Assessment of method-specific bias associated with RNA extractions for metatranscriptomics in three geographically distinct paddy soils with different origin of parent materials

Abstract:

[Objective] RNA extraction is considered the key for soil metatranscriptomics and it is expected that different methods would have generated distinct resolution of active soil microbiome. However, it remains largely unknown about the quantitative bias assessment of soil microbial communities associated with commercial Kits RNA (KR) and Manual RNA extraction (MR) methods. The aim of this study was to assess the bias of metatranscriptomics associated with total RNA extractions by commercial kits and manual methods from three geographically distinct paddy soils. [Methods] Three kinds of paddy soils were collected, representing distinctly different origin of parent materials including viscous black sand in Hailun city of Heilongjiang province, sandy loam in Binhai city of Jiangsu province, quaternary red clay in Yingtan city of Jiangxi province. Total RNA was obtained by commercial kit method and manual methods. The quantity and quality of total RNA were assessed by ultraviolet spectrophotometry and agarose gel electrophoresis. The abundance and composition of soil microbiome was analyzed by real-time quantitative PCR and high-throughput sequencing of cDNA reversely transcribed from 16S rRNA. [Results] The purity of RNA extracted by kit method was higher than that of manual method, but it did not hold true for the quantity of RNA extracts. The kit method generated higher quantity of soil RNA extract from paddy soils with higher organic matters from Heilongjiang and Jiangsu provinces, while the manual MR method recovered more RNA from paddy soil in Yingtan city of Jiangxi province with lower organic matter. High-throughput sequencing revealed a total of 27 phyla and 409 genera in three paddy soils, and 19 phyla and 181 genera were statistically significantly biased, i.e., the significant difference in relative abundance of these phylotypes was observed between manual MR and kit KR methods. These 19 phyla and 181 genera averaged for 40.4% and 44.4% of the total microbial abundance in soils. There are 11 phyla with higher relative abundance by KR than MR methods, and it is noteworthy that only Armatimonadetes phylum was KR-specifically biased, i.e., it is consistently detected in three soils with higher abundance by KR than MR methods. Similarly, the MR method also led to higher abundance of 11 phyla than KR method, and the Firmicutes is the only phylum for MR-specific bias, i.e., it is consistently observed with higher abundance in three soils by MR than KR method. At the genus level, the kit method consistently revealed two genera with higher relative abundance in all three paddy soils than manual method, and five genera were preferentially recovered by MR method in all three soils. We further assessed a total of 72 numerically dominant genera that could be recovered from all three soils having a relative abundance > 0.1%. These genera accounted for more than 80% of total microbial abundance in the three soils. The result suggested that 48 out of 72 genera have the same changing patterns of relative abundance among three paddy soils regardless of KR and MR methods. For example, the manual RNA extraction method indicated that the relative abundance of aerobic methanotrophs could be arranged in a decreasing order as Heilongjiang (1.68%) > Jiangxi (0.90%) > Jiangsu (0.59%), while the kit KR RNA extraction method has the same order as Heilongjiang (0.52%) > Jiangxi (0.18%) > Jiangsu (0.13%). [Conclusion] Among 27 phyla and 409 genera detected in paddy soils, there are only 2 phyla and 7 genera that was consistently biased in all three paddy soils by either kits KR or manual MR methods. It suggests that the biased phylotypes associated with the RNA extraction method itself could be insignificant, accounting only for 7.4% and 1.7% of the total phyla and genera respectively. Although the quantity and quality of RNA extracted from paddy soil are obviously different between manual MR and kit KR methods, the bias associated with these two RNA extraction methods has no significant impact on the biogeographic patterns of soil microbiomes in the three paddy soil tested. It is estimated that 70% of phyla and 22% of genera detected were observed with statistically significant difference between KR and MR methods. However, both RNA extraction methods could lead to the same conclusions regarding the changing patterns in relative abundance of microbial phylotpes among the three paddy soil tested. Despite the fact that the detection of every individual phylotype cannot be entirely reproduced by both methods and there are huge difference between these two methods, it seems plausible that the difference among soil types is sufficiently large so that the recovery of microbial communities would not be biased by RNA extraction method itself. These results imply that the method-specific bias of phylotype detection is much less than expected during soil RNA extraction. For future study the choice of RNA extraction method may not be of significant help, and the priority is to have experimental manipulation and treatments that would select for microbiomes with difference significantly larger than the bias associated with RNA extraction methods.