Trypanosoma cruzi specific mRNA amplification by in vitro transcription improves parasite transcriptomics in host-parasite RNA mixtures

Abstract

Background: Trypanosomatids are a group of protozoan parasites that includes the etiologic agents of important human illnesses as Chagas disease, sleeping sickness and leishmaniasis. These parasites have a significant distinction from other eukaryotes concerning mRNA structure, since all mature mRNAs have an identical species-specific sequence of 39 nucleotides at the 5' extremity, named spliced leader (SL). Considering this peculiar aspect of trypanosomatid mRNA, the aim of the present work was to develop a Trypanosoma cruzi specific in vitro transcription (IVT) linear mRNA amplification method in order to improve parasite transcriptomics analyses.

Methods: We designed an oligonucleotide complementary to the last 21 bases of T. cruzi SL sequence, bearing an upstream T7 promoter (T7SL primer), which was used to direct the synthesis of second-strand cDNA. Original mRNA was then amplified by IVT using T7 RNA polymerase. T7SL-amplified RNA from two distinct T. cruzi stages (epimastigotes and trypomastigotes) were deep sequenced in SOLiD platform. Usual poly(A) + RNA and and T7-oligo(dT) amplified RNA (Eberwine method) were also sequenced. RNA-Seq reads were aligned to our new and improved T. cruzi Dm28c genome assembly (PacBio technology) and resulting transcriptome pattern from these three RNA preparation methods were compared, mainly concerning the conservation of mRNA transcritional levels and DEGs detection between epimastigotes and trypomastigotes.

Results: T7SL IVT method detected more potential differentially expressed genes in comparison to either poly(A) + RNA or T7dT IVT, and was also able to produce reliable quantifications of the parasite transcriptome down to 3 ng of total RNA. Furthermore, amplification of parasite mRNA in HeLa/epimastigote RNA mixtures showed that T7SL IVT generates transcriptome quantification with similar detection of differentially expressed genes when parasite RNA mass was only 0.1% of the total mixture (R = 0.78 when compared to poly(A) + RNA).

Conclusions: The T7SL IVT amplification method presented here allows the detection of more potential parasite differentially expressed genes (in comparison to poly(A) + RNA) in host-parasite mixtures or samples with low amount of RNA. This method is especially useful for trypanosomatid transcriptomics because it produces less bias than PCR-based mRNA amplification. Additionally, by simply changing the complementary region of the T7SL primer, the present method can be applied to any trypanosomatid species.

Keywords: RNA amplification; RNA-Seq; T7 RNA polymerase; Transcriptomics; Trypanosoma cruzi; Trypanosomatid.

Fig. 1

T7SL IVT method. a Comparison between T7oligo(dT) (left) and T7SL IVT (right) methods. In the Eberwine method, the reverse transcription reaction is performed using a T7oligo(dT) primer, resulting in a first-strand cDNA containing the T7 promoter. The T7 RNA polymerase is added for in vitro transcription of the purified cDNA and antisense aRNA is obtained. For the second method (T7SL IVT), the reverse transcription reaction is performed using random primers and the second-strand cDNA is obtained by a DNA polymerase reaction with the T7SL primer, which also contains the T7 promoter. After cDNA purification, in vitro transcription with the T7 RNA polymerase is performed, producing sense aRNA. b Distribution of RNA lengths from IVT samples and poly(A) + RNA (control), quantified by the BioAnalyzer 2100 equipment (Agilent). c Median length of poly(A) + RNA and aRNAs from the IVT samples. d aRNA yield obtained from the two IVT methods

Fig. 2

aRNA-Seq results. a RNA-Seq coverage along annotated genes. To plot all genes in the same graph, all coding sequences were split in 100 bins (percentiles) and the number of reads aligned to each percentile were summed and plotted as a ratio against the bin with higher number of aligned reads. Dotted lines are standard deviation. b IGV genome browser visualization of RNA-Seq reads alignment along a 3 kb gene for all three methods used (specified in left). Coverage were plotted in log scale. c correlation between the technical replicates (same RNA input for different amplification reactions) of T7SL IVT and T7oligo(dT) IVT methods. d correlation between biological replicates (RNA from separate epimastigote populations) of T7SL IVT and T7oligo(dT) IVT. For all scatter plots, scales are log₂ of normalized read counts and values inside the graphs represent Pearson correlation

Fig. 3

Putative DEG detection for the three analyzed methods. a Euler diagram showing the number and overlap of detected putative DEGs (FDR < 0.01) for each method (T7SL, T7dT and poly(A)+) when comparing epimastigote to trypomastigote transcriptomes, without a fold change threshold (left diagram) or at least a fold change of two (right diagram). Graphics produced on Cytoscape 3.2.0 b Scatter plot correlating epimastigote to trypomastigote fold changes (log₂) for the different RNA-Seq methods; values inside the graphs represent Pearson correlation. Far right scatter plot correlates T7SL fold changes to the ones detected by our analysis of Li and collaborators data [44]

Fig. 4

Performance of T7SL IVT on RNA mixtures. a Length distribution profiles for aRNA produced from pure samples (T. cruzi epimastigote and HeLa), mixture of T. cruzi (epimastigote) and HeLa and blank samples (no RNA for amplification reaction). Percentage are relative mass of Epi:HeLa on RNA mixture used for T7SL IVT. b After T7SL IVT and RNA-Seq of mixture samples, reads were aligned to T. cruzi and human genomes. The percentage of aligned reads with best match on parasite or human genome were retrieved and plotted. The input mass (in ng) used in each mixture is specified at the bottom. c Scatterplot comparing T7SL Pfx IVT and mixtures of HeLA-Epi RNAs, or PolyA+ RNA against HeLA-Epi RNAs. d Scatter plots of epimastigote to trypomastigote fold changes when comparing PolyA(+) transcriptome quantification to T7SL using T.cruzi/HeLa RNA mixtures

Fig. 5

IGV view of RNA-Seq aligned reads coverage for T.cruzi-HeLa mixture samples. A ~8 kb genome region, containing six genes (blue boxes at the bottom), is showed. For each sample, y axis is log10 of coverage, with counts range specified at the left. Percentage of epimastigote RNA in the T. cruzi-HeLa RNA mixture used for T7SL Pfx amplification is indicated for each sample

Fig. 6

Minimal input RNA mass for optimal T7SL IVT. a Graph showing the correlation between the input total RNA mass (X axis) and the aRNA yield (Y axis). b aRNA length distribution obtained with different input RNA mass. Note that below 6.25 ng input, the aRNA lose the typical smooth length distribution. For all samples, the same RNA mass were applied on a BioAnalyzer chip. c RNA-Seq scatterplot with different RNA inputs (from 100 to 3 ng) for T7SL IVT. Pearson correlation is depicted, based on log normalized read counts. Note that even when 3 ng of mass input is used for RNA amplification, the transcriptome quantification is very similar across all expression levels. “Sort” sample correspond to 10⁵ epimastigotes sorted directly to RNA extraction buffer, which roughly corresponds to 68 ng of total RNA (Additional file 4, Table S2)

Fig. 7

T7SL IVT DEGs detection on mixtures and limiting mass samples. a Hierarchical clustering of all DEGs (FDR < 0.01) when comparing all epimastigotes samples (mixtures and limiting mass) to the trypomastigote stage. Blue/black/yellow color code is log₂ of expression fold change of epimastigote to trypomastigote). Supra Gene expression level is rainbow color coded based on log₂ of counts per million reads (log₂ CPM). Heatmap was created using Euclidian distance, average linkage clustering method and MeV (Multiple Experiment Viewer) v. 4.8.1 software. Note that, in general, all epimastigote samples show a similar fold change pattern for the great majority of genes, independent of initial mass used for amplification (limiting mass samples) or percentage of parasite RNA on host-parasite mixtures (mixture samples). b Principal Component Analysis (PCA) of the same samples used in a. The first component represents 84% of the total variation and is mainly due to epimastigote to trypomastigote differences; the second component represents 9% of the total variation and is mainly due to epimastigote transcriptome differences between the limiting mass experiments. PCA graph was created using Perseus v.1.5.0.31 software. c Euler diagram showing the number and overlap of detected DEGs for epimastigote to trypomastigote comparison using five different groups: (i) Poly(A)+: both parasite stages analyzed by Poly(A) + RNA; (ii) T7SL: both parasite stages analyzed by T7SL IVT amplification method from 100 ng of initial mass; (iii) Epi_3ng: limiting mass of 3 ng for epimastigote T7SL IVT; (iv) Epi_0.1%: mixture of host-parasite RNA containing only 0.1% of parasite RNA; (v) Epi_Sort: amplification of RNA obtained from 10⁵ sorted epimastigotes