Biology subjects

Richard, G.

Publications and source records attributed to Richard, G..

At least 19 recordsLinked to original sources

bioRxiv2026

Genetic diversity and phylogeography of swine influenza A virus in 2013-2022 in Europe using a harmonized genotyping nomenclature

Swine Influenza A virus (swIAV) poses constant health threat for both humans and animals. This study comprehensively analysed the genetic diversity and evolution of European swIAVs from 2013 to 2022. To address existing classification gaps, an easily accessible full-genome genotyping tool based on Nextclade was developed alongside a global nomenclature system. Application of this tool to European sequence data revealed unprecedented genetic diversity, successfully identifying 150 swIAV genotypes mixing segments of both swine and human origin. Bayesian phylogeographic models, combined with economic data, identified that the trade of live weaner pigs for fattening drove the transboundary spread of swIAVs across Europe. The study detected repeated human-to-swine reverse-zoonotic events across multiple European countries. These findings clearly highlight the role of commercial pig movements in viral dissemination and emphasize the critical importance of harmonized genomic surveillance to monitor swIAV viral evolution to inform global pre-pandemic preparedness strategies.

microbiology↗

bioRxiv2025

Spatiotemporal Gray Matter Plasticity During Chronification of Preclinical Neuropathic Pain

Chronic neuropathic pain is increasingly recognized as a brain disease characterized by time-dependent structural and functional reorganization of key neural circuits. While human imaging studies implicate widespread changes in network connectivity and gray matter density (GMD), animal models enable direct longitudinal mapping of such plasticity. Here, we applied high-resolution structural MRI in a rat model of chronic pain (spared nerve injury, SNI) and quantified GMD changes across 134 brain regions. Dynamic weight bearing analysis confirmed persistent pain in SNI rats, validating the chronic pain phenotype in our experimental cohort. Longitudinal MRI revealed significant GMD alterations in 31 regions, predominantly within limbic, prefrontal, and cingulate circuits, representing 21% of total brain volume. Among this affected volume, over 17% of brain volume demonstrated GMD increases while only ~3% showed GMD decreases, indicating a heterogeneous neuroplastic response. Specifically, the Frontal Association Cortex exhibited an approximate 10% increase in GMD, the Primary Cingular Cortex showed a modest increase of about 2%, and the Amygdalohyppocampic Area demonstrated a ~10% decrease in GMD over 28 days. Primary sensory, parietal, visual, retrosplenial, and temporal cortices remained largely unaffected. No significant changes were observed in healthy animals over the same period, highlighting the specificity of brain reorganization to persistent neuropathic pain. These findings reaffirm the ability of MRI to robustly quantify pain-induced neuroanatomical remodeling but leave open critical questions about the underlying cellular and molecular mechanisms. Future studies integrating histological and molecular approaches are needed to determine the precise substrate and reversibility of these structural changes, with the goal of identifying therapeutic targets to prevent or reverse maladaptive neuroplasticity in chronic.

neuroscience↗

bioRxiv2025

Bimodal masked language modeling for bulk RNA-seq and DNA methylation representation learning

Oncologists are increasingly relying on multiple modalities to model the complexity of diseases. Within this landscape, transcriptomic and epigenetic data have proven to be particularly instrumental and play an increasingly vital role in clinical applications. However, their integration into multimodal models remains a challenge, especially considering their high dimensionality. In this work, we present a novel bimodal model that jointly learns representations of bulk RNA-seq and DNA methylation leveraging self-supervision from masked language modeling. We implement an architecture that reduces the memory footprint usually attributed to purely transformer-based models when dealing with long sequences. We demonstrate that the obtained bimodal embeddings can be used to fine-tune cancer-type classification and survival models that achieve state-of-the-art performance compared to unimodal models. Furthermore, we introduce a robust learning framework that maintains downstream task performance despite missing modalities, enhancing the models applicability in real-world clinical settings.

bioinformatics↗

bioRxiv2025

Shivering, but not adipose tissue thermogenesis, increases as a function of mean skin temperature in cold-exposed men and women

Skin cooling results in the activation of heat generating mechanisms to counteract heat lost to the environment. Here, we aim to understand the extent that variations in cold-stimulated heat production may be driven by differences in the contribution of shivering and non-shivering thermogenesis (NST) and the interaction with biological sex. Using a novel mean skin temperature clamping technique in healthy men and women, our data shows that cold-stimulated heat production rises with increasing shivering and myocardial oxidative metabolism in a skin temperature-dependent fashion. Shivering and myocardial thermogenesis were also moderately associated. In contrast, adipose tissue NST did not increase in a linear manner to reductions in skin temperature. Men and women displayed similar thermoregulatory responses, although women presented more pronounced shivering through a greater recruitment of lower-body muscles and greater number of motor units recruited. Thus, shivering contributes proportionally to cold-induced thermogenesis whereas adipose tissue thermogenesis displays an all-or-none response.

physiology↗

bioRxiv2024

Major change in swine influenza virus diversity in France owing to emergence and widespread dissemination of a newly introduced H1N2 1C genotype in 2020

Swine influenza A viruses (swIAV) are a major cause of respiratory disease in pigs worldwide, presenting significant economic and health risks. These viruses can reassort, creating new strains with varying pathogenicity and cross-species transmissibility. This study aimed to monitor the genetic and antigenic evolution of swIAV in France from 2019 to 2022. Molecular subtyping revealed a marked increase in H1avN2 cases from 2020 onwards, altering the previously stable subtypes distribution. Whole-genome sequencing and phylogenetic analyses of H1av (1C) strains identified ten circulating genotypes, including five new genotypes, marked by a significant predominance of the H1avN2#E genotype. It was characterized by an HA-1C.2.4, an N2-Gent/84, and internal protein-encoding genes belonging to a newly defined genogroup within the Eurasian avian-like (EA) lineage, the EA-DK subclade. H1avN2#E emerged in Brittany, the countrys most pig-dense region, and rapidly became the most frequently detected swIAV genotype across France. This drastic change in the swIAV lineages proportions at a national scale was unprecedented, making H1avN2#E a unique case for understanding swIAV evolution and spreading patterns. Phylogenetic analyses suggested an introduction of the H1avN2#E genotype from a restricted source, likely originating from Denmark. It spread rapidly with low genetic diversity at the start of the epizootic in 2020, showing increasing diversification in 2021 and 2022, and exhibiting reassortments with other enzootic genotypes. Amino acid sequence alignments of H1avN2#E antigenic sites revealed major mutations and deletions compared to vaccine 1C strain (HA-1C.2.2) and previously predominant H1avN1 strains (HA-1C.2.1). Antigenic cartography confirmed significant antigenic distances between H1avN2#E and other 1C strains, suggesting the new genotype escaped from the swine population preexisting immunity. Epidemiologically, the H1avN2#E virus exhibited epizootic hallmarks with more severe clinical outcomes compared to H1avN1 viruses. These factors likely contributed to the spread of H1avN2#E within the pig population. The rapid rise of H1avN2#E highlighted the dynamic nature of swIAV genetic and antigenic diversity, underscoring the importance of adapted surveillance programs to support risk assessment in the event of new outbreaks. This also demonstrate the need to strengthen biosecurity measures when receiving pigs in a herd and to limit trading of swIAV-excreting live swine between European countries.

evolutionary biology↗

bioRxiv2024

Tricellular junction recruitment of the Wave regulatory complex by Sidekick and Lar induces protrusive activity resolving cell intercalation

Cell intercalation, a fundamental morphogenetic process characterized by the exchange of neighboring cells, plays a pivotal role in epithelial tissue development. While the initiation of new junctions remains poorly understood, recent research indicates the involvement of tricellular junction actors. In this study, we explore the contribution of the WAVE regulatory complex (WRC), a critical regulator of branched F-Actin generation, in tissue elongation and cell intercalation within the Drosophila ovarian follicular epithelium. WRC localizes at tricellular junctions, where it orchestrates the generation of highly dynamic protrusions emanating from one cell and extending between the bicellular junctions of neighboring cells. This protrusive activity is essential for the initiation of new junctions in cells located at the extremities of these junctions. Furthermore, our findings indicate that WRC recruitment at tricellular junctions is a redundant process, involving the cooperative action of the two transmembrane proteins Sidekick and Lar. Disruption of this recruitment impairs protrusive activity, cell intercalation resolution, and tissue elongation, thereby mechanistically bridging molecular, cellular and tissular scales. Consequently, this elucidates a critical mechanism underlying epithelial morphogenesis through actin polymerization at tricellular junctions.

cell biology↗

bioRxiv2024

BulkRNABert: Cancer prognosis from bulk RNA-seq based language models

RNA sequencing (RNA-seq) has become a key technology in precision medicine, especially for cancer prognosis. However, the high dimensionality of such data may restrict classic statistical methods, thus raising the need to learn dense representations from them. Transformers models have exhibited capacities in providing representations for long sequences and thus are well suited for transcriptomics data. In this paper, we develop a pre-trained transformer-based language model through self-supervised learning using bulk RNA-seq from both non-cancer and cancer tissues, following BERTs masking method. By probing learned embeddings from the model or using parameter-efficient fine-tuning, we then build downstream models for cancer-type classification and survival-time prediction. Leveraging the TCGA dataset, we demonstrate the performance of our method, BulkRNABert, on both tasks, with signifi-cant improvement compared to state-of-the-art methods in the pan-cancer setting for classification and survival analysis. We also show the transfer-learning capabilities of the model in the survival analysis setting on unseen cohorts. Data and Code AvailabilityIn this paper, we leverage the Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/), which includes bulk RNA-seq samples as well as clinical targets for each patient (cancer type, survival time). For pre-training experiments, this dataset is completed with non-cancerous bulk RNA-seq samples from GTEx (Carithers and Moore, 2015) and ENCODE (de Souza, 2012). Code available at https://github.com/instadeepai/multiomics-open-research Institutional Review Board (IRB)This research does not require IRB approval.

bioinformatics↗

bioRxiv2024

ChatNT: A Multimodal Conversational Agent for DNA, RNA and Protein Tasks

Language models are thriving, powering conversational agents that assist and empower humans to solve a number of tasks. Recently, these models were extended to support additional modalities including vision, audio and video, demonstrating impressive capabilities across multiple domains including healthcare. Still, conversational agents remain limited in biology as they cannot yet fully comprehend biological sequences. On the other hand, high-performance foundation models for biological sequences have been built through self-supervision over sequencing data, but these need to be fine-tuned for each specific application, preventing transfer and generalization between tasks. In addition, these models are not conversational which limits their utility to users with coding capabilities. In this paper, we propose to bridge the gap between biology foundation models and conversational agents by introducing ChatNT, the first multimodal conversational agent with an advanced understanding of biological sequences. ChatNT achieves new state-of-the-art results on the Nucleotide Transformer benchmark while being able to solve all tasks at once, in English, and to generalize to unseen questions. In addition, we have curated a new set of more biologically relevant instructions tasks from DNA, RNA and proteins, spanning multiple species, tissues and biological processes. ChatNT reaches performance on par with state-of-the-art specialized methods on those tasks. We also present a novel perplexity-based technique to help calibrate the confidence of our model predictions. Our framework for genomics instruction-tuning can be easily extended to more tasks and biological data modalities (e.g. structure, imaging), making it a widely applicable tool for biology. ChatNT is the first model of its kind and constitutes an initial step towards building generally capable agents that understand biology from first principles while being accessible to users with no coding background.

genomics↗

bioRxiv2024

SegmentNT: annotating the genome at single-nucleotide resolution with DNA foundation models

Genome annotation models that directly analyze DNA sequences are indispensable for modern biological research, enabling rapid and accurate identification of genes and other functional elements. This capability is paramount as the volume of sequenced genomes rapidly expands, making the need for efficient and accurate annotation methods increasingly critical, particularly in the context of genetic variant prediction and in-silico sequence design. Current annotation tools are typically developed for specific element classes and trained from scratch using supervised learning on datasets that are often limited in size. This approach constrains their performance and ability to generalize to new genomes. Here, we frame the genome annotation problem as instance segmentation and introduce a novel methodology for fine-tuning pre-trained DNA foundation models to segment 14 different genic and regulatory elements at single-nucleotide resolution. We leverage the self-supervised pre-trained model Nucleotide Transformer (NT) to develop a general segmentation model, SegmentNT, capable of processing DNA sequences up to 50kb long. By utilizing pre-trained weights from NT, SegmentNT surpasses the performance of several ablation models and baselines, including convolutional networks with one-hot encoded nucleotide sequences and large models trained from scratch. We demonstrate state-of-the-art performance on gene annotation, splice site and regulatory elements detection throughout the genome. We also leveraged our framework to accommodate two extra DNA foundation models, Enformer and Borzoi, extending the sequence context up to 500kb and enhancing performance on regulatory elements. Finally, we show that a SegmentNT model trained on human genomic elements generalizes to elements of different species, and a multi-species SegmentNT model achieves strong generalization across unseen species. Our approach is readily extensible to additional genomic elements and species. We have made our SegmentNT human and multi-species models, as well as the SegmentEnformer and SegmentBorzoi models, available on our github repository in Jax and HuggingFace space in Pytorch.

genomics↗

bioRxiv2024

The Dystrophin-Dystroglycan complex ensures cytokinesis efficiency in Drosophila epithelia

Cytokinesis physically separates daughter cells at the end of cell division. This step is particularly challenging for epithelial cells, which are connected to their neighbors and to the extracellular matrix by transmembrane protein complexes. To systematically evaluate the impact of the cell adhesion machinery on epithelial cytokinesis efficiency, we performed an RNAi-based modifier screen in the Drosophila follicular epithelium. Strikingly, this unveiled adhesion molecules and transmembrane receptors that facilitate cytokinesis completion. Among these is Dystroglycan, which connects the extracellular matrix to the cytoskeleton via Dystrophin. Live imaging revealed that Dystrophin and Dystroglycan become enriched in the ingressing membrane, below the cytokinetic ring, during and after ring constriction. Using multiple alleles, including Dystrophin isoform-specific mutants, we show that Dystrophin/Dystroglycan localization is linked with unanticipated roles in regulating cytokinetic ring contraction and in preventing membrane regression during the abscission period. Altogether, we provide evidence that, rather than opposing cytokinesis completion, the machinery involved in cell-cell and cell-matrix interactions has also evolved functions to ensure cytokinesis efficiency in epithelial tissues.

cell biology↗

bioRxiv2023

Mapping Cerebellar Anatomical Heterogeneity in Mental and Neurological Illnesses

The cerebellum has been linked to motor coordination, cognitive and affective processing, in addition to a wide range of clinical illnesses. To enable robust quantification of individual cerebellar anatomy relative to population norms, we mapped the normative development and aging of the cerebellum across the lifespan using brain scans of > 54k participants. We estimated normative models at voxel-wise spatial precision, enabling integration with cerebellar atlases. Applying the normative models in independent samples revealed substantial heterogeneity within five clinical illnesses: autism spectrum disorder, mild cognitive impairment, Alzheimers disease, bipolar disorder, and schizophrenia. Notably, individuals with autism spectrum disorder and mild cognitive impairment exhibited increased numbers of both positive and negative extreme deviations in cerebellar anatomy, while schizophrenia and Alzheimers disease predominantly showed negative deviations. Finally, extreme deviations were associated with cognitive scores. Our results provide a voxel-wise mapping of cerebellar anatomy across the human lifespan and clinical illnesses, demonstrating cerebellums nuanced role in shaping human neurodiversity across the lifespan and in different clinical illnesses.

neuroscience↗

bioRxiv2023

A Foundational Large Language Model for Edible Plant Genomes

Significant progress has been made in the field of plant genomics, as demonstrated by the increased use of high-throughput methodologies that enable the characterization of multiple genome-wide molecular phenotypes. These findings have provided valuable insights into plant traits and their underlying genetic mechanisms, particularly in model plant species. Nonetheless, effectively leveraging them to make accurate predictions represents a critical step in crop genomic improvement. We present AO_SCPLOWGROC_SCPLOWNT, a foundational large language model trained on genomes from 48 plant species with a predominant focus on crop species. We show that AO_SCPLOWGROC_SCPLOWNT can obtain state-of-the-art predictions for regulatory annotations, promoter/terminator strength, tissue-specific gene expression, and prioritize functional variants. We conduct a large-scale in silico saturation mutagenesis analysis on cassava to evaluate the regulatory impact of over 10 million mutations and provide their predicted effects as a resource for variant characterization. Finally, we propose the use of the diverse datasets compiled here as the Plants Genomic Benchmark (PGB), providing a comprehensive bench-mark for deep learning-based methods in plant genomic research. The pre-trained AO_SCPLOWGROC_SCPLOWNT model is publicly available on HuggingFace at https://huggingface.co/InstaDeepAI/agro-nucleotide-transformer-1b for future research purposes.

genomics↗

bioRxiv2023

Gap junction-dependent amino acid transfer promotes germ cell growth

Gap junctions allow the exchange of small molecules between cells. How this function could be used to promote cell growth is not yet fully understood. During Drosophila ovarian follicle development, germ cells, which are surrounded by epithelial somatic cells, undergo massive growth. We found that this growth depends on gap junctions between these cell populations, with a requirement for Innexin4 and Innexin2, in the germ cells and the somatic cells, respectively. Translatomic analyses revealed that somatic cells express enzymes and transporters involved in amino acid metabolism that are absent in germ cells. Among them, we identified an amino acid transporter required for germline growth. Its ectopic expression in the germline can compensate for its absence in somatic cells. Moreover, affecting either gap junctions or amino-acid import in somatic cells induces P-bodies in the germ cells, a feature associated with an arrest of translation. Finally, in somatic cells, innexin2 expression and gap junction assembly are regulated by the insulin receptor/PI3K kinase pathway. Overall, these results support the view that metabolic transfer through gap junction promotes cell growth and illustrate how such a mechanism can be integrated into a developmental programme, coupling growth control by extrinsic systemic signals with the intrinsic coordination between cell populations.

developmental biology↗

bioRxiv2023

Basement membrane diversification relies on two competitive secretory routes defined by Rab10 and Rab8 and modulated by dystrophin and the exocyst complex

The basement membrane (BM) is an essential structural element of tissues, and its diversification participates in organ morphogenesis. However, the traffic routes associated with BM formation and the mechanistic modulations explaining its diversification are still poorly understood. Drosophila melanogaster follicular epithelium relies on a BM composed of oriented BM fibrils and a more homogenous matrix. Here, we determined the specific molecular identity and cell exit sites of BM protein secretory routes. First, we found that Rab10 and Rab8 define two parallel routes for BM protein secretion. When both routes were abolished, BM production was fully blocked; however, genetic interactions revealed that these two routes competed. Rab10 promoted lateral and planar-polarized secretion, whereas Rab8 promoted basal secretion, leading to the formation of BM fibrils and homogenous BM, respectively. We also found that the dystrophin-associated protein complex (DAPC) and Rab10 were both present in a planar-polarized tubular compartment containing BM proteins. DAPC was essential for fibril formation and sufficient to reorient secretion towards the Rab10 route. Moreover, we identified a dual function for the exocyst complex in this context. First, the Exo70 subunit directly interacted with dystrophin to limit its planar polarization. Second, the exocyst complex was also required for the Rab8 route. Altogether, these results highlight important mechanistic aspects of BM protein secretion and illustrate how BM diversity can emerge from the spatial control of distinct traffic routes.

cell biology↗

bioRxiv2023

The Nucleotide Transformer: Building and Evaluating Robust Foundation Models for Human Genomics

Closing the gap between measurable genetic information and observable traits is a longstanding challenge in genomics. Yet, the prediction of molecular phenotypes from DNA sequences alone remains limited and inaccurate, often driven by the scarcity of annotated data and the inability to transfer learning between prediction tasks. Here, we present an extensive study of foundation models pre-trained on DNA sequences, named the Nucleotide Transformer, ranging from 50M up to 2.5B parameters and integrating information from 3,202 diverse human genomes, as well as 850 genomes selected across diverse phyla, including both model and non-model organisms. These transformer models yield transferable, context-specific representations of nucleotide sequences, which allow for accurate molecular phenotype prediction even in low-data settings. We show that the developed models can be fine-tuned at low cost and despite low available data regime to solve a variety of genomics applications. Despite no supervision, the transformer models learned to focus attention on key genomic elements, including those that regulate gene expression, such as enhancers. Lastly, we demonstrate that utilizing model representations can improve the prioritization of functional genetic variants. The training and application of foundational models in genomics explored in this study provide a widely applicable stepping stone to bridge the gap of accurate molecular phenotype prediction from DNA sequence. Code and weights available on GitHub in Jax and HuggingFace in Pytorch. Example notebooks to apply these models to any downstream task are available on HuggingFace.

genomics↗

bioRxiv2022

A comparison of intracranial volume estimation methods and their cross-sectional and longitudinal associations with age

Intracranial volume (ICV) is frequently used in volumetric brain magnetic resonance imaging (MRI) studies, both as an adjustment factor for head size and as a variable of interest. Associations with age have been reported in both longitudinal and cross-sectional studies, but results have varied, potentially due to differences in ICV estimation methods. Here, we compared five commonly used ICV estimation methods and their cross-sectional and longitudinal associations with age. T1-weighted cross-sectional MRI data was included for 651 healthy individuals recruited through the NORMENT Centre (mean age = 46.1 years, range = 12.0-85.8 years) and 2,410 healthy individuals recruited through the UK Biobank study (UKB, mean age = 63.2 years, range = 47.0-80.3 years), where follow-up data was also available with a mean follow-up interval of 2.3 years. ICV was estimated with FreeSurfer (eTIV and sbTIV), SPM12, CAT12, and FSL. We assessed Pearson correlations, performed Bland-Altman analysis, and tested the explained variance of sex, height, body weight, and age on pairwise differences between ICV estimation methods. We fitted regression models to test linear and non-linear cross-sectional associations between age and ICV. For the UKB dataset, we further assessed longitudinal ICV change using linear mixed-effects (LME) models. We found overall high correlations across ICV estimation method, with the lowest correlations between FSL and eTIV (r=0.87) and between FSL and CAT12 (r=0.89). Widespread proportional bias was found in the Bland-Altman analyses, i.e., agreement between methods varying as a function of head size. Body weight, age, and sex explained the most variance in the differences between ICV estimation methods, indicating possible confounding by these variables for some estimation methods. In the NORMENT dataset, cross-sectional associations with age were found only for FSL and SPM12, indicating a positive association. For the UKB dataset, we observed negative cross-sectional associations with age for all ICV estimation methods. Longitudinal associations with age were found for all ICV estimation methods, with estimated annual percentage change ranging from -0.291 % to -0.416 % across the sampled age range. This convergence of longitudinal results across ICV estimation methods, in the largest dataset to date, offers strong evidence for age-related ICV reductions in mid- to late adulthood. HighlightsO_LICorrelations between the five assessed estimation methods were very high (r>0.90) with the exception of FSL and eTIV (r=0.87), and FSL and CAT12 (r=0.89). C_LIO_LIExplained variance of estimated ICV differences by body weight, age, and sex indicate possible confounding for some ICV estimation methods. C_LIO_LIPositive cross-sectional associations with age, from adolescence to old age, were observed for the SPM12 and FSL estimation methods in one dataset. C_LIO_LIIn the other dataset, negative cross-sectional associations with age, from mid- to late adulthood, were found for all estimation methods. C_LIO_LILongitudinal ICV changes were observed for all estimation methods, indicating an annual percentage ICV reduction of -0.29 % to -0.42 % in mid- to late adulthood. C_LI

neuroscience↗

bioRxiv2022

High-fructose feeding suppresses cold-stimulated brown adipose tissue glucose uptake in young men independently of changes in thermogenesis and the gut microbiome

Diets rich in added sugars, especially high in fructose, are associated with metabolic diseases such as insulin resistance, and non-alcoholic fatty liver disease. Studies have shown a link between these pathologies and changes in the microbiome and its metabolites. Given the reported associations in animal models between the microbiome and brown or beige adipose tissue (BAT) function, and the alterations in the microbiome induced by high glucose or high fructose diets, we investigated the potential causal link between high glucose or fructose diets and BAT dysfunction in humans. We show that BAT glucose uptake, but not thermogenesis, is impaired by a high fructose but not high glucose diet, in the absence of changes in body mass, the gastrointestinal microbiome, and faecal short-chain fatty acids. We conclude that BAT metabolic dysfunction occurs independently from changes in gut microbiome composition, and earlier than other pathophysiological abnormalities associated with insulin resistance and dyslipidemia during fructose overconsumption in humans.

physiology↗

bioRxiv2021

Masculinization of the X-chromosome in aphid soma and gonads

Males and females share essentially the same genome but differ in their optimal values for many phenotypic traits, which can result in intra-locus conflict between the sexes. Aphids display XX/X0 sex chromosomes and combine unusual X chromosome inheritance with cyclical parthenogenesis. Theoretical and empirical works support the hypothesis that the large excess of male-biased genes observed on the aphid X chromosome compared to autosomes evolved in response to sexual conflicts, by restricting the products of sexually antagonistic alleles to the sex they benefits. However, whether such masculinization of the X affects all tissues (as expected if it evolved in response to sexual conflicts) or is limited to specific tissues remains an open question. Here, we measured gene expression in three different somatic and gonadic tissues of males, sexual females and parthenogenetic females of the pea aphid. We observed a masculinization of the X in each of the studied tissues, with male-biased genes being 2.5 to 3.5 more frequent on the X than expected. We also tested the hypothesis that gene duplication can facilitate the attenuation of conflicts by allowing gene copies to neo- or sub-functionalize and reach sex-specific optima. As predicted, X-linked copies of duplicated genes having their other copies on autosomes were more frequently male-biased (40.5% of the genes) than duplicated autosomal genes (6.6%) or X-linked single-copy genes (32.5%). These results highlight a peculiar pattern of expression of X-linked genes in aphids at the tissue level and provide further support for sex-biased expression as a mechanism to attenuate intra-locus sexual conflicts.

evolutionary biology↗