Sequence of Rat Liver a2-Macroglobulin and Acute Phase Control of Its Messenger RNA*

Six a,-macroglobulin cDNA clones were isolated from two liver cDNA libraries produced from rats undergoing acute inflammation. The coding sequence for rat a,-macroglobulin including its 27-residue sig- nal peptide and the 3‘- and part of the 5‘ nontranslated regions were determined. The mature protein consist- ing of 1445 amino acids is coded for by a 4790 f 40 nucleotide messenger RNA. It contains a typical inter- nal thiol ester region and 25 cysteine residues which are conserved between rat and human a,-macroglobu- lin. Although the amino acid sequences of rat and human a,-macroglobulin share 73% identity, two small divergent areas of 17 and 38 residues were found, corresponding to 29 and 11% identity, respectively. These areas are located in the bait region and, therefore, may confer specific proteinase recognition capa- bilities on rat a2-macroglobulin. control values and at h. h the levels had decreased to less than 30% the maximum value. Transcription rates from the a2-macro- globulin gene as measured run-on

hepatocytes, can be induced in rats up to several hundredfold in 24 to 48 h following intramuscular or subcutaneous injections of turpentine (4-7), or intraperitoneal injection of either complete Freund's adjuvant (8) or sterile barium sulfate (9). a2M is a member of the thiol ester protein family, comprising complement components C3 and C4 and the related amacroglobulins (10-12). Rat al-macroglobulin (alM) (4, 8, 13), rat a,-inhibitor I11 ((~'13) (8, 14-17), human pregnancy zone protein (PZP) (11,18,191, and probably the human pregnancy-associated plasma protein A (20, 21) also belong to this family. The a-macroglobulins have related primary structures and many of them share a characterisitic internal activatable 6-cysteinyl-y-glutamyl thiol ester bond (8,22). In addition, rat and human a-macroglobulins possess a bait region, a limited amino acid sequence in the central portion of the polypeptide chain, which exposes a number of potential target sites for a spectrum of small proteinases. Hydrolysis of one of these substrate bonds causes the activation and cleavage of the thiol ester bond resulting in the covalent attachment of the proteinase to the thiol ester group of the inhibitor (23-26). Subsequently, macrophages and other phagocytes, possessing azM receptors on their surface, clear the proteinase-inhibitor complexes from the circulation through receptor-mediated endocytosis (27, 28). The main physiological role of the a-macroglobulins is believed to be that of a scavenger for small proteinases, mediated through the thiol ester trap mechanism.
Rat a2M is a major acute phase reagent, while rat alM only increases by a factor of 2 or less (4,8) and rat al13 is a negative acute phase reagent (8, 14-17). By contrast, human azM is not an acute phase protein, its plasma concentrations remaining constant (29); the reason for this species-specific difference is not clear. We are interested in the molecular mechanism of rat a2M induction and in a comparison with the control mechanisms for rat alM, rat 01'13, and human a2M. Our working hypothesis was that the strong increase in rat aZM protein concentration following inflammation was caused by a proportional increase in cytoplasmic azM mRNA concentration and that this, in turn, was due to corresponding changes in the rate of aZM transcription.
In initial studies Northemann and colleagues (30) were able to demonstrate a 66-fold increase in translatable liver azM mRNA by cell-free translation and immunoprecipitation, following injection of turpentine. Since this increase was substantially smaller than the reported several hundredfold increase of azM plasma protein concentrations, we wished to determine the change in a2M mRNA levels by nucleic acid hybridization. To obtain the necessary hybridization probes, we constructed cDNA libraries from acute phase rat liver mRNA and then isolated azM cDNA clones (31). Here we describe the isolation and sequence analysis of cDNA clones containing the entire protein coding sequence, a comparison of this sequence with its human counterpart, and its use in time course studies to determine the acute phase changes in both a2M mRNA levels and nuclear a2M transcription rates.

EXPERIMENTAL PROCEDURES
Animals and Materials-Male Fisher 344 rats were from Simonsen Laboratories, Gilroy, CA. Laboratory chemicals and standard enzymes for molecular biology were from Baker, Mallinckrodt, Fisher, Sigma, New England Biolabs, Bethesda Research Laboratories, and Boehringer Mannheim, as described (32). Freund's adjuvant was from Gibco Laboratories. Radionucleotides were from Amersham Corp. and ICN.
Experimental Inflammation and mRNA Extraction-Experimental inflammation was induced by intraperitoneal injection of 0.35 ml of complete Freund's adjuvant into 10-to 20-week-old male Fisher 344 rats, weighing about 250 g, according to Lonberg-Holm et al. (8). Total RNA was prepared from livers excised at different times by the guanidine hydrochloride method (33,34), modified as follows. Five g of minced liver was homogenized in 100 ml of 7 M guanidine hydrochloride, 20 mM sodium acetate, 1 mM dithiothreitol, and 0.5% sarcosyl, pH 5.0. After a clearance centrifugation (30 min at 10,000 X g, -10 "C), nucleic acids were precipitated with one-half volume of ethanol and collected by centrifugation. The RNA was dissolved in 8 ml of buffer A, containing 8 M guanidine hydrochloride, 20 mM sodium acetate, 20 mM EDTA, 5 mM dithiothreitol, and 0.5% sarcosyl at pH 7.0, and reprecipitated with one-half volume of ethanol. The resuspension and reprecipitation cycle was repeated twice with 4 and 3 ml of buffer A, respectively. The RNA was finally dissolved in 10 ml of 20 mM EDTA, containing 0.5% sarcosyl at pH 7.0, extracted with an equal volume of a 2 4 1 mixture of chloroform and isoamyl alcohol, and reprecipitated with 2% volumes of ethanol overnight. The RNA was then dissolved in 5 ml of 10 mM Tris-HC1 at pH 7.4, 1 mM EDTA, and 0.2% sodium dodecyl sulfate (SDS). Polyadenylated mRNA (poly(A)+ mRNA) was then prepared by retention on o1igofdT)-cellulose (32). To enrich a2M mRNA by size, sedimentation was performed in a 5 to 25% linear sucrose gradient in 0.1 M sodium chloride, 10 mM Tris-HC1, pH 7.4, 2 mM EDTA, and 0.5% SDS for 4% h at 40,000 rpm in a Beckman Spinco SW4OTi rotor at 26 "C. RNA from individual gradient fractions was coupled to a nitrocellulose membrane (Schleicher & Schuell, BA85), and fractions containing a,M mRNA were identified by hybridization with nick-translated (32) rat a2M cDNA (pRLA2M/4B, see below).
Preparation and Screening of cDNA Libraries-One cDNA library was prepared from non-size-selected total poly(A)+ mRNA from acutely inflamed adult male rats, according to Okayama and Berg (35,36), and another from the >3,000-nucleotide RNA size fraction obtained by sucrose gradient centrifugation. One pg of total poly(A)+ mRNA gave rise to approximately 100,000 independently transformed Escherichia coli colonies containing cDNA plasmids. The size-enriched library contained approximately 70,000 independently transformed E. coli colonies. The non-size-selected cDNA library was screened with nick-translated human azM cDNA probes. One of the resulting clones (pRLA2M/4B) was sequenced, and its cDNA insert fragments were nick-translated and used to rescreen the sucrose gradients and the size-enriched cDNA library (Fig. 1, Appendix).
DNA Sequencing and Computer-assisted Data Analysis--Random subfragments of cDNA generated by sonication and specific fragments generated by restriction enzyme digestion were cloned into M13 phage vectors by standard procedures (37). DNA sequencing was performed using the dideoxy sequencing technique (38,39). DNA sequence data were collected and aligned using Staden's software (40) and analyzed with the University of Wisconsin Genetics Computer Group programs (41) on a Digital VAX 11/750 computer.
Northern Blot Analysis-For each of the time points of the acute phase response, 5 pg of poly(A)+ mRNA was denatured by heating at 55 "c for 15 min in 50% formamide, 2.2 M formaldehyde, 20 mM MOPS, 5 mM sodium acetate, 0.5 mM EDTA, pH 7.0, and separated by electrophoresis in a 1.4% agarose gel containing 0.74% formaldehyde in MOPS buffer (40 mM MOPS, pH 7.0,lO mM sodium acetate, 1 mM EDTA). Transfer of RNA from the gel to a nylon membrane (Pall, Biodyne A) was carried out by electroblotting using the procedure described by the manufacturer (Bio-Rad, Transblot Cell, manual). After transfer, the wet membrane was wrapped in cling film (Saran Wrap) and placed, with the RNA side facing up, in an ultraviolet light irradiation chamber containing four General Electric . (G 15 T8) 15-watt germicidal tubes. The tubes were at a distance of 35 cm above the membrane. The RNA was fixed on the membrane by 3 min of irradiation (43). Three probes were used to identify and quantitate mRNA on Northern blots. Two were gel-eluted cDNA fragments corresponding to rat serum albumin (RSA) (44) and al-acid glycoprotein (AGP) (45). These fragments were nick-translated to specific activities of 5 X lo7 to 2 X 1 0 8 cpm/pg. The probe for a2M was a mixture of twcr synthetic oligonucleotides, representing cDNA sequences, which were labeled with polynucleotide kinase and [y-32P]ATP to specific activities in excess of 4 X 108 cpm/pg. Pretreatment and hybridization reactions were performed in 50 mM PIPES buffer, pH 6.8, containing 100 mM sodium chloride, 50 mM sodium phosphate, 1 mM EDTA, and 5% SDS (42). Following pretreatment for 30 min at 55 "C in 20 ml of liquid, the nylon filters were hybridized overnight at 55 ' C with 5 X 10' cpm/ml of the a2M probe in 20 ml of liquid sealed in a bag and then washed four times in hybridization buffer at 55 'C for 15 min. After autoradiography, the quantities of specific RNA species were determined by excision of the hybridizing areas from the filters and liquid scintillation counting. The same procedure was used at 65 "C for the RSA and AGP probes at 2.5 X 106 cpm/ml, with additional wash steps at higher stringency where needed.
RNase H Experiments-These experiments were performed as previously described (46). Briefly, 5 pg of poly(A)+ mRNA and 500 ng of synthetic oligonucleotide were annealed for 10 min at 65 "C in 80% formamide and for an additional 90 min at 50 "C in 50% formamide and PIPES buffer (46). The nucleic acids were ethanolprecipitated, dried, redissolved in RNase H buffer (46), and incubated with 3 units of RNase H for 30 min at 37 "C. After phenol extraction and ethanol precipitation, the RNase H-resistant RNA was separated by electrophoresis in urea-acrylamide gels, transferred to nylon membranes, and visualized by Northern blot hybridization with nicktranslated cDNA probes.
Nuclear Run-on Transcription Assays-Nuclei were prepared as described (47,48), resuspended in 40% glycerol, 50 mM Tris-HC1, pH 8.3,5 mM magnesium chloride, and 0.1 mM EDTA at a concentration of lo6 nucleilpl, and stored at -70 "C. Elongation reactions were performed as described (49,50). Briefly, 2 X lo7 nuclei were incubated with 480 pl of elongation mixture containing [w3'P]UTP (49) for 45 min at 26 "C. The reaction mixture was then treated with RNase-free DNase (51) for 10 min at 37 "C and proteinase K for 30 min at room temperature. After two extractions with phenol/chloroform and chloroform, the nucleic acids were precipitated with 10% trichloroacetic acid and collected on nitrocellulose filters. The filters were treated once more with RNase-free DNase and subsequently with proteinase K. The RNA was then eluted from the filters in four steps with elution buffer (1% SDS, 1 mM EDTA, 10 mM Tris-HC1, pH 7.4, and 10 pg/ml of carrier tRNA) for 10 min each at 67 "C. Four pg of each linearized plasmid DNA, containing cDNA inserts, was coupled to 4 X 4-mm nitrocellulose filters (52). Prehybridization, hybridization with the nuclear run-on reaction product, RNase treatment, and washes were as described (47,49,50). Typically, 10 million Cerenkov cpm of nuclear run-on reaction product were used in each hybridization reaction. In control experiments, a-amanitin (Sigma) was used at 0.4 pg/ml (53). The y fibrinogen plasmid used was prA-Fib (54), containing a 1580-base pair insert.

RESULTS
Two cDNA libraries from acute phase male rat liver mRNA were constructed in the pcDVl plasmid vector. Screening of the first library, prepared from total polyadenylated RNA, by cross-hybridization with human a2M cDNA probes (55) resulted in several isolates. One of these, pRLA2M/4B, (Fig. 1) was further characterized and sequenced. Nick-translated cDNA from this clone was used to screen the second cDNA library prepared from size-enriched mRNA (see "Experimental Procedures"). Of the resulting isolates, the one containing the longest insert, pRLA2M/102 (Fig. I), was sequenced. It contained most of the protein coding sequence, starting from nucleotide 287 of the final sequence. To isolate clones representing the missing sequences, an oligonucleotide (P2) (Fig.  1) was synthesized. Rescreening of the library with the oligonucleotide produced 35 isolates. The four most intensely hybridizing clones (pRLA2M/5, pRLA2M/22, pRLA2M/26,

Structure and
Regulation of Rat az-Macroglobulin mRNA and pRLA2M/29) were analyzed further. The 700-base pair fragment extending from a BamHI site in the vector to the BamHI site in the cDNA insert at position 576 ( Fig. 1) from all four of these clones was subcloned into M13 mp8 and sequenced. All four sequences were identical and encoded azM. The complete coding sequence for the aZM polypeptide was deduced as summarized in Fig. 1. The sequence also includes the 3' nontranslated region, with the polyadenylation signal AATAAA at a distance of 17 nucleotides upstream of the polyadenylation site, and 63 nucleotides of the 5' nontranslated region (Fig. 2). These isolates were identified as aZM clones by comparison of the derived amino acid sequence with independently determined partial amino acid sequences of rat azM (56) (Fig. 3). Fourteen of fifteen residues were identical. The rat signal peptide is 4 amino acids longer than its human counterpart (27 residues as opposed to 23 (55)).
The rat precursor polypeptide pro-azM is 1472 residues long compared to 1474 for human pro-azM, and the mature rat polypeptide has 1445 residues, compared to 1451 for human azM. The calculated mass of the nonglycosylated mature proteins is 160,670 Da for rat azM and 160,798 Da for human azM (Table I).
The sequence CGEQ has been shown to form the thiol ester bond in human azM (22,56). Rat azM is also known to contain one such bond (8) and contains the corresponding sequence in the same position (residues 970-973, Fig. 2).
The extent of sequence conservation between the rat and human proteins was determined over three segments separately and for the entire protein (Table (11). They share 1067/ 1451 identical residues and are, thus 73.5% identical. Both sequences contain 25 cysteine residues in conserved positions. Thus, the disulfide bridge structure of both molecules is likely to be identical as well as the number and arrangement of the interchain disulfide bridges in the native protein dimer. shows the N terminus of mature rat azM determined from the purified protein (56). Amino acids are given in the single-letter code; identical residues are bored. The arrows indicate the signal peptidase cleavage site. The asterisk denotes the only difference between the protein sequence and the sequence predicted from cDNA.  "The amino acid numbering refers to mature human azM as obtained by subtracting 23 residues (for the signal peptide) from published numbering for pro-azM (55). The comparison was perprogram BESTFIT (41).  (57). The odd span length parameter was 25, the score parameter was 300, and the percent score mode was used. Gap I (arrow) occurs at human coordinates 632 to 648, gap 2 at 690 to 727 (see Fig. 5).

1474
puter analysis to emphasize the content in chemically equivalent amino acid sequences and thus the structural relatedness of these proteins beyond the strict amino acid identities (57) (Fig. 4). Homology is continuous throughout the sequences with the exception of the signal peptides and two internal regions. Region 1 comprises residues 632-648, and region 2 residues 690-727 of the human pro-a,M sequence (55). In these areas the extent of sequence identity is reduced to 29% (region I ) and 11% (region 2 ) . The two regions are located near and within the bait region (residues 690-730, approximately, in the human sequence), an area of central importance for the function of these proteins as proteinase inhibitors. The sequence alignment of the bait regions of rat aZM, human azM, and partially that of the related human PZP (11) is given in Fig. 5 . From this alignment and published data (II), it appears that the human PZP, like rat a z M , differs more strongly from human azM in this region than in other parts of the protein.
Both sequences were analyzed for potential N-linked carbohydrate attachment sites (Asn-X-Thr and Asn-X-Ser) (58). Twelve potential sites were found for rat aZM, and eight for human azM, with seven of these sites being conserved (Table  111).
We have determined the length of azM mRNA missing from the 5' end of the cloned cDNA. Size-enriched acute FIG. 5. Alignment of rat and human azM with human PZP in the bait region. The alignment between rat and human a2M was produced using the University of Wisconsin Computer Group Program BESTFIT (41), with the exception of region 632 to 644, which was aligned by hand in order to better accommodate the corresponding PZP sequence. The alignment between human aZM and human PZP is as published (ll), with one alteration. One gap was introduced a t residue 635 to allow for a better alignment of all three molecules in the region 635 to 641. Ra, rat; Hu, human; A2M, a2M. Rat and human pro-a2M numbering was used (55).
* Amino acids in single letter code.

310-
phase and control rat mRNA were separately annealed with the oligonucleotide P2 (Fig. 6). The reaction mixtures were then treated with RNase H, which destroys RNA contained in RNA/DNA hybrids but leaves single-stranded RNA intact. The surviving RNA fragments were then analyzed by Northern blot hybridization with radioactive a2M cDNA probes. From the length of the surviving RNA fragment located to the 5' side of the oligonucleotide (Fig. 6, track I ) , we calculated that the cap site must be located 21 f 5 nucleotides upstream of the sequence represented by cloned cDNA. From the length of the surviving fragment located to the 3' side of the oligonucleotide (data not shown), we have determined the length of the total mRNA to be 4790 f 40 nucleotides. The mean length of the poly(A)+ tail must then be 180 & 30 nucleotides. To test whether the increased a2M protein level during the acute phase response is due to an increased level of a2M mRNA, we measured the latter as a function of time after induction by nucleic acid hybridization with cloned cDNA probes (Fig. 7). Only trace amounts of a2M mRNA were detectable in control rats (not visible in Fig. 7B). We have  compared base-line mRNA levels for five control rats (data not shown). In all animals, low amounts of a2M mRNA were detected, indicating that the gene is normally expressed a t a low constitutive rate. Variation in individual base-line values is probably due to mild inflammatory reactions in some of the animals. In the experiment shown in Fig. 7 two animal livers were pooled for each time point, and the increase in a2M mRNA level in this particular experiment was 214-fold, with a maximum at 18 h after stimulation (Fig. 7 A ) . Subsequently, the a2M mRNA level decreased again and was less than 30% of the maximum value at 24 h. RSA mRNA levels were increased 1.3-fold in the experiment shown in Fig. 7A. AGP mRNA was increased 16.6-fold with a maximum at 24 h. The AGP mRNA level did not show a peak and a rapid decrease over this time period as was observed for a2M, suggesting the expression of these two strongly induced acute phase mRNA species is controlled by partially different mechanisms. In addition to showing a quantitative increase, AGP mRNA was altered qualitatively during the course of the induced response. This was manifest by the appearance of an additional mRNA species, larger by about 100 nucleotides, with an observed maximum size a t 6 h (Fig. 7A). Subsequently, the larger mRNA species became progressively smaller until it approached the size of the species seen a t time 0.
To test whether the increased accumulation of a2M mRNA is due to an increased rate of transcription, we carried out nuclear run-on experiments using nuclei from rat livers excised a t various times after inflammatory stimulation. Purified "run-on" RNA product was quantitatively hybridized with excess filter-bound cDNA specific for a2M, y fibrinogen, and RSA. The counts hybridized to a specific probe are a measure for the transcription rate of the corresponding gene, provided they are 0-amanitin-sensitive (Fig. 8). y Fibrinogen serves as a positive control because the acute phase increase of fibrinogen mRNA is known to be due to an increase in transcriptional rate (59). The rates of fibrinogen transcription were increased 3-to 4-fold (Fig. 8), with a maximum at 15 h after stimulation. The rate of RSA transcription was decreased 2-fold a t 12 h after stimulation and then returned to the initial value, suggesting RSA is a moderate negative acute phase gene. A significant rate of azM transcription was observed in control nuclei. Fifty percent of this activity was CYamanitin-sensitive (data not shown) and, thus, was due to RNA polymerase I1 activity. Early in the inflammatory response the aZM transcription rate was reduced almost 2-fold with a minimum a t 10 h after stimulation and then rose to the initial value (Fig. 8). These experiments were repeated four times with nuclear preparations from different rats. The maximum increase of a2M transcriptional rate observed was 2.3-fold over the base-line value, indicating that a small part of the overall induction of a2M mRNA is due to an increased transcriptional rate.

DISCUSSION
The objective of this study was to produce and characterize rat liver ap-macroglobulin cDNA clones and to use them to determine whether the strong acute phase induction of the protein is due to a comparably strong increase in the transcription rate from this gene.
The identity of the clones as C Y~M clones, as opposed to other members of this protein family, was established by comparison with independently determined amino acid sequences from the N terminus of rat a2M (56). Although there was a discrepancy in 1 of 15 amino acids (Fig. 3), the result was still unambiguous. The cDNA does not correspond to amino acid sequences from the N termini of rat oclM (8) and r a t C X~I~, * .~ b u t confirms a2M cDNA sequences reported earlier from our laboratory and others (31,60,61).
The value of 160,670 Da calculated from cDNA as the mass of the nonglycosylated mature protein (Table I)   protein was reported as 182,000 Da (7), corresponding to an estimated carbohydrate content of 11%. However, SDS-polyacrylamide gel electrophoresis sometimes results in an underestimation of the size of glycoproteins over 150,000 Da (8). From the sums of the masses of proteolytic fragments of rat and human azM, which could be measured with greater precision, Lonberg-Holm et al. (8) have derived so-called model molecular masses for rat and human a2M (Table I). From the difference between the cDNA-derived masses and the model molecular masses, one can estimate the carbohydrate contents of rat and human a2M to be in the order of 9 and 11%, respectively.
Rat and human a2M show extensive sequence homology along the entire length of the proteins with the exception of the signal peptides and two small areas in and near the bait region (Fig. 5 ) . In order to rule out that this could be due to sequencing or cloning artifacts, the bait region coding area was resequenced from an independent cDNA isolate containing an insert of different size. The second sequence confirmed the first one (see Fig. 1 legend). The high degree of sequence conservation between rat and human a2M reflects strong selective pressure to maintain the overall three-dimensional structure of the a2M molecule, which presumably is essential for its function as a proteinase inhibitor. The strict conservation of all 25 cysteine residues reinforces this notion, as do electron microscopic studies of the two proteins (8). Sequence divergence generally reflects the absence of selective pressure because the sequences involved are neither important for structure nor function. In the case of azM, however, sequence divergence occurs in the functionally essential bait region, which contains the recognition sequences for proteinases that are targets for inhibition. This renders sequence drift unlikely and suggests selective pressure to diverge instead. A reasonable selective advantage of sequence variation in the bait region would be the emergence of specialized a-macroglobulins that inhibit specific subsets within the overall spectrum of proteinases. We expect other a-macroglobulins, in particular rat a,M and rat aJ3, to have a similar structure and to also show divergence mainly in the variable areas of the bait region. The limited cDNA sequence data available for other rat a-macroglobulins support this predi~tion.~ A major difference between rat and human azM is that the former is an acute phase protein, while the latter is not and is always present in the circulation at high levels (29,62). Unlike most proteinase inhibitors, which interact directly with the reactive site of a particular subgroup of proteinases, azM forms complexes with numerous proteinases from all four major classes (62). Thus, human a2M may contain in its bait region cleavage sites both for proteinases present under normal conditions and for those induced under acute phase conditions, whereas rat a2M may preferentially react with the latter group of proteinases. Alternative explanations are possible and need to be tested. Human azM also binds lectins, small proteins and peptides, hormones, growth factors, and metal ions through a binding site separate from its proteinase binding site (62). It is not known at present which of these binding functions are shared by rat a2M and which functions provide a rationalization for the acute phase inducibility of rat a2M and the selective advantage of this inducibility.
Three factors influence the measurements of changes in a2M mRNA levels during an acute phase response: the inflammatory stimulus, the detection technique, and the base-line values observed in normal rats. The highest increases of rat a2M plasma protein are obtained after intramuscular and subcutaneous injection of turpentine (4-7), intraperitoneal injection of barium sulfate (9), and the combined administration of catecholamines and glucocorticoids (63). In the present study, intraperitoneal injection of Freund's adjuvant was used, because with this stimulus we encountered fewer problems in consistently recovering intact RNA in comparison with the use of barium sulfate. Increases in mRNA levels between 60and 214-fold were observed with Freund's adjuvant (Fig. 7). In earlier experiments using barium sulfate, increases of 400to 790-fold were observed (31).
A striking feature in the acute phase response of cvzM is the rapid decrease of its plasma concentrations after the maximum has been reached (8) which is preceded by a rapid decrease of a2M mRNA (Fig. 7). A rapid decrease in the amount of translatable aZM mRNA was also found in cellfree translation experiments (30). These results suggest that after 18 h of inflammation azM mRNA may have a half-life of only a few hours, but direct measurements of aZM mRNA half-life have not been reported.
The main result reported here is that the level of aZM mRNA in rat liver is increased up to several hundredfold following acute inflammation, to the same order of magnitude as the aZM level in plasma. Therefore, the amount of a2M protein is mainly controlled through the level of mRNA, and translational or post-translational controls are probably negligible.
The size difference between the normal and acute phase species of rat AGP mRNA of approximately 100 nucleotides was unexpected but reproducibly obtained. According to published data, rats have only one AGP gene (64), implying inflammation-dependent production of two distinct mRNA species from only one gene. The origin of this difference is currently being investigated in greater detail. The different kinetics of mRNA increase for AGP and aZM, the different base-line values, the different magnitude of induction, and the absence of size changes in aZM mRNA suggest that these two genes are regulated by partially different mechanisms during acute inflammation. Nevertheless, both are major acute phase proteins, and glucocorticoids and hepatocyte stimulating factors contribute to the regulation of both genes (64, 65).4 When this project was initiated, we postulated that an increase in the rate of a2M transcription was the major factor in the acute phase regulation of this mRNA. However, rate measurements by nuclear run-on experiments indicate that transcriptional rates vary less than 3-fold in our experiments and less than 6-fold in experiments reported by others (60) and that this increase of azM transcription is much lower than expected on the basis of the amplification in azM mRNA levels detected by hybridization. The simplest interpretation of our data and those of others (60) is that moderate increases in the rate of transcription contribute to the overall induction of azM mRNA but that post-transcriptional regulation plays a major role. This regulation could include both changes in intranuclear RNA processing and/or changes in cytoplasmic mRNA stability. In view of the significant azM transcription rate observed in control rat nuclei and of the low level of mature mRNA under these conditions, rapid turnover of a2M transcripts must occur in nuclei or in the cytoplasm of control rats or both. No experimental evidence for a change in a2M mRNA half-life under acute phase conditions is available. However, many other mRNA species induced by outside signals (hormones, metal ions, and heat shock) show concomitant increases in transcription rate and mRNA stability (67).
To investigate the mechanism of aZM mRNA induction in greater detail and to determine whether changes in intranu-G. Fuller, personal communication.
clear RNA processing are involved, we have isolated the rat a2M gene and have determined some of its structural features. Currently, experiments are in progress to locate the regulatory sequences of this gene and to compare the gene with other members of the rat a-macroglobulin gene family. sequencing strategy. The first clone isolated by cross-hybridization with human azM cDNA was pRLA2M.4B (4B). The 520-bp BamHI fragment containing the 3' end was cloned into the BamHI site of M13 mp8 and sequenced from both ends. The 1636-bp fragment upstream of the BamHI site at coordinate 4116 was self-ligated, sonicated, blunt-ended, and shotgun-cloned in M13 mp8. This 1636-bp fragment was also nick-translated and used to rescreen the library, resulting in clone pRLA2M.102. The 362-bp PuuII fragment (coordinates 2174-2536) and the fragment 287-637 were cloned in M13 and sequenced from both ends; the 1537-bp PuuII fragment (coordinates 637-2174) was shotgun-cloned and sequenced. From the 5' region of this sequence (coordinates 290-328), a synthetic oligonucleotide was prepared (P2) and used to rescreen the library, resulting in the four clones pRLA2M/5, pRLAZM/22, pRLA2M/26, and pRLA2M/29. The 700-bp B'-B fragment (until coordinate 576) was shotgim-cIoned and sequenced and the 777-bp HX fragment (coordinates 1440-2217) of clone pRLA2M.5 was blunt end cloned into M13 mp8 and sequenced from coordinates 2217 to 1950. The small 200-bp fragment containing the BamHI and PuuII sites at coordinates 576 and 637 was cloned and sequenced in both orientations. For the complete sequencing project, 35,880 nucleotides of total sequence data were collected, covering each nucleotide of the cDNA on an average of 7 times and over 90% of the DNA on both strands.