Identification of EPHB4 variants in dilated cardiomyopathy patients

  1. Guillermo Luxán
  2. Marion Muhly-Reinholz
  3. Simone F. Glaser
  4. Johannes Trebing
  5. Christoph Reich
  6. Jan Haas
  7. Farbod Sedaghat-Hamedani
  8. Benjamin Meder
  9. Stefanie Dimmeler

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Cardiac homeostasis relies on the appropriate provision of nutrients to the working myocardium. EPHB4 is required for the maintenance of vascular integrity and correct fatty acid transport uptake in the heart via regulating the caveolar trafficking of the fatty acid receptor CD36. In the mouse, endothelial specific loss-of-function of the receptor EphB4, or its ligand ephrin-B2, induces Dilated Cardiomyopathy (DCM) like defects. Now, we have identified six new EPHB4 variants with deleterious potential in a cohort of 573 DCM patients. Similar to the EphB4 mutant mice, EPHB4 variants carrying patients show an altered expression pattern of CD36 and CAV1 in the heart. For the first time, our data identifies EPHB4 mutations in DCM patients. This observation supports the notion that the Eph-ephrin signalling pathway, and in particular the receptor EPHB4, plays a role in the development of DCM in human patients.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1 (Evidence, reproducibility and clarity): Summary:

    This research article describes genetic identification and expression analyses of six Ephrin type-B receptor 4 (EPHB4) variants identified in patients with dilated cardiomyopathy (DCM). Variants were identified Variants were identified in a cohort of 573 patients enrolled through the multicenter DZHK-TORCH (TranslatiOnal Registry for CardiomyopatHies) study and the Institute for Cardiomyopathies Heidelberg registry. Expression of downstream molecules, CAV1 and CD36, was assessed in human cardiac tissues by immunohistochemistry. EPHB4 cardiac expression was assessed using recently published single-cell/nucleus RNA sequencing data (Nicin et al 2022) incorporating siRNA-seq data from two other studies (healthy cardiac tissue, Litvinukova et al 2020) and (hypertrophic/aortic stenosis, Nicin et al. 2020).

    We thank the reviewer for the recommendations that have improved our manuscript.

    Major Comments:

    1. Details of identified truncating RBM20 and TTN variants must be provided. These should be integrated into Table 1 alongside each co-occurring EPHB4 variant. List whether the TTN truncating variant is located in the A-band and whether these variants would be adjudicated as pathogenic/likely pathogenic, variant of uncertain significance by ACMG and/or similarly refined DCM criteria (Morales et al. 2020, Circ-Genom Precis Med).

    Details of the truncating RNM20 and TTN have been provided in the new supplementary table 1. As indicated in the table both mutations are pathogenic, and thus, most probable the cause for the disease in these patients. In case of TTN this is a truncating variant and is located in the M-band in exon 358, which is annotated with a PSI in DCM of 100% in cardiodb.org. The fact that these mutations are most probably the cause for DCM in these patients has been included in the discussion section and reads as follows:

    Although it is most probable that in the case of the patients carrying TNN and RNM20 variants this would be the cause of the disease, this study further supports, the importance of EPHB4 regulating CD36 caveolar trafficking to the membrane, whether this happens in endothelial cells or cardiomyocytes, maintaining cardiac homeostasis in humans and its implication on DCM

    1. Discuss co-occurrence of multiple EPHB4 variants in two patients (DCM1, DCM3) and identification of 2 EPHB4 variants in more than one proband.

    As shown in Figure 1A, the detected variants are found in multiple domains of the protein, hence no clear hotspot is detected. We did not yet investigate on the exact mechanisms of action, however, when we compare the two patients with multiple EPHB4 variants, the average LVEF (echo) is 17.5 compared to 38,67 for the remaining 4 patients with only one EPHB4 variant and 35,17 for the six non-EPHB4 variant-carriers. Although the sample number only allows for a semi-quantitively analysis, it still hints at a possible EPHB4-variant effect, which certainly needs verification in a larger cohort.

    Since we do not postulate the detected variants as independently disease-causing, and we also did not explicitly filter for very rare variants, it is not surprising that we find two variants in multiple patients. As stated above, we did not investigate this further, but evidence is growing that compound heterozygosity is playing a role in heritable diseases. It will be interesting to analyze e.g. phasing (Hofmeister et al., Nature Genetics, 2023) or additive (biallelic) effects, which have come to attention also in cardiomyopathies recently (Lipov et al., Nature Cardiovascular Research, 2023).

    This fact has now been included in the manuscript, both in the results and in the discussion. It reads as follows:

    Interestingly, two of the analysed patients present more than one variant of EPHB4 and we could identify the same variant in more than one patient (Table 1)

    (…)

    Nevertheless, two of the patients carrying one benign or likely benign also carry another variant classified as likely pathogenic or of uncertain significance (Table 1) and interestingly, the average LVEF of the two patients with multiple EPHB4 variants is 17.5 compared to 38,67 for the remaining 4 patients with only one EPHB4 variant and 35,17 for the six non-EPHB4 variant-carriers. Although the sample number only allows for a semi-quantitively analysis, it still hints at a possible EPHB4-variant effect, which certainly needs verification in a larger cohort.

    1. Three of the six variants (p.Lys635Asn, Val113Ile, Glu890Asp) are classified as Clinvar Benign/Likely Benign. Additionally, p.Glu890Asp has been identified in 50 homozygotes in gnomAD non-Finnish European population. These data cast doubt on the pathogenicity of these variants. These classifications, as well as VUS classification of p.Pro79Leu, should be listed in Table 1. The authors should reconcile the benign/likely benign Clinvar classifications with their presented evidence for pathogenicity in the discussion.

    We have now included the ACMG classification in Table 1. Similar to the Clinvar classification, some of the variants are classified benign or likely benign. Still, the fact that the patients that carry them also carry another variant and that the histological findings are similar among the patients carrying an EPHB4 variant and different to those that don’t and the enriched presence of EPHB4 variants in the DCM population support our hypothesis that the Eph-ephrin signalling pathway plays a role in the development of DCM.

    Nevertheless, we agree with the reviewer that the fact that some of these variants have been classified as benign, and the presence of mutations in other genes already related to DCM like TNN or RMM20 may suggest that the EPHB4 mutations may not be the only cause for the disease but rather have an additive effect. As a consequence, we have toned down our conclusions and the discussion reads now as follows:

    Finally, although not as the main disease cause, this study not only supports the role of EPHB4 in the heart, but it also corroborates the importance of CD36 and CAV1 for the cardiac health, and has the potential to improve diagnosis and risk stratification tools for DCM. In addition, as other genes crucial for fatty acid transport may be involved in cardiac disease, this study may help identify new diagnostic or therapeutic targets.

    1. CD36 and CAV1 expression are not quantified. Qualitatively, it is difficult to confirm CD36 reduction in DCM and disruption in EPHB4 variant samples as imaging parameters are not specified and do not appear to be standardized across treatments. Clearly state (either in the figure legend or in the methods) whether identical imaging parameters were used across panels 1C-1E. Note any differences in these parameters.

    We have now quantified the two IHC. It is very clear that the total CD36 is significantly reduced in both groups when compared to the healthy donor (Figure for the reviewer 1A). In case of CAV1 this is not so evident, although the signal seems reduced this is not significant (Figure for the reviewer 1B). These new data have been included in the figure of the manuscript.

    Figure for the reviewer 1. Quantification of (A) CD36 and (B) CAV1 in the immunohistochemistry analysis of patients biopsies. Data shown as mean ± SEM. (A) P value was calculated using one sample one sample Wilcoxon test for DCM and one sample t test for DCM EPHB4. Both cohorts where compared to the mean of HD. P value < 0.05 was considered significant. (B) P value was calculated with one sample t test for both cohorts. P value < 0.05 was considered significant.

    All the images have been taken in the same conditions. The observed difference in the background is due to the disease conditions of the DCM samples. Furthermore, the apparent reduced number of capillaries observed in the DCM patients are caused by the hypertrophic state of the cardiomyocytes in the diseased state. These are bigger and thus, less cells and capillaries appear per picture. The parameters have been included in the methods and read as follow:

    Immunohistochemistry was imaged in a Leica Stellaris confocal microscope. All images were obtained with 63x magnification and the same laser and gain intensities. Images were acquired using the software LAS X (Leica, version 4.4) and quantified using the Volocity Software (Quorum Technologies, version 6.5.1)

    1. Why was EPHB4 membrane localization not assessed or reported?

    We agree with the reviewer that this would be a very interesting point. Unfortunately, we had very limited amount of material and we did not have a proper working antibody.

    1. A key finding of the manuscript is that all six variants produce similar histological impacts on CAV1 and CD36 expression, denoting downstream impacts of EPHB4 genetic disruption. There is no granular data presented to support this claim. Additional discussion is also required to address how the authors anticipate variants in functionally distinct domains on either side of the plasma membrane to similarly impact downstream expression of CAV1/CD36. Mapping to available crystal structures in the Protein Data Bank (PDB) may be insightful to determine which variants may be most likely to have an impact on heterotetramer formation or to exert dominant negative effects on receptor function.

    As an appendix to this revision we have included a figure with representation images of all biopsies analysed to support our claim.

    The whole protein structure is not solved and only some individual domains are present in the Protein Data Bank making difficult to analyse the effect on the tetramer without crystallising the whole protein.

    1. Study limitations are not discussed and are significant. 5 of the 6 samples were from male patients, there are limitations to analyses of non-diverse patient ancestry, there is uncertainty regarding pathogenic contributions of variants in established DCM genes in 2/6 patients, data is limited to expression-only analyses highlighting need for additional functional modeling in cell or animal based systems.

    We have now included a limitation sections that includes all the points raised by the reviewer. It reads as follows:

    Although this study offers valuable insights to the potential implication of the Eph-ephrin signalling pathway in the development of DCM it has some limitations that need to be discussed. Despite finding increased presence of EPHB4 variants in the DCM population when compared to the healthy population, analysis of the identified variants in using different classifications (CADD and ACMG) not always predicted pathogenicity for these variants. For this reason, further experiments should be performed to determine the effect of every variant.

    It is also important to note that given the lower number of patients analysed these are not age and gender matched. The EPHB4 carrying DCM patients were younger than the DCM patients carrying a wild type EPHB4 sequence and mainly male. Finally, no biomaterial nor genetic testing from family related patients is available.

    1. Language used in conclusions overstates study findings ["our results confirm a crucial role of the Eph-ephrin signaling pathway in DCM" (page 3), "this study not only confirms the crucial role of EPHB4 in the heart..." (page 8)]. Change to "suggest" or "support".

    We have revised our discussion according to the limitations discussed in the previous remark and these words have been corrected.

    Major Methods Comments:

    1. DCM diagnostic criteria (clinical and imaging) for inclusion in the DZHK-TORCH study and the Institute for Cardiomyopathies Heidelberg registry should be stated or referenced. Likewise, describe and/or reference DCM exclusion criteria. State any relevant differences in DCM enrolment criteria for the two registries.

    We have now included our inclusion criteria in the methods and include two references to support this. The paragraph reads as follows:

    The criteria to be included in the study was reduced left ventricular ejection fraction (LVEF) <50% validated either with two independent image techniques or at two different time points with the same imaging technique. Furthermore, patients should include left ventricular dilation (LVEDD) >117% corrected with age and body surface according to the Henry-Formel formula (LVEDD= 45,3 * BSA1/3 – 0,03*Age –7,2). In both cases the heart were analysed either by echocardiography or magnetic resonance tomography (MRT)

    1. Describe how the final cohort of 573 DCM patients was reached. (All patients with DCM in the DZHK-TORCH study/Heidelberg registry? All patients with available exome data meeting QC standards and having available cardiac tissue?).

    From the 573 DCM patients, 100 have been recruited as part of the DZHK-TORCH registry and have been genome sequenced. Further 62 genomes and 411 exomes have been sequenced from patients of the cohort from the Institute for Cardiomyopathies (ICH) at the Heidelberg University Hospital.

    From this cohort, we selected 6 patients with and 6 without an EPHB4 variant and received heart tissue slides from the pathology department.

    1. State whether any family/segregation data is available for these patients.

    DCM4 has a mother and aunt (mother’s sister) who are also affected by CMP. In case of DCM6, the mother was also diagnosed with CMP. Unfortunately, no further biomaterial nor genetic testing of those individuals is available. This has been included in the new limitation sections as described above.

    1. Description of genetic testing methods are inadequate. Describe how genetic analyses were completed for each study/registry and how results were filtered/quality controlled. If sequencing methods were different across registries, state which patients were tested by which methods. If any testing was gene-targeted rather than whole exome/genome, list the specific DCM genes tested.

    All data has been sequenced using Illumina paired-end technology with either 2x100bp or 2x150bp. Exome enrichment was achieved using SureSelect Human All Exon V6 Target Enrichment (Agilent Genomics) was used. Bioinformatics analysis pipeline was based on “Best Practices Guideline” from the Genome Analysis Toolkit (GATK) (https://gatk.broadinstitute.org/hc/en-us). Besides the analysis for EPHB4, we assessed further genes associated with cardiomyopathies (ACTC1, ACTN2, ALPK3, BAG3, CRYAB, CSRP3, DES, DMD, DSC2, DSG2, DSP, FLNC, GLA, HCN4, HRAS, JPH2, JUP, KRAS, LAMP2, LDB3, LMNA, MIB1, MYBPC3, MYH7, MYL2, MYL3, MYPN, NEXN, PKP2, PLN, PRDM16, PRKAG2, PTPN11, RAF1, RBM20, RYR2, SCN5A, SHOC2, TAZ, TMEM43, TNNC1, TNNI3, TNNT2, TPM1,TTN, TTR, VCL).

    This is information has been included in the methods section.

    1. Provide additional detail for human cardiac biopsies. Was the same chamber/tissue biopsied in all samples? Is an endomyocardial biopsy available for all 573 patients included in this study? If not, were additional EPHB4 variants identified in patients without biopsy samples?

    All biopsies investigated are from left-ventricular tissue, accessed during cardiac catheterization.

    We did find additional, mainly non-coding variants in the cohort. However, as the focus on the study was on the histological analysis of the CD36 and CAV1 expression, we did restrict our analysis to our selected samples as described in the response to comment 2.

    1. Describe the source of the healthy control biopsy, alongside brief clinical detail establishing suitability as a control. Did DCM controls carry variants in known DCM genes (including truncating variants in RBM20 or TTN)? How were DCM controls selected?

    The healthy control biopsy was kindly donated by Prof. Dettmeyer from the University Gießen. This is a postmortem sample with unrelated cause of death. Cardiac biopsy was examined to discard any pathological alterations. This sample originates from a 27 years old female, and thus ideal as a healthy sample. This information has been included in the methods.

    1. List statistical analyses and associated experiments. (Page 5).

    Statistical tests have been included in the figure legend of each experiment. This reads as follows:

    (B) EPHB4 variant allelle frequency analysis. Each variant is compared in a paired wise manner between the two population. P value was calculated with a paired one-tailed Student’s t test comparing the frequencies of the different variants in the two populations.

    And

    (F) Quantification of CD36 and CAV1 expression in the immunohistochemistry analysis of patients biopsies. Data shown as mean ± SEM. In the case of CD36, P value was calculated using one sample one sample Wilcoxon test for DCM and one sample t test for DCM EPHB4. P value < 0.05 was considered significant. In the case of CAV1, P value was calculated with one sample t test for both cohorts. P value < 0.05 was considered significant. In both cases, the cohorts where compared to the mean of HD.

    1. List microscopes/equipment and software used to complete immunohistochemistry experiments. Describe imaging parameters to facilitate comparisons between treatments in Figure 1C-E.

    Immunohistochemistry was imaged in a Leica Stellaris confocal microscope. All images were obtained with 63x magnification and the same laser and gain intensities. Images were acquired using the software LAS X (Leica, version 4.4) and quantified using the Volocity Software (Quorum Technologies, version 6.5.1)

    This paragraph has now been included in the methods section.

    1. Please reword the following passage, which is almost verbatim to the same passage in Nicin et al. 2022.

    Page 4

    **"In brief, a combination of two human snRNA-seq datasets was used. Data from healthy cardiac tissue from the septum of 14 individuals in the Litvinukova et al. study and data from location-matched hypertrophic cardiac tissues from five patients with aortic stenosis."

    Nicin et al. 2022 (https://doi.org/10.1038/s44161-022-00019-7)**

    "Two human snRNA-seq datasets were used: data from healthy cardiac tissue from the septum of 14 individuals in the Litvinukova et al. study and data from location-matched hypertrophic cardiac tissues from five patients with aortic stenosis."

    We have reworded the paragraph in the methods sections. Now it reads as follows:

    Healthy cardiac tissue data was derived from the cardiac septum of 14 individuals 15. Subsequently, it was integrated with data from the septum of hypertrophc cardiac tissue from 5 patients with aortic stenosis 16.

    Minor Comments:

    1. Results: List source for Non-Finnish European Control cohort (gnomAD) (Page 5).

    The Non-Finnish European Control cohort (gnomAD) was obtained from https://gnomad.broadinstitute.org/. This information has been included in the methods section.

    1. Discussion: "all DCM patients" (page 6) requires clarification.

    We have made clear that this refers to the patients analysed in this study. The new sentence reads as follows:

    Furthermore, our results stress the importance of the endothelial CD36 in the onset of cardiac disease as all DCM patients analysed by immunohistochemistry show a downregulation of CD36 in the endothelium and warrant a more detailed assessment of genes involved in vascular function20

    1. Discussion: Define acronyms. CSF, IL4, LPS (Page 7)

    We have defined the acronyms in the discussion. The new sentence reads as follows:

    CD36 expression is upregulated by the nuclear hormone transcription factor Peroxisome Proliferator-Activated Receptor-Gamma (PPAR-ɣ), cerebrospinal fluid (CSF) cytokines and Interleukin-4 (IL4). In the other hand, lipopolysaccharides (LPS) and dexamethasone downregulate its expression In microvascular endothelial cells, CD36 is downregulated by lysophosphatidic acid.

    1. Table 1. Table is confusingly arranged. It would make more sense to organize the table by cDNA/AAchange to better correspond to Figure 1A. List the impacted protein domain for each variant in a separate column. It is also unclear how DCM allele frequencies were calculated as the reported number of patients (DCM1-6) carrying each variant do not universally correspond to the listed allele frequencies (see AFs of 0.0052 and 0.0208). Clarification should be added to the legend so it is clear to the reader how these frequencies were determined

    In case of the EPHB4 variants table, we agree with the reviewer and to make the table more understandable we have removed the first three columns, which are the same for all variants. This information has been included in the table legend. Nevertheless, this information has been kept in the new Supplementary table 1 that contains the variants on the other DCM causing genes.

    Regarding the calculation of the allele frequency we made by dividing the number of alleles found in the population by the total number of alleles in the population. This information has been included in the methods.

    We want to note that we performed a mistake in the original table. We had calculated the frequencies by dividing the number of alleles by the number of individuals in the population. We have now corrected both Table 1 and Figure 1B.

    1. Figure 1B. Add variant labels. Indicate relevant p-values for each variant. It is unclear to which comparison the p = 0.024 belongs. State in legend that 2 variants were omitted (presumably due to absence from gnomAD)

    No variants were omitted in the representation of Figure 1B. Some of them have the same allele frequency in the DCM population and thus, the individual data points appear overlapping. The variants that were not detected in the genomAD population were considered as 0 for the representation and for the analysis.

    For the comparison with P=0.024 (now corrected to 0.0011) between the two groups we have performed a one tail paired t test comparing the frequencies in both populations. The information regarding the test has been included in the figure legend and included in the methods section as indicated above.

    1. Figure 1E. Add label to indicate which EPHB4 variant is depicted.

    The DCM sample from which the images originates is now indicated in Figure 1E.

    **
    **Referees cross-commenting****

    As is, this manuscript is not ready for publication. Our comments are in complete alignment. Like the other reviewer, I also emphasize the need for other DCM genes tested to be listed. I also reiterate that any similarly worded passages to other published material must be corrected

    Reviewer #1 (Significance): This study presents genetic and expression data on a novel DCM gene candidate (EPHB4) from a European cohort of 573 DCM patients. This work is of interest as much of genetic DCM remains unexplained and identification of novel genes and pathways will be critical to advance understanding of the disease and to develop novel treatments. Reported data will be of greatest interest to cardiovascular practitioners and translational/basic researchers working with genetic heart disease/DCM. The fact that cardiac tissue was available for histological analyses for all six patients is an asset. There are considerable weaknesses to the paper, as written. There is a lack of detail in the included genetic methods and results. While the premise of the study is intriguing, additional detail is required for identified TTN and RBM20 truncating variants and additional discussion is needed to resolve confusion regarding reported allele frequencies and benign/likely benign Clinvar classifications. Because study design is restricted to genetic and expression analyses, reported data do not address possible pathogenic mechanisms. Overall, there is insufficient data presented to confirm a role for EPHB4 in causing DCM. Manuscript-specific (as-opposed to study specific) weaknesses include insufficient methods detail, a lack of clarity in the presented genetic and expression data (particularly Figure 1), insufficiently described study limitations, and overstated study conclusions. These scientific and manuscript issues will need to be addressed for the manuscript to be suitable for publication.

    Reviewer fields of expertise: cardiovascular genetics, DCM.

    Insufficient expertise to evaluate statistical methods.

    Reviewer #2 (Evidence, reproducibility and clarity):
    I reviewed a paper by Luxan et al. describing EPHB4 variants as a novel disease gene for dilated cardiomyopathy (DCM).

    The short report is interesting, however, not enough evidence is given to convince me EPHB4 is indeed a novel disease gene for DCM. More work is needed before this can be published.

    Major points:

    1. Genetics: two individuals have EPHB4 variants together with DCM causing TTN tv or RBM20 variants. Which other DCM genes were excluded for the remaining four cases? GnomAD MAF of 0.008748404 suspiciously high.

    So overall the small case number makes it hard to judge whether these are truly pathogenic variants.
    Could the authors attempt co-segregation of DCM with EPHB4 variant in families?

    Unfortunately we do not have family information from these patients. We have included this in the new limitation sections in the discussion that reads as follows:

    Although this study offers valuable insights to the potential implication of the Eph-ephrin signalling pathway in the development of DCM it has some limitations that need to be discussed. Despite finding increased presence of EPHB4 variants in the DCM population when compared to the healthy population, analysis of the identified variants in using different classifications (CADD and ACMG) not always predicted pathogenicity for these variants. For this reason, further experiments should be performed to determine the effect of every variant.

    It is also important to note that given the lower number of patients analysed these are not age and gender matched. The EPHB4 carrying DCM patients were younger than the DCM patients carrying a wild type EPHB4 sequence and mainly male. Finally, no biomaterial nor genetic testing from family related patients is available.

    1. Only CADD tools was used for pathogenicity, several tools should be used. Is the structure solved? Structural predictions on the consequences of the variants should be done.

    We have now included the ACMG classification in Table 1. As discussed above in the comments of Reviewer 1, some of the variants are classified as benign or likely benign. For this reason we have now toned down our conclusion suggesting that the EPHB4 may not be sufficient to trigger DCM but act as modifiers. This is supported by the fact that the histological analysis revealed that the patients carrying EPHB4 variants are similar among themselves and different to the other patients. Furthermore, our hypothesis is also supported by the fact that those patients carrying benign or potentially benign variants also carry another variant and the fact that they even have lower LVEF. The new classification has been included in the results and discussion sections and it reads as follows:

    Nevertheless, the classification of the variants according to the American College of Medical Genetics (ACMG) 25 suggests that two of the variants are benign, two likely benign, one likely pathogenic and one variant of uncertain significance (Table1). Nevertheless, two of the patients carrying one benign or likely benign also carry another variant classified as likely pathogenic or of uncertain significance (Table 1) and interestingly, the average LVEF of the two patients with multiple EPHB4 variants is 17.5 compared to 38,67 for the remaining 4 patients with only one EPHB4 variant and 35,17 for the six non-EPHB4 variant-carriers. Although the sample number only allows for a semi-quantitively analysis, it still hints at a possible EPHB4-variant effect, which certainly needs verification in a larger cohort.

    And

    Our analysis identified several variants in EPHB4 enriched in a cohort of DCM patients. According to the CADD score prediction, all these variants have a deleterious potential. Nevertheless, the ACMG classified some of the variants as benign or potentially benign. Also the fact, that one variant has identified in two non-related patients suggests that this variant may be benign for the protein. Nevertheless, two of the patients carrying a benign or potentially benign variant also carried another potentially pathogenic or of uncertain significance. During Eph-ephrin signalling, the binding of the ligand induces Eph receptor heterotetramers to initiate the signalling via Eph–Eph cis interactions30. Thus, variant EPHB4 molecules could have a dominant negative effect on these heterotetramers, and while maybe not completely abrogating its function, reducing the functionality of the heterotetramers. This observation could explain why the presence of one variant copy in the DCM patients of our cohort would be sufficient to reduce the activity of the Eph-ephrin signalling pathway. Although this shows that some of the variants may indeed not be the sole cause for DCM it shows that the Eph-ephrin signalling pathway, and in particular EPHB4 may be important for the development of DCM.

    Only parts of the protein have been resolved and present in the Protein Data Base.

    1. The microscopy Figure 1C-E is not convicing. Only one sample shown while 6 were available/investigated. I would not be comfortable to identify cardiomyocytes/endothelial cells from these sections

    As an appendix to this document, we included figures with images obtained from all the analysed patients. These were not included on the original figure for space reasons.

    These sections are perfect to identify cardiomyocytes and endothelial cells in cardiac tissue. First, endothelial cells, that form the microvasculature are labelled with ULEX, a well known marker of endothelial cells. Secondly, cardiomyocytes are really big cells easy to score for their size and location between the capillaries in the heart. Other cells present in the heart, like fibroblasts, macrophages, or pericytes would also be located in the space left in between cardiomyocytes but would need to be labelled for visualization. We believe that our interpretation of the immunohistochemistry pictures is correct.

    1. Functional work is needed to understand the interplay between EPHB4, CAV1 and CD36. Such as transfecting mutant EPHB4 into cells and probing for altered localisation/attachment of binding partners, most likely in endothelial - cardiomyocyte co-culture systems.

    Our study is based in our previous murine study in which we showed that the deletion of EphB4 or its ligand ephrinB2 would induce a phenotype similar to DCM in mice. At the molecular level, defects in the Ephb4 are linked to compromised caveolar function and reduced CAV1 phosphorylation, which involves the kinase Src, a known mediator of Eph receptor signalling. In the healthy heart, caveolar transport is required for the membrane translocation and correct function of fatty acid translocase FAT/CD36, which mediates the uptake of fatty acids. The objective of this follow up study was to study whether we could identify EPHB4 mutations in DCM patients. As seen in the results we have observed that there is an enrichment of EPHB4 variants in the DCM population. We think that the previous study supports our conclusions and hope that the reviewer agrees with us. Nevertheless, we agree with the reviewer that functional assays could be performed with every variant. We have included this in the new limitation sections of the manuscript described above.

    Minor points:

    1. Figure 1B does not make sense

    Figure 1B confirms the enrichment of EPHB4 mutations in the DCM population. We have corrected the labelling to make this clearer. We have now labelled the figure “EPHB4 variant allele frequency in control and DCM population”.

    1. Statistics: Which tests were performed, if normality tests were applied, which one was used?

    The tests used for every comparison are included in the figure legend. In case of EPHB4 variant allele frequency, we performed a paired one-tailed Student’s t test comparing the frequencies of the different variants in the two populations. In case of the CD36 and CAV1 quantifications, we performed a two-tailed one sample t test. In this case, we compare the expression of CD36 and CAV1 to an hypothetical healthy population with mean equal 1 as que have used this value for normalization.

    1. Please do not use contractions, e.g. 'can't' in discussion section

    Contractions have been removed from the manuscript.

    **
    ** Referees cross-commenting****

    Overall I agree with the other reviewer on the points raised.

    Reviewer #2 (Significance): Description of EPHB4 as a novel DCM gene is of interest, but the current data are not convincing enough to make this statement.

    Mechanistic work on the interplay of endothelial cells and cardiomyocytes and consequences of EPHB4 variants would make it a very compelling story.

    Reviewer #3 (Evidence, reproducibility and clarity): Summary:

    The authors of this manuscript studied the prevalence of a population of Ephrin type-B receptor 4 (EPHB4) in a cohort of 573 DCM patients and found six new EPHB4 variants, possibly pathogenic based on the Combined Annotation Dependent Depletion (CADD) score and population frequency. Moreover, the authors perform immunofluorescence (IF) and histologic analysis on 6 EPHB4 variant carrying DCM patients, 6 DCM patients with wild type EPHB4 and one healthy control biopsy and found dysregulation of Caveolin 1 (CAV1) and CD36 (which are implicated in fatty acid transport in endothelial cells and cardiomyocytes) in both groups of DCM patients.

    Major comments:

    • Additional experiments are necessary to prove the hypothesis: for example, co-IF staining with endothelial markers should be provided. IF should be supported by western blots and qPCR.

    The objective of this study was to explore whether we could identify EPHB4 mutants in a DCM cohort. Interestingly we have shown that EPHB4 mutations are enriched in the DCM population when compared to the general population. Nevertheless, we agree with the reviewer that a more in depth mechanistic study would improve the significance of the study. We have included a limitations section that reads as follows:

    Although this study offers valuable insights to the potential implication of the Eph-ephrin signalling pathway in the development of DCM it has some limitations that need to be discussed. Despite finding increased presence of EPHB4 variants in the DCM population when compared to the healthy population, analysis of the identified variants in using different classifications (CADD and ACMG) not always predicted pathogenicity for these variants. For this reason, further experiments should be performed to determine the effect of every variant.

    It is also important to note that given the lower number of patients analysed these are not age and gender matched. The EPHB4 carrying DCM patients were younger than the DCM patients carrying a wild type EPHB4 sequence and mainly male. Finally, no biomaterial nor genetic testing from family related patients is available.

    • The DCM samples with wild type EPHB4, have no CD36: the mechanism by which a mutation in another gene could affect the Eph-ephrin signaling pathway should be at least discussed.

    These patients do not have any mutation on EPHB4. Based in the literature and the previous murine study show that the Eph-ephrin signaling pathway is upstream of CD36. For these reasons we believe that our observation that shows that CD36 expression is reduced in all DCM patients confirms the important role of CD36 in cardiac homeostasis and the development of DCM. We further, as indicated in the discussion, other genes crucial for fatty acid transport may be involved in cardiac disease and thus, this study may help identify new diagnostic or therapeutic targets.

    • The authors should discuss and possibly prove the correlation between mutant EPHB4 and CD36 and CAV1 expression and localization in endothelial cells vs cardiomyocytes and explain the mechanistic implications of co-localization of CAV1 with CD36.

    In a previous study we showed that the deletion of EphB4 or its ligand ephrinB2 would induce a phenotype similar to DCM in mice. At the molecular level, defects in the Ephb4 are linked to compromised caveolar function and reduced CAV1 phosphorylation, which involves the kinase Src, a known mediator of Eph receptor signalling. In the healthy heart, caveolar transport is required for the membrane translocation and correct function of fatty acid translocase FAT/CD36, which mediates the uptake of fatty acids. We have expanded the introduction to explain the relationship between these molecules. It reads as follows:

    Mechanistically, EPHB4 deficient endothelial cells are characterized by compromised caveolar function and reduced Caveolin 1 (CAV1) phosphorylation. EPHB4 is required for the phosphorylation of CAV1 at Tyr-149. The phosphorylation of CAV1 promotes the release of caveolae from the plasma membrane10. Caveolae are required for the correct membrane translocation of the fatty acid translocase FAT/CD3611 and fatty acids are used by cardiomyocytes to obtain about 50% to 70% of their energy12. Absence of CD36 in cardiomyocytes reduces fatty acid uptake by the cardiac muscle cells13 and accelerates the progression from compensated hypertrophy to heart failure14. Finally, some cardiomyopathies a causally related to defects in the synthesis of the proteins required for fatty acid uptake in the heart15.

    • The available snRNAseq raw data are from normal subjects and aortic stenosis patients who are different from DCM patients. A better dataset would be the one from Reichart D, et al. Pathogenic variants damage cell composition and single cell transcription in cardiomyopathies. Science 2022.

    The single nucleus RNA sequencing data was used in an exploratory manner to study whether EPHB4 would also be expressed in cardiomyocytes. We did not perform any study on gene expression comparing the two groups. We believe that the use of this dataset is justified. We hope that the reviewer agrees with us.

    • Furthermore, the link between the analysis done on the published snRNA seq datasets and the authors' own data is not clearly explained.

    As we stated above and in the methods, we have used the single nucleus RNA sequencing to explore whether cardiomyocytes express EPHB4. The sentence in the methods reads as follows:

    The single-nucleus-RNA-sequencing data set generated in the paper by Nicin et al.14 was used to explore EPHB4 expression in human cardiac cells

    • DCM1 and DCM 3 carry 2 EPHB4 variants: please describe if the phenotype was more severe.

    As discussed above in the response to reviewer 1, the two patients with multiple EPHB4 variants present an average LVEF (echo) of 17.5 compared to 38,67 for the remaining 4 patients with only one EPHB4 variant and 35,17 for the six non-EPHB4 variant-carriers. Although the sample number only allows for a semi-quantitively analysis, it still hints at a possible EPHB4-variant effect, which certainly needs verification in a larger cohort.

    This information has been included in the manuscript and reads as follows:

    and interestingly, the average LVEF of the two patients with multiple EPHB4 variants is 17.5 compared to 38,67 for the remaining 4 patients with only one EPHB4 variant and 35,17 for the six non-EPHB4 variant-carriers. Although the sample number only allows for a semi-quantitively analysis, it still hints at a possible EPHB4-variant effect, which certainly needs verification in a larger cohort.

    • Provide p values on suppl table 1. The 2 groups are not matched by age and maybe gender, and this could affect the histological findings.

    We have not performed any comparison between the two groups in the characteristics shown in supplementary table 1. Nevertheless, we agree with the reviewers that the fact that the patients are not matched in age and gender is a limitation to our study. We have acknowledged this in the new included limitations section that is mentioned above.

    • Please discuss why in the DCM population the EPHB4 variant is enriched as compared with controls. Causal role? Modifiers?

    The deletion of EphB4 and its ligand ephrin-B2 induce DCM in mouse. The objective of this study was to determine whether there would be mutations in EPHB4 associated to DCM. We agree with the reviewer that in depth mechanistic studies both in vivo and in vitro would be required to determine the exact role of the here identified mutations in the development of DCM. This has been acknowledged in the new limitations sections and indicated in the discussion of the results as follows:

    Finally, this study not only supports the crucial role of EPHB4 in the heart, but it also corroborates the importance of CD36 and CAV1 for the cardiac health, and has the potential to improve diagnosis and risk stratification tools for DCM. Nevertheless, whether mutations in EPHB4 are causative or modifiers of the disease should be further studied. In addition, as other genes crucial for fatty acid transport may be involved in cardiac disease, this study may help identify new diagnostic or therapeutic targets.

    • The data and the methods are presented in such a way that they could be reproduced however,

    We thank the reviewer for the positive comment on our methods section.

    • At least 2 more healthy controls should be included, and the DCM groups should be matched by gender and age.

    Healthy donor biopsies are very rare and difficult to obtain. Although we agree with the reviewer that this could strengthen our study, we cannot add more healthy biopsies. We hope the reviewer understands this.

    As stated above, we have included a limitation section in the manuscript discussing the issue with the gender and age.

    • The causal mutation of the DCM patients should be provided.

    Only 35% of DCM cases have been related to mutations in genes encoding cytoskeletal, sarcomere or nuclear envelope proteins. In our case, the DCM patients that we use do not carry a variant in any of the DCM known genes. We have now expanded the methods sections explaining the inclusion criteria for the DCM patients including this issue:

    The criteria to be included in the study was reduced left ventricular ejection fraction (LVEF) <50% validated either with two independent image techniques or at two different time points with the same imaging technique. Furthermore, patients should include left ventricular dilation (LVEDD) >117% corrected with age and body surface according to the Henry-Formel formula (LVEDD= 45,3 * BSA1/3 – 0,03*Age –7,2). In both cases the heart were analysed either by echocardiography or magnetic resonance tomography (MRT).

    Minor comments:

    • I would explain in more detail the interactions among EPHB4, CD36 and CAV1 in the introduction, as the readers may not be familiar with this pathway.

    We have completed the introduction expanding the paragraph where the relationship between EPHB4, CD36 and CAV1 is presented. It now reads as follows:

    Mechanistically, EPHB4 deficient endothelial cells are characterized by compromised caveolar function and reduced Caveolin 1 (CAV1) phosphorylation. EPHB4 is required for the phosphorylation of CAV1 at Tyr-149. The phosphorylation of CAV1 promotes the release of caveolae from the plasma membrane10. Caveolae are required for the correct membrane translocation of the fatty acid translocase FAT/CD3611 and fatty acids are used by cardiomyocytes to obtain about 50% to 70% of their energy12. Absence of CD36 in cardiomyocytes reduces fatty acid uptake by the cardiac muscle cells13 and accelerates the progression from compensated hypertrophy to heart failure14. Finally, some cardiomyopathies a causally related to defects in the synthesis of the proteins required for fatty acid uptake in the heart15.

    • Panel B in Fig 1 shows 4 variants and not 6.

    All variants are shown in the panel As stated in the response to reviewer 1, it the fact that some variants have the same value that induces to think that only four are shown. The variants that do not appear in the genomAD have been considered 0 for this analysis.

    • IF in Fig 1: make sure that control and DCM are at the same magnification.

    Both control and DCM are at the same magnification. The reason why it looks different is the DCM phenotype. Cardiomyocytes are hypertrophic in the in the disease samples giving the impression that they are shown in a higher magnification.

    • The authors analyze snRNA seq data from available datasets and not from their own patients: so, the paragraph title in the method section should be changed as it is misleading.

    We have changed the title of this section of the methods. We have labelled it now “Analysis of single-nucleus-RNA-sequencing”.

    Reviewer #3 (Significance): Despite the main focus of the manuscript is EPHB4, dysregulation of CD36 and its interaction with CAV1 seem to be a common mechanism in the pathogenesis of all DCM. The significance of these findings is higher than the role of EPHB4 alone and should be improved.
    Metabolic abnormalities, mainly affecting the fatty acid metabolism, have been described as causes or modifiers of DCM pathogenesis but in my knowledge the role of EPHDB4, CD36 and CAV 1 have not been studied in human tissues. The discovery of the mechanisms through which dysregulation of metabolism is induced by DCM genetic mutations would be an advance in the field. However, the paper in the present form is not going to have a significant impact. There is no clear connection between the sets of experiments and more mechanistic experiments should be provided to prove causality. This may take months or even years depending on the availability of human tissues and resources.

    The type of audience interested in this research are mainly translational scientists mainly in the field of genetic cardiomyopathies. Furthermore, the elucidation of the metabolic effects of genetic mutations on DCM evolution may be of interest in the field of heart failure in general.

    The focus of my research is genetic and molecular pathogenesis of cardiomyopathies.

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Summary:

    The authors of this manuscript studied the prevalence of a population of Ephrin type-B receptor 4 (EPHB4) in a cohort of 573 DCM patients and found six new EPHB4 variants, possibly pathogenic based on the Combined Annotation Dependent Depletion (CADD) score and population frequency. Moreover, the authors perform immunofluorescence (IF) and histologic analysis on 6 EPHB4 variant carrying DCM patients, 6 DCM patients with wild type EPHB4 and one healthy control biopsy and found dysregulation of Caveolin 1 (CAV1) and CD36 (which are implicated in fatty acid transport in endothelial cells and cardiomyocytes) in both groups of DCM patients.

    Major comments:

    • Additional experiments are necessary to prove the hypothesis: for example, co-IF staining with endothelial markers should be provided. IF should be supported by western blots and qPCR.
    • The DCM samples with wild type EPHB4, have no CD36: the mechanism by which a mutation in another gene could affect the Eph-ephrin signaling pathway should be at least discussed.
    • The authors should discuss and possibly prove the correlation between mutant EPHB4 and CD36 and CAV1 expression and localization in endothelial cells vs cardiomyocytes and explain the mechanistic implications of co-localization of CAV1 with CD36.
    • The available snRNAseq raw data are from normal subjects and aortic stenosis patients who are different from DCM patients. A better dataset would be the one from Reichart D, et al. Pathogenic variants damage cell composition and single cell transcription in cardiomyopathies. Science 2022.
    • Furthermore, the link between the analysis done on the published snRNA seq datasets and the authors' own data is not clearly explained.
    • DCM1 and DCM 3 carry 2 EPHB4 variants: please describe if the phenotype was more severe.
    • Provide p values on suppl table 1. The 2 groups are not matched by age and maybe gender, and this could affect the histological findings.
    • Please discuss why in the DCM population the EPHB4 variant is enriched as compared with controls. Causal role? Modifiers?
    • The data and the methods are presented in such a way that they could be reproduced however,
    • At least 2 more healthy controls should be included, and the DCM groups should be matched by gender and age.
    • The causal mutation of the DCM patients should be provided.

    Minor comments:

    • I would explain in more detail the interactions among EPHB4, CD36 and CAV1 in the introduction, as the readers may not be familiar with this pathway.
    • Panel B in Fig 1 shows 4 variants and not 6.
    • IF in Fig 1: make sure that control and DCM are at the same magnification.
    • The authors analyze snRNA seq data from available datasets and not from their own patients: so, the paragraph title in the method section should be changed as it is misleading.

    Significance

    Despite the main focus of the manuscript is EPHB4, dysregulation of CD36 and its interaction with CAV1 seem to be a common mechanism in the pathogenesis of all DCM. The significance of these findings is higher than the role of EPHB4 alone and should be improved.

    Metabolic abnormalities, mainly affecting the fatty acid metabolism, have been described as causes or modifiers of DCM pathogenesis but in my knowledge the role of EPHDB4, CD36 and CAV 1 have not been studied in human tissues. The discovery of the mechanisms through which dysregulation of metabolism is induced by DCM genetic mutations would be an advance in the field. However, the paper in the present form is not going to have a significant impact. There is no clear connection between the sets of experiments and more mechanistic experiments should be provided to prove causality. This may take months or even years depending on the availability of huma tissues and resources.

    The type of audience interested in this research are mainly translational scientists mainly in the field of genetic cardiomyopathies. Furthermore, the elucidation of the metabolic effects of genetic mutations on DCM evolution may be of interest in the field of heart failure in general.
    The focus of my research is genetic and molecular pathogenesis of cardiomyopathies.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    I reviewed a paper by Luxan et al. describing EPHB4 variants as a novel disease gene for dilated cardiomyopathy (DCM).

    The short report is interesting, however, not enough evidence is given to convince me EPHB4 is indeed a novel disease gene for DCM. More work is needed before this can be published.

    Major points:

    1. Genetics: two individuals have EPHB4 variants together with DCM causing TTN tv or RBM20 variants. Which other DCM genes were excluded for the remaining four cases? GnomAD MAF of 0.008748404 suspiciously high.
      So overall the small case number makes it hard to judge whether these are truly pathogenic variants.
      Could the authors attempt co-segregation of DCM with EPHB4 variant in families?
    2. Only CADD tools was used for pathogenicity, several tools should be used. Is the structure solved? Structural predictions on the consequences of the variants should be done.
    3. The microscopy Figure 1C-E is not convicing. Only one sample shown while 6 were available/investigated. I would not be comfortable to identify cardiomyocytes/endothelial cells from these sections
    4. Functional work is needed to understand the interplay between EPHB4, CAV1 and CD36. Such as transfecting mutant EPHB4 into cells and probing for altered localisation/attachment of binding partners, most likely in endothelial - cardiomyocyte co-culture systems.

    Minor points:

    1. Figure 1B does not make sense
    2. Statistics: Which tests were performed, if normality tests were applied, which one was used?
    3. Please do not use contractions, e.g. 'can't' in discussion section

    Referees cross-commenting
    Overall I agree with the other reviewer on the points raised.

    Significance

    Description of EPHB4 as a novel DCM gene is of interest, but the current data are not convincing enough to make this statement.

    Mechanistic work on the interplay of endothelial cells and cardiomyocytes and consequences of EPHB4 variants would make it a very compelling story.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary:

    This research article describes genetic identification and expression analyses of six Ephrin type-B receptor 4 (EPHB4) variants identified in patients with dilated cardiomyopathy (DCM). Variants were identified Variants were identified in a cohort of 573 patients enrolled through the multicenter DZHK-TORCH (TranslatiOnal Registry for CardiomyopatHies) study and the Institute for Cardiomyopathies Heidelberg registry. Expression of downstream molecules, CAV1 and CD36, was assessed in human cardiac tissues by immunohistochemistry. EPHB4 cardiac expression was assessed using recently published single-cell/nucleus RNA sequencing data (Nicin et al 2022) incorporating siRNA-seq data from two other studies (healthy cardiac tissue, Litvinukova et al 2020) and (hypertrophic/aortic stenosis, Nicin et al. 2020).

    Major Comments:

    1. Details of identified truncating RBM20 and TTN variants must be provided. These should be integrated into Table 1 alongside each co-occurring EPHB4 variant. List whether the TTN truncating variant is located in the A-band and whether these variants would be adjudicated as pathogenic/likely pathogenic, variant of uncertain significance by ACMG and/or similarly refined DCM criteria (Morales et al. 2020, Circ-Genom Precis Med).
    2. Discuss co-occurrence of multiple EPHB4 variants in two patients (DCM1, DCM3) and identification of 2 EPHB4 variants in more than one proband.
    3. Three of the six variants (p.Lys635Asn, Val113Ile, Glu890Asp) are classified as Clinvar Benign/Likely Benign. Additionally, p.Glu890Asp has been identified in 50 homozygotes in gnomAD non-Finnish European population. These data cast doubt on the pathogenicity of these variants. These classifications, as well as VUS classification of p.Pro79Leu, should be listed in Table 1. The authors should reconcile the benign/likely benign Clinvar classifications with their presented evidence for pathogenicity in the discussion.
    4. CD36 and CAV1 expression are not quantified. Qualitatively, it is difficult to confirm CD36 reduction in DCM and disruption in EPHB4 variant samples as imaging parameters are not specified and do not appear to be standardized across treatments. Clearly state (either in the figure legend or in the methods) whether identical imaging parameters were used across panels 1C-1E. Note any differences in these parameters.
    5. Why was EPHB4 membrane localization not assessed or reported?
    6. A key finding of the manuscript is that all six variants produce similar histological impacts on CAV1 and CD36 expression, denoting downstream impacts of EPHB4 genetic disruption. There is no granular data presented to support this claim. Additional discussion is also required to address how the authors anticipate variants in functionally distinct domains on either side of the plasma membrane to similarly impact downstream expression of CAV1/CD36. Mapping to available crystal structures in the Protein Data Bank (PDB) may be insightful to determine which variants may be most likely to have an impact on heterotetramer formation or to exert dominant negative effects on receptor function.
    7. Study limitations are not discussed and are significant. 5 of the 6 samples were from male patients, there are limitations to analyses of non-diverse patient ancestry, there is uncertainty regarding pathogenic contributions of variants in established DCM genes in 2/6 patients, data is limited to expression-only analyses highlighting need for additional functional modeling in cell or animal based systems.
    8. Language used in conclusions overstates study findings ["our results confirm a crucial role of the Eph-ephrin signaling pathway in DCM" (page 3), "this study not only confirms the crucial role of EPHB4 in the heart..." (page 8)]. Change to "suggest" or "support".

    Major Methods Comments:

    1. DCM diagnostic criteria (clinical and imaging) for inclusion in the DZHK-TORCH study and the Institute for Cardiomyopathies Heidelberg registry should be stated or referenced. Likewise, describe and/or reference DCM exclusion criteria. State any relevant differences in DCM enrolment criteria for the two registries.
    2. Describe how the final cohort of 573 DCM patients was reached. (All patients with DCM in the DZHK-TORCH study/Heidelberg registry? All patients with available exome data meeting QC standards and having available cardiac tissue?).
    3. State whether any family/segregation data is available for these patients.
    4. Description of genetic testing methods are inadequate. Describe how genetic analyses were completed for each study/registry and how results were filtered/quality controlled. If sequencing methods were different across registries, state which patients were tested by which methods. If any testing was gene-targeted rather than whole exome/genome, list the specific DCM genes tested.
    5. Provide additional detail for human cardiac biopsies. Was the same chamber/tissue biopsied in all samples? Is an endomyocardial biopsy available for all 573 patients included in this study? If not, were additional EPHB4 variants identified in patients without biopsy samples?
    6. Describe the source of the healthy control biopsy, alongside brief clinical detail establishing suitability as a control. Did DCM controls carry variants in known DCM genes (including truncating variants in RBM20 or TTN)? How were DCM controls selected?
    7. List statistical analyses and associated experiments. (Page 5).
    8. List microscopes/equipment and software used to complete immunohistochemistry experiments. Describe imaging parameters to facilitate comparisons between treatments in Figure 1C-E.
    9. Please reword the following passage, which is almost verbatim to the same passage in Nicin et al. 2022.

    Page 4
    "In brief, a combination of two human snRNA-seq datasets was used. Data from healthy cardiac tissue from the septum of 14 individuals in the Litvinukova et al. study and data from location-matched hypertrophic cardiac tissues from five patients with aortic stenosis."

    Nicin et al. 2022 (https://doi.org/10.1038/s44161-022-00019-7)
    "Two human snRNA-seq datasets were used: data from healthy cardiac tissue from the septum of 14 individuals in the Litvinukova et al. study and data from location-matched hypertrophic cardiac tissues from five patients with aortic stenosis."

    Minor Comments:

    1. Results: List source for Non-Finnish European Control cohort (gnomAD) (Page 5).
    2. Discussion: "all DCM patients" (page 6) requires clarification.
    3. Discussion: Define acronyms. CSF, IL4, LPS (Page 7)
    4. Table 1. Table is confusingly arranged. It would make more sense to organize the table by cDNA/AAchange to better correspond to Figure 1A. List the impacted protein domain for each variant in a separate column. It is also unclear how DCM allele frequencies were calculated as the reported number of patients (DCM1-6) carrying each variant do not universally correspond to the listed allele frequencies (see AFs of 0.0052 and 0.0208). Clarification should be added to the legend so it is clear to the reader how these frequencies were determined
    5. Figure 1B. Add variant labels. Indicate relevant p-values for each variant. It is unclear to which comparison the p = 0.024 belongs. State in legend that 2 variants were omitted (presumably due to absence from gnomAD)
    6. Figure 1E. Add label to indicate which EPHB4 variant is depicted.

    Referees cross-commenting

    As is, this manuscript is not ready for publication. Our comments are in complete alignment. Like the other reviewer, I also emphasize the need for other DCM genes tested to be listed. I also reiterate that any similarly worded passages to other published material must be corrected

    Significance

    This study presents genetic and expression data on a novel DCM gene candidate (EPHB4) from a European cohort of 573 DCM patients. This work is of interest as much of genetic DCM remains unexplained and identification of novel genes and pathways will be critical to advance understanding of the disease and to develop novel treatments. Reported data will be of greatest interest to cardiovascular practitioners and translational/basic researchers working with genetic heart disease/DCM. The fact that cardiac tissue was available for histological analyses for all six patients is an asset. There are considerable weaknesses to the paper, as written. There is a lack of detail in the included genetic methods and results. While the premise of the study is intriguing, additional detail is required for identified TTN and RBM20 truncating variants and additional discussion is needed to resolve confusion regarding reported allele frequencies and benign/likely benign Clinvar classifications. Because study design is restricted to genetic and expression analyses, reported data do not address possible pathogenic mechanisms. Overall, there is insufficient data presented to confirm a role for EPHB4 in causing DCM. Manuscript-specific (as-opposed to study specific) weaknesses include insufficient methods detail, a lack of clarity in the presented genetic and expression data (particularly Figure 1), insufficiently described study limitations, and overstated study conclusions. These scientific and manuscript issues will need to be addressed for the manuscript to be suitable for publication.

    Reviewer fields of expertise: cardiovascular genetics, DCM.

    Insufficient expertise to evaluate statistical methods.

  5. Version published to 10.1101/2022.12.16.22283442v2 on medRxiv
  6. Version published to 10.1101/2022.12.16.22283442v1 on medRxiv