Pervasive selection for and against antibiotic resistance in inhomogeneous multistress environments

Abstract

Antibiotic-sensitive and -resistant bacteria coexist in natural environments with low, if detectable, antibiotic concentrations. Except possibly around localized antibiotic sources, where resistance can provide a strong advantage, bacterial fitness is dominated by stresses unaffected by resistance to the antibiotic. How do such mixed and heterogeneous conditions influence the selective advantage or disadvantage of antibiotic resistance? Here we find that sub-inhibitory levels of tetracyclines potentiate selection for or against tetracycline resistance around localized sources of almost any toxin or stress. Furthermore, certain stresses generate alternating rings of selection for and against resistance around a localized source of the antibiotic. In these conditions, localized antibiotic sources, even at high strengths, can actually produce a net selection against resistance to the antibiotic. Our results show that interactions between the effects of an antibiotic and other stresses in inhomogeneous environments can generate pervasive, complex patterns of selection both for and against antibiotic resistance.

Figure 1. Doxycycline potentiates selection biases for and against tetracycline resistance by diverse otherwise neutrally selective antibiotics.

(a) YFP-labelled, tetracycline-sensitive (Green) and CFP-labelled, tetracycline-resistant (Red) E. coli are mixed and grown together over a diffusing toxin gradient in agar either absent or containing a uniform level of the tetracycline antibiotic, doxycycline (−Dox, +Dox, respectively). Tetracycline resistance is induced by doxycycline in +Dox plates and, without growth defect, by anhydrotetracycline added to −Dox plates. Final strain ratios reveal deviations from background inhibition (Yellow lawn with dark zone of clearing) that bias selection towards tetracycline resistance (Red ring) or sensitivity (Green ring). (b,c) Certain compounds such as erythromycin (b, Ery) and ciprofloxacin (c, Cpr) do not directly select on tetracycline resistance alone (−Dox plates), yet combine with a sub-inhibitory background level of a tetracycline (doxycycline) to select for resistance (Ery, +Dox, Red ring) or against resistance (Cpr, +Dox, Green ring). d, Nearly identical growth responses of tetracycline-sensitive (Tet^S, Green plots Strain Wyl,) and -resistant (Tet^R, Red plots, Strain t17cl) strains by ciprofloxacin alone (−Dox), diverge significantly in the presence of a uniform, sub-inhibitory level of doxycycline (+Dox), generating a region of strong ‘threshold selection' between the MICs (grey vertical lines) where only the sensitive strain can grow (Green shading). Smoothing splines R²: 0.998 (Tet^R, −Dox), 0.997 (Tet^S, −Dox), 0.999 (Tet^R, +Dox), 0.995 (Tet^S, +Dox). e, Competing tetracycline-resistant (Red) and -sensitive (Green) bacteria are equally inhibited, and experience no change in relative growth along diffusing gradients of diverse non-tetracycline antibiotics acting alone (−Dox panels). Combining these antibiotic gradients with a uniform, sub-inhibitory level of doxycycline in the agar (+Dox panels) typically biases selection near the MIC (see Supplementary Fig. 2 for an exception) either towards tetracycline resistance (Red rings) or tetracycline sensitivity (Green rings).

Figure 2. A simple model suggests nearly all antibiotic-stress interactions exhibit regions of antibiotic-potentiated selection for or against resistance.

Antibiotics applied in combination produce interactions which are coarsely visualized by the minimum inhibitory concentration (MIC) line, which separates growth and no growth regions in drug–drug concentration space. Interactions can range from drug X suppressing, to buffering, to augmenting the effects of drug Y (a–c, respectively). MIC lines of an X-resistant strain (X^R, red) are approximated by geometrically scaling the MIC lines of the X-sensitive strain (X^S, green) along the drug x axis, reflecting the diminished effective levels of that drug to the resistant strain (grey arrow in a). While resistance to drug X does not change the MIC to drug Y alone (the green and red lines coincide along Y-axis, [X]=0), adding sub-inhibitory levels of drug X tends to separate the MIC lines of the X^R and X^S strains, exposing regions of ‘threshold selection' where one strain grows but the other is fully inhibited (a, only X^S strain can grow in the pale green region with no red stripes; c, only X^R strain can grow in the red striped region with no green background). This potentiated threshold selection occurs along gradients of drug Y at a fixed positive concentration of X (wedge at [X]=x′) for every drug interaction except for buffering (b), where the MIC lines coincide. Qualitatively, these regions of selection between sensitive and resistant MIC lines are observed in differential inhibition assays over diffusing Y gradients (Compare wedges in plots and in schematic differential inhibition assays directly above). The model can be refined to include other positive growth isoclines which, treated similarly to the MIC, generate less intense selection windows with the same dependence on the shape of the stress interaction.

Figure 3. Sub-inhibitory doxycycline potentiates selection for or against tetracycline resistance by non-antibiotic chemical and physical stresses.

(a) Differential inhibition assays point to doxycycline-potentiated selection by non-antibiotic compounds, biased in favour of tetracycline resistance for hydrogen peroxide, EDTA, and citric acid in various forms, and biased towards tetracycline sensitivity for paraquat and CaCl₂. (b) Doxycycline potentiates selection biased towards tetracycline sensitivity by osmotic stress due to high concentrations of NaCl, and to a lesser degree sucrose and PEG 8000. (c) Modified differential inhibition assays indicate that tetracycline-sensitive and -resistant bacteria stressed by transient gradients of ultraviolet radiation and heat also exhibit tetracycline-potentiated selection for resistance.

Figure 4. Non-selective toxins can substantially modify the selection on resistance by an antibiotic diffusing from a local source.

(a) A diffusing gradient of doxycycline alone selects strongly for tetracycline resistance at higher concentrations and is neutral at lower levels (Inset, −Cpr). A small inhibitory level of ciprofloxacin added uniformly to the agar (+Cpr) drastically changes the pattern of selection by doxycycline, revealing a large band of selection for sensitivity (green circle), a band of selection for resistance (red circle), and multiple regions of inhibition of both strains (black regions). (b) Since ciprofloxacin and doxycycline interact suppressively, we rationalize this picture by comparing the sampled gradient in the image (a) with a schematic of such a suppression interaction (compare blue dashed wedges), that similarly crosses regions of threshold selection for tetracycline resistance and sensitivity.