Attempted purification of single finger domains from DZFs

Since the purification of Zif268-DZF fusion proteins proved to be very challenging we took a different approach and decided to attempt to purify single finger domains derived from various DZFs for use in X-ray crystallography studies. We constructed four pET3a derived expression plasmids which expressed single N- or C-terminal ZFs of the Ikaros and the

A B

C A

B

C

Figure 5.8 Analytical RP-HPLC trace for the Zif268-Pegasus fusion protein. Bound peptides were eluted over a ~60 min period using a Buffer B gradient (shown in black). UV trace is shown in blue.

Collected peaks are numbered A, B, C but did not contain the expected peptide.

Hunchback Drosophila DZF domains (Table 5.1). These plasmids were introduced into the E. coli strain BL21(DE3)pLysS. To test, whether any of these peptides could be overexpressed we examined whole cell lysates from uninduced and IPTG induced cells using SDS PAGE and found that the peptides could not be appreciably detected and are therefore not expressed at high levels (data not shown). Efforts to overcome this problem by using different concentration of IPTG, increasing and decreasing induction time or adding glucose were not successful and the expression level could not be improved (note that plasmids containing toxic target genes can be destabilized in stationary phase cultures by expression of the T7 RNA polymerase. This expression is initiated by cAMP mediated derepression of the lacUV5 promoter and can therefore be avoided by adding glucose to the media which inhibits cAMP production).

5.6 Discussion

In order to obtain structural information about a C2H2 ZF mediated protein-protein interaction, we attempted to purify the DZF domains from several Ikaros/Hunchback transcription factors for use in crystallographic studies. When overexpressed, these DZFs accumulated in inclusion bodies and could be solubilized using denaturants (urea) and reducing agents (DTT). Further purification was successfully achieved by ion-exchange and RP-HPLC chromatography and resulted in a very clean sample containing only the respective DZF domain as judged by silver staining. Simultaneously, the denaturant was removed during the HPLC run to permit subsequent refolding of the solubilized proteins. To perform the folding reaction, the peptides were rapidly diluted into folding buffer containing either cobalt or zinc. However, although highly purified peptides were successfully obtained, various attempts to refold these peptides into active domains resulted in the formation of precipitates containing the various DZFs. We can not rule out that aggregation of these DZF peptides is due to an unknown misfolding event but believe that these peptides may just fold fine (as indicated by the blue color). The correctly folded DZFs may aggregate due to the dimerization surface, and DZFs that are usually not able to mediate dimerization may actually dimerize at these high protein concentrations (that are present in the folding reactions).

Although we do not know the precise multimerization state of the various DZF domains,

aggregation may just be a consequence of the formation of higher order oligomers as previously described for Eos (Westman et al., 2003) and Ikaros (McCarty et al., 2003).

In general, misfolding as well as aggregation competes with the correct folding event resulting in a dilution of the amount of active peptides. For example, aggregation processes can be caused by nonspecific, hydrophobic interactions of mainly unfolded polypeptide chains. Moreover, correct folding of peptides also depends on correct regeneration of covalent disulfide bonds and it is known that the presence of free thiols can cause complications during the re-folding process due to oxidation problems (reviewed in Rudolph and Lilie, 1996). Thus, various properties of the peptide to be folded can influence the folding process and have to be evaluated. In fact, analysis of the amino acid sequences of the various DZFs that were unsuccessfully refolded indicates that they all contain a highly hydrophobic N-terminal finger consisting of numerous aromatic and aliphatic residues. These residues may have decreased the solubility of the peptides during refolding which in turn resulted in aggregation of folding intermediates. Introducing silent mutations into these hydrophobic residues may help preventing aggregation. In addition, at least one cysteine residue was present in these DZFs predicted not to be involved in zinc ion co-ordination. These cysteines provide free thiol groups that may have formed incorrect disulfide bonds resulting in misfolded peptide aggregates (reviewed in Rudolph and Lilie, 1996). Thus, it may be reasonable to silently mutate all cysteine residues that are not involved in zinc binding.

Several attempts to address the problem of aggregation were undertaken that were all aimed to directly influence aggregation during the folding event. For example, because aggregation usually appears at high protein concentrations (reviewed in Rudolph and Lilie, 1996;

reviewed in Rudolph et al., 1998; reviewed in Clark, 1998), renaturation was performed at high dilutions of the protein. Furthermore, various refolding conditions were tested including variables such as buffer composition (pH and ionic strength of the folding buffer) and temperature. Several additives know to enhance the folding process were added to the folding reaction as well. Examples are urea (Orsini and Goldberg, 1978; Maeda et al., 1996), ionic and non-ionic detergents (Tandon and Horowitz, 1987) and sugars (Maeda et al., 1996; Ahn et al., 1997). Although the precise mechanism of action for these additives is not known they are believed to prevent aggregation by either destabilizing incorrect folded intermediates or by stabilizing the correct folded product (reviewed in Clark, 1998). However, none of the tested additives in combination with several buffer conditions was able to decrease the amount of aggregated DZF domains. Other additives such as L-Arginine/HCl (Buchner and

Rudolph, 1991; Brinkmann et al., 1992) or the use of chaperones (Thomas et al., 1997;

Altamirano et al., 1997) have also been described as enhancers of the folding reactions but were not tested in this study.

However, since none of the attempts to directly influence the success of the folding reaction worked, it is very likely that this problem can only be solved by using a different folding method. Various methods for refolding of proteins have been described and may be helpful for future attempts to renature the DZF domain: Dialysis is probably the most common method to remove denaturing and reducing agents and therefore allowing the peptide to renature. In doing so, the concentration of the solubilizing agent decreases slowly which allows the protein to refold properly (reviewed in Clark, 1998; reviewed in Rudolph and Lilie, 1996). Another method to remove the denaturant is pulse renaturation. Here, aliquots of denatured protein are added to the renaturation buffer at defined time points, so that the concentration of unfolded protein is kept low. This strategy is based on the observation that during refolding only the concentration of unfolded and not that of correct folded protein is critical for aggregation. The process is stopped when the concentration of denaturant reagent introduced into the renaturation buffer reaches a critical level at which even native peptides tend to aggregate (reviewed in Lilie et al., 1998).

Finally, one can try to address the solubilization problem right at the beginning by avoiding a situation where the DZF peptides aggregate in inclusion bodies in the first place. ZF proteins are generally insoluble but can be linked to a soluble peptide that forces the fusion peptide to stay in solution. In fact, previous studies described various DZF domains fused to the maltose-binding protein (MBP) which were soluble and could be used for biochemical analysis (Westman et al., 2003). In addition, subsequent studies done by this group using single C2H2 ZFs of Pegasus and Eos fused to glutathione S-transferase (GST) resulted in the NMR structure of the C-terminal ZF of the DZF domain from Eos (Westman et al., 2004).

However, despite extensive efforts this group did not succeed in crystallizing the complete Eos DZF domain or oligomers generated by this domain. In addition, other groups reported similar technical difficulties including the insolubility of DZFs (Westman et al., 2004, 2003;

McCarty et al., 2003). Thus, the crystallization of the DZF domain has proven to be very challenging and will remain the goal of future studies.

Chapter 6. Functional analysis of the Hunchback DZF domain in