DEVELOPMENTAL

BIOLOGY

Behavior

154,205-21’7

(19%)

of Structurally Divergent a-Tubulin lsotypes during Drosophila Embryogenesis: Evidence for Post-translational Regulation of lsotype Abundance E. THEURKAUF

WILLIAM Department

of Biochemistry

and Biophysics,

University

of California,

Accepted

July

San Francisco,

San

Francisco,

California

9.&g?-O&?

16, 1992

Two major or-tubulin isotypes are present during Drosophila embryogenesis: an evolutionarily divergent maternal isotype that is synthesized only in the ovary and deposited in the oocyte and a highly conserved constitutive isotype that is both maternally supplied and zygotically synthesized. A maternal isotype-specific antibody and a monoclonal antibody that recognizes both the maternal and constitutive isotypes were characterized and used to determine the distribution and abundance of a-tubulins during embryogenesis. Both isotypes are abundant and assemble into all classes of microtubules from the syncytial blastoderm stage until completion of germ band retraction. During subsequent development, however, the maternal isotype is retained only in the developing CNS, and later in a subset of connective fibers within the CNS. In contrast, total a-tubulin levels remain high in essentially all tissues throughout embryogenesis, indicating that most tissues selectively accumulate the constitutive isotype. To determine if selective accumulation of the constitutive isotype requires zygotic synthesis of this protein, mutant embryos that do not contain functional constitutive a-tubulin genes were examined. In these embryos, as in wild type, the maternal isotype decreases to background levels in tissues that retain high levels of the constitutive isotype. The constitutive isotype therefore appears to be more stable than the maternal isotype in most tissues. Differences in isotype stability may play an o 1992 Academic important role in determining the developmental pattern of isotype accumulation in Drosophila embryos. Press, Inc.

INTRODUCTION

Microtubules are ubiquitous intracellular filaments that are major components of mitotic and meiotic spindles, cilia, flagella, and the cytoskeleton (reviewed by Dustin, 1984). These filaments are assembled from heterodimeric subunits composed of an 01-and a /3-tubulin polypeptide. The diversity of microtubule-based structures led to the hypothesis that functionally distinct tubulin polypeptides are used in their assembly (Fulton and Simpson, 1976). This hypothesis was supported by the subsequent finding that (Y- and ,&tubulins are the products of multigene families that encode structurally different, yet evolutionarily conserved, tubulin isotypes (see Cleveland and Sullivan, 1986, for review). While recent studies suggest that some tubulin isotypes may differ in function (Arai and Matsumoto, 1988; Joshi and Cleveland, 1989; Hoyle and Raff, 1990), in the majority of systems structurally diverse tubulins assemble into all classes of microtubules and biochemical differences have not been detected (Kemphues et al., 1982; Schatz et al., 1986; May et al., 1985; Weatherbee et al., 1985; Lewis et al., 1987; Lewis and Cowan, 1988; Gu et al., 1988; Lopata and Cowan, 1987). The function of the multiple tubulin isotypes present in these systems is not known. 205

The a-tubulin isotypes present during early Drosophila development were examined with the hope that the developmental context would provide insights that cannot be gained from an examination of isotype function in cultured cells or in tissues from a single stage in the life cycle of an organism. The structural heterogeneity and tissue-specific expression of the cr-tubulins in Drosophila are typical of tubulin gene families (Theurkauf et aZ., 1986; Kalfayan and Wensink, 1982; Mischke and Pardue, 1982; Natzle and McCarthy, 1984). The four a-tubulin genes, formerly designated al, (~2,a3, and CY~(Kalfayan and Wensink, 1981), but now atub84b, Lutub85e, (-utub84d, and atub67c, respectively (Fyrberg and Goldstein, 1990), encode three distinct a-tubulin isotypes (Theurkauf et ak, 1986). The cutub84b and atub84d genes are expressed at all developmental stages and encode proteins that differ at only 2 of 450 residues. The constitutive isotype encoded by these genes shares 96% sequence identity with abundant mammalian tubulins. The &ub85e gene encodes a tubulin that differs from the constitutive isotype at 21 of 450 positions. Expression of this gene is restricted to the testes and a subset of cells in the peripheral nervous system (Bo and Wensink, 1989; Matthews et al., 1990). The cYtub67c mRNA is present only in ovaries and very early embryos (Kalfayan and Wensink, 1982; 0012-1606/92$5.00 Copyright All rights

Q 1992 by Academic Press. Inc. of reproduction in any form reserved.

206

DEVELOPMENTALBIOLOGY

Natzle and McCarthy, 1984), and it appears to be translated only in ovaries (Matthews et CL, 1989). This a-tubulin, referred to as the maternal isotype, has an unusually divergent amino acid sequence, sharing only 66% identity with the constitutive ru-tubulin. The constitutive and maternal isotypes account for virtually all of the a-tubulin in the early embryo, providing a relatively simple system for analysis of tubulin isotype function. Immunocytochemical analysis of the behavior of the maternal and constitutive isotypes during embryogenesis is reported here. While both isotypes assemble into all classes of microtubules in the early embryo, these studies indicate that their developmental stabilities differ. These differences, at least in part, appear to determine the pattern of isotype accumulation in embryonic tissues. METHODS

Antibody

Production

DNA fragments encoding amino acids 47 through 61 of both the constitutive (utub84b tubulin and the maternal atub6’7c tubulin were chemically synthesized and inserted into the SmaI site of the open reading frame vector pMRlO0 (Gray et aZ., 1982), restoring the reading frame and allowing a-tubulin/@-galactosidase fusion proteins to be made. Fusion protein expression was induced and cells harvested as described by Gray et al. (1982). Cells were lysed by grinding with alumina (Schleif and Wensink, 1981) in extraction buffer (0.2 M NaCl, 0.1 MNaHCO,, pH 8.3,lO mMmagnesium acetate, 1 mM@-mercaptoethanol, 0.1% sodium dodecyl sulfate). Alumina was removed by centrifugation, and 600 mg of each extract (approximately 200 mg of fusion protein) was subjected to electrophoresis through a 7.5% preparative SDS polyacrylamide gel. The fusion protein bands were excised, homogenized in phosphate-buffered saline (PBS, Karr and Alberts, 1986), and were mixed with Freund’s adjuvant. Rabbits were immunized with 100 pg of fusion protein in complete adjuvant. Subsequent boosts, at 2-week intervals, were 100 pg each in incomplete adjuvant. Sera were initially screened by probing Western blots of taxol-stabilized microtubule protein purified from early embryos (Vallee, 1982). One rabbit immunized with the maternal tubulin fusion protein produced antibodies that recognize embryonic tubulin. Immunization with the constitutive tubulin fusion protein did not elicit production of specific antibodies. Anti-maternal tubulin antibodies were purified by affinity chromatography on fusion protein coupled to CNBr-activated Sepharose 4B (Sigma). Thirty-five milligrams of fusion protein was coupled to 1 ml of Sepharose. Serum was passed over an atub84b/&galactosi-

vOLUME154.1992

dase column twice to remove anti-Sgalactosidase antibodies. The flowthrough was then applied to an atub67c/@-galactosidase column, and the bound antibodies were eluted with 0.1 M glycine, pH 2.5 (Hudson and Hay, 1980).

Fixation

and ImmunoJEuorescence

Embryos were fixed by a modification of the method of Mitchison and Sedat (1983). Manipulations were performed at room temperature unless otherwise noted. Embryos collected from Oregon R flies were dechorionated in a mixture of 2.7% sodium hypochlorite (50% bleach), 0.1% Triton X-100, and 0.7% NaCl for 2 min, then rinsed with 0.1% Triton X-100, 0.7% NaCl, and blotted dry on paper towels. The vitelline membranes were then permeabilized by shaking the embryos for 30 set in 10 ml of 100% heptane. An equal volume of 33% formaldehyde, 50 mM EGTA, pH 6.6, was added to the heptane and the two-phase mixture was incubated with gentle shaking for 5 min. Two different procedures were used to remove the vitelline membranes. The aqueous phase was removed and either 10 ml of 100% methanol or 10 ml of PBS was added. Methanol causes the vitelline membranes to rupture, and embryos without vitelline membranes sink to the bottom of the methanol phase. Floating embryos and the heptane/methanol mixture were removed and the remaining embryos were rehydrated (Karr and Alberts, 1986). If PBS was added, the vitelline membranes were removed manually, as follows: Fixed embryos were transferred from the heptane-PBS interface to the frosted portion of a glass slide. A cover glass was then placed over the embryos, which were then rolled between the two glass surfaces. This ruptures and removes the vitelline membranes from most of the embryos. The “rolled embryos” were rinsed into a test tube with PBS. An equal volume of heptane was then added, the mixture shaken, and the phases were allowed to separate. Embryos without vitelline membranes sink to the bottom of the PBS phase. Heptane and embryos floating at the heptane-PBS interface were then removed and the remaining embryos were extracted with 1% Triton X-100 in PBS for 1 to 2 hr. Embryos were rinsed in PBS containing 0.05% Triton X-100 prior to staining. Comparable immunofluorescence staining is obtained with both manual and methanol removal of vitelline membranes, although embryo morphology is generally better with the manual procedure. Alternatively, embryos were fixed with methanol as described by Kellogg et al. (1988). This procedure pro-

WILLIAM

E. THEURKAUF

duces similar results, although embryo morphology is generally poor. Double-label immunofluorescence analysis was carried out as described by Karr and Alberts (1986) with the following modifications. Incubations with both primary and secondary antibodies were for 18 hr at 4°C. Both primary and secondary antibody incubations were followed by four 30-min washes. Stained embryos were mounted in PBS containing 90% glycerol and 1 mg/ml p-phenylenediamine (mounting medium), and examined on either a Nikon microphot FXA microscope with epifluorescence attachment using a Nikon Plan 20/0.5 DIC objective, or a Nikon optiphot microscope with an MRC500 confocal imaging system (Bio-Rad), using either a Nikon Fluor 40/1.3 oil or a plan apo 60/1.4 oil objective. The morphological descriptions of staged embryos given by Campos-Ortega and Hartenstein (1985) were used to estimate the ages of the embryos. Immunojbrescence

Controls

To control for nonspecific staining, embryos were labeled with secondary antibodies alone or with inappropriate combinations of primary and secondary antibodies. Only low-level background staining was detected in these control experiments. In addition, the fluorescent secondary antibody conjugate used to detect each primary antibody was varied to further control for nonspecific staining. Similar results were obtained when double labeling was done with either rhodamine-conjugated goat anti-mouse and fluorescein-conjugated goat antirabbit secondary antibodies or fluorescein-conjugated goat anti-mouse and rhodamine-conjugated goat antirabbit secondary antibodies. The specificity of anti-maternal tubulin staining was confirmed by preincubating this antibody with a synthetic peptide corresponding to the maternal tubulin insertion in the atub6’7c/P-galactosidase fusion protein immunogen. At 10 pg/ml this peptide completely blocked microtubule staining by anti-maternal tubulin antibody in early embryos (not shown), but had no effect on staining by the monoclonal anti-cy-tubulin antibody. Low-level granular staining near the surface of later embryos is not inhibited by this peptide, indicating that it is nonspecific (see Results). Proteins cross-linked to tubulin during formaldehyde fixation could, in principle, mask the epitopes recognized by the antibodies. To control for such epitope masking embryos were fixed with methanol instead of formaldehyde before staining for a-tubulin (Warn and Warn, 1986, as modified by Kellogg et aZ., 1988). Methanol, a precipitating fixative, is unlikely to preserve specific protein-protein interactions capable of obscuring an epitope. This procedure yields comparable results to formaldehyde fixation (not shown).

Tubulin

Isotype

207

Regulation

Analysis of Embryos Hwmozygous for a Dejkiency Deletes the Constitutive a-T&&in Genes

That

The deficiency Df(3R) Antpneer17 removes the atub84b and atub84f genes, which encode the constitutive isotype, as well as the fushi tarazu (fiz) gene. To generate embryos homozygous for the deficiency, stocks were out-crossed to wild-type Oregon R flies, and offspring carrying one wild-type copy of the third chromosome and one third chromosome with the deficiency were mated to each other. Embryos collected from these crosses were fixed and stained for tubulin isotypes as described above. Homozygous Df(3R) Antp”‘+‘l’ embryos, over 8 hr old, can be identified by segmentation defects which develop due to lack of the ftx gene. Other Procedures SDS-polyacrylamide gel electrophoresis was performed as described in Maniatis et al. (1982). For Western blotting, proteins in polyacrylamide gels were transferred to nitrocellulose sheets and were probed with antibodies as described by Towbin et al. (1979). Primary antibody binding was detected using horseradish peroxidase or alkaline phosphatase-conjugated secondary antibodies (BioRad). Protein concentrations were estimated by the dye binding method (Bradford, 1976), using bovine serum albumin as a standard. Whole mount in situ hybridization using digoxygenin-labeled probes was performed as described by Tautz and Pfeiffe (1989). The Hind111 fragment containing the entire atub84b gene was used to detect constitutive tubulin message, and a specific probe for the atub67c gene was used to detect RNA encoding the maternal isotype (Kalfayan and Wensink, 1982). RESULTS

Antibody

Characterization

The specificity of affinity purified anti-maternal tubulin fusion protein antibody (see Methods) was determined by probing Western blots of fusions proteins and whole embryo extracts (Fig. 1). The purified antibody reacts with the fusion protein immunogen, but not with the constitutive tubulin fusion protein (Fig. 1, lanes l-4), demonstrating that the purified antibody recognizes maternal isotype-specific sequences, not B-galactosidase. The specificity of the anti-maternal tubulin antibody in embryos was determined by assaying fractions from a standard microtubule purification (Vallee, 1982) applied to early embryos. This antibody recognizes a single species in whole embryo extracts that copurifies with cu-tubulin, as detected with a monoclonal anti-vertebrate a-tubulin (Blose et ah, 1984), indicating that

DEVELOPMENTAL 5

BIOLOGY

6

VOLUME

sequence differences cause a decrease in electrophoretic mobility of the maternal isotype relative to the constitutive isotype, and that the anti-maternal tubulin fusion protein antibody is specific for this isotype. This antibody is hereafter referred to as anti-maternal tubulin. Because the monoclonal anti-a-tubulin detects both the maternal isotype and a major species that comigrates with vertebrate a-tubulin, and the maternal and constitutive and maternal isotypes are the only a-tubulins in early embryos, I conclude that the monoclonal antibody recognizes both maternal and constitutive isotypes. This antibody is used as a probe for total a-tubulin. Develmental

of anti-cu-tubulin antibodies. Extracts of Eschethe maternal tubulin/@-galactosidase immunogen (lanes 1 and 3) or a constitutive tubulin/&galactosidase fusion protein (lanes 2 and 4) were resolved by electrophoresis on duplicate 7.5% polyacrylamide gels which were either silver stained to reveal all proteins (lanes 1 and 2) or transferred to nitrocellulose and probed with anti-maternal tubulin antibody (lanes 3 and 4). Only the fusion protein containing maternal tubulin sequence is recognized by the anti-maternal tubulin antibody. Binding was detected with an alkaline phosphatase-conjugated goat anti-rabbit secondary antibody. To determine the specificity of the a monoclonal anti-cY-tubulin antibody, embryonic homogenate was subjected to electrophoresis on 8.5% polyacrylamide gels, transferred to nitrocellulose, and probed with either the monoclonal anti-vertebrate a-tubulin antibody (lane 5) or affinity-purified anti-maternal tubulin antibody (lane 6). The monoclonal antibody recognizes two species, one of which comigrates with the single species recognized by the anti-maternal tubulin antibody. FIG.

1. Specificity

richiu coli cells containing

cross-reaction with nonmicrotubule proteins is not significant (Theurkauf, 1988). The isotype specificity of the anti-maternal tubulin antibody was confirmed by probing Western blots of extracts prepared from early embryos. The anti-maternal tubulin fusion protein antibody detects only one of the two species detected with the monoclonal anti-vertebrate a-tubulin (Fig. 1, lanes 5 and 6). The major antigen recognized by the monoclonal antibody has the same electrophoretic mobility as vertebrate a-tubulin (data not shown), while the anti-fusion protein antibody recognizes a species with a higher apparent molecular weight. Because the maternal isotype shares only 66% sequence identity with vertebrate a-tubulins, and is larger due to the insertion of 11 amino acids near the amino terminus (Theurkauf et aL, 1986), I conclude that

154,1992

Changes in Isotype Levels

To determine the pattern of a-tubulin accumulation during embryogenesis, Western blots of extracts made from staged embryos were probed with the anti-maternal isotype antibody and the anti-a-tubulin monoclonal antibody which recognize both maternal and constitutive isotypes. Total a-tubulin levels, detected with the monoclonal anti-a-tubulin, peak between 9 and 15 hr after fertilization (Fig. 2a). This pattern of protein accumulation is similar to the expression pattern of the constitutive a-tubulin mRNA (Kalfayan and Wensink, 1982). In contrast, the maternal isotype is most abundant in early embryos. Levels of this isotype begin to decline between 6 and 9 hr and the protein is nearly undetectable after 18 hr (Fig. 2b). This is consistent with the findings that mRNA encoding the maternal

FIG. 2. Changes in maternal and total cu-tubulin levels during embryogenesis. Whole cell extracts (5 pg/lane) from staged embryos were resolved on 9.5% polyacrylamide gels, blotted onto nitrocellulose, and probed with either the monoclonal anti-o-tubulin antibody (a) or the anti-maternal tubulin antibody (b). The ages of embryos (in hours) used to prepare extracts are indicated. Only the tubulin region of the blots are shown. The maternal and constitutive isotypes do not resolve on 9.5% polyacrylamide gels.

WILLIAME.THEURKAUF

isotype is undetectable after 3 hr (Kalfayan and Wensink, 1982; Matthews et ah, 1989; Natzle and McCarthy, 1984), and that the message that is present in early embryos may not be translated (Matthews et al, 1989). The alkaline phosphatase reaction used to detect the anti-maternal isotype in Fig. 2 is more sensitive than the peroxidase reaction used to detect total cu-tubulin. The relative staining in this figure, therefore, does not reflect relative abundance. Analysis of silver-stained two-dimensional gels that resolve the early embryonic tubulins indicates that the maternal and constitutive isotypes are present at similar levels in 0- to 3-hr embryos (Theurkauf, 1988). The Lu-tubulin isotypes present in the embryo therefore changes during embryogenesis from a mixture of maternal and constitutive isotypes to primarily the constitutive isotype. CY-Tubulin Isotype Distributions in Syncytial Embryos

Embryogenesis in Drosophila begins with a series of rapid, nearly synchronous, nuclear divisions without cellularization. For the first seven divisions, nuclei accumulate in the center of the embryo. During cycles eight and nine most of the nuclei migrate to the cortex, where they form a monolayer. At approximately 3 hr postfertilization, after four additional divisions, membranes enclose the cortical nuclei (for a detailed description of nuclear behavior in early embryos see Foe and Alberts, 1983; Zalokar and Erk, 1976). Gastrulation begins immediately after formation of the cellular blastoderm. To determine if the maternal and constitutive isotypes assemble into functionally distinct microtubules during these rapid divisions, the distributions of a-tubulin isotypes in syncytial blastoderm-stage embryos were examined using both conventional and laser scanning confocal microscopy (White et al., 1987). Both antibodies label astral and spindle microtubules at metaphase, and interphase microtubule arrays (Fig. 3). Therefore, all classes of microtubules in early embryos contain the structurally divergent maternal isotype. Maternal and total a-tubulin distributions remain essentially indistinguishable from cellularization through completion of germ band retraction at approximately 8 hr (Figs. 3,4a, and 4a’, closed arrow), indicating that the blastoderm precursors of all of the tissues that develop later during embryogenesis contain both isotypes. Changes in Isotype Distribution Embryogenesis

during Later

Differences between the staining patterns obtained with the anti-maternal and anti-a-tubulin antibodies develop as the central nervous system (CNS) is established. At 8 hr, the ventral midline stains brightly with

Tuindin

Isotype

Regulation

209

the anti-maternal tubulin antibody (Figs. 4a and 4a’, open arrow), while staining throughout the rest of the embryo has become relatively weak. Maternal isotype staining in the 4.5-hr embryo in Fig. 4a (arrow) is clearly more intense than the labeling in most tissues in the 8-hr embryo in the same frame. Only the cells at the ventral midline in the 8-hr embryo retain levels of antimaternal tubulin staining that are comparable to those found earlier. These observations suggest that the maternal isoform is retained in the cells at the midline, but is degraded in most other tissues. Between 8 and 12 hr after fertilization, as the CNS is formed (Thomas et al, 1984), the level of maternal isotype-specific staining in nonneuronal tissues drops significantly (Fig. 4~). The nonneuronal staining that is observed in embryos over 12 hr old appears to be nonspecific (see below). Within the developing CNS, however, maternal isotype staining remains relatively strong (Figs. 4b and 4b’). Total cY-tubulin levels, in contrast, remain high in all tissues. The esophagus and the epidermis, particularly at segment boundaries, provide striking examples of the differences that develop between total a-tubulin and maternal a-tubulin isotype labeling outside of the nervous system (Figs. 4b, 4b’, 4c, and 4c’, arrows). To analyze developmental changes in isotype distribution in greater detail, I have used the scanning confocal microscope (White et al, 1987) to examine changes in cr-tubulin isotype distribution during neurogenesis. This instrument produces clear images within thick specimens by eliminating out-of-focus signal. In addition, the fluorescence intensity observed in different optical sections can be quantitatively compared, providing microscope parameters affecting signal intensity are not altered. These parameters were not changed between optical sections during the analyses reported here. Figure 5 shows representative confocal images of embryos between 8 and 12 hr of embryogenesis. Figures 5a and 5a’ display maternal and total cY-tubulin distributions in newly formed commissural and connective neuronal processes within the developing CNS of a lo-hr embryo, while the optical section in Figs. 5b and 5b shows isotype distributions in the epidermis of the same embryo. Maternal and constitutive staining is similar in both neuronal and nonneuronal tissues at this stage, although maternal isotype staining is somewhat more diffuse and weaker than total tubulin staining in the epidermis (Figs. 5b and 5b’). By 12 hr, maternal isotype-specific staining in the epidermis has dropped significantly, while total cr-tubulin levels remain high (Figs. 5d and 5d’). The punctate staining that is detected with the anti-maternal tubulin

210

DEVELOPMENTALBIOLOGY

vOLlJME154,1%2

WILLIAM

E. THEURKAUF

antibody in the epidermis does not colocalize with the filaments labeled with the monoclonal antibody (compare Figs. 5d and 5d’), and is not eliminated by preincubation of the antibody with a peptide corresponding to the immunogen (Figs. 5f and 5f’, and see Methods). In the CNS, in contrast, maternal isotype staining remains relatively bright, colocalizes with the fibers detected with the monoclonal anti-cu-tubulin (Figs. 5c and 5c’), and is competed by the peptide immunogen (Figs. 5f and 5f’). Thus, specific anti-maternal tubulin staining becomes restricted to the CNS between 8 and 12 hr. The distribution of maternal isotype-specific staining becomes further restricted between 12 and 15 hr. Only a single pair of connective fibers on either side of the ventral midline consistently retain high-level maternal isotype staining in 15 hr embryos (Fig. 5e, small arrow). Connectives that preferentially stain with the anti-maternal tubulin antibody remain present through 18 hr of development: after this time these processes are not observed, suggesting that either the maternal isotype is degraded and replaced by the constitutive isotype or that the processes rich in the maternal cY-tubulin isotype degenerate. Masking of the epitope recognized by the anti-maternal tubulin antibody could lead to restricted immunolabeling. This does not appear to be the case, however, because the loss of maternal isotype immunofluorescence labeling in most tissues during embryogenesis parallels the decrease in isotype levels observed on Western blots (Fig. 2). In addition, both cross-linking and precipitating fixation protocols yield similar results (see Methods). It is unlikely that protein-protein interactions capable of masking a specific epitope will be maintained after precipitating fixation. Post-translational modification of the a-tubulin isotypes could also alter antibody binding. Post-translationally modified variants of tubulin can be resolved on 2D gels (Matthews et oZ., 1987; Theurkauf, 1988). The anti-maternal isotype antibody and monoclonal anti-tubulin antibodies used in this study react with multiple isoelectric variants, and these variants account for essentially all of the species with the apparent molecular weight of ar-tubulin detected by silver staining (Theurkauf, 1988). Neither antibody, therefore, recognizes an epitope that is altered in the identifiable modified forms of cu-tubulin. Based on these

T&din

Isotype

211

Regulation

observations, I conclude that the maternal isotype is selectively lost from most embryonic tissues. Isotype Accumulation Tulndin Expression

in the Absence of Zygotic a-

The loss of maternal isotype staining from most tissues could result from uniform degradation of both maternal and constitutive a-tubulins, coupled with zygotic synthesis of only the constitutive isotype. Alternatively, selective loss of the maternal isotype could yield the observed changes in isotype levels. To determine the contribution of zygotic cY-tubulin expression, isotype behavior in mutant embryos that do not contain functional constitutive isotype genes was examined. These embryos were generated by crossing adults heterozygous for a deficiency (Df(3R)Antpn”+“‘) that removes both constitutive cr-tubulin genes. There is no detectable embryonic expression of the maternal atub67c gene, and the remaining cY-tubulin gene, cutub85e, is only transcribed in a small number of peripheral neurons (Bo and Wensink, 1989; Matthews et al, 1990). Therefore, most tissues in embryos homozygous for Df(3R)Antp”‘+‘l’ will not express a-tubulin zygotically. Perdurance of maternally supplied a-tubulin mRNA in the mutant embryos that do not express a-tubulins zygotically will affect the interpretation of isotype accumulation patterns. Previous studies demonstrate that mRNA encoding the maternal isotype is degraded before blastoderm cellularization (Matthews et aL, 1990). To determine the perdurance of maternally supplied constitutive transcript, the behavior of this mRNA in embryos which do not express the constitutive isotype zygotically was examined (Fig. 6). Twenty-five percent of the embryos obtained from a cross of flies heterozygous for the deficiency should be homozygous for the deficiency. These homozygotes can be identified by segmentation defects which appear after 8 hr, because Df(3R)AntpnSf’17 removes the segmentation gene fushi tarazu (fix). mRNA encoding the constitutive a-tubulin isotype is undetectable in homozygous embryos showing characteristic segmentation defects (Fig. 6a), indicating that most of the maternally supplied constitutive a-tubulin mRNA has been degraded by 8 hr. In addition, approximately 25% of embryos between the cellular blastoderm stage and the onset of visible segmentation

FIG. 3. The cY-tubulin distribution in early embryos. Each pair of confocal micrographs shows lower case letters) and total cu-tubulin (right, letter with prime) in the same field. (a and a’) nuclear division cycle 9. (b and b’) Pole buds at the posterior end of a cycle 10 embryo. (c and Anaphase spindles in the cortex of a cycle 12 embryo. Monoclonal anti-cy-tubulin binding was anti-mouse IgG secondary antibody. All embryos are between 1.5 and 2.3 hr old (nuclear cycle monoclonal anti-or-tubulin and anti-maternal tubulin antibodies for indirect immunofluorescence in (a’) is 25 bm. The bar in b’ is 10 pm and also applies to c, c’, d, and d’.

the distribution of maternal isotype (left, plain The anterior end of an embryo in anaphase of c’) An embryo in prophase of cycle 9. (d and d’) visualized using a rhodamine-conjugated goat 9 to 12), and were fixed and double labeled with analysis as described under Methods. The bar

212

DEVELOPMENTAL

MATERNAL

strc sysl the dist Fig

BIOLOGY

VOLUME

154,1992

TOTAL

‘IG. 4. Isotype redistribution between 4.5 and 12 hr. (a and a’) The 1lateral surface near the anterior end of a 4.5- to 5-hr form bright staining with both antibodies (closed arrow), and the ! ventral surface of an 8-hr embryo, showing midline mgly with the anti-maternal tubulin antibody (open arrow). (b an Id b’) Lateral view of a 12-hr embryo focusing on the tern. (c and c’) The embryo in b and b’, focusing instead on the epider .mis. Embryos are oriented with the anterior left, with 4.5-hr embryo in a and a’. The arrow in b and b’ indicates the esc,phagus and the arrows in c and c’ indicate segment .ributions of total cY-tubulin and the maternal cu-tubulin isotype wei *e visualized and micrographs are labeled as described .3, except that embryos were photographed on a standard fluoresce rice microscope. The bar denotes 50 pm.

embryo showing cells that stain central nervous the exception of boundaries. The in the legend to

WILLIAME.THEURKAUF MATERNAL

TOTAL

T&din

Isotype

213

Regulation

MATERNAL

TOTAL

analysis of isotype redistribution. (a and a’) Ventral view of the CNS of a 10.5-hr embryo, as the FIG. 5. Laser scanning confocal microscopic connectives and commissural processes are forming. This embryo is oriented with the anterior up. The connectives run in two tracks along the anterior-posterior axis (vertically here), and in each segment the commissures run laterally between these tracks. (b and b’) Isotype distributions in the epidermis approximately 15 wrn above the same section shown in a and a’. (c and c’) Ventral view of the CNS in an ll- to 12-hr embryo, oriented with its anterior end to the left. One of the two connectives in each neuromere that stain preferentially with anti-maternal tubulin is indicated (arrow). (d and d’) Low-level maternal isotype-specific staining in the epidermis of the same embryo shown in c and c’. (e and e’) Lateral view of the CNS in a 16- to 17-hr embryo, oriented with its anterior end to the left. A connective preferentially stained with anti-maternal tubulin is indicated (small arrow). Commissural processes that stain preferentially with the monoclonal anti-a-tubulin are also indicated (large arrow). (f and f’) Cross section of a 12-hr embryo double labeled in the presence of competing maternal isotype-specific peptide. CNS staining is competed by the peptide, while background staining seen at the surface is not. The bar in e’ is 10 pm and applies to a through e. The bar in f is 20 pm and applies to f and f’.

214

DEVELOPMENTALBIOLOGY

VOLUME154,1992

FIG. 6. cY-tubulin message distribution in embryos produced by crossing flies heterozygous for a deficiency that removes the constitutive cr-tubulin genes (Df(3R)Antp”+“‘). T wenty-five percent of the embryos resulting from this cross are predicted to be homozygous for the deficiency and therefore will not express the constitutive isotype zygotically. Homozygous mutant embryos more than 8 hr old were identified by segmentation defects associated with loss of the j?z gene (see Methods). (a) A 12-hr embryo containing the constitutive cy-tubulin genes (small arrowhead), and a homozygous mutant embryo of the same age (large arrowhead). (b) Constitutive message in 2-hr (A), 3-hr (B), and 4-hr (C) embryos. The constitutive mRNA is nearly undetectable in the 2-hr embryo (A). (c) Hybridization typical of 72%. of the 2.5- to 8-hr embryos. (d) Lack of maternal message typical of 28% of 2.5- to S-hr embryos. mRNA was detected using a digoxygenin-labeled probe. The scale bar in b denotes 50 pm, and also applies to a. The scale bar in d denotes 50 pm and applies to c and d.

also do not show detectable hybridization with the constitutive cY-tubulin probe (Fig. 6). These embryos are likely to be homozygous for Df(3R)Antp”“+“‘. These observations suggest that maternally supplied constitutive a-tubulin mRNA is degraded by 2.5 to 3 hr. Significant synthesis the constitutive and maternal isotypes is therefore unlikely in the homozygous mutant embryos after 2.5 to 3 hr of embryogenesis. The cu-tubulin isotype distributions in embryos homozygous for the deficiency removing the constitutive LYtubulin gene indicate that selective retention of the constitutive isotype is independent of zygotic expression of the conatitutive a-tubulin gene. All of the 0- to 8-hr embryos resulting from this cross stain with both maternal and total cY-tubulin antibodies (Figs. 7a and 7a’, embryo

A); thus zygotic expression is not required to maintain cr-tubulin levels in embryos to this age. In all older embryos, including those homozygous for the deficiency, there is a dramatic loss of the maternal isotype staining. Total a-tubulin levels, in contrast, remain high (Figs. 7a and ‘7a’,embryos B and C). The selective loss of maternal isotype staining between 8 and 12 hr is clearly shown in Figs. 7b and ‘7b’. The distributions of both isotypes are indistinguishable in the early gastrula at upper left (Figs. 7b and 7b’), while the maternal isotype is essentially undetectable in most tissues of the 12- to 13-hr embryo at lower right (Figs. 7b and 7b’, lower right). These observations indicate that selective retention of the constitutive isotype in most embryonic tissues reflects selective loss of the maternal isotype.

WILLIAME.THEURKAUF

TOTAL

MATERNAL

FIG. 7. Selective retention of the constitutive a-tubulin isotype in the epidermis is independent of zygotic expression of the constitutive ol-tubulin genes. (a and a’) Maternal and total or-tubulin distributions in 3.5-hr (A), S-hr (B), and 12-hr (C) embryos. Maternal and total a-tubulin distributions are the same in all 0- and S-hr embryos. The 8and 12-hr embryos are both homozygous for the deficiency. (b and b’) Details of the 8- and 12-hr embryos in a and a’. Note the dramatic loss of maternal tubulin staining between 8 hr (top left embryo) and 12 hr (bottom right embryo).

DISCUSSION

In an attempt to gain insights into the functions of the structurally divergent a-tubulin isotypes present in Drosophila embryos, I have examined their distributions during embryogenesis. Although the maternal and constitutive isotypes share only 66% sequence identity, both assemble into all classes of microtubules in syncytial embryos (Fig. 3). Therefore, structural differences between the maternal and constitutive isotypes do not direct the assembly of morphologically distinct microtubules during early embryogenesis. The maternal isotype is present throughout the cellular blastoderm stage and remains indistinguishable from total cu-tubulin in its spatial distributions until

Tulndin

Isotype

Regulation

215

neurite outgrowth begins between 8 and 9 hr. Thus, the precursors of all later embryonic tissues contain both isotypes. Between 9 and 12 hr, however, maternal isotype-specific staining is lost from most embryonic tissues, while these tissues maintain high total a-tubulin levels (Figs. 4 and 7). Only the developing CNS retains maternal isotype staining during this time. During later stages the maternal isotype becomes further restricted, with high levels only observed in a subset of connective fibers within the CNS. The loss of maternal isotype-specific staining from embryonic tissues that retain high levels of total a-tubulin could reflect specific degradation of the maternal isotype, stabilization of the constitutive isotype, or uniform degradation of all maternally deposited cr-tubulins combined with selective zygotic replacement of the constitutive isotype. Isotype behavior in embryos that do not express a-tubulin zygotically indicates that selective retention of the constitutive isotype does not require zygotic expression of the constitutive a-tubulin gene (Fig. 7). The constitutive isotype is therefore developmentally longer lived than the maternal isotype in the majority of embryonic tissues, and this post-translational difference between the isotypes appears to be sufficient to generate the change in isotype composition observed in most tissues during embryogenesis. In contrast to the majority of embryonic tissues, the developing CNS, and later to a subset of neuronal processes, continue to stain with the anti-maternal tubulin antibody during later stages of embryogenesis. Tissuespecific expression of the atub67c gene could produce this maternal a-tubulin staining. There is no biochemical evidence for zygotic expression of the maternal tubulin gene, however, or for synthesis of the maternal isotype in the embryo (Kalfayan and Wensink, 1982; Matthews et al., 1989 and 1990; Mischke and Pardue, 1982; Natzle and McCarthy, 1984). Asymmetric partitioning of the maternal isotype at cytokinesis or differences in the number of cell divisions between lineages could also produce tissue-specific accumulation of a maternally deposited protein. The tissue-specific distribution of the maternal isotype develops after the final cell division in the embryo, however, and the cells that give rise to the CNS actually go through a greater number of divisions than other embryonic precursors (Campos-Ortega and Hartenstein, 1985). It therefore seems likely that the maternal isotype is more stable in neuronal processes than in other tissues, indicating that the developmental stability of this isotype is tissue specific. The mechanism that gives rise to maternal isotype perdurance in the CNS remains to be determined. Perdurance of the maternal isotype in the CNS could result from formation of stable microtubules in neuronal processes. This seems unlikely, however, because preferen-

216

DEVELOPMENTAL BIOLOGY

tial retention of the maternal isotype is detectable in the neuronal precursors before neurite outgrowth has begun (Fig. 4). Alternatively, the degradation of this protein may be regulated in a tissue-specific manner. The cellular mechanism that leads to selective retention of the constitutive isotype in most tissues is also unknown. A likely possibility is that isotype-specific sequences are differentially recognized by the cellular degradation machinery. If isotype-specific sequences are involved in regulating &tubulin stability, this level of control could help explain the evolutionary conservation of the amino acid sequences that distinguish different isotypes. I thank Douglas Kellogg, Pieter Wensink, and Bruce Alberts for comments on the manuscript. Special thanks to Pieter Wensink and Bruce Alberts for supporting this work, which was funded by National Institutes of Health research grants to Bruce M. Alberts (GM23928) and Pieter C. Wensink (GM31234), and a grant from the Damon Runyon-Walter Winchell Cancer Research Fund to W. E. Theurkauf (DRG-945).

REFERENCES Arai, T., and Matsumoto, G. (1988). Subcellular localization of functionally differentiated microtubules in squid neurons: Regional distribution of microtubule-associated proteins and B-tubulin isotypes. J. Neurochem.

51,1825-1838.

Blose, S. H., Meltzer, D. I., and Ferimisco, J. R. (1984). lo-nm filaments induced to collapse in living cells microinjected with monoclonal and polyclonal antibodies against tubulin. J. Cell Biol. 98, 847-858. Bo, J. Y., and Wensink, P. C. (1989). The promoter region of the Dro sophila a-tubulin gene directs testicular and neural specific expression. Development 106,521-587. Bradford, M. M. (1976). A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing protein-dye binding. Anal B&hem 72,248-254. Campos-Ortega, J. A., and Hartenstein, V. (1985). “The Embryonic Development of Drosophila.” Springer-Verlag, Heidelberg. Cleveland, D. W., and Sullivan, K. F. (1986). Molecular biology and genetics of tubulin. Annu. Rev. B&hem. 54,331-365. Cleveland, D. W. (1987). Autoregulated instability of tubulin mRNAs: A novel eukaryotic regulatory mechanism. TIBS 13,339-343. Dice, J. F. (1987). Molecular determinants of protein half-lives in eukaryotic cells. FASEB J. 1, 349-357. Dustin, P. (1984). “Microtubules,” 2nd ed. Springer, New York. Foe, V. E., and Alberts, B. M. (1983). Studies of nuclear and cytoplasmic behavior during the five mitotie cycles that precede gastrulation in Drosophila embryogenesis. J. Cell Sci. 61,31-70. Fulton, C., and Simpson, P. A. (1976). Selective synthesis and utilization of flagellar tubulin: The multitubulin hypothesis. In “Cell Motility,” (R. Goldman, T. Pollard, and J. Rosenbaum, Eds.), pp. 9871005. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Fyrberg, E., and Goldstein, L. (1990). The cytoskeleton in Drosophila Annu. Rev. Cell Biol. 6,559-596. Gray, M. R., Colot, H. V., Guarente, L., and Rosbash, M. (1982). Open reading frame cloning: Identification, cloning, and expressing open reading frame DNA. Proc. Natl. Acad. Sci. USA 79,6598-6602.

VOLUME I54,1992

Gu, W., Lewis, S. A., and Cowan, N. J. (1988). Generation of antisera that discriminate among mammalian a-tubulins: Introduction of specialized isotypes into cultured cells results in their coassembly without disruption of normal microtubule function. J. CeU Biol. 106, 2011-2022. Hoyle, H. D., and Raff, E. C. (1990). Two Drosophila fl-tubulin isoforms are not functionally equivalent. J. Cell Biol 111,1009-1026. Hudson, L., and Hay, F. C. (1980). “Practical Immunology,” 2nd ed. Blackwell Scientific, Boston. Joshi, H. C., and Cleveland, D. W. (1989). Differential utilization of p-tubulin isotypes in differentiating neurites. J. Cell Biol. 109, 663674. Kalfayan, L., and Wensink, P. C. (1981). a-tubulin genes of Drosophila. Cell 24, 97-107. Kalfayan, L., and Wensink, P. C. (1982). Developmental regulation of Drosophila a-tubulin genes. Cell 29,91-98. Karr, T. C., and Alberts, B. M. (1986). Organization of the cytoskeleton in early Drosophila embryos. J. CeU Biol 102.1494-1509. Kellogg, D. R., Mitchison, T. J., and Alberts, B. M. (1988). Behavior of microtubules and actin filaments in living Drosophila embryos. Development

103,675-686.

Kemphues, K. J., Kaufman, T. C., Raff, R. A., and Raff, E. C. (1982). The testes-specific fi-tubulin subunit in Drosophila melanogaster has multiple functions in spermatogenesis. Cell 31, 3711-3722. Laemmli, Il. K. (1970). Cleavage of the structural proteins during assembly of the head of bacteriophage T4. Nature (London) 227,680685. Lewis, S. A., and Cowan, N. J. (1988). Complex regulation and functional versatility of mammalian a- and P-tubulin isotypes during differentiation of testes and muscle cells. J. CeUBiol. 106,2023-2033. Lewis, S. A., Wei, G., and Cowan, N. J. (1987). Free intermingling of mammalian @-tubulin isotypes among functionally distinct microtubules. Cell 49, 539-548. Lopata, M. A., and Cleveland, D. W. (1987). In viva microtubules are copolymers of available fl-tubulin isotypes: Localization of each of six vertebrate /3-tubulin isotypes using polyclonal antibodies elicited by synthetic peptide antigens. J. Cell Biol. 105,1707-1720. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982). “Molecular Cloning.” Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Matthews, K. A., Miller, D. F. B., and Kaufman, T. C. (1989). Developmental distribution of RNA and protein products of the Drowphila a-tubulin gene family. Dew. Biol. 132,45-61. Matthews, K. A., Miller, D. F. B., and Kaufman, T. C. (1990). Functional implications of the unusual distribution of a minor a-tubulin isotype in Drosophila: A common thread among chordotonal ligaments, developing muscle, and testis cyst cells. Dev. Bio2. 137, 171183.

May, G. S., Gambino, J., and Morris, N. R. (1985). Identification and functional analysis of P-tubulin genes by site specific integration in Aspergillus

nidulans.

J. Cell Biol

101,239-255.

Mischke, D., and Pardue, M. L. (1982). Organization and expression of a-tubulin genes in Drosophila melunogaster. J. Mol. Biol. 156,449466. Mitchison, T. J., and Sedat, J. (1983). Localization of antigenic determinants in whole Drosophila embryos. Dev. Biol. 99,261-264. Natzle, J. E., and McCarthy, B. J. (1984). Regulation of the Drosophila (Y-and fl-tubulin genes during development. Dev. Biol. 104,187-198. Schatz, P. J., Solomon, F., and Botstein, D. (1986). Genetically essential and non-essential a-tubulin genes in the yeast Saccharomyces cerevisiae encode divergent proteins. Mol. Cell. Biol. 6,3711-3721. Schleif, R. F., and Wensink, P. C. (1981). “Practical Methods in Molecular Biology.” Springer, New York.

WILLIAME.THEURKAUF Tautz, D., and Pfeiffe, C. (1989). A non-radioactive in situ hybridization method for the localization of specific RNAs in Drosophila embryos reveals translational control of the segmentation gene hunchback. Chrowwsm 98,81-85. Theurkauf, W. E. (1988). “Studies on the Structure, Function, and Developmental Regulation of o-Tubulins in L?ro.sophiZu.” Ph.D. thesis, Brandeis University. Theurkauf, W. E., Baum, H. J., Bo, J., and Wensink, P. C. (1986). Constitutive and developmentally regulated a-tubulin genes in Drosophila code for structurally distinct proteins. Proc. Natl. Acad Sci USA 83.8477-8481. Thomas, J. B., Bastiani, M. J., Bate, M., and Goodman, C. S. (1984). From grasshopper to Drosophila: A common plan for neuronal development. Nature @m&m) 310,203-207. Towbin, H., Staehelin, T., and Gordon, J. (1979). Electrophoretic

T&&n

Isotvpe Regulation

217

transfer from polyacrylamide gels to nitrocellulose sheets: Procedure and some applications. Proc. Natl. Acad Sci. USA 76, 84778418. Vallee, R. B. (1982). A taxol-dependent procedure for the isolation of microtubules and microtubule-associated proteins (MAPS). J. Cell Biol. 101,712-719. Weatherbee, J. A., May, G. S., Gambino, J., and Morris, N. R. (1985). Involvement if a particular species of fi-tubulin (B3) in conidial development in Aspergillus nidulans. J. Cell Biol. 101,712-719. White, J. G., Amos, W. B., and Fordham, M. (1987). An evaluation of confocal versus conventional imaging of biological structures by fluorescence light microscopy. J. CeU Biol. 105.41-48. Zalokar, M., and Erk, I. (1976). Division and migration of nuclei during embryogenesis of Drosophila melmogaster. J. Aiiorobiol. Cell 25,97106.