Purification of mammalian filamin. Similarity to high molecular weight actin-binding protein in macrophages, platelets, fibroblasts, and other tissues.

We have purified the high molecular weight actin-binding protein, filamin from guinea pig vas deferens. We find this mammalian filamin is very similar to chicken gizzard filamin in subunit molecular weight, amnio acid composition, actin-binding properties, immunological cross-reactivity, and the ability to be phosphorylated by cyclic AMP-dependent protein kinase. Anti-filamin antibodies cross-react with a high molecular weight macrophage actin-binding protein, and with a high molecular weight protein in platelets and fibroblasts. Furthermore like filamin, these proteins are also phosphorylated and cyclic AMP stimulates their phosphorylation. Anti-filamin antibodies do not cross-react with the erythrocyte membrane protein spectrin or with high molecular weight proteins in brain extracts. We conclude that filamin from avian and mammalian smooth muscle are very similar proteins and furthermore that many, but not all, non-muscle cells contain a protein closely related to filamin.


Institute,
Laboratory of Molecular Biology, Bethesda, Maryland 20014 We have purified the high molecular weight a&in-binding protein. filamin from guinea pig vas deferens. We find this mammalian filamin is very similar to chicken gizzard filamin in subunit molecular weight, amino acid composition, actin-binding properties, immunological cross-reactivity. and the ability to be phosphorylated by cyclic AMP-dependent protein kinase. Anti-filamin antibodies cross-react with a high molecular weight macrophage a&in-binding protein, and with a high molecular weight protein in platelets and fibroblasts.
Furthermore like filamin, these proteins are also phosphorylated and cyclic AMP stimulates their phosphorylation.
Anti-filamin antibodies do not cross-react with the erythrocyte membrane protein spectrin or with high molecular weight proteins in brain extracts. We conclude that filamin from avian and mammalian smooth muscle are very similar proteins and furthermore that many, but not all. non-muscle cells contain a protein closely related to filamin.
Extracts of many vertebrate and invertebrate cell types contain high molecular weight proteins that interact with actin (l-12). Several of these proteins have been purified and the purified proteins have been shown to be actin-binding proteins.
Thus, the erythrocyte membrane protein spectrin has been purified by several groups (for review see Ref. 13) and has been shown to bind to actin (6,7). Stossel and his associates have purified a high molecular weight actin-binding protein (ABP) from alveolar macrophages and leukemic leukocytes (8,(14)(15)(16) Filamin is another high molecular weight protein that has been isolated from chicken gizzard (9,10,17) and that has been shown to bind to F-actin (10,18,19). Filamin has also been identified immunochemically in cultured fibroblasts (9,20 (20). lyzed. For the preparation of solid tissues such as skeletal muscle, cardiac muscle, vas deferens, and brain, guinea pigs of the Hartley or NIH strains weighing 300 to 600 g were lightly anesthetized with ether and killed by cardiac excision.
Tissues were rinsed in NaCl/P, and were then quickly frozen in liquid nitrogen. Chicken gizzard extract was prepared from fresh chicken gizzard in the same manner as for guinea pig tissues. Triton X-100 (1%) was included in the extracts of vas deferens, tibroblasts, and platelets. The extract was centrifuged at 100,000 x g for 30 min, the pellet discarded, and then bovine serum albumin was added to the supernatant to a final concentration of 1.5 mgiml.
(ii) Immunoprecipitation Procedure. Samples of 300 ~1 of the extract were incubated with 80 pg of either purified anti-filamin antibody or preimmune r-globulin. The incubation was done at 1°C for 20 min. The IgG fraction of rabbit anti-goat IgG (300 /*-1 of 10 mg of protein/ml, Cappel) was then added. After incubation for 20 min at l"C, a solution of 6 M urea and Triton X-100 was added to give a final concentration of 3 M urea and 1% Triton X-100. The immunoprecipitate was sedimented at 12,000 x g for 10 min in siliconized glass tubes (10 x   KCl, all the tilamin sedimented with the actin (Fig. 3, Group  B). If the KC1 concentration was increased to 150 mM KC1 (Fig. 3, Group C) or 600 mM KC1 (Fig. 3, Group D) less fXunin sedimented with the actin. Thus, like chicken gizzard filamin, mammalian filamin binds to F-actin and this interaction can be reduced by high concentrations of KC1 (10,18).

3332
Filamin in Mammalian Tissues the specificity of the antibody checked by immunoprecipitation technique (Fig. 5, Group A). When acrylamide gels containing (Fig. 4). As shown in Fig. 4, Group A, anti-GP filamin an extract of guinea pig vas deferens were exposed to lZ51-antiantibody quantitatively immunoprecipitated the filamin from GP filamin, the tilamin band was the only protein to bind an extract of guinea pig vas deferens (Fig. 4, Group A).
anti-filamin antibody. Furthermore no other vas deferens proteins were immunopre-We next tested the cross-reactivity of guinea pig vas defercipitated by the antibody.
The specificity of the anti-GP ens fllamin and chicken gizzard filamin (Fig. 6). As shown in tilamin antibody is also demonstrated by the gel localization Fig. 6 (Fig. 6, Group B, Lane 3) and guinea pig vas deferens filamin (Fig. 6, Group A, Lane 3). In this study we did not attempt to quantitate the degree of cross-reactivity of the two proteins. However our data clearly established that chicken gizzard and guinea pig vas deferens filamin are immunologically related. Phosphorylation of Guinea Pig Vas Deferens Filamin Effect of CAMP-We previously reported that chicken gizzard and fibroblast filamin can be phosphorylated (20). Guinea pig vas deferens filamin also can be phosphorylated.
As shown in Fig.  7, Group A, incubation of extracts of vas deferens with ly-:)'j P]ATP leads to a low level of x P incorporation into filamin (Fig. I, Group A, Lane 2). Inclusion of CAMP in the incubation mixture greatly stimulated filamin phosphorylation (Fig. 7 tein (ABP) -Stossel and Hartwig have reported that extracts of macrophages contain a major protein (ABP) with an electrophoretic mobility slower than the heavy chain of myosin (8). As shown in Fig. 2, Lane C, upper band, the electrophoretic mobility of ABP is very similar to that of filamin. Furthermore, the amino acid composition of guinea pig vas deferens filamin is very similar to that reported by Stossel and Hartwig (15) for rabbit alveolar macrophage ABP (Table I).
We also used the antibody to guinea pig vas deferens fllamin to investigate the similarity of filamin to ABP. As shown in Fig. 4, Group B, anti-GP filamin antibody specifically immunoprecipitated all the ABP from extracts of macrophages. Furthermore no other macrophage proteins were immunoprecipitated. An alternative demonstration of the cross-reactivity of ABP and filamin is shown by the gel localization technique. As shown in Fig. 5, Group B, ABP binds 1251-anti-GP filamin antibodies and no other proteins in the macrophage extract cross-react with filamin. Thus macrophage ABP clearly crossreacts with anti-GP filamin antibodies. Since fdamin is a phosphoprotein we wished to determine whether macrophage ABP can be phosphorylated.
As shown in Fig. 7, Group B, incubation of extracts of macrophages with lyJIP]ATP leads to :?lP labeling of ABP, and CAMP markedly stimulated this phosphorylation (Fig. 7,  -Human platelets contain three major proteins with electrophoretic mobilities between 250,000 and 200,000 (Fig. 2, Lane E). The lower of these bands has the same mobility as myosin heavy chain and is presumably platelet myosin. The upper of these bands has an electrophoretic mobility identical to filamin (Fig. 2, Lane E). To determine whether human platelets contain filamin, extracts of platelets were immunoprecipitated with anti-GP filamin antibodies. As shown in Fig. 4, Group C, anti-GP filamin antibodies quantitatively and specifically immunoprecipitated the uppermost of the platelet high molecular weight proteins, indicating this protein is filamin. The gel localization technique confirmed this observation (Fig. 5,Group D). Furthermore as in other tissues, no other proteins reacted with the anti-filamin antibody.
Since filamin and the related ABP are phosphoproteins, and since CAMP probably plays an important role in platelet function, it was of interest to determine whether platelet filamin could be phosphorylated.
As shown in Fig. 7, Group C, extracts of platelets incubated with ly x PIATP phosphorylated filamin and CAMP greatly stimulated this phosphorylation (compare Fig. 7, Group C, Lanes 2 and 3).

Relationship
of Filamin to Fibroblast High Molecular Weight Proteins -Extracts of cultured rat fibroblasts contain a high molecular weight protein with the same mobility as filamin (Fig. 2, Lane F) that reacts weakly with anti-chicken gizzard filamin antibody (20). This protein is also immunoprecipitated by the anti-guinea pig filamin antibodies (Fig. 4, Group D). Furthermore this protein binds '""I-anti-GP filamin and is the only fibroblast protein to do so (Fig. 5, Group E).

Relationship
of Filamin to Brain High Molecular Weight Proteins -Guinea pig brain contains two very prominent high molecular weight proteins with molecular weights between 250,000 and 200,000 (Fig. 2, Lane G). Neither of these proteins has a mobility comparable to filamin and no other proteins are evident in whole brain extract that have the same mobility as filamin. Gel localization of brain extracts, Fig. 5, Group F, showed no binding of '"jI-anti-GP filamin antibody to brain proteins. Thus filamin does not appear to be present in measurable quantities in brain tissue.

Relationship
of Filamin to High Molecular Weight Proteins in Skeletal and Cardiac Muscle -SDS-gel electrophoresis of extracts of skeletal and cardiac muscle (Fig. 2, Lanes H and I) indicate that only a small amount of protein was present in the M, = 250,000 region. Whether this protein was derived from muscle cells or other cells present in muscle tissue was not investigated.

Smooth
Muscle Filamins-The objective of these studies was to obtain highly purified filamin from a mammalian source and to use it to clarify the relationship between the various high molecular weight actin-binding proteins. We have investigated these proteins in a variety of cell types and tissues of mammalian origin. As a starting point we have used the well characterized smooth muscle actin-binding protein filamin. Previous studies have been done on filamin from chicken gizzard, an avian source. Since our interest is in mammalian tissues, we purified a mammalian filamin. The purification of filamin from guinea pig vas deferens was very similar to the purification of filamin from chicken gizzard except that the vas deferens protein required a slightly higher KC1 concentration for its elution from DEAE-cellulose.
The two proteins have identical subunit molecular weights and comparable amino acid compositions. Furthermore both proteins bind to F-actin and gel solutions of F-actin, indicating some functional similarity. The cross-reactivity of both filamins to antisera prepared against the other indicates substantial immunological similarity. Finally both proteins are phosphorylated by CAMP-dependent protein kinase (201.' While both proteins are similar, they are clearly not identical. Chicken gizzard filamin is cleaved at a single site by porcine muscle Caz+-activated protease to give two fragments, heavy merofilamin (M, = 240,000) and light merofilamin" (M, = 10,000). Both these fragments are resistant to further degradation by the protease. Guinea pig vas deferens Iilamin contains a second site susceptible to the Ca'+-activated protease.' A second difference is that rabbit muscle CAMP dependent protein kinase appears to phosphorylate guinea pig vas deferens filamin faster than chicken gizzard filamin (data not shown). Although subtle differences may exist between the smooth muscle filamins of different species we think it likely that smooth muscle filamins represent a group of rather closely related proteins.

Non-Muscle Filamins
-These studies show that proteins closely related to filamin exist in some but not all mammalian cells. In particular the previously described actin binding protein of rabbit alveolar macrophages (8) is closely related to filamin. It resembles filamin in subunit molecular weight, amino acid composition, and actin binding and reacts with anti-filamin antibodies. Furthermore like filamin, macrophage ABP is a phosphoprotein whose phosphorylation can be increased by addition of CAMP to macrophage extracts. Other non-muscle tissues, particularly platelets and fibroblasts, also contain filamin and furthermore these non-muscle filamins are phosphoproteins* (20).

Comparison of Filamin and Spectrin
-Several lines of evidence suggest that filamin and spectrin are unrelated proteins. Wang et al. (9) have demonstrated the differences in electrophoretic mobility of chicken gizzard filamin and human spectrin and we have observed similar differences with mammalian smooth muscle and non-muscle filamins.
Furthermore, anti-filamin antibodies do not cross-react with human erythrocyte spectrin. Also studies with anti-spectrin antibodies (28, 29) have failed to detect spectrin in tissues rich in filamin such as platelets and fibroblasts. Therefore, the evidence seems to suggest that although filamin and spectrin are both high molecular weight proteins that interact with actin, they are unrelated molecules.

Variation in Filamin Distribution
-The content of filamin in different cell types and tissues is quite variable. Thus, filamin is a major constituent of smooth muscle and of platelets and is present in substantial amounts in macrophages and fibroblasts. However, filamin is either completely absent from or present in very low amounts in other tissues such as brain, skeletal and cardiac muscle and erythrocytes. Since the function of filamin in cells is unclear, the significance of these differences is not yet understood. Any conclusion on the role of filamin in cell function will have to take into account this marked tissue specificity.