Increased autophagy in EphrinB2-deficient osteocytes is associated with elevated secondary mineralization and brittle bone

Mineralized bone forms when collagen-containing osteoid accrues mineral crystals. This is initiated rapidly (primary mineralization), and continues slowly (secondary mineralization) until bone is remodeled. The interconnected osteocyte network within the bone matrix differentiates from bone-forming osteoblasts; although osteoblast differentiation requires EphrinB2, osteocytes retain its expression. Here we report brittle bones in mice with osteocyte-targeted EphrinB2 deletion. This is not caused by low bone mass, but by defective bone material. While osteoid mineralization is initiated at normal rate, mineral accrual is accelerated, indicating that EphrinB2 in osteocytes limits mineral accumulation. No known regulators of mineralization are modified in the brittle cortical bone but a cluster of autophagy-associated genes are dysregulated. EphrinB2-deficient osteocytes displayed more autophagosomes in vivo and in vitro, and EphrinB2-Fc treatment suppresses autophagy in a RhoA-ROCK dependent manner. We conclude that secondary mineralization involves EphrinB2-RhoA-limited autophagy in osteocytes, and disruption leads to a bone fragility independent of bone mass.

This study aims to determine the role of osteocytic ephrinB2 on bone mineralization. In addition to investigating mineralization patterns in the ephrinB2 deficient mice, the authors performed RNA sequencing and showed differential expression of autophagyassociated genes. This led them to conclude that the process of secondary mineralization uses the autophagy machinery in an ephrinB2-dependent manner. The data appear solid and the description is clear. However, the connection between mineralization defects and autophagy is missing, and no attempt was made to determine whether autophagy dysregulation is responsible for the mouse phenotype or whether modulating autophagy at least in vitro has any consequences on the phenotype of eprhinB2-deficient osteocytes. Without this evidence the data on autophagy is not relevant for the manuscript and it should be removed, and the title should be modified accordingly. Figure 7, panels B and C), (2) increased mineralization in cultured EphrinB2-deficient osteocytes ( Figure 8). This data provides compelling evidence that strongly supports our model proposed in the initial submission. This new data is described at the end of the results section (p12-14), and in some changes to the discussion (p18,19).

new Figure 7, panels D and E), and (3) abundant autophagosomes and matrix vesicles in EphrinB2 deficient osteocytes in vivo (new
Additional issues: 1-The first sentence of the introduction should be modified, since the skeleton is unique not only to the human body, and the authors are actually using a mouse model in their studies. RESPONSE: We have removed the phrase "in the human body" from this sentence.
2-References should be provided for the statement regarding the process of mineralization in the first paragraph of the Introduction. RESPONSE: We have added the following two references to this statement: • 3-Second paragraph of the introduction, lines 77-79: the statement should be made more general since not only PTHrP, but also PTH (teriparatide) are used to stimulate bone formation. RESPONSE: We have clarified this statement in the new version of the manuscript, as follows: "This PTHR1-mediated action is exploited by the pharmacological agents teriparatide (PTH) and abaloparatide (modified N-terminal PTHrP), the only pharmacological agents currently available that can increase bone mass in patients with fragility fractures 12,13 ." 4-Minor point: the reviewer cannot find the "full data availability" statement in the manuscript.

RESPONSE: This has been added to the end of the materials and methods section (p37).
The GEO dataset will be made public at the time of publication.

Reviewer #2:
Following a previous sFTIRM study of bone matrix maturation after PTH treatment, this manuscript aims to define the role of ephrinB2 in osteocytes. It shows that osteocytespecific ephrinB2-deficient mice have a brittle bone phenotype. It also provides RNA sequencing data using an osteocyte cell line treated with ephrinB2 knockdown vector and reports that a group of autophagy-associated genes were differentially expressed. It proposes that ephrinB2 is required to restrain autophagy in osteocytes and limits the secondary mineralization process.
Although the role of osteocytes in secondary mineralization is an important area of bone biology considering the limited knowledge about the mechanisms controlling the speed of continuing mineral depositions, this study is too preliminary for multiple reasons. First, it is potentially interesting that knockdown of ephrinB2 affects autophagy-related genes, but the mechanism and significance are not clear. The statement was made that autophagic processes in osteocytes directly control mineralization, but there is no such data shown. Figure 7, panels B and C), (2)  Third, the conclusions of the FTIR measurements are only poorly supported by the data presented. In particular, time-dependent descriptions such as "brittle bone phenotype is therefore associated with more rapid bone matrix maturation, including rapid accumulation of mineral and carbonate substitution and more rapid collagen compaction within the matrix" (page 9) and "both mineral accrual and carbonate substitution within the mineral occur at an accelerated rate in the absence of ephrinB2" (page 13) appear to be overinterpretation of regional data without performing timecourse experiments. 2. Figure 4B and Figure 4E are inconsistent. Wavenumber of Amide I (1588-1712) and Amide II (1500-1600) in Figure 4E is different from the shaded Wavenumber in the FTIRM spectrum ( Figure 4B). The photo shown in Figure 4A top was previously published by the authors, but this is not mentioned in the legend. Regions should include endocortical as in Figure 5. RESPONSE: The regions shown in Figure 4B were intended as a guide only, as mentioned in the legend ("approximate regions"). To avoid confusion, we have updated the shading to more accurately represent the peaks used for analysis, and clarified the legend.

RESPONSE: We now have new data showing (1) confirmation of increased autophagy in EphrinB2 knockdown osteocytes (new
We have replaced the image in Figure 4A with a new photo from the present study. Figure 4 does not include endocortical regions, as these were not analysed by sFTIRM. This is because the endocortical surface, unlike the periosteal surface, undergoes both bone resorption and formation, and is therefore not a region that can be used to study bone matrix maturation. We have added the following sentence in the results section to explain this (p9): "Unlike the endocortical surface, which is remodeled, the periosteal surface at this location undergoes continuous bone formation (bone growth), without bone resorption 20 .".
3. The difference between w/w and f/f was observed in periosteal at 0 degree polarization ( Figure 5). Why similar differences were not observed at the endosteum or 90 degree polarizing filter? RESPONSE: No difference was observed on the endocortical surface, likely because this region is remodeled, and therefore contains a mix of both mature, and immature bone; we have added a note about this in the results section (p10): "There was no reduction in amide I:II ratio on the endocortical surface; since this region is remodeled, the bone in this region would contain both old and new bone, which may mask any difference".
We have clarified our description of the difference in result between the 0° and 90° polarizing filters on p17 as follows: "Since the reduction in amide I:II was detected under the 0°, but not the 90°, polarizing filter, the increased collagen compaction may be specific to those collagen fibers aligned along the length of the bone, rather than those with radial orientation, or transverse to the bone. This suggests the altered organization of longitudinal fibers may be the main contributor to the ephrinB2-deficient phenotype, and these fibers may be more important for absorbing mechanical forces along the length of the tibiae". Figure 2) are essentially the same information as the Stress (MPa)-Strain (%) curve and related parameters (Figure 3), if the shape of bones were not altered between w/w and f/f as the authors claim. RESPONSE: Yes, and this is why it is important to show both stress-strain and loaddeformation data. The load-deformation data ( Figure 2) indicates total strength of the bone (without correcting for differences in bone size). It is because we still see a reduction in strength after correcting for bone size (in the stress-strain data - Figure 3), that we know the defect in strength is a material defect. We have added some information to clarify this in the results section (p6-7) and in the figure legends.

The Load (N)-Deformation (mm) curve and related parameters (
5. Scale bars are missing from Figure 5A, B, D, and E. RESPONSE: Scale bars have been added. 6. Figure 6D. A break in the Y axis is necessary not to start at 0. RESPONSE: Thank you for alerting us to this. This has been corrected (now Fig. 7D).
7. Does the upregulation of ephrinB2 expression by PTH treatment reduce susceptibility of osteocytes to autophagy? RESPONSE: We thank you for this suggestion. We have not tested this and have planned to explore this in the future.
8. Is there any detectable enhancement of autophagy in osteocyte-specific ephrinB2deficient mice? RESPONSE: We have now used transmission electron microscopy to detect autophagy in the osteocyte-specific ephrinB2-deficient mice (new Figure 8). This revealed a major morphological change in osteocytes in the Dmp1Cre.Efnb2 f/f mouse model. These cells exhibited both a high level of autophagy and a greater level of matrix vesicle formation. This is described in detail in the results section (p13-14), and we have added a number of sentences throughout the discussion to fit this evidence with the model already proposed (p18,19).

Reviewer #3:
In this study, the authors found that osteocyte-specific EphrinB2 knockout mice result in fragile bones could cause brittle bone disease. This phenotype was attributed to defect in bone matrix maturation by unregulated secondary mineralization. Further, the authors revealed that autophagy is enhanced in the osteocytes lacking EphrinB2. On the basis of these results, the authors claim that ephrinB2 inhibits autophagy in osteocytes for generating the appropriate mineralization. While the presented data are of interest, some additional experiments are needed to support the author's claim.
Comments 1. The title seems inappropriate since there in no direct evidence that increased autophagy is involved in the pathogenesis of brittle bone. RESPONSE: Our title does not claim a direct relationship between increased autophagy and a brittle bone phenotype, but that they are associated.

Responses to Reviewers
Reviewer #1: This study aims to determine the role of osteocytic ephrinB2 on bone mineralization. In addition to investigating mineralization patterns in the ephrinB2 deficient mice, the authors performed RNA sequencing and showed differential expression of autophagyassociated genes. This led them to conclude that the process of secondary mineralization uses the autophagy machinery in an ephrinB2-dependent manner. The data appear solid and the description is clear. However, the connection between mineralization defects and autophagy is missing, and no attempt was made to determine whether autophagy dysregulation is responsible for the mouse phenotype or whether modulating autophagy at least in vitro has any consequences on the phenotype of ephrinB2-deficient osteocytes. Without this evidence the data on autophagy is not relevant for the manuscript and it should be removed, and the title should be modified accordingly.
In response to this comment, we now have new data showing: Additional issues: 1-The first sentence of the introduction should be modified, since the skeleton is unique not only to the human body, and the authors are actually using a mouse model in their studies. RESPONSE: We have removed the phrase "in the human body" from this sentence. 3-Second paragraph of the introduction, lines 77-79: the statement should be made more general since not only PTHrP, but also PTH (teriparatide) are used to stimulate bone formation.

RESPONSE: We have clarified this statement in the new version of the manuscript, as follows:
"This PTH1R-mediated action is exploited by the pharmacological agents teriparatide (PTH) and abaloparatide (modified N-terminal PTHrP), the only pharmacological agents currently available that can increase bone mass in patients with fragility fractures 12,13 ." 4-Minor point: the reviewer cannot find the "full data availability" statement in the manuscript.
RESPONSE: This has been added to the end of the materials and methods section (p33).
The GEO dataset will be made public at the time of publication.

Reviewer #2:
Following a previous sFTIRM study of bone matrix maturation after PTH treatment, this manuscript aims to define the role of ephrinB2 in osteocytes. It shows that osteocytespecific ephrinB2-deficient mice have a brittle bone phenotype. It also provides RNA sequencing data using an osteocyte cell line treated with ephrinB2 knockdown vector and reports that a group of autophagy-associated genes were differentially expressed. It proposes that ephrinB2 is required to restrain autophagy in osteocytes and limits the secondary mineralization process.
Although the role of osteocytes in secondary mineralization is an important area of bone biology considering the limited knowledge about the mechanisms controlling the speed of continuing mineral depositions, this study is too preliminary for multiple reasons.
First, it is potentially interesting that knockdown of ephrinB2 affects autophagy-related genes, but the mechanism and significance are not clear. The statement was made that autophagic processes in osteocytes directly control mineralization, but there is no such data shown. RESPONSE: In our initial manuscript we were very careful to state only that autophagic processes MAY directly control mineralization. Since we did not yet know which types of autophagy are modified, we suggested this as model for how mineralization may be controlled.
We have now carried out additional experiments to provide mechanistic insights to this hypothesis: 1. We have confirmed that Efnb2 knockdown increases autophagy using a second method (new Figure 6, panels B and C), and now show that EphrinB2-Fc treatment suppresses autophagy (new Figure 8E) Figure 7). 3. We also now provide mechanistic data showing that the ability of EphrinB2-Fc to stimulate autophagy is RhoA-dependent (new Figure 8E). Furthermore, RhoA inhibition, like Efnb2 knockdown, results in increased mineralization (new Figure  8A-D).
These new data strongly support the hypothesis proposed in the initial submission, that EphrinB2 suppresses autophagy, and in its absence autophagic processes are increased and contribute to the process of mineralization. The new data are described at the end of the results section (p11-14), and in some changes to the discussion (p18,19).
Second, the high osteocyte lacuna density in the absence of ephrinB2 remains descriptive in nature. Determining the mechanism of this or the causal relationship to a brittle bone phenotype or greater mineral and carbonate deposition would strengthen the manuscript. RESPONSE: We have not suggested that the high osteocyte lacunar density causes the brittle bone phenotype; such a mechanism would be very difficult to prove. In answer to this reviewer and Reviewer 3, we carried out a Ploton Silver Stain to assess the osteocyte lacunar-canalicular structure; this data is included in Supplementary Figure 2. Since this revealed no change in the dendritic network, and no change was observed by TEM, we have deleted the section of our discussion that suggested autophagy may control osteocyte maturation. We have retained the section describing the increase in osteocyte density, and added a statement that it is not clear whether this is causative or secondary to the greater mineralization (p17).
Third, the conclusions of the FTIR measurements are only poorly supported by the data presented. In particular, time-dependent descriptions such as "brittle bone phenotype is therefore associated with more rapid bone matrix maturation, including rapid accumulation of mineral and carbonate substitution and more rapid collagen compaction within the matrix" (page 9) and "both mineral accrual and carbonate substitution within the mineral occur at an accelerated rate in the absence of ephrinB2" (page 13) appear to be overinterpretation of regional data without performing timecourse experiments. 2. Figure 4B and Figure 4E are inconsistent. Wavenumber of Amide I (1588-1712) and Amide II (1500-1600) in Figure 4E is different from the shaded Wavenumber in the FTIRM spectrum ( Figure 4B). The photo shown in Figure 4A top was previously published by the authors, but this is not mentioned in the legend. Regions should include endocortical as in Figure 5. Figure 4B (now Figure 3B) were intended as a guide only, as mentioned in the legend ("approximate regions"). To avoid confusion, we have updated the shading to more accurately represent the peaks used for analysis, and clarified the legend.

RESPONSE: The regions shown in
We have replaced the image in Figure 4A (now Figure 3A) with a new photo from the present study. Figure 3) does not include endocortical regions because they were not analysed by sFTIRM. This is because the endocortical surface, unlike the periosteal surface, undergoes both bone resorption and formation, and is therefore not a region that can be used to study bone matrix maturation. We have added the following sentence in the results section to explain this (p8): "Unlike the endocortical surface, which is remodeled, the periosteal surface at this location undergoes continuous bone formation (bone growth), without bone resorption 20 .".

Figure 4 (now
3. The difference between w/w and f/f was observed in periosteal at 0 degree polarization ( Figure 5). Why similar differences were not observed at the endosteum or 90 degree polarizing filter? RESPONSE: No difference was observed on the endocortical surface, likely because this region is remodeled, and therefore contains a mix of both mature, and immature bone; we have added a note about this in the results section (p10): "There was no reduction in amide I:II ratio on the endocortical surface; since this region is remodeled, the bone in this region would contain both old and new bone, which may mask any difference".
We have clarified our description of the difference in result between the 0° and 90° polarizing filters on p16 as follows: "Since the reduction in amide I:II was detected under the 0°, but not the 90°, polarizing filter, the increased collagen compaction may be specific to those collagen fibers aligned along the length of the bone, rather than those with radial orientation, or transverse to the bone. This suggests the altered organization of longitudinal fibers may be the main contributor to the EphrinB2-deficient phenotype, and these fibers may be more important for absorbing mechanical forces along the length of the tibiae". Figure 2) are essentially the same information as the Stress (MPa)-Strain (%) curve and related parameters (Figure 3), if the shape of bones were not altered between w/w and f/f as the authors claim. Figure 1) indicates total strength of the bone (without correcting for differences in bone size). It is because we still see a reduction in strength after correcting for bone size (in the stress-strain data -now in Figure 2), that we know the defect in strength is a material defect. We have added some information to clarify this in the results section (p6-7) and in the figure legends.

RESPONSE: Yes, and this is why it is important to show both stress-strain and loaddeformation data. The load-deformation data (now in
5. Scale bars are missing from Figure 5A, B, D, and E. RESPONSE: Scale bars have been added (now Figure 4A,B,D,E).
6. Figure 6D. A break in the Y axis is necessary not to start at 0. RESPONSE: Thank you for alerting us to this. This has been corrected (now Fig. 5D).
7. Does the upregulation of ephrinB2 expression by PTH treatment reduce susceptibility of osteocytes to autophagy? RESPONSE: We thank you for this suggestion. We have not tested this and have planned to explore this in the future.
8. Is there any detectable enhancement of autophagy in osteocyte-specific ephrinB2deficient mice? RESPONSE: We have now used transmission electron microscopy to detect autophagy in the osteocyte-specific EphrinB2-deficient mice (new Figure 7). This revealed a major morphological change in osteocytes in the Dmp1Cre.Efnb2 f/f mouse model. These cells exhibit both a high level of autophagy, particularly ER-phagy and a greater level of matrix vesicle formation. This is described in detail in the results section (p13-14), and we have added a number of sentences throughout the discussion to fit this evidence with the model already proposed (p18,19).

Reviewer #3:
In this study, the authors found that osteocyte-specific EphrinB2 knockout mice result in fragile bones could cause brittle bone disease. This phenotype was attributed to defect in bone matrix maturation by unregulated secondary mineralization. Further, the authors revealed that autophagy is enhanced in the osteocytes lacking EphrinB2. On the basis of these results, the authors claim that ephrinB2 inhibits autophagy in osteocytes for generating the appropriate mineralization. While the presented data are of interest, some additional experiments are needed to support the author's claim. Comments 1. The title seems inappropriate since there in no direct evidence that increased autophagy is involved in the pathogenesis of brittle bone. RESPONSE: Our title does not claim a direct relationship between increased autophagy and a brittle bone phenotype, only that they are both observed in the EphrinB2-deficient osteocytes.
2. In Fig. 6D, immunofluorescent analysis with anti-LC3 antibody is required (Yoshii SR, Mizushima N. Int J Mol Sci. 18. pii: E1865. 2017). Fig. 6B,C. This new data confirms that EphrinB2-deficient Ocy454 cells exhibit a high level of endogenous autophagy, to the same level as that observed in control cells treated with Bafilomycin. This supports our earlier data indicating a high level of autophagy in these cells. This data is described in detail on p12.

RESPONSE: We have now carried this out, and include it in a new
3. The authors should show the results of region 1 and 3 in Fig. 4B. RESPONSE: We have modified this figure to include all three regions (now Figure 3B).  The authors performed additional experiments to reinforce the autophagy data, which is partly successful. However, the revised manuscript fails because it tries to deliver too much complex information raging from the altered bone matrix material properties to enhanced autophagy mainly in cultured cell line, which do not really answer to the previous comments of this reviewer.

(Previous comment) "
The statement was made that autophagic processes in osteocytes MAY directly control mineralization, but there is no such data shown." The authors claim that new data shows "abundant autophagosomes and matrix vesicles in Dmp1Cre.Efnb2f/f osteocytes in vivo (new Figure 7)" without showing evidence that these structures are matrix vesicles. Are they loaded with calcium and inorganic phosphate ions?
2. (Previous comment) "Second, the high osteocyte lacuna density in the absence of ephrinB2 remains descriptive." The authors' finding of the greater osteocyte lacunar density in Dmp1Cre.Efnb2f/f bone is interesting and potentially affects mechanical properties of bone. At least, quantitative discussion is necessary whether the higher lacunae density contributes observed differences in bone mechanical properties.
3. (Previous comment) "Third, the conclusions of the FTIR measurements are only poorly supported by the data presented." The authors' claim that "Unlike the endocortical surface, which is remodeled, the periosteal surface at this location undergoes continuous bone formation (bone growth), without bone resorption (Ref. 20)" is not supported by data.

(Previous comment) "Why was a strength defect detectable only in females?
Were there any sex differences in other experiments?" In the revised manuscript, the authors made a very interesting observation that male Dmp1Cre.Efnb2f/f femora showed no significant modification in 222 mineral:matrix, carbonate:mineral or amide I:II ratio (Supplementary Figure 1). This indicates that observed sex difference in mechanical properties is not simply due to geometry differences of bone but due to fundamental differences in bone matrix properties between females and males. Female-male differences in other experiments including autophagy aspects should be further analyzed.
5. (Previous comment) "Does the upregulation of ephrinB2 expression by PTH treatment reduce susceptibility of osteocytes to autophagy?" This comment is not addressed by the authors.
6. The new scheme Figure 9: Model of osteocytic EphrinB2 regulation of bone matrix composition" is difficult to follow. Why mineral crystals are accumulating far away from autophagic osteocytes.
7. The references are not well selected (too many).
Reviewer #3: Remarks to the Author: In the revised manuscript, the authors have addressed and expressed my concerns raised in the previous review. I think the present manuscript is acceptable for publication in Nature Communications.

Both Reviewer 1 and Reviewer 3 requested no further modifications.
Responses to Reviewer 2: 1. COMMENT: The authors claim that new data shows "abundant autophagosomes and matrix vesicles in Dmp1Cre.Efnb2f/f osteocytes in vivo (new Figure 7)" without showing evidence that these structures are matrix vesicles. Are they loaded with calcium and inorganic phosphate ions?
RESPONSE: The presence of calcium and inorganic phosphate ions is not a defining feature of matrix vesicles. Matrix vesicles are simply extracellular membrane-invested particles released by budding from the surfaces of matrix-associated cells, including chondrocytes, osteoblasts, odontoblasts and osteocytes (see HC Anderson, Clin Orthop Rel Res 1995). These are clearly observed in Figure 7. We suspect the matrix vesicles in our images contain mineral, but confirming this will be very challenging and requires extensive experiments beyond the scope of this study. We have clarified our statement on p18 to reflect this, as follows (new text in red): "We propose that osteocytes also control mineralization by autophagy-dependent release of matrix vesicles, that may contain mineral; the observation of many matrix vesicles budding from Dmp1Cre.Efnb2 f/f osteocytes, including some containing autophagosomes (Fig. 7), is consistent with this suggestion, although at this stage, we do not know whether they contain mineral at the time of release." 2. COMMENT: The authors' finding of the greater osteocyte lacunar density in Dmp1Cre.Efnb2f/f bone is interesting and potentially affects mechanical properties of bone. At least, quantitative discussion is necessary whether the higher lacunae density contributes observed differences in bone mechanical properties.  Figure 1). This indicates that observed sex difference in mechanical properties is not simply due to geometry differences of bone but due to fundamental differences in bone matrix properties between females and males. Female-male differences in other experiments including autophagy aspects should be further analyzed.
RESPONSE: We agree that the difference in mechanical properties is unlikely to be due to geometry differences, but is due to baseline differences in bone matrix composition between females and males. Given that the male Dmp1Cre.Efnb2f/f femora show no strength phenotype, it would be very difficult to justify the use of additional animals to specifically validate whether they also exhibit no change in autophagy in their osteocytes. We have clarified our comment about sex differences in the discussion, as follows: "We suggest that changes in bone matrix composition may play a more significant role in determining bone strength in female bones because they have higher mineral content and more rapid bone remodeling than male bones." 5. COMMENT: "Does the upregulation of ephrinB2 expression by PTH treatment reduce susceptibility of osteocytes to autophagy?". This comment is not addressed by the authors.
RESPONSE: As we stated in the previous response, we plan to assess the effects of PTH treatment on autophagy. This requires an extensive set of in vivo experiments, which is beyond the scope of the current work.
Given that pharmacologic PTH treatment does not modify the mineralization process (Vrahnas et al, Bone 2016), we had commented in the discussion that, in a physiological context, any PTH1R-mediated action on autophagy is more likely to relate to actions of endogenous PTHrP. See p20, end of first paragraph, included here for convenience: "Since exogenous PTH and endogenous PTHrP promote EphrinB2 expression in both osteoblasts and osteocytes, they may promote both initiation of osteoid mineralization and restrain the rate of mineral accrual. While there is no evidence for altered mineral accrual with PTH pharmacological administration, the reduced material strength of mice with osteocyte-specific PTHrP deletion suggests PTHrP may regulate physiological mineral accrual by inhibiting autophagy through EphrinB2-dependent actions in the osteocyte." 6. COMMENT: The new scheme Figure 9: Model of osteocytic EphrinB2 regulation of bone matrix composition" is difficult to follow. Why mineral crystals are accumulating far away from autophagic osteocytes.

RESPONSE:
We have simplified the diagram, modified the legend for clarity, and changed the labelling to make it clear that the crystals are not accumulating far from the osteocytes, but likely within their lacunar space.
7. COMMENT: The references are not well selected (too many).
RESPONSE: We have removed references where possible (17 references removed).

Reviewers' Comments:
Reviewer #2: Remarks to the Author: I felt that the possible link between autophagy and mineralization is not yet robust. For example, I was not convinced with "abundant extrusions and matrix vesicles budding from the cell surface" (line 317). Are these vesicles really associated with autophagy and involved in mineralization? This is important because osteoblasts/osteocytes under non-mineralizing conditions also produce exosome-like vesicles containing bio-molecules including miRNAs (ref. 1).
Mineralization in matrix vesicles can be evaluated by analyzing composition of Ca, P, and C with scanning electron microscopy (SEM) with energy-dispersive X-ray spectroscopy (EDS, EDX or EDXS), which is a widely used analytical technique (ref.