Paper reading (八十)：Persistent microbiome alterations modulate the rate of post-dieting weight regain

论文题目：Persistent microbiome alterations modulate the rate of post-dieting weight regain

scholar 引用：169

页数：23

发表时间：2016.12

发表刊物：nature

作者：Christoph A. Thaiss, Shlomik Itav,...,Eran Segal & Eran Elinav

摘要：

In tackling the obesity pandemic, considerable efforts are devoted to the development of effective weight reduction strategies, yet many dieting individuals fail to maintain a long-term weight reduction, and instead undergo excessive weight regain cycles. The mechanisms driving recurrent post-dieting obesity remain largely elusive. Here we identify an intestinal microbiome signature that persists after successful dieting of obese mice and contributes to faster weight regain and metabolic aberrations upon re-exposure to obesity-promoting conditions. Faecal transfer experiments show that the accelerated weight regain phenotype can be transmitted to germ-free mice. We develop a machine-learning algorithm that enables personalized microbiome-based prediction of the extent of post-dieting weight regain. Additionally, we find that the microbiome contributes to diminished post-dieting flavonoid levels and reduced energy expenditure, and demonstrate that flavonoid-based ‘post-biotic’ intervention ameliorates excessive secondary weight gain. Together, our data highlight a possible microbiome contribution to accelerated post-dieting weight regain, and suggest that microbiome-targeting approaches may help to diagnose and treat this common disorder.

正文组织架构：

1. Main

2. Enhanced weight regain after dieting

3. Persistence of post-dieting microbiome alterations

4. The post-dieting microbiome contributes to weight regain

5. Microbiota composition predicts weight regain

6. Metabolites contribute to post-dieting weight regain

7. Flavonoids modulate weight regain and UCP1 expression

8. Discussion

9. Methods

正文部分内容摘录：

1. Biological Problem: What biological problems have been solved in this paper?

• Classification of obesity history

2. Main discoveries: What is the main discoveries in this paper?

• Here we identify an intestinal microbiome signature that persists after successful dieting of obese mice and contributes to faster weight regain and metabolic aberrations upon re-exposure to obesity-promoting conditions.
• We develop a machine-learning algorithm that enables personalized microbiome-based prediction of the extent of post-dieting weight regain.
• Together, our data highlight a possible microbiome contribution to accelerated post-dieting weight regain, and suggest that microbiome-targeting approaches may help to diagnose and treat this common disorder.

3. ML(Machine Learning) Methods: What are the ML methods applied in this paper?

• We further devise a machine-learning algorithm that successfully predicts the personalized propensity for recurrent diet-induced obesity solely on the basis of microbiome composition and demonstrate that faecal microbiome transplantation (FMT) or metabolite-based treatment may ameliorate exacerbated post-dieting weight regain.
• Given the above causal connection between microbiome configuration and post-dieting weight regain, we asked whether the extent of recurrent weight gain could be computationally predicted for each individual mouse based on its microbiota composition at the post-dieting nadir period. 
• We therefore profiled the microbiota composition of 25 mice that had undergone post-obesity dieting until metabolic normality and 25 weight-matched NC controls
• We first devised a machine-learning algorithm, based solely on the microbiota composition, aimed at predicting a history of obesity or lack thereof
• Notably, the derived random forest classifier predicted obesity history nearly perfectly (AUC = 0.96)

4. ML Advantages: Why are these ML methods better than the traditional methods in these biological problems?

• Classification of obesity history. Mouse obesity history was predicted using Random Forest Classification (sklearn 0.15.2) with the features being the relative abundances of 16S OTUs as outputted by QIIME. Classification was made in leave-one-out cross-validation in which each mouse was classified as negative or positive for obesity history.