Fig. 2: 1A01 and 3B05 co-culture in weak bicarbonate buffer.

a Solid arrows indicate schematic of acetate cross-feeding based on single–strain characteristics derived from Fig. 1. In a weak buffer, acetate excretion will reduce the pH and inhibit cell growth as indicated by the dashed and dotted lines. b A canonical scenario of syntrophy is realized if the growth-inhibiting effect exerted on the acid excreter (1A01, solid line) is stronger than that on the acid eater (3B05, dashed line) as pH drops. The intersection of these two lines is the fixed point describing a stable, exponentially-growing co-culture. To test this scenario, we grew the 1A01 and 3B05 co-culture in 5 mM GlcNac in weak buffer (2 mM bicarbonate), inoculated at 1:1 ratio. c shows the OD (black squares) and pH (orange circles), with the horizontal dotted line indicating the final OD reached by the same co-culture grown in strong buffer (Fig. 1g). d shows the GlcNAc (blue triangles) and acetate (red squares) concentrations in the medium. e shows the viable cell density (Supp. Fig. 3) for 1A01 (filled circles) and 3B05 (open circles). The vertical dashed line in (c)–(e) indicates the time when the increase in OD ceased according to (c). f Dependences of the growth rate of 1A01 (solid circles) and 3B05 (open circles) on the medium pH. Cells were grown in minimal medium buffered by 10 mM MES with different ratios of the acid and base form to obtain the desired pH. Glycerol was used as the sole carbon source as both strains grew on it and neither strain excreted acetate or other fermentation products which would have changed the medium pH during the course of experiment. The data shows that 3B05 is more sensitive to pH than 1A01, hence precluding the scenario of a stable, exponentially-growing co-culture depicted in (b). We developed a simple quantitative model (Supplementary Note 1) for the co-culture dynamics using single-strain characteristics summarized in Supplementary Table 1 and acid response data shown in (f). g Model output on the concentrations of GlcNAc (blue line) and acetate (red line) agree quantitatively with those measured in (d) up to the time of growth arrest. h Model output on the densities of 1A01 and 3B05 cells (solid and dashed lines) agree quantitatively with the observed viable cell densities shown in (e). The model also correctly predicted growth arrest to occur about 6 h after inoculation (position of the vertical dashed lines in (c)–(e), with about one-third of the initial GlcNAc still remaining at that time. This simple model does not predict what occurs after the growth arrest, which is the focus of the rest of the study.