Reverse transcriptase template switching during reverse transcriptase-polymerase chain reaction: artificial generation of deletions in ribonucleotide reductase mRNA

Comparative Transcriptome Profiling of Disruptive Technology, Single- Molecule Direct RNA Sequencing

Background:

The standard approach for transcriptomic profiling involves high throughput short-read sequencing technology, mainly dominated by Illumina. However, the short reads have limitations in transcriptome assembly and in obtaining full-length transcripts due to the complex nature of transcriptomes with variable length and multiple alternative spliced isoforms. Recent advances in long read sequencing by the Oxford Nanopore Technologies (ONT) offered both cDNA as well as direct RNA sequencing and has brought a paradigm change in the sequencing technology to greatly improve the assembly and expression estimates. ONT enables molecules to be sequenced without fragmentation resulting in ultra-long read length enabling the entire genes and transcripts to be fully characterized. The direct RNA sequencing method, in addition, circumvents the reverse transcription and amplification steps.

Objective:

In this study, RNA sequencing methods were assessed by comparing data from Illumina (ILM), ONT cDNA (OCD) and ONT direct RNA (ODR).

Methods:

The sensitivity & specificity of the isoform detection was determined from the data generated by Illumina, ONT cDNA and ONT direct RNA sequencing technologies using Saccharomyces cerevisiae as model. Comparative studies were conducted with two pipelines to detect the isoforms, novel genes and variable gene length.

Results:

Mapping metrics and qualitative profiles for different pipelines are presented to understand these disruptive technologies. The variability in sequencing technology and the analysis pipeline were studied.

Template-switching artifacts resemble alternative polyadenylation

BACKGROUND

Alternative polyadenylation is commonly examined using cDNA sequencing, which is known to be affected by template-switching artifacts. However, the effects of such template-switching artifacts on alternative polyadenylation are generally disregarded, while alternative polyadenylation artifacts are attributed to internal priming.

RESULTS

Here, we analyzed both long-read cDNA sequencing and direct RNA sequencing data of two organisms, generated by different sequencing platforms. We developed a filtering algorithm which takes into consideration that template-switching can be a source of artifactual polyadenylation when filtering out spurious polyadenylation sites. The algorithm outperformed the conventional internal priming filters based on comparison to direct RNA sequencing data. We also showed that the polyadenylation artifacts arise in cDNA sequencing at consecutive stretches of as few as three adenines. There was no substantial difference between the lengths of poly(A) tails at the artifactual and the true transcriptional end sites even though it is expected that internal priming artifacts have shorter poly(A) tails than genuine polyadenylated reads.

CONCLUSIONS

Our findings suggest that template switching plays an important role in the generation of spurious polyadenylation and support the need for more rigorous filtering of artifactual polyadenylation sites in cDNA data, or that alternative polyadenylation should be annotated using native RNA sequencing.

Quantification of RNA by Real-Time Reverse Transcription-Polymerase Chain Reaction (RT-PCR)

This protocol describes a real-time reverse transcription-polymerase chain reaction (RT-PCR) assay using a two-enzyme, two-tube approach, carried out using either SYBR Green I or TaqMan chemistries. The protocol uses a PCR volume of 20 µL (although most manufacturers recommend 50-µL reactions). However, if the PCR target is not very abundant (i.e., present at one to 10 copies per sample), a larger volume may yield better reproducibility between samples. Discussion on preparing high-quality RNA, choosing a priming method, selecting an enzyme, and selecting an endogenous reference gene is also included.

Transcriptional-Readthrough RNAs Reflect the Phenomenon of "A Gene Contains Gene(s)" or "Gene(s) within a Gene" in the Human Genome, and Thus Are Not Chimeric RNAs

Tens of thousands of chimeric RNAs, i.e., RNAs with sequences of two genes, have been identified in human cells. Most of them are formed by two neighboring genes on the same chromosome and are considered to be derived via transcriptional readthrough, but a true readthrough event still awaits more evidence and trans-splicing that joins two transcripts together remains as a possible mechanism. We regard those genomic loci that are transcriptionally read through as unannotated genes, because their transcriptional and posttranscriptional regulations are the same as those of already-annotated genes, including fusion genes formed due to genetic alterations. Therefore, readthrough RNAs and fusion-gene-derived RNAs are not chimeras. Only those two-gene RNAs formed at the RNA level, likely via trans-splicing, without corresponding genes as genomic parents, should be regarded as authentic chimeric RNAs. However, since in human cells, procedural and mechanistic details of trans-splicing have never been disclosed, we doubt the existence of trans-splicing. Therefore, there are probably no authentic chimeras in humans, after readthrough and fusion-gene derived RNAs are all put back into the group of ordinary RNAs. Therefore, it should be further determined whether in human cells all two-neighboring-gene RNAs are derived from transcriptional readthrough and whether trans-splicing truly exists.

Two RNAs or DNAs May Artificially Fuse Together at a Short Homologous Sequence (SHS) during Reverse Transcription or Polymerase Chain Reactions, and Thus Reporting an SHS-Containing Chimeric RNA Requires Extra Caution

Tens of thousands of chimeric RNAs have been reported. Most of them contain a short homologous sequence (SHS) at the joining site of the two partner genes but are not associated with a fusion gene. We hypothesize that many of these chimeras may be technical artifacts derived from SHS-caused mis-priming in reverse transcription (RT) or polymerase chain reactions (PCR). We cloned six chimeric complementary DNAs (cDNAs) formed by human mitochondrial (mt) 16S rRNA sequences at an SHS, which were similar to several expression sequence tags (ESTs).These chimeras, which could not be detected with cDNA protection assay, were likely formed because some regions of the 16S rRNA are reversely complementary to another region to form an SHS, which allows the downstream sequence to loop back and anneal at the SHS to prime the synthesis of its complementary strand, yielding a palindromic sequence that can form a hairpin-like structure.We identified a 16S rRNA that ended at the 4th nucleotide(nt) of the mt-tRNA-leu was dominant and thus should be the wild type. We also cloned a mouse Bcl2-Nek9 chimeric cDNA that contained a 5-nt unmatchable sequence between the two partners, contained two copies of the reverse primer in the same direction but did not contain the forward primer, making it unclear how this Bcl2-Nek9 was formed and amplified. Moreover, a cDNA was amplified because one primer has 4 nts matched to the template, suggesting that there may be many more artificial cDNAs than we have realized, because the nuclear and mt genomes have many more 4-nt than 5-nt or longer homologues. Altogether, the chimeric cDNAs we cloned are good examples suggesting that many cDNAs may be artifacts due to SHS-caused mis-priming and thus greater caution should be taken when new sequence is obtained from a technique involving DNA polymerization.

RNA-seq of human reference RNA samples using a thermostable group II intron reverse transcriptase

Next-generation RNA sequencing (RNA-seq) has revolutionized our ability to analyze transcriptomes. Current RNA-seq methods are highly reproducible, but each has biases resulting from different modes of RNA sample preparation, reverse transcription, and adapter addition, leading to variability between methods. Moreover, the transcriptome cannot be profiled comprehensively because highly structured RNAs, such as tRNAs and snoRNAs, are refractory to conventional RNA-seq methods. Recently, we developed a new method for strand-specific RNA-seq using thermostable group II intron reverse transcriptases (TGIRTs). TGIRT enzymes have higher processivity and fidelity than conventional retroviral reverse transcriptases plus a novel template-switching activity that enables RNA-seq adapter addition during cDNA synthesis without using RNA ligase. Here, we obtained TGIRT-seq data sets for well-characterized human RNA reference samples and compared them to previous data sets obtained for these RNAs by the Illumina TruSeq v2 and v3 methods. We find that TGIRT-seq recapitulates the relative abundance of human transcripts and RNA spike-ins in ribo-depleted, fragmented RNA samples comparably to non-strand-specific TruSeq v2 and better than strand-specific TruSeq v3. Moreover, TGIRT-seq is more strand specific than TruSeq v3 and eliminates sampling biases from random hexamer priming, which are inherent to TruSeq. The TGIRT-seq data sets also show more uniform 5' to 3' gene coverage and identify more splice junctions, particularly near the 5' ends of mRNAs, than do the TruSeq data sets. Finally, TGIRT-seq enables the simultaneous profiling of mRNAs and lncRNAs in the same RNA-seq experiment as structured small ncRNAs, including tRNAs, which are essentially absent with TruSeq.

High-throughput sequencing of human plasma RNA by using thermostable group II intron reverse transcriptases

Next-generation RNA-sequencing (RNA-seq) has revolutionized transcriptome profiling, gene expression analysis, and RNA-based diagnostics. Here, we developed a new RNA-seq method that exploits thermostable group II intron reverse transcriptases (TGIRTs) and used it to profile human plasma RNAs. TGIRTs have higher thermostability, processivity, and fidelity than conventional reverse transcriptases, plus a novel template-switching activity that can efficiently attach RNA-seq adapters to target RNA sequences without RNA ligation. The new TGIRT-seq method enabled construction of RNA-seq libraries from <1 ng of plasma RNA in <5 h. TGIRT-seq of RNA in 1-mL plasma samples from a healthy individual revealed RNA fragments mapping to a diverse population of protein-coding gene and long ncRNAs, which are enriched in intron and antisense sequences, as well as nearly all known classes of small ncRNAs, some of which have never before been seen in plasma. Surprisingly, many of the small ncRNA species were present as full-length transcripts, suggesting that they are protected from plasma RNases in ribonucleoprotein (RNP) complexes and/or exosomes. This TGIRT-seq method is readily adaptable for profiling of whole-cell, exosomal, and miRNAs, and for related procedures, such as HITS-CLIP and ribosome profiling.

Hypothesis: Artifacts, Including Spurious Chimeric RNAs with a Short Homologous Sequence, Caused by Consecutive Reverse Transcriptions and Endogenous Random Primers

Recent RNA-sequencing technology and associated bioinformatics have led to identification of tens of thousands of putative human chimeric RNAs, i.e. RNAs containing sequences from two different genes, most of which are derived from neighboring genes on the same chromosome. In this essay, we redefine "two neighboring genes" as those producing individual transcripts, and point out two known mechanisms for chimeric RNA formation, i.e. transcription from a fusion gene or trans-splicing of two RNAs. By our definition, most putative RNA chimeras derived from canonically-defined neighboring genes may either be technical artifacts or be cis-splicing products of 5'- or 3'-extended RNA of either partner that is redefined herein as an unannotated gene, whereas trans-splicing events are rare in human cells. Therefore, most authentic chimeric RNAs result from fusion genes, about 1,000 of which have been identified hitherto. We propose a hypothesis of "consecutive reverse transcriptions (RTs)", i.e. another RT reaction following the previous one, for how most spurious chimeric RNAs, especially those containing a short homologous sequence, may be generated during RT, especially in RNA-sequencing wherein RNAs are fragmented. We also point out that RNA samples contain numerous RNA and DNA shreds that can serve as endogenous random primers for RT and ensuing polymerase chain reactions (PCR), creating artifacts in RT-PCR.

Identification, molecular cloning, and analysis of full-length hepatitis C virus transmitted/founder genotypes 1, 3, and 4

Hepatitis C virus (HCV) infection is characterized by persistent replication of a complex mixture of viruses termed a “quasispecies.” Transmission is generally associated with a stringent population bottleneck characterized by infection by limited numbers of “transmitted/founder” (T/F) viruses. Characterization of T/F genomes of human immunodeficiency virus type 1 (HIV-1) has been integral to studies of transmission, immunopathogenesis, and vaccine development. Here, we describe the identification of complete T/F genomes of HCV by single-genome sequencing of plasma viral RNA from acutely infected subjects. A total of 2,739 single-genome-derived amplicons comprising 10,966,507 bp from 18 acute-phase and 11 chronically infected subjects were analyzed. Acute-phase sequences diversified essentially randomly, except for the poly(U/UC) tract, which was subject to polymerase slippage. Fourteen acute-phase subjects were productively infected by more than one genetically distinct virus, permitting assessment of recombination between replicating genomes. No evidence of recombination was found among 1,589 sequences analyzed. Envelope sequences of T/F genomes lacked transmission signatures that could distinguish them from chronic infection viruses. Among chronically infected subjects, higher nucleotide substitution rates were observed in the poly(U/UC) tract than in envelope hypervariable region 1. Fourteen full-length molecular clones with variable poly(U/UC) sequences corresponding to seven genotype 1a, 1b, 3a, and 4a T/F viruses were generated. Like most unadapted HCV clones, T/F genomes did not replicate efficiently in Huh 7.5 cells, indicating that additional cellular factors or viral adaptations are necessary for in vitro replication. Full-length T/F HCV genomes and their progeny provide unique insights into virus transmission, virus evolution, and virus-host interactions associated with immunopathogenesis.

Hepatitis C virus (HCV) infects 2% to 3% of the world’s population and exhibits extraordinary genetic diversity. This diversity is mirrored by HIV-1, where characterization of transmitted/founder (T/F) genomes has been instrumental in studies of virus transmission, immunopathogenesis, and vaccine development. Here, we show that despite major differences in genome organization, replication strategy, and natural history, HCV (like HIV-1) diversifies essentially randomly early in infection, and as a consequence, sequences of actual T/F viruses can be identified. This allowed us to capture by molecular cloning the full-length HCV genomes that are responsible for infecting the first hepatocytes and eliciting the initial immune responses, weeks before these events could be directly analyzed in human subjects. These findings represent an enabling experimental strategy, not only for HCV and HIV-1 research, but also for other RNA viruses of medical importance, including West Nile, chikungunya, dengue, Venezuelan encephalitis, and Ebola viruses.

Characterization of major histocompatibility complex class I allele polymorphisms in common marmosets

Currently, little information is available for major histocompatibility complex (MHC)-I that conditions the T-cell response of marmosets. In this study, 471 clones of MHC-I cDNA sequences were isolated from 12 marmosets. Twenty full-length sequences of class I G (Caja-G) alleles were obtained from these marmosets, 15 of them were novel. Among these 20 Caja-G alleles, 10 were found in individual animals while the rests were in two to four marmosets, but none was common to all animals. Ten marmosets possessed one to three Caja-G alleles, and two marmosets carried five or six alleles, which suggested that the Caja-G locus was duplicated in marmoset's genome. The high polymorphisms of Caja-G sequences provided important information helpful for understanding the cellular immune response in virus-infected marmosets.

Genetic diversity and evolution of satellite RNAs associated with the bamboo mosaic virus

Satellite RNAs (satRNAs) are subviral agents that depend on cognate helper viruses for genome replication and encapsidation. Their negative impacts on helper viruses have been exploited to control plant viral diseases. SatBaMV is a commonly found satRNA associated with Bamboo mosaic virus (BaMV) that infects diverse bamboo species in the field. To investigate the genetic diversity and evolution of satRNAs, we examined seven satBaMV populations derived from five bamboo species and cultivars from Taiwan, China, and India and one from the greenhouse. We found 3 distinct clades among the seven populations. Clade I is consisted of all satBaMV isolates, except for those from Dendrocalamus latiflorus in Taiwan and Bambusa vulgaris in India, which belong to Clades II and III, respectively. Interestingly, nucleotide diversity was lower for Clade I than II and III. However, the nucleotide diversity did not seem to depend on bamboo species or geographic location. Our population genetic analyses revealed the presence of excessive low-frequency polymorphic sites, which suggests that the satBaMV population was under purifying selection and/or population expansion. Further analysis of P20, the only satBaMV gene that encodes a non-structural protein involved in the long-distance movement of satBaMV, showed evidence of purifying selection. Taken together, our results suggest that purifying selection against defective P20 protein is responsible at least in part for the evolution of the satBaMV genome.

A comprehensive survey of non-canonical splice sites in the human transcriptome

We uncovered the diversity of non-canonical splice sites at the human transcriptome using deep transcriptome profiling. We mapped a total of 3.7 billion human RNA-seq reads and developed a set of stringent filters to avoid false non-canonical splice site detections. We identified 184 splice sites with non-canonical dinucleotides and U2/U12-like consensus sequences. We selected 10 of the herein identified U2/U12-like non-canonical splice site events and successfully validated 9 of them via reverse transcriptase-polymerase chain reaction and Sanger sequencing. Analyses of the 184 U2/U12-like non-canonical splice sites indicate that 51% of them are not annotated in GENCODE. In addition, 28% of them are conserved in mouse and 76% are involved in alternative splicing events, some of them with tissue-specific alternative splicing patterns. Interestingly, our analysis identified some U2/U12-like non-canonical splice sites that are converted into canonical splice sites by RNA A-to-I editing. Moreover, the U2/U12-like non-canonical splice sites have a differential distribution of splicing regulatory sequences, which may contribute to their recognition and regulation. Our analysis provides a high-confidence group of U2/U12-like non-canonical splice sites, which exhibit distinctive features among the total human splice sites.

Deep sequencing analysis of defective genomes of parainfluenza virus 5 and their role in interferon induction

Preparations of parainfluenza virus 5 (PIV5) that are potent activators of the interferon (IFN) induction cascade were generated by high-multiplicity passage in order to accumulate defective interfering virus genomes (DIs). Nucleocapsid RNA from these virus preparations was extracted and subjected to deep sequencing. Sequencing data were analyzed using methods designed to detect internal deletion and "copyback" DIs in order to identify and characterize the different DIs present and to approximately quantify the ratio of defective to nondefective genomes. Trailer copybacks dominated the DI populations in IFN-inducing preparations of both the PIV5 wild type (wt) and PIV5-VΔC (a recombinant virus that does not encode a functional V protein). Although the PIV5 V protein is an efficient inhibitor of the IFN induction cascade, we show that nondefective PIV5 wt is unable to prevent activation of the IFN response by coinfecting copyback DIs due to the interfering effects of copyback DIs on nondefective virus protein expression. As a result, copyback DIs are able to very rapidly activate the IFN induction cascade prior to the expression of detectable levels of V protein by coinfecting nondefective virus.

New methods as alternative or corrective measures for the pitfalls and artifacts of reverse transcription and polymerase chain reactions (RT-PCR) in cloning chimeric or antisense-accompanied RNA

We established new methods for cloning cDNA ends that start with reverse transcription (RT) and soon proceed with the synthesis of the second cDNA strand, avoiding manipulations of fragile RNA. Our 3′-end cloning method does not involve poly-dT primers and polymerase chain reactions (PCR), is low in efficiency but high in fidelity and can clone those RNAs without a poly-A tail. We also established a cDNA protection assay to supersede RNA protection assay. The protected cDNA can be amplified, cloned and sequenced, enhancing sensitivity and fidelity. We report that RT product using gene-specific primer (GSP) cannot be gene- or strand-specific because RNA sample contains endogenous random primers (ERP). The gene-specificity may be improved by adding a linker sequence at the 5′-end of the GSP to prime RT and using the linker as a primer in the ensuing PCR. The strand-specificity may be improved by using strand-specific DNA oligos in our protection assay. The CDK4 mRNA and TSPAN31 mRNA are transcribed from the opposite DNA strands and overlap at their 3′ ends. Using this relationship as a model, we found that the overlapped sequence might serve as a primer with its antisense as the template to create a wrong-template extension in RT or PCR. We infer that two unrelated RNAs or cDNAs overlapping at the 5′- or 3′-end might create a spurious chimera in this way, and many chimeras with a homologous sequence may be such artifacts. The ERP and overlapping antisense together set complex pitfalls, which one should be aware of.

Strand-specific libraries for high throughput RNA sequencing (RNA-Seq) prepared without poly(A) selection

BACKGROUND

High throughput DNA sequencing technology has enabled quantification of all the RNAs in a cell or tissue, a method widely known as RNA sequencing (RNA-Seq). However, non-coding RNAs such as rRNA are highly abundant and can consume >70% of sequencing reads. A common approach is to extract only polyadenylated mRNA; however, such approaches are blind to RNAs with short or no poly(A) tails, leading to an incomplete view of the transcriptome. Another challenge of preparing RNA-Seq libraries is to preserve the strand information of the RNAs.

DESIGN

Here, we describe a procedure for preparing RNA-Seq libraries from 1 to 4 μg total RNA without poly(A) selection. Our method combines the deoxyuridine triphosphate (dUTP)/uracil-DNA glycosylase (UDG) strategy to achieve strand specificity with AMPure XP magnetic beads to perform size selection. Together, these steps eliminate gel purification, allowing a library to be made in less than two days. We barcode each library during the final PCR amplification step, allowing several samples to be sequenced in a single lane without sacrificing read length. Libraries prepared using this protocol are compatible with Illumina GAII, GAIIx and HiSeq 2000 platforms.

DISCUSSION

The RNA-Seq protocol described here yields strand-specific transcriptome libraries without poly(A) selection, which provide approximately 90% mappable sequences. Typically, more than 85% of mapped reads correspond to protein-coding genes and only 6% derive from non-coding RNAs. The protocol has been used to measure RNA transcript identity and abundance in tissues from flies, mice, rats, chickens, and frogs, demonstrating its general applicability.

The role of canonical and noncanonical pre-mRNA splicing in plant stress responses

Plants are sessile organisms capable of adapting to various environmental constraints, such as high or low temperatures, drought, soil salinity, or pathogen attack. To survive the unfavorable conditions, plants actively employ pre-mRNA splicing as a mechanism to regulate expression of stress-responsive genes and reprogram intracellular regulatory networks. There is a growing evidence that various stresses strongly affect the frequency and diversity of alternative splicing events in the stress-responsive genes and lead to an increased accumulation of mRNAs containing premature stop codons, which in turn have an impact on plant stress response. A number of studies revealed that some mRNAs involved in plant stress response are spliced counter to the traditional conception of alternative splicing. Such noncanonical mRNA splicing events include trans-splicing, intraexonic deletions, or variations affecting multiple exons and often require short direct repeats to occur. The noncanonical alternative splicing, along with common splicing events, targets the spliced transcripts to degradation through nonsense-mediated mRNA decay or leads to translation of truncated proteins. Investigation of the diversity, biological consequences, and mechanisms of the canonical and noncanonical alternative splicing events will help one to identify those transcripts which are promising for using in genetic engineering and selection of stress-tolerant plants.

Complexity of bidirectional transcription and alternative splicing at human RCAN3 locus

Human RCAN3 (regulator of calcineurin 3) belongs to the human RCAN gene family.

In this study we provide, with in silico and in vitro analyses, the first detailed description of the human multi-transcript RCAN3 locus. Its analysis revealed that it is composed of a multigene system that includes at least 21 RCAN3 alternative spliced isoforms (16 of them identified here for the first time) and a new RCAN3 antisense gene (RCAN3AS). In particular, we cloned RCAN3-1,3,4,5 (lacking exon 2), RCAN3-1a,2,3,4,5, RCAN3-1a,3,4,5, RCAN3-1b,2,3,4,5, RCAN3-1c,2,3,4,5, RCAN3-1c,2,4,5 and RCAN3-1c,3,4,5, isoforms that present a different 5′ untranslated region when compared to RCAN3. Moreover, in order to verify the possible 5′ incompleteness of previously identified cDNA isoforms with the reference exon 1, ten more alternative isoforms were retrieved. Bioinformatic searches allowed us to identify RCAN3AS, which overlaps in part with exon 1a, on the opposite strand, for which four different RCAN3AS isoforms were cloned.

In order to analyze the different expression patterns of RCAN3 alternative first exons and of RCAN3AS mRNA isoforms, RT-PCR was performed in 17 human tissues.

Finally, analyses of RCAN3 and RCAN3AS genomic sequences were performed to identify possible promoter regions, to examine donor and acceptor splice sequences and to compare evolutionary conservation, in particular of alternative exon 1 or 1c - exon 2 junctions in different species.

The description of its number of transcripts, of their expression patterns and of their regulatory regions can be important to clarify the functions of RCAN3 gene in different pathways and cellular processes.

Single-molecule direct RNA sequencing without cDNA synthesis

Methods for in-depth genome-wide characterization of transcriptomes and quantification of transcript levels using various microarray and next-generation sequencing technologies have emerged as valuable tools for understanding cellular physiology and human disease biology and have begun to be utilized in various clinical diagnostic applications. Current methods, however, typically require RNA to be converted to complementary DNA prior to measurements. This step has been shown to introduce many biases and artifacts. In order to best characterize the 'true' transcriptome, the single-molecule direct RNA sequencing (DRS) technology was developed. This review focuses on the underlying principles behind the DRS, sample preparation steps, and the current and novel avenues of research and applications DRS offers.

Transcriptome profiling using single-molecule direct RNA sequencing

Methods for in-depth characterization of transcriptomes and quantification of transcript levels have emerged as valuable tools for understanding cellular physiology and human disease biology, and have begun to be utilized in various clinical diagnostic applications. Today, current methods utilized by the scientific community typically require RNA to be converted to cDNA prior to comprehensive measurements. However, this cDNA conversion process has been shown to introduce many biases and artifacts that interfere with the proper characterization and quantitation of transcripts. We have developed a direct RNA sequencing (DRS) approach, in which, unlike other technologies, RNA is sequenced directly without prior conversion to cDNA. The benefits of DRS include the ability to use minute quantities (e.g. on the order of several femtomoles) of RNA with minimal sample preparation, the ability to analyze short RNAs which pose unique challenges for analysis using cDNA-based approaches, and the ability to perform these analyses in a low-cost and high-throughput manner. Here, we describe the strategies and procedures we employ to prepare various RNA species for analysis with DRS.

Whole transcriptome analysis: what are we still missing?

New technologies such as tag-based sequencing and tiling arrays have provided unique insights into the transcriptional output of cells. Many new RNA classes have been uncovered in the past decade, despite limitations in current technologies. Even as the repertoire of known functional elements of the transcriptome increases and contemporary technologies become mainstream, inadequacies in conventional protocols for library preparation, sequencing and mapping continue to hamper revelation of the entire transcriptome of cells. In this article, we review current protocols and outline their deficiencies. We also provide our view on what we may be overlooking in the transcriptome, despite exhaustive investigations, and indicate future areas of technological development and research.

RNA sequencing: advances, challenges and opportunities

In the few years since its initial application, massively parallel cDNA sequencing, or RNA-seq, has allowed many advances in the characterization and quantification of transcriptomes. Recently, several developments in RNA-seq methods have provided an even more complete characterization of RNA transcripts. These developments include improvements in transcription start site mapping, strand-specific measurements, gene fusion detection, small RNA characterization and detection of alternative splicing events. Ongoing developments promise further advances in the application of RNA-seq, particularly direct RNA sequencing and approaches that allow RNA quantification from very small amounts of cellular materials.

A critical analysis of Atoh7 (Math5) mRNA splicing in the developing mouse retina

The Math5 (Atoh7) gene is transiently expressed during retinogenesis by progenitors exiting mitosis, and is essential for ganglion cell (RGC) development. Math5 contains a single exon, and its 1.7 kb mRNA encodes a 149-aa polypeptide. Mouse Math5 mutants have essentially no RGCs or optic nerves. Given the importance of this gene in retinal development, we thoroughly investigated the possibility of Math5 mRNA splicing by Northern blot, 3'RACE, RNase protection assays, and RT-PCR, using RNAs extracted from embryonic eyes and adult cerebellum, or transcribed in vitro from cDNA clones. Because Math5 mRNA contains an elevated G+C content, we used graded concentrations of betaine, an isostabilizing agent that disrupts secondary structure. Although approximately 10% of cerebellar Math5 RNAs are spliced, truncating the polypeptide, our results show few, if any, spliced Math5 transcripts exist in the developing retina (<1%). Rare deleted cDNAs do arise via RT-mediated RNA template switching in vitro, and are selectively amplified during PCR. These data differ starkly from a recent study (Kanadia and Cepko 2010), which concluded that the vast majority of Math5 and other bHLH transcripts are spliced to generate noncoding RNAs. Our findings clarify the architecture of the Math5 gene and its mechanism of action. These results have implications for all members of the bHLH gene family, for any gene that is alternatively spliced, and for the interpretation of all RT-PCR experiments.

Apparent non-canonical trans-splicing is generated by reverse transcriptase in vitro

BACKGROUND

Trans-splicing, the in vivo joining of two independently transcribed RNA molecules, is well characterized in lower eukaryotes, but was long thought absent from metazoans. However, recent bioinformatic analyses of EST sequences suggested widespread trans-splicing in mammals. These apparently spliced transcripts generally lacked canonical splice sites, leading us to question their authenticity. Particularly, the native ability of reverse transcriptase enzymes to template switch during transcription could produce apparently trans-spliced sequences.

PRINCIPAL FINDINGS

Here we report an in vitro system for the analysis of template switching in reverse transcription. Using highly purified RNA substrates, we show the reproducible occurrence of apparent trans-splicing between two RNA molecules. Other reported non-canonical splicing events such as exon shuffling and sense-antisense fusions were also readily detected. The latter caused the production of apparent antisense non-coding RNAs, which are also reported to be abundant in humans.

CONCLUSIONS

We propose that most reported examples of non-canonical splicing in metazoans arise through template switching by reverse transcriptase during cDNA preparation. We further show that the products of template switching can vary between reverse transcriptases, providing a simple diagnostic for identifying many of these experimental artifacts.

A variant of the KLK4 gene is expressed as a cis sense-antisense chimeric transcript in prostate cancer cells

In humans, more than 30,000 chimeric transcripts originating from 23,686 genes have been identified. The mechanisms and association of chimeric transcripts arising from chromosomal rearrangements with cancer are well established, but much remains unknown regarding the biogenesis and importance of other chimeric transcripts that arise from nongenomic alterations. Recently, a SLC45A3-ELK4 chimera has been shown to be androgen-regulated, and is overexpressed in metastatic or high-grade prostate tumors relative to local prostate cancers. Here, we characterize the expression of a KLK4 cis sense-antisense chimeric transcript, and show other examples in prostate cancer. Using non-protein-coding microarray analyses, we initially identified an androgen-regulated antisense transcript within the 3' untranslated region of the KLK4 gene in LNCaP cells. The KLK4 cis-NAT was validated by strand-specific linker-mediated RT-PCR and Northern blotting. Characterization of the KLK4 cis-NAT by 5' and 3' rapid amplification of cDNA ends (RACE) revealed that this transcript forms multiple fusions with the KLK4 sense transcript. Lack of KLK4 antisense promoter activity using reporter assays suggests that these transcripts are unlikely to arise from a trans-splicing mechanism. 5' RACE and analyses of deep sequencing data from LNCaP cells treated +/-androgens revealed six high-confidence sense-antisense chimeras of which three were supported by the cDNA databases. In this study, we have shown complex gene expression at the KLK4 locus that might be a hallmark of cis sense-antisense chimeric transcription.

Direct RNA sequencing

Our understanding of human biology and disease is ultimately dependent on a complete understanding of the genome and its functions. The recent application of microarray and sequencing technologies to transcriptomics has changed the simplistic view of transcriptomes to a more complicated view of genome-wide transcription where a large fraction of transcripts emanates from unannotated parts of genomes, and underlined our limited knowledge of the dynamic state of transcription. Most of this broad body of knowledge was obtained indirectly because current transcriptome analysis methods typically require RNA to be converted to complementary DNA (cDNA) before measurements, even though the cDNA synthesis step introduces multiple biases and artefacts that interfere with both the proper characterization and quantification of transcripts. Furthermore, cDNA synthesis is not particularly suitable for the analysis of short, degraded and/or small quantity RNA samples. Here we report direct single molecule RNA sequencing without prior conversion of RNA to cDNA. We applied this technology to sequence femtomole quantities of poly(A)(+) Saccharomyces cerevisiae RNA using a surface coated with poly(dT) oligonucleotides to capture the RNAs at their natural poly(A) tails and initiate sequencing by synthesis. We observed transcript 3' end heterogeneity and polyadenylated small nucleolar RNAs. This study provides a path to high-throughput and low-cost direct RNA sequencing and achieving the ultimate goal of a comprehensive and bias-free understanding of transcriptomes.

When good transcripts go bad: artifactual RT-PCR 'splicing' and genome analysis

Gene and intron prediction are essential for accurate inferences about genome evolution. Recently, two genome-wide studies searched for recent intron gains in humans, reaching very different conclusions: either of a complete absence of intron gain since early mammalian evolution, or of creation of numerous introns by genomic duplication in repetitive regions. We discuss one possible explanation: the underappreciated phenomenon of "template switching", by which reverse transcriptase may create artifactual splicing-like events in the preparation of cDNA/EST libraries, may cause complications in searches for newly gained introns in repetitive regions. We report large numbers of apparent template switching in transcript sequences from the intron-poor protists Trichomonas vaginalis and Giardia lamblia. Supplementary material for this article can be found on the BioEssays website (http://www.interscience.wiley.com/jpages/0265-9247/suppmat/index.html).

A multicenter analysis of the FIP1L1-αPDGFR fusion gene in Japanese idiopathic hypereosinophilic syndrome: an aberrant splicing skipping the αPDGFR exon 12

To study the clinical characteristics of hypereosionophilic syndrome and chronic eosinophilic leukemia (HES/CEL) in Japan, the clinical data of 29 HES/CEL patients throughout the country were surveyed. Moreover, the involvement of the FIP1L1-alphaPDGFR fusion gene resulting from a cryptic del (4)(q12q12) was examined in 24 cases. The FIP1L1-alphaPDGFR messenger RNA (mRNA) was detected in three patients (13% of patients fulfilled WHO criteria and 17% of Chusid criteria). One had a novel fusion transcript, which skipped the exon 12 of alphaPDGFR. The transcript appears to be generated by a splicing mechanism that is different from the previously reported splicing patterns. In silico analysis, the exon skipping was not related to a disruption of the exonic splicing enhancers within the exon but strongly associated with the loss of the vast majority of the FIP1L intron 8a where intronic splicing enhancers were accumulated. Unexpectedly, pseudo-chimera DNA fragments with some shared characteristic features were occasionally generated from healthy control samples by reverse transcriptase polymerase chain reaction (RT-PCR). Considering the relatively low incidence of the FIP1L1-alphaPDGFR transcript positive case, extreme care must therefore be taken when making a diagnosis using RT-PCR before imatinib therapy.

Expression pattern analysis of transcribed HERV sequences is complicated by ex vivo recombination

Background

The human genome comprises numerous human endogenous retroviruses (HERVs) that formed millions of years ago in ancestral species. A number of loci of the HERV-K(HML-2) family are evolutionarily much younger. A recent study suggested an infectious HERV-K(HML-2) variant in humans and other primates. Isolating such a variant from human individuals would be a significant finding for human biology.

Results

When investigating expression patterns of specific HML-2 proviruses we encountered HERV-K(HML-2) cDNA sequences without proviral homologues in the human genome, named HERV-KX, that could very well support recently suggested infectious HML-2 variants. However, detailed sequence analysis, using the software RECCO, suggested that HERV-KX sequences were produced by recombination, possibly arising ex vivo, between transcripts from different HML-2 proviral loci.

Conclusion

As RT-PCR probably will be instrumental for isolating an infectious HERV-K(HML-2) variant, generation of "new" HERV-K(HML-2) sequences by ex vivo recombination seems inevitable. Further complicated by an unknown amount of allelic sequence variation in HERV-K(HML-2) proviruses, newly identified HERV-K(HML-2) variants should be interpreted very cautiously.

A PCR-based method for detection and quantification of small RNAs

Recent cloning efforts have identified hundreds of thousands of small RNAs including micro RNAs (miRNAs), Piwi-interacting RNAs (piRNAs), and small nucleolar RNAs (snoRNAs). These non-coding small RNAs need to be further validated and characterized by detecting and quantifying their expression in different tissues and during different developmental courses. A simple, accurate, and sensitive method for small RNA expression profiling is in high demand. Here, we report such a PCR-based method.

Template switching within exons 3 and 4 of KV11.1 (HERG) gives rise to a 5' truncated cDNA

K(V)11.1 (HERG) channels contribute to membrane potential in a number of excitable cell types. We cloned a variant of K(V)11.1 from human jejunum containing a 171 bp deletion spanning exons 3 and 4. Expression of a full-length cDNA clone containing this deletion gave rise to protein that trafficked to the cell membrane and generated robust currents. The deletion occurred in a G/C-rich region and identical sequence elements of UGGUGG were located at the deletion boundaries. In recent studies these features have been implicated to cause deletions via template switching during cDNA synthesis. To examine this possibility we compared cDNAs from human brain, heart, and jejunum synthesized at lower (42 degrees C) and higher temperatures (70 degrees C). The 171 bp deletion was absent at the higher temperature. Our results suggest that the sequence and secondary structure of mRNA in the G/C rich region leads to template switching producing a cDNA product with a 171 bp deletion.

Reverse transcriptase template switching and false alternative transcripts

Reverse transcriptase (RT) can switch from one template to another in a homology-dependent manner. In the study of eukaryotic transcripts, this propensity of RT can produce an artificially deleted cDNA, which can be wrongly interpreted as an alternative transcript. Here, we have investigated the presence of such template-switching artifacts in cDNA databases, by scanning a collection of human splice sites (Information for the Coordinates of Exons, ICE database). We have confirmed several cases at the experimental level. Artifacts represent a significant portion of apparently spliced sequences using noncanonical splice signals but are rare in the context of the whole database. However, care should be taken in the annotation of alternative transcripts, especially when the RT used is poorly thermostable and when the putative intron is flanked by direct repeats, which are the substrate for template switching.

Generation of aberrant transcripts of and free DNA ends in zebrafish no tail gene

The zebrafish no tail gene (ntl) is indispensable for the formation of the notochord and the tail structure. In a wild-type zebrafish population, we occasionally observed adult zebrafish with a narrow or no tailfin. This led us to examine the hypothesis that the activity of ntl was somehow genetically unstable. Here we present two findings regarding the gene. First, approximately 3% of ntl transcripts were aberrant; most of them carried deletions at various positions. Second, free, DNA double-stranded ends (DSEs) were formed at an AT dinucleotide repeat in ntl. DSEs were also generated in another zebrafish gene, noggin2 (nog2). DSEs in ntl and nog2 had common characteristics, which suggested that the AT repeats in these genes elicited DSEs by blocking progression of the replication.

Analysis of mRNA transcripts from the NAD(P)H oxidase 1 (Nox1) gene. Evidence against production of the NADPH oxidase homolog-1 short (NOH-1S) transcript variant

Recent reports indicate that NAD(P)H oxidase 1 (Nox1) mRNA undergoes alternative splicing, producing a short transcript (NOH-1S) encoding a novel H+ channel. Although the H+ transport properties of NOH-1S-transfected cells resemble those of many cells, the production of a NOH-1S protein was never documented. We characterized Nox1 transcripts in colon-derived cells and present evidence that mRNA splicing does not produce NOH-1S; rather, NOH-1S appears to be an artifact of template switching during cDNA synthesis. The NOH-1S transcript was not observed by Northern blotting, despite claims of its abundance based on RNase protection assays. The shortened cDNA was generated by avian myeloblastosis virus reverse transcriptase, but not by thermally stable reverse transcriptase under conditions that produce full-length Nox1. Analysis of shortened cDNAs detected NOH-1S sequence and other variants that differ at the alleged splice junction site. Although no appropriate RNA splicing sites were found within Nox1 to account for NOH-1S formation, we found repetitive sequence elements bordering the deleted region, which could promote intramolecular template switching during cDNA synthesis. Template switching was confirmed in vitro, where the deleted cDNA was generated by avian myeloblastosis virus reverse transcriptase from a synthetic, full-length Nox1 RNA template. A survey of the expressed sequence tags database suggests that similar switching phenomena occur between repetitive elements in other Nox family transcripts, indicating such cloning artifacts are common. In contrast, genuine RNA splicing does account for another Nox1 transcript lacking the entire exon 11, which is abundant in colon cells but encodes a protein incapable of supporting superoxide production.

Inverted Gcg/CGC trinucleotide microsatellites in the 5'-region of Mus IDS mRNA: recurrent induction of aberrant reverse transcripts

An investigation was initiated to explore previously published results indicating that approximately 80 bp of the 5'-end of the iduronate sulfatase (IDS) cDNA sequence (Accession No L07291) are 100% homologous with the 3'-UTR of isoform I of the sodium hydrogen exchanger (Acc. No. U51112). 5'-RACE carried out on IDS mRNA demonstrated the apparent homology to be a cloning artifact. A sequence comparison of the IDS 5'-RACE product with a mouse BAC clone covering the region, and with various IDS ESTs, suggested that the region is highly susceptible to cloning artifacts, a common one of which is template switching by reverse transcriptase. The nucleotide sequence flanking the translation start site is unusual in containing two inverted repeats composed of the complementary trinucleotide microsatellites, (GCG)9 and (CGC)6. These likely form a highly stable stem of 20-21 nt, through which reverse transcription is compromised. Such a stem could be involved in the regulation of IDS expression by directly affecting translation, message turnover, or serving as a substrate for siRNA production. Though such mRNA features are relatively rare, they may be more abundant but overlooked due to difficulties in their reverse transcription.

Detection of a novel sense-antisense RNA-hybrid structure by RACE experiments on endogenous troponin I antisense RNA

Conformational changes in the troponin/tropomyosin complex significantly alter the mechanical properties of cardiac muscle. Phosphorylation of cardiac troponin I, part of the troponin/tropomyosin complex, reduces calcium affinity, which leads to increased relaxation of cardiac muscle. Because cardiac troponin I plays a central role in tuning the heart to different work demands, detailed knowledge of troponin I protein regulation is required. Our group previously detected naturally occurring antisense RNA for troponin I in human and rat hearts, and here, attempt to unravel the structure of rat cardiac troponin I antisense RNA. We performed rapid amplification of cDNA ends (RACE) experiments and discovered antisense sequences identical to a copy of the sense mRNA, which led us to conclude that the antisense RNA must be transcribed from troponin I mRNA in the cytoplasm. Moreover, we isolated RNA structures comprising sense and antisense sequences in one continuous molecule. As we found no homolog structures described in the literature, we called this "hybrid RNA." Because a duplex formation was demonstrated previously we concluded that hybrid RNA is a consequence of a tight interaction between sense and antisense troponin I RNA in vivo, which we discuss in the article.

Genetic alterations of the HCCS1 gene in Korean hepatocellular carcinoma

We analyzed the gene mutations and loss of heterozygosity (LOH) of the HCCS1 gene using intragenic polymorphic markers in a series of 88 primary HCCs. We found two sequence variations at exon 5 and 14 in both normal and tumor DNAs of case 50 and 51, respectively. The variation in case 50 led to a reading frameshift and a premature stop (TGA) at codon 125 and case 51 showed amino acid change at codon 448 (Val-->Ala, GTG-->GCG). Interestingly, these variations were not found in peripheral lymphocytes of 69 normal individuals and 227 cancer patients (86 HCC, 75 unselected gastric cancer, and 66 breast cancer), suggesting that these two variations are mutation, not polymorphism. In addition, we found 14 novel intragenic polymorphic sites in the HCCS1 gene. Thirty-two (47%) of sixty-eight informative cases showed allelic loss at at least one or more intragenic polymorphic sites, but there was no significant relationship between the frequency of LOH and clinicopathologic parameters. These results suggest that mutation of the HCCS1 gene might not be a main inactivation mechanism in the development of Korean HCC and that the HCCS1 gene might be involved in acceleration of the tumorigenic process in Korean HCC.

Lymphoma- and leukemia-associated chromosomal translocations in healthy individuals

Chromosomal translocations (CTs) are hallmark mutations of hematopoietic malignancy that result in the deregulated expression of oncogenes or the generation of novel fusion genes. The polymerase chain reaction (PCR) can be used to detect illegitimate recombinations of genomic DNA sequences as a more sensitive assay than cytogenetics for determining the presence of CTs. Both direct DNA-PCR and reverse transcriptase-PCR were used to examine healthy individuals for lymphoma- and leukemia-associated CTs. Two oncogene-activating CTs [t(14;18)(q32;q21) and t(8;14)(q24;q32)] and one fusion-gene CT [t(2;5)(p23;q35)] from lymphomas and five fusion-gene CTs from leukemia [t(9;22)(q34;q11), t(4;11)(q21;q23), t(15;17)(q22;q11), t(12;21)(p13;q22), t(8;21)(q22;q22)] were detected in such studies. The biological implication is that CTs associated with malignant tumors may also be found in cells that are not neoplastic. CTs are characteristic attributes of neoplastic clones but are by themselves insufficient to cause malignant transformation. A better understanding of the special biology of non-neoplastic CT-bearing cells will provide insight into their putative role as tumor precursors. Prospective epidemiological studies are needed to determine whether such cells in healthy individuals may, in some instances, become clonogenic founders of lymphoma or leukemia.

AXIN1 mutations but not deletions in cerebellar medulloblastomas

Medulloblastoma is a malignant, invasive embryonal tumour of the cerebellum which manifests preferentially in children. A subset of cases is associated with colon cancer and APC germline mutations (Turcot syndrome), and APC and beta-catenin point mutations occur in up to 10% of sporadic cases, indicating the involvement of the Wnt pathway in the development of medulloblastoma. In 39 sporadic cerebellar medulloblastomas screeened for alterations in the AXIN1 gene, another component of the Wnt pathway, we found missense AXIN1 mutations in two tumours, CCC-->TCC at codon 255 (exon 1, Pro-->Ser) and TCT-->TGT at codon 263 (exon 1, Ser-->Cys). Furthermore, the A allele at the G/A polymorphism at nucleotide 16 in intron 4 was significantly over-represented in medulloblastomas (39 cases; G 0.76 vs-A 0.24) compared to healthy individuals (86 cases; G 0.91 vs A 0.09; P=0.0027). RT-PCR revealed large deletions in the AXIN1 gene in 5/12 (42%) medulloblastomas, consistent with a previous report. However, we observed such deletions at a similar frequency also in normal brain tissue (6/12, 50%). Since there are multiple complementary, inverted sequences present in the AXIN1 gene, these large deletions may represent RT-PCR errors due to stem-loop secondary structures.