Animal lungs are filled with tiny air sacks called alveoli, where the gas exchanges that keep organisms alive can take place. Small blood vessels known as capillaries come in close contact with the alveoli, allowing oxygen to be extracted from the air into the blood, and carbon dioxide to be released from the blood into the air. The cells that line the inside of these capillaries (known as pulmonary endothelial cells) are important actors in these exchanges.

After having been damaged, for example by viruses like influenza, the lungs need to regenerate and create new capillaries. Yet, it was still unclear how pulmonary endothelial cells participate in the healing process, and if capillaries contain several populations of endothelial cells that play different roles.

To investigate this question, Niethamer et al. used an approach called single-cell analytics to examine individual endothelial cells in the alveoli of mice infected with influenza. This revealed that different subtypes of endothelial cells exist in capillaries, and that some may be able to perform slightly different jobs during lung recovery.

Niethamer et al. found that all subtypes could quickly multiply after injury to create more endothelial cells and re-establish gas exchanges. However, one newly identified group (called Car4-high ECs) was particularly primed to receive orders from damaged alveoli. These cells were also often found at the sites where the alveoli were most injured.

Lung injuries are a major cause of death worldwide. Understanding how pulmonary endothelial cells work when the organ is both healthy and injured should help to find ways to boost repair, and to create therapies that could target these cells.

The primary function of the respiratory system is to exchange oxygen and other gases with the external environment. The functional compartment or niche responsible for gas exchange in the lungs is called the alveolus and is comprised of multiple epithelial, endothelial, and mesenchymal cell lineages. The pulmonary vascular plexus interacts intimately with alveolar type 1 (AT1) cells to form a very thin, gas-diffusible interface. The lung alveolus and its gas-exchange function is seriously disrupted upon acute lung injury such as occurs during a viral infection (Gregory and Kobzik, 2015; Herold et al., 2015; Kash and Taubenberger, 2015; Kumar et al., 2011; Pociask et al., 2017; Zacharias et al., 2018). Despite the functional importance of the lung alveolus, only recently have studies begun to elucidate its overall cellular complexity (Cohen et al., 2018; Ding et al., 2018; He et al., 2018; Montoro et al., 2018; Plasschaert et al., 2018; Reyfman et al., 2019; Sabbagh et al., 2018; Treutlein et al., 2014; Xie et al., 2018; Xu et al., 2016; Zacharias et al., 2018; Zepp et al., 2017). An improved understanding of the molecular mechanisms of alveolar regeneration is critical for the development of targeted therapies that induce or improve this process to restore functional gas exchange in pulmonary disease or following acute injury to the lung.

Previous work has demonstrated that acute or serious lung injury can lead to the generation of dense epithelial remodeling resulting in hypoxic vasoconstriction in regions of severe damage (Xi et al., 2017). Proper tissue regeneration in the lung requires neovascularization of the alveolar region, and previous work has suggested that certain EC populations may be primed for proliferation when cultured in vitro (Alvarez et al., 2008; Lee et al., 2017; Taha et al., 2017). Cell surface markers such as CD34 and CD309 are enriched on mouse ECs that proliferate in culture (Alvarez et al., 2008; Asahara et al., 1997). Putative EC progenitors can be isolated from the peripheral blood or the liver and transplanted into injured regions to contribute to regeneration (Asahara et al., 1997; Wakabayashi et al., 2018). However, whether such EC subpopulations contribute to in vivo tissue homeostasis and response to injury in the adult lung remains unknown.

To address these questions and to define pulmonary EC heterogeneity at homeostasis and during regeneration, we utilized single cell RNA sequencing (scRNA-seq) analysis of the adult mouse lung, both uninjured and after acute influenza-induced viral injury. In addition to identifying microvascular (miEC) and both arterial and venous macrovascular (maEC) populations, we identified a new population we have termed Car4-high ECs that possess a unique transcriptome. Car4-high ECs express high levels of Car4 and Cd34, are found throughout the lung periphery at homeostasis, and are primed to respond to Vegfa signaling based on their high expression of Vegf receptors, which corresponds to a receptor-ligand interaction analysis between Car4-high ECs and AT1 cells, their epithelial co-partners in gas exchange. Car4-high ECs are enriched in the regenerating zones surrounding the most damaged regions of the lung following influenza- or bleomycin-induced lung injury during the subsequent tissue regeneration process. Influenza injury revealed the emergence of a unique population of highly proliferative ECs, which are closely related to Car4-low miECs in gene expression but likely arise from multiple EC populations to contribute to regeneration of the pulmonary vasculature after injury by replacing lost endothelium. These data provide critical insight into lung EC heterogeneity and the key roles that differing subtypes of pulmonary ECs play in the response to acute lung injury.

Through single-cell analysis of the pulmonary vasculature at homeostasis and after acute lung injury, we have expanded our understanding of vascular heterogeneity in the lung and begun to elucidate the cellular mechanisms by which the vasculature contributes to alveolar regeneration after injury. In combination with transcript and protein localization, our scRNA-seq analysis identified a unique population of microvascular endothelium, the Car4-high EC. Through further characterization of these cells, we have determined that they coalesce within sites of moderate-to-severe alveolar damage after influenza infection and possess a transcriptional signature suggesting they are primed to receive signals from other alveolar cell types. Gene ontology analysis also suggests a role for these cells in vasculogenesis and blood vessel development. We additionally demonstrate the emergence of a highly proliferative EC population in the lung following influenza injury. The existence of endothelial subtypes that preferentially contribute to regeneration has been investigated in many organs, but the identity and origin of putative endothelial ‘progenitor’ cells remains a topic of debate (Basile and Yoder, 2014). In the aorta and the liver, endothelial regeneration has been attributed to small subsets of endogenous ECs with enhanced proliferative capacity (McDonald et al., 2018; Wakabayashi et al., 2018). Others have suggested that circulating bone marrow-derived ECs can be delivered to regions of injury to regenerate the endothelium (Asahara et al., 1999a; Asahara et al., 1997; Asahara et al., 1999b; Lin et al., 2000; Takahashi et al., 1999), but this hypothesis has also proven controversial (Purhonen et al., 2008). In the lung, it remained unclear whether all ECs, or only a unique subpopulation of progenitors, could contribute to endothelial injury repair. Our scRNA-seq analysis identifies a proliferating EC population that arises after acute lung injury and suggests that these cells are closely related to a subset of endogenous miECs. In addition, EdU incorporation analysis indicates that both Car4-high and other ECs proliferate across the 21 day period following injury, demonstrating that regenerative potential is not restricted to a particular subset of the endothelium.

An increasingly useful mouse model to study alveolar destruction and regeneration has been the administration of a sub-lethal dose of the animalized version of the H1N1 influenza virus (PR8). This injury model has been used recently to show the death and regeneration of the alveolar epithelium, primarily by the expansion of a squamous Krt5⁺ epithelial cell (Kumar et al., 2011; Vaughan et al., 2015) and an alveolar epithelial progenitor cell (AEP) (Zacharias et al., 2018). This observed epithelial regeneration is necessary to cover the exposed basement membrane and increase the surface area available for gas exchange. To fully understand the regeneration of the alveolar compartment with respect to the recovery of gas exchange, the behavior of the capillary endothelium must be studied in parallel. While the epithelial cells have been observed as the primary infection site of the influenza virus resulting in their death both in vitro and in vivo (Short et al., 2014), there is evidence that the endothelium can also be infected and killed in vitro (Armstrong et al., 2012). However, the regeneration of the alveolar capillary plexus following influenza injury and the mechanisms by which gas exchange is re-established have remained incompletely understood. Our data sheds crucial light on the potential of the endothelium to proliferate after injury, a cellular behavior that likely serves to regenerate the pulmonary vasculature. High expression of Car4 in the EC population we have identified also suggests that these cells may play a specialized role in gas exchange in the mouse and that their proliferation after injury may contribute meaningfully to restoring the gas exchange interface in the lung: Car4 has been proposed to play a role in carbon dioxide exchange in the rat lung, where it is more broadly expressed in the microvasculature (Fleming et al., 1993). A better understanding of the processes of endothelial proliferation and repair of the gas exchange interface will be a crucial step in developing therapies to alleviate patients of hypoxemia and acute respiratory distress syndrome (ARDS) that can be developed post-influenza infection.

An essential role of the endothelium in regeneration is to reestablish oxygenation of the tissue through restoration of blood circulation. The methods the endothelium uses to repopulate areas lacking in vascularity are classically defined as vasculogenesis, or formation of new vascular structures, and angiogenesis, or expansion of the vascular network through generation of new vessels from existing ones (Potente and Mäkinen, 2017). Of the cells that contribute to sprouting angiogenesis, there are two distinct EC phenotypes termed tip cells and stalk cells. Tip cells are leading cells that direct the new vessel growth through sprouting, invasion, and migration; stalk cells follow the tip cells and elongate, proliferate and establish the lumen of the new vessel (Betz et al., 2016; Potente et al., 2011). Tip cells are distinguished from stalk cells through increased expression of Vegfr2 and Nrp1, which suppresses stalk cell fate (Aspalter et al., 2015; Simons et al., 2016), whereas stalk cells are characterized by expression of Notch1 and Vegfr1, a decoy receptor that can sequester Vegf (Blanco and Gerhardt, 2013). However, tip cell and stalk cell fates are dynamic during sprouting angiogenesis, with ECs possessing the ability to respond to different environments by adjusting Vegfr levels (Jakobsson et al., 2010). During late lung development and maturation, the lung vasculature is also thought to develop via intussusceptive angiogenesis, a poorly understood process of vessel splitting that may also contribute to regeneration in the lung (Gianni-Barrera et al., 2011). Our transcriptomic analysis of the pulmonary endothelium through scRNA-seq shows that key tip cell genes such as Vegfr2 and Nrp1 are enriched in the Car4-high EC population at homeostasis and after injury; however, the Car4-high ECs also express high levels of stalk cell gene Vegfr1 (Tables 1, 2 and 4). The Gpihbp1-high miEC population shares several reported transcripts that define stalk cells such as Hes1, Hey1 and Jagged1 (Moya et al., 2012). However, following influenza injury, the proliferating pulmonary EC population we have identified is closely transcriptionally associated with the miEC population, and endothelial proliferation is evident both in Car4-high and other ECs. This is in line with a recent report indicating that endothelial proliferation and regeneration in large vessels does not require the specification of tip or stalk cells, but is instead initially induced in all cells in close proximity to the site of injury (McDonald et al., 2018). However, McDonald and colleagues report a second phase of regeneration in which a smaller subset of proliferating ECs in the aorta continue to generate new cells as wound healing progresses. The gene expression profiles of the cells we have identified and their pseudotime trajectories suggest that a similar phenomenon may occur in the microvasculature of the alveoli during regeneration after influenza injury, with a mechanism that does not require tip or stalk cell specification.

Recently, the existence of ECs primed for proliferation in different vascular beds has gained appreciation (Taha et al., 2017). Evidence for such cells in the lung has been derived from cell cloning assays that isolate a subset of cells that are prone to proliferate (Alvarez et al., 2008; Lee et al., 2017). Alvarez and colleagues showed that the clones derived from the microvascular endothelium that were prone to proliferate, termed resident microvascular endothelial progenitors, were enriched in expression of certain progenitor cell antigens (e.g., CD34, CD309) (Alvarez et al., 2008). Studies of these ECs in vitro have benefited from the ability to clonally assess the cells’ proliferative capacity, but in vitro EC studies are problematic because of the propensity of ECs to de-differentiate once put into culture (Lacorre et al., 2004). Thus, studying such cell populations in vivo is essential to eliminate the confounding factors associated with cell culture studies. Evidence of proliferating ECs has been seen in models of idiopathic pulmonary fibrosis (Ebina et al., 2004), partial hepatectomy (Ding et al., 2010), and left unilateral pneumonectomy (Ding et al., 2011). Our studies suggest the presence of a Car4-high/Cd34-high EC subpopulation in the lung alveolus and the induction of a highly proliferative EC population after injury, but in contrast to previous studies, our data indicate that these two populations are not identical. Future work will further elucidate the relative contributions of Car4-high/Cd34-high ECs and other ECs to endothelial proliferation, the contributions of the proliferating endothelium to generation of new vasculature in the alveolar space after injury, and the roles of each cell type in re-establishing gas exchange and in other aspects of lung regeneration.

Single-cell RNA sequencing datasets have been deposited in GEO under accession code GSE128944.

1. 1. Niethamer TK
   2. Stabler CT
   3. Morley MP
   4. Babu A
   5. Morrisey EE

   (2020) NCBI Gene Expression Omnibus

   ID GSE128944. Defining the role of pulmonary endothelial cell heterogeneity in the response to acute lung injury.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE128944

1. 1. Adamson IY

   (1976) Pulmonary toxicity of bleomycin

   Environmental Health Perspectives 16:119–125.

   https://doi.org/10.1289/ehp.7616119

   • PubMed
   • Google Scholar
2. 1. Adamson IY
   2. Bowden DH

   (1974)

   The pathogenesis of bleomycin-induced pulmonary fibrosis in mice

   The American Journal of Pathology 77:185–197.

   • Google Scholar
3. 1. Alvarez DF
   2. Huang L
   3. King JA
   4. ElZarrad MK
   5. Yoder MC
   6. Stevens T

   (2008) Lung microvascular endothelium is enriched with progenitor cells that exhibit vasculogenic capacity

   American Journal of Physiology-Lung Cellular and Molecular Physiology 294:L419–L430.

   https://doi.org/10.1152/ajplung.00314.2007

   • PubMed
   • Google Scholar
4. 1. Armstrong SM
   2. Wang C
   3. Tigdi J
   4. Si X
   5. Dumpit C
   6. Charles S
   7. Gamage A
   8. Moraes TJ
   9. Lee WL

   (2012) Influenza infects lung microvascular endothelium leading to microvascular leak: role of apoptosis and claudin-5

   PLOS ONE 7:e47323.

   https://doi.org/10.1371/journal.pone.0047323

   • PubMed
   • Google Scholar
5. 1. Asahara T
   2. Murohara T
   3. Sullivan A
   4. Silver M
   5. van der Zee R
   6. Li T
   7. Witzenbichler B
   8. Schatteman G
   9. Isner JM

   (1997) Isolation of putative progenitor endothelial cells for angiogenesis

   Science 275:964–966.

   https://doi.org/10.1126/science.275.5302.964

   • PubMed
   • Google Scholar
6. 1. Asahara T
   2. Masuda H
   3. Takahashi T
   4. Kalka C
   5. Pastore C
   6. Silver M
   7. Kearne M
   8. Magner M
   9. Isner JM

   (1999a) Bone marrow origin of endothelial progenitor cells responsible for postnatal vasculogenesis in physiological and pathological neovascularization

   Circulation Research 85:221–228.

   https://doi.org/10.1161/01.RES.85.3.221

   • PubMed
   • Google Scholar
7. 1. Asahara T
   2. Takahashi T
   3. Masuda H
   4. Kalka C
   5. Chen D
   6. Iwaguro H
   7. Inai Y
   8. Silver M
   9. Isner JM

   (1999b) VEGF contributes to postnatal neovascularization by mobilizing bone marrow-derived endothelial progenitor cells

   The EMBO Journal 18:3964–3972.

   https://doi.org/10.1093/emboj/18.14.3964

   • PubMed
   • Google Scholar
8. 1. Aspalter IM
   2. Gordon E
   3. Dubrac A
   4. Ragab A
   5. Narloch J
   6. Vizán P
   7. Geudens I
   8. Collins RT
   9. Franco CA
   10. Abrahams CL
   11. Thurston G
   12. Fruttiger M
   13. Rosewell I
   14. Eichmann A
   15. Gerhardt H

   (2015) Alk1 and Alk5 inhibition by Nrp1 controls vascular sprouting downstream of notch

   Nature Communications 6:7264.

   https://doi.org/10.1038/ncomms8264

   • PubMed
   • Google Scholar
9. 1. Basile DP
   2. Yoder MC

   (2014) Circulating and tissue resident endothelial progenitor cells

   Journal of Cellular Physiology 229:10–16.

   https://doi.org/10.1002/jcp.24423

   • PubMed
   • Google Scholar
10. 1. Becht E
    2. McInnes L
    3. Healy J
    4. Dutertre CA
    5. Kwok IWH
    6. Ng LG
    7. Ginhoux F
    8. Newell EW

    (2018) Dimensionality reduction for visualizing single-cell data using UMAP

    Nature Biotechnology 37:4314.

    https://doi.org/10.1038/nbt.4314

    • Google Scholar
11. 1. Berg S
    2. Kutra D
    3. Kroeger T
    4. Straehle CN
    5. Kausler BX
    6. Haubold C
    7. Schiegg M
    8. Ales J
    9. Beier T
    10. Rudy M
    11. Eren K
    12. Cervantes JI
    13. Xu B
    14. Beuttenmueller F
    15. Wolny A
    16. Zhang C
    17. Koethe U
    18. Hamprecht FA
    19. Kreshuk A

    (2019) Ilastik: interactive machine learning for (bio)image analysis

    Nature Methods 16:1226–1232.

    https://doi.org/10.1038/s41592-019-0582-9

    • Google Scholar
12. 1. Betz C
    2. Lenard A
    3. Belting HG
    4. Affolter M

    (2016) Cell behaviors and dynamics during angiogenesis

    Development 143:2249–2260.

    https://doi.org/10.1242/dev.135616

    • PubMed
    • Google Scholar
13. 1. Blanco R
    2. Gerhardt H

    (2013) VEGF and notch in tip and stalk cell selection

    Cold Spring Harbor Perspectives in Medicine 3:a006569.

    https://doi.org/10.1101/cshperspect.a006569

    • PubMed
    • Google Scholar
14. 1. Butler A
    2. Hoffman P
    3. Smibert P
    4. Papalexi E
    5. Satija R

    (2018) Integrating single-cell transcriptomic data across different conditions, technologies, and species

    Nature Biotechnology 36:411–420.

    https://doi.org/10.1038/nbt.4096

    • PubMed
    • Google Scholar
15. 1. Chao CM
    2. Moiseenko A
    3. Kosanovic D
    4. Rivetti S
    5. El Agha E
    6. Wilhelm J
    7. Kampschulte M
    8. Yahya F
    9. Ehrhardt H
    10. Zimmer KP
    11. Barreto G
    12. Rizvanov AA
    13. Schermuly RT
    14. Reiss I
    15. Morty RE
    16. Rottier RJ
    17. Bellusci S
    18. Zhang JS

    (2019) Impact of Fgf10 deficiency on pulmonary vasculature formation in a mouse model of bronchopulmonary dysplasia

    Human Molecular Genetics 28:1429–1444.

    https://doi.org/10.1093/hmg/ddy439

    • PubMed
    • Google Scholar
16. 1. Cohen M
    2. Giladi A
    3. Gorki AD
    4. Solodkin DG
    5. Zada M
    6. Hladik A
    7. Miklosi A
    8. Salame TM
    9. Halpern KB
    10. David E
    11. Itzkovitz S
    12. Harkany T
    13. Knapp S
    14. Amit I

    (2018) Lung Single-Cell signaling interaction map reveals basophil role in macrophage imprinting

    Cell 175:1031–1044.

    https://doi.org/10.1016/j.cell.2018.09.009

    • PubMed
    • Google Scholar
17. 1. Ding BS
    2. Nolan DJ
    3. Butler JM
    4. James D
    5. Babazadeh AO
    6. Rosenwaks Z
    7. Mittal V
    8. Kobayashi H
    9. Shido K
    10. Lyden D
    11. Sato TN
    12. Rabbany SY
    13. Rafii S

    (2010) Inductive angiocrine signals from sinusoidal endothelium are required for liver regeneration

    Nature 468:310–315.

    https://doi.org/10.1038/nature09493

    • PubMed
    • Google Scholar
18. 1. Ding BS
    2. Nolan DJ
    3. Guo P
    4. Babazadeh AO
    5. Cao Z
    6. Rosenwaks Z
    7. Crystal RG
    8. Simons M
    9. Sato TN
    10. Worgall S
    11. Shido K
    12. Rabbany SY
    13. Rafii S

    (2011) Endothelial-derived angiocrine signals induce and sustain regenerative lung alveolarization

    Cell 147:539–553.

    https://doi.org/10.1016/j.cell.2011.10.003

    • PubMed
    • Google Scholar
19. 1. Ding J
    2. Aronow BJ
    3. Kaminski N
    4. Kitzmiller J
    5. Whitsett JA
    6. Bar-Joseph Z

    (2018) Reconstructing differentiation networks and their regulation from time series single-cell expression data

    Genome Research 28:383–395.

    https://doi.org/10.1101/gr.225979.117

    • Google Scholar
20. 1. Ebina M
    2. Shimizukawa M
    3. Shibata N
    4. Kimura Y
    5. Suzuki T
    6. Endo M
    7. Sasano H
    8. Kondo T
    9. Nukiwa T

    (2004) Heterogeneous increase in CD34-positive alveolar capillaries in idiopathic pulmonary fibrosis

    American Journal of Respiratory and Critical Care Medicine 169:1203–1208.

    https://doi.org/10.1164/rccm.200308-1111OC

    • PubMed
    • Google Scholar
21. 1. Fleming RE
    2. Crouch EC
    3. Ruzicka CA
    4. Sly WS

    (1993) Pulmonary carbonic anhydrase IV: developmental regulation and cell-specific expression in the capillary endothelium

    American Journal of Physiology-Lung Cellular and Molecular Physiology 265:L627–L635.

    https://doi.org/10.1152/ajplung.1993.265.6.L627

    • Google Scholar
22. 1. Gianni-Barrera R
    2. Trani M
    3. Reginato S
    4. Banfi A

    (2011) To sprout or to split? VEGF, notch and vascular morphogenesis

    Biochemical Society Transactions 39:1644–1648.

    https://doi.org/10.1042/BST20110650

    • PubMed
    • Google Scholar
23. 1. Gregory DJ
    2. Kobzik L

    (2015) Influenza lung injury: mechanisms and therapeutic opportunities

    American Journal of Physiology-Lung Cellular and Molecular Physiology 309:L1041–L1046.

    https://doi.org/10.1152/ajplung.00283.2015

    • PubMed
    • Google Scholar
24. 1. He L
    2. Vanlandewijck M
    3. Mäe MA
    4. Andrae J
    5. Ando K
    6. Del Gaudio F
    7. Nahar K
    8. Lebouvier T
    9. Laviña B
    10. Gouveia L
    11. Sun Y
    12. Raschperger E
    13. Segerstolpe Å
    14. Liu J
    15. Gustafsson S
    16. Räsänen M
    17. Zarb Y
    18. Mochizuki N
    19. Keller A
    20. Lendahl U
    21. Betsholtz C

    (2018) Single-cell RNA sequencing of mouse brain and lung vascular and vessel-associated cell types

    Scientific Data 5:180160.

    https://doi.org/10.1038/sdata.2018.160

    • PubMed
    • Google Scholar
25. 1. Herold S
    2. Becker C
    3. Ridge KM
    4. Budinger GR

    (2015) Influenza virus-induced lung injury: pathogenesis and implications for treatment

    European Respiratory Journal 45:1463–1478.

    https://doi.org/10.1183/09031936.00186214

    • PubMed
    • Google Scholar
26. 1. Jakobsson L
    2. Franco CA
    3. Bentley K
    4. Collins RT
    5. Ponsioen B
    6. Aspalter IM
    7. Rosewell I
    8. Busse M
    9. Thurston G
    10. Medvinsky A
    11. Schulte-Merker S
    12. Gerhardt H

    (2010) Endothelial cells dynamically compete for the tip cell position during angiogenic sprouting

    Nature Cell Biology 12:943–953.

    https://doi.org/10.1038/ncb2103

    • PubMed
    • Google Scholar
27. 1. Kash JC
    2. Taubenberger JK

    (2015) The role of viral, host, and secondary bacterial factors in influenza pathogenesis

    The American Journal of Pathology 185:1528–1536.

    https://doi.org/10.1016/j.ajpath.2014.08.030

    • PubMed
    • Google Scholar
28. 1. Kumar PA
    2. Hu Y
    3. Yamamoto Y
    4. Hoe NB
    5. Wei TS
    6. Mu D
    7. Sun Y
    8. Joo LS
    9. Dagher R
    10. Zielonka EM
    11. Wang deY
    12. Lim B
    13. Chow VT
    14. Crum CP
    15. Xian W
    16. McKeon F

    (2011) Distal airway stem cells yield alveoli in vitro and during lung regeneration following H1N1 influenza infection

    Cell 147:525–538.

    https://doi.org/10.1016/j.cell.2011.10.001

    • PubMed
    • Google Scholar
29. 1. Lacorre DA
    2. Baekkevold ES
    3. Garrido I
    4. Brandtzaeg P
    5. Haraldsen G
    6. Amalric F
    7. Girard JP

    (2004) Plasticity of endothelial cells: rapid dedifferentiation of freshly isolated high endothelial venule endothelial cells outside the lymphoid tissue microenvironment

    Blood 103:4164–4172.

    https://doi.org/10.1182/blood-2003-10-3537

    • PubMed
    • Google Scholar
30. 1. Lee JH
    2. Bhang DH
    3. Beede A
    4. Huang TL
    5. Stripp BR
    6. Bloch KD
    7. Wagers AJ
    8. Tseng YH
    9. Ryeom S
    10. Kim CF

    (2014) Lung stem cell differentiation in mice directed by endothelial cells via a BMP4-NFATc1-thrombospondin-1 Axis

    Cell 156:440–455.

    https://doi.org/10.1016/j.cell.2013.12.039

    • PubMed
    • Google Scholar
31. 1. Lee JY
    2. McMurtry SA
    3. Stevens T

    (2017) Single cell cloning generates lung endothelial colonies with conserved growth, angiogenic, and bioenergetic characteristics

    Pulmonary Circulation 7:777–792.

    https://doi.org/10.1177/2045893217731295

    • PubMed
    • Google Scholar
32. 1. Lin Y
    2. Weisdorf DJ
    3. Solovey A
    4. Hebbel RP

    (2000) Origins of circulating endothelial cells and endothelial outgrowth from blood

    Journal of Clinical Investigation 105:71–77.

    https://doi.org/10.1172/JCI8071

    • PubMed
    • Google Scholar
33. 1. Londhe VA
    2. Maisonet TM
    3. Lopez B
    4. Jeng J-M
    5. Xiao J
    6. Li C
    7. Minoo P

    (2011) Conditional deletion of epithelial IKKβ impairs alveolar formation through apoptosis and decreased VEGF expression during early mouse lung morphogenesis

    Respiratory Research 12:134.

    https://doi.org/10.1186/1465-9921-12-134

    • Google Scholar
34. 1. McDonald AI
    2. Shirali AS
    3. Aragón R
    4. Ma F
    5. Hernandez G
    6. Vaughn DA
    7. Mack JJ
    8. Lim TY
    9. Sunshine H
    10. Zhao P
    11. Kalinichenko V
    12. Hai T
    13. Pelegrini M
    14. Ardehali R
    15. Iruela-Arispe ML

    (2018) Endothelial regeneration of large vessels is a biphasic process driven by local cells with distinct proliferative capacities

    Cell Stem Cell 23:210–225.

    https://doi.org/10.1016/j.stem.2018.07.011

    • PubMed
    • Google Scholar
35. 1. McQuin C
    2. Goodman A
    3. Chernyshev V
    4. Kamentsky L
    5. Cimini BA
    6. Karhohs KW
    7. Doan M
    8. Ding L
    9. Rafelski SM
    10. Thirstrup D
    11. Wiegraebe W
    12. Singh S
    13. Becker T
    14. Caicedo JC
    15. Carpenter AE

    (2018) CellProfiler 3.0: next-generation image processing for biology

    PLOS Biology 16:e2005970.

    https://doi.org/10.1371/journal.pbio.2005970

    • PubMed
    • Google Scholar
36. 1. Montoro DT
    2. Haber AL
    3. Biton M
    4. Vinarsky V
    5. Lin B
    6. Birket SE
    7. Yuan F
    8. Chen S
    9. Leung HM
    10. Villoria J
    11. Rogel N
    12. Burgin G
    13. Tsankov AM
    14. Waghray A
    15. Slyper M
    16. Waldman J
    17. Nguyen L
    18. Dionne D
    19. Rozenblatt-Rosen O
    20. Tata PR
    21. Mou H
    22. Shivaraju M
    23. Bihler H
    24. Mense M
    25. Tearney GJ
    26. Rowe SM
    27. Engelhardt JF
    28. Regev A
    29. Rajagopal J

    (2018) A revised airway epithelial hierarchy includes CFTR-expressing ionocytes

    Nature 560:319–324.

    https://doi.org/10.1038/s41586-018-0393-7

    • Google Scholar
37. 1. Moya IM
    2. Umans L
    3. Maas E
    4. Pereira PN
    5. Beets K
    6. Francis A
    7. Sents W
    8. Robertson EJ
    9. Mummery CL
    10. Huylebroeck D
    11. Zwijsen A

    (2012) Stalk cell phenotype depends on integration of notch and Smad1/5 signaling cascades

    Developmental Cell 22:501–514.

    https://doi.org/10.1016/j.devcel.2012.01.007

    • PubMed
    • Google Scholar
38. 1. Nguyen NM
    2. Kelley DG
    3. Schlueter JA
    4. Meyer MJ
    5. Senior RM
    6. Miner JH

    (2005) Epithelial laminin alpha5 is necessary for distal epithelial cell maturation, VEGF production, and alveolization in the developing murine lung

    Developmental Biology 282:111–125.

    https://doi.org/10.1016/j.ydbio.2005.02.031

    • PubMed
    • Google Scholar
39. 1. Piali L
    2. Fichtel A
    3. Terpe HJ
    4. Imhof BA
    5. Gisler RH

    (1995) Endothelial vascular cell adhesion molecule 1 expression is suppressed by melanoma and carcinoma

    The Journal of Experimental Medicine 181:811–816.

    https://doi.org/10.1084/jem.181.2.811

    • PubMed
    • Google Scholar
40. 1. Plasschaert LW
    2. Žilionis R
    3. Choo-Wing R
    4. Savova V
    5. Knehr J
    6. Roma G
    7. Klein AM
    8. Jaffe AB

    (2018) A single-cell atlas of the airway epithelium reveals the CFTR-rich pulmonary ionocyte

    Nature 560:377–381.

    https://doi.org/10.1038/s41586-018-0394-6

    • PubMed
    • Google Scholar
41. 1. Pociask DA
    2. Robinson KM
    3. Chen K
    4. McHugh KJ
    5. Clay ME
    6. Huang GT
    7. Benos PV
    8. Janssen-Heininger YMW
    9. Kolls JK
    10. Anathy V
    11. Alcorn JF

    (2017) Epigenetic and transcriptomic regulation of lung repair during recovery from influenza infection

    The American Journal of Pathology 187:851–863.

    https://doi.org/10.1016/j.ajpath.2016.12.012

    • PubMed
    • Google Scholar
42. 1. Potente M
    2. Gerhardt H
    3. Carmeliet P

    (2011) Basic and therapeutic aspects of angiogenesis

    Cell 146:873–887.

    https://doi.org/10.1016/j.cell.2011.08.039

    • PubMed
    • Google Scholar
43. 1. Potente M
    2. Mäkinen T

    (2017) Vascular heterogeneity and specialization in development and disease

    Nature Reviews Molecular Cell Biology 18:477–494.

    https://doi.org/10.1038/nrm.2017.36

    • PubMed
    • Google Scholar
44. 1. Purhonen S
    2. Palm J
    3. Rossi D
    4. Kaskenpää N
    5. Rajantie I
    6. Ylä-Herttuala S
    7. Alitalo K
    8. Weissman IL
    9. Salven P

    (2008) Bone marrow-derived circulating endothelial precursors do not contribute to vascular endothelium and are not needed for tumor growth

    PNAS 105:6620–6625.

    https://doi.org/10.1073/pnas.0710516105

    • PubMed
    • Google Scholar
45. 1. Reyfman PA
    2. Walter JM
    3. Joshi N
    4. Anekalla KR
    5. McQuattie-Pimentel AC
    6. Chiu S
    7. Fernandez R
    8. Akbarpour M
    9. Chen CI
    10. Ren Z
    11. Verma R
    12. Abdala-Valencia H
    13. Nam K
    14. Chi M
    15. Han S
    16. Gonzalez-Gonzalez FJ
    17. Soberanes S
    18. Watanabe S
    19. Williams KJN
    20. Flozak AS
    21. Nicholson TT
    22. Morgan VK
    23. Winter DR
    24. Hinchcliff M
    25. Hrusch CL
    26. Guzy RD
    27. Bonham CA
    28. Sperling AI
    29. Bag R
    30. Hamanaka RB
    31. Mutlu GM
    32. Yeldandi AV
    33. Marshall SA
    34. Shilatifard A
    35. Amaral LAN
    36. Perlman H
    37. Sznajder JI
    38. Argento AC
    39. Gillespie CT
    40. Dematte J
    41. Jain M
    42. Singer BD
    43. Ridge KM
    44. Lam AP
    45. Bharat A
    46. Bhorade SM
    47. Gottardi CJ
    48. Budinger GRS
    49. Misharin AV

    (2019) Single-Cell transcriptomic analysis of human lung provides insights into the pathobiology of pulmonary fibrosis

    American Journal of Respiratory and Critical Care Medicine 199:1517–1536.

    https://doi.org/10.1164/rccm.201712-2410OC

    • PubMed
    • Google Scholar
46. 1. Rojas-Quintero J
    2. Wang X
    3. Tipper J
    4. Burkett PR
    5. Zuñiga J
    6. Ashtekar AR
    7. Polverino F
    8. Rout A
    9. Yambayev I
    10. Hernández C
    11. Jimenez L
    12. Ramírez G
    13. Harrod KS
    14. Owen CA

    (2018) Matrix metalloproteinase-9 deficiency protects mice from severe influenza A viral infection

    JCI Insight 3:99022.

    https://doi.org/10.1172/jci.insight.99022

    • Google Scholar
47. 1. Sabbagh MF
    2. Heng JS
    3. Luo C
    4. Castanon RG
    5. Nery JR
    6. Rattner A
    7. Goff LA
    8. Ecker JR
    9. Nathans J

    (2018) Transcriptional and epigenomic landscapes of CNS and non-CNS vascular endothelial cells

    eLife 7:e36187.

    https://doi.org/10.7554/eLife.36187

    • PubMed
    • Google Scholar
48. 1. Short KR
    2. Kroeze EJBV
    3. Fouchier RAM
    4. Kuiken T

    (2014) Pathogenesis of influenza-induced acute respiratory distress syndrome

    The Lancet Infectious Diseases 14:57–69.

    https://doi.org/10.1016/S1473-3099(13)70286-X

    • Google Scholar
49. 1. Simons M
    2. Gordon E
    3. Claesson-Welsh L

    (2016) Mechanisms and regulation of endothelial VEGF receptor signalling

    Nature Reviews Molecular Cell Biology 17:611–625.

    https://doi.org/10.1038/nrm.2016.87

    • Google Scholar
50. 1. Street K
    2. Risso D
    3. Fletcher RB
    4. Das D
    5. Ngai J
    6. Yosef N
    7. Purdom E
    8. Dudoit S

    (2018) Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics

    BMC Genomics 19:477.

    https://doi.org/10.1186/s12864-018-4772-0

    • PubMed
    • Google Scholar
51. 1. Taha M
    2. Cadete VJ
    3. Stewart DJ

    (2017) Macro- and micro-heterogeneity of lung endothelial cells: they may not be smooth, but they got the moves

    Pulmonary Circulation 7:755–757.

    https://doi.org/10.1177/2045893217743710

    • PubMed
    • Google Scholar
52. 1. Takahashi T
    2. Kalka C
    3. Masuda H
    4. Chen D
    5. Silver M
    6. Kearney M
    7. Magner M
    8. Isner JM
    9. Asahara T

    (1999) Ischemia- and cytokine-induced mobilization of bone marrow-derived endothelial progenitor cells for neovascularization

    Nature Medicine 5:434–438.

    https://doi.org/10.1038/7434

    • Google Scholar
53. 1. Töpfer L
    2. Menk M
    3. Weber-Carstens S
    4. Spies C
    5. Wernecke KD
    6. Uhrig A
    7. Lojewski C
    8. Jörres A
    9. Deja M

    (2014) Influenza A (H1N1) vs non-H1N1 ARDS: analysis of clinical course

    Journal of Critical Care 29:340–346.

    https://doi.org/10.1016/j.jcrc.2013.12.013

    • PubMed
    • Google Scholar
54. 1. Treutlein B
    2. Brownfield DG
    3. Wu AR
    4. Neff NF
    5. Mantalas GL
    6. Espinoza FH
    7. Desai TJ
    8. Krasnow MA
    9. Quake SR

    (2014) Reconstructing lineage hierarchies of the distal lung epithelium using single-cell RNA-seq

    Nature 509:371–375.

    https://doi.org/10.1038/nature13173

    • PubMed
    • Google Scholar
55. 1. Tripathi S
    2. Pohl MO
    3. Zhou Y
    4. Rodriguez-Frandsen A
    5. Wang G
    6. Stein DA
    7. Moulton HM
    8. DeJesus P
    9. Che J
    10. Mulder LC
    11. Yángüez E
    12. Andenmatten D
    13. Pache L
    14. Manicassamy B
    15. Albrecht RA
    16. Gonzalez MG
    17. Nguyen Q
    18. Brass A
    19. Elledge S
    20. White M
    21. Shapira S
    22. Hacohen N
    23. Karlas A
    24. Meyer TF
    25. Shales M
    26. Gatorano A
    27. Johnson JR
    28. Jang G
    29. Johnson T
    30. Verschueren E
    31. Sanders D
    32. Krogan N
    33. Shaw M
    34. König R
    35. Stertz S
    36. García-Sastre A
    37. Chanda SK

    (2015) Meta- and orthogonal integration of influenza "OMICs" Data Defines a Role for UBR4 in Virus Budding

    Cell Host & Microbe 18:723–735.

    https://doi.org/10.1016/j.chom.2015.11.002

    • PubMed
    • Google Scholar
56. 1. Vaughan AE
    2. Brumwell AN
    3. Xi Y
    4. Gotts JE
    5. Brownfield DG
    6. Treutlein B
    7. Tan K
    8. Tan V
    9. Liu FC
    10. Looney MR
    11. Matthay MA
    12. Rock JR
    13. Chapman HA

    (2015) Lineage-negative progenitors mobilize to regenerate lung epithelium after major injury

    Nature 517:621–625.

    https://doi.org/10.1038/nature14112

    • PubMed
    • Google Scholar
57. 1. Wakabayashi T
    2. Naito H
    3. Suehiro J
    4. Lin Y
    5. Kawaji H
    6. Iba T
    7. Kouno T
    8. Ishikawa-Kato S
    9. Furuno M
    10. Takara K
    11. Muramatsu F
    12. Weizhen J
    13. Kidoya H
    14. Ishihara K
    15. Hayashizaki Y
    16. Nishida K
    17. Yoder MC
    18. Takakura N

    (2018) CD157 marks Tissue-Resident endothelial stem cells with homeostatic and regenerative properties

    Cell Stem Cell 22:384–397.

    https://doi.org/10.1016/j.stem.2018.01.010

    • PubMed
    • Google Scholar
58. 1. White AC
    2. Lavine KJ
    3. Ornitz DM

    (2007) FGF9 and SHH regulate mesenchymal vegfa expression and development of the pulmonary capillary network

    Development 134:3743–3752.

    https://doi.org/10.1242/dev.004879

    • PubMed
    • Google Scholar
59. 1. Xi Y
    2. Kim T
    3. Brumwell AN
    4. Driver IH
    5. Wei Y
    6. Tan V
    7. Jackson JR
    8. Xu J
    9. Lee DK
    10. Gotts JE
    11. Matthay MA
    12. Shannon JM
    13. Chapman HA
    14. Vaughan AE

    (2017) Local lung hypoxia determines epithelial fate decisions during alveolar regeneration

    Nature Cell Biology 19:904–914.

    https://doi.org/10.1038/ncb3580

    • PubMed
    • Google Scholar
60. 1. Xie T
    2. Wang Y
    3. Deng N
    4. Huang G
    5. Taghavifar F
    6. Geng Y
    7. Liu N
    8. Kulur V
    9. Yao C
    10. Chen P
    11. Liu Z
    12. Stripp B
    13. Tang J
    14. Liang J
    15. Noble PW
    16. Jiang D

    (2018) Single-Cell deconvolution of fibroblast heterogeneity in mouse pulmonary fibrosis

    Cell Reports 22:3625–3640.

    https://doi.org/10.1016/j.celrep.2018.03.010

    • PubMed
    • Google Scholar
61. 1. Xu Y
    2. Mizuno T
    3. Sridharan A
    4. Du Y
    5. Guo M
    6. Tang J
    7. Wikenheiser-Brokamp KA
    8. Perl AT
    9. Funari VA
    10. Gokey JJ
    11. Stripp BR
    12. Whitsett JA

    (2016) Single-cell RNA sequencing identifies diverse roles of epithelial cells in idiopathic pulmonary fibrosis

    JCI Insight 1:e90558.

    https://doi.org/10.1172/jci.insight.90558

    • PubMed
    • Google Scholar
62. 1. Yamamoto K
    2. de Waard V
    3. Fearns C
    4. Loskutoff DJ

    (1998) Tissue distribution and regulation of murine von willebrand factor gene expression in vivo

    Blood 92:2791–2801.

    https://doi.org/10.1182/blood.V92.8.2791

    • PubMed
    • Google Scholar
63. 1. Zacharias WJ
    2. Frank DB
    3. Zepp JA
    4. Morley MP
    5. Alkhaleel FA
    6. Kong J
    7. Zhou S
    8. Cantu E
    9. Morrisey EE

    (2018) Regeneration of the lung alveolus by an evolutionarily conserved epithelial progenitor

    Nature 555:251–255.

    https://doi.org/10.1038/nature25786

    • PubMed
    • Google Scholar
64. 1. Zepp JA
    2. Zacharias WJ
    3. Frank DB
    4. Cavanaugh CA
    5. Zhou S
    6. Morley MP
    7. Morrisey EE

    (2017) Distinct mesenchymal lineages and niches promote epithelial Self-Renewal and myofibrogenesis in the lung

    Cell 170:1134–1148.

    https://doi.org/10.1016/j.cell.2017.07.034

    • PubMed
    • Google Scholar

Acceptance summary:

The reviewers appreciate the novelty and scientific value of mapping the heterogeneity of pulmonary endothelial cells, identification of different subpopulations of these cells, and insights into their roles. Of particular interest is the identification of a new subpopulation of pulmonary endothelial cells that are involved in lung regeneration following injury.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Defining the role of pulmonary endothelial cell heterogeneity in the response to acute lung injury" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered for publication in eLife in its present state.

While the reviewers were enthusiastic about the novelty and importance of your study, they also raised substantial concerns that would require extensive additional experimentation for a responsive revision. We would therefore encourage resubmission of your manuscript following revisions, with a commitment to solicit opinions from the original reviewers in order to expedite the re-review process.

We see the strength of the paper in the new insights into the heterogeneity of endothelial cells before and after lung injury, obtained by analysis of single cell RNA sequencing data in the murine lung. Three classes of endothelial cells that were identified and termed "macrovascular", "microvascular" and "regulatory" are of great interest. In particular, the "regulatory" endothelial cells are a novel cell type that is found in alveolar areas, characterized with high expression of CD34 and CAR4, and shown to expand following injury. The reviewers felt that these findings advance current understanding of endothelial cells in the lung, and the dynamic changes during lung injury and repair. The main weaknesses identified by the reviewers include the lack of functional data, and of statistical validation/ analysis (except for the single cell RNA-seq data). Please consult the individual reviews for more detail.

Reviewer #1:

In this study, the authors used single-cell RNA-seq and the latest bioinformatic algorithms to dissect the heterogeneity of endothelial cells in the adult lung, both during homeostasis, as well as during flu or bleomycin-induced injury. They identified three major populations, including REC, a subpopulation of the alveolar endothelium that expresses CD34. They found that CD34⁺ REC cells expanded following injury. Furthermore, following injury, they identified a new population, PEC, which is highly proliferative. Through pseudo-time analysis, they linked PEC to the miEC population, which is distinct from REC. These findings greatly advance current understanding of endothelial cell composition in lung and offer the first glimpse of the dynamic changes contributing to repair. This study will be of great interest to the general readership of eLife. Addressing the following comments will enhance the rigor and impact of the study.

1) At the start of the paper and in Abstract, endothelial cells were dissected into three major groups, maEC, miEC and REC. In Figure 4, for uninjured, endothelial cells were separated into more clusters (two for maEC, two for miEC). These additional insights into heterogeneity should be highlighted and further explored by key marker identification and spatial localization (see below).

2) Additional spatial analysis of subpopulations will be very informative. For example, how are the miEC and REC spatially distinct from each other? Where do the Ifi47-high vs Gpihbp1-high subclusters of miEC localize to? Where are PECs?

3) Following injury, there is an increase of REC, and an appearance of PEC. Localization of PEC in relationship to REC will allow an estimation of the extent of contribution of each population to injury.

4) Aside from the focus on PEC and cell cycle markers after injury, how the transcriptome of the other populations change will be informative, especially in the aspect of signaling.

5) CD34 increase should be quantified by qRT.

6) Parallel to the expansion of REC, how maEC and miEC subclusters change spatially following injury will be informative.

7) Figure 6—figure supplements 1 and 2, provide a better explanation how the different gene plots illustrate the curves.

Reviewer #2:

In this manuscript, the authors present a thorough analysis of scRNA-seq data generated from the endothelial compartment of the lung, before and after injury with influenza and bleomycin. They identify a novel endothelial cell type characterized by high expression of CD34 and of CAR4 that is present in alveolar areas and increases after injury. A role for these cells, called regulatory endothelial cells (RECs), in injury repair is inferred from these findings. The strength of this paper is the thorough mining of the scRNA-seq data. A serious weakness however is that mechanistic and functional conclusions are drawn based on mere inference from the scRNA-seq data, and therefore are speculation at best. Hence, this manuscript is merely descriptive and based on a very limited number of experiments that were maximally mined.

Specific comments:

1) The scRNA-seq studies used purified cells based on CD31 expression. Given the heterogeneity in endothelial phenotypes, does this strategy capture all endothelial cells? Following influenza infection, scRNA-seq was performed on whole lung. Were exactly the same EC populations identified in the control of that experiment, where CD31 purification was not used?

2) After injury, CD34-high ECs increase in the alveolar regions adjacent to regions of severe injury (Figure 3). It is inferred that this represents an accumulation REC, but this might as well simply represent upregulation of CD34 on other ECs. Using other markers for RECs, such as Car4, might be useful. Furthermore, increased CD34 expression is only evaluated by qualitative assessment of IF images. Could this be quantified by flow cytometry? Similarly, the increased in the abundance of these cells is shown in one IF image but is never quantified.

3) A role for proliferating ECs that might be derived from miECs is suggested in lung regeneration. Where were these located?

4) Rigorous quantification is also lacking in the data on cell cycling in Figure 5. Based on scRNA expression, cell cycle status of various populations was assessed. Quite remarkably, these data would suggest that even in steady-state, a very large fraction of all lung cell types are in cell cycle. This is surprising as the lung has been reported previously to be very quiescent, and sheds doubt on the reliability of this type of analysis to assess cell cycle status. It is then stated that after injury, in particular ECs cycle, but this appears true in steady-state as well. Furthermore, there are no statistics to support this conclusion. The “proliferating ECs” of course cycle even more, since they were identified as such by expression of proliferation markers.

5) The EdU incorporation data in Figure 5 also lack controls and statistics. The fraction of S-phase ECs seems remarkably high for steady-state, no data are provided on subpopulations of ECs (CD34-high, for example), no data are provided on other cell types in the sample, no statistical analysis is given, and a staining of PBS-treated mice (i.e. no EdU administration) is lacking as a control for EdU staining.

6) In Figure 6, lineage relations are inferred from scRNA-seq analysis, but the conclusions, while possible correct, are speculation at this point.

7) A similar comment is true for Figure 7, where ligand/receptor pair analysis suggests interactions between ECs and ATI cells. This is very likely correct, but not surprising, given the close apposition of these two cell types. This, however, does not inform on the role of either proliferating ECs or REC in the lung regeneration.

Reviewer #3:

In this manuscript Niethamer et al. report their findings regarding endothelial cell heterogeneity in normal adult and injured mouse lung as defined by single cell RNA-sequencing. Three classes of EC's are defined under homeostatic conditions including macrovascular EC's, microvascular EC's and a third group that the authors refer to as "regulatory" EC's. Parenchymal injury resulted in the appearance of a fourth cluster referred to as proliferative EC's. These findings are important to the field of lung regeneration and endothelial cell biology, but there are a number of technical and conceptual concerns as outlined below.

Specific issues/concerns:

1) Since normal endothelial single cell RNA-Seq data represent aggregated data from multiple individual mice it would be important to show the distinct contributions of these individuals to the entire data set for assessment of batch variability between individuals.

2) Even though the appearance of proliferative EC's in response to acute lung injury is reasonable and well supported by gene expression data, the use of the term regulatory EC's is speculative and not well justified based upon the data. Additional concerns include: a) One of the top differentially expressed genes appearing in REC's, Car4 or carboxypeptidase 4, has been described as broadly expressed within rat microvasculature EC's (Fleming et al., 1993). As such, it is not clear whether REC's represent a novel endothelial cell type, whether they represent a subset of microvascular ECs and/or whether there are species differences that account for this observation; b) none of the marker genes evaluated (with the exception of CD31) are endothelial cell-specific, which makes interpretation of immunofluorescence data challenging.

3) It would be helpful if data generated from uninjured adult mice vs either bleo or influenza virus injured lungs could be combined to determine how EC subsets change with injury. EC clusters generated with the combined transcriptomes of all lung cell types are most likely quite different to those generated if only EC's are evaluated in isolation. Furthermore, it is not clear (without lineage tracing and/or data aggregation) that direct relationships exist between cell types observed in the uninjured vs injured lungs. Accordingly, the authors interpretation of cellular relationships presented in the manuscript may represent an overly simplified scenario and could be somewhat misleading.

4) Reference is made to additional subsets of EC's revealed within single cell preparations of lung tissue after influenza virus infection, Figure 4—figure supplement 3, including lymphatic EC's, venous maEC's and Arterial maEC's. It is not clear how these EC subsets relate to those recovered from normal uninjured lung tissue. Are they lost following selection for CD31⁺ EC's? Candidate receptor-ligand interactions are identified between REC's and ATI cells. However, the basis for selecting these interacting cell types and the absence of validation make this analysis and interpretation rather speculative.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Defining the role of pulmonary endothelial cell heterogeneity in the response to acute lung injury" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Didier Stainier as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare the final submission.

While the reviewers agreed that the revisions of the manuscript are significant and that the revised manuscript is much improved, a few comments remained that would require your attention:

1) Please establish to what extent Car4⁺ corresponds to CD34-high ECs. For example, by sorting out CD34-high ECs (not just CD34 positive) and showing that Car4 is selectively expressed in this population.

2) Please determine the exact location of the Car4 population with respect to alveolar epithelial cells and other ECs, using Car4, not just CD34, as a marker.

3) Please perform EdU incorporation with a shorter pulse to avoid cytotoxic effect of EdU, which may affect proliferation.

4) Please clarify why is such a high fraction of both ECs and epithelial cells cycling in steady-state according to this analysis in Figure 5—figure supplement 5, and how these data relate to the other inferences from scRNAseq?

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:1) At the start of the paper and in Abstract, endothelial cells were dissected into three major groups, maEC, miEC and REC. In Figure 4, for uninjured, endothelial cells were separated into more clusters (two for maEC, two for miEC). These additional insights into heterogeneity should be highlighted and further explored by key marker identification and spatial localization (see below).

We agree that it is important to identify key markers and determine the spatial relationships of endothelial cell subsets in the distal lung. We have therefore performed RNAscope analysis using markers of each EC population in the adult lung, both at homeostasis and at 14 days post H1N1 injury. We found that miECs, marked by expression of Gpihbp1, Pecam1, and Plvap, localize to the alveolar capillary endothelium both at homeostasis and after influenza injury, while cells in large vessels do not express Gpihbp1. Arterial and venous maECs, marked by Cxcl12 and Vegfc, respectively, localize to large vessels both before and after injury. We have quantified relative numbers of Car4-high ECs at homeostasis and after H1N1 injury and determined that while they are localized throughout the capillary space at homeostasis, Car4-high ECs are increased in zones of mild to moderate as well as severe injury in H1N1-infected animals. We have added additional supplemental figures containing this analysis (Figure 3—figure supplements 2-3).

2) Additional spatial analysis of subpopulations will be very informative. For example, how are the miEC and REC spatially distinct from each other? Where do the Ifi47-high vs Gpihbp1-high subclusters of miEC localize to? Where are PECs?

We have addressed the localization of miECs, proliferating ECs, and Car4-high ECs at homeostasis and after injury using RNAscope. We show that miECs and Car4-high ECs are transcriptionally rather than spatially distinct, as both are distributed throughout the alveolar space at homeostasis and during repair. Proliferating ECs are rare at homeostasis but localize both to zones of mild to moderate tissue damage and to zones of more severe tissue damage after injury (Figure 3—figure supplements 2-3; Figure 5—figure supplement 4).

3) Following injury, there is an increase of REC, and an appearance of PEC. Localization of PEC in relationship to REC will allow an estimation of the extent of contribution of each population to injury.

We have addressed the localization of proliferating ECs and Car4-high ECs after H1N1 injury using RNAscope and find that both cell types are localized to zones 1/2 of mild to moderate injury as well as to zone 3 of more severe tissue damage, with an increase in numbers of each cell type after injury with respect to control (Figure 3—figure supplement 2; Figure 5—figure supplement 4).

4) Aside from the focus on PEC and cell cycle markers after injury, how the transcriptome of the other populations change will be informative, especially in the aspect of signaling.

To determine how heterogeneous EC populations change with injury, we have integrated the control and H1N1 datasets and performed additional analysis to directly compare cell clusters and their differential gene expression. We discovered that Car4-high ECs do not change in expression of key genes that differentiate this population from other EC populations after injury, but we do see statistically significant differences in expression of several signaling genes, genes that regulate the cytoskeleton, and the transcription factor Hes1. We have added violin plots to show these gene expression differences in Car4-high ECs and gene ontology analysis to describe genes differentially expressed across EC populations from uninjured and injured lungs. We find that after H1N1 injury, ECs upregulate genes related to translation, cell proliferation and survival, as well as cell death. This analysis is shown in two additional supplemental figures (Figure 4—figure supplements 5 and 6).

5) CD34 increase should be quantified by qRT.

We have used FACS to isolate CD34-high and CD34-low EC populations and have shown that relative Cd34, Car4, Ednrb, and Kdr expression is increased in CD34-high ECs by qRT-PCR. However, relative expression of other miEC and pan-EC markers such as Gpihbp1, Plvap, and Pecam1 is not significantly different between CD34-high and CD34-low ECs. We have added an additional supplemental figure containing this data (Figure 1—figure supplement 3). We have also quantified changes in number of CD34-high cells between control and H1N1-infected mice using flow cytometry and showed that the proportion of CD34-high ECs increases after H1N1 injury. This data forms part of the new Figure 5.

6) Parallel to the expansion of REC, how maEC and miEC subclusters change spatially following injury will be informative.

We have also addressed this with our RNAscope analysis. Please see response to comment 1 above and the new Figure 3—figure supplement 3.

7) Figure 6—figure supplements 1 and 2, provide a better explanation how the different gene plots illustrate the curves.

The feature plots associated with trajectory or curve 1 of the pseudotime analysis (Figure 6—figure supplement 1) and curve 2 (Figure 6—figure supplement 2) highlight the top differentially expressed genes that define each trajectory. We have refined this analysis to highlight the top 15 genes for each curve, which demonstrate that curve 1 is defined by expression differences between Car4-high ECs and proliferating ECs, whereas curve 2 is defined by differences between Car4-high ECs and miECs. We have added additional explanation to the main text to clarify these figures – please see paragraph four of subsection “Endothelial Cell Proliferation Increases Significantly During H1N1 Injury Response” and associated figure legends.

Reviewer #2:1) The scRNA-seq studies used purified cells based on CD31 expression. Given the heterogeneity in endothelial phenotypes, does this strategy capture all endothelial cells? Following influenza infection, scRNA-seq was performed on whole lung. Were exactly the same EC populations identified in the control of that experiment, where CD31 purification was not used?

As expected, we identified additional endothelial cell populations in the whole lung scRNA-seq experiments. These include separation of venous and arterial macrovascular endothelium and lymphatic endothelium. The additional separation of endothelial cell populations is likely the result of differences in unbiased clustering due to an increased number of cell types and additional variation in gene expression between cell types analyzed in the whole-lung experiments as well as a less biased cell isolation procedure. We have added sentences to the main text to clarify these differences and have better justified our choice of CD31 expression to purify endothelial cells.

2) After injury, CD34-high ECs increase in the alveolar regions adjacent to regions of severe injury (Figure 3). It is inferred that this represents an accumulation REC, but this might as well simply represent upregulation of CD34 on other ECs. Using other markers for RECs, such as Car4, might be useful. Furthermore, increased CD34 expression is only evaluated by qualitative assessment of IF images. Could this be quantified by flow cytometry? Similarly, the increased in the abundance of these cells is shown in one IF image but is never quantified.

We have quantified the increase in CD34-high ECs after H1N1 infection by flow cytometry and have found that the proportion of CD34-high ECs does increase significantly with influenza injury. We have added this quantification and associated statistical analysis to the new Figure 5. In addition, we have quantified numbers of Car4-expressing cells in different zones of injury following H1N1 infection by RNAscope and have seen a corresponding increase in these cells in regenerating regions, indicating that our IHC data likely does not represent simply an upregulation of CD34 protein. This quantitative analysis of Car4-expressing cells is shown in an additional supplemental figure (Figure 3—figure supplement 2).

3) A role for proliferating ECs that might be derived from miECs is suggested in lung regeneration. Where were these located?

We have addressed PEC localization after H1N1 influenza infection with RNAscope analysis and found that these cells localize to both regions of mild to moderate injury and regions of more severe injury. Please see response to reviewer 1, comments 2 and 3 above, as well as the new Figure 5—figure supplement 4.

4) Rigorous quantification is also lacking in the data on cell cycling in Figure 5. Based on scRNA expression, cell cycle status of various populations was assessed. Quite remarkably, these data would suggest that even in steady-state, a very large fraction of all lung cell types are in cell cycle. This is surprising as the lung has been reported previously to be very quiescent, and sheds doubt on the reliability of this type of analysis to assess cell cycle status. It is then stated that after injury, in particular ECs cycle, but this appears true in steady-state as well. Furthermore, there are no statistics to support this conclusion. The “proliferating ECs” of course cycle even more, since they were identified as such by expression of proliferation markers.

To assess and quantify changes in endothelial cell populations and their proliferation after H1N1 influenza injury, we have performed flow cytometry experiments at 7, 14, and 21 dpi and quantified the relative abundance of CD31⁺/ CD34-high and CD31⁺/ CD34-low cell populations and their EdU incorporation at homeostasis and during repair. We found that the proportion of CD31⁺/ CD34-high ECs increases over the course of the injury response, and that both CD34-high and CD34-low ECs incorporate EdU during regeneration. We have added an additional figure to the manuscript that includes this data (the new Figure 5 and associated Figure 5—figure supplements 1-3). We have also clarified the methods of the cell cycling analysis of our scRNA-seq datasets and have added text to the manuscript to emphasize the predictive nature of this analysis, as well as its advantages and limitations for assessing cell-cycle status. Please see paragraph three of subsection “Endothelial Cell Proliferation Increases Significantly During H1N1 Injury Response” and Figure 5—figure supplement 5.

5) The EdU incorporation data in Figure 5 also lack controls and statistics. The fraction of S-phase ECs seems remarkably high for steady-state, no data are provided on subpopulations of ECs (CD34-high, for example), no data are provided on other cell types in the sample, no statistical analysis is given, and a staining of PBS-treated mice (i.e. no EdU administration) is lacking as a control for EdU staining.

We have performed additional and more rigorous EdU incorporation experiments in control and H1N1-infected mice described in the new Figure 5 and associated Figure 5—figure supplements 1-3. We have also added a control experiment to assess the reliability of the EdU water administration and fluorescent labeling strategy we employed. We found that there is a small amount of background fluorescence in single cells of H1N1-infected mice not given EdU water but subjected to the fluorescent labeling assay. This may result in a slightly larger percentage of S-phase ECs quantified at both homeostasis and after injury but is very minor in comparison to the difference we observed in EdU incorporation between control and flu-infected mice (Figure 5—figure supplement 1). We have also analyzed proliferation in CD34-high and CD34-low ECs and found that both populations incorporate EdU over the course of injury repair. In analyzing this quantitative proliferation data, we have used statistical analysis to support our conclusions, which is shown in the new Figure 5. In addition, we analyzed proliferating non-ECs by RNAscope (Top2a⁺/Prc1⁺ cells that are Pecam1^-) and found that like proliferating ECs, these cells are increased in both zones 1/2 and zone 3 after H1N1 injury (Figure 5—figure supplement 4).

6) In Figure 6, lineage relations are inferred from scRNA-seq analysis, but the conclusions, while possible correct, are speculation at this point.

While we agree that these analyses are exploratory rather than confirmatory, we would argue that mechanistic analysis of these predicted relationships is beyond the scope of the current manuscript and forms the basis for future work. We have edited the text to clarify the role of these analyses in our conclusions as a whole and have modified our language to emphasize the speculative nature of these analyses.

7) A similar comment is true for Figure 7, where ligand/receptor pair analysis suggests interactions between ECs and ATI cells. This is very likely correct, but not surprising, given the close apposition of these two cell types. This, however, does not inform on the role of either proliferating ECs or REC in the lung regeneration.

Similar to Figure 6, the analyses presented in Figure 7 are exploratory rather than confirmatory. We have modified our language in the text to emphasize the fact that this analysis suggests predicted interactions and to clarify the role of this analysis in the manuscript as a whole.

Reviewer #3:1) Since normal endothelial single cell RNA-Seq data represent aggregated data from multiple individual mice it would be important to show the distinct contributions of these individuals to the entire data set for assessment of batch variability between individuals.

Both our CD31⁺ endothelial cell-specific and our H1N1 injury whole-lung scRNA-seq datasets contain aggregated data from multiple individual mice. We have therefore added additional supplemental figures showing the overlay of data from individual mice in each integrated dataset. The EC-specific dataset analyzed in Figure 1 is aggregated from three individual mice; in the new Figure 1—figure supplement 2, we show that all seven clusters 0-6 are identified in the dataset for each individual mouse. The H1N1 whole-lung dataset analyzed in Figure 4 is aggregated from two individual mice; in the new Figure 4—figure supplement 5, we find that integration of these two datasets with the control whole-lung dataset results in similar clustering and cell populations as does individual analysis. These figures nicely demonstrate the concordance between individual datasets used for aggregation.

2) Even though the appearance of proliferative EC's in response to acute lung injury is reasonable and well supported by gene expression data, the use of the term regulatory EC's is speculative and not well justified based upon the data. Additional concerns include: a) One of the top differentially expressed genes appearing in REC's, Car4 or carboxypeptidase 4, has been described as broadly expressed within rat microvasculature EC's (Fleming et al., 1993). As such, it is not clear whether REC's represent a novel endothelial cell type, whether they represent a subset of microvascular EC's and/or whether there are species differences that account for this observation; b) none of the marker genes evaluated (with the exception of CD31) are endothelial cell-specific, which makes interpretation of immunofluorescence data challenging.

We agree that our use of the term “regulatory endothelial cell” is a potential overstatement, and we have therefore revised the text to refer to these cells as “Car4-high ECs” in all text and figures. We have also performed further analysis of gene expression in Car4-high ECs by qRT-PCR of FACS-isolated CD34-high and CD34-low ECs and have found that CD34-high (Car4-high) ECs express higher levels of Cd34, Car4, Ednrb, and Kdr than CD34-low ECs, but similar levels of Pecam1, Gpihbp1, Plvap, and Vwf (Figure 1—figure supplement 3). We have also assessed the spatial localization of Car4-high ECs by RNAscope; similar to our CD34 IHC analysis, RNAscope for Car4 indicates that these cells are localized throughout the alveolar space at homeostasis (Figure 3—figure supplement 2). These data have led us to conclude that these cells represent a novel miEC population. We agree that high Car4 expression may not define a subpopulation of ECs in all species and have revised the Discussion to include reference to the above article. In our RNAscope analysis (please see response to reviewer 1, comment 1 above), we have used co-labeling with probes that detect pan-endothelial gene expression to provide more rigorous analysis of ECs from each subpopulation in control and H1N1-infected conditions (Figure 3—figure supplements 2-3).

3) It would be helpful if data generated from uninjured adult mice vs either bleo or influenza virus injured lungs could be combined to determine how EC subsets change with injury. EC clusters generated with the combined transcriptomes of all lung cell types are most likely quite different to those generated if only EC's are evaluated in isolation. Furthermore, it is not clear (without lineage tracing and/or data aggregation) that direct relationships exist between cell types observed in the uninjured vs injured lungs. Accordingly, the authors interpretation of cellular relationships presented in the manuscript may represent an overly simplified scenario and could be somewhat misleading.

We have added additional analysis of our scRNA-seq data to compare gene expression within EC subtypes before and after injury, including integration of control and H1N1 datasets (please see response to reviewer 1, comment 4 above). We have also added additional text to clarify the differences observed in our endothelial cell-specific and whole-lung scRNA-seq datasets (please see response to reviewer 2, comment 1 above).

4). Reference is made to additional subsets of EC's revealed within single cell preparations of lung tissue after influenza virus infection, Figure 4—figure supplement 3, including lymphatic EC's, venous maEC's and Arterial maEC's. It is not clear how these EC subsets relate to those recovered from normal uninjured lung tissue. Are they lost following selection for CD31+ EC's? Candidate receptor-ligand interactions are identified between REC's and ATI cells. However, the basis for selecting these interacting cell types and the absence of validation make this analysis and interpretation rather speculative.

We have added additional text to the manuscript to clarify these points (please see responses to reviewer 2, comment 1 and reviewer 2, comment 7 above).

[Editors’ note: what follows is the authors’ response to the second round of review.]

1) Please establish to what extent Car4+ corresponds to CD34-high ECs. For example, by sorting out CD34-high ECs (not just CD34 positive) and showing that Car4 is selectively expressed in this population.

We assessed this using the FACS strategy outlined in Figure 1—figure supplement 3, panel A, where our gating strategy separates CD34-high cells (right panel, upper right-hand box) and CD34-low/CD34^- cells (right panel, upper left-hand box). qRT-PCR comparing these two populations reveals that CD34-high cells express a significantly higher level of Car4 than CD34-lowcells (>6 fold) (panel C). Of note, these data were included in the previous version of the manuscript.

2) Please determine the exact location of the Car4 population with respect to alveolar epithelial cells and other ECs, using Car4, not just CD34, as a marker.

To address this question, we have conducted RNAscope for Car4, Pecam1 as a marker for all ECs, and either Hopx (alveolar type 1 cells) or Sftpc (alveolar type 2 cells). Spatial analysis of these images shows that both Car4-high ECs and other Pecam1⁺ ECs can be found juxtaposed with both AT2 and AT1 cells. Using these methods, we do not observe a preferential localization of Car4⁺ endothelial cells to AT1 or AT2 cells. We have added a new figure (Figure 3—figure supplement 3) as well as text to the manuscript describing these results.

3) Please perform EdU incorporation with a shorter pulse to avoid cytotoxic effect of EdU, which may affect proliferation.

To directly address the issue of EdU toxicity, we have compared endothelial cell proliferation as well as total proliferation by IHC using phosphohistone H3 (pHH3) in EdU-treated and untreated animals from the same H1N1 injury experiment at a defined time point of 14 dpi. We feel that this analysis addresses the issue of EdU toxicity directly, whereas redoing the entire experiment with a shorter EdU pulse would not completely address this (such studies would not have detected any possible toxicity) and would extend the revisions well past the 2-month time frame. Moreover, our choice of concentration (0.2 g/L) and timeline (7 days) forin vivoEdU administration in drinking water has plenty of precedent in the literature [1-2], and indeed, EdU administration in drinking water has been performed at up to fivefold higher concentration [3] as well as for longer periods up to 90 days with no reported toxicity [4]. Our new results indicate that 1) EdU treatment does not affect proliferation in the lung using the dose/timing in our experiments and 2) Total proliferation, EC proliferation, and proliferation in Car4-high ECs are all increased at 14 days post H1N1 infection as measured by IHC for pHH3 in control and injured animals, confirming our previous results. We have added a new figure (Figure 5—figure supplement 1) and text to the manuscript describing this new data. Of note, to complete this analysis, we have implemented a cell counting pipeline that utilizes the ilastic [5] and CellProfiler [6] programs to accurately and rigorously segment nuclei and assess cell numbers expressing multiple marker genes. Please refer to the Materials and methods section for a more detailed description of this new tool.

4) Please clarify why is such a high fraction of both ECs and epithelial cells cycling in steady-state according to this analysis in Figure 5—figure supplement 5, and how these data relate to the other inferences from scRNAseq?

Figure 5—figure supplement 6 (with new numbering resulting from addition of the supplemental figure described in (3) above) contains an informatic derivation of cell cycle state based on expression of genes related to cell cycle progression. This is therefore a predictive analysis rather than a functional one and does not allow us to determine whether a cell is actively cycling. Our subsequent EdU incorporation, RNAscope, and IHC analysis all confirm that the distal lung is quiescent at steady-state, with very little proliferation in endothelial cells or elsewhere. Therefore, although informatic analysis indicates that cells express some genes associated with S and G2/M phases at homeostasis, functional data indicates that most of these cells are not actively cycling. The discrepancy may relate to differences between cells poised to re-enter the cell cycle versus cells actively cycling. We have added text to the manuscript to reflect this. However, if the editors and reviewers feel that these data are confusing, we would be happy to remove them from the manuscript. The data contained in Figure 5 and its other supplements all provide more convincing functional evidence.

References

1) Meln et al.,Molecular Metabolism(2019) PMC: 6531874

2) Kryvalap et al.JBC(2016) PMC: 4697161

3) Amano et al.,Cell Metabolism(2013) PMC: 3931314

4) Clarke LE et al.,J Neurosci(2012) PMC: 3378033

5) Berg et al., Nat Methods (2019) PMID: 31570887

6) McQuin et al., PLoS Biol (2018) PMC: 6029841

Author details

1. Terren K Niethamer

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Department of Cell and Developmental Biology, University of Pennsylvania, Philadelphia, United States
   3. Penn-Children's Hospital of Philadelphia Lung Biology Institute, University of Pennsylvania, Philadelphia, United States
   4. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Project administration

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0914-994X

2. Collin T Stabler

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Department of Cell and Developmental Biology, University of Pennsylvania, Philadelphia, United States
   3. Penn-Children's Hospital of Philadelphia Lung Biology Institute, University of Pennsylvania, Philadelphia, United States
   4. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Conceptualization, Validation, Investigation, Visualization

   Competing interests

   No competing interests declared

3. John P Leach

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Department of Cell and Developmental Biology, University of Pennsylvania, Philadelphia, United States
   3. Penn-Children's Hospital of Philadelphia Lung Biology Institute, University of Pennsylvania, Philadelphia, United States
   4. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Formal analysis, Investigation, Visualization, Methodology

   Competing interests

   No competing interests declared

4. Jarod A Zepp

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Department of Cell and Developmental Biology, University of Pennsylvania, Philadelphia, United States
   3. Penn-Children's Hospital of Philadelphia Lung Biology Institute, University of Pennsylvania, Philadelphia, United States
   4. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Investigation, Visualization

   Competing interests

   No competing interests declared

5. Michael P Morley

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Penn-Children's Hospital of Philadelphia Lung Biology Institute, University of Pennsylvania, Philadelphia, United States
   3. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology

   Competing interests

   No competing interests declared

6. Apoorva Babu

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Resources, Data curation, Software, Formal analysis, Validation, Visualization, Methodology

   Competing interests

   No competing interests declared

7. Su Zhou

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States

   Contribution

   Resources, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology

   Competing interests

   No competing interests declared

8. Edward E Morrisey

   1. Department of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Department of Cell and Developmental Biology, University of Pennsylvania, Philadelphia, United States
   3. Penn-Children's Hospital of Philadelphia Lung Biology Institute, University of Pennsylvania, Philadelphia, United States
   4. Penn Cardiovascular Institute, University of Pennsylvania, Philadelphia, United States
   5. Penn Institute for Regenerative Medicine, University of Pennsylvania, Philadelphia, United States

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Project administration

   For correspondence

   emorrise@pennmedicine.upenn.edu

   Competing interests

   Reviewing editor, eLife

   "This ORCID iD identifies the author of this article:" 0000-0001-5785-1939

• 11,069

  Page views

• 1,528

  Downloads

• 97

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.