Missense mutation of a class B heat shock factor is responsible for the tomato bushy root-2 phenotype

The bushy root-2 (brt-2) tomato mutant has twisting roots, and slower plant development. Here we used whole genome resequencing and genetic mapping to show that brt-2 is caused by a serine to cysteine (S75C) substitution in the DNA binding domain (DBD) of a heat shock factor class B (HsfB) encoded by SolycHsfB4a. This gene is orthologous to the Arabidopsis SCHIZORIZA gene, also known as AtHsfB4. The brt-2 phenotype is very similar to Arabidopsis lines in which the function of AtHsfB4 is altered: a proliferation of lateral root cap and root meristematic tissues, and a tendency for lateral root cap cells to easily separate. The brt-2 S75C mutation is unusual because all other reported amino acid substitutions in the highly conserved DBD of eukaryotic heat shock factors are dominant negative mutations, but brt-2 is recessive. We further show through reciprocal grafting that brt-2 exerts its effects predominantly through the root genotype even through BRT-2 is expressed at similar levels in both root and shoot meristems. Since AtHsfB4 is induced by root knot nematodes (RKN), and loss-of-function mutants of this gene are resistant to RKNs, BRT-2 could be a target gene for RKN resistance, an important trait in tomato rootstock breeding. Gene & accession numbers SolycHsfB4a - Solyc04g078770. Supplementary Information The online version contains supplementary material available at 10.1186/s43897-022-00025-0.


Introduction
Root architecture is plastic and important for water and mineral absorption, anchorage and storage (Nibau et al., 2008). Changes in root function and architecture have resulted in enhancements for crop production (Hammer et al., 2009;Siddiqui et al., 2021), and much has been achieved to understand the genetic regulation of root system architecture and development, particularly in Arabidopsis (Motte et al., 2019). Breeding for improved root systems is of great interest for grafted vegetable production where elite scion genotypes with favourable aboveground traits are grafted onto rootstocks, especially in the solanaceous crops tomato, pepper and eggplant . When choosing rootstocks, the foremost interests are the overall yield, resistance against biotic and abiotic stresses and improved resource use efficiency (Martínez-Andújar et al., 2020). Rootstocks have been selected to challenge extreme conditions, such as low nutrient availability (Schwarz et al., 2013), hydric stress (Sánchez-Rodríguez et al., 2012), high salinity (Santa-Cruz et al., 2002) and pest control (Gregory et al., 2013;Gálvez et al., 2019). Breeding for tomato rootstocks requires an understanding of the genetic variation for root traits and the available germplasm resources that can be applied to rootstock breeding .
In tomato there have been relatively few investigations linking root traits to loci and genes. Although a recent study developed methods to rapidly identify seedling root mutants via EMS mutagenesis in the dwarf tomato cultivar Micro-Tom (Alaguero-Cordovilla and Belén Sánchez-García, 2021), the mutant collection of the C.M. Rick Tomato Genetics Resource Centre (TGRC, David, California), already includes 15 monogenic tomato mutants with distinctive root phenotypes (Table 1). For the further understanding of molecular processes impacting root development and rootstock characteristics in tomato, we have investigated several of these root mutants. One of these mutants, bushy root-2 (brt-2), possess a twisting tap root, and lateral roots were reported to arise at high density giving a bushy appearance (Voland and Zobel, 1988). The lateral and basal roots also curl and twist, and the shoot growth of brt-2 is relatively slower than other tomato lines. The brt-2 mutant was previously crossed with a series of classical tomato mutants and was found to be linked to four mutant loci: clausa, fulgens, entire and divergens, indicating that the brt-2 locus maps at 40-45 cM on chromosome 4 (Voland and Zobel, 1988). The entire locus showed the closest linkage with brt-2 and was subsequently identified as a single-base deletion in the SlIAA9 gene (Solyc04g076850), a transcriptional repressor of auxin signalling impacting leaf morphogenesis and fruit development (Zhang et al., 2007).
Here we aim to identify the causative gene for brt-2 phenotype. Next generation sequencing (NGS) technologies have made studies linking phenotype to genotype Adventitious roots on stem from soil level to considerable height above Yes unknown (Philouze, 1971) aerial roots-2 (aer-2) Abundance of root initials along the stems. Yes unknown (Kerr, 1982) baby lea syndrome (bls) Anthocyaninless; restricted root system; short internodes, leaves, and trusses No chr 3 (Clayberg et al., 1966) bushy root (brt) Radical branches early; radical and root profusely branched; root tips twisting upwards, not dense growth.

diageotropica (dgt)
Plant habit prostrate due to reduced gravitropic response; growth retarded; stems and leaves droopy; cotyledons concave. Roots grow horizontally rather than downwards; no lateral root formation.
No unknown (Stubbe, 1964) ridged (ri) Ridged leaves; retarded growth of shoots and roots. No chr 6 (Lindstrom, 1933) Root suppressed (Rs) Greatly restricted or no root development Yes chr 4 (Yu, S. -a. and Yeager, A.F., 1960) umbrosa (um) Mature leaves darker green, wilted appearance; later growth stunted; reduced root growth No chr 1 (Stubbe, 1958) wilty dwarf (wd) Greyish-green, droopy leaves; stunted plants; leaves droop when drought stressed Yes chr 9 (Rick and Khush, 1961) faster and cheaper. The tomato reference genome based on cv. Heinz 1706 (Sato et al., 2012) is extensively used for SNP identification between different genomes. SNP and InDel polymorphisms are described in over 500 accessions of tomato (Aflitos et al., 2014;Kim et al., 2014;Lin et al., 2014), and a pan genome (Gao et al., 2019) and genome-wide structural variants (Alonge et al., 2020) are also described. These data sets and low-cost whole genome NGS in tomato make fine genetic mapping of mutants highly amenable. In this study we identify an excellent candidate gene for brt-2 through sequencing and fine mapping, and use grafting and microscopic analyses to further define the brt-2 phenotype.

Observations of the brt-2 phenotype
The TGRC entry describes the bushy root-2 tomato line (LA3206) as a spontaneous mutant within an unknown genetic background showing severely stunted growth and possessing dense, bushy, twisted roots. Compared to AC, brt-2 roots exhibited a strongly decreased growth, which is already noticeable at the cotyledon stage (Fig. 1a). The four-week-old brt-2 plants also showed decreased shoot development and curly roots with visibly reduced root length density, generally lacking finer lateral roots (Fig. 1b). The established brt-2 plants possessed shorter shoot (Fig. 1c) and extremely reduced root system compared to AC with striking difference ( Fig. 1d and e). The young leaves of brt-2 were epinastic (laminar and petiole tending to curve downwards) and the leaves were observed to have a tendency to wilt in the glasshouse in well-watered conditions when evaporative demand was high (high temperature, low relative humidity and high incident solar radiation). Despite the decreased root system and wilted shoot, the brt-2 plant produced 7-8 cm sized fruit with high seed set; however they had a high tendency for radial cracking (Fig. 1f). When the mutant line was grown in an aeroponic Fig. 1 The brt-2 phenotype. a comparison of seedling growth of brt-2 and AC. b comparison of 4 weeks old AC and brt-2 lines. c 9 weeks old AC and brt-2 plants possess different shoot size. d 9 weeks old AC produces significantly more roots than brt-2 plants e in 10 L pots. f The brt-2 tomatoes usually possess cracks system, the size difference compared to AC was even more stark (data not shown).

Reciprocal grafting of AC and brt-2
In order to investigate the tissue specific impacts of the brt-2 locus on the whole plant phenotype, we made reciprocal grafting with AC in the four possible combinations including self-grafted genotypes (shoot/root): AC/ brt-2, brt-2/AC, AC/AC and brt-2/brt-2. After 9 weeks the shoot and root dry weights (DW) were obtained ( Fig. 2; Table S1). Self-grafted AC/AC plants were significantly larger than brt/brt plants, confirming the negative effect of the brt-2 mutation observed ( Fig. 1) and providing quantification of the difference; shoot DW and root DW were both 3.4-fold greater in AC/AC vs brt-2/brt-2. The AC shoot in AC/brt-2 grafts was much smaller than in AC/AC grafts, indicating that the mutant rootstocks impaired shoot growth. In contrast, both AC and brt-2 scions showed similar growth when grafted onto AC rootstocks (although the brt-2 shoot DW was 24% less than AC shoot growth, this was not statistically significant). The dry weight of brt-2 shoots was increased 3-fold by using AC rather than brt-2 as rootstock, but the AC scion was not able to increase the brt-2 root mass in the AC/brt-2 graft compared to brt-2/ brt-2. These data indicate that the effects of the brt-2 mutation are only expressed when the mutation is present in the root; a shoot containing the brt-2 mutation grows normally if its rootstock carries the wild type (WT) allele (brt-2 + ).

Genetic mapping of the brt-2 locus
The brt-2 mutant line was crossed with AC to create an F 2 population that segregated with WT (219 lines) and brt-2 (69 lines) phenotype with an approximate 3:1 ratio ( Fig. S1). This classifies brt-2 as a monogenic, recessive trait confirming the brt-2 locus description of the TGRC database.
To identify polymorphisms for fine mapping we initially used genotype-by-sequencing (Elshire et al., 2011), but since very few polymorphisms were obtained within this cross, whole genome resequencing was performed. Based on this data we designed and tested KASP markers for six SNPs polymorphic between AC and brt-2 lines between~40 and 65 Mbp on chromosome 4 (Table S2; Fig. 3). Of the 69 F 2 plants with the brt-2 phenotype, 37 were recombinant allowing brt-2 to be mapped between 59,032,422 and 65,276,012 bp (reference SL2.50), containing approximately 900 genes. Four more KASP markers were designed and scored within the recombinants and the mapping region was reduced to 1.9 Mbp between 62,760,651 and 64,623,394 bp ( Fig. 3).
Solyc04g078770 is the only candidate gene for brt-2 This 1.9 Mbp region contains approximately 250 genes, however, the NGS analyses revealed only ten SNP/InDel variations in the brt-2 parental sequence compared to AC ( Table 2). Five of the ten variations were located in intergenic regions, distantly from genes. Two of the gene-related variations were within intron sequences (Solyc04g078000, Solyc04g080010), and one was in a putative promoter sequence (Solyc04g080020). The remaining two SNPs caused amino acid modifications, a serine-cysteine change in Solyc04g078770 and a prolinehistidine conversion in Solyc04g080120 (Table 2). When all ten sequence variations are compared to the 150 Tomato Resequencing Project (Aflitos et al., 2014) and Tomato 360 Resequencing Project (Lin et al., 2014), only two of these changes were unique for the brt-2 line. One is an intergenic SNP at 62,760,651 bp, already excluded from the mapping interval by the marker at this position  (Table S1). Standard errors are indicated, Least Significant Difference (LSD) of the one-way ANOVA are marked (p < 0.05) (Fig. 3). The other is the A/T change causing a Ser to Cys (S75C) substitution in the 1st exon of Solyc04g078770 at 63,443,492 bp. With a further two KASP markers ( Fig. 3; markers at 63,443,492 and 63,565,771) we tested the linkage of the Solyc04g078770 mutation to the brt-2 phenotype in the 69 recombinant F 2 lines ( Table 2). The A/T change in Solyc04g078770 was the only polymorphism that showed 100% linkage with the brt-2 phenotype, thus formally defining the mapping interval to 0.8 Mbp (62.76 Mbp to 63.57 Mbp), a region only containing three polymorphisms, two of which could be excluded because they were common to other tomato accessions lacking the brt-2 phenotype. So, finally the 63,443,492 SNP in Solyc04g078770 was the only unique polymorphism in the mapping interval, it co-segregated with the brt-2 phenotype, and caused an amino acid change; this evidence indicates very strongly that the mutation in Solyc04g078770 causes of the brt-2 phenotype.
The S75C mutation is predicted to have a large impact on the "class B heat shock factor" protein function Solyc04g078770, also known as SolycHsfB4a (Berz et al., 2019) codes for a heat stress transcription factor, HsfB4a, which is the orthologue of the Arabidopsis SCHIZORIZA (SCZ) protein (ten Hove et al., 2010). SCZ is a member of a large gene family containing a highly conserved Hsf DNA-binding domain (DBD) motif  Unique variants are not present in the re-sequenced lines of SGN database. KASP marker accessibility and the linkage to the brt-2 phenotype are marked. The proposed causative gene for brt-2 is in bold in the first part of the coded protein (Ahn et al., 2001). We compared the DBD domain of SolycHsfB4a with orthologue proteins in other plant species to investigate the impact of the amino acid change in the brt-2 line (Fig. 4). The serine/cysteine replacement occurs in an extremely conserved part of the DBD domain, therefore we tested if the mutation has potential effects on the protein function. In the PROVEAN software, the S75C mutation scored − 4.858. PROVEAN scores < − 2.5 denote a potential functional shift (Choi et al., 2012) and so S75C is indeed predicted to cause a critical change in SolycHsfB4a function.

Microscopic analyses of brt-2 roots
The A. thaliana SCHIZORIZA gene is involved in root development (Mylona et al., 2002), therefore we investigated related root phenotypes in the brt-2 mutant. An image analyses showed that, compared to AC roots, the brt-2 line possesses a drastically increased root cap and cell division zone with a large number of extra cells especially in the division zone (Fig. 5). The presence of this extra tissue is associated with the separation from the root tip of lateral root cap cells and multicellular fragments (Fig. 5b-c); there are also numerous cells that have separated from the columella tip, remaining slightly distanced from the main tissue in Fig. 5d, whereas this was not observed for AC. It is likely that this detachment of root cells might be promoted by the microscopy sample preparation method, with the cells more easily dislodged by physical manipulation in brt-2 than in AC. Similar root cap cell separation was already described in lines with altered AtHsfB4 (SCZ) function (Begum et al., 2013;ten Hove et al., 2010); this phenotypic similarity in an orthologous gene strongly supports the genetic data that indicates that the S75C mutation in SolycHsfB4a is responsible for brt-2.

Discussion
The brt-2 mutant shows a perturbed root phenotype leading to delayed shoot development The brt-2 mutant is a member of the large monogenic mutant collection of TGRC which contains more than 1000 mutant lines. Among these, brt-2 is one of only a few lines defined as primary root mutants. Despite normal seed germination, the brt-2 mutant shows a severe phenotype with bushy, twisted roots during the early stages of seedling development, accompanied by delays in general plant growth. These curling roots lack fine lateral roots, further reducing the overall root system size. Compared to AC, fully established brt-2 mutant plants were observed to have a greater tendency to wilt in strong sunlight, likely indicating that the brt-2 root system is unable to meet the demand for water at high transpiration rates. Even though brt-2 has delayed shoot growth, it is able to undergo normal fruit development and seed set. These fruits usually exhibit cracks which is consistent with intermittent water stress due to poor root functioning; an episode of water stress may lead to reduced extensibility of the epidermis, followed by resumption of pericarp expansion, so creating increased tissue tension and cracking (Khadivi-Khub, 2015). We used the grafting capability of tomato to investigate whether the causative mutation is primarily acting through changes in the root system, or if it directly influences the scion growth as well. The self and reciprocal grafting between AC and brt-2 clearly showed that the phenotype is determined by the root genotype, indicating a local effect of the gene altering root development and a secondary effect on the shoot due to impaired root function.
Genetic mapping revealed SolycHsfB4a as candidate gene for brt-2 For mapping brt-2, we created a segregating F 2 population crossing the mutant line with a widely used, indeterminate tomato cultivar, Ailsa Craig. Resequencing by NGS of both parental lines identified a relatively low number of DNA polymorphisms on chromosome 4, such that a genotyping-by-sequencing (GBS) approach was not feasible, indicating close similarity between parents. The NGS polymorphisms allowed brt-2 to be mapped to~2 Mbp on the long arm of chromosome 4, and the low number of polymorphisms was very helpful in this case as only one clear candidate polymorphism was found.
The S75C mutation in SolycHsfB4a was a unique allele that displayed 100% co-segregation with the phenotype. No other polymorphisms were detected in the mapping interval that could explain the brt-2 phenotype, leading to the conclusion that SolycHsfB4a was the only candidate and the causative gene for brt-2. The physical distance between SlIAA9 (entire) and SolycHsfB4a is 1.68 Mbp, consistent with the low recombination frequency previously observed between the two loci (Voland and Zobel, 1988).
The brt-2 allele of SolycHsfB4a contains a uniquely recessive DNA binding domain variant The secondary structure of Hsf DNA binding domain (DBD) consists of three-helix bundles enveloped with a four-stranded antiparallel β sheet (Harrison et al., 1994). The order of these structural elements within the DBD is α1-β1-β2-α2-α3-β3-β4 located at the amino terminus of the protein; this pattern is unchanged among the different Hsf gene family members. For DNA binding to occur a trimer of Hsf polypeptides is formed.
The S75C brt-2 mutation is located in the conserved Hsf DBD domain and a blast search revealed that S75 in the sequence context SFVRQ is absolutely conserved across all eukaryotic organisms and in all members of the Hsf family (Lv et al., 2014). A crystal structure for human Hsf1 shows that the equivalent of S75 (S68 in human Hsf1) occurs within the α3 helix that contacts the major groove of the DNA within the specific nucleotide sequence of the heat shock element (HSE), forming a hydrogen bond with the DNA phosphate backbone (Neudegger et al., 2016). The S75C mutation leads to replacement of a single atom (oxygen of serine exchanged for sulphur of cysteine) and would be expected to disrupt the hydrogen bond, likely weakening the binding of BRT-2 protein to DNA. Since the HsfB4a class lacks the transcriptional activator domain of other Hsf proteins, it is believed to repress transcription by binding to and "blocking" the HSE; thus S75C is likely to reduce the repressor activity of SolycHsfB4. Interestingly, in human Hsf4 mutants that cause congenital lamellar cataracts, all known amino acid substitutions located in the DBD are dominant negative mutations because of the formation of dysfunctional heterotrimers in heterozygous cells (Berry et al., 2017;Jiao et al., 2019). However, the brt-2 mutation is highly unusual in being within the DBD, but also fully recessive. The S within SFVRQ has not been reported to be mutated in human HSFs linked to congenital disease, or in any other natural variants, and thus brt-2 appears to have a recessive DBD mutation not previously described in the extensive literature on HSFs in many eukaryotic organisms. In SCZ (discussed below), the reported allelic series of loss-of-function mutations are all outside the DBD and recessive (ten Hove et al., 2010). The recessive nature of the S75C mutation in the DBD could indicate that both DNA binding and trimer formation are disrupted, e.g. by a major disruption in protein folding, effectively creating a null mutant. Perhaps a less likely explanation, given the highly conserved motif SFVRQ, is that S75C causes of a gain-of-function of the DBD that is negated by trimerization with wild type polypeptides in the heterozygote, making it recessive. In general, recessive gain-of-function mutations occur rarely (Liu et al., 2020).

The tomato SolycHsfB4a gene is orthologous to Arabidopsis SCHIZORIZA
The closest homologue of SolycHsfB4a is At1g46264 in Arabidopsis, also known as SCHIZORIZA (SCZ) or AtHsfB4 which has its highest expression in root and shoot apices of A. thaliana (Winter et al., 2007;Begum et al., 2013). Similarly, in the TomExpress database (Zouine et al., 2017) SolycHsfB4a has high expression in root, and leaf and shoot meristematic tissues (Fig. S2, Table S3); it is therefore not root specific despite the clear evidence from reciprocal grafting that the main effect of the mutation acts in the root, evidence that has not been reported before for SCZ where grafting is more technically challenging. This appears to be an example where transcription profile and gene function do not coincide. The CoNekT database (Proost and Mutwil, 2018) shows mRNA levels in different root zones: meristematic, elongation, differentiation, root hairs and bulk root. This data (Fig. S3, Table S4) shows that, within roots, SolycHsfB4a is most highly expressed in the meristematic zone (25.8 TPM), with the next highest expression in the elongation zone (5.1 TPM). Similarly, the AtHsfB4 promoter directs GUS expression to root meristem tissue, and is specific for stele, cortex, endodermis and the quiescent centre (QC) (Begum et al., 2013).
Even though HsfB4 is a member of the large gene family of heat shock transcription factors (Hsf), heat stress activation is not the unique functional trigger among the rather diverse members (Swindell et al., 2007). Arabidopsis HsfA6a and HsfA6b have increased expression in samples treated with osmotic, salt, and cold stress, while HsfB1, HsfA2, HsfA4a, HsfA4c and HsfB2a are induced upon biotic stress (von Koskull-Döring et al., 2007). Seventeen Hsf genes were isolated from a wild diploid woodland strawberry and they were induced by various abiotic and biotic stresses (Hu et al., 2015). Arabidopsis has 21, and tomato 24 Hsf genes (Tang et al., 2016). Hsf proteins bind to heat shock elements (HSEs) within the promoters of target genes, including heat shock protein (HSP) genes that act as molecular chaperones in a wide range of plant responses to abiotic and biotic stimuli and during plant development.
The mutant of SolycHsfB4a, brt-2, has a similar phenotype to Arabidopsis lines with perturbed expression of SCZ (AtHsfB4) SCZ encodes a nuclear protein regulating cell division asymmetry in A. thaliana roots through coordinated action with SCARECROW (SCR) (Mylona et al., 2002;ten Hove et al., 2010;Pernas et al., 2010): together these two genes direct the development of the root cap, epidermis and ground tissue (cortex and endodermis). The scz transposon knock-out mutant shows a perturbed, asymmetric cell division pattern that can be seen in the root meristem from the torpedo embryo stage (Pernas et al., 2010). The knock-out mutant has a range of complex phenotypes: additional and aberrant periclinal cell divisions; subepidermal cell layers that produce root hairs; a disorganised arrangement of thrichoblasts and atrichoblasts; and supernumerary layers of epidermal cells and/ or lateral root cap cells that become less distinct and have a tendency to separate from the root as it matures (ten Hove et al., 2010;Mylona et al., 2002;Pernas et al., 2010). This disrupted root function is coupled with a reduced shoot stature (Mylona et al., 2002) and reduced root growth (ten Hove et al., 2010). It was concluded that SCZ acts to restrict epidermal fate to the outer layers of the root, and is required to maintain the stem cells that give rise to cortex and endodermis and for normal QC development (Pernas et al., 2010). Since the QC represses the differentiation of columella stem cells and the scz mutant has a defective QC, scz also exhibits precocious differentiation of columella stem cells (Pernas et al., 2010).
The above studies on SCZ mutants in Arabidopsis have used knock-out mutants rather than amino acid substitution mutants (ten Hove et al., 2010;Mylona et al., 2002;Pernas et al., 2010), thus the S75C mutation in SolycHsfB4a might behave differently if it retains some functional aspects of the protein. It is therefore necessary to consider also the phenotypes observed in Arabidopsis AtHsfB4 overexpression (OE) lines. AtHsfB4 is normally expressed specifically in ground tissue cells and the QC, so OE of AtHsfB4 using CaMV 35 s promoter would increase expression in these same tissues, but also ectopically in other root tissues (Begum et al., 2013). The AtHsfB4-OE lines showed specific root morphological changes: while the aerial parts of the plant looked normal, there was a clear delay in root growth and development of HSFB4-OE lines compared to WT. Microscopic images revealed distinct structural changes at the root surface, thickening of the meristematic region showing a rough surface and a detachment of cells in the elongation and maturation root zones. HSFB4-OE lines have additional periclinal divisions which lead to the generation of extra cell layers in the ground tissue (cortex and endodermis) and additional layers of lateral root cap cells were also observed. Thus, the production of extra abnormal layers of cells and cellular detachment is a feature common to both SCZ knock-out and over-expression.
The brt-2 line shows high cell proliferation around the root cap and the meristematic cells (Fig. 5). These extra cells are easily detached from the main root tissue, principally from the root cap columella, lateral root cap and possibly also the cell division zone. These complex features of cell proliferation and detachment are common between Arabidopsis SCZ knock-outs, SCZ ectopic overexpression lines, and the brt-2 mutant, so while this strongly supports brt-2 as orthologous to SCZ, it is not possible to be sure from the phenotype if the S75C mutation in brt-2 is causing a complete loss or an alteration of protein function. However, since the brt-2 mutant allele is recessive, it seems more likely that the brt-2 mutant protein has lost both its HSE binding function (losing its repressor function) and its ability to form trimers, so preventing it from behaving like other DBD mutants which are dominant through the poisoning of trimers in the heterozygote.
Potential for BRT-2 alleles to provide root-knot nematode (RKN) resistance RKN resistance is an essential goal in rootstock breeding to avoid significant crop losses (Okorley et al., 2018). In tomato, it was previously found that silencing HSFA1a compromised Mi-1.2-mediated RKN resistance by preventing the hypersensitive response (Zhou et al., 2018). However, the roles of other tomato HSF genes in RKN resistance are unknown.
RKNs strongly induce the expression of SCZ/AtHsfB4 in Arabidopsis roots as part of the process leading to organogenesis of root galls, specialised structures that develop in the vascular tissue of roots and provide nourishment to the RKNs (Olmo et al., 2020). Moreover, three loss-of-function scz mutants in Arabidopsis showed "a severe decrease in nematode infection and reproduction", whereas SCZ overexpression and loss-offunction of the related genes AtHSFA1a and AtHSFA1b had no effect, indicating a specific role of AtHsfB4 lossof-function in RKN resistance (Olmo et al., 2020). It was suggested that recruitment of the host pathways for root apical meristem generation is part of the mechanism by which RKNs generate galls, and that disruption of these pathways, including by inactivation of HSFB4, could provide RKN resistance (Olmo et al., 2020). SCZ loss-offunction was associated with an anecdotal reduction in shoot stature; although this has not been quantified in Arabidopsis (Mylona et al., 2002), there is a likely tradeoff between root function and RKN resistance.
Here we provide the first description of a tomato HSFB4 mutant (brt-2). Although it is highly disruptive to root and shoot growth, the report by Olmo et al. (2020) suggests that this material will be RKN resistant and there remains the possibility that natural or engineered functional or expression variants of BRT-2/ SolycHsfB4a that combine acceptable scion growth rates coupled with RKN resistance could be developed and deployed in tomato rootstock cultivars.

Plant material and growth
Tomato cultivar Ailsa Craig carrying an introgression from Solanum peruvianum on chromosome 9 with the resistant allele of the Tobacco mosaic virus resistance-2 locus (Tm-2 a ) was used as one parent (AC). This was crossed with the bushy root-2 (brt-2) mutant line (TGRC accession LA3206), generating F 1 plants that were selfpollinated to produce F 2 seeds for use as a mapping population. Seed accessions are given in Fig. S1.
Seed were extracted from red ripe tomato fruits and separated seeds and gel were incubated overnight after adding 2-3 volumes of 0.12 M hydrochloric acid (Sigma-Aldrich), 1 g L − 1 brewer's pectolase (Ritchie, Burtonupon-Trent, UK) at room temperature. The seeds were washed thoroughly in tap water and dried at room temperature for at least five days. Before germination, all seeds were sterilised in 0.45% w/v sodium hypochlorite for 30 min and then rinsed in tap water to avoid seed-borne viral transmission. Seed were germinated as described (Silva Ferreira et al., 2018) before transplanting into 8 L pots of Sinclair multipurpose compost (LBS Horticulture, Colne, UK). Pots were irrigated according to demand and were fed twice a week with Hoagland solution (5 mM K 2 SO 4 ; 1 mM H 3 PO 4 ; 5 mM Ca (NO 3 ) 2 ; 2 mM MgSO 4; 100 μM EDTA Fe-Na; 42.2 μM H 3 BO 3 ; 9.1 μM MnCl 2 ; 0.76 μM ZnSO 4 and 0.32 μM CuSO 4 , pH 5.8 adjusted with H 3 PO 4 ), at half strength before flowering and full strength after flowering. Four-week-old plants were phenotyped for the brt-2 trait and young leaf material used for DNA extractions.

Grafting experiments
Pre-germinated AC and brt-2 seeds were sown in 24module standard seed trays with multipurpose compost and were grown in the glasshouse. Three-week-old plants were grafted in all combinations following the Japanese top-grafting method using silicon tube-shaped clips (Rivard and Louws, 2006). After grafting, plants were transferred to a healing chamber shaded from direct sunlight and providing 100% humidity levels via a LT1 Mist-Wean Controller connected to a Wet Leaf electrode (Access Irrigation, Northampton, UK) and two CoolNet Pro-4 fogging heads (Netafim, Hatzerim, Israel); the controller generated 5 s of water misting repeating after a 15 min delay and influenced by a level 2 sensitivity threshold. Plants were weaned from the healing chamber six days after grafting by reducing humidity over three days, and were transplanted into 22 cm diameter, 10 L pots filled with Sinclair multipurpose compost placed on a bench in the glasshouse, and hand watered on demand. Nine-week-old grafted plants were assessed: roots were washed from the compost, and root and shoot dry weights (DW) were measured.

DNA extraction and KASP genotyping
Genomic DNA extraction from young leaves and the KASP/KBD assays were performed as described (Silva Ferreira et al., 2018). All KBD assays were developed by LGC (Teddington, UK) based on the provided nucleotide polymorphism and flanking sequence data (Table  S1). The KASP genotyping results were analysed in CFX96 qPCR machines using the "Allelic Discrimination" feature of CFX manager software (BioRad, Watford, UK).

NGS genomic data generation and sequence analysis
Genomic DNA from AC and brt-2 plants was extracted using the DNeasy plant mini kit (Qiagen; Manchester, UK), according to the manufacturer's instructions. Both lines were subjected to paired-end sequenced using Illumina HiSeq X platforms. The data comprised 389,240,368 (AC) and 431,168,724 (brt-2) 100 bp reads representing~39x and~43x coverage of average read depths, respectively. Data is available from SRA accession PRJNA750735 (NCBI). Reads were aligned to the SL2.50 (Heinz 1706) reference genome and variants were called using the "Alpheus" pipeline (Miller et al., 2008). AC possessed 1,139,329 and brt-2 193,743 sequence variants compared to Heinz 1706. The resulting VCF files were loaded into Integrative Genomics Viewer (IGV) to visualise the sequence variations between the parental genomes along the 12 chromosomes (Robinson et al., 2011) and to design KASP markers for the genetic mapping procedure.
For protein comparisons, the open source BoxShade Server (version 3.21, EMBnet) was used with the default values. The BoxShade program used the Multiple Sequence Alignment (MSF) files generated by ClustalW Multiple Alignment feature (with default values) of BioEdit version 7.2.5 (Hall, 1999). The PROVEAN (Protein Variation Effect Analyser) tool was used to predict whether an amino acid substitution would impact on the biological function of a protein (Choi et al., 2012).

Root microscopy
AC and brt-2 seeds were germinated as described above. Seeds were placed on moistened filter paper in Petri dishes sealed with Parafilm™ to maintain humidity and covered with foil to exclude light. They were left in a growth room at 22°C for 7 days. For the microscopic studies, approx. 1 cm lengths of root (with root tip) were removed with a razor blade and placed in water on a cavity slide, a coverslip was mounted on top. Microscopy was carried out with a Leica DM6 B Compound Microscope, "Brightfield" and "Differential Interference Contrast" methods were used. Images were captured using 10x and 20x objectives via a Zeiss Axiocam 506 colour (6 Megapixel) microscope camera. The image acquisition and storage software IMS V18Q4 (Imagic Imaging Ltd), was used to capture conventional single images, and extended depth of field images via the software's "multifocus live" mode to generate composite images.