Fig. 3: 4′-DTMP selectively eliminates mutants carrying the L28R mutation from TMP-resistant populations.

a Schematic representation of our experiment. A mixture of six bacterial populations with highly TMP-resistant genotypes are propagated for ~32 h, under selection with 500 µM of 4′-DTMP (n = 7 replicates), TMP (n = 7 replicates), or no drug (n = 3 replicates) as a control. DHFR mutations that were dominantly found in these cultures are listed. Cultures were grown for 4 (or 8) hours and they were serially diluted to maintain low cell density and faster growth. Samples were collected at six time points before each dilution event and the folA gene of these populations were deep sequenced using MiSeq (please see the SI for details). b Average growth rates of cultures in between dilutions are quantified by calculating number of generations per hour (number of generations = log₂[OD_final/OD_initial]). Cultures under 4′-DTMP selection had slowest growth till 12 h and then started to grow faster. Error bars show the standard deviation, center of the error bars corresponds to the mean value of the measurements. c Temporal changes in frequencies of DHFR mutations with no drug (top panel), under TMP selection (middle panel), or under 4′-DTMP selection (bottom panel) [Error bars show the standard error on the mean, center of the error bars corresponds to the mean value of the measurements]. The L28R mutation quickly plateaus under TMP selection whereas it was replaced by the D27E and F153S mutations under 4′-DTMP selection. We note that the initial increase in L28R frequency under 4′-DTMP selection is accompanied with slow growth and small population size, further supporting selective activity of 4′-DTMP against the L28R mutation.