Genome-wide transcription patterns were analyzed in aerobic and anaerobic, steady-state chemostat cultures of the prototrophic laboratory strain S. cerevisiae
) using Affymetrix Ye6100 gene chips, which represent a DNA array encompassing virtually the entire S. cerevisiae
genome. After scanning the arrays, data analysis was performed with Affymetrix GeneChip software. Transcript levels in aerobic and anaerobic cultures (which were hybridized to different gene chips) were compared after normalization. This involved division of individual fluorescence intensities through the fluorescence of the entire chip. The complete data set is available online (15
Reliability of the DNA array analysis was evaluated by comparing transcript levels of three reference genes in the aerobic and anaerobic cultures with classical Northern data from the same RNA samples (Table ). In addition, commonly used mRNA loading standards such as ACT1
), and HHO1
) exhibited the same transcript levels (<10% difference) in aerobic and anaerobic cultures. The measured aerobic/anaerobic values were 3,669/3,839 for ACT1
, 2,561/2,687 for PDA1
, and 2,083/2,071 for HHO1
. Mating type a
-specific genes (MFA1
, and STE2
) were expressed in both cultures, whereas only low transcript levels of α-specific genes (MFα1
, and STE3
) were detected. Few data are available from conventional Northern studies on transcription in aerobic and anaerobic chemostat cultures. However, published data from Northern studies for MAE1
(three- to fourfold-higher level in the anaerobic cultures) and ACS1
(present only under aerobic conditions) agreed well with our data (4
). For ACS2
(similar levels in aerobic and anaerobic cultures), a slight increase was previously reported (23
TABLE 1 Comparison of chip results to Northerndata
In the glucose-limited chemostat cultures, 5,738 (93%) of 6,171 open reading frames (ORFs) from the S. cerevisiae
genome were transcribed at a detectable level under either aerobic or anaerobic conditions (Fig. ). This fraction is higher than reported for previous genome-wide transcription studies on batch cultures of S. cerevisiae
) and may reflect the alleviation of glucose catabolite repression that occurs as a result of the low residual glucose concentration in glucose-limited chemostat cultures (9
FIG. 1 Transcript levels of 6,171 yeast ORFs represented on the Affymetrix Ye6100 gene chips in aerobic and anaerobic chemostat cultures (dilution rate = 0.10 h−1; pH 5.0; temperature = 30°C) of S. cerevisiae CEN.PK113-7D. Transcripts that were (more ...)
The majority of the yeast genes showed similar transcript levels under aerobic and anaerobic conditions (Fig. ). Only 219 genes showed a >3-fold-higher transcription level under aerobic conditions. Under anaerobic conditions, transcript levels of 140 genes were >3-fold higher than aerobically. Only a very small number of genes exhibited a >10-fold difference between aerobic and anaerobic mRNA levels (examples given in Tables and ).
Surprisingly, the majority of genes involved in respiratory sugar metabolism (e.g., those encoding enzymes of the tricarboxylic acid cycle or proteins involved in respiration) showed little or no repression under anaerobic conditions. This result appears to contradict earlier work by DeRisi et al. (10
), who found that transcription of most genes involved in respiration was strongly induced upon a switch from fermentative growth to respiratory growth. However, this contradiction is only apparent. In the experiments of DeRisi et al., the shift from fermentative metabolism to respiratory metabolism was accomplished by growing S. cerevisiae
on glucose in batch cultures. This results in a typical diauxic pattern because initially, the high sugar concentration in the medium causes glucose catabolite repression of respiratory enzymes (12
). Only when glucose is exhausted and cells start consuming ethanol this repression is relieved. In our experiments, aerobic and anaerobic growth were studied in glucose-limited chemostat cultures in which the low residual glucose concentrations alleviated glucose repression. Apparently, under these conditions, the flux through the tricarboxylic acid cycle and respiration is primarily regulated posttranscriptionally (e.g., by concentrations of intracellular metabolites and effectors).
The physiological functions of several of the 53 genes which exhibited a strongly (>10-fold) elevated transcript level under aerobic conditions could be directly linked to typical aerobic processes. This group includes genes involved in respiration [e.g., NDE2, encoding an isoenzyme of the mitochondrial external NADH dehydrogenase; YMR118c, encoding a succinate dehydrogenase; and CYB2, encoding a cytochrome b-(l-lactate cytochrome c oxidoreductase)], protection against oxygen toxicity (CTA1, encoding the peroxisomal isoenzyme of catalase), and β oxidation (PXA1, encoding a transporter involved in translocation of long-chain fatty acids across the peroxisomal membrane; and FOX2, encoding 3-hydroxyacyl coenzyme A epimerase). For some other genes that were specifically expressed under aerobic conditions, the role in aerobic metabolism was less obvious, either because they encode proteins with unknown function (Table ) or because the known functions of their protein products could not be clearly correlated with aerobic growth. This holds, for example, for the sporulation-specific gene SPS100 (Table ), which exhibited a 36-fold-higher transcript level in aerobic cultures, even though sporulation did not occur in these cultures. Also, the high expression of three genes presumed to encode formate dehydrogenases (Table ) in aerobic cultures is unclear.
A comparison of the aerobic and anaerobic transcript profiles of wild-type S. cerevisiae
does not by itself allow conclusions about the molecular mechanisms of transcriptional regulation. However, the methodology used for this study is, in principle, well suited for disentangling the regulatory network via comparison of transcript profiles in wild-type strains and strains with defined modifications in regulatory genes. Some indication as to the involvement of known regulatory mechanisms can be obtained from the presence of consensus sequences in the promoters of aerobically and anaerobically induced genes. For example, 7 of the 11 previously identified Rox1-binding-site-containing hypoxic genes (SUT1
, and COX5b
) showed elevated transcript levels under anaerobic conditions. It is not clear whether the marginal (1.3-fold) increases of the CPH1/CPR1
and the unchanged ERG11
mRNA levels are caused by the stringent anaerobic conditions in the fermentor cultures used in this study.
The functions of some of the genes exhibiting a strongly elevated transcript level under anaerobic conditions (Table ) could be directly linked to anaerobic metabolism. For example, the anaerobic induction of SUT1, encoding a protein involved in sterol uptake (Table ), can be directly linked to the strict requirement for uptake of exogenous sterols in anaerobic cultures. Similarly, the requirement of mitochondrial ATP under anaerobic conditions is reflected by the strong (28-fold) induction of AAC3, encoding a mitochondrial ATP/ADP translocator. The high transcript level of FET4, which encodes a low-affinity ferrous iron transporter, is probably related to the fact that in anaerobic cultures, iron is predominantly present as Fe(II). Indeed, aerobic cultivation resulted in the strong (13-fold) induction of FET3, encoding a cell surface ferroxidase required for high-affinity ferrous iron uptake. As for the aerobic genes of S. cerevisiae, the physiological role of many of the anaerobically induced genes remains unclear. This does not only hold for the substantial fraction of these genes that encode proteins with completely unknown function. For example, the roles in anaerobic metabolism of several genes implicated in stress response (DAN1, TIR1, TIR2, and YSR3/LBP2) or amino acid transport (AGP1 and DIP5) remain to be elucidated.
In quantitative terms, the aerobic and anaerobic transcript profiles of S. cerevisiae
exhibit little difference. This observation can be interpreted in two ways. One possibility is that only few genes contribute to this eukaryote's unique ability to grow rapidly under both aerobic and anaerobic conditions. Alternatively, genes with similar aerobic and anaerobic transcription levels may contribute to this metabolic flexibility. Discrimination between these possibilities requires a combination of the results from this study with investigations, under well-defined aerobic and anaerobic conditions, of the fitness of defined null mutants in all yeast genes. In principle, competition experiments with large sets of defined yeast mutants (3
) in aerobic and anaerobic chemostat cultures should present an excellent tool for such studies (19