Altered regulation of mesenchymal cell senescence in adipose tissue promotes pathological changes associated with diabetic wound healing

Pathologic diabetic wound healing is caused by sequential and progressive deterioration of hemostasis, inflammation, proliferation, and resolution/remodeling. Cellular senescence promotes wound healing; however, diabetic wounds exhibit low levels of senescent factors and accumulate senescent cells, which impair the healing process. Here we show that the number of p15INK4B + PDGFRα + senescent mesenchymal cells in adipose tissue increases transiently during early phases of wound healing in both non-diabetic mice and humans. Transplantation of adipose tissue from diabetic mice into non-diabetic mice results in impaired wound healing and an altered cellular senescence–associated secretory phenotype (SASP), suggesting that insufficient induction of adipose tissue senescence after injury is a pathological mechanism of diabetic wound healing. These results provide insight into how regulation of senescence in adipose tissue contributes to wound healing and could constitute a basis for developing therapeutic treatment for wound healing impairment in diabetes.


Report
This study was performed to understand the role of adipose-derived mesenchymal cell senescence for wound healing. The study aimed to show that diabetic wounds in Lperdb/db mice exhibit low levels of senescent factors and accumulate senescent cells, impairing acute wound healing. Indeed, wound healing is a dynamic process. This study reported that p15INK4B+ senescent PDGFR-α+ mesenchymal cells in adipose tissue are transiently increased during early wound healing in non-diabetic controls, promoting wound healing.
Transplantation of adipose tissue from diabetic mice to non-diabetic mice resulted in poor wound healing secondary to insufficient acute induction of senescence after injury. The study characterizes cellular senescence and SASP profiling from adipose tissue in diabetic and nondiabetic mice, and also shares data from diabetic patients.
The main limitations of this study are that senescence profiling was performed using p15INK4B marker, a newer marker of senescence compared to validated p16INK4A or p21. The overall senescence profiling therefore requires validation with known markers of senescence i.e., p16INK4A or p21. In addition, wound healing secondary to adipose tissue senescence was discussed without commentary on the role of skin senescence.
Diabetic wound healing is an important area of research as diabetic foot ulcers result in significant health care burden. This study therefore informs an unmet patient need.

Major Comments
Title: Consider including adipose tissue in the title to increase specificity: Altered regulation of mesenchymal cell senescence in adipose tissue promotes pathological changes associated with diabetic wound healing. Figure 1: The study reported that Lperdb/db mice with higher body weight and blood glucose has poor cutaneous wound healing at 0 DPW (days post-wound) to 8 DPW by gross and histological evaluation. The study also suggests that diabetic mice exhibit a diffuse distribution of PDGFR-α and α-SMA expression at 8 DPW in the subcutaneous adipose tissue in contrast to non-diabetic mice which exhibit a localized distribution in the panniculus carnosus layer (present only in vertebrates) at the wound edge. The relevance for this variability in expression pattern is unclear as it relates to epidermal wound healing.
-Please include PDGFR-α and α-SMA immunostaining in the epidermal and dermal tissue at 8 DPW for diabetic and non-diabetic mice. Figure 2: The study focuses on the role of subcutaneous adipose tissue in wound healing. Therefore, the authors transplanted adipose tissue from diabetic mice into excisional skin of nondiabetic mice. It is reported that excisional skin wound was sutured closed after transplantation to stabilize the transplanted tissue. This allows for the excisional skin covering to act as a natural barrier (versus an open wound).
-Please include a control group with excisional skin closure suture closed with sham transplant (no transplanted tissue). This will evaluate the role of skin barrier alone without adipose tissue transplant.
- Figure 2C depicts a wound healing histological score which is arbitrary and nonblinded. It does not add value.
- Figure 2D shows 'normal dermis and wound region' but this is an overmagnified area. Please include wound region with epidermis and lower magnification image.
- Figure 2E could benefit from showing the percentage of CM-Dil labeled adipose tissue at 2 DPW and 8 DPW.
-To complement the immunostaining, please include RNA expression levels for PDGFR-α and α-SMA in adipose tissue at 8 DPW for diabetic and non-diabetic mice. Figure 3: This study evaluated cellular senescence in adipose tissue from diabetic and nondiabetic mice using p15INK4B marker, which was acknowledged in the discussion as a newer marker of senescence compared to p16INK4A. Multiple senescence-related genes were analyzed (Cdkn2b, Cdkn1a, Trp53, Serpine1, Serpine2, IL-6 and TGF-β1) in adipose tissue and compared to skin tissue (supplemental figure). While this paper focuses on the role of adipose tissue in wound healing, its relation to skin will be important to consider. -Consider including skin mRNA data along with adipose mRNA data in Figure 3A in a side-by-side comparison.
-It was noted that p15INK4B, which is encoded by Cdkn2b, is transiently increased at 2 DPW in non-diabetic mice and this transient early expression contributes to normal wound healing. This also corresponds to Figure 3A mRNA expression of Cdkn2b in adipose tissue. The supplemental figure shows no change in expression for Cdkn2b in 2 DPW and 8 DPW skin between diabetic and non-diabetic mice. How do you explain the discrepancy in expression in adipose and skin tissues with this gene and other genes? The role of cell senescence during would healing in skin is previously defined. There needs to be more discussion about the interplay between adipose tissue and skin regarding senescent markers or an explanation for the variability.
-To complement the immunostaining, please include RNA expression levels for PDGFR-α and p15INK4B in adipose tissue from diabetic and non-diabetic mice.
-Recommend senescence β-Galactosidase staining for adipose tissue from diabetic and nondiabetic mice.
-Highly recommend validation with p16INK4A and/or p21 senescence markers. Figure 4: This study reports cellular senescence in adipose tissue from diabetic and nondiabetic patients. Representative images are shown from immunostaining of adipose tissue however the source of the adipose tissue is unclear. The patients are relatively agematched between diabetic and non-diabetic patients however the source of wound significantly varies from burn injury versus pressure ulcer versus chronic traumatic ulcer according to the table. There are also variances in days post-wound (less than 30 daysacute wound and greater than 30 days -chronic wound). It is very interesting that the percentage of senescence markers increases in diabetic patients with days post-wound, especially with patient 6 with diabetes mellitus who is an elderly patient with lower extremity ulcer. Therefore, the potential differences in acute wound healing (shown in this paper) versus chronic wound healing could be highlighted in the discussion section, especially to suggest that this paper focuses on the acute wound model only.
-Is the adipose tissue from patients collected from the peripheral wound site or an arbitrary nonwound site (i.e., abdomen)? Please clarify.
-Highly recommend validation with p16INK4A and/or p21 senescence markers. Figure 5: SASP profiling was performed to assess cytokines released from adipose tissue in diabetic and non-diabetic mice collected from under the wound bed.

Minor comments
The study was conducted objectively and under ethical guidelines. There could be areas for improvement in methodology details as previously suggested. There are proper controls and defined outcomes. Line 65: Reference 15 (senescence and the healing rates of venous ulcers) evaluates venous ulcers not diabetic ulcers. Please update to venous ulcers if still relevant to include here. Line 282: The study comments on the induction of chronic inflammation however the results of the study span an acute timeframe (8 DPW) therefore, recommend adjusting wording to acute inflammation. This is an interesting and insightful study provided by Kita et al that adds to the current limited literature on cellular senescence and wound repair. Overall a number of interesting experiments have been performed to elucidate the role of subcutaneous adipose tissue in repair, and its links to cellular senescence. However, the authors have oversensationalised their data in areas, and this should be addressed. Specific comments below.

Introduction
Line 52: "Increased senescent cells are observed in the human skin during normal wound healing, heart regeneration in zebrafish and neonatal mice, and tissue regeneration in salamanders and zebrafish 4-7". There are no references here that links senescence to human skin healing. Please amend the text to reflect this.
Paragraph beginning line 58 is a bit difficult to understand as there is no clear definition of how senescence in normal and chronic healing differs. This should be amended to clearly explain that senescence in normal wound healing is transient (i.e. induced to allow matrix formation and then regresses) and in chronic wounds is uncontrolled and sustained. Line 118-120: It is difficult to determine from the images alone whether there are differences in a-SMA and PDGFR-a +ve cells between the non-diabetic and diabetic mice. Please provide quantification of these cells to strengthen the provided the results. Arrows pointing the +ve cells would also guide the reader further. Was the focus only on the subcutaneous region beneath the panniculus carnosus?

Results
Line 127: It would be more informative to provide a less vague measurement of wound healing than histological score which is based on observation alone. Wound width/wound area/reepithelialisation/granulation tissue maturity would give more concrete evidence. Please provide one or more of these additional analyses to strengthen the data.
Line 133: If stating that a group has increased levels of a-SMA+ve and PDGFR-a+ve cells in a particular region then quantification of the cells should also be shown, not just representative images. Please provide this analysis.
Line 188: How might the delay in adipose tissue senescence in diabetic mice allow for accumulation of senescent cells in diabetic mice? Please move this statement to the discussion and justify it there.
Line 229/ Fig5c-d: I find it hard to believe that the fibroblast scratches have taken 6 days to close? The measurement as % Closure is also opposite to what it should be and may cause confusion. There isn't 100% closure at D0. Please analyse in the opposite direction or alternatively describe this as % open scratch area.

Discussion
Line 237: Until quantification is performed on senescent cell numbers from immunohistochemical analysis then it cannot be assumed that there is an increase in senescent cells (see results comments).
Line 251: Please specify that this is in vitro wound healing. "which affects wound healing in vitro" Line 278: It is a stretch to say that the findings of the study suggest Lepr db/+ SASP promote wound healing by inducing cell migration and proliferation, and inhibit fibrosis as the only functional readout performed was the scratch assay. Please rephrase this to state cell migration only, and that further work would be needed to assess the influence of these adipose tissue derived SASP factors on proliferation/fibrosis. Line 281: Again, it is an overinterpretation to state the Lepr db/db SASP induce chronic inflammation as this has not been assessed, either in vitro or looked at in the transplantation model. Please rephrase this to be more speculative… the adipose-tissue SASP may potentially induce chronic inflammation… Line 284: Quantitative data is needed to be able to make the claim that a transient increase in senescence is observed in the mice (see comments on results section).

Methods
Line 253 onwards -animal use. It needs to be made clear that the male mice were used for the wounding study and the female mice for fibroblast isolation for in vitro analysis (in the animal section and the figure legends). Why were only male mice used in the wounding study, and why was the opposite sex used for in vitro study? Please clarify the groupings further and reasoning for this. The limitation of using one sex should also be added to the discussion. N numbers need to be included and justified.
We thank the Reviewers for their careful consideration of our manuscript. We have revised the manuscript by taking into account each point raised by the reviewers. These changes are highlighted in "red" color fonts in the revised manuscript. We have now addressed each of the comments as outlined below.

Report
This study was performed to understand the role of adipose-derived mesenchymal cell senescence for wound healing. The study aimed to show that diabetic wounds in Lper db/db mice exhibit low levels of senescent factors and accumulate senescent cells, impairing acute wound healing. Indeed, wound healing is a dynamic process. This study reported that p15 Ink4b+ senescent PDGFR-α+ mesenchymal cells in adipose tissue are transiently increased during early wound healing in non-diabetic controls, promoting wound healing.
Transplantation of adipose tissue from diabetic mice to non-diabetic mice resulted in poor wound healing secondary to insufficient acute induction of senescence after injury. The study characterizes cellular senescence and SASP profiling from adipose tissue in diabetic and non-diabetic mice, and also shares data from diabetic patients.
Author Response: We thank Reviewer #2 for the positive comments and suggestions.
The main limitations of this study are that senescence profiling was performed using p15 Ink4b marker, a newer marker of senescence compared to validated p16 Ink4a or p21. The overall senescence profiling therefore requires validation with known markers of senescence i.e., p16 Ink4a or p21. In addition, wound healing secondary to adipose tissue senescence was discussed without commentary on the role of skin senescence.
Diabetic wound healing is an important area of research as diabetic foot ulcers result in significant health care burden. This study therefore informs an unmet patient need. in that both bind to CDK4 and CDK6, thereby inhibiting cell-cycle progression. In addition, p15 INK4B and p21 CIP1/WAF1 have been shown to be associated with TGF-β mediated cellular senescence. Although these senescence markers were not completely merged, at least over 75% of p15 INK4B  We also described the discrepancy of senescence role in the skin and adipose tissue in the Discussion.

Major Comments
Title: Consider including adipose tissue in the title to increase specificity: Altered regulation of mesenchymal cell senescence in adipose tissue promotes pathological changes associated with diabetic wound healing.
Author Response: Thank you for your suggestion. We agree with the reviewer and have changed the title to "Altered regulation of mesenchymal cell senescence in adipose tissue promotes pathological changes associated with diabetic wound healing."  This study evaluated cellular senescence in adipose tissue from diabetic and nondiabetic mice using p15 Ink4b marker, which was acknowledged in the discussion as a newer marker of senescence compared to p16 Ink4a . Multiple senescence-related genes were analyzed (Cdkn2b, Cdkn1a, Trp53, Serpine1, Serpine2, IL-6 and TGF-β1) in adipose tissue and compared to skin tissue (supplemental figure). While this paper focuses on the role of adipose tissue in wound healing, its relation to skin will be important to consider.

#2-7
-Consider including skin mRNA data along with adipose mRNA data in Figure 3A in a side-by-side comparison.
Author Response: Thank you for your suggestion. We have put the skin mRNA data along with adipose tissue mRNA data in Fig. 3 to allow for side-by-side comparison. We also added additional data of Il6 and Tgfb1 expression in skin to compare with skin and adipose tissue. We also added the principal component analysis of skin in Fig. 3b. Please see Fig. 3a-c.

#2-8
-It was noted that p15 Ink4b , which is encoded by Cdkn2b, is transiently increased at 2 DPW in non-diabetic mice and this transient early expression contributes to normal wound healing. This also corresponds to

#2-9
-To complement the immunostaining, please include RNA expression levels for PDGFR-α and p15 Ink4b in adipose tissue from diabetic and non-diabetic mice.
Author Response: Thank you for your suggestion. The level of Cdkn2b is now shown in Fig. 3a; Cdkn2b encodes p15 INK4B . We have added the level of Pdgfra mRNA and data showing that Pdgfra gene expression increased at 8 DPW in Lepr db/db mice (p = 0.065), although this increase is not statistically significant. Please see Fig. 4c, and the Results (page 10, lines 199-201).
Author Response: Thank you for your comments. We have added images and quantitative data concerning SPiDER-β-Gal, which is a newly developed fluorescence-based assay wherein activation of fluorescence is observed upon reaction with SA-β-galactosidase(Gal) 3 . We found that cells positive for SPiDER-β-Gal fluorescence increased in Lepr db/+ non-diabetic mice at 2 DPW but decreased at 8 Representative images are shown from immunostaining of adipose tissue however the source of the adipose tissue is unclear. The patients are relatively age matched between diabetic and non-diabetic patients however the source of wound significantly varies from burn injury versus pressure ulcer versus chronic traumatic ulcer according to the table. There are also variances in days post-wound (less than 30 daysacute wound and greater than 30 dayschronic wound). It is very interesting that the percentage of senescence markers increases in diabetic patients with days post-wound, especially with patient 6 with diabetes mellitus who is an elderly patient with lower extremity ulcer. Therefore, the potential differences in acute wound healing (shown in this paper) versus chronic wound healing could be highlighted in the discussion section, especially to suggest that this paper focuses on the acute wound model only.

DPW, while SPiDER-β-Gal-positive cells increased in
Author Response: Thank you for your comments. We have added the source of the adipose tissue in Table 1. We have also added Past Medical History. Please see Table 1. In this study, we focused on acute wounding in the mice model. We previously demonstrated the importance of transient senescence in the healing process of acute muscle injury; such transient senescence is not observed in chronic muscle inflammation, which can be ameliorated by inducing senescence of mesenchymal stroma cells 4,5 . Therefore, we suspected that diabetes may induce changes in the characteristics of acute senescence post-wounding on the skin. During acute wounding of diabetic mice, senescence induction was delayed in subcutaneous adipose tissue, while senescent cells were present in the subcutaneous adipose tissue of diabetic patients over 60 days post-wounding that resembles chronic wounding.
Previous studies have demonstrated the importance of transiently induced senescence for acute wound healing, while chronic senescence induces pathological conditions 6  Author Response: Thank you for pointing out this error. We have changed "diabetic ulcers" to "venous ulcers" as requested. Please see the Introduction (page 4, lines 67-68).

#2-18
Line 282: The study comments on the induction of chronic inflammation however the results of the study span an acute timeframe (8 DPW) therefore, recommend adjusting wording to acute inflammation.
Author Response: Thank you for your comment. We have changed the indicated sentence to "By contrast, Lepr db/db adipose tissue-secreted SASP factors may potentially induce wound healing impairment." Please see page 16, lines 319-321. Introduction #3-1 Line 52: "Increased senescent cells are observed in the human skin during normal wound healing, heart regeneration in zebrafish and neonatal mice, and tissue regeneration in salamanders and zebrafish 4-7".
There are no references here that links senescence to human skin healing. Please amend the text to reflect this.
Author Response: We apologize for the misdescription. We have changed the human skin to mouse skin and added a corresponding reference. Please see the Introduction (page 3, lines 52-54). Please measure the adipose tissue area within the wound and provide quantification of collagen from the Masson"s Trichrome staining.
Author Response: We added quantification data of collagen and adipose tissue. Please see Fig. 1f-g, and the Results (page 6, lines 112-113).

#3-4
Line 118-120: It is difficult to determine from the images alone whether there are differences in a-SMA and PDGFR-a +positive cells between the non-diabetic and diabetic mice. Please provide quantification of these cells to strengthen the provided the results. Arrows pointing the +ve cells would also guide the reader further.
Was the focus only on the subcutaneous region beneath the panniculus carnosus? We focused on the interstitial connective tissue and subcutaneous adipose tissue beneath panniculus carnosus, because we observed the infiltration of myofibroblasts into both the interstitial connective tissue layer and adipose tissue at the acute wound site as shown in Fig. 1.
We have also added the anatomical definition of the panniculus carnosus and interstitial connective We added these quantitative data to determine if Lepr db/db mice-derived adipose tissue transplantation induces or impairs wound healing compared with normal and diabetic wound healing as seen in Fig.   1h-1j. We found that mice transplanted with Lepr db/db -derived adipose tissue (Lepr db/db ATT) showed an increased population of cells positive for both PDGFRα and α-SMA in interstitial connective tissue, which was also observed for impaired wound healing in Lepr db/db mice. Please see Fig. 2d  Overall, we used culture inserts to perform our wound healing assay, and added control conditions.
Hence, we think our data reliably shows that adipose tissue-derived conditioned media affects fibroblast migration.
To prevent misunderstanding of our assay protocol, we changed the term "scratch assay" to "wound healing assay." Line 251: Please specify that this is in vitro wound healing. "which affects wound healing in vitro" Author Response: We have fixed it. Please see the revised manuscript, page 14, line 292.

#3-11
Line 278: It is a stretch to say that the findings of the study suggest Lepr db/+ SASP promote wound healing by inducing cell migration and proliferation, and inhibit fibrosis as the only functional readout performed was the scratch assay. Please rephrase this to state cell migration only, and that further work would be needed to assess the influence of these adipose tissue derived SASP factors on proliferation/fibrosis.
Author Response: We have rephrased this statement into "Lepr db/+ adipose tissue-secreted SASP factors promote cell migration." Please see the Discussion section, page 16, lines 318-319.

#3-12
Line 281: Again, it is an overinterpretation to state the Lepr db/db SASP induce chronic inflammation as this has not been assessed, either in vitro or looked at in the transplantation model. Please rephrase this to be more speculative… the adipose-tissue SASP may potentially induce chronic inflammation… Author Response: Reviewer #2 also suggested that this statement is overinterpretation. Therefore, we have rephrased this sentence into "By contrast, Lepr db/db adipose tissue-secreted SASP factors may potentially induce wound healing impairment." Please see the Discussion, page 16, lines 319-321.

#3-13
Line 284: Quantitative data is needed to be able to make the claim that a transient increase in senescence is observed in the mice (see comments on results section).
Author Response: As we responded to your Comment #9, we added the quantitative data of the

#3-14
Line 253 onwardsanimal use. It needs to be made clear that the male mice were used for the wounding study and the female mice for fibroblast isolation for in vitro analysis (in the animal section and the figure legends). Why were only male mice used in the wounding study, and why was the opposite sex used for in vitro study? Please clarify the groupings further and reasoning for this. The limitation of using one sex should also be added to the discussion. N numbers need to be included and justified.
Author Response: Thank you for your point out. In this study, we used female mouse-derived skin fibroblast previously developed to reduce the number of animals used. However, as you suggested, sex difference is a very important point. Several studies showed that the prevalence of diabetes ulcers was higher among males than females 17 . Furthermore, The risk of amputation is a sequela of diabetic foot ulceration, which causes huge morbidly and mortality, is higher in males than females 18 . Due to the high risk of diabetes ulcers, therefore we used male Lepr db/db mice in this study. Now we added a new result of wound healing assay using C57BL/6 male mice skin-derived fibroblasts to show the effect of diabetic and non-diabetic adipose tissue-derived secretome factors on cell migration in vitro. This study used only male mice in an in vivo study, which we acknowledge is a limitation of this study.
Hence, we added the limitation statement in the Discussion section. Please see Fig. 6c-d, the Results section (page 13, lines 258-265) and the Discussion (pages 20-21, lines 414-419). We also provide the N numbers in the legend of Figure 6.