Cold Spring Harbor Laboratory

BackgroundMathematical models that describe sequence-function relationships are widely used in computational biology. A key challenge when interpreting these models is that their parameters are not uniquely determined, i.e., many different parameter choices can encode the same sequence-function landscape. These ambiguities, which are known as “gauge freedoms,” must be removed before parameter values can be meaningfully interpreted. Doing this requires imposing additional mathematical constraints on parameter values, a procedure called “fixing the gauge.” We recently developed mathematical methods for fixing the gauge of a large class of commonly used models, but the direct computational implementation of these methods is often impractical due to the need for a projection matrix whose size scales quadratically with the number of parameters.ResultsHere we introduce GaugeFixer, a Python package that exploits the specific mathematical structure of gauge-fixing projections to achieve linear scaling in both time and memory. This dramatically increases efficiency, enabling application to models with millions of parameters. As one application, we analyzed the local structure of peaks in an empirical fitness landscape for translation initiation. GaugeFixer reveals striking similarities, but also fine-scaled variation, in ribosome binding preferences at different positions relative to the start codon, thereby aiding the interpretation of an otherwise unwieldy fitness landscape.ConclusionsGaugeFixer thus fills an unmet need in the computational tools available for the biological interpretation of sequence-function relationships.

Dietary fat composition modulates host physiology and the gut microbiome, but the long-term effects of specific fat sources and the extent to which these changes resolve after dietary reversal remain incompletely defined. Here, we present a longitudinal multi-omic resource of mice maintained for one year on a purified control diet, seven high-fat diets differing in predominant fat source, or reversal regimens in which animals were switched from high-fat to control diet after 4 or 9 months. We further incorporated two cohorts with distinct pre-existing microbiome configurations to determine how baseline community structure shapes diet-induced remodeling of the gut microbiome ecosystem. By integrating longitudinal phenotyping, fecal metagenomics, fecal metabolomics, plasma metabolomics and lipidomics, and intestinal single-cell RNA sequencing, we defined the shared and dietary fat-specific responses across host and microbiome compartments. Baseline microbiome composition strongly influenced microbial responses to diet, indicating that pre-existing community structure is a major determinant of dietary ecosystem remodeling. Although many altered features shifted toward baseline after dietary reversal, only approximately half of diet-associated microbial changes recovered within the study window. A subset of taxa exhibited persistent alterations, including sustained depletion of Lactobacillus johnsonii and Bifidobacterium pseudolongum and sustained enrichment of Alistipes finegoldii, consistent with a "microbiome memory" of prior high-fat diet exposure. This memory effect is mirrored in the host, by sustained suppression of major histocompatibility complex class II (MHC-II) gene expression in intestinal epithelial cells after dietary reversal. These findings indicate that dietary fats leave a lasting imprint on the host-microbiome interactome that survives dietary intervention. Together, these data establish a resource for defining how dietary fat source, baseline microbiome composition, and dietary history shape host-microbiome states. The entire resource is available online as an RShiny app.

Prostate cancer (PCa) is one of the most common malignancies in men, and growing evidence implicates the gut microbiome as a significant, modifiable contributor to disease evolution and management. Dysbiosis influences PCa biology through effects on inflammation, immune regulation, metabolism, and hormone signaling. Microbial imbalance can promote systemic inflammation and increase intestinal permeability, activating immune signaling pathways such as NF-κB-IL-6-STAT3. In parallel, microbiome-driven metabolic effects, including IGF-1 signaling and microbial androgen synthesis or recycling, may contribute to resistance to androgen deprivation therapy (ADT). Microbial metabolites, notably short-chain fatty acids (SCFAs) and trimethylamine N-oxide (TMAO), exert context-dependent effects on tumor growth, treatment resistance, and progression. Conversely, beneficial microbes have been associated with improved treatment sensitivity and immune regulation. Together, these insights support the gut microbiome as a potential biomarker and emerging therapeutic target in PCa. Modulation strategies, including diet, probiotics, antibiotics, and fecal microbiota transplantation (FMT), are being explored to improve treatment response and address resistance. As mechanistic evidence continues to grow, ongoing monitoring of the gut microbiome may help inform risk stratification and treatment optimization in prostate cancer.

The floral transition in maize represents a pivotal developmental switch that determines flowering time, environmental adaptation, and yield-related traits. However, the molecular mechanisms governing shoot apical meristem reprogramming and cell identity changes during this process remain poorly understood. By integrating time-course bulk RNA-seq, single-cell transcriptomics, chromatin accessibility, and transcription factor binding profiles, we construct a spatiotemporal molecular framework of the maize shoot apical meristem floral transition. Our analyses reveal global transcriptional reprogramming accompanied by pronounced cell type-specific regulation dynamics. At a global level, our transcriptional-level inference suggests that pathways associated with chromatin remodeling, environmental response, and reproductive development are sequentially activated. We further identify a ZmMADS69-ZmRap2.7-ZMM4 regulatory module that fine-tunes the floral transition within the shoot apical meristem. At single-cell resolution, we find that the floral transition is not driven by a uniform transcriptional switch, but instead emerges from the coordinated action of spatially distinct shoot apical meristem domains. Through differential expression, trajectory, and co-expression module analyses, we further identify previously unrecognized roles for the inflorescence regulators UNBRANCHED2 and UNBRANCHED3 in promoting the floral transition, suggesting that they coordinate floral induction with subsequent inflorescence development. Our study establishes a comprehensive spatiotemporal regulatory framework for themaize floral transition, providing mechanistic insights into shoot apical meristem reprogramming and offering a foundation for identifying new regulators to improve maize adaptation and yield.

Abstract Inside the cell nucleus, nuclear speckles are a predominant type of nuclear body present in the cells throughout our bodies. While first observed by Ramon y Cajal over 100 years ago, speckles have only recently been recognized as a major layer of gene regulation that boosts gene expression by assisting multiple steps of RNA production. Large genomic regions spanning 10s of megabases are consistently organized around speckles, with regulation of chromatin position at speckles occurring at finer scales of ∼10kb-1Mb. Hence, speckles may help direct functional programs within a cell by coordinating expression of large, but specific, groups of genes. In cancer, we found that speckles exist in two major states that correlate strongly with patient outcomes in neuroblastoma and clear cell renal cell carcinoma (ccRCC). These speckle states showed dramatic differences in expression of the speckle-associated gene neighborhoods. Because certain types of genes reside within speckle-associated chromatin, speckle states correlate with broad functional programs such as protein translation and oxidative phosphorylation as well as ccRCC-specific functional programs such as HIF-2alpha-induced angiogenesis and metabolism pathways. Consistent with distinct gene regulatory programs, ccRCC tumors display tumor microenvironment differences, particularly in tumor vasculature, between the two speckle states. While these data indicate that speckles can be used as a multifunctional readout for ccRCC tumor phenotypes, what drives speckle states in ccRCC is unclear. In other efforts in the lab, we combined analysis of public data with experimental approaches to uncover that transcription factors, hormone signaling, and immune signaling globally regulate expression of genes within speckle-associated chromatin. These efforts provide clues as to what may regulate nuclear speckle functions. Ultimately, by understanding the upstream drivers and downstream consequences of speckle states in ccRCC, we hope to advance nuclear speckles as potential therapeutic targets as well as biomarkers capable of stratifying patient risk and informing treatment decisions. Citation Format: Sana Mir, Hiroe Namba, Milena Stepic, Kate Alexander. Nuclear speckles in ccRCC [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Innovations in Kidney Cancer Research: From Molecular Insights to Therapeutic Breakthroughs; 2026 Mar 13-16; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(5_Suppl_2):Abstract nr IA025.

A powerful approach to understand the computations carried out by the visual cortex is to build models that predict neural responses to any arbitrary image. Deep neural networks (DNNs) have emerged as the leading predictive models1,2, yet their underlying computations remain buried beneath millions of parameters. Here we challenge the need for models at this scale by seeking predictive and parsimonious DNN models ofthe primate visual cortex. We first built a highly predictive DNN model of neural responses in macaque visual areaV4 by alternating data collection and model training in adaptive closed-loop experiments. We then compressed this large, black-box DNN model, which comprised 60 million parameters, to identify compact models with 5,000 times fewer parameters yet comparable accuracy. This dramatic compression enabled us to investigate the inner workings of the compact models. We discovered a salient computational motif: compact models share similar filters in early processing, but individual models then specialize their feature selectivity by 'consolidating' thisshared high-dimensional representation in distinct ways. We examined this consolidation step in a dot-detecting model neuron, revealing a computational mechanism that leads to a testable circuit hypothesis for dot-selective V4 neurons. Beyond V4, we found strong model compression for macaque visual areasV1 and IT(inferior temporal cortex), revealing a general computational principle of the visual cortex. Overall, our work challenges the notion that large DNNs are necessary to predict individual neurons and establishes a modelling framework that balances prediction and parsimony.

Amyloid β (Aβ) accumulation is a hallmark of Alzheimer's disease (AD). Emerging evidence suggests that impaired microglial Aβ phagocytosis is a key feature in AD, highlighting the therapeutic potential of enhancing this innate immune function. Here, we demonstrate that genetic deletion or pharmacological inhibition of protein tyrosine phosphatase 1B (PTP1B) ameliorated memory deficits and reduced Aβ burden in APP/PS1 mice. Moreover, we show that PTP1B was highly expressed in microglia, and its deficiency promoted a transcriptional shift toward immune activation and phagocytosis. Consistently, PTP1B deletion in microglia enhanced phagocytosis and energy metabolism, supported by increased AKT-mTOR signaling, a pathway essential for meeting the energy demands of activation. Mechanistically, we identified spleen tyrosine kinase (SYK), a key regulator of microglial phagocytosis, as a direct substrate of PTP1B. Inhibition of SYK showed that PTP1B modulates microglial activation in a SYK-dependent manner. These findings established PTP1B as a critical modulator of microglial activation and a potential therapeutic target for AD.