ABSTRACT
Sequencing of the 16S rRNA gene by Illumina next-generation sequencing is broadly used in microbiome studies. Different hypervariable regions of the 16S rRNA gene, V3V4 (amplified with primers 341F–805R) or V4 (V4O; primers 515F–806R), are selected, depending on the targeted resolution. However, in population-based clinical studies, combining V3V4 and V4 data from different studies for a meta-analysis is challenging. Reads generated by short-read (150-bp) high-throughput sequencing platforms do not fully recover the V4 region read-length. Here, we evaluated the compatibility of 16S rRNA V3V4 and V4 amplicons for microbiome profiling. We compared taxonomic compositions obtained by the analysis of V3V4 and V4 amplicons, and V4 fragments trimmed from V3V4 amplicons. We also evaluated an alternative V4 region (V4N; primers 519F–798R) designed for efficient stitching with 150-bp paired-end sequencing. First, we simulated a global investigation of environmental prokaryotes in silico. This revealed that V4O primers recovered the highest proportion of fragments (81.7%) and most phyla, including archaea. Empirical sequencing of standard (mock) and human fecal samples revealed biased patterns of each primer that were similar to the ones determined by in silico simulation. Further, for human fecal microbiome profiling, the between-sample variance was greater than the systematic bias of each primer. The use of trimmed V4 fragments and single-end amplicons resulted in the same systematic bias. In conclusion, paired-end V4O sequencing yielded the most accurate data for both, simulation and mock community sequencing; the V4O amplicons were compatible with trimmed V4 sequences for microbiome profiling.
IMPORTANCE Next-generation sequencing of the 16S rRNA gene is a commonly used approach for clinical microbiome studies. Different amplicons of the 16S rRNA hypervariable regions are used in different studies, which creates incompatible sequence features when comparing and integrating data among studies by using 16S denoising pipelines. Here we compared the type of data and coverage obtained when different 16S rRNA amplicons were analyzed. In silico and empirical analyses of the human fecal microbiome revealed that the V3V4 amplicons are compatible with V4 amplicons after trimming up to the same region. These observations demonstrate that reconciling the compatibility of clinical microbiome data from different studies improve not only the sample size but also the confidence of the hypothesis tested.
