The Moon undergoes periodic tidal forcing due to its eccentric and oblique orbit around the Earth1. The response to this tidal interaction drives temporal changes in the lunar gravity field and is sensitive to the satellite's internal structure2-4. We use data from the NASA GRAIL spacecraft5-9 to recover the time-varying lunar gravity field, including a degree-3 gravitational tidal Love number, k3. Here, we report our estimated value of k3 = 0.0163 ± 0.0007, which is about 72% higher than that expected for a spherically symmetric moon10. Such a large k3 can be explained if the elastic shear modulus of the mantle varies by about 2-3% between the nearside and farside4, providing an observational demonstration of lateral heterogeneities in the deep lunar interior. This asymmetric structure suggests preservation of a predominantly thermal anomaly of roughly 100-200 K in the nearside mantle that formed surface mare regions 3-4 billion years ago11 and could influence the spatial distribution of deep moonquakes12.

Choice behaviour of animals is characterized by two main tendencies: taking actions that led to rewards and repeating past actions1,2. Theory suggests that these strategies may be reinforced by different types of dopaminergic teaching signals: reward prediction error to reinforce value-based associations and movement-based action prediction errors to reinforce value-free repetitive associations3-6. Here we use an auditory discrimination task in mice to show that movement-related dopamine activity in the tail of the striatum encodes the hypothesized action prediction error signal. Causal manipulations reveal that this prediction error serves as a value-free teaching signal that supports learning by reinforcing repeated associations. Computational modelling and experiments demonstrate that action prediction errors alone cannot support reward-guided learning, but when paired with the reward prediction error circuitry they serve to consolidate stable sound-action associations in a value-free manner. Together we show that there are two types of dopaminergic prediction errors that work in tandem to support learning, each reinforcing different types of association in different striatal areas.

In quantum mechanics, empty space is not void but is characterized by vacuum-field fluctuations, which underlie phenomena such as the Lamb shift1, spontaneous emission, and the Casimir effect2. Due to their quantitatively small relative contributions in free-space atomic physics, they were traditionally overlooked in solid-state systems. Recently, however, the interplay between electronic correlations and quantum electrodynamical effects in low-dimensional systems has become a rapidly advancing area in condensed matter physics3-5, with substantial implications for quantum materials and device engineering. High-mobility two-dimensional electron gases in the quantum Hall regime6 offer an ideal platform to investigate how vacuum electromagnetic fields affect strongly correlated electronic states. Here we demonstrate that adjusting the coupling strength between a two-dimensional electron gas and the vacuum fields of a hovering split-ring resonator leads to a significant reduction in exchange splitting at odd-integer filling factors, along with an enhancement of fractional quantum Hall gaps at filling factors 4/3, 5/3 and 7/5. Theoretical analysis indicates that these effects stem from an effective long-range attractive interaction mediated by virtual cavity photons in regions with strong vacuum electric field gradients. Our findings uncover a new mechanism by which cavity vacuum fields can reshape electronic correlations in quantum Hall systems, establishing a new approach for manipulating correlated quantum phases in low-dimensional materials and paving the way for engineering tailored many-body interactions in compact devices.

The anti-tumour effect of radiotherapy beyond the treatment field-the abscopal effect-has garnered much interest1. However, the potentially deleterious effect of radiation in promoting metastasis is less well studied. Here we show that radiotherapy induces the expression of the EGFR ligand amphiregulin in tumour cells, which reprogrammes EGFR-expressing myeloid cells toward an immunosuppressive phenotype and reduces phagocytosis. This stimulates distant metastasis growth in human patients and in pre-clinical mouse tumour models. The inhibition of these tumour-promoting factors induced by radiotherapy may represent a novel therapeutic strategy to improve patient outcomes.

Electrochemical CO2 reduction into chemicals and fuels holds great promise for renewable energy storage and carbon recycling1-3. Although high-temperature CO2 electroreduction in solid oxide electrolysis cells is industrially relevant, current catalysts have modest energy efficiency and a limited lifetime at high current densities, generally below 70% and 200 h, respectively, at 1 A cm-2 and temperatures of 800 °C or higher4-8. Here we develop an encapsulated Co-Ni alloy catalyst using Sm2O3-doped CeO2 that exhibits an energy efficiency of 90% and a lifetime of more than 2,000 h at 1 A cm-2 for high-temperature CO2-to-CO conversion at 800 °C. Its selectivity towards CO is about 100%, and its single-pass yield reaches 90%. We show that the efficacy of our catalyst arises from its unique encapsulated structure and optimized alloy composition, which simultaneously enable enhanced CO2 adsorption, moderate CO adsorption and suppressed metal agglomeration. This work provides an efficient strategy for the design of catalysts for high-temperature reactions that overcomes the typical trade-off between activity and stability and has potential industrial applications.

Immune checkpoint blockade (ICB) has transformed cancer therapy1,2. The efficacy of immunotherapy depends on dendritic cell-mediated tumour antigen presentation, T cell priming and activation3,4. However, the relationship between the key transcription factors in dendritic cells and ICB efficacy remains unknown. Here we found that ICB reprograms the interplay between the STAT3 and STAT5 transcriptional pathways in dendritic cells, thereby activating T cell immunity and enabling ICB efficacy. Mechanistically, STAT3 restrained the JAK2 and STAT5 transcriptional pathway, determining the fate of dendritic cell function. As STAT3 is often activated in the tumour microenvironment5, we developed two distinct PROTAC (proteolysis-targeting chimera) degraders of STAT3, SD-36 and SD-2301. STAT3 degraders effectively degraded STAT3 in dendritic cells and reprogrammed the dendritic cell-transcriptional network towards immunogenicity. Furthermore, STAT3 degrader monotherapy was efficacious in treatment of advanced tumours and ICB-resistant tumours without toxicity in mice. Thus, the crosstalk between STAT3 and STAT5 transcriptional pathways determines the dendritic cell phenotype in the tumour microenvironment and STAT3 degraders hold promise for cancer immunotherapy.

The known fossil record of crown-group amniotes begins in the late Carboniferous with the sauropsid trackmaker Notalacerta1,2 and the sauropsid body fossil Hylonomus1-4. The earliest body fossils of crown-group tetrapods are mid-Carboniferous, and the oldest trackways are early Carboniferous5-7. This suggests that the tetrapod crown group originated in the earliest Carboniferous (early Tournaisian), with the amniote crown group appearing in the early part of the late Carboniferous. Here we present new trackway data from Australia that challenge this widely accepted timeline. A track-bearing slab from the Snowy Plains Formation of Victoria, Taungurung Country, securely dated to the early Tournaisian8,9, shows footprints from a crown-group amniote with clawed feet, most probably a primitive sauropsid. This pushes back the likely origin of crown-group amniotes by at least 35-40 million years. We also extend the range of Notalacerta into the early Carboniferous. The Australian tracks indicate that the amniote crown-group node cannot be much younger than the Devonian/Carboniferous boundary, and that the tetrapod crown-group node must be located deep within the Devonian; an estimate based on molecular-tree branch lengths suggests an approximate age of early Frasnian for the latter. The implications for the early evolution of tetrapods are profound; all stem-tetrapod and stem-amniote lineages must have originated during the Devonian. It seems that tetrapod evolution proceeded much faster, and the Devonian tetrapod record is much less complete, than has been thought.

Depression is characterized by diverse symptom combinations that can be represented as dynamic networks. While previous research has focused on central symptoms for targeted interventions, less attention has been given to whole-network properties. Here we show that 'network temperature', a novel measure of psychological network stability, captures symptom alignment across adolescence-a critical period for depression onset. Network temperature reflects system stability, with higher values indicating less symptom alignment and greater variability. In three large longitudinal adolescent cohorts (total N = 35,901), we found that network temperature decreases across adolescence, with the steepest decline during early adolescence, particularly in males. This suggests that depression symptom networks stabilize throughout development via increased symptom alignment, potentially explaining why adolescence is a crucial period for depression onset. These findings highlight early adolescence as a key intervention window and underscore the importance of sex-specific and personalized interventions.

Schizophrenia is associated with structural and functional changes in the central nervous system, including the most distal part of it, the retina. However, the question of whether retinal atrophy is present before individuals develop schizophrenia or is a secondary consequence of the disorder remains unanswered. Here we address this question by examining the association between polygenic risk scores for schizophrenia and retinal morphologies in individuals without a schizophrenia diagnosis. We used population data for 34,939 white British and Irish individuals from the UK Biobank. Our robust regression results show that higher polygenic risk scores for schizophrenia were associated with thinner overall maculae, while controlling for confounding factors (b = -0.17, P = 0.018). Similarly, we found that greater polygenic risk scores for schizophrenia specific to neuroinflammation gene sets were associated with thinner ganglion cell inner plexiform layers (b = -0.10, self-contained P = 0.014, competitive P = 0.02). These results provide new evidence for genetic factors that could predispose individuals to heightened neuroinflammatory responses. Over time, these responses could contribute to neurodegenerative processes such as retinal thinning.