henhaobio
Sign in

Nature CommunicationsGeneticsPeer-reviewed journal article

An integrated, scaled approach to resolve TSC2 variants of uncertain significance

Carina G BiariD, Ziyu R. WangiD, Nathan D. CampiD, Daniel L. HolmesiD, Melinda K. WheelockiD, Sriram PendyalaiD, Abby V. McGeeiD, Pankhuri GuptaiD, Abbye E. McEweniD, Malvika TejuraiD, Marcy E. RichardsoniD, Jamie D. WeyandtiD, Taylor Coleman, Alan F. RubiniD, Nick MooreiD, Ross StewartiD, Daniel ZeibergiD, Allyssa J. VandiiD, Samantha Dawson, Predrag Radivojac AiD, Lea M. StaritaiD, Gemma L. CarvilliD, Richard G. JamesiD, Douglas M. FowleriD, Jeffrey Dennis CalhouniD

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

Abstract

Obtaining a precise genetic tuberous sclerosis diagnosis is a challenge as many missense TSC2 variants are variants of uncertain significance (VUS). VUS in TSC2 have been resolved by one-at-a-time functional assays, but these assays cannot scale to the 3,634 TSC2 missense VUS observed so far. To address this challenge, we used massively parallel sequencing to measure the steady-state abundance of almost 9,000 TSC2 missense variants and developed an mTOR pathway activity assay using genome editing and cell sorting to generate activity scores for 391 missense variants. 1,288 of 8,891 (14.49%) missense variants assayed had altered TSC2 abundance, and 69 of 391 (17.65%) missense variants assayed had altered mTOR pathway activity. Calibration and integration of these data into classification of variants identified in a clinical cohort putatively reclassified 212 of 276 (76.8%) TSC2 missense VUS. These datasets will lead to improved genetic diagnosis of tuberous sclerosis with potential positive impacts on the clinical management of patients and their families.

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (18 Jan 2026) · CC BY 4.0

Abundance scores for 8,891 TSC2 missense variants, plus mTOR pathway activity scores for 391, were calibrated to ACMG/AMP evidence. They putatively reclassify 212 of 276 clinical-cohort missense VUS.

One-at-a-time cell-based assays are the current gold standard for TSC2 missense VUS, but they cannot keep pace with the 3,634 observed VUS. The paper pairs a protein-abundance assay with two endogenous-editing activity assays and calibrates them with ClinGen OddsPath.

Key results

  1. Shown1,288 of the assayed TSC2 missense variants had low protein abundance (14.49%). Missense scores were bimodal, and a substantial share of assayed residues were fully or partially intolerant to substitution.

    VAMP-seq: 1,288 of 8,891 (14.49%) missense variants had low abundance, 5,740 (64.56%) normal, 1,863 (20.95%) indeterminate. Synonymous mean 0.998 (n=362); missense mean 0.751 (n=8,891). 48 residues highly intolerant, 61 partially tolerant, 377 tolerant. · Development of multiplexed assays of variant…; Approximately one-third of residues assayed for…

  2. ShownActivity assays found 69 of 391 missense variants with reduced mTOR pathway activity. A subset lost activity with normal abundance, clustering at the RAP-GAP active site. T cell and HAP1 results agreed closely.

    SGE (primary T cells): 52 of 268 abnormal; cliPE (HAP1): 22 of 153. 34 variants altered both abundance and activity; 31 reduced activity without reduced abundance. SGE vs cliPE Pearson 0.92; abundance vs activity -0.57 (T cells), -0.69 (HAP1). · Abstract; Introduction; Development of multiplexed assays of variant…; TSC2 variant abundance and activity are strongly…

  3. ShownCalibration against ClinVar controls gives strong pathogenic evidence for low abundance but only supporting benign evidence for normal abundance. The activity assays give moderate evidence in both directions.

    OddsPath 33.1 (PS3_strong) for reduced abundance, 0.343 (BS3_supporting) for normal abundance; activity 4.5 and 5.6 (PS3_moderate), 0.113 and 0.094 (BS3_moderate). Controls: 54 BLB, 80 PLP; 31/80 PLP had normal abundance. Abundance specificity 98.18%, sensitivity 61.25%. · TSC2 variant effect data improve genetic…; Discussion

  4. SuggestedCombining the calibrated data with computational evidence in a decision tree putatively reclassified most missense VUS in a clinical-laboratory cohort, mostly to likely benign. These are preliminary points-based classifications.

    212 of 276 (76.8%) VUS with functional data reclassified: 199 (72.1%) likely benign, 13 (4.7%) pathogenic/likely pathogenic, 64 (23.1%) remained VUS. Functional data contributed to 193/212 (91%); computational evidence to 192/212 (90.5%). · Abstract; Introduction; TSC2 variant effect data improve genetic…; Variant classification on Ambry cohort

  5. SuggestedApplying the same scheme beyond the clinical cohort suggests most ClinVar and not-yet-observed missense VUS with data would be called likely benign. Few would reach likely pathogenic.

    Of 902 ClinVar missense VUS with variant effect data: 595 (65.96%) likely benign, 265 (29.38%) uncertain, 42 (4.66%) likely pathogenic. Of 7,465 unobserved missense variants: 4,579 (61.3%), 2,752 (36.9%), 134 (1.8%). · TSC2 variant effect data improve genetic…

Methods, data and reproducibility

Steady-state abundance of GFP-tagged TSC2 variants was measured by VAMP-seq in HEK293T cells sorted into four fluorescence bins. Activity was measured by phospho-S6 sorting after saturation genome editing in primary CD4+ T cells and curated loci prime editing (cliPE) in HAP1 cells. Scores were calibrated by OddsPath and merged in a decision tree.

  • VAMP-seq abundance: two TSC2 site-saturation libraries (residues 554-754 and 1512-1802) in HEK293T landing-pad cells; 8,891 missense variants scored · human (HEK293T cells; human TSC2 NM_000548.5) · n = 9,253 variants measured; 8,891 missense · IGVFDS5595BTYJ; score file IGVFFI9747KART
  • SGE pS6 activity in primary CD4+ T cells from healthy donors · human · n = 268 missense variants; 4 replicates (2 donors, 2 each) · IGVFDS1782FCXW; score file IGVFFI3097DFGF
  • cliPE pS6 activity in HAP1 cells · human (HAP1 cells) · n = 153 missense variants · urn:mavedb:00001201-a
  • ClinVar missense control variants used for OddsPath calibration (benign/likely benign and pathogenic/likely pathogenic) · human · n = 54 BLB, 80 PLP
  • Clinical cohort of TSC2 missense variants from Ambry Genetics · human · n = 340 variants (276 VUS with functional data)
Methods & tools
Pacybara, PEAR v0.9.11, bcl2fastq v2.20, CountESS, AlphaMissense, AlphaFold2, PacBio SMRTbell sequencing, ClinVar, gnomAD v4, TSC2 LOVD database, MAVEdb, Ambry Classifi
Reagents & environment
PEAR v0.9.11 and bcl2fastq v2.20 are named; no containers or requirement files are stated.
Reproducibility gaps
No code repository is given; the custom Python scripts for variant counting are not linked. Scores are in Table S1 and the IGVF Portal and MAVEdb entries (accessions listed under data).
Limitations (6)
  • VAMP-seq is cDNA-based, so it cannot assess splice-altering variants; only the endogenous-editing assays can. · stated by the authors
  • cliPE is retrospective: libraries are built from variants already seen in databases, so newly observed variants need further experiments. · stated by the authors
  • HAP1 cells used for cliPE were not selected for ploidy; a small effect on partial loss-of-function or dominant-negative variants cannot be excluded. · stated by the authors
  • The assays were built to separate complete loss of function from retained function. Assays calibrated for partial loss-of-function or dominant-negative variants remain future work. · stated by the authors
  • Reclassifications are putative points-based results, not validated against clinical outcomes. Ambry Genetics employees are among the authors, and the cohort is from one laboratory. · noted in this summary
  • Abundance covers only two TSC2 regions (residues 554-754 and 1512-1802), and activity covers 391 missense variants, so most of the gene lacks data. · noted in this summary

Read this if you classify TSC2 or other mTOR pathway variants, or want a worked example of calibrating abundance and activity MAVE data to ACMG/AMP evidence strengths.

Citations
3
FWCI
13.71
Field percentile
top 1%

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
THISBiar '24Garcia '25Nelson '22Niu '24Feo '26Tejura '26
74% semantic matchGenetics
Carina G Biar, Cole Pfeifer, Gemma L. Carvill+1 more·bioRxiv·Jun 7, 2024· 1 citations