A precisely adjustable, variation-suppressed eukaryotic transcriptional controller to enable genetic discovery

Conditional expression of genes and observation of phenotype remain central to biological discovery. Current methods enable either on/off or imprecisely controlled graded gene expression. We developed a 'well-tempered' controller, WTC846, for precisely adjustable, graded, growth condition independent expression of genes in Saccharomyces cerevisiae. Controlled genes are expressed from a strong semisynthetic promoter repressed by the prokaryotic TetR, which also represses its own synthesis; with basal expression abolished by a second, 'zeroing' repressor. The autorepression loop lowers cell-to-cell variation while enabling precise adjustment of protein expression by a chemical inducer. WTC846 allelic strains in which the controller replaced the native promoters recapitulated known null phenotypes (CDC42, TPI1), exhibited novel overexpression phenotypes (IPL1), showed protein dosage-dependent growth rates and morphological phenotypes (CDC28, TOR2, PMA1 and the hitherto uncharacterized PBR1), and enabled cell cycle synchronization (CDC20). WTC846 defines an 'expression clamp' allowing protein dosage to be adjusted by the experimenter across the range of cellular protein abundances, with limited variation around the setpoint.


Introduction
Since the spectacular demonstration of suppression of nonsense mutations and its application to T4 development (Epstein et al., 1963), means to express genes conditionally to permit observation of the phenotype have remained central to biological experimentation and discovery. During the 20th century, workhorse methods to ensure the presence or absence of gene products have included use of temperature-sensitive (ts) and cold-sensitive (cs) mutations within genes, for example to give insight into ordinality of cell biological events (Hereford and Hartwell, 1974). After the advent of recombinant DNA methods, conditional expression of genes into proteins, for example by derepression of lac promoter derivatives (Goeddel et al., 1979), also found application in biotechnology for production of therapeutics and industrial products (Sochor et al., 2015). In 2021, contemporaneous approaches to conditional expression in wide use include construction of transgenes activated by chimeric activators controlled by promoters whose expression is temporally and spatially restricted to different cell lineages (Brand and Perrimon, 1993), hundreds of approaches based on production of DNA rearrangements by phage-derived site specific recombination (Sauer, 1987), and triggered induction of engineered genes by chimeric transcription regulators with DNA-binding moieties based on derivatives of TetR from Tn10 (Gossen et al., 1995;Garí et al., 1997). Most of these approaches are all-or-none, in the sense that they are not intended to bring about expression of intermediate levels of protein; and the observations they enable are often qualitative.
But it has long been recognized that adjustment of protein dosage can provide additional insight into function that cannot be gained from all-or-none expression. For example, controlled expression of the bacteriophage l cI and cro gene products was key to understanding how changes in the level of those proteins regulated the phage's decision to undergo lytic or lysogenic growth . In S. cerevisiae, contemporaneous means to tune dosage include metabolite induced promoters, such as P GAL1 , P MET3 , P CUP1 (Maya et al., 2008), in which expression is controlled by growth media composition, and small molecule induced systems, such as the b-estradiol-induced LexA-hER-B112 system (Ottoz et al., 2014). Many of these depend on fusions between eukaryotic and viral activator domains and prokaryotic proteins (Garí et al., 1997;Ottoz et al., 2014;McIsaac et al., 2013;McIsaac et al., 2014) that bind sites on engineered promoters (Brent and Ptashne, 1985). These methods suffer from a number of drawbacks, including basal expression when not induced (Bellí et al., 1998;Garí et al., 1997;Ottoz et al., 2014), deleterious effects on cell growth due to sequestration of cellular components by the activation domain (Gill and Ptashne, 1988) induction of genes in addition to the controlled gene (McIsaac et al., 2013), and high cell-to-cell variation in expression of the controlled genes (Meurer et al., 2017;Elison et al., 2017;Ottoz et al., 2014).
These inducible systems rely on 'activation by recruitment' (Ptashne and Gann, 1997); the activator binds a site on DNA upstream of a yeast gene and recruits general transcription factors and regulators of the Pre-Initiation Complex (PIC). These assemble downstream at the 'core promoter' and recruit RNA polymerase II to induce transcription (Hahn and Young, 2011). An alternative to inducible activation would be to engineer reversible repression of yeast transcription by prokaryotic repressors (Brent and Ptashne, 1984;Hu and Davidson, 1987;Brown et al., 1987). For TATA-containing promoters, binding of prokaryotic proteins such as LexA and the lac repressor near the TATA sequence can repress transcription (Brent and Ptashne, 1984;Murphy et al., 2007;Wedler and Wambutt, 1995), presumably by interference with the formation of the PIC, transcription initiation, or early elongation. It has long been recognized (Brent, 1985) that prokaryotic repressors likely work through different mechanisms than mechanisms used by repressors native to eukaryotes (Wang et al., 2011;Gaston and Jayaraman, 2003).
We envisioned that an ideal conditional expression system to support genetic and quantitative experimentation would: (1) function in all growth media, (2) be inducible by an exogenous small molecule with minimal other effects on the cell, (3) manifest no basal expression of the controlled gene in absence of inducer, allowing generation of null phenotypes, (4) enable a very large range of precisely adjustable expression, and (5) drive very high maximum expression, allowing generation of overexpression phenotypes. Moreover, since differences in global ability to express genes into proteins (Colman-Lerner et al., 2005) lead to differences in allelic penetrance and expressivity (Burnaevskiy et al., 2019), the ideal controller should (6) exhibit low cell-to-cell variation at any set output, facilitating detection of phenotypes that depend on thresholds of protein dosage, and other inferences of single-cell behaviors from population responses.
Here, we describe the development of a prokaryotic repressor-based transcriptional controller of gene expression, Well-tempered Controller 846 (WTC 846 ), that fulfils the criteria outlined above. This development had three main stages. We first engineered a powerful eukaryotic promoter that is repressed by the prokaryotic repressor TetR and induced by the chemical tetracycline and its analogue Anhydrotetracycline (aTc), to use as the promoter of the controlled gene. Next, we used instances of this promoter to construct a configuration of genetic elements that show low cell-to-cell variation in expression of the controlled gene, by creating an autorepression loop in which TetR repressed its own synthesis. Third and last, we abolished basal expression of the controlled gene in the absence of the inducer, by engineering a weakly expressed 'zeroing' repressor, a chimera between TetR and an active yeast repressor Tup1. With WTC 846 , adjusting the extracellular concentration of aTc can precisely set the expression level of the controlled gene in different growth media, over time and over cell cycle stage. The gene is then 'expression clamped' with low cell-to-cell variation at a certain protein dosage, which can range from undetectable to greater abundance than wild type. We showed that strains carrying WTC 846 allelic forms of essential genes recapitulated known knockout phenotypes, and one demonstrated a novel overexpression phenotype. We constructed strains bearing WTC 846 alleles of genes involved in size control, growth rate, and cell cycle state and showed that these allowed precise experimental control of these fundamental aspects of cell physiology. We expect that WTC 846 alleles will find use in biological engineering and in discovery research, in assessment of phenotypes now incompletely penetrant due to cell-to-cell variation of the causative gene, in hypothesis directed cell biological research, and in genome-wide studies such as gene-by-gene epistasis screens.

Construction of a repressible P TDH3 promoter
Our goal was to engineer efficient repression of eukaryotic transcription by a bacterial repressor. We started with a strong (Ho et al., 2018), well-characterized, constitutive, and endogenous yeast promoter. This promoter, P TDH3 , has three key Transcription Factor (TF) binding sites, one for Rap1 and two for Gcr1 (Yagi et al., 1994;Kuroda et al., 1994) in its Upstream Activating Region (UAS), and a P 2tet P 3tet P 5tet P 7tet.1

Repressed Activity
Maximum Activity C autofluorescence / P TDH3 Repression and maximum activity of engineered P TDH3 derivatives. Diagrams above the plots display the genetic elements of strains used in B and C. Left diagram depicts strains used to test repressed activity, right diagram maximum activity. Px denotes any TetR repressible promoter. The * in TetR indicates a SV40 Nuclear Localization Sequence. In all strains, the P TDH3 derivative promoters diagrammed on the left directed the synthesis of Citrine integrated into the LEU2 locus. Grey boxes inside the diagrams denote tetO 1 TetR-binding sites. For measurement of repressed activity, an additional P ACT1 -directed TetR was integrated into the HIS3 locus. Citrine fluorescent signal was detected by flow cytometry. Fold difference refers to the median of the maximum activity divided by the median of the repressed activity signal. Fold over autofluorescence refers to median repressed activity signal divided by the median autofluorescent background signal. Maximum promoter activity is quantified as percentage of P TDH3 signal using the medians. x axis shows intensity of fluorescence signal. Plots are density distributions of the whole population, such that the area under the curve equals 1 and the y axis indicates the proportion of cells at each fluorescence value. The circles inside each density plot show the median, and the upper and lower bounds of the bar correspond to the first and third quartiles of the distribution. Repressed activity of P 3tet is above the x axis depicted in this figure, but can be seen in Figure 1-figure supplement 1. (C) Repression and maximum activity of the optimized P 7tet.1 . Diagrams and plots as in (B). P 7tet.1 contained additional binding sites for Rap1 and Gcr1 selected for higher activity, as well as an alternative TATA sequence as described in the Supplementary Information. It shows the highest fold difference, maximum activity comparable to P TDH3 , and low repressed activity. The online version of this article includes the following source data and figure supplement(s) for figure 1: Source data 1. Numerical data for Figure 1.  TATA sequence at which PIC assembles on the core promoter ( Figure 1A). Based on earlier work, we knew that binding of prokaryotic repressors to sites flanking the TATA sequence of P TDH3 repressed activity of this promoter (Wedler and Wambutt, 1995), presumably by interfering with PIC formation, transcription initiation, or early elongation. We therefore placed well characterized, 15 bp long TetR-binding sites (tetO 1 ) (Bertram and Hillen, 2008) immediately upstream and downstream of the P TDH3 TATA sequence to create P 2tet . To determine whether repressor binding could also block function in the UAS, we placed a single tetO 1 directly upstream of each Rap1 and Gcr1binding site to create P 3tet . We also combined the operators in these constructs to generate P 5tet ( Figure 1B). We integrated a single copy (Gnügge et al., 2016) of constructs bearing these promoters directing the synthesis of the fluorescent protein Citrine into the LEU2 locus (Griesbeck et al., 2001). We compared the Citrine fluorescence signal (measured by flow cytometry at wavelengths 515-545 nm) from these promoters to quantify their activity. We compared the strains Y2551[P 2tet ], Y2564[P 3tet ], and Y2566[P 5tet ] with an otherwise-isogenic strain in which Citrine was expressed from native P TDH3 (Y2683). This fluorescence signal measures Citrine expression, but also includes autofluorescent background from the yeast cells. We quantified this background by using the otherwiseisogenic parent strain Y70. Measured in this way, P 2tet had 76%, P 3tet 69%, and P 5tet 51% of P TDH3 activity ( Figure 1B). To assess repressibility of these promoters, we compared Citrine expression in these strains with expression in otherwise-isogenic strains in which a genomically integrated P ACT1 promoter drove constitutive expression of TetR (Y2562, Y2573, Y2577). By this measure, TetR repressed P 2tet by a factor of 12, P 3tet by a factor of 1.5, and P 5tet by a factor of 12 ( Figure 1B and Figure 1-figure supplement 1). Absolute repressed signal from these promoters was 4.3, 33, and 3 times the autofluorescence background. Because our aim was to create a promoter with no expression when repressed, we viewed even small reductions in repressed expression as useful and therefore decided to use P 5tet as a basis for further constructions.
Insertion of tetO 1 sites in P TDH3 to create P 5tet had reduced promoter maximum activity considerably. In order to regain the lost activity, we tested numerous constructs to find optimal placement for the tetO 1 sites, optimized Rap1, Gcr1, and TATA sequences, and increased the number of Rap1 and Gcr1 sequences (see Appendix 1 and Appendix 1- Figure 1). This work resulted in P 7tet.1 , which carried two Rap1 and three Gcr1 sites, sequence optimized to generate higher promoter activity, and an alternative TATA sequence to that of P TDH3 . By the assays described above, the new promoter P 7tet.1 (Y2661) showed comparable maximum expression to P TDH3 , 20-fold repression of Citrine signal, and absolute repressed activity (Y2663) of 4.3-fold over background ( Figure 1C). We chose P 7tet.1 as the promoter to develop our controller with.
Complex autorepressing (cAR) controller architecture expands the input dynamic range and reduces cell-to-cell variation We set out to optimize control of genes by P 7tet.1 . To do so, we tested the ability of different constructions that directed the synthesis of TetR to regulate P 7tet.1 -citrine directed fluorescence signal. Figure 2A shows the three different architectures. In Simple Repression (SR), the P 7tet.1 controlled gene was repressed by TetR expressed from a constitutive promoter. In Autorepression (AR), the P 7tet.1 controlled gene was repressed by TetR expressed from a second instance of P 7tet.1 , therefore creating a negative feedback loop. In Complex Autorepression (cAR), a second TetR gene expressed from a constitutive promoter was added to the AR architecture.
We compared the input-output relationship (i.e. dose response) for the three architectures. To do so, we constructed otherwise-isogenic strains with these architectures in which P 7tet.1 directed Citrine expression (Y2663, Y2674, and Y2741). We used flow cytometry to quantify Citrine fluorescence signal from all strains 7 hr after addition of different concentrations of aTc and fitted a log logistic model to the median fluorescence (see Materials and methods) ( Figure 2B&C).
Compared to the SR architecture, the AR architecture showed a more gradual dose response curve and a larger input dynamic range (the range of input doses for which the slope of the dose response curve was non-zero), from 3 to 400 ng/mL vs. 5-80 ng/mL aTc. This same flattening of the response curve and increased input dynamic range in autorepressing, synthetic TetR based eukaryotic systems has been described (Nevozhay et al., 2009), and we believe it operates in evolved prokaryotic systems including Tn10 and the E. coli SOS regulon, in which the TetR and LexA repressors repress their own synthesis (see Discussion). A broader input dynamic range allows more precise adjustment of protein levels, since small differences in inducer concentration (due for example to experimental errors, or differences in aTc uptake among cells) have smaller effects.
In these experiments, we also measured cell-to-cell variation (CCV) in the expression of the controlled gene. Many existing inducible gene expression systems show considerable variation in expression of the controlled gene, making it difficult to achieve homogenous phenotypes at the population level (Meurer et al., 2017;Elison et al., 2017;Ottoz et al., 2014) Figure 2. Comparison of the three controller architectures. (A) Genetic elements of the different controller architectures used in these experiments. The * next to TetR indicates SV40 Nuclear Localization Sequence and flat headed arrows indicate repression. In all cases, P 7tet.1 drives Citrine expression integrated at the LEU2 locus. In SR, the repressor of P 7tet.1 , tetR, is integrated at the HIS3 locus and is constitutively expressed. In AR, tetR is again integrated at the HIS3 locus, but is now expressed by P 7tet.1 . cAR has the same constructs as AR and an additional, constitutively expressed zeroing repressor integrated at the URA3 locus. (B) aTc dose response curves of Citrine expression for the three different architectures. Citrine fluorescence from strains bearing these architectures was measured at steady state using flow cytometry after 7 hr of induction with different concentrations of aTc. Symbols indicate the median fluorescence at each dose. Lines are fitted using a five-parameter log logistic function as explained in Materials and methods. Dashed line indicates autofluorescence signal measured from the parental strain without Citrine. (C) Slopes of the dose response curves in (B). The x axis range with non-zero slopes defines the useful input dynamic range. (D) Cell-to-cell variation of expression by these three architectures. We calculated single-reporter cell-to-cell variation (VIV) as described. Higher Residual Standard Deviation (RSD) values (y axis) correspond to greater VIV. Dot-dash line indicates the VIV of the strain where Citrine is constitutively expressed from P TDH3 and dashed line indicates VIV of autofluorescence in the parent strain without Citrine. Error bars indicate 95% confidence interval calculated using bootstrapping (n=1000) as described in Materials and methods. The online version of this article includes the following source data and figure supplement(s) for figure 2: Source data 1. Numerical data for Figure 2 panels B and D.      C. elegans, comparison of signals from strains with different constellations of reporter genes allows quantification of different sources of variation in protein dosage (Colman-Lerner et al., 2005;Pesce et al., 2018;Mendenhall et al., 2015). Here, we quantified overall variation in protein dosage by measuring the Coefficient of Variation (CoV) in fluorescent output from a single reporter (Figure 2-figure supplement 1), and we developed a second measure called Volume Independent Variation (VIV) (explained in Appendix 2) that normalized variation in dosage with respect to a key confounding variable, cell volume, to correct for its effect on protein concentration. In VIV, we estimated cell volume by a vector of forward and side scatter signals, and calculated the remaining (Residual) Standard Deviation of the single reporter output after normalization with this estimated volume ( Figure 2D and Figure 2-figure supplement 2). By both measures, strains carrying the SR architecture showed high variation throughout the input dynamic range, with a peak around the mid-point (12 ng/mL aTc). Strains bearing the AR architecture showed low overall CCV, and no peak at intermediate aTc concentrations. This diminution of CCV in synthetic, autorepressing TetR based eukaryotic systems has previously been described (Becskei and Serrano, 2000;Nevozhay et al., 2009). In the SR architecture, variations in the amount of TetR in different cells cannot be buffered.
In the AR architecture, such variations in repressor concentration are corrected for (see Discussion) and variation in expression of the controlled gene is at or around the same level as seen for constitutive expression driven by a number of native promoters (see Figure 2-figure supplement 4 for variation of commonly used promoters). This reduced cell-to-cell variation is useful for inferring single cell behaviors by observing population level responses (see Discussion). Compared with cells bearing the SR architecture, otherwise-isogenic cells bearing the AR architecture showed increased basal expression (6.3 vs. 4.1-fold over autofluorescence background). The increased basal expression was a consequence of the fact that in the AR architecture P 7tet.1 directs the synthesis of both the controlled Citrine gene and of TetR itself, so that, in uninduced cells, the steady state abundance of TetR was lower than in cells in which synthesis of TetR was driven by P ACT1 . More important, in the AR architecture, the fact that some amount of TetR expressed from P 7tet.1 was needed to repress its own synthesis meant that it would not be possible to abolish P 7tet.1driven expression of the controlled gene completely. Since ability to abolish basal expression of the controlled gene was an important design goal, we constructed strains with a third architecture, cAR, in which a different constitutive promoter drove expression of a second TetR gene in order to drive basal expression lower. Compared to otherwise-isogenic AR strains, strains expressing Citrine controlled by the cAR architecture showed reduced basal expression (4.1-fold over autofluorescence), and, compared to the otherwise isogenic SR strain, showed reduced CCV and a more gradual dose response ( Figure 2C&D and Hybrid repressor abolishes basal expression of P 7tet.1 To further decrease basal expression in the cAR architecture, we set out to create a more effective TetR derivative. Initially, we followed an approach that increased the size and nuclear concentration of TetR by fusing it to other inert bacterial proteins and nuclear localization sequences, but this approach was not enough to abolish all basal expression (see Appendix 3). P 3tet bears tetO 1 sites only in its UAS. The fact that P 3tet SR strains only showed weak repression (1.5-fold) suggested that TetR, and other inert derivatives described in the Appendix 3, exerted their effects on P 7tet.1 mostly by their action at the tetO 1 sites flanking the TATA sequence. We thus hypothesized that TetR derivatives that carried native, active yeast repressors might more effectively repress from sites in the UAS. The yeast repressor Tup1 complexes with Ssn6 (also called Cyc8) with a ratio of 4:1, forming a complex of 420 kDa (Varanasi et al., 1996), and this complex represses transcription through a number of mechanisms. These include repositioning and stabilizing nucleosomes to form an inacessible chromatin structure (Chen et al., 2013;Zhang and Reese, 2004;Ducker, 2000). Tup1 also blocks chromatin remodeling, masks activation domains, and excludes TBP (Wong and Struhl, 2011;Zhang and Reese, 2004;Mennella et al., 2003). LexA-Tup1 fusion proteins repress transcription when bound upstream of the Cyc1 promoter (Tzamarias and Struhl, 1994), and TetR-Tup1 fusions reduce uninduced expression in a dual TetR activator-repressor controller (Bellí et al., 1998). For P 7tet.1 , we imagined that as many as seven TetR-Tup1 dimers might bind to the promoter, potentially recruiting two additional Tup1 and one Ssn6 molecules per tetO 1 site. The resulting~3mDa of protein complexes might block activation by one or more of the above mechanisms. We therefore measured the ability of a TetR-nls-Tup1 fusion to repress P 7tet.1 -driven Citrine signal in SR strains. When its expression was directed from P ACT1 (Y2669), TetR-nls-Tup1 decreased uninduced fluorescence signal to background levels ( Figure 3A). Because fusion of TetR to a mammalian repressor domain in mammalian cells had shown very slow induction kinetics (Deuschle et al., 1995), we checked whether the TetR-nls-Tup1 fusion showed increased induction time compared to TetR alone but found no such effect ( Figure 3-figure supplement 1). Additionally, TetR-nsl-Tup1 abolished uninduced expression driven by P 3tet (Figure 3-figure supplement 2) (77-fold repression), compared to repression in otherwise isogenic strains by TetR, which showed basal expression reduced by only 1.5-fold ( Figure 1-figure supplement 1). By contrast, TetR-nls-Tup1 fusion repressed P 2tet , where tetO 1 flank only the TATA sequence, more strongly than TetR alone, but still showed basal expression. Our data thus suggested that the TetR-nls-Tup1 suppressed basal expression mainly by its effects in the UAS (see Discussion).
In the cAR architecture, the induction threshold, that is, the smallest concentration of inducer that can induce expression, is determined by the number of molecules of the repressors present before induction. We sought to lower the induction threshold in order to maximize the input dynamic range. Therefore, we constructed cAR controllers using TetR and TetR-nls-Tup1, to determine the lowest level of TetR-nls-Tup1 that could still abolish uninduced expression from P 7tet.1 . TetR-nls-Tup1 was driven by constitutive promoters of genes whose products were of decreasing abundance (Ho et al., 2018) (P ACT1 , P VPH1 , P RNR2 ,P REV1 ) (Y2673, Y2684, Y2749, and Y2715). The P ACT1 , P VPH1  Figure 3. Repressor optimization to abolish P 7tet.1 basal expression. A) Testing repression by the TetR-Tup1 fusion. The top diagram indicates the genetic elements of the SR architecture used to test the ability of the TetR-Tup1 fusion to abolish basal expression from P 7tet.1 . Diagrams to the left of the plot show the different repressors used. Each * indicates one SV40 Nuclear Localization Sequence. For both (A) and (B), Citrine fluorescence from P 7tet.1 repressed by the repressors indicated was measured using flow cytometry. Plots as in Figure 1. The circles inside each density plot show the median and the upper and lower bounds of the bar correspond to the first and third quartiles of the distribution. Numbers to the left of the plot indicate fold expression over autofluorescence, that is, the median of the Citrine fluorescence detected divided by the median of the autofluorescence signal. (B) Finding the lowest expression level of the zeroing repressor TetR-nls-Tup1 that abolishes basal expression from P 7tet.1 . The top diagram shows the genetic elements of the cAR architecture in the strains tested. Pc indicates a constitutive promoter. Promoters driving TetR-nls-Tup1 expression are indicated to the left of the plot. Numbers to the left of the plot as in (A). (C) Reducing expression of TetR-nls-Tup1 lowers induction threshold. The top diagram shows genetic elements of SR architecture in which synthesis of TetR-nls-Tup1 was directed by different promoters. The plot shows Citrine fluorescence measured using flow cytometry at steady state, 7 hr after induction with different aTc concentrations. Arrows indicate induction thresholds, defined as the lowest aTc dose where an increase in fluorescence signal was detected. Dashed line indicates autofluorescence control (parent strain without Citrine), circles indicate the median of the experimentally measured population, lines are fitted. Error bars indicate 95% confidence interval calculated using bootstrapping (n=1000) as explained in Materials and methods. The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Numerical data for Figure 3.   and P RNR2 strains showed no uninduced expression ( Figure 3B), while the P REV1 strain did (Figure 3figure supplement 3). Out of the three, Rrn2 protein is present at lower abundance, and the P RNR2driven TetR-nls-Tup1 has the lowest induction threshold in a dose response experiment with strains bearing SR architectures (Y2669, 2676, 2717) ( Figure 3C).
We therefore chose as our final controller the cAR architecture in which P 7tet.1 directed the expression of both TetR and of the controlled gene, while P RNR2 directed the synthesis of TetR-nls-Tup1. We constructed plasmids such that the tetR and tetR-nls-tup1 components are encoded on a single integrative plasmid, and a separate plasmid can be used to generate PCR fragments bearing P 7tet.1 for homologous recombination directed replacement of the promoter of any yeast gene. Due to its ability to give precisely regulated expression over a wide range of inducer concentrations, we called this construct a 'Well Tempered Controller' and gave it the number of Bach's first Prelude and Fugue (Bach, Johann Sebastian, 1685-1750. The Well Tempered Clavier. Book I: 24 Preludes and Fugues, BWV 846, C Maj) ( Figure 4A).

WTC 846 fulfills the criteria of an ideal transcriptional controller
We measured the time-dependent dose response of fluorescent signal in Y2759, the WTC 846 ::citrine strain during exponential growth using flow cytometry ( Figure 4B&C). Without aTc, there was no signal above background. After induction, signal appeared within 30 min. Time to reach steady state, which will be shorter for proteins that degrade more quickly (see Appendix 4), was 7 hr for the stable protein Citrine. Steady state expression was adjustable over aTc concentrations from 0.5 ng/ mL to 600 ng/mL, a 1200-fold input dynamic range. Maximum expression was similar to that for the P TDH3 -citrine strain Y2683. Direct observation of Citrine and TetR expression by Western blotting showed no expression of Citrine in absence of aTc, adjustable Citrine levels over the same input dynamic range and TetR expression synchronized with Citrine ( Figure  To better characterize the system, we also measured the shutoff speed of WTC 846 driven expression. We reasoned that the time to observable phenotypic effect of WTC 846 shutoff would depend on the speed of five processes: (i) aTc diffusion out of the cell, (ii) TetR binding to its operators, (iii) aTc sequestration by newly synthesized TetR, (iv) degradation and dilution of citrine mRNA, and (v) degradation and dilution of Citrine protein. Processes i-iii would lead to the cessation of transcript production by WTC 846 , and their speed would be the same for all WTC 846 controlled genes, whereas the speed of processes iv and v determine perdurance of the gene product and will be different for different mRNAs and proteins.
To measure shutoff, we grew the WTC 846 ::citrine strain (Y2759) to early exponential phase and then induced with a high concentration (600 ng/mL) of aTc and measured fluorescence signal every 30 min in flow cytometry. Additionally, after 30, 90, 150, and 210 min, we removed, washed, and resuspended a sample in (a) medium without aTc (to shut off expression from WTC 846 ) and (b) medium without aTc but with cyclohexamide (to shut off both WTC 846 and new protein synthesis). After shutoff, we expected to see an initial increase in signal, followed by decline from this peak. Increase in fluorescence after shutoff in (a) would depend on the time it took for WTC 846 to stop producing new mRNA (processes i-iii), the time it took for the existing mRNA to be degraded (process iv), and on continued fluorophore formation by already synthesized but immature Citrine proteins, which has a maturation time of around 30 min (Nagai et al., 2002). Whatever increase in fluorescence in (a) observed above that baseline found after shutoff in cycloheximide (b) would be due to WTC 846 shutoff speed and mRNA degradation speed.
As expected, we observed an initial increase of fluorescence in the shutoff samples ( Figure 4figure supplement 8A), which peaked for the samples in (a) (without cycloheximide) at around 60 min. A single-cell division takes 90 min and we therefore conclude that WTC 846 shutoff (events i-iii) is rapid and occurs within one cell division, and likely within 30 min given that the time between Citrine production and observable fluorescence is around 30 min. Subsequent reduction in fluorescence, which fell to half after 120 min in all samples, is an estimation of process v, that is, Citrine degradation + dilution (Figure 4-figure supplement 8B), and is consistent with the idea that the continued reduction in Citrine signal is caused by dilution by cell division. Overall, we conclude that WTC 846 (a.u.)                  shutoff is rapid, but the time required to see the phenotypic effects of the absence of the controlled gene product will primarily depend on the stability of the mRNA and expressed protein.
We also quantified the cell-to-cell variation in Citrine expression using the single reporter VIV measure for the WTC 846 ::citrine strain (Y2759) grown in YPD, and compared it to variation in a bestradiol (LexA-hER-B112) activation based transcriptional control system we previously described, and the commonly used galactose activated P GAL1 ( Figure  . At increasing concentrations of aTc, VIV initially rose to 0.63 at 8 ng/mL, similar to the VIV measured for Citrine expression repressed by P RNR2 -driven TetR-nls-Tup1 in an SR strain (Y2717, RSD of 0.67, Figure 4-figure supplement 5). At higher aTc inputs, VIV rapidly dropped below that seen in Y70, an otherwise-isogenic autofluorescence control strain, and reached the same low level (0.18) observed for Citrine whose expression was driven by P TDH3 (Y2683). Because the autofluorescence varied so greatly, absolute VIV for cells grown in different media could not be directly compared. However, under all growth conditions (Figure 4-figure supplement 7), VIV was highest at the similarly low concentrations of aTc and decreased at higher concentrations to the levels shown by the P TDH3 -citrine strain (Figure 4-figure supplement 2). We interpret the peak of VIV in the input dynamic range as arising from the fact that the WTC 846 architecture combines Simple Repression and Autorepression of the P 7tet.1 -controlled gene (here, Citrine). At low concentrations of inducer, in the SR regime, most repression of P 7tet.1 was due to the constitutively expressed TetR-nls-Tup1, and the peak VIV was similar to that found for the strain where P 7tet.1 was repressed by constitutively expressed TetR-nls-Tup1 (Figure 4-figure supplement 5 and see previous Results section). At higher concentrations of aTc, in the AR regime, P 7tet.1 is derepressed, the concentration of TetR and the ratio of TetR to TetR-nls-Tup1 is large. At these inducer concentrations, TetR controls its own synthesis and variation is suppressed by this negative feedback, resulting in much lower cell-to-cell variation throughout the dynamic range compared to routinely used transcriptional controllers. Taken together, these results indicated that WTC 846 fulfilled our initially stated criteria for an ideal conditional expression system.

WTC 846 alleles allow precise control over protein dosage and cellular physiology
We then assessed the ability of WTC 846 to direct conditional expression of endogenous genes. We selected (i) genes that are essential for growth, but for which previously generated transcriptionally controlled alleles still formed colonies on solid medium (CDC42, TOR2, PBR1, CDC20) or continued to grow in liquid medium (PMA1) under uninduced conditions, (ii) essential genes for which existing transcriptionally controlled alleles did not show the expected overexpression phenotype (IPL1), or (iii) essential genes for which conditional expression alleles did not exist (CDC28) (Mnaimneh et al., 2004;Yu et al., 2006;Dechant et al., 2014). These genes encoded proteins with a variety of functions: stable (Cdc28) and unstable (Cdc20 and Cdc42) cell cycle regulators, a spindle assembly checkpoint kinase (Ipl1), a metabolic regulator (Tor2), a putative oxidoreducatase (Pbr1), and a high abundance membrane proton pump (Pma1). The encoded proteins spanned a range of abundance from~1000 (Tor2 and Ipl1) to >50,000 (Pma1) molecules per cell (Ho et al., 2018).
We constructed strains in which WTC 846 controlled the expression of these genes. Before transformation the cells were grown in liquid medium containing aTc, and then plated on solid medium containing aTc (see Appendix 5 for a detailed protocol) ( Figure 5A and Table 1, strains labeled WTC 846-Kx ::gene_name). To make these strains, we integrated a single plasmid-borne TetR-nls-Tup1 and autorepressing TetR construct into the LEU2 locus in a BY4741 background, and replaced sequences upstream of the ATG of the essential gene with a~1940 bp casette carrying an antibiotic selection marker and P 7tet.1 , without altering the sequence of the upstream gene or its terminator. In most cases we removed between 20 and 200 bp of the endogenous gene promoter. The cassette carried one of three different 15 bp translation initiation sequences (extended Kozak sequences; K1, K2, K3) as the last 15 bases before the ATG. These were designed to enable different levels of translation of the gene's mRNA (Li et al., 2017). The predicted efficiency of the sequences was K1> K2> K3. If cells of a strain carrying a WTC 846 -controlled essential gene formed colonies on solid medium without aTc, we constructed an otherwise-isogenic strain with a lower efficiency Kozak sequence (data not shown).
We spotted serial dilutions of cultures of the final seven strains on YPD, YPE, SD, S Glycerol and SD Proline plates with and without inducer, and assessed the strains' ability to grow into visible Fluorescence x10   At high aTc concentrations, the WTC 846-K2 ::IPL1 strain formed colonies with lower plating efficiency than the parent strain. Ipl1 is a component of the kinetochore and is required for correct sister chromatid separation during mitosis. In mouse embryonic fibroblasts, overexpression of the orthologous Aurora B kinase causes aberrant chromosome segregation and increases duration of mitosis by activating the Spindle Assembly Checkpoint, which stops mitosis until correct spindle attachments to sister chromatids can be formed (González-Loyola et al., 2015). In a previous study in S. cerevisiae, however, P GAL1 -driven overexpression of Ipl1 did not decrease plating efficiency, did not cause accumulation of cells with 2 n DNA content unable to complete mitosis, and did not cause aberrant chromosome segregation as assessed by microscopy, unless Ipl1 was overexpressed simultaneously with another kinetochore component (Sli15) (Muñoz-Barrera and Monje-Casas, 2014). We asked whether WTC 846 -driven Ipl1 overexpression alone could cause missegregation phenotypes in S. cerevisiae. We cultured WTC 846-K2 ::IPL1 cells for 18 hr in YPD with a high concentration of aTc (400 ng/ mL), and measured total DNA content in flow cytometry to assess cell cycle state. In these cultures compared to the parent with WT Ipl1, many cells were in the G2/M phase with 2 n DNA content, indicative of an inability to complete mitosis, and a significant portion of the population showed aberrant chromosome numbers above 2 n ( Figure 5-figure supplement 2). That is, WTC 846 -driven Ipl1 overexpression in S. cerevisiae caused a previously undescribed phenotype, which resembled that caused by Aurora B overexpression in mammalian cells. To determine why WTC 846 -driven Ipl1 overexpression caused this phenotype while P GAL1 -driven overexpression did not, we compared WTC 846 -driven Citrine expression with Citrine driven by P GAL1 carried on a centromeric plasmid. Compared with WTC 846 -driven expression, centromeric P GAL1 plasmid expression was twofold lower, and cell-to-cell variation was~4.5-fold higher ( Figure 5-figure supplement 7). Either the lower expression or the higher variation, or both, might account for the fact that P GAL1 driven Ipl1 overexpression does not result in the mammalian Aurora B phenotype in S. cerevisiae.
We tested whether adjustable expression of metabolic and essential genes could be used to titrate growth rates. We constructed strains with WTC 846 alleles of Tor2, a low abundance, stable, essential protein necessary for nutrient signalling and actin polarization (Bartlett and Kim, 2014), Source data 1. Raw plate images for Figure 5B and raw data for Figure 5C. Source data 2. Numerical data for Figure 5D.          2683 P TDH3 -const leu2D::P TDH3 _citrine-LEU2 2551 P 2tet -const leu2D::P 2tet _citrine-LEU2 2564 P 3tet -const leu2D::P 3tet _citrine-LEU2 2566 P 5tet -const leu2D::P 5tet _citrine-LEU2 2562 P 2tet -SR leu2D::P 2tet _citrine-LEU2 his3D::P ACT1 _tetR-NLS-HIS3 2573 P 3tet -SR leu2D::P 3tet _citrine-LEU2 his3D::P ACT1 _tetR-NLS-HIS3 2577 P 5tet -SR leu2D::P 5tet _citrine-LEU2 his3D::P ACT1 _tetR-NLS-HIS3 2659 P 5tet.1 -const leu2D::P 5tet.1 _citrine-LEU2  Pma1, an abundant, essential proton pump that regulates the internal pH of the cell (Ambesi et al., 2000), and Tpi1, a highly abundant, non-essential glycolytic enzyme (Fraenkel, 2003) (Y2773, 2828, 2849). We cultured WTC 846 ::TOR2, WTC 846 ::PMA1, and WTC 846 ::TPI1 cells in different liquid media over a large input dynamic range of aTc, and measured growth by scattered light intensity in a growth reader as a proxy for culture density (Biolector or GrowthProfiler) ( Figure 5C for Tor2 and Figure 5-figure supplement 3 for all three proteins). All strains showed distinct growth rates at different aTc concentrations. For all strains, we identified an aTc concentration that resulted in the same growth rate as the otherwise-isogenic strain bearing the native gene promoter. In order to assess whether the WTC 846 ::PMA1 strain showed the expected hypomorphic phenotype of defective daughter cell separation (Cid et al., 1987), we used flow cytometry and Sytox Green staining to quantify DNA content. At low aTc concentrations, cells showed an apparent increase in ploidy and cell size and microscopic observation showed that each mother had multiple daughters attached to it ( Figure 5-figure supplement 4). Observation of WTC 846 ::TOR2 strains revealed a novel overexpression phenotype: at high aTc concentrations, cells bearing the higher translational efficiency TOR2 allele (WTC 846-K1 ::TOR2) grew more slowly than the otherwise-isogenic control parent strain with WT TOR2 (Figure 5-figure supplement 3D, compare 600 ng/mL line to blue dashed line). The strain with the less efficient WTC 846-K3 ::TOR2 allele did not show this overexpression phenotype. These results demonstrate that researchers can adjust input to WTC 846 alleles to tune protein levels and different growth rates with a level of precision not achievable until now, and that the dynamic range of phenotypic outputs can be further expanded by the ability to construct WTC 846 alleles with alternative Kozak sequences to observe phenotypes at the two dosage extremes. We then tested whether adjustable gene expression could precisely regulate cell size. In S. cerevisiae, Whi5 regulates the volume at which unbudded cells commit to a round of division and start forming buds. whi5D cells are smaller, and cells expressing Whi5 under P GAL1 control are larger than otherwise-isogenic cells (de Bruin et al., 2004). Whi5 controls cell volume by a complex mechanism and unlike most other proteins, its abundance does not scale with cell volume (Schmoller et al., 2015). Whi5 mRNA and protein are expressed during S/G2/M (in haploids, at about 2500 molecules), and Whi5 is imported into the nucleus in late M phase (Taberner et al., 2009), where it suppresses transcription of the G1 cyclins needed to commence a new round of cell division (de Bruin et al., 2004;Taberner et al., 2009). During G1, as cells increase in volume, the nuclear concentration of Whi5 falls due to dilution (Schmoller et al., 2015) and slow nuclear export (Qu et al., 2019) until a threshold is reached, after which Whi5 is rapidly exported from the nucleus, and cells enter S phase. To test whether we could control cell volume by controlling Whi5, we constructed haploid and diploid WTC 846 ::WHI5 strains (Y2791, Y2929). In these strains, we expected Whi5 to be expressed throughout the cell cycle, but that import of the protein into the nucleus during late M phase, and diminution of nuclear concentration to below the threshold needed to START as cell volume increased in G1, should remain unaffected. We expected that the volume of these cells should scale with the concentration of the aTc inducer. We grew these strains along with otherwise isogenic control strains in S Ethanol to exponential phase at different aTc concentrations, and measured cell volume using a Coulter counter. Increasing Whi5 expression resulted in increasingly larger cells ( Figure 5D). Without aTc, diploid WTC 846 ::WHI5 cells were about the same volume as haploid controls (median 27fL vs 25fL), whereas haploid WTC 846 ::WHI5 cells were only slightly smaller at 24fL. At around 10 and 12 ng/mL aTc, both haploid and diploid strains had about the same volume as controls. At full induction, both WTC 846 ::WHI5 strains had a median volume of around 60fL, almost twice as large as the diploid control, yielding a more than twofold range of possible cell volumes BY4743, whi5D::WHI5-mKokappa-HIS3/WHI5 myo1D::MYO1-mKate(3x)-KanMX/MYO1 leu2D::P RNR2 _tetR-NLS-tup1_P 7tet.1 _tetR-NLS-LEU2/leu2D0 ura3D::P RNR2 _tetR-NLS-tup1_P 7tet.1 _tetR-NLS-URA3/ura3D0 P_WHI5::P 7tet.1-K1 -HygMX/ P_WHI5::P 7tet.1-K1 -NatMX attainable using WTC 846 for both haploid and diploid cells. We also calculated the CoV of cell volume to assess cell-to-cell variation of this WTC 846 directed phenotype. For most of the volume range, the CoV was around the same level as for the control strains with WT Whi5 (Figure 5-figure supplement 5A&B). Both diploid and haploid cells (especially haploids) expressing high levels of Whi5 showed increased variation in volume. We quantified DNA content of the haploid strain in a high aTc concentration using Sytox staining and found an increase in the number of aneuploid cells (>2n) ( Figure 5-figure supplement 5C). We therefore believe that overexpression of Whi5 leads to endoreplication, and the increased variation in volume at high aTc concentrations in the haploid strain originates from these endoreplicated cells. Finally, we tested the ability of WTC 846 to exert dynamic control of gene expression by constructing a WTC 846-K3 ::CDC20 strain (Y2837) and using this allele to synchronize cells in batch culture by setting Cdc20 expression to zero and then restoring it (Juanes, 2017). Cdc20 is an essential activator of Anaphase Promoting Complex C, which once bound to Cdc20, initiates the mitotic metaphase to anaphase transition (Pesin and Orr-Weaver, 2008), and is then degraded during anaphase. Upon depletion of Cdc20, for example by shift of ts strains to the restrictive temperature, or transcriptionally controlled alleles to non-inducing medium, cells arrest in metaphase with large buds and 2 n DNA content. When Cdc20 is restored by switching to the permissive condition, cells enter the next cell cycle simultaneously (Cosma et al., 1999;Shirayama et al., 1998). For an investigator to be able to use WTC 846-K3 ::CDC20 to synchronize the cells in a culture, the investigator would need to shut off Cdc20 expression completely, and then re-express it in all the cells in a population. To test the feasibility of this, we diluted exponentially growing WTC 846-K3 ::CDC20 cells into YPD medium without aTc (0.5 million cells/mL) and took samples for Sytox staining and flow cytometry analysis for DNA content at fixed intervals. Within 480 min, the entire culture had arrested at the G2/M phase with 2 n DNA content ( Figure 5E). Microscopic inspection confirmed that cells had arrested with large buds, as is expected upon a G2/M arrest. We next added 600 ng/mL of aTc. As assayed by Sytox staining and flow cytometry and confirmed by microscopy, cells then re-entered the cell cycle within 35 min and went through one cell cycle completely synchronously. Induction of WTC 846 is thus rapid, indicating that diffusion of aTc into the cell, and TetR unbinding of tetO are also rapid.
We also determined the arrest time of cells pre-cultured with a lower concentration (3 ng/mL) of aTc ( Figure 5-figure supplement 6). These cells had a lower concentration of Cdc20 before aTc was removed, and therefore required less time to reach complete arrest (~210 min as opposed tõ 480 min). This suggests that the predominant contribution to the time to reach complete arrest is the concentration and stability of Cdc20. Given this, and the rapid shutoff kinetics of WTC 846 presented in Figure 4-figure supplement 8, we conclude that the shutoff dynamics of WTC 846 controlled phenotypes depend mostly on the speed of degradation of the controlled protein.

Discussion
Conditional expression of genes and observation of phenotype remain central to biological discovery. Many methods used historically, such as suppression of nonsense mutations, or conditional inactivation of temperature sensitive mutations, do not facilitate titration of graded or intermediate doses of protein. More current methods for graded expression do not allow experimenters to adjust and set protein levels and show high cell-to-cell variation of protein expression in cell populations, limiting their utility for elucidating protein-dosage-dependent phenotypes. Moreover, most such methods also have secondary consequences including slowing of cell growth. In order to overcome these limitations, we developed for use in S. cerevisiae a 'Well-tempered Controller'. This controller, WTC 846 , is an autorepression-based transcriptional controller of gene expression. It can set protein levels across a large input and output dynamic range. As assessed by Citrine fluorescence readout, WTC 846 alleles display no uninduced basal expression, and uninduced WTC 846 alleles of poorly expressed proteins display complete null phenotypes. WTC 846 alleles also exhibit high maximum expression, low cell-to-cell variation, and operation in different media conditions without adverse effects on cell physiology.
The central component of WTC 846 is an engineered TATA containing promoter, P 7tet.1 . We and others had shown that prokaryotic repressors including LexA (Brent and Ptashne, 1984), TetR (Murphy et al., 2007), lcI (Wedler and Wambutt, 1995) and LacI (Hu and Davidson, 1987;Figge et al., 1988) can block transcription from engineered TATA-containing eukaryotic promoters, when those promoters contain binding sites between the UAS (or, for vertebrate cells, the enhancer) and the TATA (Brent and Ptashne, 1984) or downstream of or flanking the TATA. To develop P 7tet.1 , we placed seven tetO 1 TetR-binding sites in the promoter of the strongly expressed yeast gene TDH3. Two of the sites flank the TATA sequence, the other five abut binding sites for an engineered UAS that binds the transcription activators Rap1 and Gcr1. In WTC 846 , one instance of P 7tet.1 drives expression of the controlled gene, while a second instance of P 7tet.1 drives expression of the TetR repressor, which thus represses its own synthesis.
We believe that repression of P 7tet.1 by TetR is due mainly to its action at the two tetO 1 sites flanking the TATA sequence, because TetR represses a precursor promoter that only carries such sites to the same extent. The mechanism(s) by which binding of repressors near the TATA might interfere with PIC formation, transcription initiation, or early elongation remain unknown, as well as why binding of larger presumably transcriptionally inert TetR fusion proteins results in stronger repression. However, examination of the Cryo-EM structure of TBP and TFIID bound to mammalian TATA promoters (Nogales et al., 2017) suggests that binding of TetR and larger derivatives of it to these sites might simply block PIC assembly. Studies of repression of native Drosophila melanogaster promoters by the en and eve homeobox proteins show that a similar, steric occlusion based mechanism can block eukaryotic transcription by binding of the repressors to sites close to the TATA sequence (Ohkuma et al., 1990;Austin and Biggin, 1995).
In WTC 846 , when inducer is absent, measured basal expression of the controlled gene is abolished by a second TetR derivative, a fusion bearing an active repressor protein native to yeast. Because the same TetR-nls-Tup1 fusion protein fully represses a precursor promoter that only carries TetR operators in the UAS, we believe that the main zeroing effect of TetR-nls-Tup1 is manifested through binding the tetO 1 sites in the UAS. Native Tup1 repressor complexes with Ssn6 (also called Cyc8) to form a 420 kDa protein complex (Varanasi et al., 1996), and TetR binds DNA as a dimer. In gcr1D cells, in which transcription from P TDH3 is severely diminished, the native P TDH3 promoter has two nucleosomes positioned between the UAS and the transcription start site (Pavlovic´and Hö rz, 1988). It is thus possible that in the UAS as many as five very large dimeric TetR-nls-Tup1 complexes block binding of Gcr1 and Rap1, mask their activating domains, or some combination of these, resulting in similar placement of two nucleosomes in P 7tet.1 . One of these nucleosomes could then be positioned at the 294nt stretch between the UAS and the TATA sequence. It also seems possible that binding of the TetR-nls-Tup1 repressor might shift the position of the second nucleosome further downstream, so that it obscures the transcription start site.
Both the increased input dynamic range and the lower cell-to-cell variation in expression from WTC 846 arise from the fact that the TetR protein that represses the controlled gene also represses its own synthesis. This autorepression architecture is common in prokaryotic regulons (Smith and Magasanik, 1971;Brent and Ptashne, 1980) including Tn10, the source of the TetR gene used here, and it has been engineered into eukaryotic systems (Becskei and Serrano, 2000;Nevozhay et al., 2009). In self-repressing TetR systems, the input (here, aTc) and TetR output together function as a comparator-adjustor (Andrews et al., 2016). In such systems, aTc diffuses into the cell. Intracellular aTc concentration is limited by entry. Inside the cell, aTc and TetR free concentrations are continuously compared by their binding interaction. If TetR free is in excess, it represses TetR expression, and total intracellular TetR concentration is reduced by dilution, cell division, and active degradation of DNA-bound TetR until an equilibrium determined by the intracellular aTc concentration is again reached. The consequence of this autorepression is that the WTC 846 requires more aTc to reach a given level of controlled gene expression than strains in which TetR is expressed only constitutively. Autorepression flattens the dose response curve, increases the range of aTc doses where a change in promoter activity can be observed, and buffers the effects of stochastic cell-to-cell variations in TetR concentration, thereby reducing cell-to-cell variation in expression of the controlled gene throughout the input dynamic range.
We further extended the output dynamic range of WTC 846 -controlled genes by developing three different Kozak sequences, K1, K2, and K3 (Li et al., 2017), to allow controlled genes to be translated at different levels. We used these sequences to construct strains bearing conditional alleles of the essential genes CDC28, IPL1, TOR2, CDC20, CDC42, PMA1, and PBR1. These strains all showed graded expression of growth and other phenotypes, from lethality at zero expression to penetrant expression of previously reported phenotypes at higher protein dosage (Mnaimneh et al., 2004;Yu et al., 2006;Dechant et al., 2014;Muñoz-Barrera and Monje-Casas, 2014). We used controlled expression in the WTC 846 ::CDC20 strain to bring about G2/M arrest followed by synchronous release, with low cell-to-cell variation in induction timing, demonstrating that WTC 846 can be used in experimental approaches that require dynamic control of gene expression.
Both induction and shutoff with WTC 846 are rapid, as both Citrine and Cdc20 expression occur within 30 min of induction, and shutoff of Citrine expression is observed within 60 min. However, time to steady state expression after induction, reached when degradation and dilution through cell division balance new synthesis, takes longer. Time to steady state will depend on the stability of the controlled protein. This is 6-7 hr for the stable protein Citrine, and the majority of yeast proteins have similar stability (Wiechecki et al., 2018). Those proteins with shorter half-lives will reach steady state faster. Furthermore, we showed in WTC 846 ::IPL1 strains that high level expression of this spindle assembly checkpoint kinase arrests cells at G2/M with 2 n or higher DNA content. This phenotype, thought to be due to disruption of kinetichore microtubule attachments, is displayed in mammalian cells when the homologous Aurora B is overexpressed (González-Loyola et al., 2015), but had not been observed previously in S. cerevisiae when Ipl1 was overexpressed from P GAL1 (Muñoz-Barrera and Monje-Casas, 2014). We also showed that in WTC 846 ::WHI5 strains, different levels of controlled expression of Whi5 can constrain cell sizes within different limits.
Cell-cell variation in WTC 846 -driven expression is highest at low aTc levels, because control in this regime depends mostly on the higher variation Simple Repression by TetR-Tup1 expressed from the P RNR2 . This variation at low input doses in the WTC 846 represented a trade-off between the design goals of abolition of basal expression and suppression of cell-to-cell variation. The Autorepression (AR) architecture better suppressed cell-to-cell variation in controlled gene expression at low inducer inputs, but, because of the fact that TetR and the controlled gene were both under the control of the same repressible promoter, the controlled gene still showed considerable basal expression when uninduced. Given that suppression of basal expression of the controlled gene was critical to generating 'reversible null' phenotypes, we developed the AR architecture further. The resulting cAR configuration of WTC 846 , had low cell-to-cell variation, equivalent to the variation at the lowest expression levels that AR could achieve. Importantly, because transcription of WTC 846 -controlled genes is synchronized to that of the autorepressing TetR gene, transcription and mRNA abundance of WTC 846 -controlled genes should be steady throughout the cell cycle. This autorepressing circuitry operationally defines WTC 846 as an 'expression clamp', a device for adjusting and setting gene expression at desired levels, and maintaining it with low cell-to-cell variation, and so allowing expressed protein dosage in individual cells to closely track the population average.
Taken together, our results show that WTC 846 controlled genes define a new type of conditional allele, one that allows precise control of gene dosage. We anticipate that WTC 846 alleles will find use in cell biological experimentation, for example in assessment of phenotypes now incompletely penetrant due to variable dosage of the causative gene products (Casanueva et al., 2012), and for sharpening the thresholds at which dosage dependent phenotypes manifest. We also hope that genome wide collections of WTC 846 alleles might enable genome wide gene-by-gene and gene-bychemical epistasis for interactions that depend on gene dosage. In S. cerevisiae, recent development of strains and methods (such as the SWAP-Tag [Weill et al., 2018]) that facilitate installation of defined N and C terminal genetic elements after cycles of mating, sporulation, and selection of desired haploids should allow generation of whole genome WTC 846 strains for this purpose. Epistasis screens rely on measurement of colony size on plates or culture density in liquid media. For two proteins whose effect on growth rate was identical, a one-generation difference in achievement of steady state expression could result in a twofold difference in number of cells in a colony or well, and thus in a 1.26-fold difference in colony diameter. We therefore suggest that growth rate-based assays using WTC 846 or any other inducible system pre-induce cells several generations before plating or pinning. WTC 846 alleles may find use in engineering applications such affinity maturation of antibodies expressed by yeast surface display, where precise ability to lower surface concentration should aid selection for progressively higher affinity binders. WTC 846 can also be a useful complement to boost the efficiency of methods that act at the protein level such as induced degradation or AnchorAway techniques. Such techniques could be used in conjunction with WTC 846 to achieve rapid and sustained shutoff from a well-maintained steady state level. This would also allow fast step function decreases in abundance. For example, an experimenter might simultaneously induce depletion of the product of a controlled gene by such a method while adjusting aTc downward to rapidly reset the level of an expressed protein to a new, lower level. Implementation of the WTC 846 control logic in mammalian cells and in engineered multicellular organisms should allow similar experimentation now impossible due to cell-to-cell variation and imprecise control.

Materials and methods
Plasmids Information on plasmids, and promoter and protein sequences used in this study can be found in Supplementary file 1 -Tables S2 and S4. Plasmids with auxotrophic markers were constructed based on the pRG shuttle vector series (Gnügge et al., 2016) using either restriction enzyme cloning or isothermal assembly (Gibson et al., 2009). Inserts were generated either by PCR on existing plasmids or custom DNA synthesis (GeneArt, UK). Oligos for cloning and for strain construction were synthesized by Thermofisher, UK. Plasmids used to generate linear PCR products for tagging transformations were based on the pFA6 backbone (Janke et al., 2004). Plasmids necessary to construct WTC 846 strains are available through Addgene. Plasmid structures and a detailed protocol for strain construction can be found in Appendix 5.
pRG shuttle vector series backbones used for integrative transformations have T7 and T3 promoters flanking the insert (Gnügge et al., 2016). During cloning, the insert of plasmids bearing TetR were cloned such that the insert promoter was closer to the T7 promoter and the terminator was near the T3 promoter of the backbone. In plasmids bearing Citrine, the insert was flipped onto the opposite strand, such that the insert promoter was near the T3 promoter, and the terminator near the T7 promoter. This inversion was done to avoid homologous recombination during subsequent integration of these plasmids into the same strain, since in many strains TetR and Citrine were flanked by the same promoter and the same terminator.

Strains
Strains used in this study can be found in Supplementary file 1 - Table S1. Strains used for fluorescent measurements and the WTC 846-K3 ::TPI1 strain are based on a BY4743 derivative haploid background (MATa his3D leu2D met15D ura3D lys2D). Strains where P 7tet.1 replaced endogenous promoters were based on the haploid BY4741 background with the modifications whi5D::WHI5-mKOkappa-HIS3, myo1D::MYO1-mKate(3x)-KanMX and so were resistant to G418. The oligos used to replace the promoters of the different endogenous genes with WTC 846 -controlled P 7tet.1 can be found in Supplementary file 1 - Table S3. Correct replacement of the endogenous promoter with P 7tet.1 was checked using colony PCR with the protocol from the Blackburn lab (also detailed in Gnügge et al., 2016), and subsequent sequencing (Microsynth, Switzerland). For colony PCR, we used a standard forward oligo annealing to P 7tet.1 , and gene specific reverse oligos annealing within the tagged gene. Oligo sequences for colony PCR can be found in Supplementary file 1 - Table S3. A comprehensive protocol on how to generate strains where WTC 846 controls endogenous genes can be found in Appendix 5.

Chemicals and media
YPD/YPE was prepared with 1% yeast extract (Thermofisher, 212720), 2% bacto-peptone (Thermofisher, 211820), and 2% glucose (Sigma, G8270) / ethanol (Honeywell, 02860). Synthetic (S) media except SD Proline contained 0.17% yeast nitrogen base (without amino acids and ammonium sulfate) (BD Difco, 233520) with 0.5% ammonium sulfate (Sigma, 31119) as nitrogen source, complete complement of amino acids and adenine and uracil, except for SD min which contained only the necessary amino acid complements to cover auxotrophies. SD Proline media contained 0.17% yeast nitrogen (without amino acids and ammonium sulfate), only the amino acids necessary to cover auxotrophies and 1 mg/mL proline as the sole nitrogen source. The carbon source was 2% glucose for SD and SD Proline, 2% ethanol for S Ethanol, 3% glycerol for S Glycerol (Applichem, A2957), 2% fructose for S Fructose, 2% Raffinose for S Raffinose and 2% Galactose together with 2% Raffinose for S GalRaf. Experiments were performed in YPD media unless otherwise specified. Solid medium plates were poured by adding 2% agar (BD Sciences, 214040) to the media described above.
aTc was purchased from Cayman Chemicals (10009542) and prepared as a 4628.8 ng/mL (10 mM) stock in ethanol for long term storage at À20˚C and diluted in water for experiments as necessary.
When constructing strains where P 7tet.1 replaces endogenous promoters, a PCR fragment containing P 7tet.1 and an antibiotic marker (either Nourseothricin (Werner BioAgents, clonNAT) or Hygromycin (ThermoFisher,10687010)) was transformed for homologous recombination directed replacement of the endogenous promoter. Cells were plated on YPD + antibiotic plates for selection. Whenever the promoter of an essential gene was being replaced, transformations were plated on multiple plates with YPD + antibiotic and 10/50/100/500 ng/mL aTc.

Spotting assay
For spotting assays of cell growth and viability, cells were precultured in YPD media with 20 ng/mL aTc (except for WTC 846-K2 ::IPL1 strain which was precultured in 10 ng/mL aTc) and the necessary antibiotic to stationary phase, and diluted into YPD + antibiotic without aTc at a concentration of 0.8x10 6 cells/mL. Six hr later, cells were spun down and resuspended in YPD. Cells were spotted onto plates containing different media and aTc concentrations prepared as described above such that the most concentrated spot has 2.25x10 6 cells, and each column is a 1:10 dilution. Pictures were taken after 24 hr for the YPD and SD plates, and 42 hr for SD Proline, S Glycerol and YPE plates.

Flow cytometry
Cells were diluted 1:200 from dense precultures and cultured to early exponential phase (2-5 x 10 6 cells/mL) in 96 deep-well plates at 30˚C before induction with aTc if necessary. For aTc dose responses, samples were taken at times indicated. For experiments where no dose response was necessary, cells were measured at least 4 hr after dilution of precultures, but always before stationary phase. Samples were diluted in PBS and measured using a LSRFortessa LSRII equipped with a highthroughput sampler. PMT voltages for the forward and side scatter measurements were set up such that the height of the signal was not saturated. Citrine fluorescence was quantified using a 488 nm excitation laser and a 530/30 nm emission filter. PMT voltage for this channel was set up such that the signal from P TDH3 expressed Citrine did not saturate the measurement device, except for basal level measurements in Figure 3B and Figure 3-figure supplement 3, where PMT voltage for the Citrine channel was increased to maximum. Side scatter was measured using the 488 nm excitation laser and 488/10 nm emission filter.

Western blots
Cells were grown to stationary phase with the indicated aTc concentration. 5 mL of cell culture was centrifuged and resuspended in 1 mL 70% ethanol. Fixed cells were again centrifuged, and resuspended in 200 uL Trupage LDS loading buffer (Merck, PCG3009) supplemented with 8M urea. Cells were broken using glass beads and a bead beater, and boiled at 95˚C for 30 min. Proteins were separated using SDS-Page with Trupage precast 10% gels (Merck, PCG2009-10EA) and the associated commercial buffer, and transferred onto a nitrocellulose membrane (GE Healthcare Life Sciences, 10600008).
We used mouse monoclonal primary antibodies for detecting TetR (Takara, Clone 9G9), and Citrine (Merck, G6539), both diluted 1:2000 in Odyssey Blocking buffer (PBS) (LI-COR Biosciences) + 0.2% Tween 20. The secondary antibody was the near-infrared fluorescent IRDye 800CW Goat anti-Mouse IgG Secondary Antibody from Li-Cor (926-32210), diluted 1:5000 in the same manner. We used Chameleon Duo pre-stained Protein Ladder as our molecular weight marker (928-60000). We used the SNAP i.d. 2.0 system which uses vacuum to drive reagents through the membrane, and the Odyssey CLx (LI-COR) detector for imaging. Images were processed using the Fiji software to obtain black and white images with high contrast (Schindelin et al., 2012).

Growth curves
Cells were precultured in YPD (with aTc in the case of strains where WTC 846 controlled essential genes) to stationary phase, then diluted into fresh media at a concentration of 50.000 cells per mL and induced with the necessary aTc concentrations, except for YP Ethanol and S Ethanol media where the concentration was 500,000 cells per mL. The Growth Profiler 960 (EnzyScreen) with 96well plates and 250 mL volume per well, or Biolector (m2p-labs) with 48 well plates and 1 mL volume per well was used to measure growth curves. These are commercial devices that quantify culture density by detecting the light that is reflected back by the liquid culture.
Arrest and release assay and DNA staining WTC 846 -K3::CDC20 and the appropriate control strains were precultured in YPD (pH 4) with with the indicated aTc concentration to a concentration of 2x10 6 cells/mL, then centrifuged and diluted 1:3 into YPD (pH 4) without aTc. We found that low pH (pH4) of the media was necessary for efficient mother-daughter separation upon completion of cytokinesis, potentially due to the low pH optimum of the chitinase CTS1 (Hurtado-Guerrero and van Aalten, 2007), which plays a role in separation. For the experiment presented in Figure 5E, to prevent the culture from becoming too dense, 25% of the media was filtered and returned to the culture after 4 hr of growth without aTc, which removed 1/4th of the cells. If release was performed, this was done after 8 hr of arrest by adding 600 ng/mL aTc to the culture. Samples were taken at indicated time points before, and every 5 min after aTc was added to the culture, and fixed with 70% ethanol. For the experiment presented in Figure 5E, to aid mother-daughter separation, the samples were sonicated for 1 min in a water bath before fixation.
Samples for DNA staining were digested with 5 mg/mL proteinase K for 50 min at 50˚C, followed by 2 hr of RNase A (Applichem, A2760,0500) treatment at 30˚C. Samples were stained for DNA content using SYTOX Green (Thermofisher, S7020) diluted 1:5000 in PBS, and were sonicated in a water bath for 25 s before flow cytometry. Fluorescence was detected using a 488 nm excitation laser and a 525/15 nm emission filter. The PMT voltage was set up such that the sample with the highest expected ploidy did not saturate the signal.

Shutoff assay
Cells were grown to early exponential phase(~3 million cells/mL) in YPD at 30˚C with shaking and induced with 600 ng/mL aTc. Two mL samples were taken at indicated time points. To remove excess aTc, cells were spun down for 20 s, supernatant was removed and cells were resuspended in YPD. This process was repeated three times. After the 3rd resuspension, the 2 mL sample was divided between two wells of a 96 deep-well plate. Cycloheximide was added to one of the wells at a final concentration of 70 mg/mL. The plate was continuously shaken at 30˚C. Citrine fluorescence was measured every 30 min using flow cytometry as explained above.
Flow cytometry data was not gated except when necessary to remove debris. For aTc dose response experiments, median fluorescence of the entire population was used to fit a five-parameter dose response curve with the drm() command and the fplogistic formula c þ ðdÀcÞ 1þexpðbðlogðxþ1ÞÞ p 1 þexpðlogðxþ1ÞÞ p 2 Þ from the drc package. Parameters p_1 and p_2 were fixed individually for each curve, the rest of the parameters were estimated by the drm command. Parameter values can be found in Supplementary file 1 - Table S5. The cytometry cell volume proxy was always calculated as the magnitude of the vector of the FSC-W and SSC-H signals ( ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðFSC À WÞ 2 þ ðSSC À HÞ 2 q ), since forward and side scatter signals provide information about cell volume and budding state. The forward scatter width and side scatter height were chosen because this combination (as opposed to other combinations involving FSC-H/SSC-W or area of the signals) showed the most separation between measured signal peaks corresponding to spherical calibration beads of known diameter. For single-reporter quantification of VIV, we calculated the residual standard deviation (RSD) of a linear model describing the relationship between the cytometry cell volume proxy and fluorescence of the population. To do this, the rlm() command from the MASS package was used to generate the linear model, and the residual standard deviation given by the same rlm() command was used as our measure of VIV. See Appendix 2 for a detailed explanation of the method.
Where shown, error bars for median fluorescence and the RSD were calculated using bootstrapping. The original set of data points was sampled with replacement and median fluorescence or RSD was calculated. 95% confidence intervals were calculated based on 1000 repetitions of this sampling process and plotted as error bars.
To generate a linear model describing the relationship between the volume proxy measurements done by flow cytometry and volume measurements by Coulter Counter, first the two data sets were sampled with replacement 5000 times. Then these samples were ordered by increasing volume proxy or volume and merged. The lm() command in the R package stats was used to fit the linear model. Sub-populations from Gates 4, 6, and 9 were not included in the fitting, as these medians were deemed suboptimal representations of the bimodal distributions. The resulting linear fit had a slope of 471 and an intercept of 62032.

Additional files
Supplementary files . Supplementary file 1. Table S1 Strains used in this study. Kozak sequence is the last 15 bp before the start codon. Table S2 Plasmids used in this study. * indicates plasmids available through Addgene. These plasmids are sufficient to allow construction of WTC 846 strains carrying the cAR architecture without any further construction. They can also be modified to construct strains with genes controlled by the Simple Repression (SR) and Autorepression (AR) architectures presented in this manuscript. Construction of SR strains would require deletion of the negative feedback-controlled TetR from (P2365/2370/2371/2372/2374), and construction of AR strains would require deletion of the constitutively expressed TetR-Tup1 from the same plasmids. Table S3 Oligos used in this study to create strains where WTC 846 controls endogenous gene expression. See Appendix 5 for the protocol used for endogenous gene promoter replacement through homology directed repair. These oligos were used in conjunction with P2375 (NatMX) or P2350 (HygMX) to create the linear PCR fragment necessary for promoter replacement, or as colony PCR oligos to confirm correct promoter replacement. Table S4 Sequences used in this study. (*) indicates a shortened t_CYC1 used to avoid homology in plasmids where there are more than one t_CYC1 sequences. (**) indicates the linker sequence used between TetR-nls and the fusion partners MBP and Tup1. (***) indicates the linker sequence used between TetR-nls and the fusion partner GST. Table S5 Parameters used to fit 5parameter sigmoid curves to experimental data. See Materials and methods for the 5-parameter log logistic forumula.
. Transparent reporting form

Data availability
All relevant sequences are included in the supporting files for reproducibility. All raw flow cytometry data is publicly available at https://doi.org/10.3929/ethz-b-000488967. All other source data is included in the manuscript and supporting files.
The following dataset was generated: Appendix 1-figure 1. Optimization of tetO 1 placements and endogenous transcription factor binding sites to increase maximum activity of a TetR-repressible derivative of P TDH3 . (A) Diagram of P TDH3 shows the nucleotide positions of the binding sites for the endogenous transcription factors Rap1 and Gcr1, the TATA-sequence, and the transcription start site relative to the TDH3 start codon. Gcr1 binding sites were found in reference Yagi et al., 1994 and confirmed in reference Kuroda et al., 1994. (B) Repression and maximum activity of the P TDH3 derivatives tested for optimization. Diagrams above the plots display the genetic elements of strains used (Y2565, 2575,2598,2647,2599,2648,2601,2649,2602,2650,2603,2656,70,2683). Left diagram depicts strains used to test repressed activity, right diagram maximum activity. Px denotes any tetR repressible promoter. The * in TetR indicates a SV40 Nuclear Localization Sequence. In all strains, the P TDH3 derivative promoters diagrammed on the left directed the synthesis of Citrine integrated into the LEU2 locus. Grey boxes inside the diagrams denote tetO 1 TetR-binding sites. For measurement of repressed activity, otherwise-isogenic strains carried a P ACT1 -TetR construct integrated in the HIS3 locus. Citrine fluorescent signal was detected by flow cytometry. For the measurements, 'fold difference' measures the median of the maximum activity signal divided by the median of the repressed activity. 'Fold over autofluorescence' refers to median repressed activity signal divided by the median autofluorescence background signal. Maximum promoter activity is quantified as median fluorescence signal expressed as percentage of signal from otherwise-isogenic P TDH3 -Citrine strain. For the plots, x axis shows intensity of fluorescence signal. Plots are density distributions of the whole population, such that the area under the curve equals one and the y axis indicates the proportion of cells at each fluorescence value. The circles inside each density plot show the median and the upper and lower bounds of the bar show the first and third quartiles of the distribution. (C) Repression and maximum activity of optimized P 5tet derivatives. Diagrams and plots as in (B). These promoter variants contained additional binding sites for Rap1 and Gcr1 selected for higher activity, as well as an alternative TATA sequence as described.
The online version of this article includes the following source data is available for figure 1: Appendix 1-figure 1-source data 1. Numerical data for Appendix 1-figure 1.
Single reporter quantification of CCV in fluorescent protein expression by flow cytometry using the Volume-independent variation measure In yeast and C. elegans, comparison of signals from strains with different combinations of different reporter genes allows the different contributions to variation to be independently quantified (Colman-Lerner et al., 2005;Mendenhall et al., 2015). One of these contributions, individual differences in general ability to express genes into proteins, contributes to phenotypic variation in genetic penetrance and expressivity (Burnaevskiy et al., 2019). Quantification of this and other sources of variation benefits from ability to measure output of single cells over time (Colman-Lerner et al., 2005) and, in flow cytometry, requires measurement of outputs of multiple reporters (Pesce et al., 2018). Here, however, we were interested in the overall variability rather than specific sources of variability. Therefore, we only had a single reporter protein (Citrine). However single reporter studies have a major, confounding contribution to measured variation in gene expression that multi-reporter studies don't: Fluorescent proteins in yeast are degraded very slowly unless they have degradation tags attached (Gordon et al., 2007) and therefore, if constitutively expressed, their abundance increases over time (Cookson et al., 2010). Thus, in cycling populations of budding yeast that continually express fluorescent proteins, a major source of cell-to-cell variation in fluorescent signal is that small, new-born cells have not had time to accumulate much fluorescent protein, while larger cells have. This source of variability normally affects all reporter proteins in the cell in a similar fashion, and therefore does not require correcting in multi-reporter studies. On the other hand in single reporter studies with flow cytometry in yeast, as for higher cells, this volume related variation in fluorescent protein expression is generally corrected for by gating; that is filtering the data to select only a narrow subset of cells with similar forward and side scatter, and thus volume, which increases with cell cycle progression. Such gating disregards data from the majority of the cells whose values fall outside the gated range. Here, in order to avoid discarding data, we established a singlereporter measure of cell-to-cell variation that corrects for variation due to fluorescent protein accumulation without gating. We first established that forward and side scatter signals can be used to distinguish smaller cells from larger ones. We sorted cells on a BD FACS Aria III flow cytometer. We set different gates on the FSC and SSC signals (shown in Appendix 2-figure 1) to collect 10 sorted sub-populations, each containing about 100,000 exponentially growing Y2683 cells. We then immediately measured (a) FSC and SSC from the collected subpopulations on a different instrument, the LSRII Fortessa LSR used for the flow cytometric measurements in this work, and (b) volume in fL with a Coulter Counter (Appendix 2-figure 2). The raw data acquired by the two methods can be seen in Appendix 2- figure 2A,B. For the flow cytometry data, we used the width of the FSC and height of the SSC to calculate a volume proxy using the formula ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðFSC À WÞ 2 þ ðSSC À HÞ 2 q as explained in Materials and methods. Figure 2C shows a linear relationship between the medians of the sub-populations as measured by the two methods, that is, that the flow cytometric measurement is a proxy for volume, and that two volume measurements qualitatively agree. The three sub-populations (4, 6, and 9) where a slight deviation from the linear relationship is observed are all bimodally distributed, meaning the population is a mixture of large and small cells and the median is not a good representation of this sub-population. Overall, for the cell-to-cell variation calculations outlined below, this relative relationship is enough to distinguish new-born, smaller cells from larger cells that have had time to accumulate fluorescent protein.
Appendix 2-figure 1. Sub-populations collected by FACS for validating the cell volume proxy measure. Strain Y2683 was grown to exponential phase in YPD and was run through the sorter at a concentration of 2 million cells per mL. 10 separate gates were set on the FSC-W and SSC-H signals for collecting sub-populations as depicted in the figure. Fluorescence x10 3 (a.u.) Appendix 3-figure 1. Effect of increased nuclear concentration and size of TetR on repression. The top diagram indicates the genetic elements of the SR architecture used to test the ability of various TetR derivatives to abolish basal activity of P 7tet.1 . Diagrams to the left of the plots show the different repressors used. Each * indicates one SV40 Nuclear Localization Sequence. GST refers to Glutathione S-transferase, and MBP to Maltose Binding Protein, both of E. coli. Citrine fluorescence from P 7tet.1 repressed by the repressors indicated was measured using flow cytometry (Y2657,2681,2664,2665,2666,2667,70). Plots are density distributions of the whole population, such that the area under the curve equals one and the y axis indicates the proportion of cells at each fluorescence value. The circles inside each density plot show the median and the upper and lower bounds of the bar correspond to the first and third quartiles of the distribution. Numbers to the left of the plot indicate fold expression over autofluorescence, that is, the median of the Citrine fluorescence detected divided by the median of the autofluorescence signal. Although increased Appendix 3-figure 1 continued on next page Time to steady state depends on the stability of the controlled protein In the experiments in Figure 4, WTC 846 -controlled Citrine takes around 7 hr to reach steady state concentration. Here we present a simple ODE model to demonstrate that the time to steady state will change based on the stability of the controlled protein. In the model (Equation 1) the protein of interest is produced at a constant rate a and lost with a lumped linear rate (dilution + degradation) d. Analytical solution (Equation 2) of this model shows that the only constant that affects the time variable is the degradation + dilution rate. Dependence on this variable is also evident in simulations based on this model (Appendix 4-figure 1). The smaller d is, (i.e. the more stable the protein is), the longer it takes to reach steady state. On the other hand, changes in the production rate a have no effect on time to steady state, although both rates affect the maximum level. Citrine is a remarkably stable protein (see Figure 4-figure supplement 8), but recent data suggests that most (somewhere between 50-85% depending on the data set) of the yeast proteome is just as stable Wiechecki et al., 2018. Therefore most other WTC 846 -controlled endogenous proteins will likely have a time to steady state around 7 hr, except those with a shorter half-life which will exhibit a shorter time to steady state.

A B
Appendix 4-figure 1. Simulations showing the relationship between production and degradation rates and time to steady state. The ODE model presented in Appendix 4 was simulated with (A) constant production rate a = 0.2, and a varying degradation rate d, or (B) with constant degradation rate d = 0.005 and a varying production rate. Both parameters affect the maximum level of Protein produced, but only the degradation rate determines the time required to reach the steady state expression. The lower the degradation rate (i.e. the more stable the protein is), the longer it takes to reach steady state.

Protocol for WTC 846 strain generation
The WTC 846 is a two unit transcriptional control system for S. cerevisiae (Appendix 5- figure 1A). An inducible promoter (P 7tet.1 ) is placed in front of the Gene of Interest (GOI). The promoter is based on an engineered version of the strong constitutive promoter of TDH3. It was made repressible by placing TetR-binding sites next to the binding sites for the transcriptional machinery. As a result, binding of the TetR protein can prevent binding of the endogenous proteins which normally drive transcription. The repressors TetR and TetR-Tup1 are found on one integrative repressor plasmid (Appendix 5- figure 2A). TetR is expressed under the above described promoter (P 7tet.1 ) creating an autorepression loop. TetR-nls-Tup1 abolishes the basal activity of P 7tet.1 and is expressed under the control of the weak, constitutive RNR2 promoter.
To create a functional system, we advise to first integrate the repressor plasmid. The P 7tet.1 can then be placed in front of any gene in the genome using PCR tagging Janke et al., 2004. The tagging plasmid (based on Janke et al., 2004) is used as a template (Appendix 5- figure 3). We provide two versions of the P 7tet.1 followed by a flag tag followed by a linker composed of eight glycine residues; either cloned in a plasmid providing a HygR marker (P2350), or a NAT marker (P2375). For PCR-based tagging, the 5' and 3' ends of the PCR fragment need to be complementary to a sequence upstream of the GOI and to the beginning of the GOI, respectively. This is ensured by using primers with tails complementary to these regions. We tested the plasmid for use with and without the flag tag. Circles represent the median of the fluorescence signal, lines were fitted. The dashed line indicates autofluorescence control, that is, the parent strain without any Citrine integrated.