henhaobio
Sign in

bioRxivCancer BiologyPreprint

Genome-wide functional genomics screens identify therapeutic vulnerabilities for lymphangioleiomyomatosis

Alberto Camacho‐Magallanes, Sean P. Delaney, Eric Y. LianiD, Ellis Wiljer, Julien Yockell‐Lelièvre, Lisa M. JulianiD, Adam PietroboniD, William Batoff, Arnold S. KristofiD, William L. StanfordiD

PDFDOI
In-depth full-text digest available · Jump to digest ↓

Abstract

Lymphangioleiomyomatosis (LAM) is a rare lung disease marked by cystic destruction caused by invasive LAM cells harboring loss-of-function mutations in TSC2 that exhibit dysregulated mTORC1 signaling. Rapamycin, the only approved treatment, is not curative. Therapeutic discovery has been limited by inadequate cell and animal models due to the unknown LAM cell of origin. We report a novel TSC2--/-- human pluripotent stem cell (hPSC)-derived neural crest cell (NCC) model, which replicates key LAM features. Leveraging this model, temporal RNA sequencing and genome-wide CRISPR knockout screens identified synthetic lethal genes including TWIST1 and SUN4. Antisense oligonucleotides (ASOs) targeting TWIST1 and SUN4 demonstrated selective TSC2 mutant NCC cytotoxicity and inhibited invasion. Using a "test-withdraw-monitor" clinical trial paradigm, TWIST1 and SUN4 ASOs profoundly suppressed tumor viability in a Tsc2--/-- mouse allograft model and eradicated 40% of tumors, outperforming standard of care rapamycin and highlighting their promise as novel cytoablative LAM therapeutics.

In-depth summary

Claude Opus 5.5 · xhigh effort · 2 Nov 2026 · from the full text of version 4 (17 Jul 2026) · CC BY-NC-ND 4.0

Differentiating TSC2-null human pluripotent stem cells into neural crest cells models lymphangioleiomyomatosis and reveals TWIST1 and SUN4 as synthetic lethal targets whose antisense knockdown eradicates mouse allograft tumors.

Rapamycin is the only approved therapy for lymphangioleiomyomatosis (LAM), a cystic lung disease driven by TSC2 loss, but its effects are cytostatic and lesions regrow upon treatment withdrawal. Because therapeutic discovery has been limited by models that fail to capture the pathognomonic PMEL-positive LAM cell population, an isogenic human neural crest cell platform enables genome-wide screening for cytocidal, mTORC1-independent targets.

Key results

  1. ShownGenome-wide CRISPR/Cas9 dropout screening in female wild-type (WT) and TSC2-null neural crest cells (NCCs) identified 665 synthetic lethal genetic elements, showing minimal overlap with canonical mTORC1 pathway genes and nominating SUN4 and TWIST1 as top candidates.

    Of 20,914 targeted genetic elements (19,050 protein-coding genes and 1,864 miRNAs), 665 were selectively depleted in TSC2-null NCCs over 14 days; only 13 (1.4%) overlapped with a 248-gene mTORC1 pathway set, while 293 elements overlapped between MAGeCK and BAGEL analyses with SUN4 ranked at the top. · A genome-wide CRISPR/Cas9 screen identifies…; Figure legends

  2. ShownAntisense oligonucleotide (ASO) knockdown of TWIST1 or SUN4 selectively kills TSC2-null NCCs via apoptosis, impairs clonogenic survival, and suppresses 3D extracellular matrix invasion without affecting WT cell survival.

    Treating two female isogenic pairs (H9 and H7) with 500 nM ASOs reduced target transcripts by >50% and protein by 30-50% at 48 hours, inducing cleaved caspase-3 in approximately 15% of TSC2-null NCCs after 72 hours versus minimal apoptosis in WT cells, and inhibiting invasion of 1,000-cell spheroids in Matrigel. · Inhibition of TWIST1 and SUN4 potentiates…; RNA-seq analysis reveals anti-metastatic effects…; Figure legends

  3. ShownExtended in vivo administration of TWIST1 or SUN4 ASOs suppresses Tsc2-null mouse allograft growth comparably to rapamycin during dosing and achieves complete tumor eradication in a subset of animals after treatment withdrawal.

    In a 30-day treatment (SUN4-ASO at 10 mg/kg, TWIST1-ASO at 2.25 mg/kg, or rapamycin at 1.14 mg/kg) followed by a 30-day withdrawal period in C57BL/6 mice bearing 105K allografts, ASOs eradicated 40% of tumors at necropsy whereas all rapamycin-treated tumors retained residual disease and regrew. · Abstract; TWIST1-ASO and SUN4-ASO treatments drive…; Figure legends

  4. ShownDifferentiated TSC2-null NCCs recapitulate phenotypic and sex-dimorphic transcriptional hallmarks of primary human LAM lesions, with female lines showing stronger enrichment for cancer-associated pathways than male lines.

    Day 10 TSC2-null vs WT NCCs yielded 609 differentially expressed genes (FDR < 0.05; 111 overlapping primary LAM scRNA-seq signatures); sex-stratified analysis identified 1,328 DEGs in male lines (D168, H1) and 961 DEGs in female lines (H9, H7), with only 8.6% overlap out of 2,264 total DEGs. · TSC2—/— NCCs model the PMEL component of LAM; TSC2—/— NCCs exhibit transcriptional features of…; Sexual dimorphic signatures in TSC2—/— NCCs

Methods, data and reproducibility

CRISPR/Cas9 editing knocked out TSC2 in four human pluripotent stem cell lines (two male, two female), which were differentiated into neural crest cells via a 12-day dual-SMAD inhibition protocol. Temporal bulk RNA-seq, a 14-day GeCKOv2 genome-wide CRISPR dropout screen in female H9 NCCs, gymnosis ASO knockdown assays, and subcutaneous 105K mouse allograft trials evaluated LAM phenotypes and target vulnerability.

  • Isogenic WT and TSC2-null human pluripotent stem cell lines (female H9/WA-09 and H7/WA-07; male H1/WA-01 and iPSC 168) differentiated into NCCs for temporal, sex-specific, and ASO bulk RNA-seq · human · n = 4 isogenic WT and TSC2-null cell line pairs · GSE137614, GSE338669
  • Genome-wide GeCKOv2 CRISPR/Cas9 dropout screen in Cas9-expressing female H9 WT and TSC2-null NCCs sampled at day 0 and day 14 · human · n = 20,914 targeted genetic elements (123,411 gRNAs)
  • Subcutaneous 105K Tsc2-null renal cystadenoma allograft tumors in immunocompetent male and female C57BL/6 mice (short-term 10-day and 30-day treatment/30-day withdrawal trials) · mouse · n = 25 mice (2 tumors per mouse; n = 3 to 4 per treatment group)
  • Publicly available gene signatures of primary pulmonary LAM cells (scRNA-seq), mouse NCCs, TSC-associated rAMLs, 621-101 cells, Tsc2-null MEFs, and Tsc2-null NDFs · human, mouse
Methods & tools
CRISPR/Cas9, GeCKOv2, MAGeCK-VISPR, MAGeCKFlute, BAGEL, Salmon, DESeq2, apeglm, clusterProfiler, g.Profiler, Cytoscape, CRISPOR
Reagents & environment
R v4.0.3 and v4.3.2, RStudio 023.12.1, DESeq2 v1.42.1, apeglm v1.2.1, FloJo v10.5.3, GraphPad Prism 11
Reproducibility gaps
No analysis code repository is provided; cell lines, plasmids, and non-GEO datasets are available upon request to the lead contact.
Limitations (4)
  • The hPSC-derived neural crest cell culture model does not capture the full cellular heterogeneity and microenvironmental context of primary patient LAM lesions. · stated by the authors
  • In vivo efficacy was tested using transformed, rapidly proliferating mouse 105K renal cystadenoma allografts rather than human TSC2-null NCCs, which are non-transformed, proliferate slowly, and fail to form subcutaneous xenograft tumors. · stated by the authors
  • The precise molecular mechanisms by which TWIST1 and SUN4 knockdown selectively trigger cell death in TSC2-deficient cells remain undefined. · stated by the authors
  • The genome-wide CRISPR/Cas9 dropout screen and subsequent in vitro ASO validations were conducted only in female hPSC-derived NCC lines (H9 and H7), leaving synthetic lethality in male TSC2-null backgrounds untested. · noted in this summary

Read this if you study tuberous sclerosis complex or lymphangioleiomyomatosis, use stem cell differentiation to model rare diseases, or screen for synthetic lethal antisense oligonucleotide targets independent of mTORC1.

Citations
12

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISPietrobon '21Valianou '22Olatoke '24Olatoke '22Clements '25Babaei‐Jadi '25
73% semantic matchCell Biology
A. Pietrobon, J. Yockell-Lelievre, N. Melong+15 more·bioRxiv·Jul 3, 2021