A Plasminogen-like Protein Selectively Degrades Stearoyl-CoA Desaturase in Liver Microsomes

We the purification of a 90-kDa with and characterized the inhibitor sensitivity of the protease. Here we show that the 90-kDa protein is a microsomal form of plasminogen (Pg), and that the purified SCD protease contains a spectrum of plasmin-like derivatives. The 90-kDa protein was identified as Pg by mass spectrometry of its tryptic peptides. The purified SCD protease reacted with Pg antibody, and immunoblotting demonstrated enrichment of Pg by the purification procedure established for the SCD protease. Analysis of microsomes by zymography demonstrated a single band of proteolytic activity at 70-kDa corresponding to the mobility of Pg in nonreduced polyacrylamide gels. When microsomes were incubated at 37˚C prior to zymography, an intense band of proteolytic activity developed at 30-kDa. The purified SCD protease displayed a spectrum of proteolytic bands ranging from 70- to 30-kDa. Degradation of SCD by the purified protease and by microsomes was inhibited by bdellin, a plasmin inhibitor from the medicinal leech Hirudo medicinalis . To explore the role of Pg in the degradation of SCD in vivo , we examined SCD expression and degradation in microsomes isolated from Pg deficient (Pg-/-) mice. Compared with microsomes from wild-type littermate control mice, liver microsomes from Pg-/- mice had significantly higher levels of SCD. Degradation of SCD in microsomes from Pg-/- mice was markedly diminished, whereas liver microsomes from control mice showed rapid SCD degradation similar to that observed in rat liver microsomes. These findings indicate that SCD is degraded by a protease related to Pg, and suggest that plasmin moonlights as an intracellular protease.


INTRODUCTION
The biophysical properties of membranes are critically dependent on the degree of desaturation of their hydrocarbon core (1). When environmental (dietary) sources are scarce, the pool of unsaturated fatty acyl-CoA precursors required for the assembly of membranes and other lipidbased structures is maintained by synthesis of oleoyl-CoA (18:1) and palmitolyl-CoA (16:1) in the endoplasmic reticulum (ER) by ∆ 9 stearoyl-CoA desaturase (SCD). The rate of unsaturated fatty acyl-CoA synthesis is directly related to the concentration of SCD in the ER (2); no allosteric mechanisms of SCD regulation have been identified. The concentration of SCD in the ER membrane is determined by the rates of SCD synthesis and degradation. Transcriptional regulation of SCD synthesis has been studied extensively (3,4). SCD degradation is less well understood, but is critical to the regulation of SCD since the concentration of an enzyme can be reduced rapidly only if it is short-lived. Hepatic SCD is degraded more rapidly than most ER proteins and has a half-life of about 3-4 hours (5). The physiological need to rapidly degrade SCD is not clear, but this property of SCD has been maintained during evolution since yeast SCD is also rapidly degraded (6). Presumably, excessive availability of unsaturated fatty acyl-CoAs would result in the assembly of membranes that are too disordered and be detrimental.
We have studied the degradation of SCD in two systems. Investigation of SCD-green fluorescent protein chimeras expressed in cultured animal cells identified a 33-amino acid segment at the amino terminus of SCD that is essential for optimal degradation of those constructs (7,8). We have also investigated the degradation of SCD in liver using conventional methods of subcellular fractionation, protein isolation, and reconstitution (9,10). When liver microsomes prepared from rats induced to express high levels of hepatic SCD are incubated at 37 o C, rapid and selective degradation of SCD occurs (9). The microsomal protease that degrades by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 5 SCD (SCD Pr) is tightly membrane bound, as it cannot be extracted by selective release of lumenal ER proteins or by high salt washing procedures that remove proteins from the cytosolic surface of microsomes (10). [ 35 S]SCD synthesized in vitro is also rapidly degraded by desaturase induced liver microsomes, whereas, under the same conditions, several other microsomal proteins synthesized in vitro are stable (11). Using specific degradation of [ 35 S]SCD as an assay to monitor the purification of SCD Pr, a 90 kDa protease was isolated from the Triton X-100 insoluble fraction of high salt washed microsomes (11). Upon incubation, the 90-kDa form of the protease undergoes rapid conversion to a series of smaller proteins. This conversion is associated with a marked increase in proteolytic activity. The purified protease is partially inhibited by diisopropylphosphofluoridate, dithiothreitol, and leupeptin suggesting it is a serine protease zymogen with Arg/Lys specificity and an essential disulfide bond (11). Mass spectrometry of its tryptic peptides indicated that the 90-kDa protein is a microsomal form of Pg.
This prompted us to compare SCD Pr with Pg isolated from plasma, to study the activation of Pg in microsomes, and to examine SCD expression and degradation in mice with a genetic deficiency in Pg. In light of recent evidence that Pg is expressed in a broad range of tissues (12), the significance of plasmin as an intracellular protease is discussed. Immunoblot analysis. Preparation of polyclonal antisera against SCD was described previously (10). Proteins were separated on 6% or 10% SDS-PAGE and transferred onto PVDF membranes. The membranes were incubated with 3% albumin. The transblots were incubated with antibody against SCD or rat Pg overnight at 4 o C. Bound antibody was detected with an anti-IgG-alkaline phosphatase conjugate and a phosphatase detection kit (Kirkegaard and Perry Laboratory, Gaithersburg, MD).
Preparation of microsomes and purification of the SCD protease. Desaturase-induced rat liver microsomes were prepared as described (10). Subsequent purification of the SCD protease was done at 4 o C as described previously (11). Protein concentrations were determined by the Coomassie colorimetric method (Pierce).

MALDI Mass Spectrometry.
For identification of the 90-kDa protein, gels were stained with Coomassie blue and protein bands were excised with a razor blade. Gel pieces were rinsed in water and in-gel tryptic digest was performed. Peptide digests were concentrated using ZipTip C18 micropipette tips (Millipore Corp.). MALDI-TOF and Post-Source-Decay (PSD) data were obtained on an AXIMA-CFR mass spectrometer using software version 2.0.1 (Kratos Analytical, Manchester, UK). Mass spectra were obtained at a laser power near threshold with pulsed extraction optimized for 2,000 Da. PSD data were obtained at a laser power about 15% higher than threshold. Samples were co-crystalilized with alpha-cyano-4-hydroxy-cinnamic acid (10mg/ml in 1% formic acid, 50% acetonitrile). PSD fragment ions were fitted to a generated curve that was calibrated with PSD fragments from the synthetic peptide, P 14 R. Data base searches were performed using Protein Prospector version 3.4.1. PSD fragments were searched 8 against the NCBI nonredundant data base using MS-Tag and 0.2 Da as the parent tolerance and 0.5 Da as the fragment ion tolerance (13).
Inhibition of SCD Pr by bdellin. Inhibition of SCD proteolysis was determined by the addition of bdellin to aliquots of either the translation mixture containing [ 35 S]-Met labeled SCD and SCD Pr or to desaturase induced rat microsomes. The samples were incubated on ice for 30 min followed by incubation at 37˚C for the indicated times.
The gels were washed with 50 mM Tris-HCl, pH 8.0 containing 2.5% Triton X-100 at room temperature for 1 hr and then incubated in 50 mM Tris-HCl, pH 8.0 at 37˚C either overnight for standard zymography or 1 hr for Pg zymography. The gels were stained with Coomassie blue.
Stearoyl-CoA desaturase activity The microsomal membrane homogenates were prepared as described above. Reactions were performed at 37 o C for 1 h with aliquots of microsomal protein suspension at 8µg/µl and 1.5 nmole of [1-14 C]-stearoyl-CoA (62 µCi/µmole), 100 µM of NADH, 9 impregnated TLC using benzene as developing solution. The TLC plates were then subjected to autoradiography. The enzyme activity was determined as nmole•min -1 •mg -1 protein.
HPLC analysis of SCD degradation. To identify SCD cleavage sites, 20 µg purified microsomal SCD was incubated with a 2 µl aliquot of either SCD Pr fraction or plasmin (1 µg) at 37 o C for 12 hr. Rat plasmin was generated from rat Pg by urokinase (1:20 ratio) at 37˚C for 3 hr.
Peptides were extracted with 0.1% trifluoroacetic acid (TFA) in 25% acetonitrile in water and subjected to HPLC using 0.1%TFA in water (A) and 0.1% TFA in 75% acetonitrile (B) as solvents with a linear gradient of 0-70% B in 50 min. Isolated peptides were sequenced on an Applied Biosystems 470A sequencer.
Two dimensional difference gel electrophoresis . Rat Pg and SCD Pr were labeled with Cy3 and Cy5 dye, respectively, followed by isoelectric focussing (24 cm, pH 3-10 IPG strip) as per manufacturers instructions (Amersham Biosciences). For the second dimensional separation, the IPG strips were then loaded and run on a 12.5% SDS-PAGE gel. Both samples were run on the same gel and fluorescent images were analyzed.

RESULTS
We previously reported the purification of a 90-kDa protein that selectively degrades microsomal SCD. In order to identify the 90-kDa protein, tryptic peptides obtained by in-gel digestion of SDS-PAGE gel slices were analyzed by MALDI-TOF mass spectrometry (Fig. 1A).
Eleven peptides matching the molecular weight predicted for tryptic peptides of Pg were identified ( Table 1) The presence of Pg in the SCD Pr preparation was confirmed by SDS-PAGE and immunoblot analysis. The mobility of SCD Pr was identical to Pg isolated from rat plasma ( Fig. 2A). A protein band with mobility identical to Pg was detected in microsomal fractions obtained during purification of SCD Pr by immunoblot analysis with Pg antibody, and the amount of the Pg immunoreactive band increased with progressive purification of SCD protease activity (Fig. 2B).
The mobility of both Pg and SCD Pr relative to an 111-kDa protein standard was dependant on the acrylamide concentration of the gels. In 6% SDS-PAGE gels, both Pg and SCD Pr migrated slightly slower than the 111-kDa protein standard ( Fig. 2) whereas in 10% SDS-PAGE gels they migrated faster than the 111-kDa protein standard (Fig. 3B). The reason for this difference relative to the standard is not clear, but under both sets of SDS-PAGE conditions, the mobility of SCD Pr was identical to the mobility of Pg. The SDS-PAGE protein profiles of SCD Pr under 12 reducing and nonreducing conditions were compared with Pg ( Fig. 3A and B). Pg had a significantly higher mobility in nonreduced SDS-PAGE gels as compared with reduced gels presumably because its numerous internal disulfide bonds maintain it in a compact conformation even after denaturation in SDS. SCD Pr exhibited the same SDS-PAGE mobility shift under nonreducing conditions. The SCD Pr preparation used in the experiments shown in Fig. 3A and B had been stored briefly at 4°C. We previously showed that the 90 kDa form of SCD Pr is converted to a spectrum of smaller peptides when stored at 4°C (11). A number of minor peptide bands were detected by reduced SDS-PAGE of the stored SCD Pr preparation that were not evident in nonreduced gels. This suggests that upon storage at 4°C certain peptide bonds in SCD Pr are cleaved and the resulting peptides are held together by disulfide bonds.
To evaluate their N-linked carbohydrate status, Pg and SCD Pr were digested with endoglycosidase F followed by SDS-PAGE and immunobloting ( Fig. 3C and D).
Endoglycosidase F digestion of rat Pg and SCD Pr did not alter their mobility upon SDS-PAGE.
Human Pg was used as a positive control in these experiments since it exists in two major glycoforms (17 Since SCD Pr appeared to represent a microsomal form of Pg, the ability of Pg and plasmin isolated from rat plasma to degrade SCD was investigated (Fig. 4A). There was no detectable degradation of SCD when it was incubated with 0.6 µM Pg for two hours at 37°C. By contrast, coincubation of SCD with 0.6 µM Pg and urokinase (uPA) or tissue Pg activator (tPA) led to extensive degradation of SCD within two hours (Fig. 4A, lanes 4 and 5). These results indicate that plasmin is able to degrade SCD. The results obtained with Pg alone suggest the Pg preparation was not contaminated with plasmin or Pg activators. The peptides produced by plasmic cleavage of SCD were compared with degradaton products produced by incubating SCD with SCD Pr (Fig. 4B and C). As shown, the HPLC profiles of the peptides produced by SCD Pr and plasmin are nearly identical. Sequence analysis of the identified peptides showed cleavage at Lys residues and a single peptide resulting from cleavage at His residue (Fig. 4D). No cleavage at Arg residues were identified, despite the presence of 22 Arg residues and 20 Lys residues in the SCD sequence (9).
The proteolytic activity in microsomes and microsomal fractions obtained during the purification of SCD Pr was analyzed by zymography on casein gels (Fig. 5). Zymography is a method for measuring proteolytic activity in SDS gels impregnated with a protein substrate such as casein or gelatin (14)(15)(16). Proteases that remain active in SDS or have the ability to refold into an active conformation when the SDS concentration is lowered can be analyzed by this method.
Since most of the SCD Pr activity in the hydroxyapatite fractions is latent (11) conditions. Interestingly, Pg which was free of plasmin by other criteria (see above) yielded a 70-kDa band of proteolytic activity on casein zymography (Fig. 5, lane 2). This observation suggests that Pg undergoes conversion to plasmin during zymography, or refolds into a conformation with protease activity when the SDS is removed. After Pg was incubated with uPA, a spectrum of proteolytic species from 40-kDa to 70-kDa was detected (Fig. 5, lane 3).
The primary proteolytic activity detected in microsomes by casein zymography migrated with an apparent molecular weight of 70-kD. The mobility of this protease is identical to the mobility of the protease detected in Pg (compare lanes 2 and 4 in Fig. 5). When microsomes were incubated at 37 o C, the casein hydrolysis activity of the 70-kDa protease increased, and a 30-kDa protease was generated (Fig. 5, lane 5). The same pattern was seen when high salt washed microsomes (HSWM) were analyzed, but a smaller quantity of the 30-kDa protease was produced, and it appeared as a doublet (Fig. 5, lane 7). When HSWM were solubilized with 2% Triton X-100 (TX-100) at 4°C, SDC Pr activity was recovered in the insoluble fraction (11). Since it appeared that SCD Pr was a product similar to activated Pg, we analyzed microsomal fractions for Pg activators by zymography using SDS-PAGE gels containing copolymerized casein and Pg (16). By this method, two Pg activators were detected in microsomes and in partially purified SCD Pr preparations with apparent molecular weights of 35-kD and 55-kDa.
( Fig. 6A). These bands of activity are microsomal Pg activators because they were not detected in casein gels without Pg and did not align with those of uPA or tPA (Fig. 6B). Several proteolytic bands were detected in Pg activator zymograms of SCD Pr (Fig. 6A, lane 7), however, these bands were also present in casein gels without copolymerized Pg (Fig. 6B, lane   7), indicating that they represent caseinases rather than specific Pg activators.
Further proof that SCD is degraded by a microsomal form of Pg was provided by bdellin inhibition studies. Bdellin, also known as bdellastasin, is a 59 residue, cysteine-rich serine protease inhibitor of the antistasin family which occurs naturally in the medical leech (Hirudo medicinalis) (18). Bdellin is a powerful and relatively specific proteinase inhibitor of trypsin and plasmin and displays no inhibition for a variety of other serine proteases (19). As shown in Fig.   7, 100 µM bdellin inhibited the degradation of native SCD in microsomes and inhibited the degradation of [ 35 S]SCD by purified SCD Pr.
To explore the role of Pg in the degradation of SCD in vivo, we isolated microsomes from Pg deficient (Pg-/-) mice, and compared the levels of SCD protein, SCD enzyme activity, and SCD Pr activity with wild-type control animals (Fig. 8). Liver microsomes from Pg-/-mice had significantly higher levels of SCD protein and SCD enzyme activity than microsomes from the wild-type mice. Degradation of SCD in microsomes from Pg-/-mice was slow, whereas liver by guest on March 24, 2020 http://www.jbc.org/ Downloaded from microsomes from control mice showed rapid SCD degradation similar to that observed in rat liver microsomes.

DISCUSSION
Hepatic microsomal desaturase activity is regulated by the rapid degradation of SCD. SCD is also rapidly and selectively degraded in vitro when liver microsomes are incubated at 37 o C (9).
We previously reported that the microsomal SCD protease (SCD Pr) is a 90-kDa zymogen with Arg/Lys specificity (11). The data presented here show that the 90-kDa protein is a microsomal form of Pg. Pg is secreted by hepatocytes, so it is not surprising that it can be isolated from hepatic microsomes which are derived predominantly from secretory organelles including the endoplasmic reticulum. Initially, we were concerned that Pg might also adsorb onto the surface of microsomes during tissue homogenizaton. However, several observations indicate that the source of Pg in our preparations is intracellular. SCD Pr was isolated from microsomes that had been extensively washed with high salt buffer (10 mM Tris-acetate [pH 8.0], 20% glycerol, 500 mM NaCl, 20 mM EDTA, 1 mM DTT) in order to remove trapped cytosolic proteins and peripheral membrane proteins. This washing procedure removed more than 95% of the peripheral membrane protein but did not deplete microsomal membranes of Pg or SCD Pr activity. Extensive perfusion of livers to remove plasma proteins prior to homogenization had no effect on the Pg content of microsomes. Most importantly, the observation that Pg-/-mice have elevated SCD activity suggests that Pg degrades SCD in vivo.
SCD Pr activity was recovered in the TX100 insoluble fraction when high-salt washed microsomes were solubilized with 2 % TX100. This suggests that SCD Pr is derived from a microsomal Pg fraction that is bound to the microsomal membrane surface. Earlier work showed that 40% of Pg in rough microsomes is tightly associated with microsomal membranes, and that the secretion of Pg is delayed relative to albumin (20). The mechanism by which Pg binds the microsomal membrane and is retained in the ER is unknown, however several hypotheses are consistent with current knowledge of ER protein traffic. Post-translational lipidation with palmitate or an isoprenoid derivative would explain membrane binding, and would not change the SDS-PAGE mobility significantly. Palmitoylation and isoprenylation both occur in the ER (21). Pg binds C-terminal lysyl residues in fibrin and cell surface proteins via kringle structures in its N-terminal domain. Thus, interaction with lysyl residues in microsomal proteins is a second possible mechanism for the tight binding of SCD Pr to microsomal membranes. A number of proteins, including C-reactive protein (22) and β-glucuronidase (23), which lack ER retention/retrieval signals are retained in the ER by forming a complex with lumenal esterases 1 and 2. Microsomal esterases are retained in the ER by a C-terminal HIEL motif (24). Nacent secretory glycoproteins are maintained in a monoglucosylated state and retained in the ER by lectins until folding is complete (25). Finally, the current paradigms for protein folding and ER quality control suggest that "incorrect" conformations are recognized and held in the ER by chaparones. A fraction of newly synthesized microsomal Pg molecules may have a unique conformation or structural feature that codes for ER retention. Our results address several of the possible mechanisms for retention of Pg in the ER. MALDI-TOF mass spectrometry of peptides obtained by in-gel tryptic digestion and purification by C18 micropipette tips did not identify post-translational modifications that could explain membrane binding. However, only eleven peptides were identified by this method. This is sufficient structural information to identify SCD Pr as a form of Pg; more extensive structural analysis will be necessary to exclude posttranslational modifications. Retention of Pg in the ER probably does not involve lectins, because we did not detect N-linked carbohydrate in SCD Pr or Pg by N-glycosidase F digestion.
How can the generation of plasmin, a protease with broad substrate specificity, account for the highly selective degradation of SCD in microsomes? During the dissolution of clots, fibrin is specifically degraded through the formation of a ternary complex between fibrin, Pg, and tPA (26). The formation of this complex enhances the specificity of plasmin by colocalizing the activator, zymogen, and substrate and also by conformational changes that occur when tPA and Pg bind to fibrin. In the absence of fibrin, the conversion of Pg to plasmin by activators is inefficient and the specificity toward fibrin is lost. Thus, one potential mechanism for the apparent specificity of SCD cleavage in isolated microsomes is that SCD and Pg colocalize in the ER/microsomal membrane. The microenviroment of the membrane likely influences both the activation and catalytic activity of SCD Pr.
Glu-Pg is synthesized in hepatocytes as a single chain protein of 791 amino acids with 24 disulfide bridges, and a molecular weight of 92-kDa (26). Conversion of Pg to the active serine protease plasmin requires specific cleavage of the Arg562-Val563 peptide bond in the C-terminal serine protease domain. Two physiologic Pg activators have been identified: tissue-type Pg activator (tPA) and urokinase-type Pg activator (uPA). The synthesis of Pg activators by hepatocytes has not been extensively studied. In an immunohistochemical study, uPA was identified in oval cells of the liver, but not hepatocytes (27). uPA was also identified in cultured stellate liver cells by zymography (28). We identified Pg activators of Mr 35-kDa and 55-kDa in hepatic microsomal preparations by zymography (Fig. 6). Since these proteases are candidates for intracellular Pg activators, their identity and cellular origin are of interest. Pg deficiency in mice causes severe thrombosis and is associated with a poorly understood wasting syndrome with a high mortality rate (34). Because loss of fibrinogen rescues mice from the pleiotropic effects of Pg deficiency, it has been suggested that the only essential 20 physiological role of Pg is fibrinolysis (35). More recently, Bezerra et. al. (36) reported that Pg deficiency impedes the clearance of necrotic hepatocytes after toxic liver injury, and genetically superimposed deficiency of the Aα fibrinogen chain did not correct the abnormal phenotype.
Furthermore, non-fibrin substrates of plasmin have been identified in several laboratories (reviewed in reference 12 and 37), and despite exclusion of Pg and fibrinogen by the blood-brain barrier, Pg has been implicated in several natural and experimental neurodegenerative processes (38). Our results suggest that some of the plasmin substrates in these pathologic processes may be intracellular. Leptin deficient mice ob-/-have up-regulated SCD (7-fold) and are obese (42,43 (42,43). Analysis of food intake and energy expenditure in abj/abj;ob/ob mice showed increased oxygen consumption and consumed more food than ob/ob littermates suggesting that hepatic SCD deficiency may modulate pathways that decrease fatty acid synthesis and increase lipid oxidation (42,43). While SREBP-1a and ob/ob mice have similar up-regulated hepatic SCD these two models differ significantly, since serum lipid levels in SREBP-1a mice were largely unaltered (39). In summary, hepatic SCD is a  were generated by in-gel tryptic digestion, and concentrated by reverse phase adsorption chromatography. MALDI-TOF data were obtained for the isolated peptides as described in "Experimental Procedures." The numbers in parentheses indicate the peptides shown in Table 1.