The Plant Cell, Vol. 2, 1201-1 224, December 1990O 1990American Society of Plant Physiologists Different Temporal and Spatial Gene Expression Patterns Occur during Anther Development Anna M. Koltunow, Jessie Truettner,‘ Kathleen H. COX,~Marco Wallr~th,~and Robert 6. Goldberg4 Departmentof Biology, Universityof California, Los Angeles, California90024-1606 We studied the temporal and spatial regulation of three mRNA sequence sets that are present exclusively, or at elevated levels, in the tobacco anther. One mRNA set accumulates in the tapetum and decays as the tapetum degenerates later in anther development. The second mRNA set accumulates after the tapetal-specific mRNAs, is localized within the stomium and connective, and also decays as these cell types degenerate during anther maturation. The third mRNA sequence set persists throughout anther development and is localized within most anther tissues. A tapetal-specific gene, designated as TA29, was isolated from a tobacco genome library. Runoff transcription studies and experiments with chimeric p-glucuronidase and diphtheria toxin A-chain genes showed that the TA29 gene is regulated primarily at the transcriptional leve1and that a 122-basepair 5‘ region can program the tapetal-specific expression pattern. Destruction of the tapetum by the cytotoxic gene had no effect on the differentiation and/or function of surrounding sporophytic tissues but led to the production of male-sterile plants. Together, our studies show that several independent gene expression programs occur during anther development and that these programs correlate with the differentiated state of specific anther cell types. INTRODUCTION Reproductive processes in higher plants take place within two specializedfloral organ systems, the stamen and the pistil (Esau, 1977; Raven et al., 1986; Goldberg, 1988; Drews and Goldberg, 1989). The anther compartment of the stamen contains diploid cells that undergo meiosis to form haploid microspores that differentiateinto the pollen grains or male gametophyte (Vasil, 1967; Esau, 1977; Raven et al., 1986). By contrast, sporogenouscells within the ovary of the pistil ultimately lead to the production of the female gametophyte, or embryo sac, that contains the egg cell (Esau, 1977; Raven et al., 1986). Pollination is required to transfer pollen grains containing the sperm cells to the stigma of the pistil so that fertilization can occur in the embryo sac (Esau, 1977; Raven et al., 1986). Stamensare differentiatedfrom primordiathat are specifiedwithin the floral meristemfollowing the transitionfrom a vegetative to a flowering pathway (Esau, 1977; Raven et al., 1986).Stamenprimordiaspecificationdepends upon the action of several genes (Bowman et al., 1989; Meyerowitz et al., 1989; Carpenter and Coen, 1990), some of ’Currentaddress:Departmentof CellBiology, Universityof Miami, Miami, FL 33101. * Current address: Departmentof Microbiology,Universityof Tennessee, Memphis, TN 38163. Current address: Plant Cell Research Institute, San Carlos, CA 94070. To whom correspondenceshould be addressed. which may encode transcriptional activator proteins (Sommer et al., 1990;Yanofsky et al., 1990). Following stamen primordia initiation, several highly specialized anther tissues are differentiated from cell lineages present in the floral meristem (Satina and Blakeslee, 1941) and are responsiblefor carrying out nonreproductivefunctions (e.g., support, dehiscence) and reproductive functions (e.g., spore and gamete formation). The molecular processes that are responsiblefor specifying functionally distinct anther cell types after primordia initiation are not well understood. Previously, we used RNA-excessDNA/RNA hybridizationexperimentsto show that approximately25,000 diverse genes are expressed in the tobacco anther (Kamalay and Goldberg, 1980). Comparisonsof tobacco organ system mRNAand nuclear RNA populationsshowed that about 10,000 mRNAs are anther specific andare undetectableinthe cytoplasm and nucleus of cells in other vegetative and floral organ systems (Kamalay and Goldberg, 1980, 1984).These findings suggest that the transcriptional activation of specific gene sets is required for the establishment and maintenance of differentiated cell types and functions during anther development. In this paper, we describe the temporal and spatial expression patterns of several different anther-specific genes. Our results show that distinct gene expression programsoccur during anther developmentand that these programscorrespondwiththe differentiationof functionally BBBHDBH (B) (C) (D) FLOWERVL " 5i i i i i i i i i i A«400 -300 ANTHER 0 . 4 ' i 3 0 PISTIL 1 3 5 7 9 STAGE 3 5 7 9 11 STAGE 3 5 7 9 STAGE en E Figure 1. Tobacco Flower Development. (A) Flower development from 8-mm bud to opening. Flower buds were divided into 12 stages on the basis of size [(D)] and morphological Anther Gene ExpressionPatterns 1203 distinct cell types. Transformationstudies with one antherspecific gene, designated as TA29 (Mariani et al., 1990; Seurinck et al., 1990), demonstrated that sequences responsible for the transcriptional activation of this gene in the tapetum are localized within a 122-bp 5' region. Destruction of the tapetum by a chimeric TA29 diphtheria toxin A-chain gene indicated that the tapetum is not required for the differentiation and/or function of other specializedcell types later in anther development. RESULTS Anther Development Has Two Distinct Phases Previously, we divided tobacco flower development into 12 stages to provide reference points for the expression of genesin differentfloral organ systems (Goldberg,1988). Figure 1A shows these stages, and the floral bud lengths and morphological markers used to describe each stage are summarized in Table 1. Figures 1B and 1C correlate each stage with the developmental state of the pistil and anthers. Measurements presented in Figure 1D indicate that both the flower bud (calyx plus corolla) and pistil increase in fresh weight and length continuously from stages 1 to 12. By contrast, the anthersincreaseinweight and length only until stage 5. These parameters remain relatively constant until stage 11 and then decrease precipitously when flower opening and anther dehiscence occur at stage 12 (Figures 1A and 1C). We prepared transverse anther sections at each stage to correlate changes in external morphology (Figure 1C) with the presenceof specific sporophytic tissues and cell types and with the presence of microspores and pollen grains. Bright-field photographs of representative anther sections are shown in Figure 2A, and the major tissues present and developmental events that occurred from stages 1 to 12 are summarized in Table 1. At stage 1, all major anther tissues have been specified. These included the epidermis, endothecium, vascular bundle, connective, stomium, and tapetum. In addition, meiosis was finished and the microspores were bound together as tetrads in the pollen sac. Following stage 1, the pollen grains differentiated and the anther underwent a series of events leading to dehiscence and pollen release (Figure 2A and Table 1). For example, by stage 8, the tapetum was degenerated, the connective separating the pollen sacs had begun to degrade, and the anther wall showed signs of splitting in the stomium region (Figure2A and Table 1). At stage 11, the connective was absent, the anther was bilocular,and pollengrains filled the locules (Table 1). We characterized anthers at earlier developmental stages to determine when the major histodifferentiation events occurred. Figure 1E shows pictures of flower bud development before stage 1. We divided these buds into seven stages on the basis of both length (Table 1) and morphologicalmarkers observed in the scanning electron microscope (Figure 1F). Early flower bud stages were designatedwith a minussignto indicatethat they occurred before the completion of meiosis within the anther (Table 1). The scanning electron micrographs shown in Figure 1F and the transverse sections shown in Figure 2A indicate that anther primordia were present in stage -7 flower buds and that the differentiation of specific anther tissues had begun. For example, archesporial cells, connective tissue, prevascular tissue, and epidermis were already visible at this developmental stage (Table 1). Between stages -7 and -1, the anther acquired its characteristic external shape (Figure 1F), all major anther tissues were differentiated (Table 1), and the microspore mother cells Figure 1. (continued). criteria(Table 1). Stage 1 was designated as the periodwhen tetrads are present within the anther and corresponds to an 8-mmflower bud. Stage 12 was designated as the time of flower opening and anther dehiscence. The transitionfrom stage 1 to stage 12 occurred over a 1-week period during the summer. (B)Pistil development from stage 1 to stage 12. Pistillengths and weights are presented in (D). (C)Anther development from stage 1 to stage 12. Anther lengths and weights are presented in (D). (D) Changes in organ system weight and length during flower development. Fiveflower buds were picked at each developmentalstage. and their lengths and fresh weights were measuredindividually.Similar measurementswere madefor the anthers and pistilof eachflower bud. Data points represent the meanof these five measurements.Anther fresh weights represent the collective weight of all five anthers withinthe flower bud. (E) Flower development from 0.75" bud to 7-mm bud. Young flower buds were divided into seven stages on the basis of size and developmentalevents occurring within the anther (Table 1). Stage -7 was designated as the period when tissue differentiationbegins in the anther primordium [(F)] and stage -1 was designated as the periodjust before the completion of meiosis withinthe anther (Figure2 and Table 1). Stage -7, -6, -5, -4, -3, -2, and -1 flower buds averaged 0.75 mm, 1.5 mm, 3 mm, 4 mm, 5 mm, 6 mm, and 7 mm in length, respectively. The transition from stage -7 to stage -1 occurred over approximately 5 days. Photographs were taken using a dissecting microscopewith a magnificationfactor of X75. (F) Scanning electron micrographs of anthers during the early stages of development. Flower buds at the designated stages were harvested, and their sepals and petals were removed to expose the anthers. The dissected flower buds were then photographed in the scanning electron microscope as outlined in Methods. Magnification factors for the stage -7, -6, -5, -4, and 1 anthers were x170, x120, x120, x125, and x60, respectively. 1204 The Plant Cell Table 1. Major Events during Tobacco Anther Development Flower Anther Bud Stage Length' Morphological Markersb Tissues Present" Major Events and Morphological Markersd -7 -6 -5 -4 -3 -2 -1 1 2 3 4 5 6 7 8 9 10 11 12 0.75 1.5 3 4 5 6 7 8 11 14 16 20 22 28. 39 43 45 47 46 Peta1and stamen primordia present; E, V, A, C Calyx closed; carpels not fused. E, V, Sp, P, C Carpels fused; stamenfilaments elongated; E, En, T, V, MMC, C carpels forming; calyx almost closed. petals equal in lengthwith anthers. Anthers yellowish; petals enclose anthers; E, En, T, V, MMC, C stigma forming. Style elongating; stamen filament extension E, En, T, V, MMC, C continues; anthers below stigma. Style clearly elongated; ovary expanded; anthers below stigma. Petals approachingtop of sepals. Anthers and pistil fully differentiatedand green. Calyx opens slightly at top of bud Corolla emerges from calyx. Sepals completely separated at top of Corollatube bulgejust inside calyx. Corolla tube bulge at tip of calyx. calyx. Corolla tube bulge above calyx; petals Corolla elongating; petals green and Corolla tube bulge enlarging; petal tips Corolla limb beginningto open; petaltips Corolla limbhalfway open; stigma and closed. slightly open. becoming pink. pink. anthers visible. E, En, T, V, MMC, C, S E, En, T, V, MMC, C, S E, En, T, V, TDS, C, S E, En, T, V, Msp, C, S E, En, T, V, Msp, C, S E, En, T, V, Msp, C, S E,En,T, V, Msp, C,S E, En, T, V, Msp, C, S E, En, V, Msp, C, S E, En, V, Msp, C, S E,En, V, Msp, C, S E, En, V, Msp, S E, En, V,Msp, S Flower open; anthers dehisced; corollalimb E, En, V,PG fully expanded and deep pink. Roundedprimordium; tissue differentiationbegun. lntense mitotic activity in four corners; invaginationof inner side. Wall layers including endothecium and tapetum beingformed; connective established. Tapetum and pollen sacs distinct; inner and outer tapetum morphologically different;middlelayer crushed. Meiosis begins; callose deposition betweenmicrospore mother cells evident. Meiosis in progress; tapetum large and multinucleate; stomium differentiation begins; thick callose walls between microspore mother cells. Meiosis in progress; continued stomium differentiation. Meiosis complete; microspores in tetrad; stomium differentiated; all sporophytic tissues formed. Microspores separate. Tapetum shrunken; secondary thickening in outer wall layers; pollen grains begin to form. Cells adjacent to stomium degenerated; tapetum degenerating. Secondary thickeningin outer wall layers intensified. Remnants of tapetum present; microspore nucleus dividing. Degradationof connective tissue in stomium region. Disruptionof connective tissue separating pollensac. Continued connective degradation. Connectivetissue almost fully degraded; pollen binucleate. Anthers bilocular; connective absent; locules filled with maturepollen grains. released. Anthers dehisce along stomium; pollen a Mean of five individualdeterminations expressedin millimeters. Datafrom stages 1 to 12 taken from those presentedin Figure 1D. Major markers that can be used with bud length measurements to identify stage. Tube refers to the white elongated portion of the corolla; limb refers to the top pink corolla region where the petaltips are separated from each other (Figure 1A). Markers for stages -7 to -1 taken from the scanning electron micrographs (Figure 1F); markers for stages 1 to 12 were taken from visual inspectionof buds (Figure1A). A, archesporial cells; C, connective; E, epidermis; En, endothecium;MMC, microspore mother cells; Msp, microspores; P, parietallayer; PG. pollen grains; S, stomium; Sp, sporogenous cells; T, tapetum; TDS, tetrads; V, vascular tissue. Taken from bright-fieldphotographs of transverse anther sections (Figures 2 and 14). Taken from bright-fieldphotographsof transverse anther sections (Figures 2 and 14). Anther GeneExpressionPatterns 1205 were undergoing meiosis (Figure 2A). Taken together, these data indicate that anther developmentalstages can be correlatedwith floral budlengths andthat there aretwo major phases of anther development: phase 1-tissue differentiation, microspore mother cell specification, and meiosis (stages -7 to -l), and phase 2-growth, pollen grain differentiation, and tissue degeneration and dehiscence (stages 1 to 12). Anther-Specific Clones Were ldentified in an Anther cDNA Library We constructed a stage 6 (Figures I A and 1C) anther cDNA library and then screened this library for cDNA clones that representedmRNAs present exclusively,or at elevated levels, in the anther (see Methods). We used stage 6 anther mRNA to construct the library becauseour earlier solution DNA/RNA hybridization studies indicated that stage 6 anthers contain a large number of antherspecific mRNAs (Kamalay and Goldberg, 1980,1984).We obtained 58 anther-specific cDNA clones out of a screen of 768 plasmids and then sorted 26 of these clones into groups by RNA gel blot and cross-hybridization analyses. Sixteen cDNA clones failed to cross-hybridize at a moderately stringent criterion(42OC,50% formamide, 1M Na') andwere designatedas unique.The remainderwere found to belongto four different cross-hybridizinggroups. Table 2 lists representative cDNA clones from three of the groups, as well as others that sorted into the unique category. DNA sequencing studies (Seurinck et al., 1990; J. Seurinck, J. Leemans, and R.B. Goldberg, unpublished results) showed that the TA32136 and TA56157 groups representedmRNAsencodinglipidtransfer proteins(Bouillon et al., 1987; Takishima et al., 1988)and thiol endopeptidase proteins (Miller and Huffaker, 1982; Mitsuhashiand Minamikawa, 1989), respectively. In addition, the TA131 TA29 group represented mRNAs that encoded glycinerich proteins with properties of cell wall proteins (Condit andMeagher, 1986;Kelleret al., 1988;Varner and Cassab, 1988). By contrast, the unique anther cDNA clones sequenced to date (e.g., TA20, TA25, and TA26) failed to show relatednessto any known mRNA or protein. We hybridized representative anther cDNA probes with gel blots containing floral and vegetative organ system mRNAs. As shown in Figure3, all probes producedstrong hybridization signals with anther mRNA. Prevalence estimates indicated that these mRNAs represented from 0.05% to 0.5% of the anther mRNA, depending upon the message(Table 2). ln most cases, the cDNA probesfailed to produce a detectable signal with heterologous organ mRNAs, even after the gel blots were exposed 50 times longer than that requiredto produce a weak anther signal. By contrast, three cDNA clones (TA20, TA26, TA4) produced signals with other mRNAs in addition to anther mRNA; however, in each case, the heterologous hybridization signal was at least 10-foldlower than that observed in the anther. In addition, each of these cDNA clones produced a distinct hybridizationpattern with the heterologous mRNAs (Figure 3). For example, the TA20 probe hybridizedwith pistil and peta1mRNAs,whereas the TA26 probe hybridized only with leaf mRNA. Taken together, thesedata indicatethat we haveidentifiedsevera1different anther-specific cDNA clones, that these clones represent relativelyprevalent anther mRNAs, and that most of these mRNAs are either undetectable in other organ system mRNAs at our sensitivity level (e.g., TA13, TA29, TA36) or are present at a level at least 10-fold lower than that observed in the anther (e.g., TA4, TA20, TA26). Gene Expression1s Temporally Regulatedduring Anther Development We hybridized representative anther cDNA plasmids with mRNAsfrom different anther stages to determinewhether their corresponding genes were regulated temporally duringantherdevelopment.These plasmidsincludedthe eight listed in Table 2 in addition to 11 others. Figures 4A and 4B present the resultsof RNA dot blot studies, and Figure 4C shows duplicate RNA gel Dlot experiments with some of the same plasmids. Three distinct geneexpression patternswere observed. First, Figures 4A and 4C show that most anther mRNAs investigatedaccumulated coordinately between stages 1 and4 andthen their levelsdeclinedsignificantlyafter stage 6. These mRNAs includedthose encoding the glycine-rich proteins(TA13, TA29) andthe lipidtransfer proteins(TA32, TA36). Second, Figure48 shows that the TA20 and TA25 mRNAswere presentat relativelysimilar levelsin all stages studied. Closer inspection of the data, however, indicated that these mRNAs increased slightly in prevalence between stages 2 and 3 (Figure4B). Finally, Figures 48 and 4C show that the thiol endopeptidase mRNAs (TA56, TA57) accumulated between stages 1 and 6, remained relatively constant until stage 8, and then declined to low levelsbeforedehiscenceat stage 11.Together, these data indicate that many anther-specific genes are regulated temporally and that there are timing differences in anther gene expression patterns. Gene Expression1s Spatially Regulated during Anther Development We hybridizedsingle-stranded35S-RNAprobeswith anther sections in situ (see Methods) to correlate the temporal gene expression patterns observed during anther development (Figure 4) with the presence of specific anther tissues and cell types (Figure2 and Table 1). 1206 The Plant Cell TA29/DTAWILD TYPE En Figure 2. Bright-Field Photographs of Tobacco Anther Development in Untransformed Plants and Plants Transformed with a Chimeric TA29/DTA Gene. Anthers at the designated stages were fixed, embedded with paraffin, and sliced into 10-Mm transverse sections as described in Methods. The fixed sections were stained with toluidine blue and photographed with bright-field illumination. A, archesporial cells; C, connective; E, Anther Gene ExpressionPatterns 1207 Table 2. Anther-Specific mRNAs mRNA Size" Prevalenceb Protein Encoded" TA13' TA20" TA25" TA26" TA2gd TA32' TA36' TA56O 1.1, 1.2 0.4 0.6 0.5 0.7 0.2 0.6, 0.7 0.2 1.1, 1.2 0.2 0.6, 0.7 0.05 0.6, 0.7 0.05 1.5 0.05 GIycine-rich Unknown Unknown Unknown Glycine-rich Lipid transfer Lipid transfer Thiol endopeptidase a Taken from Figure3. Percentage of stage 3 anther polysomal poly(A) mRNA. Estimated relativeto calibratedRNA standards. The TA13 and TA29 mRNA prevalences were estimated using mRNA-specific oligonucleotide probes. Translationof DNA sequence (Seurinck et al., 1990), or highest identity score with proteins in the GenBank. Different members of the same family. TA13 and TA29 mRNAs are 90% similar at the nucleotide level (J. Seurinck, J. Leemans, and R.B. Goldberg, unpublisheddata). e Not significantly similar at the nucleotide level to each other or to the other mRNAs listed in this table (J. Seurinck, J. Leemans, and R.B. Goldberg, unpublisheddata). 'Different members of the same gene family. TA32 and TA36 mRNAs are 63% similar at the nucleotide level (J. Seurinck, J. Leemans,and R.B. Goldberg, unpublisheddata), and their cDNA plasmidsonly weakly hybridizeat a moderatelystringentcriterion (42OC, 50% formamide, 1.2 M Na'). The TA56 andTA57 cDNA plasmids(Figure48) cross-hybridize at a moderately stringent criterion. Severa1DifferentAnther mRNAs Are Localized within the Tapetum Figure 5 shows the localization patterns of the TA26, TA29, and TA32 mRNAs (Table2). As seen in Figures5A to 5C, no hybridization grains above background were observedwith the TA29 mRNAcontrol probe. By contrast, Figures 5D to 5L show that the TA26, TA29, and TA32 anti-mRNA probes produced identical, tapetal-specific localization patterns, as did TA13 and TA36 anti-mRNA probes (data not shown). For example, the TA29 antimRNA probe producedan intensehybridizationsignalover the tapetum at stage 3 (Figure 5H) and did not produce any hybridizationgrains above backgroundover any other anther regions. Nor were there any detectable hybridization signals produced at stage 1 (Figure 5G)and stage 6 (Figure51). The hybridization signals at each stage correlated directly with the mRNA levels observed in the RNA dot blot and gel blot experiments(Figures4A and 4C). The absence of a detectable signal at stage 6 (Figures 5F, 51, and 5L)coincidedwith the degenerationprocessoccurring in the tapetum (Figure2 and Table 1). We hybridized 3H-poly(U)with the anther sections to localizethe distribution of total poly(A)RNA molecules. In contrast with the results obtained with the TA13, TA26, TA29, TA32, and TA36 anti-mRNA probes (Figure 5), the distribution of grains was uniform throughout the anther at each stage investigated and grains were present over all tissues (data not shown). Together, these results indicate that severa1 different mRNAs that accumulate and decay coordinately during early phase 2 of anther development are localized within the tapetum and that these mRNAs include those encoding the glycine-rich proteins (TA13, TA29) and the lipidtransfer proteins (TA32, TA36). The TA56 ThiolEndopeptidasemRNA 1s Localized wifhin the Connectiveand Stomium The TA56 mRNA localizationpattern is shown in Figure6. Figures6A to 6D show that the TA56 mRNAcontrol probe producedno hybridizationgrains above background levels with anther sections from stage 1 (Figure 6A), stage 3 (Figure6B), stage 8 (Figure6C), and stage 11(Figure6D). By contrast, Figures 6E to 6H show that the TA56 antimRNA probe produced a localized hybridizationsignal at all stages investigated and that both the location and intensity of the hybridizationgrains changed with the developmental state of the anther. Bright-field photographs of the anther regions containing the TA56 mRNA are shown in Figures 61 to 6L, and Figures 6M to 6P show high-magnification dark-field photographs of the hybridization grains at each stage. As predicted from the RNA accumulation studies (Figures 48 and 4C), a relatively weak TA56 hybridization signal was observed at stage 1 of anther development (Figures 6E and 6M). This signal was localized on both sides of the anther over a circular cluster of cells between the stomium and the connective (Figures 6E, 61, and 6M). At stage 3, the TA56 mRNA hybridizationsignal intensified Figure 2. (continued). epidermis; En, endothecium; MMC, microspore mother cells; Msp, microspores; PS, pollen sac; S, stomium; Sp, sporogenous cells; T, tapetum; TDS, tetrads; V, vascular bundle. Magnification factors for stage -7, -6,--2, 1, 3, and 8 anthers were x400, x200, x100, ~ 1 0 0 ,x100, and ~ 1 0 0 ,respectively. (A) Anther developmentin wild-type untransformed plants. (6)Anther developmentin plantstransformedwith a chimericTA29/DTA gene. Tobacco plantswere transformedwith a TA29IDTA gene as described in Methods. The TA29/DTA gene contained a TA29 5' DNA fragment containing nucleotides -1477 to +51 (Mariani et al., 1990; Seurinck et al., 1990) fused with the DTA gene coding sequence (Greenfieldet al., 1983; Maxwell et al., 1986; Palmiter et al., 1987). ORGAN SYSTEM cDNA A Pi P L S R kb TA13 TA29 TA20 TA25 TA26 TA36 TA39 TA51 TA4 _ . -1.1 1.1 -0.6 -0.7 -0.7 -0.6 -0.6 -0.7 -0.6 -0.9 -0.5 Figure 3. Representation of Anther-Specific mRNAsin Floral and Vegetative Organ Systems. Tobacco anther (A), pistil (Pi),petal (P), leaf (L),stem (S), and root (R) polysomal poly(A) mRNAs were fractionated on denaturing agarose gels, transferred to nitrocellulose paper, and hybridized with labeled plasmid DMAprobes as outlined in Methods. Anther and pistil mRNAswere isolated from stage 6 flower buds (Figure 1), whereas petal mRNA was isolated from stage 12 flowers (Figure 1). Leaf, stem, and root mRNAswere isolated from plants described in Methods. One microgram of each mRNA was used for the TA29 gel blot, and 0.5 ng of each mRNA was used for the TA13, TA20, and TA25 gel blots. By contrast, the TA26, TA35, TA51, and TA4 gel blots contained 0.1 ng of anther mRNA and 1 ^g of all other mRNAs. Film exposure times varied for each gel blot but, with the exception of the TA20 gel blot, were the same for all RNA lanes. Exposure times for the TA20 petal, leaf, stem, and root RNAlanes were approximately 30 times longer than that used for the anther and pistil RNA lanes. but remained localized over the same circular cell cluster (Figures 6F, 6J, and 6N). No hybridization grains above background levels were observed over any other anther tissues at this stage, including the epidermis, connective, endothecium, tapetum, and wall. By stage 8, however, the TA56 mRNA localization pattern changed. Figures 6G, 6K, and 6O show that the TA56 mRNA was now localized over both stomium regions and was uniformly distributed over the connective separating the pollen sacs. By contrast, TA56 mRNA was undetectable in the circular cell cluster between the stomium and connective because these cells had degenerated (Figure 6K). Finally, at stage 11 after the connective degenerated (Table 1), the TA56 mRNA hybridization signal was reduced significantly but was still localized over the stomial cells (Figures 6H, 6L, and 6P). Together, these data indicate that the TA56 thiol endopeptidase mRNA is present in specific anther cell types and that the appearance of this mRNA occurs just before the degeneration of these cells in anther development. The TA20 mRNA Is Distributed throughout the Anther and Is Localized in Specific Pistil Regions Figures 3 and 4B show that the TA20 mRNA is present throughout phase 2 of anther development and is also present in the pistil at a level 10-fold lower than that observed in anthers of the same stage flower bud. We hybridized a TA20 anti-mRNA probe with anther and pistil sections to localize TA20 molecules within both of these floral organ systems. Figure 7A shows that this probe produced an intense hybridization signal over most anther tissues at stage 2, including the connective, stomium, and wall layers. Highest concentrations of grains occurred over the connective and over cells immediately surrounding the tapetum. By contrast, the TA20 anti-mRNA probe did not produce any hybridization grains above background levels (Figure 7E) over the vascular bundle, the tapetum, or the circular cell cluster between the stomium and connective that contained the TA56 thiol endopeptidase mRNA (Figures 6M and 6N). Figures 7B to 7D show that at later developmental stages the TA20 mRNA remained localized within the same tissues, but beginning at stage 6 became progressively concentrated in connective regions adjacent to the vascular bundle (Figure 7C), and by stage 8 became less prevalent throughout the anther (Figure 7D). Figures 8A to 8C show bright-field photographs of ovary cross-sections from a stage 4 pistil (Figure 8A) and a stage 6 pistil (Figure 8B), as well as a stigma/style longitudinal section from a pistil at stage 6 (Figure 8C). Two carpels can be visualized within the ovary of each pistil, along with locules, placenta, vascular bundle, ovules, and the ovary wall (Figures 8A and 8B). A prominent stigma and a style with transmitting tissue were also visualized within the pistil (Figure 8C). Figures 8D and 8E show that the TA20 mRNA was localized within the ovary wall and a narrow cell layer that connects the ovules to the placenta. No TA20 hybridization grains above background levels (data Anther Gene Expression Patterns 1209 (A) STAGE cDNA 1 2 3 4 5 6 7 8 9 10 11 Control TA13 • • • TA29 ••• TA1 • • TA32 •• • TA26 • • • TA33 • • TA36 • • • TA39 • • • TA45 • • • TA46 • • • TA51 • • • • * • * (B) STAGE cDNA 1 2 3 4 5 6 7 8 9 10 11 Control TA20 ••••••••••• TA25 ••••••••••• TAX •••••••••• TA54 •• ••••• TA56 • • "• •_• • • TA57 • • • • • • TA30 • • • • • • (C) STAGE cDNA 1 2 3 4 5 6 7 8 10 Kb TA26 •§••• tflff -0.7 -0.6 TA32 »- TA56 -0.7 -1.5 Figure 4. Accumulation of Anther-Specific mRNAs during Anther Development. Polysomal poly(A) mRNAs were isolated from anthers at different developmental stages (Figure 1) and either spotted onto nitrocellulose paper [(A)and (B)] or fractionated on denaturing agarose gels and transferred to nytran [(C)]. Dot blots and gel blots were hybridized with labeled plasmid DMAs as outlined in Methods. Film exposuretimes differed for each dot blot or gel blot developmental series but were the same for each RNA hybridized with a given DNA probe. (A) Hybridization of anther cDNA plasmids with anther mRNAs at different developmental stages. The TA13and TA29 dots contained 0.25 fig of mRNA; the other dots contained 0.05 »g of mRNA. Control dots contained an equivalent amount of soybean embryo polysomal poly(A) mRNA. (B) Hybridization of anther cDNA plasmids with anther mRNAs at different developmental stages. The TA20 and TA25 dots contained 0.25 ^g of mRNA; the other dots contained 0.05 //g of mRNA. Control dots contained an equivalent amount of soybean embryo polysomal poly(A) mRNA. not shown) were observed over the ovules, vascular bundles, and most of the placenta. Figure 8F shows that TA20 hybridization grains were also absent from the stigma and the transmitting tissue of the style. By contrast, TA20 mRNA molecules were localized within parenchyma tissue that extended from beneath the stigma surface through the style to the ovary (Figure 8F). Together, these data show that the TA20 mRNA is present in the majority of anther cells throughout phase 2 of development and is concentrated within specific regions of the pistil. We conclude from these data and those obtained with the other mRNAs (Figures 5 and 6) that temporal differences in anther gene expression programs correlate with the differentiation and degeneration of specific anther cell types. Most Anther Genes Are Represented only a Few Times in the Tobacco Genome We hybridized stage 6 anther 32 P-cDNA with an excess of tobacco DNA in solution to estimate the copy number of genes expressed in the anther. We utilized an internal 3 Hsingle-copy DNA standard to calibrate the hybridization kinetics (Goldberg et al., 1978, 1981). Figure 9A shows that 70% of the 32 P-cDNA hybridized with a rate constant identical to that of the 3 H-single-copy DNA (0.00045 M~1 sec"1 ). By contrast, 30% of the 32 P-cDNA hybridized with a rate constant of 0.047 M~1 sec"1 , suggesting that these sequences are reiterated approximately 100 times (0.047/ 0.0045) in the tobacco genome. These studies, and those presented previously (Kamalay and Goldberg, 1980), indicate that mRNAs comprising most of the stage 6 anther mRNA mass and sequence complexity are encoded by single-copy genes, or genes present only a few times (<5) per tobacco genome. We hybridized anther cDNA plasmids with DNA gel blots to estimate the copy number of individual anther-specific genes. We utilized several tapetal-specific cDNA plasmids for these experiments (Figure 5) and included those representing the glycine-rich protein mRNAs (TA29, TA13) and the lipid transfer protein mRNAs (TA32, TA36) (Table 2). Figure 9B shows that each of these probes hybridized with only three or four EcoRI fragments in the tobacco genome at the hybridization criterion employed (42°C, 50% formamide, 1 M Na+ ). For example, the TA29 plasmid probe hybridized with two similar-sized EcoRI fragments 6.5 kb to 7.0 kb in length and two additional EcoRI fragments 8.6 kb and 10.5 kb in length, respectively. The restriction map of the TA29 gene region, shown in Figure 10A, and sequencing studies with TA13 and TA29 cDNAs (Seurinck et al., 1990; J. Seurinck, J. Leemans, and R.B. (C) Hybridization of the TA26, TA29, TA32, and TA56 cDNA plasmids with anther mRNAs at different developmental stages. Each RNAgel blot contained 0.7 ng of mRNA per lane. The TA26, TA29, TA32, and TA56 RNA gel blots were exposed for 1.5 hr, 1.5 hr, 5 hr, and 96 hr, respectively. Figure 5. Localization of the TA26, TA29, and TA32 mRNAs during Tobacco Anther Development. Anthers were fixed, embedded in paraffin, sliced into 10-Mm sections, and hybridized with single-stranded 35 S-RNA probes as outlined in Methods. Photographs were taken by dark-field microscopy, and all film emulsion exposures were for 3.5 days. Corresponding brightfield photographs are shown in Figure 2. Magnification factors for the stage 1, stage 3, and stage 6 anthers were X100, X100, and x66, respectively. C, connective; E, epidermis; T, tapetum; V, vascular bundle. (A) to (C) In situ hybridization of a TA29 mRNA control probe with anthers at stage 1 [(A)], stage 3 [(B)], and stage 6 [(C)]. White grains represent background hybridization levels; white patchy areas in the anther wall represent light-scattering effects due to dark-field illumination through stained sections. (D) to (F) In situ hybridization of a TA26 anti-mRNA probe with anthers at stage 1 [(D)], stage 3 [(E)], and stage 6 [(F)]. White grains in the stage 3 anther represent regions containing RNA/RNA hybrids. (G) to (I) In situ hybridization of a TA29 anti-mRNA probe with anthers at stage 1 [(G)], stage 3 [(H)], and stage 6 [(I)]. White grains in the stage 3 anther represent regions containing RNA/RNA hybrids. (J) to (L) In situ hybridization of a TA32 anti-mRNA probe with anthers at stage 1 [(J)], stage 3 [(K)], and stage 6 [(L)]. White grains in the stage 3 anther represent regions containing RNA/RNA hybrids. Goldberg, unpublished data) indicated that these EcoRI fragments were derived from cleavage of the TA13 and TA29 genes. DNA sequencing studies (J. Seurinck, J. Leemans, and R.B. Goldberg, unpublished results) also suggested that each TA32 and TA36 EcoRI fragment represented a different gene (Figure 9B). Together, these data indicate that each tapetal-specific mRNA investigated is encoded by a small gene family. We hybridized the TA29 cDNA plasmid with gel blots containing tobacco DNA (Nicotiana tabacum) and the Anther Gene Expression Patterns 1211 Figure 6. Localization of the TA56 mRNA during Tobacco Anther Development. Anthers were fixed, embedded in paraffin, sliced into '\0-nm sections, and hybridized with single-stranded 35 S-RNA probes as outlined in Methods. C, connective; E, epidermis; En, endothecium; S, stomium; T, tapetum; TDS, tetrads. (A) to (D) In situ hybridization of a TA56 mRNA control probe with anthers at stage 1 [(A)], stage 3 [(B)], stage 8 [(C)], and stage 11 [(D)]. Photographs taken with dark-field microscopy, and film emulsion exposure times were for 13 days. White grains represent background hybridization levels; white patchy areas represent light-scattering effects due to dark-field illumination through stained anther sections. The magnification factor for anthers at each stage was X100. (E) to (H) In situ hybridization of a TA56 anti-mRNA probe with anther sections at stage 1 [(E)], stage 3 [(F)], stage 8 [(G)], and stage 11 [(H)]. Photographs were taken with dark-field microscopy, and film emulsion exposure times were for 13 days. White grains represent regions containing RNA/RNA hybrids. The magnification factor for anthers at each stage was xlOO. (I) to (L) Bright-field photographs of anther stomium region at stage 1 [(I)], stage 3 [(J)], stage 8 [(«)], and stage 11 [(!_)]. Anther regions correspond to those showing intense hybridization in (E) through (H) and (M) through (P).The magnification factor for anthers at each stage was x400. (M) to (P) In situ hybridization of a TA56 anti-mRNA probe with anther stomium region at stage 1 [(M)], stage 3 [(N)], stage 8 [(O)], and stage 11 [(P)]. Dark-field photographs represent the stomium region of anthers shown in (E) to (H) at a magnification factor of x350. DMAs of its diploid progenitor species (N. tomentosiformis and N. sylvestris; Goodspeed, 1954)to determine whether genes encoding this tapetal-specific mRNA were derived from one or both tobacco parental lines. As seen in Figure 9C, the gel blot hybridization patterns obtained with all three tobacco DNAs were similar with the exception of 1212 The Plant Cell one or two polymorphisms. Polymorphic DNA fragments were not unexpected because our previous experiments showed that N. tomentosiformis and N. sylvestris singlecopy DNAs had diverged considerably from each other (Okamuro and Goldberg, 1985). Together, these results indicate that members of the TA13/TA29 glycine-rich protein gene family are present in both parental species of N. tabacum and that these genes were probably present in the genus before the divergence of N. sylvestris and N. tomentosiformis from a common ancestor (Goodspeed, 1954). The TA29 Gene Does not Contain Introns and Is Contiguous to a ConstitutivelyExpressed Gene We isolated a genomic clone containing the TA29 gene (Figure 10A) to begin to dissect processes that regulate the coordinate expression of the tapetal-specific gene set (Figures 3 to 5). We hybridized ATA29 phage DNA with anther mRNA under conditions that form R-loops to visualize TA29 gene structure in the electron microscope (see Methods). Figure 10A shows that the TA29 R-loops were simple in structure and lacked detectable introns, and that their average length was approximately the size of the TA29 mRNAs. The TA29 gene' sequence confirmed the absence of introns and revealed that this gene has a continuous open reading frame of 963 bp (Seurinck et al., 1990). Translation of the reading frame indicated that the TA29 gene encoded a 33-kD glycine-rich protein that has several potential glycosylation sites and is organized into two similar 120-amino acid modules that are highly hydrophobic (Seurinck et al., 1990). A schematic representation of the TA29 gene, including the positions of relevant consensus sequences, is shown in Figure 10A. We hybridized the XTA29 phage DNA with gel blots containing floral and vegetative organ system mRNAs to determine whether other genes were present in the TA29 gene region. Figure 10B shows that the phage probe produced an intense hybridization signal with the 1.1-kb and 1.2-kb TA29 anther mRNAs. In addition, this probe Figure 7. Localization of the TA20 mRNA during Tobacco Anther Development. Anthers were fixed, embedded in paraffin, sliced into 10-Mm sections, and hybridized with single-stranded 35 S-RNA probes as outlined in Methods. Photographs were taken by dark-field microscopy, film emulsion exposure times were 6 days, and the magnification factor for anthers at each stage was xlOO. Corresponding bright-field photographs are shown in Figure 2. C, connective; S, stomium; T, tapetum; V, vascular bundle; W, wall. (A) to (D) In situ hybridization of a TA20 anti-mRNA probe with anthers at stage 2 [(A)], stage 4 [(B)], stage 6 [(C)], and stage 8 [(D)]. White grains represent regions of RNA/RNA hybridization. (E) In situ hybridization of a TA20 mRNA control probe with a stage 4 anther. White grains represent background hybridization levels. Anther Gene Expression Patterns 1213 Figure 8. Localization of the TA20 mRNA within the Tobacco Pistil. Pistils were fixed, embedded in paraffin, sliced into 10-Mm sections, and hybridized with a TA20 anti-mRNA probe as outlined in Methods. The film emulsion exposure times and magnification factors were 6 days and x40, respectively. L, locule; O, ovule; P, placenta; TTi,transmitting tissue; Sti, stigma; Sty,style; V, vascular bundle; W, wall. (A) and (B) Bright-field photographs of ovary cross-sections from pistils at stage 4 [(A)]and stage 6 [(B)]. (C) Bright-field photograph of a stigma and style longitudinal section from a stage 6 pistil. (D) and (E) In situ hybridization of a TA20 anti-mRNA probe with ovaries from pistils at stage 4 [(D)] and stage6 [(E)].Photographs were taken by dark-field microscopy. White grains represent regions of RNA/RNA hybridization. (F) In situ hybridization of a TA20 anti-mRNA probe with the stigma and style of a stage 6 pistil. Photograph was taken by dark-field microscopy. White grains represent regions of RNA/ RNA hybridization. hybridized weakly with a 0.9-kb mRNA in all other organs. The prevalence of the 0.9-kb mRNA in the leaf, root, pistil, and petal was at least 2 orders of magnitude lower than that observed for the TA29 anther mRNAs. By contrast, the 0.9-kb mRNA was approximately 10-fold more prevalent in the stem than in the other organs (Figure 10B). We localized DNA sequences responsible for producing the 0.9-kb mRNA signal to a region just 3' to the TA29 gene (Figure 10A). R-loop studies with stem mRNA(Figure 10A) and the DNA sequence of this region (Seurinck et al., 1990) revealed the presence of a gene with two introns that was transcribed in the opposite direction as the TA29 gene. We designated this gene as TSJT1 (Seurinck et al., 1990), and a schematic representation of its structure and position in the TA29 gene region is shown in Figure 10A. The TSJT1 coding sequence was not related to any known gene or protein in the GenBank. We hybridized a DNA fragment containing a portion of the TSJT1 gene with gel blots containing floral and vegetative organ system mRNAs to determine whether it was also active in the anther. Figure 10C shows that a 0.9-kb TSJT1 mRNA signal was produced with anther mRNA at a level equivalent to that observed with all organ system mRNAs except the stem. Together, these data show that the TA29 tapetal-specific gene has a relatively simple structure and that it is contiguous to a convergently transcribed gene that is differentially regulated and expressed in all organ systems. The TA29 Gene Is Regulated at the Transcriptional Level We hybridized DNA gel blots containing TA29 gene sequences with 32 P-RNAssynthesized in isolated anther and leaf nuclei (see Methods) to determine whether the organ specificity of the TA29 gene was controlled by transcriptional or post-transcriptional processes. We included TSJT1 DNA sequences on these blots as a control because this gene was active in both the anther and the leaf (Figure 10C). The results are shown in Figures 11A and 11B and are summarized schematically in Figure 11 C. Figure 11 shows that both the TA29 gene (Figure 11 A) and the TSJT1 gene (Figure 11 B) hybridized with anther 32 P-nuclear RNA. The TSJT1 hybridization signal was slightly more intense than that obtained with the TA29 gene, even though the 0.9-kb TSJT1 mRNA was approximately 100-fold less prevalent than the TA29 mRNAs in the anther (Figures 10B and 10C). By contrast, no detectable TA29 gene hybridization signal was observed with leaf 32 P-nuclear RNA (Figure 11 A), even though this probe produced a signal with DNA fragments representing the TSJT1 gene (Figure 11 B). Together, these data indicate that the TA29 gene is regulated primarily at the transcriptional level and that post-transcriptional processes contribute to establishing TA29 and TSJT1 mRNA levels in the anther. 1214 The Plant Cell (A) (B) (C) TA29 TA26 TA32 TA36 I I I Copy* IX 5X I Copy * Rl H3 Rl H3 Kl H3 IX 5X Jib. Jib. kb kb -10.5 -8.6 , -7.0 '-6.5 -13.0 — -8.2 «* -6.6 '-7.4 '-4.5 :•! 10.5- 8.6- 7.0- 6.5- 1.9- 1.75- 1.65" "- 'icr1 10° EQUIVALENT COT Figure 9. Representation of Anther mRNA Sequences in the Tobacco Genome. (A) Hybridization of stage 6 anther cDNA with an excess of tobacco DMA. Trace amounts of 32 P-anther cDNA and 3 H-single-copy DMA were hybridized together with an excess (>25,000/1 mass ratio) of unlabeled total genomic DNA, and the extent of hybridization was measured by hydroxyapatite chromatography (see Methods). The single-stranded fragment size of driver and tracer DMAs was 0.3 kb. The reassociation kinetics of the driver genomic DNA weredescribed elsewhere(Zimmerman and Goldberg, 1977; Okamuro and Goldberg, 1985). Curves through the data points represent the best least-squares solution for either one second-order component (single-copy DNA) or two second-order components (anther cDNA). For these solutions, the rate constant of the slowest hybridizing (single-copy) component was fixed at 0.00045 M~' sec'1 , a value obtained previouslyfrom several independent reassociation experiments with tobacco single-copy DNA (Goldberg et al., 1978; Kamalay and Goldberg, 1980; Okamuro and Goldberg, 1985). Other solutions to these data points produced higher root mean square errors. The fraction of DNA fragments, second-order rate constant, and averagecopy number for each component were: 3 H-cDNA; component 1, 0.29, 0.047 M~1 sec~1 , 100; component 2, 0.71, 0.00045 M~' sec~\ 1; 32 P-singlecopy DNA; component 1, 1.0, 0.00045 M~1 sec"', 1. The fraction of DNA fragments in each component was normalized to 100% tracer reactivity (Kamalay and Goldberg, 1980; Okamuro and Goldberg, 1985). (B) Hybridization of tobacco DNA gel blots with labeled TA26, TA29, TA32, and TA36 plasmid DMAs. Tobacco DNA was digested with EcoRI, fractionated by electrophoresis on agarose gels, transferred to nitrocellulose, and hybridized with each labeled plasmid DNA as outlined in Methods. The reconstruction lanes contained one copy (1X) and five copy (5x) equivalents of EcoRI-digested XTA29 phage DNA (Figure 10) and were calculated using a XTA29 phage DNA length of 43 kb (Figure 10) and an N. tabacum genome size of 2.4 x 106 kb [(A) and Okamuro and Goldberg, 1985]. The TA29 probe hybridized with TA13 sequences at the criterion employed, whereas the TA32 and TA36 probes did not hybridize significantly to each other (Table 2). (C) Hybridization of labeled TA29 plasmid DNA with gel blots containing DNAs of different tobacco species. Tobacco DNAswere digested with the designated restriction endonucleases, fractionated by electrophoresis on agarose gels, and hybridized with labeled TA29 plasmid DNA as described in Methods. The one copy (1 x) and five copy (5x) reconstruction lanes were calculated on the basis of the N. tabacum genome size as described in (B). A 122-bp 5' Region ProgramsTA29 Gene TapetalSpecific Expression We showed elsewhere (Mariani et al., 1990) that the TA29 gene 5' region can direct Escherichia coli /3-glucuronidase (GUS) gene expression within the tapetum, and that the chimeric TA29/GUS gene is regulated exactly like the endogenous TA29 gene (Figures 3 to 5). We generated deletions of the TA29gene 5' region to localize sequences required to activate TA29 gene transcription within tapetal cells. These deletions were fused with the GUS gene and then the chimeric TA29/GUS genes were transferred to tobacco plants (see Methods). A schematic representation of some of the chimeric TA29/GUS genes is shown in Figure 12A. We obtained 170 transformants representing 20 different 5' deletions ranging from -5670 to +18 nucleotides relative to the TA29 transcription start site (Figures 10A and 12A).The number of individual transformants per deletion varied from one (-2070) to 19 (-1990), but averaged approximately nine. DNA gel blot studies and genetic analysis of kanamycin-resistant (Knr ) gene segregation patterns showed that most transformants had one unrearranged copy of the TA29/GUS gene (data not shown). Anther Gene Expression Patterns 1215 TA29 Anther TSJTI Stem (B) IX 100X A Pi P L S R 1.2- VI- f kb -0.9 (C) S A Pi P L S R kb 0.9Figure 10. Organization of the TA29 Anther Gene Region. (A) Structures of the TA29 anther gene and contiguous TSJT1 stem gene. The TA29 cDNA plasmid was used to isolate a 13.2kb genomic clone from an EcoRI ACharon 32 tobacco DMAlibrary (Seurinck et al., 1990). DMAgel blot experiments (Figures 9B and 9C) showed that this clone represents an unrearranged copy of the TA29 genomic region. The 6.2-kb Clal/Aval region containing the anther and stem genes was sequenced by Seurinck et al. (1990). Large arrows represent gene locations and transcriptional orientations. Solid and open regions represent exons and introns, respectively. TA29 nucleotides +1, +1141, and +1238 indicate the transcription start site and 3' gene ends as determined by both S1 nuclease and primer extension analyses. The TA29 cDNA plasmid represents nucleotides +452 to +821. The TSJT1 stem gene is numbered relative to the ATG start codon because the transcription start site has not yet been determined. Nucleotide coordinates for each exon, as well as the start and stop codons, are boxed. Fusion of the three stem gene exons results in an open reading frame of 498 nucleotides, which is approximately 60% of the mRNA length [(B) and (C)]. The TA29 gene 3' end is only 412 nucleotides from the TSJT1 gene stop codon (Seurinck et al., 1990). Only relevant XTA29 restriction endonuclease sites are shown. Sites contained within the 6.2-kb Clal/Aval region can be obtained from the DMA sequence (Seurinck et al., 1990). XL and Xn refer to the XCharon 32 left and right arms, respectively. The R-loops were formed by hybridizing a sequence excess of either anther mRNA or stem mRNA with XTA29 DMAas described by Fischer and Goldberg (1982). The bar in the electron micrographs represents 1 kb. E and I refer to exons and introns, respectively. The number-average anther gene R-loop size was 0.84 ± 0.2 kb (n = 26). The R-loop and intron sizes for the stem gene were: 0.31 kb (E1),0.22kb (11), 0.29 kb (E2), 0.42kb (12), and 0.35kb (E3) (n = 1). The anther and stem gene R-loops were oriented relative to the XTA29 restriction endonuclease map and the XCharon32 left and right arms. (B) Representationof XTA29 sequences in vegetativeand floral We used fluorimetric analysis to measure GUS enzyme activity in stage 4 anthers from 148 individual transformants. Although the GUS enzyme levels varied between transformants with a given deletion, the results showed that TA29/GUS gene transcriptional activity remained relatively constant as the 5' region was deleted from -5670 to -279 nucleotides (data not shown). By contrast, no detectable GUS enzyme activity was observed in anthers containing TA29/GUS genes with only 150 nucleotides of TA29 5' region (data not shown). We hybridized a GUS probe with gel blots containing pooled stage 4 anther mRNAs from five independent transformants per deletion (Figure 12A) to show directly that transcriptional control sequences were localized between -279 and -150 nucleotides from the TA29 gene transcriptional start site (Figures 10A and 12A). Figure 12B shows that GUS mRNA levels were similar in transformants containing TA29/GUS genes with 5270 to 279 nucleotides of the TA29 5' region. Relative to the RNA standards shown in Figure 12C, the GUS mRNA prevalence in these transformants averaged 0.1%, a value similar to that obtained for the endogenous TA29 mRNA gene at the same developmental stage (Table 2). In situ hybridization studies showed directly that tapetal cell GUS and TA29 mRNA levels were similar in TA29/ GUS transformants containing either 5270 or 1477 nucleotides of the TA29 5' region (Mariani et al., 1990; K.H. Cox and R.B. Goldberg, unpublished results). By contrast, no detectable GUS mRNA was observed in transformants containing either 150 nucleotides of the TA29 5' region or no TA29 5' region, even though endogenous TA26 mRNA levels were similar to those obtained in all of the other transformants (Figure 12B). This result indicated that the TA29/GUS gene transcriptional level dropped to below our detection limit (>100-fold) when the TA29 5' region was deleted from -279 to -150 nucleotides. Together, these organ system mRNAs. Anther (A),pistil (Pi), petal (P),leaf (L), stem (S),and root (R) mRNAs were fractionated by electrophoresis on denaturing agarose gels, transferred to nitrocellulose, and hybridized with labeled XTA29 phage DNA as outlined in Methods. Anther and pistil mRNAs were isolated from stage 6 flowers; petal mRNA was isolated from stage 12 flowers (Figure 1). Each lane contained 0.5 ^g of mRNA, but the anther mRNA gel blot (Ix) was exposed for only 1/100 of the time used to expose the other gel blots (100x). (C) Representation of the TSJT1 gene sequence in vegetative and floral organ system mRNAs. Gel blot conditions and mRNAs were the same as those used in (B). For these gel blots, a DNA fragment containing portions of the TSJT1 stem gene exon 2 and intron 2 [0.7-kb Hindlll fragment, (A)]was used as a probe. The reconstruction lanes contained 0.5 Mg (0.7%), 0.05 ^g (0.07%), and 0.005 ng (0.007%) of soybean midmaturation stage mRNA, and were hybridized with the soybean L9 lectin cDNA plasmid (Goldberg et al., 1983). Experiments published previously established that the 1-kblectin mRNA represents approximately 0.7% of the midmaturation stage embryo mRNA (Goldberg et al., 1983). All gel blot RNAlanes were exposed for 3 days. (A) (B) A+S (C) Anther TA29 TSJTI cco o •* *o o 1E 1 A A+S A+S S S ti S