Microarray Analysis of Rice Salinity Stress Responses
Supplemental Materials

Gene Expression Profiles during the initial Phase of Salt Stress in Rice (Oryza sativa L.)

Shinji Kawasaki, Chris Borchert, Michael Deyholos, Hong Wand, Susan Brazille, Kiyoshi Kawai, David W. Galbraith, Hans J. Bohnert
Department of Biochemistry and Molecular Biophysics and Department of Plant Sciences, University of Arizona, Tucson, AZ 85721, U.S.A.

A manuscript has been submitted.

We outline a strategy for the analysis of salinity stress responses in rice (Oryza sativa L., var Pokkali and IR29) by microarray analysis. The indica variety Pokkali is a land race with moderate salinity stress tolerance and even young plants (1-2 weeks of age) will survive the exposure to ~150 mM NaCl. In contrast, indica rice IR29 is extremely salt-sensitive and will irreversible wilt within 24 hours when exposed to this sodium concentration. Both to obtain a strong stress response and to maximize the responses that may distinguish the lines, we chose "salt-shock" (from 0 to 150 mM NaCl). A description of the response and response differences between the lines could, we argued, provide more information under extreme conditions than could be obtained during a gradual increase of stress, as it occurs in nature. If differences could be pinpointed, differently regulated genes could provide a set of "candidate genes" that could be analyzed in more detail and under natural salt stress, drought or high temperature conditions. After the addition of NaCl, photosynthesis and stomatal conductance in Pokkali decreased by approximately 80% within 15 minutes, reached a low point at 30 minutes after stress. Within one week, both parameters recovered to approximately 20-25% of the original values. After one week, Pokkali plants had approximately doubled their biomass. IR29 responds the same way in principle; however, the decrease in photosynthesis and stomatal conductance was slower and there was no low point but the decrease continued to zero which was reached before 24 hours of stress when the plants had died.

We include materials describing responses in gene expression to salt shock treatment by young rice plants (var. Pokkali and IR29) (slide 1). Pokkali plants were stressed by the addition of 150 mM NaCl and leaf and root and leaf cDNA libraries were generated (slide 2). ESTs were sequenced. Up to 1728 ESTs were deposited in triplicate or quadruplicate on glass slides (Deyholos and Galbraith, 2001) (slide 3). Hybridizations were performed with Cy3- and Cy5-labelled targets and regulated transcripts were identified (slide 4).

Hybridizations were performed overnight at 42°C in humidified chambers. The slides were washed in 1xSSC and 0.2% SDS (10 min) and 0.1xSSC (10 min). The slides were rinsed for 1 min in 0.01xSSC and dried by centrifugation. The fluorescent signatures were captured using a ScanArray 3000 (GSI), and analyzed using ImaGene III software (BioDiscovery). Local background was subtracted from the value of each spot on the array. Spots covered by dust particles, missing spots, spots with low signal intensity, and spots within high background areas were flagged as candidates for exclusion after further analysis. Normalization between the Cy3 and Cy5 fluorescent dye emission channels was initially achieved by adjusting the level of both image intensities to the signal intensity of exogenously added non-plant control genes and internal control genes. Transcript regulation is expressed as the ratio of intensities between stress and control (log ratio, termed LR). Five human cDNA clones accession numbers. AA418251 (zinc finger protein PLAG1), AA464627 (intestinal membrane protein A4), H28469 (IGa-2 chain C), AA 456109 (scaffold protein Pbp1) and AA485668 (integrin b-4 subunit) were transcribed in vitro. The five cRNAs were diluted each to a different (order as listed above) degree to provide 2, 1, 0.2, 0.1 and 0.02 ng of cRNA, respectively. This mixture of cRNAs was added to the plant target RNA before the incorporation of the Cy3 and Cy5 dyes. Microarray slides were printed to generate four major segments into which individual elements are printed multiple times (triplicates in most experiments). In some experiments it was necessary to vary the normalization factor in different segments due to variability in the background or variable target intensity. To reduce area-specific effects, normalization between the Cy3 and Cy5 fluorescent dye channels was achieved by calculating the ratio between the total Cy3 signal from all spots in relation to the total Cy5 signal from all spots in each segment.

Signals from triplicate spots were averaged (slide 5). The Cy3/Cy5 signal intensities were adjusted with the help of exogenously added control genes, which had been placed in different sections of the microarray slides to compensate for variable background levels. Repeat experiments indicate variability (slide 6).

We describe the changes in rice Pokkali and IR29 at various time points during salt stress: 15 minutes, 1 hr, 3 hr, 6 hr, 24 hr and 7 days. The annotation and regulation of 1728 ESTs is shown (Tables). Changes in signal intensity were converted into log-10 expression ratios, LR (Log-10 Ratio [Cy3/Cy5]), for the Cy5 and Cy3 signals of all clones on the arrays. From repeat experiments, we considered log10 +/-0.2 (1.6-fold upregulation or downregulation) as the threshold that to signified a significant change. In some cases, transcripts considered regulated in one experiment deviated from the LR+/-0.2 threshold but were still included because they irrespectively were among the most highly up- or down-regulated transcripts on the slide. A few transcripts had to be flagged, and were removed from the list of the most highly up- or down-regulated transcripts in the comparison of repeat experiments. In such experiments, comparisons of hybridizations with the same RNA target to different slides showed 99% of the ESTs to deviate by less than LR0.2 from the mean. In repeat experiments with RNA targets from different sets of plants, higher variation was observed as from 94 to 97% of the ESTs deviated by less than LR+/-0.2. Variability seemed to be random, depending on the quality of the microarray slides, on local background on the slides, and possibly also on incomplete solvation of the target which leads to concentration differences in different regions of the slides.

The tables (Tables) include clone ID and annotations for 1728 ESTs used in the microarray analysis. All ESTs derived from root tissue. Included are the LR values that distinguish the unstressed and stressed states for 15 min, 1, 3, 6 and 24 hr and 7 days for Pokkali and 1, 3, and 6 hr for IR29 from one experiment.

Microarrays may be useful to obtain not only expression ratios but absolute intensity (transcript abundance) as well (slide 7). As an example we chose the ESTs for four aquaporins (aquaporins; WCP-I to IV), one encoding a low abundance (according to RNA gel blots) PIP homolog and three more highly expressed TIP homologs found in our cDNA libraries and included in the microarrays. All four AQP ESTs declined during the initial phase of NaCl stress in rice Pokkali (circle [15 min], square [1 hr], triangle [3 hr], diamond [6 hr]). After 24 hr, most of the transcripts had increased to approximately pre-stress levels (star). After 7 days, all AQP transcripts showed a higher than pre-stress transcript amount (cross). It seems significant that all four AQP transcripts occupy a narrow range of signal intensity (x-axis) and that the lower-abundance PIP showed the lowest intensity level. The PIP aquaporin and one of the three TIPs are circled.