Laboratory evolution of S. aureus under treatment with drugs.
Adaptive Laboratory Evolution (ALE) has emerged as a prominent technique in generating drug-resistant strains, complementing clinical isolation. We selected 15 drugs, each with distinct mechanisms of action, to emulate the screening pressure (Table 1). We leveraged S. aureus ATCC25923, a wildtype strain, to conduct 10 rounds of evolution experiments (Fig. 1a). Ultimately, S. aureus exhibited significant resistance to 9 out of 15 drugs (Fig. 1b). Drug resistance manifested at varying rates in individual cases. Notably, Ampicillin (AMP), Gentamicin (GM), and Kanamycin (KM) exhibited a progressive shrinking of their bacteriostatic zones, indicating a slower rate of resistance development. Conversely, strains treated with Streptomycin (SM), Rifampicin (RIF), Fusidic Acid (FD), and Novobiocin (NV) showed pronounced resistance by the second generation. It's of significance to note that in the RIF treatment group, the filter paper's outline was barely discernible starting from the fourth evolution round. Interestingly, Cefalexin (CEX), FD, and NV, which initially had extensive bacteriostatic zones, triggered a swift development of complete bacterial resistance. Surprisingly, resistance levels to CEX, SM, and NV fluctuated periodically.
Table 1
List of all drugs used in this paper
Drug | Abbreviation | MIC of ATCC25923 (µg/ml) | Acting site |
ampicillin | AMP | 5.12 | Penicillin-binding protein |
cefalexin | CEX | 10.24 | Penicillin-binding protein |
vancomycin | VA | 0.64 | Peptidoglycan |
gentamicin | GM | 5.12 | 30S ribosomal subunit |
streptomycin | SM | 10.24 | 30S ribosomal subunit |
tetracycline | TE | 0.64 | 30S ribosomal subunit |
minocycline | MIN | 1.28 | 30S ribosomal subunit |
erythromycin | EM | 0.64 | 50S ribosomal subunit |
kanamycin | KM | 10.24 | 50S ribosomal subunit |
linezolid | LZD | 2.56 | 50S ribosomal subunit |
chloramphenicol | CAP | 5.12 | 50S ribosomal subunit |
rifampicin | RIF | 0.005 | DNA-dependent RNA polymerase |
fusidic acid | FD | 0.16 | Elongation factor G |
novobiocin | NV | 0.08 | DNA gyrase subunit B |
clofazimine | CFZ | 2.56 | Cell membrane and ROS generation |
Identification of key genes associated with drug resistance.
As delineated in the Methods section, we initially sequenced the complete genome of the ATCC 25923 parental strain, then sequenced the genomes of a total of 108 mutant strains. The raw data are available on NCBI under the access number PRJNA911958. By comparing the whole-genome sequencing results of the evolved and parental strains, a wealth of information was gathered regarding the mutation sites of the bacteria (see Table 2 and Supplementary Table 1). Mutants that occurred only once were eliminated to avoid interference. Notably, the evolved strains, when exposed to the same drug, frequently shared similar mutated genes. However, variations were still observed between the two independent evolutionary replicates.
Figure 2 provides a more comprehensive illustration of the correlation between mutations and the emergence of resistance. Genes with a higher mutation frequency were identified, and the time at which each mutation occurred was determined through Sanger sequencing of the PCR products from each strain (Supplementary Table 2). For AMP, SM, RIF, FD, and NV, development of resistance hinged on singular, stable mutations. The emergence of resistance to these drugs was attributed to mutations in pbpB, rpsL, rpoB, fusA, and gyrB respectively. The groups treated with SM, RIF, FD, and NV demonstrated rapid acquisition of mutations and drug resistances early in evolution. This trend aligns with the changes observed in the bacteriostatic zones (Fig. 1b). However, AMP-treated bacteria showed a delay in resistance emergence compared to the mutation. For EM, resistance development and mutations in rpmA and BPENGOFF-02461 were all observed from the 6th round evolution. This indicates that these two mutations are significant contributors to drug resistance. The CEX group showcased that resistance was mainly linked to fmtA mutation, which was acquired from the 5th round evolution. This corresponded to a significant decline in the bacteriostatic zone (Fig. 2). Mutations in gdpP and greA were also identified from the 8th round evolution, but with no significant changes in resistance. For GM, mutation inheritance proved to be unstable. norA mutations, despite a 58% frequency, independently did not affect MIC development, similar to the case with the bglA and odhA genes. MIC only increased once the fusA mutation was identified. Likewise, in the KM-treated group, there was increased complexity. The treR2 mutation, despite its early identification, had a minor impact on MIC. The MIC value started increasing only after the identification of fusA mutation and escalated with the mutation of gcvPA and sdhC.
Table 2
Representative mutant genes identified in drug-resistant strains
Drugs and mutated genes | Gene function | Mutation type | Frequency(mutated clones/all clones) | Group |
ampicillin | | | | |
pbpB | penicillin-binding protein | nonsynonymous | 4/12 | replicate2 |
BPENGOFF − 011573 | bacillithiol system protein YtxJ | nonsynonymous | 2/12 | replicate2 |
rpoB | DNA-directed RNA polymerase subunit beta | nonsynonymous | 2/12 | replicate1 |
cefalexin | | | | |
fmtA | teichoic acid D-Ala esterase FmtA | nonsynonymous | 12/12 | both |
gdpP | cyclic-di-AMP phosphodiesterase GdpP | stopgain | 12/12 | both |
greA | transcription elongation factor GreA | nonsynonymous | 7/12 | replicate1 |
BPENGOFF − 01589 | hypothetical protein | nonsynonymous | 5/12 | replicate2 |
BPENGOFF − 00051 | Putative DNA-binding protein | nonsynonymous | 4/12 | replicate2 |
BPENGOFF − 02465 | SH3 domain-containing protein | nonsynonymous | 4/12 | replicate2 |
gtf1-2 | glycosyl transferase, group 1 family protein | stopgain | 4/12 | replicate1 |
treR2 | trehalose operon repressor | stopgain | 4/12 | replicate1 |
BPENGOFF − 01771 | competence transcription factor ComK | nonsynonymous | 3/12 | replicate1 |
mutS2 | endonuclease MutS2 | nonsynonymous | 3/12 | replicate2 |
pnp1 | RNA-binding protein S1 | stopgain | 3/12 | replicate1 |
BPENGOFF − 01749 | RluA family pseudouridine synthase | nonsynonymous | 2/12 | replicate2 |
gentamicin | | | | |
fusA | elongation factor G | nonsynonymous | 7/12 | both |
norA | multidrug efflux MFS transporter NorA | nonsynonymous | 7/12 | both |
qoxB | cytochrome aa3 quinol oxidase subunit I | nonsynonymous | 5/12 | replicate2 |
ebh | hyperosmolarity resistance protein Ebh | nonsynonymous | 5/12 | replicate2 |
ythA | cytochrome ubiquinol oxidase subunit I | nonsynonymous | 5/12 | replicate2 |
rocD2-2 | ornithine–oxo-acid transaminase | nonsynonymous | 4/12 | both |
treR-2 | trehalose operon repressor | nonsynonymous | 3/12 | both |
bglA | aryl-phospho-beta-D-glucosidase BglA | nonsynonymous | 2/12 | replicate1 |
fatC | iron chelate uptake ABC transporter family permease subunit | nonsynonymous | 2/12 | replicate2 |
moeA | molybdopterin molybdotransferase MoeA | nonsynonymous | 2/12 | replicate2 |
odhA | 2-oxoglutarate dehydrogenase E1 component | stopgain | 2/12 | replicate1 |
oppD-4 | ABC transporter ATP-binding protein | nonsynonymous | 2/12 | replicate1 |
yxeP-3 | amidohydrolase | nonsynonymous | 2/12 | replicate2 |
yxeP-4 | amidohydrolase | nonsynonymous | 2/12 | replicate1 |
streptomycin | | | | |
rpsL | 30S ribosomal protein S12 | nonsynonymous | 12/12 | both |
rpoB | DNA-directed RNA polymerase subunit beta | nonsynonymous | 2/12 | both |
rpsD | 30S ribosomal protein S4 | nonsynonymous | 2/12 | replicate1 |
erythromycin | | | | |
rpmA | 50S ribosomal protein L27 | nonsynonymous | 5/12 | replicate2 |
csd | cysteine desulfurase | nonsynonymous | 5/12 | replicate2 |
pheT1 | phenylalanine–tRNA ligase subunit beta | nonsynonymous | 4/12 | replicate1 |
BPENGOFF − 02461 | recombination factor protein RarA | nonsynonymous | 4/12 | replicate2 |
mtlA | PTS mannitol transporter subunit IICB | nonsynonymous | 2/12 | replicate2 |
kanamycin | | | | |
treR2 | trehalose operon repressor | nonsynonymous | 11/12 | both |
fusA | elongation factor G | nonsynonymous | 11/12 | both |
gcvPA | glycine cleavage system P protein | stopgain | 6/12 | replicate2 |
malP | alpha-glucoside-specific PTS transporter subunit IIBC | stopgain | 6/12 | replicate2 |
sdhC | succinate dehydrogenase cytochrome b558 subunit | stopgain | 6/12 | replicate2 |
msrA2 | elastin-binding protein EbpS | nonsynonymous | 5/12 | replicate2 |
frdB | succinate dehydrogenase iron-sulfur subunit | nonsynonymous | 5/12 | replicate1 |
efeM | EfeM/EfeO family lipoprotein | stopgain | 5/12 | replicate1 |
atpG | ATP synthase subunit gamma | stopgain | 5/12 | replicate1 |
glvR | putative transcriptional regulator GlvR | stopgain | 5/12 | replicate1 |
BPENGOFF − 00007 | restriction endonuclease subunit S | stopgain | 3/12 | replicate1 |
rifampicin | | | | |
rpoB | DNA-directed RNA polymerase subunit beta | nonsynonymous | 12/12 | both |
BPENGOFF − 00211 | accessory regulator protein C | nonsynonymous | 6/12 | both |
agrA | LytTR family DNA-binding domain-containing protein | stopgain | 2/12 | both |
trpD-2 | anthranilate phosphoribosyltransferase | nonsynonymous | 2/12 | replicate2 |
yexP-2 | amidohydrolase | nonsynonymous | 2/12 | both |
fusidic acid | | | | |
fusA | elongation factor G | nonsynonymous | 12/12 | both |
BPENGOFF − 00211 | accessory regulator protein C | nonsynonymous | 4/12 | both |
rplF | 50S ribosomal protein L6 | nonsynonymous | 2/12 | replicate1 |
novobiocin | | | | |
gyrB | DNA gyrase subunit B | nonsynonymous | 12/12 | both |
potB | Putative ABC transport system permease protein | stopgain | 5/12 | replicate1 |
fpgS | folylpolyglutamate synthase | nonsynonymous | 2/12 | replicate2 |
Cross-resistance and relative growth rates.
To delve deeper into the role of various gene mutations in drug resistance, cross-resistance experiments were conducted (as shown in Fig. 3). We isolated three monoclonal strains in each experiment, sequencing their entire genome (refer to Table 3). We distinctly observed cross-resistance in two β-lactam antibiotics, AMP and CEX. A similar pattern was detected with GM and SM, both of which target the 30S ribosomal subunit. Strains showcasing tolerance to AMP, CEX, KM, or FD presented varying resistance levels to GM. Interestingly, in the case of EM and KM, which influence the 50S ribosomal subunit, cross-resistance did not feature. Almost all drug-resistant strains demonstrated heightened sensitivity towards FD. Strains showing tolerance to RIF, FD, or NV tended to have an increased sensitivity to AMP, especially the strains resistant to NV. An exception, however, were the AMP-tolerant strains, which also exhibited tolerance to VA, unlike other drug-resistant strains. Notably, amongst the strains with tolerance to SM, SME1 portrayed strong resistance to RIF, a trait not seen in its group members. As for the relative growth rate, most strains were unaffected, with the exception of the CEXE3 strain whose growth rate decreased by nearly 50%, and curiously, displayed increased sensitivity to many other drugs.
Table 3
Mutated genes of representative drug-resistant strains
Strains | Genes |
AMPE-1 | rpoB |
AMPE-2 | rpoB glnA |
AMPE-3 | pbpB |
CEXE-1 | fmtA gdpP dppE greA treR_2 gtf1_2 BPENGOFF_01771 BPENGOFF_00137 |
CEXE-2 | fmtA gdpP gyrA greA |
CEXE-3 | fmtA gdpP BPENGOFF_01589 BPENGOFF_02465 |
GME-1 | norA fusA bglA dagK dat fakA odhA treR_2 |
GME-2 | norA ebh korA qoxB ythA BPENGOFF_01047 |
GME-3 | norA ebh ptsG_2 qoxB ythA treR_2 |
SME-1 | rpsL sucD pepS |
SME-2 | rpsL agrA hssS |
SME-3 | rpsL rho BPENGOFF_01241 |
EME-1 | pheT_1 |
EME-2 | BPENGOFF_02230 |
EME-3 | rpmA BPENGOFF_02461 cdr csd ftsW pstA |
KME-1 | fusA atpG efeM glvR treR_2 sdhB |
KME-2 | fusA sdhC gcvPA malP treR_2 BPENGOFF_02043 BPENGOFF_01578 |
KME-3 | hemB yqfL |
RIFE-1 | rpoB rpoC BPENGOFF_00211 BPENGOFF_00846 |
RIFE-2 | rpoB agrA ebh scrB purM |
RIFE-3 | rpoB gdpP BPENGOFF_00211 |
FDE-1 | fusA |
FDE-2 | fusA |
FDE-3 | fusA scrB BPENGOFF_00211 |
NVE-1 | gyrB pepA_2 potB fmhA BPENGOFF_00044 |
NVE-2 | gyrB rimI |
NVE-3 | gyrB |
Laboratory evolution has impacted protein’s structure and function of S.aureus.
The drugs of focus for this study, RIF and NV, were selected due to the relative ease of identifying their main drug-resistant mutations. Upon docking these drugs with the respective proteins, it was observed that the mutations induced changes in the amino acids located near the antibiotic binding site. Specifically, for NV, three mutations were identified in the DNA gyrase subunit B, which is encoded by gyrB. These mutations are S128L, R144I, and G85S (refer to Fig. 4a). In contrast, for RIF, a solitary mutation, H481Y, was identified in the DNA-directed RNA polymerase subunit β, encoded by rpoB (refer to Fig. 4b).
In assessing real-world strains, we observed that the effects of gene mutation varied among strain types. Non-synonymous mutations initiated by single bases were scored via MutPred2 on a scale of “0” to “1”, with any score over 0.5 implying a consequential impact on protein function. The cumulative scores for all mutations in a specific gene enabled us to estimate the overall impact on protein function. As observed from our laboratory evolution research, the genes associated with human clinical isolate mutations notably exhibited higher MutPred2 scores when compared to those of environmental or animal isolates. (Refer to Fig. 5 and Supplementary Table 3 for more details).