Breaking the symmetry of cell contractility drives tubulogenesis via CXCL1 polarization

Significance

Tubulogenesis can be regulated by both biochemical molecules and mechanical properties of the cellular microenvironment. However, technical limitations have hindered the study of tubulogenesis directly from a biomechanical aspect. Here, we successfully established a robust system to investigate the mechanoregulatory mechanisms underlying tubulogenesis. Symmetry breaking of cell contractility via reduced matrix stiffness or inhibited myosin activity facilitated polarization of CXC chemokine ligand 1 at cell protrusions. This process promoted cell–cell association and tube elongation. These findings provide valuable insights into the complex interplay between biomechanics and cellular behaviors during tubulogenesis. The establishment of this system not only expands our knowledge of mechanobiological regulation but also paves the way for future research exploring the intricate molecular mechanisms underlying tubulogenesis.

Abstract

The intricate interplay between biomechanical and biochemical pathways in modulating morphogenesis is an interesting research topic. How biomechanical force regulates epithelial cell tubulogenesis remains poorly understood. Here, we established a model of tubulogenesis by culturing renal proximal tubular epithelial cells on a collagen gel while manipulating contractile force. Epithelial cells were dynamically self-organized into tubule-like structures by augmentation of cell protrusions and cell–cell association. Reduction and asymmetric distribution of phosphorylated myosin light chain 2, the actomyosin contractility, in cells grown on soft matrix preceded tube connection. Notably, reducing matrix stiffness via sonication of collagen fibrils and inhibiting actomyosin contractility with blebbistatin promoted tubulogenesis, whereas inhibition of cytoskeleton polymerization suppressed it. CXC chemokine ligand 1 (CXCL1) expression was transcriptionally upregulated in cells undergoing tubulogenesis. Additionally, inhibiting actomyosin contractility facilitated CXCL1 polarization and cell protrusions preceding tube formation. Conversely, inhibiting the CXCL1–CXC receptor 1 pathway hindered cell protrusions and tubulogenesis. Mechanical property asymmetry with cell–collagen fibril interaction patterns at cell protrusions and along the tube structure supported the association of anisotropic contraction with tube formation. Furthermore, suppressing the mechanosensing machinery of integrin subunit beta 1 reduced CXCL1 expression, collagen remodeling, and impaired tubulogenesis. In summary, symmetry breaking of cell contractility on a soft collagen gel promotes CXCL1 polarization at cell protrusions which in turn facilitates cell–cell association and thus tubule connection.

Cell shape change is the cumulative result of force (1). Anisotropic contraction leads to a polarized cell state that in some contexts can result in tubule morphogenesis. A biomechanical feedback loop from cell–collagen interaction restricts mechanotransduction along the linear pattern of collagen fibers by which cells break the symmetry of collagen organization and facilitate tubule-like structure formation in human mammary gland cells (2). Of note, this collagen reorganization process requires cell contractility, suggesting that symmetry breaking of cell contractility could contribute to tube formation. This process is also crucial for the development and function of various tubular organs in multicellular organisms, for example, blood vessels, kidneys, and lungs. Cell contractility is influenced by various biochemical, biophysical, and mechanical factors, and its understanding is crucial for gaining insights into the development of tubular organs and addressing related health issues (3, 4).

The actomyosin contractile unit, comprising the actomyosin cytoskeleton with the molecular motors myosins, serves as a primary source of cellular traction force that shapes cell morphology (1). Cytoskeleton, including microfilaments and microtubules, not only provides structural support but is also the driving force for cell and tissue morphogenesis (5–7). The balance between actomyosin contractility and microtubule polymerization regulating cell shape and cell–collagen interaction has been demonstrated in fibroblasts (8) but is yet to be studied during tubulogenesis.

Although it is widely accepted that extracellular matrix (ECM) proteins form networks providing both structural and biomechanical cues to cells, there is a scarcity of studies investigating the biomechanical regulation of tubulogenesis. A recent study employing liquid chromatography–tandem mass spectrometry and 3D confocal microscopy revealed a complex and dynamic expression pattern of the interstitial matrix during kidney development, indicating the potential biomechanical role of ECM in tubulogenesis (9). Supporting this notion, an in vitro study demonstrated that the rigidity of the substrate influences the differentiation of renal proximal tubular epithelial cells (10). Additionally, collagen gels derived from young versus aged rat tails exhibit distinct biophysical properties. Collagen fibril stiffness contributes to different activities during the branching morphogenesis of renal epithelium within collagen gels (11, 12). Furthermore, it has been proposed biomechanical factors affect tubule elongation through shear stress applied to the apical surface during branching morphogenesis in both the kidney and the lung (13). Therefore, biomechanical cues from the microenvironment influence the tubulogenesis of epithelium. However, the 3D culture environment of tubulogenesis models impedes the direct measurement of biomechanical signals. The current study developed a simple culture system utilizing renal proximal tubular epithelial cells grown on collagen gels. The co-axial system comprising of atomic force microscopy (AFM) and confocal microscopy enables direct measurement of the mechanical properties at a single collagen fibril level during tubulogenesis.

CXC chemokine ligand 1 (CXCL1) belongs to the CXCL class of chemokines that signal through CXC receptors (CXCRs) involved in inflammatory reactions and chemotaxis (14–16). Additionally, CXCLs/CXCRs have been implicated in the progression of endothelial morphogenesis and several cancers (14, 16–18). While the biological function of CXCL1 in a rat renal proximal tubular epithelial cell line (NRK-52E cells) has been associated with neutrophil attraction under pro-inflammatory conditions (19), the role of CXCL1 in renal tubulogenesis has yet to be investigated.

Previously, we demonstrated that matrix stiffness and composition profoundly affect cytoskeletal mechanics (20) and the morphogenesis of MDCK cells (11, 21), indicating that an outside-in biomechanical signal regulates epithelial morphogenesis. Notably, the expression level of integrins in MDCK cells positively correlates with the branching morphogenesis in collagen gel, suggesting the involvement of mechanotransduction through beta 1 integrin (CD29) (22). Consistent with these findings, perturbation of CD29 and alpha 6 integrin has been shown to affect tubule formation and mesenchyme condensation in a kidney ex vivo organ culture system (23), highlighting the significant role of CD29 in renal proximal tubule morphogenesis. Motivated by these previous studies, in the current work we show that a soft environment could drive non-branching morphogenesis of renal proximal tubular epithelial cells on a collagen gel by remodeling actomyosin contractility and CXCL1 polarization.

Results and Discussion

The In Vitro Model of Soft Collagen Gel–Induced Tubulogenesis.

To the best of our knowledge, there is currently no in vitro model available for generating renal tubules on a collagen gel substrate. Our previous study demonstrated the influence of collagen gel mechanical properties on branching tubulogenesis (11) and the potential role of alpha 3 beta 1 integrin as a mediator of mechanotransduction during tubulogenesis (24). Those studies were conducted within a collagen gel framework. Although the 3D model of tubulogenesis may capture its inherent nature, it hinders direct measurement of mechanoregulation during tubulogenesis. This study established a cell model to study the non-branching morphogenesis of epithelial tubes utilizing collagen gels derived from fish skin or rat tail and cells with different levels of biomechanical properties. To investigate the biomechanical effect on epithelial tube morphogenesis, renal proximal tubular epithelial cells were cultured either on the collagen gel–coated surface (gel-coating) or directly on the collagen gel (gel) (Fig. 1A). The collagen matrix derived from fish skin used for both gelation and surface coating exhibited similar biological activity. However, there was a significant difference in the rigidity between the gel-coating and the gel. The gel-coating appeared thinner but displayed a stiffer rigidity compared to the gel (Fig. 1B). Cells grown on stiffer surfaces (gel-coating) displayed higher levels of phosphorylated MLC2 (pMLC2) at the cell perimeter dissociated from the microtubule domains. Alternatively, pMLC2 was associated with microtubules at the cell periphery and cell protrusions on the gel (Fig. 1 C and D and SI Appendix, Fig. S1). Not only was pMLC2 changed but the microtubule distribution pattern also, suggesting a symmetry breaking of cell contractility in response to substrate stiffness.

Fig. 1.

Collagen gel stiffness modulates tubule morphogenesis. (A) Experimental setup of tubulogenesis on the collagen gel. Rat renal proximal tubular epithelial cells (NRK-52E cells) were cultured on collagen-coated dishes (gel-coating) or collagen gel (gel) for the indicated periods. (B) AFM analysis of the mechanical properties of the fish skin–derived collagen gel and gel-coating. n = 221,724 for gel-coating and n = 149,613 for gel. ***P < 0.001. (C) Representative confocal microscopy image of cells grown on the collagen gel–coating or gel for 6 h. The images show tubulin-α (green), pT18S19MLC2 (red), and nucleus (blue). The enlarged image with signal profile analysis shows strong colocalization of microtubules and pMLC2 at cell peripherals and protrusions. Arrows denote the colocalization of microtubules and pMLC2 at cell peripherals and protrusions. (D) The statistical analysis of the mean fluorescence intensity of pT18S19MLC2 per high power field (HPF) in C. There were approximately 61 and 104 cells per HPF in the gel-coating and gel group, respectively. ***P < 0.001. (E) Representative images show cells on the fish skin–derived collagen gel formed tubule-like structures, while cells on the collagen-coating formed a monolayer of cell sheets. (F) Tubulogenesis on the fish skin–derived collagen gel with various initial cell densities for 1 d and the statistical analysis thereof. ***P < 0.001. (G) Montage images extracted from a timelapse recording of tubulogenesis on the fish skin–derived collagen gel (Movie S1) presented as a 2-h-based series from 2 to 24 h (as denoted in the separate frame) after plating on the gel. (H) Representative confocal microscopy image of cells grown on the collagen gel for 1 d, showing the presence of lumen-like structures (white arrows) stained with E-cadherin (green), F-actin (red), and nucleus (blue). (I) Representative confocal microscopy image of cells grown on the collagen gel for 3 d, demonstrating the spiral elongation of tubule structures stained with E-cadherin (green), F-actin (red), and nucleus (blue).

Upon culture, cells formed a monolayer of cell sheets on the gel coating. Conversely, they organized into tubule-like structures on the gel (Fig. 1E). The formation of these tubule networks occurred rapidly within 24 h and remained remarkably stable with minimal variations over the subsequent 5 d (SI Appendix, Fig. S2). Notably, the initial cell density played a critical role in determining the complexity of the tubule networks (Fig. 1F), highlighting the importance of cell–cell communication during tubulogenesis. To capture the dynamic process of cell organization during tubulogenesis, timelapse microscopy was employed. Surprisingly, a non-branching morphogenic process was revealed (Fig. 1G and Movie S1). It was characterized by the mobilization of individual cells and subsequent cell aggregate formation through robust cell protrusions. This was followed by the connection of cell aggregates, tube elongation, and network formation (SI Appendix, Fig. S3).

After 24 h of tube formation, the tube consisted of multiple cell layers and exhibited empty spaces within the cell aggregates, as evidenced by E-cadherin and F-actin staining (Fig. 1H). However, these intercellular spaces or mini-lumena became less evident 3 d after tube formation, replaced by a spiral-shaped orientation of tubular elongation (Fig. 1I).

Taken together, this model provides valuable insights into epithelial tube non-branching morphogenesis using renal proximal tubular epithelial cells. The choice of gel-coating versus gel culture conditions resulted in the alteration of cell contractility and distinct cellular arrangements, highlighting the influence of substrate stiffness. The non-branching morphogenic process observed during tubulogenesis, along with the intercellular dynamics and tubular elongation, contributes to our understanding of the complex mechanisms driving tubule formation.

Breaking the Symmetry of Cell Contractility by Reducing Matrix Stiffness or Myosin Activity Induces Tubulogenesis.

Research has demonstrated the influence of ECM on cells, adhesion, migration, differentiation, and morphogenesis is not limited to just structural support or cell-intrinsic mechanisms. It includes matrix stiffness–mediated mechanotransduction, including integrin-mediated signaling and mechanosensitive ion channel–dependent pathways (25). Collagen gel derived from different ages of rat tails exhibits distinct matrix stiffness, which can modulate branching tubulogenesis in MDCK cells and non-branching tubulogenesis in the current study (Fig. 1 B and E). Motivated by these findings, we aimed to investigate the impact of matrix stiffness on the non-branching tubulogenesis of epithelial cells using 9-mo-old rat tail collagen that shows a higher matrix stiffness when compared with younger rats (11, 12). To reduce matrix stiffness, we employed sonication on collagen fibrils with varying amplitudes with a fixed frequency, resulting in different levels of collagen fragmentation (Fig. 2A). The protein stain analysis revealed an increase in small molecular weight fragments of collagen, indicating the successful modulation of collagen structure. The stiffness of the gel was significantly reduced by 70% following sonication as measured using AFM (Fig. 2B). Consistent with previous results using the gel and gel-coating strategy to modulate matrix stiffness (Fig. 1), cells demonstrated not only the asymmetric pMLC2 distribution (Fig. 2C and SI Appendix, Fig. S4) but also the reduction (Fig. 2D) of pMLC2 levels on the gel made from the sonicated collagen. Notably, cells exhibited a spreading phenotype and formed a monolayer of cell sheets on the control gel (Movie S2), whereas they organized into tube structures on the gel made from the sonicated collagen (Fig. 2 E and F). Compared with collagen gel derived from fish skin, the 9-mo-old rat tail gel exhibited a higher matrix stiffness similar to the gel-coating from fish skin. Interestingly, the sonicated rat-tail collagen matrix stiffness was reduced to a level similar to fish skin–derived gel which induced tubule-shaped formation. These results strongly suggest the reduction of cell contractility elicited by soft matrix promotes the reorganization of cells into tubule structures.

Fig. 2.

Matrix stiffness and myosin activity modulates tubulogenesis via symmetry breaking of cell contractility. (A) Coomassie blue stain of the SDS-PAGE of the collagen solution derived from the 9-mo-old rat tail with different sonication amplitudes. (B) AFM analysis of the mechanical properties of the 9-mo-old rat-tail collagen gel treated with or without 75% amplitude sonication before gelation. n = 77,184 for control and n = 117,794 for sonication. ***P < 0.001. (C) Representative confocal microscopy image of cells grown on 9-mo-old rat-tail collagen gel treated with or without 75% amplitude sonication before gelation for 3 h. Staining of tubulin-α (green), pT18S19MLC2 (red), F-actin (cyan), and nucleus (blue). (D) Statistical analysis of the mean fluorescence intensity of pT18S19MLC2 per high power field (HPF) in C. There were approximately 18 and 15 cells per HPF in the control and sonication groups, respectively. ***P < 0.001. (E) Representative images of tube formation by cells grown on collagen gels with different levels of sonication before gelation. (F) Analysis of total tube length observed in the cells cultured on collagen gel treated with or without 75% amplitude sonication before gelation. ***P < 0.001. (G) NRK-52E cells cultured on the fish skin–derived collagen gel treated with blebbistatin. (H) NRK-52E cells cultured on the fish skin–derived collagen gel treated with cytochalasin D. (I) NRK-52E cells cultured on the fish skin–derived collagen gel treated with nocodazole. DMSO was used as the vehicle control. ***P < 0.001. (J) Montage images extracted from timelapse recordings of tubulogenesis on the fish skin–derived collagen gel (Movie S3) with or without blebbistatin, nocodazole, and cytochalasin D.

Cytoskeleton remodeling and crosstalk, including actin microfilaments, microtubules, and intermediate filaments, are important for multiple biological processes, for example, the regulation of cell shape, cell migration, and tissue morphogenesis (6). Particularly, the coordination of microfilaments and microtubules is critical for cell protrusions, epithelial cell shape, and function (26). Considering these findings, we aimed to investigate whether cytoskeleton remodeling plays a role in the non-branching morphogenesis of epithelial tubes on the collagen gel focusing on the balance of intrinsic forces by manipulating cell contractility and cytoskeleton polymerization. The compressive forces exerted by the actin cortex and the actomyosin contractile unit serve as a physical barrier that prevents microtubule polymerization (5, 8). To explore this further, we employed the myosin inhibitor blebbistatin to suppress actomyosin contractility, along with inhibitors that interfere with cytoskeleton remodeling to study tubulogenesis. Interestingly, the suppression of actomyosin contractility resulted in enhanced tubulogenesis (Fig. 2G). Conversely, inhibiting the polymerization of microfilaments using cytochalasin D (Fig. 2H) and microtubules using nocodazole (Fig. 2I) resulted in a concentration-dependent suppression of tube formation. These observations suggest that intracellular forces generated by cytoskeleton polymerization are necessary to drive tube formation. Notably, the enhancement of tubulogenesis resulting from the suppression of cell contractility was diminished (Fig. 2J and Movie S3) when either cytochalasin D (SI Appendix, Fig. S5) or nocodazole (SI Appendix, Fig. S6) was added. This implies an antagonistic relationship between the actomyosin contractile unit and microfilaments/microtubules, highlighting their roles in modulating tubulogenesis.

These findings further underscore the importance of matrix stiffness in modulating tubulogenesis and reveal the critical involvement of cytoskeleton remodeling, particularly microfilaments, and microtubules. The enhancement of tubulogenesis upon the suppression of cell contractility, along with the subsequent antagonism observed when cytoskeleton inhibitors were introduced, provides further insights into the complex interplay between the actomyosin contractile unit and cytoskeletal structure during tubulogenesis. The reduction in gel stiffness induced by sonication is similar to the inhibition of actomyosin with blebbistatin since both reduce cell contractility. The reduction and asymmetry of cell contractility facilitated the formation of tube structures by promoting cell reorganization. These findings contribute to our understanding of the mechanobiological regulation of epithelial morphogenesis and emphasize the importance of matrix stiffness in guiding cellular behaviors and probably tissue development.

Soft Matrix Induces CXCL1 Polarization Promoting Cell Protrusions and Tubulogenesis.

Although the effect of CXCL1 in renal proximal tubular epithelial cells remains unclear, it has been shown to mediate the chemotactic migration of neutrophils (19) and endothelial cells (27) through the CXCRs (14). Activation of chemokine receptors, including CXCRs, leads to the rearrangement of the cell cytoskeleton and the formation of protrusions that guide cells toward areas of higher ligand concentration (16). In particular, CXCL1 has been shown to facilitate neo-angiogenesis in a rat corneal model (27, 28). Considering that cell protrusions and cell motility are essential steps in the tube formation process as observed in our timelapse recording (Movie S1 and Fig. 1G), we investigated the association between CXCL1 expression level and tubulogenesis by comparing its expression levels under different tube-forming conditions, e.g., on matrices with different rigidities. Our findings revealed CXCL1 expression was enhanced at both mRNA (Fig. 3A) and protein levels (Fig. 3 B–D and SI Appendix, Fig. S7) and was polarized at cell protrusions (Fig. 3C and SI Appendix, Fig. S7) in cells at the early stages of tubulogenesis on fish skin–derived collagen gel compared to gel-coating conditions. Similar results were observed when comparing CXCL1 expression levels in cells on the reduced stiffness sonicated 9-mo-old rat tail-derived collagen gels (Fig. 3 E and F). These results indicated a strong correlation between CXCL1 expression and tubulogenesis, implying the importance of CXCL1 expression and polarization in tubulogenesis.

Fig. 3.

Soft matrix induces CXCL1 polarization promoting cell protrusions and tubulogenesis. Analysis of (A) Cxcl1 mRNA expression of cells grown on the fish skin–derived collagen gel–coating or gel for 6 h and their (B) CXCL1 protein level in the culture supernatant. **P < 0.01; ***P < 0.001. (C) Representative confocal microscopy images showing CXCL1 (green), pT18S19MLC2 (red), microtubules (cyan), and nucleus (blue) in cells on the fish skin–derived collagen gel or gel-coating for 6 h. (D) Statistical analysis of mean fluorescence intensity of CXCL1 per high power field (HPF) in C. There were approximately 61 and 104 cells per HPF in the gel-coating and gel group, respectively. ***P < 0.001. Analysis of (E) Cxcl1 mRNA expression of cells grown on the 9-mo-old rat tail collagen gel with or without sonication (75% amplitude) for 1 d and their (F) CXCL1 protein level in the culture supernatant. ***P < 0.001. (G) Representative epifluorescent microscopy images of cellular staining for microtubules (green), microfilaments (red), and nucleus (blue) in cells treated with SB225002 on the fish skin–derived collagen gel for 7 h (yellow arrows: cell protrusions). (H) Montage images extracted from timelapse recordings of cells treated with siRNAs on fish skin–derived collagen gel (Movie S4) (red arrows; cell protrusions at 6 h). (I) Statistical analysis of relative cell protrusion activity per high power field (HPF) in Fig. 4H. There were approximately 310, 267, and 247 cells per HPF at 1 h in the siCntl, siCxcl1, and siCxcr1 group, respectively. **P < 0.01. (J) Tube formation on fish skin–derived collagen gel for 2 d in cells with siRNA knockdown of CXCL1. Cells with siCntl served as control. ***P < 0.001. (K) Tube formation on fish skin–derived collagen gel for 1 d in cells treated with SB225002 CXCRs inhibitor. *P < 0.05; **P < 0.01. (L) Tube formation on the fish skin–derived collagen gel for 2 d in cells with siRNA knockdown of CXCR1. ***P < 0.001.

To further investigate the requirement of CXCL1 in tubulogenesis, we utilized a pharmaceutical inhibitor, SB225002, to target CXCR1/CXCR2 (18), and siRNAs to reduce the expression of either CXCL1 or CXCR1. Since we did not detect CXCR2 expression in NRK-52E cells using the real-time PCR method, we focused on CXCR1. Successful reduction of CXCL1 expression was achieved using siRNAs, as demonstrated by mRNA expression (SI Appendix, Fig. S8) and protein levels (SI Appendix, Fig. S9). Notably, inhibition of the CXCL1–CXCRs pathway using a pharmacological inhibitor, SB225002 (Fig. 3G and Movie S4), or siRNAs targeting CXCL1 and CXCR1 (Fig. 3H and Movie S5) reduced the formation of cell protrusions at the early stages of tubulogenesis (Fig. 3I). The images captured in Fig. 3G provide visual insights into the cellular morphology and cytoskeletal organization following treatment with SB225002. Notably, the yellow arrows point out cellular protrusions, highlighting their presence before the tube connection, which is also highlighted in Fig. 1G. Additionally, within the SB225002-treated group, a minimal cell protrusion can be observed as indicated by a white arrow, suggesting the effect of the inhibitor on cellular protrusions. Therefore, these figures and movies offer valuable visual evidence to support the experimental findings and contribute to a better understanding of the impact of the CXCL1–CXCR1 pathway on cellular morphology and cytoskeletal dynamics in the context of tubulogenesis.

Importantly, the ability to form tubes was significantly inhibited in cells with CXCL1 knockdown (Fig. 3J and Movie S5). Furthermore, the use of a CXCRs inhibitor demonstrated a concentration-dependent suppression of tube formation (Fig. 3K and Movie S4), suggesting the involvement of the CXCL1–CXCRs pathway in tubulogenesis. Consistent with these findings, tubulogenesis was greatly suppressed in cells with CXCR1 knockdown (Fig. 3L, SI Appendix, Fig. S10, and Movie S5). Therefore, these data highlight the critical role of CXCL1 in tubulogenesis. The expression of CXCL1 is associated with the level of cell protrusions and tubulogenesis, as evidenced by its enhanced expression in cells undergoing tube formation. Furthermore, functional inhibition of CXCL1 or its receptor CXCR1 significantly impairs the ability to form tubes, confirming the requirement of the CXCL1–CXCR1 pathway in tubulogenesis.

Reduction of Cell Contractility Promotes Cell Protrusions.

Given the observed correlation between CXCL1 expression and its distribution along cell protrusions during tubulogenesis, we proceeded to investigate whether a reduction in cell contractility would impact cell protrusions and CXCL1 expression. Inhibiting actomyosin contractility using blebbistatin did not alter CXCL1 expression levels (Fig. 4 A and B) while promoting cytoskeleton remodeling and CXCL1 polarization at cell protrusions (Fig. 4C and SI Appendix, Fig. S11). The cell number, as indicated by nuclear staining (Fig. 4D), remained unchanged (SI Appendix, Fig. S12), but cell protrusions, as indicated by phase contrast imaging (Fig. 4D) or by staining of microfilaments and microtubules (Fig. 4C), significantly increased (Fig. 4E) preceding tube connection. These results along with the timelapse recordings (Movie S3 and Fig. 2J) suggest forces generated from cytoskeleton remodeling, resulting from a reduction in cell contractility, could be one of the driving forces behind cell protrusions.

Fig. 4.

Reduction of cell contractility promotes cell protrusions. (A) Real-time PCR analysis of Cxcl1 expression in cells treated with or without blebbistatin on fish skin–derived collagen gel for 1 d. (B) ELISA of CXCL1 protein expression in the cell culture supernatant from cells treated with or without blebbistatin on fish skin–derived collagen gel for 1 d. (C) Representative confocal microscopy images showing CXCL1 (green), microtubules (red), microfilaments (cyan), and nucleus (blue) in cells treated with or without blebbistatin and grown on fish skin–derived collagen gel for 6 h. (D) Representative epifluorescent microscopy images of CXCL1 (green), microtubules (red), and nucleus (blue) in cells treated with or without blebbistatin on fish skin–derived collagen gel for 6 h. Cell morphology was observed through the bright field channel (Ph) (yellow arrows: cell protrusions). (E) Statistical analysis of cell protrusions per high-power field (HPF) of D. ***P < 0.001.

Cell Protrusions Remodel Collagen Fibrils to Facilitate Cell–Cell Association and Result in Tubulogenesis.

As a result of the outside-in signal from matrix stiffness regulating tubulogenesis as demonstrated in Figs. 1 and 2, we were interested to know whether the inside-out signal exerting on the matrix also played a role in tube formation. Notably, the asymmetric distribution of pMLC2 in the cells on the soft matrix (Figs. 1 C and D and 3C and SI Appendix, Figs. S1 and S4) was observed, suggesting anisotropic traction forces in the cells in response to matrix stiffness. We hypothesized cells would remodel pericellular collagen fibrils through cell protrusions during tubulogenesis resulting in higher rigidity of collagen fibrils associated with cell protrusions or the tips of tubule structure. Cell protrusions refer to specific cellular domains that mediate ECM remodeling (8) and cell–cell association contributing to tube formation. Since the CXCL1–-CXCR1 pathway was critical for cell protrusions, we further investigated whether inhibiting this pathway using SB225002 could affect ECM remodeling at the early stage of tubulogenesis (Fig. 5A). Intriguingly, the control group exhibited a strong alignment of collagen fibrils along cell protrusions, as indicated by microfilaments, prior to tube connection (DMSO, Fig. 5B). In contrast, cells treated with SB225002 displayed a predominantly spherical morphology, characterized by reduced cell protrusions and collagen accumulation around cell perimeters (Fig. 5B). These findings imply cell protrusions exert mechanical forces to reshape the matrix during tube connection.

Fig. 5.

Cell protrusions remodel collagen fibrils facilitating cell–cell association resulting in tubulogenesis. (A) Schematic presentation of ECM remodeling by cell protrusions at the early stage of tube connection. (B) Representative confocal microscopy images of FITC-conjugated collagen I (FITC-Col I, green) microfilaments (red), and nucleus (DAPI, blue) in cells on collagen gel treated with or without SB225002 for 6 h. Heat map indicates expression level of FITC-Col I (white arrows: cell protrusions). (C) Schematic presentation of AFM-confocal microscopy measurement on the cell protrusions and their flank region of cell aggregates on collagen gel for 6 h. (D) AFM-confocal microscopy analysis of cellular stiffness in the cells on FITC-conjugated collagen gel for 6 h. (E) Statistical analysis of cellular stiffness for cell protrusions and flank regions of cell aggregates as shown in D. n = 2,754 for protrusions and n = 7,026 for flank. ***P < 0.001. (F) Schematic presentation of AFM-confocal microscopy measurement on different regions of round-shaped and tubule-shaped cell aggregates on collagen gel for 1 d. (G) AFM-confocal microscopy analysis of cell aggregates-matrix interaction on FITC-conjugated collagen gel for 1 d. Heat map indicates the range of height and stiffness of the measurement. Brighter colors indicate higher values. (H) Statistical analysis of matrix stiffness among different regions of cell aggregates is shown in G. n = 2,694 for 0°, n = 2,357 for 90°, n = 3,931 for tip, and n = 5,322 for flank. ***P < 0.001.

To gain further insights into the distribution of mechanical properties of cell membranes, we employed a co-axial system of AFM and confocal microscopy to reveal 3D images with rigidity maps of the cell protrusion and its flank regions. Consistently, the results showed cell protrusions formed at the early stage of tube formation (Figs. 1G and 5 C and D). They contained microfilaments, microtubules, CXCL1, and pMLC2 (Figs. 1C, 2C, 3 C and G, and 4 C and D and SI Appendix, Figs. S1, S4, and S7) and exhibited higher stiffness than their flank region (Fig. 5E). These data support anisotropic traction forces in the cell protrusions as ideal subcellular domains for ECM remodeling and inside-out signal transduction. These results suggested the involvement of cytoskeleton remodeling and CXCL1 distribution in force generation for cell protrusions and consequent ECM remodeling and tube connection.

The transition from round-shaped cell aggregates to tube-like structures, as illustrated in Fig. 5F, is attributed to the process of symmetry breaking. The association of cell–collagen fibril and rigidity map among the round-shaped and tubule-shaped cell aggregates was revealed using the AFM-confocal system (Fig. 5G). There were noticeable differences in collagen fibril organization between round-shaped and tubule-shaped structures. Specifically, near the tubule tip, collagen fibrils appeared straighter, whereas they formed a reticular pattern near the round-shaped cell aggregates or the flank region of a tube (confocal images in Fig. 5G). These results implied a difference in traction forces exerted on the matrix among round-shaped and tubule-shaped structures and near the tip and the flank region of the tubule structure. Consistent with this observation, the matrix stiffness between two perpendicular regions within the round-shaped cell aggregates was found to be similar (Fig. 5H). However, we observed an approximately 1.6-fold increase in matrix stiffness at the tip region compared to the flank region of the tubule-shaped structure (mean stiffness values of 3.633 kPa and 2.285 kPa, respectively, Fig. 5H). These results demonstrated the presence of anisotropic matrix rigidity distributed along the periphery of tubule-shaped cell aggregates, suggesting anisotropic stress from cells acting on the neighboring ECM is required for maintaining the tubule-shaped structures.

Suppressing the Mechanosensing Machinery CD29 Abolishes CXCL1 Expression and Tubulogenesis.

Integrins, the collagen receptors, sense mechanical stimulations and transduce signals through focal adhesions which regulate cytoskeleton remodeling and gene expression (29). Previously, we showed CD29 acts as mechanosensing machinery in epithelial cells in a soft environment (22). Therefore, we investigated the role of CD29 in CXCL1 expression and tube formation on collagen gels. A stable CD29 knockdown clone of NRK-52E cells was established and exhibited a significant reduction of CD29 expression at both the mRNA (SI Appendix, Fig. S13) and protein levels (Fig. 6A). Notably, the knockdown of CD29 did not impact cell proliferation (SI Appendix, Fig. S14), but it significantly impaired tube formation (Fig. 6B). This suggests a crucial role for CD29 in tube formation.

Fig. 6.

Suppressing the mechanosensing machinery CD29 abolishes CXCL1 expression and tubulogenesis. (A) Western blot analysis showing the expression of CD29 and tubulin-α in cells transfected with control (shCntl) or CD29-targeting shRNA (shCD29). (B) Tube formation on fish skin–derived collagen gel for 1 d in cells with shCntl or shCD29. ***P < 0.001. (C) Representative confocal microscopy images of cells with phalloidin (red) for microfilaments and DAPI (blue) for the nucleus on FITC-conjugated collagen gel for 1 d. (D) Analysis of Cxcl1 mRNA expression in cells grown on fish skin–derived collagen gel for 1 d. ***P < 0.001. (E) ELISA analysis of cell culture supernatant taken from cells grown on fish skin–derived collagen gel for 1 d. ***P < 0.001. (F) Flow chart shows breaking the symmetry of cell contractility by reducing matrix stiffness or myosin activity promotes CXCL1 polarization at cell protrusions. CXCL1–CXCR1 interaction at cell protrusions promotes cell polarization and ECM remodeling, which in turn, accelerates cell–cell connection and tube elongation.

Collagen remodeling was significantly altered in the CD29 knockdown cells (Fig. 6C). As shown in Fig. 6C, collagen fibrils formed a mesh-like network in the absence of cells and became condensed around the cell aggregates. In control cells, the parallel and straight alignment of collagen bundles was observed in the vicinity of cell aggregates, particularly at the tip region of the tubule-shaped structure. However, this alignment was not observed in the CD29 knockdown cells (Fig. 6C). These results support the involvement of CD29 as a collagen receptor in tubulogenesis.

CD29 has been shown to play a role in CXCL1 expression in astrocytes (30) and endothelial cells (31). Therefore, we investigated whether CD29 expression affects CXCL1 expression in NRK-52E cells. Our results demonstrated that CD29 knockdown not only reduced tube formation (Fig. 6B) but also decreased CXCL1 expression at both the mRNA (Fig. 6D) and protein levels (Fig. 6E). These results suggest mechanotransduction through CD29 signaling modulates CXCL1 expression, in turn contributing to tubulogenesis. Although the molecular mechanism underlying mechanoregulation of CXCL1 expression through CD29 in renal epithelial cells during tubulogenesis remains uncertain, several studies have demonstrated that NF-κB is involved in its transcriptional regulation (31, 32).

These results highlight a critical role of CD29 in tube formation on collagen gels. The knockdown of CD29 impairs tube formation and alters collagen remodeling. Of note, long-range collagen fibril alignment was observed between two cell aggregates (Figs. 5B and 6C). These findings suggest the presence of matrix-transmitted paratensile signaling, previously observed in fibroblast crosstalk (33), plays a role in epithelial cells during tube formation. The hypothesis that paratensile signaling is involved in the regulation of tubulogenesis, potentially associated with the function of CD29, warrants further investigation. Moreover, these data suggest a connection between CD29 and CXCL1 during the tubulogenesis of the renal proximal tubular epithelial cells, with CD29 knockdown leading to reduced CXCL1 expression. These findings provide valuable insights into the mechanistic regulation of tubulogenesis and underscore the significance of the CD29–CXCL1 axis in this process.

These results indicate that reducing cell contractility, either through the extrinsic factor of matrix stiffness or the intrinsic factor of myosin activity, promotes cytoskeleton remodeling. This, in turn, facilitates symmetry breaking of cell contractility and subsequent CXCL1 polarization, cell protrusions, and tube connection (Fig. 6F).

Conclusions

Here, we investigated how environmental cues elicited from the biophysical properties of the ECM could arouse self-assembly of the epithelial tube through modulation of cell contractility. This question is of great significance for tissue regeneration, as many internal organs, including the kidney, exhibit soft tissue properties under normal physiological conditions. However, these organs often become progressively stiffer and lose their regenerative ability during pathological transformations for example tissue fibrosis. This suggests a relatively soft environment is necessary for maintaining normal tissue physiology and regenerative capability. Of note, we have unveiled a mechanosensing mechanism that governs the activation of CD29 in epithelial cells experiencing low substratum rigidity (22). In that study, CD29 activation occurs at approximately 60 Pa and reaches saturation at 386 Pa, highlighting the threshold at which CD29 responds to mechanical cues (22). An additional previous work indicated the appropriate zone of wound bed stiffness for hair regeneration falls within the range of 5 to 15 kPa (34), emphasizing the significant role of mechanoregulation in tissue remodeling during regeneration. The current study shows renal tubulogenesis preferentially occurs in a soft environment (collagen gel made from 9-mo-old collagen solution was 294.2 ± 70.66 Pa and sonicated gel was 89.6 ± 48.31 Pa at a single collagen fibril level), and reducing matrix stiffness or myosin activity promotes tube connection, indicating a biomechanical cue that guides tubular morphogenesis. By utilizing this simple cell culture system in conjunction with the co-axial system of AFM and confocal microscopy, mechanoregulation of tubulogenesis was characterized. The biomechanics involved in tubulogenesis were directly measured. Our findings reveal that a suitable level of biomechanical signaling derived from a soft environment can modulate the CD29 pathway, thereby preventing isotropic contraction and cell spreading. Instead, CD29 triggers the CXCL1–CXCR1 pathway and disrupts the balance between the actomyosin contractile unit and microtubules. Consequently, this process induces anisotropic stress, leading to cell protrusions and ultimately culminating in the formation of tubes. These findings provide valuable insights into the molecular mechanisms coupling biomechanical and biochemical regulations underlying tubulogenesis and suggest potential therapeutic targets for modulating this process.

Materials and Methods

Cell Culture.

The NRK-52E cell line, rat renal proximal tubular epithelial cells, was used in this study. The cells were maintained in DMEM medium (HyClone, SH30003.02) supplemented with 10% fetal bovine serum. For CD29 knockdown (shCD29), shRNA was used with the targeting sequence (GCACGATGTGATGATTTAGAA), and scramble control (shCntl) with the sequence of

CCTAAGGTTAAGTCGCCCTCG. Puromycin-resistant clones of shCntl and shCD29 were isolated for further study. For transient suppression of CXCL1 and CXCR1 expression, siRNAs targeting rat Cxcl1 (M-098496-01-0005) and Cxcr1 (M-090650-01-0005) were purchased from GE Dharmacon (siGENOME) and transfected into NRK-52E cells using an electroporator (Neon Transfection System, Invitrogen) with the following program: 1,300 V, 20 ms, 2 pulses, after trypsin-mediated cell detachment. The CXCRs inhibitor, SB225002, was purchased from MedChemExpress (HY-16711). Blebbistatin, cytochalasin D, and nocodazole were purchased from Sigma-Aldrich and Cayman Chemical.

Western Blot.

Cell lysates were subjected to denatured and reduced SDS-PAGE and transferred to a PVDF membrane. After membrane blocking with 5% non-fat milk, specific antibodies against CD29 (610468, BD Biosciences) or tubulin-α (66031, ProteinTech) were applied. After washing with PBST (0.05% tween 20 in PBS), an HRP-conjugated antibody was applied. The specific signal was acquired and analyzed using the iBright Imager System (iBright 1500, Invitrogen) after washing with PBST.

Preparation of Rat Tail Collagen.

Animal care conditions and the experimental protocols (111143 and 113023) were approved by the Institutional Animal Care and Use Committee of the National Cheng Kung University. Rat tails from 9-mo-old Sprague-Dawley rats obtained from the National Cheng Kung University Animal Center were used to extract collagen following the procedures described previously (11). The procedure involved making an incision along one side of the tail to remove the skin, exposing the tendons at the distal end, and collecting the tendon fibers. The tendons were then dissolved in 0.5 M acetic acid under stirring at 4 °C. After centrifugation at 10,000×g for 30 min, the supernatant underwent collagen salt precipitation with 0.7 M NaCl at 4 °C for 24 h. The collagen pellet obtained after centrifugation was solubilized in 25 mM acetic acid and dialyzed to remove the salt. The collagen solution was then lyophilized and stored at 4 °C until use. To reduce collagen crosslinking, the collagen solution obtained from the rat tail was subjected to sonication (Q125, Qsonica) at different amplitudes with a fixed frequency (20 s on and 2 s off) for a total duration of 5 min at 4 °C. The level of collagen fibril fragmentation was analyzed using Coomassie Blue stain after SDS-PAGE.

Labeling of Collagen with FITC.

Collagen labeling with FITC (F7250, Sigma) was performed by dissolving FITC in anhydrous DMSO at a concentration of 1 mg/mL. The FITC solution was slowly added to the collagen solution (354236, Corning) with continuous stirring at a 1:100 dilution. The reaction was incubated in the dark at 4 °C for 24 h. Subsequently, the solution was dialyzed against 25 mM acetic acid to remove free FITC.

Tubulogenesis on the Collagen Gel.

Type I collagen solutions isolated from rat tail (from a 9-mo-old rat), fish skin (purchased from Ducolege Biotechnology Co., Ltd.), or FITC-conjugated collagen were used to prepare collagen gels. The collagen solution (1 mg/mL) was neutralized using NaOH, as previously described (11). For the gel-coating, collagen gel solution was used to rinse the surface for cell culture. The cells were detached from the culture dish using a cell detachment solution (AT-104, Accutase) and neutralized with culture medium. A micro-slide (ibidi, 81506) was used to study tubulogenesis and for the measurement of the total tube length. For the treatment of inhibitors, cells were treated with inhibitors on the micro-slide. The relative total tube length was measured using ImageJ software 1 to 2 d after cell culture. Timelapse recordings of tubulogenesis were captured using an inverted microscope (Leica DMI6000 B) equipped with a digital EMCCD camera (iXon 897) and objective lenses (N PLAN 10x/0.25 and HCX PL FLUOTAR L 20x/0.4) at 37 °C supplemented with 5% CO₂. The extent of cell protrusion activity in cells treated with siRNAs, including siCntl, siCxcl1, or siCxcr1, was determined through analysis of timelapse recordings. This involved quantifying the number of cell protrusions at the 6-h mark and subsequently normalizing this value to the cell count observed at the 1-h mark on the collagen gel.

Co-Axial System of AFM and Confocal Microscopy.

A co-axial system comprising a laser scanning confocal microscope (FV3000, Olympus) and an Atomic Force Microscope (BioScope Resolve, Bruker) was used to observe and scan living cells and FITC-conjugated collagen fibers. An AFM equipped with a 65 nm probe (PFQNM-LC-A-CAL) was operated in FASTForce Volume mode at a 20-Hz scanning rate. The indentation force was adjusted below 3-nm deformation. Force-distance curves were analyzed using the Sneddon model in NanoScope Analysis software. Immunofluorescent staining was performed using specific antibodies against CXCL1 (AF-515-NA, R&D Systems), tubulin-α (05-829, Merck), pMLC2 (95777, Cell Signaling Technology), and E-cadherin (610182, BD Biosciences) followed by Alexa Fluor-conjugated secondary antibodies (Invitrogen). Microfilaments were stained using Alexa Fluor-633-conjugated phalloidin. When comparing the expression levels of molecules within the group using confocal microscopy, consistent acquisition conditions and display processes were employed. The mean fluorescence intensity per high-power field was computed by dividing the fluorescence intensity, as measured using ImageJ software, by the number of cells present.

ELISA.

The CXCL1 protein expression in the conditioned medium of cell cultures was measured using the ELISA kit (DY515) from R&D Systems, following the manufacturer's instructions. The conditioned medium was collected at specified time points for the ELISA, while the cell lysate was used for RNA isolation. The relative protein expression level was calculated and normalized to the amounts of cells in each group.

Real-Time PCR.

Total RNA was isolated from cells using the Direct-zol RNA MiniPrep kit (Zymo Research), and cDNA was generated using the PrimeScript RT Reagent kit (TaKaRa). PowerUp SYBR Green Master Mix (Applied Biosystems) was used for PCR, which was performed using a StepOnePlus Real-Time PCR System (Bio-Rad). The relative mRNA expression level of the target gene was quantified using the 2^ΔΔCt method, with hypoxanthine phosphoribosyltransferase 1 (HPRT) as the internal control. The PCR primer sequences are listed in Table 1.

Table 1.

The primers for real-time PCR

GENE		Sequence
HPRT	Forward primer	TCT TTG CTG ACC TGC TGG ATT ACA
	Reverse primer	AGT TGA GAG ATC ATC TCC ACC AAT
Cxcl1	Forward primer	ACT CAA GAA TGG TCG CGA GG
	Reverse primer	CTT GGG GAC ACC CTT TAG CA
Cxcr1	Forward primer	TAT ACA GGC GAA GGA CCC GA
	Reverse primer	GCC AAG AAG GGC AGA GTC AA
CD29	Forward primer	CGT GCA TGT TGT GGA GAC TC
	Reverse primer	AAC AAT TCC AGC AAC CAC GC

Statistical Analysis.

Data are presented as mean ± SEM. Statistical analysis was performed using GraphPad Prism. Student’s t test (unpaired and two-tailed) was used to compare two groups, while one-way ANOVA followed by Bonferroni's multiple comparisons test was used for comparing more than two groups. The non-parametric statistical tests were conducted to analyze non-normally distributed datasets. Therefore, Figs. 1B, 2B, and 5E were analyzed using the Mann-Whitney test, and Fig. 5H was analyzed using the Kruskal–Wallis test. A P-value less than 0.05 was considered statistically significant.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Non-branching morphogenesis of renal proximal tubular epithelial cells. A timelapse recording of tubulogenesis on the fish skin-derived collagen gel. This movie was taken 2 hours after the cells were on the collagen gel.

Movie S2.

Reduction of matrix stiffness promotes tubulogenesis. A timelapse recording of tubulogenesis on the 9-month-old rat tail-derived collagen gel. This movie was taken 0.5 hours after the cells were on the collagen gel.

Movie S3.

Inhibiting cell contractility promotes tubulogenesis. A timelapse recording of tubulogenesis on the fish skin-derived collagen gel in the presence of blebbistatin or combination with nocodazole or cytochalasin D. This movie was taken 2 hours after cells were on the collagen gel.

Movie S4.

Inhibiting CXCL1-CXCRs suppresses tubulogenesis. A timelapse recording of tubulogenesis on the fish skin-derived collagen gel in cells treated with SB225002 (20 μM). This movie was taken 1 hour after the cells were on the collagen gel.

Movie S5.

Knockdown of CXCL1 or CXCR1 suppresses tubulogenesis. A timelapse recording of tubulogenesis on the fish skin-derived collagen gel in cells transfected with siRNA against Cxcl1 or Cxcr1. This movie was taken 1 hour after the cells were on the collagen gel.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.