INTRODUCCIÓN Y ANTECEDENTES:

The EcoR collection is a set of 72 E. coli wildtype strains isolated from a large variety of mammalian host organisms from different locations 103_{. Our initial assumption was that,} dependent on host and isolation location, the E. coli strains faced different environmental conditions such as available nutrients and should have therefore adapted their metabolic network to the respective specific environments. Such an adaption can lead to a loss of metabolic capabilities and misregulations as a consequence of mutations which however have no impact on the fitness in their natural environment.

We were interested if we could identify metabolic bottlenecks in single reactions that might even impair the general fitness of the cells in conditions that differ from the natural growth conditions. For that, we compared the growth rates of the 72 strains of the EcoR collection in M9 minimal medium with glucose as carbon source and under aerobic conditions and incubated at 37°C. In addition to these strains we also analyzed the growth of 5 commonly used laboratory strains (E.

coli MG1655, W3110, MDS42, BW25113, EMG-2) (Figure 12).

We found that a majority (66 of 77) of all strains had similar growth rates of 0.6 h-1 _{and higher,} whereas a group of 11 strains had lower growth rates. To our surprise, all laboratory strains were found to belong to this group with growth rates between 0.41 ± 0.04 h-1 _{(MG1655) and} 0.47 ± 0.18 h-1 _(MG1655).

Chapter 3 - Understanding the impact of metabolic bottlenecks on the general state of fitness

Figure 12: Growth rates of 72 wildtype isolates and 5 laboratory strains.

Blue bars indicate growth rates of wildtype isolates from the EcoR collection 103_{, red bars of laboratory strains. Shown} are the mean growth rates measured in three independent growth experiments.

To test if the reduced fitness can be linked to metabolic bottlenecks in single reactions we measured the concentrations of 94 metabolites of central carbon metabolism, nucleotide metabolism, amino acid metabolism and other parts of the metabolic network in all 77 strains (Figure 13) using high-throughput metabolomics techniques 104_.

We noticed that all laboratory strains had a similar metabolic profile, which differed from the wildtype isolates. Specifically, we observed a markedly higher abundance of three intermediates of the pyrimidine pathway, N-carbamoyl-L-aspartate, dihydroorotate and orotate (Figure 13b + c). On the other hand, the concentration of UMP, a later intermediate of the same pathway, was relatively low. These results indicate a metabolic bottleneck in a reaction between orotate and UMP and can indeed by explained by a reported frameshift mutation in the gene upstream pyrE,

rph 105_{. This frameshift leads to lower expression rates of pyrE which codes for the orotate} phosphoribosyltransferase, an enzyme that utilizes orotate and PRPP to convert it to the direct precursor of UMP, orotidine 5'-phosphate. Remarkably, the strain MDS42 possesses a minimal genome with over 700 genes deleted 106_{. However, compared to the other laboratory strains, no} difference in growth rate or metabolite pattern could be determined.

Chapter 3 - Understanding the impact of metabolic bottlenecks on the general state of fitness

Figure 13: Concentration of 94 metabolites in 72 E. coli wildtype isolate and 5 laboratory strains and metabolites of the pyrimidine biosynthesis pathway.

(A) Clustered heatmap with all 94 metabolites on the Y-axis and all tested strains on the X-axis. The blue and red bars underneath the strain labels mark laboratory strains (red) and wildtype strains (blue). The colors in the heatmap indicate an up to 10 fold increase (yellow) or decrease (blue) of metabolite concentrations compared to MG1655. On the upper part strains are clustered that share a similar metabolic profile. In red all laboratory strains are marked. On the left side metabolites with a similar appearance in the different strains are clustered together. The three rows at the bottom are the intermediates of the pyrimidine biosynthesis pathway. (B) Overview over the pyrimidine biosynthesis pathway. All intermediates that we were able to measure are marked. (C) Metabolite concentrations of the 4 intermediates normalized to MG1655. Blue bars indicate the concentrations measured in wildtype isolate strains, red bars in the laboratory strains (from left to right: MDS42, W3110, BW25113, EMG-2, MG1655).

In the natural isolate strains, linkages between growth phenotypes and metabolic bottlenecks were less apparent.

In EcoR51, one of the slowest growing strains, we could measure the highest levels of PEP. PEP is substrate and product of several metabolic reactions in E. coli, it is therefore unclear in which reaction is limited as a result of a metabolic bottleneck. However, we speculate that as a result of the general importance of these pathways for the generation of energy and precursors for amino acid biosynthesis, metabolic bottlenecks in central carbon metabolism (dephosphorylation of PEP to pyruvate catalysed by pyruvate kinase), or anaplerosis (carboxylation of PEP to oxaloacetate, catalysed by PEP carboxylase) could not just result in the accumulation of PEP but might also result in reduced growth rates.

For three other slow growing natural isolates - EcoR23, 29 and 52 - we could not identify any specific metabolic bottleneck. However, all of those show a similar metabolic profile that was very distinct to all other strains, suggesting that all these strains possess a similar metabolic network and possibly the same metabolic bottlenecks.

For the slow growing strains EcoR8 and 49 we could also not identify a metabolic bottleneck. Instead, we found that their metabolic profiles were similar to those of the faster growing strains EcoR27 and EcoR42 – both with growth rates above 0.7 h-1_.

It should be noted that only 94 of 1192 metabolites 1 _{and therefore only a tiny fraction of} metabolites could be measured here. Hence, it is possible that in the strains of which we observed reduced growth rates but could not identify a particular metabolic bottleneck like EcoR8 or 49, bottlenecks might exist in reactions of which we can measure neither product nor substrate. In addition, also the contribution of several bottlenecks to the growth reduction is imaginable as well as a limitation of transport capabilities which could limit the growth rate without causing a measurable accumulation of specific metabolites in the cell.

We decided to examine the contribution of the pyrE bottleneck to the reduced growth rates in the laboratory strains in more detail. For that, we analyzed the growth rates of two laboratory strains (MG1655 and W3110) and 3 natural isolates (EcoR18, 42 and 46) when growing in M9 minimal medium in presence and absence of 100 mM uracil (Figure 14). Uracil can be taken up

Chapter 3 - Understanding the impact of metabolic bottlenecks on the general state of fitness

by the cell and transferred to UMP, thereby bypassing the metabolic bottleneck. We found that uracil addition did not affect the growth rates of the natural isolate strains, whereas the growth rates of both laboratory strains increased by 15% (MG1655) and 18% (W3110), respectively. This suggests that the pyrimidine bottleneck indeed reduces the growth rates of the laboratory strains. It should be noted that the growth rates were not restored to the level of the wildtype cells, though. This could be due to insufficient uptake of uracil from the medium as well as a contribution of one or more other metabolic bottlenecks that are limiting the growth rates under the given growth conditions.

Figure 14: Uracil supplementations

Top: Uracil can be imported and converted to UMP, thereby bypassing the PyrE bottleneck. Bottom: Growth rates of 3 wildtype strains and 2 laboratory strains, grown in presence (yellow bars) and absence (blue bars) of 100 mM Uracil, measured in 3 independent cultivations.

Chapter 3 - Understanding the impact of metabolic bottlenecks on the general state of fitness

3.2 Inhibition of pyrE transcription with CRISPR