SlyD – Prolyl Isomerase and Folding Helper

In the first part of this thesis, the work focussed on structural properties of folding intermediate states. Therefore mutations were introduced to trap proteins in a partially folded state or NMR spectra in the presence of denaturant were recorded and analyzed, in order to fill the gap of structural information on folding intermediate states.

In the second part, we concentrate on enzymes, which prevent the accumulation of partially folded states by speeding up their folding reactions to the native state.

As mentioned in chapter 1.2, accumulation of folding intermediates raises the risk of misfolding and aggregation. To suppress this process and enhance productive folding, nature has evolved folding helper proteins. SlyD is one of them. It is an efficient peptidyl-prolyl cis/trans isomerase (PPIase) and shows in addition chaperone-like activities ¹²⁶. Already 15 years ago, SlyD was discovered in E. coli as a host factor for the ΦX174 lysis cycle ^127-130. Based on sequence comparisons, it was proposed that SlyD belongs to the FKBP family (one group of PPIases) ¹³¹, for which binding of the immunosuppressant FK506 is characteristic

132. Indeed, the isolated protein was highly active in prolyl isomerase assays. Further research results revealed chaperone activity for SlyD in an ATP independent manner.

SlyD is probably one of the most obnoxious proteins in molecular biology because it contains a cysteine- and histidine-rich tail, which is suggested to act as metal binding domain. Hence, recombinant proteins, overexpressed in E. coli and purified via IMAC chromatography are almost always contaminated with SlyD. Although various structure groups claimed that they have solved the structure of SlyD by accident, so far no structure files have been deposited in the pdb databank. The protein attracted attention in the last few years, because it could be shown, that covalent fusion of aggregation prone proteins with SlyD modules strongly enhanced cytosolic expression and solubility. Therefore, it became a valuble tool in diagnostic biotechnology ^{56; 133}.

Although this protein is well known, the mechanistic details of this enzyme are poorly understood. In addition, SlyD does not just consist of an FKBP domain, it also contains an 45 amino acid insertion in the flap region (according to the nomenclature of human FKBP12), close to the prolyl isomerase active site. The structure of SlyD from the thermophilic

organism Thermus thermophilus (tSlyD), which shows more than 50 percent sequence identity to E. coli SlyD* (1-165), but lacks the C-terminal cysteine- and histidine- rich tail, was solved by X-ray crystallography (Fig. 16). The deletion of this C-terminal part was shown to have no influence on the prolyl isomerase and chaperone activity in E. coli SlyD*.

tSlyD was also highly active in prolyl isomerase and chaperone assays. The crystal structure of tSlyD comprises a FKBP-like domain and the insertion in the flap folds into an autonomous domain (IF domain). Diffracting crystals of tSlyD were obtained under two different conditions in different space groups. Resulting structures of this two-domain protein showed variability regarding their domain orientation, while the overall shape of the domains was basically unchanged. Actually, the difference in domain orientation of these two structures is caused by a twist and short movement of the linker region, which orients the domains in closer proximity in one structure compared to the other (Fig. 16B-D). A detailed structure characterization is found in subproject D.

T. thermophilus SlyD E. coli SlyD*

T. thermophilus SlyD E. coli SlyD*

T. thermophilus SlyD E. coli SlyD*

A)

B) C) D)

IF domain

FKBP domain

Fig. 16. (A) Sequence alignment of tSlyD and E. coli SlyD* (1-165). Identical and similar residues between these two proteins are boxed or coloured in red, respectively. Both constructs carry a C-terminal histidine tag.

Histidine residues involved in Ni²⁺-ion binding are underlined. (B,C) Schematic representation of the crystal structure of tSlyD (3CGM.pdb, 3CGN.pdb in Protein Data Base) derived from two different crystal forms (crystal form A (B), crystal form B (C). The FKBP and the IF domain are indicated. (D) Superposition of the crystal structures of tSlyD demonstrates the different orientation of the two domains towards each other. The Cα

atoms of the FKBP domain were used for superimposition.

Deletion mutants and binding studies of tSlyD with various unfolded substrates followed by NMR, ITC, and fluorescence clearly showed, that the chaperone activity is located in the IF domain, whereas the FKBP domain exhibits the expected prolyl isomerase activity. Binding is mainly mediated via hydrophobic interactions (Fig. 17).

A)

∆ (ppm)δ

0.16 0.12 0.08 0.04

01 20 40 60 80 100 120 140

sequence B)

hydrophobic h.

QRRDFLKYSVALGVASALPLWSRAVFA

NH -2 -OH

hydrophobic h.philic

twin R (C)

5 4 3 2 1

[ligand] (µM)

fluorescence change (%)

0 10 20 30 40

- 100

Fig. 17. Mapping the tSlyD chaperone binding interface by NMR titration and fluorescence spectroscopy. (A) Residues of tSlyD showing chemical shift changes ∆δ > 0.08 ppm upon binding to RCM-α-lactalbumin are colored in red. (B) Backbone chemical shift changes (∆δ) of tSlyD upon binding to RCM-α-lactalbumin. (C) Sequence and schematic representation of the signalpeptide. Fluorescence increase of Trp 21 of the Tat-signalpeptide upon binding to tSlyD ({). tSlyD∆IF lacking the IF domain (z) did not bind.

Moreover, we could express both domains isolated in a folded and functional state and could directly assign the different functions (data not shown). The ¹⁵N-HSQC spectrum of tSlyD is a linear combination of the spectra of the isolated domains. By using the NMR H/D exchange technique, we were able to probe the local stability of tSlyD under native conditions.

Interestingly, amide protons of the FKBP domain, were, on average, 1000-fold more protected, than those of the IF domain. This proves, that the latter undergoes frequent local opening without unfolding the FKBP domain and is significantly less stable than the FKBP

In binding studies, single expressed domains did not interact with each other, indicating that both domains are highly flexible and can adopt different orientations in the full length protein.

This agrees well with our findings derived from the crystal structures of tSlyD and NMR structures of E. coli SlyD* (Weininger U., unpublished results). By using SAXS analysis, however, we were able to assign the most preferred domain orientation in solution to the structure of crystal form A, which shows a slightly kinked arrangement of the domains.

Interestingly, scattering curves for tSlyD in complex with the Tat-signalpeptide were not significantly altered when compared to the free form (Fig. 18). These data do not support a model where substrate binding orients both domains constantly in direct proximity.

s (Å )^-1

Intensity, relative

1000 100 10 1

0.010 0.1 0.2 0.3 0.4 0.5

A)

0.1

B) C)

Fig. 18. SAXS analysis of tSlyD. (A) Experimental intensities for free tSlyD (light grey) and a tSlyD/Tat-signalpeptide complex (black triangles). The scattering profiles are displaced along the ordinate for better visualization. The fit to the scattering pattern computed from the crystal structures of tSlyD is shown in black (chi = 1.61) (crystal form A, middle) and grey (chi= 2.43) (crystal form B, bottom). Ab initio low resolution structure models of tSlyD (B) and the tSlyD-Tat-signalpeptide complex (C) calculated from the SAXS pattern.

The balls represent the dummy atoms in the simulated annealing procedure to restore the models.

Nevertheless, the presence of the chaperone domain has an enormous impact on the activity of the prolyl isomerase. The catalytic efficiency towards proline limited refolding of ribonuclease T1 is 100-fold reduced, when the chaperone domain is absent in tSlyD (for details see subproject D). Therefore, a certain type of coupling of both domains must exist, to explain this dramatic activity change. This makes SlyD an interesting target to study because the combination of a catalytic domain with a chaperone domain is a common principle in nature. But in contrast to the most prominent member “trigger factor” ⁵⁴, which also exhibits prolyl isomerase and chaperone activity, the architecture of SlyD is much simpler .

The influence of the IF domain on a catalytic domain was further demonstrated by the chimeric protein “Thermus BP12”. This artificial protein consists of human FKBP12, but the original flap region (nine amino acid long), was replaced by the IF domain of tSlyD. The presence of the chaperone domain turned FKBP12 into an excellent catalyst of proline limited

protein folding, with a 180-fold increased activity compared to FKBP12 only. This chimeric protein is even more active than various SlyD species, tested from different organisms.

But what are the mechanistic details? Are there any limiting factors? The simplest explanation for the increased activity of chaperone domains containing prolyl isomerases, of course, comes along with the additional binding site for a substrate close to the active site. This leads to an increased local concentration and thus higher activity.

The question remains: Is it that simple? The different orientations of both domains found in crystal and NMR structures suggest a swinging arm like mechanism, where substrates bound to the IF domain are translocated close to the active site of SlyD by domain rearrangement.

Recent NMR dynamic studies on enzymes revealed, that the intrinsic dynamics of certain enzymes can be strongly coupled to the turn-over rate of their reactions ^134-136. To validate such a model for SlyD it will be of great interest but also very challenging to unravel, whether the domain dynamics of free SlyD, is somehow correlated to the turn-over rates of substrates.

Indeed, we do not expect a direct correlation of domain dynamics with the turn-over rate, rather propose, that the domain dynamics display an upper limit of reachable turn-over rates.

NMR R2-dispersion experiments in combination with single molecule FRET spectroscopy will be the most promising tools for future experiments.