Structure and dimerization of the catalytic domain of the protein phosphatase Cdc14p, a key regulator of mitotic exit in Saccharomyces cerevisiae

Abstract

In the budding yeast Saccharomyces cerevisiae, the protein phosphatase Cdc14p orchestrates various events essential for mitotic exit. We have determined the X‐ray crystal structures at 1.85 Å resolution of the catalytic domain of Cdc14p in both the apo state, and as a complex with S160‐phosphorylated Swi6p peptide. Each asymmetric unit contains two Cdc14p chains arranged in an intimately associated homodimer, consistent with its oligomeric state in solution. The dimerization interface is located on the backside of the substrate‐binding cleft. Structure‐based mutational analyses indicate that the dimerization of Cdc14p is required for normal growth of yeast cells.

Short abstract

PDB Code(s): 5XW4; 5XW5

Introduction

In eukaryotes, the regulation of mitotic exit involves inactivation of mitotic cyclin‐dependent kinases (CDKs) and activation of counteracting protein phosphatases.1 In the budding yeast Saccharomyces cerevisiae, the protein phosphatase Cdc14p is an antagonist of CDKs and is a key regulator of late mitotic events such as chromosome segregation, spindle disassembly, and cytokinesis.2 The activity of Cdc14p is regulated through nucleolar sequestration.3, 4 Cdc14p is kept inactive from G1 to metaphase by its inhibitor Net1p (also known as Cfi1p), which sequesters Cdc14p in the nucleolus. During anaphase, Cdc14p is released from Net1p by at least two signaling networks (FEAR and MEN), and spreads throughout the cell where it antagonizes mitotic CDK activity and so returns the cell to G1. Cdc14p is also required for accurate segregation of ribosomal DNA (rDNA) and telomeres,5, 6 and inhibition of transcription of yeast ribosomal genes during anaphase.7

The budding yeast CDC14 gene is essential for viability.8 The budding yeast Cdc14p is composed of two domains. The N‐terminal catalytic domain (residues 1–374) is necessary and sufficient for complementation of the cdc14 null mutation.9 The C‐terminal domain (residues 375–551) is rich in asparagine and serine residues and is not required for the critical cell cycle function of Cdc14p.9

Cdc14 homologs have been identified in a wide range of organisms from yeast to humans.10 Previously, crystal structure of the catalytic domain of human Cdc14B, one of the three Cdc14 homologs in humans, was determined, and revealed the molecular mechanism of the dephosphorylation reaction catalyzed by Cdc14 and provided structural evidence that Cdc14 selectively targets CDK phosphorylation sites (pS‐P‐x‐K/R, where “pS” stands for phosphoserine, and “x” can be any amino acid).11 Because the Cdc14 residues at the catalytic site are highly conserved,11 the catalytic mechanism is likely conserved from yeast to humans. However, there does not seem to be a conserved function of this phosphatase family throughout evolution,10 and mitotic exit in animal cells seem to rely on phosphatases of the PP1 and PP2A families, despite the presence of Cdc14 homologs.1

In this study, we report the crystal structures of the catalytic domain of the budding yeast Cdc14p that formed an intimately associated homodimer in the asymmetric unit of the crystal. Structure‐based mutational analyses provided a new biological insight that dimerization is important for the function of Cdc14p.

Results and Discussion

Structure determination

We obtained crystals of wild‐type apo Cdc14p (residues 1–374) that belonged to the space group P2₁2₁2₁ and had unit cell parameters of a = 82.85 Å, b = 95.13 Å, c = 137.76 Å. The crystals contained two Cdc14p molecules per asymmetric unit, with a solvent content of 57%. The structure was determined at 1.85 Å resolution by molecular replacement. The final model of wild‐type apo Cdc14p was refined to free and working R‐factor values of 20.3% and 17.8%, respectively (Table 1). A representative portion of the final 2F _O–F _C map is shown in Supporting Information Figure S1. Although V321 was a Ramachandran outlier, the electron density unambiguously showed the location of its carbonyl oxygen (Supporting Information Fig. S1).

Table 1.

Crystallographic Statistics

Crystal	apo Cdc14p (wild‐type)	Cdc14p (C283S) bound to the Swi6p phosphopeptide
Data collection
Space group	P212121	P212121
Unit cell dimensions
a, b, c (Å)	82.85, 95.13, 137.76	82.93, 95.93, 136.62
α, β, γ (degree)	90, 90, 90	90, 90, 90
Wavelength (Å)	0.98	1.0
X‐ray source	BL‐17A, PF	BL‐5A, PF
Resolution range (Å)a	55.79–1.85 (1.88–1.85)	30.59–1.85 (1.88–1.85)
No. of measured reflectionsa	611776 (15667)	355151 (10066)
No. of unique reflectionsa	91638 (3747)	91878 (3652)
Completeness (%)a	98.1 (82.2)	97.0 (79.2)
R merge (%)a	6.3 (61.6)	5.9 (44.2)
Mean I/σ(I)a	16.5 (1.7)	13.8 (2.3)
Mean I half‐set correlation CC(1/2)a	0.999 (0.744)	0.999 (0.809)
Multiplicitya	6.7 (4.2)	3.9 (2.9)
Wilson B factor (Å2)	22.3	19.7
Refinement
Resolution range (Å)a	55.79–1.85 (1.87–1.85)	30.59–1.85 (1.87–1.85)
R work (%)a	17.8 (27.3)	19.1 (24.8)
R free (%)a	20.3 (32.7)	21.7 (30.7)
No. of atoms
Cdc14p	5955	5857
Swi6p	‐	60
Sulfate	0	5
Water	571	666
No. of amino acids	737	734
Mean B factor (Å2)
Cdc14p	31.7	26.9
Swi6p	‐	28.7
Sulfate	‐	19.7
Water	38.1	34.8
RMSD from ideality
Bond lengths (Å)	0.004	0.006
Bond angles (degree)	0.665	0.758
Protein geometryb
Rotamer outliers (%)	0.16	0
Ramachandran favored (%)	96.02	96.94
Ramachandran outliers (%)	0.27	0.28
Cβ deviations > 0.25 Å (%)	0	0
MolProbity score (percentile)	1.29 (98)	1.29 (99)
PDB code	5XW4	5XW5

^aValues in parentheses are for the highest‐resolution shell.

^bMolProbity was used to analyze the structures.

We next obtained crystals of apo C283S mutant of Cdc14p (residues 1–374) that is catalytically inactive but is capable of binding phosphorylated substrates.9 The apo crystals of the C283S mutant were isomorphous to the wild‐type apo crystals. To determine the structure of Cdc14p (C283S) bound to a phosphorylated substrate, we used a phosphopeptide (155‐HRELG‐pS‐PLKK‐164) derived from Swi6p, which is one of the physiological substrates of Cdc14p.12 Swi6p is a transcription factor that is involved in cell‐cycle‐dependent transcription. The major cell cycle kinase Cdc28p and the S‐phase cyclin Clb6p specifically phosphorylate Swi6p at S160 and direct nuclear export of Swi6p, whereas Cdc14p dephosphorylates pS160 and stimulates nuclear import of Swi6p.12 After soaking the apo crystals of Cdc14p (C283S) with the S160‐phosphorylated Swi6p peptide, we obtained a data set to 1.85 Å resolution and determined the structure by molecular replacement. Residues 157–164 of Swi6p bound to the catalytic site of Cdc14p chain A were identified unambiguously in the electron density map [Fig. 1(A)], whereas the density map showed that a sulfate ion, instead of the phophopeptide, was bound to the catalytic site of Cdc14p chain B, probably because crystal packing hindered access of the phosphopeptide to this site of chain B. The final model of the phosphopeptide complex was refined to free and working R‐factor values of 21.7% and 19.1%, respectively (Table 1). The description of the Cdc14p structure below is based on the crystal structure of the phosphopeptide complex, but the conformation of the apo Cdc14p was essentially identical to that of Cdc14p in the phosphopeptide complex.

Figure 1.

Structure of the catalytic domain of budding yeast Cdc14p. (A) The electron density of the Swi6p phosphopeptide bound to Cdc14p. The omit F _O–F _C map covering the Swi6p phosphopeptide is shown in blue mesh (contoured at 3.0σ) with the refined model of the Swi6p phosphopeptide. (B) Superposition of budding yeast Cdc14p (green) and human Cdc14B (orange; PDB code, 1OHE). (C) Two orthogonal views showing the overall structure of the catalytic domain of budding yeast Cdc14p (C283S; ribbon representation) bound to the Swi6p phosphopeptide (stick representation). The A‐ and B‐domains are colored blue and green, respectively. The linker helix connecting the A‐ and B‐domains is colored purple. The C‐terminal extension is colored orange. (D) Primary structure of the catalytic domain of Cdc14p, with secondary structure assignment. (E) A stereo view of the interactions between Cdc14p (ribbon representation) and the Swi6p phosphopeptide (stick representation with yellow carbons) at the substrate‐binding cleft formed at the interface of the A‐domain (blue) and the B‐domain (green). The Cdc14p residues that are involved in the recognition of the phosphopeptide substrate are shown in stick representation with orange carbons. Dashed lines represent hydrogen bonds or salt bridges.

Overall structure of Cdc14p protomer

The two Cdc14p molecules in the asymmetric unit had essentially identical structure, with a root mean square deviation (r.m.s.d.) of Cα atoms of 0.285 Å. Similar to human Cdc14B [Fig. 1(B)], the catalytic domain of budding yeast Cdc14p is mainly composed of two similar domains [the N‐terminal A‐domain (residues 1‐163) and the C‐terminal B‐domain (residues 171‐342)], each with a dual‐specificity protein phosphatase (DSP)‐like fold, arranged in tandem [Fig. 1(C)]. The A‐domain has four α‐helices (α1A, α2A, α3A, and α4A) packed on one side of a β‐sheet composed of four β‐strands (β1A, β2A, β3A, and β4A) [secondary structure assignment, based on the method of Kabsch and Sander,13 is depicted in Fig. 1(D)]. The B‐domain has a β‐sheet composed of five β‐strands (β1B, β2B, β3B, β4B, and β5B) surrounded by α‐helices with two (α1B and α2B) on one side and four (α3B, α4B, α5B, and α6B) on the opposite side. In the B‐domain, the WPD loop (residues 249–258) connecting β4B with α3B bears the general acid/base D253 residue.14 Also in the B‐domain, the loop connecting β5B with α4B contains the active site sequence C‐X₅‐R (283‐CKAGLGR‐289), which is a hallmark for the protein tyrosine phosphatase (PTP) superfamily.15 The A‐ and B‐domains are connected by a linker helix, and the B‐domain is followed by a C‐terminal extension that has two short β‐strands (β1C and β2C) and a α‐helix (α1C) [Fig. 1(C)].

Recognition of S160‐phosphorylated Swi6p peptide by Cdc14p

The Swi6p phosphopeptide binds along the cleft formed at the interface of the A‐ and B‐domains of Cdc14p [Fig. 1(C)] and buries 591 Ų of interfacial surface area. Swi6p forms an extensive network of interactions with Cdc14p residues at the catalytic site [Fig. 1(E)]. The phosphate group of the pS160^Swi6p side chain forms extensive hydrogen bonds with main chain amide groups of Cdc14p residues K284, A285, L287, G288, and R289, and also forms bidentate salt bridges with the guanidinium side‐chain group of R289^Cdc14p. The aliphatic ring of the side chain of P161^Swi6p at the +1 position of the phosphorylation site (the P + 1 position) fits into a nonpolar pocket formed by Cdc14p residues F47, A130, Y132, L287, and V322. The interactions between the Cdc14p catalytic site and the phosphoserine and the proline at the P + 1 position of the substrate are quite similar to the interactions observed previously in the crystal structure of human Cdc14B bound to a phosphopeptide (A‐pS‐PRRR).11 In addition, the structure of the budding yeast Cdc14p bound to the S160‐phosphorylated Swi6p peptide also shows that the basic side chain of K163^Swi6p at the P + 3 position binds to the acidic residues of Cdc14p (E171 and D177) that cluster at the entrance of the substrate‐binding cleft. The aliphatic portion of the K163^Swi6p side chain is sandwiched between the nonpolar side chains of Cdc14p residues F47 and V173. These interactions are supplemented with hydrogen bonds between the main chain amide groups of Swi6p and the side chains of Cdc14p residues (Y132 and D253).

Dimerization of Cdc14p and its functional significance

The catalytic domain of Cdc14p forms an intimately associated homodimer in the asymmetric unit of the crystal [Fig. 2(A)]. Notably, the dimerization interface is quite extensive and buries 1668 Ų of the surface area of each chain, which is unusually large for normal crystal packing interfaces.16 This is consistent with a previous biochemical study that suggested that the catalytic domain of Cdc14p most likely forms a homodimer in vitro.9 The dimerization interface is located on the backside of the substrate‐binding cleft [Fig. 2(A)]. The dimer interface residues (41 residues in each chain) are highlighted in stick representation in Figure 2(A,B). In this symmetrical homodimer, the A‐domain of chain A mainly interacts with the A‐domain of chain B, and the B‐domain of chain A mainly interacts with the B‐domain of chain B. The long loop (residues 118–140) connecting α3A with α4A and adjacent three α‐helices (α2A, α3A, and α4A) are involved in the packing between the A‐domains in the dimer. The long α‐helix α6B and the adjacent short loop (residues 303–305) connecting α4B with α5B are involved in the packing between the B‐domains in the dimer. These interactions are supplemented with interactions between α4A in the A‐domain of one chain and α6B in the B‐domain of the second chain in the dimer. This mode of dimerization was not observed previously in the crystal structure of human Cdc14B,11 whose oligomeric state in solution and in living cells remains uncharacterized. The surface conservation profile of Cdc14 as analyzed by the ConSurf web server17 is shown in Supporting Information Figure S2. Except for the highly conserved catalytic‐site residues of Cdc14, the surface residues of Cdc14 are rather poorly conserved, indicating that the way Cdc14p chains form a dimer as observed in the crystal structures of Cdc14p may be specific to the budding yeast Cdc14p.

Figure 2.

Dimerization of Cdc14p. (A) Two orthogonal views showing the ribbon diagram of Cdc14p homodimer in the asymmetric unit. The two Cdc14p chains are distinguished by the colors dark green and orange. The dimer interface residues are shown in stick representation. The Swi6p phosphopeptide (stick representation) is bound to the catalytic site of Cdc14p chain A (green), whereas a sulfate ion (stick representation) is bound to the catalytic site of Cdc14p chain B (orange). Dashed lines represent hydrogen bonds. (B) The backside of the Cdc14p catalytic site. The dimer interface residues are shown in stick representation, together with ribbon diagram of Cdc14p chain A and stick representation of Swi6p. Chain B of Cdc14p is not shown for clarity. (C) Gel filtration analysis of the catalytic domain of wild type and mutant Cdc14p. (D) Standard curve for estimation of molecular masses (M _r) of Cdc14p oligomers. K _av = (V _e – V _o)/(V _t – V _o), where V _e = elution volume, V _o = void volume, and V _t = total volume. (E) Functional analysis of Cdc14p in yeast. Complementation of cdc14Δ by vector alone, wild type CDC14, and mutant CDC14 alleles containing the mutation(s) at the dimerization interface.

To verify that the dimer observed in the asymmetric unit of the Cdc14p crystals corresponds to Cdc14p dimer in solution, we engineered Cdc14p mutants in which the dimer interface observed in the crystal structures is specifically disrupted. We made two mutants: P123E and Y146K/Y330K. Among the dimer interface residues, P123 is located at the “periphery” of the dimer interface, and the aliphatic ring of P123 of chain A packs against that of P123 of chain B [Fig. 2(A)]. Thus, the substitution of P123 with glutamic acid would disrupt a hydrophobic interaction at the dimer interface. Y146 and Y330 are located close to the “core” of the dimer interface [Fig. 2(A)]. The phenolic side chain of Y146 of chain B contacts that of Y330 of chain A, and also forms a hydrogen bond with H327 of chain A [Fig. 2(A)]. Y330 of chain A also forms a hydrogen bond with E308 of chain B [Fig. 2(A)]. Because this is a symmetrical homodimer, Y146 and E308 of chain A interact with Y330 and H327 of chain B in the same manner. Therefore, the double mutations Y146K/Y330K would effectively disrupt the intermolecular interactions at the dimer interface at two sites. We compared the oligomeric state of the Cdc14p mutants in solution with that of wild‐type Cdc14p, using analytical gel filtration chromatography on a Superdex200 column. In gel filtration, His₆‐Cdc14p (residues 1–374, wild‐type) eluted as a single peak, and both P123E mutant and Y146K/Y330K mutant eluted later as a single peak [Fig. 2(C)]. Calibration of the Superdex200 column with molecular mass standards gave an apparent molecular mass of 76 kDa for wild‐type and 33 kDa for the mutants [Fig. 2(D)], suggesting that the wild‐type catalytic domain of Cdc14p forms a stable dimer but the mutants are unable to form a dimer under the condition tested in vitro.

We also analyzed the effects of mutations on the in vivo function of full‐length Cdc14p. A haploid cdc14Δ strain containing an extrachromosomal wild‐type copy of CDC14 in a CEN URA3 plasmid was transformed with CEN LEU2 plasmids containing wild‐type or mutant alleles of CDC14 and selected on 5‐fluoroorotic acid (5‐FOA) media for the loss of CEN URA3 [Fig. 2(E)]. As expected, wild‐type CDC14 complemented cdc14Δ lethality as opposed to vector alone [Fig. 2(E)]. Although the P123 mutation hardly affected yeast cell growth at 30°C, it severely impaired growth at 37°C [Fig. 2(F)]. The double mutant Y146K/Y330K exhibited more drastic phenotype. The mutations Y146K/Y330K severely impaired yeast cell growth both at 30°C and 37°C [Fig. 2(E)]. These data provide evidence that the dimer interface residues are important for Cdc14p function in vivo.

Conclusions

In summary, we have established the structures of the catalytic domain of budding yeast Cdc14p, the founding member of the Cdc14 family of dual‐specificity phosphatases, in both the apo state (wild type) and a phophopeptide substrate‐bound state (with a C to S substitution at the catalytic site). The structures defined the dimerization interface of Cdc14p on the backside of the catalytic site. Importantly, amino acid substitution(s) at the dimer interface of Cdc14p not only inhibited dimer formation in solution but also impaired yeast cell growth. These observations have the potential to open up new lines of inquiry into the functional role of the quaternary structure of Cdc14p.

Materials and Methods

Expression and purification of Cdc14p

N‐terminally His₆‐tagged Cdc14p (S. cerevisiae, residues 1–374; TEV site was inserted between the His₆‐tag and Cdc14p) was expressed from pRSFDuet‐1 (Novagen) in the E. coli host strain BL21‐CodonPlus(DE3)RIL (Stratagene) at 25°C in LB medium. His₆‐Cdc14p (residues 1–374) was initially purified by Ni‐NTA (Qiagen) chromatography with buffer A [30 mM Tris‐HCl (pH 7.5), 0.5 M NaCl, 10 mM imidazole, 7 mM 2‐mercaptoethanol, 0.5 mM PMSF, 0.5 mM AEBSF]. After elution (buffer A with 250 mM imidazole), the protein was finally purified by gel filtration over Superdex200 (GE Healthcare) in buffer B [10 mM Tris‐HCl (pH 8.0), 0.7 M NaCl, 0.05% glycerol, 2 mM 2‐mercaptoethanol], and concentrated to 14 mg/ml using a Millipore concentrator (Mol. wt. cutoff 10,000).

Mutants were created using the QuickChange system (Stratagene). All DNA constructs were verified by DNA sequencing.

Crystallization, data collection, and structure determination

Crystals of apo His₆‐Cdc14p (residues 1–374, wild type) were grown at 20°C from 14 mg/ml protein by hanging drop vapor diffusion against 0.1 M Bis‐Tris (pH 6.0), 0.1 M ammonium sulfate, and 6% PEG4000. Crystals were cryoprotected using mother liquor containing 13% PEG4000 and 18% glycerol, and flash‐cooled in liquid nitrogen. X‐ray diffraction datasets were collected at 95 K at Photon Factory beamline BL‐17A using an ADSC Quantum 270 CCD detector at a wavelength of 0.98 Å. Diffraction data were processed using iMOSFLM and CCP4 programs.18, 19 The structure was solved by molecular replacement using MOLREP20 using the structure of the catalytic domain of human Cdc14B (PDB code, 1OHE)11 as a search model. The structure was refined by iterative cycles of model building using COOT21 and refinement using PHENIX.22 The final model geometry was validated by MolProbity.23

The Cdc14p‐phosphopeptide complex was obtained by soaking method. S160‐phosphorylated Swi6p peptide 155‐HRELG‐pS‐PLKK‐164 was synthesized by Sigma‐Aldrich Japan. Crystals of apo His₆‐Cdc14p (residues 1–374, C283S mutant) were grown at 20°C from 14 mg/ml protein by hanging drop vapor diffusion against 0.1 M Bis‐Tris (pH 6.0), 0.1 M ammonium sulfate, and 6% PEG4000. The Swi6p phosphopeptide was soaked into the apocrystals of C283S mutant in mother liquor supplemented with 11 mM peptide for 2 h. Crystals were cryoprotected using mother liquor containing 13% PEG4000 and 18% glycerol, and flash‐cooled in liquid nitrogen. X‐ray diffraction datasets were collected at 95 K at Photon Factory beamline BL‐5A using an ADSC Quantum 315r CCD detector at a wavelength of 1.0 Å. Diffraction data were processed using iMOSFLM and CCP4 programs.18, 19 The structure was solved by molecular replacement using MOLREP20 using the structure of wild‐type apo Cdc14p as a search model. The structure was refined by iterative cycles of model building using COOT21 and refinement using PHENIX.22 The final model geometry was validated by MolProbity.23 Structural figures were produced using CCP4MG24 and PyMOL.25

ConSurf analysis of evolutionary conservation

The evolutionary conservation profile of Cdc14 was estimated using ConSurf.17 A CSI‐BLAST search for homologs of the budding yeast Cdc14p sequence was performed against the UNIREF90 database with an E‐value cutoff of 0.00001, minimal % ID of 35% for homologs and maximal % ID of 95% between sequences. A total of 150 homologous sequences were retrieved and multiply aligned using MAFFT. Calculation of position‐specific conservation scores was performed using the Bayesian method.

Analytical gel filtration chromatography

To analyze the oligomeric state of the catalytic domain of Cdc14p in solution, we gel‐filtered His₆‐Cdc14p (residues 1–374) in buffer C (10 mM Tris‐HCl pH 7.5, 0.6 M NaCl, 2 mM β‐mercaptoethanol) with a Superdex200 10/300 GL column (GE Healthcare) calibrated with globular proteins of known molecular weight. Each injection used 0.1 ml of 0.3 mg/ml His₆‐Cdc14p (residues 1–374), and the elution from the column was monitored using the ultraviolet absorbance at 280 nm. The molecular mass of His₆‐Cdc14p (residues 1–374) calculated from its amino acid sequence (including the N‐terminal tag) is 48 kDa.

In vivo functional analysis

The in vivo function of Cdc14p mutants was assessed using a plasmid shuffle technique.26 For the plasmid shuffle, plasmids (CEN, LEU2) encoding the wild type Cdc14p (full‐length) or Cdc14p mutants (full‐length) expressed from the CDC14 promoter, or a control vector (pRS315)27 were each transformed into haploid cdc14Δ yeast cells (his3Δ1 leu2Δ0 ura3Δ0 cdc14Δ::kanMX4) containing a wild‐type CDC14 maintenance plasmid (CEN, URA3). Single transformants were grown in SC‐Ura‐Leu medium at 30°C for 1 day to saturation, diluted to an OD₆₀₀ of 0.3, and grown in SC‐Ura‐Leu medium at 30°C for 4 h to an OD₆₀₀ of 1.2. Yeast cells were then serially diluted 1:6, and spotted onto 5‐fluoroorotic acid (5‐FOA) plates. 5‐FOA eliminates the URA3 maintenance plasmid encoding wild‐type Cdc14p. Plates were incubated at 30°C or 37°C for 3 days.

Accession numbers

The atomic coordinates and structure factors have been deposited in the PDB with accession codes 5XW4 [apo Cdc14p (wild type)] and 5XW5 [Cdc14p (C283S) bound to the Swi6p phosphopeptide].

Conflict of Interest Statement

The authors have no conflict of interest to declare.

Supporting information

Supporting Information