Analysis of PDE6 Function Using Chimeric PDE5/6 Catalytic Domains ଝ

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Vision Research 46 (2006) 860–868 www.elsevier.com/locate/visres

Analysis of PDE6 function using chimeric PDE5/6 catalytic domains ଝ

Hakim Muradov, Kimberly K. Boyd, Nikolai O. Artemyev ¤

Department of Physiology and Biophysics, University of Iowa College of Medicine, Iowa City, IA 52242, USA

Received 19 August 2005; received in revised form 12 September 2005

Abstract

cGMP-phosphodiesterases of the PDE6 family are expressed in retinal photoreceptor cells, where they mediate the phototransduction cascade. A system for expression of PDE6 in vitro is lacking, thus straining progress in understanding the structure–function relationships of the photoreceptor enzyme. Here, we report generation and characterization of bacterially expressed chimeric PDE5/6 catalytic domains which are highly soluble, catalytically active, and sensitive to inhibition by the PDE6 P subunit. Two Xexible PDE6 loops, H and M, impart chimeric PDE5/6 catalytic domains with PDE6-like properties. The replacement of the PDE6 H-loop into the PDE5 cata-

lytic domain increases the catalytic rate and the Km value for cGMP hydrolysis, whereas the substitution of the M-loop produces catalytic PDE domains responsive to P. Multiple PDE6 segments preventing functional expression of the catalytic domain are identiWed, supporting the requirement for specialized photoreceptor chaperones to assist PDE6 folding in vivo.  2005 Elsevier Ltd. All rights reserved.

Keywords: Phototransduction; Phosphodiesterase; PDE5; PDE6; cGMP

1. Introduction cGMP-gated channels in the plasma membrane and a neu- ronal response (Arshavsky, Lamb, & Pugh, 2002; Chabre & Photoreceptor phosphodiesterases (PDEs)1 are the eVec- Deterre, 1989). Two unique features of PDE6 enable the tor enzymes in the vertebrate visual transduction cascade. enzyme to fulWll the visual transduction tasks. Firstly, a They belong to the sixth family (PDE6) of eleven families high-aYnity interaction with the P-subunits blocks of PDEs identiWed in mammals (Beavo, 1995; Francis, cGMP-hydrolysis in the dark and thus reduces photorecep- Turko, & Corbin, 2001). Upon photoexcitation of rod or tor cell noise. Secondly, the cGMP- hydrolytic rate of cone photoreceptor cells, the GTP-bound -subunit is PDE6 is remarkably high when P is displaced by trans- released from the heterotrimeric visual G protein (transducin- and provides for robust ampliWcation of the signal. ducin) and stimulates PDE6 activity by relieving the inhibi- Despite the prominent role of PDE6 in photoreceptor tory constraint imposed by two identical inhibitory P biology, the structure–function relationships of this enzyme subunits on the enzyme catalytic dimer. Ensuing drop in family are poorly understood. The lack of understanding of intracellular concentration of cGMP leads to a closure of PDE6 at the molecular level can be mainly attributed to the failure of useful functional expression of the enzyme in var- ious systems (Piriev, Yamashita, Samuel, & Farber, 1993; ଝ This work was supported by National Institutes of Health Grant Piriev, Yamashita, Shih, & Farber, 2003; Qin & Baehr, EY-10843. 1994; Qin, Pittler, & Baehr, 1992). The most promising * Corresponding author. Tel.: +1 319 335 7864; fax: +1 319 335 7330. results were achieved using the baculovirus/sf9 cell system E-mail address: [email protected] (N.O. Artemyev). (Qin & Baehr, 1994). But even with this system, the yields of 1 Abbreviations used: PDE, phosphodiesterase; PDE6, photoreceptor W functional rod PDE6 catalytic subunits were very low (50– PDE; P , -subunit of PDE6; PDE5, cGMP-binding, cGMP-speci c PDE W (PDE5 family); cdPDE5, cdPDE6, the catalytic domains of PDE5 and 100 g/L). To date, there are no reports describing puri ca- PDE6. tion and characterization of stable recombinant PDE6

0042-6989/$ - see front matter  2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.visres.2005.09.015 H. Muradov et al. / Vision Research 46 (2006) 860–868 861 enzyme. Tangible progress had been achieved through 2.2. Cloning of the PDE5, PDE6 and PDE5/6 chimeric characterization of the chimeric enzymes between cGMP- catalytic domains binding, cGMP-speciWc PDE5 and PDE6 expressed in insect cells (Granovsky & Artemyev, 2000; Granovsky All constructs were cloned into the pET15b vector et al., 1998). Although PDE5 is not inhibited by P and its (Novagen) using NdeI and XhoI sites. cdPDE5 (aa 525– W kcat for cGMP hydrolysis is at least 100-fold lower than that 850) was ampli ed using pFastBacPDE5 as a template of PDE6, the enzymes are similar in terms of domain orga- (Granovsky & Artemyev, 2001) with primers containing nization. Like PDE5, PDE6 has two N-terminal GAF NdeI and XhoI sites (primers 1 and 2, Table 1). Sequences domains (termed for their presence in cGMP-regulated for cdPDE6 (aa 482–816), cdPDE6 (aa 480–814), and PDE, Anabaena adenylyl cyclases, and the Escherichia coli cdPDE6Ј (aa 480–814) were ampliWed using a bovine reti- protein Fh1A) and the catalytic domain located in the C- nal cDNA library kindly provided by Dr. W. Baehr (Uni- terminal part of the molecule (Aravind & Ponting, 1997; versity of Utah) (see Table 1 for primer information). Beavo, 1995). In addition, PDE5 and PDE6 share a rela- Chimera C1 (Fig. 2) was generated by a two-step PCR pro- tively high homology of the catalytic domains, speciWcity cedure. The PDE5(525–716) sequence was ampliWed with a for cGMP relative to cAMP, and sensitivity to common forward primer 1 containing a Nde site and a reverse primer catalytic-site inhibitors (Cote, 2004). PDE5/6 chimeras have 9 corresponding to PDE5(706–716)/PDE6Ј(671–681). helped to delineate the P- and noncatalytic cGMP-binding This PCR product was used as a forward primer in the sec- sites of PDE6 (Muradov, Boyd, & Artemyev, 2004; Mura- ond PCR ampliWcation with a Xho-containing reverse dov, Granovsky, Schey, & Artemyev, 2002). However, the primer 4 corresponding to PDE6Ј(804–814). The resulting utility of this approach to further elucidate the catalytic PCR product was cut with NdeI/XhoI and inserted into the determinants of PDE6 is limited since the PDE6 catalytic pET15b vector. Similar procedures were used for the con- domain (cdPDE6) does not correctly fold in the context of struction of C2 and C3 (Fig. 2). Chimera C4 was obtained PDE5/6 chimeras (Granovsky & Artemyev, 2001). by ligation of two PCR products into the pET15b vector The original report by Fink, Francis, Beasley, Grimes, cut with NdeI and XhoI. The Wrst PCR product coding for and Corbin (1999) Wrst established expression of an active, PDE5(525–652) was ampliWed with a 5Ј primer 1 and a 3Ј monomeric catalytic domain of PDE5 (cdPDE5) using the primer 12 corresponding to PDE5(643–652)/PDE6Ј(607– baculovirus system. Recent studies have demonstrated that 615). The second PCR product was obtained by amplifying the catalytic domain of PDE5 (cdPDE5) can be function- PDE5(669–850) with a forward primer 13 corresponding to ally expressed in E. coli with high yields permitting a solu- PDE6Ј(616–622)/PDE5(669–677) and a reverse primer 2. tion of its crystal structure (Huai, Liu, Francis, Corbin, & Approaches similar to those outlined above were used to Ke, 2004; Sung et al., 2003). The atomic structures of construct the remaining chimeras. The sequences of all chi- cdPDE5 facilitate the design of constructs for cdPDE6 and/ meric PDE constructs were veriWed by automated DNA or chimeric cdPDE5/6, whereas the bacterial expression sequencing at the University of Iowa DNA Core Facility. system oVers an advantage of rapid generation and screen- ing of these constructs. In this study, we examined expres- 2.3. Isolation and puriWcation of chimeric PDE5/6 catalytic sion of rod and cone cdPDE6 as well as chimeric PDE5/6 domains catalytic domains in E. coli. In contrast to poorly soluble and inactive cdPDE6, several chimeric cdPDE5/6 were Plasmids containing chimeric constructs were trans- highly soluble, catalytically active, and inhibited by P. formed into BL21-codon plus competent cells (Stratagene). Characterization of these proteins indicates that they are Single colonies were picked to inoculate LB medium con- well suited for future structural studies of the P/PDE6 taining ampicillin. The cultures were incubated overnight at interaction. 37 °C, diluted to 0.5–1.0 L (1:100) with 2£ TY medium containing ampicillin. At the cell density of OD D 1 at 600 nm 2. Materials and methods and selected temperature (typically 16 °C), the cultures were induced with 100 M IPTG. After 16–18 h incubation, cells 2.1. Materials were pelletted and stored at ¡80 °C or processed immedi- ately. Cell pellets were resuspended in 20 mM Tris–HCl [3H]cGMP was a product of Amersham Pharmacia Bio- buVer (pH 8.0) containing 500 mM NaCl, 20 mM imidazole, tech. All restriction enzymes were purchased from NEB. and Complete Mini protease inhibitor cocktail (Roche AmpliTaq DNA polymerase was a product of Applied Bio- Molecular Biochemicals) and then sonicated with six 20-s systems, and Pfu DNA polymerase was a product of Strat- pulses using a Xat tip attached to a 550 Sonic Dismembra- agene. Polyclonal anti-His6 antibodies were obtained from tor (Fisher ScientiWc). Cell lysates were cleared by centrifu- Santa Cruz. All other reagents were purchased from Sigma. gation (100,000g, 1 h, 4 °C) and loaded into NiSO4-charged Recombinant P was obtained as described previously His-bind resin (Novagen). The bound proteins were eluted (Artemyev, Arshavsky, & Cote, 1998). Synthetic peptide with 20 mM Tris–HCl pH 8.0, 500 mM NaCl, 200 mM corresponding to P-63–87 was custom-made by Sigma– imidazole. Soluble and fuctional chimeras C4, C5, C6, Genosys and puriWed by reverse-phase HPLC. and C8 were analyzed as His6-tagged proteins. Control 862 H. Muradov et al. / Vision Research 46 (2006) 860–868

Table 1 experiments showed no diVerences in stability, catalytic Oligonucleotide primers used for generation of PDE constructs properties, or P-inhibition of these chimeras prior to or Construct Primers Primer sequence after removing the His6-tag with thrombin. cdPDE5 1 GAGAATACATCATATGGAAGAAACCA GAGAGCTGCAGTCC 2.4. PDE activity assay 2 GACTTCTCGAGTCACTGCTGTTCTGCA AGAGCCTGCCATTTCTGCC 3 Ј PDE activity was measured using [ H]cGMP as cdPDE6 3 GAGAATACATCATATGTGTGAAGAAA X AACAGCTTGTCACAATTTTG described (Muradov et al., 2004). Brie y, 5–20 nM of 4 GACTTCTCGAGTCAATACTCATCAGCT cdPDE5/6 were incubated in 80 L of 20 mM Tris–HCl (pH V AGGGATTTCCATTCCACTCT 8.0) bu er containing 50 mM NaCl, 10 mM MgSO4, 2 mM cdPDE6 5 GAGAATACATCATATGTGCGAGGAAG 2-mercaptoethanol, 0.1 U bacterial alkaline phosphatase, AAGAACTGGCTGAGATCCTG 0.3 mg/mL BSA, and 5 M [3H]cGMP (100,000 cpm) at 6 GACTTCTCGAGTCAATACTCATCAGCA 3 AGCGCTTTCCACTCCTTGC 25 °C. After addition of [ H]cGMP, the reaction was cdPDE6 7 GAGAATACATCATATGTGCGAGGAGG allowed to proceed for 10 min and was stopped by the addi- ACGAGCTGGGGAAAATCC tion of AG1-X2 cation exchange resin (0.5 mL of 20% bed 8 GACTTCTCGAGTCAGTACTCATCGGCC volume suspension). Samples were incubated for 10 min at AAGGCCTTCCACTCC 25 °C with occasional mixing and spun at 9000g for 2 min. C1 1, 4, 9a TTTTTGAAACATGGTTCTCTTCTTGAA ATATAATGCTAGGTCTGTGGCTAAAA Aliquots of 0.25 mL were removed for counting in a scintil- TAGCTTGCTTGAT lation counter. To determine Km values for cGMP, PDE C2 2, 3, 10 GTTCAAAAAATTCTCCTCGTCTCTTTA activity was measured using 0.5–100 M cGMP and the TGTACAGAGCCAGGTCAGTCGCTATT W data were t to equation Y D Vmax ¤ X/(Km + X). The kcat ATCGCAACTTCAAAC values for cGMP hydrolysis were calculated as Vmax/[PDE]. C3 2, 3, 11 GCACTGATCAAAATGATGATGCTCCA TGATTGAAGAGCCGTGAAGCCTTGCT To determine Ki values for P and P -63–87, PDE activity 3 AGCGGAGATGTG was measured with 5 M [ H]cGMP in the presence of C4 1, 2, 12 CGGAGATGTGGACTTCATCTGATACA increasing concentrations of P or P -63–87. The Ki values AGTTATTGACACCACGGTGATCCAGA were calculated by Wtting data to equation: Y (%) D 100/ TCATG ٙ (1 + 10 ((X ¡ LogKi))), where X is the logarithm of total P 13 CTAGCAAGGCTTCACGGCTCTTCAATC ATGGAGCATCATCATTTTGAT or P -63–87 concentration. Fitting the experimental data to C5 1, 2, 14 GGGTTGTTGCTGCAGCACTGTTCTCTC equations was performed with nonlinear least squares crite- CAGATCTCCTTGTTCCCAAAATTCAGT ria using GraphPad Prizm 4 Software. The Km, Ki, and kcat GGCAACAAGTTCTGCTATC values are expressed as means § SE for three independent 15 ATTCCTATGATGGACAGAAACAAAAA experiments. AGATGAACTACCTAAACTTCAAGTTG GATTCATAGATGCC C6 1, 2, 16 AAAGATGAACTACCTAAACTTCAAGTT 2.5. Gel Wltration analysis GGATTTATTGACTTTGTCTGTACTCAA CTGTATGAGGCCTTGACC Aliquots of cdPDE samples (50–100 L, »1mg) were C7 1, 2, 17 GGATTTATTGACTTTGTCTGTACTTTT injected into a Superose 12 10/30 column (Amersham Phar- CTGTATGAGGCCTTGACCCATGTG C8 1, 14 macia Biotech) equilibrated at 25 °C with 20 mM Tris–HCl V C9 1, 2, 18 TTCAGTTTCCATTTTTTCACAGGCATC bu er (pH 8.0) containing 100 mM NaCl, 10 mM MgSO4, AACAATTTTTTGAAACATGGTTCGTCT and 2 mM 2-mercaptoethanol. Proteins were eluted at CTTTATGTACAGTGCTAG 0.4 mL/min, and 0.4 mL fractions were collected. Each frac- 19 GAAGAGGCCATCAAATATGTAACCAT tion was assayed by SDS–PAGE and for PDE activity. TGATCCAACCAAAAAAGAGTTGTTTTT AGCGATGCTG PDE activity was measured using 10–20 L aliquots from 3 C10 3, 4, 20 ATTTTTCATTATAAGTTCAAAAAATTC fractions and 5 M [ H]cGMP as described above. The col- TCCTCTCTTCTTGAAATATAAAGCCAG umn was calibrated with the following protein standards: 21 CAATTCAATTTGGAAGATCCTCATCA sweet potato -amylase (200 kDa), bovine serum albumin AAAGGAGATTATCATGGCAATGATG (67 kDa), chicken ovalbumin (45 kDa), and bovine erythro- ATG C11 2, 3, 20 cyte carbonic anhydrase (29 kDa). C12 2, 22 GACTATTAAGGATCATCAGGCAGTAC TCCAGGTGGTGCCTCTCCAAGATGGA 2.6. Native gel electrophoresis AGAGCCGTGAAGCCTTGCTAGCGGAG C13 4, 23 ACAAGGAGATCTGGAGAGAACAGTGC Native gel electrophoresis of cdPDE was run in precast TGCAGCAAC C14 1, 4, 24 CGTTAAACCCGTGTCGCCAGTTGTGG 7.5% acrylamide gels (Bio-Rad) according to the manufac- TAGGCGACGTTCTTCCGATAGTTCTTC ture’s protocol. Samples of cdPDE were diluted with two TTCAC volumes of sample buVer (62.5 mM Tris–HCl, pH 6.8, 40% a Sequence corresponds to a primer numbered in bold. glycerol, and 0.01% bromophenol blue), loaded on an acrylamide gel, and run at 25 °C with constant voltage of 180 V H. Muradov et al. / Vision Research 46 (2006) 860–868 863 for 35–40 min. BuVer containing 25 mM Tris and 192 mM the human cdPDE5 previously expressed in E. coli and glycine (pH 8.3) was used as electrode buVer. crystallized (Huai et al., 2004; Sung et al., 2003). Sequences for expression of the bovine cone cdPDE6Ј (aa 480–814), 2.7. Other methods rod cdPDE6 (aa 482–816), and cdPDE6 (aa 480–814) were selected based on the alignment with cdPDE5 (Figs. 1 Western blot analysis of proteins was performed follow- and 2). All four cdPDE were expressed in E. coli as His6- ing SDS–PAGE in 10% gels. Chimeric cdPDE5/6 were tagged proteins and puriWed on a His-bind resin column. A W detected using anti-His6 polyclonal antibodies (Santa Cruz) typical yield of puri ed cdPDE5 was 5–6 mg/L with the Km ¡1 (dilution 1:3000), anti-rabbit antibodies conjugated to horse- value of 6.3 M and kcat of 1.1 s . (Fig. 3, Table 2). In con- radish peroxidase (Sigma), and ECL reagents (Amersham trast to cdPDE5, the recombinant cdPDE6Ј, cdPDE6, Biosciences). The immunoblot analysis following native gel and cdPDE6 were found predominantly in insoluble electrophoresis was performed using anti-P-63–87 antibod- inclusion bodies after centrifugation of E. coli cell lysates ies (gift of Dr. R. Cote, University of New Hampshire). Pro- (100,000g, 1 h). Trace amounts of the PDE6 proteins tein concentrations were determined by the method of detected in the supernatant by Western blotting did not Bradford (1976) using IgG as a standard. Molar concentra- show enzymatic activity. Abundant formation of inclusion tions of cdPDE5/6 were calculated based on the MW values bodies in the absence of functional PDE6 proteins indi- calculated from the sequences of individual constructs. cated misfolding and aggregation of cdPDE6. Subse- quently, chimeric cdPDE5/6Ј were designed to localize 3. Results sequences preventing functional expression of cdPDE6.

3.1. Expression of the PDE5 and PDE6 catalytic domains in 3.2. Expression and characterization of chimeric PDE5/6 E. coli catalytic domains

The sequence for the construct of bovine cdPDE5 (aa The crystal structures of the PDE5 catalytic domain 525–850) (cdPDE5) was chosen to correspond exactly to show that it contains 16 helices that fold into three helical

Fig. 1. Sequence alignment of the catalytic domains of bovine rod PDE6 (6A and 6B), cone PDE6 (6C), and PDE5 (5A). The borders of the subdomains S1, S2, S3 (Sung et al., 2003), and the sequences corresponding to the metal-binding sites MI and MII, H-and M-loops, and the -hairpin region are indicated with arrows. 864 H. Muradov et al. / Vision Research 46 (2006) 860–868

S1 S2 S3 MI MII H-loopβ-hairpin M-loop 525 850 PDE5 480 814 PDE6α’ 716/671 C1 670/717 C2 622/669 C3 652/607 C4* 776/741 770/807 C5* 780/817 C6* 781/818 C7

C8* 722/677 706/742 C9 676/723 741/707 C10

C11 632/679 C12

C13 601/556 C14

Fig. 2. Schematic representation of PDE5/6 chimeric catalytic domains. Asterisks denote catalytically active chimeric PDE constructs.

Ј cdPDE5 C4 C5 C6 C8 kDa PDE6 , while C2 was reciprocal to C1 (Figs. 1 and 2). 50 Both, C1 and C2 were catalytically inactive and formed inclusion bodies, suggesting that either each of the portions

37 of cdPDE6 contains a misfolded sequence, or essential intramolecular contacts were missing in the chimeric domains. Chimera C3 containing the subdomain 1 of 25 PDEЈ was generated next (Fig. 2). This chimeric protein 20 was also insoluble and inactive (not shown). In a modiWed approach to PDE5/6 chimera design, we Fig. 3. Expression of the chimeric PDE5/6 catalytic domains in E. coli. A identiWed three regions that are likely to adopt diVerent coomassie blue-stained SDS-gel with the puriWed cdPDE5 and catalytically active chimeras C4, C5, C6, and C8. Equal fractions (0.3%) of the conformations in cdPDE5 and cdPDE6. Two such regions cdPDE preparations isolated from 1 L cultures were loaded into the gel to are counterparts of the H- and M-loops in the crystal struc- illustrate the expression yields of individual chimeras (15 g cdPDE5, ture of human PDE5 (Huai et al., 2004) (Fig. 2). The H- and 33 g C4, 18 g C5, 24 g C6, and 30 g C8). Arrows indicate correspond- M-loops adopt very diVerent conformations in otherwise ing Precision Plus Protein™ standards (Bio-Rad). well-superimposable structures of cdPDE5A and cdPDE4D subdomains, an N-terminal cyclin-fold region (S1), a linker (Xu et al., 2000). Furthermore, these loops display confor- region (S2), and a C-terminal helical bundle (S3) (Fig. 2) mational diVerences in the IBMX- and sildenaWl-bound (Huai et al., 2004; Sung et al., 2003). The connecting structures of cdPDE5 (Huai et al., 2004; Sung et al., 2003). sequence between the second and third subdomains is The H-loop and M-loop of PDE5 were substituted by the highly conserved between PDE5 and PDE6 (Fig. 1). equivalent PDE6Ј sequences in chimeras C4 and C5, Accordingly, chimera C1 was planned to include the subdo- respectively (Fig. 2). The H-loop is comprised of residues mains 1 and 2 of cdPDE5, and the subdomain 3 from 651–666 of bovine PDE5 (Huai et al., 2004). The replaced H. Muradov et al. / Vision Research 46 (2006) 860–868 865

Table 2 Properties of cdPDE5 and chimeric PDE5/6 catalytic domains a b ¡1 b b b Construct Calculated MW (kDa) Yield (mg/L) Km for cGMP ( M) kcat (s ) Ki for P ( M) Ki for P -63–87 ( M) cdPDE5 39.99 5–6 6.3 § 0.8 1.1 § 0.2 NA NA C4 39.84 10–12 15.8 § 2.2 2.4 § 0.3 NA NA C5 40.03 5–7 3.8 § 0.7 0.36 § 0.05 19.6 § 2.0 18.5 § 3.6 C6 40.07 7–8 2.5 § 0.2 0.35 § 0.04 2.7 § 0.2 1.7 § 0.2 C8 39.83 9–10 15.2 § 1.6 2.2 § 0.1 95 § 15 PDE6Јc 17–23 1200–3500 0.00017 NA – no inhibition of PDE activity was detected when using up to 0.5 mM P or P-63–87. a Including the His-tag. b Means § SE (n D 3). c Data for the full-length native PDE6Ј from Gillespie and Beavo (1988) and Granovsky et al. (1998).

PDE5 sequence in C4 corresponds to residues 653–668. Res- the M-loop have been earlier implicated in the interaction idues corresponding to positions 651 and 652 are identical with P (Granovsky & Artemyev, 2001). To probe the role in PDE5 and PDE6. Two extra residues, 667 and 668, of these residues in context of the chimeric catalytic were added to the replaced sequence to join the PDE6/ domain, two extensions of the PDE6 sequence, consisting PDE5 sequences at conserved residues (Fig. 1). The PDE5 of 10-residues and 11-residues each, were made to the M- sequence 777–806 substituted in C5 was extended by 4 loop in C6 and C7, respectively (Fig. 2). C6 was similar to residues from the M-loop sequence 777–802 (Huai et al., C5 in terms of catalytic properties, but its inhibition by P 2004) to also provide for the junction with conserved resi- was signiWcantly more potent, thus supporting the dues. Chimeras C4 and C5 were expressed as soluble and involvement of Phe777 and the surrounding residues in the catalytically active proteins. On SDS–PAGE C4 and C5 interface with the inhibitory subunit (Table 2, Fig. 4A). migrated similarly to cdPDE5 with an apparent MW of Surprisingly, addition of a single residue Phe781 led to »33 kDa, which is somewhat lower than the MW values expression of insoluble inactive chimera C7 (not shown). calculated from their sequences (Fig. 3, Table 2). Interest- The ability of the P C-terminal peptide, P-63–87, to W ingly, the level of expression and yield of puri ed C4 (10– inhibit C5 and C6 was also tested. The peptide Ki values 12 mg/L) exceeded those for cdPDE5 and C5 (5–7 mg/L) for the inhibition of C5 and C6 were 18.5 and 1.7 M, (Fig. 3, Table 2). In addition, C4 displayed »2.5-fold respectively, reXecting the stronger interaction of P-63– increases in the Km and kcat values for cGMP hydrolysis 87 with C6 (Fig. 4B, Table 2). (Table 2), which represents a shift toward a more PDE6- Based on the functionality of C4 and C5, chimera C8 Ј like enzyme. In contrast, chimera C5 had a reduced kcat of combining the H- and M-loops from PDE6 has been con- ¡1 0.36 s , whereas the Km value was modestly lower in com- structed and examined. Similarly to C4, C8 exhibited parison to cdPDE5 (Table 2). increased expression levels and elevated Km and kcat values The M-loop comprises a portion of the PDE6Ј-737– (Fig. 3, Table 2). Like C5, C8 was inhibited by P. Unex- 784 sequence, which has been previously shown to impart pectedly, the presence of the H-loop from PDE6Ј in C8 P-inhibition on the PDE5/PDE6 chimeric enzyme (Gra- reduced the eVectiveness of P inhibition (Table 2). novsky & Artemyev, 2000). The analysis of the C4 and C5 In addition to the H- and M-loops, a third region cor- inhibition by P indicated that C5, unlike C4 and the responding to the loop between helices 12 and 13 is pre- PDE5 catalytic domain, is sensitive to P (Fig. 4). Resi- dicted to be conformationally dissimilar in PDE5 and dues Phe777 and Phe781 of PDE6Ј located C-terminally to PDE6. This loop is relatively short in PDE5, but extended

A B 100 100 C5 C5 , % 75 , % 75

50 50 activity activity C6 25 C6 25 PDE PDE

0 0 -6.5 -6.0 -5.5 -5.0 -4.5 -4.0 -6.5 -6.0 -5.5 -5.0 -4.5 -4.0 Log [Pγ], M Log [Pγ-63-87], M

Fig. 4. Inhibition of chimeras C5 and C6 by P and P-63–87. The activities of C5 and C6 were measured in the presence of 5 M cGMP and increasing concentrations of P (A) or P-63–87 (B) and are expressed as a percentage of the respective PDE activity in the absence of P. Results of representative experiments are shown. The Ki values are shown in Table 2 as means § SE from three separate experiments. 866 H. Muradov et al. / Vision Research 46 (2006) 860–868

in PDE4D forming a -hairpin (Xu et al., 2000). The - 200 67 45 29 kDa hairpin region is even longer in PDE6 (Fig. 1). The diVer-

ence in the lengths of the hairpin regions represents one of 0.32 100 ) ( % activity, PDE the most obvious distinctions between PDE5 and PDE6. The loop H12–H13 of PDE5 was replaced by the -hairpin region of PDE6 in C9. This replacement produced 50 insoluble and inactive catalytic domain, indicating that 0.16 the PDE6 hairpin loop alone may impede correct folding of recombinant PDE6. This possibility was tested by 0 making the PDE6Ј catalytic domain in which the hairpin

Absorbance (280 nm) Absorbance 0 sequence was substituted by the corresponding loop of 11 22 35 48 61 PDE5 (C10) (Fig. 2). Chimera C10 too yielded inactive protein found largely in inclusion bodies. 37 kDa To further delineate regions causing misfolding and aggregation of recombinant cdPDE6, several new chimeras were designed using the template of C8. The PDE6Ј Fig. 5. Analysis of chimera C6 by gel Wltration. A sample of C6 (100 L, H-loop sequence was extended C-terminally to include 10 mg/mL) was injected on the Superose 12 10/30 column. Fractions of ᭿ the full subdomain S2 of PDE6Ј(C11) or just the adja- 0.4 mL were collected and then analyzed for PDE activity ( ) and by SDS–PAGE. The gel was stained with coomassie blue. The column was cent 10 residues (C12). In C13, the C-terminus of the calibrated with the following protein standards: sweet potato -amylase PDE6Ј catalytic domain was added to the M-loop (200 kDa), bovine serum albumin (67 kDa), chicken ovalbumin (45 kDa), sequence (Fig. 2). Expression of C11, C12, and C13 did and bovine erythrocyte carbonic anhydrase (29 kDa). not produce functional proteins. Furthermore, the N-terminal addition to the H-loop of the PDE6Ј region with AB5 5 5 5 E E E E metal-binding sites I and II also resulted in inactive and D D D D P 5 6 P 5 6 P 5 6 P 5 6 insoluble chimera C14 (Fig. 2). cd C C cd C C cd C C cd C C

3.3. Analyses of the P-binding chimera C6 by gel Wltration and native gel electrophoresis

The ability of P or the P C-terminal peptide to potently inhibit C6 indicated that this chimeric cdPDE may potentially be a useful tool for structural studies of the P/PDE6 interaction. Therefore, C6 was further analyzed by gel Wltration and native gel SDS–PAGE. The proWles of protein and PDE activity elution from a cali- γ brated Superose 12 10/30 column showed a single peak +Pγ-63-87 +P -63-87 corresponding to monomeric C6 with apparent MW of »36 kDa (Fig. 5). The elution proWle of cdPDE5 was very Fig. 6. Analysis of the P -interacting chimeras C5 and C6 by native gel electrophoresis. Samples of cdPDE5, C5, and C6 (30 M each) alone or preincu- similar to that of C5 (not shown), consistent with the pre- bated for 15 min at 25 °C with P-63–87 (100 M) were run in native 7.5% vious characterization of cdPDE5 as a monomer (Fink acrylamide gels (15 g cdPDE per lane for coomassie blue staining, 1 g for et al., 1999). immunoblotting). The gels were stained with coomassie blue (A) or ana- In addition to providing information on protein oligo- lyzed by immunoblotting with anti-P-63–87 antibodies (B). merization/aggregation state, native gel electrophoresis often allows the characterization of protein–protein inter- actions. The P-interacting chimeras C5 and C6 migrated 4. Discussion on the native gel similar to cdPDE5 as single bands with no detectable protein aggregation. Addition of P-63–87 Recent advances in bacterial expression and structural had no eVect on the migration of cdPDE5, but has analysis of the PDE1, PDE4, and PDE5 catalytic domains resulted in appearance of faster migrating species of C5 have generated the impetus to apply a similar approach to and C6 (Fig. 6A). P-63–87 is acidic (pI »4.4) and the analysis of cdPDE6 (Huai et al., 2004; Sung et al., 2003; Xu shift is consistent with the peptide/chimera complex. The et al., 2000; Zhang et al., 2004). Previous studies have estab- immunoblot analysis of the native gel conWrmed that the lished that misfolding of cdPDE6 is a primary reason for shifted C5 and C6 bands contain bound P-63–87 while the failed attempts of functional expression of the photore- no peptide co-migrated with cdPDE5 (Fig. 6B). In agree- ceptor enzymes (Granovsky et al., 1998). Our experiments ment with a stronger interaction of C6 with P-63–87, the using the bovine cdPDE5 construct have conWrmed its amount of peptide bound to C6 was greater than to C5 excellent expression in E. coli at an optimal low tempera- (Fig. 6B). ture of 16 °C. However, attempts to express rod and cone H. Muradov et al. / Vision Research 46 (2006) 860–868 867 cdPDE6, which are »40% identical (60% similar) to play a signiWcant role in PDE6 folding and stability. Substi- cdPDE5, at temperatures ranging from 12 to 37 °C pro- tutions of the Cys-residues in the CAAX box of PDE6 duced no soluble or active PDE6 proteins. Subsequently, and PDE6 abolished posttranslational isoprenylation and extensive analysis of chimeric cdPDE5/6 revealed two substantially reduced the protein expression level (Qin & important PDE6 regions allowing production of the active Baehr, 1994). Furthermore, the PDE6 isoprenylation, and enzyme and at least four regions preventing functional particularly, farnesylation of PDE6, may be essential to PDE6 folding. the ability of AIPL1 to serve as the enzyme chaperone The Wrst permissible region corresponds to the Xexible (Ramamurthy et al., 2003). Recombinant cdPDE6Ј is not H-loop of PDE6 within the C-terminal region of the S1- isoprenylated, which may explain the lack of AIPL1 eVect. subdomain. The eVect of incorporation of the PDE6 H- Yet, additional PDE6 chaperones are likely to exist. Co- loop was 2-fold: (a) an augmented expression of the active expression of isoprenylated PDE6 with AIPL1 in COS cells catalytic domain, and (b) an increase in the kcat and Km val- also failed to yield a functional enzyme (Liu et al., 2004). ues for cGMP-hydrolysis. Although, the N-terminal addi- Therefore, a successful strategy to functional expression of tion of the PDE6Ј metal-binding sites MI and MII to the PDE6 most certainly will include identiWcation of new pho- H-loop resulted in inactive protein in the context of chi- toreceptor-speciWc factors assisting PDE6 folding. mera C14, the replacement of the PDE6 segment Xanked by MI and MII into the full-length PDE5 had been shown to References produce a functional PDE chimeric enzyme Chi20 in sf9 cells (Granovsky & Artemyev, 2001). Notably, Chi20 also Aravind, L., & Ponting, C. P. (1997). The GAF domain: an evolutionary displayed a higher catalytic rate. Thus, the results of this link between diverse phototransducing proteins. Trends in Biochemical W Sciences, 22, 458–459. study combined with previous ndings suggest that the C- Arshavsky, V. Y., Lamb, T. D., & Pugh, E. N., Jr. (2002). G proteins and half of the S1-subdomain comprising the metal-binding phototransduction. Annual Reviews in Physiology, 264, 153–187. sites and the H-loop plays an essential role in determining Artemyev, N. O., Arshavsky, V. Y., & Cote, R. H. (1998). Photoreceptor the catalytic properties of PDE6. phosphodiesterase: interaction of the inhibitory subunits and cGMP W The second permissible PDE6 region corresponds to the with speci c binding sites on the catalytic subunits. Methods, 14, 93–104. Beavo, J. A. (1995). Cyclic nucleotide phosphodiesterases: functional extended M-loop within the S3 subdomain. The replace- implications of multiple isoforms. Physiological Reviews, 75, 725–748. ment of this region in cdPDE5 by the PDE6Ј sequence Bradford, M. M. (1976). A rapid and sensitive method for the quantitation produced the chimeric domain C6, which was eVectively of microgram quantities of protein utilizing the principle of protein- inhibited by P. Supporting existing evidence on the inter- dye binding. Analytical Biochemistry, 72, 248–254. action of cdPDE6 with the C-terminus of P (Granovsky, Chabre, M., & Deterre, P. (1989). Molecular mechanism of visual transduction. European Journal of Biochemistry, 179, 255–266. Natochin, & Artemyev, 1997), P -63–87 inhibited C6 as Cote, R. H. (2004). Characteristics of photoreceptor PDE (PDE6): similar- potently as the full-length P . Furthermore, the peptide Ki ities and diVerences to PDE5. International Journal of Impotence values for C6 (1.7 M) and trypsin-activated PDE6 from Research(Suppl 1), S28–S33. bovine retina (0.8 M) (Skiba, Artemyev, & Hamm, 1995) Fink, T. L., Francis, S. H., Beasley, A., Grimes, K. A., & Corbin, J. D. (1999). Expression of an active, monomeric catalytic domain of the are comparable, indicating that most of the contacts W cGMP-binding cGMP-speci c phosphodiesterase (PDE5). Journal of between cdPDE6 and the P C-terminus are recaptured in Biological Chemistry, 274, 34613–34620. C6. Gel-Wltration and native gel PAGE analyses indicate Francis, S. H., Turko, I. V., & Corbin, J. D. (2001). Cyclic nucleotide phos- that C6 is a monomeric protein well-suited for structural phodiesterases: relating structure and function. Progress in Nucleic studies of the P/PDE6 interaction. Acid Research and Molecular Biology, 65, 1–52. The regions found to interfere with a functional fold of Gillespie, P. G., & Beavo, J. A. (1988). Characterization of a bovine cone photoreceptor phosphodiesterase puriWed by cyclic GMP–sepharose cdPDE6 include the N-terminal part of the S1-subdomain, chromatography. Journal of Biological Chemistry, 263, 8133–8141. the S2-subdomain, and the -hairpin loop, and the C-termi- Granovsky, A. E., & Artemyev, N. O. (2000). IdentiWcation of the sub- nus of the S3-subdomain. In general, these regions are least unit-interacting residues on photoreceptor cGMP phosphodiesterase, conserved between cdPDE5 and cdPDE6. The multiplicity PDE6Ј. Journal of Biological Chemistry, 275, 41258–41262. and extensive nature of the misfolding segments support Granovsky, A. E., & Artemyev, N. O. (2001). Partial reconstitution of pho- W toreceptor cGMP phosphodiesterase characteristics in cGMP phos- the requirement of speci c chaperones for functional phodiesterase-5. Journal of Biological Chemistry, 276, 21698–21703. expression of PDE6. Recently, the aryl hydrocarbon recep- Granovsky, A. E., Natochin, M., & Artemyev, N. O. (1997). The -subunit tor-interacting protein-like 1 (AIPL1) was shown to be a of rod cGMP-phosphodiesterase blocks the enzyme catalytic site. Jour- specialized chaperone obligatory for expression and stabil- nal of Biological Chemistry, 272, 11669–11686. V ity of PDE6 in rod photoreceptors (Liu et al., 2004; Rama- Granovsky, A. E., Natochin, M., McEnta er, R., Haik, T. L., Franscis, S. H., Corbin, J. D., et al. (1998). Probing domain functions of chimeric murthy, Niemi, Reh, & Hurley, 2004). Mutations in AIPL1 PDE6Ј/PDE5 cGMP-phosphodiesterase. Journal of Biological Chem- cause Leber’s congenital amaurosis (LCA), a severe form of istry, 273, 24485–24490. retinal dystrophy in humans, apparently by compromising Huai, Q., Liu, Y., Francis, S. H., Corbin, J. D., & Ke, H. (2004). Crystal PDE6 expression (Sohocki et al., 2000). Co-expression of structures of phosphodiesterases 4 and 5 in complex with inhibitor 3- cdPDE6Ј with AIPL1 in E. coli produced inactive protein isobutyl-1-methylxanthine suggest a conformation determinant of inhibitor selectivity. Journal of Biological Chemistry, 279, 13095–13101. (Muradov and Artemyev, unpublished). It is very likely Liu, X., Bulgakov, O. V., Wen, X. H., WoodruV, M. L., Pawlyk, B., Yang, J., that the isoprenoid modiWcations of the PDE6 C-termini et al. (2004). AIPL1, the protein that is defective in Leber congenital 868 H. Muradov et al. / Vision Research 46 (2006) 860–868

amaurosis, is essential for the biosynthesis of retinal rod cGMP phos- Ramamurthy, V., Roberts, M., van den Akker, F., Niemi, G., Reh, T. A., & phodiesterase. Proceedings of the National Academy of Sciences USA, Hurley, J. B. (2003). AIPL1, a protein implicated in Leber’s congenital 101, 13903–13908. amaurosis, interacts with and aids in processing of farnesylated pro- Muradov, K. G., Granovsky, A. E., Schey, K. L., & Artemyev, N. O. (2002). teins. Proceedings of the National Academy of Sciences USA, 100, Direct interaction of the inhibitory -subunit of rod cGMP phosphodi- 12630–12635. esterase (PDE6) with the PDE6 GAFa domains. Biochemistry, 41, Ramamurthy, V., Niemi, G. A., Reh, T. A., & Hurley, J. B. (2004). Leber 3884–3890. congenital amaurosis linked to AIPL1: a mouse model reveals destabi- Muradov, H., Boyd, K. K., & Artemyev, N. O. (2004). Structural determi- lization of cGMP phosphodiesterase. Proceedings of the National nants of the PDE6 GAF A domain for binding the inhibitory -sub- Academy of Sciences USA, 101, 13897–13902. unit and noncatalytic cGMP. Vision Research, 44, 2437–2444. Skiba, N. P., Artemyev, N. O., & Hamm, H. E. (1995). The carboxyl termi- Piriev, N. I., Yamashita, C., Samuel, G., & Farber, D. B. (1993). Rod pho- nus of the -subunit of rod cGMP phosphodiesterase contains distinct toreceptor cGMP-phosphodiesterase: analysis of alpha and beta sub- sites of interaction with the enzyme catalytic subunits and the -subunits expressed in human kidney cells. Proceedings of the National unit of transducin. Journal of Biological Chemistry, 270, 13210–13215. Academy of Sciences USA, 90, 9340–9344. Sohocki, M. M., Bowne, S. J., Sullivan, L. S., Blackshaw, S., Cepko, C. L., Piriev, N. I., Yamashita, C. K., Shih, J., & Farber, D. B. (2003). Expression Payne, A. M., et al. (2000). Mutations in a new photoreceptor-pineal gene of cone photoreceptor cGMP-phosphodiesterase Ј subunit in Chinese on 17p cause Leber congenital amaurosis. Nature Genetics, 24, 79–83. hamster ovary, 293 human embryonic kidney, and Y79 retinoblastoma Sung, B. J., Hwang, K. Y., Jeon, Y. H., Lee, J. I., Heo, Y. S., Kim, J. H., et al. cells. Molecular Vision, 9, 80–86. (2003). Structure of the catalytic domain of human phosphodiesterase Qin, N., & Baehr, W. (1994). Expression and mutagenesis of mouse rod 5 with bound drug molecules. Nature, 425, 98–102. photoreceptor cGMP phosphodiesterase. Journal of Biological Chemis- Xu, R. X., Hassell, A. M., Vanderwall, D., Lambert, M. H., Holmes, W. D., try, 269, 3265–3271. Luther, M. A., et al. (2000). Atomic structure of PDE4: insights into Qin, N., Pittler, S. J., & Baehr, W. (1992). In vitro isoprenylation and mem- phosphodiesterase mechanism and speciWcity. Science, 288, 1822–1825. brane association of mouse rod photoreceptor cGMP phosphodiester- Zhang, K. Y., Card, G. L., Suzuki, Y., Artis, D. R., Fong, D., Gillette, S., et ase and subunits expressed in bacteria. Journal of Biological al. (2004). A glutamine switch mechanism for nucleotide selectivity by Chemistry, 267, 8458–8463. phosphodiesterases. Molecular Cell, 215, 279–286.