20.109(S09): cDNA synthesis and microarray (Day7)
Today you will use one tool, a DNA microarray, to simultaneously examine the expression of many genes. DNA microarrays are slides with DNA sequences spotted in a known order on the surface. The spots of DNA, each one smaller than the period at the end of this sentence, are placed on the slide surface with robotic arms or built one base at a time with photolithography. Each spot of DNA gets a unique address on the slide surface, and the identity and location of each spot get stored in the computerized “design file” for the array. The slide shown below is the same size as the one you’ll use (1 x 3 inches) but yours will have 4 arrays each with 44,000 spots of DNA instead of the 250 dots shown!
Consider the two spots highlighted on the microarray shown above (an array of human genes for this example). The first spot, in Row 6 Position 30, is a 60-nucleotide sequence from the human gene for glyceraldehyde-3-phosphate dehydrogenase (GAPD). This gene, which encodes an essential metabolic enzyme, has been called a “housekeeping” gene since it must be expressed in all human cells no matter how specialized. Other housekeeping genes include those for ACTB (encoding a cytoskeletal protein), TBP (encoding a general transcription factor), HPRT (encoding an enzyme required for nucleotide transport and metabolism), and PPIA (encoding an enzyme important for protein folding). The second spot highlighted on the array, in Row 4 Position 10, is a 60-nucleotide sequence from the human TERT gene. This gene encodes the protein subunit of telomerase, an enzyme that adds telomere repeats (TTGGGGTTG) to the end of chromosomes. As healthy cells age and divide, telomere repeats are lost. Cancerous cells express telomerase and so the telomeres do not shorten. Consequently, these cells “lose track” of how old they are and become immortal.
The GAPD and TERT spots can be used to illustrate how microarray data is generated and interpreted. Consider a group of “normal” human cells and a cancerous version of them. RNA from each type of cell can be isolated (you’ve seen how quick and easy it is to isolate RNA), converted from RNA into a complementary strand of DNA (called cDNA), and then “color coded.” The most commonly used molecules for color-coding are the green-fluorescing cyanine 3 (Cy3) and the red-fluorescing cyanine 5 (Cy5).For this example, the normal cells get green and the cancerous cells get red. The two colored samples are mixed and then simultaneously hybridized to a DNA microarray. The DNA spotted on the surface of the slide is in vast excess to either colored cDNA sample and so the intensity of each color will vary with the amount of RNA originally present in each sample. A gene expressed similarly in normal and cancerous cells, like the housekeeping GAPD gene, will give rise to a yellow spot in Row 6 Position 30 since equal amounts of green and red cDNA will be bound there and the merged color will appear yellow. By contrast, only red cDNA will bind at Row 4 Position 10 since cancerous cells express telomerase and normal cells do not.
NOTE: The cDNAs are not really piled on top of one another on the array. Rather they are hybridized side by side to the spot of DNA that is on the surface of the slide.
With an expensive machine, the slide is “scanned” to measure the intensity of the red and green light at each spot (remember we’re talking about thousands of spots!) and the data can then be assessed and normalized. Corrections are often made to account for differences between Cy3 and Cy5 incorporation into the cDNA as well as how much of each fluorescent molecule sticks non-specifically to different areas of the slide. These are things you will do next time with your own data.
Today you will convert the RNA isolated from your mouse cells into cDNA and hybridize the cDNA to a DNA microarray.
Part 1: cDNA synthesisCreating cDNA from RNA is done using an enzyme called reverse transcriptase. Like all DNA polymerases, this enzyme can only add sequence to an existing chain and so needs a short “primer” to begin synthesis. To perform the cDNA synthesis, you will use a kit from a company named Genisphere. The primers in this kit have a special “capture sequence” at their 5’ end. The capture sequences allow the cDNA to be reacted with Cy3 or Cy5 later, as outlined in the diagram below.
Before you begin today’s protocol, prepare your bench for working with RNA. This involves cleaning your pipetmen, retrieving your “RNase FREE” pipet tips and solutions and wiping down your bench. You should work on a fresh piece of benchpaper and remember to wear gloves when working with RNA.
For homework, you should have calculated the volume of RNA needed to obtain 1-5 μg of RNA, equal amounts for both samples, and in up to 10 μL each. If you have questions or want to double-check your work, feel free to ask the teaching faculty. Note: the cap color refers to the emission wavelengths of Cy3 and Cy5 (red and far-red), not the excitation (green and red).
|TUBE A||TUBE B|
|RNA||up to 5 ug of RNA (equal to below) from the control sample||up to 5 ug of RNA (equal to above) from the experimental sample|
|RNase-free H20||bring volume to 10 ul||bring volume to 10 ul|
|RT primer if you were Day 5 "Group 1"|| 1 ul Capture Sequence I |
vial 11, red cap
| 1 ul Capture Sequence II |
vial 11, blue cap
|RT primer if you were Day 5 "Group 2"|| 1 ul Capture Sequence II |
vial 11, blue cap
| 1 ul Capture Sequence I |
vial 11, red cap
- Heat the annealing reactions to 80°C for 10 minutes then place the tubes on ice for 2 minutes.
- Microfuge the tubes briefly to spin any condensation or droplets down to the bottom of the tube then add 1 ul of Superase•In, an RNase inhibitor (Vial 4), and 8 ul of cDNA synthesis cocktail. Because reverse transcriptase is an unstable enzyme, this cocktail must be prepared just before use. The teaching faculty will prepare some for you when you are ready for it.
- Pipet the contents of your tubes up and down to gently mix, then incubate the cDNA synthesis reactions at 42° for 1.5 hours. During this time, work with the sample array data file that is available (see Part 2 of today’s protocol).
- Microfuge the tubes briefly then add 3.5 ul of 0.5 M NaOH/0.5 M EDTA to each tube and pipet up and down to mix. Heat to 65°C for 10 minutes. This step will denature your RNA/cDNA hybrids and degrade the RNA.
- Add 5 ul of 1M Tris, pH 7 to neutralize the contents of each tube.
Part 2: practice array data analysis
Data for these exercises should be downloaded from this link. Your main goals today are to familiarize yourself with the appearance of data from microarray experiments, understand the concepts of background correction and normalization, and begin to determine the level of manipulation you would like to perform on such data. For your own benefit, you should probably take some notes on microarray data analysis in your notebook and/or Excel sheet, for reference when you analyze your real data next time. However, these will not be graded as part of your laboratory notebook today.
Small data set
Begin with the sheet called “Sample Data” which contains all the data for 21 of 22,000 spots on a tested yeast array. This sheet will familiarize you with the headings and entries you can expect to see when you examine your own data. Some important landmarks are
- Columns C and D give the address for each gene on the array
- Columns J and K list the gene name associated with each spot
- Column L gives some useful information about each gene’s function
Please answer the following questions about this data:
- Where on the array is the spot for the MFA2 gene (column and row)?
- What is the systematic name for SLG1?
- What is the cellular function for SNF2?
To look at the signal intensities for these 21 spots, you should scroll right. First, you may want to select Window → Freeze Panes while column L is selected. This allows you to see the signal intensities and gene names at the same time. Column AF reports the mean signal from the green fluorescent molecule Cy3 and Column AG reports the mean signal from the red fluorescent molecule Cy5. Columns AH and AI report the median signals. Which of these values (mean or median) should be affected less by outliers?
Use the Format menu to convert the values in the four intensity columns to numbers with no decimal places, and then answer the following questions.
- What are the green and red signal intensities for MFA2?
- What genes give the highest and lowest values in Column AF?
- Are these the same genes that give the highest and lowest values in Column AG?
- What fraction of the listed values have a larger value in Column AF than in Column AG?
- To approach this question, you may want to explore the use of Boolean statements in Excel, which can be very useful in data analysis. The basic format for an IF statement is IF(logical_test,value_if_true,value_if_false). For example, the statement =IF(A1=B1, “equal”, 0) tests if the values in cells A1 and B1 are equal, then returns the text string equal if they are, or the value 0 if they are not.
- Using the values in Column AH, find the mean and the median of the three SDC25 signals. What does the level of agreement of the mean and median tell you about the three values?
- Find the mean and the median for all the values in Column AH and AI. What is the significance that the mean and the median values are not identical? What is the significance that the mean value for Column AH is not identical to the mean value for Column AI?
Larger data set
Now you are ready to look at a bigger data set and practice some analytical methods. Look at the second sheet called “Test Array” in the Excel file. This sheet has a subset of the data (9 of the 86 columns) for a subset of the spots (1,500 of the 11,000) from a single microarray experiment.
Some of the data analysis you will perform is
- background subtraction to correct for signal intensity in areas of the array that do not have DNA spots
- normalization to correct for the physical and chemical differences in Cy3 and Cy5, and
- log2 transformations to avoid fractions when expressing signal ratios
Because microarrays are physically small, signal artifacts routinely arise. These artifacts come from tiny droplets with fluorescent molecules that remain on the array, and from scratches on the surface of the slide. Even the light that leaks into some scanners can make parts of the array appear more green or more red. The column headings in your spreadsheet that include “BG” have background measurements and these values can be used to correct the signal intensities for background artifacts.
- Determine the average red and green background signals. Do this for Column F and G (the mean signals) as well as for Column H and I (the median signals).
- Does the ratio of red and green average background intensities mirror the difference in the signal intensities themselves (Columns B and C vs F and G for example)?
- Find one green background measurement that is considerably different from the average. Is the red background measurement also different? How could you explain this?
- Insert two new columns after the background signal columns and calculate the background corrected values for the green and red signals. These corrected values are determined by subtracting the background measurement for each spot from the signal measurement.
Note that some low-intensity spots on the array may be equal to or even less than background. To avoid later artifacts in data processing, you will ultimately need to discard these values or make them very small but not equal to zero. Think about the different ways you might do this, then continue to normalization.
Now you can “normalize” the data. Many normalization methods have been suggested since microarray technology was introduced. We will practice a “global normalization” method that assumes the Cy3 and Cy5 fluorescent intensities differ by a constant factor,
R = kG where R = red (Cy5) and G = green (Cy3)
One way to determine k is to label the same RNA sample with either Cy3 or Cy5 and then compare the mean signal intensities observed on an array. Since microarray experiments are expensive to perform, this direct comparison is not often done. Instead it is assumed that arrays have the same amount of total mRNA for two samples and the difference in overall intensity is k.
- Use the mean signal intensities (data in Columns B and C) from the Test Array to calculate the average intensity for the green and red signals. What is k?
- Now use the median signal intensities (data in Columns D and E) to calculate k. Is there a difference when you calculate k using the mean and the median signal intensities? What about if you use the median of the means instead of the mean of the means, or similarly for the medians? Finally, what if you create a new column that individually calculates each intensity ratio, and then take the mean or median of that? Which of these k's should be the most reliable?
- Using the calculation method (or two) that you think is best, find k for your background-subtracted data. Why might you want to do normalization after background subtraction? On the other hand, what potential problems does background-subtracted data pose?
- Given the advantages and disadvantages above, one final option is to use a subset of your data for normalization purposes, using a Boolean statement to exclude intensities below a certain threshold. Recall that you can output a text string as the return value in such statements: " " will simply leave a cell blank, so it won't be taken into account later when calculating means or medians. You may want to learn about the AND and OR statements at this time. Note that Boolean statements can be nested.
- Example of a nested logic statement: =IF(=AND(A>5, B>5), A/B, " "). If both A and B are greater than 5, the ratio A/B is returned.
- Once you have decided on a k value, make a new column in which to calculate your normalized red intensity values by multiplication.
So far you’ve seen that microarray data must be normalized to correct for Cy3 and Cy5 differences as well as “background subtracted” to correct for artifacts on the slide. Recall that microarray experiments are designed to simultaneously compare the expression of many genes in two samples. The corrected intensities can be expressed as a ratio between the corrected signals for the two samples (Red/Green). A ratio of 4 means 4-fold gene induction and a ratio of 0.25 means four-fold repression of that gene.
To avoid the decimals associated with gene repression, the log2 of the ratios is useful. Four-fold induction is reported at log2(4) = the power of 2 needed to get 4 = 2. Four-fold repression is reported as log2(0.25) = the power of 2 needed to get 1/4 = log2(1) – log2(4) = -2. Log2 transformed data makes more sense graphically since a 4-fold induction and a 4-fold repression have the same value but different signs (i.e. +2 and –2).
- Add another column to the Test Array called “Red/Green” and calculate the ratio of the background-corrected (and normalized) red signal to the background-corrected green signal. What is the average value for the column?
- Once again, if you are comfortable with Boolean statements you may want to calculate the ratio only when your intensity values exceed a certain threshold. The question now becomes, do you want both values to exceed the threshold (AND statement) or only one (OR statement)? What kind of data will you lose if you use an AND statement? What kind of artifacts will be introduced if you use an OR statement? Can you write a compound statement that keeps all interesting data but also excludes all artifacts?
- If you are not comfortable with complex Boolean statements, you can simply ignore error messages for now. After you sort your data, they will be grouped together and easy to exclude from analysis.
- Add another column to the Test Array sheet called “LOG2 R/G” and transform the “Red/Green” data to LOG2 values. What is the average of this column?
- The exact same Boolean statement you used in the previous question can be used here. Unfortunately, if you try to take the log of a blank cell, you get an error, not a blank cell, and this will preclude you from taking an average.
- Sort the data by selecting Data → Sort while selecting all the columns, then "sort by" your LOG2 R/G column. Which 5 genes in this data set are most strongly induced and which are most strongly repressed?
- Draw a histogram that plots the sorted LOG2 values against an arbitrary axies. This gives you a quick visual of the relative number of up-regulated, down-regulated, and not differentially expressed genes.
Part 3: hybridize microarrays
The arrays we will use are the mouse whole genome microarrays from Agilent. Each slide has four arrays , each with 44,000 60-mer oligonucleotides. These “oligos” were first built on glass wafers and then printed onto the slide surface. The oligos represent more than 41,000 mouse genes, many spotted on the slide more than once.
The success of your experiment is absolutely dependent on the following:
- You must hold the slides by the edges only. If you touch the array that is printed on the slide’s surface, you will obscure the DNA that is printed there.
- Each array has a barcode printed on some stickers on one end of each slide. The array is printed on the side of the slide that says “Agilent.” If you try to hybridize your cDNA to the numbered side of the slide, there will be no array there to bind.
To hybridize the arrays
|team color||Agilent slide|
|Red||Slide 1, Array 1|
|Orange||Slide 1, Array 2|
|Yellow||Slide 1, Array 3|
|Green||Slide 1, Array 4|
|Blue||Slide 2, Array 1|
|Pink||Slide 2, Array 2|
|Purple||Slide 2, Array 3|
|Grey||Slide 2, Array 4|
- Begin by mixing the cDNA pools you have synthesized into one eppendorf tube. Why can we pool these now and not before? The total volume should be 57 ul. Add 57 ul of 2X Hybridization Buffer. Pipet up and down several times to mix the contents.
- Heat your hybridization solutions to 80° for 10 minutes then cool them to room temperature. During this time you will be shown how to assemble the hybridization chambers.
- Load your samples into the hybridization chambers as follows:
- Open the gasket and place the "Agilent" sticker facing up in the rectangular side of the holder.
- Pipet 110 ul of the hyb solution into place 1, 2, 3 or 4 (see figure) according to the information in the table above. All four groups must do this before proceeding to the next step.
- Place the microarray "Agilent" sticker down onto the hybridization solution that's in the gasket.
- Slide on the top portion of the hybridization chamber and the brace, then tighten completely.
When every group is ready, we will walk the arrays over to the BioMicroCenter in Building 68 to put them in the 60°C hybridization oven.
Here’s what will happen tomorrow:
- One of the teaching faculty will wash the unbound cDNAs off your arrays. The wash steps will be 6xSSC/0.005% Triton X-100 at 42° for 10 minutes then 0.2X SSC/0.00016% Triton X-100 at room temperature for 5 minutes.
- The slides will be dried then rehybridized with the Cy3 and Cy5 agents at 65°C for four hours. Genisphere sells these agents as “dendrimers,” essentially large fluorescent balls with an average of 850 fluorophores each. These are described in detail at this link. During the second hybridization, the Cy3 and Cy5 agents will bind the capture sequences on the cDNAs bound to the arrays.
- After 4 hours, the unbound Cy3 and Cy5 agents will be washed off your arrays. The wash steps will be 2X SSC/0.0016% Triton X-100 at 65°C for 10 minutes, 2X SSC/0.0016% Triton X-100 at room temperature for 5 minutes and 0.2X SSC/0.00016% Triton X-100 at room temperature for 10 minutes. The slides will be dried very quickly using compressed air to blow off any water droplets, then scanned in the Agilent scanner that is available in the BioMicroCenter. The data from your array will be available for you to analyze next time.
For next time
- Prepare a figure and caption showing and describing your cell cultures. In 2-3 sentences, interpret the data as you might in a Results and/or Discussion section: be sure to explicitly state what qualities of the cells you are evaluating, and whether the plasmid and/or siRNA appear to have affected them.
- Continue working on developing your Module 2 report as you see fit. You might continue writing up some results and discussion, as well as researching your mouse gene of interest.
- cDNA synthesis cocktail from Genisphere
- Superscript First Strand Buffer
- Superscript II enzyme
- 2X Hybridization Buffer from Agilent
- 20X SSC from Ambion
- 3M NaCl
- 0.3 M NaCitrate