The t = 0 was taken (between 3 and 5 mL of culture) and vacuum filtered. Estradiol was then added the continuous culture and sampled at various timepoints.
Growth protocol
All experiments were performed in chemostats (continuous cultures). We chose chemostats, in part, because the steady-state condition of chemostat cultures is a particularly useful feature for mathematical modeling. Under steady-state conditions, the levels of molecules and activities of process are not changing at a culture-wide level. Cells were grown under continuous culturing conditions in 500 mL vessels as previously described with minor adjustments (Saldanha AJ, Brauer MJ, Botstein D. Nutritional homeostasis in batch and steady-state culture of yeast. Mol Biol Cell. 2004;15: 4089–4104). Cultures were aerated with 6 L/min of humidified air at 30C, maintained at 300 mL, and stirred with a magnetic impeller at 400 RPM. For the majority of experiments, cultures were maintained with minimal medium under phosphate limitation (20 mg/L). Nitrogen-limited cultures were maintained at 40 mg/L ammonium sulfate. Methionine-limited cultures (McIsaac et al, 2012) were maintained at 7.5 mg/L methionine. Growth rates were maintained from 0.15-0.17 vol/hour. Batch growth in the chemostat vessels was initiated from a 1:60 dilution of a saturated overnight culture prior to turning on the chemostat pumps. Cells were grown to steady state, as determined by culture density, prior to the addition of 1 micromolar beta-estradiol to the culture and subsequent sampling. Chemostat experiments were performed with either the Infors Sixfors or Multifors systems.
Extracted molecule
total RNA
Extraction protocol
RNA was prepared using a standard acid phenol extraction. Lysis buffer: 600 µL of 0.5M EDTA, 1.5 mL of 10% SDS, 300 µL 1M Tris, 27.6 mL Rnase-free water.
Label
Cy5
Label protocol
Crude RNA was purified using either the QIAGEN RNeasy kit or RNAClean Ampure XP beads. 200 ng of cleaned RNA was used as an input to generate dye-labeled cRNA using the Agilent Quick-Amp Labeling Kit. 1) Prepare labeling reaction: Get clean PCR strip tubes ready; Dilute RNA to 57.2 ng/µL; Add 3.5 µl of RNA (200 ng total) to PCR tube; Make T7 Primer Mix (0.8 µL T7, 1 µL water);Add 1.8 µl of T7 Primer Mix to each tube. Each tube now has 5.3 µl; Denature the template and primer by incubating at 65°C for 10 min; Put reactions on ice for 5 min; 2) cDNA synthesis: Heat 5x First Strand Buffer 80°C for 3-4 minutes, vortex, do a quick spin, and keep at room temperature; Make cDNA master mix (2µL 5x First Strand Buffer, 1 µL 0.1M DTT, 0.5 µL 10 mM dNTP mix, 1.2 µL Affinity Script RNase Block Mix); Briefly spin sample and master mix tubes; Add 4.7 µL of cDNA Master Mix to sample tubes and mix by pipetting up and down. Each tube now has 10 µL; Incubate at 40°C for 2 hours; Incubate at 70°C for 15 minutes (inactivates the AffinityScript enzyme); If you don’t continue to next step, freeze samples -80°C; Ice for 5 min. 3) Labeled cRNA synthesis: Keep T7 RNA polymerase (red cap) on ice. Master Mix = 0.75 µL Water, 3.2 µL 5x Transcription Buffer, 0.6 µL 0.1M DTT, 1 µL NTP Mix, .21µL T7 RNA polymerase Blend; 0.24 µL Cy3-CTP or C5-CTP; Add 6 µL of Transcrption Master Mix with Dye to the cDNA. Each tube now contains a total volume of 16 µL; Incubate samples at 40°C for 2 hours. 4) SPRI cleanup of cRNA: Before beginning: Bring RNAClean XP to RT, Make up enough 80% ethanol for >2 200 µL washes per reaction, Set 96 well thermomixer to 37°C; Add 48 µL (3X) of RNAClean XP beads to each reaction and incubate for 10 minutes; Place plate on magnet and allow supernatant to clear (~5 minutes); Remove supernatant carefully, avoiding beads, and discard; Add 200 µL of 80% ethanol to each well and incubate for 30 seconds; Remove ethanol and repeat for a total of two washes; Remove ethanol, cover plate, and spin down for ~1 minute; Place plate back on magnet and use a p20 to remove residual ethanol; Dry beads at 37°C until beads no longer move around when placed on magnet (~4 minutes); Add 42.5 of RNAse-free water and resuspend beads, then incubate for 2-3 minutes; Place plate on magnet and transfer 40 µL to a new well before placing on ice
The t = 0 was taken (between 3 and 5 mL of culture) and vacuum filtered. Estradiol was then added the continuous culture and sampled at various timepoints.
Growth protocol
All experiments were performed in chemostats (continuous cultures). We chose chemostats, in part, because the steady-state condition of chemostat cultures is a particularly useful feature for mathematical modeling. Under steady-state conditions, the levels of molecules and activities of process are not changing at a culture-wide level. Cells were grown under continuous culturing conditions in 500 mL vessels as previously described with minor adjustments (Saldanha AJ, Brauer MJ, Botstein D. Nutritional homeostasis in batch and steady-state culture of yeast. Mol Biol Cell. 2004;15: 4089–4104). Cultures were aerated with 6 L/min of humidified air at 30C, maintained at 300 mL, and stirred with a magnetic impeller at 400 RPM. For the majority of experiments, cultures were maintained with minimal medium under phosphate limitation (20 mg/L). Nitrogen-limited cultures were maintained at 40 mg/L ammonium sulfate. Methionine-limited cultures (McIsaac et al, 2012) were maintained at 7.5 mg/L methionine. Growth rates were maintained from 0.15-0.17 vol/hour. Batch growth in the chemostat vessels was initiated from a 1:60 dilution of a saturated overnight culture prior to turning on the chemostat pumps. Cells were grown to steady state, as determined by culture density, prior to the addition of 1 micromolar beta-estradiol to the culture and subsequent sampling. Chemostat experiments were performed with either the Infors Sixfors or Multifors systems.
Extracted molecule
total RNA
Extraction protocol
RNA was prepared using a standard acid phenol extraction. Lysis buffer: 600 µL of 0.5M EDTA, 1.5 mL of 10% SDS, 300 µL 1M Tris, 27.6 mL Rnase-free water.
Label
Cy3
Label protocol
Crude RNA was purified using either the QIAGEN RNeasy kit or RNAClean Ampure XP beads. 200 ng of cleaned RNA was used as an input to generate dye-labeled cRNA using the Agilent Quick-Amp Labeling Kit. 1) Prepare labeling reaction: Get clean PCR strip tubes ready; Dilute RNA to 57.2 ng/µL; Add 3.5 µl of RNA (200 ng total) to PCR tube; Make T7 Primer Mix (0.8 µL T7, 1 µL water);Add 1.8 µl of T7 Primer Mix to each tube. Each tube now has 5.3 µl; Denature the template and primer by incubating at 65°C for 10 min; Put reactions on ice for 5 min; 2) cDNA synthesis: Heat 5x First Strand Buffer 80°C for 3-4 minutes, vortex, do a quick spin, and keep at room temperature; Make cDNA master mix (2µL 5x First Strand Buffer, 1 µL 0.1M DTT, 0.5 µL 10 mM dNTP mix, 1.2 µL Affinity Script RNase Block Mix); Briefly spin sample and master mix tubes; Add 4.7 µL of cDNA Master Mix to sample tubes and mix by pipetting up and down. Each tube now has 10 µL; Incubate at 40°C for 2 hours; Incubate at 70°C for 15 minutes (inactivates the AffinityScript enzyme); If you don’t continue to next step, freeze samples -80°C; Ice for 5 min. 3) Labeled cRNA synthesis: Keep T7 RNA polymerase (red cap) on ice. Master Mix = 0.75 µL Water, 3.2 µL 5x Transcription Buffer, 0.6 µL 0.1M DTT, 1 µL NTP Mix, .21µL T7 RNA polymerase Blend; 0.24 µL Cy3-CTP or C5-CTP; Add 6 µL of Transcrption Master Mix with Dye to the cDNA. Each tube now contains a total volume of 16 µL; Incubate samples at 40°C for 2 hours. 4) SPRI cleanup of cRNA: Before beginning: Bring RNAClean XP to RT, Make up enough 80% ethanol for >2 200 µL washes per reaction, Set 96 well thermomixer to 37°C; Add 48 µL (3X) of RNAClean XP beads to each reaction and incubate for 10 minutes; Place plate on magnet and allow supernatant to clear (~5 minutes); Remove supernatant carefully, avoiding beads, and discard; Add 200 µL of 80% ethanol to each well and incubate for 30 seconds; Remove ethanol and repeat for a total of two washes; Remove ethanol, cover plate, and spin down for ~1 minute; Place plate back on magnet and use a p20 to remove residual ethanol; Dry beads at 37°C until beads no longer move around when placed on magnet (~4 minutes); Add 42.5 of RNAse-free water and resuspend beads, then incubate for 2-3 minutes; Place plate on magnet and transfer 40 µL to a new well before placing on ice
Hybridization protocol
8x15k Microarray hybridization: Set up 60°C heat block; Prepare 10x Blocking Agent: Add 1250 µL of nuclease-free water to vial containing 10x lyophilized Gene Expression Blocking Agent (part #5188-5281); Vortex and spin down; Prepare hybridization samples; For 8x15k (8-pack) microarrays, mix the following components: 450 ng Cy3-amplified cRNA, 450 ng Cy5-amplified cRNA, 7.5 µL 10X Gene Expression Blocking Agent, bring volume to 36 µL with water, and then add 1.5 µL 25x fragmentation buffer; Incubate at 60°C for 30 minutes to fragment the RNA; Add 37.5 µL of 2x Hi-RPM Hybridization Buffer to strip tube and leave on ice; Add fragmented RNA directly to Hi-RPM Buffer to stop the reaction; Mix well by pipetting up and down (avoid bubbles and do not vortex); Put sample on ice and load onto array as soon as possible; Put clean gasket onto metal hybridization chamber; Add 45 µL of hybridization onto gasket; Place array (Agilent facing down) onto gasket and close the hybridization chamber; Incubate on rotisserie at 65°C for 17 hr
Scan protocol
Labeled RNA was hybridized to Agilent 8x15k microarrays, which were then washed, scanned, and processed using the Agilent Feature Extraction software with default settings and loess dye bias correction.
Data processing
Each microarray has some number of spots (usually 2) for each gene. All microarrays for the GEV experiments had identical probes, likewise for the ZEV experiments. For each spot, we computed ratio = max(red, C) / max(green, C), where C=2 in arbitrary units. This minimum value application only affects 0.3% of spots. Typical genes have red and green channel measurements above 200. These (red, green, ratio) values foreach spot serve as the raw data. For each dataset, we aggregated the data across individual spots. Specifically, for each timecourse, time point, and gene, we aggregated the spot values and measured the minimum value, maximum value, median value, and standard deviation of the values. In the usual case of 2 spots, the median value is equivalent to mean and standard deviation is equivalent to (max-min)/sqrt(2). At this stage we corrected the most extreme outlier observations. First, for a given sample, we examined the case where, for a given gene, the ratio of the spot values, was larger than four. Since the spots values do not agree, we interpolated the value with the geometric mean across bracketing time points (or neighboring time point in the case of first or last time point). The second class of outliers is where the median ratio moves by a factor of at least four between two time points, and then by a factor of four in the opposite direction for the next time point. In these cases we again replaced the central point with the geometric mean of the bracketing points. These corrections apply to less than 0.2% of the data. In processing gene expression microarrays, crosstalk between red and green channels can occur. When the red channel fluorescence is much larger than the green channel fluorescence, there can be leakage of signal, and the green channel measurement is affected. To identify instances of this occurring, we first computed the green channel ratio relative to the time zero measurement. We then measured the 30% quantile of this value within each timecourse (for time points after t = 0), as well as the 30% quantile of the log-ratio. We then flagged timecourses where the green ratio quantile exceeded a factor of eight, and the log-ratio quantile exceeds a two-fold change. These represent cases where the green channel is increasing a great deal when it should be constant. For these rare cases, we repaired the red to green ratios by duplicating the time zero green channel across the full timecourse. This affects about two dozen TF-gene timecourses (out of more than a million). The GEV and ZEV systems both have characteristic gene expression signatures, including a mild stress response. Previously, it has been shown that Singular Value Decomposition (SVD) is one way to remove such signals. Here, we used a slightly different approach. First, we computed the median time series for each gene in each class of experiments (GEV and ZEV). We then subtracted this median time series, leaving a normalized log-ratio. Because a given gene is not directly or indirectly affected by a transcription factor in most experiments, the median is an accurate reflection of any background time-dependent behavior. We refer to the dataset where outliers are removed and the GEV/ZEV signal is removed as the cleaned dataset.
