Transcriptional drifts associated with environmental changes in endothelial cells

  1. Yalda Afshar
  2. Feyiang Ma
  3. Austin Quach
  4. Anhyo Jeong
  5. Hannah L Sunshine
  6. Vanessa Freitas
  7. Yasaman Jami-Alahmadi
  8. Raphael Helaers
  9. Xinmin Li
  10. Matteo Pellegrini
  11. James A Wohlschlegel
  12. Casey E Romanoski
  13. Miikka Vikkula
  14. M Luisa Iruela-Arispe

  • Curated by eLife

    eLife logo

    eLife assessment’

    This paper is of interest to a broad audience of cell biologists, and researchers who work with cultured endothelial cells. The work uncovers the impact of culture conditions on transcriptional changes of endothelial cells and demonstrates that some of these changes can be recovered by sheer forces or coculture. The authors provide valuable datasets which will be a good resource for the community.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Environmental cues, such as physical forces and heterotypic cell interactions play a critical role in cell function, yet their collective contributions to transcriptional changes are unclear. Focusing on human endothelial cells, we performed broad individual sample analysis to identify transcriptional drifts associated with environmental changes that were independent of genetic background. Global gene expression profiling by RNA sequencing and protein expression by liquid chromatography–mass spectrometry directed proteomics distinguished endothelial cells in vivo from genetically matched culture (in vitro) samples. Over 43% of the transcriptome was significantly changed by the in vitro environment. Subjecting cultured cells to long-term shear stress significantly rescued the expression of approximately 17% of genes. Inclusion of heterotypic interactions by co-culture of endothelial cells with smooth muscle cells normalized approximately 9% of the original in vivo signature. We also identified novel flow dependent genes, as well as genes that necessitate heterotypic cell interactions to mimic the in vivo transcriptome. Our findings highlight specific genes and pathways that rely on contextual information for adequate expression from those that are agnostic of such environmental cues.

Article activity feed

  1. Version published to 10.7554/elife.81370 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    In this well-written manuscript, Afshar et al demonstrated the significant transcriptional and proteomic differences between cultured human umbilical vein endothelial cells (HUVECs) and those freshly isolated from the cords. They showed that TGFbeta and BMP signaling target genes were enriched in cord cells compared to those in culture. Extracellular matrix (ECM) and cell cycle-related genes were also different between the two conditions. Because master regulators of EC shear stress response genes, KLF2 and KLF4, were downregulated in culture, the authors sought to restore the in vivo transcriptional profile with the application of shear stress in an orbital shaker and dextran-containing media for various time periods. They showed that after 48 hours of shear stress the transcriptional profile of sheared cells correlated with in vivo transcriptional profile more significantly than static cultures. They also showed, using single cell RNAseq, that EC-smooth muscle cell cocultures resulted in changes in TGFbeta and NOTCH signaling pathways and rescued 9% of the in vivo transcriptional signatures.

    This is an important study that was elegantly executed. The authors should also be commended for making their data public; thereby, creating a valuable resource for vascular biologists.

    We much appreciate the comments and thank the reviewer for the time and effort evaluating the study.

    Reviewer #2 (Public Review):

    The authors profiled the transcriptome and proteome of human umbilical vein endothelial cells freshly isolated from in vivo and compared that with the same cells exposed to in vitro culture under different conditions, including static culture, flow, and co-culture with smooth muscle cells. The experiments were properly designed and performed. The authors also provided a reasonable and sound interpretation of their findings. This study provides valuable insights into how the culturing conditions impact on gene expression, encouraging the field to select their in vitro work setting appropriately. Overall, the manuscript is well-written and easy to follow.

    Several notable strengths include:

    1. Parallel transcriptome- and proteome-wide profiling of endothelial cells enabling the unbiased interrogation of gene expression and a genome-wide view of the impact of in vitro culture on endothelial transcriptome.
    1. The innovative experimental design and comparisons were done with genetically identical ECs (from the same donors) in vivo and in vitro.
    1. The analyses were robust and provided novel information on flow-dependent and cell context-dependent gene regulation, with the native freshly isolated cells as a baseline.
    1. The donor samples used in this study were diverse including Asian, White, Black, Latino, and American Indian samples which reduce racial background bias.

    Some points that can strengthen the study:

    A clear description of experimental and analytical details (e.g. how the comparisons were made) and more in-depth interpretation and discussion of the results, e.g. the complete genes that are rescued by flow and co-culture and potential synergy of these factors.

    We thank the reviewer for highlighting the strengths and appreciate the comments on experimental and analytical details which have been now addressed in this revised manuscript. Specifically, we have expanded the discussion and included synergy and additional comments on the rescued genes. A clear description of experimental and analytical details (e.g. how the comparisons were made) and more in-depth interpretation and discussion of the results, e.g. the complete genes that are rescued by flow and co-culture and potential synergy of these factors are now included.

    Reviewer #3 (Public Review):

    Afshar et al. performed RNA-seq and LC-MS of in vivo and in vitro HUVECs to identify the role of culture conditions on gene expression. Given the widespread use of HUVECs to study EC biology, these findings are interesting and can help design better in vitro experiments. There have been previous papers that compared in vivo and in vitro HUVECs, however, the depth of sequencing and analysis in this manuscript identifies some novel effects which should be accounted for in future in vitro experiments using ECs.

    Strengths:

    1. Major findings of distinct pathways affected by cell culture are novel and interesting. The authors identify major effects on TGFb and ECM gene expression. They also corroborate previous findings of flow response pathways, namely KLF2/4 and Notch pathway regulation.
    1. Use of multiple genomic methods to profile effects of culture conditions. The LC-MS data showed a significant correlation with RNA-seq, however, the data were not as strong so not used for subsequent analyses.
    1. Use of scRNA-seq to show the dynamic effects of co-culture and shear stress on ECs is very novel. However, the heterogeneity in the EC populations is not discussed in this manuscript.

    We would like to thank the reviewer for the in-depth analysis of our study and for highlighting the novelty and strength of the data. Note that we included comments in relation to EC heterogeneity as part of the limitations of this study (in the Discussion).

    Weaknesses:

    1. The physiological relevance of these changes in gene expression is not demonstrated in the manuscript. The authors claim the significance of their data is to improve in vitro culture to better represent in vivo biology. Is this the case with orbital shear stress? Do they rescue some functional effects in ECs with long-term shear stress? An angiogenesis, barrier function, or migration assay for HUVECs exposed to different conditions would help answer this question. A similar assay for cells after EC-VSMC co-culture would validate the importance of these stimuli.

    The reviewer is correct, our manuscript did not expand into physiological read outs, we have now clearly acknowledged this as part of the limitations of the study. Notably, there is already extensive literature on the effects of different types of flow on several physiological parameters. For example, others have shown that laminar shear stress (by orbital or other means) reduces proliferation and migration (PMID: 31831023; PMID: 22012789, PMID: 12857765, PMID: 21312062, PMID: 15886673; PMID: 17323381), reduces inflammation (PMID: 34747636; PMID: 32951280), and improves barrier function (PMID: 20543206; PMID: 32457386 ; PMID: 12577139, PMID: 27246807; PMID: 31500313 ).

    From the onset, our objective was to bring granularity to transcriptional changes associated with the transition from in vivo to in vitro. Further, it was our goal to identify the cohorts of transcripts that could and those that could not be rescued by altering culture conditions. Because we had transcriptional information from the identical samples at a time that they were in the vessel, we have been able to fulfill our goal. We feel this is important, and currently missing data, that will be of value to many investigators.

    1. One explanation for the increased expression of ECM genes in vivo is that these cells are contaminated with VSMCs/fibroblasts. This could be very likely given that cells were not sorted or purified upon isolation. Expression of other VSMC or fibroblast-specific markers (i.e. CNN1, MYH11, SMTN, DCN, FBLN1) would help determine if there is some level of non-EC contamination.

    We thank the reviewer for this comment and prompted by this, we have included a new figure (Supplemental Figure 1 and new panels in Supplemental Figure 5) that directly address this concern.

    Amongst the several pieces of data, we included scRNAseq from cells that were immediately obtained from umbilical vein – three independent experiments sequenced together and showed in one UMAP (Supplemental Figure 1C). As can be appreciated, the very large majority of cells are endothelial and the only other cell types present were blood cells (erythrocytes and CD45+ cells). No smooth muscle cells or fibroblasts were detected. These three examples are indeed representative of a large number of scRNAseq datasets (35 from cords and cultures for this and other projects). Furthermore, our cultures are also routinely evaluated by FACS (one example has been provided in Supplemental Figure 1E). We do not find, as illustrated in that example, cells that are not positive for CD31 and VE-Cadherin.

    We hope this information reveals the rigor of our studies and convinces the reviewer that the transcriptional changes observed are from endothelial cells.

    1. The use of scRNA-seq in Figure 4 is interesting. There appear to be 2 distinct EC populations in the co-cultured ECs. What are the marker genes for the 2 populations?

    Indeed, we and others (Kalluri et al., 2019) have noticed two distinct populations in the in vivo and also in cultured ECs, as pointed by the reviewer. Evaluation as to these two subpopulations reflect two transcriptionally distinct groups or different states of cyclic expression patterns, requires more thorough analysis and lineage tracing studies and distinct from the focus of this manuscript. Nonetheless, we have made a point in the revised manuscript to highlight these possibilities.

    Reference: Kalluri, AS, Vellarikkal, SK, Edelman, ER, Nguyen, L, Subramanian, A, Ellinor PT, Regev, A, Kathiresan, S, Gupta, RM. Single Cell Analysis of the Normal Mouse Aorta Reveals Functionally Distinct Endothelial Cell Populations. Circulation, 2019. 140:147-163.

    1. The modest shifts in gene expression with shear stress and co-culture could be attributed to the batch effect. The authors describe 1 batch correction method (ComBat) in the bulk RNA-seq, but no mention of batch correction was noted in the scRNA-seq methods. The authors should ensure that batch effect correction in all data is adequate, and these results should be added to the manuscript.

    We thank the reviewer for this comment. Indeed, batch effects are a particularly important consideration when samples are prepared separately and/or sequenced at distinct times, note this was not the case in this study.

    For the scRNA-seq analysis, we removed the low-quality cells, but did not use batch-effect correction methods because the samples were prepared and run at the same time. Meaning, isolation was performed in parallel, generation of cDNA libraries was done concurrently, and sequencing was run in the same gel. The quality of the data (and lack of batch effect) was subsequently verified when the two mono-culture biological replicates were evaluated by Seurat and were found to overlap on the UMAP (Figure 4), the same applies to the two co-culture biological replicates. These results clearly indicate that there’s no batch effect (as the samples were not process in distinct batches) among these samples.

    1. Table 1 shows ATAC-seq was done, however, no data from these experiments are provided in the manuscript.

    As mentioned (reviewer 2), we had performed ATACseq but decided to remove from the manuscript for several reasons and apologize for missing reference to Table 1. We have now corrected this error.

    1. Shear stress was achieved with an orbital shaker, which the accompanying citation states introduces significant heterogeneity in the ECs. This is based on the location of the culture dish. Was this heterogeneity seen in the scRNA-seq data?

    Correct. We only use the 2/3 peripheral area of the plates and discard the central aspect of the plate. We have added clarifying language to the Methods > Shear stress application to reflect this: “Orbital shear stress (130 rpm) was applied to confluent cell cultures by using an orbital shaker positioned inside the incubator as previously discussed (32). The shear stress within the cell culture well corresponds to arterial magnitudes (11.5 dynes/cm2) of shear stress. To reduce issues associated with uniformity of shear stress, the endothelial cell monolayers in 6-well plates were lysed after removing center region using cell scraper (BD Falcon #35-3085) and washing with 1X HBSS (Corning #21-022-CV). The 1.8cm blade was circumferentially used in the center of the 6-well plate to remove the center of the monolayer that did not see the higher shear stress.”

    1. It would be important to know whether the authors reproduce the findings from other papers that CD34 expression is reduced in cultured HUVECs:

    Muller AM, Cronen C, Muller KM, Kirkpatrick CJ: Comparative analysis of the reactivity of human umbilical vein endothelial cells in organ and monolayer culture. Pathobiology 1999;67:99-107. Delia D, Lampugnani MG, Resnati M, Dejana E, Aiello A, Fontanella E, Soligo D, Pierotti MA, Greaves MF: Cd34 expression is regulated reciprocally with adhesion molecules in vascular endothelial cells in vitro. Blood 1993;81:1001-1008.

    Thank you for this suggestion. Supplemental Excel 4 allows the reader to review single genes that are modulated by condition and in fact, consistent with all previous literature, CD34 expression is one of the most significantly decreased genes in cultured HUVECs (0.9, p=1E-5).

  3. eLife

    eLife assessment’

    This paper is of interest to a broad audience of cell biologists, and researchers who work with cultured endothelial cells. The work uncovers the impact of culture conditions on transcriptional changes of endothelial cells and demonstrates that some of these changes can be recovered by sheer forces or coculture. The authors provide valuable datasets which will be a good resource for the community.

  4. eLife

    Reviewer #1 (Public Review):

    In this well-written manuscript, Afshar et al demonstrated the significant transcriptional and proteomic differences between cultured human umbilical vein endothelial cells (HUVECs) and those freshly isolated from the cords. They showed that TGFbeta and BMP signaling target genes were enriched in cord cells compared to those in culture. Extracellular matrix (ECM) and cell cycle-related genes were also different between the two conditions. Because master regulators of EC shear stress response genes, KLF2 and KLF4, were downregulated in culture, the authors sought to restore the in vivo transcriptional profile with the application of shear stress in an orbital shaker and dextran-containing media for various time periods. They showed that after 48 hours of shear stress the transcriptional profile of sheared cells correlated with in vivo transcriptional profile more significantly than static cultures. They also showed, using single cell RNAseq, that EC-smooth muscle cell cocultures resulted in changes in TGFbeta and NOTCH signaling pathways and rescued 9% of the in vivo transcriptional signatures.

    This is an important study that was elegantly executed. The authors should also be commended for making their data public; thereby, creating a valuable resource for vascular biologists.

  5. eLife

    Reviewer #2 (Public Review):

    The authors profiled the transcriptome and proteome of human umbilical vein endothelial cells freshly isolated from in vivo and compared that with the same cells exposed to in vitro culture under different conditions, including static culture, flow, and co-culture with smooth muscle cells. The experiments were properly designed and performed. The authors also provided a reasonable and sound interpretation of their findings. This study provides valuable insights into how the culturing conditions impact on gene expression, encouraging the field to select their in vitro work setting appropriately. Overall, the manuscript is well-written and easy to follow.

    Several notable strengths include:

    1. Parallel transcriptome- and proteome-wide profiling of endothelial cells enabling the unbiased interrogation of gene expression and a genome-wide view of the impact of in vitro culture on endothelial transcriptome.
    2. The innovative experimental design and comparisons were done with genetically identical ECs (from the same donors) in vivo and in vitro.
    3. The analyses were robust and provided novel information on flow-dependent and cell context-dependent gene regulation, with the native freshly isolated cells as a baseline.
    5. The donor samples used in this study were diverse including Asian, White, Black, Latino, and American Indian samples which reduce racial background bias.

    Some points that can strengthen the study:

    A clear description of experimental and analytical details (e.g. how the comparisons were made) and more in-depth interpretation and discussion of the results, e.g. the complete genes that are rescued by flow and co-culture and potential synergy of these factors.

  6. eLife

    Reviewer #3 (Public Review):

    Afshar et al. performed RNA-seq and LC-MS of in vivo and in vitro HUVECs to identify the role of culture conditions on gene expression. Given the widespread use of HUVECs to study EC biology, these findings are interesting and can help design better in vitro experiments. There have been previous papers that compared in vivo and in vitro HUVECs, however, the depth of sequencing and analysis in this manuscript identifies some novel effects which should be accounted for in future in vitro experiments using ECs.

    Strengths:
    1. Major findings of distinct pathways affected by cell culture are novel and interesting. The authors identify major effects on TGFb and ECM gene expression. They also corroborate previous findings of flow response pathways, namely KLF2/4 and Notch pathway regulation.
    2. Use of multiple genomic methods to profile effects of culture conditions. The LC-MS data showed a significant correlation with RNA-seq, however, the data were not as strong so not used for subsequent analyses.
    3. Use of scRNA-seq to show the dynamic effects of co-culture and shear stress on ECs is very novel. However, the heterogeneity in the EC populations is not discussed in this manuscript.

    Weaknesses:
    1. The physiological relevance of these changes in gene expression is not demonstrated in the manuscript. The authors claim the significance of their data is to improve in vitro culture to better represent in vivo biology. Is this the case with orbital shear stress? Do they rescue some functional effects in ECs with long-term shear stress? An angiogenesis, barrier function, or migration assay for HUVECs exposed to different conditions would help answer this question. A similar assay for cells after EC-VSMC co-culture would validate the importance of these stimuli.
    2. One explanation for the increased expression of ECM genes in vivo is that these cells are contaminated with VSMCs/fibroblasts. This could be very likely given that cells were not sorted or purified upon isolation. Expression of other VSMC or fibroblast-specific markers (i.e. CNN1, MYH11, SMTN, DCN, FBLN1) would help determine if there is some level of non-EC contamination.
    3. The use of scRNA-seq in Figure 4 is interesting. There appear to be 2 distinct EC populations in the co-cultured ECs. What are the marker genes for the 2 populations?
    4. The modest shifts in gene expression with shear stress and co-culture could be attributed to the batch effect. The authors describe 1 batch correction method (ComBat) in the bulk RNA-seq, but no mention of batch correction was noted in the scRNA-seq methods. The authors should ensure that batch effect correction in all data is adequate, and these results should be added to the manuscript.
    5. Table 1 shows ATAC-seq was done, however, no data from these experiments are provided in the manuscript.
    6. Shear stress was achieved with an orbital shaker, which the accompanying citation states introduces significant heterogeneity in the ECs. This is based on the location of the culture dish. Was this heterogeneity seen in the scRNA-seq data?
    7. It would be important to know whether the authors reproduce the findings from other papers that CD34 expression is reduced in cultured HUVECs:

    Muller AM, Cronen C, Muller KM, Kirkpatrick CJ: Comparative analysis of the reactivity of human umbilical vein endothelial cells in organ and monolayer culture. Pathobiology 1999;67:99-107.

    Delia D, Lampugnani MG, Resnati M, Dejana E, Aiello A, Fontanella E, Soligo D, Pierotti MA, Greaves MF: Cd34 expression is regulated reciprocally with adhesion molecules in vascular endothelial cells in vitro. Blood 1993;81:1001-1008.

  7. Version published to 10.1101/2022.07.08.499287v1 on bioRxiv