Analysis of synthetic DZF domains using the B2H system

3.4.1 Interaction specificities of selected synthetic DZFs

To further analyze the specificity of the synthetic DZF variants, we decided to focus on four interacting pairs we identified in which at least one of the DZF variants contains subdomains derived from wild-type DZFs that do not interact with each other (Table 3.2, highlighted in red; note that these four pairs comprise a total of 6 different DZF domains). The specificity profiles of the DZF domains in these four pairs were examined using the B2H system as a reporter.

Fold-activation of β-galactosidase expression Fold-activation of β-galactosidase expression

Fold-activation of β-galactosidase expression

Fold-activation of β-galactosidase expression Fold-activation of β-galactosidase expression

Fold-activation of β-galactosidase expression Fold-activation of β-galactosidase expression

A B

C D

Figure 3.7 Interaction specificities of synthetic DZFs analyzed in the B2H system. DZFs from four pairs (A, B, C and D, respectively) were tested for hetero- (blue bars) and homo-typic (green bars) interaction as well as their ability to interact with the parental wild-type DZFs from which they were derived (orange bars). Assays were performed using the B2H system. Control experiments expressing combinations of the synthetic DZFs with either the Zif268 or the α-subunit protein with no DZF fused (indicated as ‘----’) are also shown (purple bars). Bars shown represent mean fold-activations of transcription in the B2H system calculated from three independent experiments. Standard error of the mean is indicated by the error bars. Abbreviations for identities of the synthetic DZFs are defined in Figure 3.4.

We assayed the ability of each of the six DZF domains in the various pairs to mediate hetero- and homo-typic interactions and also assessed their interactions with the wild-type DZFs from which they derived. It was found that for all four heterodimeric pairs, one DZF domain in the pair is not able to interact with itself (homotypic interaction) nor does it interact with its parental wild-type DZFs, while the second domain does both interact with itself and with its parental DZFs (Figure 3.7). The latter DZF domain is the one consisting of subdomains from wild-type DZFs which are known to interact with each other. Closer inspection of the activity profile suggest that three of the four DZF pairs tested prefer to hetero-dimerize, although one DZF domain in each pair can still mediate homodimerization. The fourth pair contained one DZF domain (Pe-Hd-Hd) that mediates both homo- and hetero-typic interactions equally well while the other pair (Hl-Eo-Eo) mediated only heterotypic interaction.

To further analyze the specificity of these six synthetic DZF domains, we tested for potential cross-interactions among them using the B2H system. The result of this experiment is summarized in Table 3.3 and indicates that all six DZF domains interact most strongly with the partners they were selected with, although some DZF domains also exhibit some

cross-1.7

Table 3.3 Analysis of cross-interactions among the six synthetic DZFs in the B2H system. Pairs of plasmids encoding the six synthetic DZFs fused to either the Zif268 DBD or the RNAP α-subunit were co-transformed into the B2H reporter strain harboring the lacZ gene and reporter gene expression was assessed by performing β-galactosidase assays. Each combination was tested in triplicates with averages and standard errors of the mean shown. Positive interactions (as defined in Figure 3.4) are marked in colored text. Bold red text indicates hetero-typic interactions between DZFs that were selected with each other and red text depicts homo-typic interactions between these DZFs. Blue bold text indicates cross-reactivity between DZFs that were identified in different selection sets. Identities of subdomains in the synthetic DZFs are abbreviated as in Figure 3.4.

reactivity and bind to at least one other synthetic DZF. Note that these DZFs were obtained from different selection sets and could not have been selected in these combinations.

Control experiments expressing either the Zif-hybrid or the α-hybrid protein alone did not show increased β-galactosidase Units (Figure 3.7). Furthermore, the previously described D18Q mutation was introduced into both the synthetic and the wild-type DZFs. This mutation has been shown to disrupt homodimerization of the Ikaros and Hunchback DZF domain (McCarty et al., 2003). All DZF domains with the exception of wild-type Pegasus were affected by this mutation in their ability to mediate homo- and heterotypic interaction (data not shown). This indicates that both interacting DZFs have to be intact and present to activate the expression of the lacZ reporter gene.

3.4.2 Anti-parallel interaction mode for synthetic DZFs

Analysis of the identities and homo- and heterotypic interaction specificities of the synthetic DZFs suggests that these domains interact in an anti-parallel fashion (i.e. the N-terminal finger of one monomer interacts with the C-terminal finger in the other monomer and vice versa). As mentioned in Chapter 1 (section 1.3.4.3), previous studies suggested that DZFs dimerize in a parallel mode (i.e. the N-terminal C2H2 ZFs in each monomer interact with each other and C-terminal C2H2 ZFs in each monomer interact with each other) (McCarty et al., 2003). This conclusion was based on the observation that a synthetic Ikaros-Hunchback hybrid DZF could efficiently homo-dimerize and on the assumption that the individual C2H2 ZFs in this synthetic hybrid display the same interaction specificities then the parental DZFs from which they were constructed (McCarty et al., 2003). This analysis was applied to the various DZFs identified in the selections. Six pairs of synthetic DZF domains are composed of subdomains that indicate a specific orientation: five pairs suggest an anti-parallel orientation while one pair is more consistent with a parallel interaction mode (Figure 3.8 and Table 3.2). The remaining pairs of chimeric DZFs contain subdomains that do not provide information in terms of an interaction mode since they can interact using either a parallel or an anti-parallel mode (Table 3.2).

However, additional confirmation of an anti-parallel interaction mode is provided by the specificity profiles of the four synthetic pairs. As mentioned above three of the four pairs prefer to interact in a heterotypic fashion and one DZF domain in each pair is not able to homodimerize at all which also rules out a strictly parallel interaction. For example, Ik-Ik-Hd and Pe-Pe-Eo can mediate robust heterodimerization but each individual DZF in this pair interacts less (Pe-Pe-Eo) or not at all (Ik-Ik-Hd) with itself.

3.4.3 Prediction and design of interactions between DZF domains

Since preliminary results demonstrated that shuffling of DZF-derived C2H2 ZFs can yield synthetic DZFs with new specificities we were also interested in creating new DZF domains by mixing certain ZFs of interest together. We decided to test whether the anti-parallel interaction mode could be used as a general rule for engineering DZF-mediated interactions.

To do this, all combinations of C2H2 ZFs and linkers from the Ikaros and Drosophila Hunchback DZFs were systematically synthesized applying the same approach used to construct our libraries. Connecting these Ikaros-Hunchback chimeras to both the Zif268 as well as the α-subunit of the RNAP resulted in eight different fusion proteins that were tested for their ability to interact with each other using the B2H system as a reporter. It was found that none of them showed a significant interaction (data not shown). On the contrary, a functional chimeric protein containing portions of the human Ikaros and Drosophila Hunchback DZFs has been described recently (McCarty et al., 2003; see also section 1.3.4.3).

Shuffled

Figure 3.8 Schematic indicating how interaction specificities of synthetic DZF domains suggest an anti-parallel interaction mode. Wild-type DZF domains used to construct the synthetic DZFs and their interaction specificity are shown on top of the Figure. Interaction specificity of a pair of synthetic DZFs is shown on the bottom (left side). Occurring interaction suggests an antiparallel interaction mode for the interaction mediated by the two synthetic DZFs.

Interactions are indicated by an arrow while red X’s indicate that no interaction was obtained. Note that in this presentation the linker region was assigned to a certain finger but can also occur in the opposite constellation. See text for details.

The reason for this discrepancy remains unknown but in the approach of McCarty and coworkers (2003) the breakpoint for separating the protein into two subdomains was defined differently. Thus, residues from the respective other subdomain were included into their chimera which were not included in our chimeric peptides. It is also important to note, that we tested this chimeric protein in the B2H system and found that it interacted only weakly as judged by our assay.

One possible explanation for these findings is that DZF-derived C2H2 ZFs can not be considered as completely modular since not all combinations of fingers can produce functional interaction domains (see also section 3.8.1 below). This rules out that interactions between DZF domains can be consistently predicted and therefore deliberately designed and further emphasizes that methods such as selections currently provide the only way to obtain such synthetic interacting DZF pairs.