Delayed commitment to evolutionary fate in antibiotic resistance fitness landscapes

  • Nature Communications volume 6, Article number: 7385 (2015)
  • doi:10.1038/ncomms8385
  • Download Citation


Predicting evolutionary paths to antibiotic resistance is key for understanding and controlling drug resistance. When considering a single final resistant genotype, epistatic contingencies among mutations restrict evolution to a small number of adaptive paths. Less attention has been given to multi-peak landscapes, and while specific peaks can be favoured, it is unknown whether and how early a commitment to final fate is made. Here we characterize a multi-peaked adaptive landscape for trimethoprim resistance by constructing all combinatorial alleles of seven resistance-conferring mutations in dihydrofolate reductase. We observe that epistatic interactions increase rather than decrease the accessibility of each peak; while they restrict the number of direct paths, they generate more indirect paths, where mutations are adaptively gained and later adaptively lost or changed. This enhanced accessibility allows evolution to proceed through many adaptive steps while delaying commitment to genotypic fate, hindering our ability to predict or control evolutionary outcomes.

Additional access options:

Already a subscriber?  Log in  now or  Register  for online access.


  1. 1.

    , , & Darwinian evolution can follow only very few mutational paths to fitter proteins. Science 312, 111–114 (2006).

  2. 2.

    et al. Stepwise acquisition of pyrimethamine resistance in the malaria parasite. Proc. Natl Acad. Sci. USA 106, 12025–12030 (2009).

  3. 3.

    et al. Compensatory mutations restore fitness during the evolution of dihydrofolate reductase. Mol. Biol. Evol. 27, 2682–2690 (2010).

  4. 4.

    et al. Evolutionary trajectories of beta-lactamase CTX-M-1 cluster enzymes: predicting antibiotic resistance. PLoS Pathog. 6, e1000735 (2010).

  5. 5.

    , & Fitness trade-offs in the evolution of dihydrofolate reductase and drug resistance in Plasmodium falciparum. PLoS ONE 6, e19636 (2011).

  6. 6.

    , , & Hidden randomness between fitness landscapes limits reverse evolution. Phys. Rev. Lett. 106, 198102 (2011).

  7. 7.

    et al. Initial mutations direct alternative pathways of protein evolution. PLoS Genet. 7, e1001321 (2011).

  8. 8.

    et al. Evolutionary paths to antibiotic resistance under dynamically sustained drug selection. Nat. Genet. 44, 101–105 (2012).

  9. 9.

    , , & Empirical fitness landscapes reveal accessible evolutionary paths. Nature 445, 383–386 (2007).

  10. 10.

    , & Revealing evolutionary pathways by fitness landscape reconstruction. Crit. Rev. Biochem. Mol. Biol. 44, 169–174 (2009).

  11. 11.

    Encyclopedia of Life Sciences John Wiley & Sons, Ltd, Chichester (2010).

  12. 12.

    , , , & Multiple peaks and reciprocal sign epistasis in an empirically determined genotype-phenotype landscape. Chaos 20, 026105 (2010).

  13. 13.

    , , , & Patterns of epistasis between beneficial mutations in an antibiotic resistance gene. Mol. Biol. Evol. 30, 1779–1787 (2013).

  14. 14.

    & One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl Acad. Sci. USA 97, 6640–6645 (2000).

  15. 15.

    , & Soluble oligomerization provides a beneficial fitness effect on destabilizing mutations. Proc. Natl Acad. Sci. USA 109, 4857–4862 (2012).

  16. 16.

    , & Perspective: sign epistasis and genetic constraint on evolutionary trajectories. Evolution 59, 1165–1174 (2005).

  17. 17.

    , , & Reciprocal sign epistasis is a necessary condition for multi-peaked fitness landscapes. J. Theor. Biol. 272, 141–144 (2011).

  18. 18.

    & Reciprocal sign epistasis between frequently experimentally evolved adaptive mutations causes a rugged fitness landscape. PLoS Genet. 7, e1002056 (2011).

  19. 19.

    New look at statistical-model identification. IEEE Trans. Automat. Control 19, 716–723 (1974).

  20. 20.

    , & Mutational reversions during adaptive protein evolution. Mol. Biol. Evol. 24, 1608–1610 (2007).

  21. 21.

    , & Long-term dynamics of adaptation in asexual populations. Science 342, 1364–1367 (2013).

Download references


We are grateful to N.D. Lord for gift of strain NDL47, and to D.M. Weinreich, D.L. Hartl and D. Landgraf for helpful discussions. This work was supported by the Novartis Institutes for Biomedical Research, US National Institutes of Health grant R01-GM081617 and the European Research Council FP7 ERC Grant 281891. A.C.P. is a James S. McDonnell Foundation Postdoctoral Fellow.

Author information

Author notes

    • Adam C. Palmer
    •  & Erdal Toprak

    These authors contributed equally to this work.

    • Adam C. Palmer

    Present address: School of Biotechnology and Biomolecular Sciences, University of New South Wales, Sydney, New South Wales 2052, Australia

    • Seungsoo Kim

    Present address: Department of Genome Sciences, University of Washington, Seattle, Washington 98195, USA


  1. Department of Systems Biology, Harvard Medical School, Boston, Massachusetts 02115, USA

    • Adam C. Palmer
    • , Erdal Toprak
    • , Michael Baym
    • , Seungsoo Kim
    • , Adrian Veres
    •  & Roy Kishony
  2. Laboratory of Systems Pharmacology, Harvard Medical School, Boston, Massachusetts 02115 USA

    • Adam C. Palmer
  3. Green Center for Systems Biology, University of Texas Southwestern Medical Center, Dallas, Texas 75390, USA

    • Erdal Toprak
  4. Department of Life Sciences, Ben-Gurion University of the Negev, Beer-Sheva 8410501, Israel

    • Shimon Bershtein
  5. Department of Biology and Department of Computer Science, Technion-Israel Institute of Technology, Haifa 3200003, Israel

    • Roy Kishony


  1. Search for Adam C. Palmer in:

  2. Search for Erdal Toprak in:

  3. Search for Michael Baym in:

  4. Search for Seungsoo Kim in:

  5. Search for Adrian Veres in:

  6. Search for Shimon Bershtein in:

  7. Search for Roy Kishony in:


E.T., A.V. and R.K. conceived the study. E.T., S.K., A.V. and S.B. synthesized the strain collection. A.C.P. and M.B. phenotyped the strain collection. A.C.P. and R.K. analysed the data and wrote the manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Roy Kishony.

Supplementary information

PDF files

  1. 1.

    Supplementary Figures, Notes and References

    Supplementary Figures 1-12, Supplementary Notes 1-2 and Supplementary References

  2. 2.

    Supplementary Data 1

    Optical density measurements over time are presented for every strain at every trimethoprim concentration. Each of three replicates are presented overlaid in red, yellow, blue. Four figures are present for the wildtype strain, as this was measured in twelve replicates. DHFR mutant genotypes are specified by a 6 character code, where '0' indicates a wildtype allele and '1' or '2' indicates a mutant allele. Thus, 000000 refers to the wildtype allele.  Strains bearing single mutations are as follows: 100000 = -35C>T 010000 = P21L 001000 = A26T 000100 = L28R 000010 = W30R 000020 = W30G 000001 = I94L Strains with multiple mutations follow this code, for example 010001 = P21L, I94L 100120 = -35C>T, L28R, W30G The numerical values extracted from these data (relative growth quantified by optical density integrated over time) are presented in Supplementary Data 2.

Zip files

  1. 1.

    Supplementary Data 2

    Table of growth measurements for every strain at every trimethoprim concentration, where relative growth is quantified by the integral of Optical Density from 0 to 30 hours. Each row specifies a single replicate set of growth measurements over a range of trimethoprim concentrations. In each row the strain genotype is listed in full (for example, 'A26' indicates wildtype allele at position 26, 'A26T' indicates a mutation at this position). Rows are grouped by threes, for the three replicate measurements, excepting wildtype which was measured in twelve replicates and is described in twelve rows. In cases where growth is 0 (strain unviable) or never exceeds the IC75 growth threshold, Log10(IC75) is described by the value of '-2'.


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.