Controlling gene expression in yeast by inducible site-specific recombination

Abstract

An intron module was developed for Saccharomyces cerevisiae that imparts conditional gene regulation. The kanMX marker, flanked by loxP sites for the Cre recombinase, was embedded within the ACT1 intron and the resulting module was targeted to specific genes by PCR-mediated gene disruption. Initially, recipient genes were inactivated because the loxP-kanMX-loxP cassette prevented formation of mature transcripts. However, expression was restored by Cre-mediated site-specific recombination, which excised the loxP-kanMX-loxP cassette to generate a functional intron that contained a single loxP site. Cre recombinase activity was controlled at the transcriptional level by a GAL1::CRE expression vector or at the enzymatic level by fusing the protein to the hormone-dependent regulatory domain of the estrogen receptor. Negative selection against leaky pre-excision events was achieved by growing cells in modified minimal media that contained geneticin (G418). Advantages of this gene regulation technique, which we term the conditional knock-out approach, are that (i) modified genes are completely inactivated prior to induction, (ii) modified genes are induced rapidly to expression levels that compare to their unmodified counterparts, and (iii) it is easy to use and generally applicable.

Figure 1

The conditional knock-out approach for controlling gene expression. (A) Methodology. The 5′- and 3′-ends of the ACT1 intron (Int5′ and Int3′, respectively) were separated by a kanMX cassette, which contains the following sequences: the kanamycin resistance gene (kan^r), the Ashbya gossypii TEF2 promoter (prom.) and terminator (term.) and tandemly oriented loxP sites. PCR is used to amplify the module and add sequences with genomic homology to the ends. The amplified fragment is used to modify YFG (your favorite gene) by one-step gene disruption (15,16), which results in loss of YFG expression. Inducible site-specific recombination is used to excise the inhibitory kanMX cassette and restore YFG expression. (B) Primer sequences for modifying genes with the intron/kanMX module. The Up and Down primers bind to the 5′-donor and 3′-acceptor sites of the ACT1 intron, respectively. Amplification of plasmid pRKO with these primers yields the self-contained intron/kanMX module with 45 bp of additional sequence on each end (X₄₅ and Y₄₅), which correspond to the integration site within the recipient gene.

Figure 2

Control of URA3 expression by the conditional knock-out approach. Three isogenic strains, PJ1 URA3, THC82 ura3::intron/kanMX and THC83 URA3::intron, were streaked on: panel 1, SC-ura; panel 2, SE+G418 drop-in media containing histidine, leucine, tryptophan, uracil and adenine (SE corresponds to synthetic media containing glutamate as a nitrogen source); panel 3, SC+G418. Rare Cre-independent recombination events between loxP sites (expected at a frequency of 1/10⁴) probably account for the infrequent appearance of ura⁺ colonies among THC82 cells.

Figure 3

Changes in DNA and RNA during URA3 activation. Nucleic acids were harvested from strain THC82 containing pSH63 (lanes 4–7) at timed intervals following the addition of galactose, according to the procedures prescribed in Materials and Methods. (A) Rearrangement of the URA3 gene. Samples were cut with NdeI prior to electrophoresis. Blot was hybridized with a randomly-primed probe corresponding to the ACT1 intron. Band intensities were measured by phosphorimaging. The percentage of recombination (% recombined) was measured by the disappearance of the ura3::intron/kanMX band, which was normalized for loading and reported relative to the value in lane 4. (B) Changes in mature URA3 mRNA levels. Blot was hybridized sequentially with probes to the 5′-UTR of URA3 and the ACT1 ORF. Values were normalized for loading and reported relative to that for the native URA3 gene in lane 1. The plasmid-free strains used as controls in lanes 1–3 were PJ1 URA3, THC82 ura3::intron/kanMX and THC83 URA3::intron, respectively.

Figure 4

Inducing DNA rearrangements with a Cre-EBD fusion protein. Strain THC82 ura3::intron/kanMX containing pSH62-EBD (lanes 3–7) was induced with galactose, as described previously. In lane 5, galactose and estradiol (C_f = 1 µM) were added simultaneously for a 2 h interval. In lane 7, galactose was added 1 h prior to the addition of estradiol. Strains THC82 ura3::intron/kanMX and THC83 URA3::intron serve as controls in lanes 1 and 2, respectively.

Figure 5

Patch mating assay for restoration of SIR3 function. Recombination was induced in strain CRC9 sir3::intron/kanMX containing pSH47. Individual colonies from the induction protocol were patch-mated with W303-1B overnight and replica-plated to SC-trp,-lys (Materials and Methods). Left panel, no galactose control. The infrequent instances of mating (2/42 isolates mated and formed patches) probably arose from low-level basal expression of the recombinase. Right panel, exposure to galactose for 2 h (40/42 isolates mated and formed patches). The percentage of mating competent cells was dependent on the length of time in galactose, indicating that restoration of SIR3 expression occurred during (and not after) the galactose induction interval (data not shown).