Biology subjects

Lee, H.

Publications and source records attributed to Lee, H..

At least 127 records · Page 7Linked to original sources

bioRxiv2025

REV1 inhibition enhances trinucleotide repeat mutagenesis

Trinucleotide repeat (TNR) instability has been implicated in the pathogenesis of numerous neurodegenerative disorders. Because TNR instability causes mutagenesis of the underlying gene, we refer to the repeat instability phenomenon as TNR mutagenesis in this study. While germline expansions destabilize TNR to cause disease anticipation, somatic cell TNR instability drives earlier onset of symptoms and further disease progression. However, the drivers behind these repeat length changes remain unclear. Current models suggest that DNA replication slippage events and the action of genome instability pathways, such as DNA repair, cause TNR mutagenesis. Whether mutagenic polymerases from the translesion synthesis (TLS) pathway result in TNR instability is unclear. TLS polymerases are best at bypassing difficult-to-replicate DNA regions due to bulky lesions or gaps in DNA. While some effects of TLS polymerases on TNR instability have been explored in lower organisms, evidence in human cells is lacking. Using a quantitative GFP reporter with expanded CAG repeats, we show that inhibition of the TLS polymerase REV1 by its inhibitor, JH-RE-06, or siRNA knockdown increases TNR instability and the underlying mutability. These results suggest that REV1 protects Trinucleotide repeat length mutagenesis through potential continuous DNA synthesis when replicative polymerases stall ahead of repeat secondary structures. Collectively, we present evidence of the role of the TLS pathway in TNR instability, with potential implications for understanding mutability mechanisms, disease biology, and therapeutic targeting.

genetics↗

bioRxiv2025

Polyclonal origins of human premalignant colorectal lesions

Cancer is generally thought to be caused by expansion of a single mutant cell1. However, analyses of early colorectal cancer lesions suggest that tumors may instead originate from multiple, genetically distinct cell populations2,3. Detecting polyclonal tumor initiation is challenging in patients, as it requires profiling early-stage lesions before clonal sweeps obscure diversity. To investigate this, we analyzed normal colorectal mucosa, benign and dysplastic premalignant polyps, and malignant adenocarcinomas (123 samples) from six individuals with familial adenomatous polyposis (FAP). Individuals with FAP have a germline heterozygous APC mutation, predisposing them to colorectal cancer and numerous premalignant polyps by early adulthood4. Whole-genome and/or whole-exome sequencing revealed that many premalignant polyps--40% with benign histology and 28% with dysplasia--were composed of multiple genetic lineages that diverged early, consistent with polyclonal origins. This conclusion was reinforced by whole-genome sequencing of single crypts from multiple polyps in additional patients which showed limited sharing of mutations among crypts within the same lesion. In some cases, multiple distinct APC mutations co-existed in different lineages of a single polyp, consistent with polyclonality. These findings reshape our understanding of early neoplastic events, demonstrating that tumor initiation can arise from the convergence of diverse mutant clones. They also suggest that cell-intrinsic growth advantages alone may not fully explain tumor initiation, highlighting the importance of microenvironmental and tissue-level factors in early cancer evolution.

cancer biology↗

bioRxiv2025

BOGO: A Proteome-Wide Gene Overexpression Platform for Discovering Rational Cancer Combination Therapies

O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/673780v2_ufig1.gif" ALT="Figure 1"> View larger version (56K): org.highwire.dtl.DTLVardef@19d5d7dorg.highwire.dtl.DTLVardef@64f7eborg.highwire.dtl.DTLVardef@d08b75org.highwire.dtl.DTLVardef@173c448_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical AbstractC_FLOATNO C_FIG Cancer drug resistance remains a major barrier to durable treatment success, often leading to relapse despite advances in precision oncology. While combination therapies are being increasingly investigated, such as chemotherapy with small molecule inhibitors, predicting drug response and identifying rational drug combinations based on resistance mechanisms remain major challenges. Therefore, a proteome-wide, single-gene overexpression screening platform is essential for guiding rational therapy selection. Here, we present BOGO (Bxb1-landing pad human ORFeome-integrated system for a proteome-wide Gene Overexpression), a robust, scalable, and reproducible screening platform that enables single-copy, site-specific integration and overexpression of [~]19,000 human open across cancer cell models. Using BOGO, we identified drug-specific response drivers for 16 chemotherapeutic agents and integrated clinical datasets to uncover proliferation and resistance-associated genes with prognostic potential. Drug response similarity networks revealed both shared and unique mechanisms, highlighting key pathways such as autophagy, apoptosis, and Wnt signaling, and notable resistance-associated genes including BCL2, POLD2, and TRADD. In particular, we proposed a synergistic combination of the BCL2 family inhibitor ABT-263 (Navitoclax(R)) and the DNA analog TAS-102 (Lonsurf(R)), which revealed that lysosomal modulation is a key mechanism driving DNA analog resistance. This combination therapy selectively enhanced cytotoxicity in colorectal and pancreatic cancer cells in vitro, and demonstrated therapeutic benefit in vivo in both cell line-derived xenograft (CDX) and patient-derived xenograft (PDX) models. Together, these findings establish BOGO as a powerful gene overexpression perturbation platform for systematically identifying chemoresistance and chemosensitization drivers, and for discovering rational combination therapies. Its scalability and reproducibility position BOGO as a broadly applicable tool for functional genomics and therapeutic discovery beyond cancer resistance.

systems biology↗

bioRxiv2025

BIK1-RIN4 complex links plant immunity and pathogen virulence

Plants deploy receptor-like cytoplasmic kinases (RLCKs) and the immune hub protein RIN4 to coordinate PAMPs-triggered immunity (PTI). Here, we show that BIK1, but not its close homolog PBL1, dynamically strengthens the interaction with RIN4 upon flg22 perception, thereby promoting phosphorylation of RIN4 at S141 and potentiating PTI responses. We further demonstrate that the bacterial type III effector AvrB disrupts the BIK1-RIN4 association through phosphorylation of RIN4 at T166, abolishing complex formation and attenuating BIK1-RIN4 complex-dependent PTI. These findings newly elucidate the BIK1-RIN4 complex as a convergent node of plant immunity and effector targeting, providing insight into the molecular arms race between host defenses and pathogen virulence strategies.

plant biology↗

bioRxiv2025

An integrated in silico-in vitro workflow for discovering high-affinity, selective antibodies to the KRAS(G12D)-MHC I complex

Antibodies that recognize peptide-loaded class I major histocompatibility complex (pMHC I) molecules could enable therapeutic targeting of intracellular oncogenic proteins, yet their discovery has been hampered by the small size of peptide antigens and allele-specificity. We describe an integrated in silico-in vitro workflow for generating high-affinity, selective antibodies to KRAS(G12D)10 presented by HLA-C*08:02, a clinically validated cancer neoantigen. In silico, multiple human antibody-derived variable fragments (Fvs) plausibly docked to the target pMHC were generated, followed by limited complementarity-determining region (CDR) sequence design. In vitro, CDR diversity was introduced at 3-4 positions per Fv to construct yeast surface display library for iterative selections. This workflow yielded antibodies with exclusive binding to KRAS(G12D)10/HLA-C*08:02 without cross-reactivity. Affinity maturation achieved nanomolar dissociation constants, and incorporation into chimeric antigen receptor T cells enabled specific activation against target-positive cells. This study establishes a practical design-to-function pipeline for TCR-like antibody discovery, and demonstrates the feasibility of therapeutic targeting against KRAS(G12D)-driven malignancies.

biochemistry↗

bioRxiv2025

Structural basis for selective remdesivir incorporation by SARS-CoV-2 RNA polymerase, and S759A resistance

Nucleoside analogs (NAs) have been successfully used to treat viral infections. dNTP analogs are primarily DNA chain terminators, while NTP analogs, such as remdesivir, can inhibit as delayed chain terminators or when in the template strand. Determining the frequency of remdesivir triphosphate (RTP) incorporation in the presence of the competing ATP can help in understanding different modes of viral RNA-dependent RNA polymerase (RdRp) inhibition by NTP analogs. We employed enzymatic assays, mass spectrometry, and cryo-EM to show that SARS-CoV-2 RdRp preferentially incorporates RTP, outcompeting 10-fold higher ATP concentration; however, successive RTP incorporations are less favoured when ATP is present. Structures of SARS-CoV-2 RdRp in this and previous studies demonstrate resilience of remdesivir:UMP base pair to translocation, explaining the reduced preference for conjugate incorporations. Together, the RTP versus ATP incorporation is driven by their relative concentration and structural rigidity of remdesivir:UMP, ultimately limiting the number of incorporated remdesivir in a fully synthesized RNA strand. The S759A mutant confers RTP resistance, and the structures of S759A RdRp catalytic complexes reveal that altered ribose-ring conformation and repositioning of the primer 3'-end nucleotide contribute to RTP resistance. These findings enhance our understanding of non-obligate NTP analogs and provide insight into S759A resistance mechanism.

biochemistry↗

bioRxiv2025

Linking Kinetochore Attachment to Checkpoint Control: The Role of Aurora B in BubR1 Acetylation

We report here that Aurora B kinase is critical in BubR1 acetylation at K250, transducing kinetochore attachment status to spindle assembly checkpoint (SAC) activity. Aurora B phosphorylates BubR1 at serines 16 and 39 in an unattachment-dependent manner, which is a prerequisite for BubR1 acetylation at lysine 250 (K250). Utilizing a novel anti-AcK250 monoclonal antibody in structured illumination microscopy (SIM), we reveal that BubR1 acetylation is associated with kinetochore expansion, evidenced by crescent-shaped volume increase of ZW10 and Mad2 surrounding the kinetochore. Furthermore, K250-BubR1 acetylation enhances its interaction with CENP-E, a kinesin crucial for chromosome alignment through facilitating end-on attachment. Disrupting Aurora B-mediated phosphorylation of Ser16/39 impaired K250 acetylation, compromising both fibrous corona polymerization and mitotic checkpoint complex (MCC) maintenance. Conversely, introducing K250 acetylation mimetic form rescued MCC and fibrous corona in cells expressing phosphorylation-deficient S16A/S39A mutants. Our findings reveal Aurora B-BubR1 acetylation axis as a critical SAC signaling pathway, interlinking unattachment sensing with kinetochore expansion and SAC activity. We propose that this stepwise phosphorylation-acetylation on BubR1 constitutes a crucial part of SAC signaling code, transducing unattachment signal to fibrous corona polymerization and SAC activity. This phosphorylation-acetylation cascade in BubR1 offers potential therapeutic targets in treating cancers of chromosome instability.

cell biology↗

bioRxiv2025

m6ASeqTools: An R toolkit for post-processing m6A sites detected by m6Anet

m6ASeqTools is an R package designed to streamline the post-processing and interpretation of site-level m6A predictions from m6Anet. It provides descriptive summaries of m6A distribution across transcripts, genes, biotypes, and transcript regions, and enables condition comparisons through a calculated weighted modification ratio. By integrating differential gene expression data, the package links methylation changes with expression differences, providing biotype-specific and region-specific insights into how m6A localization patterns relate to transcriptional regulation. Availability and implementationm6ASeqTools is freely available at https://github.com/hannalee809/m6ASeqTools Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗

bioRxiv2025

Suppression of interferon signaling via small molecule modulation of TFAM

The mitochondrial transcription factor A (TFAM) is essential for mitochondrial genome maintenance. It binds to mitochondrial DNA (mtDNA) and determines the abundance, packaging and stability of the mitochondrial genome. Because its function is tightly associated with mtDNA, TFAM has a protective role in mitochondrial diseases and supportive studies demonstrate reversal of disease phenotypes by TFAM overexpression. In addition, TFAM deficiency has been shown to cause release of mtDNA into the cytosol and activation of the cGAS/STING innate immune response pathway. As such, TFAM presents as a unique target for therapeutic intervention, but limited efforts for activators have been reported. Herein we disclose novel TFAM small molecule modulators with sub-micromolar activity. Our results demonstrate that these compounds result in increase of TFAM protein levels and mtDNA copy number. This results in inhibition of a mtDNA stress-mediated inflammatory response by preventing mtDNA escape into the cytosol. Furthermore, we see beneficial effects in cellular disease models in which boosting TFAM activity has been advanced as a disease modifying strategy including improved energetics in MELAS cybrid cells and a decrease of fibrotic markers in Systemic Sclerosis (SSc) fibroblasts. These results highlight the therapeutic potential of using small molecule TFAM activators in indications characterized by mitochondrial dysfunction.

cell biology↗

bioRxiv2025

Arm position estimates derived from motor biases

Proprioception, including position sense, is critically important for normal sensorimotor and perceptual functioning but remains poorly understood. In the laboratory and clinic, arm position sense is typically assessed using perceptual tasks, e.g., via arm position matching or reporting the arms current position against a previously sensed one. Although such assessments provide important information about position sensing, they are incomplete in that they do not directly address the role of position sense in motor planning and control. Here, we used a combination of human psychophysical experiments, forward dynamic simulations, and mathematical optimization to reverse engineer arm positions ( REAP) used during motor planning. Subjects performed arm movements in a virtual environment under conditions where visual arm position cues were either aligned with corresponding somatosensory cues or were shifted prior to reach onset in one of four directions. Under shifted conditions, subjects exhibited characteristic motor biases, i.e., systematic deviations from ideal trajectories to the visually-cued targets. The arm positions used to plan movements under shifted conditions were obtained by minimizing differences between the experimentally induced motor biases and simulated motor biases. These REAP estimates largely conformed to predictions derived from previous perceptual experiments in that they were strongly influenced by the visual cues, and in a manner that was axially dependent. The results suggest that assessments of position sense derived from motor biases can be used to augment perceptual assessments or be used in lieu of them when perceptual reporting isnt possible. In addition, the observed similarities between position estimates and weights derived from motor and perceptual tasks suggest that the brains perceptual and action systems use similar mechanisms to deduce arm position from somatosensory and visual cues.

neuroscience↗

bioRxiv2025

Computational design of a single-domain antibody that specifically recognizes WT1 peptide-loaded class I MHC

T cell receptor (TCR)-like antibodies that recognize peptide-loaded class I MHC (pMHC) complexes can enable precise targeting of cancer cells, but developing single-domain binders with high specificity and affinity is challenging. Here, we report the computational design and experimental validation of a TCR-like single-domain antibody (sdAb) that specifically recognizes the WT1-derived peptide RMFPNAPYL presented by HLA-A*02:01. Starting from the crystal structure of a Fab antibody bound to RMF/HLA- A02:01, we repurposed the VH domain into a stable, soluble Trastuzumab-derived VH scaffold. The resulting sdAb, RMFsdAb, spans all nine peptide residues and shows no binding to a control pMHC with a different peptide. Its biophysical properties were improved by fusion to human serum albumin domain III (HSA D3), yielding a monodisperse HSA D3-RMFsdAb with 81 nM affinity and specificity for RMF/HLA- A*02:01. We further engineered a bivalent format (RMFsdAb-HSA D3-RMFsdAb), which dramatically increased apparent binding affinity to 0.4 nM. When expressed on a CAR T cell, HSA D3-RMFsdAb functioned as the antigen-recognition domain to selectively activate T cells in the presence of RMF/HLA-A*02:01-positive cells. Our results demonstrate a viable strategy to develop high-specificity, peptide-focused TCR-like sdAbs for pMHC-targeted therapeutics.

biochemistry↗

bioRxiv2025

Prediction of bacterial protein-compound interactions with only positive samples

Prediction of Compound-Protein Interactions (CPI) in bacteria is crucial to advance various pharmaceutical and chemical engineering fields, including bio-catalysis, drug discovery, and industrial processing. However, current CPI models cannot be applied for bacterial CPI prediction due to the lack of curated negative interaction samples. This paper introduces a novel Positive-Unlabeled (PU) learning framework, named BIN-PU, to address this limitation. BIN-PU generates pseudo positive and negative labels from known positive interaction data, enabling effective training of deep learning models for CPI prediction. We also propose a weighted positive loss function that weights to truly positive samples. We have validated BIN-PU with multiple CPI backbone models, comparing the performance with the existing PU model using bacterial cytochrome P450 (CYP) data. Extensive experiments demonstrate the superiority of BIN-PU over the benchmark model in predicting CPIs with only truly positive samples. Furthermore, we have validated BIN-PU on additional bacterial proteins obtained from literature review, human CYP datasets, and uncurated data for its reproducibility. We have also validated the CPI prediction for the uncurated CYP data with biological and biophysical experiments. BIN-PU represents a significant advancement in CPI prediction for bacterial proteins, opening new possibilities for improving predictive models in related biological interaction tasks.

bioinformatics↗

bioRxiv2025

Plastic-hydrolytic enzyme classification using explainable deep learning

The rapid accumulation of plastic waste has emerged as a critical environmental threat, driving the need for scalable and effective biodegradation solutions. Hydrolytic plastic-degrading enzymes (PDEs) offer a promising solution, yet their functional classification remains limited by insufficient annotations and enzymatic diversity. In this study, we present an explainable deep learning framework, PEPIC, to classify nine types of PDEs directly from protein sequences. Using a curated dataset of experimentally validated enzymes and an expanded homologous dataset, we built an explainable deep learning model based on convolutional neural networks (PEPIC) for plastic-degrading enzyme prediction. We benchmarked PEPICs performance against state-of-the-art approaches. First, PEPIC demonstrated statistically significant improvements in predictive performance compared to state-of-the-art methods. Second, PEPIC calculates contribution scores for each amino acid in the protein sequence, indicating their influence on the predictions. The model interpretation revealed that regions highlighted by high contribution scores matched conserved catalytic triads and substrate-binding clefts across PET-, PCL-, and PLA-degrading enzymes. Furthermore, structural modeling confirmed the trustworthiness of PEPICs predictions. Finally, PEPIC predicted an uncurated enzyme as a PET-degrading enzyme, which was biologically validated to hydrolyze bis(2-hydroxyethyl) terephthalate (BHET). These findings demonstrate that PEPIC provides accurate and trustworthy predictions of PDEs, facilitating the discovery of novel enzymes and supporting the development of sustainable plastic biodegradation technologies.

bioinformatics↗

bioRxiv2025

Modelling complex traits with ancestral recombination graphs

Ancestral recombination graphs (ARGs) are an attractive means for quantitative genetic analysis of complex traits because they encode the realized genetic relatedness between a sample of individuals in the presence of genetic drift, recombination, and mutation. Data structures for efficiently storing ARGs can also be used to rapidly process millions of genomes, and are thus promising for fitting linear mixed models (LMMs) to large phenotype and genome datasets. Here, we study the problems of variance component estimation and prediction of genetic values with ARGs, by describing a generative model of complex traits with additive effects on an ARG, and then developing algorithms that use the ARG to solve these problems efficiently on biobank-scale datasets. We observe nearly linear scaling of runtime with sample size, which is achieved by using the succinct tree sequence representation of the ARG for implicit matrix-vector products, along with modern randomized linear algebra algorithms. We estimate variance components using restricted maximum likelihood (REML), which we find performs substantially better than the Haseman- Elston method. In simulation tests, both variance component estimation and prediction of genetic values (using the best linear unbiased predictor, BLUP) perform nearly as well with inferred ARGs as with true ARGs. We also discuss interpretations of the variance component estimates as mutational variance and additive genetic variance. We provide an implementation of the algorithms as a Python package tslmm, which leverages the tree sequence library tskit.

genomics↗

bioRxiv2025

Distinct evolutionary patterns of endemic and emerging parvoviruses, and the origin of a new pandemic virus.

Emergence of epidemic viruses in new hosts threatens both human and animal populations, and often involves virus evolution to overcome barriers that normally prevent efficient infection and spread in that host. After transfer the separated viruses will evolve in parallel as they spread within the original and new hosts. Here we examine the details of a virus involved in such a host-jumping event, where we define the natural evolution of feline panleukopenia virus (FPV) over 60 years, clarify the origins of the new pandemic canine parvovirus (CPV) that arose in the 1970s, and compare the separate evolution of those viruses over 47 years in cats or dogs. Several live-attenuated FPV vaccine viruses originated from early 1960s isolates or were a recombinant of an early virus, and many sequences in databases proved to be vaccine-derived. The sequences of wild viruses showed that FPV-like strains evolved at [~]25% the rate seen for CPV in dogs, and the higher rate of CPV evolution was consistent since 1979 when a genetic variant became widespread. The common ancestor of the CPV lineage was related to FPVs from Europe, and contained several unique host-adaptive capsid changes associated with canine transferrin receptor type-1 binding. Although the FPV vaccine strains are around 60 years old, little selection for antigenic variation was observed. The distinct evolutionary patterns of these closely related viruses circulating for decades in different hosts emphasizes the complex evolution associated with viral epidemic emergence and spread in endemic and new hosts. SIGNIFICANCE STATEMENTComparing the evolution of a virus in its reservoir host with that seen in a new host will reveal the special circumstances that allow epidemic emergence. A feline parvovirus (FPV) jumping to dogs in the mid-1970s formed canine parvovirus (CPV), which has circulated world-wide until today. The evolutionary rate of FPV in its original hosts was much lower than that of CPV in dogs, and the mutational patterns seen in the different hosts were also distinct. Early CPV isolates differed from the ancestral FPV clade in several key host range mutations. These results highlight the complex biology associated with epidemic emergence, including host-specific rates of lineage evolution and complex origins of host-adaptive mutations. (113/120)

evolutionary biology↗

bioRxiv2025

GABA neurons in the sublaterodorsal tegmental (SLD) nucleus suppress wakefulness in wild-type and narcoleptic mice

The sleep-wake cycle is generated by competing neural circuits that control the oscillation between wakefulness, rapid eye movement (REM) sleep, and non-REM (NREM) sleep. While the sublaterodorsal tegmental nucleus (SLD) is recognized for its role in REM sleep generation, the functional contribution of its GABAergic neurons (SLDGABA) to sleep-wake regulation remains poorly understood. Here, we found that SLDGABA neurons function as a suppressor of wakefulness in both wild-type and narcoleptic mice. Optogenetic silencing of SLDGABA neurons rapidly induced robust wakefulness, while enhancing cortical and motor activity. Conversely, optogenetic activation of these neurons suppressed wakefulness and promoted NREM sleep. We found that SLDGABA neurons project extensively to wake-promoting brain regions, providing an anatomical basis for their wake-suppressing effects. Importantly, we discovered that SLDGABA neurons play a pathological role in narcolepsy: their activation in orexin-deficient mice triggered characteristic sleep attacks--rapid intrusions of NREM sleep during active wakefulness-- while silencing these neurons rescued animals from both sleep attacks and cataplexy. Collectively, these findings establish SLDGABA neurons as a key regulator of arousal state transitions and identify them as a novel therapeutic target for the treatment of narcolepsy.

neuroscience↗

bioRxiv2025

TMC1 and TMC2 are cholesterol-dependent scramblases that regulate membrane homeostasis in auditory hair cells

TMC1 and TMC2, the pore-forming subunits of the mechanoelectrical transduction (MET) complex in inner ear sensory hair cells, are essential for auditory and vestibular function. Pathogenic mutations in TMC1 are a leading cause of genetic hearing loss, but their underlying cellular mechanisms remain poorly understood. Here, we reveal that TMC1 and TMC2 are cholesterol-regulated lipid scramblases whose activity modulates plasma membrane asymmetry. Using reconstituted proteoliposomes and molecular dynamics simulations, we demonstrate that both proteins facilitate phospholipid translocation across membrane bilayers, a process tuned by cholesterol and enhanced by deafness-causing TMC1 mutations. We show that this scramblase activity correlates with TMC1-dependent externalization of phosphatidylserine and membrane blebbing in murine auditory hair cells, linking TMC1-dependent membrane homeostasis dysregulation to auditory sensory cell pathology. These findings identify TMCs as a novel family of lipid scramblases, advancing our understanding of MET complex biology and offering mechanistic insight into membrane-driven forms of hereditary deafness.

neuroscience↗

bioRxiv2025

Comparative Systematic Analysis of Gray Matter BiophysicalModels on a Public Dataset

Biophysical models of diffusion tailored to characterize gray matter (GM) microstructure are gaining traction in the neuroimaging community. NEXI, SMEX, SANDI, and SANDIX represent recent efforts to incorporate different microstructural features,such as soma contributions and inter-compartment exchange, into the diffusion MRI (dMRI) signal. In this work, we present a comparative evaluation of these four gray matter models on a single, publicly available in vivo human dataset, the Connectome Diffusion Microstructure Dataset (CDMD), acquired with two diffusion times. Using the open-source Gray Matter Swiss Knife toolbox, we estimate cortical microstructure metrics in 26 healthy subjects and evaluate goodness of fit, anatomical patterns and consistency with previous studies. CDMD data yielded GM parameter estimates consistent with values reported in previous studies. This retrospective cross-model analysis establishes the feasibility of estimating exchange models from only two diffusion times and highlights trade-offs in biological specificity, model complexity, and fitting robustness, critical considerations when choosing a model for future clinical and research applications.

neuroscience↗