Bioinformatics.cau.edu.cn

Isolation of membrane-bound polysomal RNA
Maximilian Diehn
Although DNA microarrays are most often used for genome-scale analyses of gene expression, they can in principle be used to compare any nucleicacid populations that can be physically separated. Thus, one importantbiological phenomenon that can be studied with DNA microarrays is thesubcellular compartmentalization of mRNA. There are numerous documentedexamples in eukaryotes ranging from yeast to human, where the localization ofRNA within a cell plays an important biological role in the functioning of thatcell or organism. By serving as an assay for the presence or absence of thousandsof transcripts, DNA microarrays can rapidly accelerate the identification oftranscripts in various subcellular compartments. Given the biologicalimportance of such subcellular mRNA distributions it is clear that studying themwith DNA microarrays should lead to a deeper understanding of many cellularprocesses.
One of the most fundamental examples of differential subcellular localization of RNAs is the distinction between mRNAs that encode cytosolic ornuclear proteins and those that encode membrane-associated or secretedproteins. The proteins in the latter group are co-translationally inserted into therough endoplasmic reticulum (rER) by ribosomes attached to the cytoplasmicsurface of the rER while the first group is translated by ribosomes that are free inthe cytosol of the cell. This spatial separation of protein synthesis creates distinctpools of mRNAs that can be separated based on the presence or absence of anassociation with cellular membranes. Once separated, the two fractions can behybridized to DNA microarrays and subsequent analyses can yield twoimportant types of information: 1) the subcellular localization of proteinsencoded by the mRNAs can be inferred from the subcellular localization of theirtranscripts and 2) unexpected associations of mRNAs with cellular membranescan be identified and marked for further study.
There are a number of potential approaches for separating membrane- associated and free mRNA. One type of approach takes advantage of therelatively low density of cellular lipid structures by separating these from densercomponents by sedimentation equilibrium (Mechler 1987). A second set ofmethods uses differential detergent extractions coupled with sedimentationvelocity centrifugation (Stoltenburg et al. 1995) to separate the two populationsof mRNAs. It is also possible to isolate rER associated mRNAs via antibody-based purification protocols that are directed against components of the signal-peptide recognition machinery. The protocol described below employssedimentation equilibrium centrifugation in a sucrose gradient. While theseparation instructions are specific to this approach, the method of analysis thatis described can be applied to any experimental approaches that sub-fractionateRNAs.
One important consideration in designing experiments studying the subcellular compartmentalization of RNA is the inclusion of control elements inthe DNA microarrays to be used. In this case, elements representing transcripts of known subcellular localization serve as controls for two important reasons.
First, the success of the fractionation procedure can be assessed by examining thefluorescence ratio distribution for the genes of known subcellular localization.
Secondly, the known genes allow calibration of the relationship between themeasured fluorescence ratio and the likelihood that a given element encodes aprotein with a given type of localization. In the case of experiments withmembrane-associated polysomes, DNA microarrays should contain as manygenes encoding proteins of known subcellular localization as possible.
Minimally, approximately 500 of the array elements should be of this type.
MATERIALSImportant: To minimize risk of contamination of stock solutions with RNases useDEPC-treated water wherever water is needed.
Buffers and SolutionsCycloheximide (10 millig/ml) CAUTION: toxic; Store at -20<°>C.
Add cycloheximide to a final concentration of 10 microg/ml just beforeuse.
Note: 200 proof ethanol can contain fluorescent contaminants and shouldnever be used when processing samples that will be analyzed on DNAmicroarrays.
10 mM Tris-Cl, pH 7.4Hypotonic-lysis buffer (10 mM KCl, 1.5 mM MgCl , 10 mM Tris-Cl, pH 7.4) Add cycloheximide to a final concentration of 10 microg/ml just beforeuse.
Autoclave before use.
Gradient buffer (150 mM KCl, 5 mM MgCl , 50 mM Tris-Cl, pH 7.4) This solution is extremely viscous and needs to be made carefully. Add855.75 g sucrose in approximately 150 g increments to 250 ml 65<°>CGradient buffer in a 2 L beaker. Stir on a hot plate using an appropriatelysized stir bar and keep the temperature of the solution near 65<°>C. If thesolution becomes too viscous, add a small amount of 65<°>C gradientbuffer. Once all of the sucrose has been added, adjust the final volume to1 L using 65<°>C Gradient buffer. It is possible although not necessary tomeasure the concentration of sucrose in the solution using a refractometerand to adjust the solution accordingly. Each gradient requires 15 ml of the2.5 M sucrose buffer. Just before use in a gradient, add cycloheximide to afinal concentration of 10 microg/ml.
1.95 M sucrose and 1.3 M sucrose gradient buffers Make these solutions by appropriate dilution of the 2.5 M sucrose stockwith gradient buffer. Each gradient requires 13 ml of 1.95 M sucrosebuffer and 6 ml of 1.3 M sucrose buffer. Just before use in a gradient, addcycloheximide to a final concentration of 10 microg/ml.
Trizol-LS<> (Life Technologies, Inc. Cat. No. 10296)Special EquipmentBarrier tips for micropipettesBall bearing homogenizer or 5-10 ml glass dounce homogenizer2 ml microcentrifuge tubesUltra-clear 25x89mm centrifuge tubes (Beckman Instruments, Inc. Cat. No.
344058)SW-28 ultracentrifuge rotorUltracentrifuge18-gauge, 1 inch needles50 ml Oak Ridge centrifuge tubes (Nalge Nunc International)Roller bottles or spinner flasks for tissue cultureSpectrophotometer for OD 260 nm measurementsHemacytometer PROTOCOL STEPSPreparation of tissue culture cells 1. Grow up tissue culture cells using conditions appropriate for the cells of interest. Since a large number of cells is required for each gradient (5x108),it is advisable to grow cells in roller bottles or spinner flask. Keep cellsnear a concentration of 5x105 cells per ml since overcrowding may lead todown-regulation of protein synthesis and therefore lower yields ofpolysomal mRNA.
2. Add cycloheximide to a final concentration of 10 microg/ml to the cell culture and return to incubator or warm room. Incubate cells at 37<°>Cfor 5-10 minutes and proceed with pelletting procedure. If cells weregrown on plastic petri dishes, remove media, add a small volume of icecold PBS and scrape from dish using a rubber policeman. Transfer to 250ml centrifuge tube and proceed as with suspension cells.
3. Pellet cells in 250 ml centrifuge tubes for 10 minutes at 1000g (4<°>C).
From this point on all steps should be performed at 4<°>C or on ice.
4. Wash cells 2 times with 125 ml of ice cold PBS supplemented with 10 microg/ml cycloheximide. After the last wash, resuspend each pellet in10 ml ice cold PBS containing 10 microg/ml cycloheximide.
5. Pool all cells into one 250 ml centrifuge tube and count cells using a hemacytometer. Aliquot the equivalent of 5x108 cells to a 15 or 50 mlconical. Pellet cells as before. Note: It is possible to stop here and freezethe cell pellet on liquid nitrogen. Store pellets at -80<°>C. Cell pellets canbe used for up to at least 6 months after freezing.
Hypotonic Lysis and Gradient Construction 6. Resuspend cell pellet to 2.5x108 cells/ml using 2 ml of ice cold hypotonic- lysis buffer. Allow cells to swell on ice for 5-10 minutes. Note: It may benecessary or desirable to include RNase inhibitors in the lysis buffer andsubsequent gradient fractions. Since cell lines and tissues vary in theamount of RNase activity this decision must be made on a case-by-casebasis.
7. Dounce homogenize cells with 10 strokes in a dounce homogenizer or alternatively with 10 passes through a ball bearing homogenizer. Ifdesired, freeze a small aliquot of the lysate at -80<°>C for later analysis astotal cellular RNA.
8. Centrifuge lysate for 2 minutes at 2000g (4<°>C) to pellet nuclei and un- lysed cells. The pellet will be very soft and of varying size depending onthe cell type. Remove supernatant and adjust volume to 2 ml using icecold hypotonic-lysis buffer.
9. Add 2ml of lysate to 11 ml of 2.5M sucrose gradient buffer in a 50 ml 10. Construct sucrose step gradient in Ultra-Clear 25x89 mm centrifuge tubes (Beckman) for an SW-28 ultracentrifuge rotor. First place 4 ml 2.5Msucrose gradient buffer in bottom of tube. Next, carefully layer the 13 mlcontaining the lysate from the previous step onto the 2.5M sucrosecushion. Finally, successively layer 13ml 1.95 M sucrose gradient bufferand 6ml 1.3M sucrose gradient buffer onto the gradient (see Figure 1.) 11. Centrifuge for at least 5 hours at 90,000g in an ultracentrifuge. Note: It is critical to balance tubes to be spun in an ultracentrifuge. If an evennumber of gradients is being prepared, balance them on a scale using 1.3M sucrose gradient buffer. If an odd number of gradients is being set up,prepare a mock gradient using 2 ml of water instead of lysate in step #9.
After the spin, small membrane particles should be visible at the interfaceof the 1.95 M sucrose and 1.3M sucrose sections.
12. Harvest gradients by puncturing the bottom of the centrifuge tubes with an 18-gauge needle and collecting 1.5 ml fractions in 2 ml microcentrifugetubes. Alternatively, to minimize the risk of contamination with freeRNA, the membrane fraction can be isolated by successively removing 10ml from the top of the tube using a P1000 pipetman. The remainder of thegradient is then harvested by the puncture method. Note: To hold theultracentrifuge tubes during the harvesting of the gradients carefully cut apolypropylene 50 ml conical in half using a sharp knife or razor blade anddiscard the top half of the tube. Next cut off the very tip of the conicalsection remaining on the bottom of the tube. This device can besuspended over the bench using a standard clamp and ring stand and hasthe right diameter to snugly fit an ultracentrifuge tube. To collect thefractions just manually slide a microcentrifuge tube rack containing open2 ml microcentrifuge tubes under the punctured gradient tube.
13. Measure the absorbance of each fraction at 260 nm using a spectrophotometer to determine presence of nucleic acid (see Figure 2).
Free ribosomes and free mRNA will be present in the load zone whilemembrane-associated ribosomes and mRNA will be at the interfacebetween the 1.95 M and 1.3 M sucrose steps.
14. Separately pool load zone fractions (Free RNA) and 1.95 M/1.3 M interface fractions (Membrane-associated RNA) based on the OD 260 nmmeasurements and isolate total RNA from each with Trizol-LS<>Reagent. Use 3 parts Trizol-LS<> to 1 part sucrose solution and perform procedure in 50 ml Oak Ridge centrifuge tubes. Follow the protocolsupplied by the manufacturer and include the modified isopropanolprecipitation for proteoglycan and polysaccharide contamination.
15. Resuspend the final RNA pellet in 400 microl of 10 milliM Tris-HCl, pH 7.4 and perform an additional ethanol precipitation by adding 60 microl of2 M Sodium Acetate and 1 ml of 95% ethanol to the RNA in amicrocentrifuge tube. Perform one wash using 500 microl of 70 % ethanoland resuspend pellet in 40 to 200 microl of water. Yields of RNA in thetwo fractions can vary significantly depending on the cell line or tissue inquestion. The membrane-associated RNA yield can range from 10 to afew hundred microg while the free RNA yield is generally 10 to 150 timeslarger. RNA can be stored at -80<°>C for up to at least one year.
16. Perform array hybridization as described in Chapter ?. If you are using a DNA microarray, label the membrane-associated RNA with Cy5 and thefree RNA with Cy3 and hybridize them to the same array. In general, ahybridization using 30 to 50 microg of total RNA as input will work wellon a DNA microarray. If less then this amount is recovered from thegradient in the membrane fraction, it is desirable to amplify the twofractions using an in vitro transcription based amplification strategy(Chapter ?; Wang et al. 2000).
While these experiments can be analyzed using standard supervised and unsupervised methods, these data also lend themselves to specialized analysesthat allow the generation of "confidence estimates" for the likelihood that a givengene encodes a membrane-associated or free polypeptide (Diehn et al. 2000.) Thefirst step in such analyses is the identification of genes included on the array thatencode proteins that have empirically documented subcellular localization data.
Depending on the organism in question, there are various publicly accessibledatabases that curate and provide these data. Among these are SWISS-PROT(http://www.expasy.ch/sprot/), the Proteome databases(http://www.proteome.com/databases/index.html), and resources thatsynthesize data from multiple sources such as NCBI's LocusLink(http://www.ncbi.nlm.nih.gov/LocusLink/) and Stanford's SOURCE(http://genome-www.stanford.edu/source/). Using one or a combination ofthese databases, it is straightforward to generate a list of elements on the arraythat encode known membrane-associated (i.e. secreted, plasma membrane,nuclear membrane, vesicular, Golgi, or ER-associated) or free (cytoplasmic ornuclear) polypeptides.
The array elements with known subcellular localization can then be used to generate a calibration curve relating Cy5/Cy3 ratio to the fraction ofmembrane-associated genes at that region of ratio space (Figure 3A.) First, anappropriate window size (N) for the average calculation needs to be chosen. Awindow size of 75-200 elements works well, but it may need to be smaller if onlya small number of control elements were included on the microarray. Next, thefraction of membrane-associated proteins for N adjacent genes in Cy5/Cy3 ratiospace is computed and plotted as a function of the central gene in the window.
The N gene window is then moved by one gene on the Cy5/Cy3 axis and thefraction is re-calculated. This process is reiterated until the end of the Cy5/Cy3 distribution is reached. The resulting distribution should look roughly like theexample show in Figure 3B. By fitting a low degree polynomial function to thedistribution, an equation can be generated which allows direct conversion ofCy5/Cy3 ratios to the fraction of membrane-association proteins and hence anestimation of the likelihood that an unknown gene with this Cy5/Cy3 ratioencodes a membrane-associated protein.
In addition to analyzing single experiments in this fashion, data from multiple experiments can further refine the presumptive subcellular localizationof the protein products of mRNAs. For example, after analyzing membrane-associated mRNAs for multiple conditions or in different types of tissues, a list ofpotential membrane-associated proteins can be generated by including all geneswhich were present in the greater than 90% membrane-associated region on atleast one array or in the greater than 80% region on at least two arrays. The exactplace at which to draw the cutoff should be tailored to the specific goals of theapplication in question.
FIGURE LEGENDS
Figure 1. Schematic of sucrose gradient construction and RNA isolation.
Figure 2. OD 260 nm profile of sucrose gradient fractions. In this example,
MOLT 4 T cells were fractionated as described. The curve shows the OD 260 nm
measurements for each of the 1.5 ml fractions that were isolated using the
puncture method, numbered in ascending order from the bottom of the tube to
the top. Note the two peaks separated by a large region of low OD 260 nm
fractions. The right peak corresponds to the membrane-fractions while the left
peak makes up the free fractions. In this case fractions 1 through 10 were pooled
as the free fraction and 17 through 22 were pooled as the membrane fraction.
Figure 3. Generation of calibration curves for relating fluorescence ratio to the
fraction of mRNAs encoding membrane-associated proteins. (A) Schematic of
moving average algorithm for a hypothetical experiment containing 10 elements
representing genes encoding protein of known subcellular localization. The first
two iterations of the moving average calculation are shown. (B) Actual
calibration curve from an experiment using MOLT 4 T cells. The microarrays
used contained over 20,000 elements, of which approximately 1000 had
annotated subcellular localization data in Swiss-Prot. The window size was 150
elements. The horizontal red line represents the overall fraction of mRNAs
encoding membrane-associated proteins in the set of genes of known subcellular
localization.
REFERENCES
Diehn M, Eisen M.B., Botstein D, and Brown P.O. 2000. Large-scale identification
of secreted and membrane-associated gene products using DNA
microarrays. Nat Genet 25: 58-62.
Mechler B.M. 1987. Isolation of messenger RNA from membrane-bound polysomes. Methods in Enzymology 152: 241-248.
Stoltenburg R., Wartmann T., Kunze I., and Kunze G. Reliable method to prepare RNA from free and membrane-bound polysomes from different yeast
species. Biotechniques 18: 564-568.
Wang E., Miller L.D., Ohnmacht G.A., Liu E.T., and Marincola F.M. High-fidelity mRNA amplification for gene profiling. Nat. Biotechnol. 18: 457-459.

Source: http://bioinformatics.cau.edu.cn/microarray/pdfs/MembraneBoundPolysomes.pdf