Yale Virtual Center for Cellular Expression Profiling of Rice

PROGRESS (since 9/03 start)

Rice growth at Yale

We established uniform and reproducible growth conditions for rice seedlings and flowering mature plants of both japonica and indica subspecies. Growth room conditions of temperature/humidity/light level were optimized to obtain uniform seedlings for the initial leaf and root cell harvests. Greenhouse illumination, soil, fertilization, and irrigation were optimized to permit full life cycles including flowering, so we can harvest floral cell types. This optimization of cultivation was done with the assistance of Eric Larson, manager of Marsh Botanical Garden at Yale.

Profiling of rice cell types

The following table summarizes the samples for which we have obtained 4 independent biological replicate RNA samples, and produced 3 acceptable microarray hybridization datasets:

Note that these have been harvested in reproducible fashion for the same organ position, stage, age, and internal location (not annotated here), to permit the addition of other stages and locations later to the database. Each biological replicate cell harvest was continued to recover sufficient total RNA, followed by the amplification of sufficient aRNA for complete hybridizations to a 2-slide whole rice genome 70-mer oligo array. Note that each type required some optimization of histological preparation (fix/dehydrate/embed) and harvest (section thickness/beam size/beam power) conditions. We optimized labeling and purification protocols for oligo arrays by improving the reverse transcription efficiency and by adding higher-stringency purification steps

Improvements in sample isolation

Rice tissues vary in efficiency of capture and subsequent RNA recovery, requiring the capture of higher cell numbers from certain sources. The following variables were optimized in Year 1:

Fig. 1

1. Uniformity of source material:

   1. seedlings grown in growth chamber on defined medium

   2. adults grown on warming pads for uniform plastochron length and flowering time

2. Histology & RNA isolation:

   1. Acetone preferable to ethanolic acetic acid fixation

   2. RNase inhibitors can help some sources

   3. Minimize duration of warming to fix sections to slides

   4. GUS markers can assist identification of target cells

3. Cell captures:

   1. Dehumidification of environment essential

   2. High-profile caps avoid cell cross-contamination when caps repositioned over sample

   3. Non-target cells adhere to captured target cells from some sources; Post-It adhesive removes them without RNase problems

   4. Capture most efficient with high beam power; eliminates "ghost" captures

Development of protocols for cRNA labeling and hybridization for oligo microarray

RNA isolated from LCM-isolated cell populations must be amplified to obtain a sufficient amount of RNA molecules for further studies. Due to the nature of RNA amplification (Van Gelder et al., 1990), only cRNA, which is complementary to mRNA, can be generated. Since oligos are complementary to cRNA, reverse-transcribed DNA from cRNA can not hybridize to corresponding oligos in the oligo microrarray. A method to fluorescently label cRNA directly for hybridization to oligo arrays was needed to permit our cell expression profiling. We developed protocols for cRNA labeling by incorporating aminoallyl-UTP into amplified cRNA for further indirect fluorescent labeling. A protocol for cRNA hybridization to oligo array was also optimized by considering the chemical properties of the RNA labeling reaction, appropriate hybridization stringency for RNA-DNA oligo interaction and inhibition of RNA degradation.

Data evaluation, database development, website development

Fig. 2 - Examples of rice cell harvests by LCM

From left: before capture, after capture and capture

We are developing a database for this website for the storage, retrieval, display, and analysis of gene expression data. The added features for this new database include a hierarchical structure for organizing different organs, cell types, and developmental stages of rice to allow easy access and display of cell type specific expression profiles. We will link this database with other databases on rice, e.g. TIGR and BGI-RIS. In addition, statistical methods for pathway analysis that have been developed (e.g. Sun et al. 2003, Sun and Zhao 2004) and are being developed will be incorporated in this database. As soon as the data are populated into this database, we will start the statistical analysis to identify genes that are specific to cell types at specific developmental stages, and correlate these gene expression profiles with physiological characteristics. A graphical interface is being developed that provides a view of the organ anatomy and of selected/harvested cells from which each dataset was produced.

Rice cells harvests by LCM

The LCM method permits visual inspection and documentation of captured cells, prior to accepting them for subsequent RNA isolation.

Figure 2 shows from top:

1. root central metaxylem precursor

2. root cortex

3. leaf bulliform

4. embryo radicle

Reproducibility of data

Fig. 4

The table in Figure 3 shows hybridization correlation coefficients for three biological replicates of isolated leaf P2 cells. As indicated, correlation coefficients among these replicates are typically high with our microgenomic workflow.

A scatterplot is typically used to compare signals from two biological replicate hybrids at the level of individual genes. Specifically, shown here in Figure 4 is a scatterplot of repeats two and three from the leaf P1 cell type isolations. The points on the straight line (x = y) would represent genes that have identical signal strength in both replicates.

Overview of nine datasets

Fig. 5

Fig. 9

Figure 5 shows the number of expressed genes in the datasets for each of nine cell types. Additionally:

1. The Total column represents the number of genes expressed in at least one cell type.

2. The Common column represents the number of genes expressed in all nine cell types.

Fig. 6

Figure 6 examines the distribution of expression intensity among the same nine cell types. Each box in the chart shows the range between the lower quartile (25th percentile) and the upper quartile (75th percentile) with the center line showing the median distribution. Hence, the box displays the interquartile range (IQR), or the middle 50% of the body of data. The outer fences are drawn at 1.5 X IQR above the third quartile or below the first quartile. Outlying data points are stacked horizontally.

Fig. 7

Fig. 8

In Figure 7, we see a Venn diagram of the number of genes expressed among three leaf cell types: P1 (purple), p2 (blue), and p3(red).

Figure 8 shows the same among three root cell types: root central late metaxylem precursor (purple), root cortex (blue) and late root cap(red).

Next we examine the complete linkage hierarchical cluster of the expression profiles of our nine cell types. For the clustering, all expression intensities were normalized to the intensity median of the same cell type. The results show that the individual expression profiles are less distinct than the inidivual phenotypes for differing cell types. As you can see, the profiles of leaf types P1, P2, and P3 are quite similar to each other as well as being quite similar to that of SAM. Additionally, it is clear that the profiles of root cortex cells and central late root cells are more similar to each other than they are to the profile of late root cap. Finally, we observe that the profile for cultured cells is distantly similar to those of root cortex and central late root.

Project Outreach Activites

An important goal of our project has been to increase awareness of rice biotechnology in the scientific community, our local community and the global community. Highlights of our activities in these areas include:

1. We co-sponsored an international symposium on Rice Biotechnology and Policy. This conference featured two speakers from our project.

2. We recruited and trained two minority undergraduate interns.

3. We worked with Yale's Peabody Museum of Natural History to present three rice-themed "family day" activities. Activities included presentations on rice cultivation and rice biotechnology as well as hands-on DNA demonstrations for first and second grade students.