Microarray, IPA and GSEA Analysis in Mice Models

This protocol details a method to analyze two tissue samples at the transcriptomic level using microarray analysis, ingenuity pathway analysis (IPA) and gene set enrichment analysis (GSEA). Methods such as these provide insight into the mechanisms underlying biological differences across two samples and thus can be applied to interrogate a variety of processes across different tissue samples, conditions, and the like. The full method detailed below can be applied to determine the effects of muscle-specific Notch1 activation in the mdx mouse model and to analyze previously published microarray data of human liposarcoma cell lines.

significantly altered across different samples. Gene set enrichment analysis (GSEA) uses gene sets and characteristics that have been a priori associated with various diseases or pathways in order to provide biological application to the sample of interest.
The methods described below were used by Bi and colleagues to understand the effects of Notch signaling in muscle regeneration and liposarcoma, a common soft-tissue cancer type (Bi et al., 2016a). These methods probed the effects of myofiber-specific Notch activation in a Duchenne's muscular dystrophy (mdx) mouse model and discovered that over-activation of Notch in the mdx mouse model displayed similar gene-expression patterns as healthy human muscle. Similarly, Bi and colleagues performed microarray analysis, IPA and GSEA to find that over-activation of Notch in mouse inguinal white adipose tissue shares signatures of human liposarcoma (Bi et al., 2016b). Both of these studies underscore the importance of comparative analyses when using animal models and since many microarray datasets are available online, gene set enrichment analysis (GSEA) can be used to evaluate already published datasets with respect to the investigator's interest at relatively low cost. Discoveries such as these are imperative towards developing therapeutic targets and furthering our understanding of biological processes and how their perturbance may influence human disease.

6.
Add 0.2 ml of chloroform per 1 ml of TRIzol™ reagent used, and shake or vortex vigorously for 15 sec. Then incubate the sample at room temperature for 3 min.

7.
Centrifuge for 15 min at 4 °C and 12,000 × g to separate phases.

8.
Transfer the aqueous phase (clear, upper phase Figure 2) containing the RNA to a new tube using a 1 ml pipette (volume ~400-600 μl), making sure not to transfer any of the interphase or phenol-chloroform phase.

9.
Add 0.5 ml of isopropanol per 1 ml of TRIzol™ reagent used to the aqueous phase containing RNA from Step A8 and incubate at room temperature for 10 min.

10.
Centrifuge sample for 10 min at 4 °C and 12,000 × g to precipitate RNA (RNA should form a white pellet at the bottom of the tube).

11.
Discard the supernatant being careful not to disturb the pellet.

12.
Resuspend pellet in 1 ml of 75% ethanol per 1 ml of TRIzol™ reagent used to wash the RNA pellet.

13.
Gently shake sample briefly and centrifuge for 5 min at 4 °C and 7,500 × g.

14.
Carefully discard supernatant, removing as much ethanol as possible without disturbing the pellet.

15.
Air-dry pellet for maximum 5-10 min at room temperature. Make sure that the ethanol has evaporated but do not let the pellet dry for too long as residual ethanol and over-drying both may affect RNA quality.

16.
Resuspend pellet in 50 μl of RNase-free water (imperative if being used for downstream microarray analysis).

17.
Measure concentration and the ratio of A 260 /A 280 using NanoDrop Spectrophotometer. Notes:

a.
One milligram of the sample should yield roughly 1 μg of RNA.

b.
The ratio of A 260 /A 280 in the range of 1.85-1.95 is required for downstream applications, and the RNA sample quality should be checked using the Agilent bioanalyzer.

c.
Determination of the quality of RNA sample on an Agilent bioanalyzer should yield two large peaks corresponding to the 18S and 28S ribosomal subunits (Peterson et al., 2009). Examples of good and poor quality RNA are shown in Figure 3.

B. Transcriptomic Microarray Analysis
Microarray methods and analysis are adapted from the manufacturer's manuals (Agilent, Two-Color Microarray-Based Gene Expression Analysis-Low Input Quick Amp Labeling)

1.
Preparation of Spike A Mix and Spike B Mix as positive controls, all procedures should be done in an RNase-free environment to ensure stability of RNA, following steps can be done in a 1.5 ml microcentrifuge tube.

a.
Vortex and heat Spike A and B Mix at 37 °C for 5 min upon arrival and briefly centrifuge.

b.
To dilute Spike A Mix (for example) prepare first dilution by adding 2 μl of Spike A Mix to 38 μl dilution buffer, mix, and briefly spin down (dilution is 1:20).

c.
Dilute solution from Step B1 b by adding 2 μl of diluted solution to 78 μl dilution buffer, mix, and briefly spin down (dilution is 1:40).

Dilute solution from
Step B1 c by adding 2 μl of diluted solution to 30 μl dilution buffer, mix, and briefly spin down (dilution is 1:16).

e. Dilute solution from
Step B1 d by adding 4 μl of diluted solution to 28 μl dilution buffer, mix, and briefly spin down (dilution is 1:8).

g.
Repeat for Spike B Mix for other RNA samples (i.e., RNA from wild-type sample vs. RNA from experimental sample) and proceed with labeling reaction.

2.
Labeling reaction, purification and quantification of fluorescently labeled complementary RNA (cRNA) Note: For Spike A prepare with Cyanine 3-CTP and Spike B Cyanine 5-CTP dye otherwise both samples are treated the same. Have water baths or heating blocks set to 65 °C and 80 °C prior to starting procedure for Steps B2a-B2d (below).

a.
Prepare T7 primer mix by combining 1.8 μl T7 primer to 1 μl nuclease-free water per reaction.
b. Add 1.8 μl T7 primer mix to each tube and incubate at 65 °C for 10 min, gently shaking the tube every couple of minutes.

c.
Remove the reactions from heat and place on ice for 5 min.

d.
In the meantime, pre-warm the 5x first strand buffer at 80 °C for 3-4 min, vortex and spin down so that the buffer components are fully re-suspended.

e.
Assemble the following cDNA reaction on ice in a 1.5 ml microcentrifuge tube (scaling up according to the number of reactions with one reaction in excess to correct for pipetting error):

k.
Using the RNeasy Mini Kit, purify the cRNA from each reaction (protocol below as summarized per the manufacturer's instructions) i.
Bring volume of cRNA reaction to 100 μl with nuclease-free water.

ii.
Add 350 μl of RLT and mix well.

iii.
Add 250 μl of 100% ice cold ethanol and mix by gently pipetting.

iv.
Transfer reaction to spin column, place in a collection tube and centrifuge at 16,000 × g and 4 °C for 30 sec.

v.
Discard flow through and add 500 μl of RPE buffer to the column, centrifuge at 16,000 × g and 4 °C for 30 sec.

vi.
Repeat Steps B2j-B2k v but centrifuge at 16,000 × g and 4 °C for 60 sec.
vii. Place column in a fresh collection tube and centrifuge at 16,000 × g and 4 °C for 30 sec to remove and remaining buffer.
viii. Place column in a 1.5 ml microcentrifuge tube and add 30 μl of RNase-free water to the column.

ix.
Incubate at room temperature for 1-2 min and then centrifuge at 16,000 × g and 4 °C for 30 sec. x.
Optional: re-elute with eluate to increase yield.

xi.
Measure RNA concentration and quality with a NanoDrop™ ND-1000 UV-VIS Spectrophotometer version 3.2.1 or higher.

1)
Use 'Microarray Measurement' tab and select RNA-40 as sample type.

3.
Hybridization of sample and probes

a.
Preparation of 10x blocking: add 1,250 μl of nuclease-free water to the 10x gene expression blocking agent (supplied with the kit) and gently vortex to dissolve powder completely.
Note: If powder does not readily dissolve, heat blocking agent at 37 °C for 4-5 min.

b.
Assemble the following reaction in a 1.5 ml microcentrifuge tube.
Note: The reaction below and subsequent volumes are for 2pack microarray, for 1-pack, 4-pack or 8-pack refer to refer to Agilent's Two-color Microarray-Based Gene Expression Analysis Protocol.

g.
Slowly add 240 μl of sample onto the gasket well from left to right, being careful not to introduce any air bubbles.

h.
Make 1x solution from 2x Hi-RPM Hybridization Buffer in any wells that remain unused.

i.
Place the slide with the 'active' side down, such that the Agilent-labeled barcode is facing down and the numeric barcode facing upwards.

j.
Place the SureHyb chamber cover onto the slides, clamp both pieces and tighten.

k.
Double-check that there are no stationary bubbles and, if needed, tap on surface to remove.
Note: bubbles may be a source of artifacts as they may impact signal intensity l.
Load the chamber onto the rotator rack in the hybridization oven, set it to rotate at 10 rpm and hybridize at 65 °C for 17 h.
To prepare the Wash Buffers, remove outer and inner caps from container and use a pipette to add 2 ml of Triton X-102 to gene expression Wash Buffers 1 and 2.

b.
Mix by inversion and replace the original outer and inner caps with the spigot provided with the kit.

c.
Pre-warm gene expression Wash Buffer 2 to 37 °C before proceeding with washing the arrays.

d.
Wash the staining dish prior to use as follows (repeat 2 x): i.
Add slide rack and stir bar to staining dish and fill the dish with 100% isopropyl alcohol ( Figure 4A).

ii.
Turn on magnetic stir plate to wash for 5 min. iii.
Rinse staining dish with Milli-Q water multiple times.

e.
Prepare staining dishes as follows

i.
Fill slide-staining dish #1 with gene expression Wash Buffer 1.

ii.
Place slide rack into slide-staining dish #2 and add magnetic stir bar, fill with gene expression Wash Buffer 1 and place on a magnetic plate.

iii.
Place dish #3 on the stir plate, add a stir bar and only fill with pre-warmed gene expression Wash Buffer 2 immediately before use.

f.
Remove hybridization chamber from the rotating incubator and note any bubbles that may have formed during hybridization.

g.
Disassemble the hybridization chamber by placing it on a flat surface, remove the array-gasket while maintaining the numeric barcode facing up and immediately submerge it in slide-staining dish #1.

h.
Keeping the array-gasket sandwich submerged, pry open the sandwich with forceps and let the gasket slide drop to the bottom of the dish.

i.
Remove the slide and place it into the slide rack in slidestaining dish #2, being careful to only touch the slide over the numeric barcode or along the thin edges.

j.
Repeat these steps for the remaining slides.

k.
Incubate the slide on the magnetic stir plate for 1 min.

l.
Add slide rack to slide-staining dish #3 and incubate on the magnetic stir plate for 1 min.

m.
Slowly and carefully remove slide rack and place slides on the slide holder.

i.
Add the slide without the barcode label towards the edge. ii.
Active microarray surface should be facing up towards the slide cover.

iii.
Close the plastic cover.

n.
Proceed to scanning slides.

5.
Scanning microarray slides, feature selection and data collection a.
Place the slide holder containing slide into the scanner cassette.

d.
Open Agilent Feature Extraction (FE) and add the images to be extracted to the FE project (default settings for project ok).
Note: Manual grid mapping may be required.

e.
Save the Feature Extraction project as .fep via File > Save As.

f.
Select Project > Start Extracting.

C.
Validation of microarray results

1.
Gene-specific primer design for real-time quantitative PCR (qPCR): selection of amplicon size, primers and template are imperative to generating reproducible data that can accurately determine if the results from the microarray are validated Note: Pre-validated gene expression assays can be purchased from a variety of vendors and thus do not require optimization. For genes that are not available or have not been previously validated, primers efficiency should be evaluated (see below):

a.
Select genes based on the microarray results. Candidate genes should be chosen based on genes that displayed a significant change across conditions and a reference gene should be chosen that did not display any change in expression across samples (for the latter examples include 18S rRNA or β-actin).

b.
Amplicons targeted by primer should be approximately equal in size (not greater than 0.6 kb) and the secondary structures of the target sites can be determined using nucleic acid-folding software such as OligoAnalyzer 3.1 (IDT), as highly structured sequences can impact qPCR efficiency and results.

c.
Primers target sites should be analyzed by in silico PCR tools such as NCBI BLAST or UCSC Genome Browser to determine specificity.

d.
Each primer should have roughly the same melting temperature, however the exact ideal annealing temperature must be determined experimentally.
One to five microgram of template RNA from tissue samples assayed in microarray analysis required (use both biological and technical replicates here; the former being another mouse sample under the same condition and the latter the same RNA that was used for microarray analysis), and quality should be determined prior to cDNA synthesis.

b.
Thaw reagents from M-MLV RT kit, vortex and centrifuge briefly.

c.
Assemble the following reaction on ice in the respective order:

h.
Heat inactivate reaction at 70 °C for 15 min.

i.
Resulting cDNA can be used immediately for real-time PCR analysis or stored at −20 °C.

3.
Real-time qPCR of candidate target genes Note: Important to set up biological and technical replicates as well as a negative control containing no template. 3 biological replicates (i.e., three RNA/cDNA samples from three different mice), 3 technical replicates (i.e., using the same RNA to generate cDNA) suggested per sample.

a.
Primers should be re-hydrated in nuclease-free water and stored at a stock concentration of 10 pM.

b.
Thaw reagents on ice and prepare the following reaction on ice (multiply the final reaction volume by the number of PCR reactions planned plus to [to account for pipetting error] create a master mix that can be aliquoted into Roche LightCycler 480 plates).

d.
C q values can then be analyzed for primer efficiency and subsequent fold-change in target gene expression (C q values > 35 are not recommended to use).
Note: Samples requiring over 35 cycles may not provide reliable results and indicate that the cDNA quality or reaction efficiency is poor. While some very lowly expressed genes may yield a C q value between 35-40, under those conditions it is imperative the negative control produces no signal at that cycle number

Data analysis
This part of the protocol includes the analysis of the microarray results and subsequent geneset enrichment and ingenuity pathway analyses to determine candidate genes and enriched biological pathways/processes (respectively). Following candidate gene selection, real-time PCR analysis is performed to validate candidates (for an overview of workflow see Figure  5).

Analysis of microarray results
Normalization, gene alignments and calls (to correlate gene expression levels) and evaluation of genes with statistically significant gene expression changes across the evaluated samples.

1.
Download GeneSpring software to perform statistical analysis and open software.

2.
Create a new project, load the text files from the feature extractor (FE) and click 'Next', keeping the software settings as default.

3.
The data will be uploaded onto GeneSpring upon clicking 'Finish'.

4.
Assign 'Experimental Grouping' and then 'Create an Interpretation' with the respective experimental groupings properly selected.

a.
Perform an Analysis of variance (ANOVA) using the software, selecting a Tukey post-hoc test and the appropriate pairing options (depending on samples).

b.
If comparing wild-type and a knock out-sample, perform a Student's t-test (in general, the statistical test performed will depend on the samples).

6.
Fold-change analysis: elimination of probes that do not meet 1.5 foldchange.

a.
Select 'Fold-change' with the same interpretation as used for the ANOVA.

b.
Adjust fold-change to 1.5 and to determine how many probes meet this criterion.

c.
Once 'Finish' has been selected, the probe lists should appear.

7.
Combine the probes whose expression increases or decreases into one excel sheet for ingenuity pathway analysis.

B.
Ingenuity Pathway Analysis (IPA) order to understand the applicability of results to mouse models (NCBI dataset GDS3027).

4.
Microarray results containing expression data from Step C1 should be converted to the GCT file-type: Details on how to convert files to the required type for GSEA can be found on GenePattern: file format guide.

5.
A gene set database file containing a reference dataset to analyze against and a sample phenotype file must also be generated for input into GSEA (as well as an empty directory to store the output results)

a.
Gene set database file (.gmt) formatting details can be found here on GenePattern: file format guide.

b.
Sample phenotype file (.cls) formatting details can be found here on GenePattern: file format guide.

6.
For the analysis used in Bi et al. (2016a and2016b), default settings were used when calling GSEA.
Note: However, these can be changed as seen fit; see source code documentation for more details.

7.
GSEA results and graphical analysis a.
The output directory should contain GSEA summary results file; determine that the parameters meet the specified values prior to proceeding (ideal values can be found at Subramanian et al., 2005).

b.
GSEA R package contains GSEA.Analyze.Sets which generates plots of the input data, refer to the source code documentation for specific parameters.

D.
Analysis of real-time PCR results (Bustin et al., 2009) (used for validation of target genes as discovered by microarray and GSEA)
All primers used for real-time PCR should be assessed for efficiency.

b.
All primers should amplify a target site with similar efficiency with respective to the reference. Efficiency of the primer amplification can be determined by generating a calibration curve.

c.
Microarray results can be validated by real-time quantitative PCR using the same RNA used for microarray analysis. However, a power analysis should be conducted to determine the appropriate sample size for each experiment and a Student's t-test with a two-tail distribution can be to analyze results unless specified otherwise.

2.
Validation of microarray results using Gene-ontology (GO) term analysis Gene-ontology (GO) term analysis can also be performed on the list of genes generated from the microarray; however GSEA provides a rank and weight to each gene such that relative expression level in the sample is taken into consideration thus helping researchers identify candidate genes. GO term analysis does not provide gene-specific information however both GSEA and GO-term analysis will yield biological pathways that are significantly enriched in the assayed samples.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. The top panel clearly shows two peaks which correspond to the 18S and 28S ribosomal subunits, a small peak to a spike-in control and otherwise smooth lines. The middle and bottom panels represent degraded RNA in which the 18S and 28S peaks may or may not be clear and are usually preceded by multiple smaller peaks (indicative of degraded RNA). The x-and y-axis are nucleotides and fluorescence, respectively.  RNA is isolated from tissue samples (or cells) of interest. Quality of RNA is determined prior to proceeding with generation of cRNA, hybridization and data acquisition. Analysis of the microarray data is performed by the Agilent software. From there, the Gene Set Enrichment Analysis software is employed to yield genes that are significantly enriched in the assayed sample. Microarray data will also be used for Ingenuity Pathway Analysis (IPA) (and can be used for Gene Ontology [GO] analysis) to reveal pathways or biological processes that are enriched in the target sample.