Organ Polarity in Arabidopsis. NOZZLE Physically Interacts with Members of the YABBY Family

Abstract

Plant lateral organs exhibit proximal-distal and adaxial-abaxial polarity. In Arabidopsis, abaxial cell fate is regulated in part by putative transcription factors of the YABBY family, such as FILAMENTOUS FLOWER (FIL) and INNER NO OUTER (INO), by a mechanism that currently is not fully understood. NOZZLE (NZZ) encodes a plant-specific nuclear protein. Genetic evidence has shown that NZZ is involved in the positive feedback regulation of INO, thereby acting both as a temporal and spatial repressor of INO transcription. This mechanism allows the ovule primordium to complete its proximal-distal organization, prior to the onset of adaxial-abaxial development in the chalaza. During our study, we isolated FIL in a yeast two-hybrid screen using NZZ as bait. In vitro pull-down experiments confirmed the NZZ-FIL interaction. NZZ also bound INO and YABBY3, suggesting that NZZ generally interacts with YABBY proteins in vitro. The polar-charged region of NZZ was necessary and sufficient to bind to the zinc finger of INO and to interact with its C terminus carrying the high mobility group-like domain. We suggest that NZZ coordinates proximal-distal patterning and adaxial-abaxial polarity establishment in the developing ovule by directly binding to INO.

Arabidopsis ovules provide an excellent system to study pattern formation in plants (Gasser et al., 1998; Chevalier et al., 2002). Three proximal-distal elements, the funiculus, chalaza, and nucellus, can be distinguished early on (Esau, 1977; Schneitz et al., 1995). After the primordium completes its proximal-distal extension, the inner and outer integuments develop in a sequential fashion. Abaxial outer integument initiation at the chalaza is the first morphological evidence that adaxial-abaxial polarity establishment has occurred and that the radially symmetrically primordium has switched to bilateral development (Baker et al., 1997; Schneitz et al., 1997; Villanueva et al., 1999).

Some progress has recently been made toward the understanding of the molecular mechanism underlying the coordination of proximal-distal and adaxial-abaxial pattern formation in the ovule. In this context, the INNER NO OUTER (INO) gene, and its regulation, is of central importance. INO belongs to the YABBY family of putative plant-specific transcription factors (Villanueva et al., 1999). Additional members include FILAMENTOUS FLOWER (FIL) and YABBY3 (YAB3; Bowman, 2000). YABBY genes are associated with the regulation of abaxial cell fate, by a mechanism which presently is not understood (Bowman and Smyth, 1999; Sawa et al., 1999b; Siegfried et al., 1999; Villanueva et al., 1999). There is evidence coming from in vitro binding studies that FIL protein dimers bind to DNA through their high mobility group (HMG)-like (YABBY) domains in a nonsequence-specific fashion (Kanaya et al., 2002), supporting the idea that YABBY proteins act as transcriptional regulators. Ovules lacking INO function only form the inner integument and thus appear radialized (Baker et al., 1997; Schneitz et al., 1997; Villanueva et al., 1999; Balasubramanian and Schneitz, 2000, 2002; Meister et al., 2002). Before the integuments initiate, transcription of INO can be detected in a few cells on the abaxial side of the ovule and INO expression is subject to a positive autoregulatory loop (Villanueva et al., 1999; Balasubramanian and Schneitz, 2000, 2002; Meister et al., 2002, 2004). Several factors interfering with this process have been identified. SUPERMAN (SUP), encoding a putative zinc-finger transcription factor (Sakai et al., 1995), is required for abaxial INO expression, as in sup-5 loss-of-function mutants INO is misexpressed and INO transcription extends into the adaxial chalaza. This causes the loss of unequal outer integument initiation and growth, which eventually leads to a more radialized ovule (Gaiser et al., 1995; Villanueva et al., 1999; Meister et al., 2002). NOZZLE (NZZ), also known as SPOROCYTELESS, encodes a nuclear protein with limited homology to transcriptional regulators (Schiefthaler et al., 1999; Yang et al., 1999). Previous genetic evidence indicated that NZZ coordinates proximal-distal and adaxial-abaxial axis development through the temporal control of INO expression and via its involvement in the positive autoregulatory loop of INO transcription, which also includes AINTEGUMENTA (ANT; Balasubramanian and Schneitz, 2000, 2002). ANT encodes a transcription factor of the APETALA2 family and is involved in cell proliferation and growth control (Elliott et al., 1996; Klucher et al., 1996; Eshed et al., 1999; Krizek, 1999; Mizukami and Fischer, 2000; Nole-Wilson and Krizek, 2000). In addition to the temporal control of INO expression, NZZ, together with ABERRANT TESTA SHAPE (ATS; Léon-Kloosterziel et al., 1994), acts as a spatial repressor of INO transcription in the adaxial chalaza, similar to SUP (Balasubramanian and Schneitz, 2000, 2002). Thus, NZZ acts both as a temporal and spatial repressor of INO transcription during ovule development. However, the corresponding molecular mechanism remains to be elucidated. The involvement of ANT in the autoregulatory loop of INO transcription, as well as the redundancy of NZZ and ATS in spatially restricting INO expression, leave open several possibilities for how the repression of INO by NZZ can be achieved. In this study, we found the YABBY protein FIL as an interacting partner of NZZ in a yeast two-hybrid screen, and we show that NZZ can interact with several YABBY proteins in vitro, including INO. Our genetic analysis leaves open a possible NZZ-FIL interaction in vivo. By contrast, our observed in vitro NZZ-INO interaction is in accordance with previously reported genetic results. This result indicates that NZZ might coordinate proximal-distal and adaxial-abaxial axis formation in the developing ovule by directly interacting with INO.

RESULTS AND DISCUSSION

NZZ Interacts with FIL in Yeast and In Vitro

We used a yeast two-hybrid approach to identify proteins that interact with NZZ. To overcome potential problems of auto-activation of classical polymerase II reporter genes caused by the bait protein, we used an alternative two-hybrid system, which is based on a polymerase III-transcribed reporter gene (Petrascheck et al., 2001). We screened a cDNA library of approximately 5.8 times 10⁶ clones derived from inflorescences containing closed flower buds. Two of the clones that passed the selection contained fragments of the FIL cDNA. Both F263 and F207 (see Table I for details of plasmid constructs) contain the full sequence of the YABBY domain (Arg-136-Pro-183; Bowman and Smyth, 1999). Clone F207 starts at residue 23 (Asp-23) with respect to the full-length FIL protein and contains the complete zinc-finger domain according to two reports (Bowman and Smyth, 1999; Villanueva et al., 1999) but does not according to another (Kanaya et al., 2001).

Table I.

Plasmid constructs

Name	Amino Acids	Color Code						Protein	Fused Tag(s)
τ-NZZ	Met-1-Leu-314	N	PC	MR		C		NZZ	τ-138
ΔN1–227	Met-1-Tyr-227	N	PC	MR				ΔNZZ	τ-138
ΔN168–314	Ser-168-Leu-314					C		ΔNZZ	τ-138
NZZ	Met-1-Leu-314	N	PC	MR		C		NZZ	GST
ΔN1–148	Met-1-Thr-148	N	PC	MR				ΔNZZ+	GST
ΔN78–148	Glu-78-Thr-148			MR				ΔNZZ	GST
ΔN41–148	Lys-41-Thr-148		PC	MR				ΔNZZ	GST
ΔN1–84	Met-1-Ala-84	N	PC					ΔNZZ	GST
ΔI1–59	Met-1-Val-59	N	Zn					ΔINO	Xpress
ΔI1–138	Met-1-Ala-138	N	Zn	MR	P			ΔINO	Xpress
ΔI1–174	Met-1-Ala-174	N	Zn	MR	P	H		ΔINO	Xpress
INO	Met-1-Glu-231	N	Zn	MR	P	H	C	INO	Xpress
ΔI57–138	Leu-57-Ala-138			MR	P			ΔINO	Xpress
ΔI138–174	Ala-138-Ala-174					H		ΔINO	Xpress
ΔI138–231	Ala-138-Glu-231					H	C	ΔINO	Xpress
F263	Met-1-Tyr-229	N	Zn	MR	P	H	C	+FIL	Gal4DBD
F207	Asp-23-Tyr-229		Zn	MR	P	H	C	ΔFIL	Gal4DBD
FIL-myc	Met-1-Tyr-229	N	Zn	MR	P	H	C	FIL	Xpress/MYC
FIL	Met-1-Tyr-229	N	Zn	MR	P	H	C	FIL	Xpress
YAB3	Met-1-Asn-240	N	Zn	MR	P	H	C	YABBY3	Xpress

τ-138, τ-138 protein is a cofactor of RNA polymerase III; Gal4pDBD, DNA-binding domain of the Gal4 protein; N, N terminus; PC, polar-charged domain; MR, middle region; C, C terminus; Zn, zinc-finger; P, Pro-rich region; H, HMG-like domain. Δ, Indicates that a truncated form of the protein is used. The YABBY domain contains both the Pro-rich region and the HMG-like domain. The N terminus, the C terminus, and the middle region of NZZ are unrelated to the respective domains of members of the YABBY family. F263 extends further than the first Met of the normal FIL protein by an additional 34 amino acids. F207 is N-terminally truncated by 22 amino acids. Deletion of a single base within the last codon before the stop codon resulted in a frame shift and added 14 amino acids unrelated to NZZ to the C terminus of the ΔN^1–148 peptide. All tags are fused to the N terminus except for cMYC, which is fused to the C terminus.

The NZZ protein contains a domain of approximately 40 amino acids in length (Lys-41-Gln-80) toward its N terminus, containing mostly polar or charged residues (Schiefthaler et al., 1999). This polar-charged domain can be subdivided into a basic domain spanning residues 41 to 62 and into a predicted helix domain (residues 63–85; Schiefthaler et al., 1999; Yang et al., 1999). A hydrophobic middle region (Leu-81-Val-146) separates the polar-charged domain from the C terminus. We found that the N terminus of NZZ (ΔN^1–227) was necessary and sufficient for the interaction with the fusion proteins F207 and F263 in yeast (Fig. 1A). The C terminus of NZZ (ΔN^168–314) was not involved in binding to FIL. The NZZ-FIL protein-protein interaction was confirmed using in vitro pull-down assays (Fig. 1B). NZZ fused to glutathione S-transferase (NZZ), but not glutathione S-transferase (GST) alone, could bind FIL (Fig. 2, B and C) and Xpress/cMYC double-tagged FIL (FIL-myc), as detected by anti-Xpress and anti-c-myc antibodies, respectively (Fig. 1B).

Figure 1.

NZZ interacts with FIL in yeast and in vitro. A, Result of yeast two-hybrid screen. The clones FIL207 and FIL263 show interactions that depend on the presence of NZZ protein. Full-length NZZ (τ-NZZ) and the N-terminal fragment of NZZ (ΔN^1–227) interact with FIL207 and FIL263, whereas the C terminus of NZZ (ΔN^168–314) does not. +, Wild-type SNR6 was expressed as positive control; −, Gal4pDBD encoding plasmid pPC97 or empty vector yCplac111 were used as negative controls. 25°C and 37°C indicate permissive and restrictive temperature, respectively. B, NZZ and FIL interact in vitro. Two independent pull-down reactions involving GST-NZZ (NZZ) and Xpress/cMYC double-tagged FIL recombinant protein (FIL-myc) were split equally (lanes 3 and 7 and lanes 4 and 8), and the result was detected by using anti-Xpress antibodies (left) and anti-c-myc antibodies (right). FIL-myc is retained by NZZ (lanes 3, 4, 7, and 8) but not by GST (lanes 2 and 6).

Figure 2.

Summary of pull-down reactions. A, FIL (FIL, FIL-myc), INO, and YAB3 were transcribed and translated in vitro using [³⁵S]Met in the mix. The SDS-PAGE gel was dried and exposed to an x-ray film. Approximately equal amounts of FIL, FIL-myc, and YAB3 were translated. The amount of translated INO varied in different reactions (compare INO versus INO*). No translation product was detected when empty pRSET plasmid was added (−). The TNT reactions were split, and pull-down assays with equal amounts of GST protein alone or GST-NZZ (NZZ) recombinant protein were performed. B and C, The result of the pull-down reactions is shown on an x-ray film overlaying the dried gel. B, Neither the water control nor the luciferase (L) protein ever produced a signal (lanes 3, 4, 8, and 9). C, NZZ interacts with FIL, FIL-myc, YAB3, and INO, but none of the Xpress-tagged proteins interacts with GST alone (compare lanes 6 and 11 versus 1 and 10; 13 versus 12; 15 versus 14; 7, 17, and 18 versus 2 and 16). INO**, 50 μL of INO* input was used for pull-down reaction (lanes 16 and 18), whereas 20 μL of INO* input was used for pull-down reaction marked INO* (lane 17).

NZZ Can Interact with Other YABBY Proteins

NZZ was reported to be a spatial and temporal repressor of the YABBY gene INO (Balasubramanian and Schneitz, 2000, 2002). INO expression is restricted to very few cells on the abaxial side of the ovule primordium and to the outer integument (Villanueva et al., 1999; Balasubramanian and Schneitz, 2000, 2002; Meister et al., 2002; Fig. 6B). Thus, INO would be difficult to detect in a yeast two-hybrid screen. Therefore, we expressed additional members of the YABBY family using in vitro transcription and translation to directly test these proteins for their potential to bind NZZ. FIL and YAB3 represent the most recent gene duplication event within this family, whereas INO is more distantly related to FIL and YAB3 (Bowman, 2000). We performed pull-down experiments using GST-NZZ recombinant protein and GST protein alone to detect direct protein-protein interactions with different YABBY proteins. NZZ bound FIL (FIL, FIL-myc), YAB3, and INO reproducibly and independently of GST (Fig. 2, B and C). The interaction of NZZ with INO was proportional to the amount of INO that was added to the reaction (Fig. 2C, lanes 16–18). Thus, NZZ can bind directly to three tested members of the YABBY family, FIL, YAB3, and INO, in vitro, suggesting that NZZ is a general interaction partner of YABBY proteins.

Figure 6.

Expression of marker genes in floral organs of wild-type, nzz-2, fil-5, and nzz-2 fil-5 double mutant plants. In situ hybridizations of cross-sections through flowers of Landsberg erecta (A, B, E, I, M), nzz-2 (C, D, F, J, N, Q), fil-5 (G, O, R), and nzz-2 fil-5 (H, K, L, P, S, T) flowers are shown. INO mRNA can be detected in Landsberg erecta ovules around stage 2-I (A, B) but slightly earlier (stage 1-II) and at a more distal position in nzz-2 mutant ovules (C, D; Balasubramanian and Schneitz, 2000, 2002). INO expression as detected in transverse sections through ovules at stage 2-III (R) and around stage 2-III (Q, S, T) is shown. PHB transcript accumulation, as detected in longitudinal sections through stage 2-III ovules (E–H) and in transverse sections of flowers (I–L). REV expression in stage 2-II/2-III ovules is shown (M–P). Arrows point to regions showing high levels of PHB transcript accumulation. Scale bar = 20 μm. ab, abaxial; ii, inner integument; oi, outer integument; nu, nucellus; dt, distal tip; st, stamen; pe, petal; ca, carpel; o, ovule.

The Polar-Charged Domain of NZZ Is Necessary and Sufficient to Bind INO in Vitro

To map the interaction domain, we performed in vitro pull-down assays combining truncated forms of NZZ with full-length [³⁵S]Met-labeled INO protein (Fig. 3A). We found that INO was retained by all GST-fusion proteins that contained the polar-charged domain of NZZ (Fig. 3A, top, lanes 1, 2, 4, 5, and 6). The middle region of NZZ failed to bind INO (Fig. 3A, top, lane 3), and the N-terminal residues (Met-1-Leu-40) were not necessary for the NZZ-INO interaction (Fig. 3A, top, lanes 4 and 5 versus 6). Neither a Met-to-Val change at residue 64 within the predicted helix of the polar-charged domain nor a Val-to-Ala substitution at residue 29 had an effect on the INO-NZZ protein-protein interaction (Fig. 3A, lanes 4–6).

Figure 3.

Mapping of the NZZ and INO interaction domains. A, Mapping the INO-binding domain of NZZ. A, top, Result of pull-down reactions. X-ray autoradiograph shows amount of [³⁵S]Met-labeled INO protein, which was retained by various fragments of NZZ protein. A, bottom, Comparable concentrations of ΔN^1–148, ΔN^78–148, ΔN^41–148, ΔN^1–84, ΔN^1–84m2, and GST (lanes 1–6) as well as ΔN^41–148m1 (data not shown) was recovered from glutathione sepharose purified bacterial extracts, as detected on a Coomassie Blue-stained SDS-PAGE. Less GST-NZZ was recovered (lane 7) since NZZ showed a very strong tendency to accumulate in inclusion bodies. Input of the GST-fusion proteins per pull-down reaction was proportional to the input shown in A (bottom). B, Mapping of the NZZ-binding domain(s) of INO. X-ray films are shown in all three sections. B, bottom, Proportional amount of in vitro transcribed/translated fragments of INO protein is shown, as used in pull-down reactions with NZZ (top) and with ΔN^41–148 (middle). Arrowhead points at band of expected size.

NZZ Can Bind the Zinc-Finger Domain and the C-Terminal Domain of INO in Vitro

INO shares two conserved domains with the other members of the YABBY protein family, a C₂C₂-type zinc finger near the N terminus (Cys-22-Cys-53) and the YABBY domain (Lys-130-His-180; Bowman and Smyth, 1999; Villanueva et al., 1999), the latter of which is located toward the C-terminal end of the protein. The YABBY domain contains a helix-loop-helix motif with sequence similarity to the first two helices of the HMG domain (Ala-141-Pro-177 or Phe-145-His-180; Bowman and Smyth, 1999; Villanueva et al., 1999). We tested whether NZZ protein can recognize and bind to specific domains of INO. Therefore, we expressed full-length NZZ, as well as a truncated form of NZZ (ΔN^41–148), which contained only the polar-charged domain and the middle region fused to GST (Fig. 3A, bottom, lanes 7 and 3, respectively). We used NZZ and ΔN^41–148 to perform in vitro pull-down assays with several truncated INO proteins. Qualitatively, full-length NZZ and ΔN^41–148 performed the same with respect to binding INO (compare Fig. 3B, top and middle). Quantitatively, more INO was retained in the pull-down reactions involving the truncated form of NZZ (ΔN^41–148; Fig. 3B, top versus middle), likely because more ΔN^41–148 than NZZ was used per pull-down reaction (see input, Fig. 3A, bottom, lanes 3 versus 7). NZZ and ΔN^41–148, both bound to ΔI^1–138, ΔI^1–174, and INO, all of which share the zinc-finger domain (Fig. 3B, lanes 2–4, respectively). NZZ and ΔN^41–148, also bound to ΔI^1–59m, but binding was much reduced (Fig. 3B, lane 1 versus 2–4). Subsequent analysis showed that ΔI^1–59m contains a Cys-to-Ser substitution at residue 29 in the zinc-finger domain. Our result indicates that the conserved Cys-29 may be crucial for the folding of the zinc finger (Bowman and Smyth, 1999; Kanaya et al., 2001). We cannot exclude, however, that the lack of the INO middle region in ΔI^1–59m, when compared to ΔI^1–138, destabilizes the protein and causes the failure to bind NZZ. Still, the middle region (Leu-57-Ala-138), in between the zinc finger and the HMG-like domain, neither bound to NZZ nor to ΔN^41–148 (Fig. 3B, top and middle, lane 5), indicating that the middle region is not necessary for the NZZ-INO interaction.

The C terminus of INO also interacted with ΔN^41–148 and with NZZ. The plasmid ΔI^138–231 (Ala-138-Glu-231) contained the HMG domain-like helix-loop-helix motif but lacked the Pro-rich region of the YABBY domain. The in vitro transcription and translation of ΔI^138–174 failed repeatedly (Fig. 3B, bottom, lane 6). Thus, it is not clear whether the HMG-like domain alone would be sufficient for binding to either NZZ or ΔN^41–148. To summarize, NZZ as well as the truncated fragment ΔN^41–148, which contained only the polar-charged domain and the middle region, both bound to the conserved zinc finger as well as to the C terminus of INO but not to the less conserved middle region of INO.

Genetic Analysis Leaves Open a Possible In Vivo Interaction of NZZ and FIL

NZZ can interact with FIL, YAB3, and INO in vitro. The NZZ-INO interaction is supported by extensive genetic evidence (Balasubramanian and Schneitz, 2000, 2002). In this section, we present our data that relate to the in vivo relevance of the observed NZZ-FIL interaction. NZZ mRNA is present in vegetative tissues, in inflorescences, and in flowers, where NZZ transcript accumulates in carpel, stamen, petal, and ovule primordia, as well as in nectaries (Schiefthaler et al., 1999; Fig. 4, A–D, I, and Q). NZZ expression in stamens is strongest in the lateral regions of the primordium (Fig. 4, A and B). In stage 2-III ovules, NZZ transcript accumulation is strongest in the integuments, but expression can be detected throughout the developing ovule (Schiefthaler et al., 1999; Fig. 4, C, D, and Q). FIL is expressed throughout the anlagen of all aboveground lateral organs, but its expression becomes subsequently restricted to the abaxial region of organ primordia (Sawa et al., 1999b; Siegfried et al., 1999). In flowers, FIL mRNA accumulates in the abaxial domains of carpels, stamens, petals, and sepals (Sawa et al., 1999b; Siegfried et al., 1999; Fig. 4, E and F). No reproducible expression was detected in ovules (Siegfried et al., 1999; Fig. 4, G and H), even after 5-d exposure of the hybridized slides to the detection reagent (Fig. 4H). Thus, FIL and NZZ are both expressed in vegetative leaves, and their expression domains in flowers overlap in petal, stamen, and carpel primordia.

Figure 4.

NZZ and FIL mRNA expression patterns in flowers. RNA in situ hybridization detection of the NZZ transcript accumulation in Landsberg erecta (A–D, I, Q), nzz-2 (J, R), fil-5 (K, S1, S2), and nzz-2 fil-5 (L, T1, T2) floral tissue. A, Transverse section through stage 7 flower. B, Cross-section through flower around stage 6/7. C, Longitudinal section of stage 1-I ovules developing inside gynoecium. D, Ovules at stage 2-III. E to H, FIL mRNA accumulation pattern in wild-type floral organs. E, Longitudinal section through stage 5 flower. F, Transverse section through stage 7 flower. Longitudinal sections of stage 1-II ovules (G) and stage 2-III ovules (H). I to L, Transverse sections through stage 8 flowers of wild-type (I), nzz-2 (J), fil-5 (K), and nzz-2 fil-5 (L) plants. M, Longitudinal section through nzz-2 mutant flower around stage 5. N, Transverse section through a fil-5 mutant flower at stage 7. Longitudinal sections through ovules at stage 2-III (Q, S1, S2) and around 2-II/III (R, T1, T2). T1 and T2, Arrows point out elevated levels of NZZ transcript accumulation in the distal tip region. O and P, High magnification scanning electron micrograph pictures of the adaxial anther-filament junction (O) and the abaxial surface of anthers (P) of nzz-2 mutants are shown. O, Upper arrow points to irregularly shaped uniformly-sized cells. The lower arrow points to the gradual change toward files of more elongate cells typical for filaments. P, Stomata (asterisk) and irregularly shaped cells are present on the abaxial surface of nzz-2 anthers. ca, carpel; st, stamen; fst, filamentous stamen; pe, petal; o, ovule; nu, nucellus; dt, distal tip; ii, inner integument; oi, outer integument; ad, adaxial; ab, abaxial. Arrowheads points out NZZ expression in stamens and petals, respectively. Scale bar = 20 μm. Floral stages after (Smyth et al., 1990) stages of ovule development (Schneitz et al., 1995).

The phenotype of loss-of-function nzz mutants is mostly restricted to ovule and stamen development (Schiefthaler et al., 1999; Yang et al., 1999; Balasubramanian and Schneitz, 2000; Fig. 5, B, F, and J). Additionally, the inflorescence appears to be more compact (Balasubramanian and Schneitz, 2000). Mutants homozygous for strong fil loss-of-function alleles display a range of phenotypes, which mostly concern flower development. Homozygous fil mutant flowers are affected in organ number, organ positioning, and organ identity (Chen et al., 1999; Sawa et al., 1999a, 1999b; Siegfried et al., 1999). In the case of fil mutants, stamen either arrest as filamentous anther-less structures (Fig. 5K1) or occasionally develop one or two locules, which produce fertile pollen (Fig. 5K2). Ovules develop normally in fil mutants (Chen et al., 1999; Sawa et al., 1999a; Siegfried et al., 1999; Fig. 5G).Thus, both NZZ and FIL functions are necessary for normal stamen development. Plants doubly mutant for both nzz-2 and fil-5 segregated in the F₂ progeny of a cross between nzz-2 and fil-5 in the expected ratio for two recessive Mendelian traits. The nzz-2 fil-5 doubly homozygous plants were nondistinguishable from fil-5 single mutants, except for the ovules, which appeared identical to the ovules that developed in the carpels of nzz-2 mutants (Fig. 5, D, H, L1, and L2). Some stamens of nzz-2 fil-5 double mutants arrested as radialized filaments (Fig. 5L1). When anthers developed on top of the filament, the anthers appeared flat and were composed of uniformly shaped cells surrounded by epidermal cells (Fig. 5L2; data not shown). Locules never developed and we never observed pollen, as this is also the case for nzz single mutants. We did not notice any novel phenotypes in the nzz-2 fil-5 double mutant plants. The same results were obtained in crosses of nzz-2 with fil-1, fil-2, and fil-3 (data not shown).

Figure 5.

Phenotype of nzz-2 fil-5 double mutant plants. Scanning electron micrographs of wild-type (A, E, I), nzz-2 (B, F, J), fil-5 (C, G, K1, K2), and nzz-2 fil-5 (D, H, L1, L2) flowers and floral organs. A to D, Flowers around stage 13 are shown. Ovules at stage 2-III (E), 2-IV (G), and around stage 2-II/III (F and H) are shown. I, J, K1, K2, and L2, Anthers of flowers at anthesis (around stage 13) and a filamentous stamen at stage 10 (L1) are shown. Flowers of homozygous fil-5 (C) plants are affected in organ number, positioning of the floral organs, and in floral organ identity determination (Chen et al., 1999). Flowers of nzz-2 fil-5 double mutants (D) are almost indistinguishable from fil-5 (C) single mutants except for the anthers. Anthers of fil-5 single mutants either arrest as radial filamentous stamens (K1) but occasionally develop one or two fertile locules (K2). Anthers of nzz-2 fil-5 double mutants never develop locules (compare K2 versus L2) and resemble anthers of nzz-2 (J). Ovules of fil-5 mutants develop normally (compare E versus G). Ovules of nzz-2 fil-5 double mutants (H) resemble nzz-2 single mutant ovules (F) since they fail to develop a nucellus of normal size. In addition, in both nzz-2 and nzz-2 fil-5 ovules, the outer integument initiates prior to the inner integument and initiation of the outer integument occurs distally shifted. ca, carpel; st, stamen; fst, filamentous stamen; ast, arrested stamen; pe, petal; nu, nucellus; dt, distal tip; ch, chalaza; fu, funiculus; ii, inner integument; oi, outer integument; ad, adaxial; ab, abaxial. Arrowhead: A sepal has been removed to give view on the inner organs.

FIL is a promoter of abaxial identity; therefore, we examined whether anthers of nzz-2 mutants show any adaxial-abaxial polarity defects. High-resolution scanning electron micrograph images of the adaxial surface of nzz-2 anthers show irregularly shaped uniformly sized cells, which typically cover the adaxial anther surface of wild-type stamens (Fig. 4O, upper arrow). The lower arrow points to the gradual change toward files of more elongate cells typical for stamen filaments, indicating some variability in nzz-2 concerning the anther-filament transition along the proximal-distal axis (see also Fig. 5J). Stomata (Fig. 4P, asterisk) and irregularly shaped cells (Fig. 4P, arrow) are present on the abaxial surface of nzz-2 anthers, as they are present on the abaxial surface of wild-type anthers (Bowman, 1994). FIL expression in nzz-2 mutant stamen primordia was restricted to the abaxial region and reflected FIL expression pattern as seen in wild-type stamens (Fig. 4, M versus E and F). These data indicate that adaxial-abaxial polarity is maintained in nzz-2 mutant anthers. Regionally restricted accumulation of FIL transcript can also be detected in stamens of fil-5 mutant flowers, indicating that polarity is maintained in some stamens of fil-5 mutant flowers (Fig. 4N).

FIL encodes a putative transcriptional regulator, which has been shown to bind DNA in a nonsequence-specific manner (Sawa et al., 1999b; Siegfried et al., 1999; Kanaya et al., 2002). We tested whether NZZ expression pattern is dependent on FIL function. NZZ expression in fil-5 was detectable in lateral regions of stamen primordia (Fig. 4K), and spatially restricted NZZ expression was also detected in filamentous stamens and petals of fil-5 mutants (Fig. 4K). The distribution of NZZ transcript in nzz-2 fil-5 double mutant flowers resembled the pattern seen in nzz-2, fil-5, and wild-type flowers (Fig. 4, L versus I–K). Thus, NZZ expression in flowers seems to be independent of FIL function, except for the ovules, where we found some deviation from the expected NZZ expression pattern in nzz-2 fil-5 double mutants (Fig. 4, T1 and T2). NZZ expression was normal in nzz-2 (Fig. 4, Q versus R; Balasubramanian and Schneitz, 2000) and fil-5 ovules (Fig. 4, Q versus S1 and S2), respectively. In two independent in situ hybridization experiments, NZZ expression appeared up-regulated in the distal tip region of nzz-2 fil-5 ovules when compared to the level detected in the integuments (Fig. 4, T1 and T2). This result suggests that FIL might have a function during ovule development. Thus, both FIL and NZZ seem to be functionally redundant in preventing the accumulation of high levels of NZZ transcript in the nucellus. Although we failed to detect FIL mRNA in ovules, FIL might be expressed at very low levels in the ovule below the detection limit of in situ hybridization analysis. Alternatively, FIL could have some non-cell autonomous effects on ovule development.

FIL function might be masked by the action of its closest homologs in some floral organs, including ovules, but effects of its loss might be detectable as changes seen in the gene expression of some marker genes. Therefore, we examined the expression of INO, as well as two additional markers, PHABULOSA (PHB) and REVOLUTA (REV), in various floral organs of nzz-2, fil-5, and nzz-2 fil-5 double mutants. The latter two are both members of the class III homeodomain Leu-zipper (HD-ZIP) gene family (Zhong and Ye, 1999; Ratcliffe et al., 2000; McConnell et al., 2001; Otsuga et al., 2001). REV and PHB exhibit similar expression patterns in meristems, in the vasculature, and in adaxial domains of lateral organs (McConnell et al., 2001; Otsuga et al., 2001). FIL expression is lost in leaves of homozygous phb-1d gain-of-function mutants (Siegfried et al., 1999). Furthermore, it is hypothesized that adaxial and abaxial domains are established in lateral organs by mutual exclusion of the action of class III HD-ZIP, as well as the YABBY/KANADI gene families from abaxial and adaxial domains, respectively (Eshed et al., 2001; Kerstetter et al., 2001; McConnell et al., 2001; Bowman et al., 2002). A detailed description of the PHB expression pattern in ovules and the influence of NZZ function on PHB transcript accumulation is presented elsewhere (Sieber et al., 2004). Briefly, PHB mRNA in the ovule is restricted to the adaxial domain of the emerging ovule primordium (Fig. 6I, arrow) and becomes restricted to the inner integument (Fig. 6E). In ovules of nzz-2 fil-5 double mutant plants (Fig. 6H), ectopic PHB expression can be detected in the distal tip region, as this is the case in nzz-2 single mutants (Fig. 6F). In fil-5 mutant ovules (Fig. 6G), as in wild-type ovules (Fig. 6E), PHB expression is strongest in the inner integument and PHB expression can also be detected in the developing vascular cylinder. During vegetative development, PHB mRNA accumulates in the adaxial domain of leaf primordia (McConnell et al., 2001). In flowers, PHB expression in petals is stronger in the adaxial domain (Fig. 6I). In stamen primordia most PHB transcript is restricted to two lateral regions in the adaxial domain (Fig. 6I, inset, arrowhead). In nzz-2 mutants (Fig. 6J), PHB expression mimics the expression pattern seen in wild-type flowers (I versus J), with an obvious deviation from the normal pattern only detectable in the ovules after stage 1-II/2-I (Fig. 6F). In flowers of nzz-2 fil-5 double mutant plants (Fig. 6, K and L), elevated PHB transcript levels can be detected in distinct regions of petals and stamens, indicating that adaxial identity is maintained to some extent in these organs. Ovules around stage 2-I display the same ectopic PHB expression in the distal tip as seen in nzz-2 single mutant ovules (Fig. 6, L and H versus F). In wild-type ovules, REV is expressed throughout the ovule primordium. Strongest REV transcript accumulation can be seen in both integuments (Fig. 6M). In nzz-2 (Fig. 6N) and nzz-2 fil-5 ovules (Fig. 6P), REV mRNA accumulates in the distal tip region to levels comparable to the integuments, whereas in fil-5 ovules, the distribution of REV transcript matches the pattern seen in wild type (Fig. 6, O versus M). Thus, except for the expression in the ovules, both REV and PHB mRNA expression is unaltered in floral organs of nzz-2 and nzz-2 fil-5 when compared to wild type. INO expression is restricted to the abaxial cell file of the outer integument around stage 2-III. INO expression in fil-5 reflects its wild-type expression pattern, whereas INO transcript in ovules of nzz-2 fil-5 double mutants accumulates in the same pattern as INO mRNA in nzz-2 single mutants (Fig. 6, A–D and Q–T). Thus, INO, PHB, and REV all behave similarly in respect to loss of NZZ and FIL functions in the flower, respectively.

The expression pattern of the marker genes that we analyzed, namely INO, PHB, and REV, together with the morphological analysis of the nzz-2 fil-5 double mutant, did not reveal any novel findings in respect to processes that might be regulated by NZZ and FIL together. However, at this point it is interesting to notice that with REV and PHB, two members of the class III HD-ZIP gene family are prevented from being expressed at high levels in the distal tip region of nzz-2 mutant ovules. To our knowledge, NZZ is thus far the only known factor capable of regulating the spatial distribution of class III HD-ZIP transcripts, except for PHB itself, and likely the microRNAs miR165/166 (McConnell et al., 2001; Emery et al., 2003; Tang et al., 2003; Juarez et al., 2004; Kidner and Martienssen, 2004).

Taken together, NZZ and FIL may also interact in vivo, as NZZ and FIL are for example expressed in early stamens and necessary for stamen development (Chen et al., 1999; Sawa et al., 1999a, 1998b; Schiefthaler et al., 1999; Siegfried et al., 1999; Yang et al., 1999). In addition, FIL might also have a function in ovule development where it might act redundantly with NZZ, thereby acting negatively on NZZ transcript levels in the nucellus. However, our genetic analysis of NZZ and FIL interactions remains inconclusive, and we cannot rule out that the observed NZZ-FIL interaction is not occurring in vivo. Alternatively, a genetic analysis could be complicated by redundancy, as there is extensive genetic redundancy among the YABBY genes (Siegfried et al., 1999; Kumaran et al., 2002). In addition, four more homologs of NZZ (the homology being restricted to the polar-charged domain) exist in the Arabidopsis genome (P. Sieber, S. Balasubramanian, D. Chevalier, and K. Schneitz, unpublished data).

CONCLUSION

In this study, we show that NZZ can physically interact with three YABBY proteins, FIL, YAB3, and INO, in vitro. Our in vitro binding studies showed that the NZZ protein recognized both the zinc-finger domain as well as the C terminus of INO, including the HMG-like domain. The zinc-finger and the HMG-like domains are conserved among the YABBY proteins, and in the case of FIL the YABBY domain was necessary for in vitro DNA binding (Kanaya et al., 2001, 2002). The polar-charged domain of NZZ was sufficient for binding to INO protein. Thus, the polar-charged domain might recognize YABBY proteins in general.

The Repression of INO by NZZ Appears To Be Mediated by Direct Protein-Protein Interactions

Extensive genetic evidence indicates that NZZ coordinates development of the proximal-distal and adaxial-abaxial axes through the temporal and spatial repression of INO allowing proximal-distal axis formation before the initiation of the adaxial-abaxial axis and outer integument development in the chalaza (Balasubramanian and Schneitz, 2000, 2002). First, double mutant analysis showed ino-2 to be epistatic to nzz-2. Second, in nzz-2 mutant ovules, INO expression comes on precociously, correlating with premature formation of the outer integument. Third, NZZ and INO are coexpressed. Fourth, NZZ was shown to be involved in a positive feedback regulatory loop of INO. Fifth, NZZ, together with ATS, is a spatial regulator of INO transcription as well. Taken together, the genetic data strongly indicate that NZZ and INO act in the same pathway and that NZZ is a repressor of INO (Balasubramanian and Schneitz, 2000, 2002). The data presented in this study corroborate and at the same time extend this model, indicating that the repression of INO by NZZ involves direct interaction between the two proteins. The new model suggests that NZZ is an inhibitory element of the INO autoregulatory loop. Upon initiation of INO transcription, NZZ immediately binds freshly translated INO, thus rendering INO inactive. Diminishing functional INO levels attenuate the feedback loop until the repression is overcome by some unknown mechanism. There is evidence that the regulation of INO by SUP could also take place at the protein level (Meister et al., 2002). At present, it is unclear how the NZZ-INO interaction relates to the role of NZZ as a spatial repressor of INO, which NZZ performs functionally redundant with ATS or to the repression mechanism involving SUP function (Balasubramanian and Schneitz, 2002; Meister et al., 2002). One future challenge resides in the further elucidation of how these mechanisms relate to each other. Another challenge will be to resolve the basis of the proposed antirepression mechanism involved in the temporal NZZ-INO interaction.

MATERIALS AND METHODS

Plant Materials

Arabidopsis (L.) Heynh. var. Landsberg (erecta mutant) was used as wild-type accession. For crosses and in situ hybridization analysis, nzz-2, fil-1, fil-2, fil-3, and fil-5 were used (a kind gift of Garry Drews and John Bowman; Chen et al., 1999; Schiefthaler et al., 1999; Siegfried et al., 1999; Villanueva et al., 1999).

Plasmid Constructs

Standard molecular biological procedures were performed as described (Sambrook and Russell, 2001). An overview of the constructs used is given in Table I.

Yeast Two-Hybrid

The plasmid pHS13#4 (M. Petrascheck and A. Barberis, unpublished data) was digested with BamHI/NheI, and the insert was eliminated. Full-length NZZ cDNA was amplified from the NZZ cDNA (Schiefthaler et al., 1999) in a PCR using the primer combination PS18/PS21 (see Table II for details on primer sequences). The digested PCR fragment was then ligated into the BamHI/NheI-digested pHS13#4 plasmid to generate τ-NZZ. The truncated versions, ΔN^1–227 as well as ΔN^168–314, which represent the N terminus and the C terminus of the NZZ protein, have been produced analogously using primer combinations PS42/PS21 and PS18/PS35, respectively.

Table II.

Primer list

Primer	Sequence
PS18	CCGGATCCATGGCGACTTCTCTCTTCTTC
PS21	CCGCTAGCTTAAAGCTTCAAGGACAAATCAATGG
PS35	CCGCTAGCCATAAAGGAAACCTTGGCTCCT
PS42	CCGGATCCTCCAATAATCGCTGTGACACTTGC
PS47	TCCGGATCCCCCAACATGACGACAACACTC
PS48	AATGAATTCTTACTCAAATGGAGATTTTCC
PS67	TCATCGGAAGAGAGTAG
PS85	CGGGATCCAATGTCTATGTCGTCTATGTCCTCC
PS86	GGGGTACCCTTTAATAAGGAGTCACACCAACG
PS89	CGGGATCCCTTTAATAAGGAGTCACACCAACG
PS93	CGGAATTCATGTCTATGTCGTCTATGTCCTCC
PS97	CCGAATTCTTAAAGCTTCAAGGACAAATCAATGG
PS113	CGGGATCCATGTCGAGCATGTCCATG
PS114	CGGAATTCTAGTTATGGGCCACCCCAA
PS121	GGGGTACCAAAGCTATGGAGCAAAAGCTC
PS122	CGGAATTCTTAGGGTACCGAATTCAAGACCTC
PS123	GGGGTACCATAAGGAGTCACACCAACGTTAGC
PS150	GATCTGGCACGAG
PS151	CTCGTGCCA-5′phosphorylation
PS168	CGGGATCCAAGAGTCGTGGTCGGAAACC
PS169	CGGGATCCGAAAAGAAGCAACTCGCCGCC
PS170	CGGAATTCTTAGGTCTCAACAAAACCCCATGG
PS171	CGGAATTCTTAGGCGGCGAGTTGCTTCTTTTC
PS172	CGGGATCCATGACAAAGCTCCCCAACATG
PS173	CGGGATCCCTCTCTGTCAATTTGATGAAGGC
PS174	CGGAATTCTTACTCAAATGGAGATTTTCCC
PS175	CGGAATTCTTAGGCCCAATTTTTGGCAGCTAAG
PS176	CGGAATTCTTAAGCTCTTTGTCGCTTCTCAG
PS177	CGGAATTCTTAGACAGAGAGAAGGCTTGTGC
PS178	CGGGATCCGCTCCTTCAGCTTACAATTGC

GST-Fusion Proteins for In Vitro Pull-Down Assays

The plasmid NZZ was kindly provided by S. Balasubramanian. Truncated versions of NZZ were generated using primer combinations PS18/PS170, PS169/PS170, PS168/PS170, and PS18/PS171. The BamHI/EcoRI-digested PCR fragments were cloned in frame to the GST coding sequence of the BamHI/EcoRI-digested pGEX-4T-2 vector (Amersham Pharmacia, Uppsala) giving rise to ΔN^1–148, ΔN^78–148, Δ N^41–148, and ΔN^1–84, respectively. The plasmid pGEX-4T-2 was used to express GST protein, which served as a negative control in the pull-down experiments.

Xpress-Tagged Proteins for In Vitro Pull-Down Assays

The Xpress-tagged proteins FIL, FIL-myc, YAB3, and INO were generated by ligating PCR-amplified and digested cDNAs as translational fusions to the Xpress-tag of pRSET (Invitrogen, Carlsbad, CA). FIL was generated from F263, using primer combination PS85/PS86, and cloned into BamHI/KpnI-digested pRSET. Full-length YAB3 cDNA was amplified from the Gal4pDBD-cDNA library using PCR primers PS113/PS114. The enzyme combination BamHI/EcoRI was used for cloning YAB3 into pRSET to generate YAB3. INO cDNA was amplified from plasmid pRJM23 (kindly provided by Chuck Gasser) using primers PS47/PS48 and cloned into pRSET using BamHI/EcoRI to generate INO. Truncated versions of INO were constructed analogously with the primer combinations PS172/PS177 to generate ΔI^1–59, PS172/PS176 for ΔI^1–138, PS172/PS175 for ΔI^1–174, PS173/PS176 for ΔI^57–138, PS178/ PS175 for ΔI^138–174, and PS178/PS174 to produce ΔI^138–231, respectively. In order to construct an Xpress-tagged FIL with a cMYC tag fused to its C terminus, we amplified the full-length FIL cDNA from the plasmid F263, using primers PS85/PS123. Six cMYC repeats were amplified from the plasmid pGEM::6cMYC (kindly provided by Nicolas Baumberger and Beat Keller) using PCR primers PS121/PS122. We digested the FIL cDNA with BamHI/Asp718 and ligated the fragment to the Asp718/EcoRI-digested 6-cMYC PCR product. Both fragments were ligated to pRSET, which had been cut with BamHI/EcoRI.

In Situ Hybridization

Full-length cDNA of FIL was amplified from plasmid F263 using the primer pair PS93/PS89. The PCR fragment was digested with BamHI/EcoRI and ligated into the BamHI/EcoRI-digested plasmid pSK+ and pKS+ (Stratagene, La Jolla, CA) to generate pSK-FIL and pKS-FIL, respectively. From the clone pSK-FIL, a HincII fragment of approximately 0.3 kb was excised, and the vector was religated. The resulting pSK-ΔFIL does not contain the YABBY coding sequence (Siegfried et al., 1999). Full-length NZZ cDNA was amplified using the primers PS18/PS97 and ligated into BamHI/EcoRI-digested pSK+ (Stratagene) to generate pSK-NZZ. INO coding sequence was amplified from the plasmid pRJM23 (Villanueva et al., 1999), using the primers PS172 and PS174. After digesting with BamHI and EcoRI, full-length INO coding sequence was ligated into BamHI/EcoRI-digested pSK+ (Stratagene) to generate pSK-INO.

Pol III Yeast Two-Hybrid System

An RNA polymerase III-based yeast two-hybrid system was used for the screen (Petrascheck et al., 2001). The SNR6 knockout strain MPy15 carries a temperature-sensitive survival construct that allows the strain to grow at 25°C. A second plasmid carries the UASG-SNR6 reporter gene. Activation of the reporter, caused by the interaction of the fusion protein τ-NZZ with any chimeric Gal4pDBD-Y protein, leads to expression of the wild-type copy of the U6 snRNA. Thus, expression of SNR6 suppresses the temperature-sensitive phenotype and permits the yeast cells to grow at 37°C (Marsolier and Sentenac, 1999; Petrascheck et al., 2001).

Construction of an Arabidopsis Gal4pDBD-cDNA Library

Polyadenylated mRNA was isolated from Arabidopsis (ecotype Columbia) inflorescences containing flowers up to stage 13 (Smyth et al., 1990), using oligo(dT)₂₅-Dynabeads according to the manufacturer's protocol (Dynal, Oslo). Poly(A⁺) mRNA, together with marathon cDNA synthesis primer (CLONTECH, Palo Alto, CA) and moloney murine leukemia virus reverse transcriptase, were used for the first-strand synthesis, according to the manufacturer's recommendations (Stratagene). The cDNA termini were blunted after the second-strand synthesis, and a BglII-adaptor was ligated to the cDNA. For the adaptor, equimolar amounts of the two oligonucleotides PS150 and PS151 were annealed to generate BglII compatible 5′ overhangs. In a next step, the BglII ends of the cDNAs were phosphorylated using polynucleotide kinase, and the cDNAs were digested subsequently with NotI. This generated cDNAs with BglII and NotI compatible ends, respectively. After size fractionation using SizeSep400 spun columns (Amersham Pharmacia), the cDNAs were ligated into the BglII- and NotI-digested plasmid pPC97 to generate Gal4pDBD-cDNA fusions. The cDNA library in the pPC97 plasmid was then transformed into the XL2-Blue MRF′ ultracompetent Escherichia coli strain (Stratagene) and selected on Luria-Bertani medium containing 100 μg mL⁻¹ ampicillin. Approximately 9 × 10⁵ colonies were pooled and DNA was isolated, using maxiprep columns, according to the manufacturer's protocol (Genomed, Lohne, Germany).

Yeast Two-Hybrid Screen

The screening strain Mpy15 transformed with plasmid τ-NZZ was cotransformed with the Gal4pDBD-cDNA library. Standard yeast protocols were used with modifications (Petrascheck et al., 2001). Candidate Gal4pDBD-cDNA plasmids were retransformed into the screening strain as well as into MPy15 in order to confirm bait-dependent activation of the reporter gene. The plasmids were sequenced with primer PS67 (see Table II).

Expression of Recombinant Proteins

Recombinant proteins were expressed in BL21 E. coli cells (Invitrogen). The bacterial pellet was resuspended in phosphate-buffered saline buffer (137 mm NaCl, 2.7 mm KCl, 10 mm Na₂PO₄, 2 mm KH₂PO₄, pH 7.4) containing protease inhibitor (COMPLETE; Roche Diagnostics, Rotkreuz, Switzerland) and lysed on ice for 30 min with lysozyme (10 μg mL⁻¹ final concentration; Roche Diagnostics). The cell lysate was briefly sonicated on ice and centrifuged twice at 4°C with 27,000g for 30 min. The supernatant containing the cleared lysate was frozen in aliquots and kept at −80°C until used.

Xpress-tagged (Invitrogen) fusion proteins were expressed using the commercially available TNT quick coupled in vitro transcription/translation system (Promega, Madison, WI) and labeled with [³⁵S]Met (Amersham Pharmacia) according to the manufacturer's recommendations (Promega).

In Vitro Pull-Down Assays

Cleared lysate of BL21 cells expressing GST, NZZ, and truncated versions of NZZ was incubated with Xpress-tagged fusion protein, respectively, and with glutathione sepharose (Amersham Pharmacia) in PBS buffer for 2 h at 4°C on an overhead shaker. After washes, bound proteins were eluted in elution buffer (10 mm reduced glutathione in 50 mm Tris-HCl, pH 8.5) for 30 min on ice.

The eluted samples were mixed 1:1 (v/v) with 2× SDS gel-loading buffer and size separated using SDS-PAGE as described (Sambrook and Russell, 2001). For detecting [³⁵S]Met-labeled proteins, the gel was dried on a Whatman filter paper and exposed to a BIOMAX MR film (Kodak, Rochester, NY) using TRANSCREEN LE intensifying screen (Kodak). Xpress-tagged and cMYC-tagged proteins were detected as described (Spillane et al., 2000; Baumberger et al., 2001). We used the ECL plus western blotting detection system (Amersham Pharmacia) to monitor peroxidase activity.

In Situ Hybridization Experiments

The in situ hybridization experiments were performed as described (Balasubramanian and Schneitz, 2002). DIG-UTP-labeled sense and antisense NZZ RNA probes were in vitro transcribed from BamHI- and EcoRI-linearized pSK-NZZ using T7 and T3 polymerase (Roche Diagnostics), respectively. The FIL antisense probe lacking the YABBY-domain coding sequence was obtained by transcribing the XbaI-digested pSK-ΔFIL with T7 RNA polymerase. Plasmid pKS-FIL was linearized with EcoRV and transcribed with T7 RNA polymerase to produce the full-length FIL sense probe. Sense and antisense INO probes were generated from EcoRI- and BamHI-linearized pSK-INO, using T3 and T7 RNA polymerase, respectively. The plasmid pSK-REV#203 was linearized with EcoRV and SacI and subsequently transcribed with T3 and T7 RNA polymerase to generate antisense and sense REV probe, respectively. The probe contained the identical REV fragment as described (Otsuga et al., 2001). Antisense and sense PHB probes were made as described (McConnell et al., 2001).

Upon request, all novel materials described in this publication will be made available in a timely manner for noncommercial research purposes, subject to the requisite permission from any third party owners of all or parts of the material. Obtaining any permissions will be the responsibility of the requestor.