The delayed outgrowth of ∆ybeX is heterogeneous at the individual cell level
When streaking out mutant strains from glycerol stocks and overnight grown stationary phase cultures, we noticed that while the ∆ybeZstrain produces wild-type-like colonies and ∆ybeY strain has uniformly small colonies, the ∆ybeX strain produces heterogeneous colonies. Re-streaking of small and large ∆ybeX colonies resulted in well-grown large second-generation colonies, indicating that the heterogeneous phenotype is not caused by a genetic mutation (data not shown). We also tested whether the colony heterogeneity is caused by freeze-thawing in the glycerol mix by growing the ∆ybeX and wild-type cells in liquid media into stationary phase and then plating the cells directly onto LB agar plates. Again, while wild-type cells exhibited uniform colonies, the ∆ybeX strain gave heterogeneous colony growth (Fig. 2A ). Thus, it is likely that the growth heterogeneity of ∆ybeX depends on the heterogeneity of the initial physiological states of individual stationary cells.
We quantified colony radiuses of wild-type and ∆ybeX strains grown in LB and MOPS minimal media supplemented with 0.3% glucose using AutocellSeg (Khan et al. , 2018). ∆ybeX cells tend to form smaller colonies than wild-type cells when grown in LB and MOPS media (Fig. 2A ). ∆ybeX colonies are heterogeneous when grown in LB medium (Fig. 2B ), while the colony radiuses are homogeneous when ∆ybeX cells are grown in MOPS liquid medium (Fig. 2C ).
We then asked how heat shock affects cell growth and heterogeneity. Overnight cultures were diluted and plated on LB agar plates following 16-18 hours of incubation at 37°C or 42°C (Fig. S3a ). We observed fewer ∆ybeX colonies at 42°C (p < 0.0001 and p = 0.02 for LB and MOPS, respectively; Fig. 2D ).
We inspected the colony growth of ∆ybeX::kan (ybeX single deletion strain in BW25113 background constructed via lambda red recombination) and ∆ybeX/kan- (the kanamycin cassette removed from the inhouse constructed ∆ybeX::kan ) cells at 37°C for 24 and 48 hours (Fig. S3b, c ). No significant differences were observed for ∆ybeX::kan and ∆ybeX/kan- . Furthermore, although the observed tiny colonies of ∆ybeX were increasing in size over time, they consistently remained smaller than WT-like∆ybeX colonies (Fig. S3b, c ).
To better understand the nature of the observed lag phase phenotype at the individual cell level, we quantified colony radiuses of the∆ybeX and the WT cells at 37°C and 42°C using cells pre-grown for 16-18 hours in liquid LB or MOPs minimal media prior to plating. Inspection of four independent stationary phase outgrowth experiments showed, in accordance with our previous observations, that at both temperatures WT cells tend to form larger colonies, while the colony radiuses of ∆ybeX cells are heterogeneous and possibly dimorphic. These intuitions were formalized by jointly modelling means and standard deviations of colony radiuses and, in a separate model, the colony radiuses as mixtures of two normal distributions (see Materials and Methods for details). The estimated mean colony radiuses are smaller in the ybeX strain by about 1/3 (the difference, in arbitrary units, at 37°C is 3.58 [95% CI 2.89, 4.24] and at 42°C is 3.72 [3.01, 4.42]), and the ∆ybeX colony radiuses have a larger standard deviation (the difference in 37°C is 0.67 [0.36, 1.04] and at 42°C is 0.91 [0.53, 1.34]). Interestingly, modelling the colony radiuses as emanating from two distinct gaussian populations resulted in a superior out-of-sample fit, as assessed by leave-one-out-cross validation (data not shown), supporting the conjecture that ∆ybeXcells grow in at least two distinct regimes, one of which is similar to WT growth, while another results in up to two-fold smaller colonies (WT vs. ∆ybeX difference in 37°C: µ1 (estimate for the mean of the first gaussian): 1.73 [0.47, 2.93], µ2: 5.58 [4.71, 6.39], and in 42°C: µ1:1.39 [0.06, 2.81], µ2: 5.92 [4.99, 6.79]).