这是用户在 2025-4-8 9:53 为 https://app.immersivetranslate.com/pdf-pro/6eb844b6-0cb6-401c-83fd-736ad92eb1de/ 保存的双语快照页面，由 沉浸式翻译 提供双语支持。了解如何保存？

Identification and validation of cuproptosis-related molecular clusters in non-alcoholic fatty liver disease
非酒精性脂肪肝中杯突症相关分子群的鉴定与验证

Changxu Liu 1 1 ^(1){ }^{1} | Zhihao Fang 1 ( 0 ) 1 ( 0 ) ^(1)^((0)){ }^{1}{ }^{(0)} | Kai Yang 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Yanchao Ji 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Xiaoxiao Yu 1 1 ^(1)^{1} | ZiHao Guo 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Zhichao Dong 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Tong Zhu 1 , 2 ∣ 1 , 2 ∣ ^(1,2)∣{ }^{1,2} \mid Chang Liu 1 1 ^(1){ }^{1}
Changxu Liu 1 1 ^(1){ }^{1} | Zhihao Fang 1 ( 0 ) 1 ( 0 ) ^(1)^((0)){ }^{1}{ }^{(0)} | Kai Yang 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Yanchao Ji 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Xiaoxiao Yu 1 1 ^(1)^{1} | ZiHao Guo 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Zhichichao Dong 1 ∣ 1 ∣ ^(1)∣{ }^{1} \mid Tong Zhu 1 , 2 ∣ 1 , 2 ∣ ^(1,2)∣{ }^{1,2} \mid Chang Liu 1 1 ^(1){ }^{1}

1 1 ^(1){ }^{1} Department of General Surgery, Fourth Affiliated Hospital of Harbin Medical University, Harbin, China
1 1 ^(1){ }^{1} 哈尔滨医科大学附属第四医院普外科，中国哈尔滨
2 2 ^(2){ }^{2} Beijing Chaoyang Hospital Affiliated to Capital Medical University, Beijing, China
2 2 ^(2){ }^{2} 首都医科大学附属北京朝阳医院，中国北京

Correspondence  通信

Chang Liu and Tong Zhu, Fourth Affiliated Hospital of Harbin Medical University, Harbin, Heilongjiang 150001, China.
刘畅、朱彤，哈尔滨医科大学附属第四医院，黑龙江哈尔滨 150001。
Email: Ic19726666@163.com and 772337259@qq.com
电子邮件：Ic19726666@163.com 和 772337259@qq.com

Funding information  资金信息

Innovative scientific Research Fund of Harbin Medical University, Grant/Award Number: 2021-KYYWF-0260
哈尔滨医科大学创新科研基金，资助/奖励编号：2021-KYYWF-0260

Abstract  摘要

Non-alcoholic fatty liver disease (NAFLD) is a major chronic liver disease worldwide. Cuproptosis has recently been reported as a form of cell death that appears to drive the progression of a variety of diseases. This study aimed to explore cuproptosis-related molecular clusters and construct a prediction model. The gene expression profiles were obtained from the Gene Expression Omnibus (GEO) database. The associations between molecular clusters of cuproptosis-related genes and immune cell infiltration were investigated using 50 NAFLD samples. Furthermore, cluster-specific differentially expressed genes were identified by the WGCNA algorithm. External datasets were used to verify and screen feature genes, and nomograms, calibration curves and decision curve analysis (DCA) were performed to verify the performance of the prediction model. Finally, a NAFLD-diet mouse model was constructed to further verify the predictive analysis, thus providing new insights into the prediction of NAFLD clusters and risks. The role of cuproptosis in the development of non-alcoholic fatty liver disease and immune cell infiltration was explored. Non-alcoholic fatty liver disease was divided into two cuproptosis-related molecular clusters by unsupervised clustering. Three characteristic genes (ENO3, SLC16A1 and LEPR) were selected by machine learning and external data set validation. In addition, the accuracy of the nomogram, calibration curve and decision curve analysis in predicting NAFLD clusters was also verified. Further animal and cell experiments confirmed the difference in their expression in the NAFLD mouse model and Mouse hepatocyte cell line. The present study explored the relationship between non-alcoholic fatty liver disease and cuproptosis, providing new ideas and targets for individual treatment of the disease.
非酒精性脂肪肝（NAFLD）是全球主要的慢性肝病。最近有报道称，杯突症是细胞死亡的一种形式，似乎推动了多种疾病的进展。本研究旨在探索杯突相关分子集群并构建预测模型。基因表达谱来自基因表达总库（GEO）数据库。利用 50 个非酒精性脂肪肝样本研究了杯突症相关基因分子集群与免疫细胞浸润之间的关联。此外，还通过 WGCNA 算法确定了特定群组的差异表达基因。利用外部数据集来验证和筛选特征基因，并通过提名图、校准曲线和决策曲线分析（DCA）来验证预测模型的性能。最后，建立了非酒精性脂肪肝饮食小鼠模型，进一步验证了预测分析，从而为预测非酒精性脂肪肝集群和风险提供了新的见解。探讨了杯突症在非酒精性脂肪肝的发展和免疫细胞浸润中的作用。通过无监督聚类将非酒精性脂肪肝分为两个杯突相关分子集群。通过机器学习和外部数据集验证，选出了三个特征基因（ENO3、SLC16A1 和 LEPR）。此外，还验证了提名图、校准曲线和决策曲线分析预测非酒精性脂肪肝集群的准确性。进一步的动物和细胞实验证实了它们在非酒精性脂肪肝小鼠模型和小鼠肝细胞系中的表达差异。 本研究探讨了非酒精性脂肪肝与杯突症之间的关系，为该疾病的个体治疗提供了新思路和新靶点。

K E Y WORDS

cuproptosis, diagnostic, immune, non-alcoholic fatty liver disease
杯状红细胞增多症、诊断、免疫、非酒精性脂肪肝

1 | INTRODUCTION  1 简介

Globally, the number of patients with metabolic disorders has been increasing as obesity rates have steadily risen. Non-alcoholic fatty liver disease (NAFLD) is the hepatic manifestation of metabolic syndrome and is mainly characterized by excessive accumulation of fat in hepatocytes ( ≥ 5 % ≥ 5 % >= 5%\geq 5 \% ) in the absence of excessive alcohol consumption. 1 1 ^(1){ }^{1} Statistics show that around 24 % 24 % 24%24 \% of the world’s population is affected by non-alcoholic fatty liver disease (NAFLD). 2 2 ^(2){ }^{2} The pathological changes of the liver in NAFLD range from simple hepatic steatodegeneration to non-alcoholic steatohepatitis (NASH), and may even develop into liver fibrosis, cirrhosis and liver cancer in severe cases. 3 3 ^(3){ }^{3}
在全球范围内，随着肥胖率的稳步上升，代谢紊乱患者的人数也在不断增加。非酒精性脂肪肝（NAFLD）是代谢综合征的肝脏表现，主要特征是在没有过量饮酒的情况下，肝细胞内脂肪过度堆积（ ≥ 5 % ≥ 5 % >= 5%\geq 5 \% ）。 1 1 ^(1){ }^{1} 统计数据显示，全球约有 24 % 24 % 24%24 \% 的人口受到非酒精性脂肪肝（NAFLD）的影响。 2 2 ^(2){ }^{2} 非酒精性脂肪肝的肝脏病理变化从简单的肝脂肪变性到非酒精性脂肪性肝炎（NASH）不等，严重者甚至会发展为肝纤维化、肝硬化和肝癌。 3 3 ^(3){ }^{3}
However, the pathogenesis of NAFLD/NASH remains unclear. The classical theory of NASH progression is the ‘twohit hypothesis’, but this conventional view is too simplistic to account for all the molecular and metabolic changes in NASH pathogenesis. 4 , 5 4 , 5 ^(4,5){ }^{4,5} More recently, oxidative stress has been reported to play a major role in the progression of hepatic steatosis to NASH. Oxidative stress results from an imbalance between the body’s antioxidant defences and reactive oxygen species (ROS). ROS are highly reactive molecules and excessive ROS can result in oxidative stress and damage cellular components, such as DNA, proteins and lipid bilayers. The antioxidant defence includes enzymatic antioxidants such as superoxide dismutase (SOD), catalase and glutathione peroxidase (GSH-PX) and small molecules such as vitamin C and GSH. 6 6 ^(6){ }^{6} However, a highly oxidized environment can aggravate inflammation, fibrosis formation and hepatocyte death, leading to the pathological progression of NAFLD. 7 7 ^(7){ }^{7} The copper ion is essential for a number of biological processes. As a coenzyme factor, Cu 2 + Cu 2 + Cu^(2+)\mathrm{Cu}^{2+} mainly relies on mitochondrial regulation 8 8 ^(8){ }^{8} to maintain homeostasis. Copper is mainly present in mitochondria in the form of cytochrome c oxidase (COX) and superoxide dismutase (SOD1), which regulate the tricarboxylic acid (TCA) cycle and terminal oxidation. Furthermore, it participates in multiple biological processes such as reduction balance, iron utilization, oxidative phosphorylation and cell proliferation. 9 , 10 9 , 10 ^(9,10){ }^{9,10} Programmed cell death (PCD) refers to a genetically regulated cellular suicide, which plays a crucial role in tissue homeostasis and growth and is also involved in several pathological processes. 11 11 ^(11){ }^{11} Various types of programmed cell death have been found to date, including necroptosis and ferroptosis. 11 11 ^(11){ }^{11}
然而，非酒精性脂肪肝/NASH 的发病机制仍不清楚。NASH 进展的经典理论是 "两击假说"，但这种传统观点过于简单，无法解释 NASH 发病机制中的所有分子和代谢变化。 4 , 5 4 , 5 ^(4,5){ }^{4,5} 最近有报道称，氧化应激在肝脂肪变性发展为 NASH 的过程中起着重要作用。氧化应激源于机体抗氧化防御系统与活性氧（ROS）之间的不平衡。活性氧是高活性分子，过量的活性氧会导致氧化应激，损害细胞成分，如 DNA、蛋白质和脂质双分子层。抗氧化防御包括酶类抗氧化剂，如超氧化物歧化酶（SOD）、过氧化氢酶和谷胱甘肽过氧化物酶（GSH-PX），以及小分子抗氧化剂，如维生素 C 和 GSH。 6 6 ^(6){ }^{6} 然而，高度氧化的环境会加重炎症、纤维化形成和肝细胞死亡，导致非酒精性脂肪肝的病理进展。 7 7 ^(7){ }^{7} 铜离子对许多生物过程至关重要。作为一种辅酶因子， Cu 2 + Cu 2 + Cu^(2+)\mathrm{Cu}^{2+} 主要依靠线粒体调节 8 8 ^(8){ }^{8} 来维持体内平衡。铜主要以细胞色素 c 氧化酶（COX）和超氧化物歧化酶（SOD1）的形式存在于线粒体中，调节三羧酸（TCA）循环和末端氧化。此外，它还参与多种生物过程，如还原平衡、铁利用、氧化磷酸化和细胞增殖。 9 , 10 9 , 10 ^(9,10){ }^{9,10} 程序性细胞死亡（PCD）是指一种受基因调控的细胞自杀行为，它在组织稳态和生长过程中起着至关重要的作用，同时也参与了多种病理过程。 11 11 ^(11){ }^{11} 迄今已发现各种类型的程序性细胞死亡，包括坏死和铁死。 11 11 ^(11){ }^{11}
Cuproptosis is a newly identified form of programmed cell death that is distinct from other oxidative stress-regulated death events such as pyroptosis, ferroptosis and necrosis. Mitochondrial stress has been reported to be the main mechanism leading to cuproptosis, 12 12 ^(12){ }^{12} which is characterized by excessive accumulation of mitochondrial fatty acid acylase and depletion of Fe-S-cluster proteins. Moreover, a growing number of reports indicate a complex relationship between the regulatory imbalance of copper ions in NAFLD and NASH. 13 13 ^(13){ }^{13} However, the potential regulatory
杯突症是一种新发现的细胞程序性死亡形式，它有别于其他氧化应激调控的死亡事件，如热突症、铁突症和坏死。据报道，线粒体应激是导致杯突症的主要机制， 12 12 ^(12){ }^{12} 其特点是线粒体脂肪酸酰化酶过度积累和 Fe-S-簇蛋白耗竭。此外，越来越多的报告表明，非酒精性脂肪肝和非酒精性脂肪性肝炎中铜离子的调节失衡之间存在复杂的关系。 13 13 ^(13){ }^{13} 然而，在非酒精性脂肪肝和非酒精性脂肪性肝病中，铜离子的潜在调控
mechanism of cuproptosis in NAFLD remains unclear and requires further exploration. Therefore, NAFLD heterogeneity may be attributed to the molecular characteristics of cuproptosis-related genes (CRGs).
非酒精性脂肪肝的杯突症机制尚不清楚，需要进一步探讨。因此，非酒精性脂肪肝的异质性可能归因于杯突相关基因（CRGs）的分子特征。
For the first time, this study systematically investigated the differential expression of CRGs and immune infiltration characteristics between normal subjects and patients with NAFLD. Based on nine differentially expressed CRG profiles, NAFLD patients were divided into two cuproptosis-related clusters and the immune infiltration differences between the two clusters were investigated. Subsequently, cluster-specific differentially expressed gene (DEGs) were identified and the enriched biological functions and pathways were elucidated based on cluster-specific DEGs. In addition, multiple machine learning algorithms and predictive models were constructed to identify patients with different molecular clusters. The performance of the prediction model was verified using nomograms, calibration curves, decision curve analysis (DCA) and external data sets. Finally, a mouse model of NAFLD was established by a high-fat diet (HFD) to further validate the predictive analysis, providing new insights into the prediction of NAFLD clusters and risk.
本研究首次系统研究了正常人和非酒精性脂肪肝患者之间 CRG 的差异表达和免疫浸润特征。根据 9 个差异表达的 CRG 图谱，非酒精性脂肪肝患者被分为两个杯突相关群组，并研究了两个群组之间的免疫浸润差异。随后，研究人员确定了集群特异性差异表达基因（DEG），并根据集群特异性差异表达基因阐明了富集的生物学功能和通路。此外，还构建了多种机器学习算法和预测模型，以识别不同分子集群的患者。利用提名图、校准曲线、决策曲线分析（DCA）和外部数据集验证了预测模型的性能。最后，通过高脂饮食（HFD）建立了非酒精性脂肪肝小鼠模型，进一步验证了预测分析，为预测非酒精性脂肪肝集群和风险提供了新的见解。

2 | MATERIALS  2 材料

2.1 | Data collection and preprocessing
2.1 数据收集和预处理

Gene expression profiles were obtained from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo) and the NAFLD datasets GSE89632, GSE63067 and GSE48452 were downloaded from the GEO database. All the datasets originated from Homo sapiens. The GSE48452 data set used the GPL11532 data platform, with a total of 73 samples, including 14 normal liver tissue control samples, 27 obese liver tissue samples and 32 NAFLD tissue samples. The GSE89632 dataset used GPL14951 and contained 63 samples, including 24 normal liver control samples and 39 NAFLD tissue samples. All samples were included in the present. In addition, the GSE63067 dataset used GPL14877 and contained 18 samples, including 7 normal liver control samples and 11 NAFLD tissue samples. In this study, dataset GSE48452 was used as the validation set. The GSE63067 and GSE89632 data sets were merged and the batch effect was removed by the SVA package. 14 14 ^(14){ }^{14} Finally, the R package ‘Limma’ was used for analysis. 15 15 ^(15){ }^{15} The flow chart of this study is displayed in Figure S1.
基因表达谱来自基因表达总库（GEO）数据库（http://www.ncbi.nlm.nih.gov/geo），非酒精性脂肪肝数据集 GSE89632、GSE63067 和 GSE48452 是从 GEO 数据库下载的。所有数据集均来自智人。GSE48452 数据集使用 GPL11532 数据平台，共有 73 个样本，包括 14 个正常肝组织对照样本、27 个肥胖肝组织样本和 32 个非酒精性脂肪肝组织样本。GSE89632 数据集使用 GPL14951，包含 63 个样本，包括 24 个正常肝脏组织对照样本和 39 个非酒精性脂肪肝组织样本。所有样本均纳入本研究。此外，GSE63067 数据集使用了 GPL14877，包含 18 个样本，其中包括 7 个正常肝脏对照样本和 11 个非酒精性脂肪肝组织样本。本研究使用数据集 GSE48452 作为验证集。合并 GSE63067 和 GSE89632 数据集，并使用 SVA 软件包去除批次效应。 14 14 ^(14){ }^{14} 最后，使用 R 软件包 "Limma "进行分析。 15 15 ^(15){ }^{15} 本研究的流程图见图 S1。

2.2 | Evaluating the immune cell infiltration
2.2 | 评估免疫细胞浸润情况

Cibersort was used with the LM22 genetic characteristic matrix (https:/cibersort.stanford.edu/) algorithm, based on the gene expression profile assessment subtype of the immune system cells within each sample. Furthermore, the p p pp-value of the backfold product of each sample was calculated based on Monte Carlo sampling, and the Wilcoxon rank sum test was used to estimate differences in
根据每个样本中免疫系统细胞的基因表达谱评估亚型，使用 LM22 遗传特征矩阵（https:/cibersort.stanford.edu/）算法进行 Cibersort 排序。此外，还根据蒙特卡洛取样法计算了每个样本的反褶积 p p pp 值，并用 Wilcoxon 秩和检验估算了每个样本的基因表达差异。
immune cell abundance between groups. In this study, p < 0.05 p < 0.05 p < 0.05p<0.05 was considered statistically significant.
组间的免疫细胞丰度。在本研究中， p < 0.05 p < 0.05 p < 0.05p<0.05 被认为具有统计学意义。

2.3 | Correlation analysis of CRGs and immune cell infiltration
2.3 CRG 与免疫细胞浸润的相关性分析

To further confirm the relationship between CRG and NAFLDassociated immune cell properties, the relationship between CRG expression and the relative proportion of immune cells was examined. Spearman’s correlation coefficient and its associated p p pp-value were used to evaluate the correlation, with a p p pp-value < 0.05 < 0.05 < 0.05<0.05 indicating a significant association. Finally, the results were displayed using the ‘corrplot’ R package (version 0.92).
为进一步证实 CRG 与非酒精性脂肪肝相关免疫细胞特性之间的关系，研究了 CRG 表达与免疫细胞相对比例之间的关系。使用斯皮尔曼相关系数及其相关的 p p pp 值来评估相关性， p p pp 值 < 0.05 < 0.05 < 0.05<0.05 表示有显著关联。最后，使用 "corrplot "R 软件包（0.92 版）显示结果。

2.4 | Unsupervised clustering of NAFLD patients
2.4 非酒精性脂肪肝患者的无监督聚类

Cuproptosis-related genes were derived from previous studies. 16 16 ^(16){ }^{16} Based on the expression profiles of 9 cuproptosis genes with significantly different expressions, an unsupervised cluster analysis 17 17 ^(17){ }^{17} was performed to classify 50 NAFLD samples into different clusters through 1000 iterations of the k k kk-means algorithm. A maximum subtype number k ( k = 9 ) k ( k = 9 ) k(k=9)\mathrm{k}(\mathrm{k}=9) was selected, and the optimal subtype number was comprehensively evaluated based on the cumulative distribution function (CDF) curve, consensus matrix and consensus cluster score ( > 0.9 > 0.9 > 0.9>0.9 ). PCA (Principal Component Analysis) analysis showed differences in the distribution of cuproptosis among subtypes and was visualized using the ‘ggplot2’ software package.
杯突相关基因来自先前的研究。 16 16 ^(16){ }^{16} 根据 9 个表达差异显著的杯突症基因的表达谱，进行无监督聚类分析 17 17 ^(17){ }^{17} ，通过 1000 次迭代 k k kk 均值算法将 50 个非酒精性脂肪肝样本分为不同的聚类。根据累积分布函数（CDF）曲线、共识矩阵和共识聚类得分（ > 0.9 > 0.9 > 0.9>0.9 ）综合评估最佳亚型数。PCA（主成分分析）分析显示了杯突症在不同亚型中的分布差异，并使用 "ggplot2 "软件包将其可视化。

2.5 | Gene set variation analysis (GSVA) analysis
2.5 基因组变异分析（GSVA）分析

The enrichment analysis of different CRG clusters was analysed by the GSVA (Version 2.11) R package. Then, the files ‘c2.cp.kegg. v7.4.symbals’ and ‘c5.go.bp.v7.5.1.symbols’ were obtained from the MSigDB database for further GSVA analysis. The ‘LIMMA’ R package (version 3.52.1) was used to determine differential expression pathways and biological functions by comparing GSVA scores between different CRG clusters.
不同 CRG 簇的富集分析由 GSVA（2.11 版）R 软件包进行。然后，从 MSigDB 数据库中获取 "c2.cp.kegg. v7.4.symbals "和 "c5.go.bp.v7.5.1.symbols "文件，进一步进行 GSVA 分析。LIMMA "R 软件包（3.52.1 版）用于通过比较不同 CRG 集群之间的 GSVA 得分来确定差异表达途径和生物功能。

2.6 | Weighted gene co-expression network analysis (WGCNA)
2.6 加权基因共表达网络分析（WGCNA）

The R package 17 17 ^(17){ }^{17} ‘WGCNA’ (version 1.70.3) was used to identify coexpression modules. To ensure the accuracy of the study, we first grouped the samples and eliminated outliers. A soft threshold from 1 to 20 was used for topology calculation to determine the optimal soft threshold. When the minimum module size was set to 100 , a ‘dynamic tree cutting’ algorithm was used to group genes with similar patterns into modules. Finally, Pearson correlation analysis was performed to calculate the correlation between modules and traits.
我们使用 R 软件包 17 17 ^(17){ }^{17} "WGCNA"（1.70.3 版）来识别共表达模块。为确保研究的准确性，我们首先对样本进行了分组，并剔除了异常值。拓扑计算采用了 1 到 20 的软阈值，以确定最佳软阈值。当最小模块大小设定为 100 时，我们使用 "动态树切割 "算法将具有相似模式的基因归入模块。最后，进行皮尔逊相关分析，计算模块与性状之间的相关性。
Based on the correlation between modules and clinical features, the most relevant module to the disease is selected as the key module.
根据模块与临床特征之间的相关性，选择与疾病最相关的模块作为关键模块。

2.7 | Construction of multiple machine learning prediction models
2.7 | 构建多种机器学习预测模型

Based on two different CRG clusters, the ‘caret’ R package (version 6.0.91) was applied to build machine learning models, including the random forest model (RF), support vector machine model (SVM), generalized linear model (GLM) and eXtreme Gradient Boosting (XGB). RF is a machine learning technique that employs multiple independent decision trees for classification or regression predictions. 18 18 ^(18){ }^{18} In contrast, the support vector machine algorithm generates a hyperplane with a maximum margin in the feature space to distinguish positive from negative instances. 19 19 ^(19){ }^{19} The generalized linear regression model is an extension of the multiple linear regression model that flexibly estimates the relationship between normally distributed correlated features and categorical or continuous independent features. 20 20 ^(20){ }^{20} Furthermore, XGB is a gradient boosting-based boost tree ensemble that enables careful comparison between classification error and model complexity. 21 21 ^(21){ }^{21} Different clusters were used as the response variable, while cluster-specific DEGs were selected as explanatory variables. The 50 NAFLD samples were randomly divided into the training set ( 70 % , N = 35 ) ( 70 % , N = 35 ) (70%,N=35)(70 \%, N=35) and the validation set ( 30 % , N = 15 ) ( 30 % , N = 15 ) (30%,N=15)(30 \%, N=15). The parameters of these models were automatically tuned using web searches, and all of these machine learning models were run with preset parameters and evaluated by 5 -fold cross-validation. The ‘Dalex’ package (version 2.4.0) was developed to interpret the above four machine learning models and to visualize the distribution of residuals and the importance of features in these machine learning models. The area under the ROC curve was displayed using the ‘proc’ R package (version 1.18.0). Therefore, the most suitable machine learning model was evaluated and the five most important variables were identified as the main predictor genes associated with NAFLD.
基于两个不同的 CRG 集群，应用 "caret "R 软件包（6.0.91 版）建立了机器学习模型，包括随机森林模型（RF）、支持向量机模型（SVM）、广义线性模型（GLM）和极梯度提升模型（XGB）。RF 是一种采用多个独立决策树进行分类或回归预测的机器学习技术。 18 18 ^(18){ }^{18} 相反，支持向量机算法在特征空间中生成一个具有最大边际的超平面，以区分正负实例。 19 19 ^(19){ }^{19} 广义线性回归模型是多元线性回归模型的扩展，可灵活估计正态分布相关特征与分类或连续独立特征之间的关系。 20 20 ^(20){ }^{20} 此外，XGB 是一种基于梯度提升的提升树集合，可对分类错误和模型复杂度进行仔细比较。 21 21 ^(21){ }^{21} 不同的聚类被用作响应变量，而聚类特定的 DEGs 被选为解释变量。50 个非酒精性脂肪肝样本被随机分为训练集 ( 70 % , N = 35 ) ( 70 % , N = 35 ) (70%,N=35)(70 \%, N=35) 和验证集 ( 30 % , N = 15 ) ( 30 % , N = 15 ) (30%,N=15)(30 \%, N=15) 。这些模型的参数是通过网络搜索自动调整的，所有这些机器学习模型都以预设参数运行，并通过 5 倍交叉验证进行评估。我们开发了 "Dalex "软件包（2.4.0 版）来解释上述四个机器学习模型，并可视化这些机器学习模型的残差分布和特征的重要性。 使用 "proc "R 软件包（1.18.0 版）显示了 ROC 曲线下的面积。因此，对最合适的机器学习模型进行了评估，并确定了五个最重要的变量作为与非酒精性脂肪肝相关的主要预测基因。

2.8 | Analysis of the diagnostic value of biomarkers
2.8 分析生物标志物的诊断价值

The receiver operating characteristic (ROC) curve was generated using the R R RR package ’ p R O C p R O C pROCp R O C ’ and the area under the ROC curve (AUC) value was calculated. The ability of the key predictor genes to distinguish NAFLD from non-NAFLD was externally validated using the GSE48452 dataset.
使用 R R RR 软件包" p R O C p R O C pROCp R O C "生成接收者操作特征曲线（ROC），并计算 ROC 曲线下面积（AUC）值。利用 GSE48452 数据集对关键预测基因区分非酒精性脂肪肝和非酒精性脂肪肝的能力进行了外部验证。

2.9 | Construction and validation of nomogram models
2.9 名称图模型的构建与验证

Eigengenes were combined to construct nomograms using the ‘rms’ R package. Subsequently, the accuracy of the nomogram was assessed using a calibration curve, and the clinical application value of the chart was evaluated by decision curve analysis.
使用 R 软件包 "rms "将 Eigengenes 结合起来构建提名图。随后，使用校准曲线评估了提名图的准确性，并通过决策曲线分析评估了图表的临床应用价值。

2.10 | Establishment of NAFLD animal model
2.10 | 建立非酒精性脂肪肝动物模型

Six-week-old male C57BL/6J mice ( n = 14 n = 14 n=14n=14 ) weighing about 22 g were provided by Liaoning Changsheng Biotechnology Co., Ltd., and were fed with a standard diet and standard 12 h : 12 h 12 h : 12 h 12h:12h12 \mathrm{~h}: 12 \mathrm{~h} light/ dark cycle until 8 weeks old. Then, the mice were randomly divided into two groups: one group was fed a high-fat diet (research diet D12451, 45 kcal saturated fat) to induce NAFLD ( n = 7 n = 7 n=7n=7 ), while the other group was fed a normal diet (5% fat, 53 % 53 % 53%53 \% carbohydrate, 23 % 23 % 23%23 \% protein), serving as the control group ( n = 7 n = 7 n=7n=7 ). The mice were fed for 16 weeks, and one mouse was randomly selected from each group and was sacrificed by cervical dislocation. Liver samples were collected, fixed in 4 % 4 % 4%4 \% formalin for 24 h and dehydrated with a series of ethanol solutions. Liver tissues were then embedded in paraffin and sectioned at a thickness of 4 μ M 4 μ M 4muM4 \mu \mathrm{M}. The sections were stained with haematoxylin and eosin (H&E). All human and animal studies were approved by the Medical Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University, and all methods were performed in accordance with relevant guidelines and regulations.
6 周龄雄性 C57BL/6J 小鼠（ n = 14 n = 14 n=14n=14 ）体重约 22 g，由辽宁昌盛生物技术有限公司提供，以标准饮食和标准 12 h : 12 h 12 h : 12 h 12h:12h12 \mathrm{~h}: 12 \mathrm{~h} 光/暗循环饲喂至 8 周龄。然后将小鼠随机分为两组：一组饲喂高脂饮食（研究饮食 D12451，45 千卡饱和脂肪）诱导非酒精性脂肪肝（ n = 7 n = 7 n=7n=7 ），另一组饲喂正常饮食（5%脂肪、 53 % 53 % 53%53 \% 碳水化合物、 23 % 23 % 23%23 \% 蛋白质），作为对照组（ n = 7 n = 7 n=7n=7 ）。小鼠喂养 16 周后，每组随机抽取一只小鼠，颈椎脱位处死。收集肝脏样本，在 4 % 4 % 4%4 \% 福尔马林中固定 24 小时，并用一系列乙醇溶液脱水。然后用石蜡包埋肝组织，切片厚度为 4 μ M 4 μ M 4muM4 \mu \mathrm{M} 。切片用血红素和伊红（H&E）染色。所有人类和动物研究均已获得哈尔滨医科大学附属第四医院医学伦理委员会批准，所有方法均按照相关指南和法规执行。

2.11 | Cell culture and treatment
2.11 | 细胞培养和处理

Mouse hepatocyte cell line (AML12) were purchased from Procell Life Science and Technology (Wuhan, China), maintained and propagated in DMEM/F-12 with 10% FBS, ITS Liquid Media Supplement (Sigma) and 0.1 μ M 0.1 μ M 0.1 muM0.1 \mu \mathrm{M} dexamethasone at 37 ∘ C 37 ∘ C 37^(@)C37^{\circ} \mathrm{C} incubators containing 5 % CO 2 5 % CO 2 5%CO_(2)5 \% \mathrm{CO}_{2}. To establish the in vitro NAFLD cell model, AML12 were cultured in the presence or absence of 1 mM free fatty acids (FFA, containing oleic acid and palmitic acid at a 2:1 volume ratio) for 48 h and then used for the indicated assays.
小鼠肝细胞系（AML12）购自 Procell 生命科技公司（中国武汉），在含有 10% FBS、ITS 液体培养基补充剂（Sigma）和 0.1 μ M 0.1 μ M 0.1 muM0.1 \mu \mathrm{M} 地塞米松的 DMEM/F-12 培养液中培养，培养箱为 37 ∘ C 37 ∘ C 37^(@)C37^{\circ} \mathrm{C} 。为了建立体外非酒精性脂肪肝细胞模型，在有或没有 1 mM 游离脂肪酸（FFA，含油酸和棕榈酸，体积比为 2:1）的情况下培养 AML12 48 小时，然后用于指定的检测。

2.12 | Oil red O staining
2.12 | 油红 O 染色

Oil Red O staining was applied to assess lipid droplet formation in AML12 cells according to a previously described method. 22 22 ^(22){ }^{22} Briefly, cells were washed twice with PBS, fixed in 4 % 4 % 4%4 \% paraformaldehyde for 0.5 h and stained for 30 min with a 0.5 % 0.5 % 0.5%0.5 \% Oil Red O solution in 60 % 60 % 60%60 \% isopropanol. The cells were washed with PBS before analysis. The images were captured under an inverted microscope (Olympus Corporation) at 100 × 100 × 100 xx100 \times magnification.
根据先前描述的方法，用油红 O 染色法评估 AML12 细胞中脂滴的形成。 22 22 ^(22){ }^{22} 简单地说，细胞用 PBS 洗两次，在 4 % 4 % 4%4 \% 多聚甲醛中固定 0.5 小时，然后用 0.5 % 0.5 % 0.5%0.5 \% 异丙醇中的 60 % 60 % 60%60 \% 油红 O 溶液染色 30 分钟。分析前用 PBS 冲洗细胞。在倒置显微镜（奥林巴斯公司）下以 100 × 100 × 100 xx100 \times 倍率拍摄图像。

2.13 | Determination of hepatic triglycerides and total cholesterol
2.13 | 肝甘油三酯和总胆固醇的测定

Hepatic triglyceride (TG) and liver cholesterol (TC) were measured by Beijing Solarbio ScienceTechnology Co., Ltd. The experimental procedure is briefly described below. Approximately 50 mg of the liver sample was excised, homogenized and then lysed in lysis buffer for 10 min . After centrifugation at 2000 g for 5 min , the supernatant
肝脏甘油三酯（TG）和肝脏胆固醇（TC）由北京索拉生物科技有限公司测定。实验过程简述如下。切除约 50 毫克肝脏样本，匀浆后在裂解缓冲液中裂解 10 分钟。在 2000 g 转速下离心 5 分钟后，上清液
was divided into two parts: one part was used to determine liver triglyceride (TG) and liver cholesterol (TC), and the other part was used to evaluate protein concentration. Protein quantification was performed using a BCA protein assay kit (P0011, Beyotime, Shanghai, China). Finally, hepatic triglycerides (TG) and hepatic cholesterol (TC) were normalized by protein concentration.
实验分为两部分：一部分用于测定肝脏甘油三酯（TG）和肝脏胆固醇（TC），另一部分用于评估蛋白质浓度。蛋白质定量使用 BCA 蛋白检测试剂盒（P0011，上海贝奥天美生物科技有限公司）。最后，用蛋白质浓度对肝甘油三酯（TG）和肝胆固醇（TC）进行归一化。

2.14 | Quantitative RT-PCR analysis
2.14 | 定量 RT-PCR 分析

Total RNA was extracted from liver tissue and AML12 cells using Trizol reagent (Invitrogen) and reverse transcribed using the PrimeScript RT kit (Takara, Tokyo, Japan) according to the manufacturer’s instructions. Real-time quantitative polymerase chain reaction was performed on QuantStudio 3 (Thermo Fisher Scientific China) using SYBR Green PreMix Ex Taq (Takara, Japan), and data analysis was performed using the 2 − Δ Δ C t 2 − Δ Δ C t 2^(-Delta DeltaC_(t))2^{-\Delta \Delta C_{t}} method, with β β beta\beta-Actin as the internal control for normalization. The primer sequences used in RT-PCR are displayed in Table 1.
使用 Trizol 试剂（Invitrogen 公司）从肝组织和 AML12 细胞中提取总 RNA，并按照生产商的说明使用 PrimeScript RT 试剂盒（日本东京，Takara 公司）进行反转录。使用 SYBR Green PreMix Ex Taq（日本 Takara 公司）在 QuantStudio 3（赛默飞世尔科技中国公司）上进行实时定量聚合酶链反应，采用 2 − Δ Δ C t 2 − Δ Δ C t 2^(-Delta DeltaC_(t))2^{-\Delta \Delta C_{t}} 法进行数据分析，以 β β beta\beta -Actin 作为归一化的内部对照。表 1 列出了 RT-PCR 中使用的引物序列。

2.15 | Statistical analysis
2.15 | 统计分析

All data are presented as mean ± ± +-\pm standard deviation. Statistical analysis between groups was performed using an unpaired two-tailed Student’s t t tt-test. In this study, p < 0.05 p < 0.05 p < 0.05p<0.05 was considered statistically significant. All analyses were performed using GraphPad Prism 8 (GraphPad Software, San Diego, USA).
所有数据均以平均 ± ± +-\pm 标准差表示。组间统计分析采用非配对双尾学生 t t tt 检验。在本研究中， p < 0.05 p < 0.05 p < 0.05p<0.05 被认为具有统计学意义。所有分析均使用 GraphPad Prism 8（GraphPad Software，美国圣地亚哥）进行。

3 | RESULTS  3 结果

3.1 | Cuproptosis-associated gene dysregulation and immune cell infiltration in patients with non-alcoholic fatty liver disease
3.1 | 非酒精性脂肪肝患者的杯突症相关基因失调和免疫细胞浸润

To elucidate the biological functions of cuproptosis regulators in the development and progression of non-alcoholic fatty liver disease (NAFLD), the expression profiles of 19 cuproptosis-associated genes between NAFLD and non-NAFLD controls were first assessed using a merged dataset of GSE89632 and GSE63067. The dataset consisted of 50 NAFLD tissues and 31 normal liver tissues.
为了阐明杯突调节因子在非酒精性脂肪性肝病（NAFLD）发生和发展过程中的生物学功能，研究人员首先使用 GSE89632 和 GSE63067 合并数据集评估了非酒精性脂肪性肝病和非酒精性脂肪性肝病对照组之间 19 个杯突相关基因的表达谱。该数据集包括 50 个非酒精性脂肪肝组织和 31 个正常肝组织。
TABLE 1 Primers used for RT-PCR analysis.
表 1 RT-PCR 分析所用引物
  基因符号
Gene
symbol
Gene symbol| Gene | | :--- | | symbol |
 Species  物种 Forward primer  前向引物 Reverse primer  反向引物
"Gene symbol" Species Forward primer Reverse primer| Gene <br> symbol | Species | Forward primer | Reverse primer | | :--- | :--- | :--- | :--- |
Figure S 2 shows the data before batch correction ( A and B ) and after batch correction ( C C CC and D D DD ), suggesting the successful elimination of the batch effect from the pooled data. A total of nine genes were selected as cuproptosis regulators with more significant expression differences. Some showed elevated expression in NAFLD patients compared to non-NAFLD controls, including ATP7B, SLC31A1, LIAS, DLD, PDHA1, PDHB and DBT. In contrast, NFE2L2 and MTF1 showed decreased expression in NAFLD tissues compared to non-NAFLD controls (Figure 1A-C). Moreover, correlation analysis was performed between the differentially expressed cuproptosis genes, demonstrating a strong synergistic effect between LIAS and PDHA1, and a clear antagonism between MTF1 and LIAS. In addition, the correlation between these cuproptosis genes was further investigated, as shown in Figure 1D,E. Immune infiltration analysis was performed to visualize differences in the proportions of 22 infiltrating immune cell types between NAFLD controls and non-NAFLD control subjects using the CiberSort algorithm (Figure 1F). The results showed higher levels of γ γ gamma\gamma-delta T cells, M1 macrophages, M2 macrophages, resting dendritic cells and resting mast cells in NAFLD samples compared to the control group (Figure 1G), suggesting that alterations in the immune system may be involved in the development of NAFLD. Furthermore, correlation analysis results revealed that activated CD4 memory T cells, M2 macrophages and activated mast cells were all associated with cuproptosis-related genes (Figure 1H). These results suggest that CRGs may be a key factor regulating the occurrence and immune infiltration status of NAFLD patients.
图 S 2 显示了批次校正前（A 和 B）和批次校正后（ C C CC 和 D D DD ）的数据，表明成功消除了集合数据中的批次效应。共有九个基因被选为表达差异较显著的杯突症调节因子。与非非酒精性脂肪肝对照组相比，其中一些基因在非酒精性脂肪肝患者中表达升高，包括 ATP7B、SLC31A1、LIAS、DLD、PDHA1、PDHB 和 DBT。相反，与非非酒精性脂肪肝对照组相比，NFE2L2 和 MTF1 在非酒精性脂肪肝组织中的表达有所下降（图 1A-C）。此外，还对不同表达的杯突症基因进行了相关性分析，结果表明 LIAS 和 PDHA1 之间有很强的协同作用，而 MTF1 和 LIAS 之间有明显的拮抗作用。此外，还进一步研究了这些杯突基因之间的相关性，如图 1D,E 所示。利用 CiberSort 算法进行了免疫浸润分析，以观察非酒精性脂肪肝对照组和非酒精性脂肪肝对照组之间 22 种浸润免疫细胞类型比例的差异（图 1F）。结果显示，与对照组相比，非酒精性脂肪肝样本中 γ γ gamma\gamma -δT 细胞、M1 巨噬细胞、M2 巨噬细胞、静止树突状细胞和静止肥大细胞的水平更高（图 1G），这表明免疫系统的改变可能与非酒精性脂肪肝的发病有关。此外，相关性分析结果显示，活化的 CD4 记忆 T 细胞、M2 巨噬细胞和活化的肥大细胞都与杯突相关基因有关（图 1H）。这些结果表明，杯突相关基因可能是调节非酒精性脂肪肝患者发生和免疫浸润状态的关键因素。

3.2 | Identification of cuproptosis clusters in NAFLD
3.2 非酒精性脂肪肝中杯突症群的鉴定

Additionally, 50 NAFLD samples were grouped based on the expression profiles of 9 differentially expressed CRGs using a consensus clustering algorithm to identify the expression patterns of genes associated with copper death in NAFLD. The number of clusters was most stable when the k k kk value was set to 2 ( k = 2 ) 2 ( k = 2 ) 2(k=2)2(k=2) (Figure 2A), and the CDF curve fluctuated within the smallest range of consensus index 0.2 to 0.6 (Figure 2B). When k = 2 ∼ 9 k = 2 ∼ 9 k=2∼9k=2 \sim 9, the area under the CDF curve showed a difference between the two CDF curves ( k k kk and k − 1 k − 1 k-1k-1 ) (Figure 2C). Furthermore, when k = 2 k = 2 k=2k=2, the concordance scores for each subtype > 0.9 > 0.9 > 0.9>0.9. (Figure 2D). Therefore, the 50 NAFLD patients were divided into two groups, including group 1 ( n = 21 ) 1 ( n = 21 ) 1(n=21)1(n=21) and group 2 ( n = 29 ) 2 ( n = 29 ) 2(n=29)2(n=29). The results of the PCA (Principal Component Analysis) analysis showed significant differences between the two clusters (Figure 2E).
此外，根据 9 个差异表达的 CRGs 的表达谱，采用共识聚类算法对 50 个非酒精性脂肪肝样本进行了分组，以确定非酒精性脂肪肝中铜死亡相关基因的表达模式。当 k k kk 值设置为 2 ( k = 2 ) 2 ( k = 2 ) 2(k=2)2(k=2) 时，聚类数目最稳定（图 2A），CDF 曲线在共识指数 0.2 至 0.6 的最小范围内波动（图 2B）。当 k = 2 ∼ 9 k = 2 ∼ 9 k=2∼9k=2 \sim 9 时，CDF 曲线下的面积显示出两条 CDF 曲线（ k k kk 和 k − 1 k − 1 k-1k-1 ）之间的差异（图 2C）。此外，当 k = 2 k = 2 k=2k=2 时，各亚型的一致性评分为 > 0.9 > 0.9 > 0.9>0.9 .（图 2D）。因此，50 名非酒精性脂肪肝患者被分为两组，包括 1 ( n = 21 ) 1 ( n = 21 ) 1(n=21)1(n=21) 组和 2 ( n = 29 ) 2 ( n = 29 ) 2(n=29)2(n=29) 组。主成分分析（PCA）结果显示，两组之间存在显著差异（图 2E）。

3.3 | Differential expression of genes regulated by cuproptosis and immune infiltration signatures of associated cuproptosis clusters
3.3 | 杯突症调控基因的差异表达和相关杯突症集群的免疫浸润特征

The expression differences of 9 CRGs between Cluster 1 and Cluster 2 were comprehensively assessed to further explore the molecular
综合评估了 9 个 CRGs 在群组 1 和群组 2 中的表达差异，以进一步探索分子
features between subgroups. Cuproptosis Cluster1 showed high expression of MTF1 and ATP7B, while cuproptosis Cluster2 was characterized by enhanced expression of LIAS and PDHA1 (Figure 3A,B). In addition, the results of the immune infiltration assay revealed differences in the immune microenvironment between cuproptosis Cluster1 and Cluster2 (Figure 3C). Among them, the most significant changes were γ γ gamma\gamma-Delta T cells, which had a higher proportion in Cluster2 (Figure 3D).
亚群之间的特征。杯状突变簇 1 显示了 MTF1 和 ATP7B 的高表达，而杯状突变簇 2 的特点是 LIAS 和 PDHA1 的表达增强（图 3A,B）。此外，免疫浸润试验结果显示杯突病变群 1 和群 2 的免疫微环境存在差异（图 3C）。其中，变化最大的是 γ γ gamma\gamma -Delta T 细胞，其在 Cluster2 中的比例更高（图 3D）。

3.4 | Construction of gene weighted co-expression module and gene screening
3.4 基因加权共表达模块的构建与基因筛选

The WGCNA algorithm was used to establish co-expression networks and modules in normal controls and individuals with NAFLD to identify the key gene modules associated with NAFLD. Co-expressed gene modules were identified with the soft power value set to 9 and the scale-free R2 set to 0.9 (Figure 4A). The dynamic cut algorithm was used to obtain 4 co-expressed gene modules with different colours, and a topological overlap matrix (TOM) heat map was produced (Figure 4B-E). Subsequently, these genes in the 4 colour modules were applied sequentially to analyse the similarity and contiguity of the module-clinical signature (control and NAFLD) co-expression. The results revealed that the blue module was most closely related to NAFLD and included 1330 genes (Figure 4F). In addition, a strong positive correlation was found between the blue modules and module-associated genes. In addition, the WGCNA algorithm was used to analyse key gene modules closely related to cuproptosis genes. β = 9 β = 9 beta=9\beta=9 and R 2 = 0.9 R 2 = 0.9 R2=0.9\mathrm{R} 2=0.9 were screened as the most suitable soft threshold parameters for building scale-free networks (Figure 5A). Five modules were identified as significant modules, and heatmaps were generated to depict the TOMs of all module-associated genes (Figure 5B-E). Module-clinical feature (Cluster1 and Cluster2) relationship analysis revealed a high correlation between brown modules ( 556 genes) and NAFLD clusters (Figure 5F). Moreover, correlation analysis showed that brown module genes were significantly correlated with selected modules (Figure 5F).
利用 WGCNA 算法在正常对照组和非酒精性脂肪肝患者中建立共表达网络和模块，以确定与非酒精性脂肪肝相关的关键基因模块。共表达基因模块的软功率值设为 9，无标度 R2 设为 0.9（图 4A）。使用动态切割算法得到了 4 个不同颜色的共表达基因模块，并绘制了拓扑重叠矩阵（TOM）热图（图 4B-E）。随后，依次应用 4 种颜色模块中的这些基因，分析模块-临床特征（对照和非酒精性脂肪肝）共表达的相似性和连续性。结果显示，蓝色模块与非酒精性脂肪肝的关系最为密切，包含 1330 个基因（图 4F）。此外，蓝色模块与模块相关基因之间存在很强的正相关性。此外，还利用 WGCNA 算法分析了与杯突症基因密切相关的关键基因模块。 β = 9 β = 9 beta=9\beta=9 和 R 2 = 0.9 R 2 = 0.9 R2=0.9\mathrm{R} 2=0.9 被筛选为最适合构建无标度网络的软阈值参数（图 5A）。五个模块被确定为重要模块，并生成热图来描述所有模块相关基因的 TOMs（图 5B-E）。模块-临床特征（Cluster1 和 Cluster2）关系分析表明，棕色模块（556 个基因）与非酒精性脂肪肝集群之间存在高度相关性（图 5F）。此外，相关性分析表明棕色模块基因与所选模块显著相关（图 5F）。

3.5 | Identification of cluster-specific DEGs and functional annotation
3.5 集群特异性 DEGs 的鉴定和功能注释

A total of 199 cluster-specific DEGs were identified by analysing the intersection of module-associated genes of a subset of cuproptosis genes with those of NAFLD and non-NAFLD individuals (Figure 6A). GSVA analysis was used to further explore the functional differences between the two clusters associated with cluster-specific DEGs. The results showed that the galactose metabolism pathway, JAK-STAT signalling pathway, MAKP signalling pathway and adipocytokines signalling pathway were upregulated in Cluster2, while oxidative phosphorylation, base excision repair, steroid hormone biosynthesis pathway and amino sugar and nucleotide sugar
通过分析杯状突变基因子集的模块相关基因与非 NAFLD 和非 NAFLD 个体的模块相关基因的交叉点，共鉴定出 199 个集群特异性 DEGs（图 6A）。利用 GSVA 分析进一步探讨了两个集群与集群特异性 DEGs 相关的功能差异。结果显示，Cluster2 中半乳糖代谢通路、JAK-STAT 信号通路、MAKP 信号通路和脂肪细胞因子信号通路被上调，而氧化磷酸化、碱基切除修复、类固醇激素生物合成通路和氨基糖及核苷酸糖等通路被上调。
FIGURE 1 CRGs expression levels in NAFLD. (A) Boxplots showed the expression of 17 CRGs between NAFLD and non-NAFLD controls. ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(****)p < 0.01,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* *} p<0.01,{ }^{* * *} p<0.001, ns, no significance. (B) The expression patterns of 9 CRGs were presented in the heatmap. © The location of 9 CRGs on chromosomes. (D) Correlation analysis of 9 differentially expressed CRGs. (E) Gene relationship network diagram of 9 differentially expressed CRGs. (F) The relative abundances of 22 infiltrated immune cells between NAFLD and non-NAFLD controls. (G) Boxplots showed the differences in immune infiltrating between NAFLD and non-NAFLD controls. ∗ p < 0.05 , ∗ ∗ p < 0.01 ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ p < 0.01 ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(****)p < 0.01^(******)p < 0.001{ }^{*} p<0.05,{ }^{* *} p<0.01^{* * *} p<0.001, ns, no significance. (H) Correlation analysis between 9 differentially expressed CRGs and infiltrated immune cells.
图 1 非酒精性脂肪肝中 CRGs 的表达水平。(A) 方框图显示了非酒精性脂肪肝与非非酒精性脂肪肝对照组之间 17 种 CRGs 的表达情况。 ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(****)p < 0.01,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* *} p<0.01,{ }^{* * *} p<0.001 ，ns，无显著性。(B) 热图显示了 9 个 CRGs 的表达模式。9 个 CRGs 在染色体上的位置。(D) 9 个差异表达 CRG 的相关性分析。(E) 9 个差异表达 CRG 的基因关系网络图。(F）非酒精性脂肪肝与非非酒精性脂肪肝对照组之间 22 种浸润免疫细胞的相对丰度。(G）方框图显示非酒精性脂肪肝与非非酒精性脂肪肝对照组免疫浸润的差异。 ∗ p < 0.05 , ∗ ∗ p < 0.01 ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ p < 0.01 ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(****)p < 0.01^(******)p < 0.001{ }^{*} p<0.05,{ }^{* *} p<0.01^{* * *} p<0.001 ，ns，无显著性。(H）9 个差异表达的 CRG 与浸润的免疫细胞之间的相关性分析。
FIGURE 2 Identification of cuproptosis-related molecular patterns in NAFLD. (A) Consensus clustering matrix when k = 2 k = 2 k=2k=2. (B-E) Representative CDF curves when k = 2 k = 2 k=2k=2 to 9 . © Relative alterations in CDF delta area curves. (D) Consensus score in each subtype when k = 2 k = 2 k=2k=2 to 9. (E) PCA diagram demonstrates that NAFLD patients are classified into two distinct subtypes.
图 2 非酒精性脂肪肝中杯突症相关分子模式的鉴定。(A) k = 2 k = 2 k=2k=2 时的共识聚类矩阵。 (B-E) k = 2 k = 2 k=2k=2 至 9 时的代表性 CDF 曲线。 © CDF delta 面积曲线的相对变化。(D) 当 k = 2 k = 2 k=2k=2 至 9 时，各亚型的共识得分。 (E) PCA 图显示 NAFLD 患者被分为两个不同的亚型。
metabolism were upregulated in Cluster1 (Figure 6B). Moreover, the functional enrichment results indicated that Cluster2 was associated with the regulation of transsynaptic signalling, maturation of synapses and carbohydrate phosphatase activity (Figure 6C).
图 6B）。此外，功能富集结果表明，Cluster2 与跨突触信号的调控、突触的成熟和碳水化合物磷酸酶的活性有关（图 6C）。

3.6 | Machine learning model construction and evaluation
3.6 机器学习模型的构建与评估

In total, 199 genes of module-related genes from individuals with and without NAFLD were intersected based on the subset of cuproptosis genes to further identify cluster-specific genes with high diagnostic value. Subsequently, four well-established machine learning models (random forest model, support vector machine model, generalized linear model and extreme gradient boosting) were constructed. The four machine learning models were interpreted by the R package ‘Dalex’ and the distribution of the residuals of each model in the test set was plotted. The RF and SVM machine learning models yielded relatively low residuals (Figure 7A,B). Then, the top 10 significant feature variables of each model were ranked by root mean square error (RMSE) (Figure 7C). Furthermore, the discriminative performance of the four
在杯突症基因子集的基础上，对非酒精性脂肪肝患者和非酒精性脂肪肝非患者的模块相关基因共 199 个基因进行了交叉分析，以进一步确定具有高诊断价值的群集特异性基因。随后，构建了四种成熟的机器学习模型（随机森林模型、支持向量机模型、广义线性模型和极端梯度提升模型）。这四个机器学习模型由 R 软件包 "Dalex "解释，并绘制了每个模型在测试集中的残差分布图。RF 和 SVM 机器学习模型的残差相对较低（图 7A、B）。然后，根据均方根误差（RMSE）对每个模型的前 10 个重要特征变量进行了排名（图 7C）。此外，四个
machine learning algorithms was tested and evaluated on the test set by computing receiver operating characteristic (ROC) curves based on fivefold cross-validation, and the machine learning model with the largest area under the curve was selected. The support vector machine model (SVM) machine learning model showed the largest area under the ROC curve (AUC=0.950, Figure 7D), demonstrating that this algorithm had the best performance in discriminating patients from different clusters. According to these results, the five most significant variables (SLC16A1, FCAMR, RAB26, ENO3 and LEPR) were selected from the SVM model as predictive genes for further analysis. The expression differences of the signature genes were then validated in dataset GSE48452, which was derived from 32 NAFLD patients and 14 normal population controls. The results revealed that the expression differences of SLC16A1, ENO3 and LEPR were more obvious in the validation set, with SLC16A1 and LEPR being decreased in NAFLD, and ENO3 increased in NAFLD (Figure 8A,B). ROC analysis was further performed, and the results showed its good efficiency in the diagnosis of NAFLD. SLC16A1 (Figure 8C, AUC = 0.801 = 0.801 =0.801=0.801 ), LEPR (Figure 8E, AUC=0.721), ENO3 (Figure 8D, AUC=0.730). A nomogram was built to predict NAFLD progression to further evaluate the predictive power of the SVM model. In the nomogram, the formula for
通过计算基于五倍交叉验证的接收器操作特征曲线（ROC），在测试集上对机器学习算法进行了测试和评估，并选出了曲线下面积最大的机器学习模型。支持向量机模型（SVM）的机器学习模型显示出最大的 ROC 曲线下面积（AUC=0.950，图 7D），表明该算法在区分不同群组的患者方面性能最佳。根据这些结果，我们从 SVM 模型中选出了五个最显著的变量（SLC16A1、FCAMR、RAB26、ENO3 和 LEPR）作为预测基因进行进一步分析。然后在数据集 GSE48452 中验证了特征基因的表达差异，该数据集来自 32 例非酒精性脂肪肝患者和 14 例正常人群对照。结果显示，SLC16A1、ENO3 和 LEPR 在验证集中的表达差异更为明显，SLC16A1 和 LEPR 在非酒精性脂肪肝患者中降低，而 ENO3 在非酒精性脂肪肝患者中升高（图 8A,B）。进一步进行了 ROC 分析，结果表明该方法在诊断非酒精性脂肪肝方面具有良好的效果。SLC16A1（图 8C，AUC = 0.801 = 0.801 =0.801=0.801 ）、LEPR（图 8E，AUC=0.721）、ENO3（图 8D，AUC=0.730）。为进一步评估 SVM 模型的预测能力，建立了预测非酒精性脂肪肝进展的提名图。在提名图中，公式为


© (20%%
(B)

(D)
FIGURE 3 Identification of the differentiation of feature genes and immune characteristics between two cuproptosis clusters. (A) Boxplots show the expression of 9 characteristic genes between cuproptosis subtypes. ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* * *} p<0.001, ns, no significance. (B) Heatmap reveals the differential expression of 9 characteristic genes between cuproptosis subtypes. © The relative abundances of 22 infiltrated immune cells between two cuproptosis clusters. (D) Boxplots show the differences in infiltrated immune cells between cuproptosis subtypes. ∗ p < 0.05 ∗ p < 0.05 ^(**)p < 0.05{ }^{*} p<0.05, ns, no significance.
图 3 识别两个杯状突变亚型之间特征基因和免疫特征的差异。(A) 方框图显示杯突亚型之间 9 个特征基因的表达。 ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* * *} p<0.001 ，ns，无显著性。(B) 热图显示杯突亚型之间 9 个特征基因的差异表达。两个杯突亚型之间 22 个浸润免疫细胞的相对丰度。(D) 方框图显示杯突亚型之间浸润免疫细胞的差异。 ∗ p < 0.05 ∗ p < 0.05 ^(**)p < 0.05{ }^{*} p<0.05 ，ns，无显著性。
each characteristic variable corresponds to a point value, and the total value corresponds to multiple NAFLD risks, obtained by summing the values of all characteristic variables (Figure 9A). The standard curve confirmed the ability of the nomogram to accurately assess the progression of NAFLD (Figure 9B). Additionally, decision curve analysis indicated that the nomogram may provide clinical benefit to patients with NAFLD (Figure 9C). Collectively, these results suggest that this subset of cuproptosis-associated genes can accurately distinguish NAFLD from non-NAFLD cases.
每个特征变量对应一个点值，总值对应多个非酒精性脂肪肝风险，由所有特征变量值相加得出（图 9A）。标准曲线证实了提名图准确评估非酒精性脂肪肝进展的能力（图 9B）。此外，决策曲线分析表明，提名图可为非酒精性脂肪肝患者带来临床益处（图 9C）。总之，这些结果表明，杯突相关基因子集可以准确区分非酒精性脂肪肝和非酒精性脂肪肝病例。

3.7 | Verification of the expression level of Hub genes in vivo and vitro
3.7 | 验证枢纽基因在体内和体外的表达水平

In order to further verify the expression of genes ENO3, SLC16A1 and LEPR, a NAFLD mouse model was constructed. Therefore,
为了进一步验证 ENO3、SLC16A1 和 LEPR 基因的表达，我们构建了一个非酒精性脂肪肝小鼠模型。因此、
the mice were divided into two groups: the ND group (normal diet) and the NAFLD group (16-week HFD diet). H&E staining showed that HFD could induce the accumulation of lipid droplets in mouse liver cells, showing no obvious inflammatory reaction (Figure 9D). Furthermore, liver triglyceride levels and cholesterol levels were significantly elevated in mice fed a high-fat diet for 16 weeks (Figure 9E,F). Compared with the normal diet group, decreased expression of SLC16A1 and LEPR was observed in mice fed HFD, as well as increased expression of ENO3 (Figure 9G). Meanwhile, the in vitro studies were performed. Alpha mouse liver 12 (AML12) cells were treated with OA (oleate) and PA (palmitate) to induce fat accumulation, which indicates the features of hepatic steatosis. After FFA induction, AML12 cells sent a higher level of fat accumulation (Figure 10A), and the expression of ENO3 increased (Figure 10C), while the expression of SLC16A1 and LEPR decreased (Figure 10B,D).
将小鼠分为两组：ND 组（正常饮食）和 NAFLD 组（16 周 HFD 饮食）。H&E 染色显示，HFD 可诱导小鼠肝细胞中脂滴的积累，但未出现明显的炎症反应（图 9D）。此外，高脂饮食 16 周的小鼠肝脏甘油三酯水平和胆固醇水平显著升高（图 9E,F）。与正常饮食组相比，高脂饮食组小鼠的 SLC16A1 和 LEPR 表达降低，ENO3 表达升高（图 9G）。同时，还进行了体外研究。用 OA（油酸）和 PA（棕榈酸）处理阿尔法小鼠肝 12（AML12）细胞以诱导脂肪积累，这表明了肝脏脂肪变性的特征。FFA 诱导后，AML12 细胞出现了较高水平的脂肪堆积（图 10A），ENO3 的表达量增加（图 10C），而 SLC16A1 和 LEPR 的表达量减少（图 10B,D）。
FIGURE 4 Co-expression network of differentially expressed genes in NAFLD. (A) The selection of soft threshold power. (B) Cluster tree dendrogram of co-expression modules. Different colours represent distinct co-expression modules. © Representative of clustering of module eigengenes. (D) Representative heatmap of the correlations among 4 modules. (E) Correlation analysis between module eigengenes and clinical status. Each row represents a module; each column represents a clinical status. (F) Scatter plot between module membership in the blue module and the gene significance for NAFLD.
图 4 非酒精性脂肪肝中差异表达基因的共表达网络。(A) 软阈值功率的选择。(B) 共表达模块的聚类树枝图。不同颜色代表不同的共表达模块。不同颜色代表不同的共表 达模块。(D) 4 个模块之间相关性的代表性热图。(E) 模块基因与临床状态的相关性分析。每行代表一个模块，每列代表一种临床状态。(F) 蓝色模块中的模块成员与非酒精性脂肪肝基因重要性之间的散点图。
FIGURE 5 Co-expression network of differentially expressed genes between the two cuproptosis clusters. (A) The selection of soft threshold power. (B) Cluster tree dendrogram of co-expression modules. Different colours represent distinct co-expression modules. © Representative of clustering of module eigengenes. (D) Representative heatmap of the correlations among 5 modules. (E) Correlation analysis between module eigengenes and clinical status. Each row represents a module; each column represents a clinical status. (F) Scatter plot between module membership in the brown module and the gene significance for Cluster 2.
图 5 两个杯突病变群之间差异表达基因的共表达网络。(A) 软阈值功率的选择。(B) 共表达模块的簇树树枝图。不同颜色代表不同的共表 达模块。不同颜色代表不同的共表 达模块。(D) 5 个模块之间相关性的代表性热图。(E) 模块基因与临床状态的相关性分析。每行代表一个模块，每列代表一种临床状态。(F)棕色模块中的模块成员资格与第 2 组基因重要性之间的散点图。

(B)

FIGURE 6 Identifying cluster-specific DEGs and biological characteristics between two cuproptosis clusters. (A) Intersection of moduleassociated genes of a subset of cuproptosis genes with those of NAFLD and non-NAFLD individuals. (B) Differences in hallmark pathway activities between Cluster1 and Cluster2 samples ranked by t t tt-value of GSVA method. © Differences in biological functions between Cluster1 and Cluster2 samples ranked by t t tt-value of GSVA method.
图 6 在两个杯状变态反应群组之间识别群组特异性 DEGs 和生物学特征。(A）杯突症基因子集的模块相关基因与非酒精性脂肪肝和非酒精性脂肪肝个体的模块相关基因的交叉。(B) 按 GSVA 方法的 t t tt 值排列的 Cluster1 和 Cluster2 样本之间标志性通路活动的差异。簇 1 和簇 2 样本的生物功能差异按 GSVA 方法的 t t tt -值排列。

4 | DISCUSSION  4 | 讨论

Non-alcoholic liver disease (NAFLD) is the most common chronic liver disease. Nevertheless, the mechanisms of steatosis in NAFLD remain incompletely understood. Recent research has shown that minerals such as iron, copper, zinc and selenium play important roles in various biochemical processes. In addition, studies on ferroptosis and cuproptosis have further emphasized the importance of intracellular mineral homeostasis. However, mineral imbalances are common in patients with NAFLD and related diseases, and hepatic copper deficiency in NAFLD patients can lead to more pronounced steatosis, NASH and metabolic symptoms. 23 23 ^(23){ }^{23} A large cohort case-control study investigated the relationship between the severity of NAFLD and low blood copper concentrations in men. 24 24 ^(24){ }^{24} As copper is involved in mitochondrial function and fatty acid peroxisome β β beta\beta-based
非酒精性肝病（NAFLD）是最常见的慢性肝病。然而，人们对非酒精性脂肪肝中脂肪变性的机理仍不完全清楚。最新研究表明，铁、铜、锌和硒等矿物质在各种生化过程中发挥着重要作用。此外，有关铁变态反应和铜变态反应的研究进一步强调了细胞内矿物质平衡的重要性。然而，矿物质失衡在非酒精性脂肪肝和相关疾病患者中很常见，非酒精性脂肪肝患者肝铜缺乏可导致更明显的脂肪变性、非酒精性脂肪肝和代谢症状。 23 23 ^(23){ }^{23} 一项大型队列病例对照研究调查了男性非酒精性脂肪肝的严重程度与低血铜浓度之间的关系。 24 24 ^(24){ }^{24} 由于铜参与线粒体功能和脂肪酸过氧化物酶体 β β beta\beta -基
oxidation, copper deficiency can lead to mitochondrial dysfunction and oxidative stress. 25 , 26 25 , 26 ^(25,26){ }^{25,26} The mitochondrial morphology of copper-deficient rat hepatocytes was abnormally enlarged. 26 26 ^(26){ }^{26} Furthermore, direct experiments also showed that rats fed with a copper-deficient diet developed spontaneous liver steatosis. 27 27 ^(27){ }^{27} Interestingly, these studies aimed to determine the relationship between mineral disturbances and pathological features of NAFLD, including oxidative stress, mitochondrial dysfunction, inflammatory response and fibrogenesis. 12 12 ^(12){ }^{12} Cuproptosis is a recently reported copper-dependent cell death. However, the pathogenesis and regulation of cuproptosis in various diseases have not been investigated in detail. Therefore, this study aimed to elucidate the specific roles of cuproptosis-associated genes in the NAFLD phenotype and immune system infiltration. In addition, the clusters of NAFLD were predicted by using cuproptosisrelated genes for demarking.
氧化，缺铜会导致线粒体功能障碍和氧化应激。 25 , 26 25 , 26 ^(25,26){ }^{25,26} 缺铜大鼠肝细胞的线粒体形态异常增大。 26 26 ^(26){ }^{26} 此外，直接实验还表明，以缺铜饮食喂养的大鼠会出现自发性肝脏脂肪变性。 27 27 ^(27){ }^{27} 有趣的是，这些研究旨在确定矿物质紊乱与非酒精性脂肪肝病理特征（包括氧化应激、线粒体功能障碍、炎症反应和纤维化）之间的关系。 12 12 ^(12){ }^{12} 铜中毒是最近报道的一种铜依赖性细胞死亡。然而，铜氧化酶在各种疾病中的发病机制和调控尚未得到详细研究。因此，本研究旨在阐明杯突相关基因在非酒精性脂肪肝表型和免疫系统浸润中的特定作用。此外，还利用杯突相关基因预测了非酒精性脂肪肝的集群。
(A) Reverse cumulative distribution of |residual| Model — RF — xgb — sVm — glm
(A) |剩余|的反向累积分布 模型 - RF - xgb - sVm - glm

©

(B)

(D)
FIGURE 7 Construction and evaluation of RF, SVM, GLM and XGB machine models. (A) Cumulative residual distribution of each machine learning model. (B) Boxplots showed the residuals of each machine learning model. The red dot represents the root mean square of residuals (RMSE). © The important features in RF, SVM, GLM and XGB machine models. (D) ROC analysis of four machine learning models based on 5 -fold cross-validation in the testing cohort.
图 7 RF、SVM、GLM 和 XGB 机器模型的构建和评估。(A) 各机器学习模型的累积残差分布。(B) 方框图显示了每个机器学习模型的残差。红点代表残差均方根（RMSE）。RF、SVM、GLM 和 XGB 机器模型的重要特征。(D) 基于测试队列中 5 倍交叉验证的四种机器学习模型的 ROC 分析。
This study found 19 genes associated with cuproptosis (CRGs), including NFE2L2, NLRP3, ATP7B, ATP7A, SLC31A1, FDX1, LIAS, LIPT1, LIPT2, DLD, DLAT, PDHA1, PDHB, MTF1, GLS, CDKN2A, DBT, GCSH and DLST. First, the expression profile of cuproptosisrelated genes in the liver of normal people and NAFLD patients was explored. Some cuproptosis genes were significantly differentially expressed in NAFLD patients compared with normal
本研究发现了 19 个与杯突症相关的基因（CRGs），包括 NFE2L2、NLRP3、ATP7B、ATP7A、SLC31A1、FDX1、LIAS、LIPT1、LIPT2、DLD、DLAT、PDHA1、PDHB、MTF1、GLS、CDKN2A、DBT、GCSH 和 DLST。首先，研究了杯突症相关基因在正常人和非酒精性脂肪肝患者肝脏中的表达谱。与正常人相比，非酒精性脂肪肝患者肝脏中一些杯突相关基因的表达存在明显差异。
people (NFE2L2, ATP7B, SLC31A1, LIAS, DLD, PDHA1, PDHB, MTF1, DBT), suggesting that these cuproptosis-related genes play an important role in the development of NAFLD. The correlation analysis results of these genes were performed, strong synergistic effect between LIAS and PDHA1, while MTF1 and LIAS showed significant antagonism. The results of the immune infiltration analysis revealed increases in γ γ gamma\gamma - The levels of Delta T T TT
NFE2L2、ATP7B、SLC31A1、LIAS、DLD、PDHA1、PDHB、MTF1、DBT），表明这些杯突相关基因在非酒精性脂肪肝的发病中起着重要作用。对这些基因的相关性分析结果表明，LIAS 与 PDHA1 之间存在较强的协同作用，而 MTF1 与 LIAS 之间存在明显的拮抗作用。免疫浸润分析结果显示， γ γ gamma\gamma - Delta T T TT 水平升高。
FIGURE 8 Validation of correlation analysis based on the GSE48452 dataset. (A) Dataset GSE48452 was used to validate the expression of SLC16A1, FCAMR, RAB26, ENO3 and LEPR, the results of which were presented as box plots. ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* * *} p<0.001, ns, no significance. (B) The expression of SLC16A1, ENO3 and LEPR were presented in the heatmap. (C-E) The diagnostic effectiveness of the biomarkers for NAFLD by ROC analysis.
图 8 基于 GSE48452 数据集的相关分析验证。(A) 数据集 GSE48452 用于验证 SLC16A1、FCAMR、RAB26、ENO3 和 LEPR 的表达，结果以方框图表示。 ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* * *} p<0.001 ，ns，无显著性。(B）SLC16A1、ENO3 和 LEPR 的表达以热图形式显示。(C-E)通过 ROC 分析显示生物标志物对非酒精性脂肪肝的诊断效果。
cells, M1 macrophages, M2 macrophages, resting dendritic cells and resting mast cells in the liver tissue of NAFLD patients, suggesting that changes in the immune system may be the main cause of NAFLD. Previous studies have also shown that macrophages participate in the inflammatory response of NASH and have a relatively unique phenotype and function. 28 , 29 28 , 29 ^(28,29){ }^{28,29} M1 macrophages present antigens and secrete pro-inflammatory cytokines, while M2 macrophages secrete inhibitory cytokines IL-10 and transforming growth factor- β β beta\beta (TGF- β β beta\beta ), and mannose receptor (Mrc) to down-regulate the immune response. 30 , 31 30 , 31 ^(30,31){ }^{30,31} Chronic inflammation, cancer and NAFLD are mostly affected by abnormal regulation of M1/M2 macrophages. 32 32 ^(32){ }^{32} The correlation analysis results demonstrated that activated CD4 memory T cells, M2 macrophages and activated mast cells were all associated with cuproptosis-related genes (Figure 2H). These results suggest that CRGs may be key factors regulating disease development and immune infiltration status in NAFLD patients. In addition, a consensus clustering algorithm was applied to group 39 NAFLD samples according to the expression profiles of 9 significantly differentially expressed CRGs to illustrate the distinct regulatory patterns of cuproptosis gene clusters in NAFLD patients. In this process, two different cuproptosis-related clusters were determined. cuproptosis Cluster2 was characterized by the enhanced expression of LIAS and PDHA1, and the two clusters also affected the immune microenvironment. γ γ gamma\gamma-Delta T cells accounted for a high proportion
非酒精性脂肪肝患者肝组织中的细胞、M1 巨噬细胞、M2 巨噬细胞、静止树突状细胞和静止肥大细胞，表明免疫系统的变化可能是导致非酒精性脂肪肝的主要原因。以往的研究还表明，巨噬细胞参与了非酒精性脂肪肝的炎症反应，并具有相对独特的表型和功能。 28 , 29 28 , 29 ^(28,29){ }^{28,29} M1 巨噬细胞呈现抗原并分泌促炎细胞因子，而 M2 巨噬细胞则分泌抑制性细胞因子 IL-10 和转化生长因子- β β beta\beta （TGF- β β beta\beta ）以及甘露糖受体（Mrc）来下调免疫反应。 30 , 31 30 , 31 ^(30,31){ }^{30,31} 慢性炎症、癌症和非酒精性脂肪肝大多受 M1/M2 巨噬细胞调节异常的影响。 32 32 ^(32){ }^{32} 相关性分析结果表明，活化的 CD4 记忆 T 细胞、M2 巨噬细胞和活化的肥大细胞都与杯突相关基因有关（图 2H）。这些结果表明，杯突相关基因可能是调节非酒精性脂肪肝患者疾病发展和免疫浸润状态的关键因素。此外，根据 9 个显著差异表达的 CRGs 的表达谱，应用共识聚类算法对 39 个非酒精性脂肪肝样本进行分组，以说明非酒精性脂肪肝患者杯突基因簇的不同调控模式。杯突相关基因簇 2 的特征是 LIAS 和 PDHA1 的表达增强，这两个基因簇也影响了免疫微环境。 γ γ gamma\gamma -Delta T 细胞占了很高的比例。
in Cluster2. According to previous studies, γ δ γ δ gamma delta\gamma \delta T17 cells aggravate the progress of NAFLD. 33 , 34 33 , 34 ^(33,34){ }^{33,34} Furthermore, Cluster2 was mostly enriched in the MAKP signalling pathway, the JAK-STAT signalling pathway and the galactose metabolism pathway. These two pathways have also been shown by previous researchers to be related to the activation and inflammation of the immune system, 35 , 36 35 , 36 ^(35,36){ }^{35,36} and may play a role in the progress of NAFLD.
2 组中。根据以往的研究， γ δ γ δ gamma delta\gamma \delta T17 细胞会加重非酒精性脂肪肝的进展。 33 , 34 33 , 34 ^(33,34){ }^{33,34} 此外，Cluster2 主要富含 MAKP 信号通路、JAK-STAT 信号通路和半乳糖代谢通路。以往的研究表明，这两条途径也与免疫系统的激活和炎症有关 35 , 36 35 , 36 ^(35,36){ }^{35,36} ，并可能在非酒精性脂肪肝的进展中发挥作用。
Machine learning methods 37 , 38 37 , 38 ^(37,38){ }^{37,38} were applied to investigate novel disease diagnostic markers, and the predictive performance of the four machine learning classifiers (RF, SVM, GLM and XGB) was compared according to the expression profiles of cluster-specific DEGs. The results revealed that the SVM-based predictive model had the highest predictive effect in the test cohort ( A ∪ C = 0.950 A ∪ C = 0.950 A uu C=0.950A \cup C=0.950 ), indicating that SVM-based machine learning could effectively predict NAFLD clusters. Subsequently, five important variables (SLC16A1, FCAMR, RAB26, ENO3 and LEPR) were selected to construct a 5-gene-based SVM model. Three key genes (SLC16A1, ENO3 and LEPR) were identified by further validation and screening using an external data set. SLC16A1 is a well-studied member of the SLC16A family. Research has shown that SLC16A1 is distributed in almost all human tissues and is overexpressed in many cancers. In addition, up-regulation of SLC16A1 is associated with poorer prognosis in various cancer types 39 , 40 39 , 40 ^(39,40){ }^{39,40}; membrane monocarboxylate transporter 1 (Membrane Monocarboxylate Transporter 1, SLC16A1/MCT1) plays an important role in hepatocyte homeostasis and drug action and is involved in liver
应用机器学习方法 37 , 38 37 , 38 ^(37,38){ }^{37,38} 研究新型疾病诊断标志物，并根据集群特异性 DEGs 的表达谱比较了四种机器学习分类器（RF、SVM、GLM 和 XGB）的预测性能。结果显示，基于 SVM 的预测模型在测试队列（ A ∪ C = 0.950 A ∪ C = 0.950 A uu C=0.950A \cup C=0.950 ）中的预测效果最高，表明基于 SVM 的机器学习能有效预测非酒精性脂肪肝集群。随后，研究人员选择了五个重要变量（SLC16A1、FCAMR、RAB26、ENO3 和 LEPR）来构建基于 5 个基因的 SVM 模型。通过外部数据集的进一步验证和筛选，确定了三个关键基因（SLC16A1、ENO3 和 LEPR）。SLC16A1 是 SLC16A 家族中研究较多的一个成员。研究表明，SLC16A1 几乎分布在所有人体组织中，并在许多癌症中过度表达。此外，SLC16A1 的上调与各种癌症类型的预后较差有关 39 , 40 39 , 40 ^(39,40){ }^{39,40} ；膜单羧酸盐转运体 1（Membrane Monocarboxylate Transporter 1, SLC16A1/MCT1）在肝细胞稳态和药物作用中发挥重要作用，并参与肝脏
FIGURE 9 Validation of the model and the expression of SLC16A1, ENO3 and LEPR in HFD diet mice (A) Construction of a nomogram for predicting the risk of NAFLD clusters based on the SLC16A1, ENO3 and LEPR. (B, C) Construction of calibration curve (B) and DCA © for assessing the predictive efficiency of the nomogram model. (D) Haematoxylin and eosin (H&E) staining of liver slices. (E) Hepatic triglyceride (TG) levels. (F) Hepatic cholesterol (TC) levels. (G) Genes mRNA expression of SLC16A1, ENO3 and LEPR in hepatic Values are shown as the mean ± ± +-\pm s.d. ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(****)p < 0.01,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* *} p<0.01,{ }^{* * *} p<0.001 versus ND.
图 9 模型的验证以及 SLC16A1、ENO3 和 LEPR 在高密度脂蛋白饮食小鼠中的表达 (A) 根据 SLC16A1、ENO3 和 LEPR 构建预测非酒精性脂肪肝风险的提名图。(B、C）构建校准曲线（B）和 DCA © 以评估提名图模型的预测效率。(D) 肝切片的血色素和伊红（H&E）染色。(E) 肝甘油三酯（TG）水平。(F) 肝胆固醇（TC）水平。(G) 肝脏中 SLC16A1、ENO3 和 LEPR 的基因 mRNA 表达值以平均值 ± ± +-\pm s.d. ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ∗ p < 0.05 , ∗ ∗ p < 0.01 , ∗ ∗ ∗ p < 0.001 ^(**)p < 0.05,^(****)p < 0.01,^(******)p < 0.001{ }^{*} p<0.05,{ }^{* *} p<0.01,{ }^{* * *} p<0.001 与 ND 比较。
FIGURE 10 Model gene mRNA expression levels were verified in the AML12 cells. β β beta\beta-Actin was controlled. (A) Oil red O staining of cells. Magnification × 100 × 100 xx100\times 100. (B-D) Relative mRNA levels of (B) SLC16A1, © ENO3 and (D) LEPR. Relative mRNA levels were normalized to those of β β beta\beta-Actin. Values are shown as the mean ± ± +-\pm s.d. ∗ p < 0.05 ; ∗ ∗ p < 0.01 ∗ p < 0.05 ; ∗ ∗ p < 0.01 ^(**)p < 0.05;^(****)p < 0.01{ }^{*} p<0.05 ;{ }^{* *} p<0.01, ∗ ∗ ∗ p < 0.001 ∗ ∗ ∗ p < 0.001 ^(******)p < 0.001{ }^{* * *} p<0.001.
图 10 模型基因 mRNA 表达水平在 AML12 细胞中得到验证。 β β beta\beta -Actin 受控。(A) 细胞的油红 O 染色。放大 × 100 × 100 xx100\times 100 。 (B-D) (B) SLC16A1、© ENO3 和 (D) LEPR 的相对 mRNA 水平。相对 mRNA 水平与 β β beta\beta -Actin 的水平进行了归一化。数值显示为平均值 ± ± +-\pm s.d. ∗ p < 0.05 ; ∗ ∗ p < 0.01 ∗ p < 0.05 ; ∗ ∗ p < 0.01 ^(**)p < 0.05;^(****)p < 0.01{ }^{*} p<0.05 ;{ }^{* *} p<0.01 , ∗ ∗ ∗ p < 0.001 ∗ ∗ ∗ p < 0.001 ^(******)p < 0.001{ }^{* * *} p<0.001 。

pathology. 41 41 ^(41){ }^{41} Enolase 3 (ENO3) encodes enolase β β beta\beta subunits, which are distributed in various tissues, including the liver, lung, bone and heart, and has been proven to accelerate the accumulation of hepatic cholesterol ester caused by cholesterol ester synthesis. 42 42 ^(42){ }^{42} Moreover, the
病理学。 41 41 ^(41){ }^{41} 烯醇化酶 3（ENO3）编码烯醇化酶 β β beta\beta 亚基，分布于肝、肺、骨和心脏等多种组织中，已被证实可加速胆固醇酯合成引起的肝脏胆固醇酯积累。 42 42 ^(42){ }^{42} 而且
up-regulation of ENO3 in the liver may promote the progress of NASH by increasing GPX4 expression and the negative regulation of ferroptosis by lipid accumulation. 43 43 ^(43){ }^{43} Leptin receptor (LEPR) is a single transmembrane domain receptor in the cytokine receptor family. Leptin
肝脏中 ENO3 的上调可能会通过增加 GPX4 的表达和脂质积累对铁氧化的负调控促进 NASH 的进展。 43 43 ^(43){ }^{43} 瘦素受体（LEPR）是细胞因子受体家族中的一种单跨膜结构域受体。瘦素
controls satiety and maintains the balance of body energy by binding with it. 44 44 ^(44){ }^{44} Relevant studies have reported that the obesity gene LEPR is related to NAFLD. 45 , 46 45 , 46 ^(45,46){ }^{45,46} In NAFLD patients, LEPR polymorphism was found to be associated with obesity parameters, insulin resistance and blood glucose levels. 47 47 ^(47){ }^{47} Furthermore, the polymorphism of rs1137100 and rs1137101 in the LEPR locus has been related to an increased risk of NAFLD and NASH. Rs1137100 has also been proven to be related to simple steatosis and NASH. 48 48 ^(48){ }^{48}
控制饱腹感，并通过与之结合维持体内能量平衡。 44 44 ^(44){ }^{44} 相关研究报道，肥胖基因 LEPR 与非酒精性脂肪肝有关。 45 , 46 45 , 46 ^(45,46){ }^{45,46} 在非酒精性脂肪肝患者中，发现 LEPR 多态性与肥胖参数、胰岛素抵抗和血糖水平有关。 47 47 ^(47){ }^{47} 此外，LEPR 基因座中的 rs1137100 和 rs1137101 多态性与非酒精性脂肪肝和 NASH 风险增加有关。Rs1137100 也被证明与单纯性脂肪变性和 NASH 有关。 48 48 ^(48){ }^{48}
Finally, SLC16A1, ENO3 and LEPR were used to build a nomogram model to diagnose the NAFLD cluster. The model was found to have significant predictive power, suggesting its clinical utility. In addition, a NAFLD disease model was constructed with C57BL/6J mice and AML12 cells, and the gene expression was verified by RTPCR. In summary, SLC16A1, ENO3 and LEPR are satisfactory indicators to evaluate the pathological results of NAFLD cluster and NAFLD patients. Nevertheless, the limitations of the current study should be acknowledged. More detailed clinical characteristics are needed to validate the performance of predictive models. A greater number of NAFLD samples is required to elucidate the accuracy of the clustering associated with cuproptosis. Finally, the potential correlation between CRG and immune response requires further investigation.
最后，利用 SLC16A1、ENO3 和 LEPR 建立了一个诊断非酒精性脂肪肝集群的提名图模型。结果发现，该模型具有显著的预测能力，表明其具有临床实用性。此外，还利用 C57BL/6J 小鼠和 AML12 细胞构建了非酒精性脂肪肝疾病模型，并通过 RTPCR 验证了基因的表达。总之，SLC16A1、ENO3 和 LEPR 是评估非酒精性脂肪肝集群和非酒精性脂肪肝患者病理结果的理想指标。然而，本研究的局限性也应得到承认。需要更详细的临床特征来验证预测模型的性能。需要更多的非酒精性脂肪肝样本来阐明与杯突症相关的聚类的准确性。最后，还需要进一步研究 CRG 与免疫反应之间的潜在相关性。

5 | CONCLUSION  5 结论

This study explored the correlation between CRGs and infiltrating immune cells and demonstrated the heterogeneity of infiltrating immune cells among NAFLD patients with different cuproptosis gene subgroups. Three characteristic genes (SLC16A1, ENO3 and LEPR) related to cuproptosis were identified. The present study confirmed the role of cuproptosis in NAFLD, further elucidated the underlying molecular mechanisms leading to NAFLD heterogeneity and provided new ideas for the development of new targets for immunotherapy in NAFLD patients.
本研究探讨了 CRGs 与浸润性免疫细胞之间的相关性，并证明了不同杯突基因亚组的非酒精性脂肪肝患者浸润性免疫细胞的异质性。研究发现了三个与杯突症相关的特征基因（SLC16A1、ENO3 和 LEPR）。本研究证实了杯突症在非酒精性脂肪肝中的作用，进一步阐明了导致非酒精性脂肪肝异质性的潜在分子机制，并为非酒精性脂肪肝患者免疫疗法新靶点的开发提供了新思路。

AUTHOR CONTRIBUTIONS  作者贡献

Changxu Liu: Visualization (equal); writing - original draft (equal); writing - review and editing (equal). Zhihao Fang: Visualization (equal); writing - original draft (equal). Kai Yang: Software (equal); validation (equal). Yanchao Ji: Data curation (equal); formal analysis (equal). Xiaoxiao Yu: Data curation (equal); resources (equal). ZiHao Guo: Investigation (equal); resources (equal). Zhichao Dong: Formal analysis (equal); investigation (equal). Tong Zhu: Investigation (equal); methodology (equal). Chang Liu: Funding acquisition (equal); methodology (equal); supervision (equal).
刘昌旭：可视化（相同）；写作--原稿（相同）；写作--审阅和编辑（相同）。方志浩视觉效果（相同）；写作--原稿（相同）。杨凯软件（相同）；验证（相同）。纪彦超数据整理（等同）；形式分析（等同）。Xiaoxiao Yu：数据整理（相同）；资源（相同）。郭子豪：调查（相同）；资源（相同）。董志超：形式分析（相同）；调查（相同）。Tong Zhu：调查（相同）；方法（相同）。Chang Liu：资金获取（等额）；方法（等额）；监督（等额）。

ACKNOWLEDGEMENTS  致谢

This work was supported by the Innovative Scientific Research Fund of Harbin Medical University grant number: 2021-KYYWF-0260.
本研究由哈尔滨医科大学创新科研基金资助，基金号：2021-KYYWF-0260。

CONFLICT OF INTEREST STATEMENT
利益冲突声明

The authors declare that they have no competing interests.
作者声明他们没有利益冲突。

DATA AVAILABILITY STATEMENT
数据可用性声明

The datasets presented in this study can be found in online repositories. The names of the repositories and accession numbers can be found below: http://www.ncbi.nlm.nih.gov/geo/.
本研究中介绍的数据集可在在线资源库中找到。资源库名称和登录号如下：http://www.ncbi.nlm.nih.gov/geo/。

ORCID

REFERENCES  参考文献

  1. Byrne CD, Targher G. NAFLD: a multisystem disease. J Hepatol. 2015;62(1 Suppl):S47-S64.
    Byrne CD, Targher G. NAFLD：一种多系统疾病。J Hepatol.2015;62(1 Suppl):S47-S64.
  2. Younossi Z, Anstee QM, Marietti M, et al. Global burden of NAFLD and NASH: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol. 2018;15(1):11-20.
    Younossi Z, Anstee QM, Marietti M, et al. 非酒精性脂肪肝和非酒精性肝病的全球负担：趋势、预测、风险因素和预防。Nat Rev Gastroenterol Hepatol.2018;15(1):11-20.
  3. Kabbany MN, Selvakumar PKC, Watt K, et al. Prevalence of nonalcoholic Steatohepatitis-associated cirrhosis in the United States: an analysis of National Health and nutrition examination survey data. Am J Gastroenterol. 2017;112(4):581-587.
    Kabbany MN、Selvakumar PKC、Watt K 等：《美国非酒精性脂肪性肝炎相关肝硬化患病率：全国健康与营养状况调查数据分析》（Prevalence of nonalcoholic Steatohepatitis-associated cirrhosis in the United States: an analysis of National Health and nutrition examination survey data）。Am J Gastroenterol.2017;112(4):581-587.
  4. Ullah R, Rauf N, Nabi G, et al. Role of nutrition in the pathogenesis and prevention of non-alcoholic fatty liver disease: recent updates. Int J Biol Sci. 2019;15(2):265-276.
    Ullah R, Rauf N, Nabi G, et al. 营养在非酒精性脂肪肝发病机制和预防中的作用：最新进展。Int J Biol Sci.
  5. Wong VW-S, Chitturi S, Wong GLH, Yu J, Chan HLY, Farrell GC. Pathogenesis and novel treatment options for non-alcoholic steatohepatitis. Lancet Gastroenterol Hepatol. 2016;1(1):56-67.
    Wong VW-S、Chitturi S、Wong GLH、Yu J、Chan HLY、Farrell GC。非酒精性脂肪性肝炎的发病机制和新型治疗方案。Lancet Gastroenterol Hepatol.2016;1(1):56-67.
  6. García-Ruiz C, Fernández-Checa JC. Mitochondrial oxidative stress and antioxidants balance in fatty liver disease. Hepatol Commun. 2018;2(12):1425-1439.
    García-Ruiz C, Fernández-Checa JC.脂肪肝的线粒体氧化应激和抗氧化剂平衡。Hepatol Commun.2018;2(12):1425-1439.
  7. Gonzalez A, Huerta-Salgado C, Orozco-Aguilar J, et al. Role of oxidative stress in hepatic and extrahepatic dysfunctions during nonalcoholic fatty liver disease (NAFLD). Oxid Med Cell Longev. 2020;2020:1617805.
    Gonzalez A, Huerta-Salgado C, Orozco-Aguilar J, et al. 氧化应激在非酒精性脂肪肝肝内外功能障碍中的作用。Oxid Med Cell Longev.2020;2020:1617805.
  8. Baker ZN, Cobine PA, Leary SC. The mitochondrion: a central architect of copper homeostasis. Metallomics. 2017;9(11):1501-1512.
    Baker ZN, Cobine PA, Leary SC.线粒体：铜平衡的核心结构。Metallomics.2017;9(11):1501-1512.
  9. Soto IC, Fontanesi F, Liu J, Barrientos A. Biogenesis and assembly of eukaryotic cytochrome c oxidase catalytic core. Biochim Biophys Acta. 2012;1817(6):883-897.
    Soto IC, Fontanesi F, Liu J, Barrientos A. 真核细胞色素 c 氧化酶催化核心的生物发生和组装。Biochim Biophys Acta.2012;1817(6):883-897.
  10. Dennerlein S, Rehling P. Human mitochondrial COX1 assembly into cytochrome c oxidase at a glance. J Cell Sci. 2015;128(5):833-837.
    Dennerlein S, Rehling P. 人类线粒体 COX1 组装成细胞色素 c 氧化酶一览。J Cell Sci. 2015;128(5):833-837.
  11. Vanden Berghe T, Linkermann A, Jouan-Lanhouet S, et al. Regulated necrosis: the expanding network of non-apoptotic cell death pathways. Nat Rev Mol Cell Biol. 2014;15(2):135-147.
    Vanden Berghe T, Linkermann A, Jouan-Lanhouet S, et al. 调节性坏死：非凋亡细胞死亡途径网络的扩展。Nat Rev Mol Cell Biol. 2014;15(2):135-147.
  12. Wang Y, Zhang L, Zhou F. Cuproptosis: a new form of programmed cell death. Cell Mol Immunol. 2022;19(8):867-868.
    Wang Y, Zhang L, Zhou F. Cuproptosis: a new form of programmed cell death.Cell Mol Immunol.2022;19(8):867-868.
  13. Ma C, Han L, Zhu Z, Heng Pang C, Pan G. Mineral metabolism and ferroptosis in non-alcoholic fatty liver diseases. Biochem Pharmacol. 2022;205:115242.
    Ma C, Han L, Zhu Z, Heng Pang C, Pan G. Mineral metabolism and ferroptosis in non-alcoholic fatty liver diseases.Biochem Pharmacol.2022;205:115242.
  14. Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics (Oxford, England). 2012;28(6):882-883.
    Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD.用于消除高通量实验中的批次效应和其他不必要变异的 sva 软件包。生物信息学》（英国牛津）。2012;28(6):882-883.
  15. Ritchie ME, Phipson B, Wu D, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47.
    Ritchie ME, Phipson B, Wu D, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies.Nucleic Acids Res. 2015;43(7):e47.
  16. Tsvetkov P, Coy S, Petrova B, et al. Copper induces cell death by targeting lipoylated TCA cycle proteins. Science (New York, N.Y.). 2022;375(6586):1254-1261.
    Tsvetkov P, Coy S, Petrova B, et al. 铜通过靶向脂酰化 TCA 循环蛋白诱导细胞死亡。科学》（纽约州纽约市）。2022;375(6586):1254-1261.
  17. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics (Oxford, England). 2010;26(12):1572-1573.
    Wilkerson MD, Hayes DN.ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking.生物信息学（英国牛津）。2010;26(12):1572-1573.
  18. Rigatti SJ. Random forest. J Insur Med (New York, N.Y.). 2017;47(1):31-39.
    Rigatti SJ，随机森林。J Insur Med (New York, N.Y.).2017;47(1):31-39.
  19. Tan M, Pu J, Zheng B. Optimization of breast mass classification using sequential forward floating selection (SFFS) and a
    Tan M，Pu J，Zheng B. 使用顺序前向浮动选择 (SFFS) 和乳房肿块分类的优化
    support vector machine (SVM) model. Int J Comput Assist Radiol Surg. 2014;9(6):1005-1020.
    支持向量机（SVM）模型。Int J Comput Assist Radiol Surg. 2014;9(6):1005-1020.
  20. Lindsey JK. A review of some extensions to generalized linear models. Stat Med. 1999;18(17-18):2223-2236.
    Lindsey JK.广义线性模型若干扩展的回顾。Stat Med.1999;18(17-18):2223-2236.
  21. Chen T, He T, Benesty M, et al. Xgboost: Extreme Gradient Boosting. R package version 0.4-2. 1(4): p. 1-4 2015.
    Chen T, He T, Benesty M, et al：Extreme Gradient Boosting.1(4): p. 1-4 2015.
  22. Chu MJJ, Hickey AJR, Tagaloa S, et al. Ob/Ob mouse livers show decreased oxidative phosphorylation efficiencies and anaerobic capacities after cold ischemia. PloS One. 2014;9(6):e100609.
    Chu MJJ, Hickey AJR, Tagaloa S, et al. Ob/Ob 小鼠肝脏在冷缺血后显示出氧化磷酸化效率和无氧能力下降。PloS One.2014;9(6):e100609.
  23. Aigner E, Strasser M, Haufe H, et al. A role for low hepatic copper concentrations in nonalcoholic fatty liver disease. Am J Gastroenterol. 2010;105(9):1978-1985.
    Aigner E、Strasser M、Haufe H 等：低肝铜浓度在非酒精性脂肪肝中的作用。Am J Gastroenterol.2010;105(9):1978-1985.
  24. Lan Y, Wu S, Wang Y, et al. Association between blood copper and nonalcoholic fatty liver disease according to sex. Clin Nutr 2021;40(4):2045-2052.
    Lan Y，Wu S，Wang Y，et al.Clin Nutr 2021;40(4):2045-2052.
  25. Begriche K, Massart J, Robin MA, et al. Mitochondrial adaptations and dysfunctions in nonalcoholic fatty liver disease. Hepatology (Baltimore, Md.). 2013;58(4):1497-1507.
    Begriche K, Massart J, Robin MA, et al. 非酒精性脂肪肝的线粒体适应性和功能障碍。肝病学》（巴尔的摩，马里兰州）。2013;58(4):1497-1507.
  26. Antonucci L, Porcu C, Iannucci G, Balsano C, Barbaro B. Nonalcoholic fatty liver disease and nutritional implications: special focus on copper. Nutrients. 2017;9(10):1137.
    Antonucci L、Porcu C、Iannucci G、Balsano C、Barbaro B.非酒精性脂肪肝和营养影响：特别关注铜。营养素。2017;9(10):1137.
  27. Bala S. The pro-inflammatory effects of miR-155 promote liver fibrosis and alcohol-induced steatohepatitis. J Hepatol. 2016;64(6):1378-1387.
    Bala S. miR-155 的促炎作用促进肝纤维化和酒精诱导的脂肪性肝炎。J Hepatol.2016;64(6):1378-1387.
  28. Bala S, Csak T, Saha B, et al. The pro-inflammatory effects of miR155 promote liver fibrosis and alcohol-induced steatohepatitis J Hepatol. 2016;64(6):1378-1387.
    Bala S, Csak T, Saha B, et al. miR155 的促炎作用促进肝纤维化和酒精诱导的脂肪性肝炎 J Hepatol.2016;64(6):1378-1387.
  29. Kazankov K, Jørgensen SMD, Thomsen KL, et al. The role of macrophages in nonalcoholic fatty liver disease and nonalcoholic steato hepatitis. Nat Rev Gastroenterol Hepatol. 2019;16(3):145-159.
    Kazankov K, Jørgensen SMD, Thomsen KL, et al. 巨噬细胞在非酒精性脂肪肝和非酒精性脂肪性肝炎中的作用。Nat Rev Gastroenterol Hepatol.2019;16(3):145-159.
  30. Tacke F, Zimmermann HW. Macrophage heterogeneity in liver injury and fibrosis. J Hepatol. 2014;60(5):1090-1096
    Tacke F, Zimmermann HW.肝损伤和肝纤维化中巨噬细胞的异质性。J Hepatol.2014;60(5):1090-1096
  31. Lawrence T, Natoli G. Transcriptional regulation of macrophage polarization: enabling diversity with identity. Nat Rev Immunol. 2011;11(11):750-761
    Lawrence T，Natoli G. 巨噬细胞极化的转录调控：实现多样性与同一性。Nat Rev Immunol.2011;11(11):750-761
  32. Sica A, Mantovani A. Macrophage plasticity and polarization in vivo veritas. J Clin Invest. 2012;122(3):787-795.
    Sica A, Mantovani A.体内巨噬细胞的可塑性和极化。J Clin Invest.2012;122(3):787-795.
  33. Zhou D, Pan Q, Shen F, et al. Total fecal microbiota transplantation alleviates high-fat diet-induced steatohepatitis in mice via beneficial regulation of gut microbiota. Sci Rep. 2017;7(1):1529.
    Zhou D, Pan Q, Shen F, et al.粪便微生物菌群移植通过有益调节肠道微生物菌群缓解高脂饮食诱导的小鼠脂肪性肝炎。Sci Rep. 2017;7(1):1529.
  34. Castaño-Rodríguez N, Mitchell HM, Kaakoush NO. NAFLD, helicobacter species and the intestinal microbiome. Best Pract Res Clin Gastroenterol. 2017;31(6):657-668.
    Castaño-Rodríguez N、Mitchell HM、Kaakoush NO。非酒精性脂肪肝、螺旋杆菌和肠道微生物组。Best Pract Res Clin Gastroenterol.2017;31(6):657-668.
  35. Wang F, Zhang X, Liu W, et al. Activated natural killer cell promotes nonalcoholic Steatohepatitis through mediating JAK/STAT path way. Cell Mol Gastroenterol Hepatol. 2022;13(1):257-274
    激活的自然杀伤细胞通过介导 JAK/STAT 途径促进非酒精性脂肪性肝炎。Cell Mol Gastroenterol Hepatol.2022;13(1):257-274
  36. Ding J, Wu L, Zhu G, Zhu J, Luo P, Li Y. HADHA alleviates hepatic steatosis and oxidative stress in NAFLD via inactivation of the MKK3/MAPK pathway. Mol Biol Rep. 2022;50:961-970.
    Ding J, Wu L, Zhu G, Zhu J, Luo P, Li Y.HADHA 通过灭活 MKK3/MAPK 通路减轻非酒精性脂肪肝的肝脏脂肪变性和氧化应激。2022; 50:961-970.
  37. Deo RC. Machine learning in medicine. Circulation 2015;132(20):1920-1930.
    Deo RC.医学中的机器学习。Circulation 2015;132(20):1920-1930.
  38. Le NQK, Kha QH, Nguyen VH, et al. Machine learning-based Radiomics signatures for EGFR and KRAS mutations prediction in non-small-cell lung cancer. Int J Mol Sci. 2021;22(17):9254.
    Le NQK, Kha QH, Nguyen VH, et al. 基于机器学习的非小细胞肺癌 EGFR 和 KRAS 突变预测放射组学特征。2021;22(17):9254.
  39. Latif A, Chadwick AL, Kitson SJ, et al. Monocarboxylate transporter 1 (MCT1) is an independent prognostic biomarker in endometrial cancer. BMC Clin Pathol. 2017;17:27.
    Latif A、Chadwick AL、Kitson SJ 等：单羧酸盐转运体 1 (MCT1) 是子宫内膜癌的独立预后生物标志物。BMC Clin Pathol.2017;17:27.
  40. Felmlee MA, Jones RS, Rodriguez-Cruz V, Follman KE, Morris ME. Monocarboxylate transporters (SLC16): function, regulation, and role in health and disease. Pharmacol Rev. 2020;72(2):466-485.
    Felmlee MA、Jones RS、Rodriguez-Cruz V、Follman KE、Morris ME。单羧酸盐转运体（SLC16）：功能、调节以及在健康和疾病中的作用。Pharmacol Rev. 2020;72(2):466-485.
  41. Droździk M, Szeląg-Pieniek S, Grzegółkowska J, et al Monocarboxylate transporter 1 (MCT1) in liver pathology. Int J Mol Sci. 2020;21(5):1606.
    Droździk M, Szeląg-Pieniek S, Grzegółkowska J, et al 单羧酸盐转运体 1 (MCT1) 在肝脏病理学中的作用。2020;21(5):1606.
  42. Wu J, Zhou D, Deng C, Wu X, Long L, Xiong Y. Characterization of porcine ENO3: genomic and cDNA structure, polymorphism and expression. Genet Sel Evol. 2008;40(5):563-579.
    Wu J, Zhou D, Deng C, Wu X, Long L, Xiong Y.猪 ENO3 的特征：基因组和 cDNA 结构、多态性和表达。Genet Sel Evol.
  43. Lu D, Xia Q, Yang Z, et al. ENO3 promoted the progression of NASH by negatively regulating ferroptosis via elevation of GPX4 expression and lipid accumulation. Ann Transl Med. 2021;9(8):661.
    Lu D, Xia Q, Yang Z, et al. ENO3 通过提高 GPX4 表达和脂质积累负调控铁氧化酶促进 NASH 的进展。Ann Transl Med.2021;9(8):661.
  44. Zastrow O, Seidel B, Kiess W, et al. The soluble leptin receptor is crucial for leptin action: evidence from clinical and experimental data. Int J Obes Relat Metab Disord. 2003;27(12):1472-1478.
    Zastrow O, Seidel B, Kiess W, et al. 可溶性瘦素受体对瘦素作用至关重要：来自临床和实验数据的证据。Int J Obes Relat Metab Disord.2003;27(12):1472-1478.
  45. Lu H, Sun J, Sun L, Shu X, Xu Y, Xie D. Polymorphism of human leptin receptor gene is associated with type 2 diabetic patients complicated with non-alcoholic fatty liver disease in China. J Gastroenterol Hepatol. 2009;24(2):228-232.
    Lu H, Sun J, Sun L, Shu X, Xu Y, Xie D. 人类瘦素受体基因的多态性与中国 2 型糖尿病患者并发非酒精性脂肪肝的相关性。J Gastroenterol Hepatol.2009;24(2):228-232.
  46. Swellam M, Hamdy N. Association of nonalcoholic fatty liver disease with a single nucleotide polymorphism on the gene encoding leptin receptor. IUBMB Life. 2012;64(2):180-186
    Swellam M, Hamdy N. 非酒精性脂肪肝与瘦素受体编码基因单核苷酸多态性的关系。IUBMB Life.2012;64(2):180-186
  47. Aller R, de Luis DA, Izaola O, et al. Lys656Asn polymorphism of leptin receptor, leptin levels and insulin resistance in patients with non alcoholic fatty liver disease. Eur Rev Med Pharmacol Sci. 2012;16(3):335-341.
    Aller R, de Luis DA, Izaola O, et al. 瘦素受体的 Lys656Asn 多态性、瘦素水平和非酒精性脂肪肝患者的胰岛素抵抗。2012;16(3):335-341.
  48. Zain SM, Mohamed Z, Mahadeva S, et al. Impact of leptin recep tor gene variants on risk of non-alcoholic fatty liver disease and its interaction with adiponutrin gene. J Gastroenterol Hepatol. 2013;28(5):873-879.
    Zain SM、Mohamed Z、Mahadeva S 等. 瘦素受体 tor 基因变异对非酒精性脂肪肝风险的影响及其与脂肪素基因的相互作用。J Gastroenterol Hepatol.2013;28(5):873-879.

SUPPORTING INFORMATION  佐证资料

Additional supporting information can be found online in the Supporting Information section at the end of this article
其他辅助信息可在本文末尾的 "辅助信息 "部分在线查阅
How to cite this article: Liu C, Fang Z, Yang K, et al. Identification and validation of cuproptosis-related molecular clusters in non-alcoholic fatty liver disease. J Cell Mol Med. 2024;28:e18091. doi:10.1111/jcmm. 18091
本文引用方式 Liu C, Fang Z, Yang K, et al.J Cell Mol Med.2024;28:e18091. doi:10.1111/jcmm.18091
  1. Changxu Liu, Zhihao Fang, Tong Zhu and Chang Liu contributed equally to this work.
    刘昌旭、方志浩、朱彤和刘畅对本研究做出了同样的贡献。
    This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
    本文为开放存取文章，采用知识共享署名许可协议，允许在任何媒体上使用、传播和复制，但须适当引用原作。
    © 2024 The Authors. Journal of Cellular and Molecular Medicine published by Foundation for Cellular and Molecular Medicine and John Wiley & Sons Ltd.
    © 2024 作者。细胞与分子医学杂志》由细胞与分子医学基金会和 John Wiley & Sons Ltd 出版。