The acute phase response of mouse liver. Genetic analysis of the major acute phase reactants.

We have examined changes in translatable liver mRNAs during the acute phase response in several inbred mouse strains. Induction of mRNAs for the known acute phase reactants serum amyloid A, hemopexin, haptoglobin, and alpha 1-acid glycoprotein was observed. Two alpha 1-acid glycoproteins, termed AGP-1 and AGP-2, differing in size and charge, were found in all mice and appear to derive from distinct yet closely related mRNAs. Polymorphism in the structure of AGP-1 among strains was used to map the structural gene, Agp-1, to chromosome 4 near the b locus. Structural variation for serum amyloid A enabled mapping its structural gene, S alpha alpha, to chromosome 7, near Gpi-1. A wild-derived population of mice, Mus spretus, contains several variations of interest, including structural variations in many acute phase proteins and a putative regulatory variation specific to AGP-1. Further analysis of these genetic variants should provide novel insights into the acute phase response and the factors that mediate it.

A wild-derived population of mice, Mus spretus, contains several variations of interest, including structural variations in many acute phase proteins and a putative regulatory variation specific to AGP-1. Further analysis of these genetic variants should provide novel insights into the acute phase response and the factors that mediate it.
Mammals undergo a complex response to a variety of systemic injuries such as infection or inflammation. For example, injection of bacterial cell wall lipopolysaccharides or other irritants into mice causes immigration of neutrophils and monocytes to the site of injury, mitogenic stimulation and maturation of B-lymphocytes, release of a multitude of active components (interferon, interleukins, and leukotrienes) by mononuclear cells, and a dramatic increase in the levels of a variety of liver-derived serum proteins, termed the acute phase reactants (1). One well characterized acute phase reactant of the mouse, serum amyloid A, is a component of high density lipoproteins and is also the precursor to the major protein component of the amyloid fibrils that are deposited extracellularly in animals with secondary amyloidosis (2). Other prominent acute phase reactants in mice as well as in other mammals include haptoglobin, hemopexin, and a, -acid glycoprotein (3).
Extensive studies at the biochemical and cellular levels, while providing critical information on the nature of the acute phase response, have answered few questions relating to the regulation of the response and the effectors that are involved. A genetic approach offers the potential to identify and locate structural genes encoding the acute phase reactants and regulatory genes governing their expression both prior to and *This work was supported by the United States Public Health Service Grants CA26122 and GM19521. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. following the stress response; such information is crucial to understanding the regulation of this complex process.
In the present paper, we report biochemical and genetic studies on the mRNAs encoding the acute phase reactants in livers of inbred mice. In addition to defining the spectrum of the acute phase response (i.e. the spectrum of responding gene products), we have identified structural and regulatory variation for the expression of several acute phase reactants, and have mapped the structural genes for AGP' and SAA.

Animals-Laboratory
inbred strains and AKXD recombinant inbred strains derived from the strains AKR17 and DBA127 were purchased from the Jackson Laboratory, Bar Harbor, ME.
Mus spretus is from a randomly breeding colony of mice maintained by Dr. Verne Chapman. These mice were originally trapped in Southern France. Animals were used a t 2-3 months of age. An acute phase response was induced either by subcutaneous injection of 50 p1 of turpentine into the lumbar region or by intraperitoneal injection of bacterial liposaccharide in phosphate-buffered saline. A trichloroacetic acid-extracted lipopolysaccharide preparation from Escherichia coli serotype 0.27:B8 was obtained from Sigma.
Hepatocytes-Hepatocytes were obtained by collagenase perfusion of adult livers according to Seglen (4). The conditions for culturing cells and labeling with [35S]methionine will be described elsewhere (3). To obtain non-N-glycosylated secretory proteins, the cells were incubated for 2 h in medium containing 2 pg/ml of tunicamycin and labeled with [35S]methionine for 6 h. Early precursor forms of glycoproteins were purified by concanavalin A-Sepharose chromatography of extracts from cells pulse-labeled with ["Slmethionine for 10 min (5).
Extraction and Blot Analysis of RNA-Total liver RNA was extracted by the guanidine HCl procedure (6, 7). RNA was separated on agarose gels containing formaldehyde (S), transferred to nitrocellulose (9), and hybridized to 32P-labeled p10-14 DNA, which contains a cDNA insert complementary to the mRNA for rat al-acid glycoprotein (10). Hybridizing RNA species were observed by autoradiography. Mouse ribosomal RNAs were used as markers.
mRNA Selection-Plasmid DNA was immobilized on a nitrocellulose filter and used to isolate complementary RNA as described previously (11). Selected RNA was translated in a cell-free system.
Cell-free mRNA Translation-Total or plasmid-selected mRNA was added to a fractionated cell-free translation system derived from animal cells (12) in the presence of [35S]methionine.
Acrylamide Gel Analysis-Cellular and cell-free synthesized proteins were subjected to two-dimensional polyacrylamide gel electrophoresis (12,13) and were visualized by fluorography (14). Identification of spots representing haptoglobin, hemopexin, al-antitrypsin, al-antichymotrypsin, apolipoprotein A-I, major urinary proteins, albumin, and serum amyloid A has been carried out by immunoprecipitation of cell-free synthesized proteins with monospecific antibodies to the plasma proteins, mRNA selection with cloned cDNA probes, and proteolytic comparison to the appropriate plasma protein as described elsewhere (3). Proteolytic mapping of protein spots cut out of the two-dimensional gels was accomplished with Staphylococcus aureu.7 V8 protease (Miles Laboratories Inc.) according to the method of Cleveland et al. (15). Assessment of the number of N-glycans/ molecule was achieved by partial endo-8-N-acetylglucosaminidase H digestion in a polyacrylamide gel (5).

Domain of the Hepatic Acute Phase
Response-To establish the basic pattern of the acute phase response in mouse liver, we analyzed the qualitative and quantitative changes in composition of functional mRNAs in the livers of animals treated

FIG. 1. Modulation of translata-
ble mRNAs in the liver of C3H/HeJ male mice by an acute inflammation. Males of strain C3H/HeJ received on intraperitoneal injection of 300 pI of phosphate-buffered saline (control), 300 pl of phosphate-buffered saline containing 10 p g of trichloroacetic acid-extracted lipopolysaccharides, or two subcutaneous injections of 25 pl of turpentine. Liver RNAs were extracted 24 h later and translated in a cell-free system. The products in 5-pl translation mixtures were separated by two-dimensional polyacrylamide gel electrophoresis. The fluorographs were exposed for 24 h. with bacterial liposaccharides or with turpentine. Total liver RNA was extracted and translated in a cell-free system, and the products were separated by two-dimensional polyacrylamide gel electrophoresis. We have observed striking changes in the cell-free translation patterns following inflammation in several inbred strains. As a representative example, the patterns for strain C3H/HeJ are given in Fig. 1. The mRNA levels for several proteins (spots [1][2][3][4][5][6][7][8] in Fig. 1) are increased, while those for others (spots 9

Hepatic Acute
into each of the protein spots (data not shown) indicated that the magnitude of changes varied considerably, from a reduction of less than %fold (spot 11) to an increase of more than 100-fold (spots [1][2][3]. Although both irritants are capable of inducing the same major changes, we found that liposaccharides, when used in doses below 50 pg/animal, were consistently less effective than turpentine (see spots 2 , 4 , IO, and 13).
We could identify some of the protein spots as precursor forms of previously identified major plasma proteins whose circulating concentrations are known to be altered during the acute phase reaction in mice (16). Among the proteins which are increased by inflammation are serum amyloid A (spot 3), haptoglobin (spot 4), and hemopexin (spot 5 ) ; among those which are reduced are albumin (spot 9), apolipoprotein A-I (spot IO), and the major urinary proteins (spot 1 I). No significant changes in mRNA levels for al-antichymotrypsin (spot 15) and a,-antitrypsin (spot 16) were detected. The mRNA for actin (spot 14) appears to be unaffected, although it has been reported that the synthesis of this protein in liver is increased 5-fold during the acute phase response (17).
From analysis of proteins secreted from primary cultures of mouse hepatocytes, we know that q-acid glycoprotein is a major acute phase reactant (3). Indeed, we can find a protein ( Fig. 1, spot 1 ) which has an electrophoretic mobility indistinguishable from the AGP precursor of the rat (12). However, this assignment was uncertain because there was an additional inducible protein (spot 2 ) with slightly larger molecular weight and more acidic charge. T o demonstrate which spot is AGP, we took advantage of a cloned cDNA probe to rat AGP (10). When tested with mouse RNA, this probe was found to hybridize to an inflammation-inducible mRNA of length 0.85 kilobases (Fig. 2). This mRNA was isolated by capturing onto nitrocellulose filters affixed with the cDNA plasmid and was translated in the cell-free system. mRNA from the liver of a male (AKR/J x DBA/2J)F, hybrid was used because of the presence of two electrophoretic forms of spot 1 (see below). As shown in Fig. 3, both forms of spot 1 as well as spot 2 were present in translation products of the selected RNA, indicating that mRNAs encoding both polypeptides were specifically recognized by the cDNA plasmid. These results suggest the presence of mRNAs for two AGPs in the mouse.
To substantiate this finding, we compared the proteolytic digestion patterns of spot 1, spot 2, and the corresponding cellular proteins synthesized and secreted by tunicamycintreated mouse hepatocytes. Spots 1 and 2 ( Fig. 4, lanes 2 and  3), synthesized in a cell-free system, yield digestion products that are similar to each other and to authentic rat AGP (12). In addition, the patterns of the cellular forms (lanes 4 and 5 ) were indistinguishable from the cell-free forms. Thus, there are two closely related forms of AGP, henceforth termed AGP-1 and AGP-2, that are probably encoded by distinct yet closely related mRNAs.
Although it appears that AGP-1 and AGP-2 share sequence homologies, translation of their mRNAs in the presence of dog pancreatic membranes revealed a major processing difference between the two. Electrophoretic analysis of the glycosylated intermediates after their purification by concanavalin-A sepharose chromatography (12) suggested that AGP-1 acquires five glycan units, while AGP-2 acquires six (Fig. 5A). T o confirm this, we have determined the number of glycan units in cellularly synthesized precursor forms. The fully glycosylated precursor forms of AGP-1 and AGP-2 were isolated from pulse-labeled mouse hepatocytes by lectin chromatography (5) and separated by two-dimensional gel electrophoresis (Fig. 5C); as controls, nonglycosylated AGPs secreted by tunicamycin-treated cells were also analyzed (Fig. 5B). Fully glycosylated AGP-1 appears as single spot; fully glycosylated AGP-2, however, appears as two equally intensive spots with identical isoelectric points and a molecular weight

19
C cated that AGP-1 contains five N-glycans while AGP-2 comprises a mixture of forms with five and six N-glycans (Fig.  50). Genetic Variation in the Acute Phase Reactant mRNAs of Inbred Mice-In order to find genetic variation that may be useful in gaining new insights into the nature of the acute phase response, we have compared the acute phase reactant mRNAs in the livers of several inbred mouse strains. Two interesting structural variants were found. The strains express one of two charge forms of AGP-1. One form, denoted AGP-1-B, is expressed, for instance, in strain C3H/HeJ, while the other form, denoted AGP-1-A, is expressed in strains AKR/J and DE/Cv (Fig. 6). A variant form of SAA was found in strain DE/Cv. The latter expresses an acidic SAA, denoted SAA-A, while all other strains express a basic SAA, denoted SAA-B (Fig. 6). (C3H/HeJ x DE/Cv)F, hybrids express both SAA-A and SAA-B as well as AGP-1-A and AGP-1-B (Fig.  6). Thus far, DE/Cv is the only inbred strain that has been analyzed and expresses SAA-A. The electrophoretic phenotypes for SAA and AGP-1 in several inbred strains are shown in Table I. Mobility differences were also apparent by twodimensional gel analysis of native plasma proteins (data not shown).
All other acute phase reactant mRNA translation products 6J male was translated in a cell-free system in the presence of dog pancreas microsomes. The synthesized glycosylated proteins were isolated by chromatography on concanavalin A-Sepharose and separated by twodimensional electrophoresis (12). Only the region of the gel containing the AGP intermediates is reproduced. Due to a partial modification of AGPs to more basic forms by an unknown reaction in the translation mixture (12). the intermediates appear as two vertical rows of spots. The positions of the AGP forms differing by a single N-glycan unit are indicated. B, nonglycosylated AGPs synthesized by hepatocytes. Primary hepatocytes in culture were treated for 2 h in 2 pg/ml of tunicamycin, which results in 95% inhibition of ['Hlmannose incorporation (data not shown). These cells were labeled with ["'SSJmethionine for 6 h in the continued presence of tunicamycin, and the secreted proteins in the culture medium were separated by two-dimensional gel electrophoresis. Only the section of the fluorogram containing AGP-1 and AGP-2 is reproduced. C, fully N-glycosylated AGPs synthesized by hepatocytes. Primary hepatocytes were labeled with ["S]methionine for 10 min. Total cellular glycoproteins were purified by chromatography on concanavalin A-Sepharose (5) and separated by two-dimensional gel electrophoresis.
The same section of the fluorogram as in R is reproduced. D, partial endo-B-N-acetylglucosaminidase H digestions.
Glycosylated and nonglycosylated forms of AGP-1 and AGP-2, prepared as described in R and C above, were cut out of preparative two-dimensional acrylamide gels and digested for 30 min a t 37 "C with endo-P-N-acetylglucosaminidase H in an 11% polyacrylamide gel, as described (5). Lanes represent the following samples: 1-4, fully glycosylated AGP-1; 5, nonglycosylated AGP-1; 6-8, fully glycosylated AGP-2 with five N-glycan units; 9-10, fully glycosylated AGP-2 with six N-glycan units; I I , nonglycosylated AGP-2. Since the pattern of glycosylated proteins (see C) includes a series of spots which have similar molecular weights but slightly more acidic charge than AGP-2, they were included in the following lanes to demonstrate that they represent a nonrelated protein: 12, protein that is slightly more acidic than the AGP-2 with five N-glycans; 13, protein that is slightly more acidic than the AGP-2 with six N-glycans. Lanes I, 5  are identical among laboratory strains, as assessed by twodimensional gel electrophoresis. In addition, using the inflammation protocols described, we have found no evidence for quantitative variation among the mouse strains tested in the levels of any acute phase reactant mRNAs either in control or inflamed animals. The existence of structural variation for AGP-1 and for SAA enables location, by linkage analysis, of the respective structural genes in the mouse genome. Such analysis can be readily accomplished with the use of recombinant inbred lines. These are sets of inbred strains generated by inbreeding the F, progeny from a cross of two progenitor strains (18). Recombinant chromosomes containing various combinations of progenitor genes become fixed in homozygous form; thus, by comparing the segregation pattern of alleles of a variant gene with the pattern of other markers whose map locations are known, it is possible to assign the variant gene of interest to a particular location in the mouse genome.
By examining the in uitro translation products of liver mRNA from liposaccharide-induced recombinant inbred strains of the AKXD set, derived from strains AKR/J and  (20), we can calculate that the structural gene for AGP-1 is thus located 3 f 2 centimorgans from b. We call this structural gene Agp-1; Agp-1" and Agp-I " denote the alleles carried by strains DBA/2J and AKR/J, respectively. Evidence that Agp-1 is distal to the centromere from b was obtained using the AKXD recombinant inbred line segregation pattern of the Ly-ml9 locus, which is 12 centimorgans proximal from b (21). The fact that Ly-ml9 is concordant with b in 12 AKXD strains and with Agp-I in only 11 strains (Table 11) indicates that Agp-1 is most likely distal to the centromere from b. In addition, in contrast to a proximal location, a distal location of Agp-1 requires no double crossovers to generate the AKXD strain distribution patterns of the three loci in question. Thus, it is likely, although not yet definitively proven, that Agp-1 is distal to the centromere from b. Interestingly, this location places Agp-I very close to the Lps locus, which is involved in regulating all aspects of the acute phase response in mice (22). Experiments to more accurately map Agp-I are currently in progress.
Since electrophoretic variation for SAA occurs only in strain DE/Cv, no appropriate recombinant inbred strains are available for mapping the SAA structural gene. However, to determine possible linkage of the SAA structural gene to Agp-I , we made use of the backcross progeny derived from mating (C3H/HeJ X DE/Cv)F, hybrid females to DE/Cv males. Of

Strain distribution pattern for chromosome 4 genes in AKXD recombinant inbred strains
The Agp-I allele for each strain was determined by the i n uitro translation assay. The distribution for the Ly-mI9 locus has been published elsewhere (26), while that for the 6 locus was communicated to us by Dr. B. Taylor. A denotes inheritance of the AKR/J phenotype; D denotes inheritance of the DBA/ZJ phenotype; X denotes a recombinant event.   will be henceforth termed Saa, with Saab and Sua" indicating alleles carried by C3H/HeJ and DE/Cv, respectively. Genetic Variation in the Acute Phase Reactant mRNAs of Wild Mice-Several studies have indicated that wild-derived mice are richer in genetic variation compared to laboratory inbred strains (23). Therefore, we studied the acute phase reactants in wild mice, with the expectation of finding novel variants. We examined the in vitro translation patterns of liver RNA from control and turpentine-induced M. spretus, which is a random bred stock originally trapped in Southern France and Spain (24). As shown in Fig. 7 and Table I, several interesting differences from the laboratory inbred strain patterns are apparent. Although all M. spretus mice carry the Agp-I" allele, some individuals contain barely detectable levels of AGP-1 following inflammation (compare B and C in Fig. 7), indicating the presence of putative regulatory variation within the population; this is specific for AGP-1, since other acute phase reactant mRNAs, including that for AGP-2, are induced to normal levels in the same animals. Other variations found in M. spretus include an electrophoretic variant for AGP-2 that is more basic and of higher molecular weight than that found in laboratory inbred strains and a variant of haptoglobin that is basic and of lower molecular weight; both forms of AGP-2 and of haptoglobin are expressed in C3H/HeJ X M. spretus hybrids (Fig. 70). Finally, two novel SAA phenotypes, SAA-C and SAA-D, both of which involve expression of two SAA polypeptides, are segregating in the M. spretus population (Fig. 8). We are currently isolating variants for major acute phase reactants in the form of sublines that will be of use for further studies.

DISCUSSION
A genetic approach to the study of the mammalian acute phase response offers the potential to identify and map structural genes encoding acute phase reactants and regulatory genes that may modulate the response. We have undertaken genetic studies in the mouse; the availability of a large number of inbred and wild-derived stocks showing extensive genetic variation (23) clearly makes the mouse an advantageous model for such studies.
We have focused our attention on translatable acute phase reactant mRNAs of the liver. By cell-free mRNA translation, we have defined the spectrum of acute phase mRNAs and could identify several mRNAs that are induced during the acute phase response as well as several that are repressed. In contrast to rat (10, l a ) , there are two AGPs in mouse. Since both were identified in the cell-free mRNA translation products, it is likely that they are encoded by separate mRNAs. To what extent AGP-1 and AGP-2 are structurally related is unknown. Their mRNAs must share considerable sequence homology, since both cross-react with a rat AGP cDNA probe (Fig. 2). Whether these mRNAs derive from separate or overlapping structural genes can be answered only by direct analysis of corresponding genomic sequences.
Interstrain variation in AGP-1 electrophoretic mobility was identified and used to map the corresponding structural gene, Agp-I, to chromosome 4. The genetic data allowed us to tentatively place Agp-1 very close to the Lps locus, although such placement must be verified by more extensive analysis which is in progress. Lps was identified previously on the basis of a diminished response to lipopolysaccharide in C3H/ HeJ mice (22) and appears to regulate all aspects of the response, including lymphoid cell activation, hypothermia, and induction of both colony-stimulating factor and SAA levels in serum (25); in addition, we have recently noted the effect of the Lps locus on induction of mRNAs for all acute phase reactants detectable by the in vitro translation assay.:' We do not know the significance, if any, of the close proximity of Agp-I to Lps. In a preliminary analysis of (C3H/HeJ xDE/ Cv)F, hybrids, we have found that the effect of Lps on Agp-1 expression is not cis-acting.
Whatever the relationship between Agp-1 and Lps, it is clear that not all acute phase reactant structural genes are clustered on chromosome 4. Data on SAA place its structural gene, Saa, on chromosome 7; this agrees with the assignment made on the basis of a DNA polymorphism.

SAA-C / S A A -C S A A -D / S A A -0
. ) .

S A A -C / S A A -D
FIG. 8. Electrophoretic polymorphism for SAA in M. spretus. Total liver RNA was extracted from three turpentine-treated M. spretus females and translated in the cell-free system. The cell-free translation products were separated by two-dimensional gel electrophoresis using a 12% polyacrylamide gel in the second dimension. Fluorographs were exposed for 24 h. Only the section containing SAA is reproduced. The SAA phenotype of each of the three individuals is indicated.
regulatory variation, which governs the induction specifically of AGP-1 mRNA, is found in M. spretus. The location and nature of the locus responsible for this alteration is under investigation and may lead to the identification and characterization of cis-or trans-acting factors responsible for modulating the acute phase response of specific genes.