Antibiotic Resistance: Finding the right sequence of drugs

Rapidly switching between similar antibiotics may help to slow down the evolution of resistance.
  1. Anh Huynh
  2. Kevin B Wood  Is a corresponding author
  1. Department of Biophysics, University of Michigan, United States
  2. Department of Physics, University of Michigan, United States

In order to survive, many living organisms need to be able to adapt to their ever-changing environment. These past experiences shape the behavior of creatures big and small – from mutations in single genes to neurological changes that underlie memory formation in primates.

There is substantial evidence that environmental history affects how bacteria respond to antibiotics (Barbosa et al., 2019; Card et al., 2019; Nichol et al., 2019; Santos-Lopez et al., 2019; Yen and Papin, 2017). This has led researchers to suggest that switching between drugs over time could help slow down antibiotic resistance, as this will force the bacteria into a scenario where their existing solutions (resistance to the current drug) cannot protect them from tomorrow’s problem (a new drug). However, this method has led to mixed results (Abel zur Wiesch et al., 2014; Imamovic et al., 2018), and optimizing this approach is challenging, in part because it is unclear which features of the antibiotic sequence are the most important and guarantee the best results.

Now, in eLife, Hinrich Schulenburg (University of Kiel and Max Planck Institute for Evolutionary Biology) and colleagues – including Aditi Batra and Roderich Roemhild as joint first authors – report the results of experiments on the multi-drug resistant bacteria Pseudomonas aeruginosa (Batra et al., 2021). The team (who are based in Austria and Germany) exposed the bacteria to various sequences of three antibiotics that belong to commonly used classes of drugs: one class targets the ribosome, one targets DNA gyrase, and one targets the cell wall. In some cases, three drugs from different classes were used (that is, a heterogeneous sequence), and in some cases all three drugs belonged to the same class (a homogenous sequence). Batra et al. also varied the temporal properties of each sequence by switching between the drugs rapidly, slowly, or in a random order. The growth rate, phenotypic resistance levels and population genetics of the evolved populations were then analyzed to determine which sequences of drugs were the most effective at eliminating the bacteria (Figure 1A).

Experimental and biological features of effective drug sequences.

(A) Batra et al. applied different sequences of antibiotics to 756 populations of P. aeruginosa (top panel). The bacteria were treated with either a single drug (monotherapy; row 1), or three antibiotics which were switched rapidly (row 2), slowly (row 3) or in a random order (row 4): the three drugs were either from the same class (homogeneous) or from different classes (heterogeneous). This experiment revealed that fast (blue line) and random (green line) switching between three homogeneous beta-lactam drugs reduced bacteria growth and resulted in higher levels of extinction (bottom graph). (B) The effects of the different sequences are also impacted by biological features. (Top panel) When sensitive bacteria (shown in purple) are treated with the first drug, some cells will evolve genetic changes that make them resistant to the antibiotic treatment (shown in green). These evolutionary changes can lead to collateral effects that make the bacteria less (top arrow), equally (middle arrow) or more (bottom arrow) resistant to the second drug. (Bottom panel) Treatment with the first drug may also lead to negative hysteresis, when short-term physiological changes enhance the bacteria’s response to the second drug (right), leading to more cell death in the population compared to bacteria not pre-treated with the first drug (left).

Image credit: Anh Huynh.

In addition to these experimental parameters, the impact of different antibiotic sequences could also depend on how the population biologically responds to consecutive drug exposures (Figure 1B). For example, the genetic changes bacteria evolve in response to one antibiotic can lead to collateral effects that increase the population’s resistance or sensitivity to another drug. Because collateral resistance occurs more frequently between drugs of the same class, heterogeneous sequences of antibiotics are thought to be more effective at eliminating bacteria (Imamovic and Sommer, 2013; Lázár et al., 2013; Maltas and Wood, 2019; Pál et al., 2015; Lázár et al., 2014). Treatment with unrelated drugs has also been shown to favor negative hysteresis, which is when short-term physiological changes induced by one antibiotic enhance susceptibility towards another. Indeed, a recent study found that rapid switching between antibiotics from different classes promoted extinction of bacterial populations, even when the drugs were used at sub-inhibitory levels (Roemhild et al., 2018).

However, Batra et al. found that homogenous sequences of beta-lactams (a class of antibiotics that target the cell wall) were surprisingly more effective at clearing bacteria. The experiments also revealed that extinction tended to occur early in the treatment and was less effective when drugs were switched more slowly. A particular heterogeneous set of drugs also tended to not eliminate bacteria, indicating that heterogeneity, alone, does not guarantee success.

So, what are the important characteristics of a ‘good’ antibiotic sequence? To answer this question, Batra et al. used a common class of statistical models to probe for signatures of successful sequences. They found that extinction was strongly favored by two biological properties — low rates of spontaneous resistance and low levels of collateral resistance — and was also enhanced when switching between drugs was fast or random. This suggests that although thefindings of Batra et al. contradict the proposed benefits of using unrelated drugs, they still validate a portion of the underlying logic: using antibiotics with strong collateral effects and hysteresis enhances the impact of sequence therapy. It just turns out, however, that drugs with these characteristics are not always from different classes.

This study offers insight into how past antibiotic exposure shapes the response of bacterial populations to new challenges. In doing so, it provides a roadmap for future studies investigating how even the simplest organisms harbor signatures of past challenges and potential evolutionary solutions.


Article and author information

Author details

  1. Anh Huynh

    Anh Huynh is in the Department of Biophysics, University of Michigan, Ann Arbor, United States

    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8891-4234
  2. Kevin B Wood

    Kevin B Wood is in the Department of Biophysics and Department of Physics, University of Michigan, Ann Arbor, United States

    For correspondence
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0985-7401

Publication history

  1. Version of Record published: September 9, 2021 (version 1)


© 2021, Huynh and Wood

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.


  • 719
    Page views
  • 59
  • 0

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Anh Huynh
  2. Kevin B Wood
Antibiotic Resistance: Finding the right sequence of drugs
eLife 10:e72562.

Further reading

    1. Evolutionary Biology
    2. Genetics and Genomics
    Henrike Indrischek et al.
    Research Article

    Despite decades of research, knowledge about the genes that are important for development and function of the mammalian eye and are involved in human eye disorders remains incomplete. During mammalian evolution, mammals that naturally exhibit poor vision or regressive eye phenotypes have independently lost many eye-related genes. This provides an opportunity to predict novel eye-related genes based on specific evolutionary gene loss signatures. Building on these observations, we performed a genome-wide screen across 49 mammals for functionally uncharacterized genes that are preferentially lost in species exhibiting lower visual acuity values. The screen uncovered several genes, including SERPINE3, a putative serine proteinase inhibitor. A detailed investigation of 381 additional mammals revealed that SERPINE3 is independently lost in 18 lineages that typically do not primarily rely on vision, predicting a vision-related function for this gene. To test this, we show that SERPINE3 has the highest expression in eyes of zebrafish and mouse. In the zebrafish retina, serpine3 is expressed in Müller glia cells, a cell type essential for survival and maintenance of the retina. A CRISPR-mediated knockout of serpine3 in zebrafish resulted in alterations in eye shape and defects in retinal layering. Furthermore, two human polymorphisms that are in linkage with SERPINE3 are associated with eye-related traits. Together, these results suggest that SERPINE3 has a role in vertebrate eyes. More generally, by integrating comparative genomics with experiments in model organisms, we show that screens for specific phenotype-associated gene signatures can predict functions of uncharacterized genes.

    1. Evolutionary Biology
    2. Microbiology and Infectious Disease
    Isabel Nocedal, Michael T Laub
    Research Article Updated

    Gene duplication is crucial to generating novel signaling pathways during evolution. However, it remains unclear how the redundant proteins produced by gene duplication ultimately acquire new interaction specificities to establish insulated paralogous signaling pathways. Here, we used ancestral sequence reconstruction to resurrect and characterize a bacterial two-component signaling system that duplicated in α-proteobacteria. We determined the interaction specificities of the signaling proteins that existed before and immediately after this duplication event and then identified key mutations responsible for establishing specificity in the two systems. Just three mutations, in only two of the four interacting proteins, were sufficient to establish specificity of the extant systems. Some of these mutations weakened interactions between paralogous systems to limit crosstalk. However, others strengthened interactions within a system, indicating that the ancestral interaction, although functional, had the potential to be strengthened. Our work suggests that protein-protein interactions with such latent potential may be highly amenable to duplication and divergence.