2.3: Prepare RNA by IVT - Biology

2.3: Prepare RNA by IVT - Biology

We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.


So far you have prepared the DNA encoding both the 6-5 and the 8-12 aptamer fragments. However, it is the secondary structure of the RNA that actually allows the 8-12 sequence to bind to heme. In order to make RNA from DNA, you will perform an in vitro transcription (IVT) reaction.

What is needed to create RNA from DNA? In PCR, you used a heat-stable DNA polymerase and dNTPs (deoxynucleotide triphosphates) to make your DNA. In an IVT, you will use an RNA polymerase and NTPs (nucleotide triphosphates) instead. The polymerase is derived from the T7 bacteriophage, and requires that the DNA to be copied contains the T7 promoter sequence, or TAA TAC GAC TCA CTA TAG GG. The buffer conditions are also somewhat different (check out the reagent lists at the end of each day), but both contain the important co-factor Mg2+. Finally, an IVT contains pyrophosphatase, because it has been empirically found to increase efficiency. Pyrophosphate is produced during the IVT, so by Le Chatelier's principle, reaction completion may be improved by removing this product.

Now let's consider how an IVT fits into the overall scheme of SELEX, or systematic evolution of ligands by exponential enrichment. SELEX was simultaneously reported by two groups in 1990:

Ellington and Szostak described RNA that bound to select aromatic small molecules, while Tuerk and Gold isolated RNA that bound to a DNA polymerase. As shown in the figure at right, due to the stability of DNA one usually starts with a DNA library rather than directly with an RNA library. The DNA is copied into RNA by transcription, and the resulting RNA pool is run over an affinity column that has the ligand of interest bound to it. Aptamers that do not bind are washed away by flowing buffer through the column. Afterward, RNA that does bind to the ligand is washed off, either by using free ligand as a competitor or sometimes by other changes in buffer conditions (e.g., salt concentration or pH). This RNA pool is copied into DNA by reverse transcription, then amplified by PCR, often using a one-step kit these days. After RT-PCR, the new library (presumably containing many fewer species of DNA than the original library) is again transcribed to RNA. Now the new library may be further refined by a second column selection.

An optional but often important step is to perform a negative selection. To rid the RNA pool of aptamers that bind non-specifically to the column material, one can incubate the library with a column resin that lacks ligand but is otherwise identical to the affinity resin. After all, it would be a shame to think one has found several heme-binding aptamers, only to discover that they in fact bind to agarose beads! The supernatant from the negative selection is then used directly in the positive selection. Sometimes, multiple negative selections may be appropriate. For example, if one is interested in finding an aptamer that binds to a particular ligand but not to several similar ligands, one could perform negative selections using affinity columns presenting those those undesired ligands. No matter how many precautions one takes, the statistics of large numbers suggests that there will always be some false positives: RNA sequences that manage to get detectably amplified, but do not bind to the ligand of interest.

After one or more column selections, one may use cloning to select individual RNA species for sequencing or for testing in a functional assay. Again, ligating the DNA library into a plasmid that can be expressed in bacteria is easier than working with the RNA. DNA can be isolated from individual colonies of bacteria that each contain a single aptamer-encoding DNA, then sequenced and transcribed to RNA if desired. A typical starting library might have 10^13 to 10^15 sequences! In Szostak's early binding experiments, molecules that bound to the aromatic dyes were enriched from less than 1% to greater than 50% of the RNA library population after four rounds of selection, and the unique population was reduced to ~100-100,000 sequences. In our case, we already know the sequences of the two aptamers in question, and are simply interested in their ratio after a round of column selection. The functional assay that can indicate this ratio measures aptamer binding to heme by spectrophotometry.

The IVT will run during the whole lab period. In the meantime, we will discuss a journal article, both to learn more about RNA aptamers and to become comfortable reading and discussing the primary scientific literature. In 2-3 weeks, you will each present an article on your own.


Because you are preparing RNA, you will have to take special precautions today and for the rest of the module. RNA is strikingly different from DNA in its stability. Consequently it is more difficult to work with RNA in the lab. It is not the techniques themselves that are difficult; indeed, many of the manipulations are nearly identical to those used for DNA. However, RNA is rapidly and easily degraded by RNases that exist everywhere. There are several rules for working with RNA. They will improve your chances of success. Please follow them all.

  • Use warm water on a paper towel to wash lab equipment, like microfuges, before you begin your experiment. Then wipe them down with "RNase-away" solution.
  • Wear gloves when you are touching anything that will touch your RNA.
  • Change your gloves often.
  • Before you begin your experiment clean your work area, removing all clutter. Wipe down the benchtop with warm water then "RNase-away," and then lay down a fresh piece of benchpaper.
  • Use RNA-dedicated solutions and if possible RNA-dedicated pipetmen.
  • Get a new box of pipet tips from the RNA materials area and label their lid "RNase FREE" if the lid is not yet labeled.

Part 1: In Vitro Transcription

The table below lists the amount of each reaction component needed for an 80 μL IVT. First, you should calculate how much total IVT Master Mix to make (2 rxns plus 10% excess) and check your calculations with the teaching faculty if desired. Next, you can aliquot the appropriate amount of Master Mix into a number of eppendorf tubes, then add the relevant DNA to each labeled tube.

G7 buffer (2.5X stock)32
1N KOH4.48
T7 polymerase4

Place your reaction tubes on the 37 °C heat block and write the time in your notebook as well as on the sheet at the front bench. After 4 hours, the reactions will be frozen at – 20 °C until next time. When everyone is ready, we will begin the journal article discussion.

Part 2: Choose Column Conditions

Before leaving today, you and your partner should sign up for one of the column selection protocols listed in the following table:


Journal Article Discussion

We will discuss this paper:

  • Rusconi, C. P., et al. "Antidote-Mediated Control of an Anticoagulant Aptamer in vivo." Nature Biotechnology 22, no. 11 (November 2004): 1423-8.

Scientific papers are dense and often time-consuming to read and understand, but with practice, you will find strategies that improve your comprehension efficiency. Here's one tip to get you started: when reading newly reported results, be sure to refer to the associated figures frequently, because visual information is often easier to take in than purely verbal descriptions.

Technical Background

Several terms and assays in the Rusconi, et al. paper may be unfamiliar to you. (On that note, "assay" is just a term that means a type of measurement. For example, a cell viability assay measures the number of living cells in a sample.) Here we will provide some background on selected topics.

  • Bolus refers to a one-time injection of drug, as opposed to a continuous infusion (via an IV line).
  • Pharmacokinetics is the study of the route a drug takes in the body, both where it circulates to and how it is eliminated.
  • Pharmacodynamics is the study of the actual effects, both intended and unintended, of the drug on the recipient, and of the drug's mechanism and molecular-level kinetics.
  • Plasma is blood with the cells removed. (Serum is blood with both cells and clotting factors absent.)

Note that multiple pathways for blood clotting exist. An APTT assay measures the activity of the pathway that is dependent on the presence of clotting Factor IX. In Rusconi's work, buffer with or without aptamer was mixed with a plasma test sample, then activated with phospholipids and calcium ions. Adding an anti-coagulant should increase the time required for a clot to form in the activated mixture. The PT assay, on the other hand, measures the activity of a Factor IX-independent pathway. In this case, thromboplastin and calcium ions are used for plasma activation.

Discussion Topics


As you read the paper by Rusconi, et al., consider not only its scientific content, but also the authors' writing style (perhaps not all on one read!). Refer to our Guidelines for Writing Up Your Research Sketch out answers to the questions below (right on the paper if you wish). Your answers will not be collected, but you may be called on in discussion to share your ideas.

  • Which part of the text corresponds with an Introduction section? What is the topic and/or function of each paragraph? What purpose(s) do the citations serve?
  • Which part of the text corresponds with a Results section? Can you find one or more examples of paragraphs with effective introductory and concluding sentences, according to our Guidelines for results? Are there any parts in the Results that you think belong in the Discussion instead, according to our Guidelines for discussions and results vs. discussions?
  • Which part of the text corresponds with a Discussion section? What is the topic and/or function of each paragraph? What purpose(s) do the citations serve?


When you arrive in lab today, each group will be assigned one of the following topics to present to and discuss with the rest of the class. You should be somewhat familiar with the whole Rusconi, et al. paper by now, but will have some time in-class to refresh your memory and become the resident expert in one of the following areas.

  • Introduction and Figure 1a
    • What are the advantages of using an aptamer as an anti-coagulant drug?
    • What related work have the authors previously done, and what research gap is being addressed by this new paper?
    • Sumarize the sample and control aptamers used in this study. What is the aptamer target?
  • Figure 1, b-e
    • Describe the major findings portrayed in this figure.
    • What was the purpose of testing the aptamer against multiple animal sources?
  • Figure 2, a-c
    • Describe the major findings portrayed in this figure.
    • Why did the authors do these experiments, if they already had the data in Figure 1c?
  • Figure 2, d-f
    • Describe the major findings portrayed in this figure.
    • How did the authors distinguish between two hypotheses about the antidote's potent action?
  • Figure 3
    • Describe the major findings portrayed in this figure.
    • Briefly, what is a thrombus, and under what conditions is one likely to form? (May require research outside the paper.)
    • What precautions did the authors take to avoid bias during data collection (see Methods)?
  • Figure 4
    • Describe the major findings portrayed in this figure.
    • Briefly, what is a P-value? (May require research outside the paper.)
    • What does the "vehicle" refer to (see Methods) and what is its role?
  • Wrap-up
    • How does the aptamer's potential as a drug compare to the current state-of-the-art?
    • Briefly, search PubMed for whether Rusconi, et al. have published any further studies on the aptamers descrbied here, read the abstract(s), and summarize any progress.

For Next Time

  1. Day 4 of this module is poised to run long, so you should read the entire protocol and perhaps prepare some of your lab notebook in advance. In addition, to save time later you should prepare an automated worksheet (e.g., in Excel) that will perform the required calculations for that day. Your worksheet should do the following, and a copy must be handed in using the mock numbers provided:
    • The worksheet should accept 260 nm and 280 nm absorbance readings for RNA samples.
      • Mock numbers: 0.524, 0.255 for sample 1; 0.427, 0.212 for sample 2.
    • The first calculation should be the 260:280 ratio.
    • The next parameter that the user should be able to specify is the RNA dilution factor. Calculate what this is based on the Day 4 protocol.
      • If you cannot figure it out, use a value of 200 for now.
    • The next calculation should be the RNA concentration in μg/mL. Find the conversion factor for absorbance to RNA concentration in the Day 4 protocol.
    • Now calculate the RNA concentration in μM, assuming sample 1 is 6-5 and sample 2 is 8-12 aptamer. (Molecular weights are listed in - you guessed it! - the Day 4 protocol.)
    • The user should now be able to input the volume of RNA they have. Use 75 μL for both samples for now.
    • Calculate the total nmols of RNA in each sample.
    • Calculate how much total volume the RNA should be in to be at a concentration of 8 μM.
    • Finally, calculate how much volume needs to be added to the existing RNA to reach that concentration.
    • In a separate location on your sheet (or manually), calculate the volume of an 8 μM solution required to have exactly 1.4 nmol of RNA.
  2. Prepare a figure depicting your PCR gel results (from Day 2) with an appropriate caption. Also write the portion of your Results (about a paragraph) describing this experiment.
  3. To ensure that you are making steady progress on the computational assignment, you should begin the MEME analysis (exercise and question 1) and hand in a draft of the associated figure.
  4. Complete the first portion of the writing exercises from the Day 2 handout. Jargon #2, Readability #2, and Brevity #4 are due on Day 4; the rest of the exercises are due on Day 5.

Reagent List

  • G7 buffer (1X, final conc)
    • 200 M HEPES-KOH, pH 7.5
      • HEPES = 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
    • 2 mM spermidine
    • 40 mM dithiothreitol (DTT)
    • 30 mM MgCl2
    • 100 μg/mL bovine serum albumin (BSA)
  • 1N KOH
  • Pyrophosphatase, 100 U/mL in
    • 100 mM HEPES-KOH, pH 7.5
    • 1 mM DTT
    • 50% glycerol
  • T7 RNA polymerase
    • gene expressed in a pQE30 plasmid (discontinued, originally from Qiagen)
    • purified from E. coli

Tips for successful RNA Cleanup using the Monarch ® RNA Cleanup Kits

Successful RNA purification using the Monarch Total RNA Mini Prep Kit can easily be accomplished by following these helpful tips.

Part 1 Before Your Prep

1.1 Eliminate RNases From the Work Environment

Before your prep. It is important to work in an environment free from RNases. Always wear gloves, use RNase-free glass and plastic wear and clean your work areas. We recommend wiping your bench top with a cleaning agent such as RNaseZAP.

1.2 Store RNA Samples Properly

To maintain RNA integrity of your sample, prior to the prep, samples should be frozen at the time of harvest, or protected by a dedicated reagent such as the Monarch DNA and RNA Protection Reagent, which is included in our kit. Cultured cell pellets are aliquots of blood preserved with this protection reagent, should be mixed thoroughly by vortexing or pipetting prior to storage.

Small pieces of tissue less than 20mg may be stored directly in the protection reagent. However, larger tissue samples should be homogenized in protection reagent prior to storage. If samples have been preserved in the protection reagent, do not remove the liquid prior to processing, as the protection reagent also begins the lysis of the sample.

1.3 Avoid Thawing Samples

It's important to avoid thawing your sample in order to preserve it's integrity before extraction of the RNA. Frozen cultured cells can be thawed briefly before the addition of the RNA lysis buffer, but most other sample types should not be thawed in the absence of the protection reagent or the lysis buffer. These samples include tissues, blood, bacteria, yeast and plants.

1.4 Maximize Your Yield by Ensuring Complete Homogenization

Ensure that your samples are completely disrupted and homogenized in order to release all of the RNA and maximize your yield. Mechanical homogenization for example, with a bead homogenizer, can boost your yields beyond those produced by proteinase K digestion alone when you're working with tissue. Mechanical homogenization is also recommended for tough to lyse samples such as plant, bacteria, and yeast.

1.5 Multiple Rounds of Homogenization May Be Recommended

When using mechanical lysis and homogenization, multiple rounds of homogenization may be recommended. If you do use multiple rounds, place your sample on ice for approximately one minute in between rounds to prevent your sample from over heating.

1.6 For Tissues or Leukocytes Preheat the heating block

If you are working with tissues or leukocytes it can help to preheat your heating block to 55 degrees Celsius before starting your prep, so that the heat block will be at the correct temperature when you get to the proteinase K incubation. Please be aware that when working with mammalian whole blood, the proteinase K incubation is carried out at room temperature.

1.7 Prepare Master Mix

We also recommend preparing a master mix of DNase I and DNase I reaction buffer prior to beginning if you are performing multiple preps at once.

Part 2 During Your Prep

2.1 Follow NEB Guidance

Be sure to follow our guidance on the recommended sample input amounts in order to ensure that the buffer volumes are appropriate and that the columns are not overloaded. This is a common mistake that quickly can reduce yield, purity, and integrity of your RNA.

2.2 After Lysis, Perform Procedure at Room Temp

After sample lysis, perform all steps at room temperature. This will prevent detergent from precipitating in the buffers. Do not place your samples on ice after lysis.

2.3 Label Collection Tubes

Before applying your sample to the genomic DNA removal column, be sure that you label the collection tube, as you'll be discarding the column after you centrifuge. After putting your sample through the genomic DNA removal column, make sure that you've saved the flow through. This flow through contains your RNA.

2.4 Add One Volume of Ethanol to Your Flow-Through

If you want to capture total RNA including small RNAs, add one volume of ethanol to your flow through from the gDNA removal column, in preparation for binding it to the RNA purification column. If you want to exclude RNAs smaller than two hundred nucleotides, only add half the volume of ethanol to your flow through from the gDNA removal column.

2.5 Perform All Wash Steps and Spin for Two Minutes after Your Last Wash

It's very important to perform all wash steps in the protocol in order to produce highly pure RNA. Be sure to spin your column for two minutes after the last wash.

2.6 For Low Yield Samples, Elute in 50ul Nuclease Free Water

For low yield samples, including muscle or brain, or for low starting inputs, consider eluting the RNA with 50 microliters of nuclease free water in order to get concentrated RNA. It's important to have a concentration above 20 nanograms per microliter. If a microvolume spectrophotometer, such as a NanoDrop, will be used to measure concentration and purity.

Part 3 After Your Prep

3.1 Spin to Remove Silica

In some cases, silica particles from the column matrix can be found in your eluate. To ensure that this doesn't affect your OD260-230 ratio, you can centrifuge the eluate for one to two minutes at 1600 x G and pipette the aliquot from the time of the liquid for measurement on a spectrophotometer.

3.2 Add EDTA to Protect RNA and Store in Aliquots

While nucleus free water is provided in this kit for the elution of RNA, adding EDTA to a final concentration of .1 to 1 millimolar can protect RNA samples that will be stored for an extended period of time. Additionally, it's good practice to store the eluted RNA in aliquots to avoid excessive freeze-thaw cycles.

Recent advances in mRNA vaccine technology

mRNA vaccines address many of the weaknesses of conventional vaccine types.

Innovations in recent years revolutionized the applicability of mRNA vaccines.

mRNA vaccines have demonstrated a high level of efficacy in preclinical studies.

Human clinical trials of mRNA vaccines are being heavily pursued.

Messenger RNA (mRNA) vaccines represent a relatively new vaccine class showing great promise for the future. This optimism is built on recently published studies demonstrating the efficacy of mRNA vaccines in combatting several types of cancer and infectious pathogens where conventional vaccine platforms may fail to induce protective immune responses. These results would not have been possible without critical recent innovations in the field, such as the development of safe and efficient materials for in vivo mRNA delivery and advanced protocols for the production of high quality mRNA. This review summarizes the most important developments in mRNA vaccines from the past few years and discusses the challenges and future directions for the field.

Results and discussion

IVT-seq library preparation and sequencing

To generate IVT-seq libraries (for full details, please see the Materials and methods section), we produced individual glycerol stocks each harboring a single, human, fully sequenced plasmid from the Mammalian Gene Collection (MGC) [17]. Next, we extracted the plasmid DNA and plated it at 50 ng per well in 384-well plates. We mixed the contents of three 384-well plates containing a total of 1,062 cDNA clones (Additional file 1), transformed this mixture into bacteria, and plated the bacteria as single colonies. Following an overnight incubation, we scraped these plates, amplified the bacteria for a few hours in liquid culture, and purified plasmids from the bacteria as a pool (Figure 1A). Next, we linearized the plasmids, and used SP6 polymerase to drive in vitro transcription of the cloned cDNA sequences (Figure 1B). Following a DNase I treatment to remove the DNA template and RNA purification, we were left with a pool of 1,062 different human RNAs derived from fully sequenced plasmids.

Construction of IVT-seq libraries. (A) Preparation of a pool of 1,062 human cDNA plasmids. Contents of three 384-well plates containing MGC plasmids were pooled together. Pool was amplified via transformation in Escherichia coli, and resulting clones were purified and re-pooled. (B) Generation of IVT transcripts. Pool of MGC plasmids was linearized and used as a template for an in vitro transcription reaction. Enzymes and unincorporated nucleotides were purified, leaving pool of polyA transcripts. (C) Creation of IVT-seq libraries. Listed quantities of IVT RNA were mixed with mouse liver total RNA to create six pools with final RNA quantities of 1 μg. Ribosomal RNA was depleted from these pools using the Ribo-Zero Gold kit. IVT RNA and mouse RNA are now present in pools at the listed ratios, following depletion of rRNA from mouse total RNA. These pools were used to generate RNA-seq libraries using Illumina’s TruSeq kit/protocol. This entire process was performed in duplicate. Replicate libraries were pooled separately and sequenced in separate HiSeq 2000 lanes (two lanes total). IVT, in vitro transcribed MGC, Mammalian Gene Collection.

To approximate what happens in a total RNA-seq reaction, we subjected this IVT RNA to rRNA depletion and then prepared libraries using the Illumina TruSeq protocol (Figure 1C, IVT only). To account for possible carrier effects, we also mixed the IVT RNA with various amounts of mouse total RNA derived from liver. The addition of the mouse RNA gave these samples greater diversity (transcripts from approximately 10,000 genes versus 1,062) and more closely resembled a real biological sample. Also, by adding background RNA from a different species (mouse) than the IVT RNA (human), we made it easier to differentiate between the IVT transcripts and mouse sequences during downstream analysis. Because the IVT RNA did not contain rRNA sequences whereas the mouse RNA did, the quantity of mouse RNA would be significantly reduced by the rRNA depletion step. To account for this, we mixed IVT and mouse RNA such that, following rRNA depletion, we would have final pools with IVT:mouse ratios of 1:1, 1:2, and 1:10. Finally, to account for mouse RNAs potentially mapping to the human reference genome and our IVT sequences, we prepared a pool consisting of mouse RNA alone. We pooled the resulting six libraries and sequenced them using an Illumina HiSeq 2000. We performed this entire process in duplicate.

Mapping and coverage of IVT-seq data

Following sequencing and de-multiplexing, we aligned all of the data to the human reference genome (hg19) using the RNA-seq Unified Mapper (RUM) [14]. For all analyses, we only used data from reads uniquely mapped to the reference, excluding all multi-mappers (data contained in RUM_Unique and RUM_Unique.cov files). Of the 1,062 original IVT transcripts, we found 11 aligned to multiple genomic loci, while 88 aligned to overlapping loci. To avoid any confounding effects in our analyses, we filtered those transcripts from all analyses, leaving us with 963, non-overlapping, uniquely-aligned IVT transcripts. We saw excellent correlation in expression levels between replicates (transcript-level R 2 between replicates >0.95 Additional file 2: Figure S1A). Secondly, at least 90% of the 963 IVT transcripts were expressed with fragments per kilobase of exon per million mapped reads (FPKM) values ≥5 in all IVT-seq datasets, except mouse only (Table 1). In the IVT-only samples, over 80% of the IVT sequences were expressed above 100 FPKM (Additional file 2: Figure S1B). Because we prepared the MGC plasmids and IVT transcripts as pools, it is likely that the IVT transcripts showing low or zero coverage were initially present at low plasmid concentrations prior to the transformation and IVT steps. Using the IVT-seq technique, we were able to specifically detect the vast majority of the human IVT transcripts with high coverage in both the absence and presence of the background mouse RNA.

While we do see reads aligned to the human IVT transcripts in the mouse-only data, these transcripts collectively represent approximately 2% of reads (Table 1). Those transcripts with higher coverage are likely the result of mouse reads aligning to highly similar human sequences. We excluded these sequences from our analyses.

Within-transcript variation in RNA-seq coverage of IVT transcripts

Consider first the IVT-only data. Given that these transcripts were generated from an IVT reaction using cDNA sequences, these data are unaffected by splicing or other post-transcriptional regulation. Thus, most regions of transcripts should be ‘expressed’ and present at similar levels. The exceptions would be repetitive sequences that map to multiple genome locations and may be poorly represented, and the ends of the cDNAs, which are subject to fragmentation bias. To account for this, we created a simulated dataset that models the fragmentation process and deviates from uniform data only by the randomness incurred by fragmentation. We generated two such datasets using the Benchmarker for Evaluating the Effectiveness of RNA-Seq Software (BEERS) [14]. The first dataset contained all of the IVT transcripts expressed at roughly the same level of expression (approximately 500 FPKM). For the second, we used FPKM values from the IVT-only samples as a seed, creating a simulated dataset with expression levels closely matching real data (Additional file 3: Figure S2). These datasets are referred to as simulated and quantity matched (QM)-simulated, respectively. The simulated data provide an ideal result, while the QM data allow us to control for any artifacts arising from expression level (for example, transcripts with lower expression may show more variability). Next, we aligned both simulated datasets using RUM, with the same parameters as for the biological data. Thus, both simulated datasets also serve as controls for any artifacts introduced by the alignment (for example, low coverage in repeat regions). For full details on the creation of simulated data, see the Materials and methods section.

Using IVT data derived from the BC015891 transcript as a representative example, the ideal, theoretical coverage plot from the simulated data shows near-uniform coverage across the transcript’s entire length, with none of the extreme peaks and valleys characteristic of biological datasets (Figure 2A). However, our observed data showed a high degree of variability, with peaks and valleys within an exon (Figure 2B). Furthermore, these patterns were reproducible across our replicates (Additional file 4: Figure S3). We saw many other cases of extreme changes in coverage: over 50% of the IVT transcripts showed greater than two-fold changes in within-transcript coverage attributable to library preparation and sequencing (Table 2 and Additional file 5: Figure S4). For example, BC009037 showed sudden dips to extremely low levels of expression in both of its exons (Figure 2C). Both simulated datasets showed no such patterns, which indicates this coverage variability is not the result of alignment artifacts. Furthermore, the absence of this pattern in the QM-simulated data indicates these fold-change differences in coverage were not due to sampling noise introduced by transcripts with low or high coverage. In the case of BC016283, the peaks and valleys in coverage led to greater than five-fold differences in expression levels between exons (Figure 2D). Once again, these patterns were reproducible across replicates (Additional file 4: Figure S3). The SP6 polymerase cannot fall off and then re-attach at a later point in the transcript, leaving a region un-transcribed. Therefore, given that these patterns showed troughs followed by peaks, they cannot be the result of artifacts from in vitro transcription. Furthermore, we sequenced the IVT products directly and found the vast majority were transcribed with little to no bias. Taken together, these data suggest that these coverage patterns are primarily the result of technical biases introduced during library construction, rather than biology. These results are consistent with a previous study that used IVT RNA as standards in RNA-seq experiments [16], suggesting that our IVT-seq methodology is suitable for identifying technical variability in sequencing data.

Within-transcript variations in RNA-seq coverage. (A) Simulated RNA-seq coverage for a representative IVT transcript (BC015891). RNA-seq coverage plot (black) is displayed according to the gene model (green), as it is mapped to the reference genome. Blocks correspond to exons and lines indicate introns. The chevrons within the intronic lines indicate the direction of transcription. Numbers on y-axis refer to RNA-seq read-depth at a given nucleotide position. (B) The actual RNA-seq coverage plot for BC015891 in the IVT-only sample. Representative coverage plots for the IVT transcripts (C) BC009037 and (D) BC016283 are displayed according to the same conventions used above. All transcripts are displayed in the 5ʹ to 3ʹ direction.

Between-sample variation in RNA-seq coverage of IVT transcripts

In addition to this variability within transcripts, we also found many transcript regions showing extreme variability in coverage across samples (Figure 3). For example, the sixth exon of BC003355 varied wildly relative to the remainder of the transcript across all IVT:mouse dilutions. Interestingly, the overall pattern of variation relative to the rest of the transcript across the dilutions was maintained between the replicates. Almost no reads in the mouse-only sample map to this transcript, which eliminates the possibility that this variability was due to incorrect alignment of mouse RNA.

Between-sample variations in RNA-seq coverage. RNA-seq coverage plots across all samples for exons 4 to 11 of the IVT transcript BC003355. The black rectangles identify exon six, which shows extreme variability in coverage relative to the rest of the transcript when viewed across all of the samples. The ratio of IVT RNA to mouse RNA is listed to the left of each sample’s coverage plots. Coverage plots (red for first replicate blue for second replicate) are displayed according to the gene model (black), as it is mapped to the reference genome. Blocks in the gene model correspond to exons and lines indicate introns. The chevrons within the intronic lines indicated the direction of transcription. Numbers on y-axes refer to RNA-seq read-depth at a given nucleotide position.

Including BC003355, we found 86 regions of high, unpredictable coverage (hunc) spread across 65 transcripts (Additional file 6). Therefore, over 6% of the 963 IVT transcripts contained regions showing wild but reproducible variations in RNA-seq coverage between samples. While identifying these hunc regions, we used a two-stage filter to eliminate variable regions resulting from mouse reads mapped to highly similar human sequences. First, we eliminated all hunc regions coming from transcripts with FPKM ≥5 in either mouse-only dataset. Next, to account for localized misalignment of mouse reads, we filtered out all hunc regions with an average coverage ≥10 in either mouse-only dataset. We also removed those hunc regions with mouse-only coverage ≥10 in the flanking 100 base pairs (bp) on either side. Given the stringent criteria we used to identify these hunc regions (see Materials and methods section for full details), it is likely that this is an underestimate. To address the possibility that mouse RNAs may interact with homologous human RNAs and interfere with them in trans, we assayed the sequences surrounding these regions using the MEME Suite [18], but we found no sequence motifs these regions have in common. Furthermore, the depth of coverage at these regions did not follow a linear relationship with the increasing mouse RNA, which suggests it is not simply a direct interaction with the background RNA. There is no clear cause for these hunc regions, particularly since we prepared all samples from the same pool of IVT RNA and the only difference between samples was the relative ratios of IVT RNA to mouse liver RNA. We also searched for hunc regions that were divergent between the two replicates, but found none. If such regions do exist, they could be identified and overcome by creating libraries in duplicate. The hunc regions we identified above with expression patterns maintained between replicates present a greater challenge, as they could not be identified and filtered out by creating duplicate libraries. This is particularly problematic for using exon-level expression values to identify alternative splicing events or differential expression. The within-transcript and between-sample variation we see in our IVT-seq data suggests that library generation introduces strong technical biases, which could confound attempts to study the underlying biology.

Sources of variability in RNA-seq coverage

There are three potential sources for technical bias in library preparation: RNA-specific molecular biology (RNA fragmentation, reverse-transcription), RNA selection method (rRNA depletion, polyA selection), and sequencing-specific molecular biology (adapter ligation, library enrichment, bridge PCR). To identify biases introduced solely by sequencing-specific molecular biology, we created a DNA-seq library from the same MGC plasmids used as templates for the IVT-seq libraries (Additional file 7: Figure S5). In doing this, we skipped the steps specific to the IVT or RNA molecular biology. We also prepared two additional IVT-seq libraries using polyA selection or no selection, instead of rRNA depletion. By comparing our plasmid library data and the IVT-seq data using various selection methods, we could identify which coverage patterns were the result of RNA-specific molecular biology, the RNA selection method, or of some common aspect of the library-generation protocol.

We sequenced the plasmid library using an Illumina MiSeq and aligned the resulting data to the human reference genome using the same method as the IVT-seq libraries. In this plasmid data, we saw 924 of the cDNA clone sequences with FPKM values ≥5, compared to approximately 870 in both of the IVT only samples (Table 1). This small drop in coverage was likely because the IVT RNA goes through more pooling steps during library construction than the plasmids. Furthermore, the plasmids are not affected by transcription and reverse transcription efficiencies. Additionally, the plasmid data mapped to the cDNA sequences with an average, normalized coverage of 42.08, which is within the range of coverage values we see for the IVT-seq samples. We sequenced the no selection and polyA selection libraries on a HiSeq 2500. These data also show cDNA clone coverage values similar to the other IVT-seq libraries.

The plasmid data represents the ‘input’ into the IVT reaction and the no selection data represents the closest measure of its direct output. By measuring the 3′/5′ ratio in depth of coverage for each IVT transcript, we could assess the processivity of the SP6 polymerase. In a perfectly processive reaction, this 3′/5′ ratio would be 1, indicating the polymerase did not fall off the cDNA template and lead to the formation of truncated products. The median 3′/5′ ratios for the plasmid and no selection data were 1 and 0.98, respectively, indicating premature termination of the IVT reaction was not a factor in our analyses.

Effect of different RNA selection methods on coverage patterns

Our analysis is illustrated by an examination of the coverage plots for BC003355 across all of our different datasets. The high degree of variability we noted in this gene’s coverage plot from our rRNA-depleted data was absent in the no selection and plasmid data (Figure 4A). While the polyA data also showed fewer peaks and valleys than the rRNA-depleted total RNA-seq data, it was marked by the well-documented 3′ bias. These data suggest that the rRNA depletion step is likely responsible for a large quantity of the observed coverage biases.

Sources of bias in RNA-seq coverage. (A) RNA-seq coverage plots for IVT transcript BC003355 from simulated (black), plasmid (blue), no selection (green), rRNA-depleted (red), and polyA (orange) data. The gene model is displayed in black, below all of the coverage plots. Blocks correspond to exons and lines indicate introns. The chevrons within the intronic lines indicate the direction of transcription. (B) Distributions for coefficients of variation across data displayed above, with the addition of the QM-simulated data (gray). Note that while the graph is cutoff at a coefficient of variation of 1.3, the tails for the rRNA-depleted and polyA distributions extend out to 2.13 and 2.7, respectively. (C) Effect sizes for the differences in distribution of coefficients of variation between sequencing libraries and simulated data. Effect sizes are calculated as the per-transcript ratios of coefficients of variation between a given library and the simulated dataset. QM, quantity matched.

To quantify the variability for each selection method, we calculated the coefficient of variation at the single base level in coverage for all IVT transcripts across each of these datasets (Figure 4B). Using a Wilcoxon rank-sum test (plasmid n = 924, no selection n = 870, rRNA-depleted n = 869), we found the rRNA-depleted data had significantly higher variability than the no selection and plasmid data (P <2.2e -16 ). Furthermore, the rRNA-depleted and polyA libraries were more than 60% more variable on average than the plasmid library (Figure 4C). This suggests that a significant portion of the observed variability in coverage across transcripts in the IVT-seq data is the result of RNA-specific molecular biology, specifically the RNA selection step. Furthermore, after accounting for bias introduced by the sequences themselves (plasmid data) and bias introduced by the IVT reaction (no selection data), we found that 50% of transcripts had two-fold and 10% had 10-fold variation in within-transcript expression (Table 2 and Additional file 5: Figure S4). While it is well-appreciated that polyA selection introduces bias, we found that rRNA-depleted data introduced just as much if not more. Neither simulated dataset showed transcripts with a two-fold or higher change in within-transcript expression. Again, this suggests that the observed within-transcript variations are not the result of alignment artifacts or sampling due to low or high expression. One commonly acknowledged source of bias arises from random priming during library preparation [10]. When we examined the different libraries, we saw that fragments from all of the RNA-seq data showed nucleotide frequencies characteristic of random priming bias (Additional file 8: Figure S6). As expected, the plasmid data showed no such bias, since it was derived directly from DNA and did not require a cDNA-generation step. However, the significant differences between all RNA libraries suggest that bias from random priming is not the only factor. The plasmid and no selection data still contain a fair amount of variability when viewed alongside the simulated data (Figure 4A black). When we examined the entire dataset, both the plasmid and no selection data had significantly higher variation than either simulated dataset (Wilcoxon rank-sum test simulated data n = 963, QM-simulated data n = 869, plasmid n = 924, no selection n = 870 P <2.2e -16 ). These data suggest that sequencing-specific molecular biology common to all libraries we prepared (adapter ligation, library amplification via PCR) is also responsible for a portion of the observed coverage variability and sequencing bias.

Biases associated with sequence features are dependent on RNA selection method

Given these significant differences in coverage variability, we sought to identify sequence features that might contribute to this bias. We considered three quantifiable sequence characteristics: hexamer entropy, GC-content, and sequence similarity to rRNA (see Materials and methods for a detailed description of these metrics). For each sequencing strategy (plasmid, no selection, rRNA-depleted, polyA), we tested if any of the three sequence characteristics had a significant effect on variability in sequencing coverage, as measured by the coefficient of variation. While we are primarily focused on coverage variability as an indicator of sequencing bias, we also looked at depth of coverage, as measured by FPKM.

For each sequencing strategy, we sorted the transcripts by coverage variability or depth. Next, we selected the 100 most and 100 least extreme transcripts from each list. We compared the values of the sequence characteristics between the 100 most and 100 least extreme transcripts using a Wilcoxon rank-sum test. Significant P-values indicate a significant association of the sequence characteristic with coverage variability and/or depth. The results of our analysis are displayed as box-plots (Figure 5 and Additional file 9: Figure S7).

Effects of sequence characteristics on coverage variability. Distributions of (A) hexamer entropy, (B) GC-content, and (C) rRNA sequence similarity for the 100 transcripts with the highest and lowest coefficients of variation for transcript coverage from the plasmid, no selection, rRNA-depleted, and polyA libraries. Asterisks indicate the significance of a Wilcoxon signed-rank test comparing values for the listed sequence characteristics between each pair of groups from the same sample. *P <0.05 **P <0.01 ***P <0.001.

To check for any confounding effects between coverage depth and variability, we tested the least and most expressed transcripts for any correlations with variability in coverage (Additional file 10: Figure S8). The polyA library showed a significant correlation (P <2.2e -16 ) between coverage variability and depth, which indicates sequence features could be affecting coverage through variability (or vice versa). The rRNA-depleted data showed a slight, significant correlation (P = 0.04933). It is possible some feature of RNA selection affects both variability and coverage, given that we saw no significant correlations for the two remaining samples. This indicates that coverage variability and depth are independent for the plasmid and no selection data.

All three sequence characteristics had a significant association with variability and depth-of-coverage in at least one of the sequencing strategies. In particular, lower hexamer entropy, a measure of sequence complexity [19–21], was strongly associated with higher variance in all of the RNA libraries (no selection P = 4.712e -05 rRNA depletion P = 3.956e -11 polyA P = 0.003921 Figure 5A). This suggests that bias associated with hexamer entropy is due partially to RNA-specific procedures in library preparation. Furthermore, an association with lower hexamer entropy indicates there are more repeat sequences in the transcripts with higher variability. This could be indicative of complex RNA secondary structures, as repeated motifs could facilitate hairpin formation. Furthermore, the absence of this association from the plasmid data suggests that this observation was not due to mapping artifacts. The plasmid data contained the same sequences as the RNA-seq data, and would be subject to the same biases introduced by our exclusion of multi-mapped reads.

Higher GC-content was strongly associated with lower coverage variability in the no selection and polyA data (P = 5.627e -13 P = 4.914e -05 Figure 5B), suggesting that the effects of GC-bias on within-transcript variability could arise, in part, due to some RNA-specific aspects of library preparation. Also, it appears that GC-bias was not a significant contributing factor to either depth of coverage or the extreme variability in the rRNA-depleted data. Meanwhile, lower GC-content was associated with higher coverage in the plasmid data (P = 3.776e -05 ), and lower coverage depth in the no selection and polyA libraries (no selection P = 8.531e -05 polyA P = 0.0009675 Additional file 9: Figure S7B). Given that this trend switched directions between the plasmid library and the RNA libraries, this also suggests that some RNA-specific aspect of library preparation is introducing GC-bias distinct from the high GC-bias associated with Illumina sequencing [22].

Interestingly, higher rRNA sequence similarity was associated with higher coverage variability in the rRNA-depleted library (P = 9.006e -05 ) and lower variability in the no selection library (P = 0.0367 Figure 5C). It is unsurprising that similarity to rRNA sequences contributed to variability in the rRNA-depleted data, given that rRNA depletion is based upon pair-binding between probes and rRNA sequences. While it is unclear why this trend was reversed in the no selection library, it is striking given the significant increase in within-transcript variability between the no selection and rRNA-depleted libraries (Figure 4B). Furthermore, we saw a slight but highly significant correlation (Pearson R 2 = 0.308452 P <2.2e -16 ) between a transcript sequence’s similarity to rRNA and the magnitude of the difference in coverage between the no selection and rRNA-depleted libraries (Additional file 11: Figure S9 and Additional file 12). While the majority of the factors contributing to the extreme bias in sequence coverage we saw in the rRNA-depleted data remain unclear, our data suggest this could be partially due to depletion of sequences homologous to rRNA.

Taken together, our data demonstrate the utility and potential of the IVT-seq method to identify sources of technical bias introduced by sequencing platforms and library preparation protocols.

Strategies for simultaneous and successive delivery of RNA

Advanced non-viral gene delivery experiments often require co-delivery of multiple nucleic acids. Therefore, the availability of reliable and robust co-transfection methods and defined selection criteria for their use in, e.g., expression of multimeric proteins or mixed RNA/DNA delivery is of utmost importance. Here, we investigated different co- and successive transfection approaches, with particular focus on in vitro transcribed messenger RNA (IVT-mRNA). Expression levels and patterns of two fluorescent protein reporters were determined, using different IVT-mRNA doses, carriers, and cell types. Quantitative parameters determining the efficiency of co-delivery were analyzed for IVT-mRNAs premixed before nanocarrier formation (integrated co-transfection) and when simultaneously transfecting cells with separately formed nanocarriers (parallel co-transfection), which resulted in a much higher level of expression heterogeneity for the two reporters. Successive delivery of mRNA revealed a lower transfection efficiency in the second transfection round. All these differences proved to be more pronounced for low mRNA doses. Concurrent delivery of siRNA with mRNA also indicated the highest co-transfection efficiency for integrated method. However, the maximum efficacy was shown for successive delivery, due to the kinetically different peak output for the two discretely operating entities. Our findings provide guidance for selection of the co-delivery method best suited to accommodate experimental requirements, highlighting in particular the nucleic acid dose-response dependence on co-delivery on the single-cell level.

Keywords: Co-expression In vitro synthesized mRNA Integrated co-transfection Parallel co-transfection Successive transfection Transfection methods.

Conflict of interest statement

The authors declare that they have no competing interests.


Schematic overview of different transfection…

Schematic overview of different transfection methods: integrated co-transfection a refers to mixing different…

Design of experiments and description…

Design of experiments and description of transfection conditions for co-delivery of IVT-mRNA and…

Simultaneous transfection of macrophages with…

Simultaneous transfection of macrophages with EGFP and mCherry coding IVT-mRNA with two different…

Comparison of IVT-mRNA co-delivery using…

Comparison of IVT-mRNA co-delivery using liposomal and polymeric carriers in HeLa cells. Fluorescent…

Comparison of integrated and parallel…

Comparison of integrated and parallel co-transfection approaches for pDNA delivery in HeLa cells…

Successive transfection of IVT-mRNA in…

Successive transfection of IVT-mRNA in HeLa cells with LipoMM fluorescent images depicted as…

Co-delivery versus successive delivery of…

Co-delivery versus successive delivery of siRNA and IVT-mRNA in d2EGFP HeLa cells transfected…

In-vitro Transcribed mRNA Delivery Using PLGA/PEI Nanoparticles into Human Monocyte-derived Dendritic Cells

Induction of protein synthesis by the external delivery of in-vitro transcription-messenger RNA (IVT-mRNA) has been a useful approach in the realm of cell biology, disease treatment, ‎reprogramming of cells, and vaccine design. Therefore, the development of new formulations for ‎protection of mRNA against nucleases is required to maintain its activity in-vivo. It was the aim of the present study to ‎investigate the uptake, toxicity, transfection efficiency as well as phenotypic consequences of ‎a nanoparticle (NP) in cell culture. NP consists of poly D, L-lactide-co-glycolide (PLGA) and polyethyleneimine (PEI) ‎for delivery of in-vitro transcription-messenger RNA (IVT- mRNA) encoded green fluorescent protein (GFP) in human monocyte-derived ‎dendritic cells (moDCs). Nanoparticles that were synthesized and encapsulated with synthetic GFP mRNA, exhibited size distribution in this formulation, with mean particle sizes ranging between 415 and 615 nm. Zeta potential was positive (above 12-13 mV) and the encapsulation efficiency exceeded 73.5%. Our results demonstrated that PLGA/PEI NPs encapsulation of GFP mRNA had ‎no toxic effect on immature monocyte-derived ‎dendritic cells and was capable of delivering of IVT-mRNA into moDCs and was highly effective. The expression of GFP protein 48 h after transfection was confirmed by flow cytometry, microscopic examination and western blotting assay. This NP can make a way to target moDCs to express a variety of antigens by IVT- mRNA. The ‎present study introduced the PLGA/PEI NP, which provided effective delivery of ‎IVT-mRNA that encodes the GFP protein.

Keywords: Dendritic cells GFP IVT- mRNA Nanoparticle PEI PLGA.


Schematic diagram demonstrating preparation and…

Schematic diagram demonstrating preparation and synthesis of GFP mRNA- encapsulated poly (D, L…

(A) The pGE-GFP plasmid containing…

(A) The pGE-GFP plasmid containing GFP gene and cis-acting flanking structures such as…

Characterization of PLGA/PEI nanoparticle. Scanning…

Characterization of PLGA/PEI nanoparticle. Scanning electron microscope (SEM) images showing the spherical morphology…

Gel retardation assays. Electrophoretic migration…

Gel retardation assays. Electrophoretic migration of GFP mRNA complexed with PEI/PLGA at varying…

Agarose gel electrophoresis of mRNA…

Agarose gel electrophoresis of mRNA extracted from nanoparticles after treatment with nuclease enzyme.…

Intracellular uptake of CFSE-encapsulating nanoparticles…

Intracellular uptake of CFSE-encapsulating nanoparticles by moDCs. (A) Fluorescent images of uptake of…

Flow cytometry analysis, fluorescence microscopic…

Flow cytometry analysis, fluorescence microscopic images and western blotting of GFP PLGA/PEI NPs-treated…

Flow cytometry analysis of moDCs…

Flow cytometry analysis of moDCs viability, before and after transfection with NPs. The…

Don't Overdry cDNA or Nucleotides

The following is a list of common laboratory techniques that are important for the success of any procedure, but have proved to be particularly important in a multistep protocol like RNA amplification. Even if time is limited, do not cut corners with these basic techniques.

Make master mixes
Master mixes should be prepared when processing 2 or more samples simultaneously to reduce the number of pipetting steps and the potential for pipetting error. Always include

5% overage of all reagents in master mixes to cover pipetting error.

Thaw all reagents properly
All frozen reagents should be thawed completely, mixed thoroughly, centrifuged briefly, and placed on ice as necessary. However, allow IVT components to equilibrate to room temperature before setting up your reactions because spermidine in the reaction buffer may cause cDNA to precipitate at lower temperatures. To ensure optimal performance, thaw components at room temperature, and avoid higher temperatures.

Be gentle with enzymes
Never vortex enzymes. Mix by gently flicking the side of the tube to avoid inactivating the enzyme.

The success of your microarray analysis depends on the quality of the aRNA used in the hybridization. Following these tips and reminders for amplification can greatly increase the likelihood of obtaining good quality aRNA and reproducible microarray data.

4 Post-synthetic and Post-transcriptional Labeling of RNAs

Besides introduction of modified nucleotides during solid phase or enzymatic synthesis, naturally occurring RNA can also be labeled post-synthetically or post-transcriptionally. To date, a variety of different approaches has been reported employing chemical modifications 122-128 or catalytically active deoxyribozymes, 129-132 ribozymes 133-135 and enzymes 136-140 to accomplish modification reactions on unmodified RNA. Starting from pre-existing RNA sequences, that can either be prepared in vitro, by synthetic or enzymatic strategies or emerge from endogenous origin, the RNA oligonucleotides undergo labeling reactions introducing functional reporter groups such as fluorophores, affinity tags or small molecule handles for subsequent modification with reporter groups. 14 In principle, post-synthetic and post-transcriptional RNA labeling approaches can be applied to short and long sequences equally and do not face limitations concerning long RNAs. However, special attention needs to be paid addressing site-specificity and universality of such approaches regarding applicability for all types of RNA and not only for those with intrinsic distinctive features. 14 As previously mentioned, a wide range of labeling techniques for unmodified RNA were developed by now. Admittedly, some techniques include methyltransferases for the introduction of modifications at the 5′-cap structure 141 or at the 3′-terminus 142 or poly(A) tail. 143 Other approaches utilizing self-alkylating ribozymes, 144 tRNA-guanine transglycosylases (RNA-TAG) 145 or tRNA Ile2 -agmatidine synthetase (Tias) 146 are based on larger structural elements that need to be fused to the RNA of interest. As this review focuses on internal covalent and site-specific labeling of long RNA with minimal impact on RNA's native folding and function, the aforementioned approaches will not be described in the following section.

4.1 Chemical strategies for site-specific labeling of unmodified RNA

In unmodified RNA oligonucleotides, a plethora of functionalization sites are available as the 2′-OH residue of the ribose sugar possesses nucleophilic character and therefore provides multiple options for chemical modifications. 147, 148 Special care has to be taken to address site-specificity and chemoselectivity to select only a certain target site of the RNA oligonucleotide. 148 This can be achieved through the use of DNA helper strands assigned to guide a functional group to the desired site of modification and to protect the remaining RNA from chemical reagents. 123, 124, 127, 128

4.1.1 DNA-templated chemical labeling via four-way junction formation

Applying a DNA-templated chemical labeling strategy, Jahn et al. demonstrated the site-specific introduction of a thiol group for subsequent attachment of labels for both synthetic and endogenous RNA. 124 The thiol functionalization can be attached to predetermined internal nucleotides of the acceptor RNA by formation of a four-way junction with a donor DNA linked to an activated carboxylic acid group and two site-specific DNA guiding strands. The DNA donor strand will then be coupled to a specific internal 2′-OH group of the acceptor RNA via the activated carboxylic acid group. Subsequently, the adjacent disulfide bond in the donor's linker is cleaved eventually yielding the thiol-modified acceptor RNA. Afterwards, site-specifically thiol-modified target RNA can further be labeled with a reporter group which was demonstrated by coupling of a maleimide-functionalized fluorophore. 124

4.1.2 DNA-templated chemical labeling via duplex formation

An alternative chemical labeling strategy was developed by Freisinger and co-workers using a DNA-templated approach for sequence-specific generation of etheno-adducts in single-stranded DNA. 122 Thereby, a guiding DNA bearing an alkylating agent is hybridized to the target DNA to bring the target base and alkylating agent in close proximity. Using this approach, the group of Freisinger obtained only moderate yields up to 30 %. 122 Improving yields up to 65 %, Sigel, Freisinger and co-workers demonstrated DNA-templated formation of a 12-propargyl-etheno-adenine-modified single-stranded 16 mer DNA and its RNA equivalent and subsequent fluorophore labeling via CuAAC. 123 This method in principle allowed the incorporation of bioorthogonal groups into single stranded regions of both DNA and RNA of unrestricted length. Fluorescent labeling of a large RNA oligonucleotide was established for the 633 nt long D135-L14 group II intron ribozyme construct derived from the Saccharomyces cerevisiae Sc.ai5γ. 123 In 2018, they also presented site-specific dual-color labeling of the regulatory 275 nt long btuB riboswitch from Escherichia coli (E. coli) by combining internal functionalization with oxidative opening of the 3′-terminal ribose and subsequent conjugation to two different fluorophores. 128 Native folding and function of the btuB RNA riboswitch were studied using smFRET to characterize the conformational equilibrium of the btuB riboswitch upon binding of its cofactor adenosylcobalamin. The experiment proved that not only nucleotides within single-stranded regions but also within RNAs with complex secondary structures can be targeted. 128

4.1.3 RNA acylation at induced loops

More recently, Kool and co-workers developed the technique RNA acylation at induced loops (RAIL) for in vitro functionalization of RNA at specific 2′-OH groups within the sequence. 127 RAIL employs complementary helper DNA strands to expose loops or gaps at desired predefined sites within the RNA of interest to enable site-selective 2′-OH acylation with an acylimidazole reagent, e. g. nicotinyl acylimidazole azide (NAI-N3, Figure 7). The remaining RNA segments and their reactive 2′-OH groups are protected by the formed DNA-RNA duplex structures. After acylation reaction, the helper DNA can be degraded by DNase digest. 126 Utilizing this strategy, the desired mono-acylated adduct was formed predominantly and only a minor fraction of side product acylated at the adjacent nucleotide 5′ to the induced bulge was observed. 127 Addressing acylation sites in close proximity to both, 5′- and 3′-end, of the RNA revealed a lack of selectivity for the gap induction approach as reliable hybridization of a short helper DNA strand complementary to the 5′- or 3′-end of the RNA cannot be ensured. 127 RAIL was employed to selectively control the catalytic activity of a 81 nt tandem hammerhead ribozyme (TR) with dual catalytic cores (3TR and 5TR). Specific acylation at the 3TR core guided to the target site via an induced gap, subsequently suppressed the cleavage rate by 5-fold and retained 85 % of the initial rate of untreated TR with the 5TR substrate. 127 In comparison, site-specific acylation of the 5TR core facilitated by an induced loop strategy suppressed the cleavage rate by 4-fold and retained 93 % of the initial rate of untreated TR with the 3TR substrate. Thus, this method provides local control of RNA acylation in multifunctional RNA and high yields of acylation-based suppression of local RNA function with little off-target acylation. 127 Furthermore, Kool's group tested dual labeling of 65 nt small nucleolar SNORD78 RNA for FRET experiments via serial RAIL labeling. 127 Successive labeling was performed by first loop induced acylation with NAI-N3 at G14 and subsequent SPAAC reaction with Alexa488-azide, followed by a second acylation step at A49 and subsequent click reaction with tetramethylrhodamine(TAMRA)-azide. 127 Taken together, the RAIL method is well suited for internal, site-directed acylation and further functionalization particularly for longer RNAs. RNAs with several hundred nucleotides in length could in principle be selectively acylated, if helper DNAs are prepared enzymatically or several synthetic helper DNAs are arranged along the RNA of interest. 127 The major disadvantage of this approach is the formation of 20–30 % secondary acylation products adjacent to the targeted position. 127 Interestingly, the acylating reagent NAI-N3 that has so far been used for click-based conjugation as well as for blocking and caging 125 forms a reversible adduct on RNA which can be removed by bioorthogonal deacylation with phosphines. 127

Site-selective RNA labeling via RNA acylation at induced loop (RAIL) structures. 127 A complementary helper DNA strand is hybridized exposing a loop at a predefined site within the endogenous RNA of interest to enable site-selective 2′-OH acylation with an acylimidazole reagent (here: nicotinyl acylimidazole azide). Remaining RNA segments and their 2′-OH groups are shielded by the helper DNA-RNA duplex. After acylation, the helper DNA strand is degraded by DNase treatment. Site-selectively acylated RNA can be further modified by CuAAC reactions with functionalized alkine-conjugates.

4.2 Deoxyribozymes for labeling of unmodified RNA oligonucleotides

DNAzymes can not only be used to ligate RNA as described in section 2.2 but can also be employed for post-transcriptional or post-synthetic labeling of RNA. 129-132 For that purpose, special deoxyribozymes exist that can either link a modified RNA strand or a single nucleotide to a pre-existing RNA. 129, 149 A major advantage of this approach is the high labeling specificity facilitated by specially designed base pairing interactions between target RNA and DNAzyme. 129, 149 In comparison to RNA, facile and low-priced preparation of the deoxyribozyme by standard solid-phase DNA synthesis and DNA's high stability in principle allows a broad application in biochemical laboratories.

4.2.1 Deoxyribozyme-catalyzed labeling introducing phosphodiester branches

Introducing the approach of deoxyribozyme-catalyzed labeling (DECAL), Silverman and co-workers effectively utilized the 10-DM24 DNAzyme for modifying RNA at specific internal positions (Figure 8). 129 Therefore, a tagging RNA is created by standard in vitro transcription introducing a 5-aminoallyl-modified cytidine at its second position. The primary amino group on the aminoallyl-RNA is then functionalized by reaction with an NHS-ester derivative to obtain labeled tagging RNA. Using the 10-DM24 DNAzyme which hybridizes to both, target RNA and tagging RNA, the labeled tagging RNA is attached to an internal 2′-OH group of the target RNA to result in a 2′,5′-phosphodiester bond formation yielding a branched oligonucleotide. The label can be attached to different sites within the target RNA by respective modification of the DNAzyme's binding arms. 129 By this, Silverman reported the successful application of DECAL to install two fluorophores at specific internal sites of the 160 nt long P4–P6 Tetrahymena group I intron P4–P6 RNA enabling to study Mg 2+ dependent RNA folding by FRET. 129

Refined deoxyribozyme-catalyzed labeling (DECAL) of pre-existing RNA with the 10-DM24 DNAzyme applying assisting heavy metal ions and the phosphorothioate-modified GTP derivative Sp-GTP S . 130, 131 The further engineered 10-DM24 deoxyribozyme catalyzes the attachment of modified guanine mononucleotides to a 2′-OH group within the RNA strand of interest. Addition of the oligonucleotide cofactor RΔ compensates for a formerly longer oligonucleotide substrate and is necessary for sufficient conversion. The deoxyribozyme's binding arms hybridize to the target RNA and can be adapted for individual target sites to enable site-specific RNA modification. Phosphorothioate-modified Sp-GTP S is used as ligation substrate to prevent subsequent phosphodiesterase hydrolysis of the 10-DM24 catalyzed 2′,5′-branch modification due to the particular steric configuration of the newly formed linkage in 2′-Rp-P S -labeled RNA. Tb 3+ acts as an accelerating cofactor that enables efficient conversions at pH 7.5 and reduces the required concentration of modified GTP triphosphate. Thiophilic Cd 2+ ions are used to prevent phosphorothioate interfering effects on the metal ion coordination by the 10-DM24/RNA/GTP S ternary complex.

Based on the aforementioned approach, Höbartner and Silverman reported further engineering of the 10-DM24 deoxyribozyme to accept mononucleotides as ligation substrates instead of 5′-triphosphorylated oligonucleotides. 149 Catalyzed ligation of free GTP mononucleotides by the engineered 10-DM24 DNAzyme requires the auxiliary addition of an oligonucleotide cofactor to stabilize the deoxyribozyme's secondary and tertiary structure. 149

Additionally, Höbartner and co-workers showed that lanthanide cofactors, in particular Terbium (Tb 3+ ), are able to accelerate DNA-catalyzed synthesis of branched RNA. 132 They presented a general post-transcriptional labeling approach for endogenous RNA with different ribose-modified guanine triphosphates and Tb 3+ as an accelerating cofactor to install functional groups such as azides and primary amines, affinity reagents, fluorophores, spin labels and cross-linkers at 2′-OH groups of internal adenines within in vitro transcribed RNA (Figure 8). 130 This method provides several advantages: first of all, the use of commercially available modified mononucleotides ensures a small modification size compared to the earlier used tagging oligonucleotides and at the same time reduces preparatory work. 130 Secondly, utilizing Tb 3+ as accelerating cofactor for the DNAzyme enables efficient conversions at physiological pH and reduces the concentration of GTP derivatives required for labeling. Overall yields >80 % could be obtained for most labeled GTP analogs providing at the same time fast reaction rates of kobs≈1 min −1 . 130 The group of Höbartner used this method to introduce two fluorophores as FRET pair on the SAM II riboswitch RNA by two successive DNAzyme catalyzed labeling reactions to study Mg 2+ -induced folding of its 52 nt long aptamer domain in the presence and absence of SAM ligand. 130 The observed results were consistent with previously reported data, 150 proving full functionality of the labeled riboswitch. 130 Furthermore, they proved applicability of their approach to label larger RNA transcripts in vitro. For the 120 nt long spliceosomal U6 snRNA, ten different labeling positions for methylanthraniloyl-G (MANT-G) were tested in total revealing influence of the secondary structure on efficient deoxyribozyme binding and catalysis. 130 To improve DNA-catalyzed labeling of complexly folded RNAs, the use of specially designed disruptor oligonucleotides breaking up secondary structure formation might facilitate the DNAzyme's hybridization to the target RNA increasing labeling efficiency. Using this approach including a disruptor oligonucleotide, DNA-catalyzed labeling of the 156 nt long ydaO riboswitch RNA with Cy3-G and Cy5-G was successively performed yielding double labeled long RNA allowing further FRET studies. 130

4.2.2 Deoxyribozyme-catalyzed labeling introducing phosphothioate branches

In 2017, Höbartner and co-workers reported almost quick and complete hydrolysis of the 10-DM24-catalyzed 2′,5′-phosphodiester branch modification in a 209 nt long UBC4 pre-mRNA transcript after incubation with purified Saccharomyces cerevisiae debranchase (Dbr1). 131 Hence, the use of this labeling technique is restricted to in vitro applications only. However, as it is known that Dbr1 is unable to hydrolyze phosphorothioate linkages with Rp-configuration, 151 this characteristic can be exploited to prevent phosphodiesterase hydrolysis of the label. Therefore, phosphorothioate-modified, fluorescent GTP derivatives of Sp-GTP S were synthesized as 10-DM24-catalyzed labeling of RNA with Sp-GTP S results in an inversion of the phosphorothioate configuration yielding 2’-Rp-GMP S -labeled RNA. Moreover, the labeling conditions for 10-DM24 were optimized to efficiently incorporate Sp-GTP S by addition of the thiophilic metal ion Cd 2+ to prevent phosophothioate interfering effects on the metal ion coordination by the 10-DM24/RNA/GTP S ternary complex. Although reactions with Sp-GTP S derivates generally showed a slower reaction rate compared to GTP derivatives, the resulting 2’-Rp-GMP S -labeled RNA was found to be mostly resistant to debranching by yeast Dbr1. 131 Yet, applications for sensitive RNA substrates in a cellular context are limited by the need to form a three-helix junction 129 and the necessity for assisting heavy metal ions for improved deoxyribozyme activity. 130-132

4.3 Ribozymes for labeling of unmodified RNA oligonucleotides

A promising approach to label endogenous RNA in vivo can be found in utilizing genetically encoded ribozymes.

4.3.1 Ribozyme-catalyzed labeling introducing phosphodiester branches

Maghami et al. reported direct in vitro selection of trans acting 2′-5′ adenylyl transferase ribozymes for covalent and site-specific RNA labeling with N 6 -(6-aminohexyl)-ATP and N 6 -fluorophore labeled ATP analogs at specific internal 2′-OH groups resulting in 2′-5′-phosphodiester bond branched RNA. 133 In the presence of total cellular RNA, the most efficient and specific ribozyme, FH14, was employed to fluorescently label the 120 nt long E. coli 5 s sRNA at three different positions using a fluorophore-modified ATP analog. Therefore, three variants of the FH14 ribozyme were designed with individual binding arms of 8–10 nt flanking the different labeling sites. Combining two or three ribozyme variants, multiple labeling could be achieved. 133 Admittedly, it is known that the branched 2′-5′-phosphodiester bonds were readily cleaved by natural debranching enzymes resulting in loss of the installed label 131 therefore hampering the scope of this labeling approach for cellular applications. In addition, not only modified ATP analogs but also cellular, ubiquitous ATP is used as a substrate by the ribozyme and competes for incorporation. 152 In return, the N 6 -modified ATP can also be incorporated by cellular ATP dependent enzymatic reactions which will cause unspecific ribozyme-independent background labeling. 153

4.3.2 Ribozyme-catalyzed labeling introducing phosphonate ester branches

Recently, Höbartner and co-workers evolved advanced ribozymes for bioorthogonal RNA labeling attaching derivatives of the acyclic nucleoside phosphonate Tenofovir via branched phosphonate ester junctions to the 2′-OH of an adenosine within the target RNA (Figure 9). 134 Tenofovir is an antiviral lipophilic and cell-permeable prodrug used against HIV and hepatitis B virus (HBV) infections. 154 Applying the Tenofovir-transferase ribozyme FJC9 and fluorescently-modified Tenofovir, the 120 nt long E. coli 5S rRNA was efficiently labeled in the presence of cellular total RNA. Additionally, using Cy5-modified Tenofovir and Tenofovir-transferase ribozyme FJ1, six different target sites within E. coli 16S and 23S rRNA, 1500 nt and 2900 nt long, respectively were modified. 134 Notably, the phosphonate ester bonds formed by Tenofovir-transferase ribozymes were more stable towards Dbr1 compared to natural phosphodiester linkages formed during previously described labeling approaches using DNAzymes and ribozymes. This aspect taken together with the benefit of avoiding cross-interactions with non-specific cellular substrates, as described for previous nucleotide labeling analogues, 130, 132, 133, 149, 152, 153 renders this method in principle possible for future applications in a cellular context. Moreover, simultaneous orthogonal labeling installing both a phosphonate ester and phosphodiester branching modification by application of two orthogonal ribozymes could be achieved on either a single common target RNA (Figure 9) or on two individual target RNAs. More in detail, simultaneous labeling of 23S rRNA by the FJ1 ribozyme using Cy5-Tenofovir in combination with labeling of 16S rRNA by FH14 with FAM-ATP in the presence of cellular total RNA was demonstrated. 133

Site-specific orthogonal double labeling of an endogenous RNA of interest using nucleotidyl and Tenofovir transferase ribozymes. 134 The in vitro selected 2′-5′ adenylyl transferase ribozyme FH14 can be utilized for internal labeling of 2′-OH groups with N 6 -modified ATP analogs yielding 2′-5′-phosphodiester bond branched RNA. Additionally, the in vitro selected Tenofovir transferase ribozyme FJ1 can be applied for internal 2′-OH group labeling with N 6 -modified acyclic nucleoside phosphonate Tenofovir derivatives yielding 2′-5′-phosphonate ester bond branched RNA.

4.3.3 Ribozyme-catalyzed labeling by methyltransferase ribozymes

More recently, Scheitl et al. reported in vitro selection of a methyltransferase ribozyme for site-specific RNA methylation. 135 The methyltransferase ribozyme MTR1 catalyzes the transfer of a methyl group from the cofactor O 6 -methylguanine (m 6 G) to a specific adenine yielding modified 1-methyladenosine (m 1 A) in its target RNA sequence. Site-specific methylation of adenosines was shown for a target tRNA in the presence of total E. coli tRNA in vitro. 135 However, the methyltransferase ribozyme was not yet applied for site-specific RNA labeling other than methylation but represents a promising tool for future applications in RNA labeling using e. g. fluorescently labeled benzylguanine derivatives. 135 Nevertheless, when developing such an application, circumvention of undesired cross-interactions with non-specific cellular substrates needs to be addressed.

4.4 Methyltransferases for enzymatic labeling of RNA oligonucleotides

RNA methyltransferases catalyze the transfer of methyl groups to target nucleotides in RNA. 147 Many methyltransferases employ the cofactor S-adenosyl-L-methionine (AdoMet or SAM) as methyl-group donor. 147 RNA methylation can be applied for post-transcriptional modification of coding and non-coding RNA. 147, 155 Exploiting the substrate promiscuity of certain methyltransferases, these enzymes can be used for RNA labeling introducing a variety of functionalities. By that, the methyltransferase-directed transfer of activated groups (mTAG) can be used for covalent and sequence-specific labeling of pre-existing RNA. 136, 139, 140

Liu and co-workers first found that human RNA N 6 -methyladenosine (m 6 A) methyltransferase METTL3-METTL14 (METTL3-14) to a certain extend exhibits co-substrate promiscuity and is able to site-specifically transfer an allyl group to the N 6 -position of adenosine. 138 Therefore, a distinctive consensus motif needs to be present in the RNA of interest that serves as a recognition sequence for the methyltransferase and defines the methylation site. For METTL3-14, the consensus motif comprises five nucleobases in the DR A CH motif (D=A, G or U R=G or A H=A, C or U underlined A represents the modification site), in which the central A will be modified. 156 METTL3-14 is able to generate N 6 -allyl labeled RNA utilizing the synthetic cofactor allyl-SAM. 138

Recently, Rentmeister and co-workers presented sequence-specific chemo-enzymatic labeling and photocaging of RNA applying the mRNA methyltransferases METTL3-14 and METTL16 (Figure 10). 137 Utilizing the co-substrate promiscuity of the two methyltransferases, they tested different analogs of the methyltransferases’ natural co-substrate AdoMet to transfer functional groups to the N 6 -position of target adenosines within the RNA of interest. 137

Tag-free internal RNA labeling based on mRNA methyltransferases METTL3-14 and METTL-16. The approach of methyltransferase-directed transfer of activated groups (mTAG) can be used for covalent and sequence-specific modification of endogenous RNA. 144 Methyltransferases are applied to transfer various functional groups from S-adenosyl-L-methionine (SAM or AdoMet) analogs to the N 6 -position of target adenosines within a RNA of interest. Therefore, a distinct recognition sequence defining the modification site for the methyltransferase needs to be included in the RNA sequence. Additionally, METTL3-14 is able to transfer photoactive benzylic AdoMet analogs like ortho-nitrobenzyl (ONB) and 6-nitropiperonyl (NP) groups. Upon irradiation with light, the previously attached photoactive modification is removed from the RNA of interest. Both methyltransferases, METTL3-14 and METTL-16 can be used in a combined approach for orthogonal labeling of different sites within the same RNA strand. After sequential modification with a propargyl group and photoactive ONB group, the latter can be removed reversibly by irradiation with light. 137

Therefore, RNA sequences with the respective individual consensus motifs for each methyltransferase were prepared including the previously described DRACH motif for METTL3-14. The consensus motif for METTL 16 comprises the nine nucleobases UAC A GAGAA (underlined A represents the modification site) which need to be posed in a loop or hairpin structure within the target RNA. 157 Testing the transfer of propargyl groups, the selenium-based AdoMet analog SeAdoYn was used which was previously found to be more stable and to improve the transfer efficiency. 158 Both, METTL3-14 and METTL16, were shown to transfer propargyl groups with efficiencies corresponding to either 82 % or up to 69 % relative yield compared to unmodified AdoMet, respectively. METTL3-14 additionally accepted photoactive benzylic AdoMet analogs like ortho-nitrobenzyl (ONB) and 6-nitropiperonyl (NP) groups for transfer with more than 50 % relative yield compared to AdoMet. Moreover, METTL16 was able to transfer an azidobutenyl group, but only yielded low amounts of labeled RNA. 137 Subsequent functionalization of propargylated and azidobutenyl-modified RNA with different reporter groups using either CuAAC or SPAAC, respectively, was successfully demonstrated. 137 Eventually, 1900 nt long firefly luciferase (FLuc) mRNA which naturally comprises the DRACH motif was post-synthetically propargylated and then fluorescently click-labeled with Cy5-azide employing CuAAC. 137 Additionally, innovative RNA photocaging experiments proved that the installation of ONB and NP groups by METTL3-14 on short, modified RNAs is reversible as both modifications can be removed by irradiation with UV light without damaging the RNA. 137 It was further demonstrated, that both methyltransferases, METTL3-14 and METTL-16, can be used in combination for orthogonal labeling of a 26 nt long RNA at different sites. Sequential modification was performed first applying METTL16 to transfer a propargyl group followed by the transfer of an ONB group by METTL3-14. 137 The presented labeling approach is promising for future application such as dual color labeling for FRET studies on endogenous RNA as well as activation of previously photocaged RNA. Site-specificity of the presented approach is limited as the required consensus motifs are spanning only a few nucleobases. Potential interference with consensus motifs naturally present in long, functional RNA sequences 159 is imaginable and will cause unspecific labeling by methyltransferases at undesired sites. Referring to earlier in vivo metabolic labeling experiments with a propargyl-modified amino acid precursor as methyltransferase substrate eventually yielding propargylated RNA, 160 an interesting adaption for site-specific in vivo RNA labeling may be developed in the future. For this, care has to be taken to address undesired cross-interactions with cellular AdoMet or SAM substrates.

Change history

Takahashi, K. & Yamanaka, S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell 126, 663–676 (2006).

Takahashi, K. et al. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131, 861–872 (2007).

Park, I.H. et al. Disease-specific induced pluripotent stem cells. Cell 134, 877–886 (2008).

Grskovic, M., Javaherian, A., Strulovici, B. & Daley, G.Q. Induced pluripotent stem cells–opportunities for disease modelling and drug discovery. Nat. Rev. Drug Discov. 10, 915–929 (2011).

Robinton, D.A. & Daley, G.Q. The promise of induced pluripotent stem cells in research and therapy. Nature 481, 295–305 (2012).

Stadtfeld, M., Nagaya, M., Utikal, J., Weir, G. & Hochedlinger, K. Induced pluripotent stem cells generated without viral integration. Science 322, 945–949 (2008).

Yu, J. et al. Human induced pluripotent stem cells free of vector and transgene sequences. Science 324, 797–801 (2009).

Gonzalez, F. et al. Generation of mouse-induced pluripotent stem cells by transient expression of a single nonviral polycistronic vector. Proc. Natl. Acad. Sci. USA 106, 8918–8922 (2009).

Hu, K. et al. Efficient generation of transgene-free induced pluripotent stem cells from normal and neoplastic bone marrow and cord blood mononuclear cells. Blood 117, e109–e119 (2011).

Jia, F. et al. A nonviral minicircle vector for deriving human iPS cells. Nat. Methods 7, 197–199 (2010).

Narsinh, K.H. et al. Generation of adult human induced pluripotent stem cells using nonviral minicircle DNA vectors. Nat. Protoc. 6, 78–88 (2011).

Okita, K. et al. A more efficient method to generate integration-free human iPS cells. Nat. Methods 8, 409–412 (2011).

Okita, K., Nakagawa, M., Hyenjong, H., Ichisaka, T. & Yamanaka, S. Generation of mouse induced pluripotent stem cells without viral vectors. Science 322, 949–953 (2008).

Si-Tayeb, K. et al. Generation of human induced pluripotent stem cells by simple transient transfection of plasmid DNA encoding reprogramming factors. BMC Dev. Biol. 10, 81 (2010).

Woltjen, K. et al. piggyBac transposition reprograms fibroblasts to induced pluripotent stem cells. Nature 458, 766–770 (2009).

Yu, J., Chau, K.F., Vodyanik, M.A., Jiang, J. & Jiang, Y. Efficient feeder-free episomal reprogramming with small molecules. PLoS ONE 6, e17557 (2011).

Yusa, K., Rad, R., Takeda, J. & Bradley, A. Generation of transgene-free induced pluripotent mouse stem cells by the piggyBac transposon. Nat. Methods 6, 363–369 (2009).

Kim, D. et al. Generation of human induced pluripotent stem cells by direct delivery of reprogramming proteins. Cell Stem Cell 4, 472–476 (2009).

Zhou, H. et al. Generation of induced pluripotent stem cells using recombinant proteins. Cell Stem Cell 4, 381–384 (2009).

Fusaki, N., Ban, H., Nishiyama, A., Saeki, K. & Hasegawa, M. Efficient induction of transgene-free human pluripotent stem cells using a vector based on Sendai virus, an RNA virus that does not integrate into the host genome. Proc. Jpn Acad. Ser. B Phys. Biol. Sci. 85, 348–362 (2009).

Ban, H. et al. Efficient generation of transgene-free human induced pluripotent stem cells (iPSCs) by temperature-sensitive Sendai virus vectors. Proc. Natl. Acad. Sci. USA 108, 14234–14239 (2011).

Warren, L. et al. Highly efficient reprogramming to pluripotency and directed differentiation of human cells with synthetic modified mRNA. Cell Stem Cell 7, 618–630 (2010).

Diebold, S.S., Kaisho, T., Hemmi, H., Akira, S. & Reis e Sousa, C. Innate antiviral responses by means of TLR7-mediated recognition of single-stranded RNA. Science 303, 1529–1531 (2004).

Hornung, V. et al. 5′-Triphosphate RNA is the ligand for RIG-I. Science 314, 994–997 (2006).

Pichlmair, A. et al. RIG-I-mediated antiviral responses to single-stranded RNA bearing 5′-phosphates. Science 314, 997–1001 (2006).

Kariko, K., Buckstein, M., Ni, H. & Weissman, D. Suppression of RNA recognition by Toll-like receptors: the impact of nucleoside modification and the evolutionary origin of RNA. Immunity 23, 165–175 (2005).

Angel, M. & Yanik, M.F. Innate immune suppression enables frequent transfection with RNA encoding reprogramming proteins. PLoS ONE 5, e11756 (2010).

Kariko, K. & Weissman, D. Naturally occurring nucleoside modifications suppress the immunostimulatory activity of RNA: implication for therapeutic RNA development. Curr. Opin Drug Discov. Devel. 10, 523–532 (2007).

Kariko, K. et al. Incorporation of pseudouridine into mRNA yields superior nonimmunogenic vector with increased translational capacity and biological stability. Mol. Ther. 16, 1833–1840 (2008).

Anderson, B.R. et al. Incorporation of pseudouridine into mRNA enhances translation by diminishing PKR activation. Nucleic Acids Res. 38, 5884–5892 (2010).

Liptakova, H., Kontsekova, E., Alcami, A., Smith, G.L. & Kontsek, P. Analysis of an interaction between the soluble vaccinia virus-coded type I interferon (IFN)-receptor and human IFN-α1 and IFN-α2. Virology 232, 86–90 (1997).

Kormann, M.S. et al. Expression of therapeutic proteins after delivery of chemically modified mRNA in mice. Nat. Biotechnol. 29, 154–157 (2011).

Kariko, K., Muramatsu, H., Keller, J.M. & Weissman, D. Increased erythropoiesis in mice injected with submicrogram quantities of pseudouridine-containing mRNA encoding erythropoietin. Mol. Ther. 20, 948–953 (2012).

Warren, L., Ni, Y., Wang, J. & Guo, X. Feeder-free derivation of human induced pluripotent stem cells with messenger RNA. Scientific Reports 2, 657 doi:10.1038/srep00657 (2012).

McElroy, S.L. & Reijo Pera, R.A. Culturing human embryonic stem cells in feeder-free conditions. Cold Spring Harb. Protoc. 2008, doi:10.1101/pdb.prot5044 (2008).


P.K. thanks the National Natural Science Foundation of China (grant no. 31671382) and the Scientific Research Funds of Huaqiao University. F.Q. thanks the National Natural Science Foundation of China (grant no. 32000462) and the Scientific Research Funds of Huaqiao University. Y.C. thanks the Postgraduates Innovative Fund in Scientific Research from Huaqiao University. K.S.-A. thanks the New York University Abu Dhabi Research Institute and NYUAD Division of Science (funds no. 73 71210 CGSB9 and AD060) for support.


We thank Sun-Je Woo for support in the BL3 laboratory, Kwangmi Moon for advice on BL2 virus culture, and Soohyun Jang for technical support. We also thank Dr. Kwangseog Ahn for sharing reagents. This work would not have been possible without invaluable discussions with members of Narry Kim’s lab, particularly Dongwan Kim, Haedong Kim, and Dr. Hyunjoon Kim. The pathogen resource for this study was provided by the National Culture Collection for Pathogens, Korea National Institute of Health for SARS-CoV-2 (NCCP43326), and the Korea Bank for Pathogenic Viruses, Korea University Medical Center for HCoV-OC43 (KBPV-VR-8). This work was supported by the Institute for Basic Science from the Ministry of Science and ICT of Korea (IBS-R008-D1 to V.N.K.) and a BK21 Research Fellowship from the Ministry of Education of Korea (A.S.).

Author contributions

Conceptualization, S.L., Y.L., and V.N.K. Methodology and experiments, S.L., A.S., and Y.P. LC-MS/MS, J.K. and J.-S.K. Data analysis, Y.L. and Y.C. Manuscript writing, S.L., Y.L., and V.N.K. Visualization, Y.L., Y.C., and A.S. Supervision, V.N.K.

Declaration of interests

We have filed a patent relevant to this paper.

Inclusion and diversity

One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in science.

Watch the video: Leveraging the Power of In Vitro Transcription (August 2022).