Results
Principle of the screen
Rop modulates the copy number of ColE1 plasmids by facilitating the binding of the inhibitory RNA I to the priming RNA II of the ColE1 origin. Thus, RNA II is prevented from functioning as a primer for DNA replication, and plasmid copy number is reduced (Figure 1) (Cesareni et al., 1991
). We hypothesized that a cell-based screen for Rop function could be created by expressing GFP from a ColE1 plasmid. In the presence of active Rop, the copy number of ColE1 plasmid diminishes, and the amount of GFP (and hence cellular fluorescence) is expected to decrease (Figure 3a). The vector pUClacGFPuv contains the pUC19 version of the ColE1 origin (which has a point mutation in RNA II), the gene for ampicillin resistance, and an expression cassette for a GFP variant under the control of the lac promoter (see Figure 2 for vector schematics). The pUC19 variant of the ColE1 origin was effective for our screen, as opposed to the version found in plasmids such as pBR322, because the pUC19 origin confers a much larger difference in copy number with and without Rop, especially at 42°C (data not shown). This RNA II point mutation was previously known to confer an extremely high copy number at 42°C in the absence of Rop (Lin-Chao et al., 1992; Lahijani et al., 1996).
We also constructed the plasmids pAClacRop, pAClacLink and pAClacCm, which express, respectively, the gene for wild-type Rop (wt Rop) (positive control), a short linker (negative control) and the gene for chloramphenicol resistance (a long linker for ease of cloning libraries of Rop variants). These plasmids all contain an expression cassette under the control of a synthetic lac promoter, the gene for kanamycin resistance and the p15A origin from pACYC177, which is not regulated by Rop and allows for co-maintenance with ColE1 plasmids.
A screening strain was created by transforming pUClacGFPuv into DH10B E.coli, and competent cells of this strain were transformed with pAClacRop or pAClacLink. On solid LB medium, cells bearing pAClacLink (i.e. no Rop) were found to be strongly fluorescent, but those bearing pAClacRop (i.e. wt Rop) were only very weakly fluorescent, after 16 h at 42°C (Figure 3b). Thus, cellular fluorescence corresponds to the lack of active Rop under these conditions.
Positive screening with PBAD–GFP on a ColE1 plasmid
A drawback to the version of the screen described above is that cells bearing active Rop are dim, while those lacking Rop activity are fluorescent (i.e. it is a negative screen). However, one can imagine trivial reasons that might lead to diminution of fluorescence in a library experiment (such as spontaneous mutations of the GFP), and it would generally be convenient to have a screen in which brightly fluorescent cells correspond to Rop activity.
We found that this was possible under certain conditions by expressing GFPuv under the control of the arabinose promoter (PBAD) on the ColE1 plasmid. The plasmid pUCBADGFPuv is identical to pUClacGFPuv, except that the lac promoter was replaced with the arabinose promoter, including the araC gene, which is divergently transcribed from the GFPuv gene. Again, a screening strain (the ‘positive screening strain’) was created by transforming DH10B E.coli with pUCBADGFPuv, and competent cells of this strain were transformed with pAClacRop or pAClacLink. Surprisingly, on solid LB medium containing 0.0005% arabinose and 100 µM IPTG, cells grown at 42°C overnight were fluorescent when bearing wt Rop, but dim when bearing no Rop (Figure 3c). Thus, cellular fluorescence in this format of the screen corresponds to the presence of active Rop. At lower concentrations of arabinose, none of the cells were fluorescent; at higher concentrations, all of the cells were fluorescent (data not shown).
To confirm that the fluorescence phenotypes were the result of changes in plasmid copy number, we extracted and roughly quantified the plasmid DNA in the cells. Active Rop visibly reduces the amount of ColE1 plasmid that is isolated in alkaline lysis miniprep (Figure 3d). We examined cells grown in solution under conditions identical to those on agar plates, which verified that the bright, Rop-containing cells contained less of the pUCBADGFPuv ColE1 plasmid. In addition, SDS–PAGE analysis of lysates of these same cells showed that the bright, Rop-containing cells contain more GFP than the dim, linker-containing cells, and that it is mostly in the soluble fraction (data not shown).
We believe that this phenotypic reversal is the result of the mechanism of the arabinose promoter, wherein the AraC regulator acts as a repressor in the absence of arabinose but as an activator in its presence (Schleif, 2000). We hypothesize that in the absence of Rop, the high copy number of the ColE1 plasmid requires more arabinose for AraC-mediated activation of GFP expression, rendering the cells dim under these conditions. Regardless of the exact details of the mechanism, this system allows us to screen for active variants of Rop by looking for brightly fluorescent cells under these conditions.
Assay of systematically repacked Rop variants
Previously in this laboratory, Munson et al. (1994a, 1996) examined the gross structural effects of repacking the hydrophobic core of Rop with combinations of Ala, Leu and Met. Briefly, the hydrophobic core of Rop can be considered to contain eight layers composed of two amino acids from each monomer. Owing to the symmetry of the homodimer, a change in one layer concomitantly alters the symmetrically associated layer at the other end of the four-helix bundle. Thus, engineering the fourth layer to contain Ala2Leu2 simultaneously causes the fifth layer to be Ala2Leu2. This arrangement is called Ala2Leu2-2. Moreover, the core is generally composed of alternating ‘small’ and ‘large’ residues, except for in the penultimate layers at each end, which are said to be reversed. Rop variants in which the next-to-last layers are reversed are denoted ‘rev’ (Figure 4).
We examined five of these Rop variants for Rop activity by cloning the variants into the pAClacLink vector, transforming into the positive selection strain and screening as described above. Three of the variants, Ala2Leu2-2, Ala2Leu2-4 and Ala2Leu2-8-rev, were previously found (Munson et al., 1996) to have predominantly dimeric structure, CD spectra similar to wt Rop, higher Tm values than wt Rop and the ability to bind to small RNA hairpins similar to loops in RNA I and RNA II with affinities approaching that of wt Rop (as judged by gel-shift assay). However, by screen or examination of copy number by plasmid prep, the Ala2Leu2-8-rev variant is totally inactive, the Ala2Leu2-4 variant may be slightly active, and the Ala2Leu2-2 variant is strongly active. As expected, the Ala2Met2-8 variant (which was found to be weakly dimeric, to have a CD spectrum similar to wt Rop, to have a depressed Tm, and was not observed to bind RNA) and the Leu4-8 variant (apparently tetrameric, CD spectrum indicative of high helical content, non-cooperative melting transition, no RNA binding) were inactive in the cell-based assay (Figure 4).
Evidently, in vivo Rop function requires more than just the ability to bind the small stem–loop hairpins derived from the much larger RNA I and RNA II molecules. Munson et al. (1996) concluded that those Rop variants that bound the RNA hairpins were the most Rop-like in structure. However, neither the solution nor the crystal structures of these variants was ascertained, and it is therefore not known how Rop-like the variants are at atomic resolution. Also, the repacked Rop variants exhibited extremely different folding kinetics for dimerization, and this may be a factor contributing to the difference between in vivo activity and in vitro binding. Apparently, merely maintaining the presence of the RNA binding residues in a Rop variant that forms a stable, four-helix bundle is insufficient for activity. Consequently, those Rop variants that pass the screen are likely to be extremely Rop-like in structure and thermodynamic properties, possessing not only the overall Rop fold but also atomic-level details that are required for function.
Preparing for libraries: the pAClacCm vector
In order to interrogate libraries of Rop variants with the screen, one must efficiently clone genes for those variants into the pAClac vector between NdeI and BanI sites. However, we found that NdeI cuts the pAClacLink vector very inefficiently, requiring in excess of 400-fold more enzyme than anticipated (Figure 5). Since the linker in pAClacLink is very small, it is impossible to resolve singly and doubly cut vector on an agarose gel, and this leads to high background upon ligation and transformation. To combat this problem, we cloned the CmR gene for chloramphenicol acetyltransferase (CAT, from pACYC184) under the control of the lac promoter in pAClacLink. Since the gene for CAT is 
650 bp, it is possible to separate singly and doubly cut vector (Figure 5), and the background rate of recircularization is and gel purification (data not shown).
This vector also allows us to get some idea of the expression level of a gene expressed from the pAClac vector under the synthetic lac promoter. The plasmid pACYC184, which has a p15A origin and CAT gene under the control of its native promoter from Tn9, allows survival of bacteria to between 35 and 70 µg/ml chloramphenicol. However, pAClacCm confers resistance to only 2–4 µg/ml chloramphenicol. Thus, the synthetic lac promoter is weaker than the Tn9 CAT promoter in pACYC184.
Answers to all your Biology Questions
