Establishment and characterization of a human pre-osteocyte cell line: hOsteo4-E9

Abstract

Osteocytes, as the most abundant and long-lived cell type in bone tissue, have garnered significant attention due to their critical roles in bone remodeling, bone mechanobiology, bone-related diseases, and systemic homeostasis. While several well-established murine osteocyte cell lines are available, the development of a reliable human-derived osteocyte cell line remains an unmet need in the field. Here, we report the successful purification of primary bone cells from the adult human lumbar articular process and their immortalization via SV40 antigen transduction. The resulting human cell line, designated hOsteo4, exhibits high expression of early osteocyte markers (E11, DMP1), moderate levels of ALP and COL1, and low expression of late osteocyte markers (sclerostin, FGF23). Quantitative PCR analysis and in vitro osteoblastic differentiation assays indicate that hOsteo4 cells represent pre-osteocytes at the transitional stage from late osteoblasts to osteocytes. Notably, the subclone hOsteo4-E9 demonstrates robust responsiveness to fluid shear stress (FSS), characterized by dramatic morphological remodeling and upregulated focal adhesion protein expression. RNA-sequencing further reveals distinct FSS-responsive gene expression profiles when comparing human hOsteo4-E9 to the murine MLO-Y4 cell line. Collectively, hOsteo4-E9, a novel immortalized human pre-osteocyte-like cell line, represents an invaluable experimental tool that promises to advance mechanistic insights into human osteocyte biology.

Graphical Abstract

Graphical Abstract.

Introduction

Bone is one of the stiffest tissues in human body. Bone is also a complex dynamic organ composed of bone cells, bone marrow, and a rigid extracellular matrix (ECM). Among all bone cells, osteocytes represent the most abundant and long-lived cell population in adult bone, accounting for 90%-95% of total bone cells¹ with an average half-life of 20-25 yr in humans.² Over the past 3 decades, considerable attention has been focused on the structure, functions, and regulatory mechanisms of osteocytes.³ In the bone microenvironment, osteocytes engage in cross-talk with various cell types,⁴ orchestrate bone remodeling by modulating bone formation⁵ and resorption,⁶ and govern local calcium-phosphate metabolism.⁷ Notably, osteocytes serve as the primary mechanosensitive cells in bone,³ utilizing diverse molecular mechanosensors to detect and respond to mechanical stimuli—including focal adhesion (FA) proteins,^8–10 gap junctions,¹¹ cellular dendrites,¹² primary cilia,¹³ and mechanosensitive membrane channels.¹⁴

Currently, the most widely used osteocyte cell line is the MLO-Y4 cell line derived from the long bones of transgenic mice in 1997.¹⁵ Several other osteocyte-like cell lines of murine origin have also been established, including MLO-A5,¹⁶ Ocy454,¹⁷ IDG-SW3,¹⁸ and OmGFP66.¹⁹ In contrast, human-derived osteocyte models remain scarce. Bodine et al. reported the HOB-01-C1 cell line which was derived from adult cancellous bone chip,²⁰ yet despite its establishment in 1996, few follow-up research was reported. Alternative approaches include using primary human osteocytes immediately after bone sampling,²¹ differentiating iPSC-derived osteocytes,²² inducing osteoblastic-to-osteocytic differentiation in vitro using primary human osteoblasts.^23,24 Given the substantial genomic and chromatin state disparities between humans and mice,²⁵ murine osteocyte models fail to fully recapitulate human osteocyte biology. Therefore, there is an urgent need to establish a physiologically representative and reliable human osteocyte cell line for in vitro investigations.

In the present study, we purified primary bone cells from human spinal tissue and achieved their immortalization using Simian Virus 40 (SV40) large T antigens.^26,27 The resulting immortalized cell line, designated hOsteo4, was further characterized using immunofluorescence (IF), western blot (WB), and quantitative PCR (qPCR), demonstrating the co-expression of markers indicative of pre-osteocytic identity. Fluid shear stress (FSS) experiments were performed to evaluate the mechanosensitivity of hOsteo4 cells, revealing that the subclone hOsteo4-E9 exhibits robust responsiveness to mechanical stimuli—manifested by dramatic morphological remodeling and dynamic shifts in protein expressions. Collectively, we report the establishment of hOsteo4-E9, a human pre-osteocyte-like cell line which serves as a valuable experimental tool for investigating human osteocyte biology in physiological and pathological contexts.

Materials and methods

Sample collection and patient consent

The primary human bone cells were isolated from the lumbar articular process of a 22-yr-old male patient of Chinese Han ethnicity, who had normal bone mineral density, was diagnosed with spondylolisthesis, and was in overall good health. The patient underwent intervertebral disc surgery at the Department of Orthopedics, Shenzhen Nanshan People’s Hospital. The patient has been gave informed consent and agreed to participate in this project. The size of bone biopsy sample was around 0.5 × 0.5 × 0.5 cm³. This study was approved by the Ethics Committee of Shenzhen Nanshan People’s Hospital (No.KY-2021-053-01).

Primary cell isolation, purification, immortalization, and culture

Primary human bone cells were isolated using a modified protocol based on 2 previous studies.^28,29 Detailed procedures for bone cell isolation, purification, immortalization, and culture were summarized in the Tables S1 and S2. Immortalized human bone cells were maintained in α-minimum essential medium (α-MEM, Gibco, Thermo Fisher Scientific Inc., Shanghai, China) supplemented with 10% FBS (Gibco, Thermo Fisher Scientific Inc., Shanghai, China) and 1% penicillin-streptomycin (P/S, Yeasen, Shanghai, China) at 37 °C in a 5% CO₂ humidified atmosphere. The hOsteo4-E9 cell line is available upon request.

MLO-Y4 and MC3T3-E1 cell culture

The MLO-Y4 cell line, originally established by Dr. Lynda Bonewald’s laboratory,¹⁵ was kindly provided by the collaborative laboratory of Dr. Jian Q. Feng.³⁰ The MC3T3-E1 cell line, initially developed by Dr. M. Kumegawa’s laboratory,³¹ was obtained from Dr. Renny T. Franceschi’s laboratory.^32,33 Both cell lines were maintained in α-MEM (Gibco, Thermo Fisher Scientific Inc., Shanghai, China) containing 10% FBS (Gibco, Thermo Fisher Scientific Inc., Shanghai, China) and 1% P/S (Gibco, Thermo Fisher Scientific Inc., Shanghai, China) at 37 °C in a 5% CO₂ humidified atmosphere. For FSS experiments, MLO-Y4 cells were seeded on culture slips pre-coated with type I rat tail collagen (Cat# A10483-01, Gibco, Thermo Fisher Scientific, Shanghai, China; 1:50 dilution in 1 × PBS), whereas standard culture was performed on non-coated dishes.

Cell viability assay (CCK8)

Cells in the logarithmic growth phase were seeded into 96-well plates at a density of 3 × 10³ cells per well. Cell proliferation and cytotoxicity were evaluated using a Cell Counting Kit-8 (CCK8) assay (Cat# 40203ES76, Yeasen, Shanghai, China), which quantifies cellular metabolic activity by measuring cellular dehydrogenase activity in viable cells.

Quantitative PCR analysis

Total RNA was extracted using an RNA purification kit (Cat# 19221ES50, Yeasen, Shanghai, China), and RNA concentration was determined via Nanodrop spectrophotometry (Thermo Fisher Scientific Inc., Shanghai, China). Next, 1 μg of total RNA was reverse-transcribed into cDNA using an RNA reverse transcription kit (Cat# 11141ES60, Yeasen, Shanghai, China). For quantitative PCR, 1% of the cDNA was mixed with qPCR SYBR Green Master Mix (Cat# 11184ES08, Yeasen, Shanghai, China) and 0.2 μM primers, followed by analysis using the ABI 7500 Fast Real-time PCR System (Applied Biosystems, Thermo Fisher Scientific Inc., Shanghai, China). GAPDH was used as the internal reference gene, and relative gene expression was calculated using the comparative Ct method (2^−∆∆Ct). Primer sequences are listed in Table 1.

Table 1.

PCR primers.

Name	Species	Forward primer	Reverse primer
GAPDH	Human	GATTTGGTCGTATTGGGCG	CTGGAAGATGGTGATGG
GAPDH	Mouse	CAGTGCCAGCCTCGTCCCGTAGA	CTGCAAATGGCAGCCCTGGTGAC
Sclerostin	Human	ACACAGCCTTCCGTGTAGTG	GGTTCATGGTCTTGTTGTTCTCC
Sclerostin	Mouse	AGCCTTCAGGAATGATGCCAC	CTTTGGCGTCATAGGGATGGT
Integrin αv	Human	ATCTGTGAGGTCGAAACAGGA	TGGAGCATACTCAACAGTCTTTG
Integrin αv	Mouse	GCTGAAGCTGTTCTCTTTCTTGCT	ATTTGTAATGTACAGGATGGGCTTT
Integrin α5	Human	GCCTGTGGAGTACAAGTCCTT	AATTCGGGTGAAGTTATCTGTGG
Integrin α5	Mouse	CACTTTGGGCTGTGGAATCG	TGCCAAGATGATCACCCACA
Integrin α3	Human	TCAACCTGGATACCCGATTCC	GCTCTGTCTGCCGATGGAG
Integrin α3	Mouse	TGGGCAAGTGCTATGTGCGTGGCA	TCTGGGTGAAGCCGCCGCTGGT
Integrin β5	Human	GGAAGTTCGGAAACAGAGGGT	CTTTCGCCAGCCAATCTTCTC
Integrin β5	Mouse	TGAGTTCGGGAGGATCTACG	AGTGACACTCTCCATGGCCT
Integrin β7	Human	AGAATGGCGGAATCCTCACCT	TGAAGTTCAGTTGCTTGCACC
Integrin β7	Mouse	GAAGGGCTGCTCCTCCTC	AACTCACTCTGACCTCCGCC
EMP1	Human	GTGCTGGCTGTGCATTCTTG	CCGTGGTGATACTGCGTTCC
EMP1	Mouse	TTGGTGCTACTGGCTGGTCT	AGCATCTTCATTGCCGTAGGA
SNAI1	Human	TCGGAAGCCTAACTACAGCGA	AGATGAGCATTGGCAGCGAG
SNAI1	Mouse	CACACGCTGCCTTGTGTCT	GGTCAGCAAAAGCACGGTT
SNAI2	Human	TGTGACAAGGAATATGTGAGCC	TGAGCCCTCAGATTTGACCTG
SNAI2	Mouse	CATCCTTGGGGCGTGTAAGTC	GCCCAGAGAACGTAGAATAGGTC
CDH1	Human	CGAGAGCTACACGTTCACGG	GGGTGTCGAGGGAAAAATAGG
CDH1	Mouse	CAGTTCCGAGGTCTACACCTT	TGAATCGGGAGTCTTCCGAAAA
CDH2	Human	AGCCAACCTTAACTGAGGAGT	GGCAAGTTGATTGGAGGGATG
CDH2	Mouse	AGGCTTCTGGTGAAATTGCAT	GTCCACCTTGAAATCTGCTGG
CDH11	Human	AGAGGTCCAATGTGGGAACG	GGTTGTCCTTCGAGGATACTGT
CDH11	Mouse	TCTGACATTGACTCCGGTGAT	GCTCCAGTGGTCTGTTGGT

Western blotting

Total protein was extracted using RIPA buffer (Cat# P0013B, Beyotime, Shanghai, China) supplemented with protease inhibitors and phosphorylated protease inhibitors (Cat# DI101-01, Transgen, Beijing, China). A total of 20 μg of total protein were separated by 10% SDS-PAGE (Cat# 20325ES62, Yeasen, Shanghai, China) under reducing conditions, followed by transfer to PVDF membrane and blocking with 5% bovine serum albumin (BSA) in PBST (0.1% Tween 20 in 1 × PBS) for 1 h. Primary antibodies were incubated overnight at 4 °C, followed by incubation with secondary antibodies at room temperature for 1 h. All antibodies were diluted in a commercialized antibody dilution reagent (Cat# P001S-2, Beyotime, Shanghai, China). Detailed antibody information is listed in Table 2.

Table 2.

Antibodies information.

Name	Supplier	Cat #	Western blot	IF
Primary antibodies
Cx43	Abcam	ab217676	1:1000
E11	Affinity	DF12456	1:1000
Gapdh	Abcam	ab181602	1:3000
Dmp1	Invitrogen	PA5-88069	1:1000	1:500
Sclerostin (react with both Human and Mouse samples)	Invitrogen	PA5-71999	1:1000	1:200
Sclerostin (react with Mouse samples)	R&D	AF1589	1:500
Tubulin	Abcam	ab7291	1:3000	1:200
p-FAK	Abcam	ab81298	1:1000	1:200
Integrin β1	CST	34 971	1:1000
Integrin β3	CST	13166S	1:1000
ALP	Invitrogen	PA5-106391	1:3000
Collagen I	Abcam	Ab21286	1:3000
FGF23	ABclonal	A6124	1:1000	1:200
Kindlin-2	Sigma	MAB2617	1:1000	1:200
Snail1	CST	3879 T	1:1000
Cadherin-1 (E-Cadherin)	CST	14472S	1:1000	1:200
Cadherin-2 (N-Cadherin)	Abcam	ab18203	1:1000	1:200
Cadherin-11	Thermo	71-7600	1:1000	1:200
Secondary antibodies
CoraLite488-conjugated Goat Anti-Rabbit IgG(H + L)	Proteintech	SA00013-2		1:1000
CoraLite594—conjugated Goat Anti-Rabbit IgG(H + L)	Proteintech	SA00013-4		1:1000
Anti-rabbit IgG, HRP-linked Antibody	CST	#7074	1:1000

Immunofluorescence staining

Cells were fixed with 4% paraformaldehyde (PFA; Servicebio, Wuhan, Hubei, China) in 1× PBS for 15 min, then permeabilized with 0.2% Triton X-100 (Beyotime, Shanghai, China) in 1× PBS for 15 min, followed by blocking with 3% BSA (Beyotime, Shanghai, China) in 1× PBS for 1 h. Fixed cells were incubated with the primary antibodies at 4 °C overnight, followed by incubation with secondary antibodies for 1 h. All antibodies were diluted in 3% BSA (Beyotime, Shanghai, China) in 1× PBS. For F-actin staining, Rhodamine-phalloidin dye (Cat# PF00001, Proteintech Group, Inc. Rosemont, IL, USA) was used at a 1:250 dilution in 1× PBS. Nuclei were counterstained with DAPI using a mounting medium (Cat# ab104139, Abcam, Shanghai, China). Detailed antibody information is listed in Table 2.

Fluid shear stress stimulation

FSS stimulation was performed as previously described⁸ using either a Streamer System STR-4000 (Flexcell International Corporation, Burlington, NC, USA) or the Fluid Stress V2 device (Biaomianli Technology Co LTD, Hangzhou City, Jiangsu Province, China). Briefly, cells were seeded onto collagen-I pre-coated culture slips at a density of 3.0-3.5 × 10⁵ cells (cell seeding density: 160-200 cells/mm²) and allowed to adhere for 24-48 h prior to FSS treatment. Culture slips were inserted into a parallel plate flow chamber and exposed to continuous unidirectional FSS for 2 h. Applied FSS intensities were 1.5, 3.5, 5, and 10 dynes/cm². For static controls (0 dyne/cm²), cells were seeded on the identical culture slips and maintained without stress application.

Cell morphology quantification

Morphological changes in cells under FSS stimulation were analyzed using ImageJ software (https://imagej.net/ij/). Phase contrast images were acquired with a culture microscope (CKX53, Olympus, Tokyo, Japan), and immunofluorescence (IF) images were captured with a confocal laser scanning microscope (FV3000, Olympus, Tokyo, Japan). To quantify the percentage of dendritic cells and the number of dendrites per cell, the line selection tool in ImageJ was used to trace the dendritic processes. Cell spreading area was measured by outlining the 2D area of the main cell body using the free-hand area selection tool.

RNA-sequencing and data analysis

Cells were subjected to 3.5 dynes/cm² FSS for 2 h, after which total RNA was extracted with an RNA purification kit (Cat# 19221ES50, Yeasen, Shanghai, China). RNA quality was assessed by determining the RNA integrity number (RIN) using a Bioanalyzer 2100 (Agilent techniques, Beijing, China). Only samples with an RIN > 8.0 were used for library construction. Libraries were constructed using the M-MuLV reverse transcriptase system and purified with AMPure XP beads. Library fragment sizes were then accurately verified using an Agilent 4200 TapeStation. Different libraries were sequenced on the Illumina PE150 platform. To identify significant pathways, KEEG pathway enrichment analyses were performed using the clusterProfiler package (version 4.10.0) with the Benjamini-Hochberg method to correct for bias. A significance threshold of p < .05 was applied.

SEM detection

Cells were fixed using an SEM-specific fixative (Cat#G1102, Servicebio, Wuhan, Hubei, China) at 4 °C overnight. Fixed cells were dehydrated through a graded methanol series (35%, 50%, 75%, 90%, and 100%, 10 min per step), followed by complete dehydration with hexamethyldisilazane for 10 min. Samples were then air-dried overnight. Prior to imaging, samples were sputter-coated with 5-nm Au/Pd layer under high vacuum conditions (<10⁻³ Pa). SEM was performed using a high-resolution cold field-emission SEM (Hitachi Regulus 8100). The instrument was characterized by a secondary electron resolution of 1.0 nm (at 15 kV accelerating voltage, WD = 4 mm) and 1.3 nm (at 1 kV, WD = 1.5 mm). Imaging was performed at magnifications ranging from 30× to 1 000 000× with accelerating voltages adjustable between 0.1 and 30 kV.

Osteogenic differentiation assay

Cells were seeded into 12-well plates at a density of 1 × 10⁵ cells per well under standard culture conditions. On day 0, cells were maintained in culture medium, that is, α-MEM medium supplemented with 10% FBS and 1% P/S. From day 1 onwards, osteogenic differentiation was induced by replacing the culture medium with osteogenic differentiation medium (α-MEM medium with 10%FBS, 1%P/S, 50 μg/mL ascorbic acid, 10 mM β-glycerophosphate disodium, 10 nM dexamethasone), which was refreshed every 2 d. Cells were fixed with 4% PFA for 30 min at day 0, 1, 3, 5, and 7, respectively. All samples were subsequently stained using an alkaline phosphatase (ALP) color development kit (Cat# C3250S, Beyotime, Shanghai, China).

Statistical analysis

All data were analyzed with GraphPad Prism software (Version 8.0). All experiments were repeated at least 3 times. Differences between 2 groups were analyzed by Student’s t-test. For comparison involving 3 or more groups, One-way ANOVA was used, followed by Tukey’s test for pairwise comparisons. Results are expressed as mean ± SD. A p-value < .05 was considered as statistically significant.

Results

Isolation and immortalization of human bone cells

To isolate human bone cells, samples were collected from the lumbar articular process (Figure 1A). After removing all connecting soft tissues, the bone samples were cut into small pieces and subjected to 8 rounds of collagenase-EDTA digestion to eliminate non-wanted cells (for detailed protocols, see Tables S1 and S2). According to a published method,²⁸ 5 sequential digestions remove most fibroblasts and osteoblasts, while 7 digestions yield predominantly osteocytes. After 8 rounds of digestions, bone fragments were immediately transferred to culture dishes, and cell viability was monitored under microscopy. By day 21, bone cells began to migrate from the digested bone fragments (Figure 1A), reaching 50% confluency with good cell viability by day 28. Cells were then transduced with SV40 lentivirus and subjected to puromycin selection for 2-3 wk. After antibiotic treatment, the surviving cells were successfully immortalized.

Figure 1.

Primary human bone cells purification and immortalization. (A) Human bone cells were collected from human lumbar articular process, followed by sequential digestion, and cultured in vitro for 28 d before virus infection; (B) in situ F-actin staining of human osteocytes on bone samples; (C) primary human bone cells without (w/o) virus infection and with SV40 large-T antigen infection were cultured in 2D culture dish and stained with F-actin imaging.

To visualize cell morphology, rhodamine-phalloidin was used to label the F-actin cytoskeleton. In situ, osteocytes from human lumbar processes exhibited distinct dendritic processes (Figure 1B). When culturing in 2D conditions, both primary bone cells and SV40-transduced bone cells displayed a flattened morphology with dense F-actin bundles (Figure 1C). Notably, untransduced primary bone cells began to undergo cell death within one week of culture. To date, the immortalized bone cells have undergone over 50 passages and multiple frozen-thaw cycles without loss of viability.

Immortalized human bone cells (hOsteo4) were pre-osteocyte-like cells, expressing DMP1, E11, ALP, and COL1 proteins

The immortalized human bone cells were named hOsteo4 and subjected to single-cell subcloning. Four subclones were selected based on their doubling kinetics: hOste4-C7, D8, E9, and C10, for further characterization. In parallel, murine osteocyte-like MLO-Y4 (Y4) cells¹⁵ and osteoblast-like MC3T3-E1 (3T3) cells³³ were used as control groups in subsequent experiments.

Overall, all 4 hOsteo4 subclones exhibited similar cell morphology under 2D culture conditions (Figure 2A) and comparable doubling rates (Figure 2B). IF assays showed positive staining for dentin matrix protein 1 (DMP1), fibroblast growth factor 23 (FGF23), and sclerostin (Figure 2C) in all subclones. WB analysis revealed high DMP1 expression and moderate E11 expression across all subclones, with generally low sclerostin expression—slightly higher in C7 cells (Figure 2D). Notably, integrins β1 and β3 were highly expressed in all hOsteo4 subclones (Figure 2D). Additionally, these subclones expressed moderate ALP and low COL-I expression (Figure 2D). For COL-I expression (Figure 2D), we observed a band near 100 kDa size (red arrow) which was detected in MC-3 T3 cells and 4 subclones, a band above 180 kDa size (green arrow) which was detected in MLO-Y4 cells and 4 subclones of hOsteo4 cells, and 1 special bone around 130 kDa size (yellow asterisk) mainly detected in MC-3 T3 cells. The collagen heterogeneity across different cell types may result from variations in hydroxylation, glycosylation, and cross-linking of COL-I proteins.

Figure 2.

Characterization of human-derived hOsteo4 subclones. (A) Phase contrast images of 4 hOsteo4 subclones, that is, C7, D8, E9, and C10; (B) cell viability test by CCK8 assay for 4 hOsteo4 subclones and MLO-Y4 cells; N = 3 for each group; (C) immunofluorescence staining of DMP1, FGF23, and sclerostin of 4 hOsteo4 subclones, MLO-Y4 and MC3T3 cells; IgG group were MC3T3 cells and served as a negative background control; (D) western-blot results for Dmp1, Sclerostin, E11, integrin β1, integrin β3, ALP, and Col-I protein expressions from hOsteo4 subclones, MLO-Y4 cells and MC-3 T3 cells; tubulin served as a loading control.

Quantitative PCR (qPCR) further confirmed that hOsteo4 cells expressed relatively high levels of COL-1, along with moderate levels of OCN and BGLAP1/2 (Figures S1A-C). High DMP1 and moderately high E11 gene expression were also observed (Figures S1D and E), alongside robust expression of the mechanosensitive gene CX43 (Figure S1F).

Given the moderate expression of osteoblast-enriched proteins in hOsteo4 cells, we investigated their osteogenic differentiation potential using MC3T3-E1 cells as controls (Figure S2). As shown in Figure S2A and B, hOsteo4-E9 exhibited limited differentiate capacity from days 1 to 3, with progressive cell death occurring from days 5 to 7. In contrast, MC3T3 cells showed weak ALP staining from days 1 to 3, followed by a rapid increase in ALP expression from days 5 to 7. qPCR data further confirmed these observations (Figure S2C-E). Collectively, these results indicate that hOsteo4 subclones are distinct from osteoblasts and more closely resemble pre-osteocyte cells.

hOsteo4-E9 cells actively respond to FSS stimulation

Osteocytes are the primary mechanosensitive cells in bone.³⁴ Next, hOsteo4 subclones were subjected to 10 dynes/cm² FSS forces,⁸ and their morphological changes were observed. As shown in Figure 3A, all subclones displayed a fibroblast-like morphology under static conditions; upon FSS stimulation, some cells transformed into a stellate shape with long, slender dendritic processes extending from the cell body (Figure 3A, red arrows). Among all the subclones, E9 showed the highest proportion of dendritic cells (45.39 ± 5.00%) after FSS treatment (Figure 3B). Thus, hOsteo4-E9 subclone was selected for subsequent FSS characterization experiments.

Figure 3.

Subclone hOsteo4-E9 cells underwent dramatic morphological changes upon FSS treatment. (A) Phase contrast images showed the morphological changes of hOsteo4 subclones in static and FSS conditions; (B) quantitative analysis of percentages of dendritic cells of 4 hOsteo4 subclones with and without FSS; N = 3 for each group; (C) WB results of p-FAK, integrin β3, and kindlin-2 protein changes of hOsteo4-E9 subclone under different FSS stimuli; (D) phase contrast images of cell shape changes of hOsteo4-E9 subclone under different FSS stimuli; (E-G) statistical analysis of percentages of dendritic cells, number of dendrites per cell and the longest dendritic length per cell of hOsteo4-E9 subclone under FSS stimuli; quantitative results were analyzed from 3 independent experiments; (H) IF staining of F-actin cytoskeleton and DAPI nuclei in hOsteo4-E9 cells with and without (static) FSS stimuli; (I) IF images stained with F-actin, p-FAK, tubulin, and Cx43 in hOsteo4-E9 cells with and without (static) FSS stimuli. Arrowheads in figures indicated the direction of FSS forces. Results are expressed as mean ± SD. n. s. p > .05; ^*p < .05; ^**p < .01; ^***p < .001.

Notably, 10 dynes/cm² FSS dislodged a significant proportion of hOsteo4-E9 cells from culture slides, leaving the remaining cells with poor viability. To address this, FSS intensity was reduced. When exposed to 0-2 dynes/cm² FSS, hOsteo4-E9 cells displayed a gradually increased expression of mechanosensitive proteins, including phosphorylated focal adhesion kinase (p-FAK), integrin β3, and kindlin-2 (Figure 3C). However, once the FSS force reached 5 dynes/cm², hOsteo4-E9 cells showed a marked decrease in p-FAK and integrin β3 protein expression (Figure 3C). Consistent with these WB results, phase contrast images revealed that hOsteo4-E9 cells began extending long processes at 1 dyn/cm² force (Figure 3D, large red arrows), formed multiple dendrites at 2 dynes/cm² force (Figure 3D, small red arrows), but displayed a mixed flat-dendritic morphology at 5 dynes/cm² force (Figure 3D, yellow asterisks). Statistically, the percentage of dendritic cells increased in a dose-dependent manner with FSS amplitude (Figure 3E), whereas the number of dendrites per cell and dendritic length were comparable between 1 and 5 dynes/cm² FSS (Figure 3F and G). IF data further confirmed that FSS-induced morphological changes in hOsteo4-E9 cells involved remodeling of F-actin-enriched cytoskeletons (Figure 3H). Additionally, FSS treatment reorganized the distribution of p-FAK, tubulin cytoskeleton and Cx43 in these cells (Figure 3I).

Comparison between hOsteo4-E9 and MLO-Y4 cells in response to FSS

Next, we investigated the differential responses of human and murine cells to FSS. We found that hOsteo4-E9 cells had the optimal morphological changes at 3.5 dynes/cm² FSS. Then we conducted 2-h FSS experiments at 0, 1.5, and 3.5 dynes/cm² with both MLO-Y4 and hOsteo4-E9 cells. Phase contrast images showed that both MLO-Y4 cells and hOsteo4-E9 cells underwent rapid morphological changes upon FSS (Figure 4A). In hOsteo4-E9 cells, the percentage of dendritic cells increased from 3.48% at static condition to 39.12% at 1.5 dynes/cm² and 77.60% at 3.5 dynes/cm² (Figure 4B). Correspondingly, the number of dendrites per cell increased from 1.04 at static condition to 3.97 at 1.5 dynes/cm² and 6.22 at 3.5 dynes/cm². By contrast, MLO-Y4 cells—being inherently dendritic—showed no significant changes in these parameters (Figure 4B and C). When comparing the length of longest dendrite per cell, both MLO-Y4 and hOsteo4-E9 cells exhibited similar dendrite elongation upon FSS (Figure 4D). Notably, hOsteo4-E9 cells exhibited a marked reduction in spreading area under FSS: the average 2D cell area decreased from 3071 μm² at rest to 1522 μm² at 1.5 dynes/cm² and further to 670 μm² at 3.5 dynes/cm² (Figure 4E). In contrast, MLO-Y4 cells maintained a consistent spreading area of 500-600 μm² regardless of FSS (Figure 4E).

Figure 4.

hOsteo4-E9 cells showed cytoskeleton remodeling with enhanced proteins expressions of mechanosensitive genes similar to MLO-Y4 cells in response to FSS treatment. (A) Phase contrast images of MLO-Y4 and hOsteo4-E9 cells under static/0, 1.5, and 3.5 dynes/cm² FSS; (B-E) quantitative analysis of dendritic percentages, number of dendrites per cell, the longest dendritic length in single cell and spreading area changes in MLO-Y4 and hOsteo4-E9 cells under different FSS stimuli; (F) SEM images of MLO-Y4 and hOsteo4-E9 cells with or without FSS; (G) F-actin staining of MLO-Y4 and hOsteo4-E9 cells with or without FSS; (H, I) quantitative analysis of the length of secondary dendrites and the number of secondary dendrites per cell in MLO-Y4 and hOsteo4-E9 cells with or without FSS; (J) western-blot results for p-FAK, kindlin-2, integrin β3, Cx43, Dmp1, and Sclerostin in MLO-Y4, and hOsteo4-E9 cells under different FSS stimuli; tubulin served as a loading control; (K-P) statistical analysis of protein expressions of p-FAK, kindlin-2, integrin β3, Cx43, Dmp1, and Sclerostin in MLO-Y4 and hOsteo4-E9 cells under different FSS stimuli. N = 3 for each group. Arrowheads in figures indicated the direction of FSS forces. Results are expressed as mean ± SD. n.s. p > .05; ^*p < .05; ^**p < .01; ^***p < .001.

SEM images revealed that MLO-Y4 cells formed longer dendrites and denser connections with neighboring osteocytes under FSS; while hOsteo4-E9 cells transitioned from a flat to a stellate shape with multiple dendrites connecting to adjacent cells (Figure 4F). Furthermore, primary dendrites (first extensions from the cell body; Figure 4G, red arrowheads) and secondary dendrites (branches from primary dendrites; Figure 4G, yellow arrowheads) were quantified. Both cell types showed increased secondary dendrite length and number per cell upon FSS (Figures 4H and I). In addition to cell morphology, WB analysis further showed that FSS induced gradually upregulation of FA-related proteins (p-FAK, kindlin-2, integrin β3) in both cell lines (Figure 4J-M). Cx43 expression was increased in hOsteo4-E9 cells but not in MLO-Y4 cells (Figure 4J and N). The osteocyte-specific marker Dmp1 was upregulated in both cell types, though with slightly different expression kinetics (Figure 4J and O). Notably, FSS significantly reduced sclerostin expression in both MLO-Y4 and hOsteo4-E9 cells (Figures 4J and P).

RNA-sequencing indicated different FSS-responsive pathways between MLO-Y4 and hOsteo4-E9 cells

To uncover potential differences in regulatory pathways between MLO-Y4 and hOsteo4-E9 cells, RNA-seq analysis was performed with total mRNA samples from both cell lines under static and FSS conditions. Additionally, mRNA samples from primary bone cells isolated from long bones of 4-mo old male mice using our modified digestion method were included. As shown in the PCA plot (Figure 5A), intra-group variability among samples was small, whereas the inter-group variability was pronounced. A total of 10 550 genes were commonly expressed in both cell types, with 2096 genes uniquely detected in hOsteo4-E9 cells and 789 genes specific to MLO-Y4 cells (Figure 5B). The Venn diagram further revealed notable diversity in differentially expressed genes across 4 experimental groups (Figure 5C): hOsteo4-E9_FSS (E9_F), MLO-Y4_FSS (Y4_F), MLO-Y4_Static (Y4_S), and hOsteo4-E9_Static (E9_S). We also identified the functional significance of differentially expressed genes by mapping them to known biological pathways with KEGG analysis. KEGG pathway enrichment analysis was conducted for all 4 groups (All-KEGG), the hOsteo4-E9-specific group (E9-KEGG), and the MLO-Y4-specific group (Y4-KEGG). The top 15 enriched pathways in All-KEGG included PI3K-Akt signaling, endocytosis, regulation of actin cytoskeleton, etc. (Figure 5D). Y4-KEGG analysis highlighted known pathways such as endocytosis, the JAK-STAT signaling pathway, and TGF-β signaling (Figure 5F), whereas E9-KEGG uniquely enriched less characterized pathways, including neuroactive ligand-receptor interaction and cytokine-cytokine receptor interaction (Figure 5E, highlighted by red dotted lines). We also compared our RNA-seq data with published results^35,36 and found that hOsteo4-E9 cells exhibit a distinct gene profiles and express high level of several osteocyte-specific gene: DMP1, DKK1, BMP2, and BAMB1 (Figure S3).

Figure 5.

RNA-sequencing data revealed that hOsteo4-E9 cells not only share some similarity with MLO-Y4 cells, but also have distinct transcription profiles in response to FSS. (A) PCA plot of gene expression variability of human hOsteo4-E9, murine MLO-Y4, and murine long bone-derived osteocytes; (B) Venn diagram of detected genes in hOsteo4-E9 and MLO-Y4 cells; (C) Venn diagram of detected genes in 4 experimental groups, that is, E9_F, E9_S, Y4_F, and Y4_S; (D-F) top 15 of KEGG enriched pathways in all detected genes, hOsteo4-E9 specific and MLO-Y4 specific genes; (G) clustering heat map highlighted the variations of FA-associated gene expression patterns between 4 experimental groups; (H-M) qPCR validation of gene expression of integrin αν, integrin α5, integrin α3, integrin β5, integrin β7, and EMP1 in 4 experimental groups. N = 3 for each group. Results are expressed as mean ± SD. n.s. p > .05; ^*p < .05; ^**p < .01; ^***p < .001.

Next, we focused on the FA pathway changes in these 2 cell lines (Figure 5G). Consistent with the upregulation of p-FAK, integrin β1, integrin β3, and kindlin-2 observed at the protein level (Figures 3C and 4J-M), RNA-seq data indicated that integrin β1, integrin αν, integrin β6, and integrin-linked kinase were upregulated upon FSS in both cell types (Figure 5G and H). Conversely, integrin α5 and integrin α3 were downregulated upon FSS in both cell types (Figure 5G, I, and  J). Notably, a large subset of FA pathway genes was deferentially regulated between the 2 cell lines (Figure 5G). For example, integrin β5 and epithelial membrane protein 1 (EMP1) were upregulated in MLO-Y4 cells upon FSS but downregulated in hOsteo4-E9 cells (Figure 5G, K, and  M). Additionally, integrin β7 expression showed the opposite trend (Figure 5G and L). Collectively, these findings indicate conserved signaling responses to FSS in human and murine osteocyte-like cell lines, yet highlight substantial species-specific molecular regulatory differences that warrant further investigation.

hOsteo4-E9 and MLO-Y4 cells differentially regulate Snail1-Cadherin11 axis in response to FSS

Regarding specific gene regulation in human bone cells, FSS stimulation markedly upregulated SNAI1 transcription in hOsteo4-E9 cells (Figure 6A, red arrowhead). The Snail1 protein, encoded by SNAI1 gene, is a zinc-finger transcriptional factor involved in embryogenesis and cancer metastasis, serving as an upstream regulator of cadherin-mediated cell junction organization.^37–39 Snail1 controls the expressions of several cadherin (CDH) proteins, including CDH1/E-cadherin,⁴⁰ CDH2/N-cadherin,³⁹ and CDH11/osteoblast cadherin.⁴¹ In bone, CDH1 is participated in osteoclastogenesis,⁴² while CDH2 and CDH11 are important for mesenchymal cells differentiation.⁴³ Notably, the involvement of Snail and cadherin proteins in osteocyte mechanobiology remains unreported.

Figure 6.

hOsteo4-E9 and MLO-Y4 cells deferentially regulated Snail-Cadherin11 signaling pathway in response to FSS. (A) Volcano map of up-regulated and down-regulated genes in hOsteo4-E9 cells upon FSS stimulation; (B-F) qPCR analysis of mRNA levels of SNAI1, SNAI2, CDH1, CDH2, and CDH11 in hOsteo4-E9 and MLO-Y4 cells in response to FSS; (G) WB results of protein expression of Snail1, Cadherin-1, Cadherin-2, and Cadherin-11 in hOsteo4-E9 and MLO-Y4 cells in response to FSS; (H-K) quantitative analysis of protein levels of Snail1, Cadherin-1, Cadherin-2, and Cadherin-11 in hOsteo4-E9 and MLO-Y4 cells in response to FSS; (L-N) IF staining of Cadherin-1, Cadherin-2, and Cadherin-11 in hOsteo4-E9 and MLO-Y4 cells in response to FSS. Arrowheads in figures indicated the direction of FSS forces. N = 3 for each group. Results are expressed as mean ± SD. n.s. p > .05; ^*p < .05; ^**p < .01; ^***p < .001.

Then, we validated the gene expression of 2 SNAIL family genes, SNAI1 and SNAI2, in MLO-Y4 and hOsteo4-E9 cells under static and FSS conditions. Compared to the static controls, FSS upregulated SNAI1 by 50.6% in hOsteo4-E9 cells (Figure 6B), whereas SNAI2 expression remained unchanged (Figure 6C). Conversely, FSS downregulated Snai1 expression by 41.78% in MLO-Y4 cells (Figure 6B), with Snai2 expression unaffected (Figure 6C). We also examined the mRNA levels of SNAI1 downstream target genes in the CDH family via qPCR. The results showed that the mRNA levels of cdh1, cdh2, and cdh11 were decreased in MLO-Y4 cells after FSS treatment (Figure 6D-F). In hOsteo4-E9 cells, however, FSS reduced CDH1 transcription similarly to its effect in MLO-Y4 cells, but slightly upregulated the expression of both CDH2 and CDH11 (Figure 6D-F).

Furthermore, we also performed WB and IF experiments to verify protein level changes in the Snail-Cadherin signaling pathway. Our data showed that Snail1 protein level was largely reduced in MLO-Y4 cells upon FSS treatment, whereas no such change was observed in hOsteo4-E9 cells (Figure 6G and H). Notably, distinct differences in cadherin isoform expression were identified between MLO-Y4 and hOsteo4-E9 cells. Cadherin-1 protein was highly expressed in MLO-Y4 cells but barely detectable in hOsteo4-E9 cells (Figure 6G and I). Conversely, Cadherin-11 protein was highly expressed in hOsteo4-E9 cells but not in MLO-Y4 cells (Figure 6G and K). Cadherin-2 protein was more uniform across both cell types (Figure 6G and J). Moreover, FSS stimulation did not alter the expression levels of Cadherin-1 or Cadherin-2 in neither MLO-Y4 or hOsteo4-E9 cells (Figure 6I and J). However, FSS significantly inhibited Cadherin-11 expression in MLO-Y4 cells, while slightly increasing it in hOsteo4-E9 cells (Figure 6K). IF results further confirmed the differential expression patterns of cadherin isoforms between MLO-Y4 cells and hOsteo4-E9 cells (Figure 6L and M).

Discussion

In current study, human bone cells were successfully purified and immortalized as hOsteo4-E9 cell line, derived from the lumbar articular process of an adult male donor. The hOsteo4-E9 cells express high levels of early osteocyte-enriched markers (DMP1 and E11), low levels of mature osteocyte markers (Sclerostin and FGF23), and high levels of mechanosensitive proteins (p-FAK, Cx43, kindlin-2, integrin β1, and integrin β3). Moreover, hOsteo4-E9 cells exhibit moderate levels of ALP and COL1 protein but minimal osteogenic differentiation potential. These data indicate that hOsteo4-E9 cells represent pre-osteocytes at the transitional stage from late osteoblasts to osteocytes. This characteristic may explain their flat morphology under static conditions and rapid transformation into a dendritic phenotype upon FSS stimulation. Moreover, we continuously detected the similar marker gene expressions and distinct morphological changes upon FSS of hOsteo4-E9 cells after a relatively significant number of passages.

Regarding the origin of hOsteo4-E9 cells, bone samples from the lumbar articular process contained both cortical and cancellous bones. Thus, whether hOsteo4-E9 cells are derived from cortical versus trabecular bone remains undetermined. Given that the lumbar spine is a major load-bearing structure, hOsteo4-E9 cells likely acquired mechanosensitivity from their native microenvironment.

FSS is a primary mechanical stimulus for osteocytes, with fluid-structure interaction modeling estimating 33.8-75.5 dynes/cm² force around osteocytes.⁴⁴ In murine osteocyte studies, FSS of 5.0-50.0 dynes/cm² is typically used, and MLO-Y4 cells were commonly treated with 10.0 dynes/cm² flow.³ These FSS magnitudes are higher than those used in our experiments. In our study, we noticed that hOsteo4-E9 cells generated more dendrites and formed denser intercellular connections under 3.5 dynes/cm² flow. Moreover, the expression changes of FA-associated proteins and sclerostin under 3.5 dynes/cm² flow in hOsteo4-E9 cells were consistent with the regulatory patterns of these proteins in the literature.⁴⁵ These findings suggest that hOsteo4-E9 cells exhibit higher sensitivity to FSS than MLO-Y4 cells, implying species-specific responsive thresholds. In addition, the mechanosensitive properties of hOsteo4-E9 cells may distinguish them from other human osteocyte models. For instance, HOB-01-C1 cell line has been used for osteocyte-related hormonal regulation.²⁰ Other human osteocytes were differentiated from primary human osteoblasts,²³ iPSCs,²² or osteosarcoma cell line²⁴—methods that are time-consuming and prone to variability. In contract, hOsteo4-E9 cells display robust pre-osteocyte characteristics, are easy to culture, and are well-suited for osteocyte biology research.

RNA-seq and qPCR analyses revealed conserved gene regulatory patterns (eg, in FA pathways) between murine and human osteocytes under FSS, while also highlighting divergent transcriptional and translational mechanisms. Our data demonstrated that the Snail-Cadherin11 axis is tightly associated with FSS responses in osteocytes, although with differential regulation between MLO-Y4 and hOsteo4-E9 cells. These species-specific differences represent an unexplored area for future investigation.

Notably, this study utilized murine MLO-Y4 and MC3T3 cells as controls due to the lack of well-characterized human osteocyte/osteoblast cell lines. Although osteocyte-specific antibodies were validated for cross-species reactivity, we cannot rule out differential binding affinities between mouse and human proteins.

In addition, when compared our RNA-seq data with previously published results, we found that the gene expression profiles of hOsteo4-E9 cells differed from these reported studies.^35,36 We attribute these discrepancies in RNA-seq data across studies to the differences of cells used, including species, bone types, and cellular status, as well as variations in sample treatment and RNA purification protocols.

Conclusions

This newly established hOsteo4-E9 cell line represents a human pre-osteocyte model that could serve as a novel tool for in vitro studies of human osteocyte biology. Given its sensitivity to FSS, further investigation into the molecular mechanisms underline osteocyte dendrite formation and network connection during mechanotransduction would be particular insightful. Moreover, future studies could focus on osteocytes’ responses to diverse mechanical stimuli, including bending, compression, and microgravity. In addition, since hOsteo4-E9 cell line is derived from healthy human bone, it holds significant potential as a tool for exploring osteocyte crosstalk with other cell types via co-culture systems. It may also facilitate drug screening for osteocyte-related pathologies, such as osteoporosis, sclerosteosis, periodontal diseases, and bone degenerative disorders.

Author contributions

Lei Qin, Zecai Chen, and Zhen Xu contributed equally. Lei Qin (Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing—original draft, Writing—review & editing), Zecai Chen (Data curation, Formal analysis, Methodology), Zhen Xu (Data curation, Formal analysis, Investigation), Xisan Wang (Investigation, Methodology), Jie Tan (Data curation, Formal analysis), Wanrong Luo (Data curation, Validation), Kun Huang (Data curation, Methodology), Dazhi Yang (Supervision, Validation, Visualization, Writing—review & editing), Guozhi Xiao (Conceptualization, Resources, Supervision, Visualization, Writing—original draft, Writing—review & editing), Weihong Yi (Conceptualization, Funding acquisition, Resources, Supervision, Visualization)

Funding

This work was supported, in part, by the Nanshan District Science and Technology Programs Grant (NS2021018), Nanshan District Health System Major Scientific and Technological Projects Grants (NSZD2023009, NSZD2023036, NSZD2023039), Internal Grands of Shenzhen Nanshan People’s Hospital (YN2021023), Guangdong Provincial Fundamentals and Applications Joint Enterprise Fund (2022A1515220036), National Natural Science Foundation of China Grants (82302748) and the Sanming Project of Medicine in Shenzhen Nanshan People’s Hospital.

Conflicts of interest

The authors declare that they have no interests.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declaration of generative AI in scientific writing

The authors did not use any AI or AI-assisted technologies in the writing process.