"Transcriptomics & Functional Genomics"
Welcome to Transcriptomics & Functional Genomics News Letter No 7. This editions focus is on RNA Sample Amplification.
Currently only the focus topic from each newsletter is being made available on the internet (please note this material is covered by copy right and permission should be sought to reproduce any content). The full newsletter is available internally via the intranet as a pdf. If you are interested in advertising a seminar or promotion via the newsletter or sponsorship please contact : Dr K Laing (Senior Scientist Intracellullar Pathogen Cooperative).
Transcriptional profiling using the common microarray platforms, whether in-house two colour slide based, Agilent, Affymetrix or Illumina all require microgram quantities of RNA and as such impose severe restrictions on the type of biological samples that can be used by standard non-amplified protocols. However, with an ever-increasing need to utilise smaller and smaller samples and address the relevant questions in the clinic, techniques have been developed to amplify whole RNA populations from material such as clinical biopsy or laser capture microdisection and others. These techniques are now routinely used by many commercial venders. Some of the more common approaches are reviewed here.
I hope that this edition will introduce some of the common methods and concepts routinely in use within SGUL and elsewhere to enable this type of sample to be used.
If you are interested in other topics cover in other newsletters go to:
Ken Laing Intracellular Pathogen Co-operative, Cellular & Molecular Medicine, St George’s University of London
RNA sample Amplification
The concept of sample amplification is one that is now commonplace with the advent of PCR in the 1980s. However, the limitation of such a technique in its application towards expression profiling stems from its exponential nature. This comes about because the product of each round of the amplification acts as template in subsequent cycles of PCR. The consequence of this is that any errors in amplification are compounded during the process, potentially mis-representing any or all of the genes within the RNA population as a whole, resulting in a sequence or copy number dependent bias. The search for a means of amplification has focused upon two very different approaches to the problem and has opened the way forwards with solutions relying upon two commonly used techniques each existing in a multitude of flavours. These approaches are based upon in-vitro transcription1,2,3 or modification of the basic PCR technique4,5,6,7,8 to minimise the potential for bias.
In-vitro transcription as a technique dates back to the 1980s when Melton9 then described the use of SP6 Bacteriophage RNA polymerase in the production of run-off transcripts from linearised clones. But it was not until 1990 that van Gelder1 from James Eberwine’s Lab described the first use of an in-vitro transcription based technique using T7 RNA polymerase to amplify a mixed mRNA population obtained from neuronal cells, an approach that subsequently became known to as the Eberwine technique. The advantage of such a technique is that the product of the reaction is unable to act as template and the yield of any individual species within a mixed population is for the most part determined by the unchanging template concentration. The consequence of which means the technique is conceptually linear and thus would be expected to have low representational bias. In practice this is largely holds true with one important caveat and the relative RNA abundance is maintained. Where this is not the case, the bias appears consistent and systematically reproducible between samples allowing the comparison of similarly amplified RNA10,11,12.
The Eberwine technique itself relies upon the ability of bacteriophage T7 RNA polymerise to utilise linear double stranded template containing a 17 nucleotide minimal T7 promoter sequence to produce run-off transcripts. The essential steps taken to achieve amplification thus require the synthesis of a double stranded cDNA using a modified oligo dT primer. This consists of the minimal promoter sequence flanked 5’ by additional non-essential sequence and an oligo-dT tract at the 3’ end to initiate first strand synthesis (figure 1). The technique is often reported to produce around 300-1000 fold amount of RNA from an input of 1 micro gram of total RNA assuming 1-3% of this is mRNA ie a yield of 10-30 micro gram, however, yields of more than 100 micro gram are sometimes obtained. The technique has gone through numerous rounds of optimisation and modification. For example a large number of different naturally occurring promotor sequences exist and the choice of promoter/primer sequence is important for optimal promotor binding, rate of initiation and formation of a stable initiation complex13,14,15. The optimisation of reaction conditions for performing preparative scale transcription was looked at in the early 1990s16,17 pointing the way toward maximising the yield, traditionally weak tris/HCl based buffered systems were used limiting the reaction as a result of a shift in the pH as a consequence of polymerisation. On the advent of microarray technology application of the technique to gene expression profiling led to a further changes to the standard protocols some of which have been adopted other not15,18. Wang et al3 (2000) suggested the use of a template switch primer for second strand synthesis; the aim here was to promote the production of full-length ds-cDNA template reducing possible 3’ bias in the amplified RNA.
With Wang et al3 came the first critical examination of the technique applied to microarray based expression profiling. This initiated a barrage of others to further examine the technique10,11,12,15,18 along with many others too numerous to mention here. On the whole these have all come to similar conclusions that the technique works well and although it clearly produces some systematic bias in comparison with unamplified RNA the correspondence between unamplified vs unamplified and amplified vs amplified is good. Some have suggested that amplification improves the signal to noise ratio and that this characteristic of amplification compensates for the compression of the relative expression ratios generally observed. Less controversially, however on the whole there is agreement that amplification increases sensitivity and the ability to detect genes expressed at low levels and this accounts for some of the discordance between comparisons carried out between unamplified and amplified material but is by no means the whole answer to this discrepancy.
There are many modifications to the basic in-vitro transcription protocols for microarrays aimed at improving the sensitivity, representation and utility including the direct and indirect labelling of the amplified RNA and its use directly in hybridisation. There are also protocols for two round amplification procedures for use with sub microgram amounts of total RNA. The more common of these variations on the theme are listed in Table 1. The standard amplification technique used in the Ambion, Agilent, Roche and other kits, incorporates the T7 promotor into the 1st strand and produces antisense RNA (aRNA/cRNA). However, an alternative method to this more common approach incorporates the promotor into the 2nd strand synthesis using a T7-template switch primer or T7-random n6-n10 primer, as a consequence in this latter approach the sense strand is produced as a run-off transcript and is an approach used in the Clontech SMART T7 amplification kit. The choice between these two approaches has important implications since most arrays consist of single stranded reporters, usually but not always the sense strand and thus requires a labelled complementary strand for hybridisation.
In-vitro transcription based amplification has a number of drawbacks including its complexity, cost and the time required to complete, a typical protocol involves a lengthy 2 day minimum duration starting from total RNA through to amplified product. With this in mind some have returned to look at PCR based amplification and in particular modifications of the technique to adapt it to perform linear or near-linear amplification of the RNA population. Amongst these approaches include limiting the size of the PCR amplicons4 originally termed global RT-PCR, single primer amplification8 (SPA sometimes referred to as linear PCR amplification) and template switching amplification combined with limited cycling7. The first of these approaches published in 2002 Iscove et al4 is based upon a technique that was devised more than 10 years earlier18, 19. The approach has an underlying assumption that small amplicons of similar size amplified with a single pair of primers will have similar near 100% efficiency irrespective of their internal sequence and that initiation and representation of all molecules is not affected by a time limited RT reaction. Thus, the approach uses oligo dT priming of 1st strand synthesis under conditions which reportedly limit the size of the cDNA to around 300 bases followed by digestion of the RNA and polyA tailing. The single stranded product is thus used as template in PCR reaction of between 30 and 65 cycles (figure 2a). The published results of Iscove et al are impressive with amplification of 107 fold from 10pg of RNA and (despite the high cycle number) in their study maintained high correlation of amplified expression ratios with “true” unamplified ratios. The method does also have several inherent disadvantages in that it requires multiple steps involving RT, polyadenylation and PCR; but perhaps one of the major drawbacks is that the corresponding reporter sequences comprising the array have to be within the terminal 300 bases of the gene, severely limiting the array design and thus its composition in conjunction with this technique. As array design over the last few years has moved toward use of single stranded reporters covering multiple exons spanning the whole gene and its splice variants this type of approach could not be used in conjunction with such arrays. In 2003 Smith et al coined the term SPA8 and this second approach aims at transforming the exponential nature of PCR to a linear or at last near linear technique through unidirectional priming during the PCR cycling. This is done by using a modified oligo dT–non-homologous primer sequence such as the oligo dT-T7 promotor sequence to prime the 1st strand cDNA synthesis followed by conventional 2nd strand synthesis. This material is used in a PCR reaction containing a single primer comprising the non-homologous sequence (figure 2c). In the original paper, 40 cycles were used to amplify from as little as 30ng of RNA. Although amino-allyl incorporation into the amplification product is possible Smith et al converted this antisense product into labelled sense strand DNA using Klenow, which again has implications for use and array design. The study shows SPA as a “reproducible and reliable method” and clearly has the advantage of simplicity over other methods. However, SPA far exceeds the expectation of linear amplification yields, the reason for this is unclear but probably results from a combination of specific and mis-priming events. The third of these PCR based approaches uses global amplification with template switch PCR (TS-PCR21,22) combined with cycle limited amplification. Template switch amplification relies upon the characteristic terminal transferase activity of MMLV reverse transcriptase resulting from a specific mutation in RNaseH domain, this activity causes the addition of non-template nucleotides producing poly-dC tailing of the cDNA which is the used to prime 2nd strand synthesis (figure 2b). Template switching is thus used as a means of increasing the proportion of full-length dsDNA produced during cloning and other procedures requiring whole gene representation, so that in the context of microarray analysis a technique incorporating template switch should allow better 5’ representation within the amplified population as compare to other methods.
Although template switch priming has been previously used in microarray analysis, none had systematically considered how amplification affected fidelity until Petalidis et al did so in 2003. Under the conditions used in their study, amplification was exponential up to 17 cycles; when the maximum number of genes were detected above background and about 72% of amplified expression ratios compared to “true” unamplified ratios were identified, a comparative level not dissimilar to several other studies looking at in-vitro transcription and other PCR based methods.
As with in-vitro transcription, PCR based amplification also results in a compression of the ratios compared to standard reverse transcription but compression also accompanies reduced variance and greater statistical resolution. Both in-vitro transcription and PCR demonstrate increased number of genes detected above background and are supported by not wildly dissimilar reports of fidelity. However, it has to be said that PCR based methods are generally viewed with healthy scepticism rightly or wrongly and in-vitro transcription based amplification is the most common technique being routinely applied in the microarray field as an answer to the requirement for amplification of RNA from small samples. T7 based amplification is a technique universally applied to all microarray platforms including one and two-colour glass slide arrays, Affymetrix and Illumina and is the standard approach used for Affymetrix arrays in preference to direct labelling of cDNA using reverse transcription. Even so the reader should be aware there is no definitive studies comparing these methods on the same platform/array/probe set or with the same sample material and that studies have been carried out over broad time scale, the consequence of which it is difficult to truly compare these techniques particularly as the base line comparator itself, reverse transcription inevitably encompasses inherent errors in the results it produces. Does it really matter? Ultimately validation by a range of techniques is important in any scientific approach not least gene expression.
There are currently three or four main entrants in the market providing off the shelf kits for RNA amplification and almost all have adopted in-vitro transcription as the basis of their kits (see box below).