Purification and Characterization of Functional Recombinant cr-Amidating Enzyme Secreted from Mammalian Cells*

A rat a-amidating enzyme (a-AE) cDNA has been expressed in mouse C 127 cells using a bovine papilloma virus vector in which transcription was regulated by the mouse metallothionein 1 promoter. The cDNA en- coding the full length a-AE protein was modified to terminate translation at a site preceding the transmembrane and cytoplasmic domains, thereby enabling functional enzyme to be secreted into the medium. Purification of recombinant a-AE the two proteins of kDA.

A rat a-amidating enzyme (a-AE) cDNA has been expressed in mouse C 127 cells using a bovine papilloma virus vector in which transcription was regulated by the mouse metallothionein 1 promoter. The cDNA encoding the full length a-AE protein was modified to terminate translation at a site preceding the transmembrane and cytoplasmic domains, thereby enabling functional enzyme to be secreted into the medium. Purification of recombinant a-AE to homogeneity indicated that the enzyme was synthesized and secreted as two proteins of 75-77 kDA. The observed heterogeneity was due to inefficient glycosylation at Asneeo, as demonstrated by glycopeptidase F digestion. Using the synthetic peptide, dansyl-Tyr-Val-Gly, the specific activity of the recombinant enzyme at pH 7.0 was found to be 1.4 clmol/min/mg and the K,,, of the enzyme was determined to be 3 pM. The purified recombinant enzyme has maximal activity at pH 4.5-5.5; however, a rapid inactivation of the enzyme occurs in acidic solutions in vitro. This inactivation is diminished when activity is measured at pH 7.0-10.0. The availability of large amounts of readily purified, active recombinant a-AE should allow detailed probing of reaction mechanism, copper coordination chemistry, and turnover-based inactivation events.
cY-Amidating enzyme (a-AE)' is the post-translational processing enzyme that catalyzes the conversion of glycine-extended prohormone substrates to the biologically active 01amidated peptide products (l-3). Since the initial reports by Bradbury et al. (1,4) demonstrating an enzyme in porcine pituitary capable of catalyzing cu-amidation, a-AE activity has been isolated and purified from a number of tissues (5-10).
Purification from several sources has demonstrated heterogeneity in the molecular mass of the purified enzymes ranging from 34-78 kDa (6,7,11,12). In contrast, characterization of LU-AE activity from all sources studied to date has shown similar requirements for L-ascorbic acid, oxygen, and exoge- neous copper for maximal in vitro activity. These shared properties of the enzymes are consistent with a conserved and common reaction mechanism. Although the various purified enzymes have been reported to be maximally active at pH ranging from 5.0 for the rat MTC protein (6) to pH 8.5 for the bovine pituitary enzyme (ll), it is likely that a-AE resides and is active in the acidic (i.e. pH 5.5-6.0), secretory granules (13,16,17) and that these observed in vitro variations may simply reflect differences in assay conditions. Molecular cloning of (Y-AE cDNAs from bovine pituitary (18), frog skin (19, ZO), and rat medullary thyroid carcinoma (21) and atria1 tissue (22) indicates that in each of these tissues there are multiple mRNAs which encode cu-AE. In several cases, comparison of the molecular weights deduced from the cDNA sequence to the molecular weights of the purified enzyme indicates that the a-AEs are synthesized as larger precursors. In rat MTC, we have demonstrated that two mRNAs encode functional CX-AE (21). This finding is similar to what has been described for the mRNAs found in the rat atrium (22). One mRNA, designated type B, encodes a protein of a molecular mass of 105 kDa, and a second, type A, encodes a protein of 94 kDa. Inspection of the cDNA sequence for mRNAs from rat MTC and atria1 tissue indicates that both sequences encode a carboxyl-terminal transmembrane domain. We have shown in the MTC-derived cell line (CA-77) that functional secreted a-AE is generated by specific post-translational processing of the precursor proteins which contain transmembrane domains (23). Biosynthetic studies with CA-77 cells, cloning of a-AE cDNA, and purification of active enzyme from MTC, have allowed us to predict the probable site(s) contained within the precursor protein(s) which direct the post-translational processing of a precursor membrane protein to a soluble secreted enzyme (6, 23). One of our goals is to understand the mechanism of catalysis and cofactor specificity for a-AE. To study these problems, we have produced a cDNA encoding cr-AE with deleted transmembrane and cytoplasmic domains by placing a stop codon in place of Lys716, the putative site at which the type A (u-AE is normally processed (21). In this report, high level expression of this mutated cDNA and secretion of the truncated a-AE protein from mouse Cl27 cells is described. Further, we show the purification to homogeneity of the recombinant enzyme from conditioned tissue cultured medium and report that the properties of this enzyme are similar to those previously reported for soluble 75-kDa a-AE purified from rat MTC and the derived cell line, Ca-77 (6, with 20 pg of BPVMMTneo-Type Ai?, by the standard CaP04 precipitation procedure (27). After 3 days, cells were split 1:15 into loo-mm dishes and were fed with growth medium (Dulbecco's minimal essential medium plus 10% fetal bovine serum) every 2 days for the subsequent lo-14 days. Transformed foci were obvious after 7 days and were cloned on day 10. After the initial clones were grown for l-2 weeks, the cells were split into medium containing G418 (400 pg/ml) and were grown 2-3 weeks longer in this medium. At this stage, clones were screened for active, secreted (u-AE and for immunoreactive (u-AE protein. For analysis of the copy number of pdBPVMMTneo-AT5 contained in the recombinant Cl27 clones, high molecular weight DNA was prepared from Cl27 cell pellets by SDS lysis and proteinase K digestion as described (25)

AND DISCUSSION
The expression vector pdBPVMMTneo-type AT5 (LU-AE) was used to direct the expression of secreted (u-AE in mouse Cl27 cells (Fig. 1). Since BPV-based vectors efficiently transform mouse Cl27 cells, our protocol for isolating transfected cells initially involved cloning transformed foci. After the transformed clones were expanded, the cells were propagated in G418-containing medium to select for clones expressing functional neoR. A number of G4lSresistant cell lines which were isolated demonstrated a significant level of immunoprecipitable (r-AE protein that was secreted into the medium. The cell line with the highest level of enzyme activity, as well as the highest level of total (Y-AE protein recovered from the medium, was selected for further study. The selected cell line secretes greater than 3 mg/liter active CY-AE when grown in medium containing 10% fetal bovine serum. Determination of the plasmid copy number contained within the cells has revealed approximately 260 copies of the plasmid/cell (data not shown). To confirm that recombinant cr-AE was efficiently synthesized and secreted from these heterologous cells, a pulse-chase analysis of a-AE biosynthesis was performed. As shown in Fig. 2, after a short pulse with [""Slmethionine, a single protein of approximately 75 kDa was synthezied in the cells and was immunoprecipitated with an cu-AE-specific rabbit polyclonal antiserum. With increasing chase time, the 75-kDa protein was recovered from the medium. The ttlZ for secretion is approximately 1 h, consistent with an efficient transit from the cell. It is evident that the material which is secreted into the medium exhibits some heterogeneity. Most likely, this heterogeneity is due to altered glycosylation at Asn@' of the type A cDNA (21-23). This contention is supported by pretreatment of cells with tunicamycin, an inhibitor of N-linked oligosaccharide addition, prior to labeling. When cells are grown in the presence of tunicamycin (Fig. 2, lane  N), the secreted enzyme migrates as a single protein band on SDS-PAGE. The slower migrating protein has been lost. Thus, the more rapidly migrating protein (of lane M) is the non-N-glycosylated one. The difference in autoradiographic intensity of the nonglycosylated proteins (compare lanes M and N) reflects a general decrease in protein synthesis observed with tunicamycin treatment. If a longer exposure of the gel is made, there is no evidence for the slower migrating protein even when the intensity of the signal in the treated sample (lane N) is raised well beyond that observed in the untreated sample. It also appears from this analysis that the majority of the recombinant a-AE produced in this cell line is nonglycosylated (lane M). Activity determinations on the secreted protein suggest that glycosylation is not necessary for enzymatic activity. Experiments are currently in progress to directly examine the role of the single oligosaccharide chain altered oligosaccharide addition sites. To further characterize the properties of the recombinant secreted a-AE, the enzyme from conditioned cell culture medium has been purified to homogeneity. Recombinant Cl27 cells were first adapted to growth in medium containing 1% fetal bovine serum. Cells were then grown to near confluence, and conditioned medium was collected every other day for 20 days. In the purification shown, 1 liter of conditioned medium containing 825,000 units of (Y-AE activity was first subjected to ammonium sulfate precipitation at 45% of saturation (Table I). With this treatment, the recombinant LU-AE is recovered with the precipitated proteins, and a significant increase in total ol-AE activity is consistently observed. These results indicate that total activity in the starting material is an underestimate.
The calculations of recovery have been adjusted to reflect the increased activity recovered after ammonium sulfate precipitation.
We have adapted our purification scheme from the one originally standardized for purification of native a-AE from rat MTC tissue or CA-77 cell-conditioned medium (5, 6, 24). The final step in the present protocol is S-Sepharose chromatography which very effectively removes the few remaining protein contaminants from the recombinant (Y-AE. The overall purification obtained using this protocol is 468-fold, with a 41% recovery (Table I). Fig. 3 shows the analysis of samples from each step in the purification as measured by SDS-PAGE followed by silver staining (A) or Western blotting (B). It is evident from these results that purified 75-kDa o(-AE is com-2.5 5 10 20 40 prised of at least two proteins (see lane 5 in Fig. 3, A and B). However, when the purified material is subjected to digestion with glycopeptidase F, a single protein of 75 kDa is found (Fig. 4). This result confirms the observations from tunicamycin treatment demonstrating that a portion of the recombinant 75-kDa is glycosylated. Furthermore, the glycosylated 75-kDa protein appears to contain complex oligosaccharides, since a similar treatment of purified protein with endoglycosidase H does not alter the doublet seen on SDS-PAGE (not shown).
Careful inspection of the data presented in Fig. 3, A and B, indicates that there is a shift in the electrophoretic mobility of both the glycosylated and unglycosylated CY-AE proteins during the purification. This is especially obvious in Fig. 3, A and B, lane 4, where the single protein bands begin to appear as closely spaced doublets. One explanation for this increased mobility may be proteolytic cleavage at a single site within the protein which does not effect enzymatic activity. Although low levels of protease activity have been found in conditioned medium, proteolytic activity has not been detected in (w-AE samples following the S-300 chromatography where it appears that the majority of the change in SDS-PAGE mobility occurs. However, since it appears that the increased mobility may be due to a specific cleavage and not general degradation, it may not be possible to analyze for this reaction without further characterization of the cleavage site requirements. While other less attractive possibilities could also explain the observed changes on SDS-PAGE, whatever the modification, the specific activity of the purified enzyme is unaffected ( Table I). The activity of the purified (u-AE is approximately 1.4 X lo6 units/mg of protein, when assayed under standard reaction conditions (i.e. pH 7.0,2 PM CuSO.,, 3 mM ascorbate). This value agrees well with the specific activity obtained from either the rat MTC-purified (Y-AE (6) or the (Y-AE purified from CA-77 cell-conditioned medium (24). Furthermore the Km of the recombinant enzyme for the dansyl-Tyr-Val-Gly substrate is 3.3 fiM, which is essentially identical with that obtained with both the rat MTC and CA-77 cell-derived enzymes (data not shown).
As noted above, purified L~-AEs have been found to display Functional Recombinant cu-Amidating Enzyme a wide variation in the pH at which they are maximally active in uitro.
Previous results with (Y-AE purified from rat MTC or the derived CA-77 cell line indicate that the 75kDa (Y-AE has maximal activity at pH 5. 0-5.5 (6,24). To further compare the recombinant protein with the enzyme from natural sources, we have determined the pH profile for purified recombinant a-AE and found the optimum to be 4.5-5.5 (Fig. 5). When assayed at its pH optimum, the recombinant enzyme has a specific activity of about 4 x lo6 units/mg of protein, which is indistinguishable from that of the MTC-derived rat enzyme. Fig. 5 also illustrates the rapid inactivation of the enzyme at more acidic pH in uitro. In fact, due to this effect, at the later time points of the assay, the apparent pH optimum is closer to neutrality.
It appears that this inactivation at acidic pH is partially dependent on the concentration of ascorbic acid present in the reaction (data not shown). As with the native enzymes from rat thyroidal cells (6, 24), the purified recombinant a-AE is completely inactive in the absence of exogenous ascorbic acid. Other reductants may be substituted, but they support the reaction less efficiently.' The purified recombinant (u-AE was also subjected to amino-terminal sequence analysis. The data obtained from these experiments indicate that there is a single, predominant amino terminus beginning at PheZ6 (Fig. 6), which is the first amino acid of the putative "pro" region for the a-AE. However, since the sequence represented less than 50% of that expected from the purified protein input, the possibility that other, blocked amino termini are also present cannot be ruled out. These results also indicate that Cl27 cells are unable to completely process the propeptide, unlike cells which normally produce cu-AEs (18-21). This deficiency is not surprising since peptide hormone-producing cells are generally considered to be highly specialized cell types (32), containing a regulated pathway for secretion, and it is likely that proregion processing occurs in specialized intracellular compartments. Indeed, as noted above, the secretion rate from the recombinant Cl27 cells (tlj2 for secretion of 1 h; Fig. 2) is much more rapid than that observed for or-AE secretion from CA-77 cells (tal2 for secretion of 4.5 h (23)), consistent with differences between regulated and constitutive secretion (33). The enzyme with an intact proregion appears to be fully active. We have begun studies to examine the role of this region by expressing mutant cDNAs which lack this sequence.
We have demonstrated recombinant expression of a truncated cDNA encoding a fully functional a-AE. The ability to isolate and purify to homogeneity large quantities of this enzyme will greatly enhance our ability to investigate the complexity of the reaction mechanism for conversion of precursor peptides to their amidatedproducts.
In addition, during studies of cY-amidation reactions using enzyme purified from natural sources, a turnover-dependent loss of enzyme activity has been observed.3 The availability of enzyme from the recombinant cell line should allow for determination of the molecular basis of this finding.
Finally, a number of reports have indicated the presence of multiple forms of o(-AE in a number of different tissues (6-9, 11, 23, 24, 34, 35). The cloning of a-AE cDNAs from various species indicates that, at very least, the multiple forms share active sites that are well conserved (18)(19)(20)(21)(22). At this point, no compelling suggestions as to a need for these different enzymes have been advanced. With the native enzymes, careful analysis of different molecular forms under comparable conditions has been complicated by lack of starting materials and ' D. Merkler, personal communication. " N. M. Mehta, P. P. Tamhurini, and A. H. Bertelsen, unpublished observations. differences in the purification procedures required to obtain the various enzymes. We have recently reported evidence that in rat MTC tissues the lower molecular weight enzyme forms arise by translation and processing of an alternatively spliced a-AE mRNA (type B) (21,23). The availability of the 75-kDa recombinant enzyme of this report and a similarly produced 43-kDa recombinant LU-AE prepared by the truncation of a type B cDNA4 (36) should provide an opportunity to examine two forms of a-AE under very similar conditions. Perhaps these studies will help to elucidate the role of multiple forms of a-AE.