Validation analyses of the TSS-tag library. (A) Mapped positions of the TSS-tags relative to the RefSeq genes were evaluated. Population of the TSS-tags mapped at the corresponding positions indicated by the color bars in the margin is shown. The right circle graph shows the composition of the blue section in the left circle graph. (B) Distribution of the luciferase activities of the upstream 1 kb regions of the TSSs (n = 351; right). Luciferase activities of upstream regions of the TSS-tags that were not supported by any RefSeq gene models are calculated separately (n = 20; left). Luciferase activities were normalized against the average luciferase activity of randomly isolated 1 kb genomic fragments (n = 251). For further details, see the reference (27). (C) Correlation between the TSS-tag counts and the copy number estimated by real-time RT–PCR normalized by individual plasmids (n = 105). Each value is the average of three experiments. Sequences of the used primers and quantitative data are presented in Supplementary Table 10. R: correlation-coefficient calculated by linear regression. Note that, because the graph is written in log scale and the y intersect is not 0, the liner regression line is curved where the x value is small. (D) Examples of the real-time RT–PCR and independent oligo-cap RACE analyses. Experimental conditions are shown in the margin. For details, see Materials and methods section. For the primer, ‘Tag’ indicates the PCR primer targeted to the overlapping region of the TSS-tag and ‘RACE’ indicates the PCR primer targeted to the cap-replacing oligo. APID: alternative promoter ID. M: molecular marker.

Validation analyses of the fold induction. (A) Correlation between the fold changes observed using digital-expression information (vertical axis) and real-time RT–PCR (horizontal axis). In total, 57 genes out of 63 total glycolysis related genes, from which we obtained meaningful data, were used for the validation. Each value is the average of three experiments. Sequences of the used primers are shown in Supplementary Table 10. R: correlation-coefficient calculated by linear regression. (B) Validation experiments by the microarray analysis. Overlap of the microarray results and digital gene-expression profiling is shown in the bottom margin. The numbers in the parentheses indicate the number of genes induced by more than 1.5-fold (upper) and 2.5-fold (lower), although they were not detected as ‘induced’ by the criteria of 2.5-fold induction (upper) and 5-fold induction (lower). The statistical significance of the overlap by calculating hypergenometric distribution was P < 5E–67 for 2.5-fold change and P < 3E–15 for 5-fold change. (C) Comparison with data from previous studies. For the comparison with previous microarray studies, the overlap between the genes identified as ‘hypoxia induced’ by this study and the previous studies was evaluated. The ‘hypoxia responsive genes’ in the previous studies were as of those described in the corresponding papers.

Hypoxia-invoked response of the glycolysis gene network. Expression of glycolysis-enhancing enzymes (masked with pale pink) was up-regulated while that of glycolysis-suppressing enzymes (masked with pale blue) was down-regulated. Human genes assigned to the glycolysis pathway map of the KEGG database were selected, and TSS-tag numbers of the corresponding genes were evaluated. Red bars and blue bars represent TSS-tag ppm in normoxia and hypoxia respectively. Fold induction for each of the genes is shown in Supplementary Table 11. Genes included in the list of the ‘hypoxia-induced’ 120 genes are highlighted by green boxes. Note that, since we used the stricter criteria for selecting hypoxia-induced genes (>10 p.p.m., >5-fold induction), many of the genes belonging to this pathway are not directly included in the list, though they showed inductions at least to some extent.

Scheme of 5′-end sequencing using the Illumina GA Sequencer. Adaptors containing necessary sequence for the Illumina GA sequencer are represented as grey boxes. For further information, see Supplementary Data. Gppp: cap structure. AAA: polyA.

Hypoxia-induced TSCs for putatively non-protein-coding RNAs. (A) Genomic positions of the regions in which activated TSSs highly concentrated (red circle). Number of RefSeq genes overlapping the corresponding 100 kb region is shown in the left margin. Examples of regions in which large numbers of transcription initiation sites were induced by hypoxia even in intergenic regions [(B): a 100 kb region in Chromosome 8] and inside genic regions [(C): a 100 kb region in Chromosome 17]. The vertical axis represents fold induction of the TSS-tag counts. TSSs of the genic region of NM_019‱613 (WDR45-like protein gene) are shown in the bottom margin. The direction of the transcription of the RefSeq gene is represented by a red arrow. Radius of each circle represents the number of TSS-tags. Colour of each circle indicates the direction of the transcription (red: same direction with the RefSeq gene; blue: opposite direction of the RefSeq gene). Putative alternative promoters on which confirmation analysis by real-time RT–PCR is shown in (D) are indicated in red and blue letters (AP1-3). AP: Alternative Promoter. (D) Real-time RT–PCR analysis of the putative alternative promoters, AP1-3, shown in (C). Fold inductions calculated by TSS-tag counts and real-time RT–PCR are shown in the third and fourth column, respectively. Primer sequences are shown in Supplementary Table 10. Note that we used first strand single-strand cDNA as template, so that the PCR amplification should be strand-sensitive. (*) Also note that fold inductions estimated for AP3 by TSS-tag counts were the sum of the upstream promoters. As the AP3 were located inside of the last exon, it was impossible to design PCR primers which discriminate the transcript products of AP3 from those of other upstream promoters. Results of the independent oligo-cap RACE analysis for each of the APs are also shown in Supplementary Figure 6.

Hypoxia-induced TSCs in p53 family genes. (A) Count of TSS-tags mapped at the corresponding genomic regions. Red and blue solid bars represent the TSS-tag counts from normoxia and hypoxia, respectively. Exon–intron structures of the RefSeq transcripts are shown in the bottom margins. The genome regions depicted in the bar graphs are magnifications of the regions indicated by thin blue lines. Whether the transcripts from the corresponding promoters should encode the transcriptionally active (TA) or negative (DN) forms is shown in the margin. AP: alternative promoter. (B) Validation of the results shown in (A) by real-time RT–PCR analysis. Promoter ID is as of those represented in the bar graphs in (A). For p63, PCR primers were set in exon 3, so that TA-type transcripts are selectively amplified. Primer sequences are shown in Supplementary Table 10.