Ephrin-B1 regulates the adult diastolic function through a late postnatal maturation of cardiomyocyte surface crests
Curation statements for this article:-
Curated by eLife
eLife assessment
This is a valuable paper that relates the function of Ephrin-B1 to diastolic dysfunction via its actions on maturation of cardiomyocytes. The mechanisms of diastolic heart failure remain poorly understood, and this work contributes to advancing our understanding. The hypothesis is novel and the manuscript is fairly extensive and well-illustrated. The data, methods and analyses are presented to the community in a solid manner. The work represents an interesting insight into potential mechanisms of diastolic dysfunction and heart failure with preserved ejection fraction.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
The rod-shaped adult cardiomyocyte (CM) harbors a unique architecture of its lateral surface with periodic crests, relying on the presence of subsarcolemmal mitochondria (SSM) with unknown role. Here, we investigated the development and functional role of CM crests during the postnatal period. We found in rodents that CM crest maturation occurs late between postnatal day 20 (P20) and P60 through both SSM biogenesis, swelling and crest-crest lateral interactions between adjacent CM, promoting tissue compaction. At the functional level, we showed that the P20-P60 period is dedicated to the improvement of relaxation. Interestingly, crest maturation specifically contributes to an atypical CM hypertrophy of its short axis, without myofibril addition, but relying on CM lateral stretching. Mechanistically, using constitutive and conditional CM-specific knock-out mice, we identified ephrin-B1, a lateral membrane stabilizer, as a molecular determinant of P20-P60 crest maturation, governing both the CM lateral stretch and the diastolic function, thus highly suggesting a link between crest maturity and diastole. Remarkably, while young adult CM-specific Efnb1 KO mice essentially exhibit an impairment of the ventricular diastole with preserved ejection fraction and exercise intolerance, they progressively switch toward systolic heart failure with 100% KO mice dying after 13 months, indicative of a critical role of CM-ephrin-B1 in the adult heart function. This study highlights the molecular determinants and the biological implication of a new late P20-P60 postnatal developmental stage of the heart in rodents during which, in part, ephrin-B1 specifically regulates the maturation of the CM surface crests and of the diastolic function.
Article activity feed
-
-
Author Response
Reviewer #4 (Public Review):
The study employs a number of methods, including TEM morphometric analysis, immunochemistry, western blotting, genomics, genetically modified models, whole heart measurements.
However, the manuscript seems to be a collection of two unfinished works: one on the transition p20-p60 in post-natal development of the heart, second about the role of ephrinB1 in the maturation of the crests of the sarcolemma. Otherwise, it is not clear why in the first figure there is no staining for ephB1, and why there is staining for claudin 5 instead.
The reason is clearly explained in the text on page 6. The first figure explores the postnatal maturation of the CM crests and their molecular determinants and our previous paper described Claudin-5 as the first molecular determinant of the crests …
Author Response
Reviewer #4 (Public Review):
The study employs a number of methods, including TEM morphometric analysis, immunochemistry, western blotting, genomics, genetically modified models, whole heart measurements.
However, the manuscript seems to be a collection of two unfinished works: one on the transition p20-p60 in post-natal development of the heart, second about the role of ephrinB1 in the maturation of the crests of the sarcolemma. Otherwise, it is not clear why in the first figure there is no staining for ephB1, and why there is staining for claudin 5 instead.
The reason is clearly explained in the text on page 6. The first figure explores the postnatal maturation of the CM crests and their molecular determinants and our previous paper described Claudin-5 as the first molecular determinant of the crests (Guilbeau-Frugier et al, Cardiovasc Research 2019). Based on our previous demonstration of ephrin-B1 as a direct claudin-5 partner and regulator (Genet et al, Circulation Research 2012), we thus intuitively proposed ephrin-B1 as another potential molecular determinant of the crests that we explored for the first time in our current paper in revision. Moreover, ephrin-B1 is part of a large family of direct physical cell-cell communication proteins (Eph-Ephrin system), its role in the lateral crest-crest interaction was also obvious.
This is why at the beginning of the paper we explored claudin-5 and thereafter ephrin-B1 to explore more the functional role of the crests using Efnb1 KO mouse model we had already established in the lab.
The authors are trying to defend the idea that development of the heart in rats doesn't finish on postnatal day 20 and goes on for up to day 60. However, it is not convincing.
It is no surprise transcription profile is different between day 20 and day 60, I am sure as life goes on development continues into aging and any comparison of samples collected with sufficient time lapse will give transcriptional differences. Whether these differences represent a truly separate development stage is not a clear-cut story.
Most of the argument is based on morphometric study of TEM images.
But also on confocal microscopy studies and more importantly on transcriptomic data.
Whether it was evident that transcription profile is different between day 20 and day 60, then most of the studies in this postnatal field would have extended their study window over P20 which is not the case. As we mentioned it in the manuscript, most people in the field were assuming terminal maturity of the CM based essentially on its typical rod-shape which is already acquired at P20. Then growth of the heart between P20 and P60 was assumed to rely only on an increase in tissue quantitative content and not on transcriptomic changes, i.e. in qualitative content.
However, the method is not described at all. There is reference to another paper by the authors, but this paper doesn't provide a concise description of the morphometry either. It is unclear how randomisation of images and fields of view has been achieved and what statistical methods has been implemented. In TEM it is often possible to find all sorts of oddities depending on how you choose the images.
We agree with the author that TEM is often associated with “all sorts of oddities” and that‘s the reason our recent paper (Guilbeau-Frugier et al, Cardiovasc Research 2019) was dedicated to the analysis of technical pitfalls and analysis. All this paper relies on that: How to proceed the cardiac tissue to avoid artifacts on the crests/SSM visualization and how to quantify them?.
Now, instead of only citing our previous paper, we have implemented the “Material and methods” / “Transmission electron microscopy (TEM) and quantitative analysis” section (Main manuscript, page 20-21) by highly detailing all the TEM observation/quantification.
The question of randomization of images of the number fields of view is a general question in all imaging techniques and not specific at all with our TEM study. In imaging, there is no randomization.
All statistical analysis of TEM data quantifications are accurately described in all figure legends. For instance, in the figure 1: (B) Quantification of crest heights / sarcomere length (left panel), SSM number / crest (middle panel) and SSM area (right panel) from TEM micrographs obtained from P20- or P60 rat hearts (P20 n=6, P60 n=6; 4 to 8 CMs/rat, ~ 70 crests/rat). However, to better clarify the “P20 n=6, P60 n=6”, we have now specified “P20 or P60 n=6 rats”. This have been now specified in the figure legends for all statistical analysis (highlighted in yellow in the revised manuscript).
Why didn't the authors use microscopy of live isolated cells, which may be more relevant to study crest height?
We clearly explained it at the very beginning of the results section of our paper (first paragraph, second sentence (i, ii). The use of living CMs is a non-sense based on our two previous papers on this topic (Dague et al JMCC 2014 and Guilbeau-Frugier et al, Cardiovasc Research 2019). Our first paper was essentially based on AFM studies using isolated CMs and we found that rapidly after isolation, CM surface crests/SSM have a high tendency to shrink and disappear in control mice. This is why the second paper was based on an extensive characterization of the crests within the tissue using TEM experiments and the comparison of CM crests between tissue and living cells is also highlighted in this paper. More importantly, in this recent paper, we have described for the first time using high resolution imaging techniques (TEM and STEAD), the existence of intermittent physical interactions between neighboring CMs on their lateral side through crest-crest interaction via the extracellular domain of claudin-5. This crest-crest physical interaction can only be observed within the tissue since isolated adult CMs remain isolated and do not reproduce CM-CM physical interactions (through lateral or physical interactions at the longitudinal level, i.e. the intercalated disk level).
Both claudin5 and EphrinB1 seem to be expressed highly after p5, which doesn't correlate with the proposed maturation of crests at days 20 to 60.
Many processes do not rely only on gene/protein expression but on post-translational processes and localization/trafficking of proteins within the cell. This is exactly what we show with ephrin-B1 and claudin-5 proteins that traffic from the cytoplasm to the lateral membrane at the surface of the CMs between P20 and P60, as shown by our confocal images of the cardiac tissue while the global expression level of these two proteins doesn’t change (western blot results).
There is no causative relationship between the lack of ephrinb1 and crest maturing, at least to my mind.
Comparing the cardiac tissue between P20 an P60 and showing both ephrin-B1 trafficking at the CM lateral surface and crest maturation is obviously not a criterion of any relationship between these two events. However, when you delete a specific protein, i.e ephrin-B1, from a specific cell, i.e. the CM, and the phenotype of the KO mice is again a lack of crest maturation, you can at least deduce that ephrin-B1 is involved, directly or indirectly we don’t know, in the maturation process of the crests in the CM.
Now, because of the constitutive deletion of Efnb1, we couldn’t completely exclude that the phenotype of the constitutive Efnb1 CM-KO mice we described at the adult stage was directly related to specific alteration of CM surface crest/diastolic function at the adult stage or more likely related to other earlier developmental defects (secondary mechanisms). Also, to discriminate between these two possibilities, we have now used in the revision process a tamoxifen-inducible conditional-knockout (Mer-Cre-Mer) of Efnb1 in the CM (MHC promotor). This mouse model has never been reported before but its characterization (new Supplementary Figure 16) indicated that tamoxifen injection can lead to up to 50 % of Efnb1 deletion in CMs. In these conditions, deletion of Efnb1 (tamoxifen injection) was initiated at the young adult stage (2-month old) and the systolic and diastolic function (echo Doppler and LV-catheterism) but also CM crest phenotype (TEM) were examined one month later. As shown in the new Figure 7, deletion of efnb1 at the adult stage led to partial loss of CM surface crests (New Fig 7B), agreeing with the partial deletion of Efnb1, associated with a significant increase in the IVRT (echo-doppler), LVEDP (LV catheterism) with no modification of the ejection fraction (echo) compared to the control mouse littermates (tamoxifen injected) (New Fig. 7C, D). Thus, these data clearly demonstrate that ephrin-B1 is a specific determinant of the crest architecture at the CM surface and of the diastolic function at the adult stage.
-
eLife assessment
This is a valuable paper that relates the function of Ephrin-B1 to diastolic dysfunction via its actions on maturation of cardiomyocytes. The mechanisms of diastolic heart failure remain poorly understood, and this work contributes to advancing our understanding. The hypothesis is novel and the manuscript is fairly extensive and well-illustrated. The data, methods and analyses are presented to the community in a solid manner. The work represents an interesting insight into potential mechanisms of diastolic dysfunction and heart failure with preserved ejection fraction.
-
Reviewer #1 (Public Review):
The authors sought to demonstrate the specific two-stage maturation process that results in the development of cardiomyocyte crests. The authors also sought to demonstrate the importance of EphrinB1 in CM crest development and cardiac function, with disruption of EphrinB1 resulting in diastolic dysfunction and subsequently systolic dysfunction.
Strengths of the work included studies in two species (rat and mouse), cardiac-specific KO models, and careful considerations of myocardial structure-function.
-
Reviewer #2 (Public Review):
In this manuscript, Karsenty et al. describe postnatal development in the rat between P20 and P60 where crests of the cardiomyocyte lateral membranes mature. The authors previously described these crests in a Cardiovascular Res paper, where they highlighted how they facilitate interactions between cardiomyocytes in claudin-5 dependent manner. The authors previously also reported claudin-5 ephrin-b1 interactions on the cardiomyocyte lateral surface. Presently, the authors try to link together the following observations: a) crest height, mitochondrial number and area increase from postnatal day 20 to 60 in rats, and this correlates with increase claudin-5 expression, as well as gene expression accessed by microarray; b) changes in diastolic function occur in rats between days 20 and 60 postnatal life; c) …
Reviewer #2 (Public Review):
In this manuscript, Karsenty et al. describe postnatal development in the rat between P20 and P60 where crests of the cardiomyocyte lateral membranes mature. The authors previously described these crests in a Cardiovascular Res paper, where they highlighted how they facilitate interactions between cardiomyocytes in claudin-5 dependent manner. The authors previously also reported claudin-5 ephrin-b1 interactions on the cardiomyocyte lateral surface. Presently, the authors try to link together the following observations: a) crest height, mitochondrial number and area increase from postnatal day 20 to 60 in rats, and this correlates with increase claudin-5 expression, as well as gene expression accessed by microarray; b) changes in diastolic function occur in rats between days 20 and 60 postnatal life; c) cardiomyocyte-specific knockout of ephrin-b1 leads to diastolic dysfunction and a heart failure with preserved ejection fraction like phenotype.
The major strength of the paper is the detailed analysis of postnatal cardiomyocyte structure using electron microscopy and echocardiography. However, there are multiple major weaknesses of the methodology that should be clarified.
1. Most importantly, it is unclear how these mice are maintained. The authors explain the mice are in a mixed background. If this is the case, it is extremely important that littermate controls are used. Otherwise, it is very difficult to interpret the physiological data if the animals being compared are from two separate crosses.
2. The authors should clarify the echo and cath data in mice and rats. First with respect to the rat data, the end-diastolic pressure of the p60 rats reported in Figure 3 is around 9 mmHg. This is significant and abnormal, and also unexplained. More globally, the authors conclusion that "diastolic maturation" is occurring should be tempered by the fact that loading conditions at these two postnatal days are completely different, as highlighted by the blood pressures of the animals (~70/40 at p20 vs ~120/80 at P60). Third, the heart rates are significantly higher at p60 vs p20 in panel A, but not panel B. The authors report these studies were performed under isofluorane, so I imagine this reflects differences in anesthesia, which can significantly affect the heart rate and thus the diastolic parameters (particularly IVRT), so this should at least be commented on as a significant limitation.
With respect to data interpretation and whether the results are correctly interpreted, the following considerations should be taken into account:
1. The author conclusions in Figure 2d appear overstated. While the methodology of figure 2d is explained in the figure panel, it is generally looked over in the main text. Is there any precedent for utilizing human single-cell sequencing in this purpose - essentially the authors are superimposing the rat P20 and P60 data on the human UMAP plot. The authors walk back the statement in the text and clarify that these genes are expressed in the following cell types in adult human hearts.
2. The ephrin-b1 cardiomyocyte specific knockout mice are not a model of HFpEF. These mice show an accelerated death rate and clear evidence of progressive systolic impairment. Furthermore, it is completely unclear that the murine diastolic parameters are meaningfully different, and hence the overall evidence presented that this is a HFpEF phenotype is weak.
-
Reviewer #3 (Public Review):
The rod shape of cardiomyocytes (CM) as well as their distinct specialized membrane microdomains are crucial for normal cardiac function while alterations of such architecture are central to the pathogenesis of a host of cardiac diseases. However, mechanisms regulating this 3 D organization of CM during cardiac development and adult heart are still poorly known. The group C Galès had already done an important contribution to this domain by describing a distinct highly organized architecture of the lateral membrane with periodic crest containing transmembrane proteins claudin-5 and ephrin B1, of, however, unknown function. Following these previous studies, the group now investigated the maturation of this CM crest domain during the post-natal period as well as the consequences of loss of this organization on …
Reviewer #3 (Public Review):
The rod shape of cardiomyocytes (CM) as well as their distinct specialized membrane microdomains are crucial for normal cardiac function while alterations of such architecture are central to the pathogenesis of a host of cardiac diseases. However, mechanisms regulating this 3 D organization of CM during cardiac development and adult heart are still poorly known. The group C Galès had already done an important contribution to this domain by describing a distinct highly organized architecture of the lateral membrane with periodic crest containing transmembrane proteins claudin-5 and ephrin B1, of, however, unknown function. Following these previous studies, the group now investigated the maturation of this CM crest domain during the post-natal period as well as the consequences of loss of this organization on heart function. This study performed by an expert team clearly provides new and original knowledge on cardiac maturation heart development and on the relation between CM ultrastructure and cardiac function. A major finding of the study is that the protein ephrin-B1 plays a key role in adult crest-crest interactions between CM that appears to be a major determinant of the normal diastolic cardiac function. Therefore, beyond providing new insights in CM maturation, this study opens perspectives for the understanding of the pathogenesis of the so-called heart failure with preserve ejection function, a rising cause of HF with a prevalence increasing with the ageing of the population and without yet specific biomarker and therapeutic target. The article, well written and clearly illustrated, uses an impressive number of approaches including cutting edge imaging techniques.
-
Reviewer #4 (Public Review):
The study employs a number of methods, including TEM morphometric analysis, immunochemistry, western blotting, genomics, genetically modified models, whole heart measurements.
However, the manuscript seems to be a collection of two unfinished works: one on the transition p20-p60 in post-natal development of the heart, second about the role of ephrinB1 in the maturation of the crests of the sarcolemma. Otherwise, it is not clear why in the first figure there is no staining for ephB1, and why there is staining for claudin 5 instead.
The authors are trying to defend the idea that development of the heart in rats doesn't finish on postnatal day 20 and goes on for up to day 60. However, it is not convincing.
It is no surprise transcription profile is different between day 20 and day 60, I am sure as life goes on …Reviewer #4 (Public Review):
The study employs a number of methods, including TEM morphometric analysis, immunochemistry, western blotting, genomics, genetically modified models, whole heart measurements.
However, the manuscript seems to be a collection of two unfinished works: one on the transition p20-p60 in post-natal development of the heart, second about the role of ephrinB1 in the maturation of the crests of the sarcolemma. Otherwise, it is not clear why in the first figure there is no staining for ephB1, and why there is staining for claudin 5 instead.
The authors are trying to defend the idea that development of the heart in rats doesn't finish on postnatal day 20 and goes on for up to day 60. However, it is not convincing.
It is no surprise transcription profile is different between day 20 and day 60, I am sure as life goes on development continues into aging and any comparison of samples collected with sufficient time lapse will give transcriptional differences. Whether these differences represent a truly separate development stage is not a clear-cut story.
Most of the argument is based on morphometric study of TEM images. However, the method is not described at all. There is reference to another paper by the authors, but this paper doesn't provide a concise description of the morphometry either. It is unclear how randomisation of images and fields of view has been achieved and what statistical methods has been implemented. In TEM it is often possible to find all sorts of oddities depending on how you choose the images.
Why didn't the authors use microscopy of live isolated cells, which may be more relevant to study crest hight?
Both claudin5 and EphrinB1 seem to be expressed highly after p5, which doesn't correlate with the proposed maturation of crests at days 20 to 60.
There is no causative relationship between the lack of ephrinb1 and crest maturing, at least to my mind. -
Reviewer #5 (Public Review):
In this manuscript, the authors sought to study the development and functional role of cardiomyocyte crests during the postnatal period. They first utilized high resolution imaging to characterize in detail the structure and maturation of the lateral membrane surface architecture. The authors documented maturation of diastolic function coincides temporally with the formation of the lateral membrane surface structure. The authors also revealed that EphrinB1 conditional knockout exhibited defect in CM crest development and cardiac function.
Strengths of the work included detailed characterization of structural and functional maturation during postnatal stages and the use of cardiac-specific KO models. Yet, since EphrinB1 conditional knockout displayed multiple structural defects, it is difficult to discern the …
Reviewer #5 (Public Review):
In this manuscript, the authors sought to study the development and functional role of cardiomyocyte crests during the postnatal period. They first utilized high resolution imaging to characterize in detail the structure and maturation of the lateral membrane surface architecture. The authors documented maturation of diastolic function coincides temporally with the formation of the lateral membrane surface structure. The authors also revealed that EphrinB1 conditional knockout exhibited defect in CM crest development and cardiac function.
Strengths of the work included detailed characterization of structural and functional maturation during postnatal stages and the use of cardiac-specific KO models. Yet, since EphrinB1 conditional knockout displayed multiple structural defects, it is difficult to discern the primary defects and pinpoint the causal-effect relationship.
-
