High-throughput biochemical phenotyping of SHP2 variants reveals the molecular basis of diseases and allosteric drug inhibition

By applying high-throughput microfluidic enzyme kinetics to profile 190 SHP2 variants, this study reveals that dysregulated autoinhibition—not altered stability or catalysis—is the primary driver of pathogenesis and explains how allosteric inhibitors differentially stabilize specific conformations to modulate drug responses.

The Gist

Imagine your body is a massive, bustling city. Inside every cell, there are millions of tiny workers (proteins) keeping everything running smoothly. One of the most important workers is a protein called SHP2. Think of SHP2 as a traffic cop at a busy intersection. Its job is to decide when to let signals (like "grow," "divide," or "fight infection") pass through and when to stop them.

Normally, this traffic cop has a safety mechanism: a self-imposed brake. When no one is calling for help, the cop keeps the road closed (the "off" state) to prevent chaos. But when a signal arrives (like a police siren), the cop releases the brake, opens the road, and lets the traffic flow (the "on" state).

This paper is a massive investigation into what happens when this traffic cop gets mutated (when the DNA instructions for building the cop are slightly wrong). The researchers wanted to answer three big questions:

1. Why do some mutations cause developmental disorders (like Noonan Syndrome)?
2. Why do others cause cancer?
3. Why do some drugs work on some patients but fail on others?

Here is the breakdown of their findings, using simple analogies:

1. The Problem: Too Many "Unknown" Variants

Scientists have found hundreds of different typos (mutations) in the gene that builds SHP2. In the medical world, many of these are labeled "Variants of Uncertain Significance" (VUS). It's like finding a typo in a manual and not knowing if it's a harmless spelling error or a fatal instruction that will crash the system. Doctors often have to guess whether a patient's mutation is dangerous or not.

2. The Solution: A "Robot Factory" for Testing

To solve this, the researchers built a high-tech, automated factory called HT-MEK.

• The Analogy: Imagine a factory with 1,800 tiny assembly lines running at the same time. Instead of building cars, this factory builds 190 different versions of the SHP2 traffic cop, each with a different typo.
• The Process: They put these "cop robots" on a microchip, fed them fuel, and watched how fast they worked. They didn't just watch them work; they tested how well they held their brakes, how well they reacted to sirens, and how well they responded to drugs.

3. The Big Discovery: It's All About the Brake

For years, scientists thought the main problem with these mutations was that the traffic cop was either broken (couldn't work) or too strong (worked too fast).

The paper reveals a different truth: The main issue isn't how fast the cop works; it's that the brake is broken.

• The Finding: Most disease-causing mutations don't make the cop work faster or slower. Instead, they make the brake slip. The cop gets stuck in the "open" position, letting traffic flow even when it shouldn't.
• The Consequence:
  • Developmental Disorders (Noonan Syndrome): The brake is slightly loose. The cop opens the road a little too often, causing cells to grow a bit too much during development.
  • Cancer: The brake is completely gone. The cop is stuck wide open, causing cells to grow uncontrollably.

4. The Drug Mystery: Why Do Some Drugs Fail?

There are new drugs designed to fix this broken traffic cop. These drugs are like emergency handcuffs that grab the cop and force the brake back on.

• The Old Theory: Scientists thought these drugs worked by locking the cop in the "closed" (brake-on) position. They assumed that if a mutation made the cop "more open," the drug would have a harder time grabbing it.
• The New Reality: The researchers discovered the drugs actually work best on a middle state—a "half-open" position.
  • The Analogy: Imagine the cop has three positions: Locked (Safe), Half-Open (The Target), and Fully Open (Chaos).
  • The new drugs are like a specific key that fits perfectly into the Half-Open position.
  • The Twist: If a mutation makes the cop fully open (fully chaotic), the drug actually struggles to grab it because the cop is too far away from the "half-open" sweet spot. This explains why some cancer patients with very aggressive mutations don't respond to these drugs.

5. The "Three-State" Model

The paper proposes a new way to think about how SHP2 works. Instead of just "On" and "Off," imagine a dimmer switch with three settings:

1. Off (Closed): The brake is fully on.
2. Dim (Intermediate): The brake is slightly slipping. This is where the drugs work best.
3. Bright (Open): The brake is gone.

The researchers found that the most effective drugs prefer the "Dim" setting. If a mutation pushes the switch all the way to "Bright," the drug loses its grip.

Why Does This Matter?

This study is a game-changer for Precision Medicine:

• For Doctors: Instead of guessing if a mutation is dangerous, they can now look at the "biochemical fingerprint" of the mutation. If the brake is slipping, they know it's likely pathogenic.
• For Patients: It helps predict which drugs will work. If a patient has a mutation that pushes the cop to "Fully Open," doctors might know to avoid certain drugs and try a different strategy (like blocking the signal before it reaches the cop).
• For Science: It proves that looking at just one number (like "how fast does it work?") isn't enough. You need to understand the whole machine—the brakes, the sensors, and the gears—to fix it.

In short: This paper used a high-tech robot factory to test 190 broken versions of a traffic cop. They found that the real problem is usually a slipping brake, not a broken engine. They also discovered that the best drugs work on a "half-open" state, not the fully open one, which explains why some patients don't respond to treatment. This knowledge helps doctors choose the right key for the right lock.

Technical Summary

1. Problem Statement

Despite the vast number of human protein-coding variants identified through next-generation sequencing, the functional and clinical significance of most remain unresolved. Specifically for SHP2 (encoded by PTPN11), a protein tyrosine phosphatase linked to developmental disorders (Noonan Syndrome, NS; Noonan Syndrome with Multiple Lentigines, NSML) and various cancers:

• Clinical Ambiguity: Over 50% of PTPN11 missense variants in ClinVar are classified as Variants of Uncertain Significance (VUS).
• Mechanistic Gaps: It is unclear how specific mutations drive distinct phenotypes (e.g., why NS mutations often enhance basal activity while NSML mutations reduce it, yet both cause developmental issues; or why both hyperactive and hypoactive variants are linked to cancer).
• Therapeutic Uncertainty: The sensitivity of different SHP2 variants to clinical-stage allosteric inhibitors is largely unknown, and the prevailing "two-state" (open/closed) model of SHP2 regulation fails to explain observed drug resistance patterns and incomplete inhibition.
• Data Limitations: Existing high-throughput methods (like cellular fitness assays) provide composite readouts that do not isolate distinct biophysical parameters (stability, catalysis, autoinhibition, ligand affinity) required to understand complex disease mechanisms.

2. Methodology

The authors employed High-Throughput Microfluidic Enzyme Kinetics (HT-MEK) to systematically characterize 190 clinically documented SHP2 variants.

• Library Construction: A library of 189 missense variants (plus Wild Type) was generated, covering residues 1–525 (structured domains). Variants were classified based on ClinVar annotations (Pathogenic, Benign, VUS, Conflicting).
• HT-MEK Platform:
  • Utilized microfluidic devices with 1,798 reaction chambers.
  • Enabled parallel cell-free expression, purification (via EGFP-tag and anti-EGFP immobilization), and kinetic assays.
  • Allowed iterative measurement of thermodynamic and kinetic parameters with precise temporal control.
• Multi-Dimensional Assays: For each variant, the team quantified 10 distinct parameters across two constructs:
  1. Full-Length SHP2 (SHP2FL): Measured basal activity ($k_{cat}/K_M$), autoinhibition equilibrium, and response to activators (ppIRS1 peptide) and inhibitors.
  2. Catalytic Domain Only (SHP2CD): Measured intrinsic catalytic activity and thermodynamic stability ($\Delta G_{native}$) via urea denaturation curves.
• Drug Profiling: Tested sensitivity to three clinical-stage allosteric inhibitors: TNO155, RMC-4630, and GDC-1971, across varying concentrations of activating ligands (ppIRS1).
• Modeling: Used thermodynamic modeling to fit data to conformational selection models and developed a three-state model to explain drug inhibition data.

3. Key Contributions & Results

A. Decoupling Biochemical Mechanisms

The study successfully separated the effects of mutations into four distinct mechanistic categories:

1. Intrinsic Activity Changes: Mutations in the PTP domain that directly alter catalytic efficiency.
2. Autoinhibition Disruption: Mutations (often at the nSH2-PTP interface) that increase the fraction of the protein in the active "open" state ($f_{active}$).
3. Stability Defects: Mutations causing global destabilization (rarely the primary driver of activity loss in this dataset).
4. Peptide Recognition: Specific mutations altering substrate binding.

Key Finding: Dysregulated autoinhibition (increased $f_{active}$), rather than altered stability or intrinsic catalysis, is the predominant driver of SHP2-associated pathogenesis. Most pathogenic variants significantly increase the fraction of SHP2 in the open state.

B. Resolving Disease Phenotypes

• Pathogenicity Prediction: A machine learning classifier using only two biochemical features—fraction active ($f_{active}$) and activator sensitivity ($EC_{50}$)—outperformed models using all parameters or sequence-based predictors (like AlphaMissense) in distinguishing pathogenic from benign variants.
• Disease Subtyping:
  • Cancer vs. Developmental Disorders: Cancer-associated variants generally exhibit higher $f_{active}$ and activator sensitivity compared to developmental disorder variants. Strongly activating germline variants are likely embryonically lethal, explaining why they appear only as somatic mutations in cancer.
  • NS vs. NSML: Confirmed that NS variants typically have higher basal activity, while NSML variants have lower intrinsic activity. However, both share the underlying defect of impaired autoinhibition.
  • Blood vs. Solid Cancers: Blood cancers (e.g., JMML) harbor variants with higher $f_{active}$ than solid tumors. Solid tumors tolerate variants with reduced intrinsic activity, suggesting a dual role for SHP2 in oncogenesis where catalytic output is less critical than scaffolding functions.

C. Redefining Allosteric Inhibition (The Three-State Model)

The study challenged the canonical two-state model (Closed/Inactive $\leftrightarrow$ Open/Active), which posits that inhibitors stabilize the closed state.

• Observation 1: Inhibitors do not completely suppress SHP2 activity even at saturating concentrations (residual activity persists).
• Observation 2: Variants with higher $f_{active}$ (more "open") were paradoxically more sensitive to clinical inhibitors (lower $IC_{50}$), contrary to the two-state prediction.
• New Model: The authors propose a Three-Conformational-State Model:
  1. Closed ($E_c$): Fully autoinhibited.
  2. Intermediate ($E_i$): Partially open, possesses residual catalytic activity, and binds inhibitors with higher affinity than the closed state.
  3. Open ($E_o$): Fully active.
• Mechanism: Clinical-stage inhibitors (TNO155, RMC-4630, GDC-1971) preferentially stabilize the Intermediate state. Variants that populate this state more readily (due to mutations) are more sensitive to these drugs.
• Contrast with SHP099: The first-generation inhibitor SHP099 follows the two-state model (binding only the closed state), explaining its different sensitivity profile and lower potency against certain variants.

D. Therapeutic Implications

• Activator Competition: In cancer cells with high levels of endogenous activators (e.g., ppIRS1), SHP2 shifts toward the open state, reducing inhibitor potency. This creates a "therapeutic window" where inhibitors may be less effective in tumors with high upstream signaling.
• Strategy: Combining allosteric inhibitors with agents that block one SH2 domain (biasing the equilibrium toward the high-affinity intermediate state) could enhance therapeutic potency.

4. Significance

• Functional Atlas: This work provides the largest quantitative biochemical dataset for allelic variants of a disease-relevant protein to date, linking genotype directly to molecular function and clinical outcome.
• Clinical Utility: The biochemical signatures ($f_{active}$ and $EC_{50}$) offer a robust framework for reclassifying VUS, predicting disease subtypes, and stratifying patients for specific therapies.
• Drug Discovery: The discovery of the "partially active intermediate state" as the primary target for modern allosteric inhibitors fundamentally changes the understanding of SHP2 drug mechanisms. It suggests that drug optimization should focus on stabilizing this intermediate state rather than just the fully closed state.
• General Framework: The HT-MEK approach establishes a scalable, generalizable method for deciphering the biochemical logic of protein variants, applicable to other clinically relevant proteins beyond SHP2.

In summary, this paper moves beyond simple activity measurements to provide a thermodynamic and kinetic map of SHP2, resolving long-standing paradoxes in disease classification and revealing a new mechanistic basis for the efficacy of next-generation allosteric drugs.