Abstract

   Synthetic biology aims to design and construct bacterial genomes
   harboring the minimum number of genes required for self-replicable
   life. However, the genome-reduced bacteria often show impaired growth
   under laboratory conditions that cannot be understood based on the
   removed genes. The unexpected phenotypes highlight our limited
   understanding of bacterial genomes. Here, we deploy adaptive laboratory
   evolution (ALE) to re-optimize growth performance of a genome-reduced
   strain. The basis for suboptimal growth is the imbalanced metabolism
   that is rewired during ALE. The metabolic rewiring is globally
   orchestrated by mutations in rpoD altering promoter binding of RNA
   polymerase. Lastly, the evolved strain has no translational buffering
   capacity, enabling effective translation of abundant mRNAs. Multi-omic
   analysis of the evolved strain reveals transcriptome- and
   translatome-wide remodeling that orchestrate metabolism and growth.
   These results reveal that failure of prediction may not be associated
   with understanding individual genes, but rather from insufficient
   understanding of the strain’s systems biology.
     __________________________________________________________________

   Genome-reduced bacteria often show impaired growth under laboratory
   conditions. Here the authors use adaptive laboratory evolution to
   optimise growth performance and show transcriptome and translatome-wide
   remodeling of the organism.

Introduction

   Minimal genomes, containing only the necessary genes to maintain
   self-replicable life, have been constructed^[40]1–[41]3. For example, a
   native 1.08-Mbp Mycoplasma mycoides genome and its redesigned version
   (JCVI-syn3.0) was generated by de novo genome synthesis. Both genomes
   created viable organisms through genome transplantation. Specifically,
   the genome of JCVI-syn3.0 was designed based upon essential genes
   identified using transposon mutagenesis of M. mycoides^[42]3. However,
   while these fascinating minimal genomes often show unpredictable
   phenotypes such as growth retardation, a set of genes necessary for
   survival remains intact. The unexpected phenotypes highlight our
   limited understanding of bacterial genomes. For instance, all essential
   genes of M. mycoides were contained in the initial design; however, a
   viable genome could only be constructed after quasi-essential genes,
   which are not strictly essential but were required for robust growth,
   were included in the minimal genome. In contrast to this bottom-up
   approach to genome design, several Escherichia coli strains harboring
   reduced genomes have been constructed by sequential genome reduction
   mostly without growth retardation in rich
   media^[43]1,[44]2,[45]4–[46]7. However, when genome-reduced strains are
   grown in minimal medium, their growth rate is often reduced. The
   decreased growth rate has been attributed to our limited understanding
   of some bacterial genome processes, such as synthetic lethality and
   interactions between interconnected cellular components, making it
   difficult to construct minimal genomes with a top-down approach.

   To compensate for incomplete knowledge of bacterial genomes, we
   implement adaptive laboratory evolution (ALE) to allow
   self-optimization of the unknown processes encoded on a genome. It has
   been widely reported that ALE rapidly generates desired phenotypes such
   as tolerance against stresses^[47]8,[48]9, fast growth rates under
   given media^[49]10, and utilization of non-natural substrates^[50]11.
   Those phenotypes are acquired by a number of intriguing mechanisms
   during adaptation such as mutations on metabolic enzymes^[51]12,
   rewired serendipitous pathways^[52]11, and transcriptomic
   re-organization^[53]13,[54]14. Mutations on metabolic enzymes provide
   different substrate specificity and kinetic properties. As a global
   response, transcription machinery is often mutated, which have been
   reported to remodel cells’ catabolic efficiency^[55]15,[56]16.
   Moreover, ALE provides valuable insights into the genotype–phenotype
   relationship by investigating a time series of genomic changes. Thus,
   we exploit this robust method to recover the innate potential for rapid
   growth on a given medium and report a growth-recovered genome
   containing a reduced number of genes enabling rapid growth. Here, we
   apply ALE to a genome-reduced strain, named MS56, derived from the
   standard E. coli K-12 MG1655 strain, which yields growth retardation in
   minimal medium. We generated the evolved strain, named eMS57, which
   exhibits a growth rate comparable to MG1655. This is followed by
   multiple omics measurements revealing that remodeling of the
   transcriptome and translatome in eMS57 results in metabolic
   re-optimization and growth recovery. This comprehensive data provides
   valuable insights for cellular design principles for synthetic biology.

Results

ALE of a genome-reduced E. coli

   The E. coli MS56 was used as a starting strain for ALE^[57]4. MS56 was
   created from the systematic deletion of 55 genomic regions of the
   wild-type E. coli MG1655. The 55 regions had a combined length of
   approximately 1.1Mbp. No essential genes or genes expressed at a
   significant level were removed from MG1655. Although MS56 exhibited a
   comparable growth rate to E. coli MG1655 in rich medium (Fig. [58]1a),
   it showed severe growth reduction in M9 minimal medium (Fig. [59]1b).
   To reveal a molecular basis for the growth reduction, we determined
   whether MS56 could adaptively evolve to recover the growth rate of its
   MG1655 parental strain. Due to the low growth rate of MS56 in M9
   minimal medium, we initially supplemented 0.1% of LB medium (v v^−1),
   which restored the growth rate of MS56 to 2/3 that of MG1655
   (Supplementary Figure [60]1). Then, LB supplementation was reduced in a
   stepwise fashion to reach supplement-free growth (Fig. [61]1c). After
   807 generations of ALE, we isolated a clone from the evolved
   population, eMS57, which restored final cell density and growth rate to
   levels comparable to MG1655 in M9 minimal medium without any nutrient
   supplementation (Fig. [62]1d).

Fig. 1.

   [63]Fig. 1
   [64]Open in a new tab

   Adaptive laboratory evolution (ALE) of a genome-reduced strain (MS56).
   a Growth profiles of genome-reduced strain MS56 (red line) and
   wild-type E. coli MG1655 (black line) in LB medium. μ indicates
   specific growth rate. Error bars indicate standard deviation (s.d.) of
   two biological replicates. b Growth profiles of genome-reduced strain
   MS56 (red line) and wild-type E. coli MG1655 (black line) in M9 minimal
   medium. Error bars indicate s.d. of three biological replicates. c Cell
   growth trajectory showing changes in fitness during the ALE of MS56 in
   M9 minimal medium with supplementation of LB medium. Cell density was
   measured after 12 h of three individual batch cultivation (circles) and
   error bars indicate the s.d. of three individual cultures. LB
   supplementation was stepwise reduced from 0.1 to 0% over time (orange
   line). At the end of the ALE experiment, the evolved population
   exhibited restored growth rate in M9 minimal medium without any
   nutrient supplementation. d Growth profiles of a clone eMS57 (red line)
   isolated from the ALE population and wild-type E. coli MG1655 (black
   line) in M9 minimal medium. Error bars indicate s.d. of three
   biological replicates. e Morphological changes between MG1655, MS56,
   and eMS57. Upper panel, TEM images. Lower panel, SEM images. f
   Phenotype microarray characterization of MG1655 and eMS57 showing
   different nutrient utilization capability. Various carbon sources (red
   circle), phosphorus sources (black circles), nitrogen sources (yellow
   circles), and sulfur sources (blue circles). g Intracellular and
   extracellular pyruvate concentrations for MG1655 and eMG57 at 4, 6, and
   8 h after inoculation. Int: intracellular pyruvate concentration. Ext:
   pyruvate concentration in the medium. Black (MG1655) and red (eMS57)
   lines show cell density at 4, 6, and 8 h after inoculation.
   Intracellular pyruvate level is presented as mole pyruvate per 10^9
   cells and extracellular pyruvate level was measured in molar
   concentration. Individual data points are shown as blue circles and
   error bars indicate the s.d. of two biological replicates. h Pyruvate
   uptake function in MG1655 and eMS57 was examined by growth inhibition
   induced by a toxic pyruvate analog (3-fluoropyruvate, FP). Error bars
   indicate s.d. of three biological replicates. Source data are provided
   as a Source Data file

   We then sought to evaluate the phenotypic differences between eMS57,
   MS56, and MG1655. In terms of morphology, eMS57 was similar in cell
   size and length to MG1655, whereas the unevolved MS56 had a longer cell
   length than the other two strains (Fig. [65]1e). Similar morphological
   changes were observed in a previously constructed genome-reduced strain
   Δ16 that exhibited severe growth retardation^[66]7. eMS57 showed a
   narrower spectrum of nutrient utilization for carbon, nitrogen,
   phosphorus, and sulfur sources than MG1655 due to the deletion of genes
   responsible for respective nutrient utilization (Fig. [67]1f,
   Supplementary Figure [68]2, and Supplementary Data [69]1). For example,
   eMS57 did not show respiration capability on glycolate and glyoxylate
   as the sole carbon source because the genes responsible for glycolate
   utilization were removed by MD10 deletion^[70]4. There was no
   significant change in phosphorus and sulfur source utilization;
   however, MG1655 and eMS57 exhibited different nitrogen utilization
   preferences. The respiration rate of MG1655 in cytidine was much higher