Microarray analyses of gene expression during the Tetrahymena thermophila life cycle

Abstract

Background: The model eukaryote, Tetrahymena thermophila, is the first ciliated protozoan whose genome has been sequenced, enabling genome-wide analysis of gene expression.

Methodology/principal findings: A genome-wide microarray platform containing the predicted coding sequences (putative genes) for T. thermophila is described, validated and used to study gene expression during the three major stages of the organism's life cycle: growth, starvation and conjugation.

Conclusions/significance: Of the approximately 27,000 predicted open reading frames, transcripts homologous to only approximately 5900 are not detectable in any of these life cycle stages, indicating that this single-celled organism does indeed contain a large number of functional genes. Transcripts from over 5000 predicted genes are expressed at levels >5x corrected background and 95 genes are expressed at >250x corrected background in all stages. Transcripts homologous to 91 predicted genes are specifically expressed and 155 more are highly up-regulated in growing cells, while 90 are specifically expressed and 616 are up-regulated during starvation. Strikingly, transcripts homologous to 1068 predicted genes are specifically expressed and 1753 are significantly up-regulated during conjugation. The patterns of gene expression during conjugation correlate well with the developmental stages of meiosis, nuclear differentiation and DNA elimination. The relationship between gene expression and chromosome fragmentation is analyzed. Genes encoding proteins known to interact or to function in complexes show similar expression patterns, indicating that co-ordinate expression with putative genes of known function can identify genes with related functions. New candidate genes associated with the RNAi-like process of DNA elimination and with meiosis are identified and the late stages of conjugation are shown to be characterized by specific expression of an unexpectedly large and diverse number of genes not involved in nuclear functions.

Figure 1. Background determination and normalization controls.

(A) Background signal intensities during growth (L-l to L-h), starvation (S-0 to S-24) and conjugation (C-0 to C-18) stages were determined using 4308 different random probes on each array. The signal intensities shown for these probes are the average of 12924 values for triplicate growth and starvation samples and 8616 values for duplicate conjugation samples. The bars represent the standard errors. (B, C) Relative levels of HHT3 and SerH3 mRNAs at all 20 stages. The results shown here are the average of triplicate growth and starvation samples and duplicate conjugation samples, and the bars represent the standard errors.

Figure 2. Microarray-wide distribution of signal intensities as a function of the level of subtraction of the average signal intensity shown by 4308 unrelated (negative control) array probes.

As described under Materials and Methods, each curve was obtained by subtracting, from the signal intensity of every probe, the particular multiple of the negative control signal intensity (33 AUs, see Fig. 1A) shown in the box, according to the approach described in Wei . Note that at 3× subtraction, the distribution has converged to a robust profile that changes little at >3× subtraction and retains no hint of the large non-specific hybridization observed in the un-subtracted distribution. This determination was done using the 2 hr conjugation sample.

Figure 3. Comparison between microarray and northern blot expression profiles.

Top panel, microarray data; lower panel, northern blots from independent RNA preparations and hybridizations. A: DRH1, encoding a putative helicase (J. Bowen, unpublished), B: TCD1 encoding a putative chromodomain protein (W. Wang, unpublished). Values from individual conjugation experiments are shown to indicate the reproducibility of the expression of specific genes in replicate experiments. G, growing cells; S, starved cells; C, conjugating cells at 0 to 24 hr after mixing of two different mating types.

Figure 4. Number of genes expressed in at least one time point during each of the three major stages of the T. thermophila life cycle.

Expression is defined as signal intensity above 1× adjusted background (99 AUs), as described in the text. The number of genes expressed at each of the stages is shown beneath the identifier for each circle. Number of genes in composite categories discussed in the text (e.g., genes expressed during growth but not starvation) are indicated along the margins. A total of 21,178 genes are accounted for in the diagram. An additional 5,876 genes failed to show signal intensity >99 AUs at every time point and stage tested (discussed in the text). One gene had a value of exactly 99 AUs and consequently was excluded by the search criteria. The search conditions were those listed in Table S4.

Figure 5. Comparison of numbers of predicted genes specifically expressed (A) or upregulated (B) during growth, starvation and conjugation.

^{a, b}: the search conditions were those listed in Table S5.

Figure 6. Heat map of the expression of 91 growth-specific genes.

Genes expressed at levels >2× corrected background (see Figure 5A) are included. Clustering done using ArrayStar 2 (Clustering type: K-mean, Distance metric: standard Pearson). The heat map uses colors to display the relative values of all tiles within a given experimental condition wih blue indicating low expression, yellow indicating intermediate expression and red indicating high expression. The numerical values give the actual values on a log 2 scale that are associated with each color. Stages are as described in Materials and Methods. The color scale is shown by the bar at the top right corner of the figure.

Figure 7. Heat map of 90 starvation-specific genes.

Genes expressed at levels >2× corrected background (see Figure 5A) are included. Clustering parameters, conditions and other symbols are as in Figure 6.

Figure 8. Heat map of 706 genes expressed during both starvation and conjugation but not during growth.

Genes whose maximum expression levels during starvation and conjugation were >2× corrected background, and were less than 1× corrected background during growth, were included, Clustering type: Hierarchical, with Euclidean distance metric. Software, conditions and other symbols are as in Figure 6.

Figure 9. Heat map of 503 conjugation-specific genes.

Genes expressed at levels >5× corrected background (see Figure 5A) were included. Clustering parameters, conditions and other symbols are as in Figure 6. The red arrow indicates 8 genes that are expressed almost exclusively at C-0.

Figure 10. Stages of conjugation in Tetrahymena.

Figure 11. Heat map of 217 codon biased genes described in reference .

Clustering parameters, conditions and other symbols are as in Figure 6.

Figure 12. Expression profiles of five putative selenocysteine genes.

Stages are as described in Materials and Methods.

Figure 13. Expression of all seven genes in scaffold CH670435 during growth, starvation and conjugation.

The genes are given in order of their locations on the scaffold with TTHERM_01345750 located at one end and TTHERM_01345820 at the other end. Stages are as described in Materials and Methods.

Figure 14. Summary of the number of genes differentially expressed at two hour intervals during conjugation.

Values above and below the “0 line” represent the number of up-regulated and down-regulated genes, respectively.

Figure 15. The expression profiles of TWI1 (TTHERM_01161040) co-expressed genes.

Stages are as described in Materials and Methods.

Figure 16. Fifty-one Conjugation-induced/specific transcription factors of Tetrahymena with the expressed peak at 2 hr (A), 4 hr (B), 6 hr (C), and 8–14 hr (D) postmixing.

Stages are as described in Materials and Methods.