Homology-based Cloning of RGCs – A Tool for Gene Discovery

Technological primers

Homology-based cloning of RGCs has become feasible in the last several years after a number of plant R genes had been isolated through transposon tagging or MBC (Kanazin et al., 1996; Leister et al., 1996; Yu et al., 1996). Characterization of the pre- dicted protein products from these cloned R genes indicates that different plant species may use just a few classes of pro- teins in conferring resistance to various bacterial, fungal and viral pathogens, and nematodes (Hammond-Kosack and Jones

1997; Staskawicz et al., 2001). In addition, the proteins encoded by R genes seem to contain domains and motifs that are con- served across different classes of R genes. A domain that is particularly common among cloned R genes is the NBS domain (Hammond-Kosack and Jones, 1997; Meyers et al., 1999). A few highly conserved motifs within the NBS domain have allowed the development of degenerate oligonucleotide primers able to amplify DNA sequences that are very similar to the NBS–LRR class R genes. Subsequently, this polymerase chain reaction (PCR)-based approach has also been used to clone DNA sequences in the kinase or receptor-like kinase class R genes (Deng and Gmitter 2003). Several terms are used in the literature to refer to these sequences, including RGCs (resist- ance gene candidates), RGAs (resistance gene analogues) or RGLs (resistance gene- like sequences); the term RGC is used in the present discussion. The choice of this term is based not only on the strong similarities of these sequences to cloned plant R genes and their possession of the typical features of cloned R genes, but also on their increased uses in a so-called candidate gene cloning approach and their potential of being functional R genes, as confi rmed in a few recent cases (Aarts et al., 1998; McDowell et al., 1998; Pfl ieger et al., 2001).

Though many of the RGCs obtained so far correspond only to part of the full struc- tures of cloned R genes, they have found valuable uses in plant breeding and genet- ics research. Linkage mapping data have confirmed that many RGCs are tightly linked or co-segregate with known disease resistance gene loci that are responsible for resistance to bacterial, fungal or viral pathogens, or parasitic nematodes (Kanazin et al., 1996; Leister et al., 1996; Yu et al., 1996). In this sense, RGCs may provide the ultimate molecular markers for tagging plant disease resistance traits (Michelmore, 1996). Several RGCs, when used with an MBC strategy, have led to isolation of func- tional resistance genes. Examples include the cloning of lettuce Dm3 and Arabidopsis Rpp8 that are responsible for resistance to

two different downy mildew pathogens, Bremia lactucae and Pernospora parasitica, respectively (Aarts et al., 1998; McDowell et al., 1998). Thus, RGCs may serve as valu- able resources for using a candidate gene approach to clone plant R genes. Because of their close association with R genes, RGCs can also provide effective tools to facilitate understanding of some fundamental fea- tures of R genes, such as R gene organiza- tion, distribution and evolution (Michelmore, 1996).

Application of a homology-based cloning approach may be of particular sig- nifi cance to citrus and many other horticul- tural, agricultural crops and forest plants for genetic improvement or engineering of disease resistance. Transposon tagging and MBC, two approaches previously widely used in plant R gene cloning, are not read- ily accomplished in citrus and many other crops, because of their long generation times, large plant sizes or complex genetic behaviours. In addition, each plant genome may contain hundreds of potential R gene sequences to specify resistance to different pathogens. Isolating these sequences one by one, as in the case of MBC or transposon tagging, can be time consuming. A homol- ogy-based PCR amplifi cation approach may allow access to numerous candidate sequences in a genome within a short period of time, thus making it desirable, perhaps essential, in order to tag and under- stand the enormous repertories of R genes in plant genomes.

Cloning and characterization of citrus RGCs

In citrus, cloning of RGCs initially was attempted as a way to facilitate map-based isolation of Ctv. As the previous section in this chapter stated, a high resolution link- age map and molecular markers tightly linked to or co-segregating with the target disease resistance genes are prerequisites for using a map-based gene cloning strat- egy. The Ctv locus had been tagged with a dozen RAPD and several SCAR markers by 1997, but most of these markers were found

to be more distant from the resistance gene locus than expected and desired, in subse- quent mapping with a large population con- sisting of 678 backcross progeny. This indicated the need to search for more molecular markers closer to the gene. To meet this need, Deng et al. (2000) used two previously reported primers (LM637 and LM638) and four primers (F11, R11, R16 and R18) newly designed from the con- served NBS domain and amplifi ed four major DNA bands from the genomic DNA of USDA 17-47, an intergeneric CTV-resistant hybrid of C. grandis and P. trifoliata. When these bands were cloned and 39 individual clones were sequenced, they identifi ed 22 RGCs. These sequences were very similar to the NBS–LRR class R genes and contained the typical motifs (P-loop, kinase-2 and kinase-3a) of the NBS domain found in this class of R genes. The overall amino acid identity between these citrus RGCs and fi ve representative R genes (Arabidopsis RPS2, RPM1, tobacco N, fl ax L6 and tomato I2) ranged from 18 to 42%. Using a 70% amino acid identity threshold value, the 22 sequences were grouped into ten classes, RGC1–RGC10. The majority of the citrus RGCs seem to belong to the so-called non- TIR group of NBS–LRR genes, while only two of the classes, RGC1 (clone Pt6) and RGC2 (clone Pt14), fall into the TIR group. Specifi c primers developed from the diver- gent regions of 13 representative RGC sequences detected three types of polymor- phism between ‘Thong Dee’ pummelo and USDA 17-40: fragment length difference, restriction site difference, or presence and absence of amplifi ed bands. Three markers (18P33, Pt8a and Pt9a) from the RGCs were found to be closely linked to Ctv. Marker Pt8a also segregated in a population for citrus nematode resistance mapping and was located very close to Tyr1 (Deng et al., 2000; Ling et al., 2000).

This initial success in fi nding RGCs associated with disease resistance loci in Poncirus prompted a more targeted and thorough search for new RGCs more tightly linked to or even co-segregating with Ctv. Sixteen combinations of degenerate primers

designed from the NBS–LRR class R genes were screened on two genomic DNA pools each consisting of eight CTV-resistant and eight CTV-susceptible individuals, respec- tively, from a segregating backcross popula- tion (the R family). One DNA band polymorphic between the two bulks was detected on a polyacrylamide gel. This DNA fragment was highly similar to the NBS–LRR class R genes. Two markers derived from this fragment, 11R1-1a (cleaved amplifi ed polymorphic sequence (CAPS) marker) and 11R1-1a5 (restriction fragment length polymorphism (RFLP) marker), co-segregated with Ctv in high res- olution linkage mapping. Subsequently, they allowed a chromosome landing on BAC clones of the Ctv locus, thus greatly expediting the progress in constructing BAC contigs to cover the CTV resistance and susceptibility allelic regions.

Recent efforts in citrus RGC cloning have been focused on the receptor-like kinase class R genes that are represented by rice Xa21. This class of R genes is unique in that they each contain an extracellular LRR domain for signal recognition and a cyto- plasmic catalytic kinase domain for signal transduction (Song et al., 1995; Ronald, 1997). Few of this class of RGCs have been amplifi ed or cloned in plants other than rice and its related species, and Arabidopsis. Degenerate primers are not readily available for this class of R genes, because their LRRs are poorly conserved while their kinase domains are too common in many proteins responsible for diverse functions. Deng and Gmitter (2003) aligned the kinase domains of rice Xa21 and tomato Pto protein (a kinase class R gene for tomato bacterial streak resistance) and found two well-conserved motifs in the kinase subdo- mains I and VIII. Degenerate primers, kindF1 and kindR1, designed from these motifs, amplifi ed genomic DNA sequences from the intergeneric hybrid USDA 17-47 that are similar to rice Xa21. Subsequently, oligonucleotide primers derived from one of the genomic sequences (A2UP and A2LW) amplifi ed Xa21 -like sequences from citrus BACs. All 53 sequences contained

open reading frames (ORFs); their deduced peptide sequences carried the features found in the corresponding region of rice Xa21 protein and they shared 55–60% amino acid identity and 65–71% similarity. In multiple sequence alignment, the sequences were clustered into fi ve groups, designated as CRK1–5 (citrus receptor-like kinase). Each group was further divided into 2–4 subgroups, according to their sequence similarities. Rice Xa21 specifi es broad-spectrum resistance to more than 30 isolates of Xanthomonas oryzae pv oryzae (Xoo), the causal agent of rice bacterial blight. Xanthomonas axonopodis pv citri (Xac), a pathogen related to Xoo, causes the devastating disease citrus canker in citrus. Resistance to citrus canker has been observed in C. ichangensis and kumquat (Fortunella spp.). It will be an intriguing question to determine whether Xa21 -like sequences are involved in resistance to citrus canker.

The rice genome may contain more than 700 copies of NBS type sequences, and Arabidopsis may have nearly 1% of its total genomic sequences encoding NBS domains. The copy number of NBS–LRR class sequences in citrus genomes has yet to be determined, but it may be expected to be in the hundreds. Capturing some of these sequences, converting them into specifi c markers and locating them on to genetic linkage maps may provide powerful tools to tag, track and utilize gene loci for disease resistance in Poncirus, Citrus or Fortunella. As a fi rst step to realize this potential of citrus RGCs, Deng et al. (2001a) probed sev- eral BAC and transformation-competent (TAC) libraries with cloned RGC sequences to identify large-insert clones containing resistance gene-like sequences. Screening of the USDA 17-47 Bam HI BAC library with

13 representative NBS sequences from RGC1 to RGC10 yielded 322 positive clones. These clones may correspond to 40–70 genetic loci in the Citrus or Poncirus genome and contain 80–140 copies of unique NBS sequences. A similar BAC library screening with A2, a prevalent Xa21 -like sequence identified above,

resulted in 79 BAC clones containing recep- tor-like kinase-encoding capacity. These clones were assembled into 35 contigs, and contain approximately 50–70 copies of Xa21 -like sequences. These BACs can pro- vide a valuable source in citrus R gene tag- ging, mapping, cloning and characterization. Insert ends of some of these BACs have been sequenced and will be converted into specifi c PCR markers for further tests of potential association with particular resistance gene loci.

RGCs amplifi ed by PCR with NBS or kinase domain-derived primers usually are 400–650 bp in size and correspond only to part of the full-length structures of the cloned R genes. A genuine question is fre- quently raised regarding whether the amplifi ed RGCs are from structurally and functionally ‘real’ genes. Compared with the rapid increase in the number of cloned RGCs in various plants, progress in this area is lagging remarkably. This may be partly due to the ease of PCR amplifi cation and subsequent cloning, but mostly due to the tremendous amount of work required for (full gene) sequence acquisition, construct development, genetic transformation and expression analysis. So far, such work has been done only in a limited number of cases with a few plants. To answer this critical question, we obtained the upstream and downstream sequences, through primer- based walking, of four NBS-encoding regions in four BACs hybridizing with RGC Pt8, and two kinase-encoding regions in BACs hybridizing with RGC A2. Two of the DNA sequences contain ORFs that can translate into polypeptide sequences with- out any stop codons. They seem to contain all the structural features of an NBS–LRR class R gene. In a BLAST search of the GenBank database, their best hit is Arabidopsis RPS2 among cloned R genes, showing a 30–35% amino acid identity, 50–55% similarity and an Expect value less than e-110. The other two NBS–LRR class sequences seem to be pseudogenes as they each contain one stop codon in their coding regions. The two receptor-like kinase class RGCs (17o6RLKP and 26m19RLKP) con-

tained all the nine domains of rice Xa21, including a signal peptide, extracellular LRRs, transmembrane and kinase domains. Overall, they shared approximately 44% amino acid identity and 51% similarity with rice Xa21. Their kinase domains con- tained all the 12 subdomains and the serine–threonine-specifying motifs of rice Xa21, and the 15 invariant amino acids of protein kinases. The LRR domains of 17o6RLKP and 26m19RLKP consist of the same number of imperfect repeats as Xa21, and they shared approximately 44% amino acid identity and approximately 51% simi- larity. Although the functionality of these full-length RGCs remains to be determined, the above data indicate that many RGCs may have complete structures as cloned functional R genes.

General procedure and considerations

Homology-based cloning of citrus RGCs involves several steps: identifying con- served domains and motifs in cloned R genes, designing appropriate degenerate PCR primers, amplifi cation of genomic or complementary DNAs, cloning of heteroge- neous PCR fragments, characterization and selection of plasmid clones, and sequenc- ing and identifi cation of RGCs. Some of these steps such as cloning and sequencing of PCR fragments are similar to standard procedures used in molecular cloning, and therefore discussion will be focused on some of the special conditions used for RGC cloning. When further genetic, structural or functional analyses are intended for cloned RGCs, it may require acquisition of addi- tional DNA sequences, and conducting linkage mapping, gene expression study and/or genetic transformation, most of which are beyond the scope of this chapter.

Conserved motifs and degenerate primers

A number of computer software programs can be used for aligning multiple amino acid sequences to reveal highly conserved motifs in cloned plant R genes. CLUSTERX

(Thompson et al., 1997) and GCG ‘PILEUP’ are among the commonly used programs. Successful PCRs require two primers facing each other; thus at least two regions of highly conserved amino acids should be found, and each should extend for at least six amino acids, so primers of at least 17 nucleotides could be derived. The distance between the two chosen motifs needs to be at least 20 amino acids, so that PCR prod- ucts can be at least 100 bp in size. The two motifs are converted to sense and antisense nucleotide sequences, respectively. Desirable degenerate primers are expected to amplify diverse RGC sequences but few non-specifi c sequences. Therefore, it is nec- essary to control the degeneracy of designed primers. Using inosine as an alter- native base, and/or synthesizing multiple sets of primers for a given conserved motif are two approaches frequently adopted.

The NBS domain, particularly its P- loop motif, has been frequently targeted in degenerate primer design. This is because the domain appears at a high frequency in the cloned R genes and it is most conserved among the genes. In addition, an internal hydrophobic domain (HD) between the NBS and LRR domains of the NBS–LRR class of R genes also has been used exten- sively for amplifi cation of this class of RGCs. In PCRs, these primers are expected to amplify DNA fragments of the size of approximately 500 bp, provided that no introns are located between the DNA sequences encoding the two motifs.

Two short stretches of peptide sequences (FG(K/S)VYKG and GY(A/I)(A/D)PEY) in the kinase subdo- mains I and VIII were found to be relatively well conserved in the receptor-like kinase class R genes. Degenerate primers designed from these regions amplifi ed citrus DNA sequences similar to the receptor-like kinase class R genes. Attempts have been made to design degenerate primers for the LRR domain found in the NBS–LRR class and LRR class R genes, but the poor conser- vation of amino acids other than leucine has led to few successes in targeting these classes of RGCs in plant genomes.

PCR amplifi cation

When a degenerate primer is synthesized, each primer actually contains numerous oligonucleotide sequences as dictated by its degeneracy. Because of this nature, it is necessary to increase, compared with stan- dard PCRs, the amount of degenerate primers and DNA templates used in PCRs to amplify RGC sequences. In addition, appropriate annealing temperatures may need to be identifi ed to obtain the expected products.

Cloning, sequencing and sequence analysis

PCR products amplifi ed with degenerate primers are rather heterogeneous, even though they appear as a single band or a few bands on an agarose gel. It is necessary to obtain several dozen or more independ- ent recombinant DNA clones to capture the major part of the sequence diversity ini- tially detected by the degenerate primers. Restriction pattern analysis is then used to identify unique recombinant clones for sub- sequent sequencing. A BLAST search of the GenBank database seems to be a very convenient and effective way to reveal whether or not obtained sequences share any similarity to cloned R genes or the vast number of plant RGCs. In- depth analysis of RGC sequences for their coding capacities, pairwise comparisons, multiple alignment, phylogenetic relation- ship, etc. can be accomplished using vari- ous free or commercial software programs such as DNASISMAX, the GCG SEQWEB soft- ware (‘GAP’, ‘PILEUP’, etc.) and the CLUSTAL X package.

Reducing non-specifi c sequences

Many DNA sequences obtained may have no similarity to plant R genes. This phe- nomenon is common in homology-based cloning of plant R genes or genes responsi- ble for other functions. It is because of the degenerate nature of primers used, and the tremendous complexity of plant genomes. Several strategies are often used to reduce

the number of non-specifi c sequences in PCR amplifi cation and cloning. Using mul- tiple sets of primers of lower degeneracy, optimizing annealing temperatures for PCR, separation of DNA bands on agarose or polyacrylamide gels, and the choice of DNA bands are various approaches attempting to minimize the problem of non- specifi city.

Recent technical developments

Targeted isolation of RGCs for known disease resistance loci

Early studies involved cloning and sequencing of RGCs, followed by designing specifi c primers and screening them for a potential association with resistance gene loci. Numerous degenerate primers have been designed from various classes of cloned R genes in recent years, but it can be rather time consuming to follow this proce- dure, especially when it is uncertain whether or not any RGCs, or what class of RGCs, might be linked to the target resist- ance gene loci of interest. In search of new markers for Ctv, we developed a so-called BSA-RGC approach, based on the bulked segregant analysis (Michelmore et al., 1991) that has been in use for screening large numbers of random primers in tagging dis- ease resistance loci with RAPD markers. The BSA-RGC approach screens degenerate primers and their amplifi ed products with pairs of resistant and susceptible plant groups before RGCs are cloned, sequenced and converted into molecular markers. It was used in the development of markers 11R1-1a and 11R1-1a5 that co-segregated with Ctv and led to chromosome landing on Ctv -containing BACs. This approach should be applicable to other disease resistance gene loci, and to allow a large number of primer combinations to be screened in a short period of time and potential RGC sequences linked to the target gene locus to be identifi ed, before engaging with DNA fragment cloning and sequencing.

Improving RGC sequence diversity

One of the goals in RGC cloning is to obtain DNA sequences most similar to cloned R genes and from as many different chromo- somal locations as possible, i.e. maximum sequence diversity. Frequently it becomes diffi cult to fi nd new RGCs after dozens of plasmid clones have been sequenced. One recent study indicates that BACs may help overcome this diffi culty and allow for cap- turing more diverse RGC sequences. Deng and Gmitter (2003) compared the diversity of receptor-like kinase class RGC sequences amplifi ed from genomic and BAC clone DNA, respectively. The 29 sequences amplifi ed from genomic DNA with degener- ate primers kindF1 and kindR1 fell into two major groups and four subgroups in cluster analysis. The prevalent one among the 29 sequences, A2, was used to screen the USDA 17-47 Bam HI library, and positive BACs were used for amplifi cations with A2- derived specifi c primers. Surprisingly, the

23 sequences amplifi ed from these BACs fell into fi ve major groups and eight sub- groups, i.e. sequences not revealed by amplifi cation of genomic DNA were found among the BAC clone inserts.

Acquisition of full-length gene structures

The acquisition of full-length gene sequences is essential for structural and/or functional analysis of RGCs, and also bene- fi cial for marker development and genetic mapping. Long-range PCR, 5’ and 3’ RACE (rapid amplifi cation of cDNA ends), and cDNA and genomic library screening have been used in various cases for this purpose. Recent experience indicates that BAC libraries (and clones) are an excellent resource for this endeavour (Deng and Gmitter 2003). Through primer-based walk- ing on BACs, full-length gene structures have been obtained for a number of NBS–LRR class and receptor-like kinase class RGCs. BAC libraries are easy to screen and their clones contain large inserts (50–250 kb). These characteristics facilitate not only recovery of upstream and down-

stream coding or non-coding regions of spe- cifi c RGCs, but also the identifi cation and characterization of resistance gene sequence clusters. BACs in some vectors, such as pYLTAC (such clones have been called TACs), may be transferred into Agrobacterium cells and used in genetic transformation. Several BAC libraries have been constructed in the last several years, and they cover both the Poncirus and Citrus genomes with good representation (Deng et al., 2001a).