LAUSR

Increasing maize yield remains the primary goal in maize breeding in the world. Kernel and ear morphology are two important yielddetermining traits in maize (Simmons et al., 2020). However, maize kernel shape and ear size are complex quantitative traits, for which a few functional genes with potential breeding applications have been reported. Overexpression of vacuolar protein sorting 29 (ZmVPS29) can improve the yield per plant in different maize genetic backgrounds (Chen et al., 2020). Kernel number per row 6 (KNR6) can also increase the grain yield of hybrid maize either through the introduction of useful allelic variants or by raising KNR6 transcript levels (Jia et al., 2020). Moreover, genome editing of 1-aminocyclopropane-1-carboxylate oxidase 2 (ZmACO2) promoted maize ear development, resulting in an ~13.4% increase in grain yield per ear in hybrid lines (Ning et al., 2021). In a previous study, we preliminarily evaluated the potential breeding value of the quantitative trait locus kernel length 9 (qKL9), which controls kernel length and kernel weight in various populations (Gong et al., 2021). We wished to clone the causal gene responsible for the quantitative variation in maize grain yield associated with qKL9 and explore the causal polymorphisms. Our ultimate goal is to develop new functional markers and evaluate the breeding potential of the causal gene behind qKL9 to create new avenues for improving maize yield. To further fine-map qKL9, we developed a backcross population of about 20 000 individuals to screen for recombinants. Despite the size of the population, the low recombination rate over the candidate region harbouring qKL9 resulted in an ~736kb region flanked by markers IN6706 and N23, corresponding to 19 annotated genes, based on the B73 RefGen v4 genome (Figure 1a, Tables S1 and S2). An analysis of published gene expression data (Yi et al., 2019) and transcriptome deep sequencing (RNA-seq) data of seed at 6, 9 and 14 DAP (days after pollination) of the maize inbred Mc and the BC4F2 line Mc revealed Zm00001d046718 as the only differentially expressed gene in 6-DAP seeds between Mc and Mc. Another 13 genes were expressed at similar levels or at low levels in seeds (Figure 1b and Figure S1). Based on this result, we selected Zm00001d046718, which encodes a basic leucine zipper (bZIP) transcription factor, as the key candidate gene for qKL9, prompting us to name it ZmKL9. To identify the polymorphic sites in ZmKL9, we determined the genomic sequence of ZmKL9 in Mc and Mc, which revealed single nucleotide polymorphisms (SNPs) and insertion/ deletions (InDels) in the promoter and coding region, as well as the insertion of a 288and 557-bp transposable element (TE) in the 50 untranslated region (50 UTR) and the first intron, respectively, named TE1 and TE2, in Mc relative to Mc (Figure 1c, Table S3). Moreover, we genotyped the recombinants (Figure 1a) for the presence of TE1 and TE2 and established that their presence cosegregates with the kernel length phenotype (Figure 1d). We also sequenced ZmKL9 in a set of 149 diverse maize inbred lines and determined that TE1 is significantly associated with kernel length by association mapping (Figure 1e). Indeed, inbred lines harbouring TE1 (TE+) were characterized by shorter kernels and lower hundred kernel weight (HKW), on average than the lines lacking TE1 (TE–), although kernel width was not affected (Figure 1f). We investigated the possible effect of the presence of TE1 on ZmKL9 expression by performing a transient luciferase (LUC) transcriptional assay in maize protoplasts. Accordingly, we placed the firefly LUC reporter gene under the control of the ZmKL9 promoter from Mc (with TE1) or Mc (without TE1) and detected lower LUC activity derived from the construct containing TE1 (TE+ construct) relative to the construct lacking the TE (TE construct; Figure 1g). We concluded that the TE1 in the ZmKL9 50 UTR is a key polymorphism that influences the transcriptional output of ZmKL9. To elucidate the biological function of ZmKL9, we generated two overexpression lines (OE-ZmKL9-1 and OE-ZmKL9-2) and two ZmKL9 knockout lines (CR-zmkl9-1 and CR-zmkl9-2) using clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated nuclease 9 (Cas9) genome editing (Figures S2 and S3). Compared with their respective nontransgenic control lines, the two ZmKL9 overexpression lines (OEZmKL9-1 and OE-ZmKL9-2) had longer kernels, higher HKW, and longer and heavier ears (Figure 1h,i). Conversely, the two knockout lines CR-zmkl9-1 and CR-zmkl9-2 produced shorter kernels and ears, and had a lower HKW and ear weight than the corresponding wild-type lines (Figure 1j,k). These results strongly supported the notion that ZmKL9 is the causal gene underlying qKL9 and positively controls kernel and ear yield in maize. To evaluate the ability of ZmKL9 to control maize yield in different genetic backgrounds, we crossed one ZmKL9-overexpressing line and its nontransgenic control line (OE-ZmKL9-1 and WT1) to four different elite inbred lines (Chang7-2, Zheng58, PH6WC, and PH4CV). We confirmed that ZmKL9 expression is higher in F1 hybrids between each inbred and the OE line relative to the nontransgenic F1 hybrids. We also established that kernel