Characterization of an AGAMOUS homologue from the conifer black spruce (Picea mariana) that produces floral homeotic conversions when expressed in Arabidopsis

Summary

Advances in elucidating the molecular processes controlling flower initiation and development have provided unique opportunities to investigate the developmental genetics of non-flowering plants. In addition to providing insights into the evolutionary aspects of seed plants, identification of genes regulating reproductive organ development in gymnosperms could help determine the level of homology with current models of flower induction and floral organ identity. Based upon this, we have searched for putative developmental regulators in conifers with amino acid sequence homology to MADS-box genes. PCR cloning using degenerate primers targeted to the MADS-box domain revealed the presence of over 27 MADS-box genes within black spruce (Picea mariana), including several with extensive homology to either AP1 or AGAMOUS, both known to regulate flower development in Arabidopsis. This indicates that like angiosperms, conifers contain a large and diverse MADS-box gene family that probably includes regulators of reproductive organ development. Confirmation of this was provided by the characterization of an AGAMOUS-like cDNA clone called SAG1, whose conservation of intron position and tissue-specific expression within reproductive organs indicate that it is a homologue of AGAMOUS. Functional homology with AGAMOUS was demonstrated by the ability of SAG1 to produce homeotic conversions of sepals to carpels and petals to stamens when ectopically expressed in transgenic Arabidopsis. This suggests that some of the genetic pathways controlling flower and cone development are homologous, and antedate the 300-million-year-old divergence of angiosperms and gymnosperms.

Introduction

Plant evolution has been punctuated by several dramatic changes in reproductive biology. Two of the most familiar examples are the angiosperm flower and gymnosperm strobili, which are morphologically distinct despite the fact they perform the same basic functions. Arabidopsis provides a good illustration of the generalized angiosperm flower, which is bisexual and consists of four whorls of organs. The outer two whorls make up the asexual perianth, with sepals in the first whorl and petals in the second. The sexual organs are contained within the two inner whorls, with stamens developing in the third whorl and carpels in the fourth. The name angiosperm (‘a vessel seed’) is based upon development of seed within an enclosed ovary, formed by the fusion of the carpels. Gymnosperm strobili, as illustrated by the cones produced by conifers, are monosexual and lack most of the structural features seen in flowers. Female cones produce ‘naked seed’, upon which the name gymnosperm is based, which develop from ovules borne on the adaxial surface of an ovuliferous scale. The ovuliferous scale in turn develops from a primordium within the axil of an asexual bract. This structural unit is repeated in a spiral phylotaxy to form the female cone. The male cone has a similar structure but lacks asexual bracts, being made up of multiple microsporophylls arranged in a spiral phylotaxy, with each producing two microsporangia on their abaxial surface.

The last common ancestor of angiosperms and gymnosperms (a progymnosperm) is believed to have existed 285–350 million years ago (Martin et al. 1993;Munster et al. 1997;Savard et al. 1994). This large evolutionary period, combined with extensive morphological differences, make direct comparisons of cones and flowers tenuous at best, despite apparent homologies in reproductive biology. This is exacerbated in part by the fact that the evolutionary history of the angiosperm flower continues to be unresolved (Crane et al. 1995), making it difficult to determine the precise evolutionary origins of floral organs. Thus, there is a great deal of uncertainty concerning both the evolutionary relationship and potential functional homology between the reproductive structures of cones and flowers.

New insights into the functional and evolutionary aspects of angiosperm reproduction have recently provided a molecular framework upon which the developmental biology of non-angiospermous species could be examined (Doyle 1994;Meyerowitz 1994;Purugganan et al. 1995;Theissen & Saedler 1995;Theissen et al. 1996). Much of this has been based upon the developmental genetics of flower initiation and floral organ identity which has led to the establishment of the ‘ABC’ model of flower development (Coen & Meyerowitz 1991;Meyerowitz et al. 1991;Schwarz-Sommer et al. 1990). A central aspect of this model is the ability to predictably modify flower structure via modification of the ABC gene expression patterns, either through mutagenesis or by introduction of a transgene (for review see Ma 1994;Weigel & Meyerowitz 1994;Yanofsky 1995).

As described by the ABC model, sexual organ formation is dependent on the C-function such that loss-of-function mutations lead to the production of sterile flowers in which stamens and carpels are replaced by petals and sepals, respectively. Prominent examples of C-function genes are AGAMOUS from Arabidopsis (Yanofsky et al. 1990) and PLENA from Antirrhinum majus (Bradley et al. 1993). Ectopic expression of the C-function, either through gain-of-function mutations (Bradley et al. 1993;Goodrich et al. 1997;Lönnig & Saedler 1994) or by constitutive expression of AGAMOUS in transgenic Arabidopsis plants (Mandel et al. 1992;Mizukami & Ma 1992;Mizukami & Ma 1997;Mizukami et al. 1996;Riechmann & Meyerowitz 1997) produces somewhat opposite homeotic conversions of sepals to carpels and petals to stamens. The ability to generate similar ectopic phenotypes in transgenic plants has also provided evidence for the existence of cognate homologues of AGAMOUS in several other diverse angiosperm species that include Brassica napus (Mandel et al. 1992), tobacco (Kempin et al. 1993), petunia (Tsuchimoto et al. 1993), tomato (Pnueli et al. 1994) and rice (Kang et al. 1995). Furthermore, modification of flower development within transgenic plants of heterologous species, the most disparate being rice OsMADS3 in tobacco (Kang et al. 1995), demonstrates functional conservation between C-function homologues from distantly related angiosperm species.

Investigations into the molecular basis of flower initiation and development also led to the discovery that many of the ABC-function genes belong to a super-gene family of transcriptional regulators, subsequently called the MADS-box gene family, that includes the yeast MCM1 gene and serum responsive factor from mammals (Schwarz-Sommer et al. 1990;Shore & Sharrocks 1995). This presented the likelihood that non-angiospermous plants contain MADS-box genes, a contention supported by the recent descriptions of several MADS-box genes within fern (Munster et al. 1997), and three from Norway spruce, including an AGAMOUS-like gene (Tandre et al. 1995). We have attempted to explore the developmental genetics of conifers by searching for putative developmental regulators in black spruce, based upon amino acid sequence homology to those of angiosperms. The work presented here confirms not only that black spruce has a large and diverse MADS-box gene family, but also that constitutive expression of a black spruce AGAMOUS-like protein in transgenic Arabidopsis produces the same homeotic conversions as that reported for ectopic expression of AGAMOUS. Taken together with parallels in gene structure and expression patterns, these results provide evidence that the evolutionary origin of the C-function pre-dates the evolution of angiosperms, and that a significant level of functional conservation may exist between developmental regulators of conifers and angiosperms.

Results

Analysis of the black spruce MADS-box gene family via PCR cloning

To examine the general composition of the black spruce MADS-box gene family, PCR amplification with degenerate primers was used to amplify a 61 bp segment within the center of the MADS-box domain. This allowed us to simultaneously amplify this region from multiple gene family members, which produced a fragment mixture that was then subcloned to allow isolation of individual fragments. DNA sequence analysis subsequently demonstrated that > 95% of all cloned fragments encoded an authentic MADS-box segment, as established by the presence of six obligate amino acids at conserved positions (Fig. 1). All unique clones were assigned a number (SMB#) based upon the assumption that each represented a black spruce MADS-box gene. This led to the identification of 27 distinct genes (Fig. 1), some of which appear to be representative of the major classes of angiosperm MADS-box genes, including several within the ‘C’ or AGAMOUS class and two within the ‘A’ or AP1/AGL9 class (Rounsley et al. 1995).

Figure 1. Deduced amino acid sequence comparison of PCR-generated black spruce MADS-box fragments (SMB) using genomic and cDNA templates.

Conceptual translation of the intervening segment bounded by the MADS-box primers are compared with representative examples of MADS-box proteins from Arabidopsis, including the three classes known to regulate flower organ development (AP1/AGL9 = ‘A’;Pi/AP3 = ‘B’;AG = ‘C’). Invariant amino acids are denoted by vertical highlights and dots replace amino acids identical to AGAMOUS, with the dash in SMB10 representing a single amino acid deletion. MCM1: yeast mini-chromosome-maintenance1; SRF: human serum responsive factor.

To determine whether these clones represent expressed genes, we analyzed RT–PCR generated fragments using RNA isolated from various tissues and from DNA prepared from cDNA libraries (Table 1). This indicated that all the identified MADS-box genes are transcribed and that most appear to be broadly expressed.

Table 1.  Summary of PCR amplified spruce MADS-box fragments (SMB) generated from various black spruce template DNAs

Name	gDNA^a	EMB RT^b	VEG RT^b	FEM RT^b	FEM Lib^c	MAL Lib^c	Total
a gDNA: genomic DNA. ^b RT: RT–PCR using sscDNA from embryogenic callus (EMB), vegetative buds (VEG) and female cones (FEM). c Lib: DNA isolated in bulk from cDNA libraries of female (FEM) and male (MAL) cones.
SMB1	2	12					14
SMB2	26	9	6	1	1		43
SMB3	10	24	1	1			36
SMB4	10	1					11
SMB6	16	14					30
SMB8	16	19		10	4		49
SMB9		5					5
SMB10	2	2					4
SMB11	14	90					104
SMB12	23	8					31
SMB13	14	8	3	2			27
SMB14	5	3	2	6			16
SMB15	3	27					30
SMB16	2	3	1	1			7
SMB18	1	3		2	2		8
SMB19	6	6	4	8	9		33
SMB20	5	5	1	5	17		33
SMB21	9	2					11
SMB22		3					3
SMB23		10					10
SMB25		5					5
SMB26		3	4	1			8
SMB27	1	1					2
SMB35	1	1					2
SMB41						50	50
SMB42						9	9
SMB43				1	1		2
Total	166	264	22	38	34	59	583

Characterization of a spruce AGAMOUS-like gene (SAG1)

To extend the characterization of spruce AGAMOUS-like genes, a female cone cDNA library was screened with two AGAMOUS-like PCR clones (SMB11 and SMB42). This yielded a total of 12 clones that represented multiple isolates of four nearly identical cDNA clones that corresponded to SMB42 (SAG1a-d; accession numbers U69482, U46582, U69483, U69484). These clones were found to have a combined total of 15 single base pair differences, in addition to a 2 bp deletion within the 5′ UTR of SAG1b and 1 bp deletion within the 3′ UTR of SAG1a. Only one of these sequence differences produces a modification in amino acid sequence (substitution of serine for glutamine) and probably reflect polymorphisms since the cDNA library was constructed from cones collected from several individual trees. Furthermore, only two of the four cDNA clones were found to have the same 3′ terminus, indicating that SAG1 transcripts probably have multiple polyadenylation sites.

The deduced amino acid sequence of SAG1 is very similar to AGAMOUS, differing by only a single conservative substitution within the MADS-box region. Homology with AGAMOUS was examined further by determination of the intron position within the SAG1 gene. Although precise alignment of the first intron with AGAMOUS is not conclusive due to its location within the 5′ UTR region, the positions of the remaining seven introns were found to be identical to that of AGAMOUS (Yanofsky et al. 1990).

SAG1 expression

Consistent with a putative role in reproductive organ development, RNA blot analysis revealed a cone-specific 1700 bp SAG1 transcript, with no detectable expression in embryogenic callus, roots, stems, mature needles or developing vegetative buds (Fig. 2a). A progressive decrease in SAG1 expression was observed during male cone maturation from spring flush to nearly undetectable levels prior to pollen release. In contrast, developing female cones maintained a high level of expression throughout this period, a pattern similar to that reported for DAL2, a homologue of SAG1 previously isolated from Norway spruce (Tandre et al. 1995). RT–PCR analysis confirmed this trend (Fig. 2b), except for detection of a very low level of SAG1 expression in both mature needles and late vegetative buds. Genomic DNA blot analysis further indicated that black spruce probably contains two SAG1 genes (Fig. 2c), although no supporting evidence for this has been obtained.

Figure 2. SAG1 expression and genomic blot analysis.

Cone development in black spruce is completed over two growing seasons, entering a dormancy period in the autumn before overwintering. The breaking of winter dormancy is followed by 4–6 weeks of rapid growth before wind pollination in late spring. SAG1 gene expression was examined by hybridizing a SAG1-specific probe to total RNA extracted from embryogenic callus (Em), roots (R), stems (S) and mature needles (N), in addition to developing vegetative (V), male (M), and female (F) buds, collected at three stages of development following winter dormancy (1–3), the first shortly after spring flush and the last just prior to pollen release. DNA extracted from a single genotype of embryogenic callus maintained in tissue culture was used in the genomic blot analysis.

(a) SAG1 RNA blot analysis showing changes in expression levels during postdormancy development.

(b) RT–PCR analysis of the identical RNA samples used in (a). A similar expression pattern is observed, except for a very low level of SAG1 expression in both mature needles and late vegetative buds (expanding branches).

(c) Genomic DNA blot analysis hybridized to the same SAG1-specific probe as in (a). Lanes left to right: BamHI, BstE2, EcoRI, HindIII and XbaI, respectively.

To examine further the tissue-specific expression pattern of SAG1, in situ hybridization analysis was conducted on developing buds following winter dormancy (Fig. 3). This revealed a low level of SAG1 expression in male cones shortly after breaking winter dormancy in the tissue that makes up the tapetal layer, with no signal detected in either the developing pollen or central vascular region (Fig. 3c–d). SAG1 expression was undetectable in subsequent stages of male cone development (data not shown). In female cones, SAG1 expression was found to be localized within the developing ovuliferous scale, with no detectable expression within the asexual subtending bract, central vascular region or apical meristem (Fig. 3i,j), a pattern that persisted up to pollination. Differences in hybridization intensity were also apparent within the developing ovuliferous scale, with reduced signal in the peripheral regions, intense signal within the integument, contrasted by low signal within the nucellus (Fig. 3l–p). Vegetative buds showed no hybridization signal above background at any stage from postdormancy to pollination (data not shown).

Figure 3. In situ hybridization with SAG1 to longitudinal sections of developing cones from black spruce.

A series of developmental stages following winter dormancy are presented, with several Toluidine Blue O (TBO) stained longitudinal sections included for reference (for a detailed description of spruce cone development see Fraser 1966;Harrison & Owens 1983;Owens & Molder 1977; see also Introduction). The in situ hybridization signal is blue. Scale bars: 5 mm in (a and g); 500 μm in (b, c, f, h–k); and 100 μm in (d, e, l–p). Abbreviations: b, bract; in, integument;n, nucellus; o, ovule; os, ovuliferous scale; ps, pollen sac; t, tapetal tissue.

(a) A nearly mature male cone, showing multiple microsporophylls arranged in a spiral phylotaxy. (b)TBO stained longitudinal section of an immature male cone shortly after spring flush. Pollen cones emerging from winter dormancy rapidly undergo pollen mother cell meiosis, followed by callose wall degeneration, microspore release and pollen maturation. Each microsporophyll produces two microsporangia on their abaxial surface, which at the stage shown here contain large numbers of microspore mother cells. (c) In situ hybridization of a male cone in the mid-stages of pollen development which shows a low level of SAG1 expression that is restricted to the tapetal tissue. Dark spots within the vascular central region are due to lignification and do not contain any hybridization signal. (d) Close-up of two microsporophylls from (c) showing hybridization signal within transversing segments of tapetal tissue. (e) Close-up of a microsporophyll from (c) in which the plane of the section has transected the tapetal tissue separating two microsporangia. The brown-colored areas at the tip of the microsphoropyll are due to lignification and do not contain any hybridization signal. (f) TBO stained longitudinal section of a male cone shortly before pollen release; SAG1 expression is undetectable at this stage of development (data not shown). (g) A female cone opened for pollination, showing multiple ovuliferous scales that in black spruce later make up the protective outer cone scales. (h) TBO stained longitudinal section of an immature female cone shortly after spring flush. Multiple ovuliferous scales, each subtended by an asexual bract, are sequencially formed in a spiral phyllotaxy by the apical meristem before entering winter dormancy, although some cones continue to produce a few additional ovuliferous scales following spring flush. (i) In situ hybridization of a young female cone showing SAG1 expression localized to the ovuliferous scales, with lower hybridization signals within the regions that correspond to developing ovule primordia. (j) Higher magnification of the in situ hybridization pattern within the section adjacent to (i), showing absence of SAG1 expression in the cone apical meristem. Dark spots within the vascular central region are due to lignification and do not contain any hybridization signal. (k) TBO stained longitudinal section of a female cone shortly before pollination; the ovuliferous scales have opened in preparation for pollination. Two ovule primordia, composed of an outer integument and an inner nucellus, develop on the adaxial surface of each ovuliferous scale, with each ovule containing a large megaspore mother cell in the free nuclear stage that continues until pollination. (l–p) In situ hybridization of a sequential series of cross-sections through an ovuliferous scale and the subtending bract, highlighting a developing ovule before pollination. Ovules contain the nucellus surrounded by the integument, which shows an intense hybridization signal in these sections.

Ectopic expression of SAG1 in Arabidopsis

Several groups of researchers have previously reported that phenotypes similar to that produced by ectopic expression of AGAMOUS can be generated in transgenic plants constitutively expressing cognate homologues of AGAMOUS (see Introduction). The most prominent of these phenotypes are the homeotic conversion of petals to stamens, sepals to carpels, and loss of inflorescence indeterminacy. To determine if similar phenotypes could be produced by SAG1, transgenic Arabidopsis were generated containing the coding region of SAG1 driven by the 35S cauliflower mosaic virus promoter.

Of the 27 transformed lines generated, 10 produced altered phenotypes, four of which produced phenotypes nearly identical to that reported for transgenic Arabidopsis ectopically expressing AGAMOUS (Mizukami & Ma 1992;Mizukami & Ma 1997;Mizukami et al. 1996;Riechmann & Meyerowitz 1997). Phenotypes were followed for at least two generations and increased in severity in homozygous plants. These included failure of the sepals to enclose young flower buds (Fig. 4c), development of staminoid-petals (Fig. 4d,f–i and k), carpelloid-sepals (Fig. 4m,n), and loss of indeterminacy such that the inflorescence terminated with a carpelloid structure(s) or a terminal cluster of flowers (Fig. 4n). The severity of the flower phenotypes also increased acropetally, with the most extensive homeotic conversions occurring in late developing flowers.

Figure 4. Transgenic Arabidopsis ectopically expressing SAG1.

Phenotypes are presented in generally increasing severity; RNA blot analysis further demonstrated that the level of transgene expression was directly related to the level of phenotypic severity (data not shown). Scanning electron micrographs are shown in (e–j) and (k), size bar = 100 μm.

(a) Wild-type Arabidopsis flower. (b) Curled cauline leaf, entrapping the lateral inflorescence. (c) Immature early flowers in which the sepals fail to enclose the inner organs. (d) Flower with thin petals. (e) Wild-type petal. (f) Narrow petal of an early flower. (g) Staminoid-petal of a late flower. The inset shows a higher magnification of the boxed area showing the transition between the staminoid and petaloid cell types (arrow). (h) Staminoid-petal. (i) Staminoid-petal which has produced pollen. (j) Wild-type stamen. (k) Late flower with a petal containing a staminoid sector (arrow). (l) A flower lacking petals. (m) Carpelloid sepal with stigmatic papillae at the tip. (n) Premature termination of the inflorescence by a cluster of flowers with multiple carpels (arrows). All second and third whorl organs appear as staminoid organs.

The degree of staminoid-petal formation varied between and within flowers, ranging from short, narrow petals (Fig. 4d,f), to petals that developed discrete sectors of staminoid tissue, some of which produced pollen (Fig. 4g–i, k). The most severely affected lines also produced flowers lacking second whorl organs (Fig. 4l). In flowers that developed late in the inflorescence, carpelloid-sepals developed with stigmatic papillae at their tips (Fig. 4m) and occasionally with ovules on their margins (not shown). Premature loss of inflorescence indeterminacy was characterized by the production of a terminal flower with a reduced or absent pedicel, carpelloid-sepals and reduced organ number, or a cluster of terminal flowers with multiple carpelloid organs (Fig. 4n). Additional phenotypes included extensive curling of both the rosette and cauline leaves (Fig. 4b), reduced plant height, production of fewer flowers, bumpy siliques, and a reduced seed set as compared with control plants. Third and fourth whorls appeared to develop normally, except in the most severely affected flowers in which the number of stamens was reduced.

Discussion

Identification of putative developmental regulators via PCR cloning

PCR cloning has proven to be an effective method for the initial identification of either diverse members within a gene family or gene homologues from distantly related species (Feng et al. 1992;Gould et al. 1989;Kamb et al. 1989;Mieszczak et al. 1992). This has also included the identification of several diverse members of the Arabidopsis MADS-box gene family (Rounsley et al. 1995). The region selected for PCR amplification in this study encodes a 20 amino acid segment within the center of the MADS-box domain that, despite its limited size, provided useful insights into gene family composition and general expression patterns within black spruce. Although such an analysis cannot provide a comprehensive survey, it did reveal the presence of several black spruce MADS-box genes that may be involved in cone development, as based upon similarity to AGAMOUS and AP1, in addition to many others for which no comparable angiosperm MADS-box genes have been identified. Our subsequent characterization of SAG1 not only confirms the utility of this approach, but also suggests that it could be successful in the initial characterization of other gene families encoding developmental regulators from distantly related plant species.

SAG1 amino acid sequence, gene structure and overall expression pattern parallel that of AGAMOUS

SAG1 was found to share extensive sequence similarity with AGAMOUS, differing by only a single conservative substitution within the MADS-box, with 59% amino acid identity within the K-box domain, and co-linearity from the MADS-box up to and including the first 16 amino acids of the C-terminal region. As found for rice OsMADS3 and Norway spruce DAL2, SAG1 lacks the N-terminal extension present in other AGAMOUS homologues. However, deletion of this segment from AGAMOUS does not affect its function in transgenic Arabidopsis (Mizukami et al. 1996), suggesting that this N-terminal extension is not obligatory for C-function activity.

Conservation of intron position between SAG1, AGAMOUS and PLENA further supports their structural homology, and indicates that gene topography may be a distinctive feature of C-function genes. Several closely related AGAMOUS-like genes such as AGL1 and AGL5 (Heck et al. 1995), ZAG2 and ZMM1 (Theissen et al. 1995), and CUS1 (Filipecki et al. 1997) also have gene structures similar to AGAMOUS, but all lack at least the last intron. In view of additional differences in expression pattern, it is likely that many of these genes represent divergent members of an AGAMOUS-like subfamily that provide functions collateral to that of the C-function. Although the true significance of gene topology must await the availability of additional C-function gene sequences, the expression of SAG1 within both ovulate and sporangiate cones more closely parallels that of AGAMOUS than any of the other AGAMOUS-like genes. An apparent contradiction to this is the low expression of SAG1 in needles that was only detectable using RT–PCR. Although expression of AGAMOUS in wild-type Arabidopsis leaves has yet to be reported, it has been demonstrated recently that inactivation of CURLY LEAF leads to a progressive activation of AGAMOUS expression during leaf maturation (Goodrich et al. 1997). Thus, it appears that repression of AGAMOUS is required for normal leaf development in Arabidopsis. The low level of SAG1 expression in black spruce needles may therefore reflect a similar propensity for C-function expression during the late stages of needle maturation in conifers.

Although the extensive differences in the reproductive organs of conifers and angiosperms clearly make it difficult to directly compare gene expression patterns, in situ hybridization analysis did reveal that SAG1 expression has parallels with that of AGAMOUS. In postdormancy male cones, for example, SAG1 expression was only observed within tapetal tissue, with no expression detected in either vascular tissue or maturing pollen. This is similar to that observed for AGAMOUS expression during late stamen development, which is concentrated within the connective tissue separating the locules of the anther (Bowman et al. 1991). SAG1 expression was also found to be restricted to ovuliferous scale primordia, producing a nearly uniform signal that continued through to the free nuclear stage of the megagametophyte, just before pollination. Within carpel primordia of Arabidopsis, AGAMOUS expression is also uniform, a pattern that persists until the formation of ovule primordia, after which AGAMOUS expression becomes restricted to specific cell types. This later pattern includes high level expression throughout the ovule until stage 13 of flower development, at which time its expression becomes restricted to the endothelium of the inner integument (Bowman et al. 1991;Modrusan et al. 1994). High level expression of SAG1 in the integument surrounding the developing ovule, contrasted by low levels in the nucellus, may reflect a similar situation during ovule development in conifers.

Does SAG1 provide a conifer equivalent to the angiosperm C-function?

Notwithstanding the parallels in structure and expression patterns with AGAMOUS, only examination of the in vivo function can provide direct supporting evidence that SAG1 provides a C-function equivalence in spruce. The ectopic expression of SAG1 in transgenic Arabidopsis did indeed induce homeotic conversions of sepals to carpels and petals to stamens, as has been reported for several AGAMOUS homologues from angiosperms (see Introduction for references). This is in addition to other shared phenotypes that included severe curling of leaves, dramatic reduction in plant size, increased severity acropetally, and premature termination of the inflorescence. Although ectopic expression cannot provide a definitive evaluation of functional homology, these results do provide a striking demonstration that MADS-box proteins from vastly divergent species can maintain some level of functional activity when placed into the heterologous species. A more accurate assessment of functional homology will, however, require testing the ability of SAG1 to genetically complement null-mutants of AGAMOUS, or to act as a dominant negative inhibitor of AGAMOUS when lacking the C-terminal region (Mizukami et al. 1996).

Ultimately, direct examination of SAG1 function within spruce will be necessary for establishing the existence of a conifer C-function, and to define its role in cone development. The potential to produce both gain- and loss-of-function analysis for SAG1 in transgenic spruce has great promise for providing insights into the genetics processes controlling cone initiation and development, currently made inaccessible due to a lack of developmental mutants. As an initial step to provide such an analysis, our group has generated transgenic black spruce ectopically expressing SAG1. Although all lines show normal phenotypes as young seedlings (data not shown), years of accelerated growth combined with cone induction treatments may be required before a phenotypic assessment can be completed, in that black spruce normally requires 8–10 growing seasons before producing cones.

Another important aspect for contemplation of a C-function equivalence in conifers is consideration of reproductive organ homology. Similarity in the expression patterns of SAG1 within the ovuliferous scale with that of AGAMOUS within the carpel is consistent with organ homology, as supported by their common role in bearing ovules and seeds. What is less clear is the extent to which other organ homologies can be discerned. One possibility, supported by the absence of SAG1 expression in the apical meristem of the female cone, is the likelihood that a cone is representative not of a flower, but of a modified inflorescence (Harder et al. 1965). This view further proposes that the ovuliferous scale corresponds to a reduced flower, arising as a shoot in the axil of the bract. Noteworthy in this regard are the bisporangiate cones that are occasionally produced by conifers (Chamberlain 1935). Such cones generally consist of ovulate sporophylls in the upper half of the cone, with staminate sporophylls in the lower half (Caron & Powell 1990). Of particular interest is the description of bisexual sporophylls within the bisporangiate cones of Pinus maritima. These sporophylls, which are located in the transition zone between the male and female sporophylls, were found to have two microsporangia on their abaxial side and a rudimentary ovuliferous scale in their axil (Chamberlain 1935). Whether such abnormalities are reflective of a simple bisexual flower lacking a perianth is difficult to determine, but they do suggest some level of plasticity in cone development, reminiscent of that demonstrated for angiosperm flowers.

Although characterization of SAG1 has provided an important link between the reproductive biology of conifers and angiosperms, many prominent questions remain as to the extent to which the principles of the ABC model are applicable to cone development. One of the most compelling is the potential requirement of a B-function in male cone development, based upon its obligatory role in stamen development. Even though our PCR-cloning analysis did not reveal any spruce MADS-box genes with strong homology to either AP3 or Pi, the likelihood that their role in petal development co-evolved with that of the flower suggests that at least some aspects of the B-function are unique to flower development. Clearly, characterization of additional conifer MADS-box genes involved in cone development will be necessary before additional parallels with the ABC model can be more clearly defined.

Future implications

The prospect that a significant level of C-function conservation exists between conifers and angiosperms has several important implications for both basic and applied research. In particular is the potential to further exploit transgenic angiosperms for the characterization of other putative developmental regulators from conifers. It also presents the possibility to manipulate the reproductive biology of conifers via transgenic technologies modeled upon the developmental genetics of angiosperms, including such aspects as the induction of early flowering, and the production of gain- and null-mutants. The potential for engineering of reproductive sterility is another application of great interest, in that it would provide both genetic containment and potential increases in growth rate for transgenic trees (Strauss et al. 1995). The likelihood that developmental regulators from angiosperms will have some level of functionality in transgenic conifers also suggests that their utilization in the genetic engineering of conifers will enjoy a level of success previously unanticipated.

Experimental procedures

Bud collection and nucleic preparation

The collection of female, male and vegetative buds was undertaken in the spring from reproductively mature Picea mariana trees within the research forest maintained at the Petawawa National Forestry Institute, Chalk River, Ontario, Canada. RNA was prepared from buds collected on May 1, May 16 and May 26, in addition to roots, stems and needles collected from greenhouse-grown seedlings. Genomic DNA and embryonal RNA was extracted from a single line of embryogenic callus (R4F14) maintained under standard tissue culture conditions (Cheliak & Klimaszewska 1991;Gupta et al. 1993).

PCR cloning

Degenerate primers MBP1 (5′ CGGAATTCATI(G/C)A(A/G)ATIAA(A/G)(C/A)GIATIGAIAA 3′) and MBP2 (5′ GCTCTAGAGCGTCGCAIA(A/G)IACI(G/C)(T/A)IA(A/G)(T/C)TC 3′) were based upon highly conserved motifs within the MADS-box regions of AGAMOUS (Yanofsky et al. 1990) and DEFA (Sommer et al. 1990), where I = inosine and restriction sites are underlined. Electrophoretic analysis showed amplification of a single major band of the expected size (121 bp) for all amplifications (data not shown) which were restricted with EcoRI and XbaI, cloned into M13-mp18 and -mp19, and individual plaques collected for DNA sequence analysis. Sequence accuracy was ensured by analyzing multiple PCR amplifications for each template DNA, combined with the redundancy generated from sequencing multiple clones. Template DNA included genomic DNA, single-stranded cDNA prepared with the SuperScript Preamplification System (Gibco BRL), and lambda phage DNA isolated in bulk from three cDNA libraries (embryogenic callus, male cones and female cones).

cDNA cloning and gene analysis cDNA libraries (all > 10⁶ clones) were constructed using the SuperScript Lambda System (Gibco BRL) with RNA extracted from male and female cones in mid-stages of development following winter dormancy, and from embryogenic callus (R4F14). SAG1 intron position was determined by DNA sequence analysis of PCR fragments amplified from genomic DNA that encompassed the region corresponding to the largest cDNA clone and were in complete agreement DNA sequence analysis of a genomic clone containing all of the transcribed region 3′ to the second intron. Amino acid sequence alignments were constructed using Macaw (Ver. 2.0.5; NCBI).

Expression analysis

RNA and genomic DNA blots were hybridized as described by Church & Gilbert (1984), using a SAG1 specific probe that consisted of a 301 bp segment corresponding to a region from the K-box to the stop codon. RT–PCR (35 cycles: 57°C, 30′-72°C, 40′-94°C, 40′) was conducted on single-stranded cDNA (Superscript II, Gibco BRL) made from equal amounts of DNase I-treated RNA samples using SAG1-specific primers (5′ TGATGGGTGACGGGCTTACA 3′ and 5′ GCTCTAGAGCAAGCTGAAGCGTTGTTTGCT 3′) that produced a single 342 bp fragment. Bud fixation, embedding and in situ hybridization were carried out essentially as described by Leitch et al. (1994) using digoxygenin-labeled RNA probes made from a 320-nucleotide corresponding to the carboxyl portion of the SAG1 coding region; sense probes produced no signal above background. Some of the sections were stained with 0.05% Toluidine Blue O in 1% boric acid for examination of bud morphology.

Generation and analysis of Arabidopsis transformants

An NcoI-XbaI fragment containing the SAG1 coding region was generated using PCR amplification and used to replace the GUS coding region within SLJ4D4 that contains a single 35S promoter, the omega translation enhancer and OCS terminator (Jones et al. 1992). An EcoRI-HindIII fragment containing the expression cassette was inserted into pBIN19 (Clontech). Agrobacterium tumefaciens GV3850 carrying this binary vector was used to transform Arabidopsis thaliana (Columbia) plants by vacuum infiltration (Bechtold et al. 1993) and putative transformants selected on germination media containing kanamycin. Plants were grown at 22°C, 16 h light/8 h darkness, and phenotypes followed for at least two generations. The presence of the 35S-SAG1 transgene was followed by the segregation of kanamycin resistance, and confirmed in 10 lines by DNA and RNA blot analysis. Flower phenotypes were examined with a Zeiss Stemi SV8 photomicroscope and by scanning electron microscopy as described by Modrusan et al. (1994) using a Zeiss 940 A-DSN scanning electron microscope.

Image processing

Kodak CD images produced from photographic slides or captured by a Sony 3CCD video camera were processed using Adobe PhotoShop 4.0 (Adobe Systems Inc., Mountain View, CA, USA) and compiled using CorelDraw 8.0 (Corel Corp., Ottawa, Ontario, Canada).

Acknowledgements

We thank Drs Dave Ellis and Leslie Sieburth for critical reading of the manuscript; Drs Tannis Beardmore, Pierre Charest, Mike Frolich, Krystyna Klimeszewska, Beth Krisek and Hong Ma for helpful comments; Dr J.D.G. Jones for the generous gift of the SLJ4D4 vector; and Pia Kauri for technical assistance with the transgenic plants. P.F., O.N. and S.R were supported by a grant from the National Biotechnology Strategy of Canada to R.R.

GenBank accession numbers U69482, U46582, U69483 and U69484.