1. Developmental Biology
  2. Immunology and Inflammation
Download icon

Diverse homeostatic and immunomodulatory roles of immune cells in the developing mouse lung at single cell resolution

  1. Racquel Domingo-Gonzalez
  2. Fabio Zanini
  3. Xibing Che
  4. Min Liu
  5. Robert C Jones
  6. Michael A Swift
  7. Stephen R Quake  Is a corresponding author
  8. David N Cornfield  Is a corresponding author
  9. Cristina M Alvira  Is a corresponding author
  1. Division of Critical Care Medicine, Department of Pediatrics, Stanford University School of Medicine, United States
  2. Center for Excellence in Pulmonary Biology, Stanford University School of Medicine, United States
  3. Department of Bioengineering, Stanford University, United States
  4. Prince of Wales Clinical School, Lowy Cancer Research Centre, University of New South Wales, Australia
  5. Division of Pulmonary, Asthma and Sleep Medicine, Department of Pediatrics, Stanford University School of Medicine, United States
  6. Department of Chemical and Systems Biology, Stanford University, United States
  7. Chan Zuckerberg Biohub, United States
  8. Department of Applied Physics, Stanford University, United States
Research Article
  • Cited 2
  • Views 1,176
  • Annotations
Cite this article as: eLife 2020;9:e56890 doi: 10.7554/eLife.56890

Abstract

At birth, the lungs rapidly transition from a pathogen-free, hypoxic environment to a pathogen-rich, rhythmically distended air-liquid interface. Although many studies have focused on the adult lung, the perinatal lung remains unexplored. Here, we present an atlas of the murine lung immune compartment during early postnatal development. We show that the late embryonic lung is dominated by specialized proliferative macrophages with a surprising physical interaction with the developing vasculature. These macrophages disappear after birth and are replaced by a dynamic mixture of macrophage subtypes, dendritic cells, granulocytes, and lymphocytes. Detailed characterization of macrophage diversity revealed an orchestration of distinct subpopulations across postnatal development to fill context-specific functions in tissue remodeling, angiogenesis, and immunity. These data both broaden the putative roles for immune cells in the developing lung and provide a framework for understanding how external insults alter immune cell phenotype during a period of rapid lung growth and heightened vulnerability.

Introduction

Prior to birth, the lung is maintained in a fluid-filled, immune-privileged, hypoxic environment. Upon the infant’s first breath, the lung rapidly transitions to an oxygen rich environment (Koos and Rajaee, 2014), subjected to the mechanical forces of spontaneous ventilation, and exposed to diverse pathogens present in the external environment (Wirtz and Dobbs, 2000). The immune system is essential for lung homeostasis, wound-healing and response to pathogens (Lloyd and Marsland, 2017). Although the development of the murine immune system begins during early embryogenesis, little is known regarding how the dynamic physiologic changes at birth alter the lung immune cell landscape, and whether specific immune cell subpopulations influence lung growth and remodeling in addition to serving established immunomodulatory functions.

Immune cells play a central role in the development of many organs, including the kidney (Munro and Hughes, 2017), brain (Lenz and Nelson, 2018) and mammary gland (Coussens and Pollard, 2011). In highly vascularized organs, macrophages localize to the tips of vascular sprouts to enhance vascular network complexity (Rymo et al., 2011), promote angiogenesis (Riabov et al., 2014), and regulate vascular patterning (Leid et al., 2016). Although proximal lung branching occurs during early gestation, the development of distal airspaces begins only just before birth, and continues postnatally during alveolarization, the final stage of lung development characterized by rapid lung parenchymal and vascular growth (Rodríguez-Castillo et al., 2018). Disruption of late lung development contributes to pediatric lung diseases such as bronchopulmonary dysplasia (BPD), a disease of arrested lung development observed in premature infants, and pulmonary hypertension. Whether temporal regulation of specific immune populations informs lung immune function or the significant pulmonary parenchymal and vascular growth and remodeling occurring during early postnatal life remains unknown.

The prevailing notion is that the neonatal immune compartment is immature (Restori et al., 2018). Attenuated innate immunity (Sadeghi et al., 2007), poor immune-stimulatory function of antigen presenting cells (De Kleer et al., 2014), and skewed adaptive immune responses may underlie the heightened susceptibility of infants to viral and bacterial infections (Restori et al., 2018). Although the neonatal immune system can be induced to manifest adult-like responses under certain conditions (Pertmer et al., 2001), this type-2-skewed immune environment likely facilitates immune surveillance and metabolic and tissue homeostasis (Lloyd and Snelgrove, 2018). Given that lung development continues as the immune cell landscape is rapidly evolving, identifying how birth-related changes in lung physiology may influence lung immune cell maturation and function has not been addressed.

In this report, we combined single cell transcriptomics (scRNA-Seq) with fluorescent multiplexed in situ hybridization (FISH) and flow cytometry to characterize changes in composition, localization, and function of immune cells in the murine lung from just before birth through the first three weeks of postnatal life. At birth, immune cell heterogeneity increased dramatically from an embryonic landscape dominated by immature, proliferative macrophages to a complex landscape comprised of multiple types of macrophages, dendritic cells, granulocytes, and lymphocytes. Dynamic changes in macrophage heterogeneity were particularly striking, both transcriptionally and spatially, including the presence of embryonic macrophages encircling developing vessels prior to birth. After birth, these embryonic macrophages disappear, and numerous unique macrophage populations emerge, each exhibiting unique gene signatures suggesting specific roles in immunosuppression, pathogen surveillance, angiogenesis, and tissue remodeling. Multiple populations of dendritic cells, basophils, mast cells and neutrophils are also present in the postnatal lung, expressing genes important for rapid pathogen response. In contrast, although lymphocytes increase in abundance across three weeks, they remain functionally immature and skewed toward type-2 immunity. Taken together, our data demonstrate a previously underappreciated plasticity of immune cells in the perinatal and neonatal lung suggesting unique and essential roles in regulating immune function and lung structure.

Results

Diversity of the lung immune landscape increases dramatically after birth

To comprehensively define the lung immune landscape at birth, we isolated whole lungs from C57BL/6 (B6) mice at four stages of perinatal development: the early saccular (E18.5), late saccular (P1) early alveolar (P7) and late alveolar stages (P21) (Figure 1A), and quantified gene expression by scRNA-Seq. Lung tissue was isolated, the pulmonary vasculature perfused to remove circulating immune cells, and the tissue digested using an in-house optimized protocol to ensure maximal cell viability as published protocols (Nguyen et al., 2018; Jungblut et al., 2009) induced high amounts of cell death in the embryonic and early postnatal lung (Figure 1—figure supplement 1). Live CD45+ cells were sorted by FACS, processed by Smart-seq2 and sequenced on Illumina NovaSeq 6000 (Figure 1A). Gene expression was computed as previously described (Zanini et al., 2018) and 4,052 cells from eight mice, 1 female and one male for each time point were analyzed, with an average of 1.03 million mapped read pairs and ~4000 genes per cell (see Materials and methods). To quantify whether the different mice contributed spurious variation to the data, a distribution level approach was chosen. For each cell type and time point, 100 pairs of cells from either the same mouse or different mice were chosen and the distance in t-SNE space calculated. The cumulative distributions for those pairs were subsequently plotted to check whether pairs from different animals had a significantly longer distance than cells from the same mouse. We found no difference in the cumulative distributions (as also evident, on a qualitative level, by observation of the embeddings), indicating the absence of significant variation between the mice at each timepoint (Figure 1—figure supplement 2).

Figure 1 with 2 supplements see all
Diversity of the lung immune landscape increases dramatically after birth.

(A) Overview of the experimental design including the four timepoints (E18.5, P1, P7, P21) corresponding to key stages in late lung development. Lungs were isolated, perfused, and digested and immune cells isolated by fluorescence activated cell sorting (FACS) for the dead-stain-, EPCAM-, CD45+ population. (B) t-Distributed Stochastic Neighbor Embedding (t-SNE) and unsupervised clustering of over 4000 immune cells identifies fifteen distinct populations. (C) Dot plot showing level of expression (purple to yellow), and fraction of the population expressing the particular gene (dot size) for distinguishing genes expressed by the Leiden clusters broadly separated into myeloid (M), dendritic cell (DC) granulocyte (G) and lymphocyte (L) populations. (D) t-SNE of immune cell clusters identifying developmental timepoint of cell origin with E18.5 (blue), P1 (yellow), P7 (orange) and P21 (red). (E) Quantification of the abundance of specific immune subpopulations in the lung at each developmental timepoint expressed on a log (Rodríguez-Castillo et al., 2018) scale as percentage of total immune cells.

Fifteen cell clusters were identified via Leiden community detection (Traag et al., 2019) and verified by t-distributed stochastic neighbor embedding (t-SNE) (van der Maaten and Hinton, 2008; Figure 1B). Myeloid cells separated into eleven clusters, including five distinct macrophage/monocyte subpopulations with shared expression of Cd68, and distinguished by expression of Gal (Mac I), Itgax (Mac II), Car4 and Itgax (Mac III), C1qa (Mac IV), or Plac8 (Mac V). Dendritic cells (DCs) separated into three clusters, all expressing some amount of Batf3, but distinguished by the expression of Itgae (cDC1), Cd209a (cDC2), or Mreg (mig-DC). We also identified mast cells (expressing Mcpt4), basophils (Mcpt8), and neutrophils (Retnlg). Four lymphoid clusters were found, consisting of B cells (expressing Ms4a1), T cells (Cd3e), natural killer (NK) cells (Gzma), and group two innate lymphoid cells (ILC2) (Areg) (Figure 1C).

We next assessed cluster distribution across time (Figure 1D and E). Mac I cells dominate the late embryonic lung, with fewer macrophages scattered among clusters II, IV and V and an even smaller number of granulocytes and lymphocytes. After birth, immune cell heterogeneity increased explosively, concomitant with the disappearance of Mac I. Granulocyte abundance peaked just after birth and lymphocyte abundance increased progressively.

Expression of Dab2 and Plac8 broadly separates macrophages and monocytes

Clusters Mac I-V exhibited the most striking heterogeneity, so we analyzed their transcriptomes and spatial distribution in detail. All five clusters shared high expression of Cd68, indicative of macrophages or monocytes (m/m) (Figure 2A). At E18.5, Mac I comprised ~80% of m/m cells and Mac II, IV, and V each comprised 5–10% while Mac III was almost absent (Figure 2B and C). After birth, Mac I disappeared while Mac II abundance peaked to 35% of the total before decreasing again and disappearing by P21. Mac III and Mac V abundance increased steadily with Mac V being most abundant at all postnatal timepoints. The Mac IV population was relatively stable over time at ~10% of the total.

Figure 2 with 2 supplements see all
Macrophage populations present before and after birth broadly separate into two populations based on expression of Dab2 and Plac8.

(A) t-SNE plot depicting Cd68 expression in the five macrophage populations. (B) Separate embeddings for prenatal versus postnatal macrophages, identifying developmental timepoint of cell origin with E18.5 (blue), P1 (yellow), P7 (orange) and P21 (red). (C) Quantification of the abundance of each macrophage subpopulation at each developmental timepoint expressed on a log (Rodríguez-Castillo et al., 2018) scale as percentage of total macrophages. (D) t-SNE plots depicting expression of Dab2 and Plac8 within the macrophages present pre- and post-birth. (E) Multiplexed in situ hybridization to detect gene expression of Cd68 (yellow), Dab2 (red), and Plac8 (green) in lung tissue from mice at E18.5 and P7. Quantification of Dab2+ and Plac8+ cells with data = mean ± SD for n = 4–15 images from three separate FISH experiments. Calibration bar = 20 μm. For all t-SNE embeddings, the color scale is a logarithmic scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell. Source files of all fluorescent micrographs used for the quantitative analysis are available in the Figure 2—source data 1.

The Mac I-V clusters broadly separated into two groups based upon expression of the disabled two gene (Dab2), which regulates macrophage polarization (Adamson et al., 2016) and the placenta-specific eight gene (Plac8), which is related to bacterial immunity (Ledford et al., 2007). Mac I-IV cells expressed Dab2 but not Plac8 while Mac V showed the opposite pattern (Figure 2D). We confirmed by multiplexed FISH that all Cd68+ cells in the lung expressed either Dab2 or Plac8 at both E18.5 and P7 (Figure 2E and Figure 2—figure supplement 1). Quantification in situ identified 88.9 ± 8.4% of E18.5 and 65.8 ± 8.2% of P7 Cd68+ cells as Dab2+ with the remainder Plac8+. Of note, Dab2 and Plac8 expression was not consistent with previously reported markers of macrophage lineages derived from yolk sac and fetal liver (Tan and Krasnow, 2016; Guilliams et al., 2013; Figure 2—figure supplement 2).

Embryonic macrophages are proliferative and encircle developing vessels prior to birth

Mac I cells are the predominant immune population at E18.5, hence we aimed to understand their function and localization. Differentially expressed genes (DEGs) in Mac I included the proliferation markers Mki67 and Mcm5 (Figure 3A). Across macrophages and monocytes, proliferation decreased from 60% of cells at E18.5 to only 10% by P21. Most proliferating cells were Mac I prior to birth and distributed across Mac II-V after birth (Figure 3—figure supplement 1). These data are consistent with prior reports of ‘bursts’ of proliferation after recruitment of macrophages into embryonic tissues, followed by low-level self-renewal by adulthood (Kierdorf et al., 2015).

Figure 3 with 1 supplement see all
Embryonic macrophages encircle developing blood vessels prior to birth.

(A) t-SNE plots depicting expression of Mki67 and Mcm5 in the macrophage clusters with low expression in purple and high expression in yellow, with quantification of proliferating macrophages at each timepoint. In situ hybridization at E18.5 to detect: (B) Cd68 (red) and Cdh5 (green) with dotted lines identifying macrophages in circles around small vessels; (C) Epcam (green), Cd68 (red), and extracellular matrix (ECM, yellow), with white dotted circles identifying Cd68+ cells or groups of cells; (D) Quantification of the number of Cd68+ cells around vessels versus in lung parenchyma, with data = mean ± SD in n = 14 images, from five independent FISH experiments. (E) In situ hybridization at E18.5 to detect Mki67 (red), Cd68 (green), and ECM (yellow), with white dotted circles identifying Cd68+Mki67+ cells, and solid circles identify Cd68+ Mki67- cells. Quantification of the number of perivascular and parenchymal Mki67+ and Mki67- Cd68+ cells in n = 10 images with data = mean ± SD from two independent FISH experiments. (F) In situ hybridization at E18.5 to detect Cd68 (white) Dab2 (red), and Plac8 (green) with white dotted circles identifying Cd68+Dab2+ macrophages in the left panel, and arrows identifying Plac8+ cells in a separate area from the same slide (right panel). (G) In situ hybridization at E18.5 to detect Cd68 (red), Gal (white), and C1qa (green), with quantification of Gal+ and C1qa+ Cd68+ cells with data = mean ± SD from n = 12 images from three independent FISH experiments. (H) t-SNE plots of genes previously associated with a perivascular macrophage phenotype. (I) In situ hybridization of lung at P1 to detect Cd68 (red) and ECM (yellow), with white dotted circles identifying isolated macrophages around blood vessels. In each micrograph, calibration bar = 20 μm and ‘V’ denotes ‘vessel and ‘A’ ‘airway’. For all t-SNE embeddings, the color scale is a logarithmic scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell. Source files of all fluorescent micrographs used for the quantitative analysis are available in the Figure 3—source data 1, Figure 3—source data 2, and Figure 3—source data 3.

Localization of Mac I cells within the E18.5 lung revealed Cd68+ cells scattered throughout the lung parenchyma but also, surprisingly, forming almost complete rings around blood vessels of 20–30 μm in diameter found adjacent to large, conducting airways (Figure 3B). In contrast, Mac I cells were not found encircling small airways (Figure 3C). Quantification of the percentage of Cd68+ cells located around vessels versus in the lung parenchyma identified 28.4 ± 14.2% of Cd68+ cells surrounding small vessels (Figure 3D). Many of the perivascular macrophages expressed Mki67 (Figure 3E), and quantification of Mki67+ macrophages identified 55.2 ± 9.3% of proliferative perivascular Cd68+ cells, and a similar number of proliferative parenchymal Cd68+ cells (51.1 ± 16.8%). Perivascular macrophages were found on in situ analysis to express Dab2 but not Plac8 (Figure 3F). Given that Dab2+ cells at E18.5 include Mac IV cells, we aimed to distinguish these from the Mac I cells by simultaneously detecting the expression of Gal and C1qa, which were determined to be specific markers (see below). These studies demonstrated that the majority of Cd68+ macrophages surrounding the vessels were Gal+ (83.1 ± 12.8%) (Figure 3G), indicating a predominance of Mac I and a small number of Mac IV cells among the perivascular macrophage population. We then asked whether any of our Mac I-V clusters are related to previously reported perivascular macrophages that promote vascular remodeling in the developing hindbrain and retina (Pucci et al., 2009). The Mac I cluster expressed Cxcr4 and Nrp1 but failed to express many other genes characteristic of these previously reported macrophages, suggesting a distinct phenotype (Figure 3H). Moreover, concentric perivascular macrophages were not observed at P1, consistent with a function specific to Mac I prior to birth (Figure 3I). Taken together, these data suggest that within the embryonic lung, Mac I macrophages are highly proliferative and localize to small vessels, suggesting a potential role in pulmonary vascular growth or remodeling.

Distinct transcriptional profiles and spatial distribution suggest specific physiologic functions for discrete macrophage populations

Macrophage and monocyte heterogeneity increased rapidly after birth. To characterize this expanding diversity we computed DEGs and performed pathway analysis for each of the Mac I-V clusters (Figure 4A, Figure 4—figure supplement 1, and Tables 15). The most enriched pathways for Mac I were pathways associated with mitosis, cell cycle and DNA replication, consistent with the increased rate of proliferating cells within this cluster. Beyond proliferation, examination of the top DEGs identified genes associated with glycolysis, reflective of the hypoxic fetal environment compared to postnatal air-breathing life. Mac I-specific DEGs also included Crispld2, a glucocorticoid-regulated gene previously thought to be restricted primarily to the lung mesenchyme that promotes lung branching (Oyewumi et al., 2003). Crispld2 haploinsufficient mice exhibit impaired alveolarization and disorganized elastin deposition (Lan et al., 2009). Mac I cells also expressed Spint1, encoding hepatocyte growth factor activator inhibitor type 1 (HAI-1), a membrane bound serine proteinase inhibitor and regulator of angiogenesis (Figure 4A and Table 1). Loss of Spint1, results in a complete failure of placental vascularization and embryonic lethality at E10 that appears to result from a loss of basement membrane integrity (Fan et al., 2007). The most specific marker for Mac I was Gal, encoding galanin, a regulatory peptide that harbors both pro- and anti-inflammatory functions (Lang et al., 2015), promotes an anti-thrombotic phenotype in endocardial EC (Tyrrell et al., 2017), and regulates growth and self-renewal of embryonic stem cells (Tarasov et al., 2002). Galanin also inhibits inflammatory and histamine-induced vascular permeability in a number of experimental models (Ji et al., 1995; Jancsó et al., 2000; Holmberg et al., 2005), and functions as a vasoconstrictor, limiting blood flow in the cutaneous microcirculation (Schmidhuber et al., 2007). Localization identified Mac I cells throughout the lung parenchyma in addition to those found encircling small vessels (Figure 4B).

Figure 4 with 4 supplements see all
Distinct transcriptional profiles and spatial distribution suggest specific physiologic functions for discrete macrophage populations.

(A) Heatmap of select differentially expressed genes within enriched pathways illustrated. The color scale is logarithmic with a pseudocount of 0.1 counts per million, normalized to the highest expressing population for each gene, depicting 0 (none), 33 (low), 67 (mid), and 100% (high) expression compared to the highest expressing population. (B) t-SNE plots demonstrating high expression of Gal in Mac I cells, and in situ hybridization at E18.5 to detect Mac I cells that co-express Gal (red) and Cd68 (green). (C) t-SNE plots demonstrating high expression of Itgax but not Car4 in Mac II cells, and in situ hybridization to detect Mac II cells at P1 expressing Itgax, and Cd68 but not Car4 (dotted line), and additional Mac III cells in the same image co-expressing Itgax, Cd68, and Car4 (arrow). (D) t-SNE plots demonstrating high expression of Itgax and Car4 in Mac III cells, and in situ hybridization detecting Mac III cells at P7 expressing Itgax, Car4 and Cd68 now located within alveoli. (E) t-SNE plot demonstrating high expression of C1qa in Mac IV cells, and in situ hybridization to detect Mac IV cells expressing C1qa (green) and Cd68(red) at P7, with ECM marked in yellow, localizing Mac IV cells abutting vessels and large airways. Calibration bar = 20 μm for all panels. For all t-SNE embeddings, the color scale is a logarithmic scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell. (F) t-SNE plot showing a developmental gradient across Mac I-III. Pseudotime ordering of the cells identified a global trajectory from the starting cell (star) in Mac I to Mac III, and local cell states revealing multiple areas of local attraction within the Mac II and Mac III clusters. (G) RNA velocity demonstrating local group velocity from Mac I to Mac II and continuation of the vectors through Mac III.

Table 1
Top 35 differentially expressed genes in the Mac I cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 35 upregulated genes are shown for the Mac I cluser; n = 2 mice.

Mac I cluster top 35 differentially expressed genes
Gene nameStatisticP valueLog2 fold change
Gal0.61335.42E-533.6207
Crispld20.51.89E-343.6018
Ncapd20.45671.40E-283.3349
Pclaf0.494.88E-333.1388
Mcm50.50672.07E-352.9792
Cdca80.441.74E-262.8548
Lig10.459.89E-282.7655
AI5068160.52337.05E-382.7046
Asf1b0.44336.73E-272.5712
Smc20.43674.44E-262.4042
Fkbp50.55331.42E-422.3039
Mcm70.441.74E-262.1166
Cks1b0.459.89E-282.0842
A930007I19Rik0.465.23E-292.0187
Psat10.522.24E-372.0158
Pfkl0.44672.59E-271.9417
Ezh20.44336.73E-271.8453
Dkc10.47339.21E-311.8027
Cluh0.441.74E-261.7959
Scd20.441.74E-261.7455
Ruvbl20.44336.73E-271.7334
Mif0.49331.67E-331.6941
Tmem2730.441.74E-261.628
Stmn10.44336.73E-271.5905
Spint10.52672.20E-381.5781
Tpi10.50336.29E-351.562
C230062I16Rik0.459.89E-281.5462
Nt5dc20.44672.59E-271.5376
Ran0.45671.40E-281.5078
Trf0.59332.32E-491.4397
Gcnt10.441.74E-261.4095
Abcd20.465.23E-291.3537
Anp32b0.472.56E-301.3418
Dbi0.51677.04E-371.2018
Atp5g30.522.24E-371.1808
Table 2
Top 35 differentially expressed genes in the Mac II cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 35 upregulated genes are shown for the Mac II cluster; n = 2 mice.

Mac II cluster top 35 differentially expressed genes
Gene nameStatisticP valueLog2 fold change
Adgrl30.57138.85E-423.8593
Adarb10.42165.92E-232.6952
Ms4a8a0.45282.11E-262.4485
Nav20.47514.97E-292.1319
Tmcc30.42899.85E-242.0579
Ldhb0.38098.26E-192.0235
Acp50.38374.45E-192.0198
Dapk10.36582.20E-171.8704
Itgax0.58252.02E-431.8680
Csf2rb20.34591.11E-161.7791
H2-DMa0.50222.22E-321.7642
Trerf10.38503.31E-191.7402
Dmxl20.41176.59E-221.7264
Tgm20.36631.99E-171.6906
Ear20.51593.79E-341.6602
Dok20.36951.00E-171.6373
Flt10.36552.34E-171.5457
Axl0.50685.74E-331.4698
AU0202060.37235.40E-181.4207
Ece10.39622.54E-201.4075
Ear100.39384.41E-201.4069
Il180.40841.46E-211.4021
Clec7a0.49591.37E-311.3413
Spint10.37245.29E-181.3407
Mpeg10.54544.05E-381.3269
Il1rn0.34631.11E-161.2922
Clec4n0.44541.45E-251.2765
Fn10.40881.34E-211.2497
Il1b0.40021.02E-201.2281
Dst0.44846.72E-261.1994
F11r0.41294.97E-221.1982
Nceh10.35482.24E-161.1299
Neurl30.37215.74E-181.1091
Tnfaip20.37851.40E-181.0880
Plet10.44611.23E-251.0845
Table 3
Top 35 differentially expressed genes in the Mac III cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 35 upregulated genes are shown for Mac III cluster; n = 2 mice.

Mac III cluster top 35 differentially expressed genes
Gene nameStatisticP valueLog2 fold change
Slc39a20.66003.98E-629.2983
Atp6v0d20.95331.14E-1517.1784
Krt190.75331.56E-836.8066
Cd200r40.75331.56E-835.9139
Gpnmb0.69675.32E-705.0666
Ear10.62671.66E-554.9379
Car40.75001.10E-824.6965
Ly750.78672.26E-924.4964
Bhlhe410.78001.51E-904.4452
Slc7a20.79333.16E-944.1798
Lrp120.65339.21E-613.7640
Ccl60.87004.58E-1183.7161
Spp10.62671.66E-553.4907
Serpinb1a0.66003.98E-623.4778
Mgll0.80674.96E-983.4554
Il1rn0.65671.93E-613.2640
Kcnn30.76003.03E-853.2372
Ralgds0.70009.49E-713.1251
Dst0.76334.13E-863.1042
Atp10a0.66003.98E-623.0453
Ccnd20.72333.77E-762.9813
Fabp50.62337.18E-552.9667
Cd360.76003.03E-852.9174
Vat10.74677.57E-822.8682
Myof0.76003.03E-852.8483
Plet10.75672.19E-842.7934
Hvcn10.69675.32E-702.7689
Fth-ps30.69332.94E-692.7627
Lima10.65339.21E-612.7359
Itgax0.75672.19E-842.7210
Card110.63671.93E-572.6520
Mir692-20.75672.19E-842.5974
Ctsd0.81675.70E-1012.5402
Klhdc40.65004.35E-602.5367
Mir692-10.65004.35E-602.4553
Table 4
Top 35 differentially expressed genes in the Mac IV cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 35 upregulated genes are shown for the Mac IV cluster; n = 2 mice.

Mac IV cluster top 35 differentially expressed genes
Gene nameStatisticP valueLog2 fold change
Fcrls0.73672.28E-7912.2139
Pf40.79002.70E-9311.4117
Gas60.64339.38E-599.6283
C1qc0.87333.18E-1199.1247
Ccl120.64004.28E-588.7381
C1qa0.87004.58E-1188.3676
Stab10.80334.55E-977.6076
C1qb0.86001.17E-1147.0450
Sdc40.72002.32E-756.3860
Igfbp40.74335.15E-816.2878
Ms4a70.72002.32E-756.0618
Slc40a10.60333.72E-516.0358
Tmem176b0.86001.17E-1145.5402
Maf0.75001.10E-825.4450
S1pr10.63671.93E-575.3458
Igf10.67003.28E-644.8862
Tmem370.59009.02E-494.8077
Ophn10.59332.32E-494.6672
Tmem176a0.71671.41E-744.6367
C3ar10.71004.98E-734.5725
Smagp0.67003.28E-644.5052
Cxcl160.58004.98E-474.2439
Sesn10.56333.27E-444.0183
Apoe0.76003.03E-854.0149
Hpgds0.63338.62E-573.9920
Lgmn0.72676.04E-773.9551
Ms4a4a0.56679.10E-453.5567
Selenop0.62003.07E-543.4452
Cx3cr10.60001.49E-503.3845
Abca10.54335.69E-413.2834
Tcf40.54335.69E-413.1404
Blvrb0.55331.42E-423.1103
Aif10.69332.94E-693.0826
Abca90.58331.32E-473.0325
Pea15a0.54001.91E-403.0198
Table 5
Top 35 differentially expressed genes in the Mac V cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 35 upregulated genes are shown for the Mac V cluster; n = 2 mice.

Mac V cluster top 35 differentially expressed genes
Gene nameStatisticP valueLog2 fold change
I830127L07Rik0.48671.42E-327.0634
Ifitm60.88001.41E-1216.9362
Pglyrp10.68678.67E-686.2150
Serpinb100.53332.09E-396.1537
Plac80.96006.02E-1555.9647
Adgre40.60333.72E-515.8672
Nxpe40.50001.89E-344.8441
Ifitm30.92671.17E-1394.6727
Plbd10.72676.04E-773.9525
Susd30.50336.29E-353.9210
Hp0.63338.62E-573.8550
Pmaip10.47673.29E-313.7016
Adgre50.75001.10E-823.5335
Ifi27l2a0.56679.10E-453.3416
Atp1a30.49004.88E-333.3206
Ifitm20.76003.03E-853.2359
Xdh0.48334.07E-323.2003
Pou2f20.47339.21E-313.1216
Rnase60.50672.07E-353.0629
Pla2g70.68002.43E-663.0556
Thbs10.57671.86E-463.0133
Itga40.66003.98E-622.9055
Itgam0.55004.91E-422.7702
Napsa0.60001.49E-502.6112
Sorl10.50336.29E-352.5328
Wfdc170.49675.64E-342.5299
Clec4a10.53006.81E-392.4922
Slc11a10.52672.20E-382.4649
Fyb0.64004.28E-582.4454
Srgn0.66338.14E-632.2151
Cnn20.49331.67E-332.1880
Samhd10.61671.30E-532.1715
Lst10.66671.65E-632.1340
Cd300a0.51006.78E-362.1007
Nr4a10.61002.24E-522.0001

The Mac II cluster rapidly appeared after birth and expressed a gene signature suggesting a putative role in immune regulation and tissue remodeling. Pathway analysis identified top pathways regulating neutrophil homeostasis, endocytosis, apoptosis, and motility (Figure 4—figure supplement 1). Examination of the top DEGs identified shared expression of the chemokine receptors Ccr2 and Ccr5 with Mac I, molecules important for immune cell migration and localization. Also similar to Mac I and III, Mac II cells expressed genes associated with matrix remodeling and angiogenesis (Fn1, Axl, and Tgm2) (Senger and Davis, 2011; Tanaka and Siemann, 2019; Figure 4A and Table 2). We also examined expression patterns of secreted factors that regulate key signaling pathways important for lung epithelial progenitor cell function and differentiation including the Wnt, BMP, Tgf-β, Fgf and Shh pathways (Morrisey and Hogan, 2010; Figure 4—figure supplement 2). In addition to Tgfb1 and Tgfb2, Mac II cells expressed Ntn1 and Ntn4, molecules that modulate the response of the lung endoderm to FGFs, regulating epithelial bud outgrowth and shape (Liu et al., 2004), and Hhip, a Hedgehog binding protein that also modulates FGF signaling (Chuang et al., 2003). Hhip null mice exhibit defective lung branching and haploinsufficient mice develop spontaneous emphysema (Lao et al., 2016). The Mac II cluster also shared genes with Mac III important for regulating inflammation, including genes that promote intracellular killing of pathogens (Acp5 and Mpeg1) (Benard et al., 2015), but also genes that suppress inflammation (Il1rn and Dapk1) (Iannitti et al., 2016; Lai and Chen, 2014). A subpopulation within Mac II shared high expression of major histocompatibility complex (MHC) class II genes (H2-Ab1, H2-Eb1, Cd74) with a subset of Mac IV (Figure 4—figure supplement 3) suggesting a role in antigen presentation. Overall, the transient presence of Mac II together with its transcriptional signature suggests a dual role in tissue remodeling and fine tuning of immune response required immediately after birth. Studies to localize Itgax+Car4- Mac II cells in situ at P1 identified them in the distal lung parenchyma, mixed with scattered Mac III cells which were located closer to the alveolar wall (Figure 4C and Figure 4—figure supplement 4).

The Mac III cluster uniquely expressed alveolar macrophage signature genes (Car4, Bhlhe41, Trim29) (Figure 4A, Table 3). Top pathways included the inflammatory response (Figure 4—figure supplement 1), including numerous genes that constrain inflammation (Cd200r4, Gpnmb, Il1rn) (Ripoll et al., 2007; Zhang et al., 2004), and lipid catabolic processes, including the genes Lpl, Lipa, and Abcg1, indicative of their essential role in surfactant catabolism (Eckert et al., 1983; Baker et al., 2010). Tissue remodeling was also identified as a top pathway, including Ctsk, a cysteine protease important for the maintenance of airway structural integrity (Zhang et al., 2011), Serpine1, a serine protease that can either inhibit (Wu et al., 2015) or promote angiogenesis (Bajou et al., 2014), and Il18, a pro-inflammatory cytokine that drives both pulmonary vascular and parenchymal remodeling (Rodriguez-Menocal et al., 2014; Kang et al., 2012). In situ imaging of Itgax+ Car4+ cells demonstrated that they move from the distal lung parenchyma at P1 to the alveolar lumen by P7 (Figure 4D), confirming their identification as alveolar macrophages (AMs).

Mac IV uniquely expressed numerous proinflammatory genes (Pandya and Wilkes, 2014; Figure 4A), in contrast to the balanced inflammatory signature of Mac II and AMs. Top enriched pathways included leukocyte chemotaxis, inflammatory response, cytokine production and regulation of SMC proliferation (Figure 4—figure supplement 1). These included genes in the classical complement pathway (C1qa, C1qb, C1qc, C3ar1) and CCR2 ligand Ccl12 (Table 4), suggesting a role in the localization of CCR2 expressing monocytes (Griffith et al., 2014). The Mac IV cells also expressed Cxcl16, an IFNγ regulated chemokine that promotes T cell recruitment through CXCR6 (Matsumura et al., 2008). In addition to these immunomodulatory functions, the Mav IV also expressed factors with established roles in lung parenchymal and vascular homeostasis. These included Igf1, a potent mitogen and survival factor for vascular cells, recently shown to preserve lung growth and prevent pathologic vascular remodeling in an experimental model of BPD (Seedorf et al., 2020), Ang, a potent vasoconstrictor, Pf4, an angiostatic factor that inhibits EC proliferation (Maione et al., 1990), and ApoE, a lipoprotein that modulates endothelial NO production (Sacre et al., 2003). Further, Mac IV expressed secreted factors important from lung epithelial progenitors distinct from those expressed by Mac II, including Fgf10, a growth factor required for early lung branching morphogenesis that also serves to promote epithelial progenitor cell proliferation and mesenchymal cell differentiation (Sekine et al., 1999; Ramasamy et al., 2007), Bmp4, and Wnt2 and Wnt5a (Figure 4—figure supplement 2). Mac IV highly expressed Mrc1 (CD206) (Figure 4—figure supplement 3), but within the Mac IV cluster there were a group of cells with lower Mrc1 and high MHC II gene expression (H2-Ab1, H2-Eb1, and Cd74). These transcriptional differences within Mac IV are similar to two recently reported interstitial macrophages (IMs) in adult lung that can be differentiated by expression of Mrc1 and MHC II genes (Schyns et al., 2019). However, the other genes reported to distinguish those two clusters (e.g Cx3cr1, Mertk, Cclr2, Cd68) (Figure 4—figure supplement 3) were diffusely expressed throughout the Mac IV cluster. In situ imaging to localize C1qa+Cd68+ cells in the postnatal lung found Mac IV cells remained abutting small blood vessels as well as the abluminal side of large airways (Figure 4E). The characteristic location and the expression of numerous genes important for leukocyte recruitment and pathogen defense suggest that Mac IV may serve as patrollers, playing a key role in immune-surveillance, innate pathogen defense, and antigen presentation, and well as informing lung vascular and airway structure or remodeling.

The transient presence of the Mac II cells, and significant overlap with the transcriptomes of Mac I, and III, suggested that Mac II may represent an intermediate population. To determine if cells within these clusters were undergoing gradual transcriptional shifts across time, we performed pseudotime analysis on cells from clusters Mac I, II, and III. Ordering of the cells within the Mac clusters I-III by pseudotime (Figure 4F) defined a global trajectory from the Mac I to the Mac III cluster, indicating a correspondence between pseudo- and real time during perinatal development. This global trajectory notwithstanding, there were multiple local attractor states within the Mac II cluster, suggesting that cells gradually shift from Mac I to Mac II, and subsequently remain in that phenotypic state for some time before further committing to a specific fate. We also determined RNA velocity (La Manno et al., 2018) and found that the ratio of spliced to unspliced transcripts increased from the Mac I to the Mac II and III clusters along the embedding, with a clear downwards flow suggestive of a continuous development (Figure 4G). Of note, in some locations, (e.g. on the left side of Mac I), the arrows seem to exit the clusters without a sink, consistent with metastable local cell states observed in the pseudotime analysis. Taken together, our data suggest that the Mac II cells derive from Mac I and represents a transitional population that may serve a temporal-specific function before later differentiating into Mac III cells and potentially the antigen-presenting subset of Mac IV cells. However, lineage tracing studies employing a Gal-promoter driven fluorescent reporter will be required to firmly address the fate of the Mac I cells in the postnatal lung.

Mac V monocytes are characterized by developmental gene expression gradients

The Mac V cluster was characterized by expression of Plac8 (Galaviz-Hernandez et al., 2003; Figure 1E), a gene expressed by a recently identified population of ‘CD64+ CD16.2+ non-classical monocytes’ in the adult mouse lung (Schyns et al., 2019), suggesting a monocytic phenotype. Top enriched pathways included cytokine production, phagocytosis, response to virus, and both positive and negative regulation of the immune response (Figure 4—figure supplement 1). Consistent with these results, the Mac V monocytes expressed a panel of unique transcripts induced by type I interferon (IFN), important for modulating host responses to viral pathogens (Ifitm2, Ifitm3, Ifitm6, Ifi27l2a) (Zhao et al., 2018; Figure 5A), as well as genes that constrain inflammation (Cd300a, Nr4a1, Lst1) (Figure 4A and Table 5). This dual immune signature was similar to that seen in Mac II and III, emphasizing the importance of a finely tuned inflammatory response in the developing lung.

The Mac V cluster harbors distinct subpopulations with developmentally regulated gene expression patterns.

(A) t-SNE plots demonstrating high expression of Plac8 and Ifitm6 in Mac V cells. (B) t-SNE plots of Ccr1, Ccr2, Ace, and Slc12a2 suggesting the presence of two transcriptionally distinct populations within the Mac V cluster. (C) t-SNE demonstrating a developmental gradient within the Mac V sub cluster. For all t-SNE embeddings, the color scale is a logarithmic scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell. (D) Pseudotime analysis with the star indicating the starting point and the arrow denoting the trajectory across pseudotime. (E) Heatmap of differentially expressed genes within enriched pathways across pseudotime. The color scale is logarithmic with a pseudocount of 0.1 counts per million, normalized to the highest expressing population for each gene, depicting 0 (none), 33 (low), 67 (mid), and 100% (high) expression compared to the highest expressing population. (F) In situ hybridization of Cd68 (white), Ifitm6 (red), and Ccr2 (green), and to detect the ‘early’ Mac V subcluster at P1. (G) In situ hybridization of Cd68 (white), Lair1 (red), and Ifitm6 (green), to detect the ‘late’ Mac V subcluster at P21. Calibration bar = 20 μm for all panels.

Within Mac V we found a gradient of cell states between two distinct phenotypes, with some genes (e.g. Ccr1 and Ccr2) expressed at one end and other genes (e.g. Ace and Slc12a) expressed at the other end of the spectrum (Figure 5B), with a clear correspondence to real developmental time (Figure 5C). Pseudotime analysis indicated that cells with an early phenotype upregulated Ly6C (Yona and Jung, 2010), a gene expressed by classical monocytes, and genes associated with the cytoskeleton (Actb, Vim) (Rubenstein, 1990; Capote and Maccioni, 1998), matrix remodeling (Fn1, F13a1, Vcan) (Wight, 2017; Peters et al., 1996), and reduction-oxidation (Sod1, Prdx6) (Tsang et al., 2018; Fisher, 2011) in keeping with the marked physiological changes and rapid remodeling occurring in the lung during the fetal-neonatal transition (Figure 5D and E). Over both pseudo- and real time, this gene expression pattern evolved into an immunomodulatory signature, reflected by the up-regulation of genes associated with macrophage differentiation (Csf1, Spn) (Rojo et al., 2019), macrophage polarization (Tgm2, Cd36) (Rőszer, 2015), negative regulation of inflammation (Ceacam1, Ear2, Lair1) (Dankner et al., 2017), and innate immunity (Cd300ld, Treml4) (Borrego, 2013). These data indicate that only the late Mac V cells are similar to the CD64+ CD16.2+ non-classical monocytes reported by Schyns et al., 2019., while the early Mac V cells represent an additional source of heterogeneity unique to the perinatal lung.

Spatial localization of the early and late populations by detecting either Ccr2 or Lair1 in combination with Ifitm6 and Cd68 allowed the identification of early Mac V cells within the distal lung parenchyma (Figure 5F, Figure 4—figure supplement 4). Of note, these Cd68+ Ifitm6+ Ccr2+ cells were reliably found either along secondary crests or lining the developing alveoli such that one boundary of the cell was always in contact with the airspace. Despite significant changes in the gene expression, Cd68+ Ifitm6+Lair1+ were similarly found in the distal lung along the alveolar walls (Figure 5G). Taken together, these data demonstrate that within the Mac V monocytic phenotype there are functionally distinct, developmentally dynamic cell states that transition during and after birth from tissue remodeling and regulation of immune cell chemotaxis to immunomodulation and pathogen defense.

Comparison of mac I-V populations with adult murine lung data sets

Next, we determined whether our macrophage populations are similar to recently defined macrophages in the adult lung by directly comparing our data with the macrophage populations annotated by the authors of two previous reports. In comparison to Schyns et al., 2019., despite the use of different technologies, we identified clear shared patterns (Figure 6A–B). First, our Mac IV cluster embedded close to their CD206+ and CD206- macrophages, and all three express C1qa, confirming their identity as interstitial macrophages (IMs). Second, the proportion of IMs expressing H2-Eb1 is higher in their dataset. Third their monocyte cluster is located closer to our Mac V and both express Plac8, indicating a similar cell type. While a few of their cells embedded completely within our Mac V, the majority are in a distinct, yet nearby cluster, possibly reflecting developmental differences. Likewise, although our Mac III and their AMs express Car4 and cluster near one another, they remain in separate clusters, perhaps reflecting either technical or biologic differences. Comparison with Tabula Muris (Tabula Muris Consortium et al., 2018) yielded similar results (Figure 6C–D), with the Mac IV cells embedding with the cells annotated as macrophages, Mac V embedding with cells annotated as monocytes, and Mac I-III clustering separately. Taken together, these data suggest that Mac IV and the late Mac V are transcriptionally similar to adult lung interstitial macrophages and tissue monocytes, respectively, Mac III are AM with transcriptomic alterations specific to development, and Mac I and II are macrophage populations unique to the perinatal lung.

Comparison of Macrophage/Monocyte Subtypes in the Perinatal and Adult Lung.

Data sets were merged, restricted to genes that were overlapping, normalized to counts per million reads, feature selected and subjected to PCA, followed by UMAP. UMAP plots of (A) cell type and (B) data source using our data combined with Schyns et al. Data sets were merged and processed as described above, and the Northstar algorithm was used to infer the cell subtypes for our cells based on the Tabula Muris Atlas, with UMAP plots of (C) cell type and (D) data source.

Variations in macrophage fc receptor expression

Fc receptors serve as an important link between cellular and humoral immunity by bridging antibody specificity to effector cells (Hazenbos et al., 1996) and are therefore a key axis of functional heterogeneity within macrophages and monocytes. We evaluated their expression across Mac I-V and found specific patterns (Figure 7). Both Fcgr3 (encoding FcγRIII) and Fcer1g (FcεRIγ) were widely expressed by all five clusters, while Fcer1a and Fcer2a were not expressed by any cluster. Expression of Fcgr1, a high affinity receptor for IgG important for the endocytosis of soluble IgG, phagocytosis of immune complexes, and delivery of immune complexes to APC (Barnes et al., 2002) was highly expressed by Mac I and Mac IV, and in the early subcluster of Mac V. In contrast, Fcgr4, encoding an Fc receptor able to bind IgE that promotes allergic lung inflammation was highly expressed by the late sub-cluster of Mac V, in agreement with data from adult mice (Schyns et al., 2019; Mancardi et al., 2008; Hirano et al., 2007). Fcgrt, encoding the neonatal Fc receptor (FcRn), an Fc receptor with a key role in IgG recycling, was highest in Mac IV cells.

Expression of Fc receptors across macrophage clusters.

t-SNE plots of the expression of Fc receptors Fcgr1, Fcgr2b, Fcgr3, Fcgr4, Fcer1a, Fcer1g, Fcer2a, Fcgrt. For all tSNE embeddings, the color scale is a logarithmic scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell.

Dendritic cell subtypes and granulocytes are primed for rapid pathogen response

Dendritic cells (DCs) play multiple roles in the immune system including antigen presentation and regulation of tolerance and can be distinguished from other mononuclear phagocytes by the expression of Zbtb46 (Meredith et al., 2012) and Flt3 (Waskow et al., 2008). We found three clusters of cells (Figure 8A and Tables 68) expressing Zbtb46 and Flt3 (Figure 8—figure supplement 1). Classical DC1 (cDC1) cells expressed Itgae or CD103 (Figure 8B), which promotes antiviral immunity and may confer the ability for antigen cross-presentation (Helft et al., 2012), and a number of additional genes characteristic of cDC1 including Irf8, Xcr1, and Cadm1 (Figure 8—figure supplement 1). The next DC cluster expressed many genes described to be specific for classical DC2 cells, including Itgam, Sirpa, and Irf4 (Figure 8—figure supplement 1). Although cDC2 expressed Cd209a or DC-SIGN, a gene found in monocyte-derived inflammatory DC exposed to lipopolysaccharide (Cheong et al., 2010), and low levels of Fcer1g, other genes expressed by monocyte-derived DC, including Fcgr1, Ly6c1, and Ly6c2 were practically absent (Figure 8—figure supplement 1). Given that cDC2 cells can also express Cd209a, the identity of these cells was most consistent with cDC2. The final DC cluster expressed melanoregulin (Mreg), a modulator of lysosomal hydrolase maturation (Damek-Poprawa et al., 2009; Figure 8B). Mreg had not been previously identified as a marker for DC subsets, hence we examined other genes expressed by this cluster and identified Cacnb3, a voltage dependent Ca2+ channel found in stimulated Langerhans cells (Bros et al., 2011); Fscn1, which contributes to dendrite formation in maturing DC (Al-Alwan et al., 2001); Ccl5, an important chemoattractant for DC and T cells; and Ccr7, a chemokine receptor associated with trafficking to the draining lymph node (Förster et al., 2008), therefore we annotated this cluster as migratory (mig-DC) (Figure 8—figure supplement 1 and Tables 68). Quantification of the relative abundances identified cDC1 as the predominant population in the embryonic lung. cDC1 persisted postnatally to comprise between 1–2% of total lung immune cells (Figure 7C). cDC2 was present at low frequency during the first week of life and increased in abundance by P21. The mig-DC were undetectable before birth and became more abundant postnatally. Taken together, these data suggest cDC1and cDC2 are classical DC, and the mig-DC cluster a form of migratory mature DCs.

Figure 8 with 1 supplement see all
Multiple dendritic cell populations and lung granulocytes are primed for rapid pathogen response.

(A) Colored schematic and lung immune cell clustering demonstrating three separate clusters of DCs. (B) t-SNE plots of genes discriminating the three DC subclusters including Itgae (cDC1), Cd209a (cDC2) and Mreg (mig-DC). (C) Quantification of specific DC subpopulations relative to total immune cells (left) or total DC (right). (D) Colored schematic and lung immune cell clustering demonstrating the basophil, mast cell and neutrophil clusters, with high magnification of basophil and mast cell clusters and t-SNE plots of Mcpt4 and Mcpt8. (E) t-SNE plots of Il6, Hgf, Ccl4, and Osm in the basophil cluster. (F) High magnification of the neutrophil cluster with t-SNE plots of neutrophil specific genes S100a8, S100a9, Stfa1, and Stfa2. For all t-SNE embeddings, the color scale is a logarithmic scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell.

Table 6
Top 20 differentially expressed genes in the Mig-DC vs cDC2 cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 20 differentially expressed genes are shown for the migDC vs cDC2 cluster; n = 2 mice.

Mig-DC versus cDC2 cluster top 20 differentially expressed genes
Gene nameStatisticP. valueLog2 fold change
Ccl51.00003.45E-1316.9849
Il12b0.78576.97E-0715.3906
Fscn11.00003.45E-1314.5028
Nudt170.92862.48E-1013.7990
Zmynd150.92862.48E-1013.5523
Il15ra0.92862.48E-1013.5174
Cacnb31.00003.45E-1313.4891
Cd63-ps1.00003.45E-1313.4587
H2-M20.85712.06E-0813.4381
Tnfrsf40.71431.31E-0513.4073
Serpinb6b0.92862.48E-1013.2991
AW1120100.85712.06E-0813.1876
Laptm4b0.85712.06E-0812.9752
Lad10.71431.31E-0512.8948
Cd631.00003.45E-1312.8680
Dnase1l30.71431.31E-0512.7389
Apol7c0.50007.75E-0312.7302
Tm4sf50.50007.75E-0312.6086
Spint20.92862.48E-1012.2644
Zfp8720.64291.57E-0411.4802
Table 7
Top 20 differentially expressed genes in the Mig-DC vs cDC1 cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 20 differentially expressed genes are shown for the mig-DC vs cDC1 cluster; n = 2 mice.

Mig-DC versus cDC1 cluster top 20 differentially expressed genes
Gene nameStatisticP. valueLog2 fold change
H2-M20.85712.34E-0914.0574
Tnfrsf40.71433.71E-0613.6282
Zfp8720.64295.96E-0511.348
Arc0.71433.71E-0610.9369
Gbp80.64295.96E-0510.6692
Ramp30.54.51E-0310.6263
Insl60.64295.96E-0510.5787
Gbp40.92861.19E-1110.4833
Adcy60.82012.13E-0810.4176
Atxn10.71433.71E-0610.3352
Ankrd33b0.92861.19E-1110.1242
Dscaml10.64295.96E-059.8945
Src0.54.51E-039.7618
Tmem150c0.64295.96E-059.566
Slc4a80.71164.01E-069.5518
Nudt170.91016.10E-119.5137
Il21r0.69587.81E-069.3355
Fas0.62431.12E-049.3023
Clec2i0.54.51E-039.2653
Cacnb30.98152.44E-138.6751
Table 8
Top 20 differentially expressed genes in the cDC2 vs cDC1 cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 20 differentially expressed genes are shown for the cDC2 vs cDC1 cluster; n = 2 mice.

cDC2 versus cDC1 cluster top 20 differentially expressed genes
Gene nameStatisticP. valueLog2 fold change
Mgl20.78575.55E-1512.5236
Cd209d0.54764.76E-0712.1889
Cd209b0.33337.77E-0312.1445
Ms4a6d0.57141.14E-0711.7772
Ms4a4c0.54764.76E-0711.2462
Plaur0.38101.42E-0311.0707
Casp40.38101.42E-0310.0788
Il1rl10.61905.17E-0910.0357
Cd300lg0.33337.77E-039.9463
Cd330.47622.21E-059.7976
Emilin20.40485.52E-049.5336
Il1rl20.38101.42E-039.5332
Creb50.39159.39E-049.2747
Cd209c0.47622.21E-059.0161
Cd209a0.95773.33E-169.0160
Ccl90.67727.43E-119.0010
Fcrls0.38101.42E-038.9226
Il18rap0.39159.39E-048.7118
Wfdc170.72491.46E-128.5219
Tent5a0.65344.52E-108.2999

Mast cells and basophils are similar in development and function and serve as fast responders to specific immune challenges (Voehringer, 2013). Two immune clusters highly expressed Cpa3 that could be further distinguished as mast cells and basophils by the expression of Mcpt4 and Mcpt8, respectively (Lilla et al., 2011; Immunological Genome Project Consortium et al., 2016; Figure 8D). Mast cells expressed Tpsb2, which is secreted upon bacterial challenge (Thakurdas et al., 2007), and the peptidases chymase (Cma1) and tryptase (Tspab1) (Table 9). Lung resident basophils express a unique signature distinct from circulating basophils and play a key role in promoting AM differentiation (Cohen et al., 2018). Our basophils generally shared expression of many genes with lung resident basophils, including Il6, Hgf, Ccl4 and Osm (Figure 8E, Table 10). We also identified a neutrophil cluster distinguished by expression of S100a8 and S100a9, which are released during inflammation (Wang et al., 2018), and Stfa1 and Stfa2, cysteine proteinase inhibitors important for antigen presentation (Hsieh et al., 2002; Figure 8F and Table 11). Our data agree with prior work demonstrating that lung resident basophils express signaling molecules important for interaction with neighboring cells (Cohen et al., 2018), and that mast cells and neutrophils are primed for rapid innate immune responses upon pathogen challenge.

Table 9
Top 25 differentially expressed genes in the mast cell cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 25 upregulated genes are shown for the mast cell cluster; n = 2 mice.

Mast cell top 25 differentially expressed genes
Gene nameStatisticP. valueLog2 fold change
Tpsb21.00008.9E-1618.7534
Mcpt40.87502.4E-0717.4253
Tpsab10.87502.4E-0715.5899
Cma11.00008.9E-1615.0710
Slc45a30.75005.2E-0511.7029
Rprm0.87502.4E-0711.6969
Hs6st20.75005.2E-0511.1665
Maob0.86833.4E-0710.7834
Cpa30.99001.8E-1310.6394
Ednra0.75005.2E-0510.2131
Gata10.75005.2E-0510.0639
Kit1.00008.9E-169.2771
Mlph0.74675.9E-058.5668
Cma20.75005.2E-058.5455
Cyp11a10.99334.9E-148.4308
Tph10.87172.8E-078.3091
Rab27b0.87172.8E-078.1436
Il1rl10.97331.4E-117.7399
Itga2b0.83501.8E-067.6658
Gata20.99334.9E-147.5597
Poln0.73678.2E-057.2024
Slc18a20.96001.4E-106.8570
Stard130.74675.9E-056.7772
Adora30.70332.3E-046.4180
Kcne30.80506.6E-066.4033
Table 10
Top 25 differentially expressed genes in the basophil cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 25 upregulated genes are shown for the basophil cluster; n = 2 mice.

Basophil top 25 differentially expressed genes
Gene nameStatisticP. valueLog2 fold change
Mcpt80.73472.22E-1615.6663
Ms4a20.86762.22E-1612.3460
Gata20.90842.22E-169.7841
Il40.68722.22E-168.7980
Cd200r30.95252.22E-168.6767
Il60.83092.22E-168.5795
Ccl40.90212.22E-167.0748
Ifitm10.93632.22E-166.8128
Cyp11a10.97292.22E-166.6271
Ccl30.92252.22E-165.9632
Hgf0.73262.22E-165.7653
Csf10.89212.22E-165.4838
Ccl90.88252.22E-165.3108
Ifitm70.75592.22E-165.1284
Aqp90.75632.22E-165.0857
Rab440.65342.22E-165.0335
Hdc0.91332.22E-164.9976
Cd690.74542.22E-164.9828
Cdh10.71972.22E-164.9643
Il18r10.65392.22E-164.5920
Csf2rb20.69302.22E-164.3472
Lilr4b0.87882.22E-164.1383
Rgs10.70012.22E-164.0712
Il18rap0.75712.22E-164.0030
Osm0.65592.22E-163.6933
Table 11
Top 25 differentially expressed genes in the neutrophil cluster.

ScRNA-Seq was performed on live CD45+ lung cells isolated from E18.5, P1, P7 and P21 B6 pups. Top 25 upregulated genes are shown for the neutrophil cluster; n = 2 mice.

Neutrophil top 25 differentially expressed genes
Gene nameStatisticP. valueLog2 fold change
Retnlg0.78331.9E-9116.7841
Stfa20.90671.6E-13113.3347
BC1005300.90333.1E-13011.5407
Mmp90.88675.5E-12411.2071
Asprv10.75001.1E-8210.6669
Stfa2l10.93004.5E-1418.5427
S100a90.95672.7E-1538.3546
2010005H15Rik0.84672.9E-1108.1305
BC1170900.88001.4E-1217.8635
Slpi0.87672.1E-1207.5069
Stfa10.89671.1E-1276.8526
Il1r20.71338.4E-746.6465
S100a80.94671.6E-1486.2103
Slc7a110.69332.9E-695.9640
Csf3r0.74003.5E-805.7411
Cxcr20.74335.2E-815.6599
Wfdc210.77339.5E-895.3370
Pglyrp10.83674.4E-1074.9416
Mxd10.68002.4E-664.5560
Hdc0.82675.5E-1044.2867
Grina0.70331.7E-714.0795
Il1b0.65671.9E-613.8434
Thbs10.67336.5E-652.7913
S100a110.76003.0E-852.6263
Srgn0.66004.0E-622.3914

Naïve lymphocytes populate the lung at birth

Lymphocytes including ILC2s (expressing Areg), NK cells (Gzma), B cells (Ms4a1) and T cells (Cde3) were present in the lung at low frequencies (approximately 2% of immune cells) prior to birth and increased in frequency after birth. By P21, lymphocytes comprised 60% of total immune cells (Figure 1F and G), with B cells representing 30% and T cells 15% of the total (Figure 9A).

Figure 9 with 1 supplement see all
Lymphocytes populate the lung at birth but remain naïve during the first three weeks of life (A) Quantification of the abundance of B and T cells at each developmental timepoint expressed on a log (Rodríguez-Castillo et al., 2018) scale as percentage of total immune cells.

(B) Quantification of the percentage of proliferating B cells at each developmental timepoint. (C) B cell heavy and light variable (V) and joining (J) gene identity and their cumulative cell fraction. (D) t-SNE plot of Ighm and Ighd, with the logarithmic color scale with a pseudocount of 0.1 counts per million, normalized to the highest expressing cell. (E–H) At P1 (n = 7 mice) and P21 (n = 8 mice), lungs were processed to a single-cell suspension, and flow cytometry was used to assess frequencies of (F) CD4+CD3+ and CD8+CD3+ T cells, (G) Gata3+, Tbet+, Rorγt+ and Foxp3+ CD4+ T cells, and (H) IL-4-producing Gata3+CD4+ T cells or IFNγ-producing Tbet+CD4+ T cells. Data shown as mean ± SD, **p<0.01, ****p<0.00001 by Student’s t test.

B cells in the embryonic lung were rare and the majority expressed proliferation markers. After birth, B cell abundance increased, but the proliferating fraction decreased (Figure 9B). The cumulative distribution of V and J loci germline similarity revealed no somatic hypermutation (Figure 9C) and most B cells expressed Ighm and Ighd indicative of an IgM isotype (Figure 9D). Few B cells expressed activation markers (e.g. Aicda, Tbx21, Prdm1, and Ebi3, Figure 9—figure supplement 1Wang et al., 2012; Pone et al., 2012; Ma et al., 2019; Cattoretti et al., 2005), suggesting that most postnatal B cells remain naïve through late alveolarization. To test the clonality of the B cell repertoire at birth, we assembled the heavy and light chain loci and performed t-SNE on a feature-selected transcriptome limited to over-dispersed genes in B cells (Figure 9—figure supplement 1). Proliferating cells clustered together but candidate clonal families did not, suggesting primarily homeostatic B cell proliferation rather than clonal expansion (van Zelm et al., 2007).

The majority of T cells expressed Trac, suggesting αβ identity, with a few Trac- cells expressed Tcrg-C4, suggesting γδ T cell identity (Figure 9—figure supplement 1). T cell receptor diversity showed no sign of clonal expansion (data not shown). Outside the thymus, αβ T cells are usually either CD4+ (with subgroups Tbet+ or T helper (Th) 1, Gata3+ or Th2, and FoxP3+ or Th17) or CD8+, however we characterized T cell heterogeneity in neonatal lungs by flow cytometry and found that 85–90% of CD3+ cells were CD4- CD8- at both P1 and P21 (Figure 9E), confirming an earlier report suggesting this is a neonatal-specific phenotype. mRNA expression analysis qualitatively confirmed this finding (Figure 9—figure supplement 1). The frequency of total T cells was similar at P1 and P21 (Figure 9—figure supplement 1), however several T cell subsets (CD8+, CD4+ Th1, and Treg cells) increased by P21 while other subsets remained constant (Figure 9F and G). In response to stimulation with phorbol myristate acetate (PMA) and ionomycin, a greater number of CD4+ Th2 cells at P21 produced IL-4 as compared to P1 (28.3 ± 19.7 vs. 8.6 ± 5.7, p=0.0047) (Figure 8H). Few CD4+ Th1, Treg, and Th17 cells produced IFNγ, IL-10, IL-17 upon stimulation, at both P1 and P21 (Figure 9H, Figure 9—figure supplement 1).

Discussion

At birth, the lung undergoes marked physiological changes as it transitions from a fluid-filled, hypoxic environment to an air-filled, oxygen-rich environment. How these changes affect immune populations during this transition and the ensuing period of rapid postnatal lung growth remains unclear. Our study demonstrates a rapid increase in immune cell heterogeneity, especially within macrophages and monocytes. We identified five macrophage/monocyte subpopulations, each expressing a specific gene signature, spatial localization, and putative functions. Mac I, the predominant immune cell present just before birth, were highly proliferative, and completely encircled small blood vessels, suggesting a previously unrecognized role for lung macrophages in modulating lung vascular growth or remodeling during development. During the first week of life, a transitory population (Mac II) emerged from Mac I and later disappeared, transitioning into either an alveolar (Mac III) or Mac IV macrophage phenotype. One interstitial macrophage population (Mac IV) expressed complement proteins and other antibacterial molecules. Another interstitial population (Mac V) expressed antiviral molecules and spanned a gradient between two extreme phenotypes, one that expressed high levels of homeostatic genes during early postnatal development, and a second with immunomodulatory functions that resembles previously reported nonclassical monocytes. Lymphocytes increased in abundance from almost zero before birth to more than half of lung immune cells by P21, but maintained a naive phenotype skewed toward type II immunity and with predominantly Cd4- Cd8- T cells.

This comprehensive study has far-reaching implications for lung biology. Resident tissue macrophage populations are established during development, wherein progenitors undergo differentiation guided by the tissue-specific microenvironment (Sommerfeld et al., 2019). However, definitive data regarding the full complexity of lung resident macrophages and monocytes, their specific roles and functions, and how they change across development remain elusive. Although the advent of single cell transcriptomics has provided increased resolution to detect previously unrecognized immune cell populations, consensus regarding the diversity and function of lung resident macrophages has not been achieved. Cohen et al., 2018 recently performed single cell transcriptomics of the developing mouse lung from E12.5 until P7, and identified a total of three macrophage populations, and one population of resident monocytes, with alveolar macrophages representing the sole macrophage population present in the lung after P7. In contrast, Schyns et al. identified two distinct interstitial macrophages in the adult lung, and a population of nonclassical monocytes in addition to alveolar macrophages (Schyns et al., 2019). Although our results are more consistent with the report of Schyns et al, there are a number of key differences. First, the total heterogeneity in the perinatal lung far exceeds the adult lung, with the presence of two unique macrophage clusters (Mac I and Mac II, and alveolar macrophage (Mac III) and monocyte derived clusters (Mac V) that are transcriptionally distinct from their adult counterparts. Second, both the Mac IV and Mac V cluster harbor significant internal heterogeneity (in the case of Mac V, corresponding to developmental time) that cannot be easily split into transcriptionally distinct ‘subclusters’. The cells within Mac IV appear similar to the CD206+ and CD206- macrophage populations reported by Schyns et al. Although in the Schyns report those populations were reported to be distinct clusters, there was significant overlap in gene expression between the two, more consistent with our data suggesting these are not separate populations but rather a phenotypic continuum. In situ validation confirmed the presence of all five subpopulations and localized each to defined locations in the lung including the alveolar lumen, around vessels and airways, or within the distal lung parenchyma. A greater understanding of macrophage and monocyte function at birth provides an essential framework for interpreting how lung injury and developmental defects alter specific immune subpopulations and eventually influence lung growth and development.

Another key finding in our study was the unexpected presence of embryonic lung macrophages encircling small blood vessels. Vascular growth is a key driver of distal lung growth during the late saccular and alveolar stages of development (Thébaud and Abman, 2007). Macrophages support angiogenesis in other organs, promoting blood vessel formation or expansion, providing survival and migratory cues to endothelial cells, and facilitating bridging of vascular sprouts (Baer et al., 2013). In the developing hindbrain, macrophages are in close contact with endothelial cells, serving to promote vascular anastomosis (Fantin et al., 2010). Similarly, in the developing retinal vasculature, microglia connect adjacent endothelial tips cells to increase vascular plexus complexity (Rymo et al., 2011). These embryonic bridging macrophages secrete numerous genes shared by the perivascular macrophages that drive tumor angiogenesis including the angiopoietin receptor, Tek, the VEGF co-receptor Nrp1, growth factors (Fgf2, Pgf) and MMPs (Mmp2, Mmp9). Although the perivascular macrophages we observed in the embryonic lung expressed low levels of Nrp1, they appear distinct from the macrophages that influence retinal and hindbrain angiogenesis, expressing a unique set of ECM remodeling and angiogenic genes, including genes that may modulate vascular tone and permeability. Whether the macrophages form non-contiguous rings around the vessel, or sheets that completely encircle the full length of the vessel is not clear and would require antibody-based immunofluorescence with deep imaging and reconstruction of thick tissue sections. The distinctive location of these macrophages and their gene signature imply a role in vascular development. Furthermore, these encircling macrophages disappeared after birth, suggesting a function temporally restricted to prenatal development. Future studies to selectively target this subpopulation will be required to further establish their function and to delineate the signals responsible for the cessation of the macrophage-vascular interaction after birth.

Our data also revealed three distinct dendritic cell populations with transcriptional signatures indicative of classical and migratory dendritic cell phenotypes, respectively. Moreover, we identified distinct (Itgae and Cd209a) and novel (Mreg) markers superior to classical markers (Zbtb46 and Flt3) to distinguish dendritic subpopulations in the postnatal lung. Functionally, both Itgae and Cd209a/DC-SIGN can induce T cell immunity (Soilleux et al., 2002; Merad et al., 2013), suggesting that lung DCs immediately post-birth should be able to cause an effective adaptive immune response. However, despite the apparent signaling readiness of antigen-presenting DCs, T and B cell compartments showed an overall naive and rarely proliferative phenotype, lacking any clonal structure and with most αβ T cells double negative (DN) at the protein level for both CD4 and CD8, key signaling components for cell-mediated immunity. Though consistent with prior evidence showing a high proportion of pulmonary lymphocytes with unconventional phenotypes (Augustin et al., 1989), the observation of widespread DN T cells points to a yet undetermined function, perhaps related to specific immunoregulatory functions during infectious disease (Zhang et al., 2000).

The predominant stimuli that drive the increase in lung immune diversity after birth are not clear. Multiple, marked physiologic changes occur at birth including an abrupt change from hypoxia to normoxia, establishment of an air-liquid interface, cyclic stretch of the lung parenchyma, and increased shear stress in the pulmonary vasculature from the dramatic increase in pulmonary blood flow. These alterations in force and oxygen tension may influence the precise and coordinated process of lung immune cell differentiation or phenotype. For example, continuous compressive force is an essential driver of osteoclast differentiation (Sanuki et al., 2010). Cyclic stretch modulates alveolar macrophage phenotype in vitro, potentially increasing (Pugin et al., 1998) or suppressing inflammatory cytokine production (Maruyama et al., 2019). Reactive oxygen species influence the expansion, maturation, and antigen uptake and processing in DCs (Sheng et al., 2010), while hypoxia promotes macrophage differentiation (Grodzki et al., 2013) and influences macrophage polarization, resulting in alterations in inflammation, metabolism and angiogenesis (Escribese et al., 2012). However, how an abrupt transition from hypoxia to normoxia affects the phenotype of immature macrophages or myeloid progenitors has not been studied. Alternatively, differences in the lung microenvironment resulting from specific signaling pathways activated in non-immune cells during the distinct stages of development (e.g. saccular versus alveolar) may represent an additional driver of immune cell phenotype. Comparison of immune subsets in the developing mouse and human lung may provide additional clarity as to the central factors regulating diversity, given that the mouse is in the saccular stage and the human in the alveolar stage immediately prior to birth.

In addition, whether alterations in immune cell diversity may underlie the origin of pediatric lung disease remains to be determined. Many of the Mac clusters express genes with seemingly contrasting functions, highlighting a key role for these populations in the fine-tuned regulation of complex biological process such as inflammation and angiogenesis. Stimuli that disrupt this careful balance, might have profound pathologic effects and contribute to diseases such as BPD, pulmonary hypertension and asthma. For example, in the mouse, inflammatory activation of embryonic macrophages disrupts lung branching (Blackwell et al., 2011), and abnormal activation of macrophages is observed in premature infants that develop BPD (for the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network et al., 2009). The detrimental effects of prenatal macrophage activation may result both from the induction of injurious cytokines and chemokines, but also the loss of beneficial homeostatic functions. Further, alterations in the changes in specific macrophage abundance at key points in development may represent an additional origin of specific lung diseases. For example, the transient Mac II population expresses molecules that influence lung and vessel remodeling, including Tgm2, a hypoxia induced protein-crosslinker that promotes vascular smooth muscle proliferation and vascular remodeling in experimental PH (Penumatsa et al., 2017). Whether an inappropriate persistence of the Mac II cells would be associated with pathologic lung and vascular remodeling is not known. Similarly, the late Mac V subpopulation expresses Fcgr4, encoding an Fc receptor capable of binding IgE, suggesting a potential role in modulating airway inflammation and response to inhaled antigen. The alveolar macrophages (Mac III) rapidly increase from E18.5 to P21. Given that resident alveolar macrophages are anti-inflammatory and mitigate the airway hyperresponsiveness that characterizes asthma, a perinatal insult that decreases either Mac V or Mac III abundance could possess long term, pro-inflammatory effects. Further investigation into the effect of injurious stimuli on the phenotype, abundance and localization of these discrete immune cell populations is likely to provide deeper insight into the mechanisms underlying pediatric and adult lung disease.

In summary, these data highlight the marked increase in immune cell diversity after birth, with a developmental plasticity that provides distinct immune populations to fill specific roles in tissue and vascular remodeling, immunoregulation, and bacterial and viral pathogen defense. Injuries to the developing, immature lung can have profoundly untoward and life-long consequences as a significant component of lung parenchymal and vascular development occurs during late pregnancy and the first few years of postnatal life. Many of these injurious stimuli including acute infection, hyperoxia, and corticosteroids are known to have significant effects on immune cell phenotype and function. Therefore, our data provide a detailed framework that enables a more complete understanding of how disruptions of immune cell phenotype may contribute to altered lung development, both through the induction of pathologic, pro-inflammatory signaling as well as the suppression of essential homeostatic functions. Further, a deep understanding of the diversity of immune cell functions during this important window of postnatal development, and how specific immune cell phenotypes are regulated could allow for the application of immunomodulatory therapies as a novel strategy to preserve or enhance lung development in infants and young children.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Strain, strain background (M. musculus)C57BL/6JCharles Rivers Laboratories
Antibodyanti-CD3e (Hamster monoclonal)BD BiosciencesCat# 553062; RRID:AB_394595FACS (1:100)
Antibodyanti-CD4 (Rat monoclonal)BD BiosciencesCat# 552051; RRID:AB_394331FACS (1:100)
Antibodyanti- CD8a (Rat monoclonal)BioLegendCat# 100721; RRID:AB_312760FACS (1:100)
Antibodyanti-Tbet (Mouse monoclonal)BioLegendCat# 644808; RRID:AB_15955479FACS (1:100)
Antibodyanti-GATA3 (Mouse monoclonal)BD BiosciencesCat# 563510; RRID:AB_2738248FACS (1:100)
Antibodyanti-Foxp3 (Rat monoclonal)eBioscienceCat# 12-5773-82; RRID:AB_465936FACS (1:100)
Antibodyanti-Rorγt (Mouse monoclonal)BD BiosciencesCat# 562607; RRID:AB_11153137FACS (1:100)
Antibodyanti-IFNγ (Rat monoclonal)BioLegendCat# 505829; RRID:AB_10897937FACS (1:100)
Antibodyanti-IL-4 (Rat monoclonal)BD BiosciencesCat# 562045; RRID:AB_10895799FACS (1:100)
AntibodyIL-10 (Rat monoclonal)BioLegendCat# 505027; RRID:AB_2561522FACS (1:100)
AntibodyIL-17 (Rat monoclonal)Miltenyi BiotecCat# 130-095-732; RRID:AB_10828821FACS (1:100)
Commercial assay or kitFISH probe: anti-mouse Plac8Advanced Cell Diagnostics (ADC)Cat# 532701RNAscope: 50 μl per slide
Commercial assay or kitFISH probe: anti-mouse Dab2ADCCat# 558131-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse Cd68ADCCat# 316611-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-Cd68ADCCat# 316611-C3RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-Cd68ADCCat# 316611RNAscope: 50 μl per slide
Commercial assay or kitFISH probe: anti-mouse-Cdh5ADCCat# 312531-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-Mki67ADCCat# 416771-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-Car4ADCCat# 468421-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-GalADCCat# 400961-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-C1qaADCCat# 441221RNAscope: 50 μl per slide
Commercial assay or kitFISH probe: anti-mouse-EpcamADCCat# 418151-C2RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-ItgaxADCCat# 311501-C3RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-Ifitm6ADCCat# 511321-C3RNAscope: (1 μl per 50 μl)
Commercial assay or kitFISH probe: anti-mouse-Lair1ADCCat# 509151RNAscope: 50 μl per slide
Commercial assay or kitFISH probe: anti-mouse-Ccr2ADCCat# 501681RNAscope: 50 μl per slide

Mouse lung cell isolation

Request a detailed protocol

C57BL/6 mice were obtained from Charles River Laboratories. For studies using E18.5, P1, and P7 murine lungs, pregnant dams were purchased, and pups aged prior to lung isolation. At E18.5, dam was asphyxiated with CO2 and pups extracted. At P1, P7, and P21 pups were euthanized with euthanasia solution (Vedco Inc). Genetic sex of mice at developmental stages E18.5 and P1 was determined by performing PCR amplification of the Y chromosome gene Sry. P7 and P21 mice were sexed through identification of a pigment spot on the scrotum of male mice (Wolterink-Donselaar et al., 2009). For all timepoints, female and male mice were randomly selected for the studies. For all timepoints, except E18.5, the pulmonary circulation was perfused with ice cold heparin in 1x PBS until the circulation was cleared of blood. Lungs were minced and digested with Liberase (Sigma Aldrich) in RPMI for 15 (E18.5, P1, and P7) or 30 (P21) minutes at 37C, 200 rpm. Lungs were manually triturated and 5% fetal bovine serum (FBS) in 1x PBS was used to quench liberase solution. Red blood cells were lysed with 1x RBC lysis buffer (Invitrogen) as indicated by the manufacturer and total lung cells counted on Biorad cell counter (BioRad). Protocols for the murine studies adhered to American Physiological Society/US National Institutes of Health guidelines for humane use of animals for research and were prospectively approved by the Institutional Animal Care and Use Committee at Stanford (APLAC #19087).

Immunostaining and fluorescence-activated cell sorting (FACS) of single cells

Request a detailed protocol

Lungs were plated at 1 × 106 cells per well and stained with Fc block (CD16/32, 1:100, Tonbo Biosciences) for 30 min on ice. Cells were surface stained with the endothelial marker CD31 (1:100, clone: MEC3.1, eBiosciences), epithelial marker Epcam (1:100, clone: CD326, eBiosciences), and immune marker CD45 (1:100, clone: F11, eBiosciences) for 30 min on ice. The live/dead dye, Sytox Blue (Invitrogen), was added to cells and incubated for 3 min prior to sorting into 384-well plates (Bio-Rad Laboratories, Inc) prefilled with lysis buffer using the Sony LE-SH800 cell sorter (Sony Biotechnology Inc), a 100 μm sorting chip (Catalog number: LE-C3110) and ultra-purity mode. Single color controls were used to perform fluorescence compensation and generate sorting gates. 384-well plates containing single cells were spun down, immediately placed on dry ice and stored at −80C.

cDNA library generation using Smart-Seq2

Request a detailed protocol

Complementary DNA from sorted cells was reverse transcribed and amplified using the Smart-Seq2 protocol on 384-well plates as previously described (Zanini et al., 2018; Picelli et al., 2013). Concentration of cDNA was quantified using picogreen (Life technology corp.) to ensure adequate cDNA amplification. In preparation for library generation, cDNA was normalized to 0.4 ng/uL. Tagmentation and barcoding of cDNA was prepared using in-house Tn5 transposase and custom, double barcoded indices (Tabula Muris Consortium et al., 2018). Library fragment concentration and purity were quantified by Agilent bioanalyzer. Libraries were pooled and sequenced on Illumina NovaSeq 6000 with 2 × 100 base kits and at a depth of around 1 million read pairs per cell.

Data analysis and availability

Request a detailed protocol

Sequencing reads were mapped against the mouse genome (GRCm38) using STAR aligner (Dobin et al., 2013) and gene were counted using HTSeq (Anders et al., 2015). FZ has been the main maintainer of HTSeq for several years. To coordinate mapping and counting on Stanford’s high-performance computing cluster, snakemake was used (Koster and Rahmann, 2012). Gene expression count tables were converted into loom objects (https://linnarssonlab.org/loompy/) and cells with less than 50,000 uniquely mapped reads or less than 400 genes per cell were discarded. Of the 4199 immune cells that were filtered, an additional 147 were removed as suspected doublets. Doublets were removed in two ways. First, clusters that expressed epithelial, endothelial or mesenchymal genes were excluded. Clusters showing joint expression of mutually exclusive cell type markers (e.g. Epcam, Cdh5, Col6a2) with Cd45 were manually excluded as well. Counts for the remaining 4052 cells were normalized to counts per million reads. For t-distributed stochastic embedding (t-SNE) (van der Maaten and Hinton, 2008), 500 features were selected that had a high Fano factor in most mice, and the restricted count matrix was log-transformed with a pseudocount of 0.1 and projected onto the top 25 principal components using scikit-learn (Pedregosa et al., 2011). Unsupervised clustering was performed using Leiden (C++/Python implementation) (Traag et al., 2019). Singlet (https://github.com/iosonofabio/singlet) and for specific analyses, custom Python three scripts were used: the latter available at https://github.com/iosonofabio/lung_neonatal_immune/ (Domingo-Gonzalez et al., 2020; copy archived at https://github.com/elifesciences-publications/lung_neonatal_immune). Pathway analysis on the differentially expressed genes from the five macrophage clusters was performed on each the top 100 most differentially expressed genes for each cluster in a one-vs-rest comparison and evaluated the gene set enrichment via Metascape (Zhou et al., 2019). The most enriched pathways against a permutation test are shown ordered by significance from top to bottom (negative log of the P-value). Pseudotime was determined with Scanpy: https://scanpy.readthedocs.io/en/stable/api/scanpy.tl.dpt.html?highlight=pseudotime#scanpy.tl.dpt. Local states were computed by interpolating pseudotime as a scalar potential along a grid in the embedding restricted to areas populated by cells and subsequently identifying local pseudotime maxima. The 2D gradient of the potential was computed to determine the vector field (arrows) shown in the figure panel. Because this approach assumes the existence of a unique pseudotime potential, it cannot account for a possible nonzero curl in the vector field (e.g. a self-renewing embryonic population). To address this simplification, RNA velocity was performed (La Manno et al., 2018) using scVelo, which explicitly models curly developmental trajectories (https://www.biorxiv.org/content/10.1101/820936v1). To evaluate the global patterns of genes demonstrating low or noisy expression, we calculated a simmetrized k-nearest neighbor similarity graph in PC space and smoothed the count matrix twice over the graph (i.e. reaching second-order neighbors). This approach is similar in spirit but conceptually simpler than MAGIC and other scRNA-Seq imputation algorithms (van Dijk et al., 2018). It also has the advantage of working on the actual PC space and knn graph that were used for the Leiden clustering and t-SNE embedding of the cells, ensuring a self-consistent representation of the data across the analysis. T Cell receptors were assembled using TraCeR (Stubbington et al., 2016) using the default parameters of the Singularity image. B cell receptors were assembled using BraCeR (Lindeman et al., 2018) with the parameter –IGH_networks, which agreed with our in-house pipeline consisting of Basic (Canzar et al., 2017) and Change-O (Gupta et al., 2015). Northstar was used to compare our data with Tabula Muris (Tabula Muris Consortium et al., 2018; Zanini et al., 2020, https://www.biorxiv.org/content/10.1101/820928v2), while manual merging and embedding was performed to compare our data with Schyns et al., 2019. Raw fastq files, count tables, and metadata are available on NCBI’s Gene Expression Omnibus (GEO) website (GSE147668), and the gene count and metadata tables are also available on FigShare at https://figshare.com/articles/Diverse_homeostatic_and_immunomodulatory_roles_of_immune_cells_in_the_developing_mouse_lung_revealed_at_single_cell_resolution/12043365.

In-situ validation using RNAscope and immunofluorescence (IF)

Request a detailed protocol

Embryonic and post-natal mice were euthanized as described above. Female and male mice were randomly selected from the litter, and at least two litters were used to source the lung tissue for all validation studies. E18.5 lungs were immediately placed in 10% neutral buffered formalin following dissection. P1, P7, and P21 murine lungs were perfused as described above, and P7 and P21 lungs inflated with 2% low melting agarose (LMT) in 1xPBS and placed in 10% neutral buffered formalin. Following 20 hr incubation at 4C, fixed lungs were washed twice in 1xPBS and placed in 70% ethanol for paraffin-embedding. In situ validation of genes identified by single cell RNA-seq was performed using the RNAscope Multiplex Fluorescent v2 Assay kit (Advanced Cell Diagnostics) and according to the manufacturer’s protocol. Formalin-fixed paraffin-embedded (FFPE) lung sections (5 μm) were used within a day of sectioning for optimal results. Nuclei were counterstained with DAPI (Life Technology Corp.) and extracellular matrix proteins stained with hydrazide (Clifford et al., 2011). Opal dyes (Akoya Biosciences) were used for signal amplification as directed by the manufacturer. Images were captured with Zeiss LSM 780 and Zeiss LSM 880 confocal microscopes, using 405 nm, 488 nm, 560 nm and 633 nm excitation lasers. For scanning tissue, each image frame was set as 1024 × 1024 and pinhole 1AiryUnit (AU). For providing Z-stack confocal images, the Z-stack panel was used to set z-boundary and optimal intervals, and images with maximum intensity were processed by merging Z-stacks images. For all both merged signal and split channels were collected.

Quantification of RNA scope images

Request a detailed protocol

For select RNA scope experiments, the number and location of specific cells were quantified. Adjustments to the brightness and contrast of an image to increase the clarity of the signal was always applied to the entire image and never to separate areas of the image. For the determination of the percentage of Cd68+ cells located in the perivascular versus the perivascular space, all stitch-and-tile images containing large areas of parenchyma and at least on small vessel (20–30 μm) were included in the quantification. For all quantification analysis, a cell was deemed positive if it contained >1 puncta per cell for a specific gene. All studies were performed >2 times, with a total of n = 8 E18.5, n = 6 P1, n = 4 P7, and n = 4 P21mice used for the FISH studies.

Intracellular flow cytometry

Request a detailed protocol

P1 and P21 male and female murine lungs were isolated as described above. Cells were blocked for 30 min with CD16/CD32 (Tonbo Biosciences). For intracellular analyses, cells were stimulated with PMA (50 ng/ml, Sigma-Aldrich), ionomycin (750 ng/mL, Sigma-Aldrich), and GolgiStop (BD Biosciences) for 5 hr, and then surface stained with fluorochrome-conjugated antibodies for 30 min: CD3 (clone: 145–2 C11, BD Biosciences), CD4 (RM4-5, BD Biosciences), and CD8a (clone: 53–6.7, Biolegend). Cells were then permeabilized with FoxP3 Fixation/Permeabilization Kit (BD Biosciences) as indicated by the manufacturer, and stained for TBET (clone: 4B10, Biolegend), GATA3 (clone: L50-823, BD Biosciences), FOXP3 (clone: FJK-16s, eBioscience), RORγt (clone: Q31-378, BD Biosciences), IFNγ (clone: XMG1.2, Biolegend), IL-4 (clone: 11B11, BD Biosciences), IL-10 (clone: JES5-16E3, Biolegend), and IL-17 (clone: TC11-18H10, Miltenyi Biotec) for 30 min. Cells were read using an LSRII flow cytometer using FACSDiva software. Flow data was analyzed using FlowJo (Tree Star Inc). Flow cytometry analysis for this project was done on instruments in the Stanford Shared FACS Facility.

Statistical analyses

Request a detailed protocol

To identify differentially expressed genes within cell populations, Kolmogorov Smirnov tests on the distributions of gene expression were performed on all genes, and the genes with the largest test statistic (lowest P value) were chosen as the top differentially expressed genes. For the flow cytometry studies of lung lymphocytes, 7–8 mice were in each group. Data are presented as mean ± SD. Differences between groups was determined by Student’s t-test with a P value of ≤ 0.05 considered statistically significant.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
    Complement system in lung disease
    1. PH Pandya
    2. DS Wilkes
    (2014)
    American Journal of Respiratory Cell and Molecular Biology 51:467–473.
    https://doi.org/10.1165/rcmb.2013-0485TR
  77. 77
    Scikit-learn: machine learning in Python
    1. F Pedregosa
    2. G Varoquaux
    3. A Gramfort
    4. V Michel
    5. B Thirion
    6. O Grisel
    7. M Blondel
    8. P Prettenhofer
    9. R Weiss
    10. V Dubourg
    11. J Vanderplas
    12. A Passos
    13. D Cournapeau
    14. M Brucher
    15. M Perrot
    16. E Duchesnay
    (2011)
    Journal of Machine Learning Research : JMLR 12:2825–2830.
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
  91. 91
  92. 92
  93. 93
  94. 94
  95. 95
  96. 96
  97. 97
  98. 98
  99. 99
  100. 100
  101. 101
  102. 102
    Angiogenesis
    1. DR Senger
    2. GE Davis
    (2011)
    Cold Spring Harbor Perspectives in Biology 3:a005090.
    https://doi.org/10.1101/cshperspect.a005090
  103. 103
  104. 104
  105. 105
  106. 106
  107. 107
  108. 108
  109. 109
  110. 110
  111. 111
  112. 112
  113. 113
  114. 114
  115. 115
  116. 116
    Visualizing data using t-SNE
    1. L van der Maaten
    2. G Hinton
    (2008)
    Journal of Machine Learning Research : JMLR 9:2579–2605.
  117. 117
  118. 118
  119. 119
  120. 120
  121. 121
  122. 122
  123. 123
  124. 124
  125. 125
  126. 126
  127. 127
  128. 128
  129. 129
  130. 130
  131. 131
  132. 132
  133. 133
  134. 134

Decision letter

  1. Emma L Rawlins
    Reviewing Editor; University of Cambridge, United Kingdom
  2. Tadatsugu Taniguchi
    Senior Editor; Institute of Industrial Science, The University of Tokyo, Japan

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This work provides a comprehensive bioinformatic view of the immune cell landscape of the developing mouse lung at an important biological transition from in utero to the neonatal air-breathing period. The data are coupled with well-quantified spatial analysis of some of the most dynamic cell types identified in the analysis. They are also presented clearly in the context of analogous recent studies and provide a biological overview of immune cell evolution in mouse lung perinatal development.

Decision letter after peer review:

Thank you for submitting your article "Diverse homeostatic and immunomodulatory roles of immune cells in the developing mouse lung at single cell resolution" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Tadatsugu Taniguchi 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 a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

The present study by Domingo-Gonzales et al. is a comprehensive bioinformatic approach, in which they investigate immune cells in the lung perinatally by characterizing changes in composition, localization, and function before birth and through the first three weeks of life. To this end, they combined a transcriptomic analysis (single cell RNA-seq) with fluorescent in situ hybridization. They clearly identified and characterized the localization and dynamics of five macrophage/monocyte populations with very distinct functions in murine lungs. Moreover, they provide strong data on the abundance and distribution of other important immune cells (e.g. dendritic cells) that are central in lung developmental processes, but also in lung diseases. Overall the manuscript is very well written addressing an highly important and disease-relevant topic that has not been elucidated to this extent yet.

Essential revisions:

These can all be done without additional experiments.

1) The quality of the spatial information describing the location of the macrophages/monocytes needs to be improved. A detailed quantification of the in situ hybridisation images is required for clarifying the distribution of perivascular macrophage/monocytes in the tissues e.g. in Figure 2E, F all Cd68+ cells should be scored as Dab2+, Plac8+, both, or neither. In Figure 3 the % macrophages around the vessels versus scattered should be given, the relative % of type 1 versus IV around the vessels and the % of co-labelling with Mki67. Moreover, clarification of the structures seen in Figure 3 would be helpful – are the macrophages arranged in intermittent circles around the diameter of the vessels? Are they forming lines along their length (if so, from where to where)? If possible, an antibody-based immunofluorescence approach to improve the cellular resolution of the image data would be desirable.

2) The data in Figure 4A and the respective tables are very interesting, providing new insight in the functional role of the macrophages. Pathway analysis using these respective genes in the supplementary tables would give more information of the pathways and functional regulation of the individual macrophage types.

3) Figure 3 and Figure 4 suggest that some macrophage population, e.g. Mac I, might regulate and promote angiogenesis in the lung. These findings coupled with Figure 4 showing genes related to angiogenesis let me query if anti-angiogenic factors might be produced by specific macrophages. That might be of interest in regulating alveolarization, but also in the pathogenesis of lung diseases (PAH or BPD).

4) Did the authors relate macrophages or other immune cells, such as dendritic cells or B cells, to the regulation of stemness or lung progenitor cells? They are crucial in the process of alveolarization.

5) We recommend including Fc receptor data in the main manuscript. They are important in disease development and often a therapeutic target.

6) How do the present mouse data relate to human lung? At birth, murine lungs are in the saccular stage, whereas human lungs are already at the alveolar stage. What might be the main driver of macrophage differentiation: oxygen? mechanical forces? That could be further addressed in the Discussion.

7) Discussion: It would be interesting to discuss more in detail how this precise regulation of immune cell differentiation and distribution in the lung during development could underlie the origins of lung disease. For example, there are data showing that BPD is related to activation of macrophages. Could asthma originate from adverse effects on dendritic and T cells during a critical phase of differentiation? That could be discussed.

8) To make this manuscript more useful to the community further details need to be provided about the bioinformatic analysis:

a) The authors should state exactly how many cells in total were profiled and how many cells passed QC. It will be useful as part of QC metric to see the number of genes per cell state identified.

b) The link to scripts, gene count and metadata table provided in the manuscript was inactive.

c) The authors should state what is actually displayed as gene expression for all figures – is this expression value on log scale? What does min, mod, high e.g. Figure 4A refer to quantitatively?

d) Smart-seq2 is less likely to have doublet contamination but it is not clear if doublet detection and removal during analysis was attempted.

e) Was batch correction methods e.g. Harmony implemented? – this would be helpful to assess if findings were impacted by batch effects.

f) The observation of Mac I that encircles blood vessels prenatally and disappears after birth is interesting. However, the claims based on using inferred trajectory analysis of scRNA-seq data to suggest Mac I transition into Mac II and III needs to be more cautiously made as definitive lineage tracing or fate-mapping experiments were not done to validate the trajectory inference.

g) What pseudotime trajectory method was used?

9) DC nomenclature may be confusing as DC1, DC2 and DC3 are recognised subsets of conventional DCs. DCII described by authors are monocyte-derived DC and DCIII identity is unclear. This should be clarified in the text.

10) Some integration of this dataset with other existing datasets from embryonic, perinatal and adult immune cells would permit harmonization of the cell states described by the various manuscripts and enhancing the value of this manuscript from a biological advancement and resource value of the manuscript.

https://doi.org/10.7554/eLife.56890.sa1

Author response

Essential revisions:

These can all be done without additional experiments.

1) The quality of the spatial information describing the location of the macrophages/monocytes needs to be improved. A detailed quantification of the in situ hybridisation images is required for clarifying the distribution of perivascular macrophage/monocytes in the tissues e.g. in Figure 2E, F all Cd68+ cells should be scored as Dab2+, Plac8+, both, or neither. In Figure 3 the % macrophages around the vessels versus scattered should be given, the relative % of type 1 versus IV around the vessels and the % of co-labelling with Mki67. Moreover, clarification of the structures seen in Figure 3 would be helpful – are the macrophages arranged in intermittent circles around the diameter of the vessels? Are they forming lines along their length (if so, from where to where)? If possible, an antibody-based immunofluorescence approach to improve the cellular resolution of the image data would be desirable.

We very much appreciate this point from the reviewers. We have quantified the in situ hybridization images to provide the: (i) percent Dab2+ vs. Plac8+ cells at E18.5 and P7; (ii) the percentage of perivascular vs. parenchymal macrophages at E18.5, (iii) the % proliferating macrophages; and (iv) the % of Mac I vs. Mac IV found around vessels at E18.5. These data are now included in Figure 2E and Figure 3D, E, and G. Whether the macrophages are forming lines along the length of the vessels is an intriguing and important question that is difficult for us to definitively answer with our current images. Although some images that capture both the cross-section and a longitudinal cut of the vessel suggest that the macrophages may also be present along the length of the vessel, we cannot claim this to be true with confidence. We now acknowledge the limitation in the Discussion and state that antibody-based immunofluorescence followed by Z-stack imaging of thick tissue sections would be required to definitively address this important question.

2) The data in Figure 4A and the respective tables are very interesting, providing new insight in the functional role of the macrophages. Pathway analysis using these respective genes in the supplementary tables would give more information of the pathways and functional regulation of the individual macrophage types.

We agree. We have performed pathway analysis using the DEGs from the five macrophage/monocyte populations and have included these data as Figure 4—figure supplement 1, and top pathways for each population reported in the Results. Given that many top DEGs have published functions that are not included in the pathway analysis, we also continue to highlight individual genes of interest with the relevant citations.

3) Figure 3 and Figure 4 suggest that some macrophage population, e.g. Mac I, might regulate and promote angiogenesis in the lung. These findings coupled with Figure 4 showing genes related to angiogenesis let me query if anti-angiogenic factors might be produced by specific macrophages. That might be of interest in regulating alveolarization, but also in the pathogenesis of lung diseases (PAH or BPD).

We agree with the reviewer that the production of anti-angiogenic factors may be important not only for understanding alveolarization but also relevant to lung diseases such as PAH and BPD. We now also highlight some DEGs that are reported to have anti-angiogenic functions, including Serpine1, and Pf4 (subsection “Distinct transcriptional profiles and spatial distribution suggest specific physiologic functions for discrete macrophage populations”, third and fourth paragraphs).

4) Did the authors relate macrophages or other immune cells, such as dendritic cells or B cells, to the regulation of stemness or lung progenitor cells? They are crucial in the process of alveolarization.

Although it was not our initial focus, we agree that this is a very important point given the role of lung progenitor cells in all stages of lung development. We also looked at the expression of secreted growth factors and other effectors of key pathways that regulate lung progenitor cells including the Wnt, Shh, Tgf-β, and BMP pathways. Of note, the Mac II and Mac IV clusters expressed secreted molecules that regulate these pathways. The Mac II cells express both Tgf-β1 and 2, Nestin 1 and 4 and Hhip, while the Mac IV cells expressed Fgf10, Bmp4, Wnt2 and Wnt5. These data are now detailed in the Results (subsection “Distinct transcriptional profiles and spatial distribution suggest specific physiologic functions for discrete macrophage populations”, second and fourth paragraphs) and included as Figure 4—figure supplement 2.

5) We recommend including Fc receptor data in the main manuscript. They are important in disease development and often a therapeutic target.

Thank you for this recommendation. The Fc receptor data is now included in the main manuscript as Figure 7.

6) How do the present mouse data relate to human lung? At birth, murine lungs are in the saccular stage, whereas human lungs are already at the alveolar stage. What might be the main driver of macrophage differentiation: oxygen? mechanical forces? That could be further addressed in the Discussion.

We thank the reviewers for this important point. An additional paragraph has been added to the Discussion (fifth paragraph) addressing factors that might be driving immune cell diversity in the developing lung and how these data from the mouse might be compared to similar data from human lungs to better understand whether the physiologic changes at birth, as opposed to differences in the lung immune cell niche at the different stages of development might be the predominant driver of diversity.

7) Discussion: It would be interesting to discuss more in detail how this precise regulation of immune cell differentiation and distribution in the lung during development could underlie the origins of lung disease. For example, there are data showing that BPD is related to activation of macrophages. Could asthma originate from adverse effects on dendritic and T cells during a critical phase of differentiation? That could be discussed.

We agree with this excellent point. We have addressed this by including an additional paragraph in the Discussion relating our findings to pediatric lung diseases such as bronchopulmonary dysplasia, pulmonary hypertension and asthma (Discussion, sixth paragraph).

8) To make this manuscript more useful to the community further details need to be provided about the bioinformatic analysis:

a) The authors should state exactly how many cells in total were profiled and how many cells passed QC. It will be useful as part of QC metric to see the number of genes per cell state identified.

Thank you for this suggestion. Please see the breakdown of the parameters and QC during data ingestion and filtering as outlined in Author response table 1. Please note that the dataset was sequenced in combination with other studies on the same biological samples. Therefore, the number of immune cells at the second to last step is much smaller than the total number of sequenced cells, which include other cell populations processed from the same plates.

Author response table 1

b) The link to scripts, gene count and metadata table provided in the manuscript was inactive.

We would like to thank the reviewers for identifying this error. The link to the scripts was inactive due to a renaming issue which has been corrected. The metadata and gene counts are available on FigShare and on GEO as stated.

c) The authors should state what is actually displayed as gene expression for all figures – is this expression value on log scale? What does min, mod, high e.g. Figure 4A refer to quantitatively?

Thank you for this request, which we clarified in the revised manuscript. The color scheme of the gene expression tSNE embedding is logarithmic viridis color scales, with a pseudocount of 0.1 counts per million, and normalized to the highest expressing cell. The color scheme of the heatmaps plasma, logarithmic with a pseudocount of 0.1 counts per million, and normalized to the highest expressing population for each gene – i.e. every column will have one or more populations expressing “high”/yellow. The legend refers to 0%, 33%, 67%, and 100% expression (on a log scale) compared to the highest expressing population, i.e. describes equally spaced intervals of expression (on a log scale). These details have been added to the figure legends.

d) Smart-seq2 is less likely to have doublet contamination but it is not clear if doublet detection and removal during analysis was attempted.

Doublets were removed in two ways. First, clusters that expressed epithelial, endothelial or mesenchymal genes were excluded. Clusters showing joint expression of mutually exclusive cell type markers (e.g. Epcam, Cdh5, Col6a2) with Cd45 were manually excluded as well. 147 out of 4199 cells were removed, which is a slightly conservative filter compared to the expected doublet rate of around 1-2% in 384-well plates on our cell sorter. We have added this information to the Materials and methods (subsection “Data analysis and availability”).

e) Was batch correction methods e.g. Harmony implemented? – this would be helpful to assess if findings were impacted by batch effects.

We purposely did not run batch effect correction methods to avoid eliminating potentially relevant biological signals (e.g. gender and developmental effects). In our experience, such batch correction methods are primarily needed when mixing different technologies. However, during the feature selection step for calculating neighborhood graphs towards clustering and embeddings, we identified over-dispersed features within each mouse and only kept features found in several mice. This enabled us to maintain a straightforward biological interpretation for the counts (reads per million mapped uniquely) while mitigating technical artifacts in the visualizations.

f) The observation of Mac I that encircles blood vessels prenatally and disappears after birth is interesting. However, the claims based on using inferred trajectory analysis of scRNA-seq data to suggest Mac I transition into Mac II and III needs to be more cautiously made as definitive lineage tracing or fate-mapping experiments were not done to validate the trajectory inference.

Thank you for this comment. We completely agree, and we have added an additional sentence to the Results stating that definitive genetic lineage tracing studies are necessary to confirm the transition of the Mac I cells to Mac II and III. We also added RNA velocity to Figure 4G, as an additional method supporting the pseudotime analysis.

g) What pseudotime trajectory method was used?

We apologize for omitting this detail. For the pseudotime analysis we used the default algorithm in Scanpy. This information is included in the Materials and methods. We also added RNA velocity data to the manuscript, and this was performed using scVelo (Bergen et al., 2019, https://doi.org/10.1101/820936).

9) DC nomenclature may be confusing as DC1, DC2 and DC3 are recognised subsets of conventional DCs. DCII described by authors are monocyte-derived DC and DCIII identity is unclear. This should be clarified in the text.

We appreciate this comment and agree that the suspected DC identity was not clear in the prior version. We have examined numerous additional genes associated with cDC1 and cDC2 and find that the prior population we called “DCII” are most consistent with cDC2 rather than monocyte-derived DC. We have included these additional t-SNE plots in Figure 8—figure supplement 1, and renamed our DC populations: cDC1, cDC2, and migratory (mig)-DC.

10) Some integration of this dataset with other existing datasets from embryonic, perinatal and adult immune cells would permit harmonization of the cell states described by the various manuscripts and enhancing the value of this manuscript from a biological advancement and resource value of the manuscript.

Thank you for this excellent suggestion. To this end, we have analyzed our data in concert with two previously published data sets: Schyns et al., 2019, and Tabula Muris, 2018. The results from these harmonized data analyses are included in the manuscript (subsection “Comparison of Mac I-V populations with adult murine lung data sets”), and the data included in a new figure (Figure 6).

https://doi.org/10.7554/eLife.56890.sa2

Article and author information

Author details

  1. Racquel Domingo-Gonzalez

    1. Division of Critical Care Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    2. Center for Excellence in Pulmonary Biology, Stanford University School of Medicine, Stanford, United States
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Fabio Zanini
    Competing interests
    No competing interests declared
  2. Fabio Zanini

    1. Department of Bioengineering, Stanford University, Stanford, United States
    2. Prince of Wales Clinical School, Lowy Cancer Research Centre, University of New South Wales, Sydney, Australia
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Racquel Domingo-Gonzalez
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7097-8539
  3. Xibing Che

    1. Division of Critical Care Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    2. Center for Excellence in Pulmonary Biology, Stanford University School of Medicine, Stanford, United States
    3. Division of Pulmonary, Asthma and Sleep Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    Contribution
    Data curation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Min Liu

    1. Division of Critical Care Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    2. Center for Excellence in Pulmonary Biology, Stanford University School of Medicine, Stanford, United States
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Robert C Jones

    Department of Bioengineering, Stanford University, Stanford, United States
    Contribution
    Conceptualization, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7235-9854
  6. Michael A Swift

    Department of Chemical and Systems Biology, Stanford University, Stanford, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  7. Stephen R Quake

    1. Department of Bioengineering, Stanford University, Stanford, United States
    2. Chan Zuckerberg Biohub, San Francisco, United States
    3. Department of Applied Physics, Stanford University, Stanford, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Writing - review and editing
    Contributed equally with
    David N Cornfield and Cristina M Alvira
    For correspondence
    steve@quake-lab.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1613-0809
  8. David N Cornfield

    1. Division of Critical Care Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    2. Center for Excellence in Pulmonary Biology, Stanford University School of Medicine, Stanford, United States
    3. Division of Pulmonary, Asthma and Sleep Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing - review and editing
    Contributed equally with
    Stephen R Quake and Cristina M Alvira
    For correspondence
    cornfield@stanford.edu
    Competing interests
    No competing interests declared
  9. Cristina M Alvira

    1. Division of Critical Care Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, United States
    2. Center for Excellence in Pulmonary Biology, Stanford University School of Medicine, Stanford, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    Contributed equally with
    Stephen R Quake and David N Cornfield
    For correspondence
    calvira@stanford.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6921-0001

Funding

National Institutes of Health (HL122918)

  • Cristina M Alvira

National Institutes of Health (HD092316)

  • David N Cornfield
  • Cristina M Alvira

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Sai Saroja Kolluru (Stanford University) for assistance with library submission to the Chan Zuckerberg Biohub, Yuan Xue (Stanford University) for assistance with the initial single cell RNA-seq data acquisition, Astrid Gillich for technical support with the RNAscope experiments, and Maya Kumar for providing hydrazide. We thank Henry Hampton for his input and fruitful discussions. We also thank the Stanford Shared FACS Facility, Lisa Nichols, Meredith Weglarz, and Tim Knaak for assistance with the flow cytometry instrumentation and antibody panel design. Flow cytometry data was collected on an instrument in the Stanford Shared FACS Facility obtained using NIH S10 Shared Instrument Grant (S10RR027431-01). This work was supported by National Institutes of Health grants HL122918 (CMA), HD092316 (CMA, DNC), the Stanford Maternal Child Health Institute Tashia and John Morgridge Faculty Scholar Award (CMA), the Stanford Center of Excellence in Pulmonary Biology (DNC), Bill and Melinda Gates Foundation (SRQ), and the Chan Zuckerberg Biohub (DNC. and SRQ). MAS is supported by the NSF-GRFP.

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols (#19087) of Stanford University School of Medicine.

Senior Editor

  1. Tadatsugu Taniguchi, Institute of Industrial Science, The University of Tokyo, Japan

Reviewing Editor

  1. Emma L Rawlins, University of Cambridge, United Kingdom

Publication history

  1. Received: March 13, 2020
  2. Accepted: May 13, 2020
  3. Accepted Manuscript published: June 2, 2020 (version 1)
  4. Version of Record published: July 13, 2020 (version 2)

Copyright

© 2020, Domingo-Gonzalez et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,176
    Page views
  • 244
    Downloads
  • 2
    Citations

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

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Evolutionary Biology
    Meng Xie et al.
    Research Article Updated

    Growth plate and articular cartilage constitute a single anatomical entity early in development but later separate into two distinct structures by the secondary ossification center (SOC). The reason for such separation remains unknown. We found that evolutionarily SOC appears in animals conquering the land - amniotes. Analysis of the ossification pattern in mammals with specialized extremities (whales, bats, jerboa) revealed that SOC development correlates with the extent of mechanical loads. Mathematical modeling revealed that SOC reduces mechanical stress within the growth plate. Functional experiments revealed the high vulnerability of hypertrophic chondrocytes to mechanical stress and showed that SOC protects these cells from apoptosis caused by extensive loading. Atomic force microscopy showed that hypertrophic chondrocytes are the least mechanically stiff cells within the growth plate. Altogether, these findings suggest that SOC has evolved to protect the hypertrophic chondrocytes from the high mechanical stress encountered in the terrestrial environment.

    1. Chromosomes and Gene Expression
    2. Developmental Biology
    Rozenn Riou et al.
    Research Article

    Erythropoietin (EPO) is a key regulator of erythropoiesis. The embryonic liver is the main site of erythropoietin synthesis, after which the kidney takes over. The adult liver retains the ability to express EPO, and we discovered here new players of this transcription, distinct from the classical hypoxia-inducible factor pathway. In mice genetically-invalidated in hepatocytes for the chromatin remodeler Arid1a, and for Apc, the major silencer of Wnt pathway, chromatin was more accessible and histone marks turned into active ones at the Epo downstream enhancer. Activating β-catenin signaling increased binding of Tcf4/β-catenin complex and upregulated its enhancer function. The loss of Arid1a together with β-catenin signaling, resulted in cell-autonomous EPO transcription in mouse and human hepatocytes. In mice with Apc-Arid1a gene invalidations in single hepatocytes, Epo de novo synthesis led to its secretion, to splenic erythropoiesis and to dramatic erythrocytosis. Thus, we identified new hepatic EPO regulation mechanism stimulating erythropoiesis.