Microarray Sample Prep

Table of Contents

Letter from the Editor
Index of Experts
Q1: Which sample collection and storage techniques yield optimal RNA?
Q2: What techniques do you use to assess and isolate high-quality RNA?
Q3: How do you establish controls for heterogeneous or difficult samples?
Q4: How do you improve hybridization specificity and sensitivity?
Q5: What tips do you have for clean-up and handling of arrays?
List of Resources

Download the PDF version here

Letter from the Editor

Welcome to the latest installment in Genome Technology's reference guide series. In this second issue, we are pleased to present expert insights on optimizing sample prep for microarray experiments.

Sample prep can make or break your microarray experiment. Whether your samples come from the clinic or the bench, the tips compiled in the following pages will help you maximize the quality of your arrays. Since making their debut, microarrays have gained popularity at a remarkable pace. Commercial chips have found their way into labs everywhere, luring investigators with the promise of generating reliable data on thousands of genes at once. The tool will no doubt continue to reach new users as the costs involved in setting up and running a microarray experiment keep dropping, while dedicated core facilities spring up in both universities and industry.

Microarray technology may be widely available, but important issues remain. From sample selection to data analysis across platforms, groups are working to agree on a number of quality control guidelines for all steps in the process. Only in this way can microarrays go from standard tool to standardized technology.

One of the most critical steps in the process is sample preparation. Accurate and meaningful data is highly dependent on the quality of isolated nucleic acid or protein samples, and these, in turn, are contingent on correct acquisition, extraction, and purification procedures. Learning how to master correct handling of samples takes experience, which can be costly to attain (in both reagents and time).

The microarray experts on our panel offer detailed advice on issues in sample prep. Keep this guide on hand for quick answers to common problems concerning sample collection, isolation, establishing controls, and more.

— Jennifer Crebs

Index of Experts

Genome Technology would like to thank the following contributors for taking the time to respond to the questions in this tech guide.

Philippe Demougin
Biozentrum
University of Basel

Gary Hardiman
Director, BioMedical Genomics Microarray Facility
Assistant Professor, Department of Medicine
University of California, San Diego

André Ponton
Director, Microarray Laboratory
McGill University and Genome Quebec Innovation Centre

Scott Tenenbaum
Assistant Professor, University of Albany-SUNY
GEN*NY*SIS Center for Excellence in Cancer Genomics

Q1: Which sample collection and storage techniques yield optimal RNA?

The key parameter for yielding optimal total RNA quality is to immobilize tissues as fast as possible after sampling. Proceeding fast is crucial because it doesn't let time for RNA level of expression to change because of technical stress, and more importantly, RNases won't have time to get in contact with RNA molecules. Tissues can be fetched into liquid nitrogen or in RNA later (Ambion, Cat# 7020). Either way is good but RNAlater requires the samples to be small so that the solution can get into the tissue and protect RNA. RNAlater is preferred over liquid nitrogen especially in two cases: in surgery room where the use of liquid nitrogen is commonly prohibited and in case further dissection steps are required for isolating a tiny part of tissue (embryo microdissection for instance). Tissues are left in RNAlater overnight at 4°C, then stored at -20°C or -70°C. If snap-frozen in liquid nitrogen, they are stored at -70°C following snap freezing. Proceeding to tissue isolation, the use of RNAlater allows more flexibility. Since the solution penetrated the tissue before storage, samples can be thawed on ice and tissue homogenization is performed with no rush. For tissues stored at -70°C after snap freezing in liquid nitrogen, the situation is more critical: even partial or short thawing can result in dramatic RNA degradation because the tissues are destructurized by the temperature shock and RNases will be in contact with RNA. So in this case tissues must be homogenized (in a solution inhibiting RNase activity) very rapidly going from a frozen stage (transport samples on dry ice) to a completely homogenized stage within seconds.

— Philippe Demougin

I have found that both solid phase and chaotropic salt methods give adequate results. The advantage with chaotropic salt methods is that once released from the cells or tissues, the RNA is less prone to initial degradation due to the inhibitory effects of guanadium and phenol. These can later be a problem as they can interfere with the reverse transcription reaction by inhibiting the activity of the reverse transcriptase enzyme. Often I see RNA that is structurally intact but cannot be converted into high-quality target material due to the presence of contaminants. Ethanol precipitation or clean-up with a solid phase method is required. Solid phase methods work very well in that they generate very clean RNA, but there is greater risk of degradation. This means that an RNase inhibitor must be used. Additionally, the yields of RNA are generally lower.

— Gary Hardiman

We are preparing total RNA from plant tissue. For the microarray analysis and RNA preparation, to avoid the noise of gene expression by cutting, etc., we normally prepare liquid nitrogen before tissue preparation. Then whole plant (control or stresstreated plants) will be frozen in liquid nitrogen and, after completely frozen, [the] RNA preparation process will be started. This kind of treatment can avoid unexpected gene expression during sample preparation process. As for the capture of specific tissue — such as the specific cell layer, specific organ — we will capture the tissue at normal temperature and, after capture, tissue will be frozen in liquid nitrogen.

— Shoshi Kikuchi

Usually it is best to keep the RNA in RNase-free water (commercial source) and put aliquots at -80°C. It is best to avoid multiple freeze-thaw cycles.

— André Ponton

Usually, we use Qiagen's RNeasy prep or TRIzol extractions. I store my RNA in ethanol at -80°C. In my experience, you either have an RNase problem or you don't. If you don't, it's pretty stable. And if you do, it doesn't matter what you do — it's gone. The trick is to not have any RNase problems. My RNAs are usually stable, especially at -80°C in ethanol, they're good for a year, no problem.

— Scott Tenenbaum

Q2: What techniques do you use to assess and isolate high-quality RNA?

Whatever the tissue from which RNA is isolated, we systematically pass total RNA over a silica-based column (e.g. Qiagen RNeasy) in order to clean it up from genomic DNA, salts, phenol. ... We routinely used a Qiagen TissueLyser apparatus (Cat# 85220) for homogenizing the tissues. Tissue is [shaken] at high frequency in a tube in the presence of a single 5mm metal bead. The protocol applied depends on how difficult it is to disrupt the tissue and open up the cells: if the moderate lysis power of the lysis solution (RLT Buffer) provided with the RNeasy kit is sufficient, homogenate is directly passed over the silica columns. This is typically the case for cell culture or soft tissues. But a phenol-based chemical (e.g. TRIzol) is required for most tissues either because they are tougher to break up, richer in proteins, or richer in lipids. One can adjust the time and frequency of homogenization. 25Hz for two minutes is a safe starting point. One should always homogenize the minimum time necessary for a complete homogenization.

— Philippe Demougin

High-quality RNA can be isolated using either chaotropic salt and solid phase methods, or a combination of the two. The RNA can be assessed by spectrometry and gel electrophoresis if quantities permit. The ratio of the absorbance readings at 260 and 280 should be between 1.8 and 2. Typically the ratio of high to low molecular weight ribosomal RNA should be about 2. DEPC-treated water should be avoided for resuspending RNA as this can inhibit enzymatic reactions. Additionally, if there is a need to concentrate the RNA, this should be done by salt precipitation or desiccation with a low heat setting. Once RNA has been dried to completion, it can be very hard or impossible to re-suspend. Tissue provides much more of a challenge for RNA extraction than cell lines. Typically, the material needs to be well homogenized prior to extraction. When using a chaotropic salt method, it is critical to avoid organic phase contamination of the RNA supernatant. After centrifugation, one should avoid taking material near the aqueous organic interphase as this is a source of genomic DNA contamination. Agilent Bioanalyzer analysis is very useful in assessing the integrity of total RNA. The 28 and 18S ribosomal RNA species should be present as strong peaks. There should be an absence of low molecular weight products (indicating degradation) and high molecular weight contaminating genomic DNA.

— Gary Hardiman

[In order to homogenize plant tissue prior to RNA extraction] we are using Multi-Beads Shocker (Yasui Kikai). This product is the most important tool for the preparation of RNA or DNA from hard plant tissues. Many researchers in the plant field out[side] of Japan want to purchase this tool, but [it] is only purchasable in Japan.

— Shoshi Kikuchi

We received all kind of RNA from different tissues and species. In general TRIzol extraction followed by Qiagen column clean-up give good results.

— André Ponton

Q3: How do you establish controls for heterogeneous or difficult samples?

A mini-electrophoresis system such as the Agilent Bioanalyzer or the Bio-Rad Experion allows fast analysis of RNA samples. Most importantly, they allow a precise evaluation of their quality. The integrity of 28S and 18S ribosomal RNA is used to assess the quality of the total RNA sample. The sharper the peaks, the better the quality. A unique feature of the Bioanalyzer is the scoring (from 0 to 10) of the total RNA quality. For the first time researchers have a tool for assessing the RNA quality in an objective manner. RNA quality is very much dependent of the tissue, some tissues being naturally richer in RNases than others. By experience we simply know what pattern and score to get for various tissues. For difficult samples, measures are taken for shortening the times, increasing the homogenization force if necessary, working on ice, and working faster. Getting things organized in order to work faster helps a lot. A large amount of contaminated genomic DNA in an RNA sample can lead to unexpected migration profiles and misleading interpretation.

— Philippe Demougin

The most difficult samples I have found for microarray experimentation are those derived from laser capture experiments, where the yields are low and there may be contaminating genomic DNA present. The extracted RNA should be compared to a control — for example, the Stratagene or Clontech reference RNA. If sufficient RNA is available, optical density readings at 260nm and 280nm (and other wavelengths) using a NanoDrop spectrophotometer will reveal the purity of the RNA and the presence of DNA, proteins, or carbohydrates. The integrity of the RNA can be assessed using an Agilent Bioanalyzer and compared to control RNA samples.

— Gary Hardiman

We are measuring the amount of RNA by [using] NanoDrop OD measurement equipment and by the Bioanalyzer system, [by which] degradation of RNA samples is monitored.

— Shoshi Kikuchi

We use the same standard controls that most people use. Using bacterial spikes and things like that, one can average one sample from the next by gauging off of those types of controls. But I find those to be only partly useful. They tend to control for certain aspects of the technique but not the whole method. For instance, the bacterial spikes will control for labeling and hybridization issues, but not really for earlier RNA quality. There's no way to control for that, because you're adding the control a step after the process occurred. You have to do a lot of repeats. I'm funded through the ENCODE project, which is part of the NIH and the National Human Genome Research Institute, and our standard is three biological replicates, each of which is technically replicated twice. If you do literally six repeats on something, an outlier becomes pretty obviously an outlier. Then you can either selectively remove it or lessen its impact. I think the biggest criticism of the field is most researchers will err on the side of doing more time points, rather than doing fewer time points and more replicates. So they'll do one sample each at six different time points, as opposed to doing three replicates each at two time points. The data in the end is much more believable that way.

— Scott Tenenbaum

Q4: How do you improve hybridization specificity and sensitivity?

We hybridized only Affymetrix arrays. Procedures are very much standardized, and we stick to the default recommendations. Reliable arrays and [automated] procedures being the key advantages of the GeneChips versus spotted arrays.

— Philippe Demougin

For the various commercial catalog arrays, I usually adhere to standard operating protocols, as much of this has been worked out and optimized by the array manufacturer — deviating from the SOPs can mean you are on your own if the experiment fails. For homemade or boutique arrays where optimization is usually required, several components can be adjusted to improve hybridization specificity and sensitivity. Firstly, good oligonucleotide design is key. The investigator needs to ensure that the oligos are specific to the mRNA of interest, and design probes that do not contain fold-back loops. Hybridization conditions are optimized depending on the slide surface employed; sometimes a pre-hybridization step may improve the specificity. The composition of hybridization buffers is important. I favor formamide-based buffers as you can reduce the temperature and maintain hybridization specificity. One of the concerns with aqueous buffers at higher temperatures is the fear of drying out the hybridization solution. As regards sensitivity, improvements are generally seen when you move from static to agitated hybridization. The use of a reciprocal shaker at a low speed setting ensures a wave-like motion and prevents localized target depletion. The ideal is an automated hybridization/fluidic station that ensures consistent hybridization and maximum sensitivity.

— Gary Hardiman

In our case, we are using the oligoarray system produced by Agilent Technologies. We just follow the recommended protocols from them.

— Shoshi Kikuchi

This is not a concern with Affymetrix microarrays as long as the sample quality is very good.

— André Ponton

That's really a platform question. Affymetrix has made great strides in the last several years to clear up issues with their platform and get around the problems. For instance, there are only 25 base probes, but the 11 probes they now use on average per gene are very well selected. Some of the longer probe-based arrays do a better job, and NimbleGen and GE have some interesting products. For the longer 50-60mer probes, the signal to noise is better. But there are issues there too. All [platforms] inherently have a level of noise that I think most traditionally trained, reductionist biologists have to get comfortable with. There are a lot of variables, fairly involved methods, and there's a degree of fluctuation that is inherent in the system. Some researchers have a lot of issues with that; they just don't like that things are bouncing around a lot. The simplest solution I've had for that is I almost always go for the low-hanging fruit: I usually use a fourfold cutoff or greater and it cleans it up pretty quick. It's one of the nice things about working on genomic-scale stuff — there are usually so many things to look for that it's OK if you keep your stringency pretty high, because lots and lots of things will still meet that criterion.

— Scott Tenenbaum

Q5: What tips do you have for clean-up and handling of arrays?

A general precaution when dealing with fluorescent detection of arrays is to avoid getting any particles (which are usually highly autofluorescent) in contact with the arrays. Work in a dust-free environment; use only powder-free gloves; filter all solutions (0.22μm), and make sure they don't get contaminated (discard any turbid solution); use double distilled water and eventually DEPCtreated. Centrifuge the probe before loading the upper part of it onto the arrays, especially if a column-based system was used upstream during the synthesis. Particles from the column are likely to be autofluorescent.

— Philippe Demougin

The investigator should avoid touching the arrays and avoid using powdered gloves, which often create fluorescent artifacts. A clean lab or dust-free area is essential. When washing arrays, the arrays should be fully submerged or edge effects will be seen. When drying arrays, I favor centrifugation over use of nitrogen. It's also not a good idea to process too many arrays at once, as this can lead to noisy arrays. Arrays should be kept separate from each other. By this I mean not exactly side by side on a washing tray. It's not a good idea to have fluorescent target being washed off one array and binding non-specifically to an adjacent array that is also being washed.

— Gary Hardiman

Normally [by] spraying the N2 gas.

— Shoshi Kikuchi

In case of Affymetrix, we follow their recommendations.

— André Ponton

If you're getting in on the field and it's new to you, I would strongly encourage you [to] work with a core facility that has a lot of experience doing it. Core facilities that run one or two microarrays per month are really new at it themselves. You'd be better off working with people that run hundreds of arrays on a regular basis. They are much better at doing the technique and the analysis. I think people who spend a lot of time analyzing microarray data are much better suited to work with when you're getting into this field.

I have too often seen researchers who finally get enough money to run some arrays, but once they get their data back, they're really on their own. They just don't know what to do with the volumes of data one gets. There's a little bit of hand holding, but not really enough. And so they don't make any headway on what could have been some pretty good data. They spend a couple of months scratching their head and they eventually do exactly what I think they shouldn't do, and that is: they take a few genes they're interested in, and say, 'OK, that gene I know about, so I'll go focus on that.' It defeats the whole purpose of doing this big genomic-scale analysis if you're going to look at 30,000 genes and throw them all out to look at one. It seems kind of foolish. So, that would be my big suggestion: Get with a group that can give you the analysis support as well as run a lot of arrays, so they know what's going on.

— Scott Tenenbaum

List of Resources

Our microarray experts referred to a number of products, which we've compiled below. The recommended publications feature even more tips regarding sample prep and experiment design.

Products

Affymetrix: GeneChip Arrays
http://www.affymetrix.com/products/arrays/index.affx

Agilent Technologies: Bioanalyzer
http://www.chem.agilent.com/Scripts/PDS.asp?lPage=51

Oligo Array Kit
http://www.chem.agilent.com/Scripts/PDS.asp?lPage=7307

Ambion: RNAlater
http://www.ambion.com/techlib/resources/RNAlater/index.html

Bio-Rad Laboratories: Experion system
http://www.bio-rad.com/

BD Clontech: Universal Total Reference RNA
http://www.clontech.com/clontech/products/literature/pdf/brochures/UnivRNA.pdf

GE Healthcare Life Sciences: Array products
http://www.amershambiosciences.com/

NanoDrop: ND-1000 Spectrophotometer
http://www.nanodrop.com/products.html

NimbleGen Systems: Array products
http://www.nimblegen.com/

Qiagen: RNeasy Mini Spin Columns
http://www1.qiagen.com/Products/RnaStabilization

Purification/TissueLyser
http://www1.qiagen.com/Products/Accessories/Tissue Lyser/TissueLyser.aspx

Stratagene: Reference RNA
http://www.stratagene.com/products/showCategory.aspx?catId=193

Yasui Kikai: Multi-Beads Shocker (cell disruptor)
http://www.yasuikikai.co.jp/company/e_index.html
Publications

Discovering Genomics, Proteomics,and Bioinformatics
by A. Malcolm Campbell, Laurie J. Heyer (September 2002)
Benjamin Cummings; ISBN: 0805347224

DNA Microarrays and Gene Expression
by Pierre Baldi, Wesley G. Hatfield (October 2002) Cambridge University Press;
ISBN: 0521800226

Microarrays and Cancer Research
by Janet A. Warrington, Randy Todd, David Wong (June 2002)
Eaton Pub Co; ISBN: 1881299511

Microarray Quality Control

by Wei Zhang, Ilya Shmulevich, Jaako Astola(April 2004)
John Wiley & Sons; ISBN: 0471453447

Applying Genomic and Proteomic Microarray Technology in Drug Discovery
by Robert S. Matson (December 2003)
CRC Press; ISBN: 0849314690

DNA Microarrays and Gene Expression
by Pierre Baldi, Wesley G. Hatfield (October 2002)
Cambridge University Press; ISBN: 0521800226