Publications

Context:

• Low-dose interleukin-2 (IL-2), an immune signaling protein, can expand regulatory T cells, which restrain excessive immune responses; we asked how treatment changes their activity and whether a blood measurement could help anticipate response.
• We analyzed gene expression in regulatory and conventional T cells from 23 patients in an exploratory, open-label trial covering several inflammatory diseases.

Results:

• After treatment, we observed broad changes in gene activity, most prominently in regulatory T cells.
• Genes normally stimulated by interferons, signals involved in antiviral defense and inflammation, became progressively less active in both cell populations.
• A model using pretreatment interferon-related gene expression showed promise for distinguishing clinical responders from nonresponders.

Impact:

• Our results suggest that IL-2 modifies inflammatory programs beyond increasing regulatory-cell numbers. Measuring these genes before treatment could help identify likely responders, if the model’s predictive performance is confirmed in further patients.

Context:

• Some people carry malaria parasites without symptoms, and their antibodies may reveal how repeated exposure shapes immunity. We compared antibodies recognizing the parasite with antibodies recognizing the body’s own proteins.
• Our cross-sectional study in Côte d’Ivoire included 94 people with mild malaria, 33 asymptomatic carriers and 39 endemic controls; carriers and controls came from one of the three study sites.

Results:

• Asymptomatic carriers had lower overall antibody reactivity to parasite antigens but stronger IgG3 responses against a broader range of self-antigens.
• Targets distinguishing clinical groups included actin and metabolic enzymes; among mild cases, other antibody patterns differentiated settings with different transmission intensities.
• Antibody responses to a mosquito-exposure marker correlated with self-reactive antibody levels.

Impact:

• Our findings suggest that self-reactive antibodies can carry information about malaria exposure and clinical tolerance, alongside parasite-specific responses. These patterns offer candidate markers for further study, while the observational design cannot establish that the antibodies themselves prevent symptoms.

Context:

• Depression is common in systemic lupus erythematosus, a disease in which immunity attacks the body’s own tissues, but useful biological markers remain scarce. We asked whether immune cells, gut microbes and circulating metabolites together distinguish patients with major depression.
• We analyzed 99 participants from LUPIL-2 and assessed the resulting signatures in an independent cohort.

Results:

• Depression was associated with fewer naive, antigen-inexperienced CD4 and regulatory T cells, but more activated memory regulatory cells, which help control immunity; the immune profile classified patients with 94% accuracy in the analysis.
• Gut microbial diversity was lower, and metabolic changes included reduced butyrate, a short-chain fatty acid produced by intestinal microbes.
• Combined analyses linked regulatory-cell activation to microbial and metabolic changes, distinguishing lupus with and without major depression with accuracy reaching 85%.

Impact:

• Our findings identify connected biological features of depression within lupus, providing candidate markers and pathways for further investigation. They support studying immune–microbiota interactions without establishing which changes drive depression or whether modifying them improves symptoms.

Context:

• Antibody production depends on a balance between follicular helper T cells, which support B cells, and regulatory T cells, which control the response. We investigated how the inflammatory signal interleukin-1β (IL-1β) influences that balance in humans.
• We combined cell measurements, single-cell gene-expression analysis and cultures from tonsils, spleen and intestinal lymph nodes, with comparisons to samples from patients with autoimmune disease.

Results:

• Cell states and IL-1 receptor patterns varied between organs; stronger expression of the activating receptor relative to the decoy receptor was linked to a distinctive activated regulatory-cell population.
• IL-1β signaling promoted follicular T-cell activation and the transition from helper to regulatory cells in culture.
• In autoimmune patients, higher IL-1β and circulating follicular regulatory-cell levels correlated with more antibodies targeting the body’s own molecules.

Impact:

• Our work links a defined inflammatory pathway to the cellular control of human antibody responses, identifying interactions to investigate in autoimmunity without assuming that blocking this pathway will necessarily improve disease.

Context:

• Autoimmune and autoinflammatory diseases combine responses against the body’s own tissues and activation of innate inflammatory defenses in different proportions. As part of my research at Sorbonne University, we asked whether immune-cell profiles could clarify their relationships.
• We measured more than 600 blood-cell populations in 443 patients with 15 diseases and 71 healthy volunteers, allowing us to compare immune patterns across diagnoses.

Results:

• Analysis without predefined disease groups identified five clusters of disorders with related immune profiles.
• Two main dimensions distinguished these clusters: the level of inflammation and the tissues affected; regulatory T cells, which restrain immune responses, and innate lymphoid cells, which participate in rapid tissue defense, helped define these patterns.
• A smaller signature of 23 cell populations captured the differences between clusters.

Impact:

• Our findings broaden a single autoimmune–autoinflammatory spectrum into a framework that considers both inflammation and tissue involvement, providing specific cell populations to investigate as shared or disease-specific mechanisms.

Context:

• Low-dose interleukin-2 (IL-2), an immune signaling protein, stimulates regulatory T cells, which help restrain inflammation. We investigated whether treatment also changes the gut microbiota and whether those changes contribute to immune regulation.
• We profiled microbial communities in mice and humans, then transferred microbiota from treated mice into untreated recipients.

Results:

• IL-2 treatment altered the composition of the gut microbiota in both species.
• Transferred microbiota protected recipient mice against experimentally induced colitis and prevented diabetes in a susceptible mouse strain.
• Analysis of microbial genes highlighted pathways involved in producing amino acids and short-chain fatty acids, metabolites that connect microbial activity with host biology.

Impact:

• The transfer experiments indicate that altered gut communities can carry part of the treatment’s protective effect, adding a microbial component to our understanding of regulatory T-cell-directed therapies; the disease-protection experiments were performed in mice.

Context:

• Modified vaccinia Ankara (MVA) is an attenuated vaccinia virus used to deliver vaccine antigens, and the vector itself also stimulates immunity. We exposed human blood to MVA in the laboratory to ask how it changes cell-surface receptors and whether responses differ between healthy donors and people with treated HIV.

Results:

• Many receptor levels changed after exposure, including decreases in CCR5, a receptor that HIV can use to enter cells, across several immune populations.
• Production of the inflammatory signal IL-8 increased more strongly in selected monocytes and dendritic cells from participants with treated HIV.
• Conversely, some receptor decreases in their monocytes and neutrophils were less pronounced than in healthy-donor cells.

Impact:

• These results show that evaluating a vaccine vector involves understanding how it modifies the cells it encounters, as well as the antigen it delivers. Our comparisons identify the recipient’s immune context as a variable that can influence these laboratory responses.

Context:

• Neutrophils contribute to inflammation but are often lost or damaged during sample preparation, leaving them underrepresented in detailed immune studies. We used a fixation method that preserves these cells, together with mass cytometry, to profile blood from treated rheumatoid arthritis patients and controls.

Results:

• A neutrophil population with low expression of the surface protein CD11b was more abundant in patients and associated with disease duration.
• We also identified a T-cell population with a distinct adhesion and cell-killing protein profile that was potentially associated with disease activity.
• When cells were challenged with microbial-receptor stimuli, cytokine responses did not differ between patients and controls.

Impact:

• Our workflow makes it possible to include fragile innate immune cells in broad analyses of arthritis. It identifies candidate populations for follow-up studies while distinguishing changes in cell composition from the functional responses tested, which remained similar between groups.

Context:

• Complex immune disorders can involve more cell markers than a single cytometry panel can measure. We addressed this limitation in treated HIV infection by combining three mass-cytometry panels from each participant and comparing blood immune cells with those of healthy donors.

Results:

• Using CytoBackBone, we computationally assembled profiles containing information on 72 markers, expanding the molecular description available for cell analysis.
• Several immune populations showed increased expression of activation-associated proteins despite suppression of blood HIV, with particularly marked changes in monocytes and neutrophils.
• A separate experimental panel confirmed the identified changes, supporting conclusions drawn from the merged profiles.

Impact:

• The work demonstrates how integrating complementary panels can expose immune abnormalities that persist during effective antiviral treatment. It also provides an example of checking computationally combined measurements with an additional experiment before interpreting the resulting cell phenotypes.

Context:

• Cynomolgus macaques are used to study infections at mucosal surfaces, yet their resident microbial communities need to be characterized when interpreting these models. We followed vaginal and rectal bacteria weekly for 15 weeks in five females, alongside measurements of the reproductive hormone progesterone.

Results:

• The two sites contained distinct bacterial communities; rectal profiles were relatively consistent between animals, whereas vaginal profiles were highly individual.
• The abundance of several vaginal bacterial groups was associated with progesterone levels, connecting microbial variation with the hormonal cycle.
• Vaginal communities contained little Lactobacillus and resembled a human community type commonly associated with dysbiosis, or an altered microbial balance.

Impact:

• Our longitudinal description provides reference information for designing and interpreting mucosal infection experiments. It also highlights differences from commonly observed human vaginal communities that should be considered when using macaques to investigate infection susceptibility or preventive strategies.

Context:

• Flow and mass cytometry describe individual cells by measuring selected proteins, but each experimental panel captures only part of the possible profile. We developed CytoBackBone to combine information from separate panels and obtain a broader description of cell phenotypes—their measured molecular characteristics.

Results:

• The algorithm uses nearest-neighbour imputation: it searches for cells with similar profiles for the markers shared between panels and transfers information between suitable matches.
• We introduced criteria for acceptable, unambiguous matches and illustrated how the resulting R implementation merges profiles using mass-cytometry data.

Impact:

• CytoBackBone offers a practical route to analysing complementary measurements together when they cannot all be acquired in one panel. The resulting profiles contain computationally inferred information, so their interpretation should retain the distinction between markers measured on the same cell and markers linked through matching.

Context:

• Algorithms can divide cytometry data into many cell populations, but deciding whether two populations have the same characteristics remains difficult. We developed CytoCompare to make this comparison systematic, using the distributions of protein-marker measurements across cells rather than relying only on manual inspection.

Results:

• For each pair of cell populations, the package calculates a distance describing their phenotypic difference and a statistical measure of the evidence for that difference. Here, phenotype refers to the pattern of markers expressed by the cells.
• It accepts clustering results from several analysis tools, including SPADE, viSNE/ACCENSE and Citrus, allowing their outputs to be examined within a common workflow.
• Several graphical representations show marker patterns and comparison results, helping users interpret which populations resemble one another and how they differ.

Impact:

• Our method provides a practical basis for comparing clustering algorithms and their settings. It helps evaluate whether computational choices recover similar biological populations, making phenotype comparison a more explicit part of cytometry analysis.

Context:

• Cytometry measures multiple proteins in individual cells, allowing algorithms such as SPADE to group cells with similar characteristics. But obtaining these groups is only the beginning of biological interpretation. We developed SPADEVizR, an R package, to investigate which populations change and how their behaviour relates to other measurements.

Results:

• The package combines statistical comparisons with plots of cell abundance and marker expression, making it possible to examine both how many cells belong to a population and what those cells look like molecularly.
• It can integrate additional biological variables and construct several types of statistical models using population abundances to investigate measured outcomes.
• We also included ways to quantify and visualize clustering quality. Although designed around SPADE, the package can work with results from other clustering methods.

Impact:

• Our software connects cell clustering to questions about experimental conditions and biological responses. By bringing visualization, comparison and quality assessment together, it supports a fuller interpretation of the populations identified in complex cytometry datasets.

Context:

• Severe malaria can affect the brain, kidneys or several organs together, suggesting that different inflammatory processes may be involved. We measured cytokines and chemokines, signalling molecules that coordinate immune responses, in patients with falciparum malaria in Odisha, India, and in healthy or other disease comparison groups.

Results:

• Of 26 signals measured, 19 increased during malaria and helped distinguish these patients from the comparison groups, including patients with sepsis or encephalitis.
• High levels of IL-17, IP-10 and IL-10 were associated with multiple-organ dysfunction. Other combinations of signals distinguished cerebral malaria, involving the brain, from forms combining brain and other organ complications.
• Most patients with severe malaria and acute kidney failure had high IL-17 levels, drawing attention to this pathway in renal complications.

Impact:

• Our findings separate severe malaria into distinct inflammatory patterns rather than treating it as a single biological state. They identify the IL-17 pathway as a candidate for investigating kidney injury, based on associations that motivate further mechanistic work.

Context:

• Lujo and Lassa viruses can cause severe haemorrhagic disease in humans, but their behaviour in animal models differs. We compared gene activity in blood immune cells from infected cynomolgus macaques to investigate how the timing of host responses relates to disease severity.

Results:

• Lujo infection caused mild, nonlethal illness in macaques, whereas the Lassa infections examined caused severe disease. This contrast differs from the severe human cases associated with Lujo virus.
• Inflammatory, antiviral and cell-death gene programmes developed about 72 hours later during Lujo infection. The delay was especially apparent for genes promoting cell death.
• Computational analysis of cell-associated expression patterns suggested that differences in responding natural killer cells, an immune-cell population involved in antiviral defence, could contribute to these contrasting responses.

Impact:

• Our comparison identifies the timing of inflammation and cell-death responses as a potential contributor to arenavirus disease. It directs mechanistic work toward when responses occur, while showing why the macaque Lujo model cannot simply be assumed to reproduce human severity.

Context:

• The Syrian hamster is useful for infectious-disease research, but at the time of this work it lacked publicly available genomic and transcriptomic reference sequences. We set out to build a transcriptome, a catalogue of expressed RNA sequences, from six organs of three adult female hamsters.

Results:

• Sequencing and assembly produced 62,482 longer sequence fragments, alongside additional reads that could not be assembled. Together, these formed a broad reference for identifying expressed genes.
• Comparisons with mouse sequences linked the resource to 9,562 distinct genes and identified 214 almost complete transcripts; many fundamental biological pathways were represented.
• Alignments with Chinese hamster and other mammalian resources supported gene annotation and comparisons between species.

Impact:

• Our dataset supplies the sequence information needed to begin measuring gene activity systematically in this model. It supports genome annotation and the development of expression-analysis tools, making it easier to connect hamster infection experiments with known molecular processes in other mammals.

Context:

• As part of my postdoctoral work at the University of Washington, we asked whether differences in influenza severity reflect different biological responses or different speeds and strengths of the same response. We examined gene and protein measurements from the lungs of mice infected with H5N1 influenza.

Results:

• Using a new geometrical representation of the molecular data, we found that viral dose and four H5N1 mutations mainly changed the magnitude and speed of host responses, rather than which genes increased or decreased.
• Comparisons with H1N1-infected mice revealed similar effects on response timing. These patterns were associated with both illness and the amount of virus present.
• The analyses highlighted the importance of how rapidly host responses develop, particularly at the earliest stages of infection.

Impact:

• Our work shows why comparing samples only at matching times can give an incomplete picture of infection. Following the trajectory of a response helps distinguish a delayed biological programme from a fundamentally different one.

Context:

• Gene-expression analyses often ask whether a gene changes between conditions, but establishing that its activity remains similar is a different statistical question. We developed CDS to address both questions within one framework, using fold change, the ratio between measurements, together with information about measurement uncertainty.

Results:

• The test calculates error probabilities for conclusions of difference or similarity and provides confidence intervals describing the uncertainty around a fold change.
• We first derived the method for normally distributed measurements with known variances, then showed how numerical calculations could extend it to other distributions.
• When measurement variance is known, CDS can also compare single measurements from two conditions. This makes the assumptions about uncertainty especially important when repeated measurements are unavailable.

Impact:

• Our contribution makes similarity an explicit analytical outcome alongside differential expression. The same framework can therefore support questions about both changing and stable genes in large molecular datasets, provided the uncertainty in their measurements can be characterized.

Context:

• African green monkeys are important models for biomedical research, but limited sequence information made it difficult to identify and measure their genes. We built a library of expressed sequence tags, short sequences sampled from active genes, using blood immune cells from 16 animals, including 12 infected with simian immunodeficiency virus.

Results:

• Combining cells exposed to several laboratory stimulation conditions broadened the range of immune-cell activities represented. We obtained 37,787 sequences and assembled overlapping fragments into 14,410 longer sequences.
• Comparison with existing primate resources allowed us to annotate 9,029 assembled sequences and identify 506 transcripts that were almost complete.
• A further 6,576 fragments could represent sequences specific to this monkey species or primate genes that had not yet been described.

Impact:

• Our library supplies reference information for identifying genes and comparing their activity across primates. Its emphasis on blood and immune cells makes it particularly relevant to studies connecting experimental monkey models with human immunology.

Context:

• After liver transplantation, recurrent hepatitis C can progress rapidly to fibrosis, the accumulation of scar tissue that impairs organ function. We asked whether changes in gene activity could identify this progression before it became visible under the microscope, using 111 biopsies collected over time from 57 transplant recipients.

Results:

• Patients who later developed severe fibrosis shared a molecular signature in their liver tissue, linking early gene-expression patterns to subsequent disease progression.
• These changes were detectable before conventional tissue examination showed worsening disease. This timing suggests that important processes were already under way during the early infection of the transplanted liver.
• Different severe clinical outcomes also shared an earlier molecular state, pointing to common processes before their courses diverged.

Impact:

• Our work identifies a period when molecular changes precede visible tissue damage and offers candidate markers for investigating risk. It also focuses attention on early biological events that may influence the long-term fate of the graft.

Context:

• Allergen immunotherapy aims to shift immunity from allergy-driving IgE antibodies toward protective IgG antibodies. We investigated whether interleukin-1 (IL-1), an inflammatory signal, could help establish that protective response in mice.
• We combined genetic and pharmacological experiments with antibody-transfer studies in systemic and food-allergy models.

Results:

• Stronger IL-1 signaling increased allergen-specific IgG and protection against anaphylaxis, whereas blocking its activating receptor reduced IgG and worsened reactions.
• Giving IL-1β together with allergen reduced anaphylaxis in both preventive and therapeutic experiments; protection depended on IgG and the inhibitory antibody receptor FcγRIIb.
• IL-1β also increased this receptor on mast cells, key triggers of allergic reactions, adding a short-term protective effect to the antibody-mediated response.

Impact:

• Our work explains how an inflammatory signal can nevertheless strengthen a brake on allergic reactions. It identifies complementary antibody and mast-cell mechanisms that could inform allergen-immunotherapy research, while human benefit and safety remain to be established.

Context:

• Sequencing immune receptors reveals the diversity of T- and B-cell recognition, but interpreting these large datasets often requires specialized computational skills. We developed AnalyzAIRR to make advanced repertoire analysis more accessible to biologists.
• The R package combines data preparation, statistical comparisons and visualization, with a guided web interface and tutorial.

Results:

• The workflow supports exploration and filtering before comparisons, helping users choose analyses that match their biological questions.
• In a published dataset containing 18 mouse T-cell receptor repertoires, we first identified and removed a major contaminant affecting one sample group.
• Subsequent analyses revealed differences in receptor diversity consistent with the cell populations studied, as well as differences in repertoire convergence, where distinct genetic sequences encode matching receptor regions.

Impact:

• Our example shows why quality control and biological interpretation belong in the same analysis workflow. By bringing complementary measurements into a guided environment, AnalyzAIRR helps researchers investigate immune-receptor diversity without having to assemble each computational step themselves.

Context:

• Allergy treatment would ideally teach the immune system to tolerate a specific allergen without broadly suppressing its defenses. We designed virus-like particles that combine a chosen antigen with CTLA-4, a molecule involved in limiting immune activation.
• We tested how these particles changed immune-cell behavior in culture and whether they protected mice against food allergy.

Results:

• The particles shifted dendritic cells, which present antigens to T cells, toward a tolerance-promoting state and reduced their activation.
• Responses to the incorporated antigen became less active in effector T cells, while regulatory T cells were activated.
• Five daily injections protected mice against allergic symptoms and anaphylactic shock; protection was lasting, antigen-specific and transferable to other mice through regulatory T cells.

Impact:

• Our approach brings the antigen and an immune-regulating signal together in one particle, providing a preclinical strategy for redirecting a defined harmful response toward tolerance. The mouse experiments establish the mechanism’s potential, rather than efficacy in people with food allergy.

Context:

• Multiple sclerosis beginning in childhood can follow very different courses. We sought early clinical and imaging features that could help identify children likely to accumulate relapses or disability.
• We analyzed 70 children in a French cohort and tested two definitions of severe disease, because standard disability scales may miss difficulties experienced by children.

Results:

• A shorter interval between the first two relapses, older age at onset and more lesions on the initial brain scan were associated with a poorer course under both severity definitions.
• Our prediction model combined these clinical features with sex, lesions near the brain’s outer cortex and elevated white-cell counts in cerebrospinal fluid.
• Within this cohort, the model identified 68–76% of severe cases and correctly classified 84–89% of nonsevere cases, depending on the definition used.

Impact:

• These routinely collected measurements could help organize early risk assessment and treatment discussions, but the model must be tested in an independent cohort before its accuracy can be assumed for other children.

Context:

• People with similar knee-joint damage can experience different levels of osteoarthritis pain. We asked whether differences in immune regulation might help explain this variation.
• We profiled blood from 46 patients at comparable radiographic stages, examining immune-cell populations, signaling molecules and gene activity, with particular attention to regulatory T cells that normally restrain inflammation.

Results:

• Pain intensity was associated with a distinct immune profile, including changes in circulating signals linked to regulatory T-cell activity.
• Several cell populations associated with regulatory-cell activation were less abundant as pain increased.
• Regulatory cells from patients with greater pain showed increased expression of inflammatory genes, including pathways involved in innate immune responses and interferon signaling.

Impact:

• Our findings suggest that cells expected to control inflammation may themselves adopt an altered inflammatory state in painful osteoarthritis. This provides specific mechanisms to investigate beyond joint damage, while the blood-based associations do not establish whether these changes cause pain.

Context:

• Regulatory T cells need interleukin-2 (IL-2) to survive and control immune responses, but more growth does not necessarily mean stronger regulatory function. We investigated how two proteins, BLIMP-1 and CEACAM1, balance these processes in human cells.
• Using gene editing, we examined BLIMP-1, which controls gene activity, and CEACAM1, a cell-surface protein that modifies signaling.

Results:

• BLIMP-1 limited cell multiplication while supporting molecules needed for immune suppression; it reduced responsiveness to IL-2 and increased CEACAM1 expression.
• Sustained IL-2 signaling reinforced this feedback mechanism, and CEACAM1 was strongly induced on regulatory cells from autoimmune patients receiving low-dose IL-2.
• In mice carrying human immune cells, BLIMP-1 promoted regulatory protection against graft-versus-host disease, whereas CEACAM1 limited suppressive activity.

Impact:

• Our findings separate the control of regulatory-cell numbers from the control of their function, helping explain why expansion alone is an incomplete measure of an immune therapy’s effects. CEACAM1 also offers a marker of IL-2 signaling in human T cells.

Context:

• Regulatory T-cell therapies aim to restore immune restraint, but transferred cells must survive while preserving their function. We engineered these cells to produce a weaker-acting form of interleukin-2 (IL-2), their growth and survival signal.
• This partial agonist was designed to provide sustained support without the excessive signaling associated with ordinary IL-2 production.

Results:

• Engineered mouse cells expanded and survived for more than a year in mice with intact immune systems, while retaining regulatory identity and a diverse repertoire of antigen receptors.
• Preventive transfer reduced multiorgan autoimmunity in a mouse model created by depleting regulatory T cells.
• Engineered human regulatory cells also survived better in immunodeficient mice and improved control of graft-versus-host disease, in which transferred human immune cells attack mouse tissues.

Impact:

• Our results address a practical obstacle in cell therapy: providing durable support to regulatory cells without overstimulating them. The strategy improved persistence and disease control in preclinical models, establishing a basis for further development.

Context:

• Inflammatory abnormalities have been observed during bipolar depression, including reduced regulatory T cells, which normally restrain immune responses. We tested whether stimulating these cells with low-dose interleukin-2 (IL-2) could produce a measurable biological response.
• In a double-blind proof-of-concept trial, 14 patients received IL-2 or placebo in addition to their existing treatment.

Results:

• The primary outcome was met: IL-2 increased the proportion of regulatory T cells by day five and enhanced their activation.
• Depressive symptoms and overall functioning also improved from day 15 in the treated group, as assessed by secondary outcomes.
• No serious adverse events related to treatment were reported in this small trial.

Impact:

• Our findings connect a targeted change in immune regulation with an initial clinical signal, supporting larger studies of IL-2 as an additional treatment. With only ten treated participants and four receiving placebo, the study does not establish clinical effectiveness or a complete safety profile.

Context:

• Antibody responses develop in germinal centers, where follicular helper T cells support B cells and follicular regulatory T cells restrain them. We asked how interleukin-1 (IL-1), an inflammatory signal, coordinates these opposing functions.
• We selectively disabled IL-1 receptors in these T-cell populations in mice and measured responses after immunization.

Results:

• IL-1β could activate both helper and regulatory cells through the signaling receptor IL-1R1.
• Unexpectedly, inhibitory components carried by regulatory cells mainly limited those cells’ own early activation and suppressive activity, rather than directly blocking the helper-cell response.
• The same receptor system also influenced antibodies directed against the body’s own molecules and disease manifestations in an experimental lupus model.

Impact:

• Our results reveal a timing mechanism: temporarily restraining the regulatory cells allows helper cells to support an antibody response. This provides a more precise framework for investigating how inflammation can support useful antibodies or contribute to autoimmunity.

Context:

• Everyday social and adaptive abilities vary widely among autistic people, even at similar levels of core autism symptoms. We asked whether a reported maternal immune response during pregnancy was associated with this variation.
• We analyzed 295 mother–child pairs from two cohorts, comparing adaptive skills such as socialization according to pregnancy history.

Results:

• A history of maternal immune activation was reported in 14% of the pairs and was associated with greater difficulties in social and adaptive behavior.
• The association remained after accounting for available characteristics, including autism severity, and was reflected in behavioral trajectories over time.

Impact:

• Our findings suggest that pregnancy immune history may help explain differences in functional needs within autism and provides a specific question for prospective studies.
• Because the analysis relied on retrospective reports and a limited set of potential confounders, it cannot show that maternal immune activation caused the observed difficulties.

Context:

• Regulatory T cells help prevent excessive immune responses and may function inadequately in multiple sclerosis. We tested whether low-dose interleukin-2 (IL-2), a signal that supports these cells, could increase their numbers and activation.
• Thirty patients with relapsing-remitting multiple sclerosis received IL-2 or placebo in a double-blind, randomized trial lasting six months.

Results:

• Regulatory T-cell numbers did not increase at day five, the primary assessment, although the cells already showed stronger activation.
• Expansion became apparent at day 15, and the ratio of regulatory to immune-effector T cells remained elevated during treatment.
• New active brain lesions and relapses tended to be less frequent with IL-2, but differences were not statistically significant; the trial was too small to determine clinical benefit reliably.

Impact:

• Our results show a delayed, modest regulatory-cell response in this setting, giving a biological reason to investigate different dosing schedules and assessment times before drawing conclusions about treatment benefit.

Context:

• Because brain development continues after birth, we asked whether inflammatory diseases beginning in childhood are associated with neurodevelopmental difficulties, and whether disease timing matters.
• We studied 195 children with rheumatic, systemic autoimmune or autoinflammatory conditions in the ARTEMIS cohort, then combined evidence from 18 studies in a meta-analysis.

Results:

• Within ARTEMIS, neurodevelopmental disorders were more frequent when inflammatory disease began earlier and disease-modifying treatment started later.
• The subgroup with onset before approximately 11 years and a treatment delay beyond two months had the highest frequency, approaching half of patients.
• Across the published studies, childhood inflammatory diseases were associated with higher odds of neurodevelopmental disorders, and age at disease onset helped explain variation between studies.

Impact:

• Our findings support looking systematically for developmental difficulties alongside inflammatory disease, so that educational and functional needs can be addressed; these observational associations do not establish that inflammation causes the disorders or that earlier treatment prevents them.

Context:

• Hand osteoarthritis affects people in ways that pain scores alone may miss. We asked whether combinations of pain, stiffness, limited hand function and discomfort about appearance could identify distinct patient profiles.
• Using data from 389 participants in the DIGICOD cohort, we grouped patients according to these four symptoms without imposing predefined categories.

Results:

• Five groups emerged, ranging from low overall symptoms to substantial difficulties across all dimensions; another group mainly reported concerns about hand appearance.
• Only about one-third of participants belonged to the two groups with high pain, and appearance-related discomfort was associated with erosive disease and more bony nodes.
• Greater symptom intensity accompanied more joint damage and physical and psychological burden; treatment expectations also differed, although improved function was the most common priority.

Impact:

• Our profiles provide a fuller description of patients’ difficulties and priorities, offering a basis for studying whether different biological mechanisms or care needs underlie apparently similar hand osteoarthritis.

Context:

• Recovery from severe brain inflammation may look reassuring on a physical disability scale while leaving difficulties with learning and everyday functioning. We investigated this after anti-NMDA receptor encephalitis, in which antibodies target a receptor on nerve cells.
• Our retrospective French cohort included 81 children and adolescents followed for at least two years after diagnosis.

Results:

• Most children had relatively favorable outcomes on the overall disability scale, consistent with substantial motor recovery.
• Nevertheless, cognitive impairment and school difficulties remained common, and many children required rehabilitation.
• More symptoms at diagnosis were associated with poorer later cognitive outcomes, while an abnormal initial electroencephalogram was associated with academic difficulties.

Impact:

• Our findings show why physical recovery alone is an incomplete measure of outcome and support follow-up that addresses cognition, education and rehabilitation needs, with initial disease features offering possible indicators of who needs particular attention.

Context:

• Gene-expression studies often focus on which individual genes change between conditions. We explored a complementary question: can the distribution of expression across many genes reveal changes in the biological state of a sample?
• Using a public mouse-pregnancy dataset, we applied Tsallis entropy, a diversity measure that lets the analysis place different weights on abundant and less abundant transcripts.

Results:

• The adjustable measure captured information beyond that provided by conventional diversity indices.
• Changing the emphasis on abundant transcripts revealed how biological conditions affected variability between individuals.
• We also proposed a way to simplify transcriptome datasets based on maximizing differences in diversity between samples.

Impact:

• Our approach brings concepts from ecology into transcriptomics, helping researchers examine collective expression patterns and variability alongside the usual gene-by-gene comparisons.

Context:

• Severe COVID-19 can cause acute respiratory distress syndrome, in which injured lungs transfer oxygen poorly into the blood. We tested whether umbilical-cord-derived stromal cells, investigated for their immune-modulating properties, could improve oxygenation.
• In a multicenter, double-blind trial, 45 adults were randomly assigned to three cell infusions or placebo.

Results:

• The change in oxygenation over seven days, our primary outcome, did not differ significantly between groups.
• Serious adverse events occurred in both groups, but none were attributed to the cell infusions during the reported follow-up.

Impact:

• Our trial provides controlled evidence that this treatment regimen did not demonstrate the expected oxygenation benefit, an important distinction from the cells’ theoretical anti-inflammatory potential.
• These findings inform further trial design, including larger studies and the question of whether administration earlier in lung injury would change the outcome.

Context:

• Vaccination can leave lasting changes in innate immune cells, the cells responsible for rapid, first-line responses. We asked whether the route of administration influences these changes.
• In nonhuman primates, we compared a vaccine based on modified vaccinia Ankara (MVA), an attenuated virus, injected into or beneath the skin; mass cytometry measured many proteins on individual blood cells.

Results:

• The two routes produced different early immune-cell responses and patterns of circulating inflammatory signals.
• After subcutaneous vaccination, blood neutrophils retained features of greater activation or maturity long after injection; we did not observe the same persistent changes after intradermal delivery.

Impact:

• Our findings make delivery route a relevant variable when studying durable changes in innate immunity, rather than attributing those changes to vaccine composition alone.
• The observed cell profiles are consistent with possible immune training, but do not by themselves establish improved protection against a later infection.

Context:

• Follicular helper T cells guide B cells as they mature and produce antibodies. We asked whether these helper cells can develop from both naive CD4 T cells, which have not yet encountered their specific antigen, and memory cells shaped by previous responses.
• We followed this transition in a culture system using human spleen cells.

Results:

• Both starting populations acquired the surface features and functional properties of follicular helper T cells.
• The transition was accompanied by activation of genes and transcription factors that organize helper-cell development.
• Using the same system, we found evidence that HIV infection affects the naive- and memory-cell pathways differently.

Impact:

• Our work identifies more than one route to generating cells that support antibody responses and supplies a human experimental system for examining how infection changes those routes.

Context:

• Inflammation in semen may change how recipient tissues respond to viral exposure. To examine this locally, we exposed living colorectal tissue samples from cynomolgus macaques to SIV, a primate virus related to HIV.
• We compared seminal plasma, the fluid component of semen, with and without signs of genital-tract inflammation.

Results:

• Inflammatory seminal plasma contained higher levels of several immune signaling molecules.
• It increased SIV replication in the tissue cultures, and greater replication was associated with the concentrations of inflammatory and immune-regulating cytokines.
• We also observed greater accumulation of immune cells, including dendritic cells that moved into the tissue’s epithelial barrier.

Impact:

• Our experiments connect the inflammatory state of semen to changes in both viral growth and local immune-cell organization, providing a tissue-level mechanism to investigate in sexual transmission research; transmission between people was not measured.

Context:

• Does the injection site influence the kind of immunity a vaccine produces? We addressed this in nonhuman primates using modified vaccinia Ankara (MVA), an attenuated virus used as a vaccine vector.
• We compared delivery into the skin, beneath the skin and into muscle, following both early immune reactions and later antibody and T-cell responses.

Results:

• Subcutaneous delivery, beneath the skin, generated stronger neutralizing-antibody responses, which can block viral infection.
• Intradermal and intramuscular vaccination instead favored CD8 T cells capable of several immune functions.
• These differences were already reflected in the early recruitment of immune cells and activation of molecular pathways.

Impact:

• Our findings show that vaccine administration is more than a delivery detail: in this primate model, it helped determine the balance between antibody and cellular immunity, providing a variable to consider when designing vaccines for different immune goals.

Context:

• African green monkeys can maintain high levels of simian immunodeficiency virus (SIV) without the chronic intestinal damage seen in pathogenic infection. To investigate this difference, we examined CD8 T cells carrying NKG2a/c receptors, usually associated with natural killer cells, in these monkeys and in SIV-infected macaques.

Results:

• This T-cell population expanded in the blood and gut of African green monkeys, unlike in macaques; the expansion was not observed in lymph nodes.
• Its gene-expression profile included proteins linked to cell killing, movement to the gut, immune regulation and maintenance of the intestinal barrier.
• In macaques, a higher abundance of these cells was associated with lower levels of the inflammatory signal IL-23 in the intestine.

Impact:

• Our findings identify a cell population that may help balance antiviral activity with protection of the intestinal environment. They provide a specific candidate for testing how natural hosts limit inflammation despite persistent infection, with its protective function still to be established.

Context:

• USP18 limits responses to type I interferons, so its abundance influences how strongly a cell reacts to antiviral signals. We investigated whether noncoding RNAs—RNA molecules that do not encode proteins—help regulate the baseline level of USP18 in human cells.

Results:

• Several microRNAs targeted the untranslated region of USP18 RNA, a regulatory sequence outside its protein-coding section.
• Three of these small RNAs were particularly abundant in monocytes, which had low baseline USP18, linking the regulatory mechanism with a specific immune-cell population.
• We also characterized copies of the USP18 regulatory sequence within longer noncoding RNAs, including a testis-specific transcript whose expression correlated with USP18.

Impact:

• The work identifies components of a possible RNA network controlling sensitivity to interferon. It connects direct regulation by small RNAs with tissue-specific expression of longer RNAs, providing a basis for testing how cells tune their antiviral responses.

Context:

• Immune dysfunction can persist long after the start of HIV infection, raising questions about lasting changes in gene regulation. We studied DNA methylation, chemical marks associated with control of gene activity, in CD4 T cells during disease-causing simian immunodeficiency virus (SIV) infection in macaques and nonpathogenic infection in African green monkeys.

Results:

• Methylation changes appeared early and were more pronounced in lymph nodes than in blood, showing that the tissue sampled strongly affects what is detected.
• Affected genes were linked to inflammatory and metabolic pathways in macaques, whereas changes in African green monkeys involved several immune-regulatory genes.
• Selected changes were confirmed in a separate macaque cohort with circulating virus and were absent in animals controlling SIV.

Impact:

• Our analysis identifies early regulatory changes that may help explain persistent differences in immune behaviour. These genes offer targets for experiments testing how changes in methylation influence cellular metabolism and inflammation.

Context:

• Inflammation in Sjögren’s syndrome involves both salivary gland tissue and infiltrating immune cells. We investigated whether interleukin-7 (IL-7) and interferons, proteins that transmit immune signals, form a feedback loop between gland epithelial cells and T cells using patient samples and tissue cultures.

Results:

• Gland cells exposed to interferons released IL-7, providing a route by which inflammation could stimulate a new signal from the tissue.
• IL-7 then activated T cells and increased their production of type II interferon, supplying a potential return signal to the gland.
• Blocking the IL-7 receptor reduced the activity of interferon-responsive genes in cultured gland tissue.

Impact:

• Our findings connect separate immune signals into a mechanism that could sustain local inflammation. They identify IL-7 receptor blockade as a way to interrupt this interaction experimentally and provide a rationale for investigating whether the same approach can benefit patients.

Context:

• Genome sequences, gene-expression measurements and protein datasets describe different aspects of the same organism, but they are often difficult to explore together. We developed Bacnet as a platform for bioinformaticians to build websites that bring these complementary data into a common interface.

Results:

• The platform supports an interactive genome viewer and atlases connecting genes and proteins with their expression measurements.
• It also provides interfaces for co-expression networks, which group genes with related activity patterns, and for examining whether related genes are present in other organisms.

Impact:

• Bacnet supplies reusable components for creating resources tailored to a particular organism, especially species without extensive existing databases. Our aim is to make complex collections of biological data easier for experimental biologists to browse and connect when developing hypotheses.

Context:

• In Sjögren’s syndrome, salivary glands are affected by persistent immune activation, including abnormal B-cell activity. We asked whether the gland’s epithelial cells—the cells forming its glandular structures—actively support that response by culturing patient-derived cells together with B cells from healthy donors.

Results:

• Patient gland cells supported greater B-cell survival than control cells and, after stimulation, induced stronger B-cell activation.
• This support depended on soluble signals released by the cells, but blocking the individual cytokines tested did not remove the effect.
• Leflunomide and inhibitors of BTK or PI3K, proteins involved in intracellular signalling, partly reduced B-cell viability in the model.

Impact:

• Our findings place the gland tissue itself within the network sustaining immune activation. The model helps compare ways of interrupting communication between tissue cells and B cells, identifying treatment mechanisms to pursue beyond blockade of a single soluble signal.

Context:

• Low doses of interleukin-2 (IL-2) can expand regulatory T cells, which help restrain immune attacks against the body. We tested how different doses affected these cells in 24 children with recently diagnosed type 1 diabetes during a year-long randomized, double-blind, placebo-controlled trial.

Results:

• The proportion of regulatory T cells among CD4 T cells increased with dose, although the size and persistence of the response varied substantially between children.
• No serious adverse events occurred; injection-site reactions were more frequent with IL-2, while other reported events were generally mild or moderate.
• Blood-glucose control did not improve significantly compared with placebo, although stronger regulatory-cell responders showed better preservation of C-peptide, a marker of insulin production, than weaker responders.

Impact:

• The trial informs dose selection and shows why individual immune responses need to be followed over time. It supports larger studies connecting regulatory-cell changes with pancreatic function, while this small exploratory trial did not establish clinical benefit.

Context:

• Electroporation uses electrical pulses to help deliver a DNA vaccine into tissue, but it also changes the local immune environment. In nonhuman primates, we examined early skin responses to distinguish the effects of this delivery method from those of the DNA vaccine itself.

Results:

• Electroporation recruited inflammatory cells and promoted the maturation and movement of Langerhans cells, immune cells located in the epidermis.
• The DNA vaccine added signals linked to antiviral defence and immune-cell recruitment; it was also required for strong induction of AIM-2, a sensor of DNA inside cells.
• Together, the two components stimulated complementary innate responses and were followed by strong, persistent T-cell recall responses.

Impact:

• Our findings explain how delivery and vaccine composition can each contribute to the immune response. By separating these contributions, we identify early processes to investigate when refining DNA vaccination, including how the tissue detects the vaccine and communicates with immune cells.

Context:

• The interval between vaccine doses may alter the immune environment encountered by a booster. We compared injections of modified vaccinia Ankara (MVA), an attenuated vaccine virus, given two weeks or two months apart in cynomolgus macaques, measuring antibody responses alongside changes in innate cells involved in early immune defence.

Results:

• The two-month interval produced better antibody responses than the two-week schedule, which showed an impaired response to the second dose.
• Better antibody responses correlated with more activated neutrophil, monocyte and dendritic-cell populations that had accumulated before the later booster.
• These cellular changes were associated with greater production of signalling proteins, or cytokines, by blood cells examined shortly after revaccination.

Impact:

• Our results connect vaccine timing to changes that develop well after the initial inflammatory response. They suggest that following innate cells between doses can help explain differences between schedules and identify biological features to consider when choosing a booster interval.

Context:

• The skin contains immune cells that can respond to vaccination, but the earliest molecular events at the injection site are difficult to observe. We combined mass spectrometry and molecular imaging in human skin maintained in culture to examine the response to an intradermal influenza vaccine.

Results:

• Changes in proteins, lipids and metabolites were detected at four and 18 hours, allowing us to characterize the response at the injection site soon after vaccination.
• In the dermis, the deeper skin layer, protein networks were linked to cell communication and recruitment, including the neutrophil-attracting signal CXCL8.
• In the epidermis, the outer layer, the main networks involved activation of antigen-presenting cells and were associated with CXCL10.

Impact:

• Our approach distinguishes responses within different skin compartments and provides candidate molecular indicators of local vaccine activity. These measurements offer a basis for studying how the injection site initiates immunity and for testing whether early local signals predict later responses.

Context:

• Dendritic cells coordinate immune responses, while monocytes contribute to inflammation and host defence; both change during HIV infection. We used mass cytometry to distinguish their subpopulations in early infection, after a year of effective treatment and in people who naturally control the virus.

Results:

• Higher viral levels were associated with fewer dendritic-cell subsets expressing particular immune-regulatory receptors.
• Early infection was accompanied by a higher abundance of monocyte populations carrying combinations of activation, regulatory and antiviral proteins.
• Natural controllers had more of a dendritic-cell subset expressing both inhibitory receptors and molecules that present antigens to T cells.

Impact:

• Our analysis describes how immune regulation differs within broad cell categories that would otherwise appear uniform. These patterns identify specific populations to investigate in studies of viral persistence and natural control, while the observed associations do not establish that the cells cause those outcomes.

Context:

• Some childhood demyelinating disorders involve antibodies against MOG, a protein in myelin, the insulating covering of nerve fibres. We compared 27 affected children with nine macaques in which MOG triggered experimental disease to assess whether the animal model reproduces relevant human mechanisms.

Results:

• Antibody-positive children and macaques shared increased inflammatory signals in cerebrospinal fluid; this pattern was absent in antibody-negative children.
• Antibodies and complement, an immune protein system that can contribute to tissue injury, were associated with myelin in the examined antibody-positive human and macaque brain samples.
• Macaques also showed antibody deposits in lesions without immune-cell infiltration, suggesting an early role for antibodies in the attack.

Impact:

• The shared features support using this model to investigate disease development and evaluate targeted treatments. Our comparison also defines specific molecular and tissue features to examine across species, although the human brain-tissue evidence came from very few biopsies.

Context:

• Autoimmune demyelinating disease involves immune attacks on myelin, the insulating material around nerve fibres. In cynomolgus macaques, we tested whether delivering the myelin protein MOG to selected skin immune cells could induce tolerance, meaning reduced reactivity toward this specific target.

Results:

• An antibody carrying MOG and targeting a receptor on antigen-presenting cells protected against experimentally induced brain inflammation; the same strategy carrying a control antigen did not.
• Treatment induced MOG-specific regulatory T cells, which are involved in restraining immune responses.
• It also prevented the activation of inflammatory T cells and cytokine production seen in controls.

Impact:

• These experiments provide a preclinical example of redirecting immunity toward tolerance of a defined self-protein. Our findings support further development of targeted approaches for demyelinating disorders in which the relevant immune target is known, with protection demonstrated here in an animal model.

Context:

• Individuals infected with the same virus can maintain very different levels of virus in their blood. We searched for host genetic variants associated with this stable viral level after simian immunodeficiency virus (SIV) infection in Mauritian cynomolgus macaques, using candidate-gene sequencing followed by testing in larger animal groups.

Results:

• A variant that changes the protein sequence encoded by IL37 was associated with viral load in the final group of 44 macaques.
• Two haplotypes of the major histocompatibility complex (MHC), inherited combinations of genes involved in immune recognition, also contributed to the statistical model.
• Together, the IL37 variant and the two haplotypes accounted for 48% of the observed variation in viral load.

Impact:

• This analysis identifies a genetic lead for investigating why some hosts control SIV more effectively than others. It points to IL37 alongside established immune-recognition genes, while functional experiments are needed to determine how the associated variant might affect infection.

Context:

• Nerves within a tumour can influence cancer growth, but their cellular origin was unclear. We combined experiments in mouse prostate-cancer models with human tumour samples to investigate neural progenitors, immature cells capable of developing into neurons.

Results:

• In mice, progenitors left the central nervous system, circulated in the blood and entered prostate tumours, where they could generate new neurons.
• Removing these cells inhibited early tumour development, whereas transplanting them promoted tumour growth and metastasis.
• In human samples, a higher density of neural progenitors was associated with more aggressive prostate cancer and recurrence.

Impact:

• Our findings reveal a route through which the nervous system can contribute new cells to the tumour environment. The mouse experiments establish a functional role in tumour development and identify neural progenitors as a direction for therapeutic research, while the human observations provide an association to investigate further.

Context:

• A vaccine acting in the female reproductive tract encounters a local environment that changes with physiological exposures. We vaccinated nonhuman primates with modified vaccinia Ankara (MVA), an attenuated vaccine virus, to examine how seminal plasma—the fluid component of semen—from HIV-positive donors influences immune responses in these tissues.

Results:

• After exposure, we detected vaccine-specific CD4 T-cell responses that were absent from controls and stronger local CD8 T-cell responses.
• Repeated exposure did not change the antibody response, indicating that the effects differed across components of immunity.
• The cervix showed pronounced recruitment of antigen-presenting cells, which help initiate T-cell responses, alongside increased expression of inflammatory genes.

Impact:

• Our work shows that the mucosal environment can modify the response measured after vaccination. Including relevant exposure conditions in preclinical studies can therefore help interpret local immunity and guide evaluation of vaccines intended to prevent sexually transmitted infections.

Context:

• After cardiac arrest, neurological recovery is difficult to predict during the first hours of hospital care. In this study, we analysed blood gene activity in 69 comatose survivors of out-of-hospital cardiac arrest and compared it with neurological outcomes two months later.

Results:

• A set of 76 transcripts differed at admission between patients with favourable and unfavourable outcomes, and the pattern persisted the following day.
• Poorer recovery was associated with stronger innate inflammatory activity and weaker adaptive immune activity, including reduced expression of genes involved in presenting antigens to immune cells.
• A statistical model combining gene measurements predicted outcomes within this cohort.

Impact:

• The analysis identifies early systemic immune changes associated with later brain recovery and provides candidate markers for developing blood-based prognostic tools. Confirmation in larger populations is needed to determine whether these measurements add reliable information to clinical assessment.

Context:

• Vaccine responses can favour antibodies or T cells with the capacity to kill infected cells, and administration route may influence that balance. We compared three skin or muscle delivery routes for an influenza vaccine in a randomized study of 60 healthy adults.

Results:

• Responses involving cytotoxic CD8 T cells were detected in more participants after vaccination through the skin than after injection into muscle.
• Intradermal and intramuscular vaccination both generated high antibody responses against the tested influenza A strains.
• A combination of nine genes and the inflammatory protein CXCL10, measured one day after vaccination, helped distinguish the immune responses observed three weeks later.

Impact:

• Our findings connect an early blood response to the later type of immunity produced. They offer candidate markers for comparing vaccination routes and assessing whether a strategy elicits the desired immune profile, while protection against infection was not the outcome predicted here.

Context:

• Natural killer cells participate in early defence by responding to abnormal or infected cells. As part of our work on vaccination, we used mass cytometry to ask whether their protein profiles changed after the first and second doses of modified vaccinia Ankara (MVA), an attenuated virus used as a vaccine model, in cynomolgus macaques.

Results:

• The first injection altered the balance of natural killer cell subpopulations, and some changes persisted until the booster.
• Cells responding to the second injection expressed higher levels of proteins associated with cell killing, tissue homing and adhesion, suggesting a different functional state from the initial response.

Impact:

• These observations suggest that a vaccine can leave a lasting imprint on natural killer cells, alongside the better-known memory responses of T and B cells. Our results identify specific cellular features to follow when investigating how the first dose prepares the innate immune system for a booster.

Context:

• Persistent inflammation during HIV infection does not necessarily mean that immune cells respond effectively to a new threat. We tested this by exposing blood cells from people with treated or untreated HIV and healthy controls to signals detected by Toll-like receptors, which recognize microbial molecules.

Results:

• Using a 34-marker mass-cytometry panel, we measured cell identity and cytokine production together, allowing the response to be assigned to specific immune populations.
• Monocytes and conventional dendritic cells produced cytokines more slowly after stimulation through TLR4 or TLR7/8 in both HIV groups than in healthy controls.

Impact:

• The work identifies response timing as a useful dimension of immune dysfunction: cells can exist in an inflammatory environment while reacting slowly to fresh stimulation. Our findings motivate closer investigation of delayed innate signalling, including in people whose HIV infection is being treated.

Context:

• The first response to a vaccine begins at the injection site, but lasting immunity requires coordinated changes across tissues. Here, we combined imaging, cell profiling, inflammatory-protein measurements and gene-expression analysis after injecting modified vaccinia Ankara (MVA), an attenuated vaccine virus, into the skin of nonhuman primates.

Results:

• Local and systemic inflammation rose rapidly and peaked around 24 hours, accompanied by extensive recruitment of early-response immune cells into the skin.
• A brief wave of granulocytes, a family of rapidly responding white blood cells, was followed by a more persistent phase involving inflammatory monocytes.
• Gene-expression patterns confirmed activation of inflammatory pathways, while the initial response was progressively replaced by adaptive immunity directed against the vaccine.

Impact:

• By linking cell movement, molecular signals and timing, we provide a more complete picture of how a local injection initiates a wider immune response. This framework helps identify which early events to examine when comparing vaccine formulations or delivery strategies.

Context:

• CD32a, a cell-surface receptor, had been proposed as a marker of the HIV reservoir: infected cells that persist despite treatment. Its presence in healthy people raised questions about that interpretation, so we characterized CD32a-bearing CD4 T cells during early HIV infection and after effective therapy.

Results:

• Detailed protein measurements revealed several distinct cell subsets rather than a single uniform CD32a-positive population.
• Subsets with additional activating receptors were more abundant during infection, including an activated memory-cell population that was particularly prominent early on.
• Some shifts in these populations remained after effective antiretroviral treatment, while other subsets were more abundant in healthy donors.

Impact:

• Our findings place CD32a expression in the broader context of T-cell activation and differentiation. They help interpret a proposed reservoir marker by showing which cellular populations carry it, without establishing that the receptor specifically identifies cells containing persistent HIV.

Context:

• A vaccine booster encounters an immune system already modified by the first dose. To understand what this means for innate immunity, we compared neutrophils, monocytes and dendritic cells—early defence and immune-coordination cells—after initial and booster vaccination with modified vaccinia Ankara (MVA), an attenuated vaccine virus, in cynomolgus macaques.

Results:

• Inflammation in the blood was lower after the booster, and inflammation at the injection site resolved faster.
• Overall cell counts changed similarly after both injections, but mass cytometry, which measures many proteins per cell, revealed different responding subpopulations.
• Cells most characteristic of the booster response expressed stronger markers linked to activation, antigen presentation and recruitment; the population structure had already changed before the second injection.

Impact:

• These findings make the history of vaccination relevant to the interpretation of innate responses. Our approach detects changes in cell state that simple counts miss and provides a way to compare how successive doses mobilize the immune system.

Context:

• Interferons coordinate antiviral defences, but prolonged activation of their target genes is also associated with chronic inflammation. We asked which interferon types contribute to this persistent response by tracking gene activity in blood and tissues during simian immunodeficiency virus (SIV) infection in macaques.

Results:

• Both type I and type II interferons were induced early in lymph nodes, with weaker responses in rectal tissue; the same induction was not detected in circulating blood cells.
• During chronic infection, type II interferon remained detectably induced, whereas type I interferon did not.
• Many of the most strongly activated genes could respond to either interferon type, and their expression patterns changed with the stage of infection.

Impact:

• Our analysis shows why an interferon-related gene signature cannot automatically be attributed to type I interferon alone. It identifies type II interferon as another contributor to investigate when studying persistent immune activation in SIV and HIV.

Context:

• The female reproductive tract is a major entry site for HIV, yet immune responses there are not fully captured by blood measurements. We used modified vaccinia Ankara (MVA), an attenuated virus used as a vaccine model, in nonhuman primates to ask whether systemic vaccination could generate responses across the different tissues of this tract.

Results:

• Vaccine-specific CD8 T cells with several immune functions were detected throughout the reproductive tract, as well as in blood and lymph nodes, showing that the response extended beyond the circulation.
• After a second dose, antibodies against MVA were present in cervicovaginal fluid, providing evidence of a local antibody response alongside cellular immunity.
• Immune cells were distributed unevenly: memory T cells were more prominent in the vagina and cervix, whereas other immune populations predominated in the uterus and fallopian tubes.

Impact:

• Our findings show why vaccine evaluation should consider the distinct immune environments within mucosal tissues. They establish responses to the MVA vector in this model and provide an anatomical framework for further vaccine studies, including those concerned with HIV entry.

Context:

• Macaques are widely used to study human immunity, but direct comparisons are needed to understand which cellular features they reproduce. We examined blood cells involved in early immune defence from 26 healthy people and five cynomolgus macaques using mass cytometry, which measures many proteins in individual cells.

Results:

• We built broadly comparable antibody panels containing around 30 markers and applied a shared analysis workflow to group cells and compare their molecular characteristics. The panels and clustering markers were similar but not identical between species.
• Comparisons based on 16 marker measurements revealed both shared and differing staining patterns among the cell groups, providing a detailed view of where the human and macaque profiles aligned.

Impact:

• Our work supplies a comparative reference for interpreting innate immune-cell populations across these species. This helps ground the choice and interpretation of macaque models in measured cellular similarities and differences, rather than assuming that corresponding cell populations are interchangeable.

Context:

• Different vaccine designs may stimulate protective immunity against chikungunya, but their effects need to be compared in an appropriate model. We evaluated three strategies in nonhuman primates: a weakened virus, repeated genetic immunization, and a genetic first dose followed by a viral-vector booster carrying chikungunya components.

Results:

• All three approaches protected animals against experimental infection with wild-type chikungunya virus, providing evidence of protection through distinct vaccine designs.
• Although based on one viral lineage, the vaccines also generated antibodies that neutralized an isolate from a different genotype, supporting investigation of protection across viral diversity.
• The weakened-virus vaccine did not produce the clinical signs typically associated with chikungunya in this experiment, including fever, rash, reduced lymphocyte counts or joint swelling.

Impact:

• Our results provide preclinical support for advancing these approaches toward further development. They connect protection in the primate model with antibody breadth and observed tolerability, giving concrete criteria for evaluating candidates before determining their performance in humans.

Context:

• Listeria research generates complementary information about bacterial genomes, gene activity and proteins, but these datasets are often difficult to examine together. We developed Listeriomics as an interactive web platform to connect them, supporting questions about how these bacteria regulate genes and adapt to their environment.

Results:

• An interactive genome viewer places different measurements along the bacterial genome, while expression and protein atlases connect individual genes, RNAs and proteins to relevant experiments.
• A conservation tool shows how proteins are retained across Listeria species, helping users relate molecular observations to evolutionary relationships.
• Co-expression networks connect genes whose activity varies together, and experimental metadata provide the context needed to interpret those relationships across datasets.

Impact:

• Our platform brings several lines of evidence into a shared setting for biological exploration. Researchers can move from a gene of interest to its expression patterns, conservation and associated genes, generating more informed hypotheses about bacterial regulation and responses to stress.

Context:

• After internal contamination with radioactive actinides, researchers measure how these elements accumulate in tissues and leave the body over time. Comparing such experiments is difficult when datasets use different formats and conditions. We developed an R-based tool to organize these measurements and analyse them within a common workflow.

Results:

• A standardized table format connects descriptive statistics, automated figures and curve fitting, while allowing updated data and analysis outputs to be saved consistently.
• The tool includes biokinetic models, mathematical descriptions of how contaminants move between biological compartments and are eliminated, to support comparisons of their behaviour over time.
• We applied the workflow to wound-contamination experiments carried out under different conditions, demonstrating how the modules can be used together on experimental data.

Impact:

• Our approach makes archived animal experiments easier to reanalyse and compare across sources. This supports investigation of how contamination conditions affect distribution in the body and provides a more consistent analytical basis for research on treatment protocols.

Context:

• Heart-transplant rejection is usually assessed by examining small tissue biopsies under the microscope. We asked whether measuring gene activity in the same tissue could reveal rejection-related changes earlier, using 36 selected biopsies from a retrospective cohort of transplant recipients.

Results:

• Biopsies with acute rejection differed from non-rejection samples in 294 genes, mainly related to immune activation and inflammation. Rejection samples nevertheless varied in the strength of their inflammatory response.
• A separate set of 15 gene-expression measurements distinguished biopsies taken one month before rejection. It included the gene encoding titin, a muscle structural protein, suggesting that structural changes may precede overt inflammation.
• A statistical model based on those 15 measurements classified all 36 samples according to their tissue-examination categories within this selected dataset.

Impact:

• Our findings identify molecular changes that could help explain the transition toward rejection and nominate earlier warning markers. The result provides a basis for larger validation studies, rather than an independently validated clinical prediction test.

Context:

• Erythropoietin, or EPO, is a hormone that stimulates red-blood-cell production and has been proposed as an additional treatment for cerebral malaria. Because observations differed between populations, we examined circulating EPO in patients with mild malaria, severe non-cerebral malaria or cerebral malaria in Odisha, India.

Results:

• EPO levels increased with disease severity and were particularly high in fatal cerebral malaria, the form of malaria that affects the brain.
• During cerebral malaria, higher EPO also correlated with heme, a component of haemoglobin, and several inflammatory signals. Their combined associations linked EPO to a broader network of disease-related factors.
• These patterns differed from reports in African populations, indicating that the relationship between EPO and severe malaria cannot be assumed to be uniform across settings.

Impact:

• Our observations help place EPO within the biological context of cerebral malaria and inform the rationale for further treatment studies. They show why population-specific evidence matters when evaluating a proposed therapy, rather than establishing a benefit from administering EPO.

Context:

• Vaccines change the composition of B cells, the immune cells central to antibody responses, but these changes can be difficult to resolve. We used mass cytometry, which measures many protein markers in individual cells, to examine blood B cells in macaques before and after a booster with modified vaccinia Ankara (MVA), an attenuated virus used as a vaccine model.

Results:

• A two-step computational analysis grouped cells by their marker patterns and compared the resulting populations. Vaccination altered several groups, including known memory B-cell subsets and finer combinations of markers.
• One memory-cell population expanded and then contracted in parallel with changes in antibody levels against the vaccine, suggesting a cellular feature associated with the antibody response.
• A different cell-clustering algorithm produced similar findings, supporting the interpretation beyond a single computational method.

Impact:

• Our workflow connects detailed cellular descriptions with changes over time after vaccination. It provides a way to identify which B-cell populations respond and to select candidate cellular markers for more focused studies of vaccine-induced immunity.

Context:

• A viral change that increases replication may also alter how infected tissue responds and repairs itself. We investigated this relationship in mice using a 1918-like avian influenza background, comparing the effects of hemagglutinin, a surface protein, and PB2, a component of the viral replication machinery.

Results:

• Both 1918 components increased replication, but only PB2 increased lethality in this viral background. Viral growth alone therefore did not explain the difference in disease severity.
• The 1918 PB2 protein intensified inflammatory and immune responses, accompanied by greater accumulation of immune cells in the lungs.
• At the same time, it suppressed gene-expression patterns associated with Wnt signalling, a set of pathways involved in lung regeneration and repair, linking stronger inflammation with reduced repair-related signals.

Impact:

• Our experiments connect a viral genetic component with two host processes that can shape lung injury: inflammation and tissue repair. This combination provides a more complete explanation of disease mechanisms than measuring viral replication or inflammation alone.

Context:

• Influenza severity depends partly on the response the virus provokes in the host. We compared lung gene activity in mice infected with emerging H7N9 and other influenza viruses to identify shared responses associated with severe disease, and to explore whether those patterns could suggest treatment candidates.

Results:

• Responses to H7 viruses initially lay between those induced by H5N1 and pandemic H1N1, but became more similar to H5N1 as infection progressed.
• Pathogenic viruses shared three features: stronger expression of inflammatory-response genes, weaker expression of lipid-metabolism genes, and weaker expression of genes involved in blood coagulation. A similar pattern occurred with the 1918 influenza virus.
• We computationally identified six already approved drugs whose known effects might reverse aspects of the H7N9-associated response; their activity against influenza was not established by this analysis.

Impact:

• Our findings connect several severe influenza infections through common host pathways. They provide a rationale for testing whether modifying these responses, rather than targeting the virus alone, could reduce disease.

Context:

• Long non-coding RNAs are RNA molecules that do not encode proteins, and many of their biological roles remain unclear. We measured RNA in the lungs of eight mouse strains infected with influenza or SARS coronavirus to infer which of these molecules might participate in antiviral responses.

Results:

• Infection altered the expression of 5,329 long non-coding RNAs. Many followed patterns similar to protein-coding genes involved in immune responses or normal lung function.
• By comparing these expression patterns with known gene sets, we assigned potential functions to 5,295 long non-coding RNAs and predicted possible effects on nearby genes.
• Experiments using interferon-alpha, an antiviral signalling molecule, confirmed responsiveness among RNAs predicted to be interferon-stimulated, providing experimental support for that part of the analysis.

Impact:

• Our work organizes largely uncharacterized RNAs into biologically meaningful groups and makes the data accessible through MONOCLdb. This gives researchers a way to choose candidates for direct experiments on RNA function, while distinguishing predicted roles from experimentally confirmed responses.

Context:

• Comparing responses across viruses and experimental models requires access to both molecular measurements and clear descriptions of how they were obtained. We assembled a resource from a coordinated systems-biology programme studying influenza and coronaviruses, bringing together gene-expression and protein data that could be reused beyond the original experiments.

Results:

• The collection contains 47 datasets from 30 studies of infection in animals and laboratory cultures, including pandemic H1N1, avian H5N1 and SARS coronavirus.
• Experimental quality checks and comparisons of independent experiments performed under similar conditions helped assess the consistency of the data.
• We deposited primary measurements in public repositories and made processed statistical results available with standardized metadata, the information describing samples and experimental conditions.

Impact:

• Our resource enables questions that require several experiments to be considered together, such as how viral mutations or host genetic background influence responses, and where changes in RNA agree with changes in proteins. It provides a basis for comparative analyses using existing data.

Context:

• Large molecular datasets reveal many changes during infection, but only some may help explain how damaging responses are controlled. We combined gene-expression and protein measurements from influenza- and SARS-coronavirus-infected human lung cells to identify candidate regulators, molecules that could influence broader patterns of host activity.

Results:

• Our network approach combined several ways of ranking genes and proteins, using their relationships across datasets to prioritize a smaller set of potential regulators.
• We then compared the predicted regulatory patterns with data from other respiratory infection models. This identified candidates whose behaviour was conserved across different experimental settings, rather than restricted to a single dataset.

Impact:

• Our integration strategy turns a long list of molecular changes into a more focused set of hypotheses about host regulation. These candidates provide starting points for experiments testing their roles in disease and their potential relevance to treatments directed at the host.

Context:

• The 1918 influenza virus caused unusually severe disease, but efficient viral growth alone might not explain its effects. We investigated this distinction in mice by comparing the contributions of selected 1918 viral components within a contemporary H1N1 background, while measuring both lung infection and host gene activity.

Results:

• Both hemagglutinin, a viral surface protein, and the machinery that copies the viral genome promoted efficient replication in the lungs. Only hemagglutinin produced lethal disease in this comparison.
• Viruses carrying the 1918 hemagglutinin also triggered host gene-expression patterns resembling those caused by the 1918 virus, including strong acute inflammatory and protein-processing responses.
• By contrast, the 1918 replication machinery produced host-response patterns closer to those of the contemporary strain, despite supporting viral growth.

Impact:

• Our results distinguish factors that help influenza multiply from factors that provoke damaging host responses. Within this mouse model and viral background, this separation helps explain why viral replication and disease severity need to be investigated together.

Context:

• Animal models may show similar symptoms after infection without reproducing the same underlying biology. To examine this issue, we compared lung gene-expression responses to the same 2009 pandemic H1N1 influenza virus in mice, macaques and pigs, alongside clinical and virological measurements.

Results:

• Although the three species followed broadly similar clinical courses, the inflammatory molecules they activated and the timing of those responses differed. Similar illness therefore did not imply matching molecular events.
• We also found species-dependent changes in pathways involving the retinoid X receptor, a regulator of inflammatory and metabolic processes. The associated partners, signalling molecules and patterns of gene activity varied across the models.

Impact:

• Our comparison shows why evaluating an infection model requires more than matching symptoms. For research on treatments that act on the host response, these molecular differences help identify where a target discovered in one species may need reassessment in another.

Context:

• Respiratory viruses can cause damage through the host's own response as well as through infection itself. We combined results from 12 studies of influenza and SARS coronavirus to ask whether severe infections share a recognizable pattern of gene activity, despite differences between viruses and experiments.

Results:

• Among the gene sets we compared, a 57-gene signature performed best at separating levels of pathogenicity, meaning the capacity to cause disease, and predicting them in a test dataset.
• Its activity increased progressively from lower- to higher-pathogenicity conditions. Severe infections therefore appeared to amplify a shared response rather than activate an entirely separate set of genes.
• The signature included chemokines, signalling molecules that guide immune cells toward tissues, pointing to altered cell recruitment as a possible contributor to injury.

Impact:

• Our analysis connects infection severity to the strength of a common host-response programme. It also provides a focused hypothesis for experiments testing whether excessive inflammatory-cell recruitment contributes to tissue damage.

Context:

• Multidimensional scaling represents complex datasets as maps in which distances between points reflect differences between samples. Producing a faithful map can require substantial computation. Here, we investigated whether choosing a better starting arrangement could improve both efficiency and the quality of the final representation.

Results:

• Starting from the first principal components obtained by singular value decomposition, a mathematical summary of the main patterns in a dataset, improved the mapping procedure in our comparisons.
• Using these initial components was more efficient than iterating through all components. Adding randomness to the subsequent optimization did not provide the additional accuracy previously suggested.
• We also introduced a quality measure based on nearest neighbours: it checks whether points that are close in the original data remain close on the map.

Impact:

• Our approach makes it easier to construct and assess maps of high-dimensional data, including gene-expression datasets. It combines a practical starting strategy with a check on relationships that a visually convincing plot might otherwise obscure.

Context:

• The mineralocorticoid receptor allows cells to respond to aldosterone, a hormone involved in salt balance. Mutations in this receptor cause pseudohypoaldosteronism type 1, but their effects can differ. We tested four mutations found in affected families to understand how they alter hormone binding and gene activation.

Results:

• All four mutations impaired aldosterone binding, although the degree of impairment varied. Two receptors retained partial activity in reporter assays, which measure their ability to switch on a test gene.
• The other two receptors showed little or no activity in those assays and failed to activate the naturally occurring target genes examined in kidney cells.
• Individual genes did not respond uniformly: a mutation could preserve activation of some targets while weakening or disrupting others.

Impact:

• Our findings show that receptor dysfunction cannot always be summarized as a single loss of activity. The particular genes affected provide a possible molecular explanation for differences in disease features between patients.

Context:

• Recovering from an intestinal infection involves both stopping tissue damage and removing the pathogen. We used a mouse model of Salmonella diarrhoea to ask how these tasks are shared between secreted antibodies and the microbiota, the community of bacteria living in the gut.

Results:

• Antibodies directed against Salmonella limited its access to the intestinal surface and protected against inflammation when mice encountered the same pathogen again.
• However, mice lacking these antibodies cleared bacteria from the intestinal lumen, the gut's interior space, at a similar rate. Antibody protection therefore did not explain pathogen elimination.
• Mice with a simplified microbiota failed to clear the infection despite producing antibodies. Transferring a normal, complex bacterial community restored clearance.

Impact:

• Our experiments separate two complementary forms of protection: the microbiota helps eliminate intestinal Salmonella, while specific antibodies reduce disease on re-exposure. This distinction helps explain why recovery involves more than the antibody response alone.

Context:

• In this chapter, we consider future directions for animal experimentation as part of a French handbook on the general principles of research involving animals.

Results:

• Our contribution addresses perspectives on the field’s future within a handbook that presents research practice alongside animal-welfare considerations.

Impact:

• The chapter offers researchers, veterinarians and technical staff a wider perspective on the context in which animal experiments are planned and conducted, complementing the handbook’s practical and scientific guidance.

Context:

• Selecting a promising vaccine is difficult when protective immunity can only be assessed much later. In this review, we examined whether early molecular and cellular measurements can help anticipate vaccine responses.
• We brought together evidence from animal models and human populations, considering both immunogenicity—the ability to provoke an immune response—and protection against infection.

Results:

• Our synthesis describes immune signatures, or combinations of measurements, investigated as predictors across different vaccination settings.
• We explain how large-scale molecular profiling and computational models can connect early responses with later immune outcomes, alongside the contribution of preclinical models to this work.

Impact:

• The review connects experimental measurements to decisions about vaccine development and individualized vaccination, while highlighting the continuing challenge of identifying markers that reliably predict protection in humans.

Context:

• Cytometry measures many molecular markers in individual cells, producing datasets that require several complementary analysis steps. In this book chapter, we introduce the workflow for analysing such data, from storage and preprocessing to the identification and interpretation of cell populations.

Results:

• We cover manual analysis alongside automated approaches, including SPADE, Citrus and viSNE for grouping and visualizing cells, to place different analytical methods within the same workflow.
• The chapter also addresses interpretation of cell clusters, integration with additional data, mathematical modelling and sharing of the resulting datasets.

Impact:

• Our overview helps readers connect practical data handling with the biological questions addressed by cytometry. It offers a map of the analysis process so that individual tools can be considered in relation to the preceding steps and the intended interpretation.

Context:

• In this editorial, I consider the challenge of navigating high-dimensional biological data, where many measurements must be interpreted together.

Results:

• I address methodological themes at the intersection of computational biology, biostatistics and reproducibility.

Impact:

• My editorial introduces the methodological questions involved in making complex biological datasets interpretable.

Context:

• Understanding influenza severity requires looking beyond viral genes to the cellular responses they trigger. In this review, we examine how large-scale measurements of genes and proteins, combined with computational analysis, have changed the study of influenza and its interactions with the host.

Results:

• We bring together evidence linking uncontrolled inflammation to the extreme virulence of the 1918 and avian H5N1 viruses, and compare those responses with the more recent pandemic H1N1 virus.
• Genetically diverse mouse models offer a way to investigate why host genetic differences affect infection outcomes. We discuss their role in making these comparisons systematic.
• We also review screens that reduce individual genes' activity across the genome, revealing hundreds of cellular factors involved in viral replication.

Impact:

• Our synthesis connects complementary experimental approaches to a shared question: which interactions between virus and host determine disease? This framework helps organize hypotheses about host biology and the development of future drugs and vaccines.