The Wild Emmer Wheat Gene PmG16, Conferring Resistance to Powdery Mildew, is an Orthologue of Pm60 from Triticum Urartu

Wild emmer wheat (WEW), the tetraploid progenitor of durum and bread wheat, is a valuable genetic resource for resistance to powdery mildew fungal disease caused by Blumeria graminis f. sp. tritici (Bgt). PmG16 gene, derived from WEW, confers high resistance to most tested Bgt isolates. We mapped PmG16 to a 1.4 cM interval between the anking markers uhw386 and uhw390 on Chromosome 7AL. Based on gene annotation of WEW reference genome Zavitan_V1, 34 predicted genes were identied within the ~3.48 Mb target region. Six genes were annotated as associated with disease resistance, of which TRIDC7AG077150.1 was found to be highly similar to Pm60, previously cloned from Triticum urartu and residing in the same syntenic region. A functional molecular marker (FMM) for Pm60 (M-Pm60-S1) co-segregated with PmG16, suggesting that WEW PmG16 is probably an orthologue of Pm60 from Triticum urartu (designated here as TdPm60). Sequence alignment identied only eight SNPs that differentiate between TdPm60 and TuPm60. Furthermore, our results suggest that other WEW powdery mildew resistance genes MlIW172 and MlIW72, that also mapped to the same region of Chromosome 7AL, might be identical or allelic to TdPm60. Screening of 230 WEW accessions with Pm60 specic markers, 58 resistant accessions were identied from Southern Levant harboring the TdPm60 allele, while all the susceptible accessions showed no PCR amplications. Deployment of TdPm60 is clearly more advantageous over TuPm60 since it can be rapidly introgressed by classical breeding approaches into bread wheat genetic background.

Wild emmer wheat (WEW), the tetraploid progenitor of cultivated wheat, is a valuable genetic resource for powdery mildew resistance breeding. Currently, more than 24 Pm genes and one quantitative resistance loci have been identi ed and mapped in WEW. Seven of them (Pm16, Pm26, Pm30, Pm36, Pm41, Pm42 and Pm64) have been o cially designated as Pm resistance genes (Klymiuk et al. 2019;McIntosh et al. 2019;Zhang et al. 2019). Other temporarily named Pm resistance genes, MlIW72, PmG16, MlWE18, MlIW172 and HSM1 were mapped in the same genomic region of Chromosome (Chr) 7AL of WEW (at the terminal bin 0.86-1.00 of 7AL). Besides, Pm1a from bread wheat (Neu et al. 2002), PmU (Pm60) from T. urartu (Qiu et al. 2005;Zhang et al. 2018) and PmTb7A.2 from T. boeoticum (Chhuneja et al. 2015) were also mapped to the same locus on the terminal part of Chr 7AL. The sequence of Pm1a which was cloned from T. aestivum acc. Axminster is quite dissimilar to Pm60, therefore, it is unlikely that they represent orthologous genes (Hewitt et al. 2020). These ndings suggest that an R-gene cluster is present within this genomic region of cultivated and wild wheat species of different ploidy levels, conferring powdery mildew resistance (Liang et al. 2016;Zhang et al. 2018). Recently, a multi-genome comparisons showed huge differences in NLR content between wheat genotypes; Among ~ 2,500 loci with NLR signatures identi ed in the 16 wheat cultivars, 31-34% of them are shared across all genomes, while the number of unique signatures ranged from 22 to 192 per cultivar (Walkowiak et al. 2020).
Pm60 which encodes a typical CC-NB-LRR type of NLR immune receptor which was cloned from Chr 7AL of T. urartu acc. PI428309. Large sequence differences were found between the respective Pm60 alleles in PI428309 and the reference genome of T. urartu acc. G1812 (Zou et al. 2018). Two additional functional Pm60 alleles which differ in the LRR regions were identi ed from different T. urartu accessions, Pm60a in PI428210 that includes a 240-nucleotide deletion and Pm60b in PI428215 that includes a 240-nucleotide insertion as compared with the Pm60 in PI428309 (Zou et al. 2018). Two functional PCR markers for Pm60 (M-Pm60-S1 and M-Pm60-S2), covering the LRR region were developed for diagnosis of these three Pm60 alleles by amplicon size, i.e. the M-Pm60-S1 amplify 591 bp (Pm60a), 831 bp (Pm60), or 1071 bp (Pm60b), while no ampli cation product is obtained in susceptible T. urartu accessions . Marker M-Pm60-S1 ampli ed one of the three functional alleles in 78% (73 out of 93) of the tested resistant T. urartu accessions. Furthermore, 34.8% (67 out of 227) T. urartu accessions harbor the Pm60 allele, and 2.64% accessions harbor Pm60a or Pm60b alleles (6 out of 227 of each allele) indicating that the Pm60 locus is one of the major Bgt resistance loci prevalent among natural populations of T. urartu ).
The dominant gene PmG16 was identi ed and mapped using a mapping population derived from the powdery mildew resistant Israeli WEW accession G18-16 crossed with a susceptible T. durum cultivar Langdon (LDN). PmG16 conferred resistance to 34 out of 47 tested isolates from Israel and Switzerland and it was physically mapped on Chr 7AL (Ben-David et al. 2010). In the present study, we ne-mapped PmG16 and identi ed a candidate gene, which was highly similar to Pm60 and was mapped to the Pm60 syntenic locus on Chr 7AL of WEW. Sequence and functional marker analyses suggested that the candidate gene (named as TdPm60) from WEW G18-16 is an orthologous allele of TuPm60 with only eight SNPs differentiating between them. To study the geographical distribution of TdPm60, we screened a WEW collection of 230 accessions from the Fertile Crescent using a speci c selection marker for Pm60. This screening revealed that 58 WEW accessions harboring the TdPm60 allele were highly resistant to Bgt, and that most of them originated from the Southern Levant. Using marker-assisted backcrossing program, we have successfully introgressed PmG16 into the susceptible bread wheat cultivar Ruta and validated the improvement of resistance to Bgt in the progenies.

Bgt inoculation and disease assessment
Bgt isolatessix Bgt isolates used for phenotyping were collected from wild and domesticated wheat species from various agro-ecological habitats across Israel (Ben-David et al. 2010). Bgt isolate #15 (Bgt #15) collected from T. durum was used for histopathological characterization of the defense responses, phenotyping of segregating mapping populations, and screening for resistance in the WEW collection and introgression lines. Other isolates, including Bgt #70, Bgt #101 and Bgt #107 that were collected from bread wheat (T. aestivum), and Bgt #103 that was collected from WEW, were used for resistance spectrum identi cation of 29 wheat accessions with different Pm60 alleles. Isolate Bgt #SH was collected from WEW in the screening house (SH) of the University of Haifa.
Disease assessmentyoung plants at the two-leaf stage were tested for resistance to powdery mildew. Resistance results were recorded at 7 and 10 dpi. Incubation and disease assessment were as described by Hsam and Zeller (1997) and Xu et al. (2018) with slight modi cations. In brief, the rst leaf of each plant was excised into three segments and cultured in square petri dishes containing 8 g/L agar with 50 mg/L benzimidazole (Sigma-Aldrich, USA). Dishes were inoculated with fresh spores of powdery mildew in a inoculation tower, and were transferred to a climate chamber with 70% humidity and maintained in 22°C/20°C under 12 h light/12 h dark photoperiod, respectively. The reaction to Bgt was examined visually, and infection type (IT) was recorded in a scale of 0-4: 0, no visible symptoms; 0;, hypersensitive reaction; 1, disease spot of less than 1mm with a thin hypha layer and fewer spores; 2, disease spots of less than 1 mm with thick hypha layer and more spores; 3, scattered disease spots of greater than 1 mm with a thick hypha layer and more spores; and 4, continuous disease spots of greater than 1 mm with a thick hypha layer and more spores. Plants with IT of 0-2 were considered as resistant and plants with IT of 3-4 were considered as susceptible.

Bgt -wheat interactions
Bgt #15 structures were observed in the inoculated leaves under a uorescence microscope as described by Klymiuk et al. (2018) and Wei et al. (2020). Leaf segments were incubated at 37°C for 24 hours in 15 ml of 1 M KOH, neutralized with 50 mM Tris at pH 7.5, and stained with 1.0 ml of 20 µg/ml wheat germ agglutinin (WGA) in 50 mM Tris for 24 hours at 4°C. Fluorescence microscopy was carried out using an inverted uorescence microscope Leica DMi8 (Leica Microsystems, Germany), tted with a lter cube for the FITC excitation range to observe the WGA-stained fungal structures.
Reactive oxygen species (ROS) accumulation and the cell death were evaluated in inoculated wheat leaves (2 dpi). Methods were applied according to Thordal-Christensen et al. (1997) and van Wees (2008) with minor modi cations. For the ROS detection, wheat leaves placed in 1 mg mL − 1 DAB solution (w/v) with pH 3.8 (15 mins in vacuum pressure at -60 KPa, 8 hours at 28℃), decoloring for 2 days in 96% ethanol, staining of Bgt structure with 0.6% Coomassie Brilliant Blue G250 (Sigma, UK; 0.6g dissolved in 100mL absolute ethyl alcohol) for 2 mines and washing with sterile water, then stored in 50% glycerol.
For the cell death detection, wheat leaves were stained for 5 min in 0.25% Trypan blue solution (Biological Industries, Israel) with incubating in boiling water, decoloring in 2.5 g/mL Chloral hydrate (Riedel-dehaёn, Germany) for 2 days, then stored in 50% glycerol. ROS and cell death were examined under a microscope Zeiss Axio Imager M1 (Carl Zeiss, Germany).

Development of molecular markers and PCR detection
Polymorphic SNPs around the genomic region of PmG16 were detected by genotyping the 147 RILs (Table S1) of the mapping population by the 15K wheat SNP array (Trait Genetics®, Germany).
Wheat genomic DNA was extracted following the cetyltrimethylammonium bromide (CTAB) method. PCR mixture (20 µL) contained 50-100 ng of template DNA, 0.4 µM of forward and reverse primers, 10 µL of 2× Taq PCR MasterMix (Tiangen, China). PCR program for SSR ampli cation included denaturation at 94°C for 5 mins; followed by 35 cycles of 94°C for 1 min, annealing at 57-61°C for 35 s (depending on the Tm of each primer pair), and 72°C for an appropriate time depending on the length of PCR products (1 min/kb), elongation at 72°C for 10 mins, and termination at 10°C in a Veriti® 96-Well Fast Thermal Cycler (Applied Biosystems, Singapore). PCR amplicons were visualized on 1.5-3% agarose gels. KASP genotyping was performed in 5µl containing 2.5 µl KASP Master mix (LGC Genomics, USA), 0.4 µmol L − 1 of each of three primers and 200 ng of template DNA. KASP PCR ampli cations were performed in a StepOnePlus™ Real-Time PCR System (Applied Biosystem, USA) with the following ampli cation conditions: an initial step at 30°C for 1 min, 94°C for 15 mins followed by 10 cycles of 94°C for 20 s, 65°C for 1 min (drop 0.6°C per cycle); and continue with 26 cycles of 94°C for 20 s, 55°C for 1 min; terminated after 30°C for 1 min.

Development of genetic and physical maps
The genetic distances between the detected markers and PmG16 were calculated based on genotypic (15K SNP arrays and SSR markers) and phenotypic (IT to Bgt #15) data of 147 RILs by using JoinMap 4.0 (Van Ooijen et al. 2006) (Table S1). The locations and speci city of probe sequences of SNP markers included in the 15K array and SSR markers were identi ed by using a command-line tool BLAST + from NCBI (Camacho et al. 2009) against the sequences of Chr 7A of WEW Zavitan_v2.0 , durum wheat Svevo_v1.0 (Maccaferri et al. 2019) and T. urartu G1812 (Ling et al. 2018). The genetic and physical maps of PmG16 genomic region in Chr 7AL were constructed by MapChart (Voorrips et al. 2002).
Gene cloning, phylogenetic, and NLR protein structure analysis Based on the genetic and physical maps of PmG16 we have identi ed candidate genes that are residing in WEW Zavitan_v2.0 reference genome sequence between the markers anking PmG16. The sequences of candidate genes were viewed in the target genetic region of PmG16 and downloaded from GrainGenes Zavitan download page (https://wheat.pw.usda.gov/graingenes_downloads/Zavitan/). Because the WEW Zavitan_v2.0 does not contain gene annotation information, we used the annotations of these candidate genes from the WEW reference genome Zavitan_v1.0. Four PCR primer sets (Table S3) were developed based on the published 9492 bp-length nucleotide sequence around Pm60 locus in T. urartu (Tang et al. 2017) and used for cloning and sequencing of the Pm60 alleles in different WEW accessions. Sequences of PCR products were assembled and analyzed using DNAMAN software (Version 9). The Pm60 homologs in the phylogeny tree were identi ed from the GrainGenes and Ensembl Plants website, and analyzed using the Neighbor-Joining algorithm (aligning by MUSCLE) in the MEGA-X software (Kumar et al. 2018). To study the protein structure of Pm60 alleles and homologs, we used the NLR annotator , which annotates NLR genes by motifs as described in Jupe et al. (2012).

RNA Extraction and Quantitative Real-Time RT-PCR (qPCR)
Total RNA was extracted from both Bgt #15-treated and mock-treated (as control) G18-16 leaf segments collected at four different time points (12, 24, 36 and 48 hpi) using RNeasy Plant Mini Kit (Qiagen, Germany). cDNA was synthesized from total RNA using a qScript™ cDNA Synthesis Kit (Quantabio, USA).
Gene-speci c primers of the TdPm60 and the housekeeping gene Actin were used for qPCR (Table S3). qPCR ampli cations were performed in a volume of 10 µl containing 5 µl of SYBR Green FastMix (Quantabio, USA), 250 nM primers and optimized dilution of cDNA template. PCR ampli cation was performed with StepOne thermal cycler (ABI, USA). The program included an initial step at 95°C for 30 s followed by 40 cycles of 95°C for 15 s, 60°C for 30 s, and 72°C for 10 s. Relative expression of the target genes was calculated by the 2 −ΔΔCt method. All of the qPCR reactions were performed in triplicate technical repetition, and each with three independent biological repetitions.
Pm60 speci c markers and geographic distribution of Pm60 alleles in the WEW population Since several conserved SNPs were found that differentiate between functional and nonfunctional TdPm60 alleles, we developed two markers (M-Pm60-S3 and M-Pm60-S4) that can be used for diagnosis of TdPm60 alleles (Fig. S5). KASP marker M-Pm60-S4 can discriminate between the functional (795C), nonfunctional (795T) and null Pm60 alleles. For the M-Pm60-S3 marker, the forward primer targeted site (2335-2354) contained one conserved SNP. The published M-Pm60-S1 marker contains four conserved SNPs, of which three are located on the forward primer (2896-2916) and one on the reverse primer (3944-3966). Therefore, these markers can be used for diagnosing the TdPm60 locus in WEW (see Fig.  S5).
A collection of 230 WEW accessions originating from the Fertile Crescent were evaluated for their reaction to Bgt #15 and were genotyped by three TdPm60 gene-speci c markers (M-Pm60-S1, M-Pm60-S3 and M-Pm60-S4). Geographic distribution of the different alleles was plotted using ggmap R package (Kahle and Wickham, 2013).

Development of introgression lines by marker-assisted selection (MAS)
PmG16 resistance gene was introgressed from WEW G18-16 to hexaploid wheat after its rst transfer to durum wheat var. LDN, following the "durum as a bridge approach" described by Klymiuk et al. (2019).
The resistant RIL154 (F 7 of the cross of WEW G18-16 with LDN) was selected as a PmG16-donor parent for marker-assisted breeding backcross program which included four backcrosses with bread wheat cultivar Ruta, and one sel ng generation to obtain BC 4 F 2 introgression lines (ILs). Successful ILs were veri ed for the presence of the gene using M-Pm60-S1 and uhw390 markers in each generation and further phenotypic validation.

PmG16 resistance interaction with Bgt isolate #15
In a previous study, we have shown that WEW acc. G18-16 was resistant to 34 out of 47 tested Bgt isolates from Israel and Switzerland (Ben-David et al. 2010). In the current study we have selected Bgt#15 to work with since it showed virulence on numerous Pm resistance genes (Pm1a, Pm1e, Pm2, Pm3a, Pm3c, Pm3f, Pm3g, Pm4b, Pm5a, Pm5b, Pm6 and Pm7;Ben-David et al. 2010). Plant responses to Bgt #15 clearly differentiated between the parental lines used in the current study, the WEW acc. G18-16 (highly resistant) and durum cv. LDN (highly susceptible). Small Bgt colonies were observed at 4 days post-infection (dpi) in LDN that rapidly developed massive powdery mildew colonies at 7-10 dpi with infection type (IT) scale 4 ( Fig. 1a), as compared to G18-16 that was highly resistant to Bgt #15 (IT = 0), showing no visible disease symptoms nor signs even at 10 dpi (Fig. 1a). In this incompatible interaction presented by G18-16, histopathology observations have shown that a large number of cells showed H 2 O 2 accumulation (reddish-brown coloration) and cell death response (blue coloration) already at 2 dpi (10-30% of the infected epidermis cells in each leaf). In contrast during the compatible interaction, presented by LDN, only 1-3% of the epidermis cells infected with Bgt conidia showed slight visible H 2 O 2 accumulation and cell death response in each leaf at 2 dpi ( Fig. 1b and 1c).
Development of fungal structures on wheat leaves of G18-16 and LDN was observed by wheat germ agglutinin conjugated with a uorophore at a series of time points from 4 to 240 hours post-infection (hpi). The primary germ tubes (PGT), appressorium (App) and haustorial primordia (HP) could be observed in both G18-16 and LDN at 4-, 12-and 24 hpi, respectively (Fig. 1d). However, mature haustorium (MP) and hyphae (Hy) development were visible at 48 hpi only in LDN. Furthermore, the development of conidiophores and massive production of Bgt conidia were observed only in LDN from 240 hpi. In contrast, in G18-16 mature haustoria were not observed and the infection process was halted at less than 48 hpi. In only a very few cases (less than 1%), haustorial primordia were able to develop into mature haustorium and elongated Hy were observed at 72 hpi, however, their growth was suppressed 72 to 240 hpi (Fig. 1d). Taken together, these results indicate that PmG16 confers a post-haustoria resistance mechanism against Bgt #15, responsible for restraining the development of mature haustoria and further hyphal growth, accompanied by ROS accumulation and cell death, and resulting in a high resistance response against Bgt #15.

Genetic mapping of powdery mildew resistance gene PmG16
A high-density SNP genetic map was recently developed by our lab for the G18-16 × LDN RIL population (147 lines) based on the Illumina In nium 15K Wheat platform, consisting of 12,905 SNPs (Fatiukha et al. 2019). Using this genetic map, we were able to localize PmG16 on Chr 7AL within a 3.2 cM interval between the proximal anking marker wsnp_Ku_c28104_38042857 and distal marker RAC875_c8752_1079 (Fig. 2a, Table S1). From a subset of 31 RILs, nine with critical recombination events within the 3.2 cM target interval, were selected and used for further dissection of the target locus using the graphical genotypes approach (Table S2; Young and Tanksley, 1989;Distelfeld et al. 2006). For marker saturation in this region, we used ve new SSR markers developed in the current study (Table S3), as well as Xgwm344 and Xwmc809 previously identi ed as closely linked to PmG16 (Ben-David et al. 2010). Three of the SSR markers (uhw387, uhw388 and uhw389) co-segregated with PmG16 (Fig. 2b, Table S2). Moreover, these markers enabled us to map PmG16 within a 1.4 cM, at a distance of 0.3 from marker uhw386 and 1.1 cM from uhw390 (Fig. 2b).
Comparative genomics analysis of the PmG16 genetic region The physical map of PmG16 genetic region was constructed based on the physical positions of the genetic markers anchored to Chr 7AL of WEW genome Zavitan_v.2.0 and T. urartu genome G1812 ( Fig. 2c  and 2d). The marker positions on the physical map showed complete collinearity with their order on the genetic map, suggesting that the candidate gene region includes approximately 3.48 Mb interval (733339776-736815424 bp in the Chr 7A of Zavitan) anked by markers uhw386 and uhw390 (Fig. 2c).
To nd candidate gene(s) for PmG16, we searched for the annotated genes in the WEW reference genome Zavitan within the 3.48 Mb interval that covers the PmG16 gene region. A total of 34 genes were identi ed, six of which were annotated as disease resistance (R) proteins with NB-ARC and leucine-rich repeat (LRR) domains, named also as NLRs (nucleotide-binding site leucine-rich repeat) (Table S4). Two candidate NLRs TRIDC7AG076920.1 and TRIDC7AG076930.1 were highly similar to the previously cloned Pm1a (Fig. S1, Table S4). Therefore, we used the functional molecular marker (FMM) Pm1aSTS1 (Hewitt et al. 2020) to genotype G18-16, but it did not yield any amplicons, suggesting that PmG16 is not allelic to Pm1a. Moreover, we found that a third NLR, TRIDC7AG077150.1, gave the highest BLAST scores for the previously cloned Pm60 gene located in syntenic location in chromosome 7L of T. urartu genome (AA), therefore showing close evolutionary genetic relationships (Fig. S1). To support our ndings, we genotyped all the RILs that carry critical recombination events and the parental lines with the Pm60 FMM M-Pm60-S1 ). M-Pm60-S1 showed positive ampli cation in G18-16 and co-segregated with PmG16 in the RIL population (Fig. S2, Table S2). An additional marker M-Pm60-S3 that was designed to target the Pm60 LRR domain, also showed co-segregation with PmG16 (Fig. S2). Anchoring the genetic markers used in the present study to T. urartu chromosome 7AL showed full collinearity with the physical positions of the markers used for the cloning of Pm60 (Zou et al. 2018) (Fig. 2c and 2d). The content and order of all 34 PmG16 candidate genes was highly conserved between the reference genomes of T. durum Svevo, and T. dicoccoides Zavitan_V1 and_V2, with only minor rearrangements when compared with T. urartu G1812 reference genome (Fig. S3). Therefore, we hypothesized that PmG16 is probably an orthologue of Pm60 in WEW.
Homologous cloning of the TdPm60 locus from WEW For further characterization of the Pm60 locus in WEW, we used the homologous cloning strategy for cloning the full length (4365 bp) of the corresponding Pm60 locus (designated as TdPm60 hereafter) from G18-16 genome using several Pm60 cloning primers ( Fig. 3a and b). Sequence analysis of TdPm60 identi ed only eight SNPs in comparison with the original TuPm60 (MF996807.1), and the same length (4365 bp) of mRNA sequence was ampli ed also from the cDNA of G18-16, suggesting that there are no introns in the gene (Fig. 3c). Six amino acids differed between the putative proteins TdPm60 and TuPm60 (Fig. 3c). Interestingly, no PCR ampli cation was obtained in LDN using those primers indicating the absence of Pm60 orthologue or high polymorphism in positions of primers between G18-16 and LDN (Fig. 3b).
Expression patterns of TdPm60 in G18-16 was followed at 12, 24, 36, 48 and 72 hpi with Bgt #15. The obtained results showed that the relative gene expression of TdPm60 was slightly but signi cantly (p ≤ 0.05) reduced at 24, 48, and 72 hpi in the Bgt-infected leaves compared with the control set (mockinfected leaves) (Fig. 3d).

Association of Pm60 with Bgt resistance in WEW natural populations and identi cation of allele TdPm60a
For testing whether the Pm60 locus is associated with Bgt resistance in WEW natural populations, we used FMM M-Pm60-S1 ) and additional marker M-Pm60-S3 to screen 230 WEW accessions that originated from across the Fertile Crescent ( Fig. 4a and Table 5). Fifty-nine accessions were found to be positive to Pm60 markers and all of them also showed high resistance to Bgt #15, while all the 67 susceptible accessions showed negative ampli cation of Pm60 markers (Table 1, Table S5). One hundred and six accessions showed high resistance to Bgt #15, but were found to be negative to Pm60 markers, indicating the occurrence of other powdery mildew resistance genes in these WEW populations. Previously, the M-Pm60-S1 marker was used for diagnosis of functional Pm60 alleles in the T. urartu natural population and was associated with the Bgt E09 resistance . Similarly, our result suggest that the M-Pm60-S1 marker is also associated with Bgt #15 resistance in the WEW natural population. Moreover, the M-Pm60-S3 could also be used as FMM in the WEW natural population with the same ampli cation results as M-Pm60-S1.
One resistant accession TD104552 (IT = 1) had a shorter PCR product (Fig. 4a). Sequence analysis revealed a new allele containing a 240-nt deletion (from 3607 to 3368) and two SNPs compared with TdPm60 (T2394G and A3993G), resulting in two amino acid changes (H978Q and T1332A) in the LRR region. Interestingly, this WEW allele, showed the same deletion element previously reported for Pm60a from T. urartu, and therefore it was named TdPm60a.

MlIW72 and MlIW172: potential alleles of TdPm60
Previous report showed that MlIW72 (Ji et al. 2008), MlWE18 (Han et al. 2009), MlIW172 (Ouyang et al. 2014) and HSM1 (Li et al. 2014) from different Israeli WEW accessions were mapped in the same chromosome region as PmG16 (Ben-David et al. 2010; the terminal bin 0.86-1.00 of Chr 7AL). We anchored the anking markers of these ve genes to the reference genome of WEW. An overlapped chromosome interval was identi ed on Chr 7AL for these ve gene regions (from 735356949 to 736815424) (Fig. 4b). In the overlapped interval, three candidate R genes were identi ed, TRIDC7AG076920.1 and TRIDC7AG076930.1 which were similar to Pm1a, and TRIDC7AG077150.1 which was similar to Pm60 in the phylogenetic tree (Table S4, Fig. S1). We used the WEW genotypes that were originally used for the identi cation of the four resistance genes (PmG16, MlIW172, MlIW72 and HSM1) to analyze their Pm1a and Pm60 locus and their response to different Bgt isolates. We found that WEW TD116494 (MlIW172) and TD011558 (MlIW72) accessions were highly resistant to Bgt #15 with similar resistance spectrum as G18-16, and showed positive ampli cation for the FMMs M-Pm60-S1 and M-Pm60-S3 (Fig. 4a, Table S6), while no ampli cation was obtained for FMM Pm1aSTS1. Sequence analysis revealed that TdPm60 in accession MlIW72 had ve SNPs (G1242A, A1410C, T2934G, T3017C and A4304G) compared with G18-16, resulting in three amino acid changes (H978Q, P1006L and E1435G). In contrast, accession MlIW172 contained only one SNP (A3525G) as compared with TdPm60 from G18-16, with no amino acid change at the protein level. Those SNPs in the different WEW accessions, which were collected from different places in Israel, probably evolved during host-pathogen co-evolution. TA1410 (HSM1) was partially resistant to Bgt #15 and Bgt #101 (IT = 2), but it showed negative PCR ampli cation for both M-Pm60-S1 and M-Pm60-S3 (Fig. 4a, Table S6). These results indicate that of the ve powdery mildew resistance genes, which were mapped in the terminal bin 0.86-1.00 of 7AL from WEW, at least three of them (PmG16, MlIW72 and MlIW172) may represent different alleles of TdPm60, while MlWE18 locus need to be further analyzed in the future.
Phylogenetic analysis of Pm60 homologs and development of functional KASP marker BLAST search of Pm60 homologs in the annotated genome assemblies of domesticated wheat species (T. aestivum, T. spelta, T. durum and T. monococcum) and their closest wild relatives (T. dicoccoides, T. urartu and A. tauschii), using the GrainGenes and Ensembl Plants website, revealed complex evolutionary relationships at this locus. No similar genes were identi ed in the genomes of T. aestivum cv. Cadenza, Norin 61, CDC Landmark, T. monococcum cv. DV92 and A. tauschii acc. AL8/78. Two genes identi ed in T. aestivum cv. Chinese Spring (CS), TraesCS7B01G479200 and TraesCS7D01G542200, located on Chr 7B and 7D in the same physical location as Pm60, showed high nucleotide similarity (96.65% and 95.59% Identity, respectively, Table S7) to TdPm60a. In the same location, a truncated gene (1059 bp) with highcon dence annotation was found on chromosome 7A of CS with 77.68% similarity to Pm60a (Table S7, not included in the phylogenetic tree). We did not nd any complete TraesCS7B01G479200 alleles in the 7B chromosomes or any complete Pm60-orthologs in 7A chromosomes from other hexaploid wheat assemblies. In contrast, the TraesCS7D01G542200 alleles from 7D were found in 11 other hexaploid T. aestivum and one T. spelta wheat accessions with 0-2 SNPs compared with the gene in CS (Fig. S4). Interestingly, a sequence highly similar to TraesCS7B01G479200 was found in scaffold sc d078262, of T. durum cv. Kronos (possibly on 7B) that showed just one SNP (T2281G) and with a missing sequence part (3370-3420 bp). Some other annotated genes (see Fig. 4c) were also identi ed as Pm60 similar genes (with 73 ~ 80% identity, 3099-4901 bp) possibly on the same genetic location from different assemblies of related species and were further used for phylogenetic analysis (Table S7). No paralogs of Pm60 was identi ed from those accessions.
Phylogenetic analysis showed a clear clustering of functional Pm60 alleles and their TdPm60 orthologs (Pm60, Pm60a, Pm60b, TdPm60 and TdPm60a) at both levels of ploidy (AA and AABB genomes) together with the nonfunctional Pm60a' allele from T. urartu acc. PI662227 . The two homoeologous genes from T. aestivum cv. CS 7B and 7D genomes were clustered very close to the Pm60 and TdPm60 alleles (Fig. 4c). Moreover, two similar genes from T. urartu acc. G1812 (TRIUR3_00771 and TRIUR3_00770, one gene from T. dicoccoides acc. Zavitan TRIDC7AG077150.1 and an additional identi ed gene Horvu_FT11_7H01G703800 from barley were grouped on a cluster with longer branches than the Pm60 alleles (Fig. 4c). Interestingly, during the map-based cloning of Pm60, the gene sequences TRIUR3_00771 and TRIUR3_00770 (Zou et al. 2018), and TRIDC7AG077150.1 (current study) were identi ed as the candidate genes of Pm60/TdPm60 from the two susceptible wheat assemblies G1812 and Zavitan (Zou et al. 2018 , Table S6). Two genes (TRITD7Av1G277190.1 and TRITD7Bv1G230590.1) from the 7A and 7B of T. durum acc. Svevo and two genes (BRADI_1g29658v3 and BRADI_2g38192v3) from Brachypodium distachyon were grouped on two clusters, respectively, and located as an outgroup in the phylogenetic tree (Fig. 4c).
A comparison of protein structures of the functional Pm60 alleles with TraesCS7B01G479200.1, TraesCS7D01G542200.1 and TRIDC7AG077150.1 predicted proteins showed that all of them contained a complete NLR structure. Except TRIDC7AG077150.1, all the Pm60 alleles/homologs had the same motif content and order in the CC and NB-ARC domains, but differ in the motif order and number of motif repeats in the LRR domain (Table S8). Therefore, we can't rule out the option that the differences in motif order and repeat number in the LRR domain may alter the speci city of the NLR to recognize the Bgt effector. Nevertheless, CS and Zavitan were susceptible to six Bgt isolates (Table S6), suggesting that TraesCS7B01G479200.1, TraesCS7D01G542200.1 and TRIDC7AG077150.1 are not conferring resistance to those Bgt isolates. Moreover, Kronos that harbors a sequence that is highly similar to TraesCS7B01G479200.1 is also susceptible to these six Bgt isolates (Table S6). In contrast, all the functional Pm60 alleles showed highly conserved protein structure (Fig. 4c).
We aligned those identi ed Pm60 alleles/homologs, based on the SNP at position 795 (C/T), designed a KASP marker designated M-Pm60-S4, which enabled us to differentiate between the functional, nonfunctional (possibly located on genomes B and D) and null Pm60 alleles/homologs in the different wheat accessions (Fig. 4d, Fig. S5). For example, 13 WEW accessions containing TdPm60, which were con rmed by marker M-Pm60-S1, showed the functional allele (red circle). Durum cv. LDN and Svevo, bread wheat cv. Fielder and cv. Ruta, and 20 WEW accessions without TdPm60, con rmed by marker M-Pm60-S1, showed null M-Pm60-S4 allele, the same as the negative water control (Fig. 4d, Table S5 and S6). CS, Kronos and 6 WEW accessions showed nonfunctional alleles (blue circle), suggesting that those accessions might contain nonfunctional Pm60 alleles/homologs.
Only one accession in northern Israel contained the TdPm60a allele (with 240 bp deletion) which is regarded as a rare allele, while the Pm60b-similar allele (with 240bp insertion) was not identi ed among the tested WEW accessions. We have selected 10 accessions containing TdPm60 and one with TdPm60a for the resistance tests to six different Bgt isolates. The results showed that all the 11 WEW accessions showed the same resistance spectrum of high resistance to Bgt #103, #107, #101, #15 and #70, but susceptible to Bgt #SH (Table S6). By using KASP marker M-Pm60-S4, 7 Israeli accessions, 1 Syrian and 3 Turkish WEW accessions were identi ed with nonfunctional alleles/homologs. Among them, 5 accessions were highly susceptible to Bgt #15 and two of these were further found as susceptible to all the six tested Bgt isolates (Table S5 and S6). These results show that nonfunctional alleles/homologs were susceptible to Bgt #15 or all the six Bgt isolates.
A total of 160 WEW accessions (69.6%) did not show any ampli cation by any of the three markers, of which 99 accessions showed resistance to Bgt #15 suggesting that they probably contain other resistance genes.

Introgression of PmG16 into hexaploid wheat
As part of the aim to develop resistant pre-breeding genetic resource, we transferred PmG16 into bread wheat cultivar var. Ruta following the "durum as a bridge" approach (Klymiuk et al. 2019). The rst step was to test and map the gene in tetraploid durum background and subsequently to introgress the gene by a backcross program into hexaploid cultivar using marker-assisted selection (MAS) (Fig. 6a). The resistant RIL154 (F 7 RIL derived from the cross of LDN × G18-16), was crossed and backcrossed with Ruta to develop BC 4 F 2 population accompanied by MAS using the FMM M-Pm60-S1 and the closely linked co-dominant marker uhw390 (Fig. 6b). We tested the introgression lines (ILs) with Bgt #15, and con rmed that all the ILs (e.g., BC11-15-2, BC12-15-4, BC12-9-2 and BC12-9-3) that showed positive ampli cation of M-Pm60-S1 were resistant, while all lines with no ampli cation of this marker (BC12-2-2 and BC12-2-4) were susceptible (Fig. 6b). In addition, since the co-dominant marker uhw390 showed polymorphisms among G18-16, LDN and Ruta, it was used for selecting homozygote introgressions (BC11-15-2, BC12-15-4) resistant to Bgt. The heterozygote lines (BC12-9-2 and BC12-9-3) showed high resistance to Bgt #15, that further con rmed that PmG16 is a dominant gene even in the hexaploid genetic background. The cultivar Ruta was susceptible to four Bgt isolates (Bgt #103, #101, #70 and #15), while resistant to two Bgt isolates #107 and #SH, however, the Ruta ILs with PmG16 were highly resistant to all the six tested Bgt isolates (Table S6). DAB and Trypan blue staining showed that PmG16 induced ROS accumulation and cell death in the resistant IL BC12-15-4 and halted the development of mature haustorium of Bgt #15; a likely indicator of the activation of a post-haustorium resistance mechanism also in hexaploid wheat. In contrast, in Ruta, mature haustoria and mycelium structures were detected as expected for normal development of Bgt in compatible host-pathogen interactions (Fig. 6c). These results indicate that the dominant gene PmG16 found in WEW is a promising source for bread wheat resistance breeding, and that two markers (M-Pm60-S1 and uhw390) can be used for MAS in the future wheat breeding programs.

Discussion
The powdery mildew race-speci c resistance gene PmG16, was previously mapped on Chr 7AL of WEW accession G18-16 between the anking markers Xgwm344 and Xwmc809 (Ben-David et al. 2010). In the current study, based on a high-density SNP map and 8 SSR markers, PmG16 was mapped at a higher precision than in the previous study, which enabled us to narrow down genetic interval that harbors PmG16 to only 1.4 cM, residing between anking markers uhw386 and uhw390 (Fig. 2b). Based on gene annotation of WEW Zavitan genome assembly, six NLR genes have been identi ed within this interval (Table S4), one of which (TRIDC7AG077150.1) was found at the same syntenic location on chromosome 7A as Pm60, previously cloned in T. urartu (Fig. 2c, 2d and Fig. S3) (Zou et al. 2018). Moreover, we showed that the FMM of Pm60 (M-Pm60-S1) had positive ampli cation in G18-16 and co-segregated in the mapping population with the resistance conferred by PmG16 (Fig. 4S). Sequence analysis of the cloned gene from G18-16 revealed that it includes eight SNPs as compared with the original Pm60 (MF996807.1), with six amino acids changes at the protein level (Fig. 3c). Hence, we regard the NLR from WEW accession G18-16 as an ortholog of Pm60 from T. urartu and designated it as TdPm60. Screening of 230 WEW from the Fertile Crescent revealed that all the 59 WEW accessions that were positive to ampli cations with the Pm60 FMMs (M-Pm60-S1 and M-Pm60-S3) showed high resistance to Bgt #15, while no susceptible accession had positive ampli cation of the FMMs (Table 1). Since the linkage disequilibrium (LD) in WEW decays very rapidly (Sela et al. 2014, Sela et al. unpublished), it is unlikely that other genes in the interval are PmG16. Altogether, the physical location of the gene, sequence alignments and speci c Pm60 marker analyses suggest that TdPm60 is an ortholog of Pm60, a candidate gene for PmG16, associated with Bgt resistance in WEW natural populations. Ultimately, a conclusive evidence that the encoding TdPm60 sequence confers PmG16 resistance needs to be substantiated by future work using loss-of function mutants or transgenic complementation.
In the current study, we have shown that TdPm60 (PmG16) confers a post-haustorium resistance mechanism against Bgt #15, blocking the development of mature haustorium and hyphal growth. TdPm60 induced ROS accumulation and cell death defense responses, both in tetraploid and hexaploid backgrounds, similar to those induced by Pm60 in the infected T. urartu leaves (Zou et al. 2018;, resulting in high resistance to Bgt #15 (Fig. 1a-d). The same post-haustorium resistance mechanism has also been found in WEW accession G305-3M which contains PmG3M (Wei et al. 2020). Most Pm genes cloned so far are induced by Bgt at different time points from 6 to 48 hpi, such as Pm5e , Pm41 , Pm21 (He et al. 2018), Pm2 (Chen et al. 2019, and Pm24 ). On the contrary, the expression of WTK1 (Yr15) also derived from WEW is down-regulated by the presence of the yellow rust pathogen (Klymiuk et al. 2018). The expression of Pm60 in T. urartu does not signi cantly change by Bgt infection at 1-5 dpi (Zou et al. 2018), while the expression level of TdPm60 was slightly but signi cantly (p ≤ 0.05) reduced at 24, 48, and 72 hpi in Bgt-infected G18-16 leaves compared with mock-inoculated leaves (Fig. 3d).
A few cloned resistance genes in wheat were found to be orthologous to already cloned resistance genes even at different ploidy levels. For example, Pm3b cloned from Chr 1A of hexaploid wheat, which encode a CC-NBS-LRR and confer race-speci c resistance to powdery mildew (Yahiaoui et al. 2004), was found to be orthologous to other genes based on homology cloning or physical and genetic mapping. The resistance allele Pm3k from tetraploid wheat, and Pm8 from rye have been characterized as the orthologues of the Pm3 allelic series Hurni et al. 2013). Moreover, several genes that were previously mapped to the same interval in different mapping populations were found to belong to a single locus. For example, Klymiuk et al. (2018) cloned the stripe rust resistance gene Yr15 from WEW G25, and later found that the genes YrH52 and YrG303, originated from different accessions, represent the same gene ). More than 10 powdery mildew resistance genes have been previously identi ed in the Xwmc525-Xgwm344 genetic interval on Chr 7A, including Pm1a, Pm1c and Pm1e (syn. Pm22) from bread wheat (Hsam et al. 1998;Neu et al. 2002), Pm1d from T. spelta (Hsam et al. 1998), Pm1b, Mlm2033 and Mlm80 from T. monococcum (Yao et al. 2007), PmTb7A.2 from T. boeoticum (Chhuneja et al. 2015), Pm60 (PmU) from T. urartu (Zou et al. 2018;Zhang et al. 2018), and MlAG12 from T. timopheevii (Maxwell et al. 2009). Five genes from WEW were also identi ed in this region, including MlIW72 (Ji et al. 2008), HSM1 (Li et al. 2014), MlIW172 (Ouyang et al. 2014), MlWE18 (Han et al. 2009) andPmG16 (Ben-David et al. 2010; and the current study). For the ve genes identi ed in WEW, further sequencing analysis revealed that at least three of them (MlIW172, MlIW72 and PmG16 in current study) might be orthologs of Pm60 from T. urartu (Fig. 4b). Pm1a and Pm60 have been demonstrated as different genes in the same genetic region (Hewitt et al. 2020). Using Pm1a BLAST search in GrainGenes database, we identi ed six Pm1a homologs (Identity = 77.7-88.5%) in six different reference wheat genomes, which were present at 0.21-1.87 Mb apart from the Pm60 homologs towards the centromeric positions (Table S9 and S10). Additionally, we excluded the possibility of PmG16, MlIW72, MlIW172 and HSM1 as the functional Pm1a alleles by using FMM Pm1aSTS1. Further studies are needed to determine whether the other Pm genes which mapped to same genetic region, such as Pm1b-e, are truly allelic to Pm1a or possibly the orthologs of Pm60, or may even represent novel Pm genes.
Pm60 was cloned from T. urartu by positional cloning, while Pm60b and Pm60a were identi ed in T. urartu natural populations by marker selection and homology cloning (Zou et al. 2018). The accessions containing Pm60, Pm60b or Pm60a showed high resistance to Bgt E09, and all the three alleles induced cell death in N. benthamiana (Zou et al. 2018;Zhao et al. 2020). Genetic mapping in F 2 segregating populations suggested that Pm60b is a functional allele with 240 bp nucleotide (nt) insertion in the LRR region ). In the current study, 58 WEW accessions had the TdPm60 allele and only one (TD1004552) had the TdPm60a allele, while TdPm60b was not found in the WEW natural populations. The nonfunctional allele Pm60a' that had 58 SNPs and a 3-nt deletion was identi ed in 9 susceptible T. urartu accessions, suggesting that these SNPs and 3-nt deletion may be important for the resistance function . In our study, we identi ed a resistant WEW accession (TD1004552) that had a TdPm60a allele with a 240-nt deletion and one SNP in the LRR region compared with TdPm60 (Fig. 4a). Interestingly, compared with Pm60a from T. urartu, TdPm60a has the same deletion as Pm60a but almost the same SNPs as TdPm60 suggesting convergent molecular evolution of TdPm60 alleles in WEW. Therefore, it seems that the same deletion of 240-nt occurred twice, maybe due to the presence of repetitive sequences in this genomic region. The repetitive sequence may have caused additional indel in the Pm60b allele. This LRR length variation caused by repeats is one of the main sources for NLR variability (Wicker et al. 2007). The deletion in WEW accession TD1004552 (TdPm60a) showed similar resistance spectrum as G18-16, carrying the TdPm60 allele, however, it is also possible that another Pm gene is present in this accession, and further studies are still needed to verify the TdPm60a function.
Based on BLAST analysis with published sequence data of several wheat cultivars, we found that although the Pm60 homologs are highly polymorphic, the functional alleles are highly conserved (Fig. 4c). However, some similar sequences (≥ 95.59%) of Pm60 could be identi ed on several susceptible wheat accessions (i.e, Kronos and Chinese Spring; Fig. S5). The sequences found in the domesticated cultivars and CS are probably homoeologous sequences located on genomes B and D. These sequences have a deletion of 240 nt in the LRR domain relative to Pm60 (the same as the Pm60a allele) with different motif order and different number of repeats (Table S8). However, these sequences differ from Pm60 by only ~ 4% and maintain the reading frame and the important motives in the NBS domain. Therefore, they may be functional against other Bgt races or other pathogens although the CS was susceptible to all of the six tested Bgt # isolates. Sequence diversity in the LRR region was previously demonstrated for different functional Pm3 alleles (Bhullar et al. 2009). These ndings support the suggested role of the LRR domain as an effector-binding and regulatory domain in recognition speci city during the coevolution of pathogen effectors and host NLR receptors (DeYoung et al. 2006).
Previous report showed that T. urartu accessions carrying the Pm60 allele are present mostly in Lebanon (28/83) and Syria (14/39) , while T. urartu accessions collected in Turkey were more susceptible to Bgt and contained very few Pm60 alleles (3/77). In the current study, we did not identify any functional Pm60 allele in 29 Turkish WEW accessions. Among them, just two accessions were resistant to Bgt #15 ( Fig. 5 and Table S5). In contrast, 30% of the WEW accessions collected in the Southern Levant contained TdPm60 ( Fig. 5 and Table S4). Therefore, in both T. urartu and WEW natural populations, the resistant accessions baring Pm60 are highly abundant in the Southern Levant as compared with the Northern Levant regions. Previous reports using SNP based clustering show that WEW accessions from Southern (Israel, Syria, and Lebanon) and Northern (Turkey) Levant are clustered separately (Avni et al. 2017), and the same pattern is observed in T. urartu (Ling et al. 2018), which is the diploid (2n = AA) progenitor of the A sub-genome of WEW. Therefore, Pm60 in WEW was possibly inherited from T. urartu by the polyploidization event that gave birth to WEW and took place in the Southern Levant (Feldman and Kislev, 2007). Interestingly, a previous study in T. monococcum (2n = AA) natural populations (~ 270 accessions) showed that six accessions carried the Pm60 allele (Shen et al. 2018). Supported by the previous results, the current study suggest that Pm60 is an ancient gene present in three Triticum species where it shows a high level of presence/absence polymorphism typical of some R genes (Shen et al. 2006). It is possible that the low frequency of Pm60 in the Northern Levant is caused by natural selection because it does not have a selective advantage there. More evidence is needed to elucidate the genetic evolution of Pm60 in the A genome species. The distribution and variation of the functional Pm60 alleles in the domesticated emmer, durum, and common wheat should be further studied using the newly developed KASP markers, as well as the other three functional SSR markers (M-Pm60-S1, M-Pm60-S2 and M-Pm60-S3). We believe that the Pm60, as an ancient and transferable NLR gene, can be used as a classical locus for understanding the evolution and variation of the NLR gene families in wheat. These ndings can be explained also by frequent alien introgression to domesticated wheat from some relative species .
Transferring of Pm60 from T. urartu into susceptible bread wheat cv. Kn199 by genetic transformation conferred high resistance to Bgt (Zou et al. 2018). In the current study, we successfully developed several ILs which carry PmG16 and showing high resistance to Bgt #15, suggesting that PmG16 from WEW could confer resistance in both tetraploid and hexaploid genetic backgrounds and there is no suppression of the PmG16 resistance by the D genome. PmG16 induced ROS and cell death defense responses in both G18-16 and the hexaploid ILs, suggesting that PmG16 resistance response is expressed as a typical NLR gene both in WEW and in bread wheat backgrounds (Fig. 1b, c and Fig. 6c). Therefore, Pm60 orthologs are promising genetic sources for bread wheat resistance breeding. One dominant marker M-Pm60-S1 could be used for MAS in those introgression lines, while the co-dominant SSR marker uhw390 which is linked to PmG16 could be used for selection of homozygotes during MAS (Fig. 6). Moreover, since WEW is an excellent source for other disease resistance genes/alleles, in future breeding program we propose to pyramid several such genes from WEW and transfer them into durum and bread wheat varieties to obtain more durable resistance to different diseases. These genes can include powdery mildew resistance genes PmG16 and PmG3M (Xie et al. 2012;Wei et al. 2020), and the broad spectrum yellow rust resistance genes from WEW, such as Yr15 (Klymiuk et al. 2018) and Yr36 (Fu et al. 2009). Altogether, the results of our study demonstrate that WEW is a valuable genetic resource for improving disease resistance in wheat cultivars.  Tables   Table 1 Distribution   conidiophore; DP, digitate processes; HP, haustorial primordia; Hy, hyphae; MH, mature haustorium; PGT, primary germ tube. Bars = 20 µm (4-72 hpi) and 100 µm (240 hpi).  Schematic representation of the homologous cloning approach, conserved domains, SNPs relative to TuPm60 and expression patterns of TdPm60. a Locations of primers used for cloning and sequencing of TdPm60 from WEW by four overlapped PCR fragments (1-4). b PCR products of fragments 1-4 ampli ed from G18-16 and LDN. c Structures of TdPm60 conserved domains and differences from TuPm60 in nucleotides and amino acid sequences: TdPm60 (red font) and TuPm60 (bold font). d Relative expression pattern of TdPm60 after Bgt-infection in G18-16. Infected: wheat leaves infected with Bgt #15.

Declarations
Mock: wheat leaves without Bgt #15 infection.  The Geographic distribution of the TdPm60 and TdPm60a. Blue diamond-TdPm60; Purple diamond-TdPm60a; Grey triangles-nonfunctional TdPm60 alleles/homologs; Green circles-resistant accessions without TdPm60; Yellow circles-susceptible accessions without TdPm60. The resistance phenotype was checked by infection with Bgt #15. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.