Membrane-associated Carbonic Anhydrase from Rat Lung PURIFICATION, CHARACTERIZATION, TISSUE DISTRIBUTION, AND COMPARISON WITH CARBONIC ANHYDRASE IVs OF OTHER MAMMALS*

Carbonic anhydrase (CA) IV was purified to homo- geneity from rat lung microsomal and plasma membranes. The single N-terminal amino acid sequence showed 66% similarity to that reported for human CA IV. A monospecific antibody to the 39-kDa rat enzyme that cross-reacts on Western blots with CA IVs from other mammalian species was produced in rabbits.

Carbonic anhydrase (CA) IV was purified to homogeneity from rat lung microsomal and plasma membranes. The single N-terminal amino acid sequence showed 66% similarity to that reported for human CA IV. A monospecific antibody to the 39-kDa rat enzyme that cross-reacts on Western blots with CA IVs from other mammalian species was produced in rabbits. Digestion of rat lung enzyme with endoglycosidase (peptide-N-glycosidase F) reduced the M, to 36,000, suggesting that rat CA contains one N-linked oligosaccharide chain. All of eight additional mammalian CA IVs that were examined also contained oligosaccharide chains, as evidenced by reduction in M, from 62,000 (cow, sheep, and rabbit), 42,000 (pig, guinea pig, and dog), and 39,000 (mouse and hamster) to 36,000 after treatment of the respective lung microsomal membranes with peptide-N-glycosidase F. The 36-kDa human enzyme showed no change in molecular mass with this treatment. Thus, the human CA IV is the exceptional one in lacking carbohydrate.
Rat lung CA IV was found to be relatively resistant to sodium dodecyl sulfate and to be anchored to membranes by a phosphatidylinositol-glycan linkage; both properties were found to be shared by other mammalian CA IVs. Western blot analysis indicated distribution of CA IV in rat tissues other than kidney and lung where it was previously known to be present. CA IV was particularly abundant in rat brain, muscle, heart, and liver, all locations where the CA IV enzyme was not known to be present previously. None was detected in rat skin or spleen.
The carbonic anhydrase isozymes are zinc metalloenzymes that catalyze the reversible hydration of carbon dioxide (COz + H20 $ HCO: + H+). They vary in their subcellular localization, with cytoplasmic (CA I, CA 11, CA 111, and CA VII)l (1, 2), cell surface membrane (CA IV) (3)(4)(5), mitochondrial (CA V) (6), and secretory (CA VI) (7-9) forms having been described. The cytoplasmic isozymes have been studied in greatest detail. Until recently, poor yields of the membrane-GM34182 and DK40163. The costs of publication of this article were * This work was supported by National Institutes of Health Grants defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. bound enzyme (CA IV) from cumbersome purification procedures have prevented extensive studies on the structure/ function relationships of the membrane-bound isozyme.
The first membrane-associated CA purified to homogeneity was obtained from bovine lung (3). It was characterized as a disulfide bond-containing glycoprotein with an apparent molecular weight of 52 kDa and designated CA IV to distinguish it from the then known cytoplasmic isozymes CA I, CA 11, and CA I11 (3). Bovine CA IV had the unique property of being stable for up to several hours in 1-5% SDS solution, which facilitated its isolation by affinity chromatography. However, the enzyme was unstable in SDS after 24 h and could not be characterized extensively. Several years later, a different type of purification of a membrane-bound carbonic anhydrase from human kidney membranes was reported (4, 10). The apparent molecular weight was initially reported to be 68,000 (lo), but more recent purifications by this method yielded an inactive polypeptide with an M , of 34,400 on SDS-PAGE (4). The N terminus of the homogeneous protein was reported to be blocked on amino acid sequencing. Paradoxically, antiserum raised against SDS-treated 34.4-kDa enzyme reacted only with a polypeptide of 55 kDa, which was found in many tissues and was attributed to CA IV (ll), despite the difference in molecular weight from the antigen to which the antibody was raised.
Recently, Zhu and Sly (5) reported purification of CA IV to homogeneity from human lung and kidney. Their biochemical and immunological results differed substantially from those reported by Wistrand and Knuuttila (4) and Carter et al. (11). First, the 35-kDa homogeneous CA IV from human lung and kidney was catalytically active. Second, it had unblocked N termini, and amino acids could be sequenced by Edman degradation for up to 17 cycles without interruption. Third, the monospecific antiserum raised to human lung CA IV reacted only with polypeptides of appropriate M , (35,000) in membrane homogenates of lung, kidney, and other CA IVcontaining tissues. The 35-kDa human CA IV was found to contain no carbohydrate, but it did contain disulfide bonds that appeared to account for its relative stability in sodium dodecyl sulfate. The human CA IV was shown to be anchored to lung and kidney membranes through phosphatidylinositolglycan linkages.
The interesting similarities and differences between bovine lung CA IV (3) and CA IV from human lung and kidney (5) led us to purify the membrane-associated CA IV from rat lung microsomal and plasma membranes. We hoped to assess the versatility of the new purification procedure for other mammalian CA IVs, to provide antigen for immunological studies of rat CA IV, and to study the physicochemical properties of other mammalian CA IVs to define their common properties.
In this paper, we provide evidence that rat lung CA IV is a glycoprotein and is also anchored to membranes via a phosphatidylinositol-glycan linkage. We also report immunologic studies with antibody raised to the purified rat lung CA IV that show that these two properties are also shared by CA IVs from nine different species. These studies show that the large molecular weight differences between the human CA IV and the CA IVs in various other mammalian species are almost entirely explained by differences in carbohydrate content. Finally, we present the first evidence for the tissue distribution of CA IV in rat tissues other than kidney.

DISCUSSION
The successful application of the purification procedure we reported for CA IV from human lung to CA IV from rat lungs demonstrates the versatility of this procedure for producing relatively large amounts of homogeneous enzyme in a form that is active and stable to storage. The same purification has been applied successfully to CA IV from bovine lungs, porcine lungs, and rabbit lungs.3 The success of this procedure relies on one property that all the CA IVs appear to have in common, namely relative stability in 1-5% sodium dodecyl sulfate. This property allows them to be selectively adsorbed to an affinity column under conditions where nearly no other proteins (or other carbonic anhydrases) are adsorbed. The resistance to denaturation in sodium dodecyl sulfate probably reflects the stabilizing effect of one or more disulfide bonds (19), since the same level of sodium dodecyl sulfate completely inactivates CA IVs in the presence of reducing agents (3). Once the enzyme is adsorbed to the column, replacing the sodium dodecyl sulfate with a nonionic detergent before elution allows the hydrophobic CA IVs to be recovered under conditions where they are active and stable to storage (5).
A second property that mammalian CA IVs appear to have in common is a hydrophobic anchor that is sensitive to phosphatidylinositol-specific phospholipase C. The phosphatidylinositol-glycan anchor provides a means to target CA IV to the extracellular surface of the plasma membrane. In polarized cells, the phosphatidylinositol-glycan anchor has been suggested as a means of targeting protein specifically to the apical surface (22). However, in rat kidney, CA IV was identified on both the apical and basolateral surfaces of certain portions of the proximal convoluted tubules and the thick ascending limb (23). Whether it is anchored differently on the apical and basolateral membranes in the rat nephron has not yet been determined.
The presence of N-linked oligosaccharide chains was found in 9 of 10 mammalian CA IVs examined. Only human CA IV lacked N-linked oligosaccharides. The studies reported here demonstrated a wide variation in the carbohydrate content among the nonhuman CA IVs with no obvious biochemical differences between those with less and those with more carbohydrate. This observation, and the fact that the glycosylation is not conserved evolutionarily, suggests that the oligosaccharide chains do not have an important role in the The availability of the antibody described here provided the opportunity to study the tissue distribution of CA IV. One prior study on the distribution of CA IV in human tissues has been reported. However, it is not possible to interpret the results of that study with confidence. The presumptive antigen used to immunize the rabbit was a 35-kDa protein purified from human kidney that was thought to be CA IV. However, the only peptide detected with the antibody in human tissues was a 55-kDa polypeptide that was present in every tissue examined. It seems probable that the antigen surveyed in this study was a contaminant protein, rather than the 35-kDa human CA IV.
The tissue distribution of rat CA IV seen in the present work is quite extensive. Yet its absence in skin and spleen indicates that CA IV is not ubiquitous. Immunoreactive protein at 39 kDa was found in homogenates of brain, kidney, lung, liver, skeletal muscle, stomach, large intestine, and small intestine. Faint bands of larger molecular weight (50,000) cross-reacting species were also found in brain, stomach, and large intestine. Whether these bands of cross-reacting proteins have carbonic anhydrase activity is not yet clear. Nor is it clear which cell types account for the immunoreactivity found in the tissue homogenates that contained CA IV. Detailed immunohistochemical studies recently defined the cellular distribution of CA IV in rat kidney, where the CA IV is present on the apical and basolateral surfaces of specific segments of the nephron (23), and in the human eye, where the CA IV is found in a specific capillary bed, the choriocapillaris (24). Similar studies using the antibody described here can help clarify which cells account for the immunoreactivity in the additional rat tissues we found to contain CA IV in this study. is shown in Fig. 1 (lane 1). A single polypeptide of 39 kDa,was detected Physical characterization. The relative mobility Of the enzyme under nonredusing conditions on SDS-PAGE was qreater than that under reducing Conditions. This result indicates that eat lunq CA IV has a Dore conpact Structure under non-reducing conditions which may indicate that it is stabilized by disulfide bond(s) (191.
Chemical charactari=ation. Purified rat lung CA IV was subjected to N-terminal amino acid sequencing. A single N-terminal amino acid was Observed. This provided additional evidence that the enzyme preparation was homogeneous.
Comparison of the amino acid sequence Of rat lunq CA IV with human CA IV is qlven in Table I. There was 55% sequence homoloqy between the rat and human enzymes. About 19% differences in AA homology are due to sinqle base difference. based on the deduced nucleotide sequences. A highly conserved tryptophan resldue at the 5th cycle and a proline residue at 216t cycle (20) are present in rat lung CA IV. The tyrosine at the 7th cycle which has been implicated in the active site of carbonic anhydrases (20) is also consarved.

Rat -D S H W X Y E I O A K E P N S X X S G P "
M i n e acids in boxes are conserved amino acids.
x indicates non-identified residues. reaction for CA 11. Skin has no detectable CA IV, but has a very weak reaction for CA 11. The ratio between CA IV over CA I1 was maximum for heart and muscle. These results indicated that membrane bound CA IV is expressed i n many. but not all, tissues and the level of CA IV expressed in different t1ssUeJ varies with the tissue.  Table I1 Show that about 80\ Of the CA Iv enzyme activity vas released from the membrane. When the marker enzyme 5'-nuclaotidase. which is known to be phosphatidylinositol-glycan anchored, was studied under identical conditions, bo\ of the 5'-nucleotidase was solubilized (nee Table 11).  of mammalian lung CA IVs lung microsomes of several additional mammals were obtained and microsoma1 kmbranes were treated with PI-PLC Soluble and membrane-bound isozymes Were Separated by Centrifugation sAd analyzed by SDS-PAGE followed by immunoblotting. ImmUnOblOts of solubilized and membranebound enzyme are shown in Fig. 4A and 4 8 respectively. More than 8 0 % of the enzyme from lung microsomes Of each mammal Vas solubilized by PI-PLC treatment (Fig. 4 A l indicating that CA Iv in lung microsomal membranes in a11 mammals used here are anchored to the membrane vis a phosphstidylinositol-gly=~" linkage.

B K LuSP L M H SK ST U SI
In order to explore the generality Of pho6phatidylinoaitol-glycan anchoring