A serine/threonine protein kinase encoding gene KERNEL NUMBER PER ROW6 regulates maize grain yield

Increasing grain yield of maize (Zea mays L.) is required to meet the rapidly expanding demands for maize-derived food, feed, and fuel. Breeders have enhanced grain productivity of maize hybrids by pyramiding desirable characteristics for larger ears. However, loci selected for improving grain productivity remain largely unclear. Here, we show that a serine/threonine protein kinase encoding gene KERNEL NUMBER PER ROW6 (KNR6) determines pistillate floret number and ear length. Overexpression of KNR6 or introgression of alleles lacking the insertions of two transposable elements in the regulatory region of KNR6 can significantly enhance grain yield. Further in vitro evidences indicate that KNR6 can interact with an Arf GTPase-activating protein (AGAP) and its phosphorylation by KNR6 may affect ear length and kernel number. This finding provides knowledge basis to enhance maize hybrids grain yield.


4.
, the graphic symbol is not consistent with the columns. 5. Some important literatures were not cited in the manuscript, about meristem development, "AGO18b negatively regulates determinacy of spikelet meristems on the tassel central spike in maize", doi: 10.1111/jipb.12596; about the breeding, "Maize biology: From functional genomics to breeding application", doi: 10.1111/jipb.12819. 6. There are some spelling mistakes in the manuscript.

Comments for authors
Review for "KERNEL NUMBER PER ROW6 regulates grain yield by phosphorylating an Arf GTPaseactivating protein in maize" In this study, the authors cloned a gene controlling number of kernel (KNR6) in a single row in maize. They further demonstrated that the KNR6 has protein kinase activity and is capable of phosphorylating downstream genes. Moreover, the dominant allele led to the increase of yield in hybrid corn. In addition, they found an insertion of a Harbinger family transposon in the recessive allele and this transposon is not absent from the dominant allele and they consider this transposon is responsible for the difference of variation between the two alleles. Overall I think this is an interesting paper and it is practically useful. However, there are indeed a few issues that need to be clarified before publication.

Major comment
Whether "The TE-PAV Is the Causal Variant for KNR6"?
One of the important conclusions is that "The TE-PAV Is the Causal Variant for KNR6" (p.9, row 3). Of course, the TE-PAV is one of the major genetic changes occurred at this loci, so it possible TE-PAV is the causal variant. However, since there are other mutations in and around this gene, there are alternative possibilities. For example, it is possible other changes, such as the L7F switch, also contribute to the differences. In the extreme case, it is not impossible that the TE-PAV is only a marker.
In the abstract the author stated that the TE-PAV "may function as a transcriptional repressor and/or as a promoter of DNA methylation in the KNR6 upstream region to suppress expression of the gene." This statement is much more reasonable than "The TE-PAV Is the Causal Variant for KNR6" but it is not guaranteed. I agree, it is possible that it serves as a transcription repressor, based on the result from the luciferase experiment. Nevertheless, I would argue that an insertion upstream of a promoter (in the luciferase experiment) is not really the same as an insertion is present in the intron (in the KNR6 gene). Insertions in introns are not uncommon in plants and they don't always generate detectable effect.
I do suspect the methylation of upstream region of KNR6 is responsible (if not fully) for the reduced expression of this gene in the recessive allele. To test this notion, the authors should do similar association analysis as that in Fig. 3A using SNPs in upstream regions. It is unclear to me why the authors are only showing the analysis in transcribed region.
Having said that, I have problems with TE-PAV as "as a promoter of DNA methylation in the KNR6 upstream region". p.11 row 22 -23, "In changing trend, the CG and CHG methylation of sequence near the Harbingerlike TE was higher than that far from the Harbinger-like TE (Fig. 3I-J)." Based on Figure 3i, the sequences immediately flanking the TE-PAV (maybe 2kb on each side?) demonstrate very little methylation and the sequences far away (more than 2 kb) are heavily methylated. This is consistent with previous studies that most DNA transposons or genic transposons do not spread methylation.
So the above statement is largely false. Please correct it. Moreover, I don't know how you could get Figure 3J from Figure 3i. The two portions on the methylation level are not consistent at all. I guess the slide window (1200 bp) is too large to mix everything up in Figure 3j. My suggestion is to use 200 bp slide window and 100 bp step. The curve would not be as smooth but that should be fine.
In addition, given that 85% of the maize genome are transposable elements, it is a little bit surprising that in the 30 kb region in Figure 3 there is only one small Helitron (in addition to the Harbinger element). Please carefully annotate the transposons in this region, particularly the upstream region to see whether there is something readily to be methylated.
So I would like to reiterate my suggestions for this part: 1) conduct an association analysis (I assume you have the data) with SNP in the upstream region of the gene to see whether they are associated with the variation; 2) carefully annotate the transposons in this region; 3) reduce the window size in Figure 3j; 4) turn the tune down about the role of TE-PAV and discuss alternative possibilities.
Minor comments: 1. In the abstract and most places in the text, it was stated the TE-PAV is located in the 5' UTR. However, According to Fig. 1, the insertion is actually in intron. In my opinion, the impact in UTR is very different from that in intron. So please be accurate. 3. Fig. 3A. The 1 kb bar does not look right to me. According to the text, the cDNA is only 1.8 kb in length. According to this bar, it would be around 20 kb. So please make them consistent. 4. Fig. 5D, the legend inside the figure is not consistent with the columns. It should be yellow and green but it is grey and dark right now. 5. Fig. 5G, there should be a letter on top of the first green column. 6. p.18 row 3 to 4. "Sequences from 5ʹ-and 3ʹ-RACE products were assembled to obtain full-length cDNA sequence of KNR6." I looked at supplemental Fig. 4. It appears that the 5ʹ-RACE product is about 500 bp and 3ʹ-RACE product is about 250 bp. So it is unclear to me how you could get a 1.8 kb full-length cDNA through assembling 500 bp and 250 bp. Please clarify.
In addition, it would help to put the gene structure under this figure and diagram where the genespecific primers are located in the 5 RACE and 3 RACE experiments.
Reviewer #3 (Remarks to the Author): The manuscript by Jia et al. reports the cloning and molecular characterization of KRN6, a previously identified kernel row number QTL in maize. Analysis of the QTL in this manuscript is original, thorough and of high quality, leading the authors to identify a transposon insertion in a kinase gene as the causal variant for the KRN6 QTL locus, and two likely in vivo kinase substrates, one of which is an AGAP that when mutated also has a kernel number phenotype in the anticipated direction. Interestingly, the high KRN allele affects meristem size in the inflorescence. Overall the findings are substantiated by an extensive mix of well executed quantitative genetics, molecular genetics and field-based yield trial experiments that document effects of the QTL in inbred breeding and hybrid production genotypes. From these results applications of KNR6 and AGAP are easily envisioned that may increase crop grain yields. The main conclusions are robust and reliable, leading the authors to hypothesize a direct connection of the KRN6 kinase to modulation or auxin levels. This research represents a substantial advancement in our understanding of the developmental and molecular bases of a key agronomic trait, and further investigation such as of the details of effects on auxin pathways, are not necessary. All sections of the manuscript well written and referenced. The work should be of broad interest to researchers in basic and applied plant biology and in related agricultural sciences, such as quantitative genetics and plant breeding. Reviewer #1 (Remarks to the Author): Comments: In the manuscript "KERNEL NUMBER PER ROW6 regulates grain yield by phosphorylating an Arf GTPase-activating protein in maize", the authors established that KERNEL NUMBER PER ROW6 (KNR6) encodes a serine/threonine protein kinase that controls maize grain yield by determining pistillate floret number, ear length and kernel number per row, through fine-mapping, association mapping, expression analysis and transgenic validation. They found a Harbinger-like transposable element (TE) in the 5'-untranslated region of KNR6 is the causal variant to suppress expression of the gene and this TE has a strong impact on grain yield. Further, KNR6 regulates kernel number and grain yield by influencing AGAP phosphorylation to modulate ear inflorescence development. In this manuscript, the authors found a QTL KERNEL NUMBER PER ROW6 (KNR6) control kernel number per row (KNR) and ear length (EL), and elucidated its functional mechanism and the potential in upregulating grain yield. This finding will be of great significance to agricultural breeders. The whole manuscript is well organized and the results are presented with solid evidence and influent English. However, there are still some points need to be improved.
Major points: 1. The authors showed AGAP proteins are phosphorylated by KNR6 by phosphorylation assay in vitro, how about AGAP phosphorylation level in NIL qKNR and NIL qknr plants in vivo?
Response: Thanks for the suggestion. We agree that it is critical to understand the mechanism of how KNR6 acts in inflorescence meristems, and is a long-term goal of our research. Identification and isolation of this gene is just the beginning of this project. As the reviewer suggested, identifying the phosphorylation substrates of KNR6 in meristems is one of the research tasks. For this, we are breeding the key lines, including NILs and transgenic lines, at our winter maize nursery, in order to collect sufficient meristem samples for phosphoproteomics. We agree our current results are preliminary and do not fully answer the underlying mechanism. Our in vitro evidence finds AGAP as a target protein of KNR6 phosphorylation, but we agree in vivo evidence is lacking. However we think this is beyond the scope of our current paper, which focuses on the genetics of KNR6. Further evidence on how KNR6 acts in meristems, including from phosphoproteomics, will be presented in an independent paper.
2 / 9 2. It is not suitable to draw out the role of KNR6 on 14-3-3 proteins just based on previous results from Arabidopsis and no evidence in maize. In addition, in the model of Fig. 4J, the hypothetical functions of AGAP and 14-3-3 in "auxin influx" and "physiological adaptation" is exaggerated and need to be modified.
Response: Thanks for the comment, we agree there is no support for this model in maize. We therefore re-drew the putative pathway of KNR6 and removed the hypothetical functions of AGAP and 14-3-3 in "auxin influx" and "physiological adaptation" from Fig. 4J. 3. The phase "meristematic activity" appeared many times in the manuscript. How to define "meristematic activity"? In Fig.1A, the yellows bars indicated the size or the shape of the inflorescence meristem. But no activity was tested using meristem markers or other methods. I think "meristematic activity" is not accurate here.
Response: Thanks for this comment. We have not used markers to support this claim, and so we replaced the phase "meristematic activity" using appropriate words, such as " stronger ability to produce florets" and " floral production on the ear inflorescence". Response: Thanks. We agree, and deleted the phrase "two environments". Supplementary Fig. 2  Furthermore, because both TE-PAV and LTR-PAV were two major variants in two parental lines and in the association panel, and were located on KNR6 regulatory regions that are extremely important for gene expression, thus were referred to as two key genetic markers of the causal variants for qKNR6. Using the two markers, two haplotypes were identified in the association panel: Haplotype 1 (Hap1) including 48 inbred lines that carry the Harbinger-like TE and the LTR, and Haplotype 2 (Hap2) including 176 inbred lines that lack the Harbinger-like TE and the LTR in qKNR6.The average expression level of KNR6 in the Hap2 lines was significantly higher than 5 / 9 that in the Hap1 lines (p = 0.023) (Fig. 3d), and the Hap2 lines had longer ears with more KNR than the Hap1 lines (Fig. 3e, f).

In
In the abstract the author stated that the TE-PAV "may function as a transcriptional repressor and/or as a promoter of DNA methylation in the KNR6 upstream region to suppress expression of the gene." This statement is much more reasonable than "The TE-PAV Is the Causal Variant for KNR6" but it is not guaranteed. I agree, it is possible that it serves as a transcription repressor, In addition, given that 85% of the maize genome are transposable elements, it is a little bit surprising that in the 30 kb region in Figure 3 there is only one small Helitron (in addition to the 6 / 9 Harbinger element). Please carefully annotate the transposons in this region, particularly the upstream region to see whether there is something readily to be methylated.
So I would like to reiterate my suggestions for this part: 1) conduct an association analysis (I assume you have the data) with SNP in the upstream region of the gene to see whether they are associated with the variation; 2) carefully annotate the transposons in this region; 3) reduce the window size in Figure 3j Second, we identified 433 variants within a 100-kb region centered on KNR6, and performed association analysis and conditional association analysis. We found, in addition to the Harbinger-like TE PAV, the LTR-PAV was also associated with Kernel number per row (KNR).
We also analyzed the genetic effects of Haplotypes on KNR and EL (ear length). These updated results are described in the text. Third, we re-analyzed methylation levels in three contexts within an approximately 30-kb genomic region centered on KNR6 in both parents using a window size of 200 bp and a step size of 100 bp, these updated results are shown in Fig.3j and k, and Page11 Line13-25. We found, in NIL qknr6 , both the LTR-PAV and the TE-PAV were hypermethylated in CG (94.5% and 97.1%), and CHG (85.2% and 63.4%) contexts, but not in CHH (1.7% and 7.0%) (Fig. 3j). Moreover, the average methylation level in the 5 -15-kb region upstream from the Harbinger-like TE was dramatically higher in NIL qknr6 than in NIL qKNR6 (Fig. 3i, k), while the regions immediately flanking the TE-PAV had low methylation (Fig. 3i,k). This result agrees with the observation that genic transposons do not always spread methylation (Makarevitch I, et al. Transposable elements contribute to activation of maize genes in response to abiotic stress. PLoS Genetics, 2015, 11:e1004915). The heavily methylated regions 5 -15-kb upstream region from the Harbinger-like TE correspond to the LTR-PAV, suggesting methylation difference at the distal upstream region of KNR6 between the two parent lines may not be caused by the Harbinger-like 7 / 9 TE-PAV but by the LTR-PAV, which is located 5.1-kb upstream of KNR6. In addition, a high methylation level was also found within the 5 kb region downstream of KNR6 in both parents, due to the presence of a Helitron TE and an LTR retrotransposon (Fig. 3i-j).
Finally, we rewrote the paragraph on DNA methylation. The manuscript by Jia et al. reports the cloning and molecular characterization of KRN6, a previously identified kernel row number QTL in maize. Analysis of the QTL in this manuscript is original, thorough and of high quality, leading the authors to identify a transposon insertion in a kinase gene as the causal variant for the KRN6 QTL locus, and two likely in vivo kinase substrates, one of which is an AGAP that when mutated also has a kernel number phenotype in the anticipated direction. Interestingly, the high KRN allele affects meristem size in the inflorescence.
Overall the findings are substantiated by an extensive mix of well executed quantitative genetics, molecular genetics and field-based yield trial experiments that document effects of the QTL in inbred breeding and hybrid production genotypes. From these results applications of KNR6 and AGAP are easily envisioned that may increase crop grain yields. The main conclusions are robust and reliable, leading the authors to hypothesize a direct connection of the KRN6 kinase to modulation or auxin levels. This research represents a substantial advancement in our