Cloning, structure, and expression of a rat binding protein for polychlorinated biphenyls. Homology to the hormonally regulated progesterone-binding protein uteroglobin.

Certain metabolites of polychlorinated biphenyls (PCBs) are retained in the Clara cells and in the airway lumen of rodent lung due to their interaction with a secretory 13-kDa protein. Here, we report the isolation of a cDNA encoding the rat lung PCB-binding protein. The identity of the PCB-binding protein is supported by expression of the cDNA in Cos-1 cells where the homogenates from transfected cells show specific binding of 4,4'-bis([ 3H]methylsulfonyl)-2,2',5,5'-tetrachlorobiphenyl, a high affinity ligand for the PCB-binding protein. Also a monospecific antiserum to the PCB-binding protein recognizes a 13-kDa protein in the homogenates of transfected cells but not in the corresponding fraction of mock-transfected cells. Northern blot analysis of total RNA from different rat tissues demonstrates that the cDNA detects a approximately 600-base pair mRNA which appears to be solely expressed in lung. Interestingly, DNA sequence analysis and prediction of the amino acid sequence reveals that the PCB-binding protein shares 53% positional amino acid identity with uteroglobin, a progesterone-binding protein found in rabbit uterus and lung. Furthermore, amino acids shown by x-ray crystallography to delineate the central cavity of uteroglobin, which fits progesterone, are highly conserved in the two proteins.

Certain metabolites of polychlorinated biphenyls (PCBs) are retained in the Clara cells and in the airway lumen of rodent lung due to their interaction with a secretory 13-kDa protein.
Here, we report the isolation of a cDNA encoding the rat lung PCB-binding protein. Respiratory disorders have been described for human subjects accidentally or occupationally exposed to polychlorinated biphenyls (PCBs)' (Shigematsu et al., 1978;Warshaw et al., 1979) and more than 60 different methylsulfonyl-PCBs have been identified in lungs from accidentally exposed subjects (Haraguchi et al., 1984). These sulfur-containing PCB metabolites have been shown in animal studies to arise during * This work was supported by grants from the Swedish Medical Research Council, The Swedish Heart Lung Foundation, Magnus Bergvalls Stiftelse, and Draco AB. 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.  (Mio et al., 1976;Jensen and Jansson, 1976;Bakke et al., 1982;Bakke and Gustafsson, 1984). Depending on the positions of the chlorine atoms and the methylsulfonyl moieties, certain PCB methyl sulfones accumulate in the lung and kidney of rats and mice (Brandt and Bergman, 1981).
Using a radioactively labeled PCB methyl sulfone, 4,4'bis([3H]methylsulfonyl)-2,2',5,5'-tetrachlorobiphenyl, we have previously shown a selective accumulation of this compound in the cytoplasm of the nonciliated bronchiolar (Clara) cells and in the airway lumen of the rat and mouse lung (Lund et al., 1985;Brandt et al., 1985). This selective in uiuo disposition corresponds to the localization of a secretory 13-kDa protein of Clara cell origin which binds certain methylsulfonyl-PCBs with high affinity (Lund et al., 1985(Lund et al., , 1988a. The protein has been purified from rat lung and consists of two apparently identical subunits held together by disulfide bridges (Lund et al., 1988b) and a physicochemically similar protein has been characterized in human bronchoalveolar lavage fluid (Lund et al., 1986). In view of the respiratory disorders observed in PCB-exposed human subjects and the pathway for the accumulation of methylsulfonyl-PCBs in lung described above it is tempting to speculate in a causative relationship.
To test such a hypothesis would require insight into the physiological function of the PCB-binding protein.
In this paper we describe the cloning and expression of a cDNA encoding the rat lung PCB-binding protein. The DNA sequence analysis reveals a structural relationship between the PCB-binding protein and uteroglobin, a hormonally regulated steroid-binding protein in rabbit (Bailly et al., 1983). The kinship of these proteins suggest novel approaches to the elucidation of the physiological role of the PCB-binding protein and of its role in PCBinduced lung toxicity. EXPERIMENTAL PROCEDURES AND RESULTS'

DISCUSSION
In this paper we describe the cloning, structure, and expression of a cDNA corresponding to a rat lung PCB-binding protein. When expressed in Cos-1 cells, the cDNA directs the synthesis of a protein that is recognized by a monospecific antiserum to the rat PCB-binding protein. Also, the expressed protein has ligand binding properties that are indistinguishable from those of the purified rat PCB-binding protein (Lund The amino acid sequence of the rat PCB-binding protein (rUGL) was aligned with the amino acid sequence of uteroglobin (Bailly et al., 1983) using the program FASTA (Pearson and Lipman, 1988). Amino acid identities are connected by strui&t lines. Amino acids in the PCB-binding protein are numbered aboue the sequence and amino acids in uteroglobin are numbered below the sequence; residues in the putative secretory leader sequences are assigned negative numbers, whereas amino acids in the mature proteins are assigned positiue numbers, Underlined are amino acids in uteroglobin that are localized to the central ligand-binding cavity (ho/d) and the cysteins that participate in disulfide bridge formation (italic). et al, 1985, 19SSb), whereas the electrophoretic mobility of the expressed protein is slightly different from that of the purified protein, possibly due to differences in post-translational modification.
RNA blotting experiments indicate that the mRNA for the PCB-binding protein is solely expressed in the lung and genomic DNA blotting experiments suggest that the protein may be encoded by a single copy gene. The deduced sequence of the PCB-binding protein reveals a structural similarity (53% positional amino acid identity in a go-amino acid overlap) to a steroid-binding protein, uteroglobin (Menne et al., 1982;Bailly et aE., 1983). Uteroglobin has been previously thought to be exlusively expressed in rabbits and other members of the order Lugomorph (Savouret and Milgrom, 1983) although some studies suggest an immunological crossreactivity with a protein in humans (Dhanireddy et al., 1988). The observed similarity between uteroglobin and the PCBbinding protein may prove to be of particular importance with regard to our future structural and functional analysis of the PCB-binding protein. The three-dimensional structure of uteroglobin is known in detail by means of x-ray crystallography (Mornon et aZ., 1980;Morize et al., 1987;Bally and Delettre, 1989). Uteroglobin is a dimer of two identical monomers (A and B)

of 70 amino acids held together by two disulfide bridges between residues 3 and 69 (3S[A]-69S[B] and 69S[A]-3S[B]).
It is a globular protein that can be included in a sphere with a radius of 17 A. Most of the hydrophobic residues are totally or partially buried inside the protein and most of the hydrophilic groups are exposed on the surface. An oblong hydrophobic pocket is present centrally in the protein and is delimited by hydrophobic residues Ile-2, Phe-6, Val-9, Ile-10, Leu-13, Leu-14, Leu-25, Gly-38, Met-41, Leu-45, Ile-56, Met-57, Leu-59, and Ile-63 and by 2 polar residues, Tyr-21 and Thr-60. These amino acids may be important for the specificity of ligand binding. In particular the hydroxyl groups of Tyr-21 and Thr-60 may interact with oxygens in progesterone, a putative physiological ligand for the protein. It is of great interest that most of these particular residues are present in the PCB-binding protein and that they are identically spaced as in uteroglobin (Fig. 5). It is therefore our hypothesis that the three-dimensional structure of the PCB-binding protein is very similar to that of uteroglobin, keeping the conserved hydrophobic amino acids as well as the equivalents of Tyr-21 and Thr-60 in a central ligand binding cavity. Amino acids in the PCB-binding protein that through this structural analogy to uteroglobin would be assigned to the surface of the protein are not identical to the same extent. This may explain why an antiserum to the rat PCB-binding protein does not recognize a protein in lung fractions from other species and why antibodies to uteroglobin do not react with the PCB-binding protein (Lund et al., 1988b).3 Our model thus envisions two homologous proteins (i.e. of common evolutionary origin) in which divergent evolution has left a modest overall amino acid identity (53%) but where the structural basis for ligand binding has been conserved. To us the data further imply that ligand binding is linked to the function of the proteins.
The extension of this model would therefore suggest similar functions of the PCB-binding protein and uteroglobin. TO what extent can this help in the elucidation of the physiological role of the PCB-binding protein and of the toxicological implications of its PCB interactions? Preliminary results regarding the amino-terminal sequence of the PCB-binding protein prompted us to test whether methylsulfonyl-PCBs bind purified uteroglobin and if they fit its central hydrophobic cavity as studied by molecular graphics and interactive energy minimizations, and in both instances they do (Gillner et al., 1988). The limited availability of purified uteroglobin prevented a detailed analysis of the binding affinities for different PCB metabolites. Clearly both proteins bind progesterone with similar affinities (Beat0 and Baier, 1975;Lund et al., 1985). The physiological significance of the binding of progesterone to uteroglobin is not well understood, but it has been suggested that since rabbit uteroglobin is present in uterine fluids early in pregnancy it may function to protect the developing conceptus from toxic levels of progesterone (reviewed in Savouret and Milgrom, 1983). Other proposed mechanisms of action of uteroglobin include protease inhibition and immunological masking of antigens (Savouret and Milgrom, 1983). Perhaps most intriguing is the recent suggestion that uteroglobin can inhibit phospholipase A2 (Miele et al., 1988) since exposure of humans to PCBs is associated with inflammatory symptoms from the airways and an accumulation of methylsulfonyl-PCBs in the lungs of exposed subjects is well documented (Shigematsu et al., 1978, Haraguchi et al., 1984. Bronchoalveolar lavage fluid from healthy volunteers has also been shown to contain a binding protein for 4,4'-bis([3H]methylsulfonyl)-2,2',5,5'-tetrachlorobiphenyl which is physicochemically similar to the rat PCBbinding protein (Lund et al., 1986). The availability of a fulllength cDNA for the PCB-binding protein should allow for the expression of sufficient quantities of the protein to test if it is also an inhibitor of phospholipase A2 and to adress the role of ligand binding in this putative function. Finally, purification and cloning of corresponding cDNA for the human PCB-binding protein will show to what extent the shared ligand binding properties reflect a structural relation to the rat PCB-binding protein and uteroglobin.