Structural analysis of a tumor-produced sulfated glycoprotein capable of initiating muscle protein degradation.

A material of Mr 24,000 has been isolated from a cachexia-inducing mouse tumor (MAC16) and shown to initiate protein degradation in isolated gastrocnemius muscle. Biological activity was destroyed by preincubation with peptide N-glycosidase F (PNGase F) and endo-alpha-N-acetylgalactosaminidase (O-glycosidase) but not by neuraminidase or trypsin. Antibody reactivity was destroyed by treatment with periodate, indicating carbohydrate moieties to be the antigenic determinants. Antigenic activity was also reduced by treatment with PNGase F and O-glycosidase and was completely destroyed by treatment with chondroitinase ABC but was unaffected by treatment with either trypsin or chymotrypsin, confirming that the N- and O-linked sulfated oligosaccharide chains were both the antigenic and biological determinants. Biosynthetic labeling of MAC16 cells using a combination of [35S]sulfate and [6-3H]GlcN gave a single component of Mr 24,000 containing both radiolabels. Similar material could not be isolated from a cell line (MAC13) originating from a tumor that does not cause cachexia in vivo. Digestion of 3H/35S material with PNGase F produced two fragments of Mr 14,000 and 10,000 containing both radiolabels, and digestion with O-glycosidase produced three fragments of Mr 14,000, 6,000, and 4, 000, the first two contained both radiolabels and the third contained only 3H. Digestion of the fragment of Mr 14,000 released by PNGase F with O-glycosidase also gave fragments of Mr 6,000 and 4, 000. The products from both digestions were acidic as determined by anion exchange chromatography on DEAE-cellulose. The negative charge on the fragment of Mr 4,000 was removed by treatment with alkaline phosphatase. This suggests that the charge originated from phosphate residues, and this has been confirmed by biosynthetic labeling of MAC16 cells with [32P]orthophosphate, where radiolabel was incorporated into material of Mr 24,000 and into the fragment of Mr 4,000 after treatment with O-glycosidase. To determine the size of the polypeptide core MAC16 cells were biosynthetically labeled with L-[2,5-3H]His which after chemical deglycosylation produced a major component of Mr 4,000. These results suggest a model for the Mr 24, 000 material consisting of a central polypeptide chain of Mr 4,000 and with phosphate residues that may be attached to the polypeptide or a short oligosaccharide chain containing GlcN, one O-linked sulfated oligosaccharide chain containing GlcN, and of Mr 6,000 and one N-linked sulfated oligosaccharide chain of Mr 10,000 also containing GlcN. Neither chain was cleaved into disaccharides with chondroitinase ABC, suggesting that the material is a sulfated glycoprotein.

A material of M r 24,000 has been isolated from a cachexia-inducing mouse tumor (MAC16) and shown to initiate protein degradation in isolated gastrocnemius muscle. Biological activity was destroyed by preincubation with peptide N-glycosidase F (PNGase F) and endo-␣-N-acetylgalactosaminidase (O-glycosidase) but not by neuraminidase or trypsin. Antibody reactivity was destroyed by treatment with periodate, indicating carbohydrate moieties to be the antigenic determinants. Antigenic activity was also reduced by treatment with PNGase F and O-glycosidase and was completely destroyed by treatment with chondroitinase ABC but was unaffected by treatment with either trypsin or chymotrypsin, confirming that the N-and O-linked sulfated oligosaccharide chains were both the antigenic and biological determinants.

Biosynthetic labeling of MAC16 cells using a combination of [ 35 S]sulfate and [6-3 H]GlcN gave a single component of M r 24,000 containing both radiolabels. Similar material could not be isolated from a cell line (MAC13) originating from a tumor that does not cause cachexia in vivo.
Digestion of 3 H/ 35 S material with PNGase F produced two fragments of M r 14,000 and 10,000 containing both radiolabels, and digestion with O-glycosidase produced three fragments of M r 14,000, 6,000, and 4,000, the first two contained both radiolabels and the third contained only 3  Depletion of skeletal muscle is an important factor contributing to the decreased survival of cancer patients with loss of cardiac and respiratory muscles being most important. A decreased nutrient intake plays an important role in wasting of lean body mass in cancer cachexia, but it appears that it does not fully account for the changes observed (1). Although protein synthesis is decreased (2), an accelerated protein breakdown accounts in large part for the muscle wasting observed (3).
Several factors have been postulated as signals for this increased muscle proteolysis including tumor necrosis factor-␣ and interleukins 1 and 6 (4,5). While continuous infusion of the cytokines in vivo has been shown to increase protein degradation in skeletal muscle, none of the cytokines produced a direct stimulation of proteolysis when incubated in vitro (4,5).
A novel material of M r 24,000 that appears to fulfill the function of triggering muscle proteolysis during the process of cancer cachexia (6, 7) has been purified from a cachexia-inducing mouse tumor (MAC16). This material was capable of inducing muscle protein degradation in isolated gastrocnemius muscle preparations and of inducing weight loss in vivo and will be referred to as proteolysis inducing factor (PIF). 1 Similar, if not identical, material was isolated from the urine of patients with pancreatic carcinoma and weight loss (7).
Structural studies of PIF indicated a short peptide chain of M r 2,000, which was extensively glycosylated at both Asn and Ser residues (7). Enzymatic degradation suggested that some of the carbohydrate chains contained sulfate residues, whereas lectin blotting (6) indicated the presence of GlcNAc. The material bound strongly to albumin to form a species of M r 69,000, probably through the sulfate residues which would produce a strong electrostatic linkage. Recent studies (8) show that cellbound albumin binds peptidoglycan, heparin, and sulfated heparinoids as a complex of M r 70,000.
The purpose of the present investigation was to characterize PIF in terms of the number, type, and attachment of the carbohydrate chains to the peptide backbone as well as determining the role of the carbohydrate chains in antibody reactivity and protein degradative activity in isolated gastrocnemius muscle.

EXPERIMENTAL PROCEDURES
Materials-RPMI 1640 tissue culture medium with and without L-histidine and fetal bovine serum were from Life Technologies, Inc., * This work was supported by Cancer Research Campaign Grant SP1518. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom requests for reprints should be addressed.
Measurement of Protein Degradation-Female NMRI mice were killed by cervical dislocation, and their gastrocnemius muscles were quickly ligated, dissected out, and placed in ice-cold isotonic saline. To minimize diurnal variation and to ensure animals were in the fed state all animals were sacrificed between 9 and 10 a.m. The muscles were blotted, weighed, and carefully tied via tendon ligatures to stainless steel incubation supports (9). This prevents contraction and improves protein balance and energy status (10). Protein degradation was measured by tyrosine release, since tyrosine rapidly equilibrates between intracellular pools and the medium and is neither synthesized nor degraded. Muscles were preincubated in RPMI 1640 (3 ml) lacking phenol red in the presence of serum (280 l) for 30 min at 37°C in an atmosphere saturated with O 2 :CO 2 (19:1). The muscles were rinsed and incubated for a further 2 h in Krebs-Henseleit bicarbonate buffer, containing 6 mM D-glucose, 1.2 mg/ml bovine serum albumin, and 130 g/ml cycloheximide with continuous gassing. At the end of the incubation the buffer was removed, deproteinized with ice-cold 30% trichloroacetic acid (0.2 ml), centrifuged at 2800 ϫ g for 10 min, and the supernatant used for the measurement of tyrosine by a fluorimetric method (11) at 570 nm on a Perkin-Elmer LS-5 luminescence spectrometer.
Cell Culture and Radiolabeling Procedures-The two cell lines MAC16 and MAC13 were derived from the solid tumors grown in vivo and were a gift from Prof. J. A. Double and Dr. M. Bibby (University of Bradford, Yorkshire, United Kingdom). The MAC16 cells were microbiologically free and grew in suspension, while the MAC13 grew as a monolayer in RPMl 1640 medium containing 5% fetal bovine serum under an atmosphere of 5% CO 2 in air. Cells were grown in progressively lower fetal bovine serum concentrations, and a level of 1.5% was used for the labeling experiment. Cells (10 8   mice with established cachexia on tyrosine release from gastrocnemius muscle Solid MAC16 tumors excised from mice with established cachexia were fractionated on an affinity column containing the MAC16 monoclonal antibody as described (6). The enzyme-linked immunosorbent assay-positive fractions were pooled and concentrated, and portions (013A492 units 0.77 g of protein) were treated with the enzymes as described under "Experimental Procedures." All samples were preincubated with serum from non-tumor-bearing mice for 30 min at 37°C prior to determination of tyrosine release from gastrocnemius muscle as described under "Experimental Procedures." After incubation with trypsin (0.2 g/g of protein) for 20 h; ␣1-antitrypsin (8.3 g) was added prior to addition to serum. Using this procedure tyrosine release when trypsin was added to serum did not differ significantly from that observed with normal mouse serum.

Analysis of a Tumor-produced Protein Degradation Factor
containing 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 0.5 mM EGTA, and 1 mM dithiothreitol and dissociated using an ultrasonic oscillator. Debris was removed by centrifugation (15,000 rpm for 20 min), and solid ammonium sulfate (80% w/v) was added to the supernatant, and the mixture was stored overnight at 4°C. The precipitated proteins were collected by centrifugation, and the pellet was resuspended in 10 mM Tris⅐HCl, pH 8.0, containing 0.5 mM PMSF, 0.5 mM EGTA, and 1 mM dithiothreitol, and salt was removed by ultrafiltration with an Amicon filtration cell containing a membrane filter with a molecular weight cut-off of 10,000 against the same solution. The concentrated sample was loaded onto an affinity column containing mouse monoclonal antibody (6)   from the initial rate and using E ϭ 3800 M Ϫ1 for reaction products at pH 8 was 2.85 and 2.70 mol/min for 1 unit of chondroitinase ABC and AC, respectively. Chemical deglycosylation by anhydrous trifluoromethane sulfonic acid was achieved with a Glycofree deglycosylation kit according to the manufacturer's instructions. Nitrous acid deamination was carried out on the sample (in 50 l of water) kept at room temperature for 10 min. The products of the reaction were then analyzed on a column of Sephadex G-50.
Enzyme-linked Immunosorbent Plate Assay-Samples were divided into two and immobilized on a 96-well polyvinylchloride assay plate overnight at 4°C. The liquid was removed by aspiration, and the wells were washed three times with PBS ϩ 0.1% Tween 20 (200 l). Blocking solution (200 l of PBS containing 0.1% Tween 20 and 3% bovine serum albumin) was added to the wells, and the plate was incubated for 2 h at 37°C. One-half of the sample was incubated with the monoclonal antibody (10 g/ml) in blocking solution (100 l) for 1 h at room temperature, while the other half was incubated in the same solution but in the absence of the antibody. After removal of the antibody solution the wells were washed six times before the addition of a peroxidase-conjugated anti-mouse immunoglobulin, diluted 1 in 500 in blocking solution (100 l/well), and the plates were incubated for a further 1 h at 37°C. The wells were washed six times, and then the substrate solution, o-phenylenediamine dihydrochloride (0.04%), hydrogen peroxide (0.012%) in 0.15 M phosphate citrate buffer, pH 5.0 (100 l/well), was added for 30 min. The reaction was terminated by the addition of 0.2 M H 2 SO 4 (50 l/well), and the absorbance was determined at 492 nm using a microplate reader (Anthos Labtec Instruments).

RESULTS
When isolated mouse gastrocnemius muscle was incubated with serum from mice bearing the MAC16 tumor and with a weight loss between 2 and 4.4 g an increased protein degradation was observed as measured by tyrosine release (Table I). This effect was attenuated by incubation of the serum with monoclonal antibody prior to addition to the muscle preparation. An increased tyrosine release could also be produced by addition of affinity purified antigen to serum from non-tumorbearing mice (Table I), thus confirming that this material was the serum component responsible for the degradation of skeletal muscle proteins. The increased tyrosine release produced by the affinity purified antigen was abolished by preincubation with PNGase F, O-glycosidase, and sulfatase but unaffected by treatment with neuraminidase or trypsin. These results suggest that protein degradation is mediated by N-and O-linked carbohydrate chains in the molecule.
A similar relationship was obtained for antigen binding activity (Table II). Immunological activity was completely destroyed by treatment with periodate, indicating that the carbohydrate moieties are involved in the epitope. Antibody binding activity was inhibited by PNGase F, O-glycosidase, and sulfatase (Table II) Inc.) under the influence of a NaCl gradient from 0 to 0.3 M NaCl. Affinity purified biosynthetically labeled material (Fig. 2) was subjected to enzymatic deglycosylation as described under "Experimental Procedures," and the products were desalted using a microcon microconcentrator containing a filter with a molecular size cut-off of 10,000 (Amicon Corp.) and fractionated on a DEAE-cellulose column. The flow rate was 0.2 ml/min with solvent system A, 10 mM sodium phosphate buffer, pH 5.3, and solvent system B, 10 mM sodium phosphate containing 0.3 M NaCl. The gradient was for 10 min at 0% B, 40 min at 100% B, 50 min at 100% B, and 60 min at 0%t B. Absorbance was monitored at 214 nm, and the radioactivity of the individual fractions was determined using a dual counting procedure. The bands eluting at 0.28 M NaCl (B) and 0.20 M NaCl (C) were further fractionated on a Sephadex G-50 column. carbohydrate chains are the antigenic determinants. In addition antigen reactivity was reduced by treatment with chondroitinase AC and completely destroyed by treatment with chondroitinase ABC but was unaffected by treatment with endo-␤-galactosidase. Since the latter enzyme only hydrolyzes internal ␤-galactoside linkages of oligosaccharides having nonsulfated ␤-galactose residues, these results suggest that the principal immunological determinants reside in sulfated oligosaccharide chains and that the material is a glycoprotein or proteoglycan.

. Elution profile of radioactivity determined as 35 S (q) and 3 H (E) in the oligosaccharide chains released after enzymatic deglycosylation with PNGase F (A) and fractionation by high performance liquid chromatography on a DEAE-cellulose column (Applied Biosystems
To obtain information on the nature of the linkage and the number and types of glycan chains, MAC16 cells were doubly labeled with [ 35 S]sulfate and [ 3 H]GlcN, and the antigen was purified by ammonium sulfate fractionation and affinity chro-matography. The elution profile from the affinity column showed a single band of radioactivity containing both radiolabels (Fig. 1A), which represented 13% of the 35 S and 15% of the 3 H radiolabel from the ammonium sulfate precipitate. This material showed a single band of radioactivity corresponding to a M r of 24,000 on SDS-PAGE (Fig. 1B). Similar material could not be isolated from a cell line (MAC13) originating from a tumor that does not produce cachexia in vivo. Fractionation of the MAC16 material on a Sephadex G-50 column under dissociating conditions confirmed a single band of M r 24,000 that contained both 35 S and 3 H (Fig. 2A). The M r of this material was the same as that previously isolated from the MAC16 tumor using a combination of affinity and reverse phase hydrophobic chromatography (7). Identical sham incubations (with- Analysis of a Tumor-produced Protein Degradation Factor 12284 out PNGase F) showed no radioactivity eluting at positions corresponding to the released material. Re-chromatography of this material after overnight digestion with PNGase F gave two bands of radioactivity eluting at positions corresponding to M r of 14,000 and 10,000 (Fig. 2B). Both fragments contained 3 H and 35 S. Digestion of the M r 24,000 material with O-glycosidase and fractionation on Sephadex G-50 showed conversion to three bands of radioactivity corresponding to M r of 14,000, 6,000, and 4,000 (Fig. 2C). Incubation with buffer in the absence of O-glycosidase showed no degradation of the M r 24,000 material. The first two bands contained both 3 H and 35 S, and the band of M r 4,000 contained only 3 H. Treatment of the material of M r 14,000 produced from PNGase F digestion with O-glycosidase converted it into two fractions corresponding to M r of 6,000 and 4,000 (Fig. 2D).
To determine the acidity of the charged groups on each of the oligosaccharide chains, the M r 24,000 material was subjected to enzymatic digestion as above, and the products were fractionated on a DEAE-cellulose column under the influence of a linear gradient from 0 to 0.3 M NaCl. The identity of the eluted peaks was determined by exclusion chromatography on Sephadex G-50. Anion exchange chromatography after digestion with PNGase F gave two bands eluting at 0.20 M NaCl (M r 10,000) and 0.28 M NaCl (M r 14,000) (Fig. 3). Digestion with O-glycosidase gave three bands that all adhered to the DEAEcellulose column in order of elution 0.20 M NaCl (M r 6,000), 0.24 M NaCl (M r 4,000), and 0.29 M NaCl (M r 14,000) (Fig. 4). In both cases there was no further fractionation between the 3 H-and 35 S-labeled oligosaccharide chains. Although the O-linked oligosaccharide chain of M r 4,000 did not contain sulfate residues as determined by biosynthetic labeling, it was more acidic than the sulfated chains of M r 6,000 and 10,000, as determined by the concentration of NaCl required to elute it from the DEAEcellulose column.
The nature of the charged groups on this non-sulfated oligosaccharide chain was investigated by digestion with either neuraminidase or phosphatase followed by fractionation of the products by anion exchange chromatography. After treatment of the oligosaccharide with neuraminidase, the product still adhered to the DEAE-cellulose column and eluted at 0.24 M NaCl, suggesting the absence of sialic acid residues. However, after incubation with alkaline phosphatase the acid group was removed and the product no longer attached to the anion exchange column, although the apparent M r was not affected (results not shown). The M r of the intact molecule was also unaffected by alkaline phosphatase as determined by exclusion chromatography on Sephadex G-50. This suggests that phosphate residues are responsible for the negative charge on the O-glycosidase digestion product of M r 4,000. Antibody binding activity of the M r 24,000 material was reduced by 53% after treatment with alkaline phosphatase (Table II), and immunoreactivity on Western blotting was completely destroyed, suggesting that the phosphate residues are also important antigenic determinants. To confirm the presence of phosphate groups on the chain of M r 4,000, MAC16 cells were biosynthetically labeled with [ 32 P]orthophosphate, and the antigen was purified by ammonium sulfate precipitation and affinity chromatography as before. The 32 P was confined to a single band of M r 24,000 as determined by SDS-PAGE (Fig. 5A) and exclusion chromatography on Sephadex G-50 (Fig. 5B). Treatment with PNGase F (Fig. 5, A and C) or chondroitinase AC (Fig. 5A) yielded a single fraction containing 32 P of M r 14,000. Digestion of the material of M r 24,000 with either O-glycosidase (Fig. 5, A and D) or chondroitinase ABC gave a single band containing 32 P of M r 4,000. This material adhered to a DEAE column and was eluted with 0.24 M NaCl (Fig. 5E) as did the fragment from material labeled with [ 3 H]GlcN (Fig. 3).
To determine the size of the polypeptide core, MAC16 cells were labeled with L-[2,5-3 H]His, which is the amino acid at residue 16 of the core (6, 7). The radiolabel was incorporated into a single component of M r 24,000 after fractionation of the cell extract by affinity chromatography (Fig. 5B; Fig. 6). Treatment with PNGase F showed 3 H labeling of the fragment of M r 14,000 ( Fig. 5C; Fig. 6, the lower band in the autoradiograph may be an artifact). Treatment with O-glycosidase showed 3 H labeling of the M r 4,000 fragment ( Fig. 5D; Fig. 6). This fragment was the same as that found after biosynthetic labeling with 32 P since both adhered to a DEAE-cellulose column and were eluted with 0.24 M NaCl (Fig. 5E). Chemical deglycosylation with anhydrous trifluoromethanesulfonic acid showed a major band with a M r near 4,000 but lower than that formed by treatment with O-glycosidase (Fig. 6) and a minor band at M r 2,000. This suggests that the molecular weight of the polypeptide core is 4,000 and that there may be a short oligosaccharide chain (labeled with [ 3 H]GlcN) which is phosphorylated or the peptide chain itself may be phosphorylated.
To determine the nature of the oligosaccharide chains, material of M r 10,000, released by treatment of [ 35 S]sulfate-and [ 3 H]GlcN-labeled M r 24,000 with PNGase F, was treated with chondroitinase ABC followed by re-chromatography on a Sephadex G-50 column. Two bands of radioactivity corresponding to M r of 8,000 and 2,000 were obtained (Fig. 7A) but no low molecular weight material corresponding to disaccharides. Treatment of the oligosaccharide chain of M r 6,000 produced by cleavage with O-glycosidase with either chondroitinase AC or ABC (Fig. 7B) had no effect on the molecular weight. These results suggest that material of M r 24,000 does not contain glycosaminoglycan chains and that it is a sulfated glycoprotein rather than a proteoglycan. This conclusion was substantiated by analysis of the carbohydrate chains released after treatment with 0.  doubly labeled with [ 35 S]sulfate and [ 3 H]His released fragments of M r 14,000, 6,000, and 4,000 (Fig. 7D). The latter fragment was labeled only with 3 H, suggesting that it represented the polypeptide core. The elution positions of both the fragments of M r 14,000 and 6,000 were not affected by further treatment with chondroitinase ABC, AC, or nitrous acid (data not shown) again confirming that the oligosaccharide chains were not of the chondroitin, dermatan, or heparan sulfate type. DISCUSSION The data in the present study provide strong support that protein degradation in skeletal muscle during the process of cancer cachexia is mediated by a tumor-produced sulfated glycoprotein of M r 24,000. This material could be responsible for the accelerated breakdown of isolated rat diaphragm muscle observed when incubated with plasma from cancer patients with weight loss greater than 10% (12), since we have shown similar material in both human colonic carcinomas and in the urine of patients with pancreatic carcinoma and established cachexia (7). Although interleukin-I␣ and interleukin-I␤ were shown to stimulate protein catabolism in the rat diaphragm muscle bioassay, in the study of Belizario et al. (12) antibodies to the recombinant human cytokines gave only partial neutralization of bioactivity in less than half of the patients investigated, suggesting other active factors that could not be defined. The structure of the M r 24,000 material is novel, and the peptide sequence (6, 7) is distinct from the recognized cytokines. Antisera to this material was not cross-reactive with the cytokines' tumor necrosis factor-␣ and interleukin-6. In addition the ability of MAC16 cells to produce this material in vitro, as demonstrated by biosynthetic labeling studies, confirms that it arises from tumor rather than host cells. A histologically related cell line, MAC13, derived from a non-cachexia-inducing tumor was found not to be capable of producing radioactively labeled M r 24,000 material using either [6-3 H]GlcN or Na 2 35 SO 4 as the precursors. This suggests that it is only produced by some tumor cell lines which are capable of inducing cachexia.
Both functional and immunological studies provide evidence that biological activity is mediated through N-and O-linked oligosaccharide chains in the molecule. To identify the size and position of attachment of the oligosaccharide chains as well as the size of the polypeptide core, MAC16 cells have been biosynthetically labeled with 35 SO 4 , [ 3 H]GlcN, [ 3 H]His, and 32 P i followed by affinity purification to isolate the labeled product. In all cases a single component of M r 24,000 was obtained as determined by SDS-PAGE and exclusion chromatography. Similar results have previously been obtained with high performance liquid chromatography-purified material labeled with Na 125 I (7).
With regard to the molecular weight of the material and the fragments generated by enzymatic deglycosylation, these are apparent rather than real, since glycosylated molecules have hydrodynamic volumes that differ per unit of molecular weight from those of globular proteins, which have been used as molecular weight standards.
With this caveat incubation with recombinant protease-free PNGase F, which specifically cleaves the GlcNAc-Asn bond of 37°C in 250 mM Tris⅐HCl, pH 8.0. C, elution profile of radioactivity on Sephadex G-50 from affinity purified material of M r 24,000 biosynthetically labeled with 35 SO 4 (q) and [ 3 H]His (E) after incubation with 2 M NaBH 4 in 0.1 M NaOH for 16 h at 37°C. Excess NaBH 4 was destroyed with 0.25 M acetic acid in methanol, and the boric acid was removed by evaporation of the methanolic solution under a stream of nitrogen. The procedure was repeated twice with the same amount of acidified methanol and twice with methanol alone.