Dissociation, cellular isolation, and initial molecular characterization of neonatal and pediatric human lung tissues

Abstract

Human lung morphogenesis begins by embryonic life and continues after birth into early childhood to form a complex organ with numerous morphologically and functionally distinct cell types. Pulmonary organogenesis involves dynamic changes in cell proliferation, differentiation, and migration of specialized cells derived from diverse embryonic lineages. Studying the molecular and cellular processes underlying formation of the fully functional lung requires isolating distinct pulmonary cell populations during development. We now report novel methods to isolate four major pulmonary cell populations from pediatric human lung simultaneously. Cells were dissociated by protease digestion of neonatal and pediatric lung and isolated on the basis of unique cell membrane protein expression patterns. Epithelial, endothelial, nonendothelial mesenchymal, and immune cells were enriched by fluorescence-activated cell sorting. Dead cells and erythrocytes were excluded by 7-aminoactinomycin D uptake and glycophorin-A (CD235a) expression, respectively. Leukocytes were identified by membrane CD45 (protein tyrosine phosphatase, receptor type C), endothelial cells by platelet endothelial cell adhesion molecule-1 (CD31) and vascular endothelial cadherin (CD144), and both were isolated. Thereafter, epithelial cell adhesion molecule (CD326)-expressing cells were isolated from the endothelial- and immune cell-depleted population to enrich epithelial cells. Cells lacking these membrane markers were collected as “nonendothelial mesenchymal” cells. Quantitative RT-PCR and RNA sequencing analyses of population specific transcriptomes demonstrate the purity of the subpopulations of isolated cells. The method efficiently isolates major human lung cell populations that we announce are now available through the National Heart, Lung, and Blood Institute Lung Molecular Atlas Program (LungMAP) for their further study.

INTRODUCTION

Despite recent advances in lung biology, the complex cellular functions and intracellular interactions active in forming the human lung are not well understood. Embryonic and fetal lung development involves both structural and functional maturation requiring remarkable cell diversity to ensure matching of systems for ventilation and blood flow to enable successful respiration at birth. The respiratory tract continues to mature throughout childhood to achieve maximum respiratory capacity (7, 19, 21, 22, 27, 28). At least 40 different types of cells have been described in the lung (9, 15). Diverse cell types are actively involved in gas exchange, whereas others prevent pathogen entry, maintain tissue structure, manage nutrients, and provide structural support (30). Pulmonary cells can be categorized into four major cell populations: epithelial cells (EPIs), endothelial cells (ENDs), nonendothelial mesenchymal cells (MESs), and lung resident and migrating immune cells, each distinguished by specific cell-surface proteins thus allowing intact cell sorting by flow cytometry (9, 15). Although numerous cell subtypes are found within each of these four major subpopulations of pulmonary cells, the enrichment of subpopulations from dissociated mixed lung cells enhances the study of related cells by multiplexed cellular and molecular analyses. Evaluation of differential gene expression, metabolism, and function in subpopulations of cells provides a framework for elucidation of the many mechanisms influencing pulmonary development and disease. Hence, we describe a novel method to isolate simultaneously and enrich most lung cells, divided into four major components based on distinct membrane protein expression patterns using fluorescence activated cell sorting. The one-step sorting method excludes dead cells and erythrocytes and enables consistent isolation of cells with high viability during repeated testing.

MATERIALS AND METHODS

Study population.

Donor lungs were provided through the federal United Network of Organ Sharing via National Disease Research Interchange and International Institute for Advancement of Medicine. Dissociated lung cells from eleven deceased donors (4 girls, 7 boys), recovered with brief warm ischemic time as for transplantation, and entered into the LungMAP program were assessed in this study (Tables 1 and 2). The University of Rochester Institutional Review Board approved and oversaw this study (RSRB00047606).

Table 1.

Donor demographics

Donor	Age	Sex	Cause of Death	Pathologist’s Review
D019	1 day	M	Anencephaly	Normal structure and development; Aspirated squames/meconium and autolysis of bronchial epithelium.
D038	1 day	F	Anencephaly	Normal lung growth and development.
D041	6 day	M	Hypoxic ischemic encephalopathy	Normal growth; Small foci of increased macrophages and rare neutrophilic aggregates.
D044	6 day	M	Hypoxic ischemic encephalopathy	Normal alveolar development; Mild aspiration of squames, increased alveolar macrophages; Small vessel medial hypertrophy and mild muscularization.
D024	4.5 mo	M	Traumatic brain injury	Mildly deficient RAC (5.9 versus 7 expected for age); Mild inflammation with few clusters of lymphocytes and neutrophils; Increased alveolar macrophages.
D008	20 mo	F	Traumatic brain injury ($P{a}_O{}_{{}_2}$ 131†)	Normal structure and development; Patchy mild macrophage accumulation.
D011	21 mo	F	Brain injury, drowning, min pulm effects ($P{a}_O{}_{{}_2}$ 488*)	Normal structure and development; Patchy mild macrophage accumulation/inflammation.
D043	22 mo	M	Traumatic brain injury ($P{a}_O{}_{{}_2}$ 299*)	Normal alveolar structure; Moderate to severe bronchopneumonia; squamous epithelial metaplasia.
D022	29 mo	M	Traumatic brain injury ($P{a}_O{}_{{}_2}$ 191†)	Normal alveolar structure and development; Mild mixed inflammation with some eosinophils; Few arteries with medial hypertrophy.
D018	3 yr	F	Brain injury, drowning, min pulm effects ($P{a}_O{}_{{}_2}$ 550*)	Reduced lung weight but normal RAC; Aspiration with mild inflammation.
D036	8 yr	M	Traumatic brain injury ($P{a}_O{}_{{}_2}$ 131†)	Normal structure and development; Mild to moderate bronchopneumonia with aspiration; Focal eosinophilic inflammation; Increased alveolar macrophages.

F, female; M, male; $P{a}_O{}_{{}_2}$, partial pressure of oxygen; RAC, radial alveolar count.

^*Last $P{a}_O{}_{{}_2}$ in arterial blood on 100% oxygen;

^†last $P{a}_O{}_{{}_2}$ in arterial blood on 50–60% oxygen.

Table 2.

Lung warm and cold ischemia times

Donor	Type of Death	Warm Ischemia Time, h	Cold Ischemia Time, h
D019	Circulatory death	2.93	33.77
D038	Circulatory death	0.58	30.47
D041	Circulatory death	0.17	15.55
D044	Circulatory death	0.32	29.25
D024	Brain death	—	30.82
D008	Brain death	—	31.45
D011	Brain death	—	14.27
D043	Brain death	—	5.98
D022	Brain death	—	30.75
D018	Brain death	—	17.98
D036	Brain death	—	32.13

Reagents.

The tissues were digested in Dulbecco’s phosphate buffered saline (DPBS; Biowhittaker, Walkersville, MD) with 10 mM HEPES (Gibco, Gaithersburg, MD), 150 mM NaCl, 5 mM KCL, 1 mM MgCl₂, 1.8 mM CaCl₂ (each Sigma, St. Louis, MO), 2 mg/ml (0.3 units/ml) collagenase type A (Roche, Nutley, NJ), 1 mg/ml (10 units/ml) dispase II (Gibco), 0.5 mg/ml (1.5 units/ml) porcine pancreas elastase (Worthington Biochemical, Lakewood, NJ), and 2 mg/ml (800 units/ml) bovine pancreas deoxyribonuclease-I (DNase-I, Sigma). Proteases were neutralized in DPBS containing 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA; low endotoxin lot). The ammonium-chloride-potassium (ACK) buffer was from Biowhittaker. Antibody staining was performed in DPBS containing 1% wt/vol BSA (Millipore, Kankakee, IL). Antibodies used include CD31 BV605 (clone WM59), CD45 V450 (clone HI30), CD144 FITC (clone 55–7H1), CD235a PE-Cy5 (clone GA-R2), and 7-aminoactinomycin D (7-AAD) (each BD Bioscience, San Jose, CA), CD326 [epithelial cell adhesion molecule (EpCAM)] PE (clone 1B7, eBioscience, Waltham, MA), and Podoplanin AF647 (clone NC-08, Biolegend, San Diego, CA).

Lung cell dissociation.

The right-upper and right-middle lobes, beginning at the lobar bronchus, were dissected to separate airway and lung tissue, each cut into small pieces. Dissected tissues were then dissociated in 10 ml of prewarmed 37°C digestion cocktail in GentleMACS C tube on a GentleMACS Tissue Octo Dissociator [(Miltenyi Biotech, Bergisch Gladbach, Germany) 8 g of tissue/tube with 10 ml cocktail, up to 56 g total from each donor], using manufacturer-set dissociator program “mouse tumor implant program-01.01.” After 1-h incubation at 37°C, proteases were neutralized by adding 10 ml of ice-cold DPBS with 10% FBS. Cells were then passed through a sterile 100-μm cell strainer (Fisher Scientific, Agawam, MA) facilitated by massage with injection syringe plungers (BD, Franklin Lakes, NJ). Erythrocytes were lysed by incubating cells in 10 ml ACK buffer for 5 min at room temperature. After red blood cell (RBC) lysis, cells were washed in DPBS containing 10% FBS. Cells were counted by hemocytometer and viability was assessed by trypan blue exclusion. Some dissociated cells were used for tissue culture and other experiments immediately after dissociation while the rest were frozen in 10% DMSO (Sigma), 90% defined fetal bovine serum (50 × 10⁶ cells/ml). Cells were slow cooled in Nalgene Mr. Frosty (Sigma) at −80°C overnight before transfer to liquid N₂ vapor for long-term storage.

Fluorescence-activated cell sorting.

The step-by-step staining and sorting procedure is available on https://www.lungmap.net/resources/sop-search-page (SOP 705.3). Frozen mixed-lung cells were thawed, washed, pelleted, and resuspended in 1 ml staining buffer. Postfreeze cell counts and viability were determined by hemocytometer and trypan blue exclusion. Cells were then incubated in staining buffer with 2% normal mouse serum (Sigma, 5 min over ice) to block nonspecific Fc-receptor binding. Because we were staining live cells, we avoided Fc-receptor blocking using human serum to avoid specific Fc-receptor binding on leukocyte subsets and subsequent cellular activation. Cells were washed after Fc-receptor blocking and incubated at a final concentration of 1 × 10⁶ cells per 10 μl of staining cocktail containing analyte-specific antibodies and 7-AAD for 80 min at 4°C, protected from light, rewashed, and resuspended in DPBS with 1% BSA before passage through a 100-μm strainer (Falcon, Corning NY). Final mixed cells were suspended at 30 × 10⁶ cells/ml before sorting. Antibody-stained but not sorted cells [presort mixed cells (PMX)] were saved as controls.

Cell sorter.

Cells were sorted using a 4-laser, 18-color FACSAria flow cytometer (Becton Dickenson, San Jose, CA) in a Baker BioPROTECT hood (Baker, Sanford, ME) at the University of Rochester Medical Center shared-resource facility. Detailed description of the flow sorter laser and detector settings are available at https://www.lungmap.net/resources/sop-search-page/ (SOP 705.3). To ensure consistent instrument performance, the flow cytometer was calibrated with Peak 6 Rainbow Calibration Particles (Spherotech, Lake Forest, IL). A detailed description of calibration methods was previously published (24).

Cell sorting.

Antibody-stained dissociated cells were enriched into four major lung cell populations (Fig. 1). Unstained cells were used to establish cell size and background fluorescence. Both unstained and heat-killed (70°C for 30 min) 7-AAD-stained cells (100,000 cells/300 μl staining buffer) were run to establish an exclusion gate for dead cells. Single antibody-stained Simply Cellular compensation beads (Bangs Laboratory, Fishers, IN) were used to set spectral overlap compensation. Sorts were performed at 4°C using an 85-μm nozzle with a flow-rate of ~36 × 10⁶ events/hour. Sorted cells were collected in capped and sterile BSA-precoated collection tubes (VWR, Radnor, PA). Sorted cell populations were centrifuged at 800 g and 4°C for 10 min to collect pellets.

Fig. 1.

Detection and enrichment of pediatric lung cell populations by fluorescence-activated cell sorting. Schematic representation of the lung cell sorting strategy (A) and sorting template (B) are shown. 7-AAD, 7-aminoactinomycin D. A dead cell (7-AAD⁺) and erythrocyte (CD235a⁺) gate was determined on control heat-killed cells (green dots) to exclude these from the total cell populations (B, i and ii). CD45⁺ mixed immune cells (MICs) were collected from RBC-depleted live cell populations (Biii). The CD31⁺CD144⁺ endothelial cells (ENDs) were enriched from remaining CD45⁻ cells (Biv). From the leukocyte and endothelial cell depleted cells, EpCAM⁺ cells (EPI) were collected as epithelial cells (Bv). Residual live cells lacking any of the above membrane proteins were collected as nonendothelial mesenchymal cells (MESs). Cells unexposed to antibody (unstained, blue contours) are compared by overlay to antibody-incubated cells (red contour plots). In these presort plots, percentages of gated cell populations are provided as compared with total mixed cells. Viability and percentages of identified EPI, END, MES, and MIC populations were consistent in mixed-cell aliquots of the same donors, arranged left to right by age, sorted independently over time (C). CD, cluster of differentiation; EpCAM, epithelial cell adhesion molecule; D, donor; RBC, red blood cell; T1α, lung type I cell membrane-associated glycoprotein (podoplanin).

RNA extraction, real-time quantitative RT-PCR.

RNA was extracted from washed (DPBS, 800 g, 4°C, 10 min), sorted cells using the Qiagen RNeasy Microkit (Valencia, CA). RNA quantity and quality, indicated by RNA integrity number (RIN) and gel analysis, were measured by Agilent Bioanalyzer 2100 (Agilent, Santa Clara, CA). cDNA was prepared by iScript cDNA Synthesis kit (Bio-Rad). Quantitative PCR was performed with SYBR Green chemistry (ThermoFisher Scientific) and gene expression levels were calculated relative to cyclophilin A (Ppia) using ddCT methods as described previously (6). Primer sequences were obtained from PrimerBank (https://pga.mgh.harvard.edu/primerbank).

RNA sequencing.

RNA sequencing (RNAseq) was performed on PMX and sorted cells with 1 ng of total RNA per sample amplified using SMARter Ultra Low amplification kit (Clonetech, Mountain View, CA) as described previously (24). Reads were aligned using the Splice Transcript Alignment to a Reference algorithm (12), and expression values were summarized using high-throughput sequencing (1). Transcripts detected are provided on LungMAP.net and the Lung Gene Expression Anaylsis Web Portal (13).

Statistical analysis.

Statistical analyses were performed using GraphPad Prism software (GraphPad, La Jolla, CA). Central tendency values are given as means ± SE. The Mann-Whitney U-test was used to determine significant variance between parameters. Significance is marked when P < 0.05.

RESULTS

Dissociation of pediatric human lung cells.

Different combinations and concentrations of digestive enzymes and GentleMACS programs were tested with different incubation times to optimize yield and viability (data not shown). Applying our standardized multiprotease (collagenase, dispase, and elastase) digestion protocol to the combined right upper and middle lobes of donor lungs resulted in an average yield of 132.2 ± 18.7 million cells per gram of lung tissue. Freshly dissociated lung cells showed high viability: 97.6 ± 0.4% (by trypan blue exclusion). The isolated cells were expanded as EPI, stromal, and END cell cultures using selective growth media. Formation of confluent, resistant monolayers (≥300 Ω·cm²) and enhanced airway epithelial differentiation was demonstrated at air-liquid interface (Wang et al., unpublished observations). Stromal cells capable of advanced serial passage and differentiation into adipocyte, chondrocyte, and osteoblast lineages were detected (Howell et al., unpublished observations). Functional response to endotoxin was demonstrated with release of TNF-α and IL-1 β, ~5 ng/ml and 2.5 ng/ml, following 24-h culture in 1 µg/ml LPS. Dissociated cells frozen in liquid nitrogen were highly viable (97.9 ± 0.9% by trypan blue and confirmed by cytometry) upon thawing and were used to isolate lung cell subpopulations by fluorescence-based sorting.

Viable cells were enriched based on their unique membrane protein profiles.

Cell type-specific membrane proteins were chosen to distinguish and selectively label cells with specific antibodies. The sort staining and gating strategies used are outlined in Fig. 1. To ensure high viability and to exclude lysis-resistant nucleated RBCs found in neonates (10), 7-AAD⁺ dead cells, and CD235a⁺ erythrocytes were detected and subsequently excluded. Percentage of viable nonerythrocytes in the thawed mixed-cell population isolated by flow cytometry was 96.7 ± 0.2%, n = 56. From viable RBC-depleted cells, mixed immune cells (MICs) were identified by membrane CD45 expression (Fig. 1B). Percentage of leukocytes varied from donor to donor, but yields were consistent from mixed cells from the same donor sorted repeatedly (Fig. 1C). ENDs were selected by expression of CD31 and CD144. Knowing that some leukocytes, including recent thymic emigrants, express CD31 (14) the END populations were collected only after removal of CD45⁺ leukocytes. From the nonleukocyte, nonendothelial cell fractions, EPIs were identified by membrane EpCAM and collected. Selecting out dead cells, RBCs, leukocytes, and END and EPI populations leaves the 7-AAD⁻CD45⁻CD31⁻CD144⁻EpCAM⁻ cells that were then collected as enriched nonendothelial MESs. To ensure consistency, well-standardized cell processing and staining protocols were developed and followed. The protocols are available on https://www.lungmap.net/resources/sop-search-page. Calibrating the sorter with Peak 6 calibration particles before running each sort ensured consistent instrument performance. Multiple sorting experiments from the same donor tissues yielded consistent results (Fig. 1C). Adding DNase (10 μg/ml) to staining and wash buffers enhanced yields, reducing cell adhesion and clumping. After the sort, cells showed high enrichments of desired cell populations with >99% viability (Fig. 2, A and B). Approximately 7–9% of the sorted cells no longer fell in the gates defined for each population. These cells were without fluorescent marker and, we believe, are most likely to be the result of photobleaching occurring on the pass through the sorter and less likely to be contaminating cells of another type.

Fig. 2.

Flow cytometry on postsort samples confirm highly enriched four major subpopulations of dissociated lung cells. A: postsort density plots show enrichment of EpCAM⁺ (EPI), endothelial (END), nonendothelial mesenchymal (MES), and mixed immune (MIC) cells. B: bar graphs show mean postsort viability (closed bars) and purity (open bars) from multiple sorting experiments (n = 10, data shown as means ± SE). C: RNA integrity numbers (RINs) were >7 in all isolated-cell RNA. D: RNA yield (ng) per million cells is demonstrated. Significantly higher RNA yield was obtained from MICs compared with other cell types. C and D represent median, 95% of median and range; *P < 0.05 and **P < 0.01 by t-test, n = 11 per cell type. 7-AAD, 7-aminoactinomycin D; CD, cluster of differentiation; EpCAM, epithelial cell adhesion molecule; PMX, antibody-stained presort mixed cells; T1α, lung type I cell membrane-associated glycoprotein (podoplanin).

Comparison of RNA quality obtained from sorted cell populations and not-sorted mixed cells.

To determine the usefulness of sorted cells for further scientific studies, we compared the quality of extracted RNA from sorted cells with that of parallel-processed, antibody-stained, but not sorted cells (PMXs, 2 × 10⁶ cells saved before each sorting, Fig. 2C). RINs from sorted and PMXs were all above 7, although sorted EPIs had slightly decreased RIN compared with PMXs, MESs, and MICs (Fig. 2C). Notably, MIC RNA yield per million cells was significantly higher than from other sorted cell populations (Fig. 2D). The median RNA yield per million cells in MICs was also higher than that of PMXs, although it did not achieve statistical significance.

Population-specific gene expression.

RNA from sorted cell populations was assessed for population-specific gene expression by quantitative PCR and bulk RNAseq to investigate the level of enrichment and extent of purity. We demonstrated marked fold enrichment of the mRNA of the primary membrane protein markers, EpCAM, CD31, CD144, and CD45 used in the sort strategy, compared with nonsorted, antibody-stained mixed cells (PMXs), as well as of RNAs generally found to be cell type-specific, including EPI-specific surfactant protein C, club cell secretory protein (secretoglobin family 1A member 1), receptor for advanced glycosylation end products, MIC-specific CD14, and END-specific von Willebrand factor (Table 3). Platelet-derived growth factor receptor, smooth muscle actin (ACTA2), and elastin RNAs were enriched in MES subpopulations.

Table 3.

Fold increase RNA expression in sorted EPI, END, MES, and MIC cell populations compared with PMX, dissociated cells by qRT-PCR

	EPI	END	MES	MIC
EpCAM	37.3	0.3	0.6	0.1
SFTPC	77.6	0.02	0.1	0.4
SCGB1A1	47.5	0.2	0.2	0.06
RAGE/AGER	11.9	0.98	1.7	0.32
CD45	<0.01	<0.01	0.05	2.72
CD14	0.08	0.36	0.12	2.29
CD31/PECAM	0.03	5.14	0.05	0.36
CD144/VE-Cadherin	0.02	9.6	0.02	0.05
VWF	0.02	15.8	0.03	0.04
ACTA2	4.76	0.19	7.81	0.06
PDGFRβ	4.5	0.22	10.95	0.05
ELN	7.93	1.57	11.66	0.03

ACTA2, smooth muscle actin; CD, cluster of differentiation; ELN, elastin; END, endothelial cell population; MIC, mixed immune cell population; EpCAM, epithelial cell adhesion molecule; EPI, epithelial cell population; MES, nonendothelial mesenchymal cell population; PDGFR, platelet-derived growth factor receptor; PECAM, platelet endothelial cell adhesion molecule; PMX, presort mixed cells; qRT-PCR, quantitative RT-PCR; RAGE/AGER, receptor for advanced glycation endproducts/advanced glycosylation end-product specific receptor; SCGB1A1, secretoglobin family 1A member 1; SFTPC, surfactant protein C; VE-cadherin, vascular endothelial cadherin; VWF, von Willebrand factor.

The quantity and quality of RNA recovered from sorted cell types were further assessed by high-throughput RNAseq. RNAseq was performed on RNA obtained from each cell type (EPI, END, MES, and MIC). Libraries generated 24.8 ± 4.9 million raw sequence reads at a targeted sequencing depth of 10 million reads (Fig. 3A). Mapping rates (uniquely mapped sequences) across all cell types averaged at 90% and were similar within all individual cell types (EPI: 89.6.  ± 3.4; END: 91.3 ± 1.7; MES: 90.7 ± 2.0; and MIC: 90.9 ± 1.6; percent of genome mapped reads; Fig. 3B). Total number of counts mapped to the reference genome for all samples was 17.9 ± 4.2 million with little difference among cell types (EPI: 16.0 ± 3.2; END: 18.1 ± 4.7; MES: 18.4 ± 4.6; and MIC: 19.4 ± 3.7; million; Fig. 3C). Genomic coverage averaged 47% (proportion of genes expressed at the mRNA level) and was similar in all cell types (EPI: 48.6 ± 3.2; END: 44.8 ± 2.5; MES: 48.3 ± 4.6; and MIC: 46.1 ± 6.9; percent of genome mapped; Fig. 3D).

Fig. 3.

RNA sequencing (RNAseq) quality and purity. The total number of sequencing reads (A), proportion of uniquely mapped (B), total transcript count per sample (C), and proportion of the genome detected as expressed transcripts (D) for samples grouped by cell type. Normalized expression intensity of standard, relatively cell-specific marker transcripts for each of the four cell types is shown (E). Columns represent individual samples clustered by cell type. Each cell type is arranged in ascending order by age of donor (left to right). Expression patterns for individual genes are shown in rows. High expression is shown in red and low in green as indicated by the scale bar. END, endothelial cell; EPI, epithelial cell; MES, mesenchymal cell; MIC, mixed immune cell.

Normalized counts for selected cell type-specific markers were found at expected levels in each cell type (Fig. 3E), further validation of the isolation of sufficient intact RNA for high-throughput analysis with retention of the enrichment of individual populations.

DISCUSSION

In this report, we describe novel methods to enrich and isolate four cell populations simultaneously from pediatric human lung of suitable quality for further analyses. Obtaining dissociated lung single-cell suspensions and isolating one or two particular cell types have been previously described (2, 5, 8, 11, 16, 17, 26, 29). Here we present a technique to 1) efficiently isolate single-cell suspensions from lung lobes with high yield and viability and then 2) enrich four major distal lung cell subpopulations based on distinct patterns of membrane protein expression. We utilized this unique expression pattern to collect the desired cells after excluding 7-AAD-positive dead cells and CD235a-expressing RBCs. Enzymatic digestion of lung tissue to obtain single-cell suspension is known to affect cellular yield and loss of functionally important membrane proteins (2, 4). It may also cause activation of certain cell types (20). So, we optimized our enzymatic digestion protocol to get maximum viability and yield while avoiding exposure to high concentrations of proteases for prolonged times. Our sorting method enriched viable cell populations with intact RNA, as well as repeated cell proportion consistency within the donor, the latter suggestive of age-dependent trends in relative cell composition of the lung with reduced ENDs and increased immune cells with age (Fig. 1C). Quantitative RT-PCR of known population-specific transcripts is consistent with purity of the enriched cell populations. RNA sequencing of the sorted cells further substantiates the subpopulation authenticity.

Enzymatic digestion of lung tissue to isolate individual cells using different combinations of proteases has been reported by multiple research groups. The human lung is a complex organ with many different cell types. Yet most described dissociation methods are developed to isolate particular cell populations of the investigator’s interest. Although some groups described dissociation techniques useful for isolating EPIs (8, 17) or ENDs (18, 23), others used different protease combinations to obtain immune cells without cleaving lineage-specific membrane markers (11, 25). However, some recent reports have described simultaneous isolation of different subpopulations of pulmonary cells (2, 5, 16). Our objective was to develop a standardized method for isolating the majority of present lung cells but with an approach to enrich all four major subpopulations to enhance down-stream analyses. Our goal was to obtain neonatal and pediatric lung cells with consistent high viability without compromising yield that would be used for bulk RNA or single-cell RNA transcriptomic studies as well as development of in vitro models and other investigations to understand the biology of human lung development better. To develop the LungMAP repository, we calibrated a tissue digestion and cell sorting method suitable for concurrently young lungs, isolating major subpopulations with maximum yields and viability. We used GentleMACS Tissue Octo Dissociator to obtain a similar degree of mechanical tissue disruption every time, shortening time needed for isolation and reducing variability because of the human factor. Because different types of lung cells and extracellular matrixes exhibit differential levels of susceptibility to these digestive enzymes, no combination of currently available enzymes optimally releases all lung cell types without cell injury, a limitation noted by other researchers (26). As previously observed by Quatromoni et al. (26), we found that combinations of multiple enzymes at lower concentrations more efficiently produces better yield and viability than any single protease alone even in higher doses (data not shown). Our digestion cocktail consists of collagenase A, with normal balanced ratio of clostripain, tryptic, and protease activities, elastase, to disrupt abundant lung elastins and dispase II, a neutral protease. DNase-I is also added to the cocktail to neutralize extracellular DNA that causes cell clumping. Addition of dispase and elastase to the digestion cocktail increased yield, especially of EPIs, and viability (data not shown). To isolate specific cell populations from the mixture of dissociated cells by flow cytometry, we first excluded dead cells and erythrocytes that were not removed by ACK lysis. One of the challenges of working with neonatal tissue is removing residual nucleated RBCs that are resistant to ACK-mediated lysis. This is also the primary reason for adding an anti-glycophorin-A (αCD235a) antibody in our sorting panel exclusion channel, the channel that also excludes 7-AAD⁺ dead cells. Specific cell populations were then isolated based on their unique membrane protein expression pattern. Currently, the lack of a reliable MES surface marker protein remains a limitation of this method that may be overcome as candidates are identified by proteomic analysis of the MES subpopulation.

Our method of fluorescence-activated sorting of all four major cell populations from dissociated neonatal or pediatric lung cells provides a reliable and consistent technique to isolate cells for further comparative transcriptomic, proteomic, metabolic, and functional study. We have successfully employed these methods to obtain viable, specialized lung cell populations with high consistency to build and maintain a lung cell and tissue repository for the Molecular Atlas of Lung Development Program (3) that is now available to researchers for further studies.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Molecular Atlas of Lung Development Program Human Tissue Core Grants U01-HL-122700 (to G. Deutsch, T. Mariani, and G. Pryhuber), U01-HL-122642 (to S. Potter and J. Whitsett), and U01-HL-122638 (to R. Clark and S. Palmer).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.B., R.S.M., T.B., J.H.-W., P.K., G.D., T.J.M., and G.S.P. conceived and designed research; G.B., H.L.H., R.S.M., Q.W., J.M., J.L., J.R.M., M.C., P.K., G.H.D., and G.S.P. performed experiments; G.B., H.L.H., R.S.M., S.B., J.A.L., J.R.M., J.A., M.C., J.H.-W., P.K., G.D., J.A.W., Y.X., T.J.M., and G.S.P. analyzed data; G.B., H.L.H., R.S.M., S.B., Q.W., J.M., J.L., J.R.M., J.A., T.B., M.C., P.K., G.H.D., J.A.W., Y.X., T.J.M., and G.S.P. interpreted results of experiments; G.B., H.L.H., and S.B. prepared figures; G.B. drafted manuscript; G.B., H.L.H., R.S.M., S.B., J.L., J.R.M., J.A., T.B., J.A.W., Y.X., T.J.M., and G.S.P. edited and revised manuscript; G.B., H.L.H., R.S.M., S.B., Q.W., J.M., J.L., J.R.M., J.A., T.B., M.C., J.H.-W., P.K., G.D., J.A.W., Y.X., T.J.M., and G.S.P. approved final version of manuscript.

ENDNOTE

Further repository information and requests for access are available at http://www.brindl.urmc.rochester.edu/.