Matrix stiffness regulates arteriovenous differentiation of endothelial progenitor cells during vasculogenesis in nude mice

Abstract Objectives The aim of the study was to investigate the effect of matrix stiffness on arteriovenous differentiation of endothelial progenitor cells (EPCs) during vasculogenesis in nude mice. Materials and methods Dextran hydrogels of differing stiffnesses were first prepared by controlling the crosslinking reaction to generate different thioether bonds. Hydrogels with stiffnesses matching those of the arterial extracellular matrix and venous extracellular matrix were separately combined with mouse bone marrow‐derived EPCs and subcutaneously implanted on either side of the backs of nude mice. After 14 days, artery‐specific marker Efnb2 and vein‐specific marker Ephb4 in the neovasculature were detected to determine the effect of matrix stiffness on the arteriovenous differentiation of EPCs in vivo. Results Fourteen days after the implantation of the EPC‐loaded dextran hydrogels, new blood vessels were observed in both types of hydrogels. We further verified that matrix stiffness regulated the arteriovenous differentiation of EPCs during vasculogenesis via the Ras/Mek pathway. Conclusions Matrix stiffness regulates the arteriovenous differentiation of EPCs during vasculogenesis in nude mice through the Ras/Mek pathway.


| INTRODUC TI ON
The establishment of functional vascularization is key for tissue regeneration and represents one of the major challenges to the broad implementation of tissue engineering in clinical practice. 1 The formation of organ-specific vasculatures requires crosstalk between the developing tissue and specialized endothelial cells (ECs). We propose a new source of specialized ECs based on bone marrow-derived endothelial progenitors (EPCs). EPCs circulate in the bloodstream, proliferate and differentiate into mature ECs. 2 After migrating into the peripheral circulation, EPCs assemble at sites of endothelial injury in response to stimuli for revascularization and endothelial repair. Previous studies have indicated that EPCs participate in the pathogenesis of vascular diseases, such as atherosclerosis, abdominal aortic aneurysm and cardiovascular diseases. 3 EPCs may be divided into two populations as follows: early and late EPCs. Early EPCs, which can be obtained by culturing isolated mononuclear cells for 4-7 days, have limited proliferative capacity. In contrast, late EPCs exhibit higher proliferative potential in culture and survive beyond 2 weeks. Both cell types contribute to neovasculogenesis in vivo; early EPCs secrete various angiogenic cytokines, and late EPCs differentiate into specific ECs. 4 The phenotype of a late EPC is highly dependent on its microenvironment. However, the functional properties of EPCs and the molecular mechanisms of their specialized differentiation into arterial and venous subtypes are still unknown.
This limitation negatively affects the practical use of EPCs. Previous studies have indicated that fluid shear stress induces the differentiation of EPCs into arterial ECs 5 ; this finding suggests that controlling the physical characteristics of the microenvironment may represent a method for optimizing EPC-based vascularization outcomes.
The vasculature is subdivided into two interconnected, yet structurally and functionally distinct, networks of arteries and veins. They form one of the body's largest surfaces, which serves as a critical interface between the circulation and the different organ microenvironments. Among the microenvironmental cues, matrix stiffness is a universal mechanical input that profoundly controls cell behaviour, including differentiation. 6,7 The range of venous tissue stiffness in mammals, which is between the stiffness of the epithelium and cartilage, ranges from 3 to 50 kPa. [8][9][10] Unlike veins, arteries can withstand a higher blood pressure and are surrounded by several layers of smooth muscle cells and connective tissues. 11 Therefore, arteries are stiffer, with stiffness ranging from approximately 50 to 150 kPa or higher. 12,13 However, when these stiffness values are compared with those of the glass or plastic containers used for tissue culture, such as Petri dishes (~10 6 kPa), it becomes apparent that the microenvironment in which most cells are cultured in vitro is not representative of the physiological scenario. Accordingly, in order to design matrices to the specific requirements of diverse tissues, it is necessary to develop well-defined biomaterials that mimic the matrices of the vasculature and enable the reliable control of functional vascularization.
With this in mind, we first developed a stiffness-adjustable dextran hydrogel model. Mouse bone marrow-derived EPCs were cultured in the hydrogels to form vascular networks following in vivo implantation to investigate the effect of matrix stiffness on the arterial-venous differentiation of EPCs. We further explored the precise mechanism by which matrix stiffness regulates the expression of arterial and venous markers in the new vessels. Our work provides a potential method for adapting EPC-based vascularization to the specific requirements of a diverse range of tissues, thus, representing a substantial advancement in regenerative medicine.

| Fabrication of dextran hydrogels of varying stiffnesses
The hydrogel preparation method was based on the non-cytotoxic crosslinking of maleimide-modified dextran polymers with thiol-reactive groups (mal-dextran) and crosslinkers with thiol groups (CD- where E is Young's modulus, F is the force exerted on the gel, A 0 is the original cross-sectional area through which the force is applied, ΔL is the change in the thickness of the gel and L 0 is the initial thickness of the gel. The thicknesses of the hydrogel before and after force loading were determined microscopically.

| Animal model
First, EPC-containing dextran hydrogels were prepared. Based on the required ratios for the two groups of hydrogels at concentrations of 2.0 and 7.0 mmol/L (Table 1), we added water, 10 × concentration buffer, and mal-dextran to a centrifuge tube and added the resulting mixture to a suspension of passage 1 EPCs (1 × 10 6 /mL) in M199 medium Next, we subcutaneously injected the 300 μL hydrogel complexes in the left (6 kPa) and right (109 kPa) sides of the backs of the nude mice before crosslinking ( Figure 3A). After 14 days, the hydrogels were surgically removed and washed with PBS.

| Tissue sectioning
The hydrogels from the nude mice were fixed in 10% formaldehyde for 4 hour. After fixation, the hydrogels were dehydrated sequen-

| Immunohistochemical staining to identify in vivo vasculogenesis
The sections were dewaxed with xylene I for 20 minutes and xylene

| Artificial degradation of dextran hydrogels
The hydrogels were artificially degraded with dextranase (3-D Life

| Inhibitor experiments
We

| Statistical analysis
All experiments were repeated for at least three times, independently. Statistical analysis of data was performed with the SPSS statistical software version 21.0 (IBM, Armonk, NY, USA) using t test.
In each analysis, data were considered to be significantly different when the two-tailed P value was <0.05.

| Identification of EPCs
Image of the obtained cells observed by a microscope was shown in Figure 1A. Flow cytometric analysis first revealed that CD31 + /

| Assessment of the dextran hydrogels of tunable stiffness
The hydrogels with different compositional ratios were added to a 48-well plate. After 50 minutes, the hydrogels were fully crosslinked. arterial ECM, respectively. Therefore, in the following in vivo experiments, we selected these two dextran hydrogels to mimic the venous and arterial ECM for EPC cultures, which we referred to as the 6 kPa group and 109 kPa group, respectively.

| Analysis of EPC morphology on matrices of varying stiffness
First, we investigated the response of EPCs to microenvironmental stiffness via the dynamics of the cytoskeletal network. Phalloidin staining of the cell cytoskeleton indicated that there were apparent differences in cellular morphology and cytoskeletal filament arrangements on matrices of varying stiffness. The EPCs cultured on the surfaces of the soft matrices were ovoid with a small cell spreading area, and their internal cytoskeletal filaments were short, slender, and were unable to form obvious filament bundles. On stiffer matrices, the cytoskeleton filaments were elongated and formed crosslinked filament bundles and a larger EPC spreading area ( Figure 2C). In addition, we quantitatively analysed the cell spreading area and reached the same conclusion: cell spreading area increases with increasing matrix stiffness ( Figure 2D-E).

| In vivo vasculogenesis during culture of the hydrogel-EPC composites
The hydrogels with different stiffnesses were translucent, and blood vessels were visible in both the soft and hard hydrogels ( Figure 3B-C).
Next, we assessed the nature of the tubular structures within the hydrogels by immunohistochemistry. The tubular structures in the hydrogels were labelled with the anti-CD31 antibody ( Figure 3D), demonstrating that the newly formed tubular structures in the hydrogels were vascular structures.

| Detection of the arteriovenous markers Efnb2 and Ephb4
In our study, new vessels cultured on both hydrogels simultaneously expressed the arterial endothelial marker Efnb2 and the

Real-time qPCR analysis indicated that the mRNA level of Efnb2
in the 109 kPa group was 3.63 × higher than that in the 6 kPa group.
In contrast, the mRNA level of Ephb4 in the 109 kPa group was 36.4% of that in the 6 kPa group. Thus, the mRNA expression of the Efnb2 gene increased with the stiffness of the hydrogel matrix, whereas the transcription level of the Ephb4 gene showed the opposite trend ( Figure 4C).
Western blotting further confirmed that the expression of the arterial marker increased with increasing matrix stiffness, whereas the venous marker showed the opposite trend. Consistent with the results of immunofluorescence staining, the protein expression levels of Efnb2 and Ephb4 in the 109 kPa group were 1.71 × and 0.61 × those in the 6 kPa group, respectively ( Figure 4D).

| Mechanotransduction and underlying intracellular signalling pathways
Next, we investigated the potential signalling pathways involved in the mechanotransduction process. In our study, the expressions of Ras and Mek were upregulated in the 109 kPa group at both the mRNA and protein levels ( Figure 5A-C). The upregulation of RhoA could also be observed, which further contributed to the activation of the Ras pathway. 14 Each experimental value is expressed as the mean ± standard deviation; C, Gene transcript levels of the arteriovenous markers in the newly formed vessels in the hydrogels of different stiffnesses. All groups of genes were first normalized to internal references, then normalized to the control group (6 kPa group); D, Western blotting for and statistical analysis of Efnb2 and Ephb4 expression in the newly formed vessels in the hydrogels of different stiffnesses. Data shown were representative of three independent experiments; *P < 0.05, ** P < 0.01, ***P < 0.001 In this work, we found that both the mRNA and the protein levels of Notch1 were elevated in the 109 kPa group along with its downstream protein Hey1. Further, consistent with our previous study and other experimental reports, [17][18][19][20] activation of the Notch pathway can inhibit the expression of VEGFR3 ( Figure 5D-F).

| Results of inhibitor experiments
The Ras signalling inhibitor farnesylthiosalicylic acid (FTS) was used to verify the role of the Ras/Mek signalling pathway in the matrix stiffness-mediated regulation of the arteriovenous differentiation of EPCs. In the experimental 109 kPa group treated with the Ras/ Mek pathway inhibitor FTS, the Efnb2 and Ephb4 gene levels did not statistically differ from those in the control 6 kPa group ( Figure 6A).
Western blotting revealed that treatment with the inhibitor prevented significant enhancement of the expression of the arterial marker Efnb2 in the hard hydrogel and even slightly decreased its expression ( Figure 6B). There were no significant differences in the expression of Efnb2 or Ephb4 between the two groups in the pres-

| Dextran hydrogel properties
As a necessary microenvironment for cells, the ECM provides mechanical support for cells and conducts a variety of biochemical and biophysical signals, thereby affecting the biological behaviour of cells. 21

| Effect of matrix stiffness on cell morphology
The process through which cells respond to ECM stiffness is also the process through which cells achieve a dynamic balance between their contraction tension and the ECM force. 25

| In vivo vasculogenesis during culture of the hydrogel-EPC composites
Vasculogenesis is the process in which desired new blood vessels are produced by close interactions between ECM, seed cells and growth factors. 30 We performed a number of preliminary experiments in this study to determine the optimal experimental conditions for in vivo vasculogenesis in nude mice. We found that the EPC-free hydrogels implanted into nude mice did not show vascular structure formation after 14 days, even when the hydrogels were seeded with vascular growth factors, such as VEGF and bFGF, and only a small

| Mechanism exploration
Studies on the morphologies of EPCs on matrices of different stiffnesses have shed light on the important role of the ECM-integrinactin system for the transduction of matrix stiffness mechanical signals. 35 The interactions between EPCs and the ECM initiate a series of reactions to convert mechanical signals into biochemical signals that affect the outcomes of cells. External physical signals from the ECM activate the superfamily of small GTP-binding proteins through integrins. 36 The superfamily of small GTPbinding proteins can be divided into three major subfamilies as follows: Ras, Rho and Rab. 37 Ras molecules are members of the superfamily of small GTP-binding proteins, and the related signal transduction pathway Ras/Raf/Mek/Erk, also known as the Ras/ Mek signalling pathway, is considered the "master switch" of many mechanically transduced signals. 38  Based on our results and the published literature, we propose a mechanism for the regulation of arteriovenous differentiation of EPCs by matrix stiffness (Figure 6) [48][49][50][51] Notch1 and Hey1 are arterial markers, and VEGFR3 is a venous marker. The trends we detected in the changes in the expression of these markers were largely consistent with the trends in the changes of Efnb2 and Ephb4 expression. Enhanced arterial marker expression and decreased venous marker expression are associated with increased matrix stiffness. Therefore, the results of our research on the mechanism of action support our conclusion that the arteriovenous differentiation of EPCs is regulated by matrix stiffness.

| CON CLUS IONS
In summary, matrix stiffness regulates the arteriovenous differentiation of endothelial progenitor cells during vasculogenesis in nude mice through the Ras/Mek pathway: arterial lineages were obtained on stiff substrates while venous commitment predominated in the softer matrix. Therefore, the stiffness of the matrix must be considered when conducting vascularized material design, drug testing and cell therapies to meet tissue-and organ-specific vascularization requirements.

ACK N OWLED G EM ENTS
This study was supported by the National Natural Science Foundation of China (81771125, 81471803) and Sichuan Province Youth Science and Technology Innovation Team (2014TD0001).

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no competing interests.