Cell communication is essential for cellular function and relies on the faithful transmission of signals across the plasma membrane through membrane receptors. Receptor kinases constitute an important class of molecular antennas in which the extracellular signal binding module is linked to an intracellular kinase along one polypeptide chain. Perturbations in this finely coordinated system causes aberrant signaling which lead to pathological states such as cancer or developmental disorders. Despite the disease relevance and extensive therapeutic focus, we fundamentally do not understand how receptor kinases transmit a signal across the plasma membrane in the absence of full-length structures. This is particularly true for the Human Epidermal Growth Factor Receptor 2 (HER2), an orphan receptor, and the Human Epidermal Growth Factor Receptor 3 (HER3), a pseudokinase receptor which form a potent pro-oncogenic heterocomplex upon binding to extracellular ligand. Here, we present three novel high-resolution cryo-electron microscopy (cryo-EM) structures of the extracellular domain of the breast cancer receptor, HER2, engaged with its liganded co-receptor, HER3, solved in the context of near full-length receptors. As the first singly-liganded human HER receptor structures, our findings provide a missing link in the HER receptor field, offer the dimerization arm as an allosteric sensor for ligand binding, visualize HER3 in an extended state for the first time, demonstrate how the most frequent oncogenic HER2 variant, HER2 S310F, exploits dimerization arm dynamics to enhance heterodimerization, and unveil previously unknown details on how commonly prescribed biologic agents bind the heterodimer. Our studies on near full-length HER2 and HER3, when isolated alone, surprisingly reveal that HER2 is a homodimer that may adopt an autoinhibited state and HER3, contrary to dogma, homodimerizes in the presence of NRG1b. Taken together, these findings made possible through the lens of full-length receptor biophysics, explain ligand allostery, inform rational drug design, and add nuance to a model of HER2/HER3/NRG1b heterocomplex assembly.

Alternative splicing plays critical roles in normal development and can promote growth and survival in cancer. Genes that have canonical splice variants that function antagonistically can be ectopically expressed to drive malignant progression. Additionally, aberrant splicing, the production of noncanonical and cancer‐specific mRNA transcripts, can lead to loss‐of‐function in tumor suppressors or activation of oncogenes and cancer pathways. Emerging data suggests that aberrant splicing products and loss of canonically spliced variants correlate with stage and progression in malignancy. Not only do these data illuminate roles for alternative splicing in cancer and intersections between alternative splicing pathways and therapy, but they illustrate the importance of understanding the genetic basis of splicing.

Using a transgenic model of neuroblastoma, we set up a backcrossed genetic system to derive distinct virtual genomes of over 100 individual mice in which we have profiled exon expression in two different neural tissues. We identified splicing quantitative trait loci (sQTL) which map genetic control of splicing and define key splicing motifs. We identify these motifs as sites of recurrent somatic mutations in cancer. We also use this analysis to identify novel effector splicing events. Among these, we show that a triplet splicing event within FUBP1 modulates levels of the MYC oncoprotein in human neuroblastoma‐derived cell line, and correlates with outcome in neuroblastoma.