A nonS-locus F-box gene breaks self-incompatibility in diploid potatoes

Potato is the third most important staple food crop. To address challenges associated with global food security, a hybrid potato breeding system, aimed at converting potato from a tuber-propagated tetraploid crop into a seed-propagated diploid crop through crossing inbred lines, is under development. However, given that most diploid potatoes are self-incompatible, this represents a major obstacle which needs to be addressed in order to develop inbred lines. Here, we report on a self-compatible diploid potato, RH89-039-16 (RH), which can efficiently induce a mating transition from self-incompatibility to self-compatibility, when crossed to self-incompatible lines. We identify the S-locusinhibitor (Sli) gene in RH, capable of interacting with multiple allelic variants of the pistil-specific S-ribonucleases (S-RNases). Further, Sli gene functions like a general S-RNase inhibitor, to impart SC to RH and other self-incompatible potatoes. Discovery of Sli now offers a path forward for the diploid hybrid breeding program.

1. The research described in this paper is well thought out, thorough and of high quality. Unfortunately the same cannot be said of the written English in the manuscript text. While generally comprehensible, the text contains too many grammatical errors. The lack of line numbering makes it impractical to list all of these, which in any case should not be the responsibility of the reviewer. I strongly recommend having the manuscript thoroughly edited to improve grammar, word usage and sentence structure.
2. The authors fail to cite a recent similar work on Sli. On page 3, paragraph 2, the statement 'Unfortunately, the Sli gene has not been cloned yet.' is not entirely accurate. Eggers et al. (2020) reported cloning of Sli: Eggers et al. (2020) The S-Locus Inhibitor gene encodes an F-box protein with a Lectin domain and crucially enables hybrid potato breeding. Solanaceae International Online Meeting. Granted, this was an oral presentation, but the abstract is available online (ftp://ftp.solgenomics.net/sgn_homepage/sol2020/InternationalSOL_Meeting2020_AbstractsBook2 0201108.pdf).
3. The introduction section leaves the impression that SI is the only factor preventing the breeding of potatoes at the diploid level. Another hurdle is inbreeding depression caused by genetic load, which is expected to be high in tetraploid potatoes, as well as SI diploids, due to their high level of heterozygosity. Converting the breeding system of cultivated potatoes from SI to SC requires only a mutation in one gene (Sli in this case, but S-RNase or SLF mutations would also suffice), whereas overcoming inbreeding depression and purging deleterious mutations could prove more challenging in the long run if they necessitate genome wide selection across multiple generations. 4. Page 4, paragraph 2. I do not understand why F1 PI 225689 x RH would produce a/a progeny (assuming A=Sli, and a=WT), unless the RH parent was used as female. If the PI accession was used as female (as implied by the way the cross is written), and if it is fully SI, then only Slibearing pollen should be compatible. This point needs to be clarified in the text. (Or I missed something.) 5. Only two transgenic diploid potato plants expressing the NSF (Sli) transgene were obtained, and the conclusions about the function of NSF are primarily based on the phenotypes of these two plants, both of which were SC. Identical NSF sequences were found in two other SC accessions (M6 and S. chacoense), supporting their conclusion that NSF is functionally the Sli gene. Nonetheless, the small number of primary transformants is a weakness of this paper. The authors state that it was difficult to obtain diploid primary transformants due to the high frequency of tetraploid regenerates (presumably tetraploidy resulted from the transformation/regeneration conditions used, although this is not explained). Nonetheless, it would be very helpful if the authors could include data from additional transgenic plants, perhaps self-progeny of either of the two primary transgenic plants, if they have any such data. This would strengthen their functional evidence that NSF is Sli, and that it is both required and sufficient for SC in an otherwise SI genetic background.
6. The authors do not address the advantages and disadvantages of NSF/Sli based SC over other types of SC mutations. One potential drawback highlighted by this study is the fact that NSF does not interact with all S-RNases in vitro (the Y2H experiments). Thus the Sli gene may not convert SI to SC in all genetic backgrounds and S-haplotypes. This limitation contrasts with S-RNase loss of function mutations, which are known in other Solanum spp. (S. pennellii, S. habrochaites and S. arcanum) and have a stable SC phenotype because pollen retain the ability to recognize ALL nonself S-RNases. This results in an SC phenotype that is dominant over SI in heterozygotes with ANY S-haplotype (except the nonmutant S-haplotype in which an S-RNase LOF allele arose). Another limitation of NSF/Sli-based SC is that the SC mutation will segregate independently of the S-locus, which allows SC to revert to SI, for example in progeny of SI x Sli crosses. Thus from a breeding standpoint, NSF/Sli has potential limitations, based on the results of this study. The authors mention that S-RNase knockouts face the obstacle of lack of public acceptance of GMO crops, which is a valid point, however non-GMO LOF mutations have been found in natural populations of other SI species, and they may exist in diploid potatoes too.
7. Figure 1a. The pollen tubes are not easily visible, at least in the PDF version of the figure. It may be necessary to adjust the resolution or contrast of this figure. It would also help to add arrows or other symbols to indicate the position of pollen tube arrest within the styles.
8. Figure 4. The model shows NSF recognizing S2/S3 S-RNases in the pistil. I realize this is probably just for illustration purposes, but the Y2H and Luc assays show that S3 S-RNase is NOT recognized by NSF. If pollen bearing the NSF allele fail to recognize BOTH S-RNases in the pistil, the reaction will be incompatible. The model would be improved by clarifying what happens in both compatible and incompatible reactions involving NSF. Secondly, the model presents an alternative path to SC via gain-of-function mutation in an SLF that recognizes 'self' S-RNase. That's fine, but the model could be improved by incorporating another known route to SC based on loss-offunction mutations in the S-RNase (i.e. without a GOF mutation in either SLF or NSF). LOF mutations in the S-RNase are genetically simpler than GOF mutations and are the predominant route to SC, at least in the wild tomatoes, which are closely related to potato. I appreciate that Sli is more analagous to an SLF GOF mutation, since both occur in pollen, but the model would be more complete if it also included S-RNase mutations.
9. Figure 2. The numbers on the pie chart (Fig 2b) are said to represent the numbers of 'species that transfer from SI to SC after crossing with RH...'. However, each slice of the pie represents a particular species or subspecies, so I'm guessing the numbers actually represent accessions within species/subspecies. In any case the numbers don't mean much without also knowing the total number of lines that were tested within each group. In the text (page 5), '110 of the 125 species showed SC', however it is not clear to which species the 15 accessions (?) that did not convert to SC belonged. Were they all in one species, or a few in each group? Also, on Fig 2c, the pollen and style samples presumably come from mature flowers, but this is not explicitly stated in the legend.
10. The author's use of the name 'NSF' (non S-locus F-box protein) instead of, or in addition to 'Sli' creates confusion by unnecessarily adding alternative terminology to the literature. The gene symbol Sli is well established, has precedence, and reflects the SC phenotype. Furthermore, there are MANY F-box encoding genes in the genome, most of which are NOT at the S-locus. Thus the name 'Non S-locus F-box' could conceivably be understood to refer to many other F-box genes, which might create some confusion. Why not stick with Sli?
-p2, par3: define 'SLF' at first use (i.e. S-locus F-box) -p2, par3 (and elsewhere): italicize genes or mRNAs, not gene products or proteins. -p3, par1: loss of S-RNase function has been documented not just through knock-out mutations, but also from amino acid substitutions that eliminate RNase activity. -p3, par1: replace 'degradation of this gene' by 'degradation of this protein' -p3, last paragraph: the Sli gene acts gametophytically, thus it is not entirely accurate to refer to it as 'dominant'. Suggest inserting 'or gametophytic factor' after 'dominant gene'. Also, the authors should clarify the direction of the initial cross (I assume PI female x RH male). -p4, par2: the chi-square statistic should be based on an expected 1:2:1 segregation, although it doesn't hurt to also compute the goodness-of-fit to the 1:1, as shown.
-p4: replace 'narrowe' with 'narrow' -supplementary figure 2b, d, f and g: explain what M1 and M2 represent. Also the figure would be improved by having nontransformed diploid and tetraploid samples as references. The empty vector control has presumably gone through transformation and regeneration also, so could easily have undergone a change in ploidy. -supplementary figure 3. The figure shows that the sequence of NSF from RH is 100% identical to that found in S. chacoense or M6. Therefore the figure could be simplified by presenting just the sequence of RH, and in the legend stating that the other two sequences were identical.
Reviewer #2 (Remarks to the Author): I am glad to have a chance to review the manuscript, entitled "A non S-locus F-box gene breaks self-incompatibility in diploid potatoes. A diploid hybrid breeding has now become a hot topic for potato breeders. The well-known self-compatibility-inducing gene Sli has opened the way to develop diploid inbred lines, but this gene has not been isolated and its molecular function has remained unknown. This article discovered that a responsible gene is PGSC0003DMG400016861, a F-box protein gene with the F-box domain and PP2 domain and that it works like a general S-RNase inhibitor. This is certainly worth to be published.
There are two major concerns on this article. The authors identified a NSF gene that imparted SC to RH. The genomic region containing the NSF has been supposed to be a candidate region for the Sli gene (Clot et al. 2020). The sequence similarity of this region is very high not only between RH and M6 but also between these and several famous cultivars (Clot et al. 2020). This similarity has been demonstrated in Supplementary Fig. 3 where 100% similarity among SC S. chacoense, M6 and RH is shown in the sequence of the NSF gene (although I can't understand very well about this sequence because start codon and stop codon are not indicated). These indicate that the allelic variation in the NFS gene sequence is not so important because similar sequence can be found in many genotypes irrespective of being SC or SI. More important could be the presence or absence of a 536 bp insertion in the promoter region. SC S. chacoense, M6 and RH have the same size insertion present at the same position? The comparison of the promoter region is highly requested.
The second major concern is on the genetic action of the NSF/Sli gene. The authors described, "Thus, only the pollen harboring the SC gene can penetrate the self style and fulfil fertilization to produce progeny, and all the F2 progeny would carry the SC gene and exhibit self-compatible phenotype". So, they suggested that the NSF/Sli is gametophytically expressed. In contrast, a sporophytic action for the Sli gene has been proposed by Hosaka and Hanneman (1998). Based on sporophytic action hypothesis, pollen grains produced from a Sli-carrying pollen parent are all compatible in self-crossing. Thus, from a cross between SC (Sli/-) and SI (-/-), the progeny plants are segregated to a ratio of 1 SC (Sli/-) : 1 SI (-/-). Self-crossing of the SC (Sli/-) results in a ratio of 1 Sli/Sli : 2 Sli/-: 1 -/-; however, Sli/Sli was presumed to be absent due to a tight linkage with lethality (Hosaka and Hanneman 1998). Based on the sporophytic action hypothesis, the F2 population in Fig. 1d would be segregated to a ratio of 2 SC : 1 SI. If the gametophytic action is employed, only pollen grains carrying NSF/Sli are fertilized and generated only SC F2 plants. In this context, expression of NSF/Sli in pollen grains would be segregated to a ratio of 1:1; a half pollen grains would produce mRNA in in situ mRNA hybridization of NSF in Fig. 2d. in situ mRNA hybridization of NSF is suggested to conduct not with maturing pollen grains in the anther but with mature, fertile pollen grains released from the anther. Please verify which is a likely action of NSF/Sli.

Minor comments
Under the heading "SC gene colocalizes with a segregation distortion region in RH" in Results, a segregation ratio of A/A:A/a=3030:2954≈1:1 was observed for the F2 progeny from crossing PI 225689 and RH. A/A and A/a genotypes were determined by marker analysis? Why did the authors speculate that A/a is a heterozygote SC and a/a is a homozygote SI plant? This is a question relating to the major concern described above. In the following paragraph, the authors described, "In total, 6624 F2 individuals produced from the 131 selfed F1 plants were used to narrow down the SC gene to a 64.61 kb interval (Fig. 1c)". From this sentence I thought that SC and SI were segregated among F2 individuals, so that the candidate region could be narrowed down. So, not all F2 individuals were SC. If my understanding is correct, A/a (SC) and a/a (SI) were identified in the F2 population and a sporophytic action should be proposed.
Under the heading "NSF is efficient in rendering SC to SI conversion in diploid potatoes" in Results, it is described that 125 SI diploid species were pollinated. apparently, 125 is not the number of species but probably the number of accessions or clones, plants, etc. Please be clarified. In addition, S. tuberosum ssp. andigena and ssp. tuberosum are tetraploid and self-fertile. Did the authors use dihaploid genotypes of these species? In the same paragraph, the sentence "The unchanged SI phenotype of 15 diploid lines after the crossing may be caused by unsuccessful pollination as the flowering time of these lines were different from that of RH or E172." is not clear. I assume F1 hybrids were investigated for self-compatibility. Does this sentence mean that the 15 diploid lines were not true hybrids, so that they were not SC? I think Fig. 4 is not necessary.
In Methods, the affiliation of Prof. K. Hosaka is incorrect (not "Obihiro University" but "Obihiro University of Agriculture and Veterinary Medicine"). In the same sentence, what kind of genotypes for Solanum chacoense (clonal or accession ID) was provided from Prof. Hosaka?
In Methods, the heading "Polygenetic analysis of NSF/Sli" should be changed to "Phylogenetic analysis of NSF/Sli". This manuscript is presented to demonstrate that a cloned gene (NSF) is the contributing factor to over-coming self-incompatibility (SI) in diploid potato. They provide fine-mapping, pollen tube growth, quantitative RT-PCR, in situ hybridization, plant transformation and yeast 2-hybrid assay to support their work. I think the authors may have cloned a gene that contributes to overcoming self-compatibility (SC) but they are not providing seed set in the transgenic plants to support their results. Our work in SI/SC tells us that fruit set does not always correlate with seed set. We also use stylar squashes to look at pollen tube growth. In a number of instances, we see pollen tube growth to the ovules but no seed set.
Another point to address is the phenotyping of plant for SI/SC. There is a great environmental influence on this phenotype. The authors' phenotyping methods lack detail and create doubt in the results.
I question why the authors had to generated over 3,000 transgenic events. Explanation would be valuable. I am also concerned why so many tetraploid events are recovered in the 26 events studied.
The plant material used was poorly detailed. Furthermore, Clot et al. (2020) provided a means to genotype the Sli region in potato lines. I feel knowing that genotyping would add confidence in the correct interpretation to the results observed. I also feel that the authors should more clearly address their work in the introduction. In the results the authors refer to markers M-1 and M-2. These markers are not referenced. Are they based upon Clot et al.?
Reviewer #1 (Remarks to the Author): The paper by Ma et al. reports isolation of the S-locus inhibitor (Sli) gene, which confers selfcompatibility to otherwise self-incompatible diploid potato strains. They show that the gene underlying Sli, which they name 'NSF' (Non-S-locus F-box), encodes a pollen-expressed F-box protein that recognizes several different S-RNase protein variants, leading to their subsequent degradation via the ubiquitin/proteasome pathway. The research is a significant contribution as it elucidates the molecular mechanism of a well-known self-compatibility trait that has been promoted as a way to convert potato, a difficult to breed autotetraploid, into a diploid crop amenable to faster genetic improvement via conventional inbreds and F1 hybrid seed. The paper is also interesting as it describes a mutant F-box gene that apparently acquired pollen expression and recognizes a subset of stylar S-RNase proteins, unlike the S-locus F-box proteins that are normally expressed in pollen and specifically recognize only one or two S-RNases. A strong point of the paper is that it demonstrates that the NSF protein recognizes some of the many diverse S-RNase protein variants, but not others. While these interactions were assessed using in vitro methods (Y2H), which may not reflect what happens in vivo, they provide strong evidence that NSF/Sli recognizes S-RNase proteins. (They did not test for interactions with other proteins required for SI expression in the style).

Response:
We thank Reviewer for this positive comment.
Significant issues: 1. The research described in this paper is well thought out, thorough and of high quality. Unfortunately the same cannot be said of the written English in the manuscript text. While generally comprehensible, the text contains too many grammatical errors. The lack of line numbering makes it impractical to list all of these, which in any case should not be the responsibility of the reviewer. I strongly recommend having the manuscript thoroughly edited to improve grammar, word usage and sentence structure.

Response:
We feel very sorry to make many mistakes in preparing the manuscript. In the revised version, the maintext has been carefully edited by consultants from EditSprings (www.editsprings.com/en/). Online Meeting. Granted, this was an oral presentation, but the abstract is available online(ftp://ftp.solgenomics.net/sgn_homepage/sol2020/InternationalSOL_Meeting2020_Abstra ctsBook20201108.pdf).

Response:
We thank Reviewer for this suggestion. The statement was modified and the recent work was cited in the revised manuscript (Page 3, line 13-15). We submitted this manuscript together with this group, and planned to make a back-to-back publication.
3. The introduction section leaves the impression that SI is the only factor preventing the breeding of potatoes at the diploid level. Another hurdle is inbreeding depression caused by genetic load, which is expected to be high in tetraploid potatoes, as well as SI diploids, due to their high level of heterozygosity. Converting the breeding system of cultivated potatoes from SI to SC requires only a mutation in one gene (Sli in this case, but S-RNase or SLF mutations would also suffice), whereas overcoming inbreeding depression and purging deleterious mutations could prove more challenging in the long run if they necessitate genome wide selection across multiple generations.

Response:
We thank Reviewer for this suggestion and have modified the introduction section by adding another big hurdle (inbreeding depression) in diploid hybrid potato breeding. (Page 2, line 29).
4. Page 4, paragraph 2. I do not understand why F1 PI 225689 x RH would produce a/a progeny (assuming A=Sli, and a=WT), unless the RH parent was used as female. If the PI accession was used as female (as implied by the way the cross is written), and if it is fully SI, then only Slibearing pollen should be compatible. This point needs to be clarified in the text. (Or I missed something.)

Response:
Although PI 225689 is a fully SI line (a/a type), it is possible that the S-locus F-box proteins harbored by RH pollen, not the Sli allele (A), can recognize and detoxify the S-RNase expressed in PI 225689. Thus, the a/a progeny was produced from RH crossed with PI 225689. 5. Only two transgenic diploid potato plants expressing the NSF (Sli) transgene were obtained, and the conclusions about the function of NSF are primarily based on the phenotypes of these two plants, both of which were SC. Identical NSF sequences were found in two other SC accessions (M6 and S. chacoense), supporting their conclusion that NSF is functionally the Sli gene. Nonetheless, the small number of primary transformants is a weakness of this paper. The authors state that it was difficult to obtain diploid primary transformants due to the high frequency of tetraploid regenerates (presumably tetraploidy resulted from the transformation/regeneration conditions used, although this is not explained). Nonetheless, it would be very helpful if the authors could include data from additional transgenic plants, perhaps self-progeny of either of the two primary transgenic plants, if they have any such data. This would strengthen their functional evidence that NSF is Sli, and that it is both required and sufficient for SC in an otherwise SI genetic background.

Response:
Currently, gene transformation rate to produce positive diploid lines is low, as 24 positive transgenic lines were changed to tetraploids. This unexpected alteration may occur during regeneration process, since this process tends to induce genome instability 1 . We did the selfprogeny of these two transgenic lines. Each line produced a lot of seeds after selfing. This information has been added to Fig. 1g to strengthen the functional evidence of NSF/Sli as suggested. We thank Reviewer this suggestion. 6. The authors do not address the advantages and disadvantages of NSF/Sli based SC over other types of SC mutations. One potential drawback highlighted by this study is the fact that NSF does not interact with all S-RNases in vitro (the Y2H experiments). Thus the Sli gene may not convert SI to SC in all genetic backgrounds and S-haplotypes. This limitation contrasts with S-RNase loss of function mutations, which are known in other Solanum spp. (S. pennellii, S. habrochaites and S. arcanum) and have a stable SC phenotype because pollen retain the ability to recognize ALL non-self S-RNases. This results in an SC phenotype that is dominant over SI in heterozygotes with ANY S-haplotype (except the nonmutant S-haplotype in which an S-RNase LOF allele arose). Another limitation of NSF/Sli-based SC is that the SC mutation will segregate independently of the S-locus, which allows SC to revert to SI, for example in progeny of SI x Sli crosses. Thus from a breeding standpoint, NSF/Sli has potential limitations, based on the results of this study. The authors mention that S-RNase knockouts face the obstacle of lack of public acceptance of GMO crops, which is a valid point, however non-GMO LOF mutations have been found in natural populations of other SI species, and they may exist in diploid potatoes too.

Response:
We thank Reviewer for this suggestion. The limitation of using NSF/Sli to break SI and the potential application of the non-GMO LOF S-RNase in diploid potatoes have been added to the maintext (Page 7 line 4-14). As Reviewer mentioned, we found the non-GMO S-RNase LOF mutations do exist in natural potato populations, such as PG6359 2 . Since Sli is located on Ch12 and S-RNase is on Ch01, usage of SC genes from different sources in breeding can effectively avoid future genetic bottlenecks at these loci, considering that S-RNase is near centromere while Sli is near telomere.
7. Figure 1a. The pollen tubes are not easily visible, at least in the PDF version of the figure. It may be necessary to adjust the resolution or contrast of this figure. It would also help to add arrows or other symbols to indicate the position of pollen tube arrest within the styles.

Response:
The figure has been replaced by a high resolution one and arrow is used to indicate the arrest of pollen tube as suggested. We thank Reviewer for this suggestion.
8. Figure 4. The model shows NSF recognizing S2/S3 S-RNases in the pistil. I realize this is probably just for illustration purposes, but the Y2H and Luc assays show that S3 S-RNase is NOT recognized by NSF. If pollen bearing the NSF allele fail to recognize BOTH S-RNases in the pistil, the reaction will be incompatible. The model would be improved by clarifying what happens in both compatible and incompatible reactions involving NSF. Secondly, the model presents an alternative path to SC via gain-of-function mutation in an SLF that recognizes 'self' S-RNase. That's fine, but the model could be improved by incorporating another known route to SC based on loss-of-function mutations in the S-RNase (i.e. without a GOF mutation in either SLF or NSF). LOF mutations in the S-RNase are genetically simpler than GOF mutations and are the predominant route to SC, at least in the wild tomatoes, which are closely related to potato. I appreciate that Sli is more analagous to an SLF GOF mutation, since both occur in pollen, but the model would be more complete if it also included S-RNase mutations.

Response:
We thank Reviewer for this suggestion. The model was modified to show both compatible and incompatible reactions involving NSF/Sli. The SI to SC transition caused by LOF mutations of the S-RNase was also presented as suggested.
The model is modified to show loss of function mutations of S-RNases to confer SC as suggested.
9. Figure 2. The numbers on the pie chart (Fig 2b) are said to represent the numbers of 'species that transfer from SI to SC after crossing with RH...'. However, each slice of the pie represents a particular species or subspecies, so I'm guessing the numbers actually represent accessions within species/subspecies. In any case the numbers don't mean much without also knowing the total number of lines that were tested within each group. In the text (page 5), '110 of the 125 species showed SC', however it is not clear to which species the 15 accessions (?) that did not convert to SC belonged. Were they all in one species, or a few in each group? Also, on Fig 2c, the pollen and style samples presumably come from mature flowers, but this is not explicitly stated in the legend.

Response:
We thank Reviewer for this correction and feel very sorry for this mistake. It is 'accession', not 'species'. We have modified the wording in the revised manuscript. The total number we used in this experiment is 125 lines, and 15 lines failed to cross with RH or E172. The species information of the 15 exceptions was provided in Supplementary Table 1. The total number of accessions within each species was also added to the maintext (Page 5, line 27-32). We modified the legend of Fig. 2c to describe the different stages of pollen and tissue used in the qPCR analysis (Page 17, line 11-13).
10. The author's use of the name 'NSF' (non S-locus F-box protein) instead of, or in addition to 'Sli' creates confusion by unnecessarily adding alternative terminology to the literature. The gene symbol Sli is well established, has precedence, and reflects the SC phenotype. Furthermore, there are MANY F-box encoding genes in the genome, most of which are NOT at the S-locus. Thus the name 'Non S-locus F-box' could conceivably be understood to refer to many other F-box genes, which might create some confusion. Why not stick with Sli?

Response:
We do agree that 'NSF' creates confusion. NSF was used instead of using Sli mainly based on two reasons. First, a group from Wageningen was using the wild diploid potato species, Solanum chacoense Bitt., to clone Sli gene, we were using RH as a starting material. Thus, NSF was used to differentiate two studies, indicating two group applied different strategies or routes to clone Sli. Another reason is we didn't know whether the SC gene (NSF) from RH is the same to Sli from Solanum chacoense Bitt., only after we confirmed that the coding sequences of NSF and Sli are exactly the same. We changed 'NSF/Sli' to 'Sli' to avoid this confusion after two genes were confirmed to be the same gene.
-p3, par1: loss of S-RNase function has been documented not just through knock-out mutations, but also from amino acid substitutions that eliminate RNase activity.

Response:
OK. We used 'loss of function mutation of S-RNase' to replace 'knock out of S-RNase' (Page 3, line 5).
-p3, par1: replace 'degradation of this gene' by 'degradation of this protein'
-p3, last paragraph: the Sli gene acts gametophytically, thus it is not entirely accurate to refer to it as 'dominant'. Suggest inserting 'or gametophytic factor' after 'dominant gene'. Also, the authors should clarify the direction of the initial cross (I assume PI female x RH male).
-p4, par2: the chi-square statistic should be based on an expected 1:2:1 segregation, although it doesn't hurt to also compute the goodness-of-fit to the 1:1, as shown.

Response:
As mentioned below, our genetic results have proved the gametophytic action of Sli. Thus, we calculated chi-square statistic based on the 1:1 segregation, not 1:2:1 segregation.
-supplementary figure 2b, d, f and g: explain what M1 and M2 represent. Also the figure would be improved by having nontransformed diploid and tetraploid samples as references. The empty vector control has presumably gone through transformation and regeneration also, so could easily have undergone a change in ploidy.
-supplementary figure 3. The figure shows that the sequence of NSF from RH is 100% identical to that found in S. chacoense or M6. Therefore the figure could be simplified by presenting just the sequence of RH, and in the legend stating that the other two sequences were identical.

Response:
OK. As suggested by other reviewer, we provide a sequencing alignment not only the coding region, but also the promoter region among RH, M6, and E172. This figure is updated in the revised manuscript. I am glad to have a chance to review the manuscript, entitled "A non S-locus F-box gene breaks self-incompatibility in diploid potatoes. A diploid hybrid breeding has now become a hot topic for potato breeders. The well-known self-compatibility-inducing gene Sli has opened the way to develop diploid inbred lines, but this gene has not been isolated and its molecular function has remained unknown. This article discovered that a responsible gene is PGSC0003DMG400016861, a F-box protein gene with the F-box domain and PP2 domain and that it works like a general S-RNase inhibitor. This is certainly worth to be published.

Response:
We thank Reviewer for this positive comment.
There are two major concerns on this article.
The authors identified a NSF gene that imparted SC to RH. The genomic region containing the NSF has been supposed to be a candidate region for the Sli gene (Clot et al. 2020). The sequence similarity of this region is very high not only between RH and M6 but also between these and several famous cultivars (Clot et al. 2020). This similarity has been demonstrated in Supplementary Fig. 3 where 100% similarity among SC S. chacoense, M6 and RH is shown in the sequence of the NSF gene (although I can't understand very well about this sequence because start codon and stop codon are not indicated). These indicate that the allelic variation in the NFS gene sequence is not so important because similar sequence can be found in many genotypes irrespective of being SC or SI. More important could be the presence or absence of a 536 bp insertion in the promoter region. SC S. chacoense, M6 and RH have the same size insertion present at the same position? The comparison of the promoter region is highly requested.

Response:
We thank Reviewer for this suggestion. The comparison of the promoter region of NSF/Sli among S. chacoense, M6 and RH was provided. All the SC lines have an insertion at the promoter region of Sli (Page 5, line 19-25). The start and stop codon are also indicated in Supplementary Fig. 3.
The second major concern is on the genetic action of the NSF/Sli gene. The authors described, "Thus, only the pollen harboring the SC gene can penetrate the self style and fulfil fertilization to produce progeny, and all the F2 progeny would carry the SC gene and exhibit self-compatible phenotype". So, they suggested that the NSF/Sli is gametophytically expressed. In contrast, a sporophytic action for the Sli gene has been proposed by Hosaka and Hanneman (1998). Based on sporophytic action hypothesis, pollen grains produced from a Sli-carrying pollen parent are all compatible in self-crossing. Thus, from a cross between SC (Sli/-) and SI (-/-), the progeny plants are segregated to a ratio of 1 SC (Sli/-) : 1 SI (-/-). Self-crossing of the SC (Sli/-) results in a ratio of 1 Sli/Sli : 2 Sli/-: 1 -/-; however, Sli/Sli was presumed to be absent due to a tight linkage with lethality (Hosaka and Hanneman 1998). Based on the sporophytic action hypothesis, the F2 population in Fig. 1d would be segregated to a ratio of 2 SC : 1 SI. If the gametophytic action is employed, only pollen grains carrying NSF/Sli are fertilized and generated only SC F2 plants. In this context, expression of NSF/Sli in pollen grains would be segregated to a ratio of 1:1; a half pollen grains would produce mRNA in in situ mRNA hybridization of NSF in Fig. 2d. in situ mRNA hybridization of NSF is suggested to conduct not with maturing pollen grains in the anther but with mature, fertile pollen grains released from the anther. Please verify which is a likely action of NSF/Sli.

Response:
We thank Reviewer for this suggestion. According to the genetic analysis, we believe the gametophytic action should be correct for Sli. First, the genotype (Sli/Sli) was detected in the F2 progeny or the S1 population of RH. Thus, at least in our materials, Sli is not tightly linked with a lethal gene. Second, an extreme segregation distortion phenomenon (1 Sli/Sli: 1 Sli/-: 0 -/-) was observed in F2 progeny from two F1 populations (PI225689×RH and RH×Y16-90) and one S1 population of RH. In either the F2 or the S1 populations, no SI genotype (-/-) was identified. Third, pollen of RH was used as Sli donor by Dr. Chunzhi Zhang to pollinate 28 SI lines to create SC F1 individuals (Supplementary Table 1). After generates of selfing of these SC plants in creation of diploid inbred lines, she didn't observe SC progeny return back to SI, indicating the genotype (-/-) should no longer exit in these populations. In addition, the group from Wageningen university independently discovered this segregation distortion phenomenon in their populations too. Thus, we concluded that the SD in the F2 population and the S1 population of RH are caused by the gametophytic SI. On the other hand, the sporophytic action hypothesis reported in the Hosaka and Hanneman paper was based on segregation of the SC/SI phenotype in F1 and F2 populations, not based on segregation of the genotype. As SC/SI phenotype would be affected by uncontrollable environmental factors and male sterility, it is possible that segregation rate of the SC/SI phenotype is different from that of the genotype.
For Fig. 2d, our main purpose is to show Sli is highly expressed in the mature pollen tissue. Consistent to our expectations, mature pollen has the highest express of Sli, and thus was used in in situ mRNA hybridization (d4). We agree with Reviewer that a half of pollen grains would produce Sli mRNA in in situ mRNA hybridization of Sli. However, the in situ mRNA hybridization may not be a good system to reflect this fact, as a slice of sample revealed in Fig. 2d only represents a very small fragment of the mature pollen tissue. Actually, we did observe the mature pollens that didn't express Sli in the in situ experiment. Thus, we think genetic analysis is more reliable than the in situ mRNA hybridization experiment to show the gametophytic action of Sli.
In situ mRNA hybridization of NSF/Sli in the mature stamen. Pollens that did not express Sli is indicated by the arrows.

Minor comments
Under the heading "SC gene colocalizes with a segregation distortion region in RH" in Results, a segregation ratio of A/A:A/a=3030:2954≈1:1 was observed for the F2 progeny from crossing PI 225689 and RH. A/A and A/a genotypes were determined by marker analysis? Why did the authors speculate that A/a is a heterozygote SC and a/a is a homozygote SI plant? This is a question relating to the major concern described above. In the following paragraph, the authors described, "In total, 6624 F2 individuals produced from the 131 selfed F1 plants were used to narrow down the SC gene to a 64.61 kb interval (Fig. 1c)". From this sentence I thought that SC and SI were segregated among F2 individuals, so that the candidate region could be narrowed down. So, not all F2 individuals were SC. If my understanding is correct, A/a (SC) and a/a (SI) were identified in the F2 population and a sporophytic action should be proposed.

Response:
Firstly, the genotype segregation ratio of A/A:A/a=3030:2954≈1:1 was determined by the marker analysis. The state that 'A/a is a heterozygote SC and a/a is a homozygote SI plant' is mostly based on BSA-Sequence and genetic analysis of F1 population. BSA-Sequence indicated that the SC phenotype is controlled by only one locus at the end of Ch12 (Fig. 1b), which can be represented by letter 'A' (dominance) or 'a' (recessiveness). SC:SI=1:1 in F1 progeny indicates that one parent is heterozygote while the other is homozygote in this SC controlling locus. Resequencing data of the two parents, PI225689 and RH, indicates that the SC controlling locus is heterozygote in RH and homozygote in PI225689. Thus, RH is SC, and SC is dominant (A) and SI is recessive (a). The state that 'all the F2 progeny would carry the SC gene and exhibit self-compatible phenotype' is based on genotype analysis of F2 progeny that show an extreme segregation distortion. We proposed a gametophytic action of Sli for this extreme segregation distortion by stating that 'only the pollen harboring the SC gene can penetrate the self-style and fulfil fertilization to produce progeny'. As all F2 plants contain SC gene, there is no need to phenotype these F2 individuals and we just did genotype of these plants to further narrow down the target gene.
Under the heading "NSF is efficient in rendering SC to SI conversion in diploid potatoes" in Results, it is described that 125 SI diploid species were pollinated. apparently, 125 is not the number of species but probably the number of accessions or clones, plants, etc. Please be clarified. In addition, S. tuberosum ssp. andigena and ssp. tuberosum are tetraploid and self-fertile. Did the authors use dihaploid genotypes of these species? In the same paragraph, the sentence "The strengthen the functional evidence of NSF/Sli as suggested. Another point to address is the phenotyping of plant for SI/SC. There is a great environmental influence on this phenotype. The authors' phenotyping methods lack detail and create doubt in the results.

Response:
We thank Reviewer for this suggestion. Three clones for each genotype were planted, and >30 self-pollinations were carried out for each clone. The genotype was considered as SC plant when ≥2 fruits were set on all the three clones; considered as SI plant when no fruit was set on all the three clones; considered as UI (Uncertain) plant when ≤1 fruit was set on each clone. This detail was added in the revised maintext (Page 8, line 15 to 18).
I question why the authors had to generated over 3,000 transgenic events. Explanation would be valuable. I am also concerned why so many tetraploid events are recovered in the 26 events studied.

Response:
Currently, gene transformation rate to produce positive diploid lines is low, as 24 positive transgenic lines were changed to tetraploids. This unexpected alteration may occur during regeneration process, since this process tends to induce genome instability 1 . Thus, we have to enlarge the number of explants used in the Sli transformation to make sure we could get positive diploid lines.
The plant material used was poorly detailed. Furthermore, Clot et al. (2020) provided a means to genotype the Sli region in potato lines. I feel knowing that genotyping would add confidence in the correct interpretation to the results observed. I also feel that the authors should more clearly address their work in the introduction. In the results the authors refer to markers M-1 and M-2. These markers are not referenced. Are they based upon Clot et al.?

Response:
We thank Reviewer for this suggestion. We provide more details about the plant materials kindly shared by Prof. K. Hosaka ( Two InDel markers (named as M-1 and M-2) that were heterozygous in RH and homozygous in PI 225689 were designed to make sure SC locus was located between the two markers.
Introduction P2L16 Should say "Potato is consumed as a staple food." P2L29 This phrase is insufficient to address the complex way in which inbreeding depression influences self-fertility. P3L1: This additional S-RNase function has been only described in one species of the Rosaceae family, so there is not any proof that it works in Solanum. P3L15 Eggers did provide detailed methodologies. It is more appropriate to say that the molecular mechanism of Sli was not described. P3L18, L32: SC is wrongly used, since SC stands for "self-compatibility," as stated in the previous paragraphs.

Discussion
P7L15-33 Rather than belabor the model, it would be more interesting to focus on S-RNase allelic diversity and the prospect of Sli to interact with various S-RNases from a practical breeding standpoint.

Methods
The BSA method should be before, accordingly with the results section.
P8L3 It is still unclear what E172 is. More detail is needed for clone 'E'. P8L10: Specify that the PI 225689 is SI. P8L11 How many F2 individuals? P8L13 What germplasm was phenotyped with this method? P8L18: "UI" can be confused with unilateral incompatibility. P8L18 How is it that all three clones having fewer than 1 berry different than all three clones failing to set fruit? I think the English needs to be clarified. P8L20 How many pistils per genotype? P8L32 which version of the DM genome? P9L4: Again, the authors do not explain the nature and origin of markers M1 and M2. P9L5 multiple F1 populations? P9L8-9 I am not familiar with the term 'extreme partial separation' Also need to describe what the B/B haplotype represents P9L34 What is S15-65? P10L20 Which diploid lines were used for RNAseq? At what developmental stage was style tissue collected for RNAseq? Figure 1d: it seems that the SC allele is derived from the line PI 225689. Figure 2. c) Is the sample in stages 1-4 all floral organs and the pollen sample is mature pollen? Figure 4 This model is confusing and does not add to the paper Supplementary Figure 3. This alignment style is very confusing. Why are all three sequences not shown with conservation *?