The peach (Prunus persica L. Batsch) genome harbours 10 KNOX genes, which are differentially expressed in stem development, and the class 1 KNOPE1 regulates elongation and lignification during primary growth.

The KNOTTED-like (KNOX) genes encode homeodomain transcription factors and regulate several processes of plant organ development. The peach (Prunus persica L. Batsch) genome was found to contain 10 KNOX members (KNOPE genes); six of them were experimentally located on the Prunus reference map and the class 1 KNOPE1 was found to link to a quantitative trait locus (QTL) for the internode length in the peach×Ferganensis population. All the KNOPE genes were differentially transcribed in the internodes of growing shoots; the KNOPE1 mRNA abundance decreased progressively from primary (elongation) to secondary growth (radial expansion). During primary growth, the KNOPE1 mRNA was localized in the cortex and in the procambium/metaphloem zones, whereas it was undetected in incipient phloem and xylem fibres. KNOPE1 overexpression in the Arabidopsis bp4 loss-of-function background (35S:KNOPE1/bp genotype) restored the rachis length, suggesting, together with the QTL association, a role for KNOPE1 in peach shoot elongation. Several lignin biosynthesis genes were up-regulated in the bp4 internodes but repressed in the 35S:KNOPE1/bp lines similarly to the wild type. Moreover, the lignin deposition pattern of the 35S:KNOPE1/bp and the wild-type internodes were the same. The KNOPE1 protein was found to recognize in vitro one of the typical KNOX DNA-binding sites that recurred in peach and Arabidopsis lignin genes. KNOPE1 expression was inversely correlated with that of lignin genes and lignin deposition along the peach shoot stems and was down-regulated in lignifying vascular tissues. These data strongly support that KNOPE1 prevents cell lignification by repressing lignin genes during peach stem primary growth.

Introduction

Plant tree stems grow in length (primary growth; PG) and width (secondary growth; SG). During PG, the tunica and corpus layers of the shoot apical meristem (SAM) produce the protoderm, ground meristem, and procambium which, as they elongate and differentiate, create the dermal, ground, and vascular tissues, respectively (Evert and Eichhorn, 2007). In peach trees, the procambium forms the primary xylem (inward) and phloem (outward) within the eustele typically structured by a pith surrounded by bundles and parenchyma cell rays (Schneider, 1945). During PG, lignin deposition marks the latest stages of cell elongation and is observed to start in xylem vessels, tracheids, and fibres (Evert and Eichhorn, 2007). The PG to SG transition begins when the fascicular cambium and the de novo interfascicular meristem form a continuous vascular cambium, which produces secondary xylem and phloem and allows the radial growth (Spicer and Groover, 2010).

There is evidence that the homeodomain (HD) KNOTTED-like homeobox transcription factors (KNOX), necessary for the SAM functioning, play roles in tree stem development (Groover et al., 2006; Du et al., 2009). The KNOX gene classification for simple-leafed species includes these criteria: the class 1 proteins have an HD identity >73% referred to that of maize Kn1, and the genes (KNOX I) are expressed in meristems and down-regulated in leaf; the class 2 genes (KNOX II) have an intron within the ELK domain and widespread transcription; the mini KNOX genes encode HD-less proteins (Kerstetter et al., 1994; Mukherjee et al., 2009; Hay and Tsiantis, 2010). KNOX I genes are required for stem cell maintenance and inhibit cell differentiation during organogenesis (Barton and Poethig, 1993; Long and Barton, 1998; Scofield and Murray, 2006); the different regulation of KNOX I in species with simple and compound leaves subtends the diversity of foliar shape (Hay and Tsiantis, 2009). The widely studied Arabidopsis KNOX I class comprises the SHOOTMERISTEMLESS (STM), KNAT1 (or BREVIPEDICELLUS; BP), KNAT2, and KNAT6 components. STM is essential for SAM formation and maintenance (Scofield and Murray, 2006; Hay and Tsiantis, 2010). BP contributes redundantly with STM to maintain the proper function of SAM (Byrne et al., 2002) and takes part in controlling inflorescence stem (rachis) development (Byrne et al., 2002; Douglas et al., 2002; Venglat et al., 2002; Mele et al., 2003). Similarly to BP, KNAT6 contributes to SAM function and inflorescence development (Ragni et al., 2008). Finally, KNAT2 regulates flower patterning acting in the inner whorls (Yu et al., 2009). Several factors confine class 1 KNOX expression domains (Hay and Tsiantis, 2010) in Arabidopsis. Among these, the ASYMMETRIC LEAVES 1 (AS1) and AS2 repress BP, KNAT2, and KNAT6 transcription and form a complex able to bind to BP and KNAT2 promoters (Guo et al., 2008). The KNOX functions are mediated by interactions with hormones (Hay et al., 2004; Hay and Tsiantis, 2009) and specific/combinatorial dimerization with BELL (Bellaoui et al., 2001; Hackbusch et al., 2005; Rutjens et al., 2009), NAC (Zhong et al., 2008), MYB (Bhargava et al., 2010), OVATE (Li et al., 2011), LOB (Yordanov et al., 2010), and BOP transcription factors (Ha et al., 2003; Khan et al., 2011). Studies on Arabidopsis KNOX II functions (KNAT3, KNAT4, KNAT5, and KNAT7) have been limited (Truernit et al., 2006; Pagnussat et al., 2007); the KNAT7 fine regulation of secondary wall and lignin synthesis has been established (Zhong and Ye, 2007; Zhong et al., 2008).

Knowledge on KNOX I roles during stem PG and SG has been focused on Arabidopsis and aspen, respectively (Sanchez et al., 2011). The thale cress bp loss-of-function mutants showed a shortened rachis, vascular anomalies, and increased lignin content (Venglat et al., 2002; Mele et al., 2003; Douglas and Riggs, 2005); BP overexpression caused a delay in lignin deposition, and the protein bound to lignin gene promoters (Mele et al., 2003). Aspen lines overexpressing ARBORKNOX1 (ARK1), an STM orthologue, showed delayed PG–SG transition and inhibition of secondary vascular cell type differentiation (Groover et al., 2006). The overexpression of ARBORKNOX2 (ARK2), a BP-like gene, caused vascular cambium zone widening due to a delayed differentiation of xylem initials. Conversely, ARK2 silencing accelerated the PG–SG transition and the development of secondary xylem and phloem fibres (Du et al., 2009). Regarding lignin synthesis, ARK1 and ARK2 act as inducer and repressor, respectively (Groover et al., 2006; Du et al., 2009).

KNOX genes can control agro-industrial traits (e.g. trunk vigour, branch length, lignin content, and vegetative habitus), which impact on the productivity of forest and fruit arboriculture (Du and Groover, 2010). So far, KNOX genes from Rosaceae fruit trees have only been characterized in apple and peach (Watillon et al., 1997; Testone et al., 2008, 2009). The latter is of economic relevance worldwide and a model for other Prunus spp. because of the sequenced genome (International Peach Genome Initiative, www.rosaceae.org/peach/genome) and the availability of several physical and marker-saturated genetic maps (Jung et al., 2008).

In this work, 10 KNOX genes from the peach genome were characterized at the structural and expression levels, and classified. The class 1 KNOPE1 was established to map to a quantitative trait locus (QTL) for the peach internode length, which prompted the KNOPE1 function during stem primary growth to be addressed. The dynamic localization in the cortex and vascular system elements from early to late PG stages sustained the multiple roles of the gene in internode cell differentiation. Using the Arabidopsis brevipedicellus model, it was observed that KNOPE1 ectopic expression rescued the rachis internode length, lignin deposition, and transcription of lignin biosynthesis genes (LBGs), implying the gene’s involvement in elongation and lignification. The inverse correlation between KNOPE1 transcription and stem lignification/LBG expression along peach shoot stems, together with the protein binding to the typical KNOX DNA motif, strongly supported the repressive role of KNOPE1 in lignification.

Materials and methods

Plant materials and growth conditions

The trees (F1S1–18) of cultivar ‘Chiripa’ (Okie, 1998) were grown in the IBBA-CNR fields and derived from the self-pollination of clone 18 (Testone et al., 2008). Several environmental parameters were registered (2005–2007) and the mean values did not vary significantly (not shown). Plant samples were collected, frozen, and stored at –80 °C. As for Arabidopsis, the 35S:KNOPE1 genetic construct (Testone et al., 2008) was vacuum-infiltrated (Bechtold and Pelletier, 1998) into the bp4 mutant of ‘Landsberg’ erecta (Ler). Several 35S:KNOPE1 lines in the bp4 background were produced and those with severe leaf phenotypes (e.g. dramatic fringing) were dicscarded due to the occurrence of several pleiotropic effects such as rachis stunting and anomalies in flower/fruit development. The transgenic lines 1, 2, and 7 with moderate serrations of leaves and recovering epinasty and rachis length were selected, brought to the homozygous stage (T3 generation), and grown at 22–25 °C, 16/8h light/dark, 120 µmol m–2 s–1 photosynthetically ative radiation (PAR).

Isolation and sequence analysis of KNOPE genes in ‘Chiripa’

The KNOPE2, KNOPE6, STMlike1, STMlike2, and KNOPE4 genes (Table 1; Supplementary Table S1 at JXB online) were cloned through degenerate primer strategies to amplify sequences in the MEINOX and HD of class 1 and 2 genes, followed by 5′ and 3′ rapid amplification of cDNA ends (RACEs) to produce full-length genes, and further checked by whole gene sequencing. The herbaceous stem RNA was isolated (Giannino et al., 2000), DNase treated (RQ1, Promega), and 1 µg was reverse-transcribed at 55 °C by SuperscriptIII (Life Technologies). Primer pairs were: KnIFw [5′-GA(C/T) (C/G)A(A/G)TT(C/T)ATGGA(A/G) (A/G)C(A/G/C/T)TA-3′] and KnIBw [5′-AT(A/G)AACCA(A/G)TT(A/G)TT(A/T/G)AT(C/T)TG-3′] for KNOX I, and KnIIFw (5′-CCATGGAAGCAGTGATGGC-3′) and KnIIBw (5′-CTTCCTCAGTTGGATATGGCC-3′) for KNOX II. The ‘Lovell’ sequence allowed primers to be designed to clone ‘Chiripa’ KNOPE2.1, KNOPE7, and KNOPMEM fully. The PCR conditions of all experiments were: cDNA (200ng) or genomic DNA (gDNA; 500ng), and 0.2 µM primers, 0.5mM dNTPs, 2.5U of Taq DNA polymerase (Qiagen), 1× Taq buffer, 2.5mM MgCl2, in a total volume of 50 µl; starting at 95 °C for 5min, 35 cycles at 95 °C for 40 s, 58 °C for either 30 s (for cDNA) or 60 s (for gDNA), and 72°C for 30–90 s, and final extension at 72 °C for 5min. The products were cloned into pGEM-T easy (Promega); cDNA and gDNA sequences were aligned and intron size and locations established.

Table 1.

Features of ‘Chiripa’ KNOPE genes

Gene	GenBank accession no		Identity (%)a	KNOX-binding siteb
	gDNA	cDNA		Exon	Intron
Class 1
KNOPE1	DQ388033	DQ358050	KNAT1/BP (79%)	–	III-R
KNOPE2	GU144519	EF093491	KNAT2 (62%)	–	–
KNOPE2.1	—	JQ038131	KNAT2 (77%)	V-R	ND
KNOPE6	GU144515	GQ281773	KNAT6 (74%)	V-R	–
STMlike1	GU144516	GQ281774	STM (74%)	IV-D	–
STMlike2	GU144517	GQ281775	STM (81%)	–	–
Class 2
KNOPE3	EU910092	DQ786755	KNAT3 (74%)	–	–
KNOPE4	GU144518	EF107110	KNAT4 (76%)	III-D	II-D
KNOPE7	—	JQ038132	KNAT7 (87%)	III-D	ND
Class M
KNOPEM	—	JQ038133	KNATM (70%)	–	ND

a The Arabidopsis gene with the highest nucleotide identity score is reported based on the comparative analyses performed with cDNA sequences.

b Ordinal numbers were assigned to exons (I–VI) and introns (I–V) from ATG onwards. The chosen KNOX-binding sites were TGACAG(G/C)T, TGACAG(G/C), and GACAG(G/C)T, and the direct (D) or reverse (R) orientations are indicated.

ND, not detected.

Computational analyses

Alignments and phylogenetic trees were obtained as previously described (Testone et al., 2008). The KNOX protein accession numbers are listed in Supplementary Table S2 at JXB online. The ‘ePESTfind’ web tool was used to score PEST sites (http://emboss.bioinformatics.nl/).

Genetic mapping and QTL analysis

The KNOPE allelic fragments were amplified from almond ‘Texas’ (T) and peach ‘Earlygold’ (E) genotypes using ‘Chiripa’ primers and analysed for restriction site unequal distribution to create cleaved amplified polymorphic sequence (CAPS) markers (Table 2). The PCR parameters were the same as described previously (Testone et al., 2009). The digested products (enzyme 5U, final volume 30 µl) were separated (1.2% agarose). As for STMlike2, a ‘Texas’ allele insertion/deletion (INDEL) segregation was followed in the T×E F2 population by INDEL-flanking primers. The PCR conditions were: gDNA (10ng), PCR buffer (1×), MgCl2 (1.5mM), dNTPs (0.2mM), each primer (0.1 µM), Platinum Taq DNA Polymerase (0.5U, Life Technologies) in a final volume of 10 µl; starting at 95 ºC for 5min, 35 cycles of 95 ºC for 30 s, 57 °C for 30 s, 72 ºC for 30 s, and a final extension at 72 ºC for 8min. The PCR fragments were separated on a 3% high resolution agarose gel (Metaphor, Cambrex). The KNOPE segregation profiles (Supplementary Fig. S1 at JXB online) were included in the T×E map (Joobeur et al., 1998; Dirlewanger et al., 2004) using MAPMAKER EXP 3.0 (Lincoln et al., 1992) as described (Testone et al., 2009) with a LOD threshold ≥10 and recombination fraction <0.1.

Table 2.

KNOPE primers and polymorphisms in ‘Texas’ and ‘Earlygold’ parent lines

Genes	Primers (5'→3')	FS (bp)a		Pol. in T/Eb	REc	RF (bp)d
		T	E			T	E
Class 1
KNOPE1	Fw: CATCAAGAGGACACCAACCAG	2191	2195	SNP (G/C)	RsaI	1322, 598, 123, 82, 53, 17	1314, 476, 126, 123, 82, 53, 17
	Bw: ATTCTCCTGCTCCTCTTCAGATG
KNOPE2	Fw: ATCGGATGGAGGAAATGTACGG	1036	1034	INDEL (–/A)	BsrDI	539, 497	1034
	Bw: AGGAGGCGTTAGCAGATGAAGTG
KNOPE6	Fw: AAGTACAAGCAATACCAGTTGTG	681	678	SNP (C/T)	HpaII	301, 223, 157	458, 220
	Bw: AGTCCCAGCTGGTTTTTGGG
STMlike1	Fw: GAAGTGGGTCTCCAGAAGAGGTTG	823	803	SNP (T/G)	HincII	823	466, 337
	Bw: CTTTGGAGAAGCAAGGGGTAGG
STMlike2	Fw: GGCTCTACTTAGTTTTTGCTCTCA	156	144	INDEL (12bp/–)	–	–	–
	Bw: CCAATTGTTTATTTGCTTCTGG
Class 2
KNOPE3	Fw: GCCCAGCTGGCTCAGTCGC	1127	1113	INDEL (–/C)	HincII	903, 224	529, 360, 224
	Bw: AACTTCGGTGCTCAAGAGCC
KNOPE4	Fw: AGTGATGGCTTGCTGGGAGCTCG	963	963	SNP (T/A)	HinfI	963	711, 252
	Bw: CCATAAGTTGTGGACTTGTCAAGT

a FS, fragment sizes in bp amplified from the genomic DNA of ‘Texas’ (T) and ‘Earlygold’ (E) parents.

b Pol, polymorphism type. SNP, single nucleotide polymorphism; INDEL, insertion/deletion event.

c RE, restriction enzyme used to generate the gene-specific CAPS dominant marker.

d RF, restriction fragments.

For QTL association, we screened a population of 125 BC1 seedlings derived from the semi-dwarf recurrent P. persica parent IF7310828 (P) crossed with the standard Ferganensis (F) and the F1 hybrid (Verde et al., 2002). The KNOPE1 parental alleles harboured a single nucleotide polymorphism (SNP; A/G), found on the ‘Lovell’ KNOPE1 first exon (scaffold 1: 35 204 681bp, www.rosaceae.org/peach/genome), that allowed the design of primers for the KNOPE1- marker: Fw 5′-GGAAGCCATCAAAGCCAAGA-3′, Bw 5′-GGCAATCCATGTAAGCTTCCA-3′ (amplicon size 88bp). The SNP segregation was scored by high resolution melting (HRM) in duplicate assays for each sample using the 7500 Fast Real-Time PCR System. The PCR conditions were: gDNA (20ng), 2× MeltDoctor™ HRM Master Mix (Life Technologies), 0.3 µM each primer in a 20 µl final volume; start at 95 °C for 10min, 40 cycles at 95 °C for 15 s, and annealing–extension at 60 °C for 60 s. Melting curve analysis (95 °C to 60 °C ramping with a 0.1 °C incline per acquisition step) was performed by using the HRM Software v2.0.1 (Life Tecnologies). The KNOPE1- was located in the P×F map as previously described (Verde et al., 2005) using an improved map (unpublished) based on 125 seedlings elaborated through the Joinmap 4.0 software (Van Ooijen, 2006). The QTL was analysed by the MapQTL6 software (Van Ooijen, 2009) using internode measurements for 2 years as previously described (Verde et al., 2002). A permutation test (1000 iterations) assessed the significance threshold of the LOD score for the data.

qRT-PCR analyses

Total RNA isolation and cDNA synthesis were described above. The primers are listed in Supplementary Table S3 at JXB online. For peach, the shoots growing from a 1-year-old branch were sampled in May 2005–2007 (Fig. 4A). Those bearing 26 visible internodes (bottom-up numbering, following time ontogenesis) were selected (Fig. 4B) and subdivided into basal, proximal, distal, and apical sectors, which included internodes 1–6, 7–12, 13–18, and 19–26, respectively. The RNA derived from a pool (n = 3) of comparable internodes (Table 6). For Arabidopsis, the rachis internodes 1 and 2 of Ler, 35S:KNOPE1/bp4-1, and bp4 were sampled at 9 d after bolting. The RNA derived from a pool (n = 5) of comparable internodes was isolated by the RNeasy Plant Mini Kit (Qiagen) and reverse-transcribed as described above. The cDNA (60ng) was amplified using a 1× Quantimix Easy SYG Kit (Biotools) and 0.3 µM of each primer in a 20 µl final volume. The triplicate reaction conditions were: 95 °C for 180 s, 35 cycles at 95 °C for 30 s, 59 °C for 40 s, and 72 °C for 40 s. Primer specificity was checked by melting curve analysis (from 55 °C to 95 °C with a ramping of 0.5 °C s–1). The gene relative expression versus RPII (Tong et al., 2009) and ACTIN8 (An et al., 1996) constitutive markers, respectively, for peach and Arabidopsis, was elaborated by the Q-Gene program (Muller et al., 2002).

Fig. 4.

KNOPE genes are differentially expressed along peach stems. (A) Primary axis growth curve of shoots borne on ‘Chiripa’ adult plants after the vegetative bud-break (dpb, early March 2005–2007). Bars indicate standard deviations. (B) Growing shoots of 51.04±2.05cm bearing ~26 visible internodes were sampled at 60 dpb (white circle in A). The selected shoots were arbitrarily divided into apical, distal, proximal, and basal sectors, which encompassed internodes 1–6, 7–12, 13–18, and 19–26, respectively. Bar=4.5cm. (C and D) The expression of class 1 (C) and class 2/M (D) KNOPE genes was monitored from the apical to basal sectors. Bars indicate standard deviations. (This figure is available in colour at JXB online.)

Table 6

Features of model shoots at 60 d post vegetative bud-break

Sector name	Internode rangea	Sector length (cm)	Internode length (cm)b	Internode diameter (mm)	Target internodec
Apical	26–19	6.45±1.09	0.76±0.62	1.62±1.21	21–19
Distal	18–13	15.31±2.44	2.55±0.44	3.37±0.32	14–13
Proximal	12–7	16.50±3.30	2.75±0.64	4.02±0.21	9–8
Basal	6–1	15.73±2.98	2.62±0.59	5.31±0.24	4–3

a Progressive numbers were assigned according to internode ontogenesis.

b The mean ±SD was measured for 10 comparable shoots from distinct adult plants.

c Internodes sampled for expression analyses.

In situ hybridization and histological staining

The apical meristems and subtending stems of the shoot types (Fig. 4, Table 6) were paraffin embedded, cut into 8 µm sections, and hybridized at 55 °C (Cañas et al., 1994) with a digoxigenin-labelled antisense RNA probe spanning the KNOPE1 1303–1663bp cDNA (Testone et al., 2008). Safranin/Fast Green staining followed Johansen’s method (Johansen, 1940). Stem hand-made sections were stained with phloroglucinol-HCl [1% (w/v) in 6 N HCl] for a maximum of 1min.

Statistical analyses

Analysis of variance (ANOVA) was applied to transcript variation of KNOPE and peach lignin genes. Given a Fisher distribution ≥8, the variables were analysed by Tukey’s HSD test to assess mean separation significance (Supplementary Table S4 at JXB online). The Student’s t-test was performed for the Arabidopsis mutant versus the wild type, and significant values were accepted (P < 0.05).

KNOPE1 purification and electrophoretic mobility shift assay (EMSA)

The KNOPE1 cDNA (226–1508bp) was amplified (Platinum Pfx DNA Polymerase, Life Technologies) by Kn1.10Fw (5′-GGGATCCAAA-TGGAAGAGTACAACC-3′; the BamHI adaptor is underlined) and Kn1.5Bw (5′-GCAGTTTGCTTTGTCTTTTGG-3′), inserted into the pGEM-T easy vector (Promega) and BamHI–NotI cloned upstream of the GST (glutathione S-transferase) gene tag of the pGEX5X-1 vector (GE Healthcare). The sequence was checked and transferred into Escherichia coli BL21(DE3) pLysS. Protein expression was induced for 6h by isopropyl-β-d-thiogalactopyranoside (IPTG) treatment (final concentration 0.1mM in 500ml of LB broth), which started at 0.8 OD600. Cells were suspended at 4 °C in 1× phosphate-buffered saline (PBS), 1% Triton X-100, 2 µg ml–1 aprotin, 2 µg ml–1 pepstatin, 2 µg ml–1 leupeptin, 1mM phenylmethylsulphonyl fluoride (PMSF) at pH 7.2. The recombinant KNOPE1–GST (44.4 + 27.5kDa) was purified by affinity tag GST columns (Life Technologies) using 5 µl of sonicated raw cellular lysate, and the expected mol. wt of ~71.9kDa was confirmed by Coomassie blue staining. The KNOPE1 (1 µg) was incubated with: 40fmol of radiolabelled (50 000 cpm) 60-mer wild-type probe and 2 µg of poly[d(I–C)] in 1× EMSA buffer [10mM NaCl, 0.5mM dithiothreitol (DTT), 0.5mM EDTA, 1mM MgCl2, 5% glycerol, 0.15% NP-40] for 2h at 4 °C in a 20 µl reaction volume. An identical probe to the peach CCoAOMT promoter stretch (14 438 440–14 441 439 bases, scaffold 8), from the GDR database (Jung et al., 2008), was synthesized to create 5′ overhangs that were end-filled by using Klenow DNA polymerase and [α-32P]dCTP. For competition experiments, the unlabelled and mutated oligos (Fig. 10) were added to the KNOPE1–probe complex in a 10-, 50-, and 100-fold molar excess, and EMSA resolution occurred in 0.5× TBE buffer and 5% polyacrylamide gels (acrylamide: bis-acrylamide 70:1), which were dried, exposed overnight, and analysed.

Fig. 10.

The KNOPE1 protein binds to the TGACAGC 
motif. (A) The peach CCoAOMT gene was schemed into 
the promoter (black line up to 3000bp upstream of the 
arrowed ATG), exons (white boxes), and introns 
(grey boxes). Putative KNOX DNA-binding sites are 
indicated by white and black triangles in the sense 
(5'→3') or antisense orientation, respectively. The sizes of 
binding sites are in base pairs (bp). The EMSA was 
performed with a 60bp radiolabelled probe spanning 
the TGACAGC motif sited in the promoter. (B) Mutant 
(M) and wild-type (WT) cold competitor sequences. Three mismatches (underlined) were introduced in the binding 
sequence (in bold) to create a mutant competitor and assay 
the protein–promoter specificity. (C) EMSA. Radiolabelled 
probe alone (lane 1) and after incubation with KNOPE1 
protein (lane 2). The band shift in lane 2 compared with 
lane 1 indicates the KNOPE1–probe interaction. Unlabelled oligomers harbouring mutated (M, lane 3–5) or wild-type 
(WT, lane 6–8) binding sequences were used at an increasing dosage (10–100×) as competitors of the labelled 
KNOPE1–probe complex.

Results

KNOPE classification and genetic mapping

Ten KNOPE genes were cloned from the cultivar ‘Chiripa’, two (KNOPE1 and KNOPE3) previously (Testone et al., 2008, 2009) and eight in this work (five at the genomic level, KNOPE2, KNOPE6, STMlike1, STMlike2, and KNOPE4; and three at the cDNA level, KNOPE2.1, KNOPE7, and KNOPEM ). Analyses of nucleotide identity (Table 1) and structural organization (Fig. 1; Supplementary Fig. S2 at JXB online) compared with the Arabidopsis KNOX counterparts allowed the KNOPE genes to be named and grouped into six of class 1, three of class 2, and one of class M. The class 1 KNOPE1, KNOPE2, and KNOPE6 genes harboured five exons and four introns. The first and second intron fell into the KNOX1 and KNOX2 subdomains; the third and the fourth introns were sited after the KNOX2 subdomain and within the HD, respectively. In contrast, there was no intron in the KNOX2 subdomain of class 1 STMlike1 and 2. Conversely, the class 2 KNOPE3 and 4 genes contained six exons and five introns; two introns were located in the KNOX2 subdomain, the third one was in the ELK domain (a class 2-specific feature), and the fourth and fifth were placed in the HD and downstream of the sixth exon, respectively. The intron average A/T content was >64% and the GT/AG splicing consensus was strictly conserved. Interestingly, the typical KNOX DNA-binding sites were found in KNOPE1 and KNOPE4 introns and in KNOPE6 and STMlike1 exons (Table 1), suggesting that KNOPE cross-regulation might occur. KNOPE1, KNOPE2, KNOPE6, STMlike1, STMlike2, and KNOPE4 were mapped by scoring the segregation of polymorphic markers in the Prunus T×E population (Fig. 2). CAPS markers were produced for five KNOPE genes, whilst selective primers were designed for STMlike2 on an INDEL event (Table 2). The marker co-dominance was confirmed in the F1 generation, and allele segregation followed in the F2 progeny (Supplementary Fig. S1 at JXB online). Putative KNOPE–QTL associations were inferred by comparing the T×E with four QTL genetic maps which shared common markers (Table 3).

Fig. 1.

Features of KNOPE genes. The schematic diagram represents the coding sequences (open rectangles), untranslated regions (solid lines), and intron positions (filled triangles) of ‘Chiripa’ KNOPE genes and the respective Arabidopsis orthologues. Numbers indicate intron sizes in base pairs. The sequences encoding the typical KNOX domains are boxed and indicated at the bottom. In the KNOPE2 and STMlike2 genes, the introns before the ELK domain were partially sequenced and sizes were estimated by comparing PCR product sizes from cDNA and gDNA. The NCBI accession numbers of Arabidopsis KNOX sequences are: KNAT1, NM_116884; KNAT2, NM_105719; KNAT6, NM_102187; STM, NM_104916; KNAT3, NM_122431; KNAT4, NM_121144. Those of KNOPE genes are reported in Table 1.

Fig. 2.

The position of class 1 and class 2 KNOPE genes on the linkage groups (G) of the Prunus reference map.

Table 3.

KNOPE putative association with QTLs

Gene	LGa	cM	Compatible QTL	References
Class 1
KNOPE1	1	51.6	Internode length; fruit polar diameter	Verde et al. (2002); Quilot et al. (2004)
KNOPE2	1	35.7	Not found	–
KNOPE6	6	79.6	Fruit cheek diameter, weight, total sugar content, epicarp speckle	Dirlewanger et al. (1999); Etienne et al. (2002); Quilot et al. (2004)
STMlike1	3	36.4	Fruit saccharose, malic, citric, shikimic acids metabolism; fruit red overcolouring; epicarp speckle	Quilot et al. (2004)
STMlike2	4	3.5	Fruit polar diameter; fruit weight; total sugar content	Quilot et al. (2004)
Class 2
KNOPE3	1	65.4	Internode length. Leafing date.	Verde et al. (2002); Sanchez-Perez et al. (2007)
KNOPE4	7	65.6	Not found	–

a LG, linkage group in the Prunus reference map.

For the deduced proteins, the HD identities of KNOPE1, KNOPE2, KNOPE2.1, KNOPE6, STMlike1, and STMlike2 were 74–88% compared with maize Kn1, whilst those of KNOPE3, KNOPE4, and KNOPE7 were 55–58%. These values confirmed that these KNOPEs belonged to class 1 and 2, respectively. The phylogram of KNOX proteins from dicots (Supplementary Fig. S3 at JXB online) further showed that KNOPE1, KNOPE2, KNOPE2.1, KNOPE6, STMlike1, and STMlike2 fell into class 1, and KNOPE3, KNOPE4, and KNOPE7 into class 2; the mini-KNOX formed a distinct group. Regarding the N-terminal moiety upstream of the MEINOX (not shown), the intraclass identity grade (IG) was 14.3±2.1% (mean ± SD) for class 1 and 18.0±4.6% for class 2 factors. The IGs of class 1 and class 2 MEINOX were 48.9±7.6% and 60.1±5.3%, respectively. Motifs peculiar to class 1 and class 2 proteins (Fig. 3, dark and light grey, respectively) occurred within the MEINOX (Di Giacomo et al., 2008). For the stretch between the MEINOX and the ELK domain (aka GSE domain), the class 1 and 2 IGs were 28.1±9.1% and 72.0±6.9%. The region harboured PEST proteolysis signals in five KNOPEs (Fig. 3) and included class 2 unique residues (Mukherjee et al., 2009). Regarding the C-terminus, the IGs of class 1 and 2 factors were 65.2.0±5.9% and 85.7±6.3% and the KNOX characteristic residues featured (Fig. 3).

Fig. 3.

Features of the C-termini of class 1 and 2 KNOPEs. The typical KNOX domains are shown; the residues of the three α-helices are boxed; the strictly conserved stretches are in bold; and class 1- and class 2-specific residues are shaded in dark and light grey, respectively. Highly potential PEST cleavage sites (score ≥6.5) are underlined.

Finally, KNOPE transcription (Table 5) was assayed in apical tips (SAMs and surrounding leaflets), herbaceous stems, and expanded leaves (devoid of petioles) after vegetative resumption (April). The six class 1 KNOPE genes were repressed in foliar laminas, as in most class 1 KNOX genes of simple-leafed species, and were expressed at different levels in apical tips and young stems. Ubiquitous transcription of KNOPE3, KNOPE4, and KNOPE7 reflected class 2 behaviour. KNOPEM was weakly expressed in these organs consistent with the KNATM of Arabidopsis (Magnani and Hake, 2008)

Table 5.

KNOPE expression in vegetative organsa

Genes	Apical tips (MNE ±SD)	Herbaceous stems (MNE ±SD)	Mature leaves (MNE ±SD)
Class 1
KNOPE1	0.29±0.03	0.32±0.02	ND
KNOPE2	0.27±0.02	0.29±0.01	ND
KNOPE2.1	0.20±0.04	0.23±0.01	ND
KNOPE6	0.20±0.01	0.22±0.03	ND
STMlike1	0.19±0.01	0.23±0.03	ND
STMlike2	0.19±0.01	0.19±0.01	ND
Class 2
KNOPE3	0.68±0.05	0.74±0.04	0.85±0.04
KNOPE4	0.34±0.02	0.39±0.01	0.42±0.02
KNOPE7	0.42±0.03	0.49±0.03	0.45±0.04
Class M
KNOPEM	0.02±0.01	0.01±0.00	0.03±0.01

a Transcript abundances are indicated as mean normalized expression (MNE) levels with respect to the RPII reference gene.

ND, not detected after 35 amplification cycles; SD, standard deviation.

KNOPE gene expression patterns vary during stem growth

All KNOPE genes were transcribed in herbaceous stems (Table 5) and some of them might regulate shoot internode length (Table 6). To monitor the KNOPE gene expression during stem development, newly formed and comparable shoots (length, 51.04±2.05cm; number of visible internodes, 26) of adult plants were selected at 60 d (Fig. 4A, 4B) after the vegetative bud-break (dpb; occurring in early March 2005–2007). The shoot main axis was divided into apical, distal, proximal, and basal sectors (Fig. 4B, Table 6). The apical sector was assessed to be the elongating one (high variance of the mean internode length; Table 6). The remaining shoot axis was divided into ~15cm segments, bearing internodes of comparable lengths (elongation had ceased) and showing a top-down increment of the radius, due to secondary growth proceeding (Table 6). KNOPE transcription trends were graphed from the apical to basal sectors (Fig. 4C–D). Only those with significant variation (Supplementary Table S4 at JXB online) are described hereafter. For class 1 genes (Fig. 4C), KNOPE1 expression declined progressively; the KNOPE2 and KNOPE2.1 patterns consisted of parallel drop–peak–drop oscillations at the distal, proximal, and basal segments; KNOPE6 showed just a significant transcript peak at the proximal sector; and the STMlike2 mRNA level dropped at the distal portion and maintained the trend afterwards. For the class 2 genes (Fig. 4D), KNOPE3 transcription decreased linearly; the KNOPE4 mRNA level oscillated similarly to that of KNOPE2; and KNOPE7 expression was triggered at the distal sector and was constant in the subsequent moieties.

KNOPE1 maps to a QTL for internode length

To investigate further the alleged role of KNOPE1 and KNOPE3 in the stem (Table 3), the association with an internode length QTL was experimentally addressed in the P×F population. No allele diversity was found within KNOPE3, whilst the KNOPE1- SNP was analysed in the BC1 population and mapped at 77 cM in linkage group (LG) 1 (Fig. 5), where a QTL for internode length was previously identified (Verde et al., 2002). The QTL analysis was refined using an improved map (see the Materials and methods); the internode length data (Table 4; Supplementary Fig. S4 at JXB online) were normally distributed in 1998 and 1999, and the QTL LOD score threshold was found to be at 2.9 and 2.8, respectively. The previous QTL position was confirmed with an explained phenotypic variance from 19.7% to 10.9% in 1998–1999. The respective LOD peaks (5.9 and 3.1) were found to coincide with the KNOPE1- marker position (Fig. 5, Table 4).

Fig. 5.

KNOPE1 associates with a QTL for internode length. Map locations and LOD score of a QTL for internode length mapped on linkage group 1 (G1) in the peach P×F population in 1998 (In-98) and 1999 (In-99). The inner 1–LOD interval is shown as a rectangle, and the outer 2–LOD interval as lines.

Table 4.

KNOPE1 and internode length QTL association in the P×F mapping population

Internode length in BC1 progeny				QTL association
Year	Averagea	Min	Max	QTL peak (cM)	Peak closest marker (cM)	LOD	Variance (%)
1998	2.9 ± 0.33	2.2	3.9	77	KNOPE1- 681 (77)	5.9	19.7
1999	2.6 ± 0.34	1.7	3.5	77	KNOPE1- 681 (77)	3.1	10.9

a Mean ± SD (cm) was measured for 125 individuals as previously described (Verde et al., 2002).

KNOPE1 mRNA localization is dynamic in primary growing stems

Focusing on the role of KNOPE1 in peach stem, in situ hybridization was performed on sections at 0.2, 2 (PG), and 50mm (SG) beneath the SAM of the model shoots (Fig. 4). For the SAM (Fig. 6A–C), the KNOPE1 gene was expressed in the peripheral and rib zones, and in the vascular strands, whereas it was down-regulated in the pith and leaf primordia (Fig. 6B). The outermost longitudinal sections (Fig. 6C) showed that the transcript marked the procambium and procambium-derived vascular strands. At 0.2mm (Fig. 6D–F), the mRNA signal was quite intense in the procambium of leaf traces and stem bundles, and was detected in the cortex at a lower level (Fig. 6F). At 2mm (Fig. 6G–I), the pith rays (interfascicular parenchyma, Fig. 6I) separated the vascular bundles, thus marking the PG stage. The transcript was localized in the cortex layers beneath the epidermis (collenchyma) and in the vascular bundles (Fig. 6H). Close-ups of these latter showed that the signal was intense in the procambium and metaphloem regions, and faint, if not absent, in the incipient phloem fibres and xylem (Fig. 6I). At 50mm (Fig. 6J–L), the vascular cambium marked the SG stage; the KNOPE1 transcript narrowed in thin cortex layers below the epidermis and in the vascular bundles (Fig. 6K). The mRNA was detected in the cambium ring and phloem domains, but not in mature primary phloem fibres on magnification (Fig. 6L). A weak signal was found in the secondary xylem initials and xylem parenchyma cells, but not in the primary xylem (Fig. 6H, 6K). No signal above background was observed in experiments with a sense probe (Fig. 6A, 6D, 6G, 6J).

Fig. 6.

KNOPE1 mRNA localization in the SAM and subtending stem. In situ hybridization was performed on the shoot types described in Fig. 4A and 4B; the KNOPE1 mRNA appears as an intense blue signal. (A, D, G, J) Control experiments were hybridized with a digoxigenin-labelled KNOPE1 RNA sense probe. No signal above background was detected. (B) In SAM longitudinal sections, the signal marked the peripheral (PZ) and rib zones (RZ), and the provascular strands (pv), and was undetected in leaf primordia (lp) and developing leaves (dl). (C) SAM outer longitudinal sections showed that KNOPE1 was expressed in the procambium (pc) and provascular strands (pv). (E) A transversal stem section at ~0.2mm below the SAM. KNOPE1 mRNA intensely marked the leaf traces (lt) and procambium (pc) and feebly marked the cortex (cx). (F) Magnification of E; the mRNA was abundant in the leaf traces and in the procambium, whereas it was low (if not absent) in the pith (pt). (H) A transversal section at ~2mm below the SAM. The signal featured in the cortex (cx), and procambium/phloem regions (pc/p) of the stem and leaf traces, and was scarce in phloem fibres (pf), xylem (x), and pith (pt). Intense signal spots occurred in xylem cells adjacent to the pith. (I) A close-up of H focused on stem bundles showed that KNOPE1 was transcribed in metaphloem (mp) and procambium (pc) regions and down-regulated in the metaxylem (mx) region. The inferfascicular parenchyma (ip) separated the bundles and indicated a state of primary growth. (K) A transversal section at ~50mm below the SAM (secondary growth). The signal marked the cortex (cx) at the level of subepidermal layers and was intense in the phloem (p) and vascular cambium (vc) regions. (L) An enlargement of K focused on stem bundles indicated that KNOPE1 was expressed in the phloem elements (p) and vascular cambium cells (ca) but not in phloem fibres and secondary xylem (x); signal spots were also observed in xylem parenchyma cells (xpc) adjacent to the pith (pt). Size bars: 10 µm in F, 20 µm in A–C, 25 µm in B–I, 65 µm in L, 70 µm in D and E, 160 µm in G and H, 140 µm in J and K.

KNOPE1 expression rescues the bp4 rachis defects

The KNOPE1 function in stem PG was addressed by overexpressing the gene in the bp4 loss-of-function mutant, which bears typical epinastic siliques and a shortened rachis (Fig. 7A, left and right panels). The 35S:KNOPE1/bp4 selected genotypes showed moderate serrations of leaf margins (Supplementary Fig. S5 at JXB online), rescued the fruit orientation, stem length, and internode number to different extents (Fig. 7A, central panel, Fig. 7B, Table 7), and were confirmed to express KNOPE1 in the rachis (Fig. 7C). At the histological level, the lignin pattern of both 35S:KNOPE1/bp4 and wild-type genotypes was quite similar (compare Fig. 8A–D with E-H), whilst the bp4 internodes contained a thick and irregular lignin ring (Fig. 8I–L). To investigate the role of KNOPE1 in the lignin pathway further, the expression of a five gene set (covering early and late steps of lignin synthesis) was monitored in the stem inflorescence (Fig. 9). Specifically, internodes 1 and 2 of the wild-type, bp4, and 35S:KNOPE1/bp4 genotypes were found to be elongating at 9 d after bolting (Table 8) and were sampled to perform comparative analyses by qRT-PCR (Fig. 9A, 9B). The five lignin genes were significantly up-regulated in the bp4 compared with the wild-type internodes, whilst the transcript abundance of 35S:KNOPE1/bp4 did not differ from that of the wild type. These expression trends occurred in both internode 1 and 2 of the three genotypes.

Fig. 7.

KNOPE1 expression rescues the stem length of the bp4 mutant. (A) Examples of the wild type (Ler, left), bp4 mutant (right), and 35S:KNOPE1/bp4 line 1 (centre); the latter exhibits upward-pointing siliques (arrowheads) and an elongated inflorescence stem. Bar=4cm. (B) Rachis architecture of the wild-type (left), 35S:KNOPE1/bp4-1 (central), and bp4 (right) plants. Flowers, siliques, and leaves were removed to highlight rachis length and branching. Bar=1.4cm. (C) KNOPE1 and BP relative expression in transgenic and control plants. BP is expected to be transcribed in the bp4 genotype; however the mRNA was confirmed to contain the A–G substitution responsible for the mis-splicing and protein mutation (Douglas et al., 2002).

Table 7.

Phenotypes of Arabidopsis lines

Genotypes	Rachis length (mm)a	Internode number
Ler wild-type	257.0±19.8	61.8±6.9
bp4	32.7±5.9*	24.1±3.9*
35S:KNOPE1/bp4-1	210.7±24.0*	60.4±5.5*
35S:KNOPE1 bp4-2	136.5±19.8*	43.1±6.3*
35S:KNOPE1/bp4 -7	123.2±7.0*	47.2±4.1*

a Mean and standard error were measured for the central rachis at 42–45 d after bolting.

*Significant differences (Student’s t-test P < 0.05) of mutants versus the wild type plants.

Fig. 8.

KNOPE1 restores lignin deposition of the bp4 mutant. Lignin deposition patterns (phloroglucinol red staining) in transversal sections of elongating internodes I and II of wild-type (Ler), 35S:KNOPE1/bp4-1, and bp4 plants at 9 d after bolting. A, E and I are sections of the first internode in the three genotypes (indicated on the left), and B, F and J are the respective magnifications. C, G and K are sections of the second internodes, and D, H and L are the respective close-ups. In I and K, the black arrowheads indicate the irregular lignin ring; and the white arrowheads indicate the achlorophyllous stripes. In J and L, the insets magnify anomalies in xylem development. Size bars: 100 µm in A, C, E, G, I, K; 50 µm in B, D, F, H, J, L; 50 µm in J and L insets.

Fig. 9.

Lignin gene expression was rescued in 35S:KNOPE1/bp4 genotypes. The expression of five lignin genes was profiled by qRT-PCR in the first (A) and second (B) internode of the wild-type (Ler), 35S:KNOPE1/bp4-1, and bp4 lines. Significant differences compared with the wild type are marked with an asterisk (P < 0.05). C4H, cinnamate-4-hydroxylase; 4CL1, 4-coumarate:CoA ligase 1; CAD4, cinnamyl-alcohol dehydrogenase 4; LAC4, LACCASE 4; CCoAOMT, caffeoyl coenzyme A O-methyltransferase 1. Accession numbers are given in Supplementary Table S3 at JXB online.

Table 8.

Rachis elongation

Genotypes	3daba	6dab	9dab	12dab	15dab
Length of first internode (mm)
Ler wild type	7.3±2.8	17.2±5.2	28.6±8.1	36.6±13.8	37.8±14.0
bp4	3.3±1.7*	6.0±2.3*	7.2±3.3*	8.0±3.5*	9.0±3.5*
35S:KNOPE1/bp4-1	8.0±2.0	20.5±4.7	27.8±6.0	30.8±7.5	31.3±7.3
35S:KNOPE1/bp4-2	8.7±2.0	18.5±4.5	22.5±6.8*	23.9±7.7*	24.4±8.1*
35S:KNOPE1/bp4-7	8.9±1.4	19.4±7.4	23.6±9.1	26.1±9.9*	26.3±10.1*
Length of second internode (mm)
Ler wild type	1.0±0.2	5.8±1.7	14.5±8.8	23.5±8.0	27.8±7.3
bp4	–	0.7±0.2*	1.1±0.3*	2.4±0.7*	3.1±0.9*
35S:KNOPE1/bp4-1	1.3±0.5	6.1±2.3	14.8±5.1	21.3±6.1	22.3±6.5
35S:KNOPE1/bp4-2	0.7±0.1*	4.7±1.6	10.0±3.0*	13.6±4.8*	15.1±4.0*
35S:KNOPE1/bp4-7	1.2±0.4	8.8±4.1*	18.2±6.6	20.8±7.6	22.4±8.0

a dab, days after bolting; the start of bolting was fixed at when the rachis was 1–2mm long; the bolting time of bp4 (24±2 days after sowing, das) and 35S:KNOPE1 (22±1 das) genotypes occurred earlier than that of controls (36±2 das); mean± SD from 15 plants of each genotype.

*Significant differences (Student’s t-test P < 0.05) of mutants versus the wild-type plants.

KNOPE1 binds to the TGACAGC motifs harboured in lignin pathway genes

The above-mentioned Arabidopsis lignin genes harboured several KNOX-targeted binding sites (BSs), which were also present in the peach putative orthologues (Supplementary Table S5 at JXB online). To ascertain the ability of the KNOPE1 protein to recognize these BSs, EMSA experiments were performed on the peach CCoAOMT promoter moiety (60-mer) containing the TGACAGC motif (Fig. 10A, 10B). KNOPE1 recognized the radiolabelled wild-type probe (Fig. 10C, lane 2). The mutant (lanes 3–5, group M) and wild-type probes (lanes 6–8, group WT) were added as cold competitors at increasing doses to the labelled KNOPE1–wild-type probe complex. The band pattern was unchangd after adding the mutant probe, whereas a progressive shift was observed when cold wild-type probe was provided. This supported that KNOPE1 could specifically interact with the TGACAGC motif.

KNOPE1 expression is inversely correlated with peach stem lignification

The lignification grade of peach shoots increased from apical to basal internodes, as evidenced by the lignin deposition pattern (Fig. 11A). The transcript increase of the lignin synthesis genes (Fig. 11B) coincided with the decrease in KNOPE1 mRNA (Fig. 4C). Tissue staining showed that lignification started in xylem cells during PG (Fig. 11C–F, sections at 0.2mm and 2mm below the SAM) and proceeded in xylem and phloem fibres at early SG (Fig. 11G, 11H, sections at 50mm). In comparable sections, the KNOPE1 mRNA was not detected in domains where lignification occurred (Fig. 6D–L).

Fig. 11.

Lignin gene expression and deposition in peach shoot stems. (A) Phloroglucinol-stained (red) lignin sections from internodes of the model shoot (Fig. 4). Bar=1mm. (B) Peach lignin gene transcription in apical, distal, proximal, and basal sectors was inversely correlated to KNOPE1 expression (Fig. 4C). 
(C, E, G) Stem sections were stained by phloroglucinol and the respective close-ups (D, F, H) were stained by safranin/Fast Green which highlights cytoplasm and cellulosic cell walls in blue, and nuclei and lignified/suberized/cutinized cell walls in brilliant red. White arrows indicate lignified cells. Sections were performed at 0.2, 2, and 50mm beneath the SAM as in Fig. 6. cx, cortex; 
p, phloem; pc/p, procambium/phloem regions; pf, phloem fibres; pt, pith; x, xylem. Size bars: 15 µm in D, 30 µm in C, 50 µm in F, 80 µm in H, 180 µm in E, 200 µm in G.

Discussion

KNOPE genomic and deduced protein features

The peach genome contains at least 10 KNOPE genes as found in the ‘Chiripa’ and ‘Lovell’ cultivars, five of class 1, four of class 2, and one of class M (Fig. 1, Table 5; Supplementary Fig. S3 at JXB online). The KNOPE structures (Fig. 1; Supplementary Fig. S2) are remarkably similar to those of the Arabidopsis counterparts, consistent with the finding that orthologues tend to have more conserved intron positions than non-orthologues (Henricson et al., 2010). The KNOX genes appear to have a limited number in species of different genome sizes, for example 10 in peach, at least six in barrel medic (Di Giacomo et al., 2008), eight in Arabidopsis, 13 in rice, 12 in maize, 15 in poplar (Mukherjee et al., 2009), and >15 in apple (Jung et al., 2008). This is compatible with the coding theory which predicts that upper bounds for the transcription factors number exist so as to minimize cross-binding errors between transcription factors (Itzkovitz et al., 2006). Seven KNOPE genes were experimentally located on the Prunus reference map (Fig. 2), whereas KNOPE2.1, KNOPE7, and KNOPEM were computationally placed on LG5 (Jung et al., 2008). Synoptically, the KNOPE genes are sparse on six out of eight LGs, in agreement with the KNOX scattering in plant genomes (www.phytozome.net) and in contrast to homeobox gene clustering in animal genomes (Mann, 1997). Alignment matrices indicated that the STMlike1/STMlike2 and KNOPE3/KNOPE4 pairs had the highest identity (72% and 77%, respectively). These putative paralogues share similar intron–exon organizations but are located on four different LGs, suggesting that duplication and translocation events may have occurred within the genus Prunus.

The class 1, 2, and M proteins fall into separate clades which include the respective Arabidopsis orthologues, supporting the concept of shared functional equivalence. Regarding the highly variable N-termini of KNOPEs, His/Gln-rich stretches were scored (not shown) and reported to be pivotal in the autonomous transcriptional activation domains of algal KNOX (Lee et al., 2008). For the MEINOX and C-termini involved, respectively, in protein–protein and protein–DNA recognition (Hay and Tsiantis, 2010), the average IGs within class 2 factors were higher than those within class 1, suggesting that the former may have fewer combinatorial interactions. In this context, one may speculate that class 1 proteins are likely to control wider gene pools and have more diversified functions than class 2 factors. Finally, five KNOPEs harbour PEST signals and their turnover may be mediated by the cullin-RING ubiquitin ligases (Xing et al., 2010).

KNOPE genes are differentially expressed during 
stem development

The peach model shoots (Fig. 4B, Table 6) bore apical internodes undergoing elongation and subsequent radial growth (after reaching a defined length). All KNOPE genes are transcribed (Fig. 4C, 4D) in both PG and SG with diversified patterns (e.g. a linear decrease for KNOPE1, STMlike2, and KNOPE3; an increase for KNOPE7, and an oscillatory/unvaried trend for the others), suggesting complexities of their role. Consistently, the class 1 STM-like (Tioni et al., 2003; Michelotti et al., 2007) and BP-like genes (Watillon et al., 1997; Tioni et al., 2005; Tanaka et al., 2008) were observed to be down-regulated along the stem and to narrow their localization (e.g. the cambium/phloem) in mature internodes. The meristematic class 1 KNOPE gene profiles support that SAM mechanisms were co-opted for stem development and that KNOX genes are multifunctional (Spicer and Groover, 2010). For KNOPE1, the apex-to-base down-regulation supports the coercive role in lignin deposition and genes (see the previous paragraph), whilst the KNOPE1/KNOPE6 inverse correlation is reminiscent of the BP repression versus KNAT6 during stem growth (Smith and Hake, 2003; Ragni et al., 2008). Finally, the class 2 KNOPE7 trend is reminiscent of that of Arabidopsis and poplar orthologues, which control cell terminal differentiation (Zhong et al., 2008; Li et al., 2012).

KNOPE1 exerts its function in peach stem elongation and lignification

The KNOPE1 link to a QTL for the peach internode length was robust and was maintained over time (Table 4, Fig. 5). There is evidence that the valine to isoleucine substitution can modify the functionality of transcription factors in animals (Dawson et al., 1996) and the selectivity of abscisic acid (ABA) receptors in plants (Yuan et al., 2010). Considering that the KNOPE1- SNP encompasses a valine/isoleucine swap in the KNOX1 subdomain, it may be worth surveying whether it alters the KNOX–protein interactions and internode traits.

The KNOPE1 down-regulation in the SAM cental zone and forming leaves (Fig. 6A–C) agrees with previous data (Testone et al., 2008). Transcription occurs in the peripheral and rib zones, and extends to the stem cortex and vascular bundles, reminiscent of the Arabidopsis BP pattern (Lincoln et al., 1994; Semiarti et al., 2001; Lenhard et al., 2002; Khan et al., 2011). The KNOPE1 expression in the procambium/provascular strands of both the inner SAM and stem is reminiscent of that of poplar ARK2 (Du et al., 2009) and sunflower Hakn2 (Tioni et al., 2003). For the vascular net, the message marks the procambium/metaphloem and is weak in phloem fibres, xylem, and xylem fibres during PG (Fig. 6E–I). This pattern recurs in early SG (Fig. 6K, 6L) but further includes the expression in the cambium ring, phloem elements, and xylem parenchyma cells (XPCs). During PG, the transcript scarcity along the xylem side suggests that KNOPE1 down-regulation is required for several differentiation phases. Inherently, the xylem of Arabidopsis bp mutants includes anomalies in early (e.g. size, see Mele et al., 2003, Smith and Hake, 2003; organization, see insets of Fig. 8J and 8L) and late differentiation (Mele et al., 2003). During SG, the down-regulation may govern xylem early differentiation and lignification as does the aspen ARK2 (Du et al., 2009). The expression in the non-lignifying XPCs (Ros Barcelo, 2005) is in agreement with the lignin antagonism of KNOPE1 described later. The gene message persists in the phloem, although it decreases in fibres, probably due to the onset of lignification. The BP-like mRNAs occur in stem phloem (Tioni et al., 2003; Douglas and Riggs, 2005; Du et al., 2009), but their roles are still undefined during PG; complications arise because in situ experiments can also reveal sap-transported KNOX I mRNAs (Campbell et al., 2008; Ham et al., 2009) for long-distance functions (Kim et al., 2001). However, the ARK2-overexpressing and silenced lines of aspen showed delayed and precocious phloem differentiation during SG, respectively (Du et al., 2009). In peach, KNOPE1 might control the time progression of differentiation and/or vascular organization, considering that xylem is generated prior to phloem during SG (Schneider, 1945). Referring to the cortex, the KNOPE1 expression spreads in the parenchyma of subapical stems and narrows progressively to the collenchyma from PG to SG (Fig. 6E, 6H, 6K). The dynamic pattern might reflect the role of KNOPE1 in cell proliferation/differentiation required for primary stem elongation and the subsequent involvement in phelloderm identity. The BP and ARK2 mRNAs occur in the cortex of PG stems, but only the BP expression was proposed to regulate bundle spacing in Arabidopsis (Douglas and Riggs, 2005). Overall, the KNOPE1 patterns sustain the assumption that, in stem, BP-like genes may both specify the fate of meristem-derived cells and act on distinct differentiation phases (Du et al., 2009).

Regarding gene function, the KNOPE1 expression in the Arabidopsis bp4 background rescued these rachis defects during PG: internode shortening (Fig. 7, Table 7), irregular lignin deposition (Fig. 8), abnormal bundle spacing, and chlorenchyma loss (Fig. 8K, 8L). Thus, the previously assessed KNOPE1/BP functional equivalence (Testone et al., 2008) was corroborated for stem PG. The KNOPE1–internode length QTL association and the ability to restore internode extension and the number of bp4 (Tables 7, 78) strongly support the involvement in elongation. The BP (Byrne et al., 2002) and/or BP-like gene gain of function can cause stem shortening in herbs and trees (Du et al., 2009; Srinivasan et al., 2011), whilst mRNA-induced down-regulation of ARK2 produces internode lenghtening in aspen (Du et al., 2009). In this context, KNOPE1, like its orthologues, is predicted to control elongation via complex mechanisms. Stem elongation is achieved via cell division and expansion, and ceases gradually concomitantly with lignin deposition and secondary cell wall formation (Evert and Eichhorn, 2007). Inherently, hormone and cell terminal differentiation genes respond to BP-like misregulation (Du et al., 2009), and KNOX I interactions with auxins, gibberellins, and cytokinins have been extensively investigated (Hay et al., 2004). As for peach, the supply of gibberellins and biosynthesis inhibitors induces stem linear growth and shortening, respectively (Flemion, 1959; Blanco, 1988; Gonzalez-Rossia et al., 2007). Hence, the KNOPE1–gibberellin relationships merits investigation considering the repressive roles of BP and ARK2 in gibberellin biosynthesis genes (Hay et al., 2002; Du et al., 2009). Finally, the KNOPE1–cytokinin cross-talk was proposed to maintain cells in a proliferative/undifferentiated state (Testone et al., 2008), which may be required for the parenchyma layers during elongation.

A set of experiments supported that KNOPE1 prevents lignin formation by repressing several biosynthesis genes. Five of these were up-regulated in bp4 internodes, but the transcription was restored to the wild type levels by the KNOPE1 control (Fig. 9), which also caused regular lignin deposition (Fig. 8). For peach, we observed that: (i) the protein can bind to typical KNOX motifs carried in lignin gene promoters (Fig. 10; Supplementary Table S5 at JXB online); (ii) the transcription was inversely correlated with that of both lignin genes (compare Figs 4C and 11B) and deposition along the shoot stems (Fig. 11A, 11C); and (iii) the gene was down-regulated in vascular tissues undergoing lignification (compare Figs 6 and 11C). These results are in agreement with the roles of BP and ARK2 (Mele et al., 2003; Du et al., 2009). Moreover, the regulation of lignin content and composition was shown to be cell type specific (Nakashima et al., 2008), and the fine tuning of transcription factors controlling lignin pathways was proposed (Li et al., 2012). Hence, KNOPE1 might take part in the cessation of elongation by a fine and dynamic control of terminal cell differentiation.

The known peach recalcitrance to genetic transformation led to the study of the KNOPE1 function in Arabidopsis. Technologies of gene transfer by stable (Prieto, 2011) and transient approaches (Jia et al., 2010) have improved, but are still cultivar specific and time consuming. Other Prunus spp. are more viable for gene function studies (Prieto, 2011), and hybrid rootstocks are being developed (Song and Sink, 2006). Finally, the KNOPE1- marker is a genetic tool useful to screen for genotypes with desired traits, and the KNOPE1 function assessment provides information for biotechnology tools to control scion stature and maturation. Both can impact on planting density, pruning, fruit harvesting, and, eventually, costs in peach culture.

Supplementary data

Supplementary data are available at JXB online.

Figure S1. Examples of KNOPE allele segregations in the T×E population.

Figure S2. KNOPE2.1, KNOPE7, and KNOPEM gene schemes.

Figure S3. Phylogenetic analysis of KNOPE proteins.

Figure S4. Internode length of the P and F parents of the mapping population.

Figure S5. Leaf phenotype in complemented Arabidopsis lines.

Table S1. Primers and products for isolation of the KNOPE genes.

Table S2. Dicot KNOX protein list.

Table S3. Primer list for expression analyses.

Table S4. ANOVA and Tukey tests for peach stem gene expression.

Table S5. KNOX-binding sites in Arabidopsis and peach lignin genes.