Lung Single-Cell Transcriptomics in Alpha-1 Antitrypsin Deficiency

To the Editor:

Alpha-1 antitrypsin deficiency (AATD) is associated with an increased risk for emphysema in adults as a result of the reduced opposition to lung proteases such as neutrophil elastase. Severe AATD is usually caused by two copies of the (PI)*Z allele of the SERPINA1 gene (PI*ZZ genotype of serpin family A member 1). Heterozygous carriers of the PI*Z allele (PI*MZ) have an increased risk of airflow obstruction, shown particularly among cigarette smokers (1), though not as high as homozygous individuals. The AATD gene expression signature across lung cell types is not fully understood. This study sought to examine this signature using publicly available single-cell RNA-sequencing (scRNA-seq) data.

Lung scRNA-seq data from a study of chronic obstructive pulmonary disease (COPD) by Sauler and colleagues (2, 3) were obtained from the Gene Expression Omnibus database (GSE136831) and the Short Read Archive (SRP395406). Phenotypic data for each subject included age, sex, race, and ever-smoker and COPD status (3). The focus was on data from the 17 former smokers with COPD. Using the scRNA-seq read files from the Short Read Archive as the source of genetic data, the SNP rs28929474 provided the SERPINA1 Z allele, and the S (rs17580), I (rs28931570), and F (rs28929470) alleles were also examined to determine the SERPINA1 genotype (4). Cell types with zero cell counts for any of the MZ and ZZ subjects were excluded. The cleaned expression data were collapsed into gene-by-subject matrices for each cell type, creating a pseudobulk dataset. Associations between pseudobulk gene expression and SERPINA1 Z allele dosage were tested using DESeq2 (5) with the covariates age and sex. Pathway analyses were performed for results in each cell type using Gene Set Enrichment Analysis and the hallmark gene sets from the Molecular Signatures Database (6). Networks were created to highlight cell–cell relationships based on shared canonical pathway activity. See data supplement for additional details regarding the study methods.

Two subjects were found to be heterozygous for the S allele (MS genotype) and were removed, leaving 11 MM genotype subjects, 1 MZ subject, and 3 ZZ subjects (Table E1 in the data supplement). After data cleaning, scRNA-seq data were available for 38,387 cells and 41,240 genes across 15 subjects and 29 cell types (Tables E2 and E3). Higher SERPINA1 expression was observed in macrophages, alveolar epithelial type II cells, and monocytes in the bar plot of mean expression across cell types (Figure E1) and when comparing the Uniform Manifold Approximation and Projection plots with cell type labels or SERPINA1 expression shading (Figure E2).

The largest number of differentially expressed genes (false discovery rate lower than 0.05) in the tests of association between the pseudobulk expression and the Z allele dosage (Tables E4–E32) was in alveolar macrophages (72 genes; Tables 1 and E33). Volcano plots were created using the differential gene expression results for alveolar macrophages and interstitial macrophages (Figures 1A and 1B). The top results for alveolar macrophages included the genes CXCL8 (C-X-C motif chemokine ligand 8), CXCL2 (C-X-C motif chemokine ligand 2), and IL6 (interleukin 6). Intersecting the differential gene expression results with the COPD case-control findings by Adams and colleagues in this scRNA-seq dataset (3), only the ARF6 (ADP ribosylation factor 6) gene was found to be significant in both alveolar macrophage analyses, suggesting a possible overlap in disease-relevant vesicular trafficking. For interstitial macrophages, the top differentially expressed gene was FOS (Fos proto-oncogene, AP-1 transcription factor subunit) (Table 1 and Figure 1B). The genes IL6 (Figure E3) and IL1B (interleukin 1 beta) (see Table E5) were also differentially expressed in interstitial macrophages. Although not statistically significant at a false discovery rate lower than 0.05, we observed nominal differential expression (P = 0.009) of SERPINA1 in goblet cells (see Table E11).

Table 1.

Summary of Significant Differential Gene Expression Results for SERPINA1 Z Allele Dosage

Cell Type	Total Genes	No. of Genes with P < 0.05	Genes with Adjusted P < 0.05
Alveolar macrophage	9,859	920	72 genes*
Interstitial macrophage	3,995	303	26 genes*
Nonclassical monocyte	11,117	248	KLF6, ZFP36, MIR222HG, ENSG00000267519, ENSG00000274213
Classical monocyte	4,120	241	MT2P1, MTCO1P40, KLF6, ANKRD28, MBNL1, MT1E, ADD3, JUND, CAB39, CREM, ENSG00000267519, ENSG00000274213
Goblet	8,880	183	FOXO1
Conventional type 1 dendritic	10,791	228	GNG2
Natural killer	14,472	541	MT2P1
B	14,324	414	FOS, ABCA1
Myofibroblast	10,370	133	ENSG00000264281

^*See Figures 1A and 1B and Tables E4 and E5.

Figure 1.

Cell type–specific differential gene expression by Z allele dosage: volcano plots of the results for alveolar macrophages (A) and interstitial macrophages (B) and bipartite network representation of the hallmark pathway enrichment findings across all cell types in upregulated (C) and downregulated (D) results, with node size proportional to the number of edges (pathways indicated by red squares, cell types by blue circles). AT1 = alveolar epithelial type I cells; AT2 = alveolar epithelial type II cells; cMonocyte = classical monocytes; ncMonocyte = nonclassical monocyte; NK = natural killer cells; T_Cytotoxic = CD8 T cells; T_Regulatory = CD4 T cells, VE_Venous = vascular endothelial venous cells from Sauler and coworkers (2).

In the networks (Figures 1C and 1D) and heat maps (Figures E4 and E5) created using the pathway enrichment results in each cell type, upregulation of several pathways was observed in alveolar macrophages, in particular the TNFA (tumor necrosis factor-α) signaling pathway (TNFA_SIGNALING_VIA_NFKB) and INFLAMMATORY_RESPONSE pathways. The TNFA pathway was also upregulated in several other cell types, including alveolar epithelial cells and monocytes. Although not accompanied by significant differential expression among individual genes, several pathways were enriched among genes downregulated in T cells, including IFN-γ and -α in CD4⁺ T cells and unfolded protein response and oxidative phosphorylation pathways.

Alpha-1 antitrypsin has antiinflammatory properties (7). The role of TNFA signaling in AATD inflammation and COPD (8) was supported by the observed TNFA pathway upregulation with Z allele dosage across several cell types. Alveolar macrophages were observed to have the highest SERPINA1 expression and also had the highest TNFA pathway and gene upregulation with Z allele dosage, aligned with their overall role in AATD (9). IL-6 and CXCL8 were upregulated and are proinflammatory cytokines capable of recruiting neutrophils. Increased expression of NFKBIZ (NFKB inhibitor zeta) and the TNFA pathway may also reflect altered proportions of alveolar macrophage subsets that are skewed toward a proinflammatory spectrum. The presence of CD83, an activation marker that was recently reported as an inflammation checkpoint molecule (10), in the top results suggests that it may be of interest in future experiments. Although AATD does not directly affect T cells (11), differential pathway activity with Z dosage was observed, adding insight to the observations of increased lymphocytes in AATD COPD (12). Z allele dosage may impact gene expression directly in each cell type. However, we expect the transcriptomic signatures of AATD across cells to also be significantly influenced by the proinflammatory state as a result of lower levels of functional SERPINA1 with a corresponding increase in neutrophil elastase activity, as well as the proinflammatory effects of Z-AAT protein. The TNFA pathway upregulation across several cell types and the T-cell pathway findings are putative examples of such signatures.

Limitations of this study include the small populations of MZ and ZZ subjects, limiting the power to detect differentially expressed genes, and the lack of neutrophils, likely due to biases in cell dropout in single-cell sequencing (13). An enrichment of macrophages was present in the data, as outlined by Sauler and colleagues (2). Although we sought to minimize the effects of outlying gene expression, higher and more consistent cell counts across all subjects would be beneficial and will be sought in future studies. When assessing the SERPINA1 expression across cell types, the normalized data for each subject are compositional, so the levels should not be viewed as absolute. The inclusion of current smokers in future studies may provide better insight into the influence of the Z allele on the lung transcriptomic signatures of smokers.

This study has revealed genes and pathways relevant to AATD across lung cell types. We found proinflammatory changes such as enrichment of TNFA signaling, even in cell types with low SERPINA1 expression, demonstrating pervasive changes driven by AATD. Together, these findings may help improve the molecular understanding of AATD in the lung and inform future therapeutic avenues such as cytokine-related therapies (14, 15).

Some of the results of these studies have been previously reported in the form of a preprint (Zenodo, 29 Jan 2024; zenodo.org/records/10583664).