Biology subjects

Ai, M.

Publications and source records attributed to Ai, M..

12 recordsLinked to original sources

bioRxiv2026

Benchmarking foundation models for resting-state functional magnetic resonance imaging time series in cognitive aging research

Foundation models trained on large neuroimaging datasets could help smaller studies investigate cognition from brain recordings. However, whether the foundation model-derived representations improve assessment of current cognition, prediction of follow-up cognition, or characterization of cognitive change remains unclear. We evaluated nine configurations from five model families that process regional resting-state functional magnetic resonance imaging time series, using 3,103 scans from 1,653 participants across two observational and five intervention aging cohorts. Primary frozen representations showed no consistent superiority over conventional functional connectivity for concurrent cognitive associations or diagnostic classification. Changing the readout improved selected concurrent cognitive associations without updating model weights, indicating that performance depended partly on how representations were extracted. Additional post-training and embedding alignment yielded gains in selected settings but did not consistently improve performance across outcomes. In the primary longitudinal analyses, imaging representations did not establish added value beyond baseline clinical measures. These findings show that readout choice influences concurrent cognitive associations, while the added value of these representations for longitudinal outcomes remains unestablished.

bioengineering↗

bioRxiv2026

Amygdala Persistence in Aging: Replication and Extension of Puccetti et al. (2021)

Age-related differences in affect are well established, but their neural mechanisms remain unclear. In a preregistered replication of Puccetti et al. (2021) and extension, we examined age, affect, left-amygdala negative persistence, and amygdala-ventromedial prefrontal cortex (vmPFC) task-based connectivity using MIDUS 3 daily diary and fMRI data. Older age predicted higher positive and lower negative affect in the daily diary sample (N = 1,168; ps < .001). In the primary diary + fMRI sample (N = 81; ages 48-95, 61.7% female, 82.7% White), greater persistence predicted higher raw negative affect ({beta} = 0.278, 95% CI [0.012, 0.544], p = .020); associations with positive and log-transformed negative affect were nonsignificant (ps [≥] .063). Older age predicted lower persistence in the final fMRI sample (N = 127; {beta} = -0.00260, 95% CI [-0.00504, -0.00016], p = .018), but this association was weaker in the primary diary + fMRI sample (p = .200), with no evidence of mediation. Greater anterior vmPFC connectivity predicted lower raw negative affect ({beta} = -0.106, 95% CI [-0.223, 0.012], p = .039) and log-transformed negative affect (p = .014). In MIDUS Refresher 1, affect estimates were directionally similar but nonsignificant (N = 48; ps [≥] .081), whereas the age-persistence estimate reversed direction. These findings provide partial support for amygdala persistence and task-based connectivity as neural correlates of daily affect, while highlighting variability in the magnitude and replicability of fMRI-behavior associations.

neuroscience↗

bioRxiv2026

Bilingualism Protects Domain-Specific Cognitive Function in Mandarin-Speaking Older Adults

ObjectivesThis study examined whether bilingualism is associated with episodic memory (EM) and executive function (EF) in older Mandarin-speaking adults and whether associations differ by clinical diagnosis. Methods189 Mandarin-speaking older adults completed Mandarin-administered neuropsychological testing, brain MRI, and clinical diagnosis. English proficiency test was administered to determine whether they were bilinguals or monolinguals. Sensitivity analyses were conducted in an education-matched subgroup (16 years). ResultsBilingualism was associated with higher EF, with a significant bilingualism x diagnosis interaction indicating larger bilingual advantages among cognitively impaired participants. However, bilingualism was not associated with EM. Findings were preserved in education-matched analysis. ConclusionsResults support a domain-specific association between bilingualism and executive function in later life, consistent with cognitive maintenance mechanisms preferentially supporting executive processes rather than global protection.

neuroscience↗

bioRxiv2026

Foundation models for linking brain functional representations and clinical outcomes across local intervention studies targeting cognitive aging: Validity and generalizability

The benefits of interventions targeting cognitive aging vary substantially across individuals, largely owing to heterogeneity in aging-related comorbidities. It is necessary to robustly identify neural patterns underlying intervention response and test their generalizability across heterogeneous cohorts. Resting-state functional MRI (rsfMRI) offers a potential pathway, but relying on predefined summary features with conventional methods has limited capacity to capture both within-individual longitudinal variation and between-individual differences, particularly in small and heterogeneous studies. Recent rsfMRI foundation models pretrained on large observational cohorts present a promising alternative by learning transferable spatiotemporal representations from time-series signals. Yet their validity and generalizability in local intervention settings remain unclear. Here, we systematically evaluated rsfMRI foundation models using data from two independent randomized controlled trials of older adults with mild cognitive impairment, testing whether these models can robustly extract longitudinal brain representations that predict post-intervention changes in episodic memory across trials. Foundation models outperformed conventional machine learning and deep learning approaches across both trials. Clinically informed adaptation using an external Alzheimers disease cohort further improved performance and robustness to confounders (i.e., head motion, site, and intervention arm), with accuracy up to 82%. Multivariate decomposition of foundation model embeddings identified latent neural patterns associated with episodic memory change with cross-study consistency at baseline that became more spatially distributed at post-intervention. These findings show that rsfMRI foundation models can enable robust and generalizable identification of latent neural patterns linking longitudinal brain dynamics to individual intervention response, laying the foundation for precision-driven neural target discovery in cognitive aging research. Significance StatementInterventions for cognitive aging show highly variable outcomes across individuals, limiting their clinical effectiveness. This study introduces a foundation model-based approach to identify latent neural patterns underlying individual differences in intervention response using resting-state fMRI. By leveraging pretrained models and domain-adaptive fine-tuning, we demonstrate robust and generalizable prediction of cognitive improvement across independent trials. Our findings suggest that latent brain representations, rather than predefined features, provide a scalable pathway toward precision-driven intervention strategies for aging populations.

bioengineering↗

bioRxiv2025

Tirzepatide Synergizes with Leptin on Weight Loss and Restoring Metabolic Homeostasis in Diet-induced Obesity Model

Leptin resistance limits anti-obesity efficacy. We identified a leptin-sensitizing mechanism through tirzepatide (TZP), a glucagon-like peptide-1 receptor (GLP-1R) and glucose-dependent insulinotropic polypeptide receptor (GIPR) dual-agonist. Our tirzepatide clinical trial revealed that circulating leptin levels at baseline correlated with weight loss efficacy in patients with obesity, suggesting leptin and tirzepatide could interact to achieve stronger effects on weight loss. Next, we utilized the diet-induced obesity (DIO) mice and demonstrated the synergistic effects of tirzepatide and leptin combination (TZP+Lep) on weight loss. TZP+Lep treatment further improved hepatic insulin sensitivity and upregulated thermogenetic gene expression in brown adipose tissue. Metabolic profiling under thermoneutrality revealed TZP+Lep treatment further reduced food intake and increased energy expenditure. Tirzepatide sensitized leptin signaling in hypothalamic pro-opiomelanocortin (POMC) and GLP-1R expressing neurons. TZP+Lep synergistically increased POMC neuronal firing by decreasing the inhibitory postsynaptic input. Together, our work showed combining tirzepatide and leptin as a potential way for better maintenance of metabolic homeostasis in obesity management. HighlightsO_LITirzepatide and leptin synergistically promote weight loss through reduced food intake and increased energy expenditure. C_LIO_LITirzepatide and leptin synergistically improve insulin sensitivity and metabolic homeostasis with altered hepatic and adipose gene expression. C_LIO_LITirzepatide sensitizes leptin signaling in hypothalamic GLP-1R and POMC neurons. C_LIO_LITirzepatide and leptin synergistically increase POMC neuronal firing by decreasing inhibitory postsynaptic input. C_LI

neuroscience↗

bioRxiv2025

AI-assisted modeling of attentional control for intervention engagement

Digital cognitive interventions can improve cognition in older adults, but individual differences in transfer to executive function (EF) remain substantial and are difficult to anticipate before treatment completion. We developed a compact Facial Signature Score (FSS) from passive webcam recordings collected during therapeutic cognitive challenges in BREATHE (NCT04522791; development, N=87) and evaluated it in PSOPT (NCT06005038; external validation, N=39), an independent trial using a different intervention format. The FSS discriminated EF responders under nested cross-validation in BREATHE (AUROC=0.77) and generalized to PSOPT without feature reselection or reweighting (AUROC=0.78). The FSS was associated with continuous EF change in BREATHE (Spearman {rho}=0.49, p<0.001) and after outcome-independent recalibration in PSOPT ({rho}=0.32, p=0.045). Predictive performance was retained when the signature was computed from the first 25% and 50% of intervention sessions, and the FSS added value beyond baseline characteristics and conventional behavioral measures. Passive facial dynamics during therapeutic cognitive challenge may serve as a scalable intervention-process marker for early assessment of cognitive response among older adults.

neuroscience↗

bioRxiv2024

Targeting mechanistic target of rapamycin complex 2 attenuates immunopathology in Systemic Lupus Erythematosus

ObjectiveWe aim to explore the role of mechanistic target of rapamycin complex (mTORC) 2 in systemic lupus erythematosus (SLE) development, the in vivo regulation of mTORC2 by type I interferon (IFN) signaling in autoimmunity, and to use mTORC2 targeting therapy to ameliorate lupus-like symptoms in an in vivo lupus mouse model and an in vitro coculture model using human PBMCs. MethodWe first induced lupus-like disease in T cell specific Rictor, a key component of mTORC2, deficient mice by topical application of imiquimod (IMQ) and monitored disease development. Next, we investigated the changes of mTORC2 signaling and immunological phenotypes in type I IFNAR deficient Lpr mice. We then tested the beneficial effects of anti-Rictor antisense oligonucleotide (Rictor-ASO) in a mouse model of lupus: MRL/lpr mice. Finally, we examined the beneficial effects of RICTOR-ASO on SLE patients PBMCs using an in vitro T-B cell coculture assay. ResultsT cell specific Rictor deficient mice have reduced age-associated B cells, plasma cells and germinal center B cells, and less autoantibody production than WT mice following IMQ treatment. IFNAR1 deficient Lpr mice have reduced mTORC2 activity in CD4+ T cells accompanied by restored CD4+ T cell glucose metabolism, partially recovered T cell trafficking, and reduced systemic inflammation. In vivo Rictor-ASO treatment improves renal function and pathology in MRL/lpr mice, along with improved immunopathology. In human SLE (N = 5) PBMCs derived T-B coculture assay, RICTOR-ASO significantly reduce immunoglobulin and autoantibodies production (P < 0.05). ConclusionTargeting mTORC2 could be a promising therapeutic for SLE.

immunology↗

bioRxiv2024

PP2A catalytic subunit alpha is critically required for CD8+ T cell homeostasis and anti-bacterial responses

While the functions of tyrosine phosphatases in T cell biology have been extensively studied, our knowledge on the contribution of serine/threonine phosphatases in T cells remains poor. Protein phosphatase 2A (PP2A) is one of the most abundantly expressed serine/threonine phosphatases. It is important in thymocyte development and CD4+ T cell differentiation. Utilizing a genetic model in which its catalytic subunit alpha isoform (PP2A C) is deleted in T cells, we investigated its contribution to CD8+ T cell homeostasis and effector functions. Our results demonstrate that T cell intrinsic PP2A C is critically required for CD8+ T cell homeostasis in secondary lymphoid organs and intestinal mucosal site. Importantly, PP2A C deficient CD8+ T cells exhibit reduced proliferation and survival. CD8+ T cell anti-bacterial response is strictly dependent on PP2A C. Expression of Bcl2 transgene rescues CD8+ T cell homeostasis in spleens, but not in intestinal mucosal site, nor does it restore the defective anti-bacterial responses. Finally, proteomics and phosphoproteomics analyses reveal potential targets dependent on PP2A C, including mTORC1 and AKT. Thus, PP2A C is a key modulator of CD8+ T cell homeostasis and effector functions.

immunology↗

bioRxiv2023

Ventromedial Frontoinsular Connectivity is Associated with Long-term Smoking Behavior Change in Aging

A central question in the field of cognitive aging and behavioral neuroscience is what enables some individuals to successfully change their behavior more than others? Smoking is a significant risk factor for cognitive decline, particularly in vulnerable populations, including those who are at an elevated risk for Alzheimers disease (AD). Developing effective smoking reduction strategies is therefore a public health priority. The goal of the current study is to better understand the brain mechanisms underlying long-term smoking behavior change in cognitively normal, but at-risk, older adults. Neuroimaging and human lesional studies have implicated the insula and its functional network in subjective interoceptive awareness of cigarette craving and smoking-cue reactivity. We sought to characterize the extent to which anterior insular resting-state functional connectivity MRI predicted long-term smoking reduction (mean: 2.7 years, range 8 months - 4 years) using a seed-to-voxel approach. Twenty-three (18 women; 26% APOE4 carriers; 61.5 years, SD = 3.7) cognitively unimpaired older individuals who smoked cigarettes at their baseline visit and have a first-degree family history of AD (at least one parent or multiple siblings affected) were included from a prospective longitudinal cohort, PREVENT-AD (Pre-symptomatic Evaluation of Experimental or Novel Treatments for Alzheimer Disease) in the current study. We found that reduced long-term smoking behavior was associated with diminished functional connectivity between bilateral anterior insula and ventromedial prefrontal cortex (vmPFC). In a second pre-registered replication study within a larger, independent sample of one hundred and eighteen cognitively normal older adults who smoked cigarettes at baseline from the UK Biobank (73 women; 27.9 % APOE4 carriers; 60.3 years, SD = 2.7), we found that baseline diminished resting-state functional connectivity between anterior insula and vmPFC predicted long-term smoking reduction (mean 5.2 years; ranging from 3 years to 7 years). To our knowledge, this is the largest study to examine the neural substrates of long-term smoking cessation in human aging. Our results suggest that frontoinsular circuits may be a therapeutic target for smoking reduction and disease prevention in older adults at risk for AD.

neuroscience↗

bioRxiv2023

Spliceosomal mutations decouple 3' splice site fidelity from cellular fitness

The fidelity of splice site selection is thought to be critical for proper gene expression and cellular fitness. In particular, proper recognition of 3'-splice site (3'SS) sequences by the spliceosome is a daunting task considering the low complexity of the 3'SS consensus sequence YAG. Here we show that inactivating the near-essential splicing factor Prp18p results in a global activation of alternative 3'SS, many of which harbor sequences that highly diverge from the YAG consensus, including some highly unusual non-AG 3'SS. We show that the role of Prp18p in 3'SS fidelity is promoted by physical interactions with the essential splicing factors Slu7p and Prp8p and synergized by the proofreading activity of the Prp22p helicase. Strikingly, structure-guided point mutations that disrupt Prp18p-Slu7p and Prp18p-Prp8p interactions mimic the loss of 3'SS fidelity without any impact on cellular growth, suggesting that accumulation of incorrectly spliced transcripts does not have a major deleterious effect on cellular viability. These results show that spliceosomes exhibit remarkably relaxed fidelity in the absence of Prp18p, and that new 3'SS sampling can be achieved genome-wide without a major negative impact on cellular fitness, a feature that could be used during evolution to explore new productive alternative splice sites.

molecular biology↗

bioRxiv2022

Human mesenchymal stem cells and derived extracellular vesicles reduce sensory neuron hyperexcitability and pain-related behaviors in a mouse model of osteoarthritis

Osteoarthritis (OA) is a common degenerative joint disease characterized by joint pain and stiffness. In humans, mesenchymal stem cells (MSCs) and derived extracellular vesicles (MSC-EVs) have been reported to alleviate pain in knee OA. Here, we used the destabilization of the medial meniscus (DMM) mouse model of OA to investigate mechanisms by which MSCs and MSC-EVs influence pain-related behavior. We found that MSC and MSC-EV treated DMM mice displayed improved OA pain-related behavior (i.e. locomotion, digging and sleep) compared to untreated DMM mice. Improved behavior was not the result of reduced joint damage, but rather knee-innervating sensory neurons from MSC and MSC-EV treated mice did not display the hyperexcitability observed in untreated DMM mice. Furthermore, we found that MSC-EVs normalize sensory neuron hyperexcitability induced by nerve growth factor in vitro. Our study suggests that MSCs and MSC-EVs may reduce pain in OA by direct action on peripheral sensory neurons. TeaserMesenchymal stem cells and secreted extracellular vesicles normalize sensory neuron excitability to reduce pain.

neuroscience↗

bioRxiv2021

Functional characterization of ovine dorsal root ganglion neurons reveals peripheral sensitization after osteochondral defect

ObjectiveKnee joint trauma can cause an osteochondral defect (OD), a risk factor for osteoarthritis and cause of debilitating pain in patients. Modelling OD in rodents is difficult due to their smaller joint size. This study proposes sheep as a translationally relevant model to understand the neuronal basis of OD pain. MethodsUnilateral 6 mm deep OD was induced in adult sheep, 2-6 weeks after which dorsal root ganglion neurons (DRG neurons) were cultured from the control and OD side. Functional assessment of neuronal excitability and activity of the pain-related ion channels, TRPV1 and P2X3, was carried out using electrophysiology and Ca2+-imaging. Immunohistochemistry was utilized to verify expression of pain-related proteins. ResultsAn increased proportion of OD DRG neurons (sheep, n = 3, Ctrl neurons, n =15, OD neurons, n = 16) showed spontaneous electrical excitability (p = 0.009, unpaired t-test) and hyperexcitability upon TRPV1 agonist (capsaicin) application (p = 0.04, chi-sq test). Capsaicin also produced Ca2+ influx in an increased proportion of OD DRG neurons isolated (p = 0.001, chi-sq test). By contrast, neither protein expression, nor functionality of the P2X3 ion channel were altered in OD neurons. ConclusionsWe provide evidence of increased excitability of DRG neurons (which is an important neural correlate of pain) and TRPV1 function in an OD sheep model. Our data show that functional assessment of sheep DRG neurons can provide important insights into the neural basis of OD pain and thus potentially prevent its progression into arthritic pain.

neuroscience↗