Introduction

Chili pepper (Capsicum annuum) is one of world’s staple vegetables. Variable genetic sources have been developed, and classical breeding programs for pepper cultivation have been well established. However, approaches aimed at the genetic engineering of pepper have been protracted due to the difficulty in transforming pepper by Agrobacterium. While pepper regeneration itself is not an obstacle and is performed routinely (Valera-Montero and Ochoa-Alejo 1992; Ebida and Hu-C 1993; Harini and Sita 1993; Lee et al. 1993), the high level of effort invested into transferring genes into chili pepper explants has not yet yielded a successful and reproducible transformation method. Consequently, relatively few studies have been undertaken to explore methodology (Lee et al. 1993; Kim et al. 1997; Manoharan et al. 1998; S.H. Kim et al. 2001; Cai et al. 2002; Li et al. 2003). Cotyledons or hypocotyls have been the most common source of explants, and the composition of the selection media and culture conditions are similar. Cai et al. (2002) showed the successful transformation of CP genes from CMV and TMV in the same vector and tested resistance against the virus with T3 progeny. The most successful group in the areas of chili pepper transformation, led by K.H. Paek, has transferred several different genes over the past 10 years into pepper plants (CMV satellite RNA: Lee et al. 1993; Kim et al. 1997; CP of CMV and ToMV: Shin et al. 2002a; Tsi 1: Shin et al. 2002b). Although transgenic plants have been tested for functional phenotypes, transformation efficiencies have been very low and the transformation itself was neither consistent nor repeatable (K.H. Paek, personal communication). In addition, successful transformation studies have been reported in several short meeting proceedings (Dong et al. 1995; Szasz et al. 1997; Arous et al. 2001; Y.H. Kim et al. 2001), but no follow-up papers were ever published.

Taken together, the procedures described in the literature for chili pepper transformation are not very helpful in terms of achieving routine transformation. To the contrary, successful cases are regarded as rare events. Two major factors clearly inhibit efficient transformation of chili pepper: (1) the shoot regeneration rate of peppers is genotype-dependent (Christopher and Rajam 1996; Kim et al. 2002) and genotype specificity affects the transformation rate; (2) the very low efficiency of transformation indicates that gene transfer via Agrobacterium infection into cut or injured cotyledon and hypocotyl tissues hardly ever occurs for reasons that are not well understood.

We report here a protocol useful for pepper transformation with a suitable selection method. As the transformed pepper shoots can be obtained from callus-mediated shoots, a strong selection pressure should be exerted so that shoots or multi-shoots growing directly from wounded tissue are eliminated during the selection process.

Materials and methods

Plant materials

Seeds from commercially important pepper (Capsicum annuum) inbred lines (P915, P409, P410, and P101 from the Nong Woo Bio Co, Yeoju, South Korea) were used. Of the 30 inbred lines tested by Kim et al. (2002), these four lines showed a very high rate of regeneration. The seeds were surface-disinfected in 95% EtOH for 30 s and in 50% bleach (Yuhanrox) for 10 min and then rinsed three times with sterilized water. The sterilized seeds were then placed in half-strength MS medium (Murashige and Skoog 1962) and allowed to germinate in under light or dark conditions at 25°C. The cotyledons and hypocotyls from 8- to 10-day-old plants were excised and used as explants.

Pre-culture and inoculation with Agrobacterium

Explants were transferred to a pre-culture medium consisting of MS medium supplemented with 2 mg/l zeatin and 0.05 mg/l NAA or 0.1 mg/l IAA and placed in a growth chamber under light conditions at 25°C for 2–36 h. For transformation, Agrobacterium strain EHA105 or LBA4404 and pCAMBIA 2300 vectors harboring coding regions for the TMV-CP gene (L35074; Park et al. 1997) and the PPI1 gene (AF430372; Lee et al. 2002) were used. Agrobacterium was grown in YEP media supplemented with 50 mg/l kanamycin, 50 mg/l rifampicin and 100 μM AS. The Agrobacterium culture was centrifuged and then diluted with MS to an optical density (O.D.) of 0.3–0.5. This bacterial suspension was then mixed with MS liquid containing 100 μM AS and inoculated into the explants for 10–20 min, co-cultured in the dark for 38–96 h and then washed three times with either 500–800 mg/l cefotaxime or 500–800 mg/l lilacilline.

Shoot formation and regeneration

Explants were transferred to a regeneration medium consisting of a modified MS medium supplemented with 2 mg/l zeatin and 0.05 mg/l NAA or 0.1 mg/l IAA. The hormone compositions and antibiotic concentrations used to select transgenic shoots are described in Table 5. The shoot formation rate was measured by comparing the number of shoots transferred to elongation medium to the total number of explants. Explants were incubated on selection medium (2.0 mg/l zeatin + 0.05 mg/l NAA or 0.1 mg/l IAA) for 6–8 weeks, on elongation medium (2.0 mg/l zeatin + 0.01 mg/l NAA or 0.01 mg/l IAA) for 7–10 weeks, and on rooting medium (hormone-free) for 6–8 weeks. The regenerated plants were acclimated for 2 weeks in zippy pot soil at 25°C under a 16/8-h (light/dark) photoperiod.

PCR, Southern, and Northern blot analysis

Pepper genomic DNA was isolated as previously described (Lee et al. 2002). The primer sequences used in the PCR analyses for detecting the TMV-CP gene insertion were 5′-ATGACGCACAATCCCACTAT-3′ (sense: 35S promoter region at 3,185–3,204 bp of accession no.X84105) and 5′-CGAACCCCTGAAAATAAT-3′ (antisense: TMV-CP gene at 648–631 bp of accession no. L35074); for detecting the PPI1 gene insertion the primer sequences were 5′-ATGACGCACAATCCCACTAT-3′ (sense: 35S promoter region) and 5′-GTACCACTTGAAGAAGC-3′ (antisense: PPI1gene at 587–570 bp of accession no. AF430372). PCR analysis were carried out using 0.65 μM primers, 299 μM dNTP, 1 U/μM Taq DNA polymerase (New England BioLabs, Beverly, Mass.) in 50 mM KCl, 1.5 mM MgCl2, and 10 mM Tris-HCl, pH 8.3. The PCR program consisted of 35 cycles of 94°C, 55°C, and 72°C, each for 1 min. For Southern blot analysis, DNA samples from transgenic peppers (T0) were isolated and 30 μg DNA was digested to completion with restriction enzymes. For Northern blot analysis, total RNA was extracted from pepper leaves (Choi et al. 1996) and 30 μg RNA was used.

Southern and Northern blots were performed as previously described (Church and Gilbert 1984) using Hybond N membranes (Amersham Biosciences, Piscataway, N.J.) and hybridized to random-primed probes following the manufacturer’s instructions (Amersham Biosciences). The probes labeled with [32P]-dCTP were a 678-bp fragment of the TMV-CP gene and a 1,211-bp fragment of the PPI1 gene.

Resistance test for TMV

A total of 408 T1 pepper seedlings (four-leaf stage) segregated from two different T0 plants were initially exposed to TMV by carborundum and then re-exposed to TMV 2 weeks later. The TMV strain (isolated from Korean peppers) was provided by the KT&G Central Research Institute (Daejeon, Korea) (Park et al. 1997). A leaf disk from each T1 plant was taken, and ELISA was performed using the indirect ELISA test as described by Shin et al. (2002b). A reading was done at an absorbance of 405 nm using an ELISA Thermo Max Microplate Reader (Molecular Devices, Sunnyvalley, Calif.).

Results

Direct shoot versus indirect shoot formation

In order to develop a protocol of general use for pepper transformation, we used about 190,000 explants from four different inbred lines and carefully studied the shoot formation process. Two general patterns of pepper shoot formation were identified. First, a shoot or multi-shoots formed directly from the wound or cut region of the explants (direct regeneration) (Fig. 1). This pattern was observed in many cases, and we expected to find transformed shoots among these grown shoots. Five stages of shoot development were observed: direct shoot formation, multi-shoot, elongation, single-shoot elongation, and root formation. In general, 19–26 weeks of shoot development after co-culture was required before that acclimation treatment. Second, shoots were selected from callus tissues that had formed around the cut after 4–5 weeks of culture on the shoot selection medium (indirect regeneration) (Fig.  2). These cases are unusual because calli are not easily formed from the wounded region of cotyledons. Five stages of shoot development were determined: callus formation, callus development, shoot formation, single-shoot elongation and root formation.

Fig. 1a–e
figure 1

Development of direct shoot formation following co-culture. a Shoot formation (5 weeks on selection medium), b multi-shoot formation (7 weeks old), c multi-shoot elongation (9 weeks old), d single-shoot elongation from a multi-shoot (11 weeks old), e root formation (14 weeks old). The red dotted line indicates placement of the cut for removing the single shoot to the next culture

Full size image
Fig. 2a–e
figure 2

Development of indirect shoot formation following co-culture. a Callus formation (5 weeks on selection medium), b callus development (7 weeks old), c shoot formation (9 weeks old), d shoot elongation (11 weeks old), e root formation (14 weeks old). The red dotted line indicates placement of the cut for removing the tissue to the next culture

Full size image

Callus types for shoot formation

We identified six different types of callus that developed from the explants. Shoot regeneration capability was callus-type-dependent. Those able to form shoots were designated type A (white, hard surface and green tissue inside) and type B (dark-green callus with hard surface, but a little bit moist) (Fig. 3). The regeneration rate of type A was about 90% higher than that of type B (data not shown). The callus types that were not able to produce shoots were designated types C (yellow and brittle), D (yellow and hard surface), E (moist and a little transparent), and F (green, moist, and brittle). All six types appeared in the four pepper lines.

Fig. 3
figure 3

Callus types for shoot formation (a, b) and for non-shoot formation (c–f). a Type A, white hard surface and green tissue inside; b type B, dark-green callus with a hard surface but a little bit moist; c type C, yellow and very brittle; d type D, yellow and with a hard surface; e type E, moist and slightly transparent; f type F, green, moist and very brittle

Full size image

Frequencies of direct shoot and indirect (callus-mediated) shoot formation following co-culture

In order to determine the direct shoot formation rate after transformation, 151,700 explants from the four different lines were transformed with the TMV-CPand PPI1 genes. The rate of developing a direct shoot in the shooting medium after co-culture was 5.3% (8,089/151,700) (Table 1). The number of shoots surviving in the rooting medium decreased by a factor of 6 (0.93%: 1,407/151,700). Among the four lines, P915 line showed the highest rate of shoot development.

Table 1 Frequency of direct shoot formation
Full size table

To find the indirect shoot formation rate, we transformed 37,500 explants from four different lines with the TMV-CPand PPI1 genes. The frequency of callus generation from the explants was 1.2% (459/37,500) (Fig. 4, Table 2). However, not all calli produced a shoot. The frequency of shoot development from callus was 11.6% (53/459). The frequency of root formation from the callus-mediated shoots was 52.8% (28/53). Of the four lines, P915 produced 68% (19/28) of the total shoots with root development in the rooting medium, suggesting a genotype preference of pepper regeneration after transformation treatment.

Fig. 4
figure 4

a PCR analysis of peppers transformed with the TMV-CP gene (T0). Lanes: 1–12 transformed, N1–3 non-transformed, P1–2 positive control (cloned bacterial cell and pTMV-CP, respectively). b PCR analysis of peppers transformed with the PPI1 gene (T0). Lanes: 69, 76, 83, 87, 90, 92 transformed, N1–2 non-transformed, P1–2 positive control (cloned bacterial cell and pPPI1, respectively)

Full size image
Table 2 Frequency of callus development and shoot formation from callus
Full size table

Transformation rate

A total of 1,407 direct shoots grown in the rooting media were tested by PCR to identify transformed pepper plants; none of the shoots contained TMV-CP or PPI1 inserts (Table 3). To test the transformation rate of indirect callus-mediated shoots, we analyzed 28 shoots in the rooting medium by PCR (Table 3). The transformation rate was 0.19% (15/37,500) for the P915 line and 0.03% (3/37,500) for the P409 line. However, the transformation rate as determined by the number of PCR-positive versus the number of callus-mediated shoots was 34% (18/53), indicating that shoots grown from the callus could be transformed at a higher probability. Therefore, the selection of a callus-mediated shoot from among a large number of shoots growing on selection medium tended to discriminate transformed plants from non-transformed plants.

Table 3 Transformation efficiency of direct and indirect shoot formation. Percentages were obtained by dividing the number of PCR-positive (+) with the number of explants used
Full size table

Southern and Northern analysis of T0 peppers

DNA samples from transformed peppers (T0) randomly chosen from the PCR-positive peppers were isolated, and 30 μg DNA was digested to completion with XbaI for TMV-CP and with BglII for the PPI1 gene. Figure 5a shows a Southern blot analysis of TMV-CP-transformed peppers digested by XbaI. The transformed peppers showed different TMV-CP insertion sites and those had one copy of the CP gene inserted. Northern blot analysis showed that T0 plants produced TMV-CP transcripts at 678 bp (Fig. 5b) but that non-transformed plants did not synthesize the transcript

Fig. 5
figure 5

a Southern blot analysis of peppers transformed with the TMV-CP gene (T0). Lanes: 5, 63 transformed, N1–2 non-transformed. b Northern blot of peppers transformed with the TMV-CP gene (T0). Lanes: 1, 5, 21 transformed, N non-transformed

Full size image

The PPI1 insert digested with BglII was also localized at different sites in PPI1 transformed peppers (Fig. 6). The 5 kb band on the Southern blot was present in all lanes, and represented that the pepper PPI1 gene internally embedded in the genome, whereas the other inserts represented newly transformed loci. Interestingly, all of the transformed peppers had only one copy of the PPI1 gene inserted.

Fig. 6
figure 6

Southern blot analysis of transformed peppers (T0) with PPI1 gene. Lanes: 69, 76, 83, 87 transformed, N1–2 non-transformed. The 5.0-kb band belongs to the endogenous gene of PPI1

Full size image

Resistance against TMV infection

To examine the resistance efficiency of TMV-CP-transformed pepper against infection by pepper-isolated TMV, T0 plants were self-crossed and a total of 408 T1 plants were exposed to TMV. The leaves were inoculated twice at a 2-week intervals. Two weeks after the second inoculation, ELISA was performed. Leaf patches with a mosaic pattern identified susceptible plants, whereas no mosaic spot was developed on resistant plants. Twenty-eight T1 plants resistant to TMV infection were obtained (Table 4) and 380 T1 plants were susceptible. PCR analysis showed that all of the 28 resistant T1 plants contained the TMV-CP insert (data not shown). The non-transformed control plants that were susceptible did not possess the TMV-CP insert.

Table 4 Resistance test was performed by exposing T1 plants twice to TMV at 2-week intervals. Resistance was determined by ELISA using the absorbance threshold of the control (non-treated) (0.39±0.01)
Full size table

A protocol leading to successful transformation

Several cases of successful transformation protocols are summarized in Table 5. The one we report here is not significantly different from the others previously reported. Our successes were obtained using two lines (P915 and P409) and two different genes (PPI1 and TMV-CP), indicating a recurrence of transformation. Two Agrobacterium strains, EHA105 and LBA 4404, were used preferentially. The duration of pre-culture and co-culture was not necessarily fixed because the transformation was successful as long as the experiments were performed under the conditions described in Table 5. As an auxin source, NAA or IAA could be used in the medium, but the protocol does require the use of zeatin as a cytokinin source. The most important factor in our protocol for pepper transformation is the identification of shoots generated from callus in the selection medium and the elimination of the shoots grown directly. This selection pressure saves time, labor, and cost.

Table 5 A protocol for pepper transformation and selection
Full size table

Discussion

Pepper is known to be a difficult plant to transform and has resisted the efforts of many laboratories for many years. To obtain a successful transformation system for pepper plants, we developed a protocol based on the selection of shoots grown from calli that had developed from cotyledon or hypocotyl tissue. Most of the shoots we obtained seemed to grow well directly near or on the wounded surface of explants on selection medium at a rate that depended on the conditions used. However, we never obtained a transformed pepper plant from shoots grown directly from the explants (Table 3). In contrast, some shoots grew from callus tissues that had formed around the cuts we made on the explants. Shoot formation from callus started 2–3 weeks after callus development (Table 5) and was an unusual phenomen as callus is normally not easily formed from the wounded epidermis. These indirect shoots that grew from the callus showed a high probability of being transformed (Table 3). This difference between the origin of the shoots provides a means of selection that avoids unnecessary testing of non-transformed shoots.

Another important factor for selection was to determine the right callus type because not all calli produced shoots (Table 2). We observed six different callus types (Fig. 3) and of these, type A generated the most shoots—for some yet unknown reason. Consequently, we can state here that a researcher would have the most likelihood of finding putative transformed shoots from a pepper transformation process by looking for a shoot growing on callus type A.

Why were the callus-mediated or indirect shoots transformed while the direct shoots were not? It seems that the infection of Agrobacterium into pepper tissues or cell layers on the wounded surface of explants is not possible. However, the callus cell layer is not the same as the differentiated cell layer, rather it is a non-differentiated cell mass, and it is the altered tissue specificity of this non-differentiated cell mass that may help Agrobacterium invade the cells. This initial callus cell layer would have developed during the pre-culture period and could have provided a place for Agrobacterium to penetrate during co-culture. This hypothesis is reasonable because the transformation of monocot plants such as rice and grass, which initially presented difficulties, was successful with shoots generated from induced calli (Hiei et al. 1994; Rashid et al. 1996; Kusano et al. 2003).

The majority of earlier investigations reported the occasional successful transformation of one or two plants but not continuous success with different genes and lines and the production of many transformed plants (Lee et al. 1993; Kim et al. 1997; Manoharan et al. 1998; S.H. Kim et al. 2001; Shin et al. 2002a,b). However, Cai et al.’s review (2002) describes a large number of T generation progenies that stably expressed CMV-CPor the TMV-CP gene or both CP genes. Since this review does not provide details of transformation efficiency nor describe the factors critical for generating T0 plants, our results cannot be easily compared. However, the general procedure for transformation and culture that is described in this review (Cai et al. 2002) is similar to the data we show in Table 5, although the pH of the MS medium in the former was 7.0 whereas that is our investigation was 5.8.

Our transformation efficiency was low. From the 37,500 explants, 18 peppers were transformed independently with two different genes, TMV-CP and PPI1, into two different pepper lines (Table 3). However, the present report is the first to demonstrate pepper transformation using a new selection tool—i.e. the isolation of a callus-mediated shoot (indirect shoot formation). The total number of T0 peppers originally obtained was 45, but 27 of these were developed by propagating calli (data not shown). Therefore, transformed callus was able to be sub-cultured and the shoots could be grown from the propagated calli. This indicates that a certain T0 plant can be maintained and preserved as a callus form for multiplication. Of the 45 T0 plants 18 were obtained from independent calli by transforming independent explants.

The selection of genotypes for the transformation is also an important factor. Four lines (P915, P409, P410, and P101) that are known to have high regeneration rates (Kim et al. 2002) were used for the transformation, and only two of these lines (P915 and P409) were successfully transformed. The P915 proved to be by far the best line for transformation, as 68% of the transformed peppers were developed from this line (data not shown).

TMV-resistant peppers were selected from among T1 plants after a sequential but time-delayed double inoculation with TMV followed by ELISA (Table 4). Phenotypically no HR (hypersensitive response) or mosaic was found in the resistant peppers, whereas susceptible plants had mosaic leaves. All 28 resistant peppers possessed the insert, and the resistance rate was about 7% (28/408). PCR analysis was carried out with 408 self-crossed T1 plants after TMV inoculation. Of these, 380 plants were susceptible, and of these 380 plants, 315 were PCR-positive (data not shown). However, many transformed peppers that contained the TMV-CP insert were susceptible, whereas non-transformed peppers did not contain the insert. We do not know why so many PCR-positive plants were susceptible, but several hypotheses can be made. First, the gene silencing caused by the inserted gene transcription (RNA-mediated resistance) would not occur: the Northern blot of T0 (Fig. 5b) and some T1 peppers (data not shown) showed a detectable steady-state accumulation of the CP gene. Second, the coat protein that was supposed to interact with a viral RNA to prevent replication (protein-mediated resistance) was not expressed in T1 peppers. For this, coat-protein levels should be examined in TMV-resistant peppers. Third, transgenic plants are generally somaclonal variants. Therefore, the resistance levels might be dependent just on how the TMV-CP gene product segregating in the T1 plants was influenced by somaclonal variation. Very recently, our laboratory observed similar phenomena from transformation experiments with CGMMV and WMV coat-protein genes to watermelon. Many T1 plants contained the CP gene insert, but only some plants of T1 with the CP gene insert were identified as resistant plants (4–5%, data not shown). Further studies with T-generation peppers could support any of these conjectures.

We present here a protocol and a selection method for pepper transformation which is successful and dependable (Table 5). The most important selection pressure for pepper transformation is to maintain the callus-mediated shoots and to eliminate the direct shoots during culture. Recently, our laboratory has been involved in the development of a callus induction system using hormones to monitor the transformation efficiency of Agrobacterium using green fluorescent protein expression. Preliminary data show that the transformation rate of induced calli is dramatically increased. Studies on just how to enhance the regeneration rate from the induced calli are underway.