Protein Targets of Monocrotaline Pyrrole in Pulmonary Artery Endothelial Cells*

A single administration of monocrotaline to rats results in pathologic alterations in the lung and heart similar to human pulmonary hypertension. In order to produce these lesions, monocrotaline is oxidized to monocrotaline pyrrole in the liver followed by hematog-enous transport to the lung where it injures pulmonary endothelium. In this study, we determined specific endothelial targets for 14 C-monocrotaline pyrrole using two-dimensional gel electrophoresis and autoradiographic detection of protein metabolite adducts. Selective labeling of specific proteins was observed. Labeled proteins were digested with trypsin, and the resulting peptides were analyzed using matrix-assisted laser desorption ionization mass spectrometry. The results were searched against sequence data bases to identify the adducted proteins. Five abundant adducted proteins were identified as galectin-1, protein-disulfide isomerase, probable protein-disulfide isomerase (ER60), b - or g -cytoplasmic actin, and cytoskeletal tropomyosin (TM30-NM). With the exception of actin, the proteins identified in this study have never been identified as potential targets for pyrroles, and the majority of these proteins have either received no or minimal attention as targets

The pyrrolizidine alkaloid monocrotaline (MCT) 1 is a phytotoxin used experimentally to cause a pulmonary vascular syndrome in rats characterized by proliferative pulmonary vasculitis, pulmonary hypertension (PH), and cor pulmonale (1)(2)(3). Although MCT intoxication is used as a model for studying human PH, the initiating mechanism(s) by which this agent produces PH have remained elusive. To produce pulmonary insult, MCT must first be activated by the liver to the putative electrophile monocrotaline pyrrole (MCTP) (4,5) which has characteristics of a bifunctional cross-linking agent and has a half-life of ϳ3 s in aqueous environments near neutral pH (6). Stabilization of MCTP by red blood cells facilitates subsequent transport to the lung (7). The evidence for the involvement of the pulmonary endothelium as the target for MCT intoxication is supported by the circulatory proximity of the liver to the lung endothelium, evidence of increased thymidine uptake and decreased 5-hydroxytryptamine clearance by endothelial cells, and extravasculature leakage of large macromolecules (2,8,9). Human primary PH is hypothesized to be an inheritable dysfunction of the pulmonary vascular endothelial cells (10). In primary PH, disturbance of the endothelial cell surface is suspected to be the initiating factor in the formation of platelet aggregates (11) and cause the presence of in situ thrombosis (12,13). In vitro experiments with bovine pulmonary artery endothelial cells have shown that MCTP can cause a moderate decrease in their ability to act as a permeability barrier (14), cell proliferation is inhibited (15), prolonged cell cycle arrest in G2-M occurs (16,17), and cells unable to correct or compensate for electrophilic insult often undergo apoptosis (18). Apoptosis has recently been shown to occur in rat pulmonary artery endothelial cells following the in vivo administration of MCT (19).
Previous work has supported the involvement of endothelial cells as the target for MCT-induced pulmonary hypertension; however, the mechanism(s) by which these cells lose their ability to function correctly is unknown. With respect to pyrrole adduct formation this has been restricted to the measurement of covalent binding to endothelial cell DNA (16,20). MCTP has been shown to react in a facile manner with the thiol groups of cysteine and glutathione (21)(22)(23)(24)(25). A carbonium ion can be generated at both the C7 and the C9 positions on the pyrrole ring with the pyrrole structure being stabilized by resonance structures that share a charge with the bridge head nitrogen (26). This delocalization of charge for MCTP confers soft electrophile characteristics (27,28) in line more with reactivity toward soft nucleophile protein side chains than with nucleic acids, which are harder nucleophiles. It has previously been shown that MCTP reacts with thiol groups on proteins such as hemoglobin (22,29,30). Of the limited number of proteins identified as specific targets for MCTP, cytochrome P450 3A, which is responsible for the dehydrogenation of MCT, has also been shown to form adducts with pyrroles (31).
In this study we have coupled the use of two-dimensional gel electrophoresis and matrix-assisted laser desorption ionization (MALDI) to identify five major MCTP target proteins in human lung endothelial cells and discuss their potential relevance to the enigmatic process of pulmonary hypertension.

MATERIALS AND METHODS
Production of 14 C-Monocrotaline Pyrrole-All reagents unless otherwise indicated were obtained from Fisher. Crotalaria spectabilis was grown under a confined atmosphere of 14 CO 2 , and the 14 C-monocrotaline ( 14 C-MCT, purity Ͼ98%) was extracted and purified as described before (32). 14 C-MCT was converted to 14 C-monocrotaline pyrrole ( 14 C-MCTP, 1.95 mCi/mmol) by the method of Mattocks et al. (33) using tetrabromo-1,2-benzoquinone (Aldrich); the product was recrystallized in hexane:diethyl ether. Using fast atom bombardment mass spectrometry the conversion of MCT to MCTP was found to be complete; daughter spectra contained ions characteristic of the pyrrole (34). 14 C-MCTP was stored in N,N-dimethylformamide at Ϫ80°C until just prior to use.
Tissue Culture and Sample Preparation-Normal human pulmonary artery endothelial cells (passage 7-8) (Clonetics, San Diego, CA), from a 34-year-old female were grown to 80 -90% confluence prior to treatment. Cells in 175-cm 2 flasks (Falcon polystyrene) were incubated at 37°C, 5% CO 2 with humidity in EGM-2 medium (Clonetics). The medium was replaced with EGM-2 without 10% fetal bovine serum immediately before exposing cells to 105 M 14 C-MCTP delivered in 12.9 l of N,N-dimethylformamide/25 ml of media. Cells were removed from flasks using 39-cm cell scrapers (Sarstedt, Newton, NC). After pelleting, the cells were washed three times with isotonic phosphate-buffered saline (136.89 mM NaCl, 2.68 mM KCl, 10.14 mM Na 2 HPO 4 , and 1.76 mM KH 2 PO 4 , pH, 7.4). The supernatant was removed, and the cells from five flasks were combined, lysed in a 9 M urea (ultra pure) solution containing 4% CHAPS (99%), all obtained from Amersham Pharmacia Biotech plus 40 mM Tris and ␣-toluenesulfonyl fluoride (Aldrich). The latter was added just prior to use in 50 l of absolute ethanol to give a final concentration of 8 mM. Cells were disrupted with 500 l of the above, and the lysate was maintained at room temperature for 1 h. The bicinchoninic acid method, as described by the manufacturer (Pierce), was used to determine protein concentrations. Following protein determination, dithiothreitol (Amersham Pharmacia Biotech) was added to give a final concentration of 64.9 mM; lysate was retained at room temperature for an additional 25 min followed by centrifugation (100,000 ϫ g) for 1 h at 10°C. Supernatants were stored under N 2 at Ϫ80°C until needed.
Two-dimensional Gel Electrophoresis and Autoradiography-Protein samples (60 l, 400 g of protein) were diluted to 250 l with 8 M urea, 2% CHAPS, 18 mM dithiothreitol, and IPG buffer pH 4 -7 was added directly to give a final concentration of 2%. Electrophoresis reagents and hardware, including the Multiphor II platform and EPS 3500XL power supply were purchased from Amersham Pharmacia Biotech. A trace of bromphenol blue was added, and samples (250 l) were placed in a Immobiline DryStrip reswelling tray. Immobiline DryStrips (IPG) (13 cm, pH 4 -7) were placed gel side down in sample, and a layer of mineral oil was added. Samples were allowed to equilibrate with the strips overnight. Isoelectric focusing was performed on a MultiPhor II platform at 15°C, and strips were covered with mineral oil with the following voltage program: 360 V, 3 h; 1400 V, 0.5 h; 2600 V, 29 h, maximum current 1 mA. IPG strips were stored under N 2 in acidwashed glass tubes at Ϫ80°C prior to separation on the second dimension. After removal from storage, strips were allowed to just thaw before being first equilibrated in 30 ml of 50 mM Tris (pH 6.8) containing 30% v/v glycerol (ultra pure), 2% SDS, 6 M urea, 16.2 mM dithiothreitol, and a trace of bromphenol blue, followed by equilibration in the same buffer system (30 ml) with the replacement of dithiothreitol with 135 mM iodoacetamide (97%, Aldrich). Each equilibration phase was of 10-min duration and performed by placing IPG strips in a Petri dish with agitation generated by a rotary shaker. IPG strips along with a 0.5 ϫ 4-mm agarose cylinder containing 0.5 g of each molecular weight markers (14,000 -66,000, Sigma no. MW-SDS-70L) were sealed on top of a 145 ϫ 140 ϫ 1-mm Laemmli (35) separating gel (11% T, 2.7% C, acrylamide (99.9%) and bis-N,NЈ-methylenebisacrylamide were both obtained from Bio-Rad using 0.5% agarose (SeaKem Gold, FMC, Rockland, ME) dissolved in 125 mM Tris, pH 6.8, 0.1% SDS at 55°C. Electrophoretic separations were performed at 4°C, 10 mA/gel. Prior to protein transfers to Sequi-Blot PVDF membranes (0.2 m, Bio-Rad), gels and membranes were equilibrated for 15 min in transfer buffer (25 mM Tris, 192 mM glycine, 10% methanol). Overnight transfers were conducted using a Transphor (TE 52) tank blotting unit (Hoefer Scientific Instruments, San Francisco, CA), using the above buffer (4°C), and the electrical conditions were 30 V, 200 mA. Membranes were transferred to glass Pyrex dishes and extensively rinsed with numerous changes of distilled water followed by drying under a stream of N 2 and then a final dehydration step in a vacuum desiccator under house vacuum. All glassware used in this process and subsequent digestion protocols were thoroughly cleaned employing sonication in dilute Alconox (Alconox, Inc., New York, NY), and then 0.5 N HCl, distilled water rinses following each procedure coupled with a terminal methanol (HPLC grade) wash. The dried membrane was placed in an autoradiography cassette FBXC and covered with a BioMax TranScreen-LE intensifying screen containing BioMax MS film (Eastman Kodak). Cassettes were stored at Ϫ80°C until the film was developed. After film development, proteins were matched to autoradiographic film spots by staining the membranes for 60 s with 0.005% w/v sulforhodamine B (laser grade, Aldrich) in 30% v/v aqueous methanol containing 0.1% acetic acid with gentle agitation, followed by a 30-s water rinse (36). Proteins containing 14 C-pyrrole adducts were excised for tryptic digestion.
Tryptic Digestion for Peptide Mass Fingerprinting-Sections of membrane containing protein were added to microreaction vessels (0.3 ml) fitted with PTFE-lined caps (Supelco, Bellefonte, PA) and washed with 300 l of distilled water. Protein spots were then covered with 7 l of 50 mM ammonium bicarbonate containing 1% n-octyl-␤-D-glucopyranoside (Calbiochem) and 280 ng of sequencing grade-modified trypsin (Promega, Madison, WI). Vials were sealed and placed in a secondary container, which was subsequently immersed overnight in a water bath (28°C). Peptides were extracted from the PVDF matrix by adding 50 l of a 50% ethanol (Aldrich, spectrophotometric grade): 50% formic acid (purity 99%) to the digestion medium and sonicating vials in a Branson ultrasonic cleaner (Shelton, CT) for 30 min. The extract was concentrated to dryness in 100-l glass Accuform Micro-Vials (Kimble-Kontes, Vineland, NJ) using a Savant Speed Vac concentrator (Laboratory Equipment Company, Hayward, CA); samples were stored at Ϫ80°C until they were analyzed by MALDI.
Mass Spectrometry-Molecular masses of tryptic peptides were determined using a Voyager-DE STR MALDI-TOF mass spectrometer (Perseptive Biosystems, Framingham, MA), using a nitrogen laser (337 nm) for ionization. A 10 mg/ml solution of recrystallized ␣-cyano-4hydroxycinnamic acid (Aldrich) matrix was prepared in 50% aqueous acetonitrile containing 0.3% trifluoroacetic acid. Peptide digests were dissolved in 10 l of 50% aqueous acetonitrile containing 0.1% formic acid, and a 0.5-l aliquot was mixed on the sample target with 0.5 l of matrix solution and allowed to dry under ambient conditions. Multipoint mass axis calibration was performed using external standards angiotensin I, ACTH(1-17), ACTH (18 -39), ACTH , and bovine insulin. Initial screening was performed in linear mode, and more accurate monoisotopic peptide masses were determined in reflector mode. To determine the identity of selected spots, MS-Fit and in some cases MS-Tag were used to search data bases for peptide mass fingerprints and to match fragment ions observed in post-source decay (PSD) spectra, respectively (37,38). In addition to the above data bases we have used the ExPASy Proteomics tool ProtScale to accomplish hydropathicity calculations (39). These calculations were used to determine the potential accessibility of cysteine residues to MCTP.
Western Analysis of PDI and Galectin-1-An additional blot and the corresponding autoradiographic analysis were performed for antibody detection of PDI and galectin-1. PVDF membranes were stained with sulforhodamine B, and the proteins were matched with autoradiographic spots. The location of spots previously identified by peptide mass fingerprinting to correspond to galectin-1 and PDI were marked by recording their horizontal and vertical position prior to destaining with 100% methanol and a final rinse with 70% acetonitrile. Blots were sectioned in half, and one portion developed for galectin-1 and the other PDI. Membranes were first blocked with 3% nonfat milk (Bio-Rad) in 150 mM NaCl, 50 mM Tris buffer (TBS), pH 7.4, for 30 min at ambient conditions. This was followed by an overnight incubation with primary antibodies at 4°C. The primary antibodies were polyclonal rabbitraised against rat galectin-1 (generously provided by D. N. W. Cooper from UCSF), used at a concentration of 100 l of serum/20 ml of 3% milk TBS and a mouse monoclonal raised against rat protein-disulfide isomerase synthetic peptide (amino acids 499 -509) with a working concentration of 2 g/ml. The latter was obtained from StressGen Biotechnologies Corp., Victoria, British Columbia, Canada and is known to cross-react with human PDI and calreticulin. After the overnight incubation the blots were washed four times with TBS, and the secondary antibody either goat anti-rabbit or anti-mouse conjugated with alkaline phosphatase (Bio-Rad) was applied at a 1/2000 dilution in 3% milk TBS. Incubations were carried out for 1.5 h at ambient conditions, and the blots were subsequently washed with 0.05% Tween-20 in TBS (two times) followed by four washes with TBS. Blots were devel-oped for 10 min with alkaline phosphatase conjugation substrate kit (Bio-Rad). Fig. 1 show that a selective number of proteins form covalent adducts with MCTP. This pattern was consistent from separation to separation, with six separate autoradiographic profiles showing the identical pattern of labeling. Of the 13 labeled spots, seven were chosen for analysis based on the amount of radioactivity associated with them and the intensity of the corresponding sulforhodamine B stain. The latter was used to pick protein spots that were in sufficient quantity and purity to have a reasonable chance of producing an unambiguous match with protein data bases. Of these seven spots, five were identified (Table I) as probable protein-disulfide isomerase precursor ER-60 (EC 5.3.4.1, Swiss-Prot P30101), protein-disulfide isomerase precursor (PDI, EC 5.3.4.1, Swiss-Prot P07237), ␤or ␥-cytoplasmic actin (Swiss-Prot P02570 and P02571, respectively), cytoskeletal tropomyosin (TM30-NM, Swiss-Prot P12324), and galectin-1 (Swiss-Prot P09382). As recorded in Table I, both the apparent molecular weights and estimated pIs from the two-dimensional separations were in close agreement to data base values with the actual tryptic peptide masses showing excellent coverage and agreement with the expected theoretical values. Typical MALDI mass spectra for tryptic digests of galectin-1 (linear mode) and PDI (reflector mode) are shown in Figs. 2 and 3, respectively. PSD spectra were generated to confirm assignments of tryptic fragments from each of these two proteins. The PSD ions described here follow the nomenclature of Biemann (40); fragmentation designation is followed in parenthesis by the m/z value. For galectin-1 the peptide sequence DSNNLCLHFNPR was chosen. The PSD ions showed the expected amino acid immonium ions and numerous other expected fragments. Some of the more prominent ions were the C-terminal ions y 2 (272), y 2 -NH 3 (255), and y 11 (1372); N-terminal ions b 4 (431), b 6 -H 2 O (686), and c 5 (561); and internal fragment ions NLCL (b ion, 501) and CLHFN (b ion, 672). For PDI the peptide ILFIFIDSDHTDNQR was subjected to PSD. Fragmentation for this peptide was not as informative as that observed for the galectin-1 peptide because of the complexity of the PSD spectrum, but the immonium ions for histidine, isoleucine/leucine, and phenylalanine were present. Some of the more prominent ions included m/z 1790, which represented the MH ϩ -45 (loss of the threonine side chain), y 3 , 1190), and FIFIDSDHTD (a ion, 1164). Spots three and five produced no meaningful matches considering apparent molecular weight, estimated pI, or the peptide mass fingerprint. There are also a number of discrete radiolabeled spots that travel with the dye front. Because the number of known metabolites and break down products of MCTP (24,41,42) cannot account for this heavily labeled area, we suspect that these are proteolysis products derived from adducted proteins.

Two-dimensional Gel Electrophoresis, Autoradiography, and Peptide Mass Fingerprinting-Results recorded in
Antibody Recognition of Galectin-1 and PDI-Because these are two of the more interesting proteins found to be adducted by pyrroles, we determined if for future experiments they could be more simply identified through the use of antibody techniques. Results are recorded in Fig. 4. The commercially available antibody to PDI was found to react with the protein previously identified using MALDI. The antibody also reacted with another protein, which had an estimated pI and molecular weight that corresponded to calreticulin, a protein specified by the manufacture to also react with this antibody. This spot has been tentatively identified as such in Fig. 4. The polyclonal rabbit antibody raised against rat lung galectin-1 was found to also cross-react with human endothelial cell galectin-1 previously identified by MALDI. The antibody did not cross-react with any other proteins contained within the section of the membrane designated for testing. DISCUSSION Autoradiographic analysis showed that MCTP forms covalent adducts with specific proteins. Of these, we identified the following abundant endothelial proteins as targets for pyrrole adduct formation: galectin-1 (Swiss-Prot P09382), PDI precursor (EC 5.   1. A, two-dimensional protein gel transferred to a PVDF membrane and subsequently stained with sulforhodamine B. Proteins were derived from human pulmonary endothelial cells previously exposed to 105 M 14 C-MCTP for 24 h. The numbers on the figure correspond to radiolabeled regions excised for analysis by MALDI; excised portions take on a squared appearance. The following numbers correspond with the following proteins: 1, PDI; 2, ER60; 4, ␤or ␥-cytoplasmic actin; 6, cytoskeletal tropomyosin; 7, galectin-1. Library searches performed for the peptide mass fingerprints of spots 3 and 5 resulted in the absence of definitive matches. B represents the corresponding autoradiograph of the membrane done prior to the staining with sulforhodamine B and the subsequent excising of spots for tryptic digestion and MALDI analysis. Note that images have been inverted (white spots on a black background) to enhance spot visualization.   Gel images in Fig. 1, A and B, confirm this selectivity; numerous abundant proteins, in and around the 66-kDa zone, are devoid or show nondiscrete traces of associated radiolabel. This is readily observable in the abundant protein just above spot 3. The lower left quadrant, a portion of the blot containing proteins closer to an overall neutral pI, contains numerous protein spots; however, they are not associated with detectable levels of 14 C. Other prominent examples of selective adduct formation can be seen in the area of spot 6. This area contains many sulforhodamine B-stained spots, yet only spot 6 shows distinct and reproducible evidence of radiolabel incorporation. Among the seven spots that we selected for identification, it is also apparent that the degree of labeling does not correspond with the amount of protein present. Gels stained with silver (not shown) or blots developed with sulforhodamine B show that spot 5 is an abundant protein, which is qualitatively equal in content to spots 1-4, 6, and 7. However, the degree of label associated with spot 5 appears to be less when compared with these spots. Of the protein spots that consistently contain 14 C, the two unmarked spots (not identified) near spot 7 also show intense radiolabeling and are not as abundant as the proteins we have identified or attempted to characterize. The degree of selectivity observed is not surprising considering earlier work. Dehydroretronecine, a pyrrole formed from the reaction of water with MCTP, could alkylate, in decreasing order of reactivity, cysteine, tryptophan, and histidine. Adduct formation with glutathione was more pronounced than with cysteine alone, indicating that the environment of a thiol group can increase its reactivity to pyrrole (21). Adduct formation would thus be more favorable for proteins containing a higher percentage of accessible cysteine residues and a microenvironment that aids in thiol activation. Such selectivity has been observed for a number of electrophiles, one of the best studied being the arylation of proteins by acetaminophen-derived electrophile, N-acetyl-p-benzoquinone imine (43). The proteins identified in this study serve functions that are potentially important in the maintenance of the endothelial cell barrier. Alterations in these functions could be a mechanistic link between MCTP treatment and the progression of vascular remodeling evident in the lung. However this statement is tempered by the fact that for most compounds attempts have failed to definitely link the event of covalent adduct formation with cellular proteins to a toxic event observed in a particular organ. Even for well studied compounds such as acetaminophen this link is casual in that covalent binding is merely seen to parallel the degree of hepatotoxicity (44).
Additional peaks that could be identified as adducted tryptic peptides were not observed in the MALDI mass spectra. Recoveries of pyrrole adducts were expected to be low owing to the lability of the pyrrole adducts during work up in 50% formic acid solutions or because of incomplete digestion of proteincontaining intramolecular pyrrole cross-links. We have previously shown 2 that pyrrole-glutathione conjugates and pyrrole adduct peptides of glutathione transferases are not stable to  separations on silica-based reverse phase HPLC supports using trifluoroacetic acid in the mobile phase. Previous examination of adducts formed from the reaction of MCTP with synthetic peptides have revealed that the amino acids cysteine, tryptophan, and histidine are the most likely targets for pyrrole adduct formation (21,30). Because these groups are present in many proteins, accessibility of side chain nucleophiles and local microenvironments that enhance side chain nucleophilicity are expected to form the basis of selectivity in adduct formation. Galectin-1 contains six cysteine residues (4.5% of total amino acids), all of which are in the reduced state (45). Hydropathicity calculations have indicated that at least 50% of these cysteines are potentially exposed to the aqueous environment. Both PDI and ER60 have two thioredoxin domains/protein molecule that are redox active (46) and would provide reduced cysteines to bind with pyrroles. Another interesting facet of both PDI and ER60 is the presence of an acidic group that is positioned close to the cysteines involved in the catalytic center of these proteins. Chivers and Raines (47) have proposed that aspartate or glutamate participate in a general acid/base catalysis scheme at the active site (Cys-Xaa-Xaa-Cys) involved in the thiol:disulfide oxidoreduction carried out by Escherichia coli thioredoxin. These authors pointed out that this feature is also present in human PDI. Examination of ER60 also shows a similar sequence of amino acids and the presence of glutamate. For example E. coli thioredoxin (Swiss-Prot no. P00274) has the sequence DFWAEWCGPCKM (residues 26 -37), human PDI precursor (Swiss-Prot no. P07237) is EFYAPWCGHCKA, (residues 47-58), and ER60 (Swiss-Prot # P30101) is EFFAPWCGHCKR (residues 51-62). The importance of the proximity of the glutamate residue to the thiol group stems from the potential for the carboxylate to enhance thiol reactivity via general base catalysis. Because the reactivity of MCTP or dehydroretronecine toward protein nucleophiles involves first the release of the alkaloid ester or hydroxyl groups (respectively) producing a carbonium ion the presence of acidic groups could increase the alkylating potential of pyrroles (48). Protons donated by glutamic or aspartic acid drive this release under the general mechanism of acid catalysis. Therefore, proton donors near cysteine would be expected to  Table I Table I. increase their susceptibility to alkylation by pyrroles through an S N 1 mechanism. Based upon nucleic acid sequencing, PDI has six cysteine residues in its mature form. The two cysteine residues that are not in thioredoxin-like sites are reactive to iodoacetamide (49) but only under denaturing conditions; this would leave only the hyperreactive Cys-Xaa-Xaa-Cys motifs for MCTP to target. Both ␤-actin cytoplasmic 1 and cytoskeletal tropomyosin contain 1.6% cysteine, and both also have acidic pIs, 5.29 and 4.75, respectively. Hydropathicity calculations indicate that the cysteine residues of cytoskeletal tropomyosin at positions 154, 226, and 233 are more negative than the ␤-actin cytoplasmic 1 residues at 257 and 272, which are suspected to be in contact with the aqueous environment. From the autoradiograph it also appears that tropomyosin is more extensively labeled, which would correlate with its more accessible thiol groups.
The potential biological relevance of these proteins in an emerging model of MCTP-induced pulmonary hypertension will be discussed with an emphasis being placed on exploring the known functions of these proteins and the observed pulmonary physiological alterations, which typify the disease.
Galectin-1 is a lectin or carbohydrate-binding protein, which in its dimeric state possesses two galactoside binding sites (50). The dimeric state can thus participate in both intramolecular and intermolecular cross-linkings through the interaction of more than one sugar residue (50). Intermolecular cross-linking capabilities have been used as an assay for galectin-1 activity by exploiting its agglutination of erythrocytes (45). The proposed biological functions of galectin-1 include adhesion in cell-cellandcell-extracellularmatrixinteractions,concentrationdependent induction and inhibition of cell proliferation, and the ability to induce apoptosis (50). In rat lung, the ␤-galactoside-binding proteins are expressed in smooth muscle cells, type I alveolar epithelial cells, and are concentrated extracellularly in elastic fibers of the pulmonary parenchyma and blood vessels (51,52). Human aortic and umbilical vein endothelial cells have also been shown to express galectin-1 in vitro (53). In human aortic endothelial cells, the bulk of galectin is located within the interior of the cell with ϳ5% on the surface. Stimulation of human aortic endothelial cells with minimally oxidized low density lipoprotein resulted in a 47-79% increase in galectin on the cellular surface (53). Similar findings were reported for human umbilical vein endothelial cells. Both smooth muscle cells and pulmonary artery endothelial cells have been observed to undergo cellular proliferation when exposed to various lectins including galectin (54). It has been suggested that galectin-1 could play a significant role in stimulating smooth muscle growth in developing alveolar wall vessels and the development of pulmonary capillaries (54). Mitogenic versus sugar binding activity may be related to the state of galectin sulfhydryl groups, because formation of two intramolecular disulfide bonds caused the protein to act as a mitogen but lacks sugar binding properties (55). The protein may exist in two different states; as a monomer it has transforming growth factor qualities, and as a dimer it acts as a lectin (55). MCTP binding could either result in intramolecular or intermolecular cross-linking of sulfhydryls with functional consequences for the sugar binding or growth factor activities.
The conversion of galectin-1 to a mitogen could explain some of the observed pulmonary changes following the in vivo administration of MCT. Feeding of C. spectabilis seeds to rats results in a significant proliferative response in endothelial cells, fibroblasts, and smooth muscle cells as measured by the uptake of [ 3 H]thymidine (8). Galectin-1 has also been shown to form associations with extracellular matrix (ECM) proteins such as laminin and fibronectin (50) and promote cell adhesion to the ECM. Its interaction with ECM proteins has been implicated in the modulation of the spread and migration of vascular smooth muscle cells (56). Alteration in the ECM because of changes in galectin-1 may result in a matrix unable to act as a sufficient anchor for endothelial cells resulting in either loss of cells or the microvascular leakage that is observed in the MCT model.
Galectin-1 has also been shown to effect the release of cytotoxin from murine macrophage and human monocytes (57). One of the earliest morphological changes in the pulmonary vasculature of MCT-exposed rats is the accumulation of mononuclear inflammatory cells in the small intraacinar vessels and this is associated with or precedes the accumulation of edema fluid and inflammation (58).
Galectins have been implicated as participants in cell adhesion, cell growth, immunomodulation, inflammation, embryogenesis, apoptosis, pre-mRNA splicing, and metastasis. Although there are numerous biochemical and molecular studies relative to galectins, their in vivo role is basically unknown (59); however, the properties of galectin-1 make it an attractive and never before considered participant in the pathology induced by MCT treatment.
PDI is located in the lumen of the endoplasmic reticulum at concentrations approaching mM levels (49) and is responsible for the insertion of disulfides into folding proteins as well as correcting errors in disulfide formation (60). This process is achieved by a mechanism of thiol/disulfide exchange (60). PDI has two active sites each containing two cysteines separated by glycine and histidine (61)(62)(63). PDI also has been shown to act as a chaperone that binds to unfolded proteins and prevents their aggregation with other proteins (64 -67). It acts as a subunit for prolyl 4-hydroxylase as well as microsomal triglyceride transfer protein (68). Because of the numerous functions FIG. 4. Two-dimensional gel of human endothelial proteins transferred to PVDF membrane and subsequently reacted with the primary antibodies to PDI and galectin-1. Regions located by the use of a secondary antibody coupled to alkaline phosphatase. Prior to localizing proteins through antibody detection, the blot was stained with sulforhodamine B, spots containing 14 C (determined autoradiographically) and previously known to be galectin-1 and PDI by MALDI peptide mass fingerprinting were marked (still visible) for both their horizontal and vertical positions on the edges of the membrane. The majority of sulforhodamine dye was removed prior to antibody analysis by washing with methanol followed by 70% acetonitrile. In the higher mass region of the blot containing PDI, an additional spot is visible that has been tentatively identified as calreticulin according to its apparent molecular weight, pI, and the manufacture's designation of cross-reactivity. of PDI, alteration or loss of activity because of alkylation or indirect perturbation in the redox environment by pyrroles could rapidly endanger the homeostasis of the endothelial cell.
Peptide mapping of spot 4 indicated the presence of either the ␤or ␥-form of cytoskeletal actin but could not distinguish between the two. Previously we have shown, using SDS-polyacrylamide gel electrophoresis, Western blotting, and antibodies, that ␤-actin was a potential target for pyrrole adduct formation (69). Actin is important in the maintenance of the endothelial permeability barrier. Its contractile interactions with myosin regulates the endothelial permeability barrier and serves as an effector for inflammatory and procoagulant-induced vascular leak. At cell-cell adherens junctions, actin is linked to the plasma membrane through interactions with ␣-actinin, which associates with vinculin; vinculin is linked to catenins and plakoglobin, which both interact with membraneassociated cadherins. Cadherins in turn associate with cadherins from other cells through their extracellular domains. In the subendothelial matrix, fibronectin or vitronectin are bound to integrins, which bridge the plasma membrane and in turn bind through talin to vinculin, which is linked to ␣-actinin thus to actin filaments (70,71). Alterations in actin or other proteins involved in the normal functioning of the endothelial barrier could result in extravascular leakage of plasma proteins and fibrin, which could stimulate vascular wall remodeling.
Studies with nonmuscle isoforms of cytoskeletal tropomyosin have shown it to be important in the process of actin filament formation and structure and in the defining of domains along the actin filament (72). Tropomyosin can block the bundling of actin by villin (73), the interaction of actin with filamin and ␣-actinin (74,75), and inhibits fragmentation by gelsolin (76). In budding yeast, the loss of the single copy gene for tropomyosin results in the disappearance of cytoplasmic actin cables and reduced cell growth (77). These functions make tropomyosin essential for normal cell division, locomotion, and shape changes (72). This protein, like the actin it influences, could be essential for preservation of the osmotic barrier provided by functional endothelial cells and, along with DNA adducts, could potentially explain some of the observed perturbations observed in the cell cycle following MCTP treatment. Using methodologies similar to those employed in this study, acetaminophen has recently been shown to form adducts with the hepatic derived cytoskeletal tropomyosin (37).
The limited number of proteins that are bound by MCTP suggests particular motifs that may be important for the binding of pyrroles and thus narrows the search for other proteins that may be important to the toxicology of MCT. Only a limited number of proteins possess thioredoxin sites Cys-Xaa-Xaa-Cys (78). A few of these that were not observed in this study, which are present in endothelial cells, are fibronectin, thioredoxin, von Willebrand factor, and human RNA polymerase. Fibronectin, which has PDI activity, is an important component of the ECM and is necessary for endothelial integrity (79). The von Willebrand factor is produced by endothelial cells and plays an important role in platelet adherence to the interstitial matrix following endothelial damage (80). One of the manifestations of primary PH is the evidence of circulating platelet aggregations (11) and the presence of in situ thrombosis in pulmonary lesions (12,13). Human RNA polymerase contains three Cys-Xaa-Xaa-Cys regions with one of these groups falling within a zinc finger. Perturbation in the transcription of DNA into RNA would have obvious consequences. Some of the 14 C spots that remain to be identified could contain proteins with the Cys-Xaa-Xaa-Cys motif or accessible reactive cysteine(s). Additional proteins may be identified using different pH gradients and blotting conditions. The technique employed in this paper probably missed many of the higher molecular weight proteins because of the soaking technique employed and the gel composition of commercially available IGP strips (81,82).
In the search for important molecular target(s) for MCTP, it should be considered that only small quantities of electrophile (MCTP) are essential to elicit pulmonary insult. Doses as low as 1 mg/kg, administered intravenously in N,N-dimethylformamide, were found to affect pulmonary function in rats (83). With a half-life in aqueous solutions of ϳ3 s (6), a significant portion of the dose would also be expected to be lost before exacting an effect. Therefore proteins that are abundant and structural in nature may not receive enough adduct to significantly alter cellular function. However, as in the case of actin, adducts to actin alone may not present problems to the cell but actin cross-links to DNA may have more dire consequences (84). Proteins that are found in low levels or function in enzymatic or signaling processes are potentially better candidates. Alterations in a small percentage of the PDI pool may be amplified if adducted PDI can still participate in enzymatic processes resulting in the incorrect formation of disulfides or fails to act as a proper chaperone causing the aggregation of unfolded proteins. ER60 has previously been found to be associated with the internal nuclear matrix and participates in the anchorage of the DNA loops at the matrix (85). Such matrix DNA interactions have been shown to be important in the control of gene expression (86). Adducts that result in hypersensitivity reactions may also be important. PDI has been implicated in halothane hepatitis along with other adducted proteins (87)(88)(89). Autoantibodies to PDI were detected in rats exposed to D-galactosamine and acetaminophen or carbon tetrachloride when the latter two were combined with diethylmaleate (90). A similar mechanism may be relevant to MCTP toxicosis.
The proteins observed to form covalent adducts with MCTP and their resulting partial or total loss of function could be important elements in the puzzling model MCT-induced PH. Further exploration of these proteins and their role in both chemically precipitated or natural occurring PH warrant future investigation.