Mechanical stress determines the configuration of TGFβ activation in articular cartilage

Our incomplete understanding of osteoarthritis (OA) pathogenesis has significantly hindered the development of disease-modifying therapy. The functional relationship between subchondral bone (SB) and articular cartilage (AC) is unclear. Here, we found that the changes of SB architecture altered the distribution of mechanical stress on AC. Importantly, the latter is well aligned with the pattern of transforming growth factor beta (TGFβ) activity in AC, which is essential in the regulation of AC homeostasis. Specifically, TGFβ activity is concentrated in the areas of AC with high mechanical stress. A high level of TGFβ disrupts the cartilage homeostasis and impairs the metabolic activity of chondrocytes. Mechanical stress stimulates talin-centered cytoskeletal reorganization and the consequent increase of cell contractile forces and cell stiffness of chondrocytes, which triggers αV integrin–mediated TGFβ activation. Knockout of αV integrin in chondrocytes reversed the alteration of TGFβ activation and subsequent metabolic abnormalities in AC and attenuated cartilage degeneration in an OA mouse model. Thus, SB structure determines the patterns of mechanical stress and the configuration of TGFβ activation in AC, which subsequently regulates chondrocyte metabolism and AC homeostasis.

1. The importance of talin in the mechanosensory system of chondrocytes cannot be proven without conditional KO mice. The authors should show the phenotypes of chondrocyte-specific talin deficient mice.
2. Because the human sample is available, the authors should provide the data of talin expression in chondrocytes isolated from human articular cartilage. 3. The authors should check the synovial fluid levels of TGFβ from the human and the mice in this study.
Reviewer #2 (Remarks to the Author): In this study, the authors establish a relationship between subchondral bone architecture, mechanical stress in the articular cartilage, integrin activation in chondrocytes, and the disruption of TGFb homeostasis. The study includes an impressively large range of experimental systems and methods. However, essential information about experimental details is often lacking. Specifically, I found not details on what a structural model index or a trabecular pattern factor is, or how the bone surface/volume fraction was measured. The data base for Fig. 1 is unclear (how many cells from how many fields of view from how many bone samples) and should be specified in the figure legend or directly in the figures. Information like "n=10" is insufficient as it is unclear what "n" refers to. Scale bars for the µCT images are missing. Also for many of the other measurements in the following figures, it is unclear how many independent experiments were conducted, how many cells were analyzed ect. The scale bar in Fig. 2a is missing. The color bar for Fig. 2b is difficult to read, and perhaps the authors could consider using the same scale for sham and ACLT operated mice so that the differences become more clear. It is not clear to me how an increased bone roughness results in an increased stress and strain, and how this may affect the average stresses and strains over a larger region. It is also not clear to me how the load of 10 µN was distributed over the cartilage surface, or how the authors establish that stresses were generally increased in the anterior and interior regions, and generally reduced in the center region. Were the pSmad2/3 immunofluorescence data from the individual samples (n = ?) paired with the corresponding µCT/FEA data? It is not clear how shear stress was applied to cultured cells. The authors state "120 rpm", but what item was rotated at 120 rpm and what shear stress that translates to is not specified.

Reviewer #3 (Remarks to the Author):
This is a very interesting paper indicating that activation of TGF-beta is altered in OA cartilage due to changes in mechanical forces in the cartilage due to changes in the subchondral bone. The pare indicates that cartilage stays healthy by the activation of a "normal" level of TGF-beta and that changes outside of this range will lead to disease. The work appears to be carried out reliable although some details are missing in the M & M sections about the used procedures. The paper contributes to our understanding of the enigmatic role of TGF-beta in osteoarthritis.
The authors indicate that change in te subchondral bone lead to changes in TGF-beta activation. The question remains how these changes in the subchondral bone are initiated. This should be discussed.
Immortalized human chondrocytes-SV40 were purchased from Applied Biological Systems. What was the phenotype of the cells used. Do these cells still have a chondrocyte phenotype compared to freshly isolated human chondrocytes. These cells should be characterized better.
A major function of TGF-beta in chondrocytes is inhibition of hypertrophy. Data on synthesis of aggrecan in articular chondrocytes are less clear, in contrast to effects on stem cells, and it is well known that T^GF-beta inhibits IGF-I signaling. A factor crucial for aggrecan synthesis by chondrocytes. In my opinion this should be mentioned in the introduction.
"Indeed, superposition of mechanical compression impairs the anabolic effect of TGFβ on chondrocyte matrix production". Is this a well-accepted phenomenon. IS this not a result of te fact that mechanical compression induces TGF-beta synthesis and activation. It is not well described how human cartilage was categorized. What were the criteria to categorize these samples? What was the size of the cartilage sections examined? How was the normal cartilage determined to be normal? Typo in line 252.
Line 325. Another explanation could be that normal TGF-beta activation is inhibited and that this lead to the observed histological changes.
Line 33. "In this study, we show that temporal-spatial activation of 334 TGFβ to maintain TGFβ activity within an appropriate range is essential for the maintenance of 335 AC metabolic homeostasis and structural integrity;". This is not really demonstrated in this study but has been demonstrated by others. This study shows how aberrant TGF-beta activation might lead to pathology.
Mice were treated with tamoxifen. Not clear how tamoxifen was administered.
In the shear stress experiments. Could the cells in the center be used as controls?
Line 611. "We determined the concentration of active TGF-β1 in the 612 conditioned media by the ELISA". Are these samples activated by acid or heat or not before measurement?
We would like to thank the reviewers for their thoughtful and constructive comments regarding our manuscript. We have addressed all of their concerns and questions brought forth through additional experimentation and clarification. The changes are marked at the left margin in the manuscript.

Reviewer #1 (Remarks to the Author):
In this manuscript, the authors investigated how subchondral bone alterations affect articular cartilage homeostasis. The authors proposed that mechanical stress stimulates talin-centered cytoskeletal reorganization and the consequent increase of cell contractile forces and cell stiffness of chondrocytes, which triggers αV integrin-mediated TGFβ activation. This manuscript is potentially interesting because this is the first report showing that talin has a key role in the mechanosensory system of chondrocytes. However, the present study has critical concerns. Importantly, the authors did not use chondrocytespecific talin deficient mice. In addition, although the human sample was available, the authors did not provide the data of talin expression in chondrocytes isolated from human articular cartilage.

Response:
We designed additional experiments to validate the role of talin1 in OA animal models by knocking down the expression of talin1 in chondrocytes through intra-articular injection of talin-1 siRNA. Particularly, immunostaining of talin1 in human articular cartilage to characterize the expression of talin1 in human specimens. Detailed descriptions have been provided below.
Major concerns: 1. The importance of talin in the mechanosensory system of chondrocytes cannot be proven without conditional KO mice. The authors should show the phenotypes of chondrocyte-specific talin deficient mice.

Response: [Redacted]
Because this is the only talin-1 flox mouse strain available, we performed intra-arti cular injection of Talin1 siRNA to knockdown the expression of talin1 in chondrocytes. The in vivo administration of siRNA is a very mature technique and has been tested in several clinical trials 1,2 (https://clinicaltrials.gov/ct2/show/NCT01591356; https://clinicaltrials.gov/ct2/show/NCT00716014). Before the intra-articular injection of talin1 siRNA to the OA animal model, we performed the dosedependent and time-dependent experiment to determine the optimal dose, frequency, and treatment window. We found that the expression of talin in chondrocytes was significantly inhibited when 10 -2 nM of siRNA was administrated intraarticularly every three days. These results validate that the intra-articular injection of si-RNA successfully mimics the genetic deletion of talin1 in the mouse model with better translational potential (Fig A). Our in vitro experiments already showed that siRNA talin inhibited of αV integrin-RGD bond under mechanical stress (Fig 4c in the manuscript). In the revised manuscript, we further tested the role of talin in mediating mechanical stress-induced TGFβ activation in the OA mouse model. We found that suppressing the expression of talin by intra-articular injection of talin siRNA significantly downregulated the TGFβ signaling in the articular cartilage of mice that subjected to ACLT surgery ( Fig. 4i-k in the revised manuscript). This phenomenon was not observed in the sham operated mice. The finding validates that talin is an essential component in the process of αV integrin mediated TGFβ activation in articular cartilage particularly under mechanical stress.

2.
Because the human sample is available, the authors should provide the data of talin expression in chondrocytes isolated from human articular cartilage.
Response: In the revised manuscript, we performed immunohistochemical staining of Talin1 in the human articular specimens. Our results demonstrated that the talin was expressed by the articular cartilage chondrocytes of both OA-M and OA-S specimens with slightly increased expression in the OA-S specimen (Fig B,  supplemental Fig. 7). The results suggest that possibly both activation and increased expression of talin play a role in αV integrin-mediated TGFβ activation during the progression of OA. This finding is consistent and complementary to our in vitro experiments that siRNA talin inhibited its function to enhance the strength of αV integrin-RGD bond under mechanical stress (Fig 5g).

The authors should check the synovial fluid levels of TGFβ from the human and the mice in this study.
Response: Because of the extremely limited synovial fluid in the mouse knee joint, we were not able to draw synovial fluid from mouse knee joints. We therefore measured the levels of TGFβ in the lavage fluid that collected from ACLT mice and sham-operated mice to show TGFβ in the synovial fluid by performing ELISA assay. We found that the levels TGFβ in lavage fluid of the synovial capsule in ACLT mice were significantly elevated relative to that of shamoperated mice (Fig. C).

In this study, the authors establish a relationship between subchondral bone architecture, mechanical stress in the articular cartilage, integrin activation in chondrocytes, and the disruption of TGFβ homeostasis. The study includes an impressively large range of experimental systems and methods. However, essential information about experimental details is often lacking. Specifically, I found no details on what a structural model index or a trabecular pattern factor is, or how the bone surface/volume fraction was measured.
Response: All of the 3D structural parameters of CT analysis were generated by CTAn software (Bruker, Kontich, Belgium) based on the high-resolution CT images. The structural model index (SMI) is widely used to measure rods and plates in trabecular bone 3 . It exploits the change in surface curvature that occurs as a structure varies from spherical (SMI=4) to cylindrical (SMI=3) to planar (SMI=0). It has been shown previously that subchondral trabecular rod loss and plate thickening are essential indicators for the development of osteoarthritis 4 . The trabecular pattern factor (Tb.pf) is a μCT parameter that has been defined to reflect the connectedness of trabeculae 5 . The basic idea is that the connectedness of structures can be described by the relation of convex to concave surfaces. A lot of concave surfaces represent a wellconnected spongy lattice, whereas a lot of convex surfaces indicate a badly connected trabecular lattice.
Our previous studies showed that the connectivity of trabecular bone in the subchondral bone of the OA animal model and the human specimen was significantly decreased 6 . The Bone surface/volume fraction (BS/BV) is the ratio of the bone surface area to the volume of mineralized bone (bone microarchitecture analysis manual, https://analyzedirect.com/documents/BMA_Manual.pdf). Elevated BS/BV may indicate less well-connected bony tissues. We therefore used these parameters to reflect the micro-architecture of subchondral bone from different angles. The description of these parameters has been added into the method sections of the revised manuscript.
The database for Fig. 1 is unclear (how many cells from how many fields of view from how many bone samples) and should be specified in the figure legend or directly in the figures. Information like "n=10" is insufficient as it is unclear what "n" refers to. Scale bars for the µCT images are missing. Also for many of the other measurements in the following figures, it is unclear how many independent experiments were conducted, how many cells were analyzed, etc. The scale bar in Fig. 2a is missing.
Response: For the quantitative analysis of all the immunostaining, we firstly averaged the positive cell numbers in 3 randomly selected fields of view for each specimen. The final values shown in the bar chart are the average value of all specimens. The n value represents the sample size (how many specimens) that have been used in the experiments. For the in vitro experiments, the data were generated based on three independent repeats. These descriptions have been added into the figure legends or method section accordingly. The Scale bars for the μCT images have been added. Fig. 2b is difficult to read, and perhaps the authors could consider using the same scale for sham and ACLT operated mice so that the differences become more clear. It is not clear to me how an increased bone roughness results in increased stress and strain, and how this may affect the average stresses and strains over a larger region. It is also not clear to me how the load of 10 µN was distributed over the cartilage surface, or how the authors establish that stresses were generally increased in the anterior and interior regions, and generally reduced in the center region.

Response:
We have re-constructed Fig 2b and modified color bar with the same scale for sham and ACLT groups accordingly. Increased roughness would result in localized stiffening of the underlying subchondral bone, which would in turn result in stress concentrations in the overlying articular cartilage and uneven distribution of strain, as some areas of the cartilage would experience more stress/strain than before while other will experience lower stress/strain. During the simulations, all nodes on the top surface of the articular cartilage were loaded by an equal amount that summed to be 10μN in total transmitted through the joint. As stress/strain plots are displayed on the same color scale, comparing the color between sham and ACLT at the corresponding regions can be representative of the stress/strain values. We chose a 10μN as the loading value because it generated an average deformation of about 10% (an average strain of 0.1 along axial direction) of the original cartilage thickness, which is on a similar scale as other papers 7,8 .
Were the pSmad2/3 immunofluorescence data from the individual samples (n = ?) paired with the corresponding µCT/FEA data?
Response: The pSmad2/3 immunofluorescence data were generated from three independent specimens. whereas the FE model was established based on the μCT image of one of the specimens. The description has been added in the method sections of the revised manuscript.
It is not clear how shear stress was applied to cultured cells. The authors state "120 rpm", but what item was rotated at 120 rpm and what shear stress that translates to is not specified.

Response:
The shear stress across the cells on the periphery of the dishes was calculated as τmax=a√ρη(2πf)3, where a is the radius of orbital rotation (1.75 cm), ρ is the density of the medium (1.0 g/ml), η is the viscosity of the medium (7.5 × 10−3 dynes·s/cm2), and f is the frequency of rotation (rotations/second). Using this equation, shear stress of 6.58 dynes/cm2 is achieved at a rotating frequency of 118 rpm. The calculation method has been described in the method section of the manuscript.
Reviewer #3 (Remarks to the Author): This is a very interesting paper indicating that activation of TGF-beta is altered in OA cartilage due to changes in mechanical forces in the cartilage due to changes in the subchondral bone. The pare indicates that cartilage stays healthy by the activation of a "normal" level of TGF-beta and that changes outside of this range will lead to disease. The work appears to be carried out reliable although some details are missing in the M & M sections about the used procedures. The paper contributes to our understanding of the enigmatic role of TGF-beta in osteoarthritis.
The authors indicate that change in the subchondral bone leads to changes in TGF-beta activation. The question remains of how these changes in the subchondral bone are initiated. This should be discussed.
Response: In the previous study 6 , we found that osteoclastic bone resorption was significantly elevated in the subchondral bone at the early stage of osteoarthritis. As a result, excessive active TGF-beta was liberated and accumulated in the bone marrow cavity. The mesenchymal stem cells and the osteoprogenitors clustered in the bone marrow cavity and resulted in aberrant bone formation because they can't be recruited to the bone resorption site following the normal TGF-beta gradient. We further validated our findings by conditionally knocking out the TGF-beta type II receptor in the mesenchymal stem cells. We found that the aberrant bone formation and structural changes in the subchondral bone and the degeneration of articular cartilage were significantly attenuated in the OA mouse models. These findings were published in Nature Medicine (2013) and have been discussed in the present manuscript.