key: cord-0277402-ah6zgz26
authors: Li, Xin; Kolling, Fred W.; Aridgides, Daniel; Mellinger, Diane; Ashare, Alix; Jakubzick, Claudia V.
title: ScRNA-seq Expression of APOC2 and IFI27 Identifies Four Alveolar Macrophage Superclusters in Cystic Fibrosis and Healthy BALF
date: 2022-01-30
journal: bioRxiv
DOI: 10.1101/2022.01.30.478325
sha: df1e0035c102decd0cccf1870b19127a0ebb5029
doc_id: 277402
cord_uid: ah6zgz26

Rationale Alveolar macrophages (AMs) reside on the luminal surface of the airways and alveoli, ensuring proper gas exchange by ingesting cellular debris and pathogens, and regulating inflammatory responses. Recent studies have highlighted the heterogeneity of airspace macrophages from healthy bronchoalveolar lavage fluid and cystic fibrosis sputum. Understanding the heterogeneity and diverse roles played by AMs and recruited monocytes is critical for treating CF and other airway diseases. Objectives To identify and compare the cellular composition and functional diversity between CF and healthy airspace macrophages and monocytes. Methods We performed single-cell RNA sequencing (scRNA-seq) on 113,213 BAL cells from four healthy and three mild CF subjects. Transcriptional data and TotalSeq technology were used to confirm cell surface markers that distinguish resident AMs from recruited monocytes. Measurements and Main Results Unbiased clustering identified four AM superclusters based on the expression of APOC2 and IFI27 genes. Each supercluster contains four shared subclusters named after their differentially expressed gene(s): IGF1.AMs, CCL18.AMs, CXCL5.AMs, and Cholesterol.AMs. AM Supercluster 4 contained two additional subclusters. Beyond the superclusters, several additional AM clusters were identified: Chemokine, Metallothionein, Interferon, and Cycling AMs. Interestingly, the Chemokine cluster further divides into four subclusters, each selectively expressing a unique combination of chemokines. Lastly, minimal differences were observed between mild CF and healthy individuals in airspace cellular composition. Conclusions ScRNA-seq identified four AM superclusters. Several fundamental questions require further investigation, including these clusters’ locations, functions, transcription factors that regulate their distinct programming, and whether similar macrophage superclusters can be found in other organs.

Alveolar macrophages (AMs) are the primary phagocytes within the airspace of the lungs. In response to inflammatory stimuli, AMs secrete cytokines and chemokines to initiate the immune response. Subsequently, AMs and recruited monocytes contribute to the resolution of the inflammatory response to prevent excess damage (1) (2) (3) (4) . In addition, AMs play a major role in maintaining lipid homeostasis in the lung, which is essential for adequate gas exchange and alveolar epithelial integrity.

AMs may hold keys to understanding, preventing, and curing several diseases.

They derive from prenatal monocytes, self-renew, and have shown remarkable plasticity by altering their transcriptome in response to the environmental changes (3, (5) (6) (7) (8) . Their transcriptional profile is altered in various inflammatory lung diseases, including asthma (2, 9) , chronic obstructive pulmonary disease (10) , cystic fibrosis (CF), and cancer (11) (12) (13) (14) (15) (16) . Therefore, enhancing our current understanding of the normal cellular composition and functional diversity of AM subtypes will help characterize disease-related transcriptional changes more precisely and assess targeting specifically a given AM subtype to restore immune balance.

Single-cell RNA sequencing (scRNA-seq) technology is critical for accomplishing these goals. AMs have traditionally been thought to be a uniform population of cells that can be activated by different disease states (5) . However, recent scRNA-seq studies have revealed a rich diversity in AMs in bronchoalveolar lavage fluid (BALF) from healthy subjects, with multiple subpopulations that have yet to be characterized in CF and many other diseases (17) . Another study investigated differences in immune population in sputum from CF subjects and found that sputum immune cells exhibited differential activation and maturity (18) .

Our study was designed to investigate differences in the immune cell populations of BALF obtained from individuals with CF and healthy controls (HC). Using scRNA-seq, we found minor differences in cell populations between CF subjects with no significant lung disease and healthy controls, suggesting that normalizing the CF environment via treatments results in healthy-like airspace macrophages. Interestingly, unbiased analysis of this dataset identified four AM superclusters based on the expression of APOC2 and IFI27 genes, which have not been previously described. Each supercluster contains four shared subclusters named after their differentially expressed genes (DEGs): IGF1.AMs, CCL18.AMs, CXCL5.AMs, and Cholesterol.AMs. Beyond the four AM superclusters, we found four additional clusters. The most striking was the Chemokine cluster, which further divides into four subclusters with each selectively expressing a distinct combination of chemokines. Hence, there appears to be a clear division of labor in leukocyte recruitment. Moreover, it also suggests that there are distinct transcription factors (based on our regulon analysis) that regulate the differentiation and programming of each Chemokine AM cluster. This pattern is analogous to what is found in T cell biology, where distinct transcription factors regulate the cell differentiation pathway to produce a defined combination of cytokines known as Th1, Th2, Th17, etc. Overall, our dataset opens many new areas for further investigation, including addressing how resident macrophage and monocyte subtypes are altered during disease and what factors regulate their programming.

Four HC subjects and three subjects with CF underwent research bronchoscopy as previously described (19, 20) . Briefly, after local anesthesia to the posterior pharynx and intravenous conscious sedation, a flexible fiberoptic bronchoscope was inserted transorally and passed through the vocal cords into the trachea. The bronchoscope was sequentially wedged into tertiary bronchi in the right upper lobe (RUL), right middle lobe (RML), and right lower lobe (RLL), saline was instilled, and BALF was collected. An aliquot of 10 mL of BALF from the RML was used for the studies outlined here. All subjects provided written informed consent, and this study was approved by the Institutional Review Board of Dartmouth-Hitchcock Medical Center (protocol #22781).

Sample processing and library construction were performed immediately after lavage.

1X phosphate-buffered saline (PBS) containing 0.5 mM EDTA and 0.1% bovine serum albumin (BSA) was added to the BALF. Samples were filtered through 100-micron cell strainers and centrifugated at 300 x g for five minutes at 4 °C, and supernatants were discarded. TotalSeq antibodies were added at concentrations recommended by the manufacturer and stained for 30 min on ice (21) , details in online supplement. After staining, chilled Hank's balanced salt solution (HBSS) supplemented with 0.5% BSA was used to wash the sample. Samples were again filtered through 100-micron cell strainers and spun as above. Supernatants were aspirated, and pellets were resuspended with HBSS plus 0.5% BSA at an approximate concentration of 7.5 × 10 5 cells/mL. Cell quality and viability were assessed with a Cellometer K2 (Nexcelom Bioscience, Lawrence, MA). All samples had viability >80%. Single cells were then processed using the Chromium Next GEM Single Cell 3' Platform (10X Genomics, Pleasanton, CA). Approximately 30,000 cells were loaded on each channel with an average recovery rate of 24,000 cells. Libraries were sequenced on NextSeq 500/550 (Illumina, San Diego, CA) with an average sequencing depth of 50,000 reads/cell. Data Preparation: Raw sequencing reads were demultiplexed, mapped to the GRCh38 human reference genome, and gene-expression matrices were generated using CellRanger v6.1 (10X Genomics, Pleasanton, CA). The following analyses were conducted in R 4.1 (22) and Python 3.6. Seurat package v4.0 was used for downstream data analyses (23) , and figures were produced using the package ggplot2. Following a standard workflow, the gene-expression matrix was filtered to discard cells with less than 200 genes, as well as genes that were expressed in less than 3 cells. Samples then underwent quality control to remove cells with either too many or too few expressed genes (average around 2000 and 7000) and cells with too many mtRNA (average around 10%), resulting in a total of 113,213 cells. Then, "SCTransform" was applied with the "glmGamPoi" method to normalize gene expression data (24, 25) , and a CLR transformation was applied to normalize protein data within each cell. After individual preparation, all the samples were introduced into a combined Seurat object via "FindIntegrationAnchors" and "IntegrateData" functions (24) . Then scaled values of variable genes were then subject to principal component analysis (PCA) for linear dimension reduction. A shared nearest neighbor network was created based on Euclidean distances between cells in multidimensional PC space (the first 50 PC were used) and a fixed number of neighbors per cell (30 neighbors 

To assess the cellular composition in the non-diseased airspace versus that with CF, four HCs and three subjects with CF underwent BAL (donors, Table 1 ). BALF samples were spun, washed, and immediately loaded onto the 10X platform for scRNA-seq.

Sequenced cells were processed, normalized, and integrated using the Seurat package (details in Methods). Uniform manifold approximation and projection (UMAP) of all seven samples illustrates a similar cluster distribution across all subjects ( Figure 1A ).

Using a curated gene list, we identified twelve major cell types (Complete list of genes: Table E1 and Figure 

To define AM heterogeneity, we isolated and re-clustered the AM population. Unbiased clustering defined 8 AM clusters. Some clusters were previously described (37) , such as the Metallothionein, Chemokine, IFN-reacting, and Cycling AM clusters. (Figure 2A-2D ).

In addition, we identified 4 novel AM superclusters based on the differential expression of two genes: APOC2 and IFI27 ( Figure Next, we performed gene ontology and regulon analysis on the monocyte clusters. The most notable GO term was for FCN1.Mono, whose genes were enriched for bacterial defense mechanisms suggesting the importance of this monocyte subtype during pulmonary bacterial infections ( Figure 5E and Figure E2 ). The regulon analysis illuminated the overrepresentation of zinc finger transcription factors suggesting a common dependency on this structural motif for the differentiation of monocyte subtypes into functionally distinct macrophages ( Figure 5F and Figure E2 ).

ScRNA AMs.S1 and AMs.S2 seem to evolve much later than their counterparts AMs.S3 and AMs.S4, even though they share 4 subclusters and most gene expression, except IFI27

and APOC2 ( Figure 6C ).

Although scRNA-seq has revolutionized our knowledge of macrophage heterogeneity, there are instances where investigators need to isolate bulk populations to perform other functional or morphological analyses. Therefore, flow cytometric antibodies that can be used to sort and enrich bulk AMs and recruited monocytes are in need. Using flow cytometry, we demonstrate that, in addition to high side-scatter for AMs and low side-scatter for monocytes, investigators can use antibodies against CD43

(SPN) and CD169 to define AMs and CD93, CD36, and CD14 to identify monocytes ( Figure 6D ) (43) (44) (45) (46) (47) . ScRNA-seq and TotalSeq (for single-cell level protein expression assessment) also support the use of CD169 and CD43 for AMs compared to CD93, CD36, and CD14 for monocytes ( Figure 6E-F) . Thus, several forms of technology validate that higher expression of CD43 and CD169 in BAL cells define AMs over recruited monocytes.

Prior studies have demonstrated functional differences in CF compared to non-CF macrophages (48) (49) (50) . To investigate the impact of aberrant CFTR function on cell heterogeneity and transcriptional differences in cells isolated from BALF, healthy and CF subjects with preserved lung function (Table 1) , limited lung inflammation, and not currently on CFTR modulator treatment were enrolled. Two subjects had previously been on CFTR modulator treatment but stopped due to side effects, while the third subject declined CFTR modulator treatment due to overall good health. The subjects with CF in this study were stable on their medication regimens and reported no exacerbations within the prior 24 months. We observed differential expression of several genes within the total cell population of BALF from subjects with CF versus HCs ( Figure 7A ), including increased expression of APOC1, APOC2, CCL18, and SOD2 in CF cells. Increased expression of APOE and CCL18 in CF AMs compared to HC AMs is of particular interest given prior reports showed that CCL18 is associated with alternatively activated macrophages, and people with CF have a higher proportion of alternatively active macrophages ( Figure 7B ) (37, 51, 52) . Among AM superclusters, CF AMs had increased expression of APOC1 in AMs.s2, AMs.s3, and AMs.s4; APOE and CCL18 in AMs.s4; and S100A9 in AMs.s3 ( Figure 7C ). Similar to the bulk monocyte population, monocyte subclusters from subjects with CF did not demonstrate differential expression of inflammatory genes ( Figure 7D ). Although our unbiased AM analysis in Figure 2B Many lines of evidence point to impaired innate immune cell function in the CF lung (48-50, 53, 54) . Our study was originally designed to investigate the cellular and transcriptional differences in cells extracted from BALF between a cohort of CF subjects with normal lung function compared to healthy controls. Such differences may contribute to the phenotypic differences described in prior studies (48-50, 53, 54) . 

All authors declare that there are no competing interests. 

33% (1) 50% (2) Average age, years (SD) 28 ± 4.6 29 ± 6.1 FEV1, percent predicted (SD) 89.8 ± 11 Pa colonization % (n)

67% (2) Values are means ± standard deviation (SD); n = number of subjects; FEV1 = forced expiratory volume in 1 second; Pa = Pseudomonas aeruginosa; CF = cystic fibrosis; HC = healthy control. 

Alveolar macrophage transcriptomic profiling in COPD shows major lipid metabolism changes

Big Eaters": The Versatile Role of Alveolar Macrophages in Health and Disease

Fas Determines Differential Fates of Resident and Recruited Macrophages During Resolution of Acute Lung Injury

Cell Origin Dictates Programming of Resident versus Recruited Macrophages during Acute Lung Injury

Single cell RNA sequencing identifies unique inflammatory airspace macrophage subsets

Fate mapping reveals origins and dynamics of monocytes and tissue macrophages under homeostasis

Minimal differentiation of classical monocytes as they survey steady-state tissues and transport antigen to lymph nodes

Monocyte-derived alveolar macrophages drive lung fibrosis and persist in the lung over the life span

Macrophage dysfunction in the pathogenesis and treatment of asthma

Alveolar Macrophage Immunometabolism and Lung Function Impairment in Smoking and Chronic Obstructive Pulmonary Disease

Disease-causing mutations in the cystic fibrosis transmembrane conductance regulator determine the functional responses of alveolar macrophages

Macrophage Dysfunction in Cystic Fibrosis: A Therapeutic Target to Enhance Self-Immunity

Epigenetic reprogramming of airway macrophages promotes polarization and inflammation in muco-obstructive lung disease

Cystic Fibrosis Lung Immunity: The Role of the Macrophage

Targeting resident macrophages in cancer

Tissue-resident macrophages provide a pro-tumorigenic niche to early NSCLC cells

Airspace Macrophages and Monocytes Exist in Transcriptionally Distinct Subsets in Healthy Adults

Single-Cell Transcriptional Archetypes of Airway Inflammation in Cystic Fibrosis

Analysis of Lung Microbiota in Bronchoalveolar Lavage, Protected Brush and Sputum Samples from Subjects with Mild-To-Moderate Cystic Fibrosis Lung Disease

Functional and metabolic impairment in cigarette smokeexposed macrophages is tied to oxidative stress

Simultaneous epitope and transcriptome measurement in single cells

R: A language and environment for statistical computing. R Foundation for Statistical Computing

Integrated analysis of multimodal single-cell data

Comprehensive Integration of Single-Cell Data

glmGamPoi: fitting Gamma-Poisson generalized linear models on single cell count data

Three Unique Interstitial Macrophages in the Murine Lung at Steady State

Single-cell RNA-seq reveals new types of human blood dendritic cells, monocytes, and progenitors

Human dendritic cell subsets: an update

Human dendritic cell subsets

Single-cell transcriptomics of blood reveals a natural killer cell subset depletion in tuberculosis

Transcriptional Basis of Mouse and Human Dendritic Cell Heterogeneity

Transcriptome analysis highlights the conserved difference between embryonic and postnatal-derived alveolar macrophages

Gene-expression profiles and transcriptional regulatory pathways that underlie the identity and diversity of mouse tissue macrophages

Characterization of dendritic cell subtypes in human cord blood by single-cell sequencing

Human and Mouse Transcriptome Profiling Identifies Cross-Species Homology in Pulmonary and Lymph Node Mononuclear Phagocytes

Characterizing the replicability of cell types defined by single cell RNA-sequencing data using MetaNeighbor

Cell-surface calreticulin initiates clearance of viable or apoptotic cells through trans-activation of LRP on the phagocyte

SCENIC: single-cell regulatory network inference and clustering

The single-cell transcriptional landscape of mammalian organogenesis

Reversed graph embedding resolves complex single-cell trajectories

Single-cell mRNA quantification and differential analysis with Census

The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells

Flow Cytometry Reveals Similarities Between Lung Macrophages in Humans and Mice

Flow Cytometric Analysis of Myeloid Cells in Human Blood, Bronchoalveolar Lavage, and Lung Tissues

Isolation and Characterization of Human Lung Myeloid Cells

Flow Cytometric Analysis of Mononuclear Phagocytes in Non-diseased Human Lung and Lung-draining Lymph Nodes

A Consistent Method to Identify and Isolate Mononuclear Phagocytes from Human Lung and Lymph Nodes

Altered iron metabolism in cystic fibrosis macrophages: the impact of CFTR modulators and implications for Pseudomonas aeruginosa survival

Low levels of insulin-like growth factor-1 contribute to alveolar macrophage dysfunction in cystic fibrosis

Genome-wide DNA methylation profiling shows a distinct epigenetic signature associated with lung macrophages in cystic fibrosis

Characterization of macrophage activation states in patients with cystic fibrosis

Chemokine-Induced Macrophage Polarization in Inflammatory Conditions

CFTR regulates phagosome acidification in macrophages and alters bactericidal activity

Monocyte derived macrophages from CF pigs exhibit increased inflammatory responses at birth

Extracellular Vesicles from Pseudomonas aeruginosa Suppress MHC-Related Molecules in Human Lung Macrophages