Biology subjects

Tong, F.

Publications and source records attributed to Tong, F..

At least 19 recordsLinked to original sources

bioRxiv2026

AI-enabled Spatial Profiling of Circulating Tumor-Immune Ecosystems Predicts Patient Outcomes Across Cancers

Circulating tumor cells (CTCs) and immune cells form dynamic multicellular ecosystems in blood, but their spatial organization and clinical relevance have not been systematically characterized. We developed the Cell and Cluster Identification Program (CCIP), an artificial intelligence-based framework that analyzes routine multiplex immunofluorescence blood scans to segment cells, identify CTCs and five immune lineages with high accuracy, and quantify multicellular clusters and tumor-immune interactions. Applying CCIP to 2,693 blood scans from 1,399 patients, we profiled over 60 million cells (>7 million multi-cell clusters) and linked imaging-derived features to patient outcomes. Correlated with circulating-tumor DNA mutation burdens, a 14-feature image model predicted overall survival in breast cancer, outperformed clinicopathologic variables and CTC enumeration, and generalized to prostate cancer. Prognostic imaging signatures were also associated with therapy response-related progression-free survival as well as with single-cell RNA sequencing-derived immune suppression states, connecting circulating tumor-immune architecture with systemic immune dysfunction.

cancer biology↗

bioRxiv2026

Exposure to naturalistic occlusion promotes generalized, human-like robustness in deep neural networks

Human object recognition is robust to challenging conditions, such as when ones view of an object is fragmented due to an occluding foreground object. In comparison, deep neural networks (DNNs) are typically more susceptible to occlusion, suggesting that human vision relies on distinct mechanisms. Here, we investigated the role of visual diet in the emergence of these mechanisms by asking whether human-like robustness might arise in DNNs when trained with image datasets that better reflect the properties of occlusion in natural vision. We trained convolutional and transformer DNNs to classify clear images only, images augmented with artificial occluders (i.e., geometric shapes) or natural occluders (objects segmented from photographs). We then evaluated DNN occlusion robustness and compared their performance profiles with 30 human participants. We found that DNNs trained with artificial occluders remained vulnerable to natural occlusion and exhibited less human-like performance than those trained with natural occlusion. Our findings suggest that human robustness to visual occlusion arises from learning to disentangle natural objects from each other rather than simply learning to recognize objects from partial views. They also imply that commonly used forms of artificial occlusion are unsuitable for the evaluation or promotion of robustness to real-world occlusion in DNNs.

neuroscience↗

bioRxiv2026

A live-cell autophagy reporter reveals reversible vacuolation in naked mole-rat skin fibroblasts under lysosomal stress

Naked mole-rats (NMRs, Heterocephalus glaber) display unusual longevity and resistance to age-related decline, and accumulating evidence suggests that their autophagy-lysosome pathway (ALP) is regulated differently from that of conventional mammalian models. However, most studies in NMR cells have relied on static biochemical or ultrastructural readouts, leaving the dynamic organisation of autophagy in living cells poorly defined. Here, we establish a stable tandem fluorescent autophagy reporter in NMR skin fibroblasts using an mCherry-EGFP-LC3NMR construct to enable live-cell, single-cell resolution analysis of ALP dynamics. Under basal conditions, NMR skin fibroblasts exhibit a greater abundance of LC3-positive structures than HeLa cells, together with a mixed population of autophagosomes and autolysosomes, indicating a distinct steady-state organisation of the ALP. Chloroquine (CQ)-induced lysosomal stress caused the expected accumulation of LC3-positive structures but also triggered the formation of large cytoplasmic vacuoles in NMR skin fibroblasts. Importantly, this vacuolation was not associated with acute cytotoxicity and progressively resolved following CQ removal, accompanied by reorganisation of LC3-positive compartments and recovery of lysosomal acidity. Electron microscopy showed that CQ-induced vacuoles are membrane-bound, containing internal material and co-existing with multiple ALP-related vesicular compartments. Primary NMR skin fibroblasts display a similar vacuolation phenotype, indicating that this response is not an artefact of immortalisation or reporter expression. Together, these findings establish a live-cell platform for analysing autophagy in NMR cells and identify a distinctive, reversible vacuolation response to lysosomal stress, consistent with dynamic remodelling of the lysosomal system within NMR skin fibroblasts.

cell biology↗

bioRxiv2025

Photoproximity labeling of c-Myc reveals SLK as a cancer specific co-regulator

Transcription factors (TFs) have long been aspirational therapeutic targets for the treatment of diseases, as their dysregulation is a common mechanism for altered cell states. Despite this, many TFs implicated in disease have disordered structures and lack canonical binding pockets, rendering them non-trivial targets for small molecule-based therapies. Directly inhibiting TF function has proven difficult, but indirect inhibition by targeting the effector molecules that modulate TF function is a promising, yet underexplored, alternative approach. Here we report a strategy for capturing cancer-specific protein-protein interactions using context-dependent {micro}Map photoproximity labeling. Using an intein-based method for catalyst conjugation in biochemically intact nuclei, we demonstrate that we can capture unique protein interactomes of c-Myc in healthy and cancerous prostate cell lines, and that these unique interactors can be mined to identify druggable vulnerabilities. We find that a cancer specific Myc interactor, SLK, selectively promotes c-Myc stabilization at the protein level, drives epithelial morphology, and is essential for tumorigenesis, validating it as a viable therapeutic target. Importantly, this stabilization is driven by a change in SLK splicing rather than expression at the protein or RNA levels. Furthermore, analysis of cancer patient data shows a strong correlation between this splice isoform and expression of c-Myc targets, suggesting this novel regulatory axis is operative across human cancer.

cancer biology↗

bioRxiv2025

Rare Subset of T Cells Form Heterotypic Clusters with Circulating Tumor Cells to Foster Cancer Metastasis

The immune ecosystem is central to maintaining effective defensive responses. However, how immune cells in the periphery blood interact with circulating tumor cells (CTCs) - seeds of metastasis - remains largely understudied. Here, our analysis of the blood specimens (N=1,529) from patients with advanced breast cancer revealed that over 75% of the CTC-positive blood specimens contained heterotypic CTC clusters with CD45+ white blood cells (WBCs). Detection of CTC-WBC clusters correlates with breast cancer subtypes (triple negative and luminal B), racial groups (Black), and decreased survival rates. Flow cytometry and ImageStream analyses revealed diverse WBC composition of heterotypic CTC-WBC clusters, including overrepresented T cells and underrepresented neutrophils. Most strikingly, a rare subset of CD4 and CD8 double positive T (DPT) cells showed an up to 140-fold enrichment in the CTC clusters versus its frequency in WBCs. DPT cells shared part of the profiles with CD4+ T cells and others with CD8+ T cells but exhibited unique features of T cell exhaustion and immune suppression with higher expression of TIM-3 and PD-1. Single-cell RNA sequencing and genetic perturbation studies further pinpointed the integrin VLA4 (4{beta}1) in DPT cells and its ligand VCAM1 in tumor cells as essential mediators of heterotypic WBC-CTC clusters. Neoadjuvant administration of anti-4 (VLA4) neutralizing antibodies markedly blocked CTC-DPT cell clustering and inhibited metastasis for extended survival in preclinical mouse models in vivo. These findings uncover a pivotal role of rare DPT cells with immune suppressive features in fostering cancer dissemination through direct interactive clustering with CTCs. It lays a foundation for developing innovative biomarkers and therapeutic strategies to prevent and target cancer metastasis, ultimately benefiting cancer care. Brief summaryOur findings uncover a fostering role of immune-suppressive T cells in contact with circulating tumor cells and identify therapeutic approaches to eliminate devastating cancer metastasis. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=191 SRC="FIGDIR/small/646421v1_ufig1.gif" ALT="Figure 1"> View larger version (67K): org.highwire.dtl.DTLVardef@13b7862org.highwire.dtl.DTLVardef@15c97b2org.highwire.dtl.DTLVardef@1eb4fd2org.highwire.dtl.DTLVardef@1c588e1_HPS_FORMAT_FIGEXP M_FIG C_FIG

cancer biology↗

bioRxiv2025

Mesoscale proximity labeling to study macro changes to chromatin occupancy

Proximity labeling traditionally identifies interactomes of a single protein or RNA, though this approach limits mechanistic understanding of biomolecules functioning within complex systems. Here, we demonstrate a strategy for deciphering ligand-induced changes to global biomolecular interactions by enabling proximity labelling at the mesoscale, across an entire cellular system. By inserting nanoscale proximity labelling catalysts throughout chromatin, this system, MesoMap, provided new insights into how HDAC inhibitors regulate gene expression. Furthermore, it revealed that the orphaned drug candidate, SR-1815, regulates disease-linked Syngap1 gene expression through direct inhibition of kinases implicated in both neurological disorders and cancer. Through precise mapping of global chromatin mobility, MesoMap promotes insights into how drug-like chemical probes induce transcriptional dynamics within healthy and disease-associated cellular states.

molecular biology↗

bioRxiv2024

Curcumin attenuates microglia-mediated chronic neuropathic pain through CDK5 /p35 signaling pathway

Curcumin is a phenolic compound derived from turmeric, one of the main ingredients of curry powder, which is widely used for its antioxidant, anti-inflammatory and immunomodulatory effects. Curcumin has been reported to help relieve pain, such as neuropathic pain caused by injury or disease, but the specific mechanism of its antinociceptive effect on pathological pain is unclear. Cyclin-dependent kinase 5 (Cdk5) is a key control point for the release of neurotransmitters from presynaptic vesicles. Cdk5 and its activator p35 couple to regulate key signaling cascades, thereby participating in the pain process. In this study, we established a NP model by chronic constriction injury (CCI) of the bilateral sciatic nerve in rats and evaluated behavioral hyperalgesia using mechanical and hot and cold tests. Protein expression and distribution were evaluated using western blotting and immunofluorescence. The results showed that Iba-1 and Cdk5/p35 were co-localized in the dorsal horn and dorsal root ganglia, respectively. After CCI, the expression of Cdk5 and p35 was upregulated in the dorsal horn and dorsal root ganglia, while intraperitoneal injection of curcumin significantly reversed the activation of Cdk5/p35 protein and alleviated the hyperalgesia in rats. In addition, the injection of curcumin reduced the co-localization expression of Iba-1 and Cdk5/p35, indicating that curcumin inhibited the activation of Cdk5/p35 protein in the dorsal horn and dorsal root ganglia, thereby affecting the activation of microglia, thereby having a destructive effect on the neuronal cell plasticity and synaptic structure remodeling in the development of NP. Our study provides new evidence that Cdk5/p35 in the dorsal horn and dorsal root ganglia is related to the occurrence of NP, introduces microglia as the basis for the long-term maintenance of NP, and provides insights into the molecular mechanisms involved in the analgesic effect of curcumin. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=188 SRC="FIGDIR/small/623498v1_ufig1.gif" ALT="Figure 1"> View larger version (80K): org.highwire.dtl.DTLVardef@125fc20org.highwire.dtl.DTLVardef@cba7d3org.highwire.dtl.DTLVardef@772291org.highwire.dtl.DTLVardef@17e87e3_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2024

Engineering customized viral receptors for various coronaviruses

Coronaviruses display versatile receptor usage, yet in-depth characterization of coronaviruses lacking known receptor identities has been impeded by the absence of feasible infection models1,2. Here, we developed an innovative strategy to engineer functional customized viral receptors (CVRs). The modular design relies on building receptor frameworks comprising various function modules and generating specific epitope-targeting viral binding domains. We showed the key factors for CVRs to efficiently facilitate spike cleavage, membrane fusion, pseudovirus entry, and authentic virus propagation for various coronaviruses, resembling their native receptors. Applying this strategy, we delineated the accessible receptor binding epitopes for functional SARS-CoV-2 CVR design and elucidated the mechanism of entry supported by an amino-terminus domain (NTD) targeting S2L20-CVR. Furthermore, we created CVR-expressing cells for assessing antibodies and inhibitors against 12 representative coronaviruses from six subgenera, most of which lacking known receptors. Notably, a pan-sarbecovirus CVR supported entry of various sarbecoviruses, as well as propagation of a replicable HKU3 pseudovirus and the authentic strain RsHuB2019A3. Through combining an HKU5-specific CVR with reverse genetics, we successfully rescued and cultured wild-type and fluorescence protein-incorporated HKU5, a receptor-unidentified merbecovirus. Our study demonstrated the great potential of CVR strategy in establishing native receptor-independent infection models, paving the way for studying various viruses that are challenging to culture due to the lack of susceptible cells. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=162 SRC="FIGDIR/small/583237v1_ufig1.gif" ALT="Figure 1"> View larger version (52K): org.highwire.dtl.DTLVardef@7cb9ccorg.highwire.dtl.DTLVardef@8a2855org.highwire.dtl.DTLVardef@5e14f0org.highwire.dtl.DTLVardef@b0b993_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphic abstractC_FLOATNO C_FIG

microbiology↗

bioRxiv2023

Identification of ACE2 as the Entry Receptor for Two Novel European Bat Merbecoviruses

The angiotensin-converting enzyme 2 (ACE2) receptor is shared by various coronaviruses with distinct receptor-binding domain (RBD) architectures, yet our understanding of these convergent acquisition events remains elusive. Here, we report that two European bat MERS-related coronaviruses (MERSr-CoVs) infecting Pipistrellus nathusii (P.nat), MOW15-22 and PnNL2018B, use ACE2 as their receptor, with narrow ortholog specificity. Cryo-electron microscopy structures of the MOW15-22 RBD-ACE2 complex unveil an unexpected and entirely distinct binding mode, mapping 50[A] away from that of any other known ACE2-using coronaviruses. Functional profiling of ACE2 orthologs from 105 mammalian species led to the identification of host tropism determinants, including an ACE2 N432-glycosylation restricting viral recognition, and the design of a soluble P.nat ACE2 mutant with potent viral neutralizing activity. Our findings reveal convergent acquisition of ACE2 usage for merbecoviruses found in European bats, underscoring the extraordinary diversity of ACE2 recognition modes among coronaviruses and the promiscuity of this receptor. Graphic abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=112 SRC="FIGDIR/small/560486v2_ufig1.gif" ALT="Figure 1"> View larger version (45K): org.highwire.dtl.DTLVardef@1967686org.highwire.dtl.DTLVardef@13f8c8borg.highwire.dtl.DTLVardef@1241e38org.highwire.dtl.DTLVardef@3bfd7a_HPS_FORMAT_FIGEXP M_FIG C_FIG

microbiology↗

bioRxiv2023

Convolutional neural network models of neuronal responses in macaque V1 reveal limited non-linear processing

Computational models of the primary visual cortex (V1) have suggested that V1 neurons behave like Gabor filters followed by simple non-linearities. However, recent work employing convolutional neural network (CNN) models has suggested that V1 relies on far more non-linear computations than previously thought. Specifically, unit responses in an intermediate layer of VGG-19 were found to best predict macaque V1 responses to thousands of natural and synthetic images. Here, we evaluated the hypothesis that the poor performance of lower-layer units in VGG-19 might be attributable to their small receptive field size rather than to their lack of complexity per se. We compared VGG-19 with AlexNet, which has much larger receptive fields in its lower layers. Whereas the best-performing layer of VGG-19 occurred after seven non-linear steps, the first convolutional layer of AlexNet best predicted V1 responses. Although VGG-19s predictive accuracy was somewhat better than standard AlexNet, we found that a modified version of AlexNet could match VGG-19s performance after only a few non-linear computations. Control analyses revealed that decreasing the size of the input images caused the best-performing layer of VGG-19 to shift to a lower layer, consistent with the hypothesis that the relationship between image size and receptive field size can strongly affect model performance. We conducted additional analyses using a Gabor pyramid model to test for non-linear contributions of normalization and contrast saturation. Overall, our findings suggest that the feedforward responses of V1 neurons can be well explained by assuming only a few non-linear processing stages.

neuroscience↗

bioRxiv2023

Improved modeling of human vision by incorporating robustness to blur in convolutional neural networks

Whenever a visual scene is cast onto the retina, much of it will appear degraded due to poor resolution in the periphery; moreover, optical defocus can cause blur in central vision. However, the pervasiveness of blurry or degraded input is typically overlooked in the training of convolutional neural networks (CNNs). We hypothesized that the absence of blurry training inputs may cause CNNs to rely excessively on high spatial frequency information for object recognition, thereby causing systematic deviations from biological vision. We evaluated this hypothesis by comparing standard CNNs with CNNs trained on a combination of clear and blurry images. We show that blur-trained CNNs outperform standard CNNs at predicting neural responses to objects across a variety of viewing conditions. Moreover, blur-trained CNNs acquire increased sensitivity to shape information and greater robustness to multiple forms of visual noise, leading to improved correspondence with human perception. Our results provide novel neurocomputational evidence that blurry visual experiences are very important for conferring robustness to biological visual systems.

neuroscience↗

bioRxiv2023

Structure-based discovery of cannabinoid-1 receptor agonists with reduced side effects

Large library docking can reveal unexpected chemotypes that complement the structures of biological targets. Seeking new agonists for the cannabinoid-1 receptor (CB1R), we docked 74 million tangible molecules, prioritizing 46 high ranking ones for de novo synthesis and testing. Nine were active by radioligand competition, a 20% hit-rate. Structure-based optimization of one of the most potent of these (Ki = 0.7 {micro}M) led to 4042, a 1.9 nM ligand and a full CB1R agonist. A cryo-EM structure of the purified enantiomer of 4042 ( 1350) in complex with CB1R-Gi1 confirmed its docked pose. The new agonist was strongly analgesic, with generally a 5-10-fold therapeutic window over sedation and catalepsy and no observable conditioned place preference. These findings suggest that new cannabinoid chemotypes may disentangle characteristic cannabinoid side-effects from their analgesia, supporting the further development of cannabinoids as pain therapeutics.

pharmacology and toxicology↗

bioRxiv2023

Acetylated dietary protein contributes to the acetylome in cells via KARS-mediated intra-translational deposition of acetyl-lysine in nascent protein

N6-acetyl-L-lysine residue is abundant in dietary protein but less is known about its potential influences on the diet-consumers. We herein report that KARS mediates intra- translational deposition of diet-derived N6-acetyl-L-lysine in nascent proteins to contribute the acetylome in cells. Acetylated dietary protein is a direct source of N6-acetyl-L-lysine that can widely and substantially contribute the acetylome in multiple organs of mice. By analyzing the co-crystal structure of Lysyl-tRNA synthetase (KARS) in complex with N6- acetyl-L-lysyl-AMP and pyrophosphate, together with in vitro biochemical assays, we learned that KARS can utilize N6-acetyl-L-lysine to produce N6-acetyl-L-lysyl-AMP and transfers the N6-acetyl-L-lysyl-moiety to lysine cognate tRNA to generate N6-acetyl-L- lysyl-tRNA, which introduces N6-acetyl-L-lysine into growing nascent polypeptide and intra-translationally results in protein acetylation. This undocumented protein modification mechanism is inherently different from post-translational modification (PTM) and termed as intra-translational modification (ITM). ITM can functionally mimic PTM mechanisms to deposit acetylation in histones to decondense chromatin. It can also modify PTM- inaccessible regions that are buried inside and functionally important to proteins. ITM is expected to extend the repertoire of acetylome and improve our understandings in protein modification modes in cells.

biochemistry↗

bioRxiv2022

Determinants restricting ACE2 recognition of MERS-related coronaviruses in bats

Phylogenetically distant coronaviruses have evolved to use ACE2 as their common receptors, including NL63 and many Severe acute respiratory syndrome (SARS) coronavirus-related viruses. We recently reported two Middle East respiratory syndrome coronavirus (MERS-CoV) closely related bat merbecoviruses, NeoCoV and PDF-2180, use Angiotensin-converting enzyme 2 (ACE2) for entry. However, their host range and cross-species transmissibility remain unknown. Here, we characterized their species-specific receptor preference by testing ACE2 orthologs from 49 bats and 53 non-bat mammals. Both viruses exhibited broad receptor recognition spectra and are unable to use ACE2 orthologs from 24 species, mainly Yinpterochiropteran bats. Comparative analyses of bat ACE2 orthologs underscored four crucial host range determinants, all confirmed by subsequent functional assays in human and bat cells. Among them, residue 305, participating in a critical interaction, plays a crucial role in host tropism determination. NeoCoV-T510F, a mutation that enhances human ACE2 recognition, further expanded the potential host range via tighter interaction with an evolutionary conserved hydrophobic pocket. Our results elucidated the molecular basis for the species-specific ACE2 usage of MERS-related viruses across mammals and shed light on their zoonotic risks.

microbiology↗

bioRxiv2022

Close relatives of MERS-CoV in bats use ACE2 as their functional receptors

Middle East Respiratory Syndrome coronavirus (MERS-CoV) and several bat coronaviruses employ Dipeptidyl peptidase-4 (DPP4) as their functional receptors1-4. However, the receptor for NeoCoV, the closest MERS-CoV relative yet discovered in bats, remains enigmatic5. In this study, we unexpectedly found that NeoCoV and its close relative, PDF-2180-CoV, can efficiently use some types of bat Angiotensin-converting enzyme 2 (ACE2) and, less favorably, human ACE2 for entry. The two viruses use their spikes S1 subunit carboxyl-terminal domains (S1-CTD) for high-affinity and species-specific ACE2 binding. Cryo-electron microscopy analysis revealed a novel coronavirus-ACE2 binding interface and a protein-glycan interaction, distinct from other known ACE2-using viruses. We identified a molecular determinant close to the viral binding interface that restricts human ACE2 from supporting NeoCoV infection, especially around residue Asp338. Conversely, NeoCoV efficiently infects human ACE2 expressing cells after a T510F mutation on the receptor-binding motif (RBM). Notably, the infection could not be cross-neutralized by antibodies targeting SARS-CoV-2 or MERS-CoV. Our study demonstrates the first case of ACE2 usage in MERS-related viruses, shedding light on a potential bio-safety threat of the human emergence of an ACE2 using "MERS-CoV-2" with both high fatality and transmission rate.

microbiology↗

bioRxiv2021

Convolutional neural networks trained with a developmental sequence of blurry to clear images reveal core differences between face and object processing

Although convolutional neural networks (CNNs) provide a promising model for understanding human vision, most CNNs lack robustness to challenging viewing conditions such as image blur, whereas human vision is much more reliable. Might robustness to blur be attributable to vision during infancy, given that acuity is initially poor but improves considerably over the first several months of life? Here, we evaluated the potential consequences of such early experiences by training CNN models on face and object recognition tasks while gradually reducing the amount of blur applied to the training images. For CNNs trained on blurry to clear faces, we observed sustained robustness to blur, consistent with a recent report by Vogelsang and colleagues (2018). By contrast, CNNs trained with blurry to clear objects failed to retain robustness to blur. Further analyses revealed that the spatial frequency tuning of the two CNNs was profoundly different. The blurry to clear face-trained network successfully retained a preference for low spatial frequencies, whereas the blurry to clear object-trained CNN exhibited a progressive shift toward higher spatial frequencies. Our findings provide novel computational evidence showing how face recognition, unlike object recognition, allows for more holistic processing. Moreover, our results suggest that blurry vision during infancy is insufficient to account for the robustness of adult vision to blurry objects.

neuroscience↗

bioRxiv2020

Noise-robust recognition of objects by humans and deep neural networks

Deep neural networks (DNNs) for object classification have been argued to provide the most promising model of the visual system, accompanied by claims that they have attained or even surpassed human-level performance. Here, we evaluated whether DNNs provide a viable model of human vision when tested with challenging noisy images of objects, sometimes presented at the very limits of visibility. We show that popular state-of-the-art DNNs perform in a qualitatively different manner than humans - they are unusually susceptible to spatially uncorrelated white noise and less impaired by spatially correlated noise. We implemented a noise-training procedure to determine whether noise-trained DNNs exhibit more robust responses that better match human behavioral and neural performance. We found that noise-trained DNNs provide a better qualitative match to human performance; moreover, they reliably predict human recognition thresholds on an image-by-image basis. Functional neuroimaging revealed that noise-trained DNNs provide a better correspondence to the pattern-specific neural representations found in both early visual areas and high-level object areas. A layer-specific analysis of the DNNs indicated that noise training led to broad-ranging modifications throughout the network, with greater benefits of noise robustness accruing in progressively higher layers. Our findings demonstrate that noise-trained DNNs provide a viable model to account for human behavioral and neural responses to objects in challenging noisy viewing conditions. Further, they suggest that robustness to noise may be acquired through a process of visual learning.

neuroscience↗

bioRxiv2020

Alterations of human lung and gut microbiome in non-small cell lung carcinomas and distant metastasis

BackgroundNon-small cell lung cancer (NSCLC) is the leading cause of cancer-related deaths worldwide. Although dysbiosis of lung and gut microbiota have been associated with NSCLC, their relative contributions are unclear; in addition, their roles in distant metastasis (DM) are still illusive. ResultsWe surveyed the fecal and sputum (as a proxy for lung) microbiota in healthy controls and NSCLC patients of various stages, and found significant perturbations of gut- and sputum-microbiota in patients with NSCLC and DM. Machine-learning models combining both microbiota (mixed models) performed better than either dataset in patient stratification, with the highest area under the curve (AUC) value of 0.842. Sputum-microbiota contributed more than the gut in the mixed models; in addition, sputum-only models performed similarly to the mixed models in most cases. Several microbial-biomarkers were shared by both microbiota, indicating their similar roles at distinct body sites. Microbial-biomarkers of distinct disease stages were mostly shared, suggesting biomarkers for distant metastasis could be acquired early. Furthermore, Pseudomonas aeruginosa, a species previously associated with wound infections, was significantly more abundant in brain metastasis, indicating distinct types of DMs could have different microbial-biomarkers. ConclusionOur results indicate that alterations of sputum-microbiota have stronger relationships with NSCLC and distant metastasis than the gut, and strongly support the feasibility of metagenome-based non-invasive disease diagnosis and risk evaluation.

microbiology↗