Machine learning deciphers the significance of mitochondrial regulators on the diagnosis and subtype classification in non-alcoholic fatty liver disease.

Background: Non-alcoholic fatty liver disease (NAFLD) is a highly prevalent liver disease worldwide and lack of research on the diagnostic utility of mitochondrial regulators in NAFLD. Mitochondrial dysfunction plays a pivotal role in the development and progression of NAFLD, especially oxidative stress and acidity ß-oxidative overload. Thus, we aimed to identify and validate a panel of mitochondrial gene expression biomarkers for detection of NAFLD. Methods: We selected the GSE89632 dataset and identified key mitochondrial regulators by intersecting DEGs, WGCNA modules, and MRGs. Classification of NAFLD subtypes based on these key mitochondrial regulatory factors was performed, and the pattern of immune system infiltration in different NAFLD subtypes were also investigated. RF, LASSO, and SVM-RFE were employed to identify possible diagnostic biomarkers from key mitochondrial regulatory factors and the predictive power was demonstrated through ROC curves. Finally, we validated these potential diagnostic biomarkers in human peripheral blood samples and a high-fat diet-induced NAFLD mouse model. Results: We identified 25 key regulators of mitochondria and two NAFLD subtypes with different immune infiltration patterns. Four potential diagnostic biomarkers (BCL2L11, NAGS, HDHD3, and RMND1) were screened by three machine learning methods thereby establishing the diagnostic model, which showed favorable predictive power and achieved significant clinical benefit at certain threshold probabilities. Then, through internal and external validation, we identified and confirmed that BCL2L11 was significantly downregulated in NAFLD, while the other three were significantly upregulated. Conclusion: The four MRGs, namely BCL2L11, NAGS, HDHD3, and RMND1, are novel potential biomarkers for diagnosing NAFLD. A diagnostic model constructed using the four MRGs may aid early diagnosis of NAFLD in clinics.

Genome-wide identification and transcriptome profiling expression analysis of the U-box E3 ubiquitin ligase gene family related to abiotic stress in maize (Zea mays L.).

BACKGROUND: The U-box gene family encodes E3 ubiquitin ligases involved in plant hormone signaling pathways and abiotic stress responses. However, there has yet to be a comprehensive analysis of the U-box gene family in maize (Zea mays L.) and its responses to abiotic stress. RESULTS: In this study, 85 U-box family proteins were identified in maize and were classified into four subfamilies based on phylogenetic analysis. In addition to the conserved U-box domain, we identified additional functional domains, including Pkinase, ARM, KAP and Tyr domains, by analyzing the conserved motifs and gene structures. Chromosomal localization and collinearity analysis revealed that gene duplications may have contributed to the expansion and evolution of the U-box gene family. GO annotation and KEGG pathway enrichment analysis identified a total of 105 GO terms and 21 KEGG pathways that were notably enriched, including ubiquitin-protein transferase activity, ubiquitin conjugating enzyme activity and ubiquitin-mediated proteolysis pathway. Tissue expression analysis showed that some ZmPUB genes were specifically expressed in certain tissues and that this could be due to their functions. In addition, RNA-seq data for maize seedlings under salt stress revealed 16 stress-inducible plant U-box genes, of which 10 genes were upregulated and 6 genes were downregulated. The qRT-PCR results for genes responding to abiotic stress were consistent with the transcriptome analysis. Among them, ZmPUB13, ZmPUB18, ZmPUB19 and ZmPUB68 were upregulated under all three abiotic stress conditions. Subcellular localization analysis showed that ZmPUB19 and ZmPUB59 were located in the nucleus. CONCLUSIONS: Overall, our study provides a comprehensive analysis of the U-box gene family in maize and its responses to abiotic stress, suggesting that U-box genes play an important role in the stress response and providing insights into the regulatory mechanisms underlying the response to abiotic stress in maize.

Ionic Liquid Bridge Assisting Bifacial Defect Passivation for Efficient All-Inorganic Perovskite Cells with High Open-Circuit Voltage.

Serious open-circuit voltage (Voc) loss originating from nonradiative recombination and mismatch energy level at TiO2/perovskite buried interface dramatically limits the photovoltaic performance of all-inorganic CsPbIxBr3-x (x = 1, 2) perovskite solar cells (PSCs) fabricated through low-temperature methods. Here, an ionic liquid (IL) bridge is constructed by introducing 1-butyl-3-methylimidazolium acetate (BMIMAc) IL to treat the TiO2/perovskite buried interface, bilaterally passivate defects and modulate energy alignment. Therefore, the Voc of all-inorganic CsPbIBr2 PSCs modified by BMIMAc (Target-1) significantly increases by 148 mV (from 1.213 to 1.361 V), resulting in the efficiency increasing to 10.30% from 7.87%. Unsealed Target-1 PSCs show outstanding long-term and thermal stability. During the accelerated degradation process (85 °C, RH: 50∼60%), the Target-1 PSCs achieve a champion PCE of 11.94% with a remarkable Voc of 1.403 V, while the control PSC yields a promising PCE of 10.18% with a Voc of 1.319 V. In particular, the Voc of 1.403 V is the highest Voc reported so far in carbon-electrode-based CsPbIBr2 PSCs. Moreover, this strategy enables the modified all-inorganic CsPbI2Br PSCs to achieve a Voc of 1.295 V and a champion efficiency of 15.20%, which is close to the reported highest PCE of 15.48% for all-inorganic CsPbI2Br PSCs prepared by a low-temperature process. This study provides a simple BMIMAc IL bridge to assist bifacial defect passivation and elevate the photovoltaic performance of all-inorganic CsPbIxBr3-x (x = 1, 2) PSCs.

Identification and Expression Analysis of the Nucleotidyl Transferase Protein (NTP) Family in Soybean (Glycine max) under Various Abiotic Stresses.

Nucleotidyl transferases (NTPs) are common transferases in eukaryotes and play a crucial role in nucleotide modifications at the 3' end of RNA. In plants, NTPs can regulate RNA stability by influencing 3' end modifications, which in turn affect plant growth, development, stress responses, and disease resistance. Although the functions of NTP family members have been extensively studied in Arabidopsis, rice, and maize, there is limited knowledge about NTP genes in soybeans. In this study, we identified 16 members of the NTP family in soybeans, including two subfamilies (G1 and G2) with distinct secondary structures, conserved motifs, and domain distributions at the protein level. Evolutionary analysis of genes in the NTP family across multiple species and gene collinearity analysis revealed a relatively conserved evolutionary pattern. Analysis of the tertiary structure of the proteins showed that NTPs have three conserved aspartic acids that bind together to form a possible active site. Tissue-specific expression analysis indicated that some NTP genes exhibit tissue-specific expression, likely due to their specific functions. Stress expression analysis showed significant differences in the expression levels of NTP genes under high salt, drought, and cold stress. Additionally, RNA-seq analysis of soybean plants subjected to salt and drought stress further confirmed the association of soybean NTP genes with abiotic stress responses. Subcellular localization experiments revealed that GmNTP2 and GmNTP14, which likely have similar functions to HESO1 and URT1, are located in the nucleus. These research findings provide a foundation for further investigations into the functions of NTP family genes in soybeans.

Fine-Grained Accident Detection: Database and Algorithm.

This paper presents a novel fine-grained task for traffic accident analysis. Accident detection in surveillance or dashcam videos is a common task in the field of traffic accident analysis by using videos. However, common accident detection does not analyze the specific particulars of the accident, only identifies the accident's existence or occurrence time in a video. In this paper, we define the novel fine-grained accident detection task which contains fine-grained accident classification, temporal-spatial occurrence region localization, and accident severity estimation. A transformer-based framework combining the RGB and optical flow information of videos is proposed for fine-grained accident detection. Additionally, we introduce a challenging Fine-grained Accident Detection (FAD) database that covers multiple tasks in surveillance videos which places more emphasis on the overall perspective. Experimental results demonstrate that our model could effectively extract the video features for multiple tasks, indicating that current traffic accident analysis has limitations in dealing with the FAD task and that further research is indeed needed.

PEG-GNPs aggravate MCD-induced steatohepatitic injury and liver fibrosis in mice through excessive lipid accumulation-mediated hepatic inflammatory damage.

Rapid development of gold nanoparticles (GNPs) in delivering pharmaceutics and therapeutics approaches still linger the concerns of their toxic effects. Nonalcoholic steatohepatitis (NASH) is characterized by excessive lipid accumulation and overt hepatic inflammatory damage, and is the leading cause of chronic liver disease worldwide. This study aimed to assess the potential hepatic effects of GNPs on NASH phenotype and progression in mice. Mice were fed a MCD diet for 8 weeks to elicit NASH and then intravenously injected with PEG-GNPs at a single dose of 1, 5, and 25 mg/kg-bw. After 24 h and 1 week of administration, the levels of plasma ALT and AST, and the number of lipid droplets, the degree of lobular inflammation and the contents of triglycerides and cholesterols in the livers of the NASH mice significantly increased compared with the untreated NASH mice, indicating that the severity of MCD diet-induced NASH-like symptoms in mice increased after PEG-GNP administration. Moreover, the aggravated hepatic steatosis in a manner involving altered expression of the genes related to hepatic de novo lipogenesis, lipolysis, and fatty acid oxidation was observed after PEG-GNP administration. Additionally, the RNA levels of biomarkers of hepatic pro-inflammatory responses, endoplasmic reticulum stress, apoptosis, and autophagy in MCD-fed mice increased compared with the untreated NASH group. Moreover, PEG-GNP-treated NASH mice displayed an increase in MCD diet-induced hepatic fibrosis, revealed by massive deposition of collagen fiber in the liver and increased expression of fibrogenic genes. Collectively, these results suggest that hepatic GNP deposition after PEG-GNP administration increase the severity of MCD-induced NASH phenotype in mice, which is attributable to, in large part, increased steatohepatitic injury and liver fibrosis in mice.

The effect of economic growth pressure on green technology innovation: Do environmental regulation, government support, and financial development matter?

Green technology improvement is critical in promoting green development and mitigating negative externalities. Exploring the effect of economic growth pressure (EGP) on green technology innovation (GTI) is important for coordinated economic growth and green transformation. Using the data from 285 cities in China during 2006-2018, this study investigates the influence of EGP on GTI by taking the difference between economic growth target and previous year's actual growth rate to represent the EGP. The results indicate that EGP negatively affects GTI. When there is a 1% increase in EGP, green patent applications will fall by 3.2%. Furthermore, the heterogeneity analysis indicates that the negative effect of EGP is especially significant in western China compared with eastern and central regions. In addition, we find various nonlinear moderating effects between EGP and GTI by using panel threshold model. Specifically, EGP and GTI show an inverted U-shaped relationship with EGP increasing. Meanwhile, only when environmental regulation, government support, and financial development cross the thresholds will EGP have a significant role in promoting GTI. This study provides helpful implications for decision-makers to adopt a more reasonable combination of policy tools to achieve economic growth targets and low-carbon transformation.

In-plane charged domain walls with memristive behaviour in a ferroelectric film.

Domain-wall nanoelectronics is considered to be a new paradigm for non-volatile memory and logic technologies in which domain walls, rather than domains, serve as an active element. Especially interesting are charged domain walls in ferroelectric structures, which have subnanometre thicknesses and exhibit non-trivial electronic and transport properties that are useful for various nanoelectronics applications1-3. The ability to deterministically create and manipulate charged domain walls is essential to realize their functional properties in electronic devices. Here we report a strategy for the controllable creation and manipulation of in-plane charged domain walls in BiFeO3 ferroelectric films a few nanometres thick. By using an in situ biasing technique within a scanning transmission electron microscope, an unconventional layer-by-layer switching mechanism is detected in which ferroelectric domain growth occurs in the direction parallel to an applied electric field. Based on atomically resolved electron energy-loss spectroscopy, in situ charge mapping by in-line electron holography and theoretical calculations, we show that oxygen vacancies accumulating at the charged domain walls are responsible for the domain-wall stability and motion. Voltage control of the in-plane domain-wall position within a BiFeO3 film gives rise to multiple non-volatile resistance states, thus demonstrating the key functional property of being a memristor a few unit cells thick. These results promote a better understanding of ferroelectric switching behaviour and provide a new strategy for creating unit-cell-scale devices.

Extracellular vesicle-derived LINC00511 promotes glycolysis and mitochondrial oxidative phosphorylation of pancreatic cancer through macrophage polarization by microRNA-193a-3p-dependent regulation of plasminogen activator urokinase.

OBJECTIVE: The involvement of tumor-derived extracellular vesicles (EVs) in macrophage polarization has been reported. In our present study, we tried to discuss the regulatory role of LINC00511 encapsulated in pancreatic cancer (PCa) cell-derived EVs in the development and progression of PCa. METHODS: EVs from PCa cell line BxPC-3 culture medium were collected and subsequently identified by electron microscopy and nanoparticle tracking analysis. The expression pattern of LINC00511 in PCa cell-derived EVs was determined. The interaction among LINC00511, microRNA-193a-3p, and plasminogen activator urokinase (PLAU) was explored. After co-culture of PCa cell-derived EVs with macrophages, the regulatory roles of LINC00511 in macrophage polarization, PCa cell functions, glucose consumption, lactate production, glycolysis, and mitochondrial oxidative phosphorylation were investigated. RESULTS: PCa cell line BxPC-3 had highly expressed LINC00511 and LINC00511 could be internalized by macrophages. LINC00511 affected macrophage polarization through miR-193a-3p-dependent regulation of PLAU expression. Besides, EV-derived LINC00511 accelerated glycolysis and promoted mitochondrial oxidative phosphorylation of PCa cells through macrophage polarization, thus inducing invasion and migration of PCa cells. CONCLUSION: LINC00511 encapsulated in PCa cell-derived EVs facilitates glycolysis of PCa cells through regulation of macrophage polarization in the tumor microenvironment.

The Impact of Childhood Socioeconomic Status on Adolescents' Risk Behaviors: The Role of Physiological and Psychological Threats.

Adolescence is a period of high levels of risk behavior. The present research aims to examine the influences of childhood socioeconomic status (SES) on risk behaviors in gain or loss domains among adolescents and the roles of threats in this effect. In experiment 1, a total of 107 adolescents (Mage = 14.80; SDage = 1.15) were asked to complete the childhood socioeconomic status scale before they took part in a risk behavior task under the gain and loss situation. A total of 149 adolescents (Mage = 14.24; SDage = 1.11) in experiment 2a and 139 adolescents (Mage = 13.88; SDage = 1.09) in experiment 2b completed the childhood socioeconomic status scale before they took part in a risk behavior task under the gain and loss situation under physiological threats and psychological threats, respectively. The results showed that high-childhood-SES adolescents tend to take more risks than low-childhood-SES adolescents in the gain domain, while low-childhood-SES adolescents tend to take more risks than high-childhood-SES adolescents in the loss domain. Threats amplified the impact of childhood socioeconomic status on adolescents' risk behaviors in the gain and loss domains. When a physiological threat or psychological threat was primed, compared to the control group, in the gain situation, the extent to which high-childhood-SES adolescents showed greater risk seeking than low-childhood-SES adolescents became larger; in the loss domain, the extent to which low-childhood-SES adolescents showed greater risk seeking than high-childhood-SES adolescents became larger.

Powerful molecule generation with simple ConvNet.

MOTIVATION: Automated molecule generation is a crucial step in in-silico drug discovery. Graph-based generation algorithms have seen significant progress over recent years. However, they are often complex to implement, hard to train and can under-perform when generating long-sequence molecules. The development of a simple and powerful alternative can help improve practicality of automated drug discovery method. RESULTS: We proposed a ConvNet-based sequential graph generation algorithm. The molecular graph generation problem is reformulated as a sequence of simple classification tasks. At each step, a convolutional neural network operates on a sub-graph that is generated at previous step, and predicts/classifies an atom/bond adding action to populate the input sub-graph. The proposed model is pretrained by learning to sequentially reconstruct existing molecules. The pretrained model is abbreviated as SEEM (structural encoder for engineering molecules). It is then fine-tuned with reinforcement learning to generate molecules with improved properties. The fine-tuned model is named SEED (structural encoder for engineering drug-like-molecules). The proposed models have demonstrated competitive performance comparing to 16 state-of-the-art baselines on three benchmark datasets. AVAILABILITY AND IMPLEMENTATION: Code is available at https://github.com/yuh8/SEEM and https://github.com/yuh8/SEED. QM9 dataset is availble at http://quantum-machine.org/datasets/, ZINC250k dataset is availble at https://raw.githubusercontent.com/aspuru-guzik-group/chemical_vae/master/models/zinc_properties/250k_rndm_zinc_drugs_clean_3.csv, and ChEMBL dataset is availble at https://www.ebi.ac.uk/chembl/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Tumor-associated macrophage-derived exosomes transmitting miR-193a-5p promote the progression of renal cell carcinoma via TIMP2-dependent vasculogenic mimicry.

Previous studies have investigated whether tumor-associated macrophages (TAMs) play tumorigenic and immunosuppressive roles to encourage cancer development, but the role of TAMs in regulating vasculogenic mimicry (VM) in clear-cell renal cell carcinoma (ccRCC) cells has not been completely clarified. We conducted immunostaining of the tumor-associated macrophage biomarkers CD68/CD163 and double staining for PAS/CD31 in ccRCC human specimens to find that higher TAM infiltration was positively correlated with VM formation. Then we demonstrated that TAM-derived exosomes downregulate TIMP2 expression in RCC cells to promote VM and invasion by shuttling miR-193a-5p. Mechanistic analysis indicated that HIF-1α upregulation in macrophages could transcriptionally increase miR-193a-5p expression. Exosome-shuttled miR-193a-5p then targeted the 3' untranslated region (UTR) of TIMP2 mRNA to suppress its translation. A preclinical study using an in vivo orthotopic xenograft model of ccRCC in mice substantiated that TAM-derived exosomes enhance VM and enable tumor progression, which confirmed our in vitro data. Suppressing TAM-derived exosomal miR-193a-5p successfully inhibited tumor progression and metastasis. Overall, miR-193a-5p from TAM-derived exosomes downregulates the TIMP2 gene to facilitate the development of RCC, which provides a novel perspective for developing therapeutic strategies for RCC.

Identification of RNA helicases in Medicago truncatula and their expression patterns under abiotic stress.

RNA helicase catalyzes the denaturation of DNA or the unwinding of double-stranded RNA. It is vital to RNA splicing, transport, editing, degradation and the initiation of protein translation. However, the function of RNA helicase in Medicago truncatula has rarely been reported. In this study, 170 putative RNA helicase genes were identified in the M. truncatula genome, and classified into three subfamilies based on the presence of either a DEAD-box (52 genes), DEAH-box (38 genes), or DExD/H-box (80 genes) in their coding regions. Additionally, conserved helicase_C domains and other functional domains (e.g., the HA2, DUF, and ZnF domains) were also present in these genes. Chromosomal mapping and synteny analyses showed that there were tandem and segment duplications of RNA helicase genes. Furthermore, transcriptome and real-time PCR analysis showed that the expression of 35 RNA helicase genes was affected by abiotic stress. To be specific, 17, 12 and 19 genes were regulated by salt, drought and cold stress, respectively. It is worth noting that MtDEAD8, MtDEAH3, MtDExD/H18 and MtDExD/H23 responded to all three types of stress. These results provide valuable information for understanding the RNA helicase genes in M. truncatula and their abiotic stress-related functions. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s12298-021-01087-y.

Variations of stress field and stone fracture produced at different lateral locations in a shockwave lithotripter field.

During clinical procedures, the lithotripter shock wave (LSW) that is incident on the stone and resultant stress field is often asymmetric due to the respiratory motion of the patient. The variations of the LSW-stone interaction and associated fracture pattern were investigated by photoelastic imaging, phantom experiments, and three-dimensional fluid-solid interaction modeling at different lateral locations in a lithotripter field. In contrast to a T-shaped fracture pattern often observed in the posterior region of the disk-shaped stone under symmetric loading, the fracture pattern gradually transitioned to a tilted L-shape under asymmetric loading conditions. Moreover, the model simulations revealed the generation of surface acoustic waves (SAWs), i.e., a leaky Rayleigh wave on the anterior boundary and Scholte wave on the posterior boundary of the stone. The propagation of SAWs on the stone boundary is accompanied by a progressive transition of the LSW reflection pattern from regular to von Neumann and to weak von Neumann reflection near the glancing incidence and, concomitantly, the development and growth of a Mach stem, swirling around the stone boundary. The maximum tensile stress and stress integral were produced by SAWs on the stone boundary under asymmetric loading conditions, which drove the initiation and extension of surface cracks into the bulk of the stone that is confirmed by micro-computed tomography analysis.

Multiscale Synchrotron-Based Imaging Analysis for the Transfer of PEGylated Gold Nanoparticles In Vivo.

High spatial resolution imaging analysis is urgently needed to explore the biodistribution, transfer and clearance profiles, and biological impact of nanoparticles in the body, which will be helpful to clarify the efficacy of nanomedicine in clinical applications. Herein, by combination with multiscale synchrotron-based imaging techniques, including X-ray fluorescence (XRF) spectrometry, Fourier transform infrared (FTIR) spectroscopy, and micro X-ray phase contrast computed tomography (micro-XPCT), we visually displayed the transfer patterns and site-specific distribution of PEGylated gold nanoparticles (PEG-GNPs) in the suborgans of the liver, spleen, and kidney after an intravenous injection in mice. A combination of XRF and FTIR imaging analysis showed that the PEG bands presented similar distribution patterns with Au in the intraorgans, suggesting the stability of PEGylation on GNPs. We show that the PEG-GNPs presented heterogeneous distribution in the hepatic lobules with a large amount around the portal vein zone and then a gradient decrease in the sinusoidal region and the CV zone; in the spleen, it gradually accumulated in the splenic red pulp over time; and in the kidney, it quickly transported via the bloodstream to the renal pyramids and renal pelvis, and parts of PEG-GNPs finally accumulated in the renal medulla and renal cortex. Multidimensional micro-XPCT images further show that the PEG-GNP transfer in the liver induced hepatic blood vessel dilatation while they transferred in the liver, providing evidence of GNP transport across the blood vessel endothelial barrier.

Interaction of Humic Acid with Graphene Oxide: Relation to Antibacterial Activities Against Escherichia coli.

Graphene oxide (GO) sheets attracted great attention as effectively antibacterial agents in water treatment and environmental remediation applications. In the study, the interaction of humic acid (HA) as the model of natural organic matter (NOM) with GO and their antibacterial activities against Escherichia coli (E. coli) was investigated. The interaction between GO and HA molecules was analyzed by isothermal titration calorimetry (ITC) and fluorescence spectroscopy analysis. The study demonstrated that GO reaction with HA was a spontaneously exothermic process, which enabled formation of stable and well dispersed GO-HA complex in aqueous solution. Both GO and GO-HA could significantly inhibit the growth of E. coli and present dose-dependent bactericidal property. GO and GO-HA showed more obvious antibacterial activity in saline solution than in LB broth. We suggest the surface wrinkles of GO and GO-HA could contribute to the firm wrapping of E. coli, which is the principle factor for the antibacterial activity of GO and GO-HA. Especially, GO-HA exhibit less surface wrinkles in comparison with GO, corresponding to its reduced antibacterial activity in saline solution.

TCP5 controls leaf margin development by regulating KNOX and BEL-like transcription factors in Arabidopsis.

Development of leaf margins is an important process in leaf morphogenesis. CIN-clade TCP (TEOSINTE BRANCHED1/CYCLOIDEA/PCF) transcription factors are known to have redundant roles in specifying leaf margins, but the specific mechanisms through which individual TCP genes function remain elusive. In this study, we report that the CIN-TCP gene TCP5 is involved in repressing the initiation and outgrowth of leaf serrations by activating two key regulators of margin development, the Class II KNOX factor KNAT3 and BEL-like SAW1. Specifically, TCP5 directly promotes the transcription of KNAT3 and indirectly activates the expression of SAW1. We also show that TCP5 regulates KNAT3 and SAW1 in a temporal- and spatial- specific manner that is largely in accordance with the progress of formation of serrations. This regulation might serve as a key mechanism in patterning margin morphogenesis and in sculpting the final form of the leaf.

Circular RNA circTMEM45A Acts as the Sponge of MicroRNA-665 to Promote Hepatocellular Carcinoma Progression.

Mounting evidences indicate that circular RNAs (circRNAs) play vital roles in the development and progression of various cancers. However, the detailed functions and underlying mechanisms of circRNAs in hepatocellular carcinoma (HCC) remain largely unknown. The expression profile of circRNAs was screened by circRNA microarrays. Quantitative real-time PCR was used to determine the level-10 circRNAs selected from the top five upregulated (hsa_circ_0001955, hsa_circ_0001535, hsa_circ_0061395, hsa_circ_0000502, and hsa_circ_0066659) and top five downregulated circRNAs (hsa_circ_0046366, hsa_circ_0003418, hsa_circ_0026134, hsa_circ_0005692, and hsa_circ_0014130). The effects of circTMEM45A in HCC cells were studied both in vitro (in a Cell Counting Kit-8 assay, apoptosis analysis, and cell cycle assays) and in vivo (by means of tumor xenografts in nude mice). Luciferase reporter, RNA immunoprecipitation (RIP), and rescued assays were used to confirm the interactions between circTMEM45A, miR-665, and insulin growth factor 2 (IGF2). We found that the level of circTMEM45A was significantly upregulated in HCC and was positively correlated with clinicopathological features and poor prognosis of patients with HCC. Functionally, circTMEM45A promoted cell mobility in vitro, as well as in vivo tumorigenesis. Mechanistically, circTMEM45A acted as a miR-65 sponge to relieve the repressive effect of miR-665 on its target IGF2. Moreover, circTMEM45A was upregulated in serum exosomes from HCC patients. circTMEM45A promotes HCC progression through the miR-665/IGF2 axis and may serve as a novel diagnostic marker and target for treatment of HCC patients.