New sequencing methodologies reveal interplay between multiple RNA-binding proteins and their RNAs

Abstract

It is now established that base-pairing regulatory RNAs are key players in post-transcriptional regulatory networks where they affect the translation and/or stability of their target RNAs. In many cases, the base-pairing between two RNAs is facilitated by an RNA-binding protein (RBP) that serves as an RNA chaperone. Recent advances in sequencing methods have revealed the RNA populations bound by the RBPs, yielding insights valuable into regulatory networks. Further analyses of these networks can improve our understanding of the roles played by RBPs in the regulation of gene expression by regulatory RNAs, especially when multiple RBPs are involved. For example, using an RNA sequencing-based methodology that captures RNA-RNA pairs on RBP, an interplay between two RBPs in bacteria that compete on the same RNA-RNA pair was revealed. In this case, one protein promotes negative regulation of the target RNA while the second protein can block this regulation. In this mini-review, I outline the exciting future directions that can be taken to deepen our understanding of the roles played by RBPs in post-transcriptional regulation, and discuss how the different sequencing methods can assist in deciphering the relationships among RBPs, and between the RBPs and the RNAs they bind. Having a more detailed picture of the RBPs-RNAs network will elucidate how bacteria can have nuanced control of gene expression, critical for survival in the varied environments in which bacteria live.

In the last two decades there has been increasing awareness of the important roles RNAs play in cells as regulators of gene expression, primarily as post-transcriptional regulators, in addition to their traditional roles as messenger RNAs, ribosomal RNAs and transfer RNAs. The roles played by these RNAs add another layer of regulation on top of the extensive transcriptional regulation taking place (Bidnenko and Bidnenko 2018, Browning and Busby 2016, Ray-Soni, et al. 2016). Two of the most studied groups of regulatory RNAs are microRNAs (miRNAs) in eukaryotes and small RNAs (sRNAs) in prokaryotes (Fabian, et al. 2010, Hör, et al. 2020, Storz, et al. 2011). These RNAs share a common mode of action as they usually base-pair with their target RNAs and affect their translation and/or stability.

Important players in RNA-mediated post-transcriptional regulation are RNA-binding proteins (RBPs). These proteins, among other diverse biological functions, facilitate the regulation by miRNAs and sRNAs (Beckmann, et al. 2016, Holmqvist and Vogel 2018, Kilchert, et al. 2019, Nechay and Kleiner 2019). miRNAs are part of an RNA-induced silencing complex (RISC) containing multiple RBPs including the key member Argonaute, which promotes base-pairing between a miRNA and its target mRNAs (Fabian, et al. 2010, Pratt and MacRae 2009). In bacteria, the broadly-conserved RBP Hfq similarly promotes sRNA base pairing with targets (Sobrero and Valverde 2012, Vogel and Luisi 2011). Hfq is a doughnut-shaped homohexamer and has three main binding sites for RNAs, a proximal face that mainly binds sRNAs, a distal phase that mainly binds mRNAs, and a rim that helps promote base-pairing between two RNAs (Updegrove, et al. 2016, Woodson, et al. 2018). The importance of Hfq to the bacteria is underscored by the extensive set of RNAs bound by Hfq (Melamed, et al. 2016) and by the observations that Hfq-mediated regulation affects almost every aspect of bacterial physiology (Dutta and Srivastava 2018, Papenfort and Vogel 2010, Papenfort and Vogel 2014). Interestingly, in addition to its role in promoting sRNA-target base-pairing, Hfq was found to affect translation via rRNA processing and tRNA maturation (dos Santos, et al. 2019). Another group of bacterial RBPs gaining attention in the last few years is the ProQ/FinO-domain family of proteins found in many bacterial species and also suggested to play a role as RNA matchmakers (Olejniczak and Storz 2017). The few members of this family that have been studied include the plasmid encoded FinO, which plays a role in antisense regulation of F plasmid conjugation in Escherichia coli (Glover, et al. 2015), the chromosomal encoded RocC, which mediates translational repression in Legionella pneumophila (Attaiech, et al. 2016), and the chromosomal encoded ProQ shown to interact with hundreds of transcripts in E. coli and Salmonella typhimurium (Holmqvist, et al. 2018).

A more detailed discussion on the characteristics of Hfq, ProQ and other RBPs in bacteria can be found elsewhere (Holmqvist and Vogel 2018). Here, I review recently-developed transcriptome-wide sequencing approaches that have contributed significantly to the understanding of the characteristics and the roles of the RBPs as well as sRNAs in regulatory networks and are setting the stage for future studies. The methods discussed in this mini-review are summarized in Figure 1.

Figure 1.

High throughput methods to decipher the interplay among RBPs and between RPB and their target RNAs. CLIP-seq (crosslinking immunoprecipitation coupled with high-throughput sequencing) reveals the RNA population bound to the RBP and enables the detection of binding sites of RNAs to an RBP at single-nucleotide resolution. RIL-seq (RNA interaction by ligation and sequencing) allows transcriptome-wide identification of RNA-RNA interactions involving RBP-associated RNA and provides the RNA population bound to the RBP. CLASH (crosslinking, ligation and sequencing of hybrids) also allows the detection in vivo of RNA–RNA pairs found on an RBP. Rloc-seq (RNA localization by fractionation and sequencing) provides information about the distribution of the bacteria transcriptome in different subcellular localizations. Proximity-CLIP (compartment-specific protein biotinylation with ultraviolet crosslinking, which crosslinks RNA with RBPs in intact cells) allows the detection of localized RBPs and RNAs, quantification of the localized transcriptome, and the identification of RBP-occupied RNA loci. ChIPPAR-seq (Chromatin immunoprecipitation with cells treated or untreated with rifampicin followed by DNA sequencing) identifies nascent transcripts bound to RBPs.

Gray icons represent RBPs, mRNAs are colored red, sRNAs are colored blue; squiggly lines correspond to RNAs and straight lines correspond to fragments sequenced. DNA is represented by squiggly gray lines.

RNA populations and RNA-RNA interactomes on RBPs

Early studies addressed the role of RBPs such as Hfq by identifying sRNAs and mRNAs that where co-immunoprecipitated with the proteins (Zhang, et al. 2003). Based on these initial studies, methods that incorporate co-immunoprecipitation with deep sequencing were developed to further explore the global role of RBPs in regulation of gene expression. One such method is CLIP-seq, in which the protein of interest (e.g. Hfq) is in-vivo crosslinked to the RNAs it binds by ultraviolet light, the RNA-protein complex is isolated, and the associated RNAs are sequenced (Holmqvist, et al. 2016). This approach provides information on both the RNA population found on the protein and the binding sites on the RNAs since the position of UV crosslinking can be detected by a change in the sequence. CLIP-seq and other high throughput methods enhanced our understanding of RNA-protein interactions and the biological role RBPs play in the regulation of gene expression on a global level. A more detailed review on RNA-seq as a tool to study regulation of gene expression at multiple levels can be found elsewhere (Hör, et al. 2018).

A major breakthrough in the study of RBP-mediated regulation was achieved by the development of methods which capture RNA-RNA pairs found in close proximity on the RBPs and provide detailed information on the RNA-RNA interactome associated with these RBPs. These methods include RIL-seq (Melamed, et al. 2018, Melamed, et al. 2016) and CLASH (Waters, et al. 2017). The key experimental steps involve crosslinking of the RNAs to the RBP, purification of the RBP with its bound RNAs followed by a ligation of RNAs found in proximity on each RBP ultimately allowing the detection of RNA pairs upon RNA sequencing. A key step for these methods, as well as it is for other deep sequencing-base approaches, is the discrimination between physiologically relevant RNA-RNA interactions and sporadic or unspecific interactions. RIL-seq and CLASH address this aspect in two different ways, computationally and experimentally, respectively. The computational component of the RIL-seq protocol involves filtration of the data for RNA-RNA pairs that are statically significantly over-represented in order to eliminate periodic or unspecific interactions. The CLASH protocol, on the other hand, includes long and stringent washes of the RNA-protein complex that may reduce the number of detected RNA-RNA pairs but also enrich the fraction of the RNAs covalently bound to the RBP, possibly better representing the RNA population found on the RBP. Application of RIL-seq and CLASH to Hfq and RNase E (a key player in many RNA-RNA interactions facilitated by Hfq), respectively, to different E. coli strains revealed dynamic and extensive network of RNA-RNA interactions, novel regulatory circuits, and that sRNAs can originate from any genetic location (Melamed, et al. 2016, Waters, et al. 2017).

Relationships between RBPs

The first RIL-seq and CLASH experiments contributed significantly to our understanding of the biological roles of a single RBP. However, RBPs do not exist in isolation. Furthermore, an increasing number of RBPs have been discovered in recent years by global identification approaches (Queiroz, et al. 2018, Smirnov, et al. 2016, Urdaneta, et al. 2019), emphasizing the importance of studying the interplay between RBPs in the cell. By using RIL-seq to compare the RNA interactomes of ProQ and Hfq under the same growth conditions, Melamed and Adams et al, documented that two RBPs can have overlapping and competing roles in E. coli, allowing nuanced regulation of gene expression (Melamed, et al. 2020). First, Hfq and ProQ were found to bind different sets of RNAs, with Hfq binding mRNAs and sRNAs equally, and ProQ primarily binding mRNAs. Second, in many cases ProQ had a tendency to bind to mRNAs internal to the coding sequence or even at the 3’ untranslated region, which differs from the canonical Hfq binding to mRNAs at the 5’ untranslated region or in close proximity to the start codon. These observations hint that ProQ and Hfq might have different roles in bacterial cells.

Surprisingly, comparison of the RNA-RNA pairs found on each of the RBPs revealed that a significant fraction of the RNA-RNA pairs captured on ProQ were also present on Hfq, suggesting the two proteins may have overlapping and/or competing roles. One example of overlapping RNA-RNA pairs is between RybB, a previously-characterized sRNA known to act under cell envelope stress, and a novel sRNA RbsZ. Intriguingly, RbsZ corresponds to a 3’ untranslated region, a region that in recent years unexpectedly has been shown to modulate gene expression (El Mouali and Balsalobre 2019) and found to be a source of many sRNAs (Miyakoshi, et al. 2015). Studies of the RybB-RbsZ pair revealed that RbsZ regulates RybB levels by base pairing, leading to RybB degradation and affecting RybB ability to regulate its own targets. This regulation was mediated by Hfq. In contrast, the effects of RbsZ were blocked by ProQ resulting in elevated RybB. Examination of changes in the E. coli transcriptome upon pulse overexpression of ProQ showed that the protective effect of ProQ was true for many RNAs. This suggested that E. coli ProQ has a global role in protecting or stabilizing RNA transcripts, adding another layer of regulation to the already complex post-transcriptional regulatory network.

Factors affecting relationships between RBPs

Further permutations of RNA sequencing-based approaches are uncovering more relationships between RBPs and elucidating the factors that affect these relationships. For example, a recent study in Pseudomonas aeruginosa revealed how two RBPs can act together to affect a subset of genes (Kambara, et al. 2018). Here the authors used a modified ChIP-seq approach (termed ChIPPAR-seq) to detect nascent transcripts bound by an RBP. In this method, chromatin immunoprecipitation of tagged Hfq or Crc, another post-transcriptional regulator, was examined for cells grown with or without treatment with the RNA polymerase inhibitor rifampicin, followed by high-throughput DNA sequencing. It was found that Hfq can bind to nascent transcripts together with Crc and that the two RBPs can influence one another’s association with the transcripts. This mode of action provides an efficient means of controlling the translation of specific mRNAs where transcription and translation are coupled. Another example of one RBP affecting other RBPs comes from a study performed in the budding yeast Saccharomyces cerevisiae, where it was found that an RPB named Puf3p post-transcriptionally modulates the levels of other RBPs by regulating their translational efficiencies (Wang, et al. 2019).

Another aspect that can affect the outcome of an interplay between RBPs is growth or stress conditions. For instance, ProQ was reported to be critical in osmoprotection (Kerr, et al. 2014), suggesting that the interplay between Hfq and ProQ might be different under osmotic stress. Studying the elements that determine the outcomes of these competitions will be extremely valuable for obtaining reliable models for the relationships between RBPs. Factors that are likely to contribute include the RNA binding affinities of the RBPs, RNA stability, and the concentration of the RBPs. Further information about all of these parameters undoubtably will be gained from additional sequencing experiments, but modeling these regulatory networks, consisting of multiple RBPs and hundreds of bound RNAs will be a challenge in the coming years.

Finally, the subcellular localization of the RBPs likely also contributes to their various regulatory roles and their relationships with other RBPs and the RNAs they bind. A recent study by Kannaiah et al, which used a combination of cell fractionation with deep-sequencing (termed Rloc-seq) to determine the subcellular localization of each transcript, suggested that there is a spatiotemporal organization of the E. coli transcriptome (Kannaiah, et al. 2019). This work indicated that upon stress, sRNAs are enriched at the poles in an Hfq-dependent manner. In eukaryotes, it has already been established that both coding and non-coding RNA transcripts are asymmetrically localized in the cells, affecting the regulation of gene expression (Lecuyer, et al. 2007, Wilk, et al. 2016). Studying the contribution of RBPs to asymmetric RNA distribution may lead to deeper understanding of the purpose of having multiple RBPs binding the same RNA or RNA pairs. Proximity-CLIP (Benhalevy, et al. 2018), which provides a snapshot of protein-occupied RNA elements in subcellular compartments, also can be used to clarify these location-dependent roles. Proximity-CLIP is based on APEX2-mediated proximity biotinylation of proteins with a photoactivatable ribonucleoside. For cells expressing the APEX2-tagged protein, ribonucleoside-enhanced crosslinking can then be used to profile both the proteome and the transcriptome bound by RBPs in a specified location in the cell. As a proof of principle, proximity-CLIP was applied to cytoplasmic and nuclear compartments of human cells but, with a few modifications, can be adapted to study subcellular compartments in bacteria.

Perspectives

The recent advances in the sequencing methodologies described here indicate we are only at the beginning of the journey in terms of understanding the relationships between RBPs in the cells. The complex network of RBPs regulating overlapping or unique subsets of RNAs undoubtedly is important for the bacterial prosperity in a host or in the environment, where the environmental conditions are subjected to rapid changes. It allows the bacteria more nuanced control of gene expression and provides them the advantage of quick adjustment to changes in their surroundings. As more and more interplay between RBPs is revealed, the world of RBPs is only expected to become more complex and more thrilling to explore.