Single-cell multiomic analysis identifies a HOX-PBX gene network regulating the survival of lymphangioleiomyomatosis cells

Abstract

Lymphangioleiomyomatosis (LAM) is a rare, progressive lung disease that predominantly affects women. LAM cells carry TSC1/TSC2 mutations, causing mTORC1 hyperactivation and uncontrolled cell growth. mTORC1 inhibitors stabilize lung function; however, sustained efficacy requires long-term administration, and some patients fail to tolerate or respond to therapy. Although the genetic basis of LAM is known, mechanisms underlying LAM pathogenesis remain elusive. We integrated single-cell RNA sequencing and single-nuclei ATAC-seq of LAM lungs to construct a gene regulatory network controlling the transcriptional program of LAM cells. We identified activation of uterine-specific HOX-PBX transcriptional programs in pulmonary LAM^CORE cells as regulators of cell survival depending upon HOXD11-PBX1 dimerization. Accordingly, blockage of HOXD11-PBX1 dimerization by HXR9 suppressed LAM cell survival in vitro and in vivo. PBX1 regulated STAT1/3, increased the expression of antiapoptotic genes, and promoted LAM cell survival in vitro. The HOX-PBX gene network provides promising targets for treatment of LAM/TSC mTORC1-hyperactive cancers.

INTRODUCTION

Lymphangioleiomyomatosis (LAM) is a rare metastasizing lung disease that almost exclusively affects women. Pulmonary manifestations of LAM are characterized by the proliferation of abnormal smooth muscle–like cells, leading to progressive cystic lung destruction. LAM is caused by deleterious mutations in TSC1 or TSC2 (1), which results in the hyperactivation of the mechanistic target of rapamycin complex 1 (mTORC1). This discovery led to clinical trials demonstrating that mTORC1 inhibitors (mTORi) stabilize lung function in most patients with LAM (2). However, challenges remain in optimizing treatments for patients with LAM. Because mTORi are cytostatic, some patients are poorly responsive. Moreover, because sustained treatment is required, side effects can develop secondary to chronic drug exposure (2, 3). The development of better treatments for LAM will require a more comprehensive understanding of the molecular pathogenesis of the disease.

LAM lesions are stimulated by female hormones and are thought to be metastatic, most likely originating from uterine cells (4–6). The most definitive evidence supporting a metastatic mechanism for the disease is the recurrence of LAM lesions composed of recipient cells in the donor allograft following lung transplantation (7). A uterine source of LAM is supported by predominance in women, histopathological characteristics of LAM lesions consistent with uterine smooth muscle–like cells, a gradient of axial lymph node involvement from pelvis to thorax (8), and multiple case reports and case series providing evidence of LAM lesions in the uterus (4, 5). Recently, we provided further evidence supporting the notion of a uterine source based on single-cell transcriptomic analysis of explant lungs from patients with LAM and identified a unique population of cells (termed LAM^CORE cells) that are readily distinguishable from endogenous lung cell types and share strong transcriptomic similarity with uterine cells (9). LAM^CORE cells selectively express a panel of uterine-specific transcription factors (TFs), including homeobox TFs HOXA10, HOXA11, HOXD11, PITX2, EMX2, and pre–B cell leukemia homeobox 1 (PBX1), known to play important roles in reproductive and endocrine system development and diseases (9). The TF PBX1 forms dimers with other HOX proteins to influence transcriptional regulation of target genes (10). Particularly relevant to the estrogen dependence of LAM, HOX/PBX dimers have been implicated in oncogenesis of steroid-responsive cancers and are of interest as potential therapeutic targets (11, 12).

To explore further the implications of PBX1 induction in LAM, in the present study, we used single-cell multiomic data to construct a LAM cell–specific gene regulatory network and performed biochemical and functional validation experiments using LAM models in vitro and in vivo. We identify the molecular interplay among HOX and PBX TFs in LAM and define a role for HOX/PBX1 in regulating signal transducer and activator of transcription 1/3 (STAT1/3) signaling to promote LAM cell survival and lung colonization.

RESULTS

Integrative single-cell analysis of gene expression and DNA accessibility identifies the activation of uterine-selective homeobox family of TFs in the LAM lung

Using single-cell RNA sequencing (scRNA-seq) of lung samples from patients with LAM, a unique population of cells, termed LAM^CORE, was identified, which was readily distinguishable from endogenous lung cell types and shared the closest transcriptomic similarity to uterine myocytes (9). In the present study, we generated paired measurements of single-cell chromatin accessibility and gene expression using 10x Multiome, single-nucleus assay for transposase-accessible chromatin with sequencing (snATAC-seq) and existing 10x scRNA-seq (9) to produce an integrated analysis of single-cell gene expression and chromatin accessibility in human LAM tissues. Mapping of integrated LAM single-cell data using the LungMAP CellRef (13), 33 different cell types were identified in LAM lung, including the previously identified LAM^CORE cell population (Fig. 1A). Integrative analyses enabled the identification of LAM cells, cell type–specific chromatin accessibility peaks, and enriched TF binding sites or motifs in a cell type–specific manner. We found that LAM^CORE cell–specific ATAC-seq peaks accessible regions were highly enriched in motifs for uterine selective homeobox family of TFs, including PBX1, PBX3, HOXA9, HOXA10, and HOXD11, Melanocyte Inducing Transcription Factor (MITF), STAT family (STAT1 and STAT3), Nuclear Receptor Subfamily 2 Group F Member 1, Nuclear Receptor Subfamily 2 Group F Member 2 (NR2F2), and TEA Domain Transcription Factor (TEAD)family (TEAD1, TEAD2, and TEAD3) (Fig. 1B). HOXD11, PBX1, MITF, and TEAD2 were predicted as signature genes with enriched expression in LAM^CORE cells (Fig. 1C, pink bar), while MYH11, ACTA2, STAT1, and STAT3 were expressed in multiple cell types, with their average expression levels increased in LAM^CORE cells (Fig. 1C, blue bar). MITF is a known oncogenic driver for kidney angiomyolipomas (AMLs) (14), and NR2F2 is a nuclear receptor recently identified by genome-wide association studies as a modifying genetic locus in LAM (15). After obtaining differentially accessible peak sets of LAM^CORE cells, Hypergeometric Optimization of Motif EnRichment (HOMER) was used to assign peaks to the nearest or overlapping genes, enabling the association of chromatin accessibility with gene expression. The nearest genes associated with LAM^CORE cell–selective peaks were functionally enriched in muscle structure development and extracellular matrix organization (Fig. 1D). Among the nearest genes, expression of uterine TFs (PBX1, PITX2, ESR1, and HAND2), genes encoding muscle and collagen proteins (ACTA2, MYH11, COL1A2, and COL4A1), and genes involved in apoptosis (CRYAB, PTGIS, and PDE3A) were increased in LAM (Fig. 1C). ATAC accessibility and RNA expression patterns across all cell types were highly concordant (Fig. 1E). We analyzed peak-to-gene (P2G) links in genomic regions of HOXA10 and PBX1 (Fig. 1F) and identified unique accessible peaks in the promoter regions of HOXA10 and HOXA11 in LAM^CORE cells, consistent with the selective expression of HOXA10 and HOXA11 RNAs in LAM^CORE cells (Fig. 1F). PBX1 is predicted as one of the signature genes of LAM^CORE cells, its gene expression and chromatin accessible peaks were detected in other cell types, and the selectivity of PBX1 in LAM^CORE cells was likely influenced by its dimerization with HOX proteins (Fig. 1F). Thus, the present integrative scRNA-seq/snATAC-seq data analyses identified an increased expression and activity of uterine-specific transcriptional programming (i.e., homeobox family of TFs) in the LAM^CORE cells, supporting the concept that activation of HOX-PBX signaling plays an important role in the transcription and epigenetic regulation of LAM^CORE cells….