Báo cáo khoa học: The crystal structure of human WD40 repeat-containing peptidylprolyl isomerase (PPWD1) pdf

In the crystal, the N-terminus of one isomerase domain is bound in the active site of a neighboring isomerase molecule in a manner analogous to substrate.. Results The crystal structure

peptidylprolyl isomerase (PPWD1)

Tara L Davis1,2, John R Walker1, Hui Ouyang1, Farrell MacKenzie1, Christine Butler-Cole1,

Elena M Newman1, Elan Z Eisenmesser3and Sirano Dhe-Paganon1,2

1 Structural Genomics Consortium, Banting Institute, University of Toronto, Canada

2 Department of Physiology, University of Toronto, Canada

3 Department of Biochemistry and Molecular Genetics, University of Colorado, Denver and Health Sciences Center, Aurora, CO, USA

Cyclophilins (Cyps) are one of three subfamilies of the

peptidylprolyl isomerases (PPIases; E.C 5.2.1.8),

together with the structurally unrelated FK506-binding

proteins and parvulins [1,2] Biologists and clinicians

initially focused on the specific and high affinity of

PPIases for the immunosuppressants ciclosporin A,

FK506 and rapamycin [3–5] These immunosuppressive

effects were eventually determined to be uncoupled

from the enzymatic function of the PPIases, which

involves the reversible cis–trans isomerization of Xaa–

Pro peptide bonds [2,6] The evolutionary importance

of this fold and⁄ or function may be inferred from the

broad distribution of PPIases throughout Eukaryota,

Eubacteria, Archaea and, recently, a viral genome

[2,7,8] However, there is very little direct evidence

to show that the enzymatic function encoded by the

PPIases is the only, or even the primary, physiological role of these proteins in cells There are a subset of proteins whose association with Cyps is necessary for correct folding and structural integrity: for instance, maturation of steroid receptor complexes in conjunc-tion with Hsp90⁄ Hsc70 [9–11] and the function of

Nin-aA in Drosophila rhodopsin folding [12] Cyp-catalyzed isomerization can also play a part in host response events, including the participation of host CypA in binding the capsid protein Gag of HIV-1 during infec-tion in humans and packaging into HIV virions, and the association of a host CypB–viral polymerase com-plex leading to viral replication during infection by hepatitis C [13,14]

However, more recent studies of PPIases have focused on their roles as signal transducers rather than

Keywords

crystal structure; cyclophilin; peptidyl-prolyl

isomerase; spliceosome; WD40

Correspondence

S Dhe-Paganon, Structural Genomics

Consortium, Banting Institute, University of

Toronto, 100 College Street, Room 511,

Toronto, ON, Canada M5G 1L5

Fax: +1 416 946 0588

Tel: +1 416 946 3876

E-mail: sirano.dhepaganon@utoronto.ca

(Received 18 January 2008, revised 3 March

2008, accepted 6 March 2008)

doi:10.1111/j.1742-4658.2008.06381.x

Cyclophilins comprise one of the three classes of peptidylprolyl isomerases found in all eukaryotic and prokaryotic organisms, as well as viruses Many of the 17 annotated human cyclophilins contain the catalytic domain

in tandem with other domains, and many of the specific functions of a par-ticular cyclophilin or its associated domains remain unknown The struc-ture of the isomerase domain from a spliceosome-associated cyclophilin, PPWD1 (peptidylprolyl isomerase containing WD40 repeat), has been solved to 1.65 A˚ In the crystal, the N-terminus of one isomerase domain is bound in the active site of a neighboring isomerase molecule in a manner analogous to substrate NMR solution studies show that this sequence binds to the active site of the cyclophilin, but cannot be turned over by the enzyme A pseudo-substrate immediately N-terminal to the cyclophilin domain in PPWD1 could have wider implications for the function of this cyclophilin in the spliceosome, where it is located in human cells

Abbreviations

Cyp, cyclophilin; IC 50 , 50% inhibitory concentration; pNA, p-nitroaniline; PPIase, peptidylprolyl isomerase; PPWD1, peptidylprolyl isomerase containing WD40 repeat; suc, succinyl.

as chaperones or protein foldases Specific examples

include the participation of CypD in the mitochondrial

permeability transition, leading to necrotic cell death

[15], the isomerase-dependent regulation of ligand

specificity for the SH2 domain of the non-receptor

tyrosine kinase Itk [16,17], and the role of multiple

Cyps in regulating transcriptional and spliceosomal

events [18–20] In these cases, the substrate proteins

are already folded and performing a subset of their

normal functions; an isomerase interacts with this

sub-strate conformation, thereby allowing for a new set of

molecular interactions to occur with the product

con-former [1,17] The molecular switch function of Cyps

may well have different sequence determinants than

their isomerization function, but very few studies have

been performed to probe the sequence specificity of

binding versus catalysis of Cyps Early evidence

indi-cated that CypA is able to catalyze the isomerization

of a wide array of Xaa-Pro sequences with nearly

iden-tical catalytic efficiency [21] In contrast, phage display

and subsequent amino acid substitution analysis to

identify CypA binding specificity identified a clear

preference for sequences N- and C-terminal to the

Xaa-Pro target culminating in a consensus sequence of

FG*PXLp This work indicates that preferential

bind-ing of target proteins may be dictated by a select

sub-set of amino acids distinct from sequences that are

substrates for catalysis [22] The relevance of this

find-ing is dependent on studies showfind-ing an in vivo context

for peptides capable of binding to PPIases without

being used as isomerization substrates; to date, no

such peptide sequence has been described in the

liter-ature

Further complicating the field of Cyp biology is the

fact that many of the 17 Cyps annotated in the human

genome have not been thoroughly studied in terms of

either enzymatic or other functional significance For

some of these Cyps, their location inside the cell

provides the only clue as to their function The large

subset of Cyps found to be stably associated with

spliceosomes provides an example of the issues

involved in studying the complexity of Cyp function

Spliceosomes are multi-megadalton complexes

contain-ing hundreds of proteins and five essential small

nuclear RNAs, whose function is to excise out

non-coding regions (introns) of translated pre-mRNAs

[23,24] Recent technological advances in the

purifica-tion of spliceosomal complexes, coupled with advances

in mass spectrophotometric methodology, have led to

a massive increase in the identification of proteins

found to be associated with spliceosomal complexes

[25–28] At least 11 of the 17 human PPIases have

been found to be associated with intermediate

splice-osomal subcomplexes [26,29] One of the splicesplice-osomal Cyps was identified using yeast two-hybrid screens for known spliceosomal components; another was recog-nized as a spliceosomal component based on sequence homology to an SR repeat containing splicing factors [20,30] For two other spliceosomal Cyps, functional and structural studies eventually identified their cog-nate spliceosomal binding partners: PPIL1 binds to the SNW domain-containing protein 1 (SKIP), and CypH binds to the small ribonucleoprotein Prp4 [18,31–33] Interestingly, in both of these cases, the interaction with spliceosomal components was found on surfaces not involving the isomerase active site and did not affect cis–trans isomerase activity, indicating that the function of the isomerase in the spliceosomal complex may involve simultaneous active site and second-site interactions The binding partners for the other seven spliceosomal Cyps are undetermined Moreover, the physiological function of the spliceosomal Cyps remains unclear, which perhaps is not surprising, con-sidering that so many proteins encoding the same enzymatic function are found simultaneously in splice-osomal complexes It is theorized that this high level

of complexity and seeming duplication of effort in the spliceosomal complexes are a function of the exquisite sophistication of dynamic networks needed to properly regulate and proofread the splicing process [23,34] PPWD1 (peptidylprolyl isomerase containing WD40 repeat) was cloned in 1994 [35] and later purified as part of the catalytically competent form of the spliceo-some C complex [26] This polypeptide encodes an N-terminal WD40 repeat domain and a C-terminal domain homologous to Cyps As part of an attempt to structurally characterize the spliceosomal Cyps, we have determined the high-resolution X-ray crystal structure of the isomerase domain of PPWD1 and monitored its activity via both UV kinetics and NMR solution experiments In this structure, PPWD1 forms distinct intermolecular interactions within the asym-metric unit with an internal peptide containing a Gly–Pro sequence Interestingly, the Pro residue is found in trans, an unusual circumstance for a substrate peptide Further experiments have shown that PPWD1

is indeed a functional isomerase against a standard substrate sequence, but that, surprisingly, a peptide containing the internal sequence is able to bind, but is not catalyzed by the isomerase, suggesting that it is not a substrate Both the intermolecular interaction and lack of enzymatic turnover were confirmed using NMR solution studies of the PPWD1 protein This work represents the first structural and biochemical characterization of a WD40 repeat-containing spliceos-omal Cyp, and the first instance of a Pro-containing

sequence that is sufficient to bind specifically to the

active site of a Cyp, but is not a substrate for cis–trans

isomerization

Results

The crystal structure of the isomerase domain of

human PPWD1 (utilizing a construct encoding residues

473–646) was determined at 1.65 A˚ resolution The

final model of PPWD1 comprises three polypeptide

chains; each chain is disordered from residues 473–482,

such that the first interpretable density is for residue

Gln483 All other residues encoded by the PPWD1

construct are present in the final model In addition,

the model contains a single glycerol molecule most

probably contributed from the cryoprotectant and 388

water molecules With an architecture consisting of

eight antiparallel b-sheets forming a closed b-barrel

with two a-helices packed against the core (Fig 1A),

the structure of PPWD1 is similar to the structure of

the canonical CypA as well as to CypH and PPIL1,

two other spliceosomal Cyps (Fig 1B) The Ca atoms

of CypA align to the isomerase domain of PPWD1

with an rmsd of 1.34 A˚ over 124 atoms, corresponding

well to the 60% sequence similarity between the two

isomerase domains The main conformational

differ-ences lie in the a2–b8, b1–b2 and b4–b5 loops The

b1–b2 loop (corresponding to residues Thr498–

Asp502) is shorter by five residues in PPWD1

com-pared with CypA This deletion also occurs in six of

the 17 currently annotated human Cyps, including

PPIL1–PPIL4 (Fig 2), but the significance of the

resul-tant shorter b1–b2 loop is not understood

PPWD1 crystallized with three protein molecules in

the asymmetric unit Unexpectedly, there was an

inter-molecular interaction, propagated throughout the

crys-tal, in which the active sites of all three molecules in

the asymmetric unit were bound to seven residues

(QAEGP487KR) of an adjacent molecule (Fig 1C)

The three polypeptide chains in the asymmetric unit

are conformationally identical, with rmsd values of 0.4

and 0.2 A˚ over all atoms in 176 residues In addition,

the N-terminal peptide is oriented in very similar

fash-ion across all three molecules, also with rmsd values of

0.4 and 0.2 A˚ over all atoms for the first seven

resi-dues Pro487 from one monomer is buried within the

active site of another, and is less than 3 A˚ from the

side-chain of the conserved Arg535 (Arg55 in CypA)

(2.95 A˚ for NH1 and 2.85 A˚ for NH2) In addition,

Pro487 of the peptide is bound in a hydrophobic

pocket composed of Phe540, Phe593, Met541, Leu602

and His606, all of which are conserved between CypA

and PPWD1 (Figs 2 and 3) The Nd1 atom of the

conserved Trp602 (Trp121 in CypA) is coordinating the backbone oxygen of Lys488 of the QAEGPKR peptide (Fig 3A), and Phe540 is coordinating Lys488 through a stacking interaction The other interactions between the side-chain of the QAEGPKR peptide in the active site are centered about residue Glu485; in addition to specific interactions between the backbone nitrogen of Glu485 with the oxygen of Gly551, the side-chain of Glu485 is nestled in a deep pocket of the enzyme (Fig 3B) An oxygen of the c-carboxyl group forms hydrogen bonds with the backbone amide nitro-gen of Asn582 and a water molecule buried in the pocket; the other c-carboxyl oxygen is involved in a network of water-mediated hydrogen bonds Two resi-dues in the PPWD1 active site, Gln543 and Gln591, contribute to the complementary polar nature of this pocket The average B-factor for all three of the QAEGPKR regions is 48 A˚2, and is comparable with

an all-atom B-factor of 45 A˚2 for all molecules in the asymmetric unit These observations indicate a stable interaction, although PPWD1 runs as a monomer using size exclusion chromatography, which suggests that the interaction is either low affinity or has a high off rate

This mode of interaction and the trans conformation

of the peptide in the active site mimic the enzyme–sub-strate interaction observed in the CypA–HIV-1 Gag complex [36] Pro487 is found to be bound to PPWD1

in the trans configuration, which is also the conformer found in the complexes of the capsid protein Gag and

in some peptides derived from GAG protein with CypA For these complexes, the requirement for an X-Gly-Pro (X „ Pro) sequence was proposed to obtain stable binding of the trans isomer into the Cyp active site; the PPWD1 structure confirms this observa-tion and, indeed, any other amino acid at that posiobserva-tion would adopt /,w angles such that Cb would clash with the catalytic Arg535 and destabilize the trans confor-mation [36] Although the backbone amide of the )2 position has been shown to be involved in a configura-tion-dependent hydrogen bond with the b4–b5 loop, the function of the side-chain at this position has not been exhaustively studied As opposed to the mainly hydrophobic residues found in the context of HIV-1 capsid variants (containing Ala, Val, or Met at posi-tion )2), PPWD1 contains Glu at this position (Glu485), which is pointing into a region of charge complementarity in the PPWD1 active site that would not be well accommodated by hydrophobic side-chains The relevance of finding a peptide bound in trans in the active site of isomerases is ambiguous; in the case of the HIV capsid, it has been proposed that

an X-Gly-Pro sequence represents a poor substrate for

catalysis, explaining the capture of the trans conformer

in the crystal [36] However, as the HIV capsid protein

is indeed an efficient substrate for isomerization by

CypA [37], it cannot yet be stated definitively that

Gly-Pro sequences are the minimal determinant for

binding versus catalysis for Cyps, or what role sequence variation amongst the Cyps may play In summary, the interactions seen between the QAEGPKR peptide and PPWD1 active site in the cur-rent structure might be a product of an inability of the

Fig 1 Structure of PPWD1 and comparison with other Cyp structures (A) The structure of the isomerase domain of PPWD1 is shown in ribbon format The N-terminus and secondary structural elements are labeled (B) Structural alignment of PPWD1 (magenta) with Ca stick representations of the canonical CypA (2RMA) in forest green and the spliceosomal CypH (1QOI) in light green Unless otherwise men-tioned, all figures were generated using PyMOL (http://pymol.sourceforge.net/) (C) The arrangement of molecules in the asymmetric unit in the PPWD1 crystal N- and C-termini are labeled The N-terminal peptide is bound in the active site of an adjacent protein The homotypic interaction between the N-terminal peptide and the active site of a neighboring molecule could be a crystal contact; that is, it may be an arti-fact of the crystallization condition Unfortunately, crystallographic methods cannot distinguish between crystal contacts and specific pro-tein–protein interactions; however, the results presented here indicate that the homotypic interaction modeled in the crystal structure is supported by solution methods.

Fig 2 PPWD1 domain structure and sequence alignment of selected human isomerase domains Top: a graphical overview of the PPWD1 gene The internal peptide sequence QAEGP487KR is highlighted Bottom: an alignment of selected human isomerase domains; output is from MULTALIN [52] Residues conserved at 90% or more are indicated by capital letters, and between 50% and 90% by lower case letters; (I ⁄ V) conserved positions are indicated by !; (N ⁄ D ⁄ E) conserved positions are indicated by # All Cyps shown in this alignment (with the exception of CypA) have been experimentally described as being associated with spliceosomes, as referenced in the text Green boxes high-light the conserved positions in the active site of CypA [2]; blue boxes outline the positions of the spliceosomal binding site on CypH [32]; pink boxes outline the main spliceosomal interaction regions on PPIL1 [31] Asterisks (*) and crosses (+) above the alignment indicate the positions of the S1¢ and S2¢-S3¢ subsites as described in [38].

enzyme to catalyze the isomerization of this particular

sequence; alternatively, the trans conformer of the

sub-strate Gly–Pro sequence might simply have a higher

affinity for the PPWD1 binding pocket Further

analy-sis is therefore necessary to attempt to distinguish

between these two possibilities

To determine whether PPWD1 is catalytically

com-petent to isomerize Cyp substrates, we conducted

bio-chemical and biophysical assays Using a colorimetric

coupled assay against

succinyl-Ala-Ala-Pro-Phe-p-nit-roaniline (suc-AAPF-pNA), a well-characterized CypA

substrate, we found that the isomerase domain of

PPWD1 (using the crystallographic construct

compris-ing residues 473–646) is indeed an active isomerase

with a catalytic efficiency of 1.5 lm)1Æs)1, similar to

the values obtained in previous studies (Fig 4A)

[21,38] This is not surprising as the active site residues

of PPWD1 are nearly completely conserved when

com-pared with CypA (Fig 2) Furthermore, we found that

ciclosporin A binds tightly to PPWD1 with a 50%

inhibitory concentration (IC50) between 1 and 2 nm, a

value similar to that for CypA (Fig 4B) These

enzy-matic characteristics are not significantly different for

a truncated PPWD1 construct without the N-terminal

sequence (encoding residues 493–646), implying that

this region does not interfere with the active site at the

pico- to micromolar concentrations of protein used in

the assay As a result of the technical limitations of the

enzyme coupled assay described above, we may not

have detected enzymatic activity against low-affinity substrates in the millimolar KD range, nor could we detect binding without catalysis To address this issue,

we conducted direct NMR measurements, as described previously [39] PPWD1 (residues 473–646) was active

on the standard Suc-AAPF-pNA substrate, as indi-cated by a collapse of the peaks contributed by the cis and trans conformers (caused by enzyme-catalyzed isomerization that is rapid compared with the chemical shift differences between the cis and trans resonances) (Fig 4C, red peaks) PPWD1 also bound to a syntheti-cally derived QAEGPKR peptide, as shown by the small shifts for resonances on addition of the enzyme, especially for the +2 Arg resonances, which correlate well with the Arg interactions seen in the crystal struc-ture However, PPWD1 did not catalyze the isomeriza-tion of this peptide sequence, as it does for the model peptide substrate AAPF (Fig 4C, black peaks; Fig 4D) Finally, NMR-based experiments were undertaken to validate the intermolecular association

of PPWD1 in the context of the protein construct used

to obtain the crystal structure.1H,15N-HSQC measure-ments were conducted on the PPWD1 construct which crystallized (residues 473–646), as well as an N-termi-nally truncated construct (residues 493–646) As expected, the overall spectra of these two constructs were similar However, it is clear that the line widths

of most resonances in the spectrum of the longer con-struct are broader than those in the shorter concon-struct

Fig 3 Details of QAEGP 487 KR peptide interaction (A) Cartoon model of PPWD1 and stick representation of the amino terminal peptide Active site residues are shown in stick format and water molecules are shown as spheres Note the water molecules in a hydrogen bonding network with Glu486 of the N-terminal peptide (B) Electrostatic surface representation (+10 kTÆe)1, red; )10 kTÆe )1, blue) of PPWD1,

gener-ated by the APBS software package [53], showing the side-chains of Pro487 and Glu486 buried in deep pockets formed by the PPWD1 active site.

(Fig 5) The simplest interpretation of this line

broad-ening is that it is a result of a weak interaction

between protein molecules, and it is quite possible that

the interaction seen in solution using NMR is caused

by the specific interactions captured in the crystal

structure Taken together, these data suggest that the

intermolecular interaction trapped in the crystal

struc-ture can indeed be recapitulated using solution

meth-ods, and that this intermolecular interaction is a

consequence of binding, but not efficient catalysis, of

the QAEGPKR sequence by PPWD1

Discussion

The QAEGPKR peptide is capable of binding the PPI domain of PPWD1 without being an efficient substrate for cis–trans isomerization It is clear from earlier work that there is very little selectivity for substrates at the )1 or +1 position in the Cyp active site, with the caveat that Gly–Pro may pro-mote the shift of equilibrium binding in the active site to the trans over the cis conformer (although with very little difference in catalytic efficiency)

Fig 4 PPWD1 is an active isomerase and binds to QAEGPK, but does not catalyze its isomerization (A) Increasing amounts of PPWD1 accel-erate the isomerization of suc-AAPF-pNA in a standard colorimetric assay The bottom curve (squares) shows the unaccelaccel-erated cleavage of peptide relative to the catalyzed reaction (top curve, circles) (B) Ciclosporin A inhibits PPWD1 activity against suc-AAPF-pNA (C) Binding and catalysis of isomerization for the standard substrate suc-AAPF-pNA Black and red spectra of the Ala residues in the peptide are obtained from the uncatalyzed and catalyzed reaction, respectively Acceleration of the cis–trans isomerization of the peptide results in the collapse of these resonances into a single set of peaks (D) Binding, but not isomerization, of the N-terminal peptide of PPWD1 A synthetic peptide correspond-ing to the seven residues seen in the crystal structure (QAEGPKR) was added to the PPWD1 protein (residues 473–646) to assess bindcorrespond-ing as

in (C) Notice that the peak shift observed for Arg resonances is more pronounced than that for Lys (labeled with R and K, respectively), indicat-ing a change in the chemical environment for this residue; this is confirmed by the model of the crystallographic data, which shows Arg489 pointing into the active site of the isomerase, whereas Lys488 points into the solvent There are chemical shifts on addition of the enzyme, but

no collapse of cis and trans peaks, suggesting that PPWD1 binds the N-terminal peptide but does not catalyze isomerization.

[21,22,38] Therefore, it is reasonable to look

N-ter-minal and C-terN-ter-minal to these positions to find

potential binding specificity in Cyps Previous studies

have delineated binding determinants for Cyps

con-tained outside of the )1 and +1 positions using

dif-ferent methodologies For instance, by comparing

binding affinities with the ability to inhibit PPIase

activity for peptide-based inhibitors, it was found

that the active site of CypA could be separated into

two distinct subsites (called S1¢ and S2¢–S3¢),

pre-sumably indicating binding determinants on the

pro-tein distal to the )1 position [38] In addition, a

phage display experiment, which independently

con-firmed the preference for binding of Gly–Pro at the

central position, described a binding preference for

Phe at the )2 position and various amino acids at

+3 and +4 for CypA [22] These studies indicated

that there might be separate sequence determinants

for binding that are stricter than those for substrate

turnover, and that these determinants are somehow

dictated by residues outside the minimal active

sur-face of the Cyps The structure of PPWD1 indicates

that the isomerase domain may bind tightly to a

subset of protein targets with a polar side-chain at the )2 position, without losing the ability to be an efficient isomerase for substrates containing alterna-tive residues at this position (A⁄ G-P, for instance),

as opposed to the hydrophobic preference of CypA

at this position Although residues that form the skeletal S2 pocket are identical between CypA and PPWD1, those adjacent to it, particularly in the b4– b5 loop (including residues Gly551–Gly553, Glu555 and Gly560–Glu561) are different between CypA and PPWD1, and may influence the side-chain specificity

at the)2 position Indeed, this region of the active site, including the b5–b6 loop (residues Ala583–Thr587), is the most divergent when comparing all Cyp active sur-faces, and we therefore predict that these regions will dictate the greatest differences in specificity at the )2 position In addition, using this structural rationale, there is some preference for a bulky side-chain at the +2 position because of potential stacking interactions with Trp601 and Phe540 Our current data cannot address positions C-terminal to +2, as our QAEGPKR peptide is constrained by its attachment

to a neighboring Cyp molecule

Fig 5 PPWD1 protein shows association in solution Two overlain 1 H, 15 N-HSQC spectra of PPWD1 proteins are shown The 473–646 con-struct (black) contains the amino terminal sequence QAEGPKR, but 493–646 (red) does not The 473–646 concon-struct is the same protein that led to the crystal model Experiments were conducted in parallel and under conditions of identical buffer and protein concentration.

As mentioned earlier, many of the residues that

make up the distal subsites on PPWD1 are conserved

between the spliceosomal Cyps and CypA (Fig 2)

Two exceptions are Ala583, which is variously changed

to Ser, Arg or Asn in spliceosomal Cyps, and Trp601

which is largely conserved amongst the

non-spliceoso-mal Cyps, but is variably changed to His, glutamate,

or Tyr in the spliceosomal subclass of Cyps (Fig 2)

Interestingly Ala583 is part of the S1¢ subsite, and

Trp601 and Phe540 are part of the S2¢–S3¢ subsite

described previously [38] It is reasonable to propose

that the ability of PPWD1 to bind to sequence

deter-minants that are not substrates for isomerization may

be quite relevant to the larger biological function of

Cyps if it is found to be a more general phenomenon

The intermolecular interaction seen in the crystal

structure of PPWD1 may imply a new function within

the spliceosome, where it possibly plays a role in

spliceosomal assembly or activity Homotypic

interac-tions are also observed in the crystal structure of

another spliceosomal Cyp, CypH (SnuCyp-20), where

isomerase domains interact with each other, again

through an extended loop region N-terminal to the

isomerase domain PPWD1 and CypH have no

sequence similarity in this region and crystallize under

different conditions and with different spacegroups

[32] It is interesting in the context of this dissimilarity

to observe that the two spliceosomal Cyps exhibit a

similar type of intermolecular association in structural

studies Although the solution properties of the

homo-typic CypH interaction were not studied, the crystal

structure of CypH bound to a peptide derived from

the spliceosome shows that it binds a surface directly

opposite to the isomerase active site (Fig 2),

suggest-ing that the homotypic interaction seen in the structure

would not be impaired by spliceosomal association

[32,33] The overall sequence similarity between the

isomerase domain of CypH and the isomerase domain

of PPWD1 is reasonably high (55% over

approxi-mately 140 residues), but many of the residues that

form the spliceosomal binding site of CypH are not

conserved in PPWD1 (blue boxes, Fig 2) In the case

of CypJ (PPIL3), another spliceosomal Cyp whose

interaction surface with the spliceosomal protein SKIP

has been probed using NMR, the interaction with the

spliceosomal component was again found to be distinct

from the active site (pink boxes, Fig 2) [31] Again,

the spliceosomal interacting region of CypJ is not well

conserved amongst spliceosomal Cyps, including

PPWD1 Although the spliceosomal target or

interact-ing region of PPWD1 has not been isolated, it is

rea-sonable to believe, based on the cases of CypH and

CypJ, that this region may lie well outside the active

site residues, and that these surfaces may be variable

in terms of sequence amongst the spliceosomal Cyps in order to target isomerase binding to distinct spliceoso-mal substituents It may be that the bifunctional isom-erase domains of spliceosomal Cyps may be directed towards internal sequences in order to regulate the activity of these enzymes or to serve as a signal trans-duction element in addition to the isomerase function Indeed, it may be that isomerization must be prevented

in order for spliceosomal Cyps to perform these addi-tional functions, and perhaps the homotypic interac-tions seen in the spliceosomal Cyps are indicative of peptide sequences that are binding determinants, but not efficient substrates, as in the case of PPWD1

Experimental procedures

Cloning, expression and purification

Full-length cDNA encoding human PPWD1 was obtained from the Mammalian Gene Collection (BC015385) Con-structs based around the predicted isomerase domain boundaries were cloned into pET28a-LIC and transformed into BL21 Gold (DE3) cells (Stratagene, La Jolla, CA, USA) Cultures were grown in Terrific Broth medium

at 37C to D  6, and induced at 15 C with the addition

of 50 lm isopropyl thio-b-d-galactoside Pellets were resus-pended in 20 mL of lysis buffer (50 mm Tris, pH 8.0,

500 mm NaCl, 1 mm phenylmethanesulfonyl fluoride and 0.1 mL general protease inhibitor; Sigma P2714, St Louis,

MO, USA) and lysed by sonication; lysates were then cen-trifuged for 20 min at 69 673 g The supernatant was loaded onto nickel nitrilotriacetic acid resin (Qiagen, Valen-cia, CA, USA), washed with five column volumes of lysis buffer and five column volumes of low imidazole buffer (lysis buffer + 10 mm imidazole, pH 8), and eluted in

10 mL of elution buffer (lysis buffer + 250 mm imidazole,

pH 8, and 10% glycerol) One unit of thrombin (Sigma) per milligram of protein was added to remove the tag overnight at 4C For gel filtration, an XK 16 · 65 column (GE Healthcare, Piscataway, NJ, USA) packed with HiLoad Superdex 200 resin was pre-equilibrated with gel filtration buffer (lysis buffer + 5 mm b-mercaptoethanol and 1 mm EDTA) Peak fractions were pooled and concentrated using Amicon concentrators (10 000 molecular mass cut-off; Millipore, Danvers, MA, USA) The protein was used at 15 mgÆmL)1for crystallization studies

Crystallization, data collection and structure solution

A construct of the PPWD1 isomerase domain containing residues 473–646 crystallized in 1.7 m NH4SO4, 0.1 m sodium cacodylate, pH 5.7, and 0.2 m sodium acetate

Crystals were harvested into mother liquor with 15%

glyc-erol and frozen in liquid nitrogen Diffraction data were

collected on an FR-E SuperBright Cu rotating anode⁄ Raxis

IV++ detector (Rigaku Americas, The Woodlands, TX,

USA), and integrated and scaled using the hkl2000

pro-gram package [40,41] For structure solution and

refine-ment, the program phaser [42] was used as part of the

ccp4 suite [43] to find the molecular replacement solution

in the resolution range between 20 and 2.8 A˚, using PDB

ID 1XO7 as a search model Following phaser, a round of

maximum-likelihood refinement and phase extension to

1.65 A˚ was carried out with refmac5 [43,44], and the

phases were then input into arp⁄ warp [45] for automatic

model building and iterative refinement The models were

completed using the graphics program o [46], and further

rounds of refinement using refmac5 resulted in an R-factor

of 19.9% (Rfree= 24.5%) for data from 44.32 to 1.65 A˚

The model has excellent stereochemistry as judged by

procheck[47], with no residues in disallowed or

unfavor-able regions of Ramachandran space The final model of

PPWD1 comprises three polypeptide chains corresponding

to the three molecules in the asymmetric unit; each chain is

disordered from residues 473–482, such that each model

reflects residues Gln483–Lys646 Data collection and

refine-ment statistics are provided in Table 1 Coordinates and

structure factors have been deposited in the Protein Data

Bank (PDB) with ID 2A2N

PPIase colorimetric activity

PPIase activity was measured using a standard

protease-cou-pled assay [3,48] adapted to a 96-well format The reaction

mixture contained 64 pm to 1 lm of PPIase and 200 nm

chy-motrypsin (Sigma) in reaction buffer (35 mm Hepes, pH 7.8,

368 mm trifluoroethanol, 50 mm NaCl2, 10 mm LiCl2, 5 mm

b-mercaptoethanol) The reaction was performed at 25C

using 33–400 lm suc-AAPF-pNA (Bachem Americas, King

of Prussia, PA, USA), and read at 390 nm on a BioTek

Syn-ergy 2 plate reader using flat-bottomed well plates (Costar

3695) Initial velocities were plotted against the substrate

peptide concentrations to compare the uncatalyzed

chymo-trypsin rate with the isomerase-catalyzed reaction

NMR experiments

Cells were inoculated into 20 mL of modified M9 medium

containing (15NH4)2SO4, trace elements and Kao &

Mich-ayluk vitamin solution (Sigma) Growth and induction were

performed as above, except that cells were induced at

D> 3 with 100 lm isopropyl thio-b-d-galactoside

Pro-tein was purified as above.15N-labeled proteins at 1 mm in

20 mm Hepes, pH 7.5, 100 mm NaCl, 10 mm dithiothreitol

and 10% D2O were pre-centrifuged at 35 000 g for 10 min,

and then subjected to NMR using a cryoprobed Bruker

AV500 spectrometer (Bruker, Milton, Canada) All spectra

were recorded at 25C For1

H,15N-HSQC experiments, a pulse sequence with ‘flip-back’ water suppression was used Typically, sweep widths of 8000 and 2000 Hz were used for the F2 and F1 dimension, respectively The data were pro-cessed with Topspin [49] or NMRpipe [50] software

All samples aimed at assessing PPWD1 binding and cataly-sis of peptides were diluted to 300 lL with 5% D2O and placed into a Shigemi microcell (Allison Park, PA, USA) in

50 mm Hepes, pH 7, 500 mm NaCl and 1 mm dithiothreitol Samples contained 3 mm peptide and either 0.5 mm suc-AAPF-pNA (Bachem) or TQAEGPKR (Sigma-Genosys), with and without 1 mm PPWD1-Cyp Spectra were collected

at 10C on a Varian 600 or 900 MHz spectrometer (Palo Alto, CA, USA) Spectra were acquired using standard Var-ian BioPack sequences, processed using NMRpipe software [50] and visualized using ccpn software [51]

Acknowledgements

Some of the NMR instrumentation used in the current study belongs to the Ontario Center for Structural

Table 1 Data collection, phasing and refinement statistics [atomic coordinates were deposited in the Protein Data Bank (PDB) (http:// www.rcsb.org): 2A2N].

Unit cell (A ˚ ) Unit cell (deg)

139.658, 39.893, 115.638 90.00, 122.33, 90.00

Refinement

Average B-factor (A˚2 ) 46.22 Ramachandran plot

a

Highest resolution shell is shown in parentheses.bR sym = 100 · sum(|I ) <I>|) ⁄ sum(<I>), where I is the observed intensity and <I>

is the average intensity from multiple observations of symmetry-related reflections. cR free value was calculated with 5% of the data.