Targeted rRNA depletion enables efficient mRNA sequencing in diverse bacterial species and complex co-cultures

Abstract

Microbes present one of the most diverse sources of biochemistry in nature, and mRNA sequencing provides a comprehensive view of this biological activity by quantitatively measuring microbial transcriptomes. However, efficient mRNA capture for sequencing presents significant challenges in prokaryotes as mRNAs are not poly-adenylated and typically make up less than 5% of total RNA compared with rRNAs that exceed 80%. Recently developed methods for sequencing bacterial mRNA typically rely on depleting rRNA by tiling large probe sets against rRNAs; however, such approaches are expensive, time-consuming, and challenging to scale to varied bacterial species and complex microbial communities. Therefore, we developed EMBR-seq+, a method that requires fewer than 10 short oligonucleotides per rRNA to achieve up to 99% rRNA depletion in diverse bacterial species. Finally, EMBR-seq+ resulted in a deeper view of the transcriptome, enabling systematic quantification of how microbial interactions result in altering the transcriptional state of bacteria within co-cultures.

Keywords: bacterial mRNA sequencing; fungal and bacterial co-cultures; lignocellulose deconstruction; non-model microbial sequencing; rRNA depletion.

Fig 1

EMBR-seq+ depletes rRNA from E. coli total RNA at high efficiency without introducing technical biases using a small number of rRNA-specific probes. (A) Schematic for EMBR-seq+ shows that a combination of blocking primers and RNase H-based targeting is used to deplete rRNA. This panel was generated using Biorender.com. (B) Cumulative percentage of reads mapping along E. coli 16S and 23S rRNA for different depletion methods. Bold lines indicate mean values, and shaded regions indicate the minimum and maximum over three independent experiments. Inverted triangles indicate location of hotspot regions targeted by blocking primers. Insets highlight the depletion of reads from hotspot locations along the 16S and 23S rRNA sequence. (C) Percent of sequencing reads mapping to rRNA and mRNA in E. coli. Bars indicate mean values and error bars indicate standard deviations over three independent experiments. rRNA depletion in the “RNase H,” “EMBR-seq,” and “EMBR-seq+” conditions is all statistically significant from the “No depletion” condition (permutation test, P < 10⁻⁵). (D) Number of genes detected above different gene expression thresholds for E. coli. Points indicate mean values, and error bars indicate standard deviations over three independent experiments. (E) Correlation of gene expression between the “No depletion” and “EMBR-seq+” conditions (Pearson’s r = 0.95). Reads per million (RPM) is computed after removal of rRNA reads. The x- and y-coordinates of each point indicate mean values over three independent experiments in the two conditions. For consistent comparison across methods and to control for the variability in sequencing depth across samples, panels B and D show data that have been downsampled to 1 million sequencing reads.

Fig 2

EMBR-seq+ can be easily extended to diverse bacterial species. (A) Percent of sequencing reads mapping to rRNA in F. succinogenes strain UWB7 and G. metallireducens for different rRNA depletion methods. Bars indicate mean values and error bars indicate standard deviations over three independent experiments. rRNA depletion in the “RNase H,” “EMBR-seq,” and “EMBR-seq+” conditions is statistically significant from the “No depletion” condition (permutation test, P < 10⁻⁵). (B) Cumulative percentage of reads mapping to 16S and 23S rRNA of F. succinogenes strain UWB7 for different rRNA depletion methods. Bold lines indicate mean values, and shaded regions indicate the minimum and maximum over three independent experiments. Inverted triangles indicate the location of hotspots targeted by blocking primers. (C) Number of genes detected above different gene expression thresholds for F. succinogenes strain UWB7. Points indicate mean values, and error bars indicate standard deviations over three independent experiments. (D) Correlation of gene expression between the “No depletion” and “EMBR-seq+” conditions for F. succinogenes strain UWB7 (Pearson’s r = 0.94). RPM is computed after removal of rRNA reads. The x- and y-coordinates of each point indicate mean values over three independent experiments in the two conditions. For consistent comparison across methods and to control for the variability in sequencing depth across samples, panels B and C show data that have been downsampled to 1 million sequencing reads.

Fig 3

EMBR-seq+ efficiently depletes rRNA of F. succinogenes strain UWB7 grown in co-culture with A. robustus. (A) Schematic of fungal mRNA and rRNA removal from co-culture-derived total RNA samples for more efficient bacterial mRNA sequencing. This panel was generated using Biorender.com. (B) Percentage of reads mapping to the bacterial and fungal genomes using the three strategies described in panel A: total RNA isolated from co-culture pellets treated with EMBR-seq to remove bacterial rRNA [“Unprocessed total RNA (EMBR-seq)”], fungal poly-adenylated mRNA-depleted total RNA treated with EMBR-seq to remove bacterial rRNA [“Poly-A depleted (EMBR-seq)”], and “Poly-A depleted (EMBR-seq)” library additionally treated with RNase H to remove fungal and bacterial rRNA [“Poly-A depleted & RNase H treated (EMBR-seq+)”]. For the “Unprocessed total RNA (EMBR-seq)” library, n = 1. For “Poly-A depleted (EMBR-seq)” and “Poly-A depleted & RNase H treated (EMBR-seq+)” libraries, n = 3. Bars indicate mean values, and error bars indicate standard deviation from the mean. (C) Cumulative percentage of sequencing reads from the “Poly-A depleted (EMBR-seq)” library that align to the C. churrovis genome scaffold 672. The three rRNA hotspots were inferred to be the 3′-ends of 18S, 28S, and 5.8S rRNA for C. churrovis. Black inverted triangles and italicized text indicate target locations of the five RNase H probes used for fungal rRNA depletion in the EMBR-seq+ libraries. (D) Percent of sequencing reads mapping to the rRNA of F. succinogenes strain UWB7 for different depletion methods when grown in co-culture with A. robustus or C. churrovis. All four samples are poly-A depleted to remove fungal mRNA. Bars indicate mean values, and error bars indicate standard deviation over three independent experiments. For co-cultures of F. succinogenes strain UWB7 and A. robustus, rRNA depletion in the “RNase H,” “EMBR-seq,” and “EMBR-seq+” conditions is statistically significant from the “No depletion” condition (permutation test, P < 0.05). For co-cultures of F. succinogenes strain UWB7 and C. churrovis, rRNA depletion in the “RNase H,” and “EMBR-seq+” conditions is statistically significant from the “No depletion” condition (permutation test, P < 0.05). (E) Volcano plot of differentially expressed genes of F. succinogenes strain UWB7 grown in monoculture vs. co-culture with A. robustus, for the “No depletion” condition and after rRNA depletion using EMBR-seq+. n = 3 for both culture conditions and for both the library preparation conditions. Colored points indicate differentially expressed genes, determined by thresholds of |log₂ (fold change)| > 0.8, P_adj < 0.05, and RPM > 2. (F) Heatmap of differentially expressed CAZymes in F. succinogenes strain UWB7 grown in monoculture vs. co-culture with A. robustus. Genes are colored on the left side of the heatmap based on the CAZyme family they are associated with.