Introduction

An important horticultural quality criterion is a compact growth habit. However, many important ornamental plants exhibit elongated growth in their natural habitat. Hence, in commercial production, plant growth is controlled by the application of chemical compounds (Rademacher 2000). However, the use of efficient growth retardants has been prohibited due to potential harmful effects on the environment and human health (De Castro et al. 2004; Sørensen and Danielsen 2006). Consequently, there is an urgent need for producers of potted ornamental plants to find alternatives to chemical growth regulation. Stable compact plants can be obtained by molecular breeding methods using cisgenic and transgenic approaches (Lütken et al. 2010, 2011) or by physical growth retarding methods using variations in temperature, light and mechanical stress (Clifford et al. 2004; Latimer and Thomas 1991). However, a complete solution for the avoidance of chemical growth retardants has not been obtained and although promising results have been achieved through molecular breeding methods in Kalanchoë, the use of these plants is restricted due to GMO legislation in several countries (Lütken et al. 2012a).

Transformation with the naturally occurring soilborne bacterium Agrobacterium rhizogenes is a biotechnological breeding method to obtain compact plants. Plants produced by this strategy are independent of recombinant DNA techniques and according to present legislation not classified as GMOs in Denmark (EU 2001). Agrobacterium rhizogenes is a plant pathogen that gives rise to the development of characteristic “hairy roots” at the infection site. The root inducing (Ri) phenotype is the result of transfer, integration and expression of bacterial transfer-DNA (T-DNA) in the transformed plant. In agropine strains of A. rhizogenes, the T-DNA is split into left (TL) and right (TR) fragments. The TL-DNA comprises 18 open reading frames (ORFs) of which four r oot o ncogenic l oci (rol-genes), termed rolA, rolB, rolC and rolD are the major determinants for the obtained Ri phenotype. The TR-DNA contains six ORFs of which the auxin-homeostasis genes, designated aux1 and aux2, have been characterised (Bulgakov 2008; Ozyigit et al. 2013). The two T-DNA’s are separated from each other by about 15 kb of non-T-DNA and they are transferred and integrated independently (Chandra 2012; Vilaine and Casse-Delbart 1987).

Kalanchoë has been successfully transformed with A. rhizogenes, regenerated and characterised in subsequent F1 and F2 generations (Christensen et al. 2008; Lütken et al. 2012c). These plants exhibit characteristics such as short internodes, increased branching and compact stature. These traits are all useful for quality improvement in ornamental plants (Casanova et al. 2005; Lütken et al. 2012a).

The horticulturally important Campanula genus comprises more than 300 herbaceous species found throughout the northern hemisphere. Many species and hybrids belonging to this genus are used as both garden plants and indoor potted plants. Although different species of the Campanula genus vary greatly in terms of flower size and morphology, only a few species, e.g. C. portenschlagiana, C. poscharskyana and C. carpatica, are produced in larger quantities as potted plants. Molecular knowledge has been obtained through transformation and regeneration protocols of C. carpatica and C. punctata (Sivanesan et al. 2011; Sriskandarajah et al. 2008) and the closely related medicinal plant Platycodon grandiflorus (Park et al. 2011).

In the present work, we describe the utilisation of a non-GM breeding approach using a wild type agropine strain of A. rhizogenes to transform Campanula and subsequently regenerate plants. This is an example of a transformation method for ornamental plants, with the aim to develop unregulated compact plants.

Materials and methods

Plant material and sterilisation

Plants in the vegetative stage of C. portenschlagiana Schultes ‘Blue GET MEE®,(Cp), C. punctata Lam × C. takesimana Nakai ‘Pretty MEE®, (Chybr) and C. glomerata L. ‘Glory MEE®, (Cg) were obtained from Gartneriet PKM A/S, (Odense, Denmark). The plants were kept in a growth chamber with 8 h light at an intensity of 110 µmol m−2 s−1 in a 20 °C day/18 °C night cycle. Prior to transformation, plants were placed in darkness for 7 days to develop signs of etiolation. On the day of transformation, petioles (2–3 cm long) showing signs of etiolation were excised and sterilised in 70% (v/v) EtOH for 1 min., followed by 10 min. in 1.4% (v/v) NaOCl, 0.03% (v/v) polysorbate 20 (Tween® 20, Sigma-Aldrich, MO, USA) and rinsed three times in sterile water.

Transformation with A. rhizogenes

The A. rhizogenes used in this study was A4, containing plasmid pRiA4 (Jouanin et al. 1987; Slightom et al. 1985). In all experiments performed, bacteria were grown in MYA (0.4% mannitol, 0.5% (w/v) yeast extract, 0.05% (w/v) casamino acids and 0.2% (w/v) ammonium sulfate, pH 6.6), at 180 rpm and 28 °C. Bacteria were sub-cultured every 16 h and always initiated at OD600 = 0.03 to keep the culture in log-phase. A 100 ml culture started at OD600 = 0.03 would in a 16 h period (O/N) usually result in an OD600 = 0.4. This culture was diluted to OD600 = 0.2 and 74 µM acetosyringone (AS, Sigma-Aldrich) was added. Following an additional 4 h of cultivation, the culture was ready for transformation.

Bacterial inoculation was performed at OD = 0.4 in MYA. Explants were sterilised Campanula petioles, where the bleached petiole ends had been removed, producing fresh wounding sites. Explants were placed in the bacterial solution for 30 min with gentle agitation. As controls, explants submerged in MYA without bacteria were used. Following transformation, explants were dried on sterile filter paper and subsequently co-cultivated with the bacteria at 20 °C in darkness for 3 days on the basic medium (BM; 0.22% (w/v) Murashige and Skoog including vitamins (MS), 3% (w/v) sucrose, 0.7% (w/v) plant agar and pH 6.3. After autoclaving, AS was added to a final concentration of 74 µM. Following co-cultivation, bacterial growth was reduced by dipping the explants in 100 mg L−1 ticarcillin disodium/clavulanate potassium (15:1 mixture, timentin). Explants were subsequently moved to root growth media (RGM; 0.22% (w/v) MS, 1.5% (w/v) sucrose, 0.05% (w/v) 2-(N-morpholino) ethanesulfonic acid (MES), 0.2% gelrite, pH 6.3 and 100 mg L−1 timentin was added after autoclaving).

Regeneration

Explants were cultivated on RGM for approximately 6 weeks in darkness in 20 °C day/18 °C night. Transformation efficiencies were quantified at 2, 4 and 6 weeks after transfer to RGM. Hairy roots which developed to 20 mm or longer were excised from petioles and placed on regeneration medium (REM) containing 42 µM 6-(γ,γ-dimethylallylamino)-purine (2iP) and 0.67 µM 1-naphthaleneacetic acid (NAA) for induction of shoots. As part of developing procedures for plant regeneration, plant growth regulators were tested in combinations using cytokinin ± 0.67 µM NAA. Thidiazuron, kinetin, zeatin, dihydrozeatin or 6-benzyl-aminopurine were applied in the range of ~0.90–23 µM, N-(2-chloro-4-pyridyl)-N′phenylurea in the range of 0.80–34 µM and 2iP in the range of 1.0–25 µM. Once plant growth regulator treatments were initiated, roots were moved to 8 h photoperiod, 20 °C day/18 °C night with a light intensity of 120–150 µmol m−2 s−1. Cg plantlets with 4–6 leaves developing in response to 42 µM 2iP and 0.67 µM NAA were transferred to plant boxes containing plant growth medium (PGM; 0.23% (w/v) MS modification No. 1B (Murashige and Skoog 1962), 3.0% (w/v) sucrose, 0.05% (w/v) 2-(N-morpholino) ethanesulfonic acid (MES), 0.2% gelrite, pH 6.3. After autoclaving 0.67 µM NAA was added). Following 2–3 months of culturing on PGM, plantlets were transferred to pots containing autoclaved peat (Pindstrup Substrate no. 1, Pindstrup Mosebrug A/S, Kongerslev, Denmark). High humidity around the transferred plantlets was ensured by covering with the plant box lid for 1 week. In total, 14 independent Cg lines produced plantlets and 10 of these lines survived the transfer to soil. All 14 lines were analyzed by PCR to determine their genotype. MS (M0222, M0233), MES, plant agar, gelrite, antibiotics and hormones were purchased from Duchefa Biochemie B.V., Haarlem, The Netherlands.

PCR reactions

DNA was extracted from representative hairy root lines of Cp and Chybr and from 14 regenerated plant lines of Cg to verify the transfer of the pRiA4 T-DNA TL and TR fragments. Campanula ACT was used as a positive control (Jensen et al. 2016). Primers targeting rolB and aux2 were used as indicators for the presence of TL- and TR-DNA, respectively (Lütken et al. 2012b). The presence of VirD2 was used to detect any residual bacteria remaining on hairy roots and generated plants (Christensen et al. 2009). PCR reactions were performed using Ex taq polymerase (Takara, Japan) according to manufactures recommendations but with 2% (v/v) DMSO in the final reactions. PCR program for ACT: 4 min 94 °C, 35 cycles of [30 s 94 °C, 1 min 60 °C, 1 min 72 °C] and finally 7 min 72 °C. The same conditions were used with rolB, aux2 and virD primers but with primer annealing temperatures of 55.6, 55.6 and 63 °C respectively. In each PCR run, positive and negative controls were the pRiA4 plasmid and H2O, respectively. PCR product sizes are ACT; 196 bp, rolB; 182 bp. aux2; 233 bp and virD2; 445 bp. Due to the migration of virD2 PCR products in Fig. 3, all virD2 products were verified by sequencing (Eurofins Genomics, Germany).

Statistical analysis

Statistical analysis was done in SigmaPlot v. 13 (Systat Software, San Jose, CA, USA) by pairwise one way analysis of variance using the Holm-Sidak method. Statistically significant differences among time points, treatments or species were defined as P ≤ 0.05.

Results

Three Campanula species; Cp, Chybr and Cg were transformed by A. rhizogenes. Petioles of vegetative soil grown plants, showing signs of etiolation, were sterilised and used as explants. The first indications of transformation could be detected after 2–4 weeks as hairy roots developing at the ends and on the sides of the petioles (Fig. 1a, b). In contrast, hairy roots did not develop on controls (Fig. 1c, d; Table 1). Transformation efficiencies were evaluated on hairy roots attached to explants 2, 4 and 6 weeks after transformation. The transformation response was found to be species-specific with 90% of Cp and Chybr explants producing hairy roots [89.6 and 89.9%, respectively (Table 1)]. Cg hairy root development was the least effective with only 32.9% explants producing hairy roots and a 2 weeks delay in root development was observed when compared to the other Campanula explants (Table 1). Among the transformed hairy root lines we observed differential root growth patterns. While some roots grew vigorously with multiple lateral roots, others were slow-growing single roots without lateral roots or root hairs. Only vigorously growing hairy root lines were used in subsequent experiments.

Fig. 1
figure 1

Root development of Campanula transformed by A. rhizogenes. C. portenschlagiana (Cp) is used as a representative to illustrate hairy root development in response to A. rhizogenes transformation. Cp root development 2 (a) and 4 weeks (b) after transformation. Root development on control explants treated with media not containing bacteria after 2 (c) and 4 weeks (d)

Full size image
Table 1 A. rhizogenes transformation efficiencies in different Campanula species
Full size table

Plant regeneration from hairy roots in response to plant growth regulator treatments were dependent on the specific Campanula species. Roots of Cp and Chybr produced calli, but rarely produced chlorophyll (Fig. 2a, b). Regardless of treatment, with various cytokinins with or without auxin, no plants could be regenerated from Cp or Chybr hairy roots. In contrast, hairy roots originating from transformed Cg developed green calli in response to 42 µM 2iP and 0.67 µM NAA. Shoot development occurred from green calli with a frequency of up to 10%, calculated as the number of obtained regenerated plant lines per number of petioles included in the transformation (Fig. 2c–e).

Fig. 2
figure 2

Developing calli and plant regeneration of Campanula transformed by A. rhizogenes. a, b Cp callus cultured on regeneration medium containing 42 µM 2iP and 0.67 µM NAA (REM). c Cg calli producing chlorophyll which in some cases produced regenerated plantlets when cultured on REM. d, e Regenerated Cg plantlets transformed with A. rhizogenes placed on plant growth medium (PGM). f Cg plantlet before transfer from PGM to soil. g, h Transformed Cg plants growing in soil

Full size image

Shoots with 4–6 leaves were transferred as plantlets to cytokinin free PGM medium in plant boxes (Fig. 2f). 2–3 months later plantlets could be transferred to autoclaved soil. Among 14 physically independent Cg hairy root lines producing shoots, 10 lines survived the transfer to soil (Fig. 2g, h). PCR was used to verify the transfer of bacterial T-DNA to hairy root lines (Cp and Chybr) and for all regenerated plant lines of Cg. Bacterial residue was detectable on some hairy root lines of Cp and Chybr when DNA was extracted 8 weeks after transformation (Fig. 3g, h), but was not found on plantlets of Cg after multiple sub-cultivations (Fig. 3i). All tested hairy root lines of Cp and Chybr had rolB integrated in their genome (Fig. 3a, b). However, aux2 was only present in some hairy root lines of Cp and Chybr (Fig. 3d, e). In Cg, DNA was extracted from 14 plantlets regenerated from hairy roots, in total. Results from six representative lines of Cg are presented in Fig. 3c, f, i, l, lanes 1–6. Off the 14 lines, two were found to be negative for the presence of rolB (Fig. 3c) even though they originally developed from hairy roots. Also, in contrast to results obtained in Cp and Chybr, for hairy root lines, aux2 could not be detected in any of the Cg plantlets (Fig. 3f). Of the 14 Cg hairy root lines producing plantlets, genotypes all were tested by PCR and none of these where tested positive for aux2.

Fig. 3
figure 3

Detection of rolB and aux2 in transformed hairy roots or regenerated plants from Campanula transformed with A. rhizogenes. Representative hairy root lines of C. portenschlagiana (Cp; a, d, g, j) and C. takesimana × C. punctata (Chybr; b, e, h, k). In the case of C. glomerata, 14 regenerated plant lines were PCR tested, six of these lines are included (Cg; c, f, i, l). rolB (a–c) and aux2 (d–f) represents TL- and TR-DNA, respectively. virD2 (g–i) determine any residual bacteria on explants. As controls ACT (j–l), pRiA4 (P) and H2O (H) were included when appropriate. PCR product sizes are ACT; 196 bp, rolB; 182 bp. aux2; 233 bp and virD2; 445 bp. The position of the 500 bp band in the DNA ladder is marked using a black bar, bands below 500 bp occur in 100 bp increments

Full size image

Discussion

In this study, we have established a method for transformation with A. rhizogenes of three morphologically distinct Campanula species; C. portenschlagiana (Cp), C. takesimana × C. punctata (Chybr) and C. glomerata (Cg). Earlier work on C. carpatica plant regeneration from A. tumefaciens transformed tissue (Sriskandarajah et al. 2004) found a higher response to hormone treatments when etiolated starting material was used. Similarly, dark treatment had a positive effect on regeneration of plants from Buddleia leaves (Dai and Castillo 2007). Hence, a 7 day dark treatment period prior to the transformation event was included in the method described here.

Efficient transformations of Campanula with A. rhizogenes have not been previously reported, but similar results have been observed in Hibiscus rosa-sinensis (60.5%) and Catharanthus roseus (85%) (Christensen et al. 2009; Zhou et al. 2012). As in Hibiscus, control explants of Campanula, not exposed to A. rhizogenes, did not produce hairy roots (Fig. 1c, d; Table 1). Therefore, all developing roots are theoretically derived from a successful transformation event. However, as hairy root development is responsive to a combination of changes in auxin sensitivity and biosynthesis (Lima et al. 2009) and because hairy roots have been reported to show elevated auxin levels (Chriqui et al. 1996), it is possible that development of non-transformed roots may have been stimulated by auxin exudates from neighboring transformed explants with vigorously growing roots. This was, however, not verified by PCR as these root lines were omitted in subsequent sub-cultivations and regenerations.

Successful plant regeneration from hairy roots is dependent on plant species, plant growth regulator treatment and abiotic factors such as light, temperature and growth medium (Christensen et al. 2008; Zhou et al. 2012). Molecular and environmental factors influencing plant regeneration are comprehensively reviewed by Ikeuchi et al. (2016). A major obstacle in regeneration from hairy roots of A. rhizogenes transformations is that for each new plant species included, the optimal combination and concentration of plant growth regulators has to be identified. In the present study, we found variations within the Campanula genus to be a determining factor for plant shoot regeneration. Cp and Chybr were recalcitrant to regeneration in response to plant growth regulator treatments, whereas Cg developed shoots in response to REM containing 42 µM 2iP and 0.67 µM NAA. A range of other shoot regeneration media were tested, however none showed to be successful. This supports earlier results showing that variation in response to exogenous phytohormones can be high and that transformation procedures in new plant genera should include several species and/or cultivars to maximise the chance of successful regeneration (Choi et al. 2004; Jian et al. 2009; Majumdar et al. 2011). Earlier work with Campanula supports that some cultivars are notoriously difficult (Sriskandarajah et al. 2004) or not (Seglie et al. 2011; Sriskandarajah et al. 2008) to cultivate in vitro when transformations have been done with A. tumefaciens. Thus, there is still a need for development of protocols to facilitate efficient in vitro culture and regeneration.

We investigated the genomic integration of TL- and TR-DNA via PCR amplification of rolB and aux2, respectively. As the hairy root phenotype is primarily a result of integration and expression of rolA-C genes, not surprisingly, all hairy roots tested from Cp and Chybr were positive for TL-DNA (Fig. 3a, b). However, aux2 located on the TR-DNA was not equally present (Fig. 3d, e). Transformations with wild type A. rhizogenes generally use hairy roots as phenotypical markers during selection whereas there is no clear marker trait specific for TR-DNA. This could explain why we did not observe plant lines solely containing TR-DNA (Fig. 3). Similar results were obtained in studies of Catharanthus roseus where TR-DNA always integrated with TL-DNA in transformed plant lines (Taneja et al. 2010; Zhou et al. 2012). Interestingly, our results show that even though Cg plantlets were generated from hairy root lines some lines contained neither TL- nor TR-DNA (Fig. 3c, f). Since controls for the transformation did not produce hairy roots (Fig. 1c, d), the T-DNA inserts may have been lost in the regenerated plant lines during in vitro culture—possibly in the callus phase. Furthermore, the lack of TR-DNA in Cg (Fig. 3), could indicate an uncharacterised selection pressure against TR-DNA specifically in Cg. Similar results were observed in root cultures of coffee where TR-DNA was not integrated among 55 tested root clones (Alpizar et al. 2008). As transformed Cg was the only Campanula producing plantlets in response to plant growth regulator treatments, the presence of TR-DNA in Cp and Chybr may represent a critical barrier in plant regeneration which should be considered in future studies when transforming with wild type A. rhizogenes. This could be due to TR-DNA localized auxin synthesis genes resulting in a changed auxin to cytokinin ratio which could be unfavorable to shoot regeneration (Skoog and Miller 1957).

This study adds the ornamental plant Campanula to the list of plant genera which can be efficiently transformed with wild type A. rhizogenes using solely hairy root development as a selective marker. The utility of wild type A. rhizogenes transformation as a biotechnological method to obtain stable compact plants is highly dependent on the final plant status as a GMO or non-GMO (Ferrante et al. 2015; Lütken et al. 2012a). With respect to plant products resulting from transformation using wild type A. rhizogenes, the occurrence of partial fragments originating from Agrobacterium T-DNA in the genomes of wild plants is of pivotal importance for claiming that plants produced by the current breeding approach are non-GMO. Recently, both the wild Linaria vulgaris and cultivated sweet potato have been reported to contain genes homologous to Agrobacterium T-DNA (Kyndt et al. 2015; Matveeva et al. 2012). Also, homologues to rol-genes, termed Ngrol, have been identified and characterised thoroughly in Nicotiana species (Intrieri and Buiatti 2001). As sequencing of plant genomes and transcriptomes becomes widespread, it is plausible that additional natural events of transformation will be discovered.

Conclusions

We describe a new transformation method for Agrobacterium rhizogenes of Campanula. Our results obtained from regenerated C. glomerata present new insight of TL- and TR-DNA transfers to transformed plants.