Homologous recombination-mediated knock-in targeting of the MET1a gene for a maintenance DNA methyltransferase reproducibly reveals dosage-dependent spatiotemporal gene expression in rice

Authors

  • Takaki Yamauchi,

    1. National Institute for Basic Biology, Okazaki 444-8585, Japan
    2. Graduate School of Science and Technology, Chiba University, Matsudo 271-8510, Japan
    Search for more papers by this author
    • Present address: Graduate School of Agricultural and Life Sciences, The University of Tokyo, Tokyo 113-8657, Japan.

  • Yasuyo Johzuka-Hisatomi,

    1. National Institute for Basic Biology, Okazaki 444-8585, Japan
    2. Graduate School of Nutritional and Environmental Sciences, University of Shizuoka, Shizuoka 422-8526, Japan
    Search for more papers by this author
  • Sachiko Fukada-Tanaka,

    1. National Institute for Basic Biology, Okazaki 444-8585, Japan
    Search for more papers by this author
  • Rie Terada,

    1. National Institute for Basic Biology, Okazaki 444-8585, Japan
    Search for more papers by this author
  • Ikuo Nakamura,

    1. Graduate School of Horticulture, Chiba University, Matsudo 271-8510, Japan
    Search for more papers by this author
  • Shigeru Iida

    Corresponding author
    1. National Institute for Basic Biology, Okazaki 444-8585, Japan
    2. School of Pharmaceutical Sciences, University of Shizuoka, Shizuoka 422-8526, Japan
    Search for more papers by this author

*(fax +81 54 264 5503; e-mail shigiid@u-shizuoka-ken.ac.jp).

Summary

Although homologous recombination-promoted knock-in targeting to monitor the expression of a gene by fusing a reporter gene with its promoter is routine practice in mice, gene targeting to modify endogenous genes in flowering plants remains in its infancy. In the knock-in targeting, the junction sequence between a reporter gene and an endogenous target promoter can be designed properly, and transgenic plants carrying an identical and desired knock-in allele can be repeatedly obtained. By employing a reproducible gene-targeting procedure with positive–negative selection in rice, we were able to obtain fertile transgenic knock-in plants with the promoterless GUS reporter gene encoding β-glucuronidase fused with the endogenous promoter of MET1a, one of two rice MET1 genes encoding a maintenance DNA methyltransferase. All of the primary (T0) transgenic knock-in plants obtained were found to carry only one copy of GUS, with the anticipated structure in the heterozygous condition, and no ectopic events associated with gene targeting could be detected. We showed the reproducible, dosage-dependent and spatiotemporal expression of GUS in the selfed progenies of independently isolated knock-in targeted plants. The results in knock-in targeted plants contrast sharply with the results in transgenic plants with the MET1a promoter-fused GUS reporter gene integrated randomly in the genome: clear interindividual variation of GUS expression was observed among independently obtained plants bearing the randomly integrated transgenes. As our homologous recombination-mediated gene-targeting strategy with positive–negative selection is, in principle, applicable to modify any endogenous gene, knock-in targeting would facilitate basic and applied plant research.

Introduction

Although performing various targeted modifications of endogenous genes, including knock-in to monitor the expression of a gene by fusing a reporter gene with its endogenous promoter, are routine practice in mice (Cohen-Tannoudji and Babinet, 1998; Sorrell and Kolb, 2005), gene targeting (GT) by homologous recombination (HR) in flowering plants remains in its infancy (Iida and Terada, 2005; Tzfira and White, 2005; Iida et al., 2007; Johzuka-Hisatomi et al., 2008b). Since the generation of simple knock-out plants by the introduction of a selective drug resistance gene into a target locus, leading to gene disruption, there have only been three reports describing GT of endogenous genes that resulted in the generation of fertile transgenic plants, i.e. one and two genes in Arabidopsis and rice, respectively. These three reports described only one transgenic Arabidopsis plant, with the AGL5 MADS-box regulatory gene disrupted by introducing the nptII gene for kanamycin resistance (Kempin et al., 1997), and several independent transgenic rice plants, with targeted disruptions of the Waxy and Adh2 genes, respectively encoding granule-bound starch synthase and alcohol dehydrogenase, by inserting the hpt gene for hygromycin B resistance (Terada et al., 2002, 2007). In addition, gene-targeted plants carrying modified endogenous genes, the PPO gene for protoporphyrinogen oxidase in Arabidopsis and the ALS gene for acetolactate synthase in rice, were also repeatedly obtained by gene-specific direct selection, through which targeted plants acquired herbicide resistance (Hanin et al., 2001; Endo et al., 2007). Recently, the Arabidopsis Cruciferin gene encoding a seed-storage protein was also reported to be reproducibly disrupted by inserting the visually screenable GFP gene encoding the green fluorescent protein (Shaked et al., 2005). As the GFP reporter gene was intended for use in screening to identify the Cruciferin gene-targeted seeds displaying green fluorescence, such a strategy appears to be powerful for identifying targeted plants, but is unlikely to be widely applicable to monitor the spatiotemporal expression of a gene of interest. Agrobacterium-mediated transformation was used in all of these successful GT events in plants.

Although the analysis of the expression of a reporter gene fused with an appropriate promoter segment in transgenic plants, often termed promoter-reporter gene fusion analysis, is widely used to characterize the cloned promoter sequences, interindividual variation of the reporter gene expression among transgenic plants, which is attributed mainly to the insertion sites and/or copy number of the transgene and epigenetic gene silencing, hampers the proper spatiotemporal evaluation of the promoter activity (Taylor, 1997; Matzke and Matzke, 1998; van Leeuwen et al., 2001; De Buck et al., 2004; Schubert et al., 2004; Butaye et al., 2005). To circumvent such problems, an approach known as a ‘promoter trap’ is also employed: a transformant with a promoterless reporter gene, which is carried by appropriately modified T-DNA sequences or transposons, integrated into an exon of an endogenous target gene in the proper orientation, and resulting in the reporter gene expression by transcriptional fusion, is to be isolated for promoter analysis among the transformants containing the randomly inserted transgenes (Springer, 2000; An et al., 2005). Because the promoter trap is based on random insertional mutagenesis, the isolation of an appropriate promoter-trapped mutant relies on a fortuitous integration of the transgene into the target gene of interest, even though the promoter trap is much more likely to accurately reflect the expression of a gene than promoter–reporter gene fusion. Moreover, the simultaneous integration of additional transgene copies may cause a potential problem, because multiple insertions may complicate the interpretation of spatiotemporal expression patterns (Springer, 2000).

As is the case in mouse (Cohen-Tannoudji and Babinet, 1998; Sorrell and Kolb, 2005), the HR-promoted knock-in targeting strategy is expected to be a powerful tool to monitor the promoter activity accurately in plants. Knock-in targeting appears to offer at least two advantages over the promoter trap: the junction sequence between a reporter gene and an endogenous target promoter can be designed properly, and transgenic plants carrying an identical and desired knock-in allele can be repeatedly obtained. We have developed a reproducible gene-targeting procedure with positive–negative selection, and have succeeded in obtaining fertile transgenic knock-out rice plants with endogenous target genes disrupted by inserting hpt as a positive-selection marker gene (Terada et al., 2002, 2007; Johzuka-Hisatomi et al., 2008a,b). As all of the primary (T0) transgenic knock-out plants examined carried only one copy of hpt with the anticipated structure in the heterozygous condition, targeted T0 knock-in plants with a promoterless reporter gene fused with an endogenous promoter via HR are expected to contain only one copy of the reporter gene in the heterozygous condition. Indeed, we were able to obtain 15 independent transgenic rice plants that have the promoterless GUS reporter gene encoding β-glucuronidase (Jefferson et al., 1987) fused with the endogenous promoter of MET1a, one of two rice MET1 genes encoding a maintenance DNA methyltransferase (Yamauchi et al., 2008), and could compare the GUS staining patterns with transgenic rice plants with the MET1a promoter-fused GUS reporter gene integrated randomly in the genome. Although the results of such a promoter–reporter gene fusion analysis varied in the individual transgenic plants, the spatiotemporal gene expression of GUS was reproducibly observed in a dosage-dependent manner among independently isolated knock-in plants.

Results

Experimental design for knock-in targeting of the MET1a gene, and isolation of transgenic plants carrying a knock-in allele

We chose the MET1a gene as a model gene for knock-in targeting, because no alternative splicing was detected at its 5′ region, including two untranslated exons in different tissues, whereas the 5′ splicing patterns of MET1b can vary in different tissues (Yamauchi et al., 2008). We essentially adapted the same strategy used for the knock-out targeting of Adh2, employing a large-scale Agrobacterium-mediated transformation with strong positive–negative selection (Terada et al., 2007), using a knock-in targeting vector, pJHY-Ki-MET1a, in which the ATG initiation codon of GUS is placed together with the 8-bp SrfI site at the MET1a initiation codon (Figure 1). The targeted double crossovers by two consecutive HR events at the 3.1-kb homologous segments were expected to lead to the fusion of the promoterless GUS reporter gene with the endogenous MET1a promoter. The junction fragments 5′-JF and 3′-JF, which were generated by HR between the homologous segments in the introduced vector and the endogenous target gene, were not only PCR-amplified, but were also sequenced at both ends for confirmation that the PCR-amplified fragments were indeed the anticipated junction fragments (Terada et al., 2007). By identifying the calli producing the junction fragments (Figures 1c and 2a), we were able to obtain 15 independent targeted calli with a frequency of 5.3 × 10−2 per surviving callus with positive–negative selection, and 15 independent transgenic T0 plants were subsequently isolated from the 15 calli. Southern blot analysis of genomic DNAs from these 15 T0 plants with four probes, Ma5, GUS, HPT, and Ma3, revealed that they are all heterozygotes, with only one copy of GUS fused with the endogenous MET1a promoter (the knock-in allele) through HR (Figures 1c and 2b). In AflII digests, the Ma5 probe identified the 9.4-kb wild-type MET1a allele and the 6.0-kb knock-in allele, whereas the GUS probe showed only the latter knock-in allele. Similarly, the Ma3 probe identified the 13.4-kb wild-type MET1a allele and the 7.0-kb knock-in allele in BamHI digests, whereas the HPT probe showed only the 8.7-kb knock-in allele in SphI and HindIII double digests. No bands other than those corresponding to the wild-type and knock-in alleles were detectable, indicating that no additional ectopic events, such as a random integration of an additional copy of the transgene, occurred. Subsequently, seven of those setting more than 150 seeds were chosen for further examination: their knock-in alleles were transmitted to the next T1 generation in a Mendelian fashion (data not shown), and their segregants gave the anticipated Southern blot hybridization patterns (Figure 2c). These results clearly showed that we have succeeded in obtaining independent transgenic plant lines carrying the anticipated knock-in allele. From the heterozygously and homozygously targeted T1 plants, we also obtained selfed T2 progenies.

Figure 1.

 Strategy for the knock-in modification of MET1a in rice.
(a) Genomic structure of MET1a on chromosome 3. The boxes and horizontal arrow indicate MET1a exons and transcriptional initiation, respectively. The green segments are also carried by pJHY-Ki-MET1a. The red vertical arrowhead and the red horizontal bars under the map indicate the oligonucleotides that are 50 bases long and the RNA probes for in situ hybridization in roots and embryos, respectively.
(b) T-DNA region of pJHY-Ki-MET1a. Key: RB, right border; DT-A, diphtheria toxin A fragment gene for negative selection; GUS, β-glucuronidase gene; hpt, hygromycin phosphotransferase gene for hygromycin B resistance; ΔEn, the 3′ terminal part of the maize En element; LB, left border.
(c) Genomic structure of the knock-in targeted MET1a locus. The arrowheads with UF/UR, GF/GR and MF/MR indicate the primers for the RT-PCR analysis of the two 5′ untranslated exons, GUS and MET1a, respectively, and those with 5M1/2 and 5G1/2 indicate the primers for the 5′-RACE analysis of MET1a and GUS, respectively (Table S1). The regions to be amplified for qRT-PCR analysis are indicated by brackets. The 5′- and 3′-JFs indicate the 5′- and 3′-junction fragments generated by HR between homologous segments in the vector pJHY-Ki-MET1a and the endogenous target gene, respectively. The flanking small arrowheads of the 5′- and 3′-JFs, S5F/SGR and SLF/S3R, represent primers for PCR analysis (Table S1). Restriction sites: A, AflII; B, BamHI; H, HindIII; S, SphI. The horizontal dark-gray bars with Ma5, GUS, HPT and Ma3 indicate DNA probes for Southern blot hybridization.
(d) T-DNA region of pTY-MET1a. The 3.1-kb MET1a promoter segment is indicated under the map. The arrowheads with 5PF/GR indicate the primers for the PCR analysis to detect the presence of the entire MET1a promoter-GUS segment integrated randomly in the genome (Table S1).
(e) Junction sequences between the MET1a promoter and the coding sequences of MET1a and GUS. The horizontal arrows indicate the 5-bp perfect inverted repeat containing the SrfI site.

Figure 2.

 Structural analysis of the knock-in targeted MET1a locus.
(a) PCR-based screening for 15 knock-in targeted calli; C, control plasmid pJHY-Ki-MET1aC carrying both 5′-JF and 3′-JF; N, Nipponbare.
(b) Southern blot analysis of 15 primary (T0) knock-in targeted rice plants. The restriction enzymes to cleave the genomic DNA samples and the probes used are indicated without and within brackets, respectively, in each panel.
(c) Southern blot analysis of segregants with the homozygously targeted GUS/GUS (G/G), heterozygously targeted MET1a/GUS (+/G) and homozygously wild-type MET1a/MET1a (+/+) alleles in the selfed progeny of seven T0 plants.

To compare the promoter analysis in the knock-in plants obtained with the promoter–reporter gene fusion analysis in randomly integrated transgenic plants, we also isolated transgenic rice plants with the GUS reporter gene fused with the 3.1-kb MET1a promoter segment (Figure 1d). The junction between the MET1a promoter and GUS in random integration is identical to that in knock-in targeting (Figure 1e). Among the hygromycin-resistant T0 calli obtained, we randomly picked up 120 calli that contained the entire MET1a promoter segment fused with the GUS gene by PCR analysis with the primers 5PF/GR (Figure 1d). Of these, 36 T0 plants were regenerated for further analysis.

Spatiotemporal expression of the GUS reporter gene in knock-in plants

Of the 15 targeted calli obtained (Figure 2a), 10 examined gave uniform GUS-positive staining (Figure 3a), and the remaining five calli showed similar results in separate experiments (data not shown). We employed seven independently isolated transgenic plant lines containing the knock-in alleles (Figure 2c), and examined their histochemical GUS-staining patterns (Figure 3b,c,e,f,h). The reproducible and dosage-dependent expression of GUS was observed in the GUS-staining patterns of T1 and T2 roots (1 month after germination), as well as T1 and T2 embryos (24 h after imbibition of mature seeds), and the GUS-staining patterns in roots and embryos coincided with the in situ hybridization patterns of the MET1a transcripts in wild-type rice (Figure 3d,g). The results indicate that the GUS transgene is stably inherited by subsequent generations with no apparent gene silencing. The GUS-staining patterns of various tissues in T1 segregants, including seed-derived calli, were also examined (Figure 4). Apparent dosage effects were observed between the heterozygous and homozygous T1 segregants (Figure 4a,d,e), and virtually no differences were observed in the GUS-staining patterns of seedling roots, meristems, and flowers among the T1 segregants of the three knock-in-targeted lines examined (data not shown). The GUS staining of pollens within anthers was more obvious in the flowers treated with ethanol than in those without ethanol. As expected, all and half of the pollens in the homozygously and heterozygously targeted plants, respectively, were GUS stained (data not shown). As MET1a is the maintenance DNA methyltransferase associated with DNA replication (Chan et al., 2005), tissues containing actively replicating and dividing cells, such as shoot and root meristems and calli, were heavily stained, whereas no GUS staining was detectable in leaves (Figure 4c). The histochemical results, including ovary and pollen, are consistent with those of the RT-PCR analysis previously reported (Yamauchi et al., 2008). Although Arabidopsis plants with the disrupted MET1 gene showed severe phenotypes (Saze et al., 2003; Xiao et al., 2006; Mathieu et al., 2007), no apparent phenotypes could be detected in the homozygously targeted rice plants. Because the transcripts of MET1b accumulate more abundantly than those of MET1a in all of the tissues examined (Yamauchi et al., 2008), MET1b must play more essential roles than MET1a in rice.

Figure 3.

 Reproducible and dosage-dependent expression of GUS in knock-in targeted calli and plants.
(a) GUS activity of heterozygously knock-in targeted calli (T0); N, Nipponbare.
(b) Roots in T1 plants.
(c) Longitudinal sections of homozygously and heterozygously knock-in targeted roots in the T1 line (L51).
(d) Whole-mount in situ hybridization pattern of MET1a in roots of Nipponbare.
(e) Roots in T2 plants.
(f) Embryos in T1 plants.
(g) In situ hybridization pattern of MET1a in embryos of Nipponbare.
(h) Embryos in T2 plants.
For roots and embryos in T2 plants, simultaneously GUS-stained embryos in the T1 line (L83) and Nipponbare were included for comparison. The symbols G/G, +/G and +/+ are used as described in Figure 2 in the selfed progeny of the heterozygously targeted T1 and T2 plants, whereas G2/G2 represents a T2 plant obtained by selfing a T1GUS/GUS plant; N, Nipponbare.

Figure 4.

 Histochemical GUS staining patterns of various tissues in T1 plants.
(a) Seed-derived calli.
(b) Lateral roots of seedling.
(c) Seedling leaves.
(d) Meristems of mature plants.
(e) Flowers.
To display the GUS expression in seedling leaves and flowers clearly, chlorophyll was removed from the tissue samples, indicated by EtOH in parentheses. Other symbols are as in Figure 3.

Expression of the GUS and MET1a genes in knock-in plants by RT-PCR analysis

To examine whether the reproducible and dosage-dependent GUS expression in knock-in plants properly reflects the MET1a expression, we performed the RT-PCR analysis of the T1 and T2 seedlings (2 weeks after germination), which contained both shoot and root meristems. The dosage-dependent expression of both MET1a and GUS is reproducibly observed in the seedlings of two independently isolated transgenic lines, L83 and L119, and their selfed progenies: the intensity of the bands in homozygous conditions was stronger than that in heterozygous conditions (Figure 5a). Similar results were also obtained reproducibly in the T1 segregants (G/G, +/G, and +/+) of two other lines, L51 and L110, and the heterozygous T1 segregants (+/G) of the remaining L140, L193, and L236 lines (data not shown). The dosage-dependent expression of MET1a and GUS in the independently isolated lines was further demonstrated by quantitative RT-PCR (qRT-PCR) analysis: the plants with homozygous alleles had approximately twice as many accumulated transcripts as those with heterozygous alleles (Figure 5b). The residual MET1a transcripts detected in the homozygously targeted plants appeared to be a mixture of truncated short transcripts, most of which seemed to initiate from the 3′ region to the GUS-hpt insertion site (data not shown). It is noteworthy that the rice Actin 1 promoter-driven hpt gene had been fused with the 35S terminator, together with the 1.0-kb segment at the 3′ end of the maize En element (Figure 1b,c), in order to minimize readthrough transcription from the hpt gene (Terada et al., 2002), and that no such readthrough transcripts were detectable in the homozygously targeted plants (Figure S1). Virtually no GUS transcripts could be detected in Nipponbare and the wild-type segregants (Figure 5b). The 5′ rapid amplification of cDNA ends (RACE) analysis indicated that the GUS transcripts from the targeted knock-in allele contain the same two 5′-untranslated exons as the MET1a transcripts (Figure 1a,c; Yamauchi et al., 2008). Consequently, the two 5′-untranslated exons common to the MET1a and GUS transcripts remain constant (Figure 5a). The results clearly demonstrated that the MET1a gene is intrinsically expressed in a dosage-dependent manner, and that the expression of the GUS gene fused with the endogenous MET1a promoter accurately reflects the expression of MET1a. Moreover, the MET1a disruption did not appear to affect the expression of MET1b.

Figure 5.

 Expression of MET1a and GUS in knock-in targeted plants.
(a) RT-PCR analysis of MET1a and GUS in two knock-in-targeted lines (L83 and L119). RT-PCR analysis of MET1b and CDKB1 was performed with appropriate primers (Table S1). The numerals in parentheses indicate the cycles of PCR amplification.
(b) Dosage-dependent expression of MET1a and GUS assessed by qRT-PCR analysis. The value with the asterisk, which is the average of three and five observed values obtained in T1 and T2 plants, respectively, was set to 1.0 in each set of experiments. Error bars represent SD (n = 3 and 5 for T1 and T2 plants, respectively). Other symbols are as in Figure 3.

GUS staining patterns in transgenic plants carrying the randomly integrated GUS reporter gene fused with the MET1a promoter segment

We also performed the promoter–reporter gene fusion analysis in transgenic rice plants with the GUS reporter gene fused to the 3.1-kb MET1a promoter segment (Figure 1d), integrated randomly in the genome. To this end, we first isolated 120 calli carrying at least one copy of the entire MET1a promoter segment, fused with the GUS gene and the active hpt gene. These calli displayed highly variable GUS activities, and more than half of the calli exhibited stronger GUS staining than the calli containing the heterozygous knock-in allele (Figure S2). We subsequently compared the GUS staining patterns in the roots of 36 independently obtained transgenic plants with those of the knock-in targeted plants in either the homozygous or heterozygous conditions, and classified them into six groups (Figure 6). The GUS staining patterns of 22 plants belonging to groups I–IV were similar to those of the control knock-in targeted plants, whereas eight and six plants in groups V and VI exhibited no GUS activity (−) and aberrant GUS staining patterns (ab), respectively. Of the 22 plants displaying apparently normal staining patterns, i.e. the patterns similar to those observed in the knock-in targeted plants, 11 group-I plants exhibited much stronger GUS activity (+++) than the homozygously knock-in targeted control plant (++), and two group-IV plants showed weaker GUS activity than the heterozygously knock-in targeted control plant (+). Four and five of the remaining nine plants in groups II and III exhibited very similar intensity of GUS staining to the homozygously (++) and heterozygously (+) knock-in targeted control plants, respectively. Thus, interindividual variation was clearly observed in promoter–reporter fusion analysis, which generally agreed with the previous findings (Butaye et al., 2005). We noticed that the GUS activities of the transgenic plants did not always coincide with those of the calli (Figures 6 and S2), which appears to be in contrast with that of the knock-in targeted plants (Figure 3).

Figure 6.

 GUS staining patterns of roots in T0 plants carrying the randomly integrated GUS gene fused with a MET1a promoter segment.
GUS staining patterns of roots in 36 T0 plants carrying the randomly integrated GUS gene with a MET1a promoter segment were compared with those in the knock-in targeted T1 segregants (line L110) bearing the knock-in targeted allele homozygously (++), the knock-in-targeted allele heterozygously (+) and the wild-type allele homozygously (−). Judging from the intensity and patterns of the GUS staining, the T0 plants were classified into six groups (see text). The numerals indicate individual plants regenerated from the corresponding calli numbered in Figure S2. Other symbols are as in Figure 3.

Southern blot analysis revealed that almost all of the transgenic plants carry more than two copies of the GUS gene, integrated randomly (Figure S3). The plants displaying no GUS activity and aberrant GUS staining patterns tended to bear higher copies of the GUS transgene than the plants showing apparently normal staining patterns, and the plants with strong GUS activity (+++) appeared to contain more GUS copies than the others. Southern blot analysis also indicated that three plants (#77, #80, and #114) and only one plant (#104), which showed similar intensity of the GUS staining to the homozygously (++) and heterozygously (+) knock-in targeted control plants, appear to carry two and one copies of the GUS transgenes, respectively.

Discussion

Although two approaches, promoter–reporter gene fusion and promoter trap, have often been employed to monitor the promoter activity of a gene, one of the real and/or potential problems of both techniques is that multiple integrations of the reporter gene may hamper the proper spatiotemporal evaluation of the promoter activity (Taylor, 1997; Matzke and Matzke, 1998; Springer, 2000; De Buck et al., 2004; Schubert et al., 2004; Butaye et al., 2005). Based on the reproducible gene-targeting procedure with positive–negative selection (Terada et al., 2002, 2007; Johzuka-Hisatomi et al., 2008a), we succeeded in obtaining 15 transgenic knock-in rice plants with the GUS reporter gene fused to the endogenous MET1a promoter, all of which carry only one copy of the GUS reporter gene with the anticipated structure in the heterozygous condition, without detecting any associated ectopic events (Figures 1 and 2). This characteristic feature of the transgene integration in our GT procedure has also been observed in the targeting of the Waxy and Adh2 genes (Terada et al., 2002, 2007), and contrasts sharply with the feature of promoter–reporter gene fusion through random integration, in which most of the transgenic plants examined carry more than two copies of the GUS transgene (Figure S3). We have speculated that the presence of the DT-A gene at both ends of the T-DNA segment, adjacent to its border sequences of the targeting vector used in our GT system (Figure 1b), plays an important role in generating such precise GT events without the simultaneous occurrence of ectopic events (Iida and Terada, 2005; Johzuka-Hisatomi et al., 2008a).

In promoter–reporter gene fusion analysis, the GUS staining patterns in the roots of the transgenic plants were variable, and 14 out of 36 plants in groups V and VI exhibited either no GUS activity or aberrant GUS staining patterns, probably because of gene silencing, as they tended to carry relatively higher GUS copies (Figure S3; Butaye et al., 2005). We also noticed that the GUS activities in the calli were sometimes different from those in the roots (Figures 6 and S2). We do not know the exact reasons for the apparent differences in individual cases; gene silencing must play a significant role in some cases. We can further speculate that such a callus might have been a mixture of active and stochastically silenced transformants for GUS expression, and that the regenerants characterized might be derived from a minor transformant. As no such interindividual transformant variation of the GUS expression was observed among independently obtained calli, and their regenerated plants in knock-in targeting (Figure 3), it is clear that the knock-in targeting is far superior to the promoter–reporter gene fusion.

The junction sequence between a reporter gene and an endogenous target promoter can be precisely designed in knock-in targeting, whereas the formation of a junction sequence between a reporter gene and an exon of a target gene in the promoter trap depends on a fortuitous isolation of an appropriate insertion mutant. Indeed, the junction sequence between the endogenous MET1a promoter and the ATG initiation codon of the GUS reporter gene in this study contains only an additional 8-bp SrfI site, GCCCGGGC, compared with the genuine junction sequence between the promoter and the ATG initiation codon of MET1a (Figure 1e). It remains to be examined whether the presence of the 10-bp palindrome in front of the ATG initiation codon of GUS, which could result in the formation of a snap-back structure in the GUS mRNA, has some effect on GUS expression. Nevertheless, the reproducible, dosage-dependent and spatiotemporal expression of the GUS reporter gene could be observed among the independently isolated transgenic rice plants carrying the knock-in allele (Figures 3 and 4): tissues containing actively dividing cells, such as shoot and root meristems and calli, were heavily stained, as is expected because MET1a is the maintenance DNA methyltransferase associated with DNA replication (Chan et al., 2005). The results of dosage-dependent and spatiotemporal GUS staining were also consistent with the results of RT-PCR analysis (Figure 5; Yamauchi et al., 2008). Although slight but significant expression of MET1a was detectable in leaves by RT-PCR analysis, no GUS staining could be detected in the leaves of homozygously knock-in targeted plants (Figure 4c). Presumably the level of GUS expression driven by the MET1a promoter is too weak to confer positive GUS staining in leaves. Although the expression of an endogenous gene in dosage-dependent manner may easily be presumed, it has not been well documented in higher plants. Although dosage-dependent GUS expression, and its stable transmission into subsequent generations without gene silencing, was also reported in transgenic rice lines generated by the site-specific Cre-mediated integration of a single copy of the maize constitutive ubiquitin promoter-driven GUS into a designated genomic site carrying a lox sequence (Chawla et al., 2006), such an experiment was merely intended to minimize the quantitative expression variability of transgenes (Butaye et al., 2005).

It was argued that the promoter trap is more likely to accurately reflect the expression of a target gene than promoter–reporter gene fusion because all regulatory elements should be in place (Springer, 2000). Our HR-promoted gene-targeting strategy with positive–negative selection is, in principle, applicable to the modification of any endogenous gene (Johzuka-Hisatomi et al., 2008b). Thus, the knock-in targeting described here must be better than the promoter trap for monitoring gene expression, because the junction between a reporter gene and an endogenous target promoter can be designed precisely, and because knock-in targeted T0 plants carrying only one copy of the reporter gene with the anticipated structure in the heterozygous condition can be repeatedly obtained. Even when a cis-element located approximately 100 kb upstream of a transcriptional initiation site controlled the expression of a target gene (Chandler, 2007), our knock-in targeting could have detected the state of the element by monitoring the activity of the appropriately inserted GUS reporter gene fused with the target promoter. As the hpt gene residing next to the GUS gene (Figure 1c) carries the highly active rice Actin1 promoter (Figure S1; Terada et al., 2002), however, cis-elements within the Actin1 promoter might affect GUS expression. To exclude such a possibility, it would be feasible to remove the hpt gene with its Actin promoter, by employing site-specific recombination systems, to generate selectable marker-free transgenic plants (Hohn et al., 2001; Gilbertson, 2003). With the same knock-in targeting strategy, the generation of transgenic plants carrying a translationally fused endogenous gene to produce a chimeric protein would also be attainable. By choosing a suitable endogenous promoter and/or gene to be fused with an appropriate transgene, knock-in targeting can serve not only for monitoring gene expression, but also for crop improvement and the production of pharmaceutical substances in plants.

Experimental procedures

Nucleic acid procedures

General nucleic acid procedures, including plasmid preparation, plant DNA and RNA preparation, PCR and RT-PCR amplification, 5′-RACE analysis, and Southern blot and DNA sequencing analyses were performed as described previously (Terada et al., 2007; Johzuka-Hisatomi et al., 2008a; Yamauchi et al., 2008). The RT-PCR analysis of MET1a and MET1b, as well as 5′-RACE analysis of MET1a, were described previously (Yamauchi et al., 2008). RT-PCR analysis of GUS and CDKB1 encoding a B-type cyclin-dependent kinase (Boudolf et al., 2004; Guo et al., 2007), the expression of which is associated with the cell cycle, was performed with primers CDKBF/CDKBR (Table S1). For qRT-PCR analysis, the expression levels of MET1a and GUS relative to CDKB1 were measured with ABI PRISM 7000 using the SYBR Green Master Mix (Applied Biosystems, http://www3.appliedbiosystems.com/index.htm). The thermal cycling conditions used for PCR amplification with appropriate primers (Table S1) were: 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec and 64°C for 1 min. For Southern blot hybridization, the probes Ma5, GUS, HPT and Ma3 were prepared by PCR amplification with the appropriate primers (Table S1).

Vector construction

The knock-in targeting backbone vector, pJHY-Ki2, is a derivative of pINA134 (Terada et al., 2002), which has the promoterless GUS reporter gene (Jefferson et al., 1987) inserted in front of the hpt gene, and carries multiple cloning sites at both ends of the GUS-hpt segment. The knock-in targeting vector pJHY-Ki-MET1a (Figure 1b) and the control plasmid pJHY-Ki-MET1aC for detecting authentic 5′-JF and 3′-JF (Figure 1c) have 3.1- and 3.2-kb HindIII-SrfI fragments containing the MET1a promoter inserted into the HindIII and SrfI cloning sites, between the DT-A and GUS genes, as well as 3.1- and 3.5-kb PmeI-AscI segments containing the 5′ parts of the MET1a coding region inserted into the PmeI and AscI cloning sites, between the hpt and DT-A genes, respectively. As a result, the SrfI site was inserted at the junction between the MET1 a promoter and the GUS coding region (Figure 1e). The cloned DNA fragments of MET1a were prepared from the Nipponbare genome by PCR amplification (Terada et al., 2007) with appropriate primers (Table S1). To avoid the presence of PCR-induced base changes (Johzuka-Hisatomi et al., 2008a), the cloned segments in pJHY-Ki-MET1a were sequenced to verify that their sequences were identical to those of the Nipponbare genome.

In promoter–reporter gene fusion analysis, the vector, pTY-MET1a, is a derivative of pCAMBIA 1305.1 (CAMBIA, http://www.cambia.org), carrying the 3.1-kb MET1a promoter region fused with the GUS reporter gene and the rice Actin 1 promoter-derived hpt gene with the 0.3-kb RbcS-E9 terminator (TE9) (Figure 1d). To construct pTY-MET1a, the 7.7-kb segment of pJHY-Ki-MET1a containing the 1.8-kb rice MET1a promoter, 5′ untranslated exons 1 and 2, together with introns 1 and 2 of MET1a, the GUS reporter gene, and the rice Actin 1 promoter-derived hpt gene coding region was first cloned into the T-DNA region of pCAMBIA 1305.1. Subsequently, a 1.9-kb fragment containing the MET1a promoter upstream region was prepared from the Nipponbare genome by PCR amplification, and replaced the 0.6-kb segment at the 5′ end of the 1.8-kb MET1a promoter carried by the pCAMBIA derivative. The TE9 terminator derived from the pINDEX 4 vector (Ouwerkerk et al., 2001) was cloned into the resultant plasmid carrying the 3.1-kb MET1a promoter segment to yield pTY-MET1a. To avoid the presence of PCR-induced base changes, the substituted 1.9-kb promoter distal segment was sequenced to verify that its sequence is identical to that of the Nipponbare genome.

Plant transformation

In knock-in targeting, Agrobacterium-mediated rice transformation and screening for targeted calli were performed as described previously (Terada et al., 2007): about 107 g (fresh weight) of vigorously growing calli was obtained from 3300 mature rice seeds, and 15 targeted calli, with no additional undesirable ectopic events, out of 284 surviving calli with positive–negative selection could be isolated. Through multiple shoots (Terada et al., 2002), 10–20 transgenic plants were obtained from each targeted callus.

In promoter–reporter gene fusion analysis, Agrobacterium-mediated transformation with the vector pTY-MET1a was performed as described previously (Shimatani et al., 2009). Out of about 5 g (fresh weight) of vigorously growing calli, 144 transformants exhibiting hygromycin B resistance were isolated and subjected to PCR analysis with primers 5PF and GR to screen for transformants containing the 6.2-kb GUS segment, including the 3.1-kb MET1a promoter segment. Of the 120 calli obtained, 36 calli were randomly picked up for regeneration, and one transgenic plant bearing the 6.2-kb GUS segment was obtained from each transformed callus for further analysis.

Histochemical analyses

Histochemical GUS staining was carried out with 1 mm X-gluc (Wako Pure Chemical Industries, http://www.wako-chem.co.jp/english), as described previously (Jefferson, 1987; Shimatani et al., 2009). After treatment with X-gluc, chlorophyll was removed from leaf and flower samples by immersion in 100% ethanol. Imbibed seeds and meristems of mature plants of knock-in targeted lines were cut into halves with a razor, and were histochemically stained for GUS activity. Roots were mounted on a Microslicer DTK-3000W (Dousaka, http://www.kyoto.zaq.ne.jp/dkaih504/), and were then sectioned longitudinally (100-μm thick) for GUS staining.

For whole-mount in situ hybridization (Figure 3d; Küpper et al., 2007) with antisense (or sense) oligonucleotides, 50-bases long (Figure 1a, Table S1) and labeled at the 3′ end with digoxygenin (DIG), roots were fixed in a fixation solution [4% (w/v) paraformaldehyde and 4% DMSO in PBS, pH 7.0] overnight at 4°C. After proteinase K treatment, the tissues were treated with 0.5% acetic anhydride in 100 mm triethanolamine, pH 8.0, for 10 min, and were washed three times for 5 min with distilled water. After prehybridization for 1 h at 50°C in a hybridization buffer (20% formamide, 300 mm NaCl, 10 mm Tris–HCl, pH 7.0, 4 mm EDTA, 0.5% SDS, 1 x Denhardt’s solution, 400 μg ml−1 salmon sperm DNA), hybridization was performed in the hybridization buffer with oligonucleotide probes (100 nm) for 16 h at 50°C. Detection of the hybridized probe was carried out using a commercially available anti-DIG antibody linked to alkaline phosphatase (Kouchi and Hata, 1993). In situ hybridization (Figure 3g) with a DIG-labeled RNA probe (Figure 1a) in embryos was performed according to the published method of Kouchi and Hata (1993), with the following modifications. Hybridization was performed in a hybridization buffer [50% formamide, 300 mm NaCl, 10 mm Tris–HCl, pH 7.0, 1 mm EDTA, 0.1% SDS, 10% (w/v) dextran sulfate, 1 x Denhardt’s solution, 100 μg ml−1 salmon sperm DNA, 100 μg ml−1 yeast tRNA] with RNA probes (0.2 μg ml−1) for 16 h at 45°C. Note that the nucleotide sequence identities between MET1a and MET1b in the oligonucleotide and RNA probes used are 74 and 66%, respectively, and no apparent cross hybridization was likely to occur under the in situ hybridization conditions employed.

Acknowledgements

We thank Miwako Matsumoto and Hisayo Asao for technical assistance, Pieter Ouwerkerk for providing pINDEX4, Akemi Ono, Satoru Moritoh, and Atsushi Hoshino for discussion, and Masahiro Mii, Mikio Nakazono, Hirokazu Kobayashi, and Hiroshi Noguchi for encouragements. The work was supported by grants from the Program for Promotion of Basic Research Activities for Innovative Biosciences (PROBRAIN) from Bio-oriented Technology Research Advancement Institution (BRAIN) in Japan (to SI), from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (No. 17207002 to SI and No. 18-40089 to YJ-H). YJ-H was supported by an RPD fellowship from the Japan Society for the Promotion of Science (JSPS).

Ancillary