Promoter Variation of the Key Apple Fruit Texture Related Gene MdPG1 and the Upstream Regulation Analysis

MdPG1 encoding polygalacturonase in apple (Malus × domestica) is a key gene associated with fruit firmness and texture variations among apple cultivars. However, the causative variants of MdPG1 are still not known. In this study, we identified a SNPA/C variant within an ERF-binding element located in the promoter region of MdPG1. The promoter containing the ERF-binding element with SNPA, rather than the SNPC, could be strongly bound and activated by MdCBF2, a member of the AP2/ERF transcription factor family, as determined by yeast-one-hybrid and dual-luciferase reporter assays. We also demonstrated that the presence of a novel long non-coding RNA, lncRNAPG1, in the promoter of MdPG1 was a causative variant. lncRNAPG1 was specifically expressed in fruit tissues postharvest. lncRNAPG1 could reduce promoter activity when it was fused to the promoter of MdPG1 and a tobacco gene encoding Mg-chelatase H subunit (NtCHLH) in transgenic tobacco cells but could not reduce promoter activity when it was supplied in a separate gene construct, indicating a cis-regulatory effect. Our results provide new insights into genetic regulation of MdPG1 allele expression and are also useful for the development of elite apple cultivars.


Introduction
Fruit softening is a complex process of cell wall disassembly that is induced by ethylene in climacteric fruits. The biosynthesis of ethylene, a gaseous plant hormone, starts just prior to the initiation of ripening in the climacteric fruit, such as apple, and produces an ethylene burst as ripening progresses [1][2][3]. After the ethylene burst, the fruit becomes soft and loses its storage ability and then its commercial value.
Ethylene biosynthesis is sequentially catalyzed by two enzymes, 1-aminocyclopropane-1-carboxylate synthase (ACS) and 1-aminocyclopropane-1-carboxylate oxidase (ACO). In apple, MdACS1 and MdACO1 are highly expressed during fruit ripening and responsible for the production of ethylene. Genetic variants that affect gene expression level have been identified in the two genes. For example, a SINE type transposable element in the MdACS1 promoter region is known to inhibit MdACS1 expression [4]. Similarly, a deletion of a 62 bp fragment in the third intron of MdACO1 reduces MdACO1 expression [5]. Apple cultivars that are homozygous for these MdACS1 and MdACO1 variants produce little ethylene during fruit ripening and have a prolonged storage and shelf life [5][6][7]. Ethylene signaling relies on a linear signal transduction pathway. The ethylene response factors (ERFs) that act downstream are the last component of the ethylene signal transduction pathway to regulate the ethylene-responsive gene's expression. They function as transcriptional modulators by binding to the promoter elements such as GCC box or DRE/CRT motif in many plant species [8][9][10][11]. In apple, EIL and COLD BINDING FACTOR (CBF) type ERFs could transactivate the promoter of MdPG1 (polygalacturonase 1) in response to cold-and ethylene-induced fruit ripening and promote fruit softening [12]. MdERF2

Genetic Variants in the Promoter of MdPG1
To identify the causative genetic variants that may be responsible for MdPG1 expression variations in different apple cultivars, the DNA sequences covering the coding and promoter regions of MdPG1 alleles in four apple cultivars ("Meiba", "Huahong", "Huaxing" and "Huashuo") (Table S1) were PCR-amplified. A 4.8 kb and a 3.5 kb DNA promoter fragments in different apple cultivars were obtained. The four apple cultivars could be classified into three insertion genotypes, homozygous showing the 4.8 kb DNA fragment, heterozygous showing both 4.8 kb and 3.5 kb fragments, and without insertion showing the 3.5 kb fragment ( Figure 1A). The size of DNA fragments covering the coding regions in the four cultivars was the same ( Figure 1B). The PCR fragments of two MdPG1 alleles in all the culti- vars were cloned and fully sequenced ( Figure S1). The alignment of the two allele sequences in the "Meiba" and "Huaxing" promoter showed a 1.3 kb insertion in the larger fragment (named as allele 1) compared to the small fragment (named as allele 2), and the insertion was flanked by an 8-bp direct repeat ( Figure S1A,E). "Huashuo" was shown to be homozygous for the insertion ( Figure S1G). After confirming the presence of an insertion at 2.5 kb upstream from the start codon ATG of MdPG1 in some apple cultivars, variations were further identified in the 2.5 kb region upstream from the start codon of MdPG1. These variations included 3 indels and 34 SNPs in the promoter sequences of MdPG1 ( Figure S1A,C,E). In the same upstream region, three AP2/ERF and six EIN3/EIL cis-elements ( Figure S1C) were identified using the Plantpan2 database (http://plantpan2.itps.ncku.edu.tw, accessed on 1 June 2020) [56]. One of the SNPs identified above was within an ERF-binding element (AGAGTCGGCA/(C)A) at 411 bp upstream of the ATG start codon ( Figure S1C). Furthermore, two alleles of MdPG1 for differential TF binding were scanned using the CIS-BP database (http://cisbp.ccbr.utoronto.ca/TFTools.php, accessed on 1 March 2023) [2], and the results showed that AP2 was more likely to bind to the sequence (AGTCGGCAA, −418-−410) of allele 1, but not allele 2 (Table S2). This SNP may be a key variation related to MdPG1 expression. The haploid type containing the insertion and the SNP within the ERF-binding element for all four apple cultivars are shown in Figure 1C.
Plants 2023, 12, x FOR PEER REVIEW 3 of 18 the 3.5 kb fragment ( Figure 1A). The size of DNA fragments covering the coding regions in the four cultivars was the same ( Figure 1B). The PCR fragments of two MdPG1 alleles in all the cultivars were cloned and fully sequenced ( Figure S1). The alignment of the two allele sequences in the "Meiba" and "Huaxing" promoter showed a 1.3 kb insertion in the larger fragment (named as allele 1) compared to the small fragment (named as allele 2), and the insertion was flanked by an 8-bp direct repeat ( Figure S1A,E). "Huashuo" was shown to be homozygous for the insertion ( Figure S1G). After confirming the presence of an insertion at 2.5 kb upstream from the start codon ATG of MdPG1 in some apple cultivars, variations were further identified in the 2.5 kb region upstream from the start codon of MdPG1. These variations included 3 indels and 34 SNPs in the promoter sequences of MdPG1 ( Figure S1A,C,E). In the same upstream region, three AP2/ERF and six EIN3/EIL cis-elements ( Figure S1C) were identified using the Plantpan2 database (http://plant-pan2.itps.ncku.edu.tw, accessed on 1 June 2020) [56]. One of the SNPs identified above was within an ERF-binding element (AGAGTCGGCA/(C)A) at 411 bp upstream of the ATG start codon ( Figure S1C). Furthermore, two alleles of MdPG1 for differential TF binding were scanned using the CIS-BP database (http://cisbp.ccbr.utoronto.ca/TFTools.php, accessed on 1 March 2023) [2], and the results showed that AP2 was more likely to bind to the sequence (AGTCGGCAA, −418-−410) of allele 1, but not allele 2 (Table S2). This SNP may be a key variation related to MdPG1 expression. The haploid type containing the insertion and the SNP within the ERF-binding element for all four apple cultivars are shown in Figure 1C. in "Meiba", "Huahong", "Huaxing", and "Huashuo". PCR analysis of the four cultivars detected a larger DNA fragment in three cultivars indicating DNA insertion in the MdPG1 promoter. The arrow indicates a 1.3 kb insertion in the larger fragment of the MdPG1 promoter region in "Meiba", "Huaxing", and "Huashuo". (C) The SNPs and lncRNAPG1 were phased into allele 1 and allele 2 after genomic PCR fragments covering the promoter and coding regions of the two alleles of MdPG1 were cloned and fully sequenced. The triangle represents a 1.3 kb insertion (lncRNAPG1) at 2356 bp upstream of the ATG start codon. The diamond represents an ERF-binding element, and the causative SNP was within the ERF-binding element at 411 bp upstream of the ATG start codon.

Differentially Expression of MdPG1 Alleles Is Regulated by MdCBF2
To determine if this SNP within the ERF-binding element affected MdPG1 expression, "Huahong" was selected as material for analyzing because it contained this key SNP but no structure variations that may also affect gene expression. MdEIL2 and MdCBF2 were shown to transactivate the MdPG1 promoter in the presence of ethylene or under cold conditions in a previous study [12]. We tested MdEIL2, MdEIL3, MdCBF2 and MdAP2D32 to determine whether they differently activate the MdPG1 promoters of "Huahong" with and coding regions (B) in "Meiba", "Huahong", "Huaxing", and "Huashuo". PCR analysis of the four cultivars detected a larger DNA fragment in three cultivars indicating DNA insertion in the MdPG1 promoter. The arrow indicates a 1.3 kb insertion in the larger fragment of the MdPG1 promoter region in "Meiba", "Huaxing", and "Huashuo". (C) The SNPs and lncRNA PG1 were phased into allele 1 and allele 2 after genomic PCR fragments covering the promoter and coding regions of the two alleles of MdPG1 were cloned and fully sequenced. The triangle represents a 1.3 kb insertion (lncRNA PG1 ) at 2356 bp upstream of the ATG start codon. The diamond represents an ERF-binding element, and the causative SNP was within the ERF-binding element at 411 bp upstream of the ATG start codon.

Differentially Expression of MdPG1 Alleles Is Regulated by MdCBF2
To determine if this SNP within the ERF-binding element affected MdPG1 expression, "Huahong" was selected as material for analyzing because it contained this key SNP but no structure variations that may also affect gene expression. MdEIL2 and MdCBF2 were shown to transactivate the MdPG1 promoter in the presence of ethylene or under cold conditions in a previous study [12]. We tested MdEIL2, MdEIL3, MdCBF2 and MdAP2D32 to determine whether they differently activate the MdPG1 promoters of "Huahong" with different nucleotides at the SNP site within the ERF-binding element. MdEIL2 and MdEIL3 are closely related to AtEIL2 ( Figure S2A), and MdCBF2 and MdAP2D32 are closely related to AtCBFs ( Figure S2B). In dual-luciferase transient expression assays, the luciferase gene was driven by the 2.5 kb promoter of the SNP A or SNP C allele of MdPG1 (Figure 2A-C). MdEIL3 did not significantly increase the luciferase activity compared to the pSAK778 empty vector control. MdEIL2 and MdAP2D32 slightly enhanced the luciferase activity driven by the 2.5 kb promoter of both MdPG1 alleles. Interestingly, MdCBF2 significantly enhanced the luciferase activity driven by the promoter of the SNP A allele ( Figure 2B). Furthermore, the above experiment was repeated using a fusion promoter of minimal CaMV35S and a short (0.3 kb) MdPG1 promoter containing the SNP site. MdCBF2 also significantly enhanced the luciferase activity driven by the 0.3 kb promoter sequences of the SNP A allele compared to the SNP C allele (p < 0.01) ( Figure 2C). These results suggested that the SNP A within an ERF-binding element has a significant effect on the activation of MdPG1 expression via interaction with MdCBF2 transcription factor. luciferase gene was driven by the 2.5 kb promoter of the SNP A or SNP C allele of MdPG1 (Figure 2A-C). MdEIL3 did not significantly increase the luciferase activity compared to the pSAK778 empty vector control. MdEIL2 and MdAP2D32 slightly enhanced the luciferase activity driven by the 2.5 kb promoter of both MdPG1 alleles. Interestingly, MdCBF2 significantly enhanced the luciferase activity driven by the promoter of the SNP A allele ( Figure 2B). Furthermore, the above experiment was repeated using a fusion promoter of minimal CaMV35S and a short (0.3 kb) MdPG1 promoter containing the SNP site. MdCBF2 also significantly enhanced the luciferase activity driven by the 0.3 kb promoter sequences of the SNP A allele compared to the SNP C allele (p < 0.01) ( Figure 2C). These results suggested that the SNP A within an ERF-binding element has a significant effect on the activation of MdPG1 expression via interaction with MdCBF2 transcription factor.
To further determine whether SNP A/C could alter MdCBF2 binding ability to the ERFbinding element, we performed yeast-one-hybrid experiments. The results showed that the yeast cells transformed with the SNP A allele (2.5 kb or 0.3 kb) grew well under aureobasidin A (200 ng/mL) selection, whereas the yeast cells transformed with the SNP C allele grew poorly under the same selection condition ( Figure 2D,E), indicating that MdCBF2 binding was much stronger to the promoter of the SNP A allele than to the promoter of the SNP C allele of MdPG1. Taken together, these results suggest that MdCBF2 directly binds to the promoter of the SNP A allele of MdPG1 to enhance MdPG1 expression. The SNP site in an ERF−binding element is differentially regulated by MdCBF2. (A) Cisacting regulatory elements for AP2/ERF and EIN3/EIL transcription factors were identified in the MdPG1 promoter of "Huahong" using the Plantpan2.0 (http://plantpan2.itps.ncku.edu.tw, accessed on 1 June 2020). An SNP A/C site in an ERF-binding element was located 411 bp upstream of the start codon. (B,C) Dual-luciferase assays were carried by co-infiltration of Nicotiana benthamiana leaves with effector gene constructs (CaMV35S promoter driving no CDS or is driving the CDS of MdEIL2, MdEIL3, MdCBF2 or MdAP2D32) and reporter gene constructs (2.5 kb with SNP A or SNP C , and 0.3 kb plus minimal CaMV 35S with SNP A or SNP C fused to the LUC coding sequence). The error bars are standard deviations of the mean from six repeats. Asterisks indicate significant differences by ttest: ** p < 0.01. (D,E) Yeast-one-hybrid assay showed that MdCBF2 specifically bound to 2.5 kb (D) and 0.3 kb (E) promoter sequence of MdPG1 allele containing the SNP A within the ERF-binding element. Yeast cells co-transformed with the constructs named on the left were cultured on nonselective medium SD/-Leu/-AbA (left panel) and selective medium SD/-Leu/+AbA 200 ng/mL (right Figure 2. The SNP site in an ERF−binding element is differentially regulated by MdCBF2. (A) Cis-acting regulatory elements for AP2/ERF and EIN3/EIL transcription factors were identified in the MdPG1 promoter of "Huahong" using the Plantpan2.0 (http://plantpan2.itps.ncku.edu.tw, accessed on 1 June 2020). An SNP A/C site in an ERF-binding element was located 411 bp upstream of the start codon. (B,C) Dual-luciferase assays were carried by co-infiltration of Nicotiana benthamiana leaves with effector gene constructs (CaMV35S promoter driving no CDS or is driving the CDS of MdEIL2, MdEIL3, MdCBF2 or MdAP2D32) and reporter gene constructs (2.5 kb with SNP A or SNP C , and 0.3 kb plus minimal CaMV 35S with SNP A or SNP C fused to the LUC coding sequence). The error bars are standard deviations of the mean from six repeats. Asterisks indicate significant differences by t-test: ** p < 0.01. (D,E) Yeast-one-hybrid assay showed that MdCBF2 specifically bound to 2.5 kb (D) and 0.3 kb (E) promoter sequence of MdPG1 allele containing the SNP A within the ERF-binding element. Yeast cells co-transformed with the constructs named on the left were cultured on nonselective medium SD/-Leu/-AbA (left panel) and selective medium SD/-Leu/+AbA 200 ng/mL (right panel), in a dilution series of 5 0 , 5 −1 ,5 −2 , 5 −3 , 5 −4 , and 5 −5 (i-vi). Rec-p53 and the p53-promoter were used as positive controls. The empty vector and the promoter fragments of MdPG1 were used as negative controls.
To further determine whether SNP A/C could alter MdCBF2 binding ability to the ERFbinding element, we performed yeast-one-hybrid experiments. The results showed that the yeast cells transformed with the SNP A allele (2.5 kb or 0.3 kb) grew well under aureobasidin A (200 ng/mL) selection, whereas the yeast cells transformed with the SNP C allele grew poorly under the same selection condition ( Figure 2D,E), indicating that MdCBF2 binding was much stronger to the promoter of the SNP A allele than to the promoter of the SNP C allele of MdPG1. Taken together, these results suggest that MdCBF2 directly binds to the promoter of the SNP A allele of MdPG1 to enhance MdPG1 expression.

Identification and Characterization of an lncRNA in the Promoter of MdPG1
The 1.3 kb insert identified above showed a high level of sequence homology to a previously identified but not characterized lncRNA in apple and pear when it was Blast searched against Genbank databases [57] ( Table 1). The lncRNA gene contained two introns and three exons ( Figure 3A). qRT-PCR analyses showed that the lncRNA was transcribed in mature apple fruit tissues after harvest ( Figure 3B). The transcript sequence of the lncRNA was PCR-amplified from cDNA of "Huashuo" fruit flesh and cloned for sequencing. Of the four sequenced clones, one contained a 911 bp and the other three contained an 894 bp cDNA fragment ( Figure S3). As "Huashuo" was shown to be homozygous for the lncRNA gene, identification of two cDNA fragments with different length indicated alterative splicing. The lncRNA transcript was analyzed for the presence of open reading frames (ORFs). Four short ORFs were found, ORF1-4, potentially encoding peptides of 68, 28 102, and 38 amino acids in length, respectively. ORF4 is enclosed within the sequence of ORF3 but uses a different reading frame ( Figure 3C). All protein sequences of these ORFs had no significant similarity to any other proteins as determined by BLAST [57]. Furthermore, we used this sequence to BLAST search GDDH13 (v1.1) [58] and HFTH (v1.0) [59] apple reference genomes [60]. The search identified five homologs of this sequence with 96.25% to 99.55% homology and located on chromosome 0, 1, 4, 5, and 11 (Table S3), but not located on chromosome10, suggesting that there was no insertion in the promoter region of MdPG1 (chromosome10) in "Golden Delicious" and "Hanfu" cultivar. The four homologs were PCRamplified from genomic DNA of "Huashuo". The locus was heterozygous on chromosome 0 and 1, homozygous on chromosome 5 and 11 (Table S4). Sequence alignments showed that there were several SNPs and indels among these sequences ( Figure S4), so that these homologs can be distinguished from each other. According to the apple lncRNA database information (https://www.tobaccodb.org/plncdb/, accessed on 1 March 2023) [61], six highly similar sequences were annotated as putative nonprotein-coding genes ( Figure S5). The protein-coding potential of the lncRNA was predicted as non-coding by the Coding Potential Calculator (http://cpc2.gao-lab.org, accessed on 1 March 2021) ( Figure 3D) [62], further suggesting that the transcript should be considered as an lncRNA, which was named as lncRNA PG1 . The expression pattern of lncRNA PG1 was further examined in different apple tissues including root tips, shoot tips, leaf, shoot bark, and fruit flesh of "Huashuo" using qRT-PCR analysis. LncRNA PG1 transcript was detected in fruit at 10 and 20 days after harvest (DAH), with the highest level at 20 DAH. Conversely, lncRNA PG1 transcript was hardly detected in root tips, shoot tips, leaf, shoot bark, or fruits before harvest at 30, 100 and 110 days after full bloom (DAFB) ( Figure 3B). These results indicated that lncRNA PG1 was expressed in fruit after harvest. days after full bloom (DAFB) ( Figure 3B). These results indicated that lncRNAPG1 was expressed in fruit after harvest.

lncRNAPG1 Inhibited MdPG1 Promoter Activity
To directly verify the importance of lncRNAPG1 in suppressing MdPG1 expression, the MdPG1 promoter (2.5 kb with SNP A ) was fused to the LUC reporter gene in pGreenII-0800-LUC to form the ProPG1-LUC reporter construct. The luciferase activity from this reporter construct in tobacco leaves was compared to that from CaMV35S-LUC and promoterless-LUC construct after Agrobacterium infiltration. CaMV35S-LUC construct as a positive control showed an extremely high level of luciferase activity. Conversely, the promoterless-LUC construct as a negative control showed low basal activity. The ProPG1-LUC construct showed significantly higher luciferase activity than the promoterless-LUC construct (p < 0.01) ( Figure 4A). However, when the lncRNAPG1 was linked to the upstream to the MdPG1 promoter, luciferase activity of the lncRNAPG1-ProPG1-LUC construct was reduced to the level similar to that of the promoterless-LUC construct ( Figure 4A). Therefore, lncRNAPG1 has a significant repression effect on the activity of the MdPG1 promoter. This inhibitory effect was slightly enhanced when CaMV35S-lncRNAPG1 was linked to the upstream to the MdPG1 promoter in the construct CaMV35S-lncRNAPG1-ProPG1-LUC ( Figure  4A).
To test whether the function of the lncRNAPG1 is related to its cis-location, CaMV35S-lncRNAPG1 in a separate vector was co-infiltrated together with ProPG1-LUC into Nicotiana benthamiana leaves. This co-infiltration did not reduce the luciferase activity driven by the MdPG1 promoter ( Figure 4A). The above experiments were repeated using an imaging

lncRNA PG1 Inhibited MdPG1 Promoter Activity
To directly verify the importance of lncRNA PG1 in suppressing MdPG1 expression, the MdPG1 promoter (2.5 kb with SNP A ) was fused to the LUC reporter gene in pGreenII-0800-LUC to form the ProPG1-LUC reporter construct. The luciferase activity from this reporter construct in tobacco leaves was compared to that from CaMV35S-LUC and promoterless-LUC construct after Agrobacterium infiltration. CaMV35S-LUC construct as a positive control showed an extremely high level of luciferase activity. Conversely, the promoterless-LUC construct as a negative control showed low basal activity. The ProPG1-LUC construct showed significantly higher luciferase activity than the promoterless-LUC construct (p < 0.01) ( Figure 4A). However, when the lncRNA PG1 was linked to the upstream to the MdPG1 promoter, luciferase activity of the lncRNA PG1 -ProPG1-LUC construct was reduced to the level similar to that of the promoterless-LUC construct ( Figure 4A). Therefore, lncRNA PG1 has a significant repression effect on the activity of the MdPG1 promoter. This inhibitory effect was slightly enhanced when CaMV35S-lncRNA PG1 was linked to the upstream to the MdPG1 promoter in the construct CaMV35S-lncRNA PG1 -ProPG1-LUC ( Figure 4A).

lncRNAPG1 Inhibited NtCHLH Promoter Activity in Transgenic Tobacco Plants
To test whether lncRNA inhibits promoter activity of other genes, we generated stable transgenic tobacco plants expressing the β-glucuronidase (GUS) report under the control of the promoter of the gene encoding Mg-chelatase H subunit (NtCHLH) (ProCHLH-GUS) or the lncRNAPG1 connected upstream of ProCHLH (lncRNA PG1-ProHLH-GUS) ( Figure 5A). NtCHLH was chosen because it encodes a key enzyme involved in chlorophyll synthesis and its promoter drives strong gene expression in tobacco leaves. For each construct, eight independent T0 transgenic plants were examined by GUS staining. Blue GUS staining was detected in leaf tissues from the eight transgenic plants of each construct but not in leaf tissues from four WT plants. The blue staining was stronger in leaves containing the ProCHLH-GUS construct than in leaves containing the lncRNAPG1-ProCHLH-GUS construct ( Figure 5A). Furthermore, GUS activity was quantified in the four WT plants, 12 transgenic lines of ProHLH-GUS, and seven transgenic lines of lncRNAPG1-ProHLH-GUS by using the 4-Methylumbelliferyl-beta-D-glucuronide (MUG) assay. The four WT plants showed no GUS activity, and 10 of the 12 ProHLH-GUS transgenic lines showed higher GUS activity than all seven lncRNAPG1-ProHLH-GUS transgenic lines ( Figure 5B). The mean GUS activity of the 12 ProHLH-GUS lines was 3.4 folder higher than that of the seven lncRNAPG1- To test whether the function of the lncRNA PG1 is related to its cis-location, CaMV35S-lncRNA PG1 in a separate vector was co-infiltrated together with ProPG1-LUC into Nicotiana benthamiana leaves. This co-infiltration did not reduce the luciferase activity driven by the MdPG1 promoter ( Figure 4A). The above experiments were repeated using an imaging system to reveal luciferase activity in tobacco leaves. The luminescence signal of CaMV35S-LUC construct was strongest, followed by ProPG1-LUC alone or co-infiltration of 35S-lncRNA PG1 -ProPG1-LUC. The signal was weak for the other three constructs, promoterless-LUC, lncRNA PG1 -ProPG1-LUC, and CaMV35S-lncRNA PG1 -ProPG1-LUC ( Figure 4B). The similar results were obtained by using only threes constructs to compare the ProPG1-LUC to lncRNA PG1 -ProPG1-LUC or CaMV35S-lncRNA PG1 -ProPG1-LUC ( Figure 4C). These results together indicate that lncRNA PG1 s inhibitory effect on MdPG1 promotor depends on its cis-location to the gene promoter.

lncRNA PG1 Inhibited NtCHLH Promoter Activity in Transgenic Tobacco Plants
To test whether lncRNA inhibits promoter activity of other genes, we generated stable transgenic tobacco plants expressing the β-glucuronidase (GUS) report under the control of the promoter of the gene encoding Mg-chelatase H subunit (NtCHLH) (ProCHLH-GUS) or the lncRNA PG1 connected upstream of ProCHLH (lncRNA PG1 -ProHLH-GUS) ( Figure 5A). NtCHLH was chosen because it encodes a key enzyme involved in chlorophyll synthesis and its promoter drives strong gene expression in tobacco leaves. For each construct, eight independent T 0 transgenic plants were examined by GUS staining. Blue GUS staining was detected in leaf tissues from the eight transgenic plants of each construct but not in leaf tissues from four WT plants. The blue staining was stronger in leaves containing the ProCHLH-GUS construct than in leaves containing the lncRNA PG1 -ProCHLH-GUS construct ( Figure 5A). ProHLH-GUS lines ( Figure 5C). Together, these results revealed that lncRNAPG1 reduced but not completely inhibited the promoter activity of NtCHLH in tobacco leaves.

An SNP in an ERF-Binding Element of MdPG1 Promoter Causes Changes of MdPG1 mRNA Level
Although DNA markers were developed for different MdPG1 alleles that were associated with different transcript levels of MdPG1 and fruit texture properties [22,23,25], the causative genetic variants were still unknown. A previous QTL analysis showed that an SNP in the MdPG1 coding sequence was associated with fruit firmness. The SNP was heterozygous (G/T) in cultivars with soft fruit, whereas it was homozygous (T/T) in cultivars with firm fruit [19]. Our study showed a similar result with this SNP position ( Figure 1C). As the SNPs in the coding sequence are unlikely to cause changes of gene expression, we investigated a potential cis-regulatory DNA element that may affect MdPG1 expression.
We identified variations in the 2.5 kb region upstream start codon of MdPG1 by comparing the sequences of four apple cultivars. These variations included 3 indels, and 34 SNPs in the promoter sequences of MdPG1 ( Figure S1). One of the SNPs was within an ERF-binding element ( Figure 2A) and was shown to be a key variant altering MdPG1 expression. Promoter elements play an important role in regulating gene expression and any change of sequences in these elements may alter gene expression levels and cause phenotypic variations [63,64]. For example, a 3 bp deletion within a putative W-box element of the ALMT9 promoter abolishes the binding between the promoter and the transcription factor WRKY42 that is a negative regulator of ALMT9 expression, and promotes a high Furthermore, GUS activity was quantified in the four WT plants, 12 transgenic lines of ProHLH-GUS, and seven transgenic lines of lncRNA PG1 -ProHLH-GUS by using the 4-Methylumbelliferyl-beta-D-glucuronide (MUG) assay. The four WT plants showed no GUS activity, and 10 of the 12 ProHLH-GUS transgenic lines showed higher GUS activity than all seven lncRNA PG1 -ProHLH-GUS transgenic lines ( Figure 5B). The mean GUS activity of the 12 ProHLH-GUS lines was 3.4 folder higher than that of the seven lncRNA PG1 -ProHLH-GUS lines ( Figure 5C). Together, these results revealed that lncRNA PG1 reduced but not completely inhibited the promoter activity of NtCHLH in tobacco leaves.

An SNP in an ERF-Binding Element of MdPG1 Promoter Causes Changes of MdPG1 mRNA Level
Although DNA markers were developed for different MdPG1 alleles that were associated with different transcript levels of MdPG1 and fruit texture properties [22,23,25], the causative genetic variants were still unknown. A previous QTL analysis showed that an SNP in the MdPG1 coding sequence was associated with fruit firmness. The SNP was heterozygous (G/T) in cultivars with soft fruit, whereas it was homozygous (T/T) in cultivars with firm fruit [19]. Our study showed a similar result with this SNP position ( Figure 1C). As the SNPs in the coding sequence are unlikely to cause changes of gene expression, we investigated a potential cis-regulatory DNA element that may affect MdPG1 expression.
We identified variations in the 2.5 kb region upstream start codon of MdPG1 by comparing the sequences of four apple cultivars. These variations included 3 indels, and 34 SNPs in the promoter sequences of MdPG1 ( Figure S1). One of the SNPs was within an ERF-binding element ( Figure 2A) and was shown to be a key variant altering MdPG1 expression. Promoter elements play an important role in regulating gene expression and any change of sequences in these elements may alter gene expression levels and cause phenotypic variations [63,64]. For example, a 3 bp deletion within a putative W-box element of the ALMT9 promoter abolishes the binding between the promoter and the transcription factor WRKY42 that is a negative regulator of ALMT9 expression, and promotes a high level of malate accumulation in tomato fruit [65]. An A/G SNP within the TCT-motif of the GBP1 promoter increases the GBP1 expression levels and leads to earlier flowering time and maturity in soybean [66]. An A/G SNP created a TCA element in the PbrmiR397a promoter inducing the PbrmiR397a expression and reducing the lignin content and stone cell number in pear fruit [67]. Our result added another example for SNP variation in promoter affecting gene expression.

MdCBF2 Regulates MdPG1 Expression by Binding to the ERF Element
After showing that the SNP C variant reduces the binding of the transcription factor MdCBF2, thus reducing MdPG1 expression, we carried out experiments to answer how the SNP may affect MdPG1 expression. Based on the knowledge that MdPG1 gene expression responds to ethylene signals and the SNP is located in an ERF-binding element, we decided to test whether the SNP changes the binding ability of EILs and ERFs. After testing two EILs and two ERFs, we showed that the ERF transcription factor MdCBF2 could strongly bind to and activate the promoter fragments containing the SNP A but not the promoter containing the SNP C (Figure 2). MdPG1 is known to be trans-activated by MdCBF2 from a previous study [12]. Our study further showed that this activation is dependent on a correct sequence of the ERF-binding element in the promoter of MdPG1. In Arabidopsis, it has been shown that the ethylene signal cascade ultimately leads to the stabilization of the primary responsive transcription factors EIN3/EILs, which have been shown to bind and active the secondary responsive transcription factors ERFs [68]. The AP2/ERF family is a large group of plant-specific transcription factors involved in plant developmental processes and multiple environmental stimuli. This family also includes the famous members CBFs, which are strongly cold-regulated [69]. AtEIN3 acts as a negative regulator of freezing stress by directly regulating the expression of AtCBF1-AtCBF3 [70]. The data presented here show that EIL2 and EIL3 hardly transactivate the MdPG1 promoter in a transient system. There is a possibility that the endogenous tobacco EILs are inhibited, so that CBF2 can play its role in transactivating MdPG1. Further work will be needed to explore whether other transcription factors are also involved in the regulation of MdPG1 expression, and whether they work in synergy with MdCBF2.
ERFs are known to bind to cis-acting elements, such as GCC box (GCCGCC), DRE (TACCGACAT), and CTR (TGGCCGAC) motifs, to regulate gene expression. For example, in apple, MdERF2 suppresses MdACS1 expression, whereas MdERF3 promotes its expression by binding to the DRE motif in its promoter. In tomato, LeERF2 activates the expression of ethylene biosynthesis genes by binding to the GCC box or DRE motif in their promoters. In papaya, CpERF9 represses CpPME1/2 and CpPG5 expression by binding to the GCC box in their promoters [8]. In peach, PpERF3 promotes the expression of the ABA biosynthesis gene PpNCED2/3 by binding to the ERF-binding motif in its promoter [9]. Our study identified and functionally tested a new ERF-binding element in apple.

LncRNA PG1 cis-Regulates Nearby Genes
Although lncRNAs may cisor trans-regulate gene expression [71], we demonstrated here that lncRNA PG1 cis-regulated the expression of the nearby gene MdPG1 (Figure 4). Luciferase transient expression assay in tobacco leaves showed that luciferase activity was reduced by fusing lncRNA PG1 to the MdPG1 promoter driving luciferase coding sequence. However, this reduction in luciferase activity was not detected when lncRNA PG1 was separated from the MdPG1 promoter by using two different constructs ( Figure 4A,B). Our results further showed that lncRNA could also reduce the promoter activity of NtCHLH in stable transgenic tobacco plants when the lncRNA and promoter were fused within one gene construct ( Figure 5). The data presented in this study suggest that lncRNA PG1 act as a cis-regulator.
There are at least three potential mechanisms for lncRNAs cis-acting gene expression. First, lncRNAs may modulate the action of the protein-coding genes involved in epigenetic patterning and chromatin remodeling or function as scaffolds. A plant lncRNA, COLDAIR, acts in the same way as Xist [72], which serves as scaffolds for the recruitment of PRC2 complexes to specific loci and induces epigenetic silencing [73]. Second, transcription or splicing of lncRNA may modulate gene expression in cis. This mechanism is exemplified by lncRNA Blustr (bivalent locus [Sfmbt2] is upregulated by splicing and transcribing an RNA). Both deletion and insertions in the promoter of the Blustr substantially reduced the expression of a neighboring gene [74,75]. Third, lncRNA locus may contain DNA elements acting as an enhancer to regulate neighboring gene expression. Examples for this type of lncRNAs include lincRNA-p21 in human [76], Bendr in mouse [74], and cis-NAT PHO1;2 in rice [40]. Their functions can largely be ascribed to conventional cis-acting DNA elements embedded within their gene body sequences.
In our case, the presence of a high level of lncRNA PG1 transcripts may not be required for its inhibitory effect. The inhibitory effect on the MdPG1 promotor was shown when a promotorless lncRNA PG1 was used but was not further enhanced when lncRNA PG1 was expressed from the strong CaMV35S promoter ( Figure 4A). Studies have shown that some lncRNA functions could result from processes that are not mediated by the lncRNA transcripts themselves, but instead involved general processes associated with their production, including enhancer-like activity of gene promoters, the process of transcription, and the splicing of the transcript [74]. It is also worth noting that, in the establishment of transcriptional gene silencing by cis-acting lncRNAs, continuous transcription might be more important than the production of mature RNA [74,77,78].

Cloning MdPG1 Promoter and Coding Sequence
Genomic DNA was extracted from apple cultivar "Meiba", "Huahong", "Huaxing" and "Huashuo" using a DNeasy Plant Mini Kit (Tiangen, Beijing, China) and used to amplify the DNA fragment of two MdPG1 alleles. The overlapping fragments covering the coding and promoter regions of each allele were amplified using two pairs of gene-specific primers PG-F1 and PG-R1, PG-F2 and PG-R2 (Table S5). The amplified products were separated by electrophoresis on 1% agarose gel and photographed under UV illumination, and the amplified fragments of each MdPG1 allele in the four apple cultivars were cloned into the p-blunt vector (TranGen, Beijin, China) and finally sequenced. Multiple alignments of DNA sequences that cover the promoter and coding regions of MdPG1 were performed using Geneious software v9.1.4 [79].

Cis-Element Analysis
The AP2/ERF and EIN3/EIL cis-elements in the 2.5 kb promoter sequence of both MdPG1 alleles were identified using the Plantpan2.0 database (http://plantpan2.itps.ncku. edu.tw, accessed on 1 June 2020) [56] and CIS-BP database (http://cisbp.ccbr.utoronto.ca/ TFTools.php, accessed on 1 March 2023) [80]. We accessed the MdPG1 promoter via the promoter analysis function in Plantpan2.0 database. Transcription factor binding sites were predicted from all species in the database. In addition, we directly used sequences of two alleles and compared their recognition scores through estimating respective false positive rates. Scan two sequences for differential TF binding tool were used and Malus_domestica specie was selected in CIS-BP database. This method identifies TFs with maximum E-score (which provide comprehensive scores for all possible eight base sequences) >0.45 for one allele, and maximum E-score < 0.45 for the other.

Dual-Luciferase Reporter Assay
A 2.5 kb fragment (upstream of the ATG start codon) of the MdPG1 promoter containing the ERF-binding SNP was amplified using PCR primers listed in Table S5, as described above. A fusion promoter of a 0.3 kb MdPG1 fragment (between −133 to −438 from the ATG start codon, containing the SNP site) and the minimal CaMV 35S promoter sequence (−46 to −1) [81] were synthesized by Tsingke Biotechnology Co., Ltd. (Beijing, China). These four promoters (2.5 kb with SNP A or SNP C , and 0.3 kb plus minimal CaMV 35S with SNP A or SNP C ) were cloned into the reporter vector pGreenII 0800-LUC. The full-length coding sequence of MdCBF2, MdAP2D32, MdEIL2, and MdEIL3 were amplified from fruit flesh cDNA of "Huahong" and cloned into the effector vector pSAK778 under the control of the CaMV35S promoter. The reporter and effector vectors were separately transferred into Agrobacterium tumefaciens GV3101 (pSoup) cells. A mixture of Agrobacterium cells containing the reporter and effector constructs (1:8, reporter: effector) was used to infiltrate young Nicotiana benthamiana leaves [82]. After infiltration for 72 h, LUC (Firefly luciferase) and REN (Renilla luciferase) activities were assayed using the Dual-Luciferase ® Reporter Assay System (Promega, Madison, WI, USA) and the SpectraMax @ i3x Platform (MOLECULAR DEVICES, San Jose, CA, USA). Six biological replicates were performed for each assay.

Yeast One-Hybrid Assay
The yeast one-hybrid assay was performed using a Matchmaker TM Gold Yeast One-Hybrid Library Screening System Kit (Clontech, San Francisco, CA, USA). The full-length coding sequence of MdCBF2 was cloned into the pGADT7 vector, and the 2.5 kb and 0.3 kb promoter sequences of MdPG1 with SNP A or SNP C were cloned into pAbAi vector to construct the bait vectors. The bait vector was introduced into Y1HGold yeast cells that were subsequently selected on SD/-Ura medium. The yeast cells containing the MdPG1 promoter-pAbAi bait vector were re-transformed with the MdCBF2-pGADT7 prey vector and selected on SD/-Leu medium. Yeast colonies that were confirmed to contain both bait and prey vectors were grown to reach a cell density of OD 600 1.5 and then diluted in 5-fold gradient. After the dilution, 5µL of suspension cells was spotted on SD/-Leu medium with or without the addition of AbA (aureobasidin A) and incubated for 3 to 5 days at 30 • C.

Analyzing lncRNA PG1 Sequence and Expression Pattern
Fruit of "Huashuo" were harvested at 30, 100, and 110 days after full bloom (DAFB). The fruit harvested at 110 DAFB were stored at room temperature for 10 and 20 days after harvest (DAF). For quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis, different tissues including root tips, shoot tips, leaf, shoot bark and fruits (30 DAFB, 100 DAFB, 110 DAFB, 10 DAF and 20 DAF) were collected from "Huashuo". Total RNA was isolated using a Total RNA Kit (Sangon, Shanghai, China) according to the manufacturer's instructions. First-strand cDNA was synthesized from the RNA samples with a Reverse Transcriptase Kit (Tiangen, Beijing, China). Real-time PCR was performed using SYBR Green PCR Master Mix on Roche LightCycler 480 system (Roche LightCycler, Roche, Basel, Switzerland). Relative gene expression was analyzed using apple reference genes MdEF1a and MdActin [83]. Normalization factors were calculated by taking the geometric mean of the two reference genes as determined by geNorm v3.4 [84]. The primers used for qRT-PCR were listed in Table S5.

Transient Assay of MdPG1 Promoter in Nicotiana benthamiana
To verify the inhibitory effect of lncRNA PG1 on the MdPG1 promoter, dual-luciferase reporter assays were carried out using six gene constructs. The 2.5 kb MdPG1 promoter fragment was fused to the LUC report gene in pGreenII 0800 [82] to form the ProPG1-LUC construct. The CaMV35S-lncRNA PG1 construct was generated by cloning 1.3 kb lncRNA PG1 gene sequence into pSAK778 [85] between the CaMV35S and OCS terminator in a sense orientation. CaMV35S-lncRNA PG1 and lncRNA PG1 were separately inserted into the ProPG1-LUC vector to form two vectors, CaMV35S-lncRNA PG1 -ProPG1-LUC and lncRNA PG1 -ProPG1-LUC. The positive and negative control vectors, CaMV35S-LUC and promoterless-LUC, were previously constructed [86]. The above constructs were transferred into Agrobacterium tumefaciens GV3101 cells by electroporation and were infiltrated into Nicotiana benthamiana leaves, as described previously [87]. At 72 h after infiltration, Firefly and Renilla luciferase activities were assayed using the Dual-Luciferase ® Reporter Assay System (Promega, USA) according to the manufacturer's instructions, and the activity of the MdPG1 promoter was expressed as the ratio of firefly to Renilla luciferase activities. The luciferase assay was carried out using SpectraMax @ i3x Platform (MOLECULAR DEVICES, San Jose, CA, USA) in three independent experiments. In each experiment, three biological replicates were analyzed. The luminescence signal was also detected on images by using a Tanon-5200Multi machine (Biotanon, Shanghai, China). The primers used for making the gene constructs were listed in Table S5.

Tobacco Transformation
The 1.6 kb promoter sequence of Mg-chelatase H subunit (NtCHLH) was amplified from genomic DNA of Nicotiana tabacum "NC89" and inserted into pBI121-GUS vector [88] to replace the CaMV35S promoter. The sequence of lncRNA PG1 was inserted upstream of the NtCHLH promoter sequence into the above construct. The resulting ProCHLH-GUS or lncRNA PG1 -ProCHLH-GUS constructs were transferred into Agrobacterium tumefaciens GV3101 cells separately by electroporation, and then into tobacco "NC89" plants using a leaf disk transformation protocol as described previously [89]. The transformed tobacco plants were selected using 200 mg/L kanamycin. Transgenic and wide-type (WT) tobacco plants were grown in a growth chamber (temperature of 23 • C, 16 h light/8 h dark, and 80% relative humidity).

Histochemical GUS Staining and Fluorometric Assays
For histochemical staining of GUS, 1 cm leaf discs of transgenic tobacco leaves were immediately treated with 5-bromo-4-chloro-3-indolyl b-D-glucuronide (X-Gluc) at 37 • C for 24 h using the Gusblue kit (Huayueyang Biotech Co., Ltd., Beijing, China). Stained samples were bleached with 70% (v/v) ethanol to remove the chlorophyll before photographing. GUS activity was determined by measuring the fluorescence of 4-methylumbelliferone produced by GUS cleavage of 4-methylumbelliferyl-β-D-glucuronide using the GUS Gene Quantitative Detection Kit (Coolaber, Beijing, China) according to the manufacturer's instructions. The protein concentration in the supernatant was determined using the Bradford procedure with bovine serum albumin (Sigma, St. Louis, MO, USA) as a standard.

Statistical Analyses
All analyses were carried out using analysis of variance (ANOVA) of statistical analysis system, and SPSS 17.0 Statistics (SPSS Inc., Chicago, IL, USA) [90]. Significance levels in comparison of the means were determined by p < 0.01 (Student's t-test). Post hoc differences between means were determined using Fisher's least significant difference (LSD) test [91] at the 5% significance level.

Conclusions
In conclusion, despite MdPG1 playing an important role in apple fruit softening, the molecular mechanism regulating MdPG1 remains largely unclear. Here, we have identified a causative SNP within an ERF-binding element in the MdPG1 promoter to affect MdPG1 expression through altering the binding of MdCBF2 ( Figure 6). In addition, we also identified a novel lncRNA, lncRNA PG1 , which is located in the promoter of MdPG1, and demonstrated that lncRNA PG1 negatively regulates the expression of MdPG1 through cis-regulation. These findings establish a novel lncRNA PG1 -MdPG1 regulatory network in apple fruit softening ( Figure 6). All analyses were carried out using analysis of variance (ANOVA) of statistical analysis system, and SPSS 17.0 Statistics (SPSS Inc., Chicago, IL, USA) [90]. Significance levels in comparison of the means were determined by p < 0.01 (Student's t-test). Post hoc differences between means were determined using Fisher's least significant difference (LSD) test [91] at the 5% significance level.

Conclusions
In conclusion, despite MdPG1 playing an important role in apple fruit softening, the molecular mechanism regulating MdPG1 remains largely unclear. Here, we have identified a causative SNP within an ERF-binding element in the MdPG1 promoter to affect MdPG1 expression through altering the binding of MdCBF2 ( Figure 6). In addition, we also identified a novel lncRNA, lncRNAPG1, which is located in the promoter of MdPG1, and demonstrated that lncRNAPG1 negatively regulates the expression of MdPG1 through cis-regulation. These findings establish a novel lncRNAPG1-MdPG1 regulatory network in apple fruit softening ( Figure 6). Figure 6. Proposed mechanism of MdPG1 allele-specific expression in apple fruit softening. MdCBF2 directly bind to the promoter of the SNP A allele of MdPG1 to enhance MdPG1 expression. lncRNAPG1 suppresses MdPG1 mRNA level through a cis-regulation mechanism. The diamond represents an ERF-binding element.

Supplementary Materials:
The following supporting information can be downloaded at: www.mdpi.com/xxx/s1, Figure S1. Sequence alignments of MdPG1 alleles of "Meiba", "Huahong", "Huaxing" and "Huashuo". (A-H) The sequences covering the promoter (A,C,E,G) and coding regions (B,D,F,H) of two alleles of MdPG1 were PCR-amplified from genomic DNA of "Meiba" (A,B), "Huahong" (C,D), "Huaxing" (E,F) and "Huashuo" (G,H). The amplified fragments were cloned and sequenced. Pairwise sequence alignment of the promoter and gene body regions of MdPG1 was performed using Geneious software v9.1.4. The start (ATG) and stop codon (TAA) of MdPG1 gene are shown in red box. AP2/ERF and EIN/EIL3 binding elements are highlighted with yellow and red color, respectively. An SNP within an ERF-binding element is using blue color. The SNPs in the 5′ UTR and CDS of two MdPG1 alleles are highlighted. The red sequence represents the 1.3 kb insertion (A,E,G); Figure S2. Phylogenetic analysis of the EIL and AP2 proteins. Rooted neighborjoining phylogenetic tree were constructed using protein sequences of members of the EIL (A) and AP2 (B) family from apple, Arabidopsis and kiwifruit. The GenBank or Arabidopsis TAIR accession numbers or Gene ID in GDR used are listed in Table S6; Figure S3. cDNA sequence of two lncRNAPG1 transcript isoforms. Two lncRNA cDNA sequences of 911 bp and 894 bp were cloned from "Huashuo" fruit flesh. lncRNA DNA and lncRNA cDNA sequences were aligned using Geneious software v9.1.4. The blue diamond showed the three exons of lncRNA; Figure S4. Sequence alignments of lncRNAPG1 homologs. LncRNAPG1 homolog sequences were identified from on chromosome 0, 1, 5, 11 and 10 by Blast search, PCR-amplified from genomic DNA of "Huashuo", cloned, Figure 6. Proposed mechanism of MdPG1 allele-specific expression in apple fruit softening. MdCBF2 directly bind to the promoter of the SNP A allele of MdPG1 to enhance MdPG1 expression. lncRNA PG1 suppresses MdPG1 mRNA level through a cis-regulation mechanism. The diamond represents an ERF-binding element.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/plants12071452/s1, Figure S1. Sequence alignments of MdPG1 alleles of "Meiba", "Huahong", "Huaxing" and "Huashuo". (A-H) The sequences covering the promoter (A,C,E,G) and coding regions (B,D,F,H) of two alleles of MdPG1 were PCR-amplified from genomic DNA of "Meiba" (A,B), "Huahong" (C,D), "Huaxing" (E,F) and "Huashuo" (G,H). The amplified fragments were cloned and sequenced. Pairwise sequence alignment of the promoter and gene body regions of MdPG1 was performed using Geneious software v9.1.4. The start (ATG) and stop codon (TAA) of MdPG1 gene are shown in red box. AP2/ERF and EIN/EIL3 binding elements are highlighted with yellow and red color, respectively. An SNP within an ERF-binding element is using blue color. The SNPs in the 5 UTR and CDS of two MdPG1 alleles are highlighted. The red sequence represents the 1.3 kb insertion (A,E,G); Figure S2. Phylogenetic analysis of the EIL and AP2 proteins. Rooted neighbor-joining phylogenetic tree were constructed using protein sequences of members of the EIL (A) and AP2 (B) family from apple, Arabidopsis and kiwifruit. The GenBank or Arabidopsis TAIR accession numbers or Gene ID in GDR used are listed in Table S6; Figure S3. cDNA sequence of two lncRNA PG1 transcript isoforms. Two lncRNA cDNA sequences of 911 bp and 894 bp were cloned from "Huashuo" fruit flesh. lncRNA DNA and lncRNA cDNA sequences were aligned using Geneious software v9.1.4. The blue diamond showed the three exons of lncRNA; Figure S4. Sequence alignments of lncRNA PG1 homologs. LncRNA PG1 homolog sequences were identified from on chromosome 0, 1, 5, 11 and 10 by Blast search, PCR-amplified from genomic DNA of "Huashuo", cloned, sequenced, and aligned using Geneious software v9.1.4; Figure S5. Six highly similar sequences of lncRNA were annotated as putative nonprotein-coding genes. According to the apple lncRNA database information, six highly similar sequences of lncRNA were annotated as putative nonprotein-coding genes. PLnc DB (https://www.tobaccodb.org/plncdb/, accessed on 1 March 2023); Table S1. Information of the four apples; Table S2. Scan two alleles of MdPG1 for differential TF binding; Table S3. LncRNA PG1 homologs identified in apple genome; Table S4. Genotyping of lncRNA PG1 homologs in eight apple cultivars; Table S5. The list of primer sequences used in this study; Table S6. Accession numbers of proteins to generate the phylogenetic tree.