(Received for publication, October 7, 1996, and in revised form, January 15, 1997)
From the Department of Biochemistry, Dartmouth Medical School,
Hanover, New Hampshire 03755 and Department of Molecular
Biotechnology, University of Washington,
Seattle, Washington 98195
A 97-kDa protein present in the glucose transporter (GLUT4 isotype)-containing vesicles from rat adipocytes has been isolated, the sequences of two tryptic peptides were obtained, and on the basis of these its cDNA partially cloned. The 97-kDa protein is almost certainly identical to a major integral glycoprotein of this size in the rat adipocyte plasma membrane, since its predicted N-terminal sequence is the same as that recently determined for this glycoprotein by amino acid sequencing. Moreover, the predicted partial sequence (322 amino acids) of the 97-kDa protein is highly homologous to the corresponding region of a human placental amine oxidase, which was cloned simultaneously and proposed to be a secreted protein. The amino acid sequence of the 97-kDa rat/human amine oxidase indicates that the protein consists of a very short N-terminal cytoplasmic domain followed by a single transmembrane segment and a large extracellular domain containing the catalytic site. Thus this study establishes the 97-kDa rat/human amine oxidase as the first integral membrane amine oxidase to be cloned. The membrane amine oxidase was more abundant in the plasma membranes than the low density microsomes of the adipocyte, and in contrast to some other proteins found in GLUT4 vesicles, it did not redistribute to the plasma membrane in response to treatment of the cells with insulin.
Primary amine oxidases form a family of enzymes that catalyze the oxidation of primary amines by molecular oxygen, to yield the corresponding aldehyde, ammonia, and hydrogen peroxide. Typically these enzymes consist of two identical subunits; each contains an oxidized tyrosine residue, known as topa quinone, and one atom of copper, both of which participate in catalysis (reviewed in Refs. 1 and 2). Several secreted members of this family, including bovine serum amine oxidase and kidney diamine oxidase, have been cloned and extensively characterized (3, 4). By contrast considerably less is known about one (or possibly more) suspected member of this family. This is the membrane-bound amine oxidase that is highly susceptible to inhibition by semicarbazide, and is often referred to as the "tissue-bound semicarbazide-sensitive" amine oxidase (5-7). This enzyme, as detected by its activity against benzylamine, has been found to be present in a large number of tissues and cell types, including vascular smooth muscle cells, white and brown adipocytes, and skin fibroblasts (5-7). Its cloning as such (see below) has not been reported, and whether it is also of the topa quinone and copper-containing type has not been established.
The present study began as part of our investigation of insulin stimulation of glucose transport. Treatment of fat and muscle cells with insulin causes a rapid elevation in glucose transport due to a rapid increase in the amount of the glucose transporter (GLUT4 isotype)1 in the plasma membrane. The basis for this increase in amount is largely the enhanced trafficking of GLUT4 from intracellular locations to the plasma membrane (reviewed in Ref. 8). We have developed a procedure for isolating the intracellular membranes containing GLUT4 (referred to as GLUT4 vesicles) from adipocytes and are characterizing other proteins present in these vesicles (9). In the course of this work, we partially cloned a 97-kDa vesicle protein. This protein, which we initially referred to as vp97, has proven to be a membrane-bound, semicarbazide-sensitive amine oxidase that is almost certainly identical to a major integral glycoprotein in the adipocyte plasma membrane. As our investigation of this protein was nearing completion, the cloning of its human homolog was reported (10). However, the human homolog, whose properties as protein were not examined experimentally, was proposed to be another secreted oxidase, rather than a membrane-bound one. Thus, our results identify and characterize the first cloned membrane amine oxidase.
Rat adipocytes were obtained and then treated with insulin or left in the basal state, as described in (9). GLUT4-containing vesicles were isolated from the adipocytes by immunoadsorption with antibodies against GLUT4, and the vesicle proteins were released from the adsorbent by solubilization with 0.5% octaethylene glycol dodecyl ether, as presented in detail in Mastick et al. (9). Plasma membranes and low density microsomes were isolated from basal and insulin-treated adipocytes by differential centrifugation (9). The effectiveness of insulin was checked by quantitative immunoblotting of the fractions for GLUT4. As expected from earlier results (9), insulin treatment caused at least a 2.5-fold increase in GLUT4 in the plasma membranes and a 30% decrease of GLUT4 in the low density microsomes.
Gel Electrophoresis and BlottingProteins were held at 100 °C for 3 min in SDS sample buffer (4% SDS, 20 mM dithiothreitol, 1 mM EDTA, 10% glycerol, 95 mM Tris-Cl, pH 6.8) and separated on a gradient gel of 8-16% acrylamide. The gel was either stained directly for protein with Coomassie Blue, or the proteins were transferred electrophoretically to nitrocellulose in 25 mM Tris, 192 mM glycine, 0.01% SDS, 20% MeOH for 18 h. The nitrocellulose was stained for protein with either colloidal gold stain (Bio-Rad) or 0.1% (w/v) Amido Black in methanol/acetic acid/water (5/1/4, by volume), or it was stained specifically for quinone-containing proteins with nitroblue tetrazolium in potassium glycinate, as described in Paz et al. (11).
Large Scale Preparation of vp97 and Peptide SequencingThe GLUT4 vesicle proteins in 0.5% (w/v) octaethylene glycol dodecyl ether were isolated from the adipocytes of 78 rats, as described above. The proteins (approximately 400 µg) were separated from membrane lipids and the detergent by precipitation from chloroform/methanol according to the procedure in Wessel and Flügge (12), and dissolved in 300 µl of SDS sample buffer. The proteins were separated by SDS-gel electrophoresis in two 8-mm lanes, transferred to nitrocellulose, and stained with Amido Black, as described above. vp97 was digested with trypsin, and the peptides were isolated by high performance liquid chromatography and sequenced essentially as described in Mastick et al. (9), with the exception that tandem mass spectra were determined for each eluting peptide with a model TSQ 7000 tandem mass spectrometer (Finnigan MAT) (13). The complete sequences of two peptides (designated peptide 1 and 2) were obtained (see "Results").
Cloning of vp97 cDNATotal RNA from rat adipocytes was isolated with the Trizol reagent (Life Technologies, Inc.) according to the manufacturer's instructions, with the exception that the triglyceride in the cells necessitated the use of a larger volume of the reagent (24 ml for the packed adipocytes from six rats, which are approximately 3 × 107 cells) (14). The yield of RNA from adipocytes of six rats was 0.4 mg.
The close homology of peptides 1 and 2 with sequences in bovine serum oxidase (see "Results") served as the basis of a PCR approach to cloning. First, reverse transcription was carried out with an antisense degenerate oligonucleotide based on the sequence of the more C-terminal peptide (peptide 2) (GNCCYTCHGGRTGRAAYTG where N = A, T, G, C; Y = C, T; R = A, G; H = A, C, T). Total RNA (1 µg) was denatured at 90 °C for 5 min, and then reverse transcribed for 10 min at 20 °C followed by 45 min at 42 °C, with the antisense degenerate oligonucleotide (25 pmol) and Superscript II (200 units) (Life Technologies, Inc.) in 20 µl volume. Following RNase digestion, the single-stranded cDNA was purified from the mixture using the GlassMax kit (Life Technologies, Inc.). It was then amplified by PCR with the Ultma DNA polymerase (Perkin-Elmer), using the antisense degenerate primer given above (50 pmol) and a sense degenerate primer based on the sequence of the more N-terminal peptide (peptide 1, primer GARTAYCARGAYATHCARGA, 50 pmol) in a 100-µl reaction mixture (30 cycles of 30 s at 94 °C, 60 s at 45 °C, and 30 s at 72 °C, followed by 7 min at 72 °C). This yielded a fragment of about 450 bp. A second round of amplification was performed with a nested sense degenerate primer based on peptide 1 (GAYATHCARGARATGATHTTYCA) and the same antisense primer. The resulting fragment of about 450 bp was cloned into pCR-Script (Stratagene) according to the manufacturer's instructions. The inserts in four clones were sequenced on both strands.
The 5 end of the cDNA was cloned by a 5
-RACE procedure using the
Life Technologies Inc. 5
-RACE kit according to the manufacturer's instructions. Reverse transcription on total RNA was carried out with
the antisense oligonucleotide for nucleotides 978-995 (see Fig. 2),
and the cDNA was tailed at the 3
end with Cs. It was then
amplified with the abridged anchor primer provided with the kit and a
nested antisense primer (for nucleotides 937-957). The product was
then used as the template for amplification of a smaller fragment with
the abridged universal amplification primer of the kit and another
nested antisense primer (for nucleotides 666-687). The resulting
700-bp fragment was cloned into pCR-Script, and the inserts of three
clones were sequenced on both strands.
DNA sequencing was performed on the Applied Biosystems 373 DNA sequencing system using Perkin-Elmer DNA sequencing kit FS. Data were analyzed using the software supplied by Applied Biosystems. Searches of the protein sequence data bases were performed by using BLAST (15).
Amine Oxidase AssayThis enzyme was assayed through its conversion of benzylamine to benzaldehyde, which was monitored by the increase in adsorbance at 250 nm (16). Reaction was initiated by the addition of the enzyme-containing fraction (50 µg in 80 µl or less) to 950 µl of assay solution, in a cuvette thermostated at 37 °C. The assay solution consisted of 1 mM benzylamine in 100 mM potassium phosphate, pH 7.2, with 0.1% (w/v) octaethylene glycol dodecyl ether. The nonionic detergent was included to solubilize the membranes, so that the substrate would be readily accessible to the oxidase.
Vesicles containing GLUT4 were isolated from the low
density microsomal fraction of rat adipocytes by immunoadsorption with antibodies against GLUT4. The vesicle proteins were then released by
solubilization of the vesicles bound to the adsorbent with a nonionic
detergent. Lane 2 of Fig. 1 shows the
polypeptides of the GLUT4 vesicles in the 70-250-kDa range. Among the
vesicle proteins is a prominent band at 97 kDa (designated vp97). A
large preparation of GLUT4 vesicles was carried out, and this
preparation yielded vp97 in sufficient amount to obtain the sequences
of two tryptic peptides, EYQDIQEMIFHR (peptide 1) and APPLQFHPEGPR
(peptide 2).
Partial Cloning of vp97
A search of the sequence data bases
revealed that bovine serum amine oxidase contained two sequences that
were very similar (3). Peptide 1 matched residues 174-185 of the serum
oxidase in 8 of 12 positions and peptide 2 matched residues 310-321 in 10 of 12 positions. These matches, together with the fact that serum
oxidase is a protein of about the same size, strongly suggested that
vp97 was a member of the amine oxidase family and indicated that
oligonucleotides derived from the two peptide sequences should yield a
product of approximately 440 bp upon PCR amplification of reverse
transcribed rat adipocyte mRNA. In fact, when this was done, the
major product was a 450-bp fragment of DNA. Subsequently, the remaining
5 sequence of the cDNA was cloned by a 5
-RACE method (see
"Experimental Procedures").
Fig. 2 presents the nucleotide sequence for the 5
portion of the cDNA of vp97 and the corresponding amino acid
sequence. An open reading frame initiated by a methionine at nucleotide 86 proceeds to the end of the sequence. This methionine must be the
site of initiation of translation, since the 5
-upstream region contains an in-frame stop codon at nucleotide 59, and as described below, the purified protein contains the predicted N-terminal sequence
as determined by amino acid sequence analysis. The cDNA sequence
surrounding the initiation codon conforms well to the consensus
sequence for the initiation of translation in rat (17).
The partial amino acid sequence of vp97 from residues 35-320 shows high homology with the corresponding region of bovine serum amine oxidase (residues 34-319) (3), with 78% identity (data not shown). This finding thus indicated that vp97 is an amine oxidase, and the data in lane 1 of Fig. 1 provide more direct evidence that this is the case. Amine oxidases of this type contain a topa quinone residue, which is derived from the oxidation of a specific tyrosine (1). Application of a staining method for quinone-containing proteins, which has been used to detect the bovine serum oxidase (11), to the GLUT4 vesicle proteins gave specific staining of vp97.
One small region of the vp97 sequence differs markedly from the corresponding region of the serum oxidase. This region is the N-terminal 35 amino acids, which show no homology with the N-terminal 34 amino acids of the serum oxidase. In the case of the serum oxidase it has been shown that the N-terminal 16 amino acids constitute a leader sequence that is cleaved upon processing of this secreted protein (3). The explanation for the different N terminus of vp97 became apparent when we discovered by data base search that vp97 is almost certainly the same as a major integral membrane glycoprotein in rat adipocyte plasma membranes (18). This 100-kDa protein, the function of which was not known, had recently been purified and its N-terminal sequence of 19 amino acids determined by Edman degradation (18). This sequence is identical to that of residues 2-20 of vp97. Thus, vp97 is an integral membrane glycoprotein. Consistent with this expectation, analysis of its sequence for potential membrane-spanning domains (TopPred II program) (19) predicts a single one, which is the stretch of hydrophobic amino acids at the N terminus (residues 6-26, underlined in Fig. 2).
At this stage of our cloning effort, the complete sequence of a new amine oxidase, whose cDNA was cloned from a human placental library by a PCR approach starting with primers based upon conserved sequences in other oxidases, appeared in the data base (GenBankTM accession no. U39447[GenBank] and now in Zhang and McIntire (10)). This human placental amine oxidase (HPAO) is predicted to consist of 763 amino acids and have a mass of 84,621 Da. The partial sequence of vp97 is highly homologous with that of HPAO over its entire length, including the very N terminus, with 83% identity (Fig. 2). Thus, vp97 is the rat homolog of this human amine oxidase. Because of its sequence similarity to the bovine serum amine oxidase, HPAO was proposed to be a secreted protein, with its different N-terminal sequence also being a leader sequence (10). However, on the basis of the characterization of vp97 as an integral plasma membrane protein described above (also see below), a substantial portion of HPAO is probably membrane-bound with its N terminus uncleaved.
These considerations thus lead to the conclusion that vp97/HPAO is a membrane amine oxidase and subsequently it will be referred to as such. Its topology in the membrane can be deduced as follows. Analysis of the complete sequence of HPAO shows that it also has only one segment that is predicted as highly likely to be membrane spanning; that is the one at the N terminus (residues 6-26). Thus, the membrane amine oxidase is expected to be anchored by this single transmembrane segment, with only 5 residues on one side of the membrane and the bulk of the protein (residue 27 and beyond), including the catalytic site, on the other side. Since the membrane amine oxidase is a glycoprotein, as demonstrated by its strong binding to the lectin wheat germ agglutinin (18), and the sites of glycosylation for plasma membrane proteins are invariably extracellular, the bulk of the protein must be on the extracellular side of the membrane. Residues 1-5, the predicted cytoplasmic domain, contain no potential sites for Asn N-glycosylation, whereas the predicted large extracellular domain contains six such sites (10), three of which are present in the partial sequence in Fig. 2, at positions 137, 232, and 294.
Subcellular Distribution of the Membrane Amine OxidaseThe distribution of the oxidase was examined by staining the proteins in plasma membranes and low density microsomes from basal and insulin-treated adipocytes for quinone-containing proteins (Fig. 1, lanes 3-6). These fractions showed a single major quinoprotein at 97 kDa and a much weaker band at 180 kDa. The latter may be unreduced dimer, since the major glycoprotein of the rat adipocyte plasma membrane, identified here as the amine oxidase, has been reported to occur as a disulfide-linked dimer (18). On the basis of the relative staining intensities, the membrane amine oxidase appears to be about 3-fold more abundant in the plasma membranes than in the low density microsomes. This distribution between the two fractions agrees with the relative amounts of the protein that were isolated from these two fractions by adsorption with wheat germ agglutinin (18).
Some proteins found in GLUT4 vesicles, including GLUT4 itself, translocate markedly from the low density microsomes to the plasma membrane in response to insulin (9). However, insulin had no effect on the subcellular distribution of the membrane amine oxidase (Fig. 1, lanes 3-6). Further evidence for the lack of an effect was the fact that the amount of the amine oxidase isolated with the GLUT4 vesicles did not decrease in response to insulin; by contrast, the 165-kDa protein in these vesicles (Fig. 1, lane 2), which we have previously characterized as an aminopeptidase that translocates to the plasma membrane in response to insulin (9, 20), decreased by about half (data not shown).
The membrane amine oxidase, as detected by quinone staining, coincided exactly in its electrophoretic mobility with a major protein in the adipocyte plasma membranes detected by staining with Coomassie blue (Fig. 1, lanes 7 and 8). We therefore take this protein to be the amine oxidase, and as expected from the quinone staining, there is less in the low density microsomes. By visual comparison of the intensity of the amine oxidase band with known amounts of standard proteins (data not shown), the plasma membranes contain about 23 µg of oxidase/mg of protein or 2.3% of the total; the oxidase is thus an abundant protein. Since a single adipocyte contains about 90 pg of plasma membrane (14), there are approximately 1.4 × 107 copies of the oxidase on the surface of the cell.
Amine Oxidase Activity in Adipocyte Subcellular FractionsThe
plasma membranes and low density membranes from basal and
insulin-treated adipocytes were assayed for amine oxidase activity with
benzylamine as the substrate. Fig. 3 presents
representative data for a set of these subcellular fractions isolated
at the same time from one batch of cells, which had been divided in
half and treated with insulin or left in the basal state. The amine oxidase activity in the plasma membrane from basal cells was 2.3 times
larger than that in the low density microsomes from basal cells.
Insulin treatment had no significant effect on the activity in either
fraction. Similar results were obtained with a second set of
subcellular fractions isolated from another batch of cells.
Two observations indicated that the activity in the spectrophotometric assay was due to a quinone-containing oxidase that converted benzylamine to benzaldehyde. First, semicarbazide completely inhibited the reaction (2). When either the plasma membranes or the low density microsomes were first incubated with 100 µM semicarbazide for 30 min at 37 °C and then the assay was initiated with benzylamine, there was no increase in adsorbance over a 30-min period. In the corresponding control experiment, where the membrane fraction was first incubated for 30 min at 37 °C in the absence of semicarbazide, the rate was the same as that obtained without this preincubation. The benzylamine substrate protected the enzyme against inactivation by semicarbazide, since in contrast to the inhibition after preincubation in the absence of substrate, no inhibition occurred when the semicarbazide was added 15 s after the initiation of the enzymatic assay. Second, the product of the reaction was identified as benzaldehyde by its ultraviolet spectrum. This spectrum, which was obtained by subtracting the adsorbance of an assay inhibited by semicarbazide from that of the corresponding assay without semicarbazide, showed the expected maximum at 250 nm.
The combined results from this study, the partial characterization of a major integral glycoprotein in adipocyte plasma membranes (18), and the cloning of HPAO (10) show that vp97/HPAO is a semicarbazide-sensitive membrane amine oxidase. Moreover, since comparison of the complete amino acid sequence of HPAO with that of other amine oxidases revealed that it is a topa quinone and copper-containing enzyme (10), the membrane amine oxidase has this type of catalytic domain.
Raimondi et al. (21) originally reported that white adipocytes contained a membrane-bound, semicarbazide-sensitive amine oxidase, for which benzylamine was a good substrate (Km = 12 µM). Our observation that the activity is highest in the plasma membrane agrees with the results of these authors, who found the highest activity in the crude membrane fraction that would be most enriched in plasma membranes (21). Moreover, localization of the oxidase in pig adipocytes by immunofluorescence with cross-reacting antibodies raised against the bovine serum oxidase showed the protein to be on the cell surface (22).
We isolated the membrane-amine oxidase as a component of the GLUT4 vesicles. Other components of these vesicles, including GLUT4 itself and the 165-kDa membrane aminopeptidase, translocate to the plasma membrane in response to insulin (9). However, as assessed both by staining for the protein and by activity, the membrane amine oxidase did not do so. One likely explanation for the difference derives from the nature of the GLUT4 vesicles isolated by immunoadsorption with anti-GLUT4. A recent study has shown that the vesicles are not homogeneous; a portion are fragmented endosomal membranes, and another portion may be specialized small secretory vesicles (23). Insulin probably increases the rate at which vesicles traffic from the endosomal system to the plasma membrane (24) and also may increase the rate at which the specialized secretory vesicles fuse with the plasma membrane (25), but insulin treatment may not significantly reduce the membrane area of the endosomal system. If the latter is the case, the amount of GLUT4 vesicles derived from fragmented endosomes in insulin-treated cells may be about the same as for untreated cells. Thus if the intracellular portion of the membrane amine oxidase is primarily in endosomes rather than specialized secretory vesicles, then the amount of oxidase found in the GLUT4 vesicles, and the low density microsomes from which they are derived, will not decrease in response to insulin. The finding that the oxidase is more abundant in the plasma membranes than the low density microsomes is consistent with the fact that its short cytoplasmic domain contains neither the tyrosine nor the dileucine motif required for selective endocytosis from the plasma membrane (26).
The physiological role of the membrane amine oxidase is not known. It may function to degrade bioactive amines, such as histamine, or amine products of intermediary metabolism, such as methylamine and aminoacetone (5-7). These compounds have been shown to be substrates in vitro (5-7). The identification and cloning of the membrane amine oxidase should open the way to investigate its role in vivo. For example, targeted disruption of its gene in mice is now feasible.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U72632[GenBank].
We are indebted to Dr. Judith Klinman for helpful advice about amine oxidases, and to Dr. William McIntire for both advice and preprints of articles in press. We thank Joshua Sparling for technical assistance and Mary Harrington for expert secretarial assistance.