Localization of Myosin, Actin, and Tropomyosin

in Rat Intestinal Epithelium :

I mm unohistochem ical Studies at the Light

and Electron Microscope Levels

DETLEV DRENCKHAHN and UTE GRÖSCHEL-STEWARTDepartment of Anatomy, Universität Kiel, D-2300 Kiel, Federal Republic of Germany, and Departmentof Zoology, Technische Hochschule Darmstadt, D-6100 Darmstadt, Federal Republic of Germany

ABSTRACT

Myosin, tropomyosin, and actin were localized in the epithelial cells of rat intestineby means of specific antibodies to chicken gizzard smooth muscle myosin, tropomyosin, andactin by immunohistochemical studies at both the light and electron microscope levels(unlabeled antibody enzyme technique) . The pattern of antibody staining was the following :(a) Anti-actin was associated with the microfilament bundles of the microvilli in their entirelength, as well as with the microfilament network in the terminal web . (b) Anti-myosin wasconcentrated along the rootlets of the microvillar microfilament bundles and within thefilamentous feltwork forming the terminal web. (c) Anti-tropomyosin showed a distributionsimilar to that of anti-myosin . In addition, the three antibodies also labeled the subplasmalem-mal web underneath the cell membrane bordering on the basal lamina . Utilizing the aboveultrastructural findings, we wish to propose a functional model of microvillar contraction .

In recent years much interest has been focused on the force-generating apparatus responsible for the motility of intestinalepithelial brush border . This highly ordered structure provedto be an ideal model for studying the structural basis ofnonmuscle cell motility .By phase-contrast microscopy, epithelia of rat intestine and

of kidney tubules were shown to perform fast microvillarmovements (31, 35) . Studies on isolated apical segments ofintestinal epithelium demonstrated rapid microvillar retractionor contraction of the whole apical segments in response toATP, Caz+, and Mgt + (26, 30) . Biochemical analysis indicatedthe presence of myosin, tropomyosin, actin, and other associ-ated proteins in the apical cytoplasm (3, 26, 28) . Recently, thesebiochemical findings have been confirmed by the immunoflu-orescent localization of myosin, tropomyosin, and actin in theapical cytoplasm ofmouse, rat, and chicken intestinal epithelialcells (4, 13, 14, 28).

Ultrastructurally, the microvilli of rat intestinal epitheliumcontain an orderly axial bundle of actin filaments in parallelalignment, which penetrates -0.5-1 pin into the apical cyto-plasm to form a rootlet, which terminates abruptly (5) . Asjudged by heavy meromyosin binding, the actin filaments are

Tne JOURNAL Of CELL Bioro(,y " Voruti+e 86 AUGUST 1980 475-482©The Rockefeller University Press - 0021-9525/80/08/0475/08 $1 .00

attached to the apical membrane of the microvilli with thesame polarity seen in actin filaments attached to the Z line ofstriated muscle (l, 23, 27) . The apical cytoplasm contains adense network ofactin filaments (terminal web) and individualshort myosin-like filaments, which appear to be associated withthe rootlets of the microvillar filament bundles (27, 30) . On thebasis ofthese findings, it has been suggested that movement ofthe entire microvillar filament bundle could be effected byinteraction of myosin with the rootlet filaments (26, 27, 30).

In the immunocytochemical study presented here, we com-bined light and electron microscopy to obtain more detailedinformation on the distribution of myosin, tropomyosin, andactin in the brush-border region of rat intestine .

MATERIALS AND METHODSPreparation and Specificity of AntibodiesThe antigens myosin, tropomyosin, and actin were extracted from the muscle

layer of chicken gizzard and were purified as described elsewhere (8, 18, 19) . Anadditional chromatography step of myosin on hydroxyapatite (29) has recentlybeen added . Control antigens (myosin and tropomyosin from chicken striatedpectoral muscle) were prepared in the same manner. Antisera to the abovecontractile proteins were raised in rabbits as previously described (8, 18, 19) . The

475

on April 10, 2019jcb.rupress.org Downloaded from http://doi.org/10.1083/jcb.86.2.475Published Online: 1 August, 1980 | Supp Info:

immunoglobulin fraction of the immune sera and pre-immune controls wereprepared according to Harboe and Ingild (20). SDS polyacrylamide gel electro-phoresis (10% gels) separated such fractions into IgG heavy and light chains andno impurities were noted.

Anti-gizzard myosin was shown to be specific for smooth muscle myosin byimmunodiffusion, ATPase inhibition test, and immunofuorescence (I8) . It re-acted also with nonmuscle myosin extracted from rat corneal epithelium (11).Anti-gizzard actin reacted with both smooth and striated muscle actin in immu-nodiffusion and immunofluorescence . It was shown to inhibit the actin-activatedMgt. myosin ATPase activity (l9). Anti-gizzard tropomyosin wasshown to reactin immunodiffusion with smooth and striated muscle tropomyosin but not withmyosin (8) . All three gizzard antibodies reacted in immunofluorescence also withvarious nonmuscle cells (14, 17) . Their different affinities towards 3T3 cells,differentiated and dedifferentiated smooth muscle cells, fibroblasts, and endothe-lial cells were used as a tool to discriminate between these cell types in tissueculture (6-9, 17) . Antibodies to striated muscle myosin and tropomyosin did notshow any reaction with the corresponding smooth muscle and nonmuscle anti-gens, by immunodiffusion, immunofluorescence, and ATPase inhibition tests (17,18). and were thus used as controls.

ImmunocytochemistryImmunocytochemical staining was performed before embedding (pre-embed-

ding staining) using the unlabeled antibody enzyme technique according toSternberger (for review, see reference 33). All attempts to localize myosin, actin,and tropomyosin on thin plastic sections (postembedding staining) of unfixed(freeze-substituted) and fixed (2% formaldehyde in phosphate-buffered saline[PBS], no osmication) tissue, using the procedures ofSilverman (32) and Booyseet al. (2), were unsuccessful .

Young adult rats of either sex (Wistar Hannover strain) were anaesthetizedwith ether, the upper half of the small intestine was removed, cut longitudinally,and washed briefly with isotonic PBS, pH 7.2. Strips of this tissue were treated intwo ways: (a) Fixation with a solution containing 0.1% glutaraldehyde and 2%paraformaldehyde in PBS (pH 7.2) for 15 min or (b) exposure ofunfixed tissuefor l-5 min to 0.05-0.1% (vol/vol) Triton X-100 (previously filtered throughAmberlite MB-1, Serva Feinbiochemica, Heidelberg, W. Germany) or to 0.05%Nonidet P40 (Fluka A.G ., Basel, Switzerland) in PBS (containing 0.2 mM Mg2.)

followed by fixation as in a . After fixation, all pieces were rinsed for 30-90 minwith PBS (three changes) and were cut with razor blades or with a vibratingmicrotome (Vibratome, Oxford Laboratories, Foster City, Calif) into 100- to500-um-thick tissue slices. To further facilitate antibody penetration into theepithelial cytoplasm, pieces of tissue processed as described above were crushedwith forceps to obtain fragmented epithelial cells.

Tissue slices and crushed tissue were incubated at 4°C with antibodies andsera in the following sequence : (a) Normal inactivated goat serum (NordicImmunological Laboratories, Tilburg, The Netherlands, 1 :5, 10 min) followed bya 30-min wash with PBS (three changes); (b) antibodies to gizzard myosin (2 ug/ml), tropomyosin (10 lag/ml), and actin (10 lag/ml) (14 h), followed by washeswith PBS (10 h, three to five changes) ; (c) goat anti-rabbit IgG (Miles Labora-tories, Inc., Elkhart, Ind., 0.3 mg/ml, 14 h); wash with PBS (10 h, three to fivechanges); (d) rabbit horseradish peroxidase anti-horseradish peroxidase (PAP)complex (Dako Immunoglobulins, Copenhagen, Denmark, No . Z 113/087, I :100, 14 h) . All antibodies were diluted with PBS containing normal inactivatedgoat serum in a concentration of 1 :50. After a final wash with PBS (6 h), tissueslices were incubated for 5 min in 100 ml Tris-HCI buffer (0 .05 M, pH 7.2)containing 12 .5 mg diaminobenzidine (Fluka) and 20 ul 30% H-20 2 (Perhydrol,Merck Chemical Div., Merck &Co ., Inc., Rahway, N. J.). After a final rinse withdistilled water (10 min, two changes), tissue slices were postfixed with unbuffered2% Os0, for 30 min, dehydrated in a graded ethanol series, and embedded inAraldite . Unstained semithin (1 um) sections were examined with the lightmicroscope, and thin (50-100 nm) sections (no heavy metal counterstain) wereviewed with a Philips 300 electron microscope. Some of the thin sections werecounterstained with uranyl acetate and lead citrate to allow examination of theoverall cellular ultrastructure and the distribution of filaments.

RESULTS

Effect of the Fixation Procedure and DetergentTreatment on the Ultrastructure and on theImmunocytochemical Staining Patterns

Fixation, detergent treatment, and the immunocytochemicalstaining procedure did not alter the pattern of thin (5-8 nm)and intermediate (10 nm) filaments in the apical part of theintestinal epithelium . However, exposure to detergents led to

476

THE JOURNAL OF CELL BIOLOGY " VOLUME 86, 1980

considerable changes in the structure of the cytomembranes .Swelling ofmitochondria and the formation ofnumerous smalland large vacuoles were noted as first signs of detergent pene-tration . In the more progressive stages, the structure of themicrovillar plasma membrane was affected, as noted by un-dulation of the membrane, appearance of demembranatedmicrovillar segments, and irregularly curved microvilli.More severe changes of the microvillar structure, such as

fragmentation and fmal loss of the microvilli were numerousin response to exposure to 0.1% Triton X-100, but were consid-erably reduced or missing in tissue treated with 0.05% TritonX-100 or Nonidet P 40 .

In fixed tissue not exposed to detergents, immunoreactivitywas only seen in a few cells scattered along the surface of thesection . The number of immunoreactive cells was very muchincreased by exposure of the tissue to detergents . The majorityof immunoreactive cells showed more or less pronounceddetergent-induced changes of the brush border (demembran-ated microvillar segments and irregular curving of microvilli) .Only few of the immunoreactive cells possessed rather well-preserved microvilli .

Control Experiments

In tissue samples treated as described above, no immuno-staining was observed when the various control antibodies wereapplied, such as anti-striated muscle myosin or tropomyosin,pre-immune Ig, or antibody adsorbed to the homologous an-tigen . This is documented in Fig . 1 .

Light MicroscopyIn 1-ttm Araldite sections oftissue slices incubated with anti-

smooth muscle myosin, specific immunoperoxidase stainingwas confined to a narrow band in the apical cytoplasm locatedbeneath the brush border. At higher magnification, this im-munoreactive zone was seen to be composed of intenselystained dots and streaks separated by less reactive spacings(Fig. 2) . In addition, a thin immunoreactive band was regularlynoted at the base ofthe epithelial cell, presumably representingthe basal subplasmalemmal web (Fig . 2) . Anti-actin stainedboth the microvilli of the brush border and a narrow band inthe apical cytoplasm (Fig . 3), as well as the basal subplasma-lemmal web region . The staining pattern of anti-smooth muscletropomyosin (Fig. 4) was similar to that of anti-myosin ; i .e ., adotted line within the terminal web region, a narrow band atthe base of the epithelial cell, and no detectable staining of themicrovilli .

Electron MicroscopyANTI-MYOSIN:

The interrupted staining pattern seen in lightmicroscopy corresponds to a tight packing of PAP complexesalong the rootlets of microvillar filament bundles (Figs. 5 and6) . In most preparations, the staining was too dense to distin-guish individual PAP complexes ; in less intensely stained root-lets, however, the immunocomplexes were seen to be arrangedalong the periphery ofthe filament bundles (Fig. 6, inset) . Thelabel terminated abruptly at the base of the microvilli, andthere was no association of myosin immunoreactivity with corefilaments, neither in completely demembranated microvilli norin microvilli with an ultrastructurally intact plasma membrane .Within the terminal web, the myosin-specific label was lessdensely packed .

ANTI-ACTIN : Anti-actin stained the entire microvillar fila-

FIGURE 1

Electron micrographs (no heavy metal counterstain) of the apex of intestinal epithelium incubated with the followingcontrol Igs using the unlabeled antibody peroxidase method : (a) anti-striated muscle tropomyosin, (b) anti-gizzard myosinpreviously absorbed with the antigen, (c) pre-immune IgG . Tissue was exposed to Triton X-100 (0 .05-0.1% for 5 min) beforefixation and immunocytochemical staining procedure . No immunoperoxidase label is seen . MV, microvilli ; R, rootlet of microvillarfilament bundles ; D, spot desmosome . Bar, 0.5 ,um . (a) x 46,000; (b and c) x 37,500 .

FIGUREs 2-4

Light micrographs of unstained 1-Rm-thick Araldite sections of intestinal epithelium incubated with anti-gizzardmyosin (Fig . 2), anti-actin (Fig . 3), and anti-tropomyosin (Fig . 4) before embedding (exposure to 0.1% Triton X-100 for 5 min beforefixation and immunostaining) . Bar, 10Rm . x 1,600 .

FIGURE 2 The myosin-specific label is concentrated in a small band within the apical cytoplasm below the microvilli and in asmall zone along the base of the epithelial cells (arrowheads) . Arrow points to interruptions of the apical immunoreactive zone .

FIGURE 3 Anti-actin stains microvilli as well as a narrow zone within the apical cytoplasm .

FIGURE 4

The distribution of tropomyosin-specific stain is identical to that seen with anti-myosin .

ment bundles from the tip down to the rootlets, and also theterminal web (Figs . 7-9) . There was a considerable variationin the staining intensity of both rootlets and terminal web,independent of the intensity observed in the microvilli of thesame preparation . These findings will be discussed below .ANTI-TROPOMYOSIN : Anti-tropomyosin stained the rootlet

filaments and the terminal web (Figs . 10 and 11) in a pattern

similar to that described for anti-myosin . Generally, the mi-crovillar core was devoid of PAP label when the tissue waspretreated with 0.05% TritonX-100 (Fig. 10) or 0.05% NonidetP 40 (not shown) . However, when tissue had been exposed to0.1% Triton X-100 for 5 min, the microvilli of many cellsshowed a rather weak, irregular staining along their entirelength (Fig . 11) . It appears that the more rigorous treatment

DRENCKHAFIN AND GR6SCHFF-STEWAR1

Brush-border Contractile Apparatus

477

FIGURE 5

Electron micrograph showing the distribution of gizzardmyosin immunoreactivity in a cell located at the cut edge of theVibratome section (no detergent treatment) . Immunoperoxidaselabel is confined to the microvillar rootlets (appearing as dots andstreaks in this oblique section) and the terminal web (arrows) . Notethe absence of myosin label in the remaining cytoplasm. Bar, 0 .5gm . X 10,200 .

FIGURE 6 Ultrastructural distribution of gizzard myosin-specificimmunoreactivity in the brush-border region (exposure to 0.1%Triton X-100 for 5 min before fixation and staining) . The PAPcomplexes (small arrows) are located along the rootlets (R) and inthe terminal web (TW) . Inset shows a cross-sectioned rootlet whichis surrounded by individual PAP complexes (arrows) and aggregatesof PAP complexes (asterisks) . Bar, 0.5 gm . x 45,000 . Inset: Bar, 0.1gm . X 125,000 .

with a higher concentration ofdetergent leads to a displacementof tropomyosin from its site in the terminal web or rootlets tothe microvilli .

Desmosomal Zone of the Apical Filament WebThe desmosomal or basal zone of the apical filament web

478

THE JOURNAL Of CELL BIOLOGY - VOLUMi 86, 1980

FIGURE 7 Low-power electron micrograph of a whole intestinalepithelial cell illustrating the distribution of actin-specific immu-noreactivity (same tissue preparation as in Fig. 6) . Immunostainingis confined to the microvilli (M), the apical terminal web, and thebasal subplasmalemmal web (BW), whereas the remaining cyto-plasm including the nucleus (N) is devoid of immunoreactivity . Bar,1 gm . X 7,500.

(5, 22) contains mainly intermediate filaments with only a fewthin filaments showing. Thus, none or very few immunocom-plexes specific for actin, myosin, or tropomyosin were seen inthis zone .

CytomembranesCytomembranes adjacent to the immunoreactive apical cy-

toplasm (membranes of vacuoles, mitochondria, and the

plasma membrane) usually displayed an irregular electron-dense staining (Figs . 6, 8-11), although the cytomembraneswere virtually devoid of attached immunocomplexes (PAP) .This strongly indicates that the staining of the cytomembranesis caused by a nonspecific binding ofthe lipophilic and osmio-philic peroxidase reaction product (a phenazine polymer), orig-inating from adjacent filament-bound PAP complexes.

Basal Subplasmalemmal Web

A delicate network of thin filaments (4-8 nm in diameter)

FIGURE 8 Anti-actin label is concentrated along the core of themicrovilli and their rootlets . The dotlike PAP complexes are clearlyvisible . In this preparation (0 .05% Nonidet P 40 before fixation andstaining), the terminal web (TW) is only weakly labeled. Bar, 0.5pm . X 37,000 .

was seen in the basal cytoplasm of the intestinal epithelium . Itwas located underneath the basal plasmalemma and was oc-casionally intermingled with some intermediate filaments (Figs .12 and 13) . This web of thin filaments was clearly distinguish-able from the intermediate filament bundles traversing thebasal cytoplasm or running parallel to the plasmalemma . Inthe basal cytoplasm, the antibodies to contractile proteins weremainly associated with this subplasmalemmal web (Figs . 7 and14, see also Fig. 2), but not with the intermediate filamentbundles .

DISCUSSIONThe microfilament organization in the brush border of intes-tinal epithelial cells has previously been characterized by im-munofluorescence microscopy using antibodies to various con-tractile proteins (4, 13, 28) . The limited resolution of lightmicroscopy, however, demanded the use ofthe electron micro-

FIGURE 9 Portion of the apical cytoplasm illustrating the actin-specific label within the terminal web (same processing as in Fig . 8) .Arrows point to PAP complexes, some of which exhibit a ringlikesubstructure. Bar, 0.5 ftm . X 46,000.

FIGURE 10

Distribution of gizzard tropomyosin immunoperoxidase label . The immunocytochemical staining pattern is similar tothat obtained with anti-gizzard myosin (labeling of rootlets and the terminal web) . Tissue was exposed to 0.05% Triton X-100before fixation and staining . Bar, 0.5lxm. X 46,000.

FIGURE 11

Irregular anti-tropomyosin staining of microvilli in demembranated brush border exposed to 0.1% Triton X-100 for 5min before fixation and staining . Bar, 0.5 pm . X 35,000 .

DRENCKHAHN AND GROSCHEL- STEWART

Brush-border Contractile Apparatus

479

FIGURE 12

Basis of intestinal epithelium incubated with pre-immune Ig (B, basal lamina) . Asterisks indicate the subplasmalemmallayer of interwoven thin filaments measuring 4-8 nm in diameter. Small arrows point to cross-sectional profiles of individual thinfilaments; larger arrows point to bundles of intermediate filaments (8-10 nm). Uranyl-acetate and lead citrate counterstain . Bar, 0 .2gm . x 83,000 .

FIGURE 13

Oblique section of the basis of intestinal epithelium incubated with pre-immune Ig (B, basal lamina) . The subplas-malemmal web of thin filaments extends into a tangentially sectioned basal process (B, basal lamina) . Arrows point to individualintermediate filaments projecting into the subplasmalemmal web. Uranyl acetate and lead citrate counterstain . Bar, 0 .2 gm . x83,000 .

scope in conjunction with highly specific antibodies. Thistechnique requires considerable compromises in the handlingof the tissue to be examined, in order to guarantee both optimalpenetration of antibody and maximum preservation of theantigenic structure.

In the study presented here, the unlabeled antibody enzymemethod was chosen as the most satisfactory technique. Onlyimmunological bonds were used here to attach the detectormolecule (horseradish peroxidase) to the antigenic site in thetissue section . Covalent labeling of the antibodies, which maysignificantly impair their specificity, is thus avoided (for review,see reference 33). In addition, the PAP method exceeds thelabeled antibody techniques in sensitivity by several orders ofmagnitude (34) . Lastly, the cyclic structure of the PAPcomplexconfers high stability to the detector molecule and allows thedirect visualization of the 20- to 30-nm diameter rings or dotsin the electron microscope . These characteristic structures areeasily distinguishable from the nonspecific electron-dense per-oxidase reaction products often deposited near the site of

480

Till JOURNAL or CM BioIUGV - Votuml 86, 1980

reaction ; and they were never observed when the first incuba-tion step was performed with Ig fractions derived from pre-immune sera, from specific antibodies exhaustively absorbedto the homologous antigen, or from heterologous non-cross-reacting antibodies (e .g . anti-striated muscle myosin and tro-pomyosin). With the specific antibodies' to smooth musclemyosin, actin, and tropomyosin the immunoreactivity wasconsistently confined to the apical and basal cytoplasm, irre-spective whether tissue sections or fragmented epithelial cellswere used, and the patterns closely resembled those seen inimmunofluorescence (4, 13).

In the electron micrographs, however, we sometimes ob-

' It should be mentioned here that three different antisera to actin andmyosin have been used with identical results . Anti-myosins wereprepared to conventionally purified myosin and to hydroxyapatitechromatographed myosin . The IgG fraction of one of the myosinantibodies was separated by affinity chromatography with Sepharose-6B-fixed myosin .

FIGURE 14

Infranuclear portion (N, nucleus) of an intestinal epi-thelial cell incubated with anti-gizzard tropomyosin (5 min exposureto 0.05% Triton X-100 before fixation and staining) . Note immuno-staining of the basal subplasmalemmal web (arrows) . Bundles ofintermediate filaments (IF) are not stained . Bar, 0 .5 pin . x 46,000 .

served some irregularities in the staining pattern, such asfollows :

(a) In some cells the anti-actin stain was confined to themicrovilli and was practically absent in the terminal web (Fig .8) . This paradox or false negative staining may be caused bydiffusional barriers . Thus, one might postulate that the inter-microvillar plasma membrane seems to be more resistant todetergent treatment than the microvillar plasma membrane.

(b) Displacement of antigenic sites may cause false positivestaining patterns . The occurrence of irregular anti-tropomyosinstaining along many microvilli after exposure to 0.1% TritonX-100 (but not in sections treated with 0.05% detergent or inuntreated sections) could possibly be explained this way.

Being aware of the special problems posed by immunoelec-tron microscopy, we nevertheless believe that our ultrastruc-tural studies well supplement and extend the previous immu-nocytochemical results of the localization ofmyosin, actin, andtropomyosin in the intestinal epithelium . Actin, as expected,was localized in the microvilli and the terminal web . Myosinwas strictly confined to the terminal web region, where indi-vidual immunocomplexes were seen along the microvillar root-lets, accounting for the interrupted pattern noted in lightmicroscopy . The similarity of the anti-tropomyosin stainingpattern to that of myosin is striking, and one is tempted toquestion whether its role is solely that of a stabilizing factor inactin filaments (24) .On the basis of our ultrastructural immunocytochemical

observations and the data presented by others (27, 30), we wish,at this point, to propose a model for the functional organizationof myosin, actin, and tropomyosin in the brush-border region(Fig. 15), which resembles that of Rodewald et al. (30) . Con-traction might possibly involve a simultaneous interaction ofsmall myosin aggregates with both the terminal web filaments

FIGURE 15 A model for the functional organization of myosin,tropomyosin, and actin in the brush-border region of intestinalepithelium . Contraction of microvilli is suggested to occur in twosimultaneous steps including : (1) actin-myosin interaction in theterminal web (T) in order to generate tension of the interwovenfilaments between the zonula adherens (Z) ; and (2) interaction ofmyosin with rootlet filaments (R) and the tightened terminal webfilaments . Tension of the terminal web filaments, which are assumedto serve as anchoring system for the myosin units, must be strongenough to overcome the rigidity of the microvillar membrane towhich the core filaments (C) are attached (A) . The polarity of actinfilaments (1, 23, 27) is indicated by arrowheads . D, desmosome .

and the rootlet filaments . The microfilaments of the terminalweb might serve as an anchoring system for the myosin units,preventing them from sliding up the rootlet filaments . Theterminal microfilament web, in turn, is thought to be restrainedfrom moving towards the cell surface by loops of intermediatefilament bundles attached to the spot desmosomes (5, 22) .Further stability might be conferred to the terminal web byinterconnections of the thin filaments with a-actinin. Thisprotein, which is known to be present in the Z lines of skeletalmuscle, has recently been shown by immunofluorescent andimmunoferritin labeling to be scattered throughout the termi-nal web (l0, 16). Cross links between a-actinin and the terminalweb filaments would permit changes in filament polarity andthis might provide a feltwork of interconnected actin filamentswith changing polarity .An isometric tension arising within the terminal web might

be strong enough to overcome the rigidity of the microvillarmembrane (to which the core filaments are attached), andmovement of the microvilli would follow . In the model pre-sented by us, the terminal web attached to the intercellularjunctions represents thepunctumfixum, and thepunctum mobileis represented by the microvillar filament-membrane complex .This hypothetical contractile mechanism might also explainthe ATP-induced microvillar retraction observed in isolatedcompletely demembranated brush-border regions (26) : the ter-minal web, detached from the junctional membrane, mightshrink during contraction until it becomes attached to its mostadjacent row of microvillar rootlets, which will then serve as asecondary or auxiliary punctum fixum for the detached web;the rigidity of which might then be strong enough to allowmicrovillar sliding . The centripetal force generated in the ter-minal web would cause the rootlets to approach each other andthe tips of the microvilli to spread . As a matter of fact, this isexactly the picture that is regularly seen in the isolated con-tracted brush border (26, 30) . This contraction and spreadingof microvilli is less readily explained by the interaction ofsplayed rootlet filaments of adjacent microvilli, as originally

DRENCKHAHN AND GROSCItrt-STEWART

Brush-border Contractile Apparatus

481

suggested by Mooseker and Tilney (27) . Their model wouldrequire an additional restraining system (i .e., the terminal web)to prevent myosin units from moving upward during contrac-tion . Moreover, one has to postulate a high cytoplasmic viscos-ity to explain why the demembranated microvilli are not pulledtogether (instead of moving down) during the interaction ofsplayed rootlet filaments .One should not forget, however, that all the proposed models

for microvillar contraction are highly speculative, as the actualmovements of the microvilli in vivo are still unknown to us . Inaddition, it should be kept in mind, that the contractile proteinsfound in the brush-border region might not only serve inmicrovillar contraction, but may also be involved in otherdynamic events, such as endo- or exocytosis. As it has beenrecently shown, contractile proteins are also concentrated inthe terminal web region of acinar and duct cells of variousexocrine glands and in liver cells (12, 14, 15, 2l).

Contractile elements are not confined to the brush-borderregion, as recently they have also been localized in the basalcytoplasm by immunofluorescence (see figures shown in ref-erences 4 and 13) . In the study presented here, we were able toshow that antibodies to myosin, actin, and tropomyosin willbind to a very delicate network of 4- to 8-nm filaments justbeneath the basal plasmalemma. It is our suggestion that thesecontractile elements may participate in the upward movementof epithelial cells from their original sites in the crypt ofLieberkiihn to the tip of the intestinal villi (25) .

The excellent technical assistance of Miss Renate Steffens is gratefullyacknowledged .

This investigation was supported by grants from the DeutscheForschungsgemeinschaft (Dr 91/2 and Ste 105/20) .

Received for publication

17December 1979, and in revised form1 April 1980.

REFERENCES

I . Begg, D. A., R. Rodewald, and L. I. Rebhun. 1978. The visualization of actin filamentpolarity in thin sections . Evidence for the uniform polarity of membrane-associatedfilaments . J Cell Biol. 79:846-852.

2 . Booyse, F . M ., L . A. Sternberger, D. Zschmke, and M . E. Rafelson . 1971 . Ultrastructurallocalization of contractile protein (thrombosthenin) in human platelets using an unlabeledantibody-peroxidase staining technique . J. Hisiochem . Cytochem. 19:540-550 .

3 . Bretscher, A ., and K . Weber. 1978. Purification of microvilli and an analysis of the proteincomponents of the microfilament core bundle. Exp. Cell Res. 116:397-407 .

4 . Bretscher, A., and K . Weber . 1978 . Localization of actin and microfflamem-associatedproteins in the microvilli and terminal web of the intestinal brush border by immunoflu-orescence microscopy . J Cell Biol. 79 :839-845 .

5 . Brunser, O., and J . H . Loft. 1970 . Fine structure of the apex of absorptive cells from ratsmall intestine . J. Ulimsiruct . Res. 31 :291-311 .

6 . Burkl, B., C . Mahlmeister, U . Gröschel-Stewart, J . Chamley-Campbell, and G . Campbell .1979. Production of specific antibodies to contractile proteins and their use in immunoflu-orescence microscopy. III . Antibody against human uterine smooth muscle myosin .Histochemisiry. 60 :135-143 .

482

THE JOURNAL OF CELL BIOLOGY - VOLUME 86, 1980

7 . Chamley, J . H ., U . Gröschel-Stewart, G . R. Campbell, and G . Burnstock . 1977 . Distinctionbetween smooth muscle, fibroblasts and endothelial cells in culture by the use of fluores-ceinated antibodies against smooth muscle actin. Cell Tissue Res . 177 :445-457 .

8 . Chamley-Campbell, 1 ., G . R . Campbell, U . Gröschel-Stewart, and G . Burnstock . 1977 .FITC-labelled antibody staining of tropomyosin-containing fibrils in smooth, cardiac andskeletal muscle cells, prefusion myoblasts, fibroblasts, endothelial cells and 3T3 cells inculture. Cell Tissue Res. 183 :153-166 .

9. Chamley-Campbell, 1 ., G. R . Campbell, and R. Ross. 1979 . The smooth muscle cell inculture. Physiol. Rev. 59:1-61 .

10. Craig, S . W., and J . V . Pardo. 1979 . Alpha-actinin localization in the junctional complexof intestinal epithelial cells . J Cell Biol. 80:203-210 .

11 . Drenckhahn, D ., and U. Gröschel-Stewart . 1977 . Localization of myosin and actin inocular nonmuscle cells . Immunofluorescence-microscopic, biochemical and electron-mi-croscopic studies . Cell Tissue Res. 181 :493-503 .

12. Drenckhahn, D., U . Gröschel-Stewart, and K . Unsicker. 1977 . Immunofluorescence-microscopic demonstration of myosin and actin in salivary glands and exocrine pancreasof the rat. Cell Tissue Res. 183 :273-279 .

13 . Drenckhahn, D., R. Steffens, and U . Gröschel-Stewart . 1980 . Immunocytochemical local-ization of myosin in the brush border region of intestinal epithelium. Cell Tissue Res. 205 :163-166 .

14. Drenckhahn, D ., and U . Gröschel-Stewari . 1979. Immunocytochemical localization ofactin, myosin and tropomyosin in corneal epithelium, intestinal epithelium and variousother epithelial cells . Hoppe-Se vier's Z. Physiol. Chem . 360:1370 .

15. Gabbianai, G ., R . Montesano, B . Tuchweber, M . Salas, and L . Orci . 1975 . Phalloidin-induced hyperplasia of actin filaments in rat hepatocytes . Lab. Invest. 33 :562-569 .

16 . Geiger, B., K . T . Tokuyasu, and S . 1 . Singer . 1979 . Immunocytochemical localization ofa-actinin in intestinal epithelial cells . Pror. Nad. Arad. Sci. U. S. A . 76:2833-2837 .

17 . Gröschel-Stewart, U. 1980. Immunochemistry of cytoplasmic contractile proteins . Int.Rev. Cytol. 65. I n press .

18 . Gröschel-Stewart . U ., J. Schreiber, C. Mahlmeister, and K. Weber. 1976 . Production ofspecific antibodies to contractile proteins, and their use in immunofluorescence micros-copy. 1. Antibodies to smooth and striated chicken muscle myosins . Hislochemistrv. 46:229-236 .

19. Gröschel-Stewart, U ., S . Ceurremans, I . Lehr, C . Mahlmeister, and E . Paar . 1977 .Production of specific antibodies to contractile proteins, and their use in immunofluores-cence microscopy . II . Species-specific and species-non-specific antibodies to smooth andstriated chicken muscle actin . Histochemistry. 50:271-279.

20. Harboe, N ., and A. Ingild . 1973 . Immunization, isolation of immunoglobulins, estimationof antibody titre. A manual of quantitative immunoelectrophoresis. Methods and appli-cations. Scand. J. Immunol. 2(Suppl. 1) :161-164,

21 . Heine, W . D., H . W. Altmann . and U . Gröschel-Stewart . 1976 . Nachweis and Lokahsationyon Aktin and Myosin in Leberepithelzellen yon Mensch and Ratte. Immunfluoreszen-zuntersuchungen. Verh. Disch. Ges. Pathol. 60:321 .

22. Hull, B . E ., and L . A . Staehelin, 1979 . The terminal web . A reevaluation of its structureand function . J. Cell Biol. 81 :67-82 .

23 . Ishikawa, H., R . Bischoff, and H . Holtzer . 1969. Formation of arrowhead complexes withheavy meromyosin in a variety of cell types. J. Cell BiN. 43 :312-328 .

24 . Lazarides, E . 1976 . Two general classes of cytoplasmic actin filaments in tissue culturecells: The role of tropomyosin . J. Supramol. Stma. 5 :531-563 .

25 . Leblond, C . P ., and B . Messier . 1958 . Renewal of chief cells and goblet cells in the smallintestine as shown by radioautography after injection of Ihymidine-H" into mice . Anat.Rec. 132 :247-259 .

26. Mooseker, M . S . 1976 . Brush border motility . Microvillar contraction in Triton-treatedbrush borders isolated from intestinal epithelium . J. Cell Biol. 71 :417-433 .

27 . Mooseker, M . S., and L . G . Tilney . 1975 . organizatio n of an actin filament-membranecomplex . Filament polarity and membrane attachment in the microvilli of intestinalepithelial cells. J. Cell Biol. 67 :725-743.

28 . Mooseker, M . S., T . D . Pollard, and K . Fujiwara . 1978 . Characterization and localizationof myosin in the brush border of intestinal epithelial cells . J. Cell Biol. 79 :444453 .

29 . Pollard, T. D ., and E . D . Korn . 1973 . Acanthamoeba myosin. 1 . Isolation from Acantha-moeba castellanii of an enzyme similar to muscle myosin . J. Biol. Chem . 248 :4682-4690.

30. Rodewald, R., S . B . Newman, and M . J. Karnovsky . 1976 . Contractio n of isolated brushborders from the intestinal epithelium . J. Cell Biol. 70:541-554.

31 . Sandstrom, B . 1971 . A contribution to the concept of brush border function. Observationon intestinal epithelium in tissue culture . Cvtobiolope. 3 :293-297 .

32 . Silverman, A. J . 1976. Ultrastructural studies on the localization of neurohypophysialhormones and their carrier proteins . J. Hisiochem . Cytochem . 24:816-827.

33 . Sternberger, L. A. 1979 . Immunocytochemistry. Joh n Wiley & Sons. New York . 2ndedition .

34. Stemberger, L. A ., P . H. Hardy, 1 . 1 . Cuculis, and H . G . Meyer. 1970. The unlabeledantibody enzyme method of immunohistochemistry . Preparation and properties of solubleantigen-antibody complex (horseradish peroxidase-antihorseradish peroxidase) and its usein identification of spirochetes . J. Hisiochem. Cytochem. 18 :315-333 .

35 . Thuneberg, L ., and J . Rostgaard . 1969 . Motility of microvilli . A film demonstration . l.Ultrastruct. Res . 29 :578 .