Comparison of three artificial rumen systems for rumen microbiome modeling

Abstract

Abstract Background The rumen contains a complex mixture of microbes, which are crucial for ruminant health and feed fermentation. During the fermentation process some of the feed-derived carbon becomes carbon dioxide and methane, which are released into the atmosphere where they act as greenhouse gases and contribute to climate change. There is growing interest in reducing the loss of feed-derived carbon and making it available to the animal, improving animal productivity, while also reducing the carbon footprint of the ruminant industry. To this end, artificial rumen systems (ARS) have been used for evaluating novel feed additives for their effect on the rumen microbiome and rumen function prior to conducting resource intensive animal trials. Whereas ARS are capable of predicting the response of the rumen and its microbiome, it is unclear how accurately different in vitro systems simulate the natural system and how results compare between the artificial systems that are being employed. Here we evaluated physical, chemical and microbiome metrics of three ARS over five days and compared them to those metrics in the in vivo rumen. Results Over a 48 hrs sampling period, the batch style platform (Ankom) was able to replicate pH, volatile fatty acid profile, and bacterial and fungal microbiome of the in vivo rumen, but its accuracy of mimicking in vivo metrics dropped significantly beyond 48 hrs. In contrast, the semi-continuous RUSITEC models, RUSITEC PP and RUSITEC prime, were able to mimic the volatile fatty acid profile and microbiota of the in vivo rumen for up to 120 hrs of rumen simulation. Comparison of gas production across vessel types demonstrated that the semi-continuous RUSITEC platforms display less variability among vessel replicates and time compared to the Ankom system. Conclusions In this study, we found that three widely used ARS were able to simulate the rumen ecosystem adequately for the first 48 hrs, with predictions from the more advanced semi-continuous ARS being more accurate when simulations extended over 48 hrs. Findings of this study will help to select the appropriate in vitro system for evaluating the response of the complex rumen microbiome to feed additives. Further work is necessary to improve the capabilities of these platforms and to standardize the methodology for large-scale application.