- Email: [email protected]
Expression, purification, and inhibition of in vitro proteolysis of human AMPD2 (isoform L) recombinant enzymes
Protein Expression and Purification 27 (2003) 293–303 www.elsevier.com/locate/yprep
Expression, purification, and inhibition of in vitro proteolysis of human AMPD2 (isoform L) recombinant enzymes Amy Louise Haas and Richard L. Sabina* Department of Biochemistry, Medical College of Wisconsin, 8701 Watertown Plank Road, Milwaukee, WI 53226, USA Received 8 July 2002, and in revised form 25 September 2002
Abstract AMP deaminase (AMPD) is a multigene family in higher eukaryotes whose three members encode tetrameric isoforms that catalyze the deamination of AMP to IMP. AMPD polypeptides share conserved C-terminal catalytic domains of 550 amino acids, whereas divergent N-terminal domains of 200–330 amino acids may confer isoform-specific properties to each enzyme. However, AMPD polypeptides are subject to limited N-terminal proteolysis during purification and subsequent storage at 4 °C. This presents a technical challenge to studies aimed at determining the structural and functional significance of these divergent sequences. This study describes the recombinant overexpression of three naturally occurring human AMPD2 proteins, 1A/2, 1B/2, and 1B/3, that differ by N-terminal extensions of 47–128 amino acids, resulting from the use of multiple promoters and alternative splicing events. A survey of protease inhibitors reveals that E-64 and leupeptin are able to maintain the subunit structure of each AMPD2 protein when they are included in extraction and storage buffers. Gel filtration chromatography of these three purified AMPD2 enzymes comprised of intact subunits reveals that each migrates faster than expected, resulting in observed molecular masses significantly greater than those predicted for native tetrameric structures. However, chemical crosslinking analysis indicates four subunits per AMPD2 molecule, confirming that these enzymes have a native tetrameric structure. These combined results suggest that AMPD2 N-terminal extensions may exist as extended structures in solution. Ó 2002 Elsevier Science (USA). All rights reserved.
AMP deaminase (AMPD; EC 3.5.4.6) catalyzes the hydrolytic deamination of adenosine-50 -monophosphate (AMP) to inosine-50 -monophosphate (IMP) and ammonium ion. The AMPD enzyme is an important regulator of cellular energy metabolism through its participation in purine nucleotide catabolism [1–3]. Multiple isoforms have been isolated from different human tissues and are named after the source of their purification: isoforms M, muscle; L, liver; and E, erythrocyte [4]. Subsequent cloning of three human AMPD genes has revealed the molecular basis for these different isoforms: AMPD1, M; AMPD2, L; and AMPD3, E [5–8]. It is likely that the three AMPD genes arose from duplication of a common primordial gene [9], and subsequently, acquired differences via divergent evolution. Consistent with this hypothesis, AMPD isoforms contain both conserved and divergent domains. * Corresponding author. Fax: 1-414-456-6510. E-mail address: [email protected] (R.L. Sabina).
The three AMPD polypeptides share a similar 550 amino acid C-terminal end (62–70% identical; [6,7]) that contains a motif SLSTDDP believed to be the catalytic center of the enzyme [10,11]. Conversely, each AMPD polypeptide differs by divergent N-terminal sequences of 200–330 amino acids with less than 36% identity to each other. In addition, differential promoter use and alternative splicing add extensions or substitutions of four (AMPD1; [5,12]), 47–128 (AMPD2; [13]), and 7–9 (AMPD3; [7]) amino acids at the distal N-terminal end of each AMPD polypeptide. Available information suggests that different N-terminal domains and distal Nterminal variations in each AMPD polypeptide contribute to isoform-specific behaviors of this enzyme [14]. AMPD proteins from human and other sources are susceptible to proteolysis in vitro [11,15–20] and proteolytic events have been localized exclusively to the N-terminal region of these polypeptides [20–23]. Consequently, it is likely that prior studies using purified AMPD enzymes have been performed with subunits
1046-5928/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 6 - 5 9 2 8 ( 0 2 ) 0 0 6 3 6 - 8
294
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
lacking varying amounts of N-terminal sequence. It is therefore important to explore the possibility that unique N-terminal domains may affect the function of each AMPD enzyme. Although proteolysis has not yet been demonstrated in vivo, the possibility also exists that intracellular processing of these proteins could serve to remove unique N-terminal sequence and thereby abrogate any related function conferred to the respective proteins. Examination and comparison of intact and proteolyzed proteins can yield information about the impact of N-terminal extensions on AMPD2 structure and function, and may provide a basis for further studies of these proteins in vivo. Recombinant technology affords the means by which intact AMPD proteins can be expressed and purified. This method has been used to generate predominantly full-length human AMPD1 and AMPD3 enzymes, which has facilitated studies designed to investigate the functional significance of divergent N-terminal sequences in these polypeptides [20,24,25]. However, AMPD1 and AMPD3 recombinant proteins are susceptible to proteolysis during extended storage at 4 °C [20]. Expression and purification of full-length, human AMPD2 enzymes have not yet been reported. However, N-terminal proteolysis removes as much as 25–35 kDa from these proteins when they are isolated from endogenous sources [4,13,26,27]. Consequently, little is known about how the N-terminal domain of AMPD2, and extensions conferred on it by the shuffling of 50 exons, affect the function of this isoform. This study describes the recombinant expression of three naturally occurring AMPD2 variants (1A/2, 1B/2, and 1B/3) and identifies protease inhibitors that can be used to minimize in vitro proteolysis during their purification and subsequent characterization. This information should be useful to all investigators working with AMP deaminases.
Materials and methods Materials Restriction endonucleases were obtained from Roche (HindIII, BamHI, and BglII), Gibco Life Sciences (SacI), and New England Biolabs (NcoI and SapI). GraceÕs insect cell culture medium, DMEM, fetal bovine serum, lipofectamine, DNA primers, and DNA ligase were obtained from Gibco Life Sciences. The baculoviral transfer plasmid pBlueBacIII (pBB3) was purchased from Invitrogen. BaculoGold modified baculoviral DNA was obtained from Pharmingen. The mammalian expression plasmid pCMV-FLAG and antibody to the FLAG peptide epitope tag (monoclonal M2) were obtained from Stratagene. Goat anti-mouse horseradish peroxidase (HRP)-conjugated IgG (ImmunoPure),
chemiluminescence reagents (Supersignal), and the primary amine crosslinker BS3 (Bis(sulfosuccinimidyl) suberate) were purchased from Pierce. Goat anti-rabbit HRP-conjugated IgG was obtained from Santa Cruz Biochemicals. Pansorbin was purchased from Calbiochem. Sephacryl S300-HR size exclusion chromatography resin and globular gel filtration calibration standards were obtained from Amersham Pharmacia. Phosphocellulose resin (P-11) was supplied by Whatman, Ltd. Glass chromatography columns ð1:5 30 cmÞ and 4-chloro-naphthol were purchased from Bio-Rad. Hydrogen peroxide (30%) and adenosine-50 -monophosphate (AMP) were purchased from Sigma–Aldrich Chemical Company. The protease inhibitors phosphoramidon, acetyl phenylmercuric acetate (APMA), phenylmethylsulfonyl fluoride (PMSF), aprotinin, Na tosyl-phe-chloromethyl ketone (TPCK), Na -tosyl-lyschloromethyl ketone (TLCK), pepstatin, and epsilon amino caproic acid (EACA) were gifts of Sally Twining, Ph.D. (Medical College of Wisconsin, Milwaukee, WI). Leupeptin, E-64, ethylenedinitrilotetraacetic acid (EDTA), and ethyleneglycol-bis-(b-aminoethyl ether) NNN 0 N 0 -tetraacetic acid (EGTA) were obtained from Roche Biochemicals. Glycerol was purchased from Shelton Scientific (Shelton, CT). DNA and protein sequencing were performed by the Medical College of Wisconsin Protein and Nucleic Acid Facility, Milwaukee, Wisconsin. Construction of baculoviral transfer vectors for expression of wild-type AMPD2 variants A baculoviral transfer vector was constructed that contained the region encoding the unique 1A/2 N-terminal extension (pBB3-1A/2). This was accomplished by PCR amplification of a partial 1A/2 clone [13] using the primers 50 -AATAGATCTCCACCATGGCCTCAG AGGCTC-30 and 50 -CTTCTCCCGGATGAAGAGC-30 . This added a BglII restriction site at the 50 end of the PCR product (underlined) and maintained an internal SapI site (italics). The PCR product was digested with BglII and SapI, and then ligated with a SapI–HindIII fragment containing the remainder of the AMPD2 coding sequence into the baculoviral transfer plasmid pBB3 (Invitrogen). A baculoviral transfer vector containing the region encoding the unique 1B/2 N-terminal extension (pBB31B/2) was produced by PCR amplification of an existing partial 1B/2 clone [13] using the primers 50 -TATA GGATCCAGATCT GCCACCATGAGAAATCGTGG CCAG-30 and 50 -GGATGCCATGGCTGGGACC-30 . This added BamHI and BglII restriction sites at the 50 end (underlined and underlined with italics, respectively) and maintained an NcoI site at the 30 end of the PCR product (italics only). Following digestion with BglII and NcoI, the restricted PCR product was ligated with
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
an NcoI–HindIII fragment of an existing AMPD2 clone containing the remainder of the coding sequence [24] into BglII–HindIII digested pBB3. The coding region for the unique 1B/3 N-terminal extension was reconstructed from existing clones by overlap PCR in order to produce a 1B/3 baculoviral transfer vector (pBB3-1B/3) as follows: for the first round of PCR, the primers 50 -TATAGGATCCAGATCT GC CACCATGTGGCAGAGCCAG-30 and 50 -GGAGCT GCAGTGGAGGCCTGCAGGC-30 were used to amplify the exon 1B-encoded fragment of the 1B/3 coding region from pBB3-1B/2. This primer added a BamHI recognition site (underlined) and a BglII site (underlined with italics) to the 50 end of the PCR product. The primers 50 -GCCTGCAGGCCTCCACTGCAGCTCC30 and 50 -CAG GAAATCTTGCTTGG-30 [6] were then used to amplify the exon 1B–exon 3 junction from a partial 1B/3 clone [13]. The products of these two PCR reactions were then used as templates for overlap PCR amplification with the two external primers from the initial amplifications. This produced a cDNA fragment encoding the entire 1B/3 extension that was used as a template for a second round of overlap PCR. This was accomplished using a second fragment generated by PCR amplification of the region encoding exon 3 from pBB31B/2 with the primers 50 -CCAAGCAAGATTTCCTG-30 and 50 -CAGACCCAAGCCCAGGT-30 [6]. The overlap PCR extended the 1B/3 cDNA 30 of a SapI recognition site that was used in the final cloning step. This product was digested with BglII and SapI and ligated into pBB3 together with a SapI–HindIII fragment encoding the remainder of the AMPD2 coding sequence. All DNA plasmids were amplified and then purified by cesium chloride banding. cDNA identity and integrity were confirmed by restriction digest and DNA sequencing. Baculoviral transfer vectors for the AMPD2 deletion mutants of the 1B/2 polypeptide, DM55, DE129, and DD184, have been described previously [24]. Construction of plasmids for mammalian expression of AMPD2 fusion proteins An expression plasmid for N-terminally tagged 1A/2 (pCMV-FLAG-1A/2) was generated by ligation of a BglII fragment from pBB3-1A/2 in-frame into pCMVFLAG version 2B (Stratagene). A DNA fragment beginning with a SacII restriction endonuclease site 375 base pairs 50 to the stop codon and ending with the 30 end of the AMPD2 cDNA coding region that excluded the stop codon was used to generate an expression plasmid for C-terminally tagged 1A/2 (pCMV-1A/2FLAG). This was produced by PCR using the 50 primer 50 -CCTGTCCCGCGG CCTCA-30 , which contains the SacII site (italics), and the 30 primer 50 -TATA AAGCTTTTGAGGCCCTGGGCTCATGG-30 , which removes the stop codon by mutagenesis and adds a
295
HindIII site to the 30 end of the DNA fragment (underlined). The PCR product was digested with SacII and HindIII, and then ligated with a BamHI–SacII DNA fragment isolated from pBB3-1A/2 into BamHI–HindIII digested pCMV-FLAG version 4A. All expression plasmids were confirmed by DNA sequencing and purified by cesium chloride banding. Production, isolation, and purification of AMPD2 recombinant baculovirus Baculoviral transfer vectors encoding 1A/2, 1B/2, or 1B/3 were co-transfected with modified baculoviral genomic DNA into Sf9 cells by calcium phosphate precipitation to generate AMPD2-expressing virus by homologous recombination, as described previously [20]. Viral clones were purified by plaque assay and amplified by successive infection of Sf9 cells. Expression and purification of AMPD2 proteins Twelve T-185 flasks of Sf9 cells at 95% confluence were infected with high titer virus amplified from plaque-purified AMPD2 clones. Phosphocellulose purification was performed, as described previously [20]. Protease inhibitors were included as indicated. Briefly, cells were incubated with virus for four days, then loosened from the flasks by rapping, and collected by centrifugation at 1000g in a tabletop centrifuge at 4 °C. Viral supernatant was removed and cells were resuspended in cold column buffer (20 mM KPO4 , pH 6.7, 100 mM KCl, and 0.1% b-ME), supplemented with 5 lg=ml leupeptin and 10 lM E-64, unless otherwise indicated. Cells were lysed by sonication and lysates were clarified by centrifugation for 10 min at 10,000g at 4 °C. Crude homogenate was batch adsorbed to buffer-equilibrated phosphocelluose for 30 min with rotation at 4 °C. The resin was then washed with two volumes of column buffer and poured into a 1:5 cm 30 cm column. Proteins were eluted with a linear gradient of 0.1–2 M KCl in 20 mM KPO4 , pH 6.7, in the presence of protease inhibitors. Fractions were assayed for protein concentration [28] and AMPD activity. Fig. 1 schematically illustrates encoded polypeptides for all purified human AMPD2 recombinant enzymes used in this study. In addition to the naturally occurring variants of isoform L (1A/2, 1B/2, and 1B/3), three N-terminal deletion enzymes were prepared. All are numbered according to the 1B/2 polypeptide: DM55ð1B=2Þ is truncated to an in-frame methionine codon in the exon 1B sequence; DE129ð1B=2Þ begins with exon 3-encoded sequence and represents the core AMPD2 polypeptide common to all three naturally occurring variants of isoform L; DD184ð1B=2Þ starts near to identified protease sites in the rat brain AMPD2 polypeptide [23]. These N-truncated enzymes were used in comparative
296
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
Expression of AMPD2 fusion proteins in mammalian cells
Fig. 1. Schematic of human AMPD2 (isoform L) polypeptides. The three naturally occurring variants, 1A/2, 1B/2, and 1B/3, have N-terminal extensions of 47, 128, and 53 residues, respectively. These polypeptides are encoded by alternative AMPD2 mRNAs produced by the regulated expression of different promoters upstream of exons 1A and 1B and a cassette-type alternative splicing event involving exon 2 that results in the use of a different reading frame in exon 1B when this sequence is spliced directly to exon 3. Three additional N-truncated recombinant enzymes with progressive deletions of the 1B/2 polypeptide (DM55, DE129, and DD184) are included for comparative analyses with the naturally occurring variants of isoform L. Polypeptides are numbered from the C-terminus in order to better illustrate the three Nterminal extensions.
analyses together with the three wild-type variants as a means of assessing the structural roles of divergent N-terminal extensions in each AMPD2 polypeptide. Polyacrylamide gel electrophoresis Crude lysates and purified proteins were separated by SDS–PAGE using an 8% gel according to the method of Laemmli [28]. Protein bands were detected by staining with Coomassie blue R or by Western blotting. AMPD enzyme assays Enzyme assays were performed at 37 °C in 25 mM imidazole, pH 6.5, 100 mM KCl, and 0.2 mg/ml BSA, using 20 mM AMP. Substrate and product (IMP) were separated by anion-exchange HPLC using an ammonium dihydrogen phosphate (5–750 mM) and pH (2.8– 4.0) gradient and quantified by peak integration. Assays were performed at initial velocity conditions, with product not exceeding 10% of substrate. Inhibition of in vitro proteolysis Aliquots of purified protein (1.5 U; 30 lg, in 100 ll total volume) were incubated with protease inhibitors (see text for a complete list and concentrations used) at 4 °C for two to eight weeks. Proteins were separated by SDS–PAGE and visualized by Coomassie blue staining or by Western blotting.
HeLa cells were maintained in DMEM supplemented with 10% FBS at 37 °C with 5% CO2 . For transfection, cells were grown to 80–90% confluence in T-75 flasks and transiently transfected with plasmid DNAs encoding FLAG-1A/2 or 1A/2-FLAG via lipofectamine (GibcoBRL) according to manufacturerÕs protocol. Cells were then maintained at 37 °C with 5% CO2 and harvested 48 h post-transfection. For harvest, cells were washed twice in PBS, then covered with 1 ml MTMSB (50 mM KCl, 10 mM KPO4 , pH 6.7, 1% NP40, and 1 mg/ml BSA), and dislodged from the plate by scraping. Resuspended cells were lysed by sonication and clarified by centrifugation at 4 °C for 10 min at 10,000g. Lysates were assayed for protein concentration and enzyme activity. Immunoprecipitation and Western blotting For AMPD2 Western blots, crude lysate or purified protein was separated by SDS–PAGE in an 8% gel, using the method of Laemmli [29], and electroblotted onto a nitrocellulose membrane. Membranes were blocked in 2.5% milk buffered in TBS with 0.1% Tween 20 for 1 h at room temperature with agitation and then washed four times for 1 min each with TBS containing 0.5% Tween 20. Blots were then incubated with a 1:500 dilution of polyclonal rabbit antisera raised against an N-truncated human AMPD2 recombinant enzyme [24] for 1 h at room temperature with agitation and washed as before. Bound primary antibody was then complexed with a 1:2000 dilution of HRP-conjugated goat antirabbit secondary antibody for 1 h at room temperature with agitation, and unbound complexes were washed from the blot as before. Protein bands were detected by incubation with H2 O2 (0.013%) and 4-chloro-naphthol. For FLAG Western blots, clarified lysates from pCMV-FLAG-1A/2 or pCMV-1A/2-FLAG-transfected or nontransfected HeLa cells were immunoprecipitated with 3 ll undiluted preimmune or anti-AMPD2 sera for 2 h at 4 °C with rotation in 10 mM KPO4 , pH 6.7, 150 mM KCl, 1% NP40, and 1 mg/ml BSA. Immunocomplexes were incubated with 50 ll Pansorbin beads for 30 min at 4 °C with rotation and washed three times. Immunoprecipitates were separated on an 8% SDS– PAGE gel as before and transferred to nitrocellulose. Blots were blocked with 2.5% milk diluted in TBS with 1% Tween 20 and washed as before to remove excess blocker. The blots were then incubated with a 1:1000 dilution of monoclonal antibody to the FLAG epitope (M2, Stratagene), washed, and incubated with a 1:10,000 dilution of HRP-conjugated goat anti-mouse antibody. Protein bands were then detected by enhanced chemiluminescence. Clarified lysates ð150 lgÞ were also assayed as above to confirm the functional integrity of each tagged enzyme.
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
297
Gel filtration chromatography Gel filtration chromatography was performed, as described previously [20]. Briefly, 15 U of each AMPD activity was diluted to 1 ml in 50 mM imidazole, pH 6.5, containing 500 mM KCl and then loaded onto a 1 cm 90 cm Sephacryl S300-HR gel filtration column equilibrated in the same buffer at 4 °C. Fractions (0.5 ml) were collected and assayed for AMPD activity. For column calibration, the elution volumes of four known globular protein standards (aldolase, 158 kDa; catalase, 232 kDa; ferritin, 440 kDa; and thyroglobulin, 669 kDa) as well as blue dextran ðVo Þ were measured. Ve =Vo values were subjected to regression analysis to determine the equation logMW ¼ 2:3ðVe =Vo Þ þ 8:4, which was used to calculate the observed molecular mass of each AMPD protein. Chemical crosslinking Five micrograms of purified human AMPD recombinant protein was incubated with the homobifunctional, noncleavable, primary amine-crosslinking agent BS3 at 10- to 50-fold molar excess with respect to protein (final concentration 10 lM–10 mM) in 1.7 mM sodium citrate, pH 5.0, 1 M KCl, for 2 h on ice. Reactions were quenched by addition of Tris–HCl, pH 7.4, to a final concentration of 25 mM for 15 min on ice, then diluted in 2 SDS—PAGE sample buffer [29] and separated in a phosphate-buffered 3.5% acrylamide gel. Protein bands were visualized by Coomassie blue staining.
Results Inhibition of in vitro proteolysis of AMPD2 proteins Human AMPD1 and AMPD3 recombinant enzymes can be purified in the absence of protease inhibitors with subunits predominantly intact, but are susceptible to proteolysis during extended storage at 4 °C [20]. A representative Western blot analysis of the purified 1B/3 protein is shown in Fig. 2A and indicates that AMPD2 recombinant proteins are relatively more sensitive to proteolysis. Data in the right panel reveal that substantial proteolysis occurs during protein purification (lane 1) and then proceeds during an overnight dialysis at 4°C (lane 2). These results warranted efforts to prevent degradation of each AMPD2 polypeptide as a prerequisite to subsequent characterization of the three naturally occurring variants of isoform L. A preliminary experiment was performed in which a variety of protease inhibitors were screened for their ability to inhibit proteolysis of the 1B/2 protein. These included EDTA (1 mM), EGTA (1 mM), phosphoramidon ð10 lMÞ,
Fig. 2. Leupeptin and E-64 effectively inhibit the in vitro proteolysis of human AMPD2 recombinant proteins. (A) Western blot analysis of 1B/3 purified in the absence of protease inhibitors (lane 1) and after subsequent overnight dialysis at 4 °C in 50 mM imidazole, pH 6.5, containing 500 mM KCl and 1 mM dithiothreitol (lane 2). Left panel, preimmune serum; right panel, anti-AMPD2 serum. (B) Western blot analysis (immune serum only) of 1A/2 purified in the presence of leupeptin and E-64. Lane 1, stored at )20 °C for 62 days in SDS– PAGE loading buffer. Lanes 2–6, stored at 4 °C for 62 days in the presence of no inhibitor (lane 2), EDTA (lane 3), aprotinin (lane 4), leupeptin (lane 5), and E-64 (lane 6). Molecular size markers are indicated.
E-64 ð10 lMÞ, APMA (5 mM), PMSF (1 mM), aprotinin ð5 lg=mlÞ, TPCK ð50 lMÞ, TLCK ð50 lMÞ, pepstatin ð1 lMÞ, EACA (5 mM), glycerol (1%, 10%, or 30%), leupeptin (5 lg/ml), or combinations thereof. Each inhibitor was added to a separate aliquot of the 1B/2 protein immediately following purification. A two-week incubation at 4 °C indicated that only E-64 and leupeptin were effective at limiting additional proteolysis (data not shown). With the exception of aprotinin (see below), all other inhibitors were without observable effect compared to the control aliquot that received no protease inhibitor. Therefore, a phosphocellulose purification of the 1A/2 protein was performed in the presence of both E-64 and leupeptin. One aliquot was immediately frozen in SDS–PAGE sample loading buffer, while others were stored at 4 °C for 62 days in the presence of selected inhibitors. Each 4 °C sample received a weekly supplement in an attempt to maintain the presence of active inhibitor. Western blot data presented in Fig. 2B demonstrate that inclusion of E-64 and leupeptin maintains the integrity of the 1A/2 protein during purification (lane 1). Moreover, these two protease inhibitors effectively maintain 1A/2 polypeptide integrity when added individually during prolonged storage at 4 °C (compare lanes 5 and 6 to lane 2). Notably, aprotinin appears to promote proteolysis beyond
298
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
that observed when no additional inhibitors are added following purification (compare lanes 2 and 4). The reason for this latter result is not readily apparent and we are unaware of any reports citing a similar outcome with other proteins. Confirmation of full-length, nonproteolyzed recombinant AMPD2 protein in mammalian cells N-terminal sequencing of the purified 1B/2 protein was attempted in order to directly confirm that E-64 and leupeptin inhibit N-terminal proteolysis of the naturally occurring AMPD2 proteins. However, these efforts indicated that the N-terminus of the polypeptide was blocked. Therefore, an alternative approach was developed to evaluate the retention of N- and C-terminal epitope tags (FLAG) constructed onto the 1A/2 protein. HeLa cells were transfected with mammalian expression plasmids encoding FLAG-1A/2 or 1A/2-FLAG, and lysates prepared in the presence of E-64 and leupeptin
and immunoprecipitated with either anti-AMPD2 or preimmune serum. Fig. 3 shows the results of Western blot analysis of these immunoprecipitates with antibody to the FLAG epitope tag. The presence of a FLAG-reactive band in anti-AMPD2 immunoprecipitates of lysates containing N-terminally tagged 1A/2 protein demonstrates the retention of the N-terminal epitope tag on this AMPD2 polypeptide (lane 2). A similarly sized anti-FLAG immunoreactive band is seen in antiAMPD2 immunoprecipitates of lysates containing Cterminally tagged AMPD2 polypeptide (lane 4). These results indicate that the C-terminal end of the immunoprecipitated protein is intact and demonstrate that most or all of the polypeptides isolated by this method are full length (compare lanes 2 and 4). Furthermore, these bands migrate at a subunit molecular mass that is similar to the untagged 1A/2 polypeptide (compare Fig. 3, lanes 2 and 4, with Fig. 2B, left lane), suggesting that the latter is also full-length. Expression and purification of AMPD2 proteins
Fig. 3. Identification of full-sized AMPD2 proteins in transfected HeLa cells. FLAG-tagged 1A/2 proteins were transiently expressed in HeLa cells and sonicates were prepared in the presence of leupeptin and E-64. Lysates containing N-terminally tagged 1A/2 (left lanes) or C-terminally tagged 1A/2 (right lanes) were immunoprecipitated with preimmune (lanes 1 and 3) or anti-AMPD2 (lanes 2 and 4) serum. Immunoprecipitated protein bands were detected by Western blot analysis using anti-FLAG monoclonal antibody. Molecular size markers are indicated.
The three naturally occurring N-terminal variants of isoform L, 1A/2, 1B/2, and 1B/3, were expressed in Sf9 insect cells infected with baculovirus engineered for the expression of these proteins. AMPD protein was purified from infected lysates by phosphocellulose chromatography in the presence of E-64 and leupeptin to final specific activities of 33–208 U/mg protein (Table 1). A sample elution profile of 1B/3 (Fig. 4A) shows that this AMPD2 recombinant enzyme is resolved from endogenous Sf9 AMPD activity by phosphocellulose chromatography. Due to higher expression, it was not possible to resolve the 1A/2 and DD184ð1B=2Þ proteins from the endogenous Sf9 AMPD activity; however, it is reasonable to assume from the position of elution of the endogenous activity in several other protein preparations that the peak 1A/2 and DD184ð1B=2Þ fractions contain little or no endogenous Sf9 activity. The identity and integrity of each protein were confirmed by SDS–PAGE and Coomassie blue staining, or by Western blot using antiserum specific to the AMPD2
Table 1 Purification of human AMPD2 recombinant enzymesa
DD184ð1B=2Þ 1A/2 1B/2 1B/3 a
Crude Phosphocellulose Crude Phosphocellulose Crude Phosphocellulose Crude Phosphocellulose
Total protein (mg)
Units enzyme
S.A. (U/mg)
252 1.2 203 1.4 217 0.9 214 0.8
944 657 822 291 99 42 141 26
3.7 548 4.0 208 0.5 47 0.7 33
Yield (%) 70 35 42 18
Human AMPD2 recombinant enzymes were expressed in Sf9 cells and purified by batch adsorption to phosphocellulose with KCl gradient elution, as described in Materials and methods. Representative results from a single purification are shown.
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
299
Fig. 5. SDS–PAGE analysis of human AMPD2 recombinant proteins. One hundred milliunits of enzyme activity from phosphocellulose peak fractions was fractionated on a 8.5% acrylamide gel and stained with Coomassie blue. Lane 1, molecular size markers; lane 2, 1A/2; lane 3, 1B/2; lane 4, 1B/3, and lane 5, DD184.
protein bands loaded for equivalent amounts of enzyme activity (Fig. 5). This analysis removes the component of contaminating proteins in each preparation and suggests that the 1A/2, 1B/3, and DD184ð1B=2Þ enzymes similarly have specific activities higher than 1B/2. Native molecular mass of AMPD2 proteins
Fig. 4. Resolution of human AMPD2 recombinant and Sf9 AMPD endogenous enzymes by phosphocellulose chromatography. (A) Elution profiles of enzyme activities isolated from sonicates prepared from noninfected Sf9 cells ( ) and Sf9 cells infected with 1B/3 recombinant virus ( ). (B) Left panel: SDS–PAGE analysis of 2 lg total protein from crude sonicate (lane 1) and phosphocellulose peak activity fraction (lane 2) of 1B/3 recombinant virus-infected Sf9 cells. Gel is stained with Coomassie blue. Right panels: Western blot analysis of 50 mU of AMPD activity from crude sonicates of noninfected (lane 1) and 1B/3 recombinant virus-infected Sf9 cells (lane 2). Left blot, preimmune serum; right blot, anti-AMPD2 serum. Molecular size markers are indicated.
polypeptide. A representative analysis of 1B/3 is shown in Fig. 4B. Specific activity comparison of the AMPD2 proteins Table 1 shows that the 1A/2 and DD184ð1B=2Þ proteins are purified to specific activities higher than either the 1B/2 or 1B/3 proteins. However, this difference in specific activity may have been due to the relative lower expression of 1B/2 and 1B/3, which may result in a proportionately higher content of protein contaminants in these latter preparations. In order to address this issue, the intrinsic specific activity of each AMPD2 protein was estimated by Coomassie blue staining of
Two methods were used to measure the native molecular mass of each AMPD2 protein: gel filtration chromatography and chemical crosslinking. Fig. 6 shows gel filtration elution profiles of each naturally occurring AMPD2 variant and three deletion mutants (DM55½1B=2 , DE129½1B=2 , and DD184½1B=2 ). Only the DD184ð1B=2Þ enzyme elutes from this column in a volume that is consistent with a native tetrameric mass. All other AMPD2 proteins exhibit aberrant elution profiles. The inconsistency between calculated and observed molecular mass for these proteins can be correlated with N-terminal amino acid composition (Table 2). Proteins that contain both exon 2- and exon 3-encoded sequence (1A/2, 1B/2 and DM55½1B=2 ) elute with observed molecular masses 56–71% greater than expected, while those that contain only exon 3-encoded sequence (1B/3 and DE129½1B=2 ) elute 30–31% larger than expected. Fig. 6 also shows that some of each naturally occurring variant of isoform L elutes in the void volume of the column, although the relative proportion that behaves in this fashion varies, i.e., 1B=3 > 1B=2 > 1A=2. This aberrant gel filtration behavior can be correlated with the relative high proline content of each unique Nterminal extension. For example, the 1B/3 protein has the largest proportion of enzyme activity eluting in the void volume and the highest percentage of prolines in its
300
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
Fig. 6. Gel filtration analysis of human AMPD2 recombinant enzymes. Approximately 15 U of each enzyme was loaded onto a 1:5 cm 90 cm glass column packed with Sephacryl S300-HR resin (exclusion volume >2000 kDa). Void volume (Vo ) was determined by the elution of blue dextran and the resolving range of the column was calibrated with known globular standards. Fractions were assayed for AMPD activity, as described in Materials and methods. Ve, elution volume (observed); Vexp, elution volume (expected).
N-terminal extension ( 25% or 13/58 residues), compared to 1B/2 ( 12% or 15/128 residues) and 1A/2 ( 13% or 6/47 residues). In contrast, the AMPD2 core polypeptide has a proline content of only 6% (45/760 residues).
Aberrant gel filtration behaviors notwithstanding, chemical crosslinking analysis indicates that 1A/2 and 1B/2 enzymes are composed of four subunits, similar to the DD184ð1B=2Þ protein (Fig. 7). In the absence of crosslinker, all three AMPD2 proteins migrate
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
301
Table 2 Quantitative comparison of the gel filtration chromatography behaviors of human AMPD2 recombinant enzymesa AMPD2 variant
Exon 2
Exon 3
Calculated MW (kDa)
Observed MW (kDa)
Difference (kDa)
% Increase
1A/2 1B/2 DM55ð1B=2Þ 1B/3 DE129ð1B=2Þ DD184ð1B=2Þ
+ + + ) ) )
+ + + + + )
368 403 380 372 353 322
630 630 650 490 460 330
262 227 270 118 107 8
71 56 71 31 30 2
a Exon 2 (+) denotes the presence of protein sequence contributed by exon 2; Exon 3 (+) denotes the presence of sequence contributed by exon 3. Calculated MW indicates molecular mass derived from amino acid composition. Observed MW was calculated from the Ve of each protein in gel filtration analysis, using the equation logMW ¼ 2:272ðVe =39:2Þ þ 8:4051, derived from plotting the elution volumes of blue dextran and four globular protein standards.
Fig. 7. Phosphate-buffered acrylamide gel and chemical crosslinking analysis of human AMPD2 recombinant proteins. Five micrograms of each protein was treated with increasing concentrations (0–2 mM) of BS3 , a primary amine-reactive noncleavable crosslinking reagent. Products were fractionated on a 3.5% denaturing gel and stained with Coomassie blue. Molecular size markers are indicated. #,monomer; *, tetramer.
through the gel as a single predominant band, while incubation with increasing amounts of crosslinker results in the appearance of higher molecular weight species consistent with dimers, trimers, and tetramers. The purity of the 1B/3 protein was not sufficient to
allow for successful crosslinking experiments with this protein. However, it is reasonable to hypothesize that the 1B/3 protein is also a tetramer because it shares the core 734 amino acid AMPD2 polypeptide with 1A/2 and 1B/2.
302
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
Discussion AMP deaminase (AMPD) polypeptides are sensitive to limited proteolysis during purification and subsequent storage at 4 °C. This degradation was observed initially in freshly prepared rabbit skeletal muscle AMPD after storage at 4 °C [15]. Further evidence of proteolysis was provided by a comparative analysis between reported subunit molecular masses for purified human AMPD isoforms and those predicted from cloned sequences [14]. Protease cleavage sites have been identified in AMPD polypeptides isolated from a variety of sources and available data indicate that in vitro proteolysis is restricted to N-terminal sequences [20–23]. Therefore, it is important to consider that N-terminal sequences missing from purified AMPD enzymes might impact the behavior of this enzyme. For example, human AMPD1 and AMPD3 recombinant proteins can be isolated with subunits essentially intact and these enzymes exhibit significant differences in catalytic and contractile protein binding behaviors compared to their respective N-truncated derivatives [20,24]. Since most of the available information on isoform L has been generated with partially proteolyzed enzymes, further characterizations of these intact AMPD2 proteins are warranted. This study has demonstrated that leupeptin and E-64 can be used to effectively inhibit the in vitro proteolysis of human AMPD2 recombinant enzymes. The addition of these two protease inhibitors to buffers during the isolation, purification, and subsequent storage of these enzymes prevents detectable reductions in subunit molecular mass that readily occur in their absence. Leupeptin and E-64 are both inhibitors of calpain, a ubiquitous neutral cysteine protease (reviewed in [30]), which exhibits a preference for hydrophobic residues at the P2 position [31] and typically performs limited rather than digestive proteolysis [32]. In these respects, it is worth noting that inspection of reported proteolytic cleavage sites in the N-terminal sequences of all mammalian AMPD polypeptides reveals a preference for hydrophobic residues at the P2 position [20,23]. Calpain is therefore a potential candidate for a processor of AMPD enzymes. Gel filtration and chemical crosslinking data together suggest that all three naturally occurring AMPD2 variants are tetrameric enzymes that exhibit deviation from a compact globular structure in their N-terminal extensions. Each activity migrates faster than expected through a gel filtration column, while an AMPD2 deletion mutant comprised of the core polypeptide migrates as expected. In addition, some of each naturally occurring variant elutes in the void volume of the gel filtration column. These unexpected behaviors can be correlated with the unique configuration of 50 exon-encoded sequence, and with the relatively high proline
content, respectively, in each alternative AMPD2 polypeptide. Regardless of the structural basis for aberrant gel filtration behavior, the combined data suggest that each unique N-terminal extension adopts an extended nonglobular conformation. These structures would disproportionately expand the surface area of each enzyme, promote greater exclusion from molecular sieving beads, and also provide greater access to proteases. Crosslinking analysis using bis(sulfosuccinimidyl) suberate ðBS3 Þ indicates that AMPD2 enzymes have tetrameric structures and also suggests that abnormal gel filtration behavior is not due to higher-order structure. Previous gel filtration and sucrose gradient sedimentation analyses have also indicated a tetrameric structure for AMPD [4,33], but crosslinking data presented here are the first direct demonstration that the AMPD2 molecule consists of four subunits. With the identification of protease inhibitors that can be used to facilitate the isolation of intact AMPD2 proteins, the question of how unique N-terminal extensions affect the functional behaviors of isoform L can now be addressed. This will extend previous studies on the significance of divergent N-terminal sequences across different isoforms. In addition, the inclusion of E64 and leupeptin in the purification of human AMPD1 and AMPD3 recombinant enzymes can be used to maintain the integrity of their subunits during storage at 4 °C (data not shown). Therefore, the results of this study should be useful to other investigators working with AMP deaminase.
Acknowledgments This work was supported by Public Health Service Grant DK-50902 from the National Institutes of Health.
References [1] J. Lowenstein, K. Tornheim, Ammonia production in muscle: the purine nucleotide cycle, Science 171 (1971) 397–400. [2] A.G. Chapman, D.E. Atkinson, Stabilization of adenylate energy charge by the adenylate kinase reaction, J. Biol. Chem. 248 (1973) 8309–8312. [3] H. Zimmerman, 50 nucleotidase: molecular structure and functional aspects, Biochem. J. 285 (1992) 345–365. [4] N. Ogasawara, H. Goto, Y. Yamada, T. Watanabe, T. Asano, AMP deaminase isozymes in human tissues, Biochim. Biophys. Acta 714 (1982) 298–306. [5] R.L. Sabina, T. Morisaki, P. Clarke, R. Eddy, T.B. Shows, C.C. Morton, E.W. Holmes, Characterization of the human and rat myoadenylate deaminase genes, J. Biol. Chem. 265 (1990) 9423– 9433. [6] M.T. Bausch-Jurken, D.K. Mahnke-Zizelman, T. Morisaki, R.L. Sabina, Molecular cloning of AMP deaminase isoform L:
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
sequence and bacterial expression of human AMPD2 cDNA, J. Biol. Chem. 267 (1992) 22407–22413. D.K. Mahnke-Zizelman, R.L. Sabina, Cloning of human AMP deaminase isoform E cDNAs: evidence for a third AMPD gene exhibiting alternatively spliced 50 exons, J. Biol. Chem. 267 (1992) 20866–20877. D.K. Mahnke-Zizelman, F. Van den Bergh, M.T. Bausch-Jurken, R. Eddy, S. Sait, T.B. Shows, R.L. Sabina, Cloning, sequence, and characterization of the human AMPD2 gene: evidence for transcriptional regulation by two closely spaced promoters, Biochim. Biophys. Acta 1308 (1996) 122–132. T. Morisaki, R.L. Sabina, E.W. Holmes, Adenylate deaminase: a multigene family in humans and rats, J. Biol. Chem. 265 (1990) 11482–11486. Z.Y. Chang, P. Nygaard, A.C. Chinault, R.E. Kellems, Deduced amino acid sequence of Escherichia coli adenosine deaminase reveals evolutionarily conserved amino acid residues: implications for catalytic function, Biochemistry 30 (1991) 2273–2280. M. Gross, H. Morisaki, T. Morisaki, E.W. Holmes, Identification of functional domains in AMPD1 by mutational analysis, Biochem. Biophys. Res. Commun. 205 (1994) 1010–1017. I. Mineo, P.R.H. Clarke, R.L. Sabina, E.W. Holmes, A novel pathway for alternative splicing: identification of an RNA intermediate that generates an alternative 50 splice donor site not present in the primary transcript of AMPD1, Mol. Cell. Biol. 10 (1990) 5271–5278. F. Van den Bergh, R.L. Sabina, Characterization of human AMP deaminase 2 (AMPD2) gene expression reveals alternative transcripts encoding variable N-terminal extensions of isoform L, Biochem. J. 312 (1995) 401–410. R.L. Sabina, D.K. Mahnke-Zizelman, Towards an understanding of the functional significance of N-terminal domain divergence in human AMP deaminase isoforms, Pharm. Therap. 87 (2000) 279– 283. M. Ranieri-Raggi, A. Raggi, Effects of storage on activity and subunit structure of rabbit skeletal muscle AMP deaminase, Biochem. J. 189 (1979) 367–368. M. Raneiri-Raggi, A.J.G. Moir, A. Raggi, Interaction with troponin T from white skeletal muscle restores in white skeletal muscle AMP deaminase those allosteric properties removed by limited proteolysis, Biochim. Biophys. Acta 827 (1985) 93–100. R. Marquetant, R.L. Sabina, E.W. Holmes, Identification of a noncatalytic domain in AMP deaminase that influences binding to myosin, Biochemistry 28 (1989) 8744–8749. D.J. Merkler, A.S. Wali, J. Taylor, V.L. Schramm, AMP deaminase from yeast: role in AMP degradation, large scale purification, and properties of the native and proteolyzed enzyme, J. Biol. Chem. 264 (1989) 21422–21430. O.P. Chilson, A.E. Kelly-Chilson, N.R. Siegel, AMP deaminases from chicken and rabbit muscle: partial primary sequences of
[20]
[21]
[22]
[23]
[24]
[25]
[26] [27]
[28]
[29]
[30] [31] [32]
[33]
303
homologous 17 kDa CNBr fragments: autorecognition by rabbit anti-chicken AMPD, Comp. Biochem. Phys. 116B (1997) 371– 377. D.K. Mahnke-Zizelman, P.C. Tullson, R.L. Sabina, Novel aspects of tetramer assembly and N-terminal domain structure and function are revealed by recombinant expression of human AMP deaminase isoforms, J. Biol. Chem. 273 (1998) 35118– 35125. S.L. Meyer, K.L. Kvalnes-Krick, V.L. Schramm, Characterization of AMD, the AMP deaminase gene in yeast. Production of amd strain, cloning, nucleotide sequence, and properties of the protein, Biochemistry 28 (1989) 8734–8743. R. Ronca, M. Raneiri-Raggi, P.E. Brown, A.J.G. Moir, A. Raggi, Evidence of a species-differentiated regulatory domain within the N-terminal region of skeletal muscle AMP deaminase, Biochim. Biophys. Acta 1209 (1994) 123–129. K. Knecht, K. Wiesmuller, V. Gnau, G. Jung, R. Meyermann, K.G. Todd, B. Hamprecht, AMP deaminase in rat brain: localization in neurons and ependymal cells, J. Neurosci. Res. 66 (2001) 941–950. D.K. Mahnke-Zizelman, R.L. Sabina, Localization of N-terminal sequences in human AMP deaminase isoforms that influence contractile protein binding, Biochem. Biophys. Res. Commun. 285 (2001) 489–495. B. Sims, D.K. Mahnke-Zizelman, A.A. Profit, G.D. Prestwich, R.L. Sabina, A.B. Thiebert, Regulation of AMP deaminase by phosphoinositides, J. Biol. Chem. 274 (1999) 25701– 25707. B. Ashby, H. Holmsen, Platelet AMP deaminase: purification and kinetic studies, J. Biol. Chem. 256 (1981) 10519–10523. B. Ashby, H. Holmsen, Platelet AMP deaminase: regulation by Mg-ATP2- and inorganic phosphate and inhibition by the transition state analog coformycin, J. Biol. Chem. 258 (1983) 3668–3672. M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein using the principle of dyebinding, Anal. Biochem. 72 (1976) 248–254. U.K. Laemmli, Cleavage of structural proteins during the assembly of the head of bacteriophage T4, Nature 227 (1970) 680–685. H. Sorimachi, K. Suzuki, l-Calpain, in: Handbook of Proteolytic Enzymes, Academic Press, New York, 1998. R.L. Mellgren, T. Murachi, Intracellular calcium-dependent proteolysis, CRC press, Boston, 1990. T.C. Saido, H. Sorimachi, K. Suzuki, Calpain: new perspectives in molecular diversity and physiological-pathological involvement, FASEB J. 8 (1994) 814–822. S. Yun, C.H. Suelter, Human erythrocyte 50 -AMP aminohydrolase: purification and characterization, J. Biol. Chem. 253 (1978) 404–408.
Expression, purification, and inhibition of in vitro proteolysis of human AMPD2 (isoform L) recombinant enzymes Amy Louise Haas and Richard L. Sabina* Department of Biochemistry, Medical College of Wisconsin, 8701 Watertown Plank Road, Milwaukee, WI 53226, USA Received 8 July 2002, and in revised form 25 September 2002
Abstract AMP deaminase (AMPD) is a multigene family in higher eukaryotes whose three members encode tetrameric isoforms that catalyze the deamination of AMP to IMP. AMPD polypeptides share conserved C-terminal catalytic domains of 550 amino acids, whereas divergent N-terminal domains of 200–330 amino acids may confer isoform-specific properties to each enzyme. However, AMPD polypeptides are subject to limited N-terminal proteolysis during purification and subsequent storage at 4 °C. This presents a technical challenge to studies aimed at determining the structural and functional significance of these divergent sequences. This study describes the recombinant overexpression of three naturally occurring human AMPD2 proteins, 1A/2, 1B/2, and 1B/3, that differ by N-terminal extensions of 47–128 amino acids, resulting from the use of multiple promoters and alternative splicing events. A survey of protease inhibitors reveals that E-64 and leupeptin are able to maintain the subunit structure of each AMPD2 protein when they are included in extraction and storage buffers. Gel filtration chromatography of these three purified AMPD2 enzymes comprised of intact subunits reveals that each migrates faster than expected, resulting in observed molecular masses significantly greater than those predicted for native tetrameric structures. However, chemical crosslinking analysis indicates four subunits per AMPD2 molecule, confirming that these enzymes have a native tetrameric structure. These combined results suggest that AMPD2 N-terminal extensions may exist as extended structures in solution. Ó 2002 Elsevier Science (USA). All rights reserved.
AMP deaminase (AMPD; EC 3.5.4.6) catalyzes the hydrolytic deamination of adenosine-50 -monophosphate (AMP) to inosine-50 -monophosphate (IMP) and ammonium ion. The AMPD enzyme is an important regulator of cellular energy metabolism through its participation in purine nucleotide catabolism [1–3]. Multiple isoforms have been isolated from different human tissues and are named after the source of their purification: isoforms M, muscle; L, liver; and E, erythrocyte [4]. Subsequent cloning of three human AMPD genes has revealed the molecular basis for these different isoforms: AMPD1, M; AMPD2, L; and AMPD3, E [5–8]. It is likely that the three AMPD genes arose from duplication of a common primordial gene [9], and subsequently, acquired differences via divergent evolution. Consistent with this hypothesis, AMPD isoforms contain both conserved and divergent domains. * Corresponding author. Fax: 1-414-456-6510. E-mail address: [email protected] (R.L. Sabina).
The three AMPD polypeptides share a similar 550 amino acid C-terminal end (62–70% identical; [6,7]) that contains a motif SLSTDDP believed to be the catalytic center of the enzyme [10,11]. Conversely, each AMPD polypeptide differs by divergent N-terminal sequences of 200–330 amino acids with less than 36% identity to each other. In addition, differential promoter use and alternative splicing add extensions or substitutions of four (AMPD1; [5,12]), 47–128 (AMPD2; [13]), and 7–9 (AMPD3; [7]) amino acids at the distal N-terminal end of each AMPD polypeptide. Available information suggests that different N-terminal domains and distal Nterminal variations in each AMPD polypeptide contribute to isoform-specific behaviors of this enzyme [14]. AMPD proteins from human and other sources are susceptible to proteolysis in vitro [11,15–20] and proteolytic events have been localized exclusively to the N-terminal region of these polypeptides [20–23]. Consequently, it is likely that prior studies using purified AMPD enzymes have been performed with subunits
1046-5928/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 6 - 5 9 2 8 ( 0 2 ) 0 0 6 3 6 - 8
294
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
lacking varying amounts of N-terminal sequence. It is therefore important to explore the possibility that unique N-terminal domains may affect the function of each AMPD enzyme. Although proteolysis has not yet been demonstrated in vivo, the possibility also exists that intracellular processing of these proteins could serve to remove unique N-terminal sequence and thereby abrogate any related function conferred to the respective proteins. Examination and comparison of intact and proteolyzed proteins can yield information about the impact of N-terminal extensions on AMPD2 structure and function, and may provide a basis for further studies of these proteins in vivo. Recombinant technology affords the means by which intact AMPD proteins can be expressed and purified. This method has been used to generate predominantly full-length human AMPD1 and AMPD3 enzymes, which has facilitated studies designed to investigate the functional significance of divergent N-terminal sequences in these polypeptides [20,24,25]. However, AMPD1 and AMPD3 recombinant proteins are susceptible to proteolysis during extended storage at 4 °C [20]. Expression and purification of full-length, human AMPD2 enzymes have not yet been reported. However, N-terminal proteolysis removes as much as 25–35 kDa from these proteins when they are isolated from endogenous sources [4,13,26,27]. Consequently, little is known about how the N-terminal domain of AMPD2, and extensions conferred on it by the shuffling of 50 exons, affect the function of this isoform. This study describes the recombinant expression of three naturally occurring AMPD2 variants (1A/2, 1B/2, and 1B/3) and identifies protease inhibitors that can be used to minimize in vitro proteolysis during their purification and subsequent characterization. This information should be useful to all investigators working with AMP deaminases.
Materials and methods Materials Restriction endonucleases were obtained from Roche (HindIII, BamHI, and BglII), Gibco Life Sciences (SacI), and New England Biolabs (NcoI and SapI). GraceÕs insect cell culture medium, DMEM, fetal bovine serum, lipofectamine, DNA primers, and DNA ligase were obtained from Gibco Life Sciences. The baculoviral transfer plasmid pBlueBacIII (pBB3) was purchased from Invitrogen. BaculoGold modified baculoviral DNA was obtained from Pharmingen. The mammalian expression plasmid pCMV-FLAG and antibody to the FLAG peptide epitope tag (monoclonal M2) were obtained from Stratagene. Goat anti-mouse horseradish peroxidase (HRP)-conjugated IgG (ImmunoPure),
chemiluminescence reagents (Supersignal), and the primary amine crosslinker BS3 (Bis(sulfosuccinimidyl) suberate) were purchased from Pierce. Goat anti-rabbit HRP-conjugated IgG was obtained from Santa Cruz Biochemicals. Pansorbin was purchased from Calbiochem. Sephacryl S300-HR size exclusion chromatography resin and globular gel filtration calibration standards were obtained from Amersham Pharmacia. Phosphocellulose resin (P-11) was supplied by Whatman, Ltd. Glass chromatography columns ð1:5 30 cmÞ and 4-chloro-naphthol were purchased from Bio-Rad. Hydrogen peroxide (30%) and adenosine-50 -monophosphate (AMP) were purchased from Sigma–Aldrich Chemical Company. The protease inhibitors phosphoramidon, acetyl phenylmercuric acetate (APMA), phenylmethylsulfonyl fluoride (PMSF), aprotinin, Na tosyl-phe-chloromethyl ketone (TPCK), Na -tosyl-lyschloromethyl ketone (TLCK), pepstatin, and epsilon amino caproic acid (EACA) were gifts of Sally Twining, Ph.D. (Medical College of Wisconsin, Milwaukee, WI). Leupeptin, E-64, ethylenedinitrilotetraacetic acid (EDTA), and ethyleneglycol-bis-(b-aminoethyl ether) NNN 0 N 0 -tetraacetic acid (EGTA) were obtained from Roche Biochemicals. Glycerol was purchased from Shelton Scientific (Shelton, CT). DNA and protein sequencing were performed by the Medical College of Wisconsin Protein and Nucleic Acid Facility, Milwaukee, Wisconsin. Construction of baculoviral transfer vectors for expression of wild-type AMPD2 variants A baculoviral transfer vector was constructed that contained the region encoding the unique 1A/2 N-terminal extension (pBB3-1A/2). This was accomplished by PCR amplification of a partial 1A/2 clone [13] using the primers 50 -AATAGATCTCCACCATGGCCTCAG AGGCTC-30 and 50 -CTTCTCCCGGATGAAGAGC-30 . This added a BglII restriction site at the 50 end of the PCR product (underlined) and maintained an internal SapI site (italics). The PCR product was digested with BglII and SapI, and then ligated with a SapI–HindIII fragment containing the remainder of the AMPD2 coding sequence into the baculoviral transfer plasmid pBB3 (Invitrogen). A baculoviral transfer vector containing the region encoding the unique 1B/2 N-terminal extension (pBB31B/2) was produced by PCR amplification of an existing partial 1B/2 clone [13] using the primers 50 -TATA GGATCCAGATCT GCCACCATGAGAAATCGTGG CCAG-30 and 50 -GGATGCCATGGCTGGGACC-30 . This added BamHI and BglII restriction sites at the 50 end (underlined and underlined with italics, respectively) and maintained an NcoI site at the 30 end of the PCR product (italics only). Following digestion with BglII and NcoI, the restricted PCR product was ligated with
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
an NcoI–HindIII fragment of an existing AMPD2 clone containing the remainder of the coding sequence [24] into BglII–HindIII digested pBB3. The coding region for the unique 1B/3 N-terminal extension was reconstructed from existing clones by overlap PCR in order to produce a 1B/3 baculoviral transfer vector (pBB3-1B/3) as follows: for the first round of PCR, the primers 50 -TATAGGATCCAGATCT GC CACCATGTGGCAGAGCCAG-30 and 50 -GGAGCT GCAGTGGAGGCCTGCAGGC-30 were used to amplify the exon 1B-encoded fragment of the 1B/3 coding region from pBB3-1B/2. This primer added a BamHI recognition site (underlined) and a BglII site (underlined with italics) to the 50 end of the PCR product. The primers 50 -GCCTGCAGGCCTCCACTGCAGCTCC30 and 50 -CAG GAAATCTTGCTTGG-30 [6] were then used to amplify the exon 1B–exon 3 junction from a partial 1B/3 clone [13]. The products of these two PCR reactions were then used as templates for overlap PCR amplification with the two external primers from the initial amplifications. This produced a cDNA fragment encoding the entire 1B/3 extension that was used as a template for a second round of overlap PCR. This was accomplished using a second fragment generated by PCR amplification of the region encoding exon 3 from pBB31B/2 with the primers 50 -CCAAGCAAGATTTCCTG-30 and 50 -CAGACCCAAGCCCAGGT-30 [6]. The overlap PCR extended the 1B/3 cDNA 30 of a SapI recognition site that was used in the final cloning step. This product was digested with BglII and SapI and ligated into pBB3 together with a SapI–HindIII fragment encoding the remainder of the AMPD2 coding sequence. All DNA plasmids were amplified and then purified by cesium chloride banding. cDNA identity and integrity were confirmed by restriction digest and DNA sequencing. Baculoviral transfer vectors for the AMPD2 deletion mutants of the 1B/2 polypeptide, DM55, DE129, and DD184, have been described previously [24]. Construction of plasmids for mammalian expression of AMPD2 fusion proteins An expression plasmid for N-terminally tagged 1A/2 (pCMV-FLAG-1A/2) was generated by ligation of a BglII fragment from pBB3-1A/2 in-frame into pCMVFLAG version 2B (Stratagene). A DNA fragment beginning with a SacII restriction endonuclease site 375 base pairs 50 to the stop codon and ending with the 30 end of the AMPD2 cDNA coding region that excluded the stop codon was used to generate an expression plasmid for C-terminally tagged 1A/2 (pCMV-1A/2FLAG). This was produced by PCR using the 50 primer 50 -CCTGTCCCGCGG CCTCA-30 , which contains the SacII site (italics), and the 30 primer 50 -TATA AAGCTTTTGAGGCCCTGGGCTCATGG-30 , which removes the stop codon by mutagenesis and adds a
295
HindIII site to the 30 end of the DNA fragment (underlined). The PCR product was digested with SacII and HindIII, and then ligated with a BamHI–SacII DNA fragment isolated from pBB3-1A/2 into BamHI–HindIII digested pCMV-FLAG version 4A. All expression plasmids were confirmed by DNA sequencing and purified by cesium chloride banding. Production, isolation, and purification of AMPD2 recombinant baculovirus Baculoviral transfer vectors encoding 1A/2, 1B/2, or 1B/3 were co-transfected with modified baculoviral genomic DNA into Sf9 cells by calcium phosphate precipitation to generate AMPD2-expressing virus by homologous recombination, as described previously [20]. Viral clones were purified by plaque assay and amplified by successive infection of Sf9 cells. Expression and purification of AMPD2 proteins Twelve T-185 flasks of Sf9 cells at 95% confluence were infected with high titer virus amplified from plaque-purified AMPD2 clones. Phosphocellulose purification was performed, as described previously [20]. Protease inhibitors were included as indicated. Briefly, cells were incubated with virus for four days, then loosened from the flasks by rapping, and collected by centrifugation at 1000g in a tabletop centrifuge at 4 °C. Viral supernatant was removed and cells were resuspended in cold column buffer (20 mM KPO4 , pH 6.7, 100 mM KCl, and 0.1% b-ME), supplemented with 5 lg=ml leupeptin and 10 lM E-64, unless otherwise indicated. Cells were lysed by sonication and lysates were clarified by centrifugation for 10 min at 10,000g at 4 °C. Crude homogenate was batch adsorbed to buffer-equilibrated phosphocelluose for 30 min with rotation at 4 °C. The resin was then washed with two volumes of column buffer and poured into a 1:5 cm 30 cm column. Proteins were eluted with a linear gradient of 0.1–2 M KCl in 20 mM KPO4 , pH 6.7, in the presence of protease inhibitors. Fractions were assayed for protein concentration [28] and AMPD activity. Fig. 1 schematically illustrates encoded polypeptides for all purified human AMPD2 recombinant enzymes used in this study. In addition to the naturally occurring variants of isoform L (1A/2, 1B/2, and 1B/3), three N-terminal deletion enzymes were prepared. All are numbered according to the 1B/2 polypeptide: DM55ð1B=2Þ is truncated to an in-frame methionine codon in the exon 1B sequence; DE129ð1B=2Þ begins with exon 3-encoded sequence and represents the core AMPD2 polypeptide common to all three naturally occurring variants of isoform L; DD184ð1B=2Þ starts near to identified protease sites in the rat brain AMPD2 polypeptide [23]. These N-truncated enzymes were used in comparative
296
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
Expression of AMPD2 fusion proteins in mammalian cells
Fig. 1. Schematic of human AMPD2 (isoform L) polypeptides. The three naturally occurring variants, 1A/2, 1B/2, and 1B/3, have N-terminal extensions of 47, 128, and 53 residues, respectively. These polypeptides are encoded by alternative AMPD2 mRNAs produced by the regulated expression of different promoters upstream of exons 1A and 1B and a cassette-type alternative splicing event involving exon 2 that results in the use of a different reading frame in exon 1B when this sequence is spliced directly to exon 3. Three additional N-truncated recombinant enzymes with progressive deletions of the 1B/2 polypeptide (DM55, DE129, and DD184) are included for comparative analyses with the naturally occurring variants of isoform L. Polypeptides are numbered from the C-terminus in order to better illustrate the three Nterminal extensions.
analyses together with the three wild-type variants as a means of assessing the structural roles of divergent N-terminal extensions in each AMPD2 polypeptide. Polyacrylamide gel electrophoresis Crude lysates and purified proteins were separated by SDS–PAGE using an 8% gel according to the method of Laemmli [28]. Protein bands were detected by staining with Coomassie blue R or by Western blotting. AMPD enzyme assays Enzyme assays were performed at 37 °C in 25 mM imidazole, pH 6.5, 100 mM KCl, and 0.2 mg/ml BSA, using 20 mM AMP. Substrate and product (IMP) were separated by anion-exchange HPLC using an ammonium dihydrogen phosphate (5–750 mM) and pH (2.8– 4.0) gradient and quantified by peak integration. Assays were performed at initial velocity conditions, with product not exceeding 10% of substrate. Inhibition of in vitro proteolysis Aliquots of purified protein (1.5 U; 30 lg, in 100 ll total volume) were incubated with protease inhibitors (see text for a complete list and concentrations used) at 4 °C for two to eight weeks. Proteins were separated by SDS–PAGE and visualized by Coomassie blue staining or by Western blotting.
HeLa cells were maintained in DMEM supplemented with 10% FBS at 37 °C with 5% CO2 . For transfection, cells were grown to 80–90% confluence in T-75 flasks and transiently transfected with plasmid DNAs encoding FLAG-1A/2 or 1A/2-FLAG via lipofectamine (GibcoBRL) according to manufacturerÕs protocol. Cells were then maintained at 37 °C with 5% CO2 and harvested 48 h post-transfection. For harvest, cells were washed twice in PBS, then covered with 1 ml MTMSB (50 mM KCl, 10 mM KPO4 , pH 6.7, 1% NP40, and 1 mg/ml BSA), and dislodged from the plate by scraping. Resuspended cells were lysed by sonication and clarified by centrifugation at 4 °C for 10 min at 10,000g. Lysates were assayed for protein concentration and enzyme activity. Immunoprecipitation and Western blotting For AMPD2 Western blots, crude lysate or purified protein was separated by SDS–PAGE in an 8% gel, using the method of Laemmli [29], and electroblotted onto a nitrocellulose membrane. Membranes were blocked in 2.5% milk buffered in TBS with 0.1% Tween 20 for 1 h at room temperature with agitation and then washed four times for 1 min each with TBS containing 0.5% Tween 20. Blots were then incubated with a 1:500 dilution of polyclonal rabbit antisera raised against an N-truncated human AMPD2 recombinant enzyme [24] for 1 h at room temperature with agitation and washed as before. Bound primary antibody was then complexed with a 1:2000 dilution of HRP-conjugated goat antirabbit secondary antibody for 1 h at room temperature with agitation, and unbound complexes were washed from the blot as before. Protein bands were detected by incubation with H2 O2 (0.013%) and 4-chloro-naphthol. For FLAG Western blots, clarified lysates from pCMV-FLAG-1A/2 or pCMV-1A/2-FLAG-transfected or nontransfected HeLa cells were immunoprecipitated with 3 ll undiluted preimmune or anti-AMPD2 sera for 2 h at 4 °C with rotation in 10 mM KPO4 , pH 6.7, 150 mM KCl, 1% NP40, and 1 mg/ml BSA. Immunocomplexes were incubated with 50 ll Pansorbin beads for 30 min at 4 °C with rotation and washed three times. Immunoprecipitates were separated on an 8% SDS– PAGE gel as before and transferred to nitrocellulose. Blots were blocked with 2.5% milk diluted in TBS with 1% Tween 20 and washed as before to remove excess blocker. The blots were then incubated with a 1:1000 dilution of monoclonal antibody to the FLAG epitope (M2, Stratagene), washed, and incubated with a 1:10,000 dilution of HRP-conjugated goat anti-mouse antibody. Protein bands were then detected by enhanced chemiluminescence. Clarified lysates ð150 lgÞ were also assayed as above to confirm the functional integrity of each tagged enzyme.
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
297
Gel filtration chromatography Gel filtration chromatography was performed, as described previously [20]. Briefly, 15 U of each AMPD activity was diluted to 1 ml in 50 mM imidazole, pH 6.5, containing 500 mM KCl and then loaded onto a 1 cm 90 cm Sephacryl S300-HR gel filtration column equilibrated in the same buffer at 4 °C. Fractions (0.5 ml) were collected and assayed for AMPD activity. For column calibration, the elution volumes of four known globular protein standards (aldolase, 158 kDa; catalase, 232 kDa; ferritin, 440 kDa; and thyroglobulin, 669 kDa) as well as blue dextran ðVo Þ were measured. Ve =Vo values were subjected to regression analysis to determine the equation logMW ¼ 2:3ðVe =Vo Þ þ 8:4, which was used to calculate the observed molecular mass of each AMPD protein. Chemical crosslinking Five micrograms of purified human AMPD recombinant protein was incubated with the homobifunctional, noncleavable, primary amine-crosslinking agent BS3 at 10- to 50-fold molar excess with respect to protein (final concentration 10 lM–10 mM) in 1.7 mM sodium citrate, pH 5.0, 1 M KCl, for 2 h on ice. Reactions were quenched by addition of Tris–HCl, pH 7.4, to a final concentration of 25 mM for 15 min on ice, then diluted in 2 SDS—PAGE sample buffer [29] and separated in a phosphate-buffered 3.5% acrylamide gel. Protein bands were visualized by Coomassie blue staining.
Results Inhibition of in vitro proteolysis of AMPD2 proteins Human AMPD1 and AMPD3 recombinant enzymes can be purified in the absence of protease inhibitors with subunits predominantly intact, but are susceptible to proteolysis during extended storage at 4 °C [20]. A representative Western blot analysis of the purified 1B/3 protein is shown in Fig. 2A and indicates that AMPD2 recombinant proteins are relatively more sensitive to proteolysis. Data in the right panel reveal that substantial proteolysis occurs during protein purification (lane 1) and then proceeds during an overnight dialysis at 4°C (lane 2). These results warranted efforts to prevent degradation of each AMPD2 polypeptide as a prerequisite to subsequent characterization of the three naturally occurring variants of isoform L. A preliminary experiment was performed in which a variety of protease inhibitors were screened for their ability to inhibit proteolysis of the 1B/2 protein. These included EDTA (1 mM), EGTA (1 mM), phosphoramidon ð10 lMÞ,
Fig. 2. Leupeptin and E-64 effectively inhibit the in vitro proteolysis of human AMPD2 recombinant proteins. (A) Western blot analysis of 1B/3 purified in the absence of protease inhibitors (lane 1) and after subsequent overnight dialysis at 4 °C in 50 mM imidazole, pH 6.5, containing 500 mM KCl and 1 mM dithiothreitol (lane 2). Left panel, preimmune serum; right panel, anti-AMPD2 serum. (B) Western blot analysis (immune serum only) of 1A/2 purified in the presence of leupeptin and E-64. Lane 1, stored at )20 °C for 62 days in SDS– PAGE loading buffer. Lanes 2–6, stored at 4 °C for 62 days in the presence of no inhibitor (lane 2), EDTA (lane 3), aprotinin (lane 4), leupeptin (lane 5), and E-64 (lane 6). Molecular size markers are indicated.
E-64 ð10 lMÞ, APMA (5 mM), PMSF (1 mM), aprotinin ð5 lg=mlÞ, TPCK ð50 lMÞ, TLCK ð50 lMÞ, pepstatin ð1 lMÞ, EACA (5 mM), glycerol (1%, 10%, or 30%), leupeptin (5 lg/ml), or combinations thereof. Each inhibitor was added to a separate aliquot of the 1B/2 protein immediately following purification. A two-week incubation at 4 °C indicated that only E-64 and leupeptin were effective at limiting additional proteolysis (data not shown). With the exception of aprotinin (see below), all other inhibitors were without observable effect compared to the control aliquot that received no protease inhibitor. Therefore, a phosphocellulose purification of the 1A/2 protein was performed in the presence of both E-64 and leupeptin. One aliquot was immediately frozen in SDS–PAGE sample loading buffer, while others were stored at 4 °C for 62 days in the presence of selected inhibitors. Each 4 °C sample received a weekly supplement in an attempt to maintain the presence of active inhibitor. Western blot data presented in Fig. 2B demonstrate that inclusion of E-64 and leupeptin maintains the integrity of the 1A/2 protein during purification (lane 1). Moreover, these two protease inhibitors effectively maintain 1A/2 polypeptide integrity when added individually during prolonged storage at 4 °C (compare lanes 5 and 6 to lane 2). Notably, aprotinin appears to promote proteolysis beyond
298
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
that observed when no additional inhibitors are added following purification (compare lanes 2 and 4). The reason for this latter result is not readily apparent and we are unaware of any reports citing a similar outcome with other proteins. Confirmation of full-length, nonproteolyzed recombinant AMPD2 protein in mammalian cells N-terminal sequencing of the purified 1B/2 protein was attempted in order to directly confirm that E-64 and leupeptin inhibit N-terminal proteolysis of the naturally occurring AMPD2 proteins. However, these efforts indicated that the N-terminus of the polypeptide was blocked. Therefore, an alternative approach was developed to evaluate the retention of N- and C-terminal epitope tags (FLAG) constructed onto the 1A/2 protein. HeLa cells were transfected with mammalian expression plasmids encoding FLAG-1A/2 or 1A/2-FLAG, and lysates prepared in the presence of E-64 and leupeptin
and immunoprecipitated with either anti-AMPD2 or preimmune serum. Fig. 3 shows the results of Western blot analysis of these immunoprecipitates with antibody to the FLAG epitope tag. The presence of a FLAG-reactive band in anti-AMPD2 immunoprecipitates of lysates containing N-terminally tagged 1A/2 protein demonstrates the retention of the N-terminal epitope tag on this AMPD2 polypeptide (lane 2). A similarly sized anti-FLAG immunoreactive band is seen in antiAMPD2 immunoprecipitates of lysates containing Cterminally tagged AMPD2 polypeptide (lane 4). These results indicate that the C-terminal end of the immunoprecipitated protein is intact and demonstrate that most or all of the polypeptides isolated by this method are full length (compare lanes 2 and 4). Furthermore, these bands migrate at a subunit molecular mass that is similar to the untagged 1A/2 polypeptide (compare Fig. 3, lanes 2 and 4, with Fig. 2B, left lane), suggesting that the latter is also full-length. Expression and purification of AMPD2 proteins
Fig. 3. Identification of full-sized AMPD2 proteins in transfected HeLa cells. FLAG-tagged 1A/2 proteins were transiently expressed in HeLa cells and sonicates were prepared in the presence of leupeptin and E-64. Lysates containing N-terminally tagged 1A/2 (left lanes) or C-terminally tagged 1A/2 (right lanes) were immunoprecipitated with preimmune (lanes 1 and 3) or anti-AMPD2 (lanes 2 and 4) serum. Immunoprecipitated protein bands were detected by Western blot analysis using anti-FLAG monoclonal antibody. Molecular size markers are indicated.
The three naturally occurring N-terminal variants of isoform L, 1A/2, 1B/2, and 1B/3, were expressed in Sf9 insect cells infected with baculovirus engineered for the expression of these proteins. AMPD protein was purified from infected lysates by phosphocellulose chromatography in the presence of E-64 and leupeptin to final specific activities of 33–208 U/mg protein (Table 1). A sample elution profile of 1B/3 (Fig. 4A) shows that this AMPD2 recombinant enzyme is resolved from endogenous Sf9 AMPD activity by phosphocellulose chromatography. Due to higher expression, it was not possible to resolve the 1A/2 and DD184ð1B=2Þ proteins from the endogenous Sf9 AMPD activity; however, it is reasonable to assume from the position of elution of the endogenous activity in several other protein preparations that the peak 1A/2 and DD184ð1B=2Þ fractions contain little or no endogenous Sf9 activity. The identity and integrity of each protein were confirmed by SDS–PAGE and Coomassie blue staining, or by Western blot using antiserum specific to the AMPD2
Table 1 Purification of human AMPD2 recombinant enzymesa
DD184ð1B=2Þ 1A/2 1B/2 1B/3 a
Crude Phosphocellulose Crude Phosphocellulose Crude Phosphocellulose Crude Phosphocellulose
Total protein (mg)
Units enzyme
S.A. (U/mg)
252 1.2 203 1.4 217 0.9 214 0.8
944 657 822 291 99 42 141 26
3.7 548 4.0 208 0.5 47 0.7 33
Yield (%) 70 35 42 18
Human AMPD2 recombinant enzymes were expressed in Sf9 cells and purified by batch adsorption to phosphocellulose with KCl gradient elution, as described in Materials and methods. Representative results from a single purification are shown.
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
299
Fig. 5. SDS–PAGE analysis of human AMPD2 recombinant proteins. One hundred milliunits of enzyme activity from phosphocellulose peak fractions was fractionated on a 8.5% acrylamide gel and stained with Coomassie blue. Lane 1, molecular size markers; lane 2, 1A/2; lane 3, 1B/2; lane 4, 1B/3, and lane 5, DD184.
protein bands loaded for equivalent amounts of enzyme activity (Fig. 5). This analysis removes the component of contaminating proteins in each preparation and suggests that the 1A/2, 1B/3, and DD184ð1B=2Þ enzymes similarly have specific activities higher than 1B/2. Native molecular mass of AMPD2 proteins
Fig. 4. Resolution of human AMPD2 recombinant and Sf9 AMPD endogenous enzymes by phosphocellulose chromatography. (A) Elution profiles of enzyme activities isolated from sonicates prepared from noninfected Sf9 cells ( ) and Sf9 cells infected with 1B/3 recombinant virus ( ). (B) Left panel: SDS–PAGE analysis of 2 lg total protein from crude sonicate (lane 1) and phosphocellulose peak activity fraction (lane 2) of 1B/3 recombinant virus-infected Sf9 cells. Gel is stained with Coomassie blue. Right panels: Western blot analysis of 50 mU of AMPD activity from crude sonicates of noninfected (lane 1) and 1B/3 recombinant virus-infected Sf9 cells (lane 2). Left blot, preimmune serum; right blot, anti-AMPD2 serum. Molecular size markers are indicated.
polypeptide. A representative analysis of 1B/3 is shown in Fig. 4B. Specific activity comparison of the AMPD2 proteins Table 1 shows that the 1A/2 and DD184ð1B=2Þ proteins are purified to specific activities higher than either the 1B/2 or 1B/3 proteins. However, this difference in specific activity may have been due to the relative lower expression of 1B/2 and 1B/3, which may result in a proportionately higher content of protein contaminants in these latter preparations. In order to address this issue, the intrinsic specific activity of each AMPD2 protein was estimated by Coomassie blue staining of
Two methods were used to measure the native molecular mass of each AMPD2 protein: gel filtration chromatography and chemical crosslinking. Fig. 6 shows gel filtration elution profiles of each naturally occurring AMPD2 variant and three deletion mutants (DM55½1B=2 , DE129½1B=2 , and DD184½1B=2 ). Only the DD184ð1B=2Þ enzyme elutes from this column in a volume that is consistent with a native tetrameric mass. All other AMPD2 proteins exhibit aberrant elution profiles. The inconsistency between calculated and observed molecular mass for these proteins can be correlated with N-terminal amino acid composition (Table 2). Proteins that contain both exon 2- and exon 3-encoded sequence (1A/2, 1B/2 and DM55½1B=2 ) elute with observed molecular masses 56–71% greater than expected, while those that contain only exon 3-encoded sequence (1B/3 and DE129½1B=2 ) elute 30–31% larger than expected. Fig. 6 also shows that some of each naturally occurring variant of isoform L elutes in the void volume of the column, although the relative proportion that behaves in this fashion varies, i.e., 1B=3 > 1B=2 > 1A=2. This aberrant gel filtration behavior can be correlated with the relative high proline content of each unique Nterminal extension. For example, the 1B/3 protein has the largest proportion of enzyme activity eluting in the void volume and the highest percentage of prolines in its
300
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
Fig. 6. Gel filtration analysis of human AMPD2 recombinant enzymes. Approximately 15 U of each enzyme was loaded onto a 1:5 cm 90 cm glass column packed with Sephacryl S300-HR resin (exclusion volume >2000 kDa). Void volume (Vo ) was determined by the elution of blue dextran and the resolving range of the column was calibrated with known globular standards. Fractions were assayed for AMPD activity, as described in Materials and methods. Ve, elution volume (observed); Vexp, elution volume (expected).
N-terminal extension ( 25% or 13/58 residues), compared to 1B/2 ( 12% or 15/128 residues) and 1A/2 ( 13% or 6/47 residues). In contrast, the AMPD2 core polypeptide has a proline content of only 6% (45/760 residues).
Aberrant gel filtration behaviors notwithstanding, chemical crosslinking analysis indicates that 1A/2 and 1B/2 enzymes are composed of four subunits, similar to the DD184ð1B=2Þ protein (Fig. 7). In the absence of crosslinker, all three AMPD2 proteins migrate
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
301
Table 2 Quantitative comparison of the gel filtration chromatography behaviors of human AMPD2 recombinant enzymesa AMPD2 variant
Exon 2
Exon 3
Calculated MW (kDa)
Observed MW (kDa)
Difference (kDa)
% Increase
1A/2 1B/2 DM55ð1B=2Þ 1B/3 DE129ð1B=2Þ DD184ð1B=2Þ
+ + + ) ) )
+ + + + + )
368 403 380 372 353 322
630 630 650 490 460 330
262 227 270 118 107 8
71 56 71 31 30 2
a Exon 2 (+) denotes the presence of protein sequence contributed by exon 2; Exon 3 (+) denotes the presence of sequence contributed by exon 3. Calculated MW indicates molecular mass derived from amino acid composition. Observed MW was calculated from the Ve of each protein in gel filtration analysis, using the equation logMW ¼ 2:272ðVe =39:2Þ þ 8:4051, derived from plotting the elution volumes of blue dextran and four globular protein standards.
Fig. 7. Phosphate-buffered acrylamide gel and chemical crosslinking analysis of human AMPD2 recombinant proteins. Five micrograms of each protein was treated with increasing concentrations (0–2 mM) of BS3 , a primary amine-reactive noncleavable crosslinking reagent. Products were fractionated on a 3.5% denaturing gel and stained with Coomassie blue. Molecular size markers are indicated. #,monomer; *, tetramer.
through the gel as a single predominant band, while incubation with increasing amounts of crosslinker results in the appearance of higher molecular weight species consistent with dimers, trimers, and tetramers. The purity of the 1B/3 protein was not sufficient to
allow for successful crosslinking experiments with this protein. However, it is reasonable to hypothesize that the 1B/3 protein is also a tetramer because it shares the core 734 amino acid AMPD2 polypeptide with 1A/2 and 1B/2.
302
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
Discussion AMP deaminase (AMPD) polypeptides are sensitive to limited proteolysis during purification and subsequent storage at 4 °C. This degradation was observed initially in freshly prepared rabbit skeletal muscle AMPD after storage at 4 °C [15]. Further evidence of proteolysis was provided by a comparative analysis between reported subunit molecular masses for purified human AMPD isoforms and those predicted from cloned sequences [14]. Protease cleavage sites have been identified in AMPD polypeptides isolated from a variety of sources and available data indicate that in vitro proteolysis is restricted to N-terminal sequences [20–23]. Therefore, it is important to consider that N-terminal sequences missing from purified AMPD enzymes might impact the behavior of this enzyme. For example, human AMPD1 and AMPD3 recombinant proteins can be isolated with subunits essentially intact and these enzymes exhibit significant differences in catalytic and contractile protein binding behaviors compared to their respective N-truncated derivatives [20,24]. Since most of the available information on isoform L has been generated with partially proteolyzed enzymes, further characterizations of these intact AMPD2 proteins are warranted. This study has demonstrated that leupeptin and E-64 can be used to effectively inhibit the in vitro proteolysis of human AMPD2 recombinant enzymes. The addition of these two protease inhibitors to buffers during the isolation, purification, and subsequent storage of these enzymes prevents detectable reductions in subunit molecular mass that readily occur in their absence. Leupeptin and E-64 are both inhibitors of calpain, a ubiquitous neutral cysteine protease (reviewed in [30]), which exhibits a preference for hydrophobic residues at the P2 position [31] and typically performs limited rather than digestive proteolysis [32]. In these respects, it is worth noting that inspection of reported proteolytic cleavage sites in the N-terminal sequences of all mammalian AMPD polypeptides reveals a preference for hydrophobic residues at the P2 position [20,23]. Calpain is therefore a potential candidate for a processor of AMPD enzymes. Gel filtration and chemical crosslinking data together suggest that all three naturally occurring AMPD2 variants are tetrameric enzymes that exhibit deviation from a compact globular structure in their N-terminal extensions. Each activity migrates faster than expected through a gel filtration column, while an AMPD2 deletion mutant comprised of the core polypeptide migrates as expected. In addition, some of each naturally occurring variant elutes in the void volume of the gel filtration column. These unexpected behaviors can be correlated with the unique configuration of 50 exon-encoded sequence, and with the relatively high proline
content, respectively, in each alternative AMPD2 polypeptide. Regardless of the structural basis for aberrant gel filtration behavior, the combined data suggest that each unique N-terminal extension adopts an extended nonglobular conformation. These structures would disproportionately expand the surface area of each enzyme, promote greater exclusion from molecular sieving beads, and also provide greater access to proteases. Crosslinking analysis using bis(sulfosuccinimidyl) suberate ðBS3 Þ indicates that AMPD2 enzymes have tetrameric structures and also suggests that abnormal gel filtration behavior is not due to higher-order structure. Previous gel filtration and sucrose gradient sedimentation analyses have also indicated a tetrameric structure for AMPD [4,33], but crosslinking data presented here are the first direct demonstration that the AMPD2 molecule consists of four subunits. With the identification of protease inhibitors that can be used to facilitate the isolation of intact AMPD2 proteins, the question of how unique N-terminal extensions affect the functional behaviors of isoform L can now be addressed. This will extend previous studies on the significance of divergent N-terminal sequences across different isoforms. In addition, the inclusion of E64 and leupeptin in the purification of human AMPD1 and AMPD3 recombinant enzymes can be used to maintain the integrity of their subunits during storage at 4 °C (data not shown). Therefore, the results of this study should be useful to other investigators working with AMP deaminase.
Acknowledgments This work was supported by Public Health Service Grant DK-50902 from the National Institutes of Health.
References [1] J. Lowenstein, K. Tornheim, Ammonia production in muscle: the purine nucleotide cycle, Science 171 (1971) 397–400. [2] A.G. Chapman, D.E. Atkinson, Stabilization of adenylate energy charge by the adenylate kinase reaction, J. Biol. Chem. 248 (1973) 8309–8312. [3] H. Zimmerman, 50 nucleotidase: molecular structure and functional aspects, Biochem. J. 285 (1992) 345–365. [4] N. Ogasawara, H. Goto, Y. Yamada, T. Watanabe, T. Asano, AMP deaminase isozymes in human tissues, Biochim. Biophys. Acta 714 (1982) 298–306. [5] R.L. Sabina, T. Morisaki, P. Clarke, R. Eddy, T.B. Shows, C.C. Morton, E.W. Holmes, Characterization of the human and rat myoadenylate deaminase genes, J. Biol. Chem. 265 (1990) 9423– 9433. [6] M.T. Bausch-Jurken, D.K. Mahnke-Zizelman, T. Morisaki, R.L. Sabina, Molecular cloning of AMP deaminase isoform L:
A.L. Haas, R.L. Sabina / Protein Expression and Purification 27 (2003) 293–303
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
sequence and bacterial expression of human AMPD2 cDNA, J. Biol. Chem. 267 (1992) 22407–22413. D.K. Mahnke-Zizelman, R.L. Sabina, Cloning of human AMP deaminase isoform E cDNAs: evidence for a third AMPD gene exhibiting alternatively spliced 50 exons, J. Biol. Chem. 267 (1992) 20866–20877. D.K. Mahnke-Zizelman, F. Van den Bergh, M.T. Bausch-Jurken, R. Eddy, S. Sait, T.B. Shows, R.L. Sabina, Cloning, sequence, and characterization of the human AMPD2 gene: evidence for transcriptional regulation by two closely spaced promoters, Biochim. Biophys. Acta 1308 (1996) 122–132. T. Morisaki, R.L. Sabina, E.W. Holmes, Adenylate deaminase: a multigene family in humans and rats, J. Biol. Chem. 265 (1990) 11482–11486. Z.Y. Chang, P. Nygaard, A.C. Chinault, R.E. Kellems, Deduced amino acid sequence of Escherichia coli adenosine deaminase reveals evolutionarily conserved amino acid residues: implications for catalytic function, Biochemistry 30 (1991) 2273–2280. M. Gross, H. Morisaki, T. Morisaki, E.W. Holmes, Identification of functional domains in AMPD1 by mutational analysis, Biochem. Biophys. Res. Commun. 205 (1994) 1010–1017. I. Mineo, P.R.H. Clarke, R.L. Sabina, E.W. Holmes, A novel pathway for alternative splicing: identification of an RNA intermediate that generates an alternative 50 splice donor site not present in the primary transcript of AMPD1, Mol. Cell. Biol. 10 (1990) 5271–5278. F. Van den Bergh, R.L. Sabina, Characterization of human AMP deaminase 2 (AMPD2) gene expression reveals alternative transcripts encoding variable N-terminal extensions of isoform L, Biochem. J. 312 (1995) 401–410. R.L. Sabina, D.K. Mahnke-Zizelman, Towards an understanding of the functional significance of N-terminal domain divergence in human AMP deaminase isoforms, Pharm. Therap. 87 (2000) 279– 283. M. Ranieri-Raggi, A. Raggi, Effects of storage on activity and subunit structure of rabbit skeletal muscle AMP deaminase, Biochem. J. 189 (1979) 367–368. M. Raneiri-Raggi, A.J.G. Moir, A. Raggi, Interaction with troponin T from white skeletal muscle restores in white skeletal muscle AMP deaminase those allosteric properties removed by limited proteolysis, Biochim. Biophys. Acta 827 (1985) 93–100. R. Marquetant, R.L. Sabina, E.W. Holmes, Identification of a noncatalytic domain in AMP deaminase that influences binding to myosin, Biochemistry 28 (1989) 8744–8749. D.J. Merkler, A.S. Wali, J. Taylor, V.L. Schramm, AMP deaminase from yeast: role in AMP degradation, large scale purification, and properties of the native and proteolyzed enzyme, J. Biol. Chem. 264 (1989) 21422–21430. O.P. Chilson, A.E. Kelly-Chilson, N.R. Siegel, AMP deaminases from chicken and rabbit muscle: partial primary sequences of
[20]
[21]
[22]
[23]
[24]
[25]
[26] [27]
[28]
[29]
[30] [31] [32]
[33]
303
homologous 17 kDa CNBr fragments: autorecognition by rabbit anti-chicken AMPD, Comp. Biochem. Phys. 116B (1997) 371– 377. D.K. Mahnke-Zizelman, P.C. Tullson, R.L. Sabina, Novel aspects of tetramer assembly and N-terminal domain structure and function are revealed by recombinant expression of human AMP deaminase isoforms, J. Biol. Chem. 273 (1998) 35118– 35125. S.L. Meyer, K.L. Kvalnes-Krick, V.L. Schramm, Characterization of AMD, the AMP deaminase gene in yeast. Production of amd strain, cloning, nucleotide sequence, and properties of the protein, Biochemistry 28 (1989) 8734–8743. R. Ronca, M. Raneiri-Raggi, P.E. Brown, A.J.G. Moir, A. Raggi, Evidence of a species-differentiated regulatory domain within the N-terminal region of skeletal muscle AMP deaminase, Biochim. Biophys. Acta 1209 (1994) 123–129. K. Knecht, K. Wiesmuller, V. Gnau, G. Jung, R. Meyermann, K.G. Todd, B. Hamprecht, AMP deaminase in rat brain: localization in neurons and ependymal cells, J. Neurosci. Res. 66 (2001) 941–950. D.K. Mahnke-Zizelman, R.L. Sabina, Localization of N-terminal sequences in human AMP deaminase isoforms that influence contractile protein binding, Biochem. Biophys. Res. Commun. 285 (2001) 489–495. B. Sims, D.K. Mahnke-Zizelman, A.A. Profit, G.D. Prestwich, R.L. Sabina, A.B. Thiebert, Regulation of AMP deaminase by phosphoinositides, J. Biol. Chem. 274 (1999) 25701– 25707. B. Ashby, H. Holmsen, Platelet AMP deaminase: purification and kinetic studies, J. Biol. Chem. 256 (1981) 10519–10523. B. Ashby, H. Holmsen, Platelet AMP deaminase: regulation by Mg-ATP2- and inorganic phosphate and inhibition by the transition state analog coformycin, J. Biol. Chem. 258 (1983) 3668–3672. M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein using the principle of dyebinding, Anal. Biochem. 72 (1976) 248–254. U.K. Laemmli, Cleavage of structural proteins during the assembly of the head of bacteriophage T4, Nature 227 (1970) 680–685. H. Sorimachi, K. Suzuki, l-Calpain, in: Handbook of Proteolytic Enzymes, Academic Press, New York, 1998. R.L. Mellgren, T. Murachi, Intracellular calcium-dependent proteolysis, CRC press, Boston, 1990. T.C. Saido, H. Sorimachi, K. Suzuki, Calpain: new perspectives in molecular diversity and physiological-pathological involvement, FASEB J. 8 (1994) 814–822. S. Yun, C.H. Suelter, Human erythrocyte 50 -AMP aminohydrolase: purification and characterization, J. Biol. Chem. 253 (1978) 404–408.