CalUCSF | iGEM 2025

Project Description

NeuroSplice

Abstract

RNA splicing is essential in modulating gene expression and even marginal deviations of the splice variants could contribute to disease. For example, in Multiple Sclerosis (MS), one splicing alteration is found in the interleukin-7 receptor (IL7R) gene. The interleukin-7 receptor is expressed in the form of a splice variant, sIL7R, and has been linked to MS susceptibility and disease severity. Currently, there are limited tools to detect RNA variants, and those that do, are expensive, located only in the lab setting, and cannot be used to screen large populations.

In response to this void, we developed a low-cost, paper-based, cell-free biosensor called NeuroSplice to detect disease-associated RNA splice variants such as sIL7R accurately. NeuroSplice implements synthetic biology tools including, toehold-based sensors and a cell-free transcriptional system to distinguish between healthy and disease-associated RNA signatures. We also developed a computational pipeline for predicting and modeling RNA splice junctions, making the biosensor rapid and reliable for design.

Diagram illustrating the toehold switch mechanism: OFF state (hairpin structure) and ON state (unfolded structure upon RNA trigger binding).

NeuroSplice provides inexpensive and portable detection of RNA variants to revolutionize the diagnosis of neurological disease (e.g., MS), molecular testing closer to the patient, and progress towards early and accessible precision medicine.

Introduction

The Core Mechanism of Splicing

While every cell in the body has the same DNA sequence, their identities and functions are vastly different. The variation comes not just from which genes are expressed, but also from how RNA transcripts are processed after they are transcribed. One important processing event is splicing, which is the removal of non-coding introns with the joining of coding exons for the mature form of a messenger RNA (mRNA). Through alternative splicing, a gene can produce multiple mRNA isoforms, and the potential for functional diversity seems infinite. However, even slight perturbations in this finely-tuned process can produce abnormal splice variants associated with disease. (Liu et al., 2022)

Diagram showing pre-mRNA splicing to generate different mRNA isoforms (alternative splicing).

MS, sIL7R, and the Specificity Challenge

Abnormal splicing of the interleukin-7 receptor (IL7R) gene has emerged as a key genetic risk factor in Multiple Sclerosis (MS). A single-nucleotide polymorphism (SNP), rs6897932, located in exon 6 of IL7R, drives exon skipping and creates a soluble form of IL7R, sIL7R, which lacks a transmembrane domain. (Nastaran Majdinasab et al., 2014) The high levels of sIL7R are capable of altering IL-7 signaling and may serve to modulate T-cell activity, therefore driving the inflammatory pathology in MS.

Diagram contrasting the transmembrane IL7R protein (healthy) with the soluble sIL7R protein lacking the transmembrane domain (disease-associated).

The Challenge: Conventional molecular diagnostics struggle distinguishing these splice variants, as they differ at the exon-exon junctions by only several nucleotides. NeuroSplice was designed specifically to overcome this, providing high specificity and low cost.

Methodology: Design, Build, Test, Learn (DBTL) Cycle

Methodology Overview

This study aimed to develop a cell-free, toehold switch–based diagnostic biosensor capable of detecting the soluble IL7R (sIL7R) splice variant. The project combined computational modeling and experimental validation using a design–build–test–learn (DBTL) framework. We iteratively refined our constructs across ten trials, integrating data from both in silico and in vitro analyses to improve specificity, signal strength, and reproducibility.

Infographic illustrating the Design-Build-Test-Learn (DBTL) cycle used in synthetic biology.

Toehold Design and Computational Modeling

The design began with identifying the splice junction unique to the sIL7R isoform (the junction between exons 5 and 7). RNA folding simulations were performed using NUPACK and RNAstructure to evaluate secondary structures and thermodynamic stability.

  • The OFF state was optimized for a stable hairpin structure (delta G approx -10 to -20 kcal/mol), concealing the ribosome binding site (RBS).
  • The ON state was modeled to confirm efficient hairpin opening upon trigger binding, exposing the translation initiation region.
  • Designs were refined to reduce background expression (leakiness) and improve dynamic range, optimizing the reporter gene from amilCP to GFP, and finally to sfGFP.

DNA Template Preparation

Linear DNA templates (gBlocks) containing the toehold–reporter constructs were synthesized by IDT. Stocks were prepared to a 160 nM concentration (higher than standard 96 nM) to offset the volume of added RNA oligos. Purity was confirmed by A260/A280 and A260/A230 ratios.

RNA Trigger and Control Preparation

Synthetic sIL7R trigger RNA and randomized non-cognate control oligos were resuspended to a 200 uM master stock, followed by a 20 uM working stock preparation to ensure complete dissolution and minimize degradation.

Cell-Free TXTL Expression Reactions & Controls

Experiments used the myTXTL® Pro Kit and were assembled on ice in 12 uL reactions. Each trial included critical controls:

  • Experimental (Trigger-ON): Toehold + sIL7R RNA trigger (200 nM stock).
  • Negative Control 1: No RNA added (nuclease-free water).
  • Negative Control 2: Randomized (non-cognate) RNA trigger.
  • Positive Control: deGFP DNA template (kit supplied).

Reactions were incubated at 27 degrees C for 16–18 hours and monitored for color (amilCP) or fluorescence (GFP/sfGFP).

Image of a paper-based diagnostic sensor strip changing color/fluorescing.

Measurement and Analysis

Fluorescence was measured kinetically using a SpectraMax i3 plate reader (Ex: 488 nm, Em: 509 nm). Data were background-subtracted and normalized against the no-RNA control. Key performance indicators were the ON/OFF ratio and time-to-half-maximum signal (T1/2). Safety protocols were strictly followed according to UC Berkeley guidelines.

Results

Across ten iterative trials, we systematically optimized a toehold switch designed to detect the IL7R exon-6 skipping variant, progressively improving activation strength, fluorescence precision, and reproducibility.

Initial Validation (amilCP)

The initial amilCP construct confirmed specific response to the sIL7R RNA trigger. Experimental wells reached approx 0.34 a.u. compared to approx 0.12 a.u. in controls, yielding a highly significant difference (p = 1.43 x 10^-11) and a normalized 2.8x increase (p approx 0.0004). Slow maturation prompted the switch to GFP.

Fluorescence Optimization (GFP to sfGFP)

  • GFP (Trial 2): Confirmed fast, specific activation with approx 400,000 a.u. vs. 200,000–250,000 a.u. in controls (normalized 2.5x higher).
  • Structural Refinements (Trials 3 & 4): Introducing upstream buffer and lower G-content reduced OFF-state leakiness, resulting in normalized increases of 1.4x and 1.6x, respectively.
  • sfGFP (Trial 5): Replacement with superfolder GFP provided brighter, faster signals, achieving a normalized 1.8–2x ratio with high significance (p = 7.87 x 10^-25).

Reproducibility and Stability (Trials 6-10)

Subsequent trials confirmed the stability of the optimized sfGFP construct. Trials 6 through 10 consistently maintained strong ON/OFF separation:

  • Trial 6 showed a 1.9x normalized ratio (p = 5.76 x 10^-44).
  • Trials 9 and 10 demonstrated long-term stability with 60,000–70,000 a.u. ON-state intensity and highly significant separation (normalized approx 1.8x to 2x).
Placeholder for Bar Graph showing ON/OFF Ratio over ten trials.

Discussion: Engineering Synthetic Biology for Diagnostics

Through ten iterative trials, we systematically improved the performance and reliability of a cell-free toehold switch designed to detect the IL7R exon-6 skipping variant. Each stage of the process built upon the limitations and insights of the previous one, reflecting an engineering-based approach to biological design.

Iterative Refinement & Reporter Optimization

The early amilCP construct (Trial 1) demonstrated selective recognition but highlighted the need for a faster, more sensitive reporter system due to slow color development. Switching to GFP (Trial 2) allowed for real-time tracking, but revealed minor leaky translation due to an unstable hairpin structure.

Structural Tuning for Stability

Trials 3 and 4 addressed leakiness by exploring structural refinements, including introducing upstream buffer regions and reducing G-content downstream of the reporter. These changes confirmed that both upstream and downstream sequence contexts significantly influence RNA folding and translational dynamics, effectively reducing background activity.

The Final Optimized Construct

A major step forward occurred in Trial 5 with the introduction of superfolder GFP (sfGFP), which provided faster maturation and greater fluorescence intensity. Trials 6 through 10 validated the reproducibility and stability of this optimized sfGFP-based construct. The final construct maintains consistent ON-state activation, low OFF-state leak, and statistically significant separation, demonstrating a robust and sensitive RNA-sensing platform for diagnostic applications.

Future Takeaway: Clinical Validation and Deployment

Translating to Patient Samples

With the freeze-dried IL7R diagnostic platform now fully developed, the next critical step is to validate its function using RNA extracted from peripheral blood mononuclear cells (PBMCs). PBMCs (immune cells like T cells and B cells) are an ideal model for studying MS, as IL7R is naturally expressed in them and the sIL7R variant is associated with autoimmune pathogenesis. (Kleiveland, 2015)

Testing the freeze-dried diagnostic directly on RNA isolated from PBMCs of both MS patients and healthy controls would determine whether our toehold switch can accurately detect endogenous sIL7R transcripts, rather than synthetic triggers alone. Results could be cross-validated using qRT-PCR or RNA sequencing of the same samples.

Demonstrating reliable detection from patient-derived PBMC RNA would mark a crucial step toward a field-ready, portable, and equipment-free molecular diagnostic tool for multiple sclerosis.

Placeholder for image of the final portable diagnostic kit cassette.

References

  • Liu, Qi, et al. “Alternative Splicing and Isoforms: From Mechanisms to Diseases.” Genes, U.S. National Library of Medicine, 24 Feb. 2022, pmc.ncbi.nlm.nih.gov/articles/PMC8951537/. Accessed 09 Oct. 2025.
  • Nastaran Majdinasab, Behbahani, M. H., Hamid Galehdari, & Mohaghegh, M. (2014). Association of interleukin 7 receptor gene polymorphism rs6897932 with multiple sclerosis patients in Khuzestan. Iranian Journal of Neurology, 13(3), 168. https://pmc.ncbi.nlm.nih.gov/articles/PMC4240935/
  • Kleiveland, C. R. (2015). Peripheral Blood Mononuclear Cells. Springer EBooks, 161–167. https://doi.org/10.1007/978-3-319-16104-4_15

Ready to dive into the molecular details?

See Our Engineering Strategy →