- © 1999 American Society of Plant Physiologists
Abstract
The Rp1-D gene for resistance to maize common rust (Puccinia sorghi) is a member of a complex locus (haplotype) composed of Rp1-D and approximately eight other gene homologs. The identity of Rp1-D was demonstrated by using two independent gene-tagging approaches with the transposons Mutator and Dissociation. PIC20, a disease resistance (R) gene analog probe previously mapped to the rp1 locus, detected insertion of Dissociation in an Rp1-D mutation and excision in three revertants. Independent libraries probed with the PIC20 or Mutator probes resulted in isolation of the same gene sequence. Rp1-D belongs to the nucleotide binding site, leucine-rich repeat class of R genes. However, unlike the rust resistance genes M and L6 from flax, the maize Rp1-D gene does not encode an N-terminal domain with similarity to the signal transduction domains of the Drosophila Toll protein and mammalian interleukin-1 receptor. Although the abundance of transcripts of genes from the rp1 complex changed with leaf age, there was no evidence of any change due to inoculation with avirulent or virulent rust biotypes. A set of 27 Rp1-D mutants displayed at least nine different deletions of Rp1-D gene family members that were consistent with unequal crossing-over events. One mutation (Rp1-D*-24) resulted in deletion of all but one gene family member. Other unique deletions were observed in the disease lesion mimic Rp1-D*-21 and the partially susceptible mutant Rp1-D*-5. Different rp1 specificities have distinct DNA fingerprints (haplotypes). Analysis of recombinants between rp1 specificities indicated that recombination had occurred within the rp1 gene complex. Similar analyses indicated that the rust R genes at the rp5 locus, 2 centimorgans distal to rp1, are not closely related to Rp1-D.
INTRODUCTION
The rp1 locus for resistance to maize common rust maps to the distal end of the short arm of maize chromosome 10 (Rhoades, 1935). Fourteen different resistances have been given the rp1 designation on the basis of map position (Saxena and Hooker, 1968), and a number of these have been genetically recombined, suggesting that they are encoded by members of a gene cluster (Saxena and Hooker, 1968; Hulbert and Bennetzen, 1991). Different rp1 genes spontaneously mutate to susceptibility at frequencies between 0.016 and 0.5% (Pryor, 1987; Bennetzen et al., 1988). It has been proposed that the instability of rp1 genes is due to gene conversion or unequal crossing-over events between mispaired sequence repeats at the rp1 locus during meiosis (Sudupak et al., 1993; Hu and Hulbert, 1994). Mutants of rp1 also include disease lesion mimics (Hu et al., 1996) and mutants with novel resistance specificities (Richter et al., 1995). The molecular analysis of these mutants promises to shed light on the processes underlying resistance and the way in which natural plant populations generate variability at resistance loci.
During the last 6 years, a number of gene-for-gene–type plant disease resistance (R) genes have been isolated (reviewed in Baker et al., 1997). The majority of these genes encode a putative nucleotide binding site (NBS) and a leucine-rich repeat (LRR) region. NBS-LRR R genes have been isolated from a range of monocot and dicot plant species and confer resistance to bacteria, viruses, fungi, nematodes, and insects (Baker et al., 1997; Milligan et al., 1998; Rossi et al., 1998; Vos et al., 1998). Conserved amino acid motifs in and around the putative NBS of NBS-LRR resistance proteins have facilitated the polymerase chain reaction (PCR) amplification of R gene–like sequences from plant genomes (Kanazin et al., 1996; Leister et al., 1996, 1998; Yu et al., 1996; Aarts et al., 1998; Shen et al., 1998; Speulman et al., 1998). Using this approach, we identified a sequence that hybridized with a small gene family at the rp1 locus (Collins et al., 1998). In this study, we report the identification of the member of this gene family that confers the Rp1-D rust resistance specificity by using two independent transposon-tagging procedures. We also report that rp1 transcription is unaffected by rust inoculation, describe deletion mutants of Rp1-D derived from probable unequal crossing-over events, and show that the rp5 rust resistance gene located 2 centimorgans distal to rp1 is not closely related to Rp1-D.
RESULTS
Tagging with Mutator
Mutator (Mu) tagging experiments were designed to generate mutants of the Rp1-A, Rp1-B, or Rp1-D R genes. Plants heterozygous for rp1 genes (Rp1-A/Rp1-D and Rp1-B/Rp1-D) and containing multiple active Mu elements were pollinated by Rp1-J homozygotes to produce two families (1 and 2, respectively), and the progeny were screened with the rust biotype IN2, which is virulent on Rp1-J–containing plants but not on plants containing Rp1-A, Rp1-B, or Rp1-D. In family 1, 30 susceptible plants were identified from ~100,000 plants; in family 2, 27 susceptible plants were identified from ~45,000 plants.
To distinguish potential Mu insertions from recombination events between the R genes present in the heterozygous parents, we scored each susceptible individual for the restriction fragment length polymorphism (RFLP) markers bnl3.04 and ksu3, which closely flank the rp1 locus (Hulbert and Bennetzen, 1991). Twenty-seven susceptible plants derived from family 1 were recombinant for flanking markers, whereas the remaining three (NCO1, NCO2, and NCO3) contained flanking marker alleles from the Rp1-D chromosome, indicating that they were derived from the Rp1-D gene by non-cross-over events (data not shown). Of the susceptible individuals identified in family 2, one was missing both flanking markers from Rp1-B and Rp1-D (suggesting that it was the result of a large deletion), 23 were recombinant, one had the flanking marker alleles of the Rp1-B parent, and the remaining two both contained the flanking marker alleles of the Rp1-D parent (NCO4 and NCO5; data not shown). The NCO1, NCO2, NCO3, and NCO4 individuals were crossed to maize lines homozygous for detectable rp1 alleles (Rp1-J or Rp1-C) but lacking Mu elements, and the hybrids were backcrossed one or two times to these lines to reduce the number of Mu elements. Backcross families segregating for the mutant alleles were then subjected to RFLP analysis with Mu probes, and in one family of 90 progeny segregating for NCO 4, a HindIII restriction fragment of ~5.0 kb cosegregated with the mutant allele (data not shown).
Tagging with Activator and Dissociation
Experiments designed to obtain Activator (Ac) or Dissociation (Ds) insertion mutants of Rp1-D were performed concurrently with Mu tagging experiments. These tagging experiments, partially described by Pryor (1993), are summarized in Figure 1. The first mutant screen was performed by pollinating lines homozygous for Rp1-D and heterozygous for Ac with an Rp1-M line and inoculating the progeny with rust race R1, which is avirulent on Rp1-D but virulent on Rp1-M. Of ~171,000 progeny screened, 23 mutants with altered resistance phenotype were identified (Figure 1). These included 21 fully susceptible mutants, one partially susceptible mutant (Rp1-D*-5), and one susceptible mutant with a disease lesion mimic phenotype (Rp1-D*-21; Hu et al., 1996).
Only a small proportion of these mutants were expected to have been caused by Ac/Ds activity, because the frequency with which the mutations were obtained (0.014%) was similar to the mutation frequency of 0.016% observed for Rp1-D in a genetic background not known to contain any active transposons (Pryor, 1987). To help distinguish mutants caused by transposon insertion from those caused by other mutation events, such as deletions due to unequal crossing over, we tested the mutants for their ability to revert in the presence of Ac (Pryor, 1993). The disease lesion mimic Rp1-D*-21 and 17 fully susceptible mutants were tested, and all were found to be stable. Although no resistant revertants were obtained from Rp1-D*-5, this partially susceptible line gave rise to fully susceptible progeny at a frequency of 0.23% in the presence of Ac (Figure 1). No such mutants of Rp1-D*-5 were identified in the absence of Ac in a similar-sized sample (Figure 1), indicating that the instability of this line was Ac dependent.
Homozygous mutant lines were obtained by self-pollinating the F1 mutant plants and using resistance assays to select F2 or F3 progeny lacking the Rp1-M gene derived from the pollen parent (Pryor, 1993). With rust race R1, the F2 or F3 progeny showed the same resistance phenotypes in the homozygotes as were observed in the original F1 seedlings, except for Rp1-D*-13, which had been selected as a fully susceptible F1 seedling but was fully resistant and indistinguishable from the Rp1-D parent in the homozygote. We hypothesized that Rp1-D*-13 had contained a transposon in the Rp1-D gene, which excised to give a functional resistance allele before the homozygote was recovered, and that such a transposon may have moved to a location close to Rp1-D, given that transposons tend to transpose to linked sites (Greenblatt, 1984; Dooner and Belachew, 1989).
Anticipating that a closely linked transposon in Rp1-D*-13 may give rise to a relatively high frequency of transposon insertions in Rp1-D, further tagging experiments were performed with the resistant Rp1-D*-13 line. Rp1-D*-13 gave rise to susceptible mutants at frequencies of 0.14 and 0.04% in the presence and absence of Ac, respectively (Figure 1), indicating Ac-dependent instability. Two of the mutants obtained from Rp1-D*-13 in the presence of Ac (Rp1-D*-13-2 and Rp1-D*-13-3) were tested for their ability to revert. Whereas Rp1-D*-13-2 gave rise to resistant revertants at a frequency of 0.21%, no revertants were obtained for Rp1-D*-13-3 (Figure 1). Thus, along with the partially susceptible Rp1-D*-5 mutant, Rp1-D*-13-2 was identified as a potential Ds insertion mutant.
DNA gel blot analysis with Ac and Ds probes identified a number of newly transposed elements in the Rp1-D*-13-2 and Rp1-D*-5 mutants. However, when segregating families were examined, none of these was found to be linked to rp1 (data not shown). This search for transposable elements in the mutant genes was hampered by the extremely complex hybridization pattern shown by the probes, which reflected the high copy number of Ac- and Ds-related sequences in the maize genome.
Summary of Ac/Ds Tagging Experiments.
Shown is a diagram summarizing experiments, performed previously (Pryor, 1993) and in this study, the goal of which was to tag Rp1-D with Ac or Ds. Mutants were identified in genetic backgrounds with (+Ac) or without (−Ac) Ac, either directly from wild-type Rp1-D homozygotes or in second- or third-round mutation experiments. Susceptible (susc.), partially susceptible, and disease lesion mimic mutants were identified, together with resistant (res.) revertants. The numbers of the mutants observed and the size of the testcross populations screened are indicated, together with the corresponding percentage of mutation frequencies. The Rp1-D*-13 line was identified as a susceptible seedling in the initial screen, but when a line homozygous for the mutant chromosome was obtained, it was resistant—hence, the 1/1 reversion frequency. Mutation frequencies significantly greater than that previously observed for Rp1-D in the absence of any known transposons (0.016%; Pryor, 1987) were observed in Rp1-D*-5 and Rp1-D*-13, at either the P < 0.01 level (**) or the P < 0.05 level (*).
In a previous exercise, we used PCR to isolate 11 noncross-hybridizing classes of sequences from maize that encode products with similarity to the NBS regions of NBS-LRR resistance proteins (Collins et al., 1998). One of these R gene–like sequences (PIC20) hybridized with a small gene family of approximately nine members at the rp1 locus. An examination of the potential Ds insertion mutants of Rp1-D and parental lines with the PIC20 probe detected no differences between wild-type Rp1-D and the resistant Rp1-D*-13 line (Figure 2). However, in the susceptible Rp1-D*-13-2 mutant, one hybridizing 6.9-kb NcoI restriction fragment was replaced by another fragment ~400 bp longer, and in three resistant revertants of Rp1-D*-13-2, the hybridization pattern was once again indistinguishable from that of wild-type Rp1-D (Figure 2). Thus, the PIC20 probe revealed the presence or absence of an insertion at the rp1 locus in Rp1-D*-13-2 and its revertants that was perfectly correlated with loss or gain of Rp1-D resistance, respectively. PIC20 revealed deletions in three other mutants of Rp1-D*-13, in the partially susceptible line Rp1-D*-5 and in two fully susceptible derivatives of Rp1-D*-5 (Figure 2).
Cloning the Rp1-D Gene
A λ genomic DNA library of the NCO4 Rp1-D mutant containing the Mu insertion was made, and a clone containing the rp1-linked 5.0-kb HindIII fragment was isolated from it using a Mu probe. Sequencing showed that the clone contained a 1736-bp Mu element almost identical to Mu1.7 (Taylor and Walbot, 1987), which, like other Mu insertions, was flanked by 9 bp of directly duplicated target sequence, in this case, 5′-ATCCTGGGT-3′. A genomic DNA library of the Rp1-D*-13-2 mutant was also made and screened with PIC20, and one positive 13-kb clone was found by sequencing to contain a member of the Ds1 family of transposable elements (Gerlach et al., 1987) within the region covered by the PIC20 probe. The sequence of the 399-bp Ds element was 95 to 99% identical to those of other Ds1-type transposons represented in the databases, and like other Ds insertions, it was flanked on either side by 8 bp of directly duplicated target sequence, in this case, 5′-GCTGGAAG-3′. Sequencing of the entire sequence flanking the Mu element in the 5.0-kb HindIII clone and 5.6 kb of the sequence flanking the Ds element showed that the two transposable elements were inserted into the same NBS-LRR gene (GenBank accession number AF107293), with the Mu insertion site located 1219 bp downstream relative to the Ds insertion site. Thus, the two independent Rp1-D cloning approaches identified the same gene.
DNA Gel Blot Analysis of Potentially Ac/Ds-Tagged Rp1-D Mutants.
DNA of mutants and the parent Rp1-D line were cut with NcoI and probed with the PIC20 probe, which is derived from the NBS-encoding region of Rp1-D. Arrows indicate restriction fragments altered in size by an insertion/excision correlated with loss/gain of Rp1-D resistance in the susceptible Rp1-D*-13-2 mutant and its resistant revertants Rp1-D*-13-2-R2, -R5, and -R6. Mutants Rp1-D *-5, Rp1-D*- 5-2 and -5-3, and Rp1-D*-13-3, -13-4, and -13-5 all showed deletions of Rp1-D gene family members. DNA length markers (in kilobases) are shown at left.
cDNA Analysis
RNA isolated from leaves of Rp1-D plants was used to construct a cDNA library, and clones hybridizing with the PIC20 probe were identified from a screen of ~250,000 plaques. Ten of the positive clones were identical in sequence to the Rp1-D gene, and together with a clone of the 5′ end of the transcript obtained by PCR (see Methods), these provided complete coverage of the transcribed region and enabled the ends of the mRNA and the introns to be defined. Although no introns were identified in the coding region, one intron of 142 bp was identified in the 5′ untranslated region, and another of 79 bp was identified in the 3′ untranslated region 28 bp after the translation stop codon. The first intron was absent in the clone of the 5′ end of the transcript obtained by PCR but was present in the only Rp1-D cDNA clone from the library that spanned this region, indicating that the first intron is not always spliced from the Rp1-D transcript.
Two other cDNA clones of NBS-LRR genes were obtained in the cDNA library screen and were 94% identical to one another and to Rp1-D. The remaining positive clones were derived from a gene designated rp1-Cin4 (GenBank accession number AF107294), composed of a 5′ domain related to Rp1-D and a 3′ region of 475 bp with similarity to the Cin4 family of retrotransposon-like elements (i.e., 64% identity to the sequence with GenBank accession number Y00086; Schwarz-Sommer et al., 1987). Translation of rp1-Cin4 results in a truncated protein containing a putative NBS but no LRRs (data not shown). Truncated proteins containing an NBS but no LRRs are predicted products of genes from other NBS-LRR resistance loci (Whitham et al., 1994; Dinesh-Kumar et al., 1995; Lawrence et al., 1995; Parker et al., 1997; Ayliffe et al., 1999). However, it has not been determined whether these presumptive truncated proteins play a role in resistance.
The RP1-D Protein
Rp1-D encodes a 1292–amino acid protein containing the P loop and kinase-2 nucleotide binding motifs (Traut, 1994) and an LRR domain (Figure 3) and therefore belongs to the NBS-LRR class of plant R genes (Baker et al., 1997). Amino acid sequence similarity to other NBS-LRR resistance proteins was also observed throughout the protein, especially in regions around the putative NBS previously shown to be well conserved in this class of resistance protein (Grant et al., 1995; Lawrence et al., 1995; Ori et al., 1997). Unlike the flax rust resistance proteins L6 and M (Lawrence et al., 1995; Anderson et al., 1997), the RP1-D protein lacks an N-terminal TIR domain (Baker et al., 1997) with similarity to the Drosophila Toll protein and the mammalian interleukin-1 receptor. The RP1-D protein also lacks a leucine zipper motif, such as the ones present at the N termini of the RPM1, RPS2, and MI-1 resistance proteins (Bent et al., 1994; Mindrinos et al., 1994; Grant et al., 1995; Milligan et al., 1998). The leucine-rich domain of the RP1-D protein contains ~24 repeats that resemble the consensus for cytoplasmic LRRs proposed by Jones and Jones (1997) (Figure 3). The repeats in RP1-D are irregular and are interrupted by a region that cannot be easily arranged into repeats (Figure 3).
cDNA analysis identified a class of Rp1-D transcripts that retained intron 1. In this class of transcripts, an in-frame ATG start codon within intron 1 could initiate translation of a protein with an N-terminal 16–amino acid extension relative to the protein translated from the fully spliced transcript. A number of plant genes are known to retain introns in a fraction of their transcripts, and this phenomenon is thought to reflect a relatively inefficient intron splicing mechanism in plants (Nash and Walbot, 1992; Simpson and Filipowicz, 1996). Therefore, the longer alternative version of the RP1-D protein may simply be the consequence of an inefficient splicing mechanism and may play no specialized role in resistance.
The RP1-D Protein.
Underlined sequences in the N-terminal half of the protein correspond to the P loop and kinase-2 nucleotide binding motifs (Traut, 1994). The N-terminal extension resulting from retention of intron 1 is shown in lowercase letters. The C-terminal portion of the protein contains two sections with irregular LRRs (bracketed) separated by a region that cannot be easily arranged into repeats. The repeats resemble the consensus L--L--L--L-L--(N/C/T)-(-)L--IP-- proposed by Jones and Jones (1997) for cytoplasmic LRRs (where a hyphen is any residue). Residues in boldface indicate where the RP1-D repeats conform to this consensus or where the L and I residues are substituted for the residues F, I, M, or V.
Analysis of rp1 Transcription
RNA gel blot analysis using the PIC20 probe (NBS region) identified transcripts of two sizes that were present in approximately equal abundance in leaves of Rp1-D maize plants (Figure 4A). Only the larger transcript class also hybridized with a probe for the LRR-encoding region of Rp1-D, indicating that this transcript class is derived from NBS-LRR genes (Figure 4A). This transcript class represents a mixture, because the cDNA library screen showed that at least two other NBS-LRR members of the Rp1-D gene family were transcribed in addition to Rp1-D. The smaller transcript class was shown to be derived from the rp1-Cin4 gene, because it hybridized with a probe made from the Cin4-related sequence in rp1-Cin4 (Figure 4A).
Transcript levels of the Rp1-D gene family members were examined in leaves of Rp1-D homozygous seedlings inoculated with incompatible (R1) or compatible (R2) rust races. The levels of rp1-related transcripts were determined in seedlings at 0, 12, 36, and 60 hr after rust inoculation or mock inoculation and in untreated controls. The results of one experiment are presented in Figure 4B. Transcript abundance increased at the 36-hr time point in the incompatible (R1) interaction and then decreased. However, the same pattern of transcript accumulation occurred in both the untreated and mock-treated control seedlings. In the experiment shown in Figure 4B, transcript accumulation was not observed in the equivalent compatible (R2) interaction. In other experiments (data not shown), transcript levels did accumulate following innoculation with the R2 race. The reason for this discrepancy is unknown. We conclude that whereas rp1 transcript levels may be influenced by the stage of leaf development, there is no clear correlation between increases in levels of rp1 transcripts and innoculation with incompatible or compatible rust races.
Analysis of rp1 Transcription.
(A) Identification of NBS-LRR and rp1-Cin4 transcripts. A single lane containing ~3.0 μg of poly(A)+-enriched leaf RNA from homozygous Rp1-D plants was blotted and probed sequentially with probes for the NBS and LRR coding sequences of Rp1-D and the Cin4-related sequence present in rp1-Cin4. The NBS probe identified transcripts of two different sizes. The LRR and Cin4 probes indicated that the larger and smaller transcript classes were derived from NBS-LRR and rp1-cin4 members of the Rp1-D gene family, respectively. The position of the 28S rRNA is indicated.
(B) Testing for effects of rust on rp1 transcription. Ten-day-old homozygous Rp1-D seedlings were inoculated separately with the rust biotypes R1 (avirulent on Rp1-D plants) and R2 (virulent on Rp1-D plants), mock inoculated (sprayed with water minus rust spores), or left untreated (no tr.). The first leaf and the end half of the second leaf, corresponding to tissue exposed to rust spores in the inoculated samples, were taken for RNA extraction at the time of inoculation (0 hr) and at 12, 36, and 60 hr after inoculation. Some seedlings that were left unharvested were scored 8 days after inoculation to monitor the effectiveness of rust infection. Blots containing 20 μg of total RNA per lane were probed with the NBS PIC20 probe (top gel). The same blots were also probed with the rRNA probe rTA71 (bottom gel). The abundance of NBS-LRR or rp1-Cin4 transcripts was not noticeably greater in the treated samples than in the untreated samples, indicating that transcription of Rp1-D gene family members is not induced by rust infection or by the inoculation procedure. In general, the abundance of both transcript types in treated and untreated seedlings increased between the 0- and 36-hr time points and then decreased between the 36- and 60-hr time points.
Rp1-D Deletion Mutants
Most of the mutations identified in the Ac/Ds tagging studies did not revert in the presence of Ac, suggesting that they were caused by events other than transposon insertion. To determine the nature of these mutations, we performed DNA gel blot analysis with the PIC20 probe. All of the 27 mutants contained fewer hybridizing restriction fragments than did wild-type Rp1-D plants (Figure 5) but still contained the flanking RFLP markers bnl3.04 and ksu3 (data not shown), indicating that they had resulted from interstitial deletions. The Rp1-D coding region contains a 2.6-kb BglI fragment, a 2.0-kb HincII fragment, and a 1.6-kb HindIII-NsiI fragment detectable using the PIC20 probe. Each mutant was missing at least one of these fragments (data not shown), indicating that the Rp1-D coding region had been deleted or rearranged.
On the basis of the hybridization patterns obtained with the restriction enzymes AccI, BglI, BglII, DraI, NcoI, and NsiI, the mutants were of at least nine different types. Rp1-D*-5, -17, -19, -21, and -24 were each unique, whereas the other mutant classes contained two (Rp1-D*-4 and -13-3), four (Rp1-D*-7, -12, -28, and -5-2), five (Rp1-D*-10, -11, -14, -16, and -29), and 11 (Rp1-1, -2, -3, -8, -20, -22, -23, -27, -13-4, -13-5, and -5-3) representatives. Representatives within each of these groups could not be distinguished by using the six restriction enzymes. Despite the instability shown by Rp1-D*-13 and Rp1-D*-5, susceptible mutants derived from these lines (with the exception of Rp1-D-13-2 tagged with Ds) were indistinguishable from mutants obtained directly from Rp1-D plants. An examination of families segregating for mutations from each class with the PIC20 probe demonstrated that PIC20 only identifies sequences at the rp1 locus (data not shown). Hence, differences in the hybridization patterns between the mutants reflected the type of deletion that occurred at rp1. Rp1-D*-24 carried the largest deletion, because only one hybridizing band was detected in this mutant with each restriction enzyme. The smallest deletion was carried by Rp1-D*-5, which, depending on the restriction enzyme, showed a loss of zero to two hybridizating bands. Fragments that differed in size from those in wild-type Rp1-D plants were identified in all of the deletion mutants, except for Rp1-D*-4, -7, -12, -17, -21, -28, -5-2, and -13-3 (e.g., the AccI restriction fragments indicated by arrows in Figure 5). Detection of the novel restriction fragments with the PIC20 probe indicated that the majority of the deletions involved recombination or other rearrangements within or close to the coding regions of Rp1-D gene family members.
DNA Gel Blot Analysis of 27 Rp1-D Mutants.
DNA of wild-type Rp1-D plants and mutants was cut with the restriction enzyme AccI and probed with the PIC20 probe for the NBS-encoding region of Rp1-D. The mutants showed a range of deletions of Rp1-D gene family members (Rp1-D*-5 is indistinguishable from wild-type Rp1-D plants on this blot, but other restriction enzymes showed that this mutant contains a deletion). Mutants are arranged in order of decreasing deletion size. Novel-sized restriction fragments in the mutants are indicated by arrows. DNA length markers (in kilobases) are shown at left.
rp1 Locus Haplotypes
The rp1 genes described by Saxena and Hooker (1968) are available in a series of near-isogenic lines in the R168 background. DNA gel blot analysis with Rp1-D probes was used to examine DNAs of 13 of these R168 lines following digestion with a range of restriction enzymes (e.g., Figure 6). Some of the lines gave hybridization patterns that could not be clearly distinguished. Specifically, the Rp1-C, Rp1-L, and Rp1-N lines could not be distinguished from one another, nor could the Rp1-E, Rp1-I, and Rp1-K lines or the Rp1-A and Rp1-F lines (Figure 6). Those lines that showed indistinguishable hybridization patterns with Rp1-D probes could also not be distinguished by using our current set of 11 rust races (Hulbert et al., 1991). The fact that each of these lines contains unique flanking RFLP marker genotypes (Hong et al., 1993; S. Hulbert, unpublished data) makes it unlikely that the lines were mislabeled during handling. All of the remaining R168 lines showed unique hybridization patterns with Rp1-D probes. Of these lines, only Rp1-J and Rp1-H cannot be distinguished from one another by using our current collection of rust races (Hulbert et al., 1991).
The reason we could not distinguish the resistance specificities of some of the R168 lines is unknown. It may reflect limitations of the current set of rusts, but the identical hybridization patterns given by these lines with Rp1-D probes raise other possibilities. For example, the original set of rusts (no longer available) used by Hooker's group to characterize rp1 genes may have been capable of differentiating the apparently similar haplotypes. Alternatively, some of the original rust resistance phenotypes may have been the result of identical rp1 haplotypes, which were modified by unlinked genes that were eliminated during introgression into the R168 background.
Recombination Analysis of Rust Resistance Specificities in the rp1 Region
Different rp1 rust resistance specificities can be genetically recombined, and chromosomes recombinant for two different rp1 specificities (either double resistant or double susceptible) have been selected from testcross progeny by using appropriately chosen rust biotypes (Saxena and Hooker, 1968; Hulbert and Bennetzen, 1991; Hulbert et al., 1993). Rp1-D/Rp1-J recombinants subjected to DNA gel blot analysis with an Rp1-D probe for the NBS region display combinations of fragments from both parents, indicating that the recombination events had occurred within the cluster of Rp1-D homologous genes from each parent (Figure 7). Other recombinants for different pairs of rp1 genes also showed combinations of hybridization bands from both parents when examined with Rp1-D probes (data not shown). Together, these data support the hypothesis that the other uncloned rp1 specificities are encoded in gene families related to Rp1-D.
We have performed a similar analysis of lines recombinant for Rp1-D and the Rp5 rust R gene, which is located 1 to 3 centimorgans distal to the rp1 locus (Wilkinson and Hooker, 1968; Hulbert and Bennetzen, 1991). With the restriction enzyme EcoRV, the hybridization patterns of the double-resistant recombinants were identical to that of the Rp1-D parent, whereas the hybridization pattern of the double-susceptible recombinant was identical to that of the Rp5 parent (Figure 7).
DNA Gel Blot Analysis Profiles of rp1 Locus Haplotypes.
DNA of near-isogenic lines containing different rp1 genes in the R168 background was cut with the restriction enzyme EcoRV and analyzed with a probe for the NBS-encoding domain of Rp1-D. Overall, the hybridization patterns reflected our ability to distinguish these lines by using our current collection of rust biotypes (see text for details). Although Rp1-B does not appear different from Rp1-C, Rp1-L, and Rp1-N here, it was clearly different from the other lines when other restriction enzymes were used. DNA length markers (in kilobases) are shown at left.
Although some of the parental EcoRV fragments shown in Figure 7 are monomorphic, analysis of these and other Rp5/rp1 R gene recombinants with a range of restriction enzymes has failed to identify an Rp1-D–related restriction fragment associated with Rp5 resistance (data not shown). Analysis of randomly chosen progeny from segregating populations has also failed to identify any Rp1-D–related sequences located more than a fraction of a centimorgan proximal or distal to the rp1 genes (S. Hulbert, unpublished data). These data strongly suggest that the Rp5 rust resistance gene is not closely related to the rp1 genes.
DNA Gel Blot Analysis of Rp1-D/Rp1-J and Rp1-D/Rp5 Recombinants.
Lines homozygous for chromosomes and recombinant for pairs of rust resistance genes (Rp1-D/Rp1-J and Rp1-D/Rp5) were subjected to DNA gel blot analysis alongside the Rp1-D, Rp1-J, and Rp5 parent lines. The restriction enzyme EcoRV and the PIC20 probe for the NBS region were used. Of the Rp1-D/Rp1-J recombinants, two are double resistant (RR; containing both Rp1-D and Rp1-J) and three are double susceptible (SS; containing neither Rp1-D nor Rp1-J). Of the Rp1-D/Rp5 recombinants, two are double resistant (RR; containing both Rp1-D and Rp5) and one is double susceptible (SS; containing neither Rp1-D nor Rp5). Each Rp1-D/Rp1-J recombinant displays a mixture of hybridization bands from both parents, suggesting that the Rp1-J gene belongs to the Rp1-D homologous gene cluster present in the Rp1-J parent. However, only bands from the Rp1-D parent are evident in each of the two double-resistant Rp1-D/Rp5 recombinants, and only bands from the Rp5 parent are evident in the double-susceptible Rp1-D/Rp5 recombinant, suggesting that the nearby Rp5 R gene is not an Rp1-D homolog. DNA length markers (in kilobases) are shown at left.
DISCUSSION
Testing R Gene Identity
Regions of sequence conservation between the putative nucleotide binding domains of NBS-LRR resistance proteins have facilitated the PCR isolation of R gene–like sequences from a number of dicot and monocot species. Whereas some of these sequences have been shown to cosegregate with known R genes (Leister et al., 1996, 1998; Collins et al., 1998), verification of R gene identity requires either complementation in transgenic plants or a genetic approach such as point mutation or transposon tagging. In this study, independent insertions of Ds and Mu transposable elements into the same NBS-LRR gene were observed in Rp1-D mutants. In the case of the Ds insertion, excision of the Ds element was coincident with the Ac-dependent reversion to Rp1-D resistance. In lieu of a transgenic complementation test, these genetic data allowed unequivocal identification of the gene encoding the Rp1-D resistance specificity from among members of an NBS-LRR gene family that initially was identified by PCR.
Mutations of Rp1-D
Previous genetic studies of the rp1 locus revealed the following features. Genes for different rp1 resistance specificities can be combined in cis, suggesting that they belong to a gene cluster (Saxena and Hooker, 1968; Hulbert and Bennetzen, 1991). Lines homozygous for rp1 rust resistance specificities give rise to susceptible individuals at high frequencies, indicating that rp1 genes are unstable (Pryor, 1987; Bennetzen et al., 1988). The instability of rp1 genes involves chromosome pairing, because rp1 genes are stable in hemizygotes that have one copy of the rp1 region deleted (T. Pryor, unpublished data). Most losses of rp1-mediated resistance in homozygotes involve meiotic recombination, as evidenced by the exchange of RFLP markers flanking the locus (Sudupak et al., 1993; Hu and Hulbert, 1994). The conclusions drawn from these data were that rp1 resistance haplotypes probably consist of clusters of homologous genes that can undergo imperfect pairing and recombination in homozygotes or “unequal crossing-over,” resulting in deletions of active gene family members.
Our molecular analysis supports these conclusions. On the basis of DNA gel blot analysis with a probe for the NBS-encoding region, the Rp1-D haplotype consists of a cluster of approximately nine homologous genes, including Rp1-D. In addition, 27 of the mutants obtained from homozygotes showed deletions of Rp1-D gene family members consistent with unequal crossing-over events. Theoretically, deletions caused by mispairing between a tandem array of repeated sequences could vary in size, depending on the degree of mispairing. The Rp1-D*-24 mutant, which contains only one member of the Rp1-D family, appears to be a consequence of the most extreme form of mispairing, in which Rp1-D gene family members from opposite ends of the cluster paired and recombined, thereby deleting the intervening genetic material. Because the phenotype of the Rp1-D*-24 mutant is normal, except for its rust susceptibility, this deletion suggests that there are no essential genes within the rp1 complex. The other mutants showed a range of smaller deletions of at least six different sizes, presumably resulting from different, less extreme mispairing events.
The tomato Cf-9 and Cf-4 haplotypes that mediate resistance to Cladosporium fulvum each contain four homologs in addition to the respective R genes Cf-9 and Cf-4 (Parniske et al., 1997). Five recombination events between the Cf-9 and Cf-4 genes in Cf-4/9 heterozygotes were identified by Thomas et al. (1997), and all of these arose from the same chromosome pairing configuration. The recombination events were found to have occurred in intergenic regions at sites showing relatively extensive similarity (Parniske et al., 1997). Therefore, it was proposed that the bias observed toward certain pairing/recombination events was due to the presence of these islands of intergenic sequence conservation that facilitated these events. This concept was supported by the study of Dooner and Martinez-Ferez (1997), in which points of recombination between alleles of the maize bronze gene were found to be significantly biased against regions that contained insertion/deletion differences or a high density of sequence polymorphisms. Although the Rp1-D haplotype appears to have mispaired in many different ways, the inability to distinguish many of the independent mutants by DNA gel blot analysis provided evidence for a bias toward certain mispairing/recombination events.
Recombination at the rp1 locus can also give rise to disease lesion mimics (Hu et al., 1996). Rp1-D*-21 and other lesion mimic mutants of rp1 show loss of parental resistance and express a leaf-spotting phenotype, which histochemically resembles the hypersensitive lesions involved in the normal resistance response, suggesting that these mutants of rp1 result from modification of rp1 genes. Consistent with this observation is the fact that Rp1-D*-21 contains a partial deletion of the Rp1-D haplotype, which is unique among the rp1 mutants. A unique deletion was also identified in the partially susceptible mutant Rp1-D*-5. The low level of resistance in Rp1-D*-5 shows the same specificity as wild-type Rp1-D resistance across a range of rust biotypes (Richter et al., 1995), suggesting either that this mutant may still contain the part of the Rp1-D gene involved in pathogen recognition or that two other homologs have combined to form a gene with the same resistance specificity as Rp1-D. Studies currently are under way to characterize the Rp1-D*-21 and Rp1-D*-5 mutations in further detail.
R Gene Transcription
Rust inoculation resulted in no detectable increase in the levels of Rp1-D–related mRNAs in seedling leaves. However, these transcripts increased and then decreased in both inoculated and uninoculated plants during the course of the experiment, suggesting that the expression of Rp1-D gene family members changes with development. Transcripts of the Xa1 NBS-LRR bacterial resistance gene in rice were also reported to accumulate after pathogen inoculation and mock inoculation (Yoshimura et al., 1998). However, because controls receiving no treatment were not described, the possibility that this effect was due to plant development rather than pathogen inoculation cannot be ruled out completely. Possible developmental influences on transcription should be tested in any such experiment, especially in light of the present findings. Studies with the Arabidopsis PTO bacterial resistance gene, which encodes a protein kinase, and the L6 flax rust resistance NBS-LRR gene have shown that transcription of these genes is not increased by pathogen inoculation (Martin et al., 1993; Ayliffe et al., 1999).
Relationship between Rp1-D and Other Rust R Genes in the rp1 Region
Maize lines near-isogenic for different rp1 resistance specificities display distinct and complex DNA hybridization patterns when an Rp1-D probe is used, indicating that these lines contain multiple related and closely linked genes. The analysis of recombinants between several pairs of rp1 specificities using the Rp1-D probe demonstrated that recombination occurred within the rp1 gene complex. From this observation, it is inferred that other specificities that map to the rp1 locus may be encoded by genes closely related to Rp1-D. Confirmation will require mutational analysis or complementation in transgenic plants. A similar analysis of recombinants between rp1 and rp5, which maps 1 to 3 centimorgans distal to rp1, did not identify any Rp1-D–related sequences that cosegregated with rp5. From this result, we infer that rp5 is not closely related to rp1, although it may still be a member of the NBS-LRR superfamily of R genes.
Close genetic linkage has been observed between noncross-hybridizing classes of NBS-encoding R gene–like sequences in a number of plant species (Kanazin et al., 1996; Collins et al., 1998; Leister et al., 1998; Spielmeyer et al., 1999). From these observations and those made in this article, the picture is emerging that plant chromosomes contain clusters of NBS-LRR genes. Without insertional mutagenesis or highly efficient cloning/complementation technology, the identification of specific R genes by positional information alone will be difficult in most crop plants.
METHODS
Maize Lines
Lines with the Pvv (variegated pericarp) allele of the P gene containing an active Activator (Ac) element (Barklay and Brink, 1954) obtained from J. Kermicle (University of Wisconsin, Madison) were used as the source of Ac in the Ac/Dissociation (Ds) tagging experiments. Because Ac transposition is highest in maize lines containing only one Ac copy (Brink and Nilan, 1952), tagging was performed in lines heterozygous for Pvv over Pww, an allele of P that lacks Ac. Disease resistance (R) gene stability in the absence of Ac was tested in homozygous Pww lines. Absence of Ac activity was determined by testcrosses to a tester line (supplied by J. Kermicle), which carried a chromosome-breaking Ds located proximally to the gene R (colored aleurone). Absence of variegation for colored aleurone due to chromosome breakage indicated the absence of Ac activity. Lines containing multiple copies of Mutator (Mu) for use in Mu tagging experiments were obtained from G. Johal (University of Missouri, Columbia). R168 near-isogenic lines containing different rp1 genes and Rp5 were obtained from A. Hooker (formerly of the University of Illinois, Urbana).
cDNA and Genomic DNA Libraries
Genomic and cDNA libraries were constructed as recommended by the manufacturers of the vectors. HindIII fragments of 4 to 6 kb from the noncrossover event 4 (NCO4) Rp1-D mutant were purified using the GeneClean II kit (Bio-101, Vista, CA) and cloned into the λ DASH vector (Stratagene, La Jolla, CA). The Rp1-D*-13-2 genomic library was made using partially digested Sau3AI DNA fragments that were size fractionated by glycerol gradient centrifugation and cloned into the BamHI site of the EMBL3 λ vector (Promega). To make the Rp1-D cDNA library, we used poly(A)+-enriched RNA prepared from total leaf RNA by using Dynabeads oligo(dT)25 (Dynal, Oslo, Norway) to make cDNA with the HybriZAP cDNA synthesis kit (Stratagene), and the cDNA was cloned into the HybriZAP two-hybrid λ vector (Stratagene). Plasmid stocks of purified cDNA clones were obtained by excising pAD-GAL4 phagemids from the HybriZAP vector in vivo and using them to transfect Escherichia coli, as recommended by the supplier of the vector.
Cloning of Transcript 5′ Ends
The 5′ ends of Rp1-D gene family member transcripts were amplified by using the SMART polymerase chain reaction (PCR) cDNA synthesis and Advantage cDNA PCR kits (Clontech, Palo Alto, CA), essentially using the protocol supplied with the kits, except that a nested PCR approach using several downstream primers complementary to the Rp1-D gene were used instead of the downstream primer provided. The sequences of the nested primers for sites increasingly close to the 5′ end of Rp1-D were P5 (5′-TTTGCTGCCAAAGGACATTGT-3′), P21 (5′-GCCCTCAAGGACATTGTACTC-3′), and P16 (5′-GGCCGCTTGAATCACCAGCTCGAACTG-3′). First-strand cDNA was made from total leaf RNA of Rp1-D plants in reactions containing the P5 primer and the SMART oligonucleotide. PCR was then performed in a 20-μL reaction containing one-tenth of the first-strand cDNA as template, the SMART oligonucleotide, and the P21 primer by using 35 cycles of 95°C for 30 sec, 50°C for 30 sec, and 72°C for 1.5 min. Using the same amplification conditions, we used one-thousandth of this reaction as a template in a 60-μL PCR containing the SMART oligonucleotide and the P16 primer. In the last amplification, Taq polymerase was used instead of the polymerase from the kit to facilitate cloning of the products into the T-easy plasmid vector (Promega).
Sequencing
λ DNA of purified genomic clones was prepared as described by Sambrook et al. (1989). λ DNA further purified by precipitation with polyethylene glycol or plasmid DNA of λ subclones or cDNA clones purified by the boiling method (Sambrook et al., 1989) was used as a template for sequencing using the Big-Dye sequencing system (Applied Biosystems, Foster City, CA). Sequence data were analyzed using computer programs from the Genetics Computer Group (Madison, WI) software package (Devereaux et al., 1984).
DNA and RNA Gel Blot Analyses
Gel blot analysis of maize genomic DNA was performed as previously described (Collins et al., 1998). Total maize leaf RNA for RNA gel blot analysis was extracted using the method of Logemann et al. (1987) and enriched for poly(A)+ RNA by using the PolyATract mRNA isolation system (Promega). RNA samples were quantified by absorbance spectroscopy at 260 nm, subjected to denaturing agarose gel electrophoresis in 1.0% agarose gels containing formaldehyde (Sambrook et al., 1989), and blotted to a Hybond N membrane (Amersham), as recommended by the manufacturer of the membrane. Hybridization and autoradiography for RNA gel blot analysis were the same as given for DNA gel blot analysis. Membranes were hybridized with α-32P-dCTP–labeled DNA probes overnight at 65°C and given four washes of ~7 min in 0.5 × SSC (1 × SSC is 0.15 M NaCl and 0.015 M sodium citrate) at 60°C before autoradiography. Clone rTA71 (Gerlach and Bedbrook, 1979) was used for the detection of rRNAs.
ACKNOWLEDGMENTS
We thank Val Ryle, Judy Cassells, Luch Hac, and Jeff Coleman for excellent technical support. We are grateful to David Jones for his assistance in preparing the manuscript and to Arthur Hooker, Jerry Kermicle, and Guri Johal for supplying plant material. Portions of the work were supported by grants from the U.S. Department of Agriculture (project No. 9300604), the National Science Foundation (Grant No. MCB-9728490), and the Grains Research and Development Corporation of Australia (project No. CSP197).
- Received November 30, 1998.
- Accepted April 5, 1999.
- Published July 1, 1999.
REFERENCES
