The Genetic Landscape of a Metabolic Interaction

Enzyme abundance, catalytic activity, and ultimately sequence are all shaped by the need of growing cells to maintain metabolic flux while minimizing accumulation of deleterious intermediates. While much prior work has explored the constraints on protein sequence and evolution induced by physical protein-protein interactions, the sequence-level constraints emerging from non-binding functional interactions in metabolism remain unclear. To quantify how variation in the activity of one enzyme constrains the biochemical parameters and sequence of another, we focused on dihydrofolate reductase (DHFR) and thymidylate synthase (TYMS), a pair of enzymes catalyzing consecutive reactions in folate metabolism. We used deep mutational scanning to quantify the growth rate effect of 2,696 DHFR single mutations in 3 TYMS backgrounds under conditions selected to emphasize biochemical epistasis. Our data are well-described by a relatively simple enzyme velocity to growth rate model that quantifies how metabolic context tunes enzyme mutational tolerance. Together our results reveal the structural distribution of epistasis in a metabolic enzyme and establish a foundation for the design of multi-enzyme systems.


I. Supplemental Figures
Figure 1: Growth rates, epistasis, and analysis of model significance for a focused set of mutants.Table 1: Steady-state DHFR kinetics parameters.

Figure 3 :
Figure 3: Reproducibility of relative growth rate measurements.

1 :
Growth rates, epistasis, and analysis of model significance for a focused set of mutants.a. Heatmap of experimentally measured growth rates for seven DHFR point mutants in four TYMS backgrounds.White indicates WT-like growth, blue indicates deleterious growth rates.b.Heatmap of computationally predicted growth rates using the best fit model for the data in (a).c.Heatmap of experimentally measured epistasis for the same set of DHFR and TYMS mutant combinations.Values were calculated from the data in (a).White indicates no epistasis, green indicates negative (amplifying) epistasis, and pink indicates positive (buffering) epistasis.d.Heatmap of computationally predicted epistasis using the growth rates determined in (b).e. R-squared values describing the goodness of fit for 50 models trained using randomly shuffled kinetics parameters (shuffling kcat and Km across all DHFR and TYMS variants).The Rsquared value for the true (non-shuffled) data is indicated with a dashed blue line.f.RMSD values describing the deviation between the model and data for 50 models trained using randomly shuffled kinetics parameters (as in e).The RMSD value for the true (nonshuffled) data is indicated with a dashed blue line.g.Example model predictions for one particular instance of shuffling DHFR kinetic parameters while retaining the experimentally measured TYMS kinetics parameters.Each point describes a DHFR/TYMS mutant combination, points are color-coded by TYMS background.Error bars in the x axis represent SEM across 3 replicate measurements, error bars on the y axis were obtained by jackknife resampling and refitting the data.The dotted black line marks y=x. h.Example model predictions for one particular instance of shuffling TYMS kinetic parameters while retaining the experimentally measured DHFR kinetics parameters.Error bars in the x axis represent SEM across 3 replicate measurements, error bars on the y axis were obtained by jackknife resampling and refitting the data.The dotted black line marks y=x.Supplemental Figure 2: Mutational scanning library completeness.a. Heatmaps of log10 sequencing counts (reads) for all single mutations of DHFR in each TYMS background at the start of the experiment (t=0).Representative data are shown for one of three replicate measurements.DHFR positions are indicated along the vertical axis, while amino acids (sorted by physiochemical similarity) are along the horizontal axis.White boxes indicate missing mutations (no counts) and small dots indicate the WT residue identity.b.Distribution of sequencing counts (reads) for all single mutations of DHFR in the WT TYMS background.Median and mean reads are indicated in red and black respectively.Representative data are shown for one of three replicate measurements (same replicate as in A). c. Distribution of sequencing counts (reads) for all single mutations of DHFR in the Q33S TYMS background.Median and mean reads are indicated in red and black respectively.Representative data are shown for one of three replicate measurements (same replicate as in A). d.Distribution of sequencing counts (reads) for all single mutations of DHFR in the R166Q TYMS background.Median and mean reads are indicated in red and black respectively.Representative data are shown for one of three replicate measurements (same replicate as in A).Supplemental Figure 3: Reproducibility of relative growth rate measurements.a-c) Correlation in relative growth rate measurements for individual DHFR mutations across three experimental replicates in the WT TYMS background.The cyan dashed line is the line of best fit; black dashed line marks x=y.The data were normalized such that WT growth is one.d-f) Same as A-C, but for the R166Q TYMS background g-i) Same as A-C, but for the Q33S TYMS background Supplemental Figure 4: Epistasis of DHFR mutations to TYMS background a.A heatmap of epistasis for all DHFR positions to TYMS Q33S.Values displayed represent the average across three replicates.Grey pixels indicate mutations that did not have three replicates in both the WT and Q33S TYMS backgrounds.Amino acid mutations are arranged by physiochemical similariy (along the rows), positions are indicated along the columns.b.A heatmap of epistasis for all DHFR positions to TYMS R166Q.Grey pixels indicate mutations that did not have three replicates in both the WT and R166Q TYMS backgrounds.Amino acid mutations are arranged by physiochemical similariy (along the rows), positions are indicated along the columns.

Table 4 :
Relative Growth Rates (and error) for all DHFR mutations in each TYMS background.(provided as a separate excel file due to size)

Table 5 :
Epistasis (and p-values)for all DHFR mutations in the TYMS Q33S and R166Q backgrounds.(provided as a separate excel file due to size)

Table 6 :
Epistatic residue groups determined by k-means clustering.

Table 7 :
Statistical association between evolutionary conservation and DHFR positions associated to catalysis

Table 1 :
Steady-state DHFR kinetics parameters.Mean and standard deviation are reported across N=3 replicates when available.

Table 2 :
Steady-state TYMS kinetics parameters.Mean and standard deviation are reported across N=3 replicates when available.The signal-to-noise ratio for spectrophotometric measurements of R166Q TYMS activity is too low (given the slow rate of reaction) to allow accurate steady-state kinetics measurements.In this case, we assigned an arbitrarily slow kcat and a higher Km.We observed that the model fit quality did not change appreciably for other values of R166Q provided they were substantially slower than WT.