Communication Vol. 255, No. 22, lasue of November 25, pp. 10551-10554, 19Bo THE JOURNAL OF BIOLOGICAL CHEMISTRY

Printed in U.S.A.

Calmodulin-binding Proteins of the Microfilaments Present in Isolated Brush Borders and Microvilli of Intestinal Epithelial Cells*

(Received for publication, September 3, 1980) John R. Glenney, Jr.4 and Klaus Weber From the Max-Planck Institute for Biophysical Chemistry, 0-3400 Gottingen, West Germany

Isolated microfilament cores of intestinal microvilli are known to contain actin and four major associated proteins among which is calmodulin. Immunofluores- cence microscopy reveals that calmodulin is present in the microvilli prior to biochemical fractionation of in- testinal cells and thus is not bound artifactually during the isolation procedure. Identification of the major mi- crovillus calmodulin-binding protein was achieved by the use of an [‘26TJcalmodulin gel overlay technique. Proteins of microvilli or brush borders were separated by sodium dodecyl sulfate-polyacrylamide gel electro- phoresis. After removal of sodium dodecyl sulfate, di- rect binding of radiolabeled calmodulin to the sepa- rated polypeptides was assayed by autoradiography. Three calmodulin-binding polypeptides are detected in brush borders. Two polypeptides (apparent M, = 280,000 and 140,000) show Ca2+-dependent binding, whereas the third polypeptide (Mr = 110,000) can bind calmodulin in the presence or absence of Caz+. Micro- villus core filaments contain only the latter species. Microvillus cores treated with 25 l l z ~ M%+ retain cal- modulin and the 110,000 polypeptide, whereas the other two associated proteins are greatly reduced, consistent with the hypothesis that the 110,000 protein is the major calmodulin-binding protein of the core filament structure. We discuss the currently documentable structure of the core filaments and evaluate the general usefulness of the calmodulin gel overlay technique.

Local fluxes in the intracellular free Ca2+ concentration are thought to modulate many cellular events. One of the most important cytoplasmic Ca2+ receptors is calmodulin, the small, acidic, and ubiquitous Ca2+-binding protein regulating a vari- ety of key enzymatic activities (1). Recently, Howe et al. (2) reported that calmodulin is a major component of the isolated microfilament core of intestinal microvilli and we have con- firmed their finding (3). Calmodulin is one of the four major F-actin associated proteins found in the purified membrane- free core filaments (2, 3). It can be reversibly removed by the phenothiazine derivative trifluoperazine without visibly af- fecting the core filament structure (3) and does not bind to F- actin itself (2).

* 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.

$ Recipient of Postdoctoral Fellowship Grant PF-1819 from the American Cancer Society.

Given the simple protein composition of the core filaments we aimed at the identification of the calmodulin-binding pro- tein(s) present in this structure. Here we present evidence suggesting that the 110,OOO protein assumed to be responsible for the attachment of the core filaments to the inner side of the plasma membrane (4) is a calmodulin-binding protein. By developing a [‘251]calmodulin gel overlay technique we show that this protein binds calmodulin in the presence and absence of Ca2+ which is in agreement with the characteristics of calmodulin binding in isolated core filaments (2,3). Addition- ally, by using this technique we detect at least two terminal web polypeptides, which, however, bind calmodulin only in the presence of Ca2’.

MATERIALS AND METHODS

Immunofluorescence Microscopy-Mouse intestinal epithelial cells were isolated and processed for indirect immunofluorescence micros- copy as described (5), using calmodulin-specific rabbit IgGs purified by antigen-affiity chromatography (6).

Isolation of Brush Borders, Microvilli, and Core Filaments- Chicken brush borders were isolated using a sucrose step gradient (7). However, such brush border preparations were often contaminated by a small number of nuclei which could be effectively removed by a second flotation gradient. The brush border preparation was sus- pended in 63% (w/v) sucrose, overlayed with 50% (w/v) sucrose, and centrifuged for 1 h at 17,000 rpm in a Beckman SW-27 rotor. After this second centrifugation, nuclei could not be detected by phase microscopy, and SDS’ gels revealed only marginal amounts of his- tones indicative of a very low degree of contamination by chromatin. Microvilli were prepared (7) from these specially purified brush borders and from brush borders available after the first sucrose gradients. Microvilli were demembranated in 1% Triton X-100 for 15 min at room temperature and treated with buffers containing CaZ+ or Mg” as described (3). Solution I was used to prepare brush borders and fractions derived from them (4).

Iodinated Calmodulin Gel Overlay Technique-All treatments were at room temperature unless stated otherwise. Slab gels (I-mm thick) were fixed in 40% methanol, 10% acetic acid for 30 min, rinsed briefly in distilled water, and further incubated in 10% ethanol for 2 h or overnight to remove SDS. Gel slices were then incubated with shaking first in 0.1 M imidazole, pH 7.0, for 10 min and then in Solution G (20 nm imidazole, 0.2 M KCl, 0.5% gelatin, 0.02% NaNa, pH 7.0) containing either 1 mM CaCL, 1 mM EGTA, or 1 mM CaC12 plus 0.1 M chlorpromazine for 10 min. Gel slices were now incubated with gentle shaking at 37OC for 4 to 8 h in the above solutions containing 2.5 pCi of ‘251-labeled calmodulin/ml. Unbound calmodulin was removed by shaking in a large volume of Solution G (five changes) over a 2-day period. Gels were dried and subjected to autoradiogra- phy.

Miscellaneous Methods-Homogeneous bovine brain calmodulin (6) and villin (8) were isolated as described. Histones were purchased from Sigma. Calmodulin was iodinated by the method of Bolton and Hunter (9) to a final specific activity of 2 mCi/mg of protein. The labeled protein, analyzed by SDS-PAGE and autoradiography, co- migrated with unlabeled calmodulin, and no additional bands could be detected even in heavily overexposed autoradiograms. Slab gel electrophoresis was performed using the buffer system of Laemmli (10). Sample buffer contained 10 mM EGTA. Negative staining and electron microscopy was as in Ref 8.

RESULTS

Previous reports (2, 3) have demonstrated that calmodulin is associated with isolated microvilli core filaments. In order to assess the cellular distribution of calmodulin and to explore

The abbreviations used are: SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; EGTA, ethylene glycol bis(P- aminoethyl ether)-N,N,N‘,N’-tetraacetate.

10551

10552 Microcillus Calmodulin

FIG. 1. Immunofluorescence microscopic localization of cal- modulin in intestinal epithelial cells. Mouse intestinal cells were processed for indirect immunofluorescence microscopy using calmod- ulin-specific IgGs. A, fluorescence micrograph; B, same field using phase contrast optics. Note strong staining of brush border.

;; "

A B C D E F G H FIG. 2. SDS-PAGE analysis (6 to 20% gradient gel) of chicken

brush border microvillus and microvillus core polypeptides. Brush borders were isolated by normal sucrose gradient centrifugation ( F ) and microvilli ( E ) and microvillus core filaments (D) prepared from these brush borders. Brush borders were subjected to an addi- tional flotation gradient (C) and microvilli ( B ) and core filaments ( A ) prepared from these. G represents intestinal nuclei and H a r e histone standards. Note the loss of histone contamination after the flotation gradient. Microvillus core proteins indicated on the left are the 110,OOO polypeptide (IZOK). villin (VI. fimbrin (F), actin (A) , and calmodulin (C).

the possibility that calmodulin may be redistributed during brush border or microvilli isolation, intact epithelial cells were processed for indirect immunofluroescence microscopy using antibodies to calmodulin. As demonstrated in Fig. 1, very strong specific calmodulin fluorescence is observed through- out the brush border part of the cells. Since fluorescent staining of antigens present only in the terminal web displays a distinct crescent-shaped fluorescence which stops at the terminal web/microvillus junction ( 5 ) , the images suggest that the majority of the calmodulin is present in the microvilli of the intact isolated cells. Thus the biochemical characterization of calmodulin in isolated microvilli core filaments seems not to result from a redistribution during cellular fractionation.

Chicken brush border preparations are often contaminated by varying amounts of nuclei. Only a small amount of contam- inating nuclei observed light microscopically results in a large amount of contaminating chromatin as revealed by the histone bands seen in SDS gels (Fig. 2). Microvilli derived from such brush border preparations still contain a significant amount of histones even though nuclei cannot be detected by light microscopy. Brush borders which have been processed

TABLE I Release of micrordlus coreproteins by treatment with Ca2' or

Isolated microvilli core filaments in Solution I (75 mM KCI, 1 mM EGTA, 0.1 mM MgC12, 10 mM imidazole, pH 7.3) were resuspended and aliquots were added to Solution I containing the above additions at the final free concentration listed. Total protein concentration was 1.2 mg/ml. After incubation (15 min a t room temperature), the suspension was centrifuged 10 min a t 12.000 X g (Eppendorf centri- fuge). Pellets were dissolved directly in SDS sample buffer after brief sonication. Supernatant proteins were precipitated in 10% trichloro- acetic acid for 1 h at 4"C, redissolved in the same volume of SDS sample buffer as the pellet fraction, and neutralized with Tris base. Equivalent volumes of supernatant and pellet fractions were analyzed on 6 to 15% gradient slab gels. stained with Coomassie blue, and quantitated by densitometry.

Mg2+

Amount of protein found in low meed oellet Treatment

I I O . 0 0 0 Villin Fimbrin Actin Ca*modu- lin 7

Control >95 90 >95 85 >95 10 mM EGTA >95 90 >95 a5 >95 10-5 Ca?' 85 15 <5 45 65 5 mM Ca" 85 15 t 5 45 65 25 mM MgCIz >95 40 25 80 >95

FIG. 3. Negative stain analysis of microvillus core filaments treated with 25 mM MgClz as described in the legend to Table I (x 50,000).

through an additional flotation gradient step are essentially nuclei-free and histone-contamination is greatly reduced (Fig. 2). Microvilli or core filaments prepared from these purified brush borders reveal only a very low level of histones (Fig. 2), yet their calmodulin content as well as the ratio of the other major microvillus proteins remains unchanged.' This rela- tively variable amount of histone contamination of isolated microvilli and their core filaments explains the report by Matsudaira and Burgess of the presence of significant amounts of polypeptides with molecular weights of 12,000 to 16,000 being present in their microvilli preparations (4). Other re- ports document gels in which these bands are greatly dimin- ished indicative of a much lower chromatin contamination (see for instance Ref. 2).

Calmodulin is stably bound in the isolated microvillus core filament organization even when treated with high concentra- tions of EGTA (2, 3) (see also Table I). However, when microvillus cores are treated with micromolar concentrations of free Cat+, the core structure rapidly disassembles, yet two- thirds of the calmodulin along with greater than 80% of the 110,000 protein are sedimentable in the residual actin pellet obtained by low speed centrifugation (Table I and Refs. 2 and 3). This suggests that these two proteins may be associated, since F-actin itself does not bind calmodulin (2). Treatment of microvilli cores with high concentrations of Mg2+ (25 to 50 mM), which is known to give rise to actin paracrystals in vitro (11) and to enhance the visualization of membrane to core filament cross-bridges (12) on microvilli as judged by electron microscopy, causes a strong diminution of the accessory pro- teins villin and fimbrin. Under these conditions villin is re- duced by 60% and fimbrin by 75%, although all of the calmod- ulin and 110,000 protein remains sedimentable in a structure which resembles the tightly bundled F-actin filaments of the core structure (Fig. 3), again indicating that most of the

' J. R. Glenney, Jr . and K. Weber, unpublished observations,

Microvillus Calmodulin 10553

M.W X 10-3

A B C

- -0

D E F

" 0

G H I

FIG. 4 (lefr). Identification of calmodulin-binding polypep- tides from brush borders, microvilli, and microvilli cores (cal- modulin overlay technique). Brush borders (A, D, G, and J) microvilli (B , E , H, and K ) , or microvilli core filament (C, F, I, and L ) peptides were separated by SDS-PAGE (6 to 20%). A single gel containing these samples was cut into 4 strips. One strip (A to C ) was stained for proteins with Coomassie dye and the other 3 were sequen- tially (1) futed in 40% methanol, 10% acetic acid (2). incubated in 10% ethanol (31, washed in 0.1 M imidazole, pH 7.0, and (4) rinsed in Solution G comprised of 10 mM imidazole, 0.2 M KCI, 1 mM MgC12, 0.5% gelatin, 0.02% NaNn. pH 7.0. The gel strips were then incubated in Solution G containing 1 mM CaCL (D to F ) , 1 mM EGTA (G to I), or 1 mM CaC12 plus 0.1 mM chlorpromazine (J to L ) . ['p51]Calmodulin was added to each, incubated at 37°C for 4 h, the unbound calmodulin

calmodulin may be bound to the 110,OOO protein present in the core structure.

Direct evidence for an association between calmodulin and the 110,OOO protein comes from experiments in which the separated polypeptides present in SDS gels were subjected to a gel overlay technique using radioactively labeled calmodulin. In these experiments, binding of calmodulin to the 110,OOO polypeptide is observed both in the presence or absence of free Ca2'. Binding is abolished in the presence of 0.1 mM chlorpromazine which is known to interact with calmodulin (14) (Fig. 4) and acts as a phenothiazine derivative able to remove calmodulin from isolated core filaments in vitro (3). In addition, two other calmodulin-binding polypeptides (Mr = 250,000 and 140,000) are observed when whole brush borders rather than microvilli are assayed. This suggests that these two proteins are present in the underlying terminal web microfilaments. Their binding of calmodulin is Ca"-depend- ent and in this respect similar to the behavior of the calmod- ulin-binding polypeptides detected in chicken gizzard and chicken brain (Fig. 5). Since brush borders are normally isolated in EGTA-containing buffers, one would expect the two terminal web proteins not to be charged with calmodulin in the standard preparation. Thus most of the calmodulin present in isolated brush borders seems to be bound via the 110,OOO protein of the microvilli core filaments.

DISCUSSION

The gel overlay technique with radioactively labeled cal- modulin performed on separated polypeptides can be expected to be of general biochemical use. The convenient procedure performed in its current form requires a t least partial rena- turation of an originally SDS-denatured polypeptide and is therefore not quantitative, and indeed may miss calmodulin- binding proteins which are present at either very low concen- trations or which renature only very poorly. Another potential shortcoming is the possible nonspecific interaction of a rather

J K L

25* - "

was washed off with the above solutions, and the specifically bound calmodulin was visualized by autoradiography. The microvillus core filament peptides are labeled as in Fig. 2. Small arrows indicate top and bottom of gels used for autoradiography.

FIG. 5 (right). Binding of ['2KI]calmodulin to chicken brain and gizzard polypeptides. Standard acetone powders of chicken gizzard and brain were dissolved in SDS sample buffer and polypep- tides of gizzard (A, C, and E ) and brain (B , D, and F ) separated by SDS-PAGE (10% gels). The gel overlay technique with ["'I]calmod- ulin was performed as described in the legend to Fig. 4. Coomassie stain ( A and B ) , incubation in the presence of 1 mM CaCI2 (C and D), or 1 mM EGTA ( E and F ) . Note the Cap'-dependent binding of calmodulin to gizzard (130,000 and 190,OOO) and brain (47,000 and 56,000) polypeptides.

basic polypeptide with the rather acidic calmodulin.3 In spite of these problems, some of which may be controlled in the future, we have identified in chicken gizzard and brain similar polypeptides as major calmodulin-binding proteins as Grand and Perry (13) using the more cumbersome urea gel filtration procedure. In addition, in agreement with their results, these polypeptides recognized show only Ca2+-dependent calmodu- lin binding.

In the case of the 110,OOO polypeptide of microvillus core filaments, calmodulin binding in the overlay assay occurs both in the presence and absence of Ca", consistent with the known binding features revealed on intact microvilli in two independent studies (2, 3). Furthermore, as in the core fila- ments, binding is abolished by the presence of a thiophenazine derivative known to bind to calmodulin (14). The assignment of calmodulin binding to the 110,OOO polypeptide is independ- ently corroborated by the characterization of the core filament derivative obtained a t 25 to 50 mM Mg'. Although the villin and fimbrin content has greatly diminished, the complement of calmodulin and the 110,OOO polypeptide is retained. A further argument for this relation is the information detecta- ble in the gels of Matsudaira and Burgess (4) that ATP- induced loss of 110,OOO protein from core filaments is accom- panied by a corresponding loss of a 16,500 polypeptide mean- while identified as calmodulin (2, 3).

Although all data point to the 110,OOO protein as the cal- modulin-binding protein, previous stoichiometric data suggest more than one site (2,3). Since we have been unable to detect any binding to the other associated proteins, i.e. fimbrin (15) or villin (8), either in the absence or presence of Ca" or F- actin (2), the 110,000 protein may bind more than one cal- modulin molecule in agreement with the previously noted complex release pattern of calmodulin from the isolated cores (2, 3). Because the 110,OOO protein has so far defied attempts of solubilization and isolation, more detailed studies on its interaction with calmodulin have to await further progress.

."J. R. Glenney, Jr. and K. Weber, unpublished results.

10554 Microvillus Calmodulin

The emerging picture of the intestinal microvillus core has been advanced by this study. For the four associated proteins present in addition to pardely oriented F-actin (12) we begin to understand certain molecular interactions. The 110,OOO protein is most likely the cross-bridge to the plasma mem- brane (4 ) , and villin (95,000) is an F-actin bundling protein responsible for tight bundling at low Ca2+ concentrations (8). Vain’s transition to an F-actin severing protein at micromolar Ca2+ (8) is in line with its Ca2+-binding constant (3) and the documented Ca2+ sensitivity of the core (2, 3). Calmodulin may now be assigned to the cross-bridges toward the plasma membrane provided by the 110,OOO protein and could be acting beyond its usually perceived enzyme regulatory func- tions (1) as a Ca2+-buffering system protecting core filament integrity (3). Very little is known about the fourth major associated protein, fimbrin, which is a general microfilament- associated protein present also in various cellular protrusions of cultured cells (15). Further progress along the lines of specific interaction between the five major proteins of the core filament should provide a molecular model of this microfia- ment organization underlying the plasma membrane.

Acknowledgments-We thank W. Koch for technical assistance

and Dr. Anthony Bretscher for assistance with immunofluorescence microscopy.

REFERENCES 1. Cheung, W. Y. (1980) Science 207, 19-27 2. Howe, C. L., Mooseker, M. S., and Graves, T. A. (1980) J. Cell

3. Glenney, J. R., Bretscher, A., and Weber, K. (1980) Proc. Natl.

4. Matsudaira, P. T., and Burgess, D. R. (1979) J. Cell Biol. 83,667-

5. Bretscher, A., and Weber, K. (1978) J. Cell Biol. 79, 839-845 6. Andersen, B., Osborn, M., and Weber, K. (1978) Cytobiologie 17,

7. Bretscher, A., and Weber, K. (1978) Exp. Cell Res. 116,397-407 8. Bretscher, A., and Weber, K. (1980) Cell 20,839-847 9. Bolton, A. E., and Hunter, W. M. (1973) Biochern. J. 133, 529-

Biol. 86,916-923

Acad. Sci. U. S. A. in press

673

354-361

539 10. Laemmli, U. K. (1970) Nature (Lond.) 227,680-684 11. Hanson, J. (1973) Proc. Roy. SOC. Lond. B Biol. Sci. 1 8 3 , 3 9 4 5 12. Mooseker, M. S. (1976) in Cell Motility (Goldman, R., Pollard,

T., and Rosenbaum, J., ed) Vol. B, pp. 631-650, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York

13. Grand, R. J. A., and Perry, S. V. (1979) Biochern. J. 183, 285-295 14. Levin, R. M., and Weiss, B. (1977) Mol. Pharrnacol. 13.690-698 15. Bretscher, A., and Weber, K. (1980) J. Cell Biol. 86,335-340