrupts the cell cycle progression of floating hHFMSCs^OCT4.

The AJ pathway regulates the gene networks involved in cytoskeleton remodelling and haematopoiesis-related self-renewal

To explore the potential target genes of the AJ pathway that could govern the self-renewal, we generated a gene regulatory network in floating hHFMSCs^OCT4 (versus to hHFMSCs) using the resources provided by the NCBI website and STRING database (Fig. 8). The DEGs were subsequently classified into three distinct groups according to their functions. Cluster 1 included seven genes TJP1(ZO-1), TJP3, JUP, CDH2, CDH5, CTNNB1 (β-catenin), and CCNE1, that are primarily associated with the AJ pathway. Additionally, the AJ pathway can modulate the cytoskeleton via the CDH2, CDH5, and JUP genes, which interact with the ACTA2, WASL, ARPC2, and PFN1 genes from Cluster 3, thereby facilitating the aggregation and branching of F-actin. Concurrently, the genes CDH2, CDH5, TJP1, CTNNB1, and CCNE1 within the AJ pathway can augment the self-renewal capacity of floating subsets by synergistically interacting with genes from Cluster 2, which are predominantly responsible for impeding the self-renewal of HSPCs.

Fig. 8

Network analysis. A gene network regulates haematopoiesis-related proliferation of floating hHFMSCs^OCT4. The solid line shows gene relationships within the group, while the dashed line shows relationships between genes in different groups

Notably, except for that of CCNE1 and CDK4, the expression of haematopoietic genes (FOS, STAT3, CEBPA, KIT, FLT3, and STAT5B) may be directly influenced by the activation of CTNNB1. Additionally, the SPI1 and ZFPM1 (FOG1) genes, which play crucial roles in the selection and differentiation of the erythrocyte lineage, were identified as potential targets indirectly regulated by β-catenin.

Hence, the AJ pathway, which comprises TJP1, serves as the central component within these networks, facilitating interactions with cytoskeleton remodelling genes to initiate and transmit signals associated with the haematopoietic potential and self-renewal of floating hHFMSCs^OCT4. This process is achieved through the regulation of proliferation-related genes specific to pre-HSCs and HSPCs. Subsequent investigations will focus on elucidating the mechanisms underlying the interactions between AJ protein molecules and their targets related to haematopoietic potential.

Discussion

This study investigated how cytoskeletal remodel promotes in vitro self-renewal of Pre-HSCs derived from OCT4-reprogrammed hHFMSCs. First, FLT3LG and SCF induced the emergence of floating cells, which can step by step differentiate into enucleated erythrocytes. Furthermore, we found that increased OCT4 levels led to decreased nucleus/cytoplasm ratio and reduced adhesion in floating hHFMSCs^OCT4 through AJ pathway, likely due to upregulation of ZO-1. This was shown by slight depolymerization of F-actin and downregulation of β-catenin expression after ZO-1 knockdown. Additionally, cell cycle genes and HSPCs related genes may be the potential targets of AJ pathway in floating subsets, potentially activating self-renewal capacities and pluripotency of differentiation into erythroid lineage. Overall, ZO-1 may promote HSPCs-like proliferation by regulating the AJ pathway in floating hHFMSCs^OCT4.

Previously, RNA sequencing data were examined to investigate the involvement of the TJ pathway in linking pluripotency and haematopoietic differentiation in a ZO-1-dependent manner [7]. However, the function of ZO-1 in this context have not been elucidated. The self-renewal of stem cells and the maintenance of stem cell properties depend on transcriptional regulatory networks [20, 21], and transduction of the factor OCT4 alone can initiate the conversion of somatic cells into pluripotent cells by inducing the expression of various pluripotency-specific genes [22,23,24]. Moreover, the level of OCT4 expression in embryonic stem cells plays a crucial role in determining whether the cells will continue to self-renew or undergo differentiation [25]. In a similar vein, our investigation revealed heightened expression of OCT4 in the floating subset of cells, which exhibited a greater propensity to undergo trans-differentiation into haematopoietic lineages when exposed to media supplemented with various haematopoietic cytokines [6] and sustained self-renewal in ESC-suitable media [7]. Therefore, we hypothesized that the establishment of a balanced state between proliferation and haematopoiesis can be achieved through the activation of adhesion-related signalling pathways.

A substantial proportion of the specific genes targeted by OCT4 in human germ cell tumours are associated with focal adhesion and extracellular matrix tissue [26]. In our present investigation, we observed that increased expression of OCT4 potentially results in reduced adhesion of hHFMSCs and the appearance of floating hHFMSCs^OCT4, which could be attributed to cytoskeletal remodelling. The cytoskeleton comprises a complex arrangement of interconnected biopolymers and crosslinked molecules. Perturbations or modifications to specific cytoskeletal components, such as actin [27, 28] or vimentin intermediate filaments [29], can significantly impact cellular differentiation processes. The downregulation of cytoskeletal proteins results in cytoskeletal remodelling and a change in the proliferation of human umbilical cord mesenchymal stem cells [30].

Previous research has verified that ZO-1 directly interacts with F-actin, while E-cadherin indirectly associates with the cytoskeleton through α-catenin/β-catenin, which act as a binding protein for F-actin [13, 31, 32]. Researchers have hypothesized that the simultaneous binding of F-actin and the membrane localization of ZO-1 can mutually reinforce each other [33]. This interaction plays a crucial role in anchoring β-catenin to E-cadherin within the cell membrane, thereby impeding the nuclear transport of β-catenin [13, 33]. In our research, we confirmed the regulatory role of ZO-1 in the expression of the above AJ pathway members, as well as its influence on the structure and distribution of the F-actin protein, which may further affect self-renewal potential. The branching of F-actin promotes the translocation of the PDZ domain of ZO-1 to the cytoplasm, which leads to β-catenin translocation into the nucleus [33,34,35]. As mentioned earlier, nuclear F-actin not only affects the subcellular localization and transcriptional function of β-catenin but also plays a role in the regulation of Wnt/β-catenin signalling [36, 37]. Subsequent investigations will focus on exploring the interplay between AJ signals by employing immunoelectron microscopy or coimmunoprecipitation (co-IP).

Interestingly, nuclear expression of β-catenin showed consistent high levels, similar to the levels of OCT4 and ZO-1 in the floating subpopulation, on the other hand, the expression of Cyclin D1, was not downregulated upon ZO-1 knockdown. To identify the other potential target genes of β-catenin, we constructed a gene regulatory network, and the data showed that the AJ pathway was linked to cytoskeletal assembly and cell proliferation in floating hHFMSCs^OCT4. Our current data showed that increased accumulation of β-catenin in the nucleus may affect proliferation and the self-renewal capacity by activating proliferation-related genes specific to HSPCs, as well as erythropoiesis-related genes.

Additional research is warranted to clarify the regulatory mechanism through which AJ impacts its potential targets. Subsequent investigations will examine the impact of ZO-1 reduction on erythropoiesis, with the objective of elucidating the critical role of AJ in generating pre-HSCs from adult stem cells in a laboratory setting. The outcomes of our study will establish a solid basis for utilizing floating hHFMSCs^OCT4 as starter cells for in vitro haematopoiesis.

Conclusions

This study demonstrates that the overexpression of OCT4 in hHFMSCs, followed by pre-induction with FLT3-LG and SCF, facilitates erythropoiesis through ZO-1-mediated cytoskeletal remodeling. ZO-1 enhances the self-renewal capacity of floating hHFMSCs^OCT4 in vitro by modulating the expression of F-actin, E-cadherin, and the nuclear translocation of β-catenin. The interaction between AJ pathways and HSPC genes is essential for achieving both hematopoietic potential and self-renewal capability. This provides a foundational basis for employing floating hHFMSCs^OCT4 as pre-HSCs for in vitro haematopoiesis.