SPATULA and ALCATRAZ, are partially redundant, functionally diverging bHLH genes required for Arabidopsis gynoecium and fruit development


(fax +61 3 9905 5613; e-mail


The Arabidopsis gynoecium is a complex organ that facilitates fertilization, later developing into a dehiscent silique that protects seeds until their dispersal. Identifying genes important for development is often hampered by functional redundancy. We report unequal redundancy between two closely related genes, SPATULA (SPT) and ALCATRAZ (ALC), revealing previously unknown developmental roles for each. SPT is known to support septum, style and stigma development in the flower, whereas ALC is involved in dehiscence zone development in the fruit. ALC diverged from a SPT-like ancestor following gene duplication coinciding with the At-β polyploidy event. Here we show that ALC is also involved in early gynoecium development, and SPT in later valve margin generation in the silique. Evidence includes the increased severity of early gynoecium disruption, and of later valve margin defects, in spt-alc double mutants. In addition, a repressive version of SPT (35S:SPT-SRDX) disrupts both structures. Consistent with redundancy, ALC and SPT expression patterns overlap in these tissues, and the ALC promoter carries two atypical E-box elements identical to one in SPT required for valve margin expression. Further, SPT can heterodimerize with ALC, and 35S:SPT can fully complement dehiscence defects in alc mutants, although 35S:ALC can only partly complement spt gynoecium disruptions, perhaps associated with its sequence simplification. Interactions with FRUITFULL and SHATTERPROOF genes differ somewhat between SPT and ALC, reflecting their different specializations. These two genes are apparently undergoing subfunctionalization, with SPT essential for earlier carpel margin tissues, and ALC specializing in later dehiscence zone development.


The female reproductive unit of flowering plants, the carpel, evolved from a highly modified leaf to facilitate the transmission of the male gametes while protecting the recipient ovules within (Crawford and Yanofsky, 2008; Girin et al., 2009). There are two carpels in the Arabidopsis gynoecium, fused along their edges where ovules and the transmitting tract arise. Upon fertilization, the gynoecium develops into a silique, or fruit, housing the developing seeds until maturity and dispersal. Over the past 2 decades, Arabidopsis gynoecium and fruit development have been extensively studied, with many transcription factor and auxin genes now assigned developmental roles (reviewed in Crawford and Yanofsky, 2008; Girin et al., 2009; Østergaard, 2009; Roeder and Yanofsky, 2006). These include early acting organ identity genes, and genes that define polarity in the apical-basal, lateral-medial, and outer-inner orientations.

Once polarity is established, a second suite of genes coordinates development of the newly differentiating tissues arising from the medial regions where the two carpels are joined. Internally, medial ridge tissues produce two outgrowths that unite at the centre of the gynoecium to produce the false septum that includes the transmitting tract (Crawford and Yanofsky, 2008). Formation of the septum and transmitting tract requires the activity of the transcription factor genes SPATULA (SPT) (Alvarez and Smyth, 1999, 2002), NO TRANSMITTING TRACT (NTT) (Crawford et al., 2007), HECATE (HEC1, HEC2 and HEC3) (Gremski et al., 2007), SHATTERPROOF (SHP1 and SHP2) (Colombo et al., 2010) and HALF FILLED (Crawford and Yanofsky, 2011). All of these (except NTT) are also responsible to varying extents for development of the apical style and stigma that commences shortly after septum initiation. This process is dependent upon high auxin levels in apical regions (Nemhauser et al., 2000). At maturity, the ovary has a clearly defined replum, valves, and valve margins (VM), capped by the style and stigma.

Following fertilization, the fruit extends as the seeds develop within, and the VM differentiates into the dehiscence zone (DZ) during the final stages of silique development. Within the DZ, a lignified layer develops adjacent to the valve, and a separation layer next to the replum. Genes involved in VM differentiation are prevented from acting in the valves by the FRUITFULL (FUL) gene (Ferrándiz et al., 2000), and from the replum by REPLUMLESS (RPL) (Roeder et al., 2003). The SHP1 and SHP2 genes are required for VM differentiation itself (Liljegren et al., 2000). They act upstream and in parallel with INDEHISCENT (IND) which is also required later for both the lignified and separation layers of the DZ (Liljegren et al., 2004), and are upstream of ALCATRAZ (ALC), a key regulator of the separation layer (Rajani and Sundaresan, 2001).

The SPT gene was initially identified as a regulator of growth of the septum, style and stigma during gynoecium development as loss-of-function defects were found in these carpel margin tissues (Alvarez and Smyth, 1999, 2002). SPT is not only expressed within the developing gynoecium, but also in many other locations (Heisler et al., 2001; Groszmann et al., 2010). These include the VM and the later DZ of developing siliques, although no disruptions have been reported in these tissues. Identification of additional developmental roles for SPT in these other regions may be hindered by functional redundancy of related genes, a recurrent theme in developmental biology.

The Brassicaceae lineage has experienced several recent whole genome duplication (WGD) events, extensive gene shuffling, and chromosomal rearrangements, associated with the large-scale duplication of genes (reviewed in Franzke et al., 2011). Many of these duplicates have been lost through non-functionalization (Walsh, 1995; Lynch and Conery, 2003), but sometimes the duplicated genes are retained through neofunctionalization (acquisition of a new function), subfunctionalization (subdivision of ancestral gene function), or they simply continue to act in a fully redundant manner (Wagner, 1998, 1999; Walsh 2003). Genes progressing through neofunctionalization or subfunctionalization may exhibit partial or unequal redundancy associated with the ancestral role (reviewed by Briggs et al., 2006).

Previous work using a repressive version of SPT (35S:SPT-SRDX) identified enhanced disruptions to septum and gynoecium fusion beyond those seen in strong spt single mutants (Groszmann et al., 2008). SRDX fusions can reveal shared loss-of-function phenotypes even when fused to only one member of redundantly acting gene families (Hiratsu et al., 2003). Here we report the identification of ALC, the nearest relative of SPT, as a partially redundant partner in gynoecium development. ALC was previously known only to be required later for fruit DZ development (Rajani and Sundaresan, 2001). The results also show that SPT has a previously hidden later role in VM and DZ development that is redundant with ALC. Phylogenetic analysis has revealed that ALC originates from a recent duplication of a SPT-like ancestor indicating that their reciprocated partial redundancy is a consequence of subfunctionalization. ALC may represent a larger group of genes now specialized for VM/DZ development in the fruit that have origins associated with earlier carpel margin development.


ALCATRAZ, the closest relative to SPATULA, originates from a relatively recent gene duplication event

Phylogenetic analysis of all bHLH proteins in Arabidopsis places ALC as the closest relative to SPT (Bailey et al., 2003; Heim et al., 2003; Toledo-Ortiz et al., 2003), suggesting that they have arisen through a recent gene duplication event. A recognisable duplicated region that spans 29 genes on chromosome 4 and 32 genes on chromosome 5, encompasses each of SPT and ALC (Figure 1a). In this short block, only seven genes are now shared/duplicated between the two chromosomes and in each case the duplicated genes are the closest relatives to each other in Arabidopsis. A more extensive duplicated block exists between the same region surrounding ALC on chromosome 5 and a partner region on chromosome 3 consisting of approximately 160 genes with 53 retained duplicates, but ALC has been lost (Blanc et al., 2003). This chromosome 3–5 block is younger than the SPT-ALC block having emerged from the most recent (At-α) WGD event approximately 40 million years ago (MYA) near the emergence of the Brassicaceae (reviewed in Franzke et al., 2011). Therefore the SPT-ALC duplication is older than 40 MYA and pre-dates the emergence of the Brassicaceae and the diversification of a wide variety of species that develop dehiscent seed pods (Spence et al., 1996). An extensive database search across 20 species identified numerous SPT relatives (i.e. sequences more similar to AtSPT than to AtALC) across a variety of angiosperms, but failed to identify an ALC relative until at least after the Brassicaceae-Caricaceae split approximately 70 MYA (Figures 1b, S1, S2 and S3; Groszmann et al., 2008). This places the SPT-ALC duplication event between approximately 40–70 MYA, a period coinciding with the At-β WGD event (Figure 1b; Franzke et al., 2011). Also, it reveals that whereas the ALC sequences have diverged following the duplication, the SPT sequences have been relatively conserved and retain close similarities across the dicotyledons (Figure S1; Groszmann et al., 2008), as expected if their function has been maintained.

Figure 1.

ALCATRAZ is a close relative of SPATULA.
(a) Duplicated regions of chromosomes 4 and 5 containing ALC and SPT.
(b) Estimated duplication date of SPT and ALC coincides with the At-β WGD event.
(c) Protein sequence alignment between AtSPT, AtALC, and three ALC orthologues; AlSPT, Arabidopsis lyrata Genbank EFH41286; BnSPT, Brassica napus GenBank EV029403 (BnaC.ALC.a); and RsALC, Raphanus sativus GenBank EX904678. Identical residues: black; similar residues: grey; solid triangles: introns in both SPT and ALC; open triangles: SPT introns only.

ALC and SPT have 43 out of 49 identical amino acids within the bHLH region (Figure 1c). Of the six differing amino acids, two are conservative substitutions (Lys-193 and Lys-197 in SPT are Arg-111 and Arg-115 in ALC), and one is a Ser-207 (SPT) to Lys-125 (ALC) difference (Figure 1c). The remaining three differing residues are within the DNA binding basic domain.

ALC is the only bHLH in Arabidopsis that shares significant similarity with SPT outside the bHLH domain. SPT has three additional functional domains (Groszmann et al., 2008), and ALC shares the amphipathic helix near its N-terminus and a beta strand immediately C-terminal of the bHLH, but not the acidic domain (Figures 1c and S1). The loss of the acidic domain appears due to sequence erosion as opposed to a deletion given that some limited similarity remains including conservation of an intron/exon boundary within it (Figure 1c). ALC is a shorter protein than SPT mostly due to seven small deletions and a truncated C-terminus associated with a stop codon which in AtALC, AlALC, and BnALC is directly adjacent to the start of the fifth SPT intron (Figure 1c).

ALCATRAZ shows partial redundancy with SPATULA in early gynoecium development

The enhanced defects of 35S:SPT-SRDX transgenics over the spt single mutant (Groszmann et al., 2008) indicates the existence of a functionally redundant partner(s). Given its close homology to SPT, ALC is a likely candidate. To test this, the strong spt-2 mutant (see Alvarez and Smyth, 2002) was combined with alc-1, which has no reported gynoecium defects (Rajani and Sundaresan, 2001). A significant strengthening of carpel fusion and stylar defects was observed (Figure 2), confirming redundancy.

Figure 2.

 Comparison between stage 17 siliques from alc-1, spt-2, and spt-2 alc-1 double mutant.
alc-1 has no defects related to gynoecium fusion, but the spt-2 alc-1 double mutant gynoecia are more unfused than the spt-2 single mutant.The spt-2 and spt-2 alc-1 siliques shown are the second formed fruits in each case and reveal the more severe disruptions seen in early formed fruits. Scale bar = 1 mm.

To define when spt-2 alc-1 double mutants first differ from spt-2, we examined developing flowers by scanning electron microscopy (SEM; Figure 3). During stage 7 (floral stages defined by Smyth et al., 1990), vertical growth of the extending gynoecial tube is retarded in medial regions in spt-2 mutants compared with the wild type (WT) (Alvarez and Smyth, 2002; Figure 3a,d), and this is also apparent in the spt-2 alc-1 double mutant (Figure 3g). A difference between spt-2 alc-1 and spt-2 first becomes apparent during stages 8–9 when growth in this medial apical region is more severely retarded in the double mutant (Figure 3h), compared with the spt-2 single (Figure 3e). This difference is more marked through stages 9–10 (Figure 3i compared with Figure 3f).

Figure 3.

 Early development of the spt-alc double mutant gynoecium.
(a–c) Wild type development through stages 7–10. Arrows indicate the raised rim of the extending gynoecial tube in medial regions.
(d–f) spt-2 development, showing retarded growth in this region (arrows) compared with the wild type. This slowing of growth commences at mid stage 7 (see also Alvarez and Smyth, 2002).
(g–i) Gynoecium development in the spt-2 alc-1 double mutant, revealing more pronounced retardation of medial growth (arrows) than in the spt-2 single mutant. This is first apparent during stages 8–9 (h), and is further enhanced through stages 9–10 (i). All images are from early produced flowers. Scale bars = 25 μm.

By stage 11 the WT gynoecium is normally fully closed and capped with clearly developing stigmatic papillae (Figure 4a,d). At the same stage the spt single mutant has not closed over and differentiation of stigmatic papillae is only occasionally observed (Figure 4b,e), at least in early flowers (Figure S4a). The apex of spt-alc gynoecium was never closed and the further retardation of style development resulted in an almost complete loss of the stigmatic tissue (Figure 4c,f), although occasional tufts did develop on the tips of the malformed style in later developing flowers (Figure S4a). By stage 13, the mature WT and spt gynoecium have clearly defined repla and valves, and WT gynoecia have fully developed stigmatic papillae (Figure 4g,j). At this stage the spt gynoecium failed to fuse through the style and apical region of the ovary, and had poorly developed stigmatic papillae, which were shorter and less abundant than in WT (Figure 4h,k). The stage 13 spt-alc gynoecium was radically more defective and the further lack of apical fusion and inward involution of the ovary along the lateral plane resulted in internal tissues such as placentae and ovules, being observed externally (Figure 4i,l). The first several apical ovules often failed to develop correctly, instead becoming horn-like projections. The cell morphology at the apex revealed that while a style is present, the demarcation between the valves and style was not clearly distinct (Figure 4i,l). The VM, particularly near the style and gynophore were not defined, lacking the characteristic constricted growth of the VM in WT (Figures 4i,l and S4b), spt and alc (Rajani and Sundaresan, 2001; Alvarez and Smyth, 2002). The replum was difficult to distinguish in the unfused region (Figure 4i,l).

Figure 4.

 Later development of the spt-alc double mutant gynoecium.
(a–c) Medial view of stage 11 gynoecia.
(d–f) Top view of stage 11 gynoecia.
(g–i) Medial view of mature stage 13 gynoecia.
(j–l) Colourised scanning electron microscopy (SEM) images from g-i identifying tissue types: purple: stigma; red: style; green: values; orange: replum; yellow: placentae; light blue: ovules. Medial clefts at the apex continue to be more pronounced in spt-2 alc-1 double mutant (c, f, i, l) compared with the spt-2 single mutant (b, e, h, k). The wild type controls show complete fusion in these regions (a, d, g, j). All images are of early produced flowers. Scale bars = 50 μm (a–f); 100 μm (g–l).

Finally, early formed flowers lacked stigma (Figure S4a) and transmitting tract (Figure S4c) and were female sterile. However, as with spt, the spt-alc double mutant showed an acropetal decrease in severity such that a few seeds were produced by very late arising flowers.

SPT has a redundant function in valve margin development

To investigate whether SPT has a redundant role covered by the action of ALC later in fruit development, a 35S:SPT-SRDX transgenic line (Groszmann et al., 2008) was examined, focusing on VM and DZ development. Because 35S:SPT-SRDX plants are sterile, we compared stages 16–17 siliques that exhibited severely affected apical regions with siliques of unfertilized WT controls of equivalent age (Figure 5a,d). WT siliques showed normal VM development, with small cells producing a well defined indentation between the valves and replum (Figure 5b,c). By contrast, this boundary in 35S:SPT-SRDX siliques was poorly defined (Figure 5e,f), and similar to that in spt-alc double mutants (Figure S4b), and the indehiscent siliques of ind and shp1 shp2 mutants (Liljegren et al., 2000, 2004; Arnaud et al., 2010). Overall, these results are consistent with SPT playing a redundant role with ALC in VM development.

Figure 5.

SPT-SRDX siliques show reduced valve margin development.
(a) Apical region of an unfertilized mature WT silique.
(b, c) Constricted growth and clearly defined VM (arrows) characteristic of a WT silique.
(d) Sterile 35S:SPT-SRDX silique showing severe disruptions to apical tissues.
(e, f) Poorly defined VM development in 35S:SPT-SRDX silique (arrows). Scale bar = 100 μm.

ALC is widely expressed and has considerable overlap with SPT

If SPT and ALC function redundantly and cell autonomously in gynoecium development, ALC expression must presumably overlap with SPT. To date, expression of ALC has been reported in detail only later for the VM (Rajani and Sundaresan, 2001). We examined the spatial and temporal expression domains of ALC using plants heterozygous for the alc-1 GUS gene trap line (alc-1/+ are WT in appearance). The results were consistent with publicly available expression array data (Schmid et al., 2005; Figure S5).

ALC gynoecium expression was first detected during stage 8, appearing diffusely within the inner (adaxial) margin of the developing medial ridge and strongly in the outer (abaxial) medial domain where the later replum differentiates (Figure 6a). At stage 9 the adaxial expression became confined to the newly differentiating septum, while the abaxial expression expanded uniformly to encompass the entire outer epidermal layer (Figure 6b). Expression continued within the developing septum, becoming confined to the epidermis during stage 11 and was weaker where the two septum halves were fusing to form the transmitting tract (Figure 6c). Strong expression was also detected within the ovules, corresponding to the embryo sac (Figure 6c). By stage 12, expression was present throughout the entire ovule, continued within the septum, and was present at the stigma/transmitting tract boundary, within the epidermis of the valves, and strongly in the VM (Figure 6d). During anthesis, pollination, and fertilization (stages 13–15), expression faded in the stigma and valves, but remained strong within the ovules and VM (Figure 6e,f). VM expression continued strongly through stage 16 (Figure 6g), becoming localized to the separation layer as the DZ differentiated during stage 17 (see Rajani and Sundaresan, 2001).

Figure 6.

 Expression of the ALC:GUS reporter. ALC:GUS expression in WT (a–p) and in spt-2 (q–s), in transverse sections (expression is pink) and in whole mounts (expression is blue).
(a–c) Transverse sections through the upper regions of carpels. (a) Stage 8 flower: Expression occurs within the septum (arrowhead) where the replum will differentiate [r], and within the connective of the developing anthers [a]. (b) Stage 9 gynoecium: Expression is seen in the differentiated septum (arrowhead) and within the outer epidermis [oe]. (c) Stage 11 gynoecium: Expression is located within the outer epidermis [oe], embryo sac [e], and along the epidermis of the septum [*], but is weaker at the point of septum fusion (arrowhead).
(d–g) Stage 12, 14, 15 and 16 flowers: Expression occurs in the stigma [st], valve [v], along the valve margin [vm] and the nectaries [n] at stage 12, declining in the stigma and valves by stage 15 but continuing in the valve margins.
(h) Late stage developing anther: Expression is seen in the degrading tapetum [t].
(i) Basal region of a petal [p]: Expression occurs along lateral margins.
(j) Expression in the cryptic abscission zone between the pedicel and main stem (arrows). Inset shows dissected junction.
(k) Recently germinated seedling shows expression in the SAM.
(l, m) 7-day-old seedling: Expression occurs within developing vasculature of leaves [l], cotyledons [c], and hypocotyl [h], and lateral margins [lm] of young leaves.
(n–p) Roots: Expression is associated with lateral root emergence (arrowhead in n), the primary/lateral root junction [*] but not the root tip [rt] (o), and the stele [s] (o, p).
(q) Late stage 8 spt-2 gynoecium: Expression is in the developing septum (arrowhead).
(r) Stage 10 spt-2 gynoecium: Expression occurs weakly in the reduced septum (arrowhead), outer epidermis of the ovary, and in ovules.
(s) Stage 12 spt-2 flower: Expression is present in the nectaries, valve margins, stigma, and valves, although weaker than in WT (d).

Outside the gynoecium, ALC expression was detected within the connective of developing anthers during stage 8 (Figure 6a). By stage 11, expression was present in the tapetum (Figure 6h) until it degenerated during stages 12–13. ALC was also present in nectaries throughout their development (Figure 6d–f), along the margins of the petal claw (Figure 6i), and across the cryptic pedicel-stem abscission zone (Figure 6j). In vegetative regions, ALC was expressed in young seedlings, the margins of leaves, and roots (Figure 6k–p).

In the strong spt-2 mutant, ALC expression remained in the developing septum and outer adaxial medial regions during late stages 8–9 (Figure 6q). This expression continued during stage 10 (Figure 6r) although weaker than seen in WT, possibly because of the disrupted development of these tissues. At stage 12, expression was still present in the VMs, the valves themselves, and faintly in the poorly developed stigma (Figure 6s). Thus ALC gynoecium expression is largely independent of SPT function.

Comparison with SPT (Groszmann et al., 2010) revealed that ALC and SPT co-express in a range of tissues throughout development, including numerous tissue types associated with cell separation events (Table S1 and Figure S5). Within the gynoecium, co-expression occurs in the medial ridge from stage 8 and in the subsequent septal outgrowth and transmitting tract, as well as the stigma from stage 11. Both genes begin to express within the region of the VM as early as stage 9, becoming more localized once the VM is fully differentiated. Expression continues in the VM during silique development with ALC expressing more strongly (Figure S5). In the mature silique, both gene products become localized to the separation layer of the DZ (Table S1 and Figure S5).

Potential ALC regulatory elements identified through orthologue comparisons

As SPT and ALC are co-expressed in carpel and fruit margins, ALC is likely to share common regulatory elements with the SPT promoter (Groszmann et al., 2010). To investigate this, conserved upstream sequences of AtALC and five Brassica ALC orthologues were compared with SPT sequences (Figures 7 and S6). Within the 1447 bp AtALC upstream intergenic region (up to the neighbouring gene) three annotated transposons occur (At5TE96480; At5TE96485; At5TE96490). These commence 475 bp upstream of the transcriptional start site (TSS) and are conserved between the ALC orthologues but absent from the SPT promoter region.

Figure 7.

 Conserved regions of the ALC promoter. AtALC promoter regions conserved with orthologues from Arabidopsis lyrata (1), Brassica oleracea (3), and Brassica rapa (1) (Figure S6). A transposable element (At5TE96480) occurs at the same position in all sequences examined. This and two adjacent TEs are not expressed (Lister et al., 2008) (a) Homology between AtALC and the consensus sequence was determined using a sliding 9 nt window (red line). The central bolded dotted line represents 50% identity averaged across the 9 nt window (i.e. central nucleotide ± 4 nt). Seven distinct peaks were identified.
(b) AtALC promoter sequence with conserved regions highlighted. Identification of the TSS (+1) was based on the longest Arabidopsis ALC cDNA available (Genbank AY084632). Several discernable regulatory sequences are boxed: E-boxes, TATA-box, AtREG589, and a GA element.

An alignment of this 475 bp region of the ALC promoters generated a consensus sequence (Figure S6) which revealed seven conserved regions (Figures 7 and S6). The first four are in close proximity to the TSS and a conserved TATA-box. Region 1 is situated within the 5′ UTR and contains a GA element (Figure 7). Region 2 is ALC specific and contains a regulatory element (AtREG589) identified through the Plant Promoter Database (Yamamoto and Obokata, 2008). Conserved regions 5–7 occur further upstream and have no known discernible regulatory sequences. Strikingly, however, both regions 3 and 4 contain a conserved atypical E-box (CACGCG) element (Figure 7) identical to the one in the SPT promoter required for VM/DZ expression (Groszmann et al., 2010).

SPT can form homodimers, and heterodimers with ALC

bHLH proteins may function as either homo- and/or heterodimers (Murre et al., 1994; Massari and Murre, 2000). Consistent with their overlapping expression patterns and strong conservation in the HLH dimerization domain, SPT was able to heterodimerize with ALC in a yeast two-hybrid system (Figure 8a,b). These results were confirmed by co-expression experiments in onion epidermal cells. The GFP signal emitted by a SPT(ΔNLS)-GFP fusion protein lacking the NLS was present throughout the cell (Groszmann et al., 2008; Figure 8c and Table S2), but when co-expressed with either SPT or ALC it became restricted to the nucleus (Figure 8d,e and Table S2).

Figure 8.

 SPT can form homodimers, and heterodimers with ALC.
(a, b) Yeast two-hybrid growth plates (a) and assay (b), showing interaction between BD-SPT bait, and AD-SPT or AD-ALC prey. Positive control: BD-p53 + AD-T-antigen, Negative control: BD-Lam + AD-T-antigen. (BD-SPT alone does not auto-activate significantly (Groszmann et al., 2008)).
(c–e) Biolistic transient expression in onion epidermis. (c) 35S:SPT(ΔNLS)-GFP alone, showing GFP signal throughout the cell. (d, e) Co-expression of 35S:SPT (d) or 35S:ALC (e) with 35S:SPT(ΔNLS)-GFP, resulting in nuclear localization of GFP indicating dimerization and nuclear sequestering of SPT(ΔNLS)-GFP.

Over-expression of ALC can partially compensate for a loss of SPT function, and over-expression of SPT can fully compensate for loss of ALC function

To test if the partially redundant SPT and ALC proteins could fully substitute for each others’ function, we first ectopically over-expressed ALC in a spt-2 mutant background. Previous work has shown that 35S:SPT can fully complement the strong spt-2 mutant (Groszmann et al., 2008). One vigorous representative T2 plant from each of 10 independent 35S:ALC spt-2 transformed lines was selected. The first 25 siliques on the main shoot were measured (Table 1). 35S:ALC spt-2 plants showed a significant improvement in fruit length and seed set, and although less than the WT and 35S:SPT spt-2 plants, they were now comparable with the weaker spt-1 mutant. 35S:ALC was able to consistently restore apical fusion between the two carpels, enhance differentiation and growth of stigmatic papillae, and improve septum union in the basal half of the gynoecium and fruit (e.g. Figure 9a). We also tested if ALC expression driven by the SPT promoter could complement spt-2 defects. 10 complementing T1 plants carrying pSPT-1262:ALC were obtained, and the four most strongly complementing showed some increased silique elongation and seed set, although considerably less than the 35S:ALC construct (Table 1). The weak complementation may have been stronger if the longer pSPT-6253 promoter region carrying general enhancers (Groszmann et al., 2010) had been used.

Table 1. ALC complementation of the spt-2 mutant
 Mean seeds per silique ± SEMMean silique length (mm) ± SEM
  1. SEM, standard error of the mean. aFrom Groszmann et al. (2008).

WTa56.8 ± 5.912.0 ± 0.4
spt-2a1.8 ± 0.43.6 ± 0.1
spt-1a18.2 ± 2.37.6 ± 0.3
35S:ALC spt-222.5 ± 2.87.6 ± 0.4
pSPT-1262:ALC spt-25.7 ± 1.64.9 ± 0.4
35S:SPT spt-2a58.6 ± 1.211.3 ± 0.1
Figure 9.

 Reciprocal complementation test of spt and alc single mutants using 35S:ALC and 35S:SPT respectively.
(a) Mature siliques showing partial complementation of an otherwise severely affected spt-2 mutant silique (left) by 35S:ALC (right). Complemented siliques now resemble the weaker spt-1 mutant (centre).
(b) Mature dehiscent WT silique, indehiscent alc-1 silique, and a complemented and now dehiscent 35S:SPT alc-1 silique.
(c) Transverse sections through mature siliques. In the WT (top), the valves [v] detach from the replum [r] along the separation layer directly adjacent to the lignified cells (stained blue). In the alc-1 indehiscent silique (centre) the valves are held by a lignified bridge (arrows). In the complemented 35S:SPT alc-1 silique (bottom) no lignified bridge forms and valves detach from the replum.

To test for reciprocal redundancy of SPT in alc mutant plants during silique development, indehiscent alc-1 plants were transformed with 35S:SPT. Ten examined independent T1 plants were now fully or partially dehiscent (Figure 9b). The DZ of strongly complemented stage 17B siliques were examined in thin sections (Figure 9c). 35S:SPT alc-1 siliques were indistinguishable from WT, and showed clear separation between valves and replum, with the ‘lignified bridge’ present in the alc-1 mutants and disrupting the separation layer clearly absent. Thus 35S:SPT can fully rescue loss of ALC function defects in the DZ.

Interactions between SPT and other genes involved in VM/DZ development

Expression of ALC in silique VM regions is regulated by FUL and SHP1/2. To determine if SPT expression is also regulated in parallel with ALC given its shared role in VM/DZ development described here, SPT:GUS expression was examined in ful single and shp1 shp2 double mutant backgrounds.

FUL normally functions to confer valve identity by excluding VM gene expression from the valves proper. The pSPT-1262:GUS reporter, which is expressed in the VM but not the valves themselves (Figure 10a), was crossed into a ful mutant background. Significantly, pSPT-1262:GUS was now ectopically expressed throughout the ful mutant valves (Figure 10b–d), demonstrating that FUL can repress SPT expression in the valves, just as it does ALC. (The expression of SPT normally seen in valves and conferred by promoter sequences upstream of –1262 bp (Groszmann et al., 2010) may be regulated independently of FUL.)

Figure 10.

pSPT-1262:GUS expression in siliques of ful and shp1 shp2 mutants.
(a) Expression in WT VM (arrows; [v] valves, [r] replum).
(b–d) Expression in the ful-2 mutant showing ectopic expansion of SPT:GUS VM expression into the valves.
(e) SPT:GUS expression is still present in the shp1 shp2 double mutant, although more diffuse than in WT (a).

As alc can partially suppress the defective silique elongation of the ful-5 mutant (Liljegren et al., 2004), we also examined the interaction between spt and ful looking for a similar effect. However, spt-2 ful-5 double mutant plants revealed earlier disruptions to gynoecium development, which were now shorter, more severely unfused at the apex, with elongated divergent styles, and did not set seeds (Figure S7a–c). Thus SPT and FUL share functions early in gynoecium development apparently not shared by ALC and FUL.

SHP1 and SHP2 act to define the VM region, where SPT is expressed. Despite the loss of VM identity in shp1-1 shp2-1 double mutants, pSPT-1262:GUS expression was still detected in the VM, but usually weaker and more diffuse compared with expression in WT (Figure 10e). Thus SPT expression is only partially dependent on SHP activity, and in this regard is different from fully dependent ALC expression, although combined mutant studies reveal that some ALC function does occur independently of SHP (Liljegren et al., 2004).

SHP1 and SHP2 expression also overlaps with SPT earlier during gynoecium development, and SHP function is required for carpel margin tissue development (Savidge et al., 1995; Flanagan et al., 1996; Heisler et al., 2001; Colombo et al., 2010; Groszmann et al., 2010). Therefore a spt shp1 shp2 triple mutant was generated to determine if the spt single mutant phenotype was enhanced. However, the triple mutant displayed an additive phenotype, producing indehiscent, spatula-shaped siliques with no obvious enhancement of spt associated defects (Figure S7d). Thus SHP1/2 function apparently does not support SPT function redundantly early in gynoecium development.


The emergence and maintenance of ALCATRAZ

The two bHLH proteins SPT and ALC are each other’s closest relative in Arabidopsis, but whereas SPT relatives are found throughout seed plants (Groszmann et al., 2008), ALC related sequences were recoverable only from the Brassicaceae. Analysis of genomic regions surrounding the two genes revealed that they are part of larger duplicated segment, likely remnants from the At-β WGD that occurred in a common ancestor of the Brassicaceae (and related families Cleomaceae and Capparaceae) within the order Brassicales.

The function of the ancestral gene is not known, although conservation of the SPT sequences after the duplication suggests that SPT function may still broadly reflect the ancestral role. Consistent with this hypothesis, SPT-like genes are expressed in the ovary and developing fruit of peach (Prunus persea) (Tani et al., 2011), and in strawberry fruits (Fragaria × ananassa) (Tisza et al., 2010), both within the Rosaceae. In peach, expression occurs in the margins of the developing fruit endocarp which will later dehisce. While each gene still seems to be involved in defining both carpel and VM, SPT is now the major player in carpel margin development, while ALC function has become more important in DZ specification. This reciprocated partial redundancy indicates that SPT and ALC diverged through subfunctionalization.

It is of interest that two structures specified by SPT and ALC are novel features of the Brassicaceae family – the false septum of the gynoecium on the one hand, and the dehiscent silique that has led to a novel seed dispersal mechanism on the other. An extension to our hypothesis is that the two recently duplicated genes were involved together in the de novo origin of both these structures in the Brassicaceae lineage some time after gene duplication (neofunctionalization), and that their functions are now diverging (subfunctionalization). The fact that the originally duplicated genes escaped non-functionalization, a fate of up to 70% of genes following WGDs (Bowers et al., 2003), points to the importance of their diverging functions. Relevant here is that they encode transcription factors which appear disproportionately maintained after duplication events (Blanc and Wolfe, 2004).

Analysis of the ALC protein sequence reveals that it has lost about 40% of the length of SPT orthologues, occurring as short internal deletions and a substantial C-terminal truncation. Significantly, domains conserved in SPT are also retained in ALC with one exception, the acidic domain. This is required in SPT for its non-redundant functions in early gynoecium development (Groszmann et al., 2008), but is presumably not required for later VM/DZ development. Its absence in ALC is likely to account for the inability of ectopically expressed 35S:ALC to fully compensate for the loss of SPT function in carpel margin specification. On the other hand, ALC retains the conserved amphipathic helix which possibly recruits co-activators as has been implied for SPT (Groszmann et al., 2008). One possible candidate is the putative transcription factor ALC-INTERACTING PROTEIN1 (ACI1) which interacts with the N-terminal region of ALC and is co-expressed with both ALC and SPT in mature gynoecia (Wang et al., 2008).

Both SPT and ALC share the His-5, Glu-9, Arg-12 basic domain configuration in the bHLH, which suggests that each binds the canonical G-box (CACGTG) element. However, differences amongst other residues of the basic domain could mean SPT and ALC have differing affinities for specific G-box elements predicated on flanking residues (Atchley and Fitch, 1997; Atchley et al., 1999; Martinez-Garcia et al., 2000; Massari and Murre, 2000). This could have facilitated the specialization of ALC by allowing regulation of a sub-set of genes specifically involved in DZ development.

Regulation of ALCATRAZ expression

Promoter regions evolve much more rapidly than do protein coding sequences, and the ALC promoter region shares little similarity with that of SPT. However, they do share conserved atypical E-box elements (CACGCG), one in SPT and two in ALC, all within 80 bp of the transcription start site. The element in SPT is required for VM and DZ expression, and this expression is lost in mutants of another bHLH gene IND (see below; Groszmann et al., 2010). It will be of interest to determine if the elements in ALC are similarly required for VM/DZ expression. If so, the presence of duplicate elements may account for its relatively strong expression.

SPT and ALC also share a repeated GA element located in their 5′ UTRs (Groszmann et al., 2010). These are found in around 20% of all plant promoters (Yamamoto et al., 2009) and can be bound by GA-binding proteins (GBP) influencing transcription of their targets by modulating the state of chromatin (Sangwan and O’Brian, 2002; Lehmann, 2004; Pauli et al., 2004; Kooiker et al., 2005; Yamamoto et al., 2009).

No obvious further promoter homology exists between SPT and ALC in the 475 bp ALC upstream region, including the three additional upstream regions conserved amongst ALC orthologues. Therefore it is likely that the overlap in ALC and SPT expression is conferred, in part, by the core promoter regions consisting of the E-box, TATA-box, and GA elements, a region absolutely required for basal SPT expression (Groszmann et al., 2010). Functional analysis of the ALC promoter region is required to test the importance of these shared elements, as well as testing if other shared regulatory elements occur outside of the 475 bp ALC upstream region analysed in detail here.

Collectively, SPT and ALC are expressed in many other regions. Some of these, including the DZ, also express genes associated with cell separation (Gonzalez-Carranza et al., 2007; Ogawa et al., 2009; Sun and van Nocker, 2010). It is possible that SPT and ALC function more generally in coordinating differentiation of such tissues, which may reflect an ancestral role. SPT and ALC are also expressed in some developing tissues where cell proliferation is occurring. These include cotyledons, leaves and petals, all of which are larger in spt mutants (Penfield et al., 2005; Ichihashi et al., 2010). Whether a concurrent loss of alc enhances the spt associated defects in these organs remains to be investigated.

The hidden role of ALC in septum and medial carpel development

The spt-alc double mutant reveals the full impact of losing SPT function on medial carpel development while exposing a hidden early developmental role for ALC. The alc associated enhancement of spt gynoecium defects first appears during stage 8, coinciding with the first detectable expression of ALC. Expansion of ALC expression using the 35S promoter significantly ameliorates the spt mutant phenotype. This indicates that temporal expression and/or gene dosage of ALC is also a limiting factor in its redundancy with SPT early in development. Although SPT and ALC are partitioning the proposed ancestral role, some redundancy may persist, stabilising development of carpel marginal tissues.

The hidden role of SPT in valve margin and dehiscence zone development

Both alc and spt single mutants have a clearly defined VM in the mature gynoecium (Rajani and Sundaresan, 2001; Alvarez and Smyth, 2002). However the abnormal VMs of SPT-SRDX transgenic plants, and in spt-alc double mutants, have revealed a previously unknown function for SPT in VM development, while confirming an earlier VM role for ALC (Liljegren et al., 2004). The individual contributions between SPT and ALC can be separated in the ful mutant background where ful alc double mutants have reduced ectopic VM development despite a functional SPT being present (Liljegren et al., 2004). This suggests that there is a SPT/ALC dosage requirement for VM development which can be met by either gene alone in the WT but not in the extreme case of ful mutants.

The SPT/ALC VM redundancy does not extend to the later differentiating DZ as only alc is strongly indehiscent despite their apparently identical expression patterns within the VM/DZ. Even so, a subtle remnant function for SPT in the DZ may remain as modest dehiscent defects have recently been observed in spt mutants in the Columbia background (T. Girin and L. Østergaard, personal communications). SPT has all of the conserved protein domains contained within ALC and is fully capable of complementing alc when under the control of the strongly expressing 35S promoter. Therefore, it seems likely that the gene dosage requirements persist into DZ development, but that here SPT levels are insufficient to compensate for loss of the more highly expressed ALC in alc mutants.

The gene dosage requirements may be established through the recently identified interactions between DELLA proteins with SPT and ALC (Arnaud et al., 2010; Gallego-Bartolome et al., 2010). High levels of gibberellins (GA) result in DELLA degradation and the release of SPT or ALC from the suppressing complex. Plants deficient in GA have defective DZ differentiation which can be directly linked to DELLA-ALC interactions (Arnaud et al., 2010). They also lack clearly defined VMs which cannot simply be attributed to ALC suppression but are more likely due to excessive DELLA suppression of both ALC and SPT.

It may also be relevant to dosage effects that SPT and ALC can heterodimerize as well as homo-dimerize, although whether all dimer types are functional, and if targets of the different dimer types are the same, is not known.

SPT VM expression is partially but not absolutely dependant on SHP activity, and is negatively regulated by FUL. This association with SHP and FUL is identical to that seen for the bHLH gene IND, a key regulator of VM and later DZ development (Ferrándiz et al., 2000 (in which IND was known as GT140); Liljegren et al., 2004). IND functions upstream of SPT in VM/DZ development (Groszmann et al., 2010). Its regulation of SPT expression may occur directly through binding of IND to the atypical E-box element in the SPT promoter required for VM expression (Groszmann et al., 2010), and indirectly through promoting high GA levels and therefore DELLA degradation (Arnaud et al., 2010). ALC also acts predominantly downstream of IND (Liljegren et al., 2004), with IND control of ALC presumably exerted through similar means to its regulation of SPT, including the atypical E-box sequences in ALC defined here.

Links between carpel margin genes and dehiscence zone genes

There is another example of closely related bHLH genes controlling tissues derived from carpel margins on the one hand, and the DZ on the other. The HECATE (HEC) proteins are required for the former (Gremski et al., 2007), and IND for the latter (Liljegren et al., 2004). Interestingly, the HEC proteins can dimerize with SPT (Gremski et al., 2007), and IND with ALC (Liljegren et al., 2004), and other combinations may occur. Thus the co-evolution of these temporally separate processes within developmentally related regions may have involved parallel subfunctionalization events. It will be interesting to further unravel the genetic and regulatory networks involved, and their evolutionary origin.

Experimental Procedures

Plant strains

The Landsberg erecta background was used. The spt-1 and spt-2 mutants (Alvarez and Smyth, 1999, 2002), and 35S:SPT-SRDX, 35S:SPT (Groszmann et al., 2008), and pSPT-1262:GUS (Groszmann et al., 2010) transgenic lines have been previously described. ful-2, ful-5, and shp1-1 shp2-1 seeds were kindly provided by Marty Yanofsky and alc-1 by Venkatesan Sundaresan.

Generation of 35S:ALC and pSPT-1262:ALC constructs

The ALC cDNA clone was kindly provided by Sarojam Rajani and Venkatesan Sundaresan. It was amplified using 5′-GGGGTACCATGGGTGATTCTGACGTCGGTGAT-3′, which has a KpnI site (underlined) just 5′ of the ALC ATG, and 5′-CCATCGATTCAAAGCAGAGTGGCTGTGGAAAA-3′ with a ClaI site (underlined) downstream of the ALC stop codon. This was placed into the KpnI and ClaI sites of pBluescript. From here it was excised and placed into pART7 (Gleave, 1992) and the 35S:ALC:OCS cassette was removed using NotI and placed into the pMLBART transformation vector (Groszmann et al., 2008). The ALC insert in pBluescript was also placed downstream of the pSPT-1262 promoter region present in a derivative of BJ36, and then both transferred into pMLBART (Groszmann et al., 2010).

Plant transformation, and reporter gene analysis

35S:ALC and 35S:SPT constructs were transformed into spt-2 and alc-1, respectively, using the floral dip method. GUS staining and subsequent analysis followed Groszmann et al. (2010). Biolistic transfection and analysis of nuclear localization of GFP was carried out as described in Groszmann et al. (2008).

Yeast two-hybrid analysis

The Clontech Yeast two-hybrid Matchmaker 3 kit ( was used. SPT cDNA was amplified from cDNA9 (Heisler et al., 2001) using 5′-CGCCCGGGAATGATATCACAGAGAGAA-3′, which has a SmaI site (underlined) just 5′ of the SPT ATG, and 5′-CGGGATCCTCAAGTAATTCGATCTTTTAG-3′ with a BamHI site (underlined) downstream of the stop codon. This was placed into the SmaI and BamHI polylinker downstream of the GAL4 DNA binding domain (BD) of the pGBKT7 vector creating the BD-SPT bait hybrid protein. The same amplified SPT cDNA was placed into the SmaI and BamHI sites downstream of the GAL4 activation domain (AD) of the pGADT7 vector, creating the AD-SPT prey hybrid protein. The ALC cDNA was amplified using 5′-CGCCCGGGGATGGGTGATTCTGACGTC-3′ and 5′-CCGGATCCTCAAAGCAGAGTGGCTGTG-3′ which incorporated a SmaI site upstream (underlined) and BamHI site downstream (underlined) of the ALC coding region, respectively. This product was placed into the SmaI and BamHI downstream of the GAL4 AD of the pGADT7 vector, creating the AD-ALC prey hybrid protein.

Scanning electron microscopy (SEM) and resin sections:

SEM of WT, spt, and spt-alc double mutant gynoecia were performed as described in Alvarez and Smyth (2002). WT and SPT-SRDX unfertilized mature siliques were prepared as above but were not sputter coated. Enhanced cellular contrast of the valve margin region was obtained using a 4 quadrant backscattered electron detector in a Zeiss EVO LS15 SEM. Tissue was viewed using 20 kV accelerating voltage under variable pressure mode, with 10 Pa chamber pressure. For resin sections, tissue was prepared, sectioned, and stained as described in Alvarez and Smyth (2002).

Identification of ALC orthologues

Sequences with similarity to ALC were identified using the 62 amino acids comprising the bHLH domain with four additional N-terminal and 9 C-terminal amino acids. This sequence was used as a query for BLAST searches against all available datasets from the following databases:

The 1447 bp upstream intergenic region of AtALC was used as the query in BLASTN searches performed against the first two databases listed above, and the TIGR B. oleracea shot gun sequence database (accessed through: ALC orthologues were identified through matching the available downstream amino acid sequence with that of AtALC.


We thank Marty Yanofsky for the ful-2, ful-5, and shp1-1 shp2-1 seed, and Venkatesan Sundaresan and Sarojam Rajani for alc-1 seed and ALC cDNA. This work was supported in part by the Australian Research Council, grant A19927094.