Consideraciones éticas

The results in previous sections indicate that reduced flux through the pyrimidine biosynthetic pathway activates an epigenetic feedback loop, the operation of which generates phenotypic switching. Extensively studied in a number of species, the UTP biosynthetic pathway is long and complex, with multi-layered and (relatively) poorly understood control mechanisms (see Figure 6.1). Thus, in order to identify potential feedback loops, it was necessary to more closely define the region of the pathway containing the genes and/or intermediates directly responsible for switching (the ‘switch unit’). The upstream boundary in the search for the switch unit was indicated by 1w4- reN1.4 - the ability of a lone mutation in pyrH to cause phenotypic switching indicated that reduced flux downstream of pyrH was the primary cause of bistability. In an experiment designed to define the second, downstream boundary of the switch unit, genes downstream of pyrH were sequentially over-expressed, and the effect on switching observed. The expectation was that if reduction in a particular intermediate contributed to switching, addition of that intermediate (through increased enzyme expression) would alleviate switching. Beyond the downstream boundary of the switch unit, the level of intermediates would have no bearing on switching, and thus over- expression of downstream genes would no longer alleviate switching. To this end, the wild-type carB (as a positive control), pyrH, ndk and galU genes were sequentially over-expressed from the pSX plasmid in the 1w4, SBW25 and 1w4-reN1.4 backgrounds.

To achieve this, each of the four biosynthetic genes was amplified from SBW25 genomic DNA using primer pairs CarBOE-f/r (Elongase; 56˚C annealing temperature, 3.5 minutes extension time), PyrHOE-f/r (58˚C, 1 minute), NdkOE-f/r (58˚C, 30 seconds) and GalUOE-f/r (57˚C, 1 minute), respectively. Each product was ligated into pCR8/GW/TOPO, and the resulting construct used to transform chemically competent

E. coli. Clones containing mutation-free fragments were selected, and the fragments retrieved via double digestion with NdeI and BamHI. Isolated fragments were independently ligated into the pSX vector digested with NdeI/BamHI, giving the constructions pSX-carB, pSX-pyrH, pSX-ndk and pSX-galU. Along with the empty pSX vector, each construct was used to transform chemically competent SBW25, 1w4 and 1w4-reN1.4 cells (see section 2.2.3.1). Transformants were checked for presence of

the insert via PCR. This resulted in the following 13 genotypes (three independent biological replicates were produced per genotype): SBW25-pSX, SBW25-pSX-carB, SBW25-pSX-pyrH, SBW25-pSX-ndk, SBW25-pSX-galU, 1w4-pSX, 1w4-pSX-carB, 1w4-pSX-pyrH, 1w4-pSX-ndk, 1w4-pSX-galU, Re1.4-pSX, Re1.4-pSX-carB and Re1.4- pSX-pyrH. Despite several independent attempts, pSX-ndk and pSX-galU could not be used to transform chemically competent 1w4-reN1.4, indicating that over-expression of these genes was severely deleterious in the presence of the pyrH mutation.

Independent capsule counting assays were performed on each of the three sets of 13 genotypes (see section 2.2.11.4). Except in the cases of control strains (i.e. those not containing pSX or derived construct), 10 µg ml-1 Gm was added to the medium. Inducer (IPTG) was not added to the medium, as a preliminary study indicated that phenotypic effects were clearly visible through leaky expression achieved without induction. The mean and standard error of the proportion of cells capsulated in three (non-biological) replicates of each genotype were calculated in each set of 13 genotypes. Following collection of data for all three sets, the means of biological replicates were compared (Appendix A4.6). For every genotype, all biological replicates were comparable – that is, the smallest and largest of the three replicate means and associated standard errors overlapped. Given this, the final ‘composite’ mean and standard error of the proportion of capsulated cells was calculated using all nine replicates (three non-biological replicates for each of three biological replicates) of each genotype, unless otherwise stated (Figure 6.12, Appendix A4.6). Subsequently, P-values for differences between the means of genotypes of interest were calculated using two-sample t-tests or, where normality assumptions were not satisfied, M-W-W tests (see section 2.2.12.1).

Figure 6.12: Graphs showing the proportion of cells capsulated in 1w4 (A), SBW25 (B) and 1w4-

reN1.4 (C) populations with the indicated genes over-expressed from pSX. In most cases, bars

represent mean values of nine replicates (three non-biological replicates of each of three biological replicate strains; see Appendix A4.6 for details). Error bars indicate one standard error of all nine (biological and non-biological) replicates. Stars indicate a statistically significant difference between the mean and that of the corresponding genotype containing empty pSX (*<0.05, **<0.01, ***<0.001).

A: 1w4 B: SBW25 C: 1w4-reN1.4 *** *** *** * ** *** *** ***

The results of this experiment provided significant insight into the molecular mechanisms underlying capsule switching. Firstly, addition of the empty pSX vector significantly lowered the proportion of capsulated cells in both 1w4 and 1w4-reN1.4 backgrounds (P<0.001). It is possible that this resulted from Gm addition, as similar effects were noted in the presence of Tc. Alternatively, it is possible that the protocol used to render P. fluorescens cells chemically competent (or the subsequent heat shock) caused lasting alterations to the plasma membrane, which in turn alter capsule expression. Given the significant magnitude of the empty pSX effect on capsule switching, this phenomenon should be investigated further. The Gm effect could be investigated by creating a 1w4-derived genotype with a Gm resistance gene incorporated into the chromosome. If such a strain also demonstrated reduced capsule expression in the presence of Gm (but absence of the pSX vector and transformation protocol, see below), this would support the Gm hypothesis. In order to investigate the heat shock hypothesis, an ori locus could be incorporated into the pSX vector to enable conjugation of pSX overexpression constructs from E. coli cells, circumventing the need to use transformation protocols directly on Pseudomonas genotypes.

Consistent with the hypothesis that the upstream boundary of the switch unit lies beyond carB, carB over-expression had no effect on capsule expression in SBW25 or 1w4-reN1.4 (P>0.1), but alleviated switching caused by the carB mutation in 1w4 (P=4 x 10-4). In 1w4, pyrH and ndk over-expression both lowered capsule expression below that observed for the control genotype, 1w4-pSX (P<0.001), while galU over-expression did not alter 1w4 capsule expression (P=0.351). A similar pattern was observed on a smaller scale in SBW25, where pyrH and ndk over-expression significantly lowered capsule expression (P<0.05), and galU over-expression did not (P=0.489). In 1w4- reN1.4, pyrH over-expressionsignificantly reduced capsules (P=0.0168), demonstrating that the pyrH mutation causes phenotypic switching in 1w4-reN1.4 (see section 6.3.1.2). In summary, phenotypic switching in each genetic background is influenced by the over-expression of only two genes: pyrH and ndk. Collectively, the results of this experiment strongly suggest that the switch unit lies within the pyrH-ndk-pyrG segment of the pyrimidine biosynthetic pathway.

6.4 Discussion