Abstract：
OBJECTIVE: Disruption of pancreatic islet function and glucose homeostasis can lead to the development of sustained hyperglycaemia, beta cell glucotoxicity and subsequently type 2 diabetes. In this study, we explored the effects of in vitro hyperglycaemic conditions on human pancreatic islet gene expression across 24 h in six pancreatic cell types: alpha; beta; gamma; delta; ductal; and acinar. We hypothesised that genes associated with hyperglycaemic conditions may be relevant to the onset and progression of diabetes.
METHODS: We exposed human pancreatic islets from two donors to low (2.8 mmol/l) and high (15.0 mmol/l) glucose concentrations over 24 h in vitro. To assess the transcriptome, we performed single-cell RNA-seq (scRNA-seq) at seven time points. We modelled time as both a discrete and continuous variable to determine momentary and longitudinal changes in transcription associated with islet time in culture or glucose exposure. Additionally, we integrated genomic features and genetic summary statistics to nominate candidate effector genes. For three of these genes, we functionally characterised the effect on insulin production and secretion using CRISPR interference to knock down gene expression in EndoC-βH1 cells, followed by a glucose-stimulated insulin secretion assay.
RESULTS: In the discrete time models, we identified 1344 genes associated with time and 668 genes associated with glucose exposure across all cell types and time points. In the continuous time models, we identified 1311 genes associated with time, 345 genes associated with glucose exposure and 418 genes associated with interaction effects between time and glucose across all cell types. By integrating these expression profiles with summary statistics from genetic association studies, we identified 2449 candidate effector genes for type 2 diabetes, HbA1c, random blood glucose and fasting blood glucose. Of these candidate effector genes, we showed that three (ERO1B, HNRNPA2B1 and RHOBTB3) exhibited an effect on glucose-stimulated insulin production and secretion in EndoC-βH1 cells.
CONCLUSIONS: The findings of our study provide an in-depth characterisation of the 24 h transcriptomic response of human pancreatic islets to glucose exposure at a single-cell resolution. By integrating differentially expressed genes with genetic signals for type 2 diabetes and glucose-related traits, we provide insights into the molecular mechanisms underlying glucose homeostasis. Finally, we provide functional evidence to support the role of three candidate effector genes in insulin secretion and production.
METHODS: The scRNA-seq data from the 24 h glucose exposure experiment performed in this study are available in the database of Genotypes and Phenotypes (dbGap; https://www.ncbi.nlm.nih.gov/gap/ ) with accession no. phs001188.v3.p1. Study metadata and summary statistics for the differential expression, gene set enrichment and candidate effector gene prediction analyses are available in the Zenodo data repository ( https://zenodo.org/ ) under accession number 11123248. The code used in this study is publicly available at https://github.com/CollinsLabBioComp/publication-islet_glucose_timecourse .
METHODS: We exposed human pancreatic islets from two donors to low (2.8 mmol/l) and high (15.0 mmol/l) glucose concentrations over 24 h in vitro. To assess the transcriptome, we performed single-cell RNA-seq (scRNA-seq) at seven time points. We modelled time as both a discrete and continuous variable to determine momentary and longitudinal changes in transcription associated with islet time in culture or glucose exposure. Additionally, we integrated genomic features and genetic summary statistics to nominate candidate effector genes. For three of these genes, we functionally characterised the effect on insulin production and secretion using CRISPR interference to knock down gene expression in EndoC-βH1 cells, followed by a glucose-stimulated insulin secretion assay.
RESULTS: In the discrete time models, we identified 1344 genes associated with time and 668 genes associated with glucose exposure across all cell types and time points. In the continuous time models, we identified 1311 genes associated with time, 345 genes associated with glucose exposure and 418 genes associated with interaction effects between time and glucose across all cell types. By integrating these expression profiles with summary statistics from genetic association studies, we identified 2449 candidate effector genes for type 2 diabetes, HbA1c, random blood glucose and fasting blood glucose. Of these candidate effector genes, we showed that three (ERO1B, HNRNPA2B1 and RHOBTB3) exhibited an effect on glucose-stimulated insulin production and secretion in EndoC-βH1 cells.
CONCLUSIONS: The findings of our study provide an in-depth characterisation of the 24 h transcriptomic response of human pancreatic islets to glucose exposure at a single-cell resolution. By integrating differentially expressed genes with genetic signals for type 2 diabetes and glucose-related traits, we provide insights into the molecular mechanisms underlying glucose homeostasis. Finally, we provide functional evidence to support the role of three candidate effector genes in insulin secretion and production.
METHODS: The scRNA-seq data from the 24 h glucose exposure experiment performed in this study are available in the database of Genotypes and Phenotypes (dbGap; https://www.ncbi.nlm.nih.gov/gap/ ) with accession no. phs001188.v3.p1. Study metadata and summary statistics for the differential expression, gene set enrichment and candidate effector gene prediction analyses are available in the Zenodo data repository ( https://zenodo.org/ ) under accession number 11123248. The code used in this study is publicly available at https://github.com/CollinsLabBioComp/publication-islet_glucose_timecourse .
摘要：
目的：胰岛功能和葡萄糖稳态的破坏可导致持续性高血糖的发展，β细胞葡萄糖毒性和随后的2型糖尿病。在这项研究中,我们探讨了体外高血糖条件对6种胰腺细胞类型24小时内人类胰岛基因表达的影响：α;β;γ;δ;导管;和腺泡。我们假设与高血糖相关的基因可能与糖尿病的发病和进展有关。
方法：我们在体外24小时内将来自两个供体的人胰岛暴露于低（2.8mmol/l）和高（15.0mmol/l）葡萄糖浓度。为了评估转录组，我们在7个时间点进行了单细胞RNA-seq(scRNA-seq).我们将时间建模为离散变量和连续变量，以确定与培养或葡萄糖暴露中胰岛时间相关的转录的瞬时和纵向变化。此外,我们整合了基因组特征和遗传汇总统计数据来提名候选效应基因.对于其中三个基因，我们使用CRISPR干扰敲低EndoC-βH1细胞中的基因表达对胰岛素产生和分泌的影响进行了功能表征，然后进行葡萄糖刺激的胰岛素分泌测定。
结果：在离散时间模型中，我们在所有细胞类型和时间点鉴定了1344个与时间相关的基因和668个与葡萄糖暴露相关的基因.在连续时间模型中，我们确定了1311个与时间相关的基因，在所有细胞类型中，345个与葡萄糖暴露相关的基因和418个与时间和葡萄糖之间的相互作用相关的基因。通过将这些表达谱与遗传关联研究的汇总统计数据相结合，我们确定了2449个2型糖尿病的候选效应基因，HbA1c,随机血糖和空腹血糖。在这些候选效应基因中，我们展示了三个(ERO1B，HNRNPA2B1和RHOBTB3）对EndoC-βH1细胞中葡萄糖刺激的胰岛素产生和分泌有影响。
结论：我们的研究结果提供了人胰岛在单细胞分辨率下对葡萄糖暴露的24小时转录组反应的深入表征。通过整合差异表达基因与2型糖尿病和葡萄糖相关性状的遗传信号，我们提供了对葡萄糖稳态的分子机制的见解。最后,我们提供了功能证据来支持三个候选效应基因在胰岛素分泌和产生中的作用.
方法：来自本研究中进行的24小时葡萄糖暴露实验的scRNA-seq数据可在基因型和表型数据库中获得（dbGap;https://www.ncbi.nlm.nih.gov/gap/），登录号：phs001188.V3.p1.研究差异表达的元数据和汇总统计，基因集富集和候选效应基因预测分析可在Zenodo数据存储库（https://zenodo.org/）中获得，登录号为11123248。本研究中使用的代码可在https://github.com/CollinsLabBioComp/publication-islet_葡萄糖_timecourse上公开获得。
方法：我们在体外24小时内将来自两个供体的人胰岛暴露于低（2.8mmol/l）和高（15.0mmol/l）葡萄糖浓度。为了评估转录组，我们在7个时间点进行了单细胞RNA-seq(scRNA-seq).我们将时间建模为离散变量和连续变量，以确定与培养或葡萄糖暴露中胰岛时间相关的转录的瞬时和纵向变化。此外,我们整合了基因组特征和遗传汇总统计数据来提名候选效应基因.对于其中三个基因，我们使用CRISPR干扰敲低EndoC-βH1细胞中的基因表达对胰岛素产生和分泌的影响进行了功能表征，然后进行葡萄糖刺激的胰岛素分泌测定。
结果：在离散时间模型中，我们在所有细胞类型和时间点鉴定了1344个与时间相关的基因和668个与葡萄糖暴露相关的基因.在连续时间模型中，我们确定了1311个与时间相关的基因，在所有细胞类型中，345个与葡萄糖暴露相关的基因和418个与时间和葡萄糖之间的相互作用相关的基因。通过将这些表达谱与遗传关联研究的汇总统计数据相结合，我们确定了2449个2型糖尿病的候选效应基因，HbA1c,随机血糖和空腹血糖。在这些候选效应基因中，我们展示了三个(ERO1B，HNRNPA2B1和RHOBTB3）对EndoC-βH1细胞中葡萄糖刺激的胰岛素产生和分泌有影响。
结论：我们的研究结果提供了人胰岛在单细胞分辨率下对葡萄糖暴露的24小时转录组反应的深入表征。通过整合差异表达基因与2型糖尿病和葡萄糖相关性状的遗传信号，我们提供了对葡萄糖稳态的分子机制的见解。最后,我们提供了功能证据来支持三个候选效应基因在胰岛素分泌和产生中的作用.
方法：来自本研究中进行的24小时葡萄糖暴露实验的scRNA-seq数据可在基因型和表型数据库中获得（dbGap;https://www.ncbi.nlm.nih.gov/gap/），登录号：phs001188.V3.p1.研究差异表达的元数据和汇总统计，基因集富集和候选效应基因预测分析可在Zenodo数据存储库（https://zenodo.org/）中获得，登录号为11123248。本研究中使用的代码可在https://github.com/CollinsLabBioComp/publication-islet_葡萄糖_timecourse上公开获得。
