Figure 1.
Illustration of the master strain and library construction procedure.
A master strain was constructed such that it will contain two main constructs in the HIS deletion locus: a constant control construct with mCherry driven by the TEF2 promoter and terminated by a constant ADH1 terminator; and a test construct with a YFP gene driven by Gal1/10 promoter. Following master strain construction, a library of PCR products containing the downstream intergenic regions of 85 tested genes was amplified from the genome by PCR and extended to also contain the URA3 promoter and start codon. This library of DNA sequences was then integrated into the master strain such that only integrations in the exact genomic location would result in an intact selection marker.
Figure 2.
Effect of the 3′ end sequences on YFP accumulation in batch measurements.
(A) YFP measurements of clones with three different 3′ end sequences. Shown are YFP measurements of three different strains, each with a unique 3′ end sequence. Lines of the same color represent measurements of different clones from the same type of 3′ end sequence, demonstrating that the effect of the different constructs on expression is above the variability of our experimental system. The lowest expressing strain (red) contains the COX17 3′ end and serves as a positive control for our experimental system. (B,C,D) Plate fluorometer measurements over time. Following inoculation of the cells in a fresh media containing 2% galactose, optical density (OD), mCherry and YFP are measured over time (B,C and D respectively). Note that as expected, OD and mCherry measurements remain highly similar between different library strains, while YFP expression varies considerably.
Figure 3.
The effect of 3′ end sequences on expression is large and is correlated with endogenous mRNA levels.
(A) Dynamic range of YFP levels of library strains at different galactose induction levels. YFP production per cell per second was measured and calculated in different Galactose concentrations resulting in different promoter activation levels for all library strains at every galactose concentration. Shown are YFP measurements of the 3′ end library strains. Note that the ratio between the highest and lowest strain at the highest induction level (0.1% galactose) shows a fold difference of more than 10-fold. (B) Comparison of the span of expression values between promoter and 3′ end strains for the same group of genes. A box plot is added to show the difference in IQR between the groups. (C) Comparison of YFP levels in the 3′ end library (y-axis) with endogenous mRNA levels measured by RNA-seq (x-axis). The Pearson correlation is given (inset). (D) Same as (C) but for a different strain library in which promoters of the same respective genes are fused to a YFP reporter.
Figure 4.
Higher A/T content upstream of the polyadenylation site is associated with higher YFP expression.
(A) The correlation between A/T content and YFP levels in different window sizes and different locations with respect to the main polyadenylation site. Each point in the matrix represents a different window size (y-axis) centered on a different location (x-axis) with respect to the polyadenylation site. Colors represent the Pearson correlation coefficient (side bar). (B) Same as A using genome wide sequence and mRNA levels [6] data. (C) Shown is the average G/C content of three sets of genes grouped by their mRNA expression levels (0.2 percentile of the lowest and highest expressing genes and the intermediate group contains all the rest) as a function of the distance from mapped transcription end sites, in windows of 20 bp centered around each point.
Figure 5.
YFP expression is correlated with noise strength.
(A) For several different galactose concentrations (represented by different colors), shown is the YFP expression of each 3′ end library strain (x-axis) versus its noise (y-axis, expression variance divided by mean expression squared). Each point represents the noise computed from single cell flow cytometry measurements of the corresponding 3′ end strain. (B) Same as panel (A) only with noise strength (expression variance divided by mean expression) on the y-axis.