A single-cell transcriptional atlas reveals resident progenitor cell niche functions in TMJ disc development and injury

The biological characteristics of the temporomandibular joint disc involve complex cellular network in cell identity and extracellular matrix composition to modulate jaw function. The lack of a detailed characterization of the network severely limits the development of targeted therapies for temporomandibular joint-related diseases. Here we profiled single-cell transcriptomes of disc cells from mice at different postnatal stages, finding that the fibroblast population could be divided into chondrogenic and non-chondrogenic clusters. We also find that the resident mural cell population is the source of disc progenitors, characterized by ubiquitously active expression of the NOTCH3 and THY1 pathways. Lineage tracing reveals that Myh11+ mural cells coordinate angiogenesis during disc injury but lost their progenitor characteristics and ultimately become Sfrp2+ non-chondrogenic fibroblasts instead of Chad+ chondrogenic fibroblasts. Overall, we reveal multiple insights into the coordinated development of disc cells and are the first to describe the resident mural cell progenitor during disc injury.

The team uncovered 4 cell types, whereas 2 cell types were previously historically known as fibroblasts and chondrocyte-like cells. However it is unclear whether chondrocytes were discovered in their analyses as chondrocytes are not mentioned. This is an important point to make in defining cell types and distinguishing the disc from the condyle and also for reconciling previous studies to this new scRNAseq data.
The majority 88% of the disc population are fibroblasts. This population is highly studied in regenerative medicine/tissue engineering. While a section of the study describes the 6 FB clusters, the authors missed the opportunity to clearly define and name those FB populations, their markers, and describe their speculated functions. Furthermore, since multiple time points are collected authors should expand on how they propose each FB population contributes to disc development, maintenance and possible age-dependent disease. A proposed model and/or table of the different FB populations at varying timepoints and their putative functions would be highly valuable to the field.
The description of mural cells as a both a niche and as progenitor cells is a bit confusing. Do the authors mean perivascular niche? If the authors argue MCs are niche cells, which cells do they provide a niche for? Please clarify.
The title reads a bit awkward. It's unclear if the authors are referring to the perivascular location as the niche or the NOTCH3+/THY1+ MCs themselves. NOTCH3+/THY1+ MCs as both a niche cell and progenitor cell is confusing. MC/progenitor cell function during disc development was not directly validated in this study, which would require loss of function studies. Unless authors can validate developmental function during development, its recommended to remove from the title.
In Figure 3c and 3f, immunostainings for PCAM1 and C1QA are shown separately and not co-localized as described in results section line 552. TMJ disc tissue is quite thin and cell constituents can vary dependent on different sections. Its difficult to conclude ECs and Macrophages work together in the anterior band without showing the two populations in the same section. It seems that PCAM1 is localized solely in the disc lining the superior joint cavity, whereas C1QA is found in both super/inferior joint cavity. How does this specific location correlate to their proposed function and relationship to each other? Figure 4 is very interesting. However, what type of "functional disc cells" do MCs become? Please define "functional cell". This goes back to the question of what is a disc FB?
In Figure 5i-5K. The histology and pentachrome/chemical staining for bone and cartilage is not very convincing. In 5i Thy+ pellets seems highly variable in expressing Acan and pellet staining is unconvincing. In Fig 5K THY+ cells, the cartilage area (presumably blue color) appears to be missing and the lowest panel that shows "bone" looks more like fibrous tissue. It seems likely that THY+ cells have an inclination toward bone but not cartilage. More markers and/or better sections need to be shown to support the conclusions that THY+ are multipotent. To this end, it is unclear how quantification in Fig. 5K of the bone/cartilage/stroma was performed with the unconvincing histology. Please clarify the quantification.
The Parabiosis model is not sufficiently described in the written results section.
The authors should consider using more recent nomenclature and change the term "mesenchymal progenitors" to "skeletal progenitors" to align with current skeletal stem/progenitor cell biology.
In Fig 8: The Myh11-CreER mouse model needs to be validated in the TMJ disc by In situ or IHC for Mhy11. Please add pulse/chase timeline to figure since results and interpretation are highly dependent on recombination timing. The disc injury model is new and seems very technical for a small animal model. Its not clear how reproducibility was controlled. Please clarify surgical standardization methods. Moreover, the injury in Figure 8g-8i seems very small and its unclear how the location of the injury was determined via histology vs tissue artifact. The injury is in the middle and distant from the anterior region. Why was this location chosen over anterior region where most of the MCs/progenitor cells were located? Were the authors testing secreted factors to encourage migration/differentiation to injury site? Please choose another pseudo-color for NOTCH given yellow also denotes overlap of green/red signal.

Fig 5 shows
MCs have a propensity to make bone but how does that property relate to their function? Is bone found in the disc in disease or injury? During injury do Myh11-expressing Notch+ Thy1+ cells transition to bone cells? It seems there's a disconnect between the experiments that need some finetuning.
The discussion was ok. It seems redundant of the results. More effort should be made summarize the 4 populations and their respective subpopulations and to provide speculation on their respective functions as it relates to disc tissue development, homeostasis and disease.
While this study focuses on one progenitor cell population among MCs, it seems this is a very small population with not that many progeny per lineage tracing experiment. Thus there are likely other progenitor cell populations that give rise to other mature cell types that need to be discussed here. A discussion of TMJ disc progenitor cells in the context of skeletal stem/progenitor cells would be warranted here and broaden the scope of this work.
Please discuss how this work would improve patient care/diagnoses as suggested in intro. It may also be helpful to discuss how these newly defined cell populations may shift during aging and disease.
Reviewer #2 (Remarks to the Author): In this manuscript, Bi et al. performed Single-cell RNA-Seq of the TMJ disc and demonstrated that TMJ disc tissue contains 9 distinct clusters and 4 types of cells: fibroblasts, endothelial cells, macrophages and mural cells. The authors showed spatial-temporal heterogeneity of fibroblast cells throughout TMJ disc tissue during postnatal development and aging and determined the role of macrophages and endothelial cells in TMJ development. The authors also demonstrated that the Notch/Thy1 pathway are active in mural cells during TMJ postnatal development and aging and characterized mesenchymal progenitor capacities of Thy1+/Notch3+ mural cells. Overall, this is an interesting study revealing the functional role of 4 types of cells in TMJ disc tissue. However, some questions, listed below, need to be addressed. Figure 6c, the thresholds of flow cytometry should be kept in consistent. Please repeat this experiment.

1)For
2)The authors reported that diameters of the TMJ discs were increased at different postnatal stages. Does the cell size of TMJ disc also increase at different postnatal stages?
3)Please elaborate the relationship between distribution of different fibroblast clusters and ECM features, especially the different types of collagen. It would be interesting to detect different fibroblast clusters by IHC or IF in TMJ disc tissues. Figure 3c and 3d showed that macrophages were gradually decreased comparing the newborn mice to the aged mice. Have the authors analyzed macrophage polarization in this process? It is not cleat if M1 and M2 markers are both decreased from the newborn stage to the aged stage.

4)
5)It has been shown that Notch signaling participates in the onset and development of TMJ-OA and inhibition of Notch signaling pathway temporally postpones the cartilage degradation of TMJ arthritis in mice (DOI: 10.1016/j.jcms.2018.04.026). In this study, the authors found that Notch signaling was activated at all postnatal stages in mural cells. Are these mice suffering from spontaneous TMJ-OA? Please discuss more about Notch signaling in TMJ-OA development.
Reviewer #3 (Remarks to the Author): The paper "Single-cell transcriptional atlas revealed that the resident progenitor cell niche functions in TMJ disc development and injury" is well executed and clearly written.

Reviewer #1 (Remarks to the Author):
This study from the Zhu lab aims to delineate a comprehensive roadmap of TMJ disc cells at various post-natal ages in mice. The study team uses singlecell RNAseq to identify 4 cell types, including fibroblasts, endothelial cells, macrophages and mural cells, with 9 distinct clusters. Most notably, NOTCH3+/THY1+ MCs were identified to be progenitor cells that self-renew and have multi-potent ability both in vitro and in vivo. The authors use a TMJ disc injury model and lineage tracing to argue that MCs differentiate into other disc cell types during repair.
Overall this is an important study that addresses a critical medical problem.
TMJ biology is severely understudied and treatment for TMDs are limited. The sequencing data provides key information that would advance the understanding of heterogenous TMJ disc cell types and thus represents a significant advance to the field. The computational analyses are comprehensive.
However there are several weaknesses that need to be addressed.
The team uncovered 4 cell types, whereas 2 cell types were previously historically known as fibroblasts and chondrocyte-like cells. However it is unclear whether chondrocytes were discovered in their analyses as chondrocytes are not mentioned. This is an important point to make in defining cell types and distinguishing the disc from the condyle and also for reconciling previous studies to this new scRNAseq data. This is a very important comment. We totally agree that identification of chondrocyte-like cells in our scRNA-Seq data would be very crucial for establishing unified theoretical system of TMJ disc types from previous studies and our scRNA-Seq results. To further clarify diverse roles of FB clusters at different postnatal stages, we chose our 3d scRNA-seq data, which is the most active stage for cell development and differentiation, and used CytoTRACE (v.0.3.3) to predict differentiation states of different FB clusters at this stage (8). We found that the non-chondrogenic fibroblast 2 (proliferation) cluster showed the highest CytoTRACE value, which was estimated with optimum developmental potential (8). On the other hand, we also found the chondrogenesis related FB1 and FB2 were at the relatively later order of CytoTRACE value ( Figure S4a-e). These findings enlighten us again that the chondrocyte-like FBs in TMJ disc are terminally differentiated cells with limited capacity of self-repair and being associated with aging and injury (9,10).
The description of mural cells as a both a niche and as progenitor cells is a bit confusing. Do the authors mean perivascular niche? If the authors argue MCs are niche cells, which cells do they provide a niche for? Please clarify.
The reason of describing MCs as niche cells is because we found that MCs have active signaling crosstalk with other cell types. We totally agree this speculation lacks solid evidence and made our description of MCs functions confusing. Therefore, we deleted the description of 'cell niche provided by MCs'.
Thank you for this thoughtful comment. Followed your suggestion, the title has been changed as follows: Single-cell transcriptional atlas revealed that the resident progenitor cells function in mouse temporomandibular joint disc injury.
In Figure 3c and 3f, immunostainings for PCAM1 and C1QA are shown separately and not co-localized as described in results section line 552. TMJ disc tissue is quite thin and cell constituents can vary dependent on different sections. Its difficult to conclude ECs and Macrophages work together in the anterior band without showing the two populations in the same section. It seems that PCAM1 is localized solely in the disc lining the superior joint cavity, whereas C1QA is found in both super/inferior joint cavity. How does this specific location correlate to their proposed function and relationship to each other?
Our speculation that 'The colocalization of C1QA + MPhs and PECAM-1 + ECs in this area suggests that a close interaction between ECs and MPhs may contribute to the modulation of vascular function and the inflammatory response' is based on the relatively closed spatial relationship between the two cell types in the anterior band, as well as previous reports considering signal interactions between MPhs and ECs may participate in immune response tissue injury (11,12). However, we totally agree with your concern that our current evidence is very limited to support this speculation, a series of experiments and much more comprehensive studies are needed to clarify the signal and functional connection between these two cell types. Since our current study is focusing on the functional role of mural cells as progenitors in the disc, we deleted the confusing sentence and leave this speculation to be elucidated in future work. 'Functional cells' was used to define cells that participate in disc tissue repair, which cells were supposed to be part of FBs. To verify this speculation, we performed combined FISH and antibody staining to co-localize different FBs clusters and Myh11 + MCs lineage. Interestingly, we found that when Myh11 + MCs lineage migrated toward disc injury site, considerable number of MCs lineage started to express Sfrp2 (RFP + / Sfrp2 + ), but much fewer RFP + /Chad + cells were observed (Figure 8l). In addition to our other experiments that showed RFP + MCs lineage loss its original specific expressions of NOTCH3, THY1 and Myh11 during disc injury repair (Figure 8a-k, S9a-c), we conclude that MCs were able to differentiate toward fibroblasts for mouse TMJ disc injury repair, specifically transformed toward non-chondrogenic fibroblasts instead of chondrogenesis related cells.
In Figure 5i-5K. The histology and pentachrome/chemical staining for bone and cartilage is not very convincing. In 5i Thy+ pellets seems highly variable in expressing Acan and pellet staining is unconvincing. In Fig 5K THY+ cells, the cartilage area (presumably blue color) appears to be missing and the lowest panel that shows "bone" looks more like fibrous tissue. It seems likely that THY+ cells have an inclination toward bone but not cartilage. More markers and/or better sections need to be shown to support the conclusions that THY+ are multipotent. To this end, it is unclear how quantification in Fig. 5K of the bone/cartilage/stroma was performed with the unconvincing histology. Please The strategy of Movat pentachrome staining is based on previous studies (13)(14)(15). Previous studies define collagen + area with yellow color as bone tissue, mucin + area with blue color as cartilage tissue and fibronectin + area with red color as stroma/marrow. To be more cautious about our occlusion, we switched the definition of 'bone', 'cartilage' and 'stroma/marrow' to 'collagen', 'mucin' and 'fibronectin'.
We have added immunostaining of SOX9 and Aggrecan on the THY1 + / THY1pellets (Figure6 i-j). Consistent with the pentachrome staining results, THY1 + disc cells showed a significantly stronger cartilaginous capacity than THY1disc cells under chondrogenic transduction, manifesting as higher expression of Aggrecan and Sox9. In the ex vivo transplantation of THY1 + MCs under the renal capsules of NOD-SCID mice, four-week grafts of THY1 + MC populations formed more collagen (yellow), fibronectin (red) and mucin (blue), however, there was no mature chondrocytes found in the renal capsule ossicles.
Therefore, we speculate that it is hard for MCs to form mature cartilage without chondrogenesis induction, suggests a limited chondrogenesis capacity of MCs for disc fibrocartilage repair.
The Parabiosis model is not sufficiently described in the written results section.
Followed your suggestion, we described the parabiosis model as follows: To rule out the possibility of a circulating source of disc tissue repair, we used a parabiosis model that generated circulation between GFP + mice and TMJ injured non-GFP mice ( Figure S7a). We found there was no circulating GFP + cells participating in joint disc repair ( Figure S7b-c) The authors should consider using more recent nomenclature and change the term "mesenchymal progenitors" to "skeletal progenitors" to align with current skeletal stem/progenitor cell biology.
According to your advice, we have changed the 'mesenchymal progenitors' to 'skeletal progenitors'.
In In our pilot study, we randomly selected histological sections from individual mice and performed HE staining, finding that the injury sites were located at similar areas in the anterior band (red rectangle), which showed that under our standardized surgical process, the TMJ disc injury model has good reproducibility.
The reason of choosing this area as puncture site (in the middle of anterior band instead of the area near attachment) is because we do not want to incidentally damage the blood vessels within the attachment during puncture, which may bring exogenous progenitors from circulation, leading to interference for our observation.
According to your advice, we have changed the pseudo-color for NOTCH from yellow to magenta. Thank you. TMJ disc calcification, although was not a common disease, were still described by several previous studies (16)(17)(18). At the early stage of disc calcification, it usually manifests with asymptom and is difficult to be found, but it may lead to restricted joint function at later stage, including joint clicking, TMJ pain, and limitation of mouth opening.
The etiology of TMJ disc calcification remains unknown. Jibiki et al. (16) revealed that the incidence of disc calcifications was particularly high in cases of perforated discs and from elder human cadavers, suggesting inflammatory microenvironment, excessive mechanical stress and aging could be potential cause of the articular disc calcification. We agreed that MCs could not be the only progenitor source in the TMJ disc.
In our lineage tracing experiments, we found that the Myh11-CreER+ MCs appears to differentiate toward Sfrp2 + non-chondrogenic fibroblasts instead of Please discuss how this work would improve patient care/diagnoses as suggested in intro. It may also be helpful to discuss how these newly defined cell populations may shift during aging and disease.
Followed your suggestion, we have discussed the potential significance for TMD clinical diagnosis and treatment: The Sfrp2 + FBs, which cells were considered as potential functional cells for disc tissue repair, were also found mainly located in this area. This finding highlights the importance of homeostatic anterior attachment for disc function and improve the understanding of the vascular cell niche within the anterior band attachment area. More importantly, these findings suggest that this area could be a critical site for early clinical diagnosis and for specific target therapy of TMJ disc disarrangement.
Reviewer #2 (Remarks to the Author): In Please repeat this experiment.
In our flow cytometry experiments, we set blank group for each individual experimental group at each time point, and the threshold of each group at each timepoint is according to different blank that may has various threshold. This is for minimizing the effect from autofluorescence from different cells we prepared at each time. Followed your suggestion, we chose the plots with similar gate threshold at each timepoint as our representative images for easier reading.
2)The authors reported that diameters of the TMJ discs were increased at different postnatal stages. Does the cell size of TMJ disc also increase at different postnatal stages?
We analyzed the diameter of disc cells at different stages using pentachromic staining sections. We found that there was no significant difference of cell diameter between different stages. However, when we analyzed cell density at different stages using ImageJ, we found that cells were significantly denser at 3d than other stages, which suggests that the increased size of TMJ disc during postnatal development were mainly attributed to the accumulation of extracellular matrix.  (Figure 2e). When we further compared expressions of these chondrocyte related markers between FB subclusters, we found that expressions of these genes were mainly enriched in FB5 and FB6 (Figure 2f). Guided by GO analyses, FB5 and FB6 were also found closely related to chondrocyte differentiation and ossification ( Figure   S2b). Therefore, we defined these 2 FB clusters as chondrogenesis related  Representative flow cytometry plots of the percentage of m1/m2 macrophages in TMJ discs. (a-b) Gating strategies for the analysis of mouse TMJ disc single cells isolated from C57BL/6 mice at different stages (3 d, 3 w, 16 w, 78-82 w), FSC-A/SSC-A was used to distinguish between cell groups and exclude cell debris, and FCS-A/FSC-H was used to remove adhesion cells and screen out single cells. (c) Representative graphs of the percentages of CD86 + cells or CD206 + cells at different stages is presented (right, n = 2-3 biological replicates per group,). The proportion of CD86 + cells (dot on the right side of the gate)and the proportion of CD206 + cells (dot on the upper side of the gate) gradually decreased with age, while there was a rebound tendency at 78-82w group. (d-f) Proportion change of different types of macrophages at different stages. Each dot represents a biological duplication, one-way ANOVA followed by multiple comparisons were used for statistical analysis. Followed your advice, we have expanded our Discussion about NOTCH signaling in TMJ homeostasis and OA: NOTCH signaling pathway was found with a dual role during joint cartilage metabolism, as well as being identified as a potential regulator of both catabolic and anabolic molecules in the cartilage ECM during development (25)(26)(27).
Transient activation of NOTCH signaling in postnatal chondrocytes results in increased synthesis of cartilage ECM and joint maintenance, while overexpression of NOTCH signaling activates the pathway in OA cartilage (28).
In TMJ, our previous study showed that NOTCH signaling was excessively activated during the onset and development of TMJOA, while partial blocking NOTCH signaling by preventing NICD release could alleviated the cartilage destruction (29). In addition, NOTCH activation was found critical for chondrogenic progenitor specifications in condylar fibrocartilage (30). These findings, in addition to our current study showing that NOTCH is active at all postnatal stages in TMJ disc, suggest potential functions of NOTCH both as the 'identity' and the regulator of MCs characteristics during TMJ cartilage development and injury repair.
Reviewer #3 (Remarks to the Author): The paper "Single-cell transcriptional atlas revealed that the resident progenitor cell niche functions in TMJ disc development and injury" is well executed and clearly written.
Thank you very much for your praise for our work. As for TMJ, since it's a proprietary abbreviation in the craniofacial research field, we would love to keep TMJ for more succent expression.
*Please indicate how many animals or cells from individual animals were used in each set of data. It is hard at this point to evaluate rigour.
Followed your suggestion, we have added animal numbers for each experiment in our figure legend, therefore reviewers and readers can clearly know the exact number of animals used in each set of data. Thank you.