Overexpression of PhEXPA1 increases cell size, modifies cell wall polymer composition and affects the timing of axillary meristem development in Petunia hybrida


Author for correspondence:
Mario Pezzotti
Tel: +39 045 8027951
Email: mario.pezzotti@univr.it


  • Expansins are cell wall proteins required for cell enlargement and cell wall loosening during many developmental processes. The involvement of the Petunia hybrida expansin A1 (PhEXPA1) gene in cell expansion, the control of organ size and cell wall polysaccharide composition was investigated by overexpressing PhEXPA1 in petunia plants.
  • PhEXPA1 promoter activity was evaluated using a promoter-GUS assay and the protein’s subcellular localization was established by expressing a PhEXPA1-GFP fusion protein. PhEXPA1 was overexpressed in transgenic plants using the cauliflower mosaic virus (CaMV) 35S promoter. Fourier transform infrared (FTIR) and chemical analysis were used for the quantitative analysis of cell wall polymers.
  • The GUS and GFP assays demonstrated that PhEXPA1 is present in the cell walls of expanding tissues. The constitutive overexpression of PhEXPA1 significantly affected expansin activity and organ size, leading to changes in the architecture of petunia plants by initiating premature axillary meristem outgrowth. Moreover, a significant change in cell wall polymer composition in the petal limbs of transgenic plants was observed.
  • These results support a role for expansins in the determination of organ shape, in lateral branching, and in the variation of cell wall polymer composition, probably reflecting a complex role in cell wall metabolism.


Expansins are pH-dependent proteins that regulate cell wall enlargement in growing plant cells (McQueen-Mason et al., 1992; Lee & Kende, 2001; Cosgrove et al., 2002) and cell wall loosening in situations such as fruit softening (Brummell et al., 1999; Powell et al., 2003; Kitagawa et al., 2005), abscission (Cho & Cosgrove, 2000), pollen tube invasion (Pezzotti et al., 2002), the emergence of root hairs (Cho & Cosgrove, 2002), xylem differentiation (Milioni et al., 2001), symbiosis and pathogen infection (Giordano & Hirsch, 2004; Balestrini et al., 2005; Fudali et al., 2008). There are two expansin families (EXPA and EXPB) that facilitate wall stress relaxation and wall extension, also known as ‘wall creep’. EXPA is the most extensive family, with 26 genes in Arabidopsis and 34 in rice (Oryza sativa). There are also two further expansin-like families (EXLA and EXLB) whose functions are unknown (Sampedro & Cosgrove, 2005; Choi et al., 2006).

The mechanism of expansin activity is unclear, although wall extension occurs without the hydrolysis of major cell wall components (McQueen-Mason & Cosgrove, 1995). Expansin activity is usually studied in crude extracts, but such studies cannot resolve the complex molecular network of the cell wall or identify initiation sites. It is thought that expansins may dissociate the polysaccharide complex that links microfibrils together, without changing wall structure and the degree of cross-linking (McQueen-Mason & Cosgrove, 1994; Cosgrove, 2000).

Many expansin genes are expressed in patterns consistent with their involvement in the growth of specific organs or cell types (Reinhardt et al., 1998; Brummell et al., 1999; Lee & Kende, 2001; Wu et al., 2001). The expression of some expansins is also regulated by phytohormones (Catala et al., 2000; Lee & Kende, 2002), reflecting the presence of auxin, gibberellin, ethylene and abscisic acid response elements in the corresponding promoters (Lee et al., 2001). The genetic modification of expansin gene expression has been used to elucidate their specific involvement in wall growth and loosening, and has shown that morphological changes reflect differences in cell size rather than number (Cho & Cosgrove, 2000; Choi et al., 2003; Zenoni et al., 2004; Gray-Mitsumune et al., 2008).

Expansins can also alter the physical stress pattern in the meristem, so that tissue bulging occurs and cells achieve primordium identity. Fleming et al. (1997) showed that expansins applied to discrete topical regions on the flanks of tomato (Solanum lycopersicum) vegetative meristems induced the initiation of aberrant primordia that produced abnormal leaves. By contrast, the localized induction of expansin transgene expression on the flanks of tobacco vegetative meristems not only induced the formation of primordia but also reiterated the whole process of leaf development, giving rise to phenotypically normal leaves (Pien et al., 2001). The spatiotemporal pattern of expansin expression may therefore play an important role in regulating early organ development (Fleming, 2006).

Down-regulation of Petunia hybrida expansin A1 (PhEXPA1), an expansin preferentially expressed in petal limbs during development, affected cell expansion and final organ size (Zenoni et al., 2004). Limb epidermal cells in the transgenic lines lacked the normal contents of deposited crystalline cellulose, leading us to speculate that the role of expansins may be to disrupt noncovalent bonds between cellulose microfibrils and cross-linking glycans, thus preparing the cell wall for further cellulose deposition. Here, we show that PhEXPA1 is present in the cell wall of expanding tissues, where it plays a role in organ development (particularly the development of the petal limb) by remodelling the cell wall. The constitutive overexpression of PhEXPA1 in transgenic petunia plants had a significant impact on organ size, and also changed the architecture of the plant by initiating premature axillary meristem outgrowth, suggesting that expansins are involved in organ morphogenesis. Fourier transform infrared (FTIR) microspectroscopy and fractionation analysis of the components of the petal limb cell wall indicated that the polymer composition is different in wild-type and transgenic plants, confirming the role of expansins in cell wall metabolism and providing insights into their molecular functions.

Materials and Methods

Plant material and growth conditions

Petunia hybrida (var. Mitchell) plants were grown under glasshouse conditions (50–70% relative humidity, 24–30°C and a photoperiod of 14 h light and 10 h dark), and segregation analysis was carried out as previously described (Zenoni et al., 2004). The P. hybrida W138 Juliet insertion library (Vandenbussche et al., 2003) was prepared as described by Koes et al. (1995).

Promoter sequence isolation

A clone comprising the first 1008 bp upstream of the PhEXPA1 gene was isolated by two sequential screens of the Juliet library (dTph1-insertion lines) using a modified family signature insertion screening strategy (Vandenbussche et al., 2003). The first PhEXPA1 specific primer (reverse orientation) was 5′-ACC TGG AAT TCT TGC TTC AAC TGT TAG-3′, and the second (reverse orientation) was 5′-GTG CGA TTG AAA CTA GTA GGA CCC TAA-3′. Primer labelling, PCR screening of the insertion library, fragment isolation and sequencing were carried out as described by Vandenbussche et al. (2003).

Cloning and plant transformation

The PhEXPA1::GUS reporter construct, in which the Escherichia coliβ-glucoronidase-encoding gene uidA was driven by a 972-bp PhEXPA1 promoter fragment, was created by amplifying the promoter sequence directly from genomic DNA using primers PROfor (5′-CAC CCC AAA TAG CTT TCA CAT TGA GCT TCG-3′), containing four additional bases (underlined) required for directional cloning, and PROrev (5′-TTC TAT AAA AGG GAA CAG AAA GAG-3′). This product was transferred to the pENTR™/D-TOPO vector (Invitrogen, Carlsbad, CA, USA) for sequencing, and then used for LR recombination with the promoter analysis Gateway destination vector pKGWFS7 containing uidA (Karimi et al., 2005).

The 35S::PhEXPA1-eGFP construct, containing a PhEXPA1 fusion with the gene for green fluorescent protein (eGFP), was created by amplifying the PhEXPA1 coding region from petal cDNA using primers EXPfor (5′-CAC CAT GGG GTT TTT CAA GAA TG-3′) and EXPrev (5′-AAT TCT AAA GTT CTT TCC C-3′). The PCR product was inserted into the pENTR™/D-TOPO vector using the directional cloning strategy described above for PhEXPA1::GUS reporter construct, sequenced, and transferred to the C-terminal eGFP fusion Gateway destination vector pK7FWG2 (Karimi et al., 2005).

The 35S::PhEXPA1 construct, with the PhEXPA1 coding region under the control of the cauliflower mosaic virus (CaMV) 35S promoter, was created by amplifying the PhEXPA1 coding region from petal cDNA using forward primer EXPfor in combination with reverse primer EXPSTOPrev (5′-TTA AAT TCT AAA GTT CTT TCC C-3′). The product was transferred to the pENTR™/D-TOPO vector (Invitrogen) and then the Gateway pK7WG2 vector (Karimi et al., 2005) by LR recombination.

The 35S::GUS construct was created by LR recombination between the pENTR™-gus Positive Control vector (Invitrogen) and the Gateway pK7WG2 vector.

Transformation of P. hybrida var. Mitchell plants was carried out as described by Zenoni et al. (2004). After regeneration under kanamycin selection, transgenic lines were identified by genomic PCR using primers PRO1F (5′-CTC ATT GGA GAC TGA ACC ATAC-3′) and GUS2 (5′-GGA TAG TCT GCC AGT TCA GT-3′) for PhEXPA1::GUS, EXP2 (5′-AGT GGT GGA GCA TGG CAA A-3′) and GFP2 (5′-CTT CAG CTC GAT GCG GTT CA-3′) for 35S::PhEXPA1-eGFP, 35S (5′-AAG GAA GGT GGC ACC TAC AA-3′) and STOP (5′-TTA AAT TCT AAA GTT CTT TCC C-3′) for 35S::PhEXPA and 35S and GUS2 for 35S::GUS.

GUS reporter analysis

The PhEXPA1 expression profile was investigated by staining tissues from transgenic plants containing a single copy of the PhEXPA1::GUS reporter transgene for GUS activity. Histochemical assays were performed as described by Jefferson et al. (1987) using the substrate 5-bromo-4-chloro-3-indolyl-β-d-glucuronide. The tissues were cleared in 70% ethanol before analysis.

Subcellular localization of PhEXPA1-eGFP

Limb flower tissues from wild-type and 35S::PhEXPA1-eGFP transgenic plants were cut into small sections, washed with distilled water and counterstained with 0.1% calcofluor white for 15 min to reveal the cell walls. Plasmolysis was induced by suspending the sections in 2 M sucrose for at least 30 min to allow discrimination between the cell wall and plasma membrane. Flower tissues were examined using a Zeiss Axioplan microscope and a Leica TCS SP2 confocal microscope (Heidelberg, Germany). Confocal images of the abaxial epidermis were taken by sequentially acquiring fluorescent signals from GFP (488-nm Ar/Ar-Kr laser) and calcofluor (351-nm Ar-UV laser) from single optical sections. The fluorescent signals were analysed using Leica Confocal Software (v. 2.61.1537).

Semiquantitative real-time RT-PCR

Total RNA was extracted from P. hybrida petal limbs, ovaries and leaves, and cDNA was synthesized as described by Zenoni et al. (2004). Petal limbs and ovaries were collected 140, 170, 190 and 250 h after flower bud appearance (AFBA) and leaves were collected at three different developmental stages. Semiquantitative real-time RT-PCR was carried out on pooled samples from all developmental stages for each organ, using an Mx3000P QPCR (Stratagene, La Jolla, CA, USA) with SYBR green PCR Master Mix reagent (Applera, Foster City, CA, USA). Specific primers for PhEXPA1 were designed in the 5′ untranslated region: EXP1 (5′-ATG GGG TTC AAG AAT G-3′) and EXPRT (5′-CCA CCA TAA AAA GTA GCA T-3′). For PhEXP2 we used primers described by Zenoni et al. (2004). For PhEXP4 and PhEXP5 we designed specific primers in the 3′ untranslated region: EXP4for (5′-TTC TAT ACC TTT CCT CTT CAC CTT TAC-3′) and EXP4rev (5′-TTT ATA GGG AAT GGG TTT GTT ACT-3′), and EXP5for (5′-TAA GCG GGT GCT GAC CGA ACA CAA-3′) and EXP5rev (5′-CTT TGA CTA TTG AAA TCA TCT C-3′). All experiments were carried out three times with independent RNA samples using a thermal cycling profile described by Zenoni et al. (2004). Quantitative values were normalized to those of actin (Zenoni et al., 2004) amplified under the same conditions using primers ACT1 (5′-ATC CCA GTT GCT GAC AAT AC-3′) and ACT2 (5′-GGC CCG CCA TAC TGG TGT GAT-3′). Each real-time assay was tested using a dissociation protocol to ensure that each amplicon was a single product. Data were analysed using MxPro QPCR software (Stratagene), and reaction efficiencies were calculated using LinRegPCR (Ramakers et al., 2003).

Measurement of expansin activity

Crude protein extracts were prepared from 10 g of wild-type and 35S::PhEXPA1 (line 1) leaves, pooled from the second to the eighth internodes from the apex, as described previously (http://homes.bio.psu.edu/expansins/Protocols/Extraction.htm). The protein extracts were dispersed in 50 mM sodium acetate buffer (pH 4.5) to a final concentration of 0.625 μg μl−1. Cellulose/xyloglucan composites from Acetobacter xylinus were used for cell wall extension assays essentially as described by Nieuwland et al. (2005). The composites were cut into 2 × 20 mm strips and placed under a constant load of 20 g in 800 μl of 50 mM sodium acetate buffer (pH 4.5). Extension was measured every 30 s until c. 15 min after the extension rate became stable. The buffer was then replaced with a protein extract and expansin activity was calculated as the difference in extension rates 10 to 0 min before and 5 to 15 min after addition of the protein extract.

Analysis of epidermis cell area and total cell number

Petal limbs from wild-type and transgenic plants were collected at anthesis and divided into three numbered zones representing the five petals of corolla limbs. Each zone was further divided into basal (a) and distal (b) portions (designated as (a) and (b) respectively, in Fig. S5a,b). The total epidermis surface area and abaxial epidermis cell number were determined as described by Zenoni et al. (2004).

FTIR microspectroscopy

Mid-infrared spectra were acquired in the 1800–900 cm−1 range using a Vertex 70 Bruker spectrometer (Bruker Optics Inc., Billerica, MA, USA) coupled to a Hyperion 3000 vis/IR microscope (Bruker Optics Inc., Billerica, MA, USA) equipped with a photoconductive Mercury Cadmium Telluride (MCT) detector and a 20X Germanium attenuated total reflection (ATR) crystal objective (c. 100 μm diameter). Twelve petal limbs from six different flowers were randomly selected for each of the three plant lines (wild type, 35::antisensePhEXPA1 and 35S::PhEXPA1). For each petal limb, at least 10 point-by-point spectra were acquired by co-adding 64 scans at 4 cm−1 resolution (30 s acquisition time), without prior sample preparation. Absorbance spectra were corrected for the wavelength dependence of penetration depth inside the sample and area normalized after baseline subtraction using the Bruker opus 6.0 software (Bruker Optics Inc., Billerica, MA, USA), to allow the evaluation of relative changes in cell wall composition. Principal component analysis (PCA) was applied to the average spectra of each petal limb using the statistical packages of the Matlab© software (Natick, MA, USA) environment.

Cell wall fractionation and analysis

A sequential chemical extraction protocol was adapted to separate the components of the petal limb cell walls, as described by Li et al. (2006). Final, dried ‘crude wall material’ was then extracted in 90% DMSO for 48 h at room temperature to dissolve starch (Carpita & Kanabus, 1987), then three times with 4 ml of 50 mM EDTA (pH 6.5) at 100°C to extract pectic substances, and then with 5 ml of 24% (w/v) KOH containing 0.02% NaBH4 at room temperature to extract hemicellulose. The remaining insoluble residue was designated the cellulose fraction. The wall fractions were dialysed and lyophilized before the analysis of neutral sugars, uronic acid and xyloglucan. Uronic acid content was determined by the m-hydroxybiphenyl method (Blumenkrantz & Asboe-Hansen, 1973; van den Hoogen et al., 1998). Neutral sugar content was calculated as described by Sakurai et al. (1987). The xyloglucan content was determined by iodine staining (Wakabayashi et al., 1991). The cellulose fraction underwent Seaman hydrolysis before the neutral sugar content was measured (Shatalov et al., 1999).


PhEXPA1 is expressed during limb development and in axillary branch emergence zones

The PhEXPA1 promoter was isolated by screening the P. hybrida W138 dTph1 insertion library (Vandenbussche et al., 2003) using gene-specific reverse primers, amplifying a fragment containing 1008 bp of sequence upstream of the PhEXPA1 coding region. The boundaries of the putative 5′-UTR were determined based on the average length of dicot 5′-UTR sequences (Kochetov et al., 2002) and were confirmed by screening the plant 5′-UTR sequence database (http://utrdb.ba.itb.cnr.it/). Putative TATA box, transcriptional initiation site and cis-acting elements in the upstream PhEXPA1 region were identified by searching the PLACE (Higo et al., 1999) and PlantCARE (Rombauts et al., 1999) databases (Supporting Information Fig. S1). Analysis of the PhEXPA1 promoter sequence revealed the presence of putative gibberellin and auxin response elements and several core binding sequences for DOF (DNA-binding with one finger) zinc finger proteins, three of which were clustered in the region −850 to −660 bp along with a core motif for MYB, a petal epidermis-specific protein involved in the regulation of flavonoid biosynthesis in petunia (Solano et al., 1995). Interestingly, two CArG consensus binding sites were also found in the PhEXPA1 promoter, the first a binding site for AGAMOUS-like 15 (Tang & Perry, 2003) and the second found in various promoters including those of SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) and FLOWERING LOCUS C (FLC ), components of the vernalization pathway (Hong et al., 2003).

To investigate the spatial and temporal expression of PhEXPA1 during plant development, we created transgenic petunia plants in which a region of 972 bp of genomic DNA upstream of the PhEXPA1 coding region (−873 to +99 bp) was used to drive GUS. This putative regulative region shares most of the characteristic hormone-responsive elements reported for other expansin promoters (Cho & Cosgrove, 2000, 2002) and its length is consistent with recently analysed expansin promoters (Ogasawara et al., 2009; Won et al., 2010).

Three independent PhEXPA1::GUS transgenic lines homozygous for a single copy of the transgene were analysed for GUS activity, revealing major sites of promoter activity in petals, ovaries and stigmas (Fig. 1a), roots (Fig. 1c) and stems (Fig. 1d–f). This is consistent with our previous real-time RT-PCR data (Zenoni et al., 2004). Fig. 1(b) shows GUS activity during flower development, when expression is restricted to specific portions of petal limbs at different stages, as defined by Reale et al. (2002). At stages 7 and 9, GUS activity was distributed in the unfolded parts of the limbs (Fig. 1b). At stage 11, GUS activity was visible in the subapical portion of the petal limbs, whereas at stage 13 it was detected in the boundary region between the limb and the tube. Interestingly, Reale et al. (2002) showed that cell division is confined to the folded part of petunia petal limbs at stages 7 and 9. GUS staining in petal development therefore provides further evidence linking PhEXPA1 expression and cell expansion during limb growth.

Figure 1.

Expression profile of Petunia hybrida expansin A1 (PhEXPA1). Petunia hybrida organs were histochemically stained to reveal GUS activity resulting from the expression of the β-glucoronidase-encoding gene uidA under the control of the PhEXPA1 promoter. (a) Flower organs. (b) Petal development, showing petals at stages 7, 9, 11 and 13 of flower development as reported by Reale et al. (2002). (c) Root (90 d old). (d) Longitudinal section of the apex, when flower and inflorescence meristems are clearly separated. FM, floral meristem; IM, inflorescence meristem. (e) Longitudinal section of the third node from the apex. FS, floral stalk; S, stem; AM, axillary meristem. (f) Expression during axillary meristem outgrowth (redrawn from Souer et al. (1998)). Flowers are indicated by closed circles, apical inflorescences by closed triangles, and vegetative axillary meristems by open triangles; smaller size indicates stronger dormancy. The position of leaves and bracts is shown by large and small ovals, respectively.

In roots, PhEXPA1 GUS activity was localized mainly in the apical undifferentiated cells and in the vascular cylinder (Fig. 1c). GUS activity was also detected in stems, although it was limited to the nodes at the base of the pedicel (floral stalk) (Fig. S2). GUS activity was barely detectable in the most acropetal nodes but increased progressively in nodes further from the apex (Fig. 1f). In the apex, when floral and inflorescence meristems were clearly separated, there was low GUS activity in the ovary and in the two stalks (Fig. 1d). Notably, PhEXPA1 expression in the third node from the apex predominantly localized to the axil of the pedicel at the base of the axillary meristems (Fig. 1e).

PhEXPA1 protein is detected in the cell wall

The PhEXPA1 protein has a 25-amino acid N-terminal signal peptide, suggesting that it is secreted to the cell wall (Zenoni et al., 2004). To investigate the subcellular localization of PhEXPA1, we monitored GFP fluorescence in transgenic petunia plants expressing PhEXPA1-eGFP under the control of the strong CaMV 35S promoter, the native promoter proving unsuitable. Preliminary analysis by confocal microscopy indicated that the fusion protein was localized in the cell wall and cytoplasm, but not in the nucleus or vacuole (data not shown). To define the subcellular localization of PhEXPA1-eGFP in more detail, wild-type and 35S::PhEXPA1-eGFP transgenic epidermal limb cells were stained with the β-glucan-specific fluorochrome calcofluor white and monitored after plasmolysis (Fig. 2). Weak background blue autofluorescence was observed in wild-type plants (Fig. 2a,c), whereas transgenic plants showed strong GFP fluorescence and the signal was distributed in both protoplasts and cell walls (Fig. 2d–f). Indeed, the partial co-localization of GFP fluorescence with the calcofluor signal confirmed that some of the PhEXPA1-eGFP fusion protein was localized within the cell wall.

Figure 2.

Subcellular localization of the Petunia hybrida expansin A1 (PhEXPA1)-eGFP fusion protein. Plasmolysed Petunia hybrida petal limb abaxial epidermis cells from wild-type and 35S::PhEXPA1-eGFP transgenic plants were analysed with a confocal laser scanning microscope. (a) GFP signal (green) in wild-type cells. (b) Calcofluor fluorescence (red) in wild-type cells. (c) Merged image of the GFP (green) and calcofluor (red) signals. (d) GFP signal (green) in 35S::PhEXPA1-eGFP transgenic cells. (e) Calcofluor fluorescence (red) in 35S::PhEXPA1-eGFP transgenic cells. (f) Merged image of the GFP (green) and calcofluor (red) signals. Bar, 19 μm.

Overexpression of PhEXPA1 affects organ size by stimulating cell expansion

The down-regulation of PhEXPA1 expression reduces the size of petal limbs in comparison to wild-type plants and reduces the deposition of crystalline cellulose in epidermal cells (Zenoni et al., 2004). In order to provide further insight into the role of PhEXPA1, we created transgenic plants in which the PhEXPA1 gene was controlled by the constitutive CaMV 35S promoter, with the 35S:GUS construct used as a control. We obtained three independent 35S:PhEXPA1 transgenic lines and four independent 35S:GUS transgenic lines homozygous for a single copy of the transgene. All plants were analysed by real-time RT-PCR, revealing no significant differences in PhEXPA1 expression levels between 35S:GUS lines and wild-type plants (data not shown). In 35S:PhEXPA1 lines, PhEXPA1 expression levels were similar in leaves, petals and ovaries (data not shown), and were much higher than the levels in wild-type plants. Consistent with tissue-specific PhEXPA1 expression in wild-type plants (Zenoni et al., 2004), transgene expression was c. 20-fold higher in petals (Fig. 3a), c. 50-fold higher in ovaries (Fig. 3b) and c. 150-fold higher in leaves (Fig. 3c) in 35S:PhEXPA1 lines (t-test, P < 0.01).

Figure 3.

Analysis of Petunia hybrida expansin A1 (PhEXPA1) expression by semiquantitative real-time RT-PCR in three single-copy 35S::PhEXPA1 transgenic Petunia hybrida lines and wild-type (WT) controls. Total RNA extracted from petals, ovaries and leaves at different developmental stages was pooled for each organ. Actin was used as a control gene. (a) Petals; (b) ovaries; (c) leaves. Error bars represent standard deviation from the mean.

Transgenic 35S:GUS lines showed no morphological differences from wild-type plants, and we used the latter for characterization of the 35S:PhEXPA1 phenotype. All 35S:PhEXPA1 homozygous transgenic plants showed similar morphological phenotypes: petal limbs, leaves and internodes in the transgenic lines were significantly larger than their wild-type counterparts (t-test, P < 0.01) (Fig. 4a,c,e,f) and the average height of the transgenic lines was greater (Fig. 4g). This difference reflected an increase in internode length (t-test, P < 0.01), whereas the number of internodes was unchanged (Fig. 4d). The size of other plant organs, and germination and flowering times, were not affected (data not shown).

Figure 4.

Morphological changes in 35S::PhEXPA1 transgenic lines (PhEXPA1, Petunia hybrida expansin A1). (a) Comparison of Petunia hybrida petal limb surface area at anthesis. Pale grey bar, WT; dark grey bars, 35S::PhEXPA1 lines. (b) Comparison of fully expanded leaf surface area. Pale grey bar, WT; dark grey bars, 35S::PhEXPA1 lines. Leaves were analysed at the same internode insertion (eighth from the apex). Error bars represent standard errors; = 20 for both limbs and leaves, from 10 different plants. (c) Comparison of internode lengths. Pale grey bar, WT; dark grey bars, 35S::PhEXPA1 lines. Error bars represent standard errors; = 100 from 10 different plants. (d) Comparison of internode number. Pale grey bar, WT; dark grey bars, 35S::PhEXPA1 lines. Error bars represent standard errors; = 10 from 10 different plants. (e) Comparison of petal limbs at anthesis from 35S::PhEXPA1 line 1 and wild-type plants. (f) Comparison of fully expanded leaves at the same internode insertion (eighth from the apex) from 35S::PhEXPA1 line 1 and wild-type plants. (g) Comparison of whole plants (35S::PhEXPA1 line 1 and wild type) during vegetative growth, 60 d after germination. Asterisks indicate significantly different from wild type (t-test, P < 0.01).

To determine whether the overexpression of PhEXPA1 affected other expansin genes, we analysed the expression of (leaf-specific) PhEXP2, PhEXP4 and PhEXP5 (Zenoni et al., 2009) in the leaves of the three transgenic petunia lines by real-time RT-PCR, but there was no significant effect (t-test, P < 0.01) (Fig. S3).

To evaluate whether higher PhEXPA1 mRNA levels correlated with an increase in expansin activity, expansin activity in cell wall protein extracts obtained from wild-type and 35S::PhEXPA1 line 1 leaves was determined by extensometer measurements. The expansin activity in crude cell wall extracts from 35S::PhEXPA1 line 1 was significantly higher (t-test, P < 0.05) than that in wild-type extracts on cellulose-xyloglucan composites in vitro, confirming that PhEXPA1 is an active expansin and that the morphological differences in transgenic lines constitutively overexpressing PhEXPA1 reflect increased expansin activity (Fig. S4).

To determine whether the larger size of the petal limbs correlated with differences in the number or size of epidermal cells, petals from transgenic and wild-type flowers at anthesis were compared by light microscopy (Fig. 5a). There was no significant difference (t-test, P < 0.01) in the total cell number in the abaxial limb epidermis (data not shown). However, the limb abaxial epidermal cells in the transgenic plants had a significantly greater area than the corresponding wild-type cells (t-test, P < 0.01; Fig. 5a,b). To localize this aberrant developmental behaviour, limb tissues were divided into six portions based on organ morphology (Fig. S5a), but each portion appeared to contribute equally to the increase in size (Fig. S5b). Together with results from the antisense transgenic plants (Zenoni et al., 2004), these data suggest that PhEXPA1 is involved in determining final organ size by controlling cell expansion.

Figure 5.

Petunia hybrida petal limb morphology of 35S::PhEXPA1 line 1 transgenic and wild-type plants, correlating with the increase in abaxial epidermal cell surface area in the transgenic plants (PhEXPA1, Petunia hybrida expansin A1). (a) Comparison of abaxial petal limb epidermal cells at anthesis. Bar, 30 μm. (b) Surface area of petal limb abaxial epidermal cells at anthesis. Pale grey bar, WT; dark grey bar, 35S::PhEXPA1. Error bars represent standard errors; = 100 for epidermal cells, from five different flowers. The asterisk indicates significantly different from wild type (t-test, P < 0.01).

Overexpression of PhEXPA1 affects the initiation of branching

Petunia axillary meristems give rise to branches, which develop from basipetal nodes. The transition from vegetative to reproductive development is characterized by the formation of an inflorescence meristem that generates two leaf-like bracts and divides into two portions (Souer et al., 1998), one of which remains as a meristem while the other develops into a flower. When the flower and the inflorescence have grown out, axillary meristems form in the axil of the two bracts (Fig. 6b). These meristems are initially vegetative, forming leaves before they transform into an inflorescence meristem. In wild-type petunia, both meristems remain in a dormant state at the acropetal nodes, whereas starting at the third or fourth node from the apex, the axillary meristems in the axil of the flower and bract grow out (Fig. 6a).

Figure 6.

Altered branching morphology in 35S::PhEXPA1 and 35S::antisense PhEXPA1 transgenic lines of Petunia hybrida (PhEXPA1, Petunia hybrida expansin A1). (a) Branching architecture of whole 35S::PhEXPA1, wild-type and 35S::antisense PhEXPA1 plants. (b) Wild-type and 35S::PhEXPA1 transgenic plants, showing acropetal nodes. Note the premature emergence of the axillary meristem in the 35S::PhEXPA1 transgenic plant.

The timing of lateral branch initiation appeared to be disrupted in 35S::PhEXPA1 transgenic plants; that is, axillary meristem outgrowth was premature and occurred at the second node in both the vegetative (Fig. 4g) and reproductive phases (Fig. 6a,b). The axillary meristem developed into a vegetative shoot that produced leaves before changing into an inflorescence, as in wild-type plants (Fig. 6a). Interestingly, similar alterations affecting organ size and branching were observed in 35S::PhEXPA1-eGFP transgenic plants (data not shown), suggesting that a significant amount of the PhEXPA1-eGFP protein reached its correct subcellular destination and was functional.

Only one axillary meristem develops during natural flower abortion, but in 35S:PhEXPA1 transgenic plants the axillary meristem of both bracts grew out (Fig. S6). Taken together, these results show that the outgrowth of axillary buds, which is normally controlled by apical dominance (Souer et al., 1998), is stimulated by the overexpression of PhEXPA1 during early meristem development. We therefore carried out a more detailed morphological analysis of alterations in the 35S::antisensePhEXPA1 plants described by Zenoni et al. (2004). As anticipated, we found that the timing of lateral branching was delayed and prolonged compared with wild-type plants (Fig. 6a), with the axillary meristem growing out only at the fifth or sixth node from the apex. We also noticed that internodes in 35S::antisensePhEXPA1 plants were slightly shorter than wild-type internodes (Fig. S7). This may facilitate escape from apical dominance, thus leading to axillary shoot development.

PhEXPA1 levels determine the relative amounts of different cell wall polymers

Petal limb abaxial epidermis samples from wild-type, 35S:PhEXPA1 and 35S::antisensePhEXPA1 plants were analysed by FTIR microspectroscopy in the mid infrared range in ATR mode (Griffiths & de Haseth, 1986). FTIR microspectroscopy can be a rapid, easy to use and accurate tool with which to simultaneously study the main biochemical components in plants and plant cells (Schulz & Baranska, 2007). Moreover, the ATR technique was chosen for its cell wall sensitivity (Burattini et al., 2008), because of the low penetration depth of the infrared photon beam inside the sample. In fact, in the present configuration, the infrared photon beam penetrates the sample to a depth of 0.4 μm (at 1800 cm−1) to c. 0.7 μm (at 900 cm−1), less than the cell wall thickness in P. hybrida, which can be estimated to be 1.2 ± 0.2 μm. The 35S::antisensePhEXPA1 line was included to confirm earlier results obtained by FTIR microspectroscopy in transmission mode (Zenoni et al., 2004).

The average and area-normalized absorbance spectra from 1800 to 900 cm−1 for each plant line showed the relative changes in the main biochemical components of the cell wall (Fig. 7a). These were identified after second derivative and by comparison with the literature (McCann et al., 1992; Kacurakova et al., 2000; Wilson et al., 2000; Schulz & Baranska, 2007). The most significant differences involved the bands at 1150–950 cm−1, related to polysaccharides (cellulose, pectins and hemicelluloses), which showed a relative decrease in the transgenic plants with respect to the wild type, and the bands at 1700–1500 cm−1, related to adsorbed water (c. 1645 cm−1) and proteins (amide I at c. 1655 cm−1 and amide II at c. 1550 cm−1), which showed a relative increase in the transgenic plants with respect to the wild type. Interestingly, in the 35S:PhEXPA1 transgenic plants, the 1150–950 cm−1 absorption bands appeared strongly modified in shape, suggesting that the relative polymer levels in these plants had changed.

Figure 7.

Behaviour of infrared absorption bands in wild-type, 35S::PhEXPA1 and 35S::antisense PhEXPA1 transgenic Petunia hybrida plants (PhEXPA1, Petunia hybrida expansin A1). (a) Average and area-normalized Fourier transform infrared (FTIR) spectra of wild-type (blue), 35S::antisense PhEXPA1 (green) and 35S::PhEXPA1 (red) lines. The main differences are seen in the bands at 1150–950 cm−1, related to polysaccharides (cellulose, pectins and hemicelluloses), and at 1700–1500 cm−1, related to adsorbed water (c. 1645 cm−1) and proteins (amide I at c. 1655 cm−1 and amide II at c. 1550 cm−1). (b) PC1 and PC2 score plots after the first derivative on the average spectrum of each petal limb (12 petal limbs from six flowers for each different plant) for wild-type (blue circles), 35S::antisense PhEXPA1 (green triangles) and 35S::PhEXPA1 (red triangles) lines. The percentage of captured variance is also indicated.

PCA was applied to the average area-normalized spectra for each petal limb from each of the plant lines after first derivative. The corresponding PC1/PC2 score plot confirmed that the three lines can be distinguished on the basis of cell wall composition (Fig. 7b). To investigate in more detail the specific spectral differences between the wild type and transgenic lines, PCA was also carried out on the area-normalized spectra of each petal limb for each transgenic line compared with wild-type plants. In both PC1/PC2 score plots (Fig. 8a,b), separation between the wild type and transgenic lines can be seen in the first principal component PC1, whereas corresponding loadings (Fig. 8c,d) indicate that this separation was attributable to a relative reduction in the overall polysaccharide content (as a result of the positive loadings at 1150–950 cm−1) and a relative increase in the protein and water content (as a result of the negative loadings at 1700–1500 cm−1) in the transgenic lines.

Figure 8.

Relative reduction in the overall polysaccharide content in 35S::PhEXPA1 and 35S::antisense PhEXPA1 transgenic Petunia hybrida plants (PhEXPA1, Petunia hybrida expansin A1). (a) PC1 and PC2 score plots on the average spectra of each petal limb in wild-type (blue circles) and 35S::PhEXPA1 lines (red triangles) and in (b) wild-type (blue circles) and 35S::antisensePhEXPA1 lines (green triangles). The percentage of captured variance is also indicated. (c, d) Loadings of PC1 from the principal components analysis (PCA) on (c) the wild-type and 35S::PhEXPA1 lines and (d) the wild-type and 35S::antisensePhEXPA1 lines. Together with the PC1–PC2 score plots, the loadings indicate a relative decrease in polysaccharides (1150–950 cm−1 region) and a relative increase in adsorbed water (c. 1645 cm−1) and proteins (amide I and amide II bands in the 1700–1500 cm−1 region) in the transgenic lines compared with wild type.

These results suggested a deeper, although time-consuming analysis following stepwise chemical fractionation of crude cell wall extracts, to allow the separate measurement of neutral sugars, uronic acid and xyloglucan in the pectin, hemicellulose and cellulose fractions. Measurements of three independent biological replicates showed significant differences in polymer contents among the three lines, and also significant differences in the relative levels of different polymers in each fraction (Fig. 9 and Table S1). Cell wall extracts from 35S::antisensePhEXPA1 plants contained 30–35% less cellulose than wild-type extracts, whereas 35S::PhEXPA1 extracts contained normal levels of cellulose. Extracts from both transgenic lines contained less pectin than wild type, with the antisense line showing the most significant loss. Extracts from 35S::PhEXPA1 plants contained 35% lower levels of hemicelluloses than wild-type extracts, whereas 35S::antisensePhEXPA1 extracts contained normal levels. These data support the distinct shapes of the 1150–950 cm−1 spectral window in the FTIR spectra for plants overexpressing PhEXPA1, and suggest that altering PhEXPA1 expression leads to relative changes in polysaccharide levels. However, differences in extractability cannot be completely ruled out.

Figure 9.

Transgenic 35S::PhEXPA1 and 35S::antisense PhEXPA1 lines show distinct cellulose, pectin and hemicellulose profiles in Petunia hybrida petals compared with wild type (PhEXPA1, Petunia hybrida expansin A1). (a) Cellulose content; (b) pectin content; (c) hemicellulose content. Error bars represent standard errors; = 3 pools of 10 petals. Asterisks indicate significantly different from wild type (t-test, < 0.01).


PhEXPA1 developmental expression and subcellular localization

Plants have many expansin paralogues that are closely related in sequence but they have diverse and complex spatiotemporal expression profiles often regulated by external stimuli such as hormones and light (Hong et al., 2003; Reyes et al., 2004; Jiao et al., 2005). This suggests that the expression profile may be more important than functional differences between proteins when it comes to determining the roles of individual genes (Fleming, 2006). In turn, this indicates that promoter analysis is likely to provide clues as to the function of individual expansin genes, so we started this study by cloning the promoter region of the PhEXPA1 gene and analysing its sequence. Like other expansin promoters (Lee et al., 2001; Cho & Cosgrove, 2002), the 1 kb of PhEXPA1 promoter contains response elements to gibberellin and auxin. The regulation of expansin genes by gibberellins and auxins is well documented mainly from studies on deepwater rice (Oryza sativa; Lee & Kende, 2002), tomato (Catala et al., 2000; Vogler et al., 2003), soybean (Glycine max; Downes et al., 2001), and chickpea (Cicer arietinum; Sanchez et al., 2004). PhEXPA1 promoter activity was studied using a GUS reporter assay, showing a pattern of activity consistent with the spatiotemporal profile determined by real-time RT-PCR (Zenoni et al., 2004). In flowers, GUS activity was detected in petals, sepals, ovaries, styles and stigmas, and was coincident with areas of cell expansion during limb development rather than zones of cell division as mapped by Reale et al. (2002). There was no GUS activity in anthers. In stems, GUS staining was restricted to the nodes at the base of axillary meristems, suggesting that PhEXPA1 has a role in promoting stem elongation and possibly axillary bud outgrowth. These processes are, at least partially, under the control of hormones, mainly auxin and gibberellin (Benkova et al., 2003; Reinhardt et al., 2003; Fleet & Sun, 2005). Although we have not demonstrated the direct influence of phytohormones on PhEXPA1 expression, the presence of corresponding response elements in the PhEXPA1 promoter, together with its restricted expression pattern in the stem, suggests that PhEXPA1 acts downstream of phytohormones in balancing meristem maintenance and organ initiation.

In order to determine the intracellular distribution of PhEXPA1, we generated transgenic plants constitutively expressing a PhEXPA1-eGFP fusion protein. Expansins have previously been localized to the cell wall by immunolocalization and electron microscopy (Cosgrove et al., 2002; Fudali et al., 2008). However, because EXPA-specific antibodies have yet to be developed, it is only possible to look at expansin distribution in a general context, rather than the distribution of specific gene products. Furthermore, in a recent study, Mohanty et al. (2009) showed that an EXPA-YFP fusion protein was correctly targeted to the cell wall in maize (Zea mays). We confirmed using confocal microscopy that the PhEXPA1-eGFP fluorescence signal overlapped with the cell wall-specific calcofluor signal in plasmolysed limb abaxial epidermis tissue, strongly suggesting that PhEXPA1-eGFP is localized in the cell wall. A strong GFP signal was also detected inside the protoplast, although this could reflect the high activity of the constitutive CaMV 35S promoter producing larger quantities of the protein than normally found in planta.

Constitutive overexpression of PhEXPA1 increases organ size and changes the timing of meristem development

The overexpression of PhEXPA1 resulted in morphological changes affecting petal limb size, internode length and leaf area. PhEXPA1 overexpression also increased the extractable expansin activity in leaves, suggesting that changes in petunia expansin activity affect final organ size.

Analysis of the number and size of petal limb cells showed that PhEXPA1 overexpression affected cell enlargement but not cell division. Several previous studies have shown a correlation between the abundance of expansin and organ size, including our own earlier study (Zenoni et al., 2004). Rice plants transformed with sense and antisense OsEXP4 constructs also showed corresponding changes in terms of coleoptile elongation (Choi et al., 2003), and the ectopic expression of PttEXPA1 in hybrid aspen (Populus tremula × Populus tremuloides) increased stem internode elongation and leaf growth through the promotion of cell wall expansion in primary and secondary tissues (Gray-Mitsumune et al., 2008).

The correlation between higher expansin mRNA levels and organ growth was not proportional, as the leaves of the transgenic plants produced c. 150-fold more PhEXPA1 mRNA than wild-type plants but the leaf area increased by just 20%. This suggests that there is a physical limitation to the extent of cell expansion in leaves, or that leaf-specific expansins determine the final size of this organ and that the ectopic expression of PhEXPA1 increases local cell expansion to only a minor extent. This is supported by the observation that PhEXPA1 overexpression had no effect on the expression levels of at least three leaf-specific expansins. The overexpression of PhEXPA1 in the stem induced a 20% increase in stem internode length. We previously reported that PhEXPA1 antisense lines showed no evident reduction of internode length, but the more detailed morphological analysis carried out in the current investigation revealed a c. 15% decrease in internode length in the antisense plants. These results, together with the observation that PhEXPA1::GUS expression is restricted to nodes, indicate that PhEXPA1 not only controls petal limb size, but could also play a role in internode elongation.

The changes in the timing of axillary meristem outgrowth in transgenic plants provide evidence that PhEXPA1 may have a further role in development. The overexpression of PhEXPA1 promoted axillary meristem release, whereas PhEXPA1 silencing generated the opposite phenotype. However, it is possible that the silencing of PhEXPA1 by antisense RNA expression might have a direct or indirect effect on the expression of other members of the expansin family, so it is not possible to attribute the control of branching unequivocally to PhEXPA1. The alteration of branching patterns has been studied by characterizing Arabidopsis more axillary growth (max ) mutants, where homeostasis and transport were impaired for both auxin and strigoloctones (Umehara et al., 2008).

The application of exogenous expansin proteins or the manipulation of expansin gene expression in model plant systems often results in abnormal development (Fleming et al., 1997; Pien et al., 2001; Choi et al., 2003). The altered branching phenotype reported in our study, together with developmental alterations described in other plant species, supports the hypothesis that expansins may have a role in the specification of plant architecture. The altered development of the axillary meristem could reflect the defective regulation of cell enlargement by expansins that may promote the transition from proliferative to expansion-based growth. Alternatively, altered branching may indirectly result from the modification of normal internode length, which in turn affects apical dominance. Petunia meristem maintenance could therefore be regulated by expansin activity directly, through the control of cell expansion during the initial development of the axillary meristem, and/or indirectly, through the determination of internode length and therefore distance from the apex (Fig. 10).

Figure 10.

Model of expansin activity in Petunia hybrida branching pattern determination. Apical dominance inhibits the development of the lateral meristem and decreases towards the base of the plant. Expansin expression decreases towards the tip of the plant and may contribute to axillary meristem outgrowth and/or internode length. Flowers are indicated by closed circles, apical inflorescences by closed triangles, vegetative axillary meristems by open triangles, and bracts by ovals. Smaller size indicates stronger dormancy (redrawn from Souer et al. (1998)).

Changes in the relative levels of cell wall polymers indicate that expansin may promote wall remodelling during cell extension

Current models of expansin activity suggest that they may promote slippage between polysaccharides during turgor-driven cell growth by disrupting noncovalent bonds between cell wall polysaccharides (Cosgrove, 2005). The precise sites and mechanism of expansin activity remain unknown, but expansins bind noncrystalline polysaccharides (McQueen-Mason & Cosgrove, 1995), and synergically enhance the hydrolysis of cellulose by cellulases (Cosgrove, 2005). Many studies show that expansins are involved in cell wall disruption in vivo (Brummell et al., 1999; Cho & Cosgrove, 2000; Powell et al., 2003). We used FTIR microspectroscopy in attenuated total reflection mode to investigate how PhEXPA1 affects the distribution of cell wall polymers. We found that the polymer profiles were altered in both transgenic lines, showing that both the up-regulation and down-regulation of PhEXPA1 had an impact on the substructure of the cell wall. The 1150–950 cm−1 absorption range was particularly informative, showing that overexpression of PhEXPA1 had no effect on cellulose but significantly reduced the amount of pectin and hemicellulose in the cell wall. In comparison, the loss of PhEXPA1 expression reduced the levels of cellulose and pectin, but levels of hemicellulose remained unchanged. Similar results were obtained in tomato plants overexpressing LeEXPA1, a ripening-specific expansin (Brummell et al., 1999). Hemicellulose content was reduced, and the level of polyuronide fell by a moderate amount later in the ripening process. The suppression of LeEXPA1 did not prevent depolymerization of hemicelluloses, but there was a substantial delay in polyuronide depolymerization. LeEXPA1 may therefore play a role in hemicellulose and polyuronide catabolism during ripening by facilitating access to endoglucanase, xylotransglycosylase/hydrolase and polygalacturonase substrate sites. We propose that PhEXPA1 overexpression has the same effect, promoting cell wall loosening by providing greater access to substrate sites for catabolic enzymes. As there is no increase in hemicellulose when PhEXPA1 is suppressed, we conclude that physiological levels of hemicellulose can be maintained by basal amounts of expansin, or that there is a degree of functional redundancy among the petunia expansins allowing other expansins to complement PhEXPA1 activity.

The suppression of PhEXPA1 led to a loss of pectin in the cell wall, and overexpression of PhEXPA1 had a similar albeit less severe effect. We propose that the overexpression of PhEXPA1 reduces pectin levels by exposing more polygalacturonase substrate sites, as is the case for LeEXPA1 (Brummell et al., 1999), whereas the reduction that accompanies PhEXPA1 suppression probably reflects the loss of cellulose, as the ratio between these polymers appears to be important to maintain the structural integrity of the cell wall. The suppression of PhEXPA1 reduces the deposition of crystalline cellulose, resulting in thinner cell walls that are resistant to expansion (Zenoni et al., 2004). Because overexpression does not increase the thickness of the cell wall and its cellulose content, it is likely that there are physical limitations affecting the abundance of cellulose and that cell expansion does not critically depend on the cellulose content. Moreover, we showed that altering PhEXPA1 expression levels does not affect the extractability of cell wall components. The loss of hemicelluloses in plants overexpressing PhEXPA1 could promote cell extension because cellulose is linked with fewer xyloglucans (Takeda et al., 2002), and hydrogen bonds between cellulose and hemicellulose are more easily disrupted.

In summary, we provide direct evidence that the overexpression of a specific expansin changes the cell wall polymer composition and the manner of polymer interactions in expanding tissues. These results support our earlier hypothesis that PhEXPA1 not only promotes wall creep by disrupting the noncovalent bonds between microfibrils and cross-linking glycans, but also affects cellulose deposition during wall expansion. It remains to be determined whether this reflects the direct metabolic activity of EXPA1 on cell wall polysaccharides or indirect changes in other cellular processes that control cell wall deposition.

Accession numbers

The GenBank accession numbers for the genes mentioned in this article are HQ453182 (PhEXP4) and HQ453183 (PhEXP5).


We thank Anne-Marie Digby for her help with the molecular characterization of transgenic plants, Flavia Guzzo for her assistance with light microscopy and Fabio Finotti for technical assistance in the glasshouse. We thank Ivo Rieu for his assistance in the expansin activity data interpretation.