Integrative screening identifies functional variants and VNTRs underlying GWAS signals at the 5p15.33 multi-cancer susceptibility locus

By integrating statistical fine-mapping with functional assays, this study identifies eight multi-cancer functional variants and a key intronic VNTR within the 5p15.33 locus that mediate antagonistic pleiotropy in cancer susceptibility through differential regulation of TERT and CLPTM1L via Hippo-pathway transcription factors.

The Gist

The Big Picture: A Genetic "Hotspot" with a Split Personality

Imagine the human genome as a massive, sprawling city. Most of the city is quiet, but there are certain intersections known as "high-risk zones" where accidents (cancers) happen more often.

One of the most notorious intersections is a spot on Chromosome 5 called 5p15.33. Scientists have known for years that this spot is linked to many different types of cancer, including pancreatic, lung, melanoma, and bladder cancer.

But here is the weird part: This intersection has a split personality.

• A specific genetic variation (a "risk allele") might make you more likely to get pancreatic cancer.
• Yet, that exact same variation might actually protect you from getting lung cancer.

This is called antagonistic pleiotropy. It's like a traffic light that turns green for cars going North (helping them) but turns red for cars going South (hurting them). For years, scientists knew this traffic light existed, but they didn't know how it worked or exactly which part of the signal was broken.

The Detective Work: How They Solved the Mystery

The researchers in this paper acted like a team of high-tech detectives. They wanted to find the specific "broken wires" in the genetic code that cause these mixed effects. They used a three-step investigation strategy:

1. The Shortlist (Fine-Mapping)

First, they looked at massive databases of DNA from hundreds of thousands of people. They narrowed down the "suspects" from millions of genetic variations to a shortlist of about 116 prime suspects (called Credible Causal Variants) that were most likely to be the culprits.

2. The Speed Test (MPRA)

Next, they put these suspects through a "speed test." Imagine taking a snippet of DNA and plugging it into a machine that measures how much it turns on a gene (like a volume knob). They tested these snippets in four different types of cancer cells (pancreas, lung, skin, bladder).

• Result: They found that some snippets acted like volume knobs, turning genes up or down. They identified 8 "Multi-Cancer Functional Variants" (MCFVs) that were the most active troublemakers.

3. The "Silence" Test (CRISPRi)

To be sure, they used a tool called CRISPRi (a genetic "mute button"). They went into the cells and silenced specific parts of the DNA to see what happened.

• The Discovery: When they silenced a specific spot near the TERT gene (which helps cells live forever, a hallmark of cancer), the cells stopped growing. This confirmed these spots were essential for cancer cell survival.

The Twist: It's Not Just Single Letters (SNPs)

For decades, scientists thought genetic diseases were caused by single-letter typos in the DNA code (like changing an 'A' to a 'G'). These are called SNPs.

However, this paper found something much more complex hiding in the shadows: VNTRs (Variable Number Tandem Repeats).

The Analogy:

• SNPs are like a typo in a sentence: "The cat sat on the mat." (Changed 'm' to 'n').
• VNTRs are like a stutter or a repeated phrase: "The cat sat on the mat mat mat mat."

The researchers found a specific "stutter" (a VNTR) inside a gene called CLPTM1L. This stutter wasn't just a random glitch; it was a powerful volume booster (an enhancer).

• The Mechanism: This stutter acts like a magnet for specific proteins (Hippo pathway transcription factors). When the stutter is long, it grabs these proteins tightly, turning up the volume on genes that help cancer grow.
• The Proof: They used a special microscope (long-read sequencing) to count the repeats and found that people with longer "stutters" were at higher risk for pancreatic cancer.

The "Why": Context is King

The most fascinating part of the paper explains the "split personality" (why it helps one cancer but hurts another).

Imagine the DNA as a light switch.

• In Pancreatic cells, the switch is wired to turn the lights ON (promoting growth).
• In Lung cells, the same switch is wired to turn the lights OFF (stopping growth).

The researchers found that the genetic variation rs421629 sits on this switch.

• When they silenced it in Pancreatic cells, the cancer genes went down.
• When they silenced it in Lung cells, the cancer genes went up.

This suggests that the same genetic "part" can act as an accelerator or a brake depending on the cell type it is in. It's like a key that starts a car in one garage but locks the doors in another.

The Takeaway

This paper is a breakthrough because it changes how we look at genetic risk:

1. It's not just single letters: We need to look at "stutters" (VNTRs) and repeated patterns in our DNA, not just single typos.
2. Context matters: A genetic risk factor isn't always "bad." It depends on the tissue it lives in.
3. The Solution: By understanding exactly how these switches work (like the Hippo pathway proteins grabbing onto the VNTR stutter), we might be able to design drugs that flip the switch back to the "safe" position, regardless of which cancer it is.

In short: The researchers found the specific broken wires and the "stuttering" patterns in our DNA that cause a genetic intersection to be a danger zone for some cancers and a safe haven for others. They didn't just find the problem; they figured out the wiring diagram.

Technical Summary

1. Problem Statement

Chromosome 5p15.33 is one of the most replicated cancer susceptibility loci, harboring at least 10 distinct association signals across multiple cancer types (including pancreatic, lung, melanoma, and bladder cancer). A defining feature of this locus is antagonistic pleiotropy, where risk alleles for one cancer type often act as protective factors for another.

• The Gap: While Genome-Wide Association Studies (GWAS) have identified these statistical signals, the functional mechanisms and causal variants (CCVs) driving these associations remain largely unresolved.
• The Challenge: Traditional post-GWAS analyses often focus on Single Nucleotide Polymorphisms (SNPs) and may overlook complex structural variants, such as Variable Number Tandem Repeats (VNTRs), which are increasingly recognized as drivers of GWAS signals. Furthermore, the cell-type-specific nature of regulatory elements makes it difficult to pinpoint how a single locus influences diverse cancer phenotypes.

2. Methodology

The authors employed a multi-step, integrative functional genomics framework to dissect the 5p15.33 locus:

• Statistical Fine-Mapping:

  • Performed ancestry-specific fine-mapping using SuSiE on GWAS summary statistics for four cancer types: Pancreatic Ductal Adenocarcinoma (PDAC), Melanoma, Lung Adenocarcinoma (European and East Asian), and Bladder Cancer.
  • Generated 95% credible sets of causal variants (CCVs) and integrated them with previously reported ASSET signals to create a panel of 116 CCVs.

• Massively Parallel Reporter Assays (MPRA):

  • Tested the allele-specific regulatory activity of the 116 CCVs across eight cancer cell lines (representing the four cancer types).
  • Measured transcriptional output (RNA/DNA ratios) to identify variants with significant allelic effects (FDR ≤ 0.01, fold change ≥ 15%).

• Tiled CRISPRi Proliferation Screens:

  • Conducted an agnostic, tiled CRISPR-inhibition (CRISPRi) screen using 25,778 sgRNAs covering the entire 5p15.33 locus (chr5:1,233,287–1,380,000).
  • Performed in the same eight cell lines engineered to express dCas9-KRAB-ZIM3.
  • Screened over 25 cumulative cell doublings to identify proliferation-essential regulatory elements using the CRISPR-SURF algorithm.

• Integration and Validation:

  • Defined Multi-Cancer Functional Variants (MCFVs) as variants that were both MPRA-hits and colocalized with CRISPRi-identified essential elements in at least one cancer type.
  • Long-Read Sequencing: Used PacBio HiFi sequencing to genotype VNTRs in the CLPTM1L intron 9 region across 1,810 PDAC cases/controls and 1kGP samples.
  • Imputation: Imputed VNTR genotypes into a larger PDAC meta-GWAS dataset to test for association.
  • Mechanistic Studies: Validated enhancer activity via luciferase assays and identified binding transcription factors (TFs) using DNA pulldowns coupled with mass spectrometry and Western blotting.

3. Key Contributions

1. Identification of MCFVs: Successfully identified eight Multi-Cancer Functional Variants (MCFVs) across three distinct GWAS signals, bridging statistical genetics with functional biology.
2. Discovery of VNTRs as Causal Variants: Demonstrated that a Variable Number Tandem Repeat (VNTR) in CLPTM1L intron 9 is a potent causal variant underlying the rs465498 signal, challenging the SNP-centric view of GWAS causality.
3. Mechanism of Antagonistic Pleiotropy: Provided a cis-regulatory explanation for antagonistic pleiotropy, showing that specific regulatory elements (e.g., rs421629) can have opposing effects on gene expression (TERT) depending on the cellular context (pancreatic vs. lung).
4. Hippo Pathway Mediation: Uncovered that the VNTR enhancer activity is mediated by Hippo pathway transcription factors (TEAD1/TEF-1, TEAD4, and TAZ/WWTR1).

4. Key Results

• Functional Variants:

  • The MPRA and CRISPRi integration identified 8 MCFVs.
  • rs421629 (part of the CLPTM1L signal) showed opposing effects on TERT expression: CRISPRi repression increased TERT in pancreatic cells (PANC-1) but decreased it in lung cells (A549). This suggests a context-dependent regulatory switch, potentially explaining why the risk allele is protective for one cancer and a risk factor for another.
  • Variants in TERT intron 2 (e.g., rs7705526 signal) consistently reduced TERT expression across cell lines.

• VNTR Discovery and Association:

  • VNTR1 (in CLPTM1L intron 9) was identified as a proliferation-essential element.
  • Long-read sequencing revealed a trimodal distribution of VNTR1 alleles.
  • Association: Longer VNTR1 alleles (specifically the 2,934 bp allele and "Cluster 3") were significantly associated with increased PDAC risk (OR ≈ 1.20, P < 10⁻¹⁴).
  • Conditioning analysis showed that the VNTR signal largely explains the association of the lead SNP rs31490, suggesting the VNTR is the true causal variant or in very high LD with it.

• Regulatory Mechanism:

  • Enhancer Activity: VNTR1 acts as a potent enhancer for both TERT and CLPTM1L. Luciferase assays showed that longer VNTR1 alleles (Cluster 3) had higher enhancer activity than shorter alleles.
  • Transcription Factors: DNA pulldown and mass spectrometry confirmed robust binding of TEF-1 (TEAD1), TEF-3 (TEAD4), and TAZ (WWTR1) to the VNTR repeat units.
  • Hippo Dependence: Treatment with verteporfin (an inhibitor of TEF-YAP binding) abolished VNTR1-driven reporter activity, confirming that the enhancer function is dependent on the Hippo signaling pathway.

5. Significance

• Resolving Pleiotropy: The study provides a concrete molecular mechanism for antagonistic pleiotropy at 5p15.33, demonstrating that cell-type-specific wiring of cis-regulatory elements can toggle a variant between risk and protective states.
• Beyond SNPs: It highlights the critical importance of including structural variants (VNTRs) in GWAS follow-up studies. The findings suggest that a significant portion of "missing heritability" in complex loci may be driven by repeat polymorphisms that are poorly captured by standard SNP arrays.
• Therapeutic Insights: By identifying the Hippo pathway as the mediator of these risk variants, the study suggests potential therapeutic targets (e.g., YAP/TAZ inhibitors) for cancers driven by 5p15.33 susceptibility.
• Methodological Framework: The paper establishes a robust, generalizable framework for integrating fine-mapping, MPRA, CRISPRi, and long-read sequencing to resolve complex, multi-cancer susceptibility loci.

In summary, this work moves beyond statistical association to define the functional landscape of 5p15.33, revealing that both SNPs and VNTRs, mediated by the Hippo pathway, drive cancer susceptibility through context-dependent regulation of TERT and CLPTM1L.