Resolving host-pathogen interactions by dual RNA-seq

Abstract

The transcriptome is a powerful proxy for the physiological state of a cell, healthy or diseased. As a result, transcriptome analysis has become a key tool in understanding the molecular changes that accompany bacterial infections of eukaryotic cells. Until recently, such transcriptomic studies have been technically limited to analyzing mRNA expression changes in either the bacterial pathogen or the infected eukaryotic host cell. However, the increasing sensitivity of high-throughput RNA sequencing now enables "dual RNA-seq" studies, simultaneously capturing all classes of coding and noncoding transcripts in both the pathogen and the host. In the five years since the concept of dual RNA-seq was introduced, the technique has been applied to a range of infection models. This has not only led to a better understanding of the physiological changes in pathogen and host during the course of an infection but has also revealed hidden molecular phenotypes of virulence-associated small noncoding RNAs that were not visible in standard infection assays. Here, we use the knowledge gained from these recent studies to suggest experimental and computational guidelines for the design of future dual RNA-seq studies. We conclude this review by discussing prospective applications of the technique.

Fig 1. Methods for RNA sequencing of bacterial infections.

A. Concept of dual RNA-seq. Total RNA is extracted from infected cells and analyzed by RNA-seq. The mixed sequencing reads are assigned to their originating genomes in silico. B. Different approaches to quantify gene expression of bacteria in context with mammalian host cells. Traditionally, host material was depleted prior to analysis, either by detergent-mediated lysis of host cells (left) or by sequence-specific removal of host transcripts (middle). Instead, dual RNA-seq omits host depletion (right) and analyzes pathogen and host gene expression in parallel.

Fig 2. A generic dual RNA-seq workflow analyzing total mixed RNA after double rRNA depletion that discovered the role of PinT small regulatory RNA (sRNA) during Salmonella infection of host cells [13].

Salmonella having gfp stably integrated in their chromosome and expressed from a constitutive promoter were used to infect cultures of HeLa cells. RNA-seq of the bacterial input (1) or mock-infected HeLa cells (2) served as reference controls for Salmonella or human gene expression analysis, respectively. Infection was carried out in parallel with wild-type and sRNA mutant Salmonella strains, and samples were taken over a time-course of infection. The resulting cell samples constituted a mixed population consisting of both invaded (GFP-positive) and uninfected bystander (GFP-negative) cells (3). To obtain a homogeneous population of invaded cells, the samples were sorted based on the emitted GFP fluorescence (4). Total RNA was extracted from the thus enriched cells, rRNA from both infection partners was depleted (5), and rRNA-free samples were converted into cDNA libraries and sequenced. The resulting sequencing reads were mapped in parallel against the Salmonella and human (core and mitochondrial) genome. Differential expression analysis of the time course revealed the strong induction over time of a novel Salmonella sRNA, PinT, and comparative analysis unraveled the molecular footprint of this sRNA in the bacterial transcriptome (6). Likewise, comparison of the host transcriptome between wild-type and ΔpinT infections revealed PinT-dependent changes in the immune response, including a differential activation of Janus kinase-Signal Transducer and Activator of Transcription (JAK-STAT) signaling as well as changes with respect to the expression of host long noncoding RNAs (lncRNAs) and microRNAs (miRNAs) (7). In addition, the pinT status of the infecting bacterium influenced mitochondrial gene expression, and infection with ΔpinT Salmonella led to the relocalization of mitochondria in invaded host cells (8).

Fig 3. Illustration of biological insights obtained from dual RNA-seq studies in four different bacterial infection models.

HEp-2 cells infected with obligate intracellular Chlamydia trachomatis [10], primary airway epithelial cells with nontypeable Haemophilus influenzae [16], primary murine bone marrow macrophages with uropathogenic E. coli (UPEC) [11], and diverse human, mouse, and porcine cell lines with Salmonella Typhimurium [13]. See main text for details.

Fig 4. Bioinformatic analysis pipeline for dual RNA-seq datasets.

Quality-filtered RNA-seq reads are aligned in parallel against the respective host and pathogen replicons. Reads mapping equally well to both reference organisms (“cross-mappings”) are quantified and discarded from downstream analyses. Reads unequivocally mapped to either the bacterial or host reference are used for quantification and functional analyses. Dual RNA-seq enables a wide range of downstream analyses, discussed in detail in the text. “MT,” mitochondrial genome.