Identification of endothelial and mesenchymal FOXF1 enhancers involved in alveolar capillary dysplasia

Mutations in the FOXF1 gene, a key transcriptional regulator of pulmonary vascular development, cause Alveolar Capillary Dysplasia with Misalignment of Pulmonary Veins, a lethal lung disease affecting newborns and infants. Identification of new FOXF1 upstream regulatory elements is critical to explain why frequent non-coding FOXF1 deletions are linked to the disease. Herein, we use multiome single-nuclei RNA and ATAC sequencing of mouse and human patient lungs to identify four conserved endothelial and mesenchymal FOXF1 enhancers. We demonstrate that endothelial FOXF1 enhancers are autoactivated, whereas mesenchymal FOXF1 enhancers are regulated by EBF1 and GLI1. The cell-specificity of FOXF1 enhancers is validated by disrupting these enhancers in mouse embryonic stem cells using CRISPR/Cpf1 genome editing followed by lineage-tracing of mutant embryonic stem cells in mouse embryos using blastocyst complementation. This study resolves an important clinical question why frequent non-coding FOXF1 deletions that interfere with endothelial and mesenchymal enhancers can lead to the disease.

c, Sub-clustering of CAP1 (gCAP) cells indicates that these cells consist of four distinct cell clusters: sub-clusters 1, 2, 3 and 4. Sub-cluster 1 contains the majority of CAP1 cells (~60%).d, All CAP1 cells express Gpihbp1 and Kit that are well-defined markers for CAP1.e, Violin plots show that sub-cluster 1 expresses high levels of Scn7a and Sparcl1.Sub-cluster 2 expresses high levels of Tmem100 and Sema3c.Subcluster 3 is a highly proliferative cell subset as shown by the enrichment of Mki67 and Top2a.Adgrg6 and Fbln5 mRNAs are selectively expressed in sub-cluster 4. The alignment of snATAC data from multiome E18.5 sequencing to the publicly available bulkATACseq obtained from different developmental stages: E8.5, E14.5, E15.5, E16.5 and E18.5.There are 4 shared accessible regions in Foxf1 upstream sequences in addition to an accessible region in the Foxf1 promoter.The FEL1, FEL3 and FEL4 are accessible at late developmental stages, whereas FEL2 is accessible at early developmental stages.TSS is transcriptional start site.
Sub-clustering of matrix fibroblasts indicates that these cells consist of two cell clusters.b, The heatmap shows representative markers enriched in each cell subcluster.
The integrated projection of single-cell RNAseq library (10X v3 3' dual index) and single-nuclei GEX (RNA or gene expression) library from multiome sequencing.Both libraries were prepared from lungs of E18.5 embryos that were enzymatically digested to obtain single cell suspensions.GFP-expressing cells were FACS-sorted from Foxf1-GFP reporter mouse line.Cell clusters were identified using unsupervised clustering UMAP projection.b, The UMAP embedding of singlenuclei RNA sequencing (10X multiome GEX library).c, The UMAP embedding of single-cell RNA sequencing (10X v3 chemistry dual index library).d, Pie-plots compare the cellular distribution from 10X multiome GEX library and 10X v3 dual index library.e, Dot-plots show the consistency of clustering between the two libraries when cell cluster-specific markers are compared.
The sequence alignment of the distal regulatory element FEL4.The fasta sequences of the FEL4 enhancer in different species are downloaded using UCSC genome browser and the sequences were aligned using the ClustalW2 algorithm.The EBF-binding site is shown by the dark blue box.b, The sequence alignment of the distal regulatory element FEL2.The GLI-binding site is shown by the purple box.
Schematic representation of mutations to disrupt FOX and ETS motifs in FEL1 and FEL3 enhancers.b-c, The dual luciferase assay shows that both FOX-binding site and ETS-binding site contribute to the enhancer activity in FEL1 and FEL3 regulatory regions.The FOXF1 expression vector cooperates with either FLI1 or ERG expression vectors to activate FEL1 and FEL3, * is p<0.05.Source data are provided as a Source Data file.Regulation of FEL2 and FEL4 enhancers by GLI1 and EBF1.a, siRNA-mediated inhibition of Ebf1 does not change the FEL4 activity.b, Real-time RT-PCR analysis shows that a 50% inhibition of Ebf1 is not sufficient to reduce Foxf1 mRNA.c, The diagram shows the mutation to disrupt the putative GLI-binding motifs in the FEL2 enhancer.d, The luciferase assay indicates that the GLI1 expression vector significantly increases FEL2 activity.FEL2 activity is dependent on the GLI-binding site.The activity of FEL4 is not changed after transfection with the GLI1 expression plasmid, * is p<0.05,** is p<0.01,ns is not significant.Source data are provided as a Source Data file., curated subset (NM_*, NR_*, NP_* or YP_*) -Annotation Release NCBI RefSeq GCF_000001405.40-RS_2023_10 (2023-10-11) ENCODE Candidate Cis-Regulatory Elements (cCREs) combined from all cell types The alignment of 21 genomic 16q24 deletions in ACDMPV patients.Twelve genomic deletions do not affect the FOXF1 coding sequence, 9 genomic deletions disrupt both FOXF1 coding sequence and non-coding sequences.The vertical line marks the proximal FOXF1 promoter.b, The alignment of FOXF1 regulatory elements in 12 ACDMPV cases with non-coding genomic deletions.The previously reported 75kb shared deletion region (SDR) is shown by the black box.c, The diagram shows the relation of the previously characterized SRO (~60kb) regulatory region with newly identified FEL1, 2, 3 and 4 endothelial and mesenchymal enhancers.The FEL1 enhancer is shown as 2kb interval flanking the core sequence for better visualization.FEL2, 3, and 4 enhancers are shown as 1.5kb intervals.Endothelial enhancers (FEL1 and FEL3) are shown in blue.Mesenchymal enhancers (FEL2 and FEL4) are shown in green.The SRO region is shown in red.FEL1 is localized upstream of non-coding RNA LINC 01082 and overlaps with the SRO region.The integrative UMAP projection of GEX libraries from 10X multiome sequencing of ACDMPV patient P1 and patient P2.Similar cell sub-clusters are seen in both datasets.b, Schematic diagram shows the genomic deletion (grey) in P1 ACDMPV patient in the context of 16q24.1 region.The FOXF1 genomic region is highlighted by the orange box.The green box indicates the 75kb shared deletion region (SDR).The illustration shows an overlap of the FOXF1 genomic deletion in patient P2 which includes FEL2-4 but not FEL1.The annotated coordinates are derived from the human genome assembly GRCh38.c, The UMAP joint projection and unsupervised clustering of non-hematopoietic lung cells (corresponding to CD45negative cells in flow cytometry).d, The genome alignment of FOXF1 regulatory elements is performed using singlenuclei ATACseq from the multiome sequencing from patient P1.The accessibility of the FOXF1 proximal promoter region is detected, but the accessibility of endothelial regulatory elements FEL1 and FEL3 is undetectable.The alignment of FOXF1 regulatory elements is performed using single-nuclei multiome sequencing from healthy donor lung.The 20kb adjacent regions are shown for each regulatory element.The joint clustering of ACDMPV patient P2 multiome (GEX + ATAC) data.b, The featureplots of CA4, EDNRB, APLN (all are aCap markers) and KIT, GPIHBP1, APLNR (all are gCAP markers) show the absence of aCAP and gCAP endothelial cells in the ACDMPV dataset.c, The human FOXF1 genome region corresponding to mouse Foxf1 regulatory elements is identified using comparative genomic analysis from UCSC genome browser.For each region, 20kb sequences flanking the core regulatory region are shown.Clustering of non-hematopoietic cells in the snATACseq dataset from healthy human donor lung.The annotation of cell clusters is adapted from GSE161383 dataset.b, Human FOXF1 genome regulatory elements corresponding to mouse Foxf1 regulatory elements are identified based on the comparative genomic analysis.For each region, 20kb sequences flanking the core regulatory region are shown.c, Schematic shows the FEL2 enhancer, the size of which is ~400bp based on evolutionary conservation.The narrow peak harboring the potential GLI-binding site (65bp, orange box) is located within a region of increased chromatin accessibility in both pulmonary endothelial cells and lung fibroblasts.The diagram of the syntenic region flanking FOXF1 in human (top) and Foxf1 in mouse genomes (bottom).The genome organization is highly conserved upstream and downstream of Foxf1 mouse and FOXF1 human genes.b-e, Gins2 and Banp are two genes in the proximity to Foxf1 in both mouse and human.The open chromatin profile in Gins2 promoter is similar between human (b) and mouse genomes (d).The chromatin accessibility in the Banp promoter is similar between human (c) and mouse genomes (e).
CCTTGTTCGT TGTCACGGTA-3' 3'-GGAACAAGCA ACAGTGCCAT-5' 5'-GCTTTTTATC TGTCTTAGGA-3' 3'-CGAAAAATAG ACAGAATCCT-5' The schematic diagram shows the strategy to disrupt the FEL1 enhancer using the multiplex genome editing with CRISPR-Cpf1.b, Sequences show guide RNAs used to disrupt the FEL1 regulatory element.c, The chromatogram of Sanger DNA sequencing of the PCR fragment flanking FEL1 genomic fragment.Black line shows the genome break point after CRISPR genome editing.d, The detailed sequence of the upstream and downstream break point in genomic DNA from mutant ESCs.CCCATGTCCC TGAGACTCCT-3' 3'-GGGTACAGGG ACTCTGAGGA-5' 5'-AGTTGAGGCC AGTGTGGGCT-3' 3'-TCAACTCCGG TCACACCCGA-5' The schematic diagram shows the strategy to disrupt the FEL4 enhancer using the multiplex genome editing with CRISPR-Cpf1.b, Sequences show guide RNAs used to disrupt the FEL4 regulatory element.c, The chromatogram of Sanger DNA sequencing of the PCR fragment flanking FEL4 genomic fragment.Black line shows the genome break point after CRISPR genome editing.d, The detailed sequence of the upstream and downstream break point in genomic DNA from mutant ESCs.

b FEL1 FEL3 Supplementary Figure 6. DNA sequences of distal regulatory enhancers FEL1 and FEL3.
a EBF binding site and motif b GLI binding site and motif