Comparative Transcriptomic Analysis Revealed the Suppression and Alternative Splicing of Kiwifruit (Actinidia latifolia) NAP1 Gene Mediating Trichome Development

Kiwifruit (Actinidia chinensis) is commonly covered by fruit hairs (trichomes) that affect kiwifruit popularity in the commercial market. However, it remains largely unknown which gene mediates trichome development in kiwifruit. In this study, we analyzed two kiwifruit species, A. eriantha (Ae) with long, straight, and bushy trichomes and A. latifolia (Al) with short, distorted, and spare trichomes, by second- and third-generation RNA sequencing. Transcriptomic analysis indicated that the expression of the NAP1 gene, a positive regulator of trichome development, was suppressed in Al compared with that in Ae. Additionally, the alternative splicing of AlNAP1 produced two short transcripts (AlNAP1-AS1 and AlNAP1-AS2) lacking multiple exons, in addition to a full-length transcript of AlNAP1-FL. The defects of trichome development (short and distorted trichome) in Arabidopsis nap1 mutant were rescued by AlNAP1-FL but not by AlNAP1-AS1. AlNAP1-FL gene does not affect trichome density in nap1 mutant. The qRT−PCR analysis indicated that the alternative splicing further reduces the level of functional transcripts. These results indicated that the short and distorted trichomes in Al might be caused by the suppression and alternative splicing of AlNAP1. Together, we revealed that AlNAP1 mediates trichome development and is a good candidate target for genetic modification of trichome length in kiwifruit.


Introduction
Kiwifruit (Actinidia chinensis) is a popular and health-beneficial fruit because of its high vitamin C content and balanced nutrients, including dietary fiber, various minerals, and other metabolites [1,2]. Kiwifruit consumption improves immune, digestive, and metabolic health and even provides anticancer effects [3,4].
The ARP2/3-mediated nucleation of actin filaments is involved in trichome development [31,32]. The SCAR/WAVE complex is the major activator of ARP2/3-mediated F-actin nucleating and branching [33]. A defect in Nck-associated protein 1 (NAP1), a subunit of the SCAR/WAVE complex, causes short and distorted trichomes in Arabidopsis [34,35]. In soybean, two mutants with shorter and distorted trichome were revealed by map-based cloning to be caused by loss-of-function mutations in Glycine max NAP1 (GmNAP1) [36,37]. These results indicated that NAP1 plays an important role in trichome development.
Transcriptomic analysis has been widely used to discover the genes mediating trichome development in different species, such as tomato [38], cucumber [39], Lilium pumilum [40], and medicinal cannabis [41]. In this study, we analyzed two kiwifruit species (A. eriantha [42] with long bushy trichomes and A. latifolia [43] with short, sparse trichomes) by second-and third-generation transcriptome analysis. Our results indicated that the expression level of the NAP1 gene was much lower in A. latifolia than that in A. eriantha. Meanwhile, we discovered that there is alternative splicing of NAP1 mRNA in A. latifolia, which might be responsible for the shorter trichomes in A. latifolia.

Morphologic Observation of Trichomes in Two Kiwifruit Species
The fruits of A. latifolia (Al) are sparsely covered by short trichomes ( Figure S1A), while A. eriantha (Ae) fruits are densely covered by long intertwining trichomes ( Figure S1A), which is consistent with the previous observation [44]. The contrasting distribution patterns between Al and Ae trichomes were also observed in the petiole of mature leaves ( Figure 1A-D). The epidermis of Al petiole is attached with short trichomes ( Figure 1D,F), and the average length of trichomes is 100 µm ( Figure 1G). The trichomes of Ae petioles are much longer than that in Al (Figure 1C,E,G). Additionally, the trichome density in Ae petioles is much higher than that in Al ( Figure 1H). It is worth mentioning that Al trichomes are mildly distorted and brown ( Figure 1F), while Ae trichomes are straight and colorless ( Figure 1E). These observations indicated that the morphologic characterizations of Al and Ae trichomes are quite different. and Al (D). Trichome of petioles of Ae (E) and Al (F) was observed in microscope. Trichome length (G) and density (H) are shown in the mean values ± SD from three replicates. Significant differences were determined by Student's t-test (** indicates p < 0.001).

The Third-and Second-Generation Transcriptome Analysis
Since the time of flowering and fruit setting is quite different between these two species, the climatic conditions will dramatically affect the gene expression if we harvest the fruits at different times to perform the transcriptome analysis, which will obstruct our digging for the differentially expressed genes mediating trichome development. Considering that trichome distribution patterns on the fruits are similar to that on the petiole of mature leaves in two species (Figure 1), we used the petioles of mature leaves from the and Al (D). Trichome of petioles of Ae (E) and Al (F) was observed in microscope. Trichome length (G) and density (H) are shown in the mean values ± SD from three replicates. Significant differences were determined by Student's t-test (** indicates p < 0.001).

The Third-and Second-Generation Transcriptome Analysis
Since the time of flowering and fruit setting is quite different between these two species, the climatic conditions will dramatically affect the gene expression if we harvest the fruits at different times to perform the transcriptome analysis, which will obstruct our digging for the differentially expressed genes mediating trichome development. Considering that trichome distribution patterns on the fruits are similar to that on the petiole of mature leaves in two species (Figure 1), we used the petioles of mature leaves from the plants of two species grown in the same field for the RNA sequencing. Because the genome sequence of Al is lacking, the third-generation (full-length) RNA sequencing of mature leaves, including the petioles from Al and Ae, was performed to facilitate gene annotations. Table 1, we obtained 135,286 and 156,752 high-quality isoforms from third-generation RNA sequencing of Al and Ae, respectively. Among them, there are 126,102 and 93,226 non-redundant transcripts for Al and Ae, respectively (Table 1). These transcript sequences will contribute to gene annotations and further functional characterization. Meanwhile, three biological replicates of the epidermis of petioles samples from Al (Al-q1, -q2, and -q3) and Ae (Ae-q1, -q2, and -q3) were harvested for the second-generation RNA sequencing. The PCA analysis indicated that there is a high correlation among the three samples of each cultivar, suggesting a high repeatability of RNA sequencing results ( Figure S2). Next, we discovered 88,270 differentially expressed transcripts (DETs) (listed in Table S1), including 39,879 up-regulated and 48,391 down-regulated transcripts between Al and Ae samples ( Table 2). These DETs were also mapped to the genome of 'White' [42], which is provided by the kiwifruit genome database (http://kiwifruitgenome.org/, accessed on 3 October 2021) [45], and 12,950 differentially expressed genes (DEGs) were identified ( Table 2). KEGG analysis of DEGs indicated that the pathways of RNA metabolisms, including RNA processing (spliceosome), transport, degradation, surveillance, and polymerase (synthesis), were enriched ( Figure 2). No pathway of trichome development has been identified in KEGG enrichment.  Next, we screened out the homologous genes of trichome development-regulating genes in Arabidopsis from all the DEGs and compared their expressions between Al and Ae ( Figure 3A). Unexpectedly, the expression levels of most genes encoding a positive regulator of trichome development in Al, such as GL3, TTG1, TTG2, Constitutive Expressor of Pathogenesis-Related Genes5 (CPR5) [46], Zinc Finger Protein5 (ZFP5) [47], and GENERAL CONTROL NON-REPRESSED PROTEIN5 (GCN5) [48], are higher than that in Ae ( Figure 3A) while most of the genes encoding negative regulators, such as CPC, SPINDLY (SPY) [28], Jasmonate ZIM (JAZ) [29], and SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 9 (SPL9) [30], are down-regulated in Al but up-regulated in Ae ( Figure 3A). These results were opposite to our expectation since Al with fewer trichomes are supposed to down-regulate the positive regulators or up-regulate the negative regulators. The expression level of UPL3, a negative regulator, appeared to be lower in Ae than in Al ( Figure 3A), which requires further verification. Only one positive regulator, NAP1, as highlighted in yellow in Figure 3A, showed low expression in Al but a high expression in Ae, which is consistent with the phenotypic observation. Our qRT-PCR analysis confirmed that the expression patterns of NAP1 and other regulators of trichome development are consistent with the results of RNA seq ( Figure 3B). These results indicated that down-regulated NAP1 in Al might contribute to the phenotype that Al has short, sparse trichomes. Next, we screened out the homologous genes of trichome development-regulating genes in Arabidopsis from all the DEGs and compared their expressions between Al and Ae ( Figure 3A). Unexpectedly, the expression levels of most genes encoding a positive regulator of trichome development in Al, such as GL3, TTG1, TTG2, Constitutive Expressor of Pathogenesis-Related Genes5 (CPR5) [46], Zinc Finger Protein5 (ZFP5) [47], and GENERAL CONTROL NON-REPRESSED PROTEIN5 (GCN5) [48], are higher than that in Ae ( Figure 3A) while most of the genes encoding negative regulators, such as CPC, SPINDLY (SPY) [28], Jasmonate ZIM (JAZ) [29], and SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 9 (SPL9) [30], are down-regulated in Al but up-regulated in Ae ( Figure  3A). These results were opposite to our expectation since Al with fewer trichomes are supposed to down-regulate the positive regulators or up-regulate the negative regulators. The expression level of UPL3, a negative regulator, appeared to be lower in Ae than in Al ( Figure 3A), which requires further verification. Only one positive regulator, NAP1, as highlighted in yellow in Figure 3A, showed low expression in Al but a high expression in Ae, which is consistent with the phenotypic observation. Our qRT-PCR analysis confirmed that the expression patterns of NAP1 and other regulators of trichome development are consistent with the results of RNA seq ( Figure 3B). These results indicated that down-regulated NAP1 in Al might contribute to the phenotype that Al has short, sparse trichomes.

Alternative Splicing of AlNAP1 mRNA
To clone the NAP1 gene in kiwifruit, we used the samples for RNA-seq as a template and the same pair of primers targeting the 5′ and 3′ UTRs of both AeNAP1 (DTZ79_14g00610, accession number of Ae genome) and AlNAP1 to perform RT-PCR. The results showed that Ae has a clear band with the expected size ( Figure 4A), while Al has two alternatively spliced bands, named AlNAP1-AS1 and AlNAP1-AS2, in addition to the

Alternative Splicing of AlNAP1 mRNA
To clone the NAP1 gene in kiwifruit, we used the samples for RNA-seq as a template and the same pair of primers targeting the 5 and 3 UTRs of both AeNAP1 (DTZ79_14g00610, accession number of Ae genome) and AlNAP1 to perform RT-PCR. The results showed that Ae has a clear band with the expected size ( Figure 4A), while Al has two alternatively spliced bands, named AlNAP1-AS1 and AlNAP1-AS2, in addition to the full-length band AlNAP1-FL ( Figure 4A). These bands were cloned and analyzed by Sanger sequencing. The results indicated that the encoding sequences of AeNAP1 and AlNAP1-FL are almost the same except for two SNPs causing the substitution of two amino acids ( Figure S3).
We identified the 23 exons of AlNAP1 by comparing its cDNA sequence with the genome sequence of AeNAP1 Locus ( Figure 4B). The sequencing results of AlNAP1-AS1 and AlNAP1-AS2 indicated that AlNAP1-AS1 is short of the exons from 16th to 22nd, while AlNAP1-AS2 lacks most exons but remains the 1st and 23rd exons linked with two introns ( Figure 4B). The motif-based sequence analysis [49] indicated that AeNAP1 and AlNAP1-FL contain 15 conserved motifs that are present in all NAP1 proteins from Arabidopsis, rice, and soybean ( Figure 4C). AlNAP1-AS1 encodes a short protein lacking the conserved motif 1, 6, 11, 12, and 15 ( Figure 4C), while AlNAP1-AS2 encodes a very short peptide having no similarity with NAP1, suggesting that AlNAP1-FL might have the conserved function in trichome development while the AlNAP1-AS1 and -AS2 should lose the function. The qRT−PCR indicated that the expression level of all transcripts of AlNAP1, including AlNAP1-FL, -AS1, and -AS2 is about 40% of that of AeNAP1 ( Figure 4D). The expression of AlNAP1-FL is only 8% of that of AeNAP1 ( Figure 4D). These results indicated that the alternative splice of AlNAP1 further decreases the level of functional proteins. Moreover, the alternatively spliced bands of NAP1 were also amplified by RT−PCR from the epidermal tissues of Al young fruits but not from that of Ae fruits ( Figure S1B), suggesting that the alternative splice of AlNAP1 occurs in both leaves and fruits.
full-length band AlNAP1-FL ( Figure 4A). These bands were cloned and analyzed by Sanger sequencing. The results indicated that the encoding sequences of AeNAP1 and AlNAP1-FL are almost the same except for two SNPs causing the substitution of two amino acids ( Figure S3). We identified the 23 exons of AlNAP1 by comparing its cDNA sequence with the genome sequence of AeNAP1 Locus ( Figure 4B). The sequencing results of AlNAP1-AS1 and AlNAP1-AS2 indicated that AlNAP1-AS1 is short of the exons from 16th to 22nd, while AlNAP1-AS2 lacks most exons but remains the 1st and 23rd exons linked with two introns ( Figure 4B). The motif-based sequence analysis [49] indicated that AeNAP1 and AlNAP1-

Complementation of Arabidopsis nap1 Mutant Was Achieved by AlNAP1-FL but Not by AlNAP1-AS1
To verify the function of AlNAP1 in trichome development, we tried to express AlNAP1-FL and AlNAP1-AS1 in the mutant nap1-2 (SALK_014298, nap1 for short) of Arabidopsis [34]. A his-tag was fused to the C terminus of AlNAP1-FL and AlNAP-AS1, respectively, and the fused genes were driven by the CaMV35S promoter ( Figure 5A).
The nap1 mutant was transformed by the agrobacteria containing the construct of 35S:: AlNAP1-FL-His and 35S::AlNAP1-AS1-His, respectively. Three independent transgenic lines for each construct were identified by Western blot analysis using anti-His antibodies ( Figure 5B). The nap1 mutant showed shorter and distorted trichomes [34] compared with the Col-0 ( Figure 5C,D). Three independent transgenic lines of nap1/AlNAP1-FL (−1, −2, and −3) showed similar phenotypes in that their trichomes are long and straight ( Figure 5C just showed the trichomes of nap1/AlNAP1-FL-1). The trichomes of nap1/AlNAP1-FL-1 are much longer than that of nap1 and indistinguishable from that of Col-0 plants ( Figure 5D). However, three transgenic lines of nap1/AlNAP1-AS1 (−1, −2, and −3) showed short and distorted trichomes ( Figure 5C just showed the trichomes of nap1/AlNAP1-AS1-1), which are as short as that of nap1 mutant ( Figure 5D). The trichome density of the nap1 mutant is not altered by AlNAP1-FL or AlNAP1-AS1 overexpression ( Figure 5D). These results indicated that AlNAP1-FL, but not AlNAP1-AS1, rescued the defects of trichome development in the nap1 mutant.

Discussion
The trichomes protect plants from mechanical damage, insect predation, UV radiation, and excess transpiration [9]. In kiwifruit, trichomes might play a role in the resistance to bacterial canker caused by Pseudomonas syringae pv. Actinidiae [50]. Commercially, trichomes affect the kiwifruit popularity in the market [5]. Although brushing with commer-

Discussion
The trichomes protect plants from mechanical damage, insect predation, UV radiation, and excess transpiration [9]. In kiwifruit, trichomes might play a role in the resistance to bacterial canker caused by Pseudomonas syringae pv. Actinidiae [50]. Commercially, trichomes affect the kiwifruit popularity in the market [5]. Although brushing with commercial brushes has been practically used to remove the surface trichomes of kiwifruit, brushing promotes kiwifruit softening and reduces the shelf life [6]. Therefore, it is better to breed new kiwifruit cultivars with shorter and fewer trichomes or genetically modify the popular cultivar by targeting the key genes mediating trichome development. The genes regulating trichome development have been extensively identified in Arabidopsis [10][11][12]. To the best of our knowledge, no such gene has been characterized in kiwifruit.
In this study, we observed that Al has short, mildly distorted, and brown trichomes ( Figure 1F), while Ae has long, straight, and colorless ones ( Figure 1E). The trichome density in Al petioles is much lower than that in Ae ( Figure 1H). By the second-and third-generation transcriptome analysis, we revealed that the mRNA levels of most DEGs mediating trichome development are contrary to the observed phenotypes of trichome ( Figure 3). Only NAP1 expression, which is lower in Al than in Ae, is consistent with the trichome phenotypes ( Figure 3). NAP1, a subunit of the SCAR/WAVE complex, is required for the activation of the Arp2/3 complex, which is a crucial regulator of actin nucleation and branching. A defect on NAP1 gene causes short and distorted trichomes in both Arabidopsis [34,35] and soybean [36,37], although the underlying mechanisms of regulating trichome development remains largely unknown. In our study, the alternative splicing of AlNAP1 mRNA occurs in both leaves and fruits of Al but not in that of Ae (Figures 4 and S1B). Sanger sequencing indicated that two spliced mRNA, AlNAP1-AS1 and AlNAP1-AS2, lost 7 exons and 21 exons, respectively ( Figure 4B). AlNAP1-AS2 encodes a totally different protein that has no similarity with NAP1. AlNAP1-AS1 encodes a short protein lacking the conserved motif 1, 6, 11, 12, and 15 ( Figure 4C) and cannot rescue the defects of trichome development in mutant nap1 of Arabidopsis ( Figure 5), suggesting that AlNAP1-AS1 is not a functional protein. The alternative splicing further down-regulated the expression level of AlNAP1-FL ( Figure 4D), which has been verified to have functions in regulating trichome development ( Figure 5). Defects on NAP1 genes cause shorter and distorted trichome in Arabidopsis [34] and soybean [36,37]. Therefore, we speculate that the suppression and alternative splicing of AlNAP1 is responsible for the short, mildly distorted trichome of Al. Considering that the trichome densities in nap1 mutant and nap1/AlNAP1-FL plants are not altered compared to that of Col-0 ( Figure 5), we believe that low trichome density in Al has nothing to do with the AlNAP1 gene. Of course, verification of our speculation requires further investigations, such as genetic modification of the NAP1 gene in Al, which is currently not practicable because the method of Al transformation is not established yet.
As for the reason why alternative splicing of NAP1 exists in Al but not in Ae, we revealed that the genes belonging to spliceosome were most significantly enriched in our transcriptomic analysis (Figure 2). Differentially expressed spliceosome genes in Al might contribute to the alternative splicing of AlNAP1. To date, no spliceosome gene has been demonstrated to participate in trichome development. It will be interesting to figure out which spliceosome gene plays a key role in the processing of AlNAP1 in the future.
Among the negative regulators identified by our transcriptome, UPL3 appeared to have lower levels in Ae than in Al ( Figure 3A). UPL3 is an E3 ligase mediating ubiquitin/26S proteasome-dependent degradation of GL3 and EGL3 [27]. Overexpression of GL3 significantly increased the trichome number in Arabidopsis [16] and even had a stronger effect on trichome density in the upl3 mutant [27]. Therefore, we speculate that the lower expression of UPL3 in Ae might cause the increased protein abundance of GL3, which might be responsible for the high trichome density in Ae.
Together, our results indicate that the shorter and distorted trichomes in Al might be caused by the suppression and alternative splicing of AlNAP1, which can be a potential target for genetic modification of trichome length in kiwifruit.

Plant Material and Growth Conditions
The kiwi species "Actinidia eriantha" and "Actinidia latifolia" were grown on a kiwi plantation at Anhui Agricultural University. Hefei, China. The leaves and petioles of "Actinidia eriantha" and "Actinidia latifolia" with similar development periods and states were cut down and immediately frozen in liquid nitrogen and stored at −80 • C.
The Arabidopsis nap1 mutant strain SALK_014298 was ordered from the AraShare Science website (www.arashare.cn, accessed on 10 November 2021). Homozygous plants were identified by the triple primer method, as described previously. Arabidopsis Columbia ecotype (Col-0) and nap1 mutant were sterilized and vernalized for 3 d at 4 • C and then were sown on MS medium for 10 d under 14 h light/10 h dark cycle at 22 • C. Then, Arabidopsis seedlings were transferred to a soil mixture of vegetative soil and vermiculite (3:1, v/v) and grown at a long-day condition (22 • C, 14/10-h light/dark).

RNA Extraction and qRT−PCR Assays
RNA was extracted from the leaves of "Actinidia eriantha" and "Actinidia latifolia" using an RNA plant plus Reagent kit (TianGen, Beijing, China) and transcribed into cDNA using the PrimeScript ® RT reagent kit (Perfect Real Time, TaKaRa). Fluorescent quantitative primers were designed by Primer-BLAST in NCBI (www.ncbi.nlm.nih.gov/tools/primerblast/, accessed on 25 November 2021). The qRT-PCR was carried out using the CFX connectTM Real-Time System (BIORAD, US), qRT-PCR reactions were as follows: predenaturation at 95 • C for 10 min, denaturation at 95 • C for 15 s, annealing at 56 • C for 15 s, extension at 65 • C for 10 s, 40 cycles, relative expression values were analyzed via the cycle threshold (Ct) 2 −∆∆CT method [51]. Three biological replicates for each sample and three technical replicates for each biological replicate. The qRT−PCR primers are listed in Supplemental Table S2.

Cloning of AlNAP1 and AeNAP1
The full-length cDNA of AlNAP1 and AeNAP1 were amplified with the first strand cDNA using the forward primer 5 -TCCAACAATCGGCCTTCCCTAC-3 and the reverse primer 5 -AGGACCAGACCTTGACACAGC-3 , respectively. The genes were amplified by PrimeSTAR ® Max DNA Polymerase (Takara, Dalian, CN), and PCR reaction conditions were pre-incubation at 98 • C for 3 min, followed by 35 cycles of 98 • C (30 s), 55 • C (30 s), and 72 • C (5 min), with a final extension at 72 • C for 5 min. The PCR product was recovered, ligated to the pESI-Blunt simple vector, and then transferred into E. coli. Recombinant clones were identified by PCR and sequencing.

Vector Construction and Plant Transformation
A His-tag (6×His) was fused to the coding sequence of AlNAP1-FL and AlNAP1-AS1, respectively, by PCR. The fused genes AlNAP1-FL-His and AlNAP1-AS1-His were transformed into the expression vector pCAMBIA1302. The constructed vectors 35S::AlNAP1-FL-His and 35S::AlNAP1-AS1-His were transformed into Arabidopsis nap1 mutant strain using the floral dip transformation method as described previously [52]. Transgenic lines were then screened with 60 mg/L hygromycin. Two independent transgenic plants for each vector were identified and used for further experiments.

Sequence Alignment and Protein Domain Analysis
Sequence alignments were performed using ApE software, a phylogenetic tree was performed using the neighbor-joining method by the MEGA (ver. 7.1), and protein domains were analyzed using the NCBI Conserved Domain Search website (www.ncbi.nlm.nih.gov/ Structure/cdd/wrpsb.cgi, accessed on 8 January 2022). The motifs of the NAP1 gene were further analyzed using the MEME-MEME Suite website (meme-suite.org, accessed on 10 January 2022) [49].

Third-and Second-Generation Transcriptome Sequencing
Since the time of flowering and fruit setting is quite different between these two species, the climatic conditions will dramatically affect the gene expression if we harvest the fruits at different times to perform the transcriptome analysis, which will obstruct our digging for the differentially expressed genes mediating trichome development. Considering that trichome distribution patterns on the fruits were similar to that on the petiole of mature leaves in two species, we used the mature leaves and the epidermal tissues of the petiole for third-and second-generation transcriptome analysis, respectively. For the third-generation transcriptome, we just had one sample, which is a mixture of three leaves from different trees. Total RNA was extracted from the mature leaves (including petioles) of "Actinidia eriantha" and "Actinidia latifolia" and used for third-generation transcriptome analysis. For the second-generation transcriptome, we have three samples (biological replicates) for each specie, which means that three petioles were detached from three different trees. Total RNA was also extracted from epidermal tissues of petioles and used for second-generation transcriptome analysis. The RNA sequencing and analysis were performed by Biomarker biotech Co., Ltd. (Qingdao, China). Briefly, after total RNA was extracted, eukaryotic mRNA was enriched by magnetic beads with Oligo (dT) and then was randomly interrupted by fragmentation buffer. One-stranded cDNA was synthesized from the interrupted mRNA template using six-base random primers. Then, the second cDNA strand was synthesized by adding buffer, dNTPs, RNase H, and DNA polymerase I. The resulting double-stranded cDNA was purified using AMPure XP beads, and the terminal was repaired, added with A tail, and connected with the sequencing connector, and then AMPure XP beads were used to select the fragment size, and the cDNA library was obtained by PCR enrichment. The constructed library was sequenced using the Illumina platform to generate 150 bp double-ended reads; the total read is above 6G bp. The RNA sequencing data were obtained from three biological replicates, and PCA analysis was used as an assessment indicator of biological repeat relevance. PCA analysis was performed using the base package stats in R (v3.1.1), and visualization was performed using the ggplot2 package (v1.0.1). The data were obtained from three biological replicates, use of Spearman's correlation coefficient R as an assessment indicator of biological repeat relevance. The expressed genes were analyzed; log2 fold change values > 1 and log 2 fold change values < −1 were considered significant. The significantly induced (>2-fold) and repressed (<0.5-fold) genes are listed in Supplemental Table S1.

Protein Extraction and Western Blotting
Approximately 500 mg of Arabidopsis seedlings were ground in a 2× Laemmli sample buffer containing 100 mM Tris-HCl (pH6.8), 4% SDS, 0.2% BPB (bromophenol blue), 2% beta-ME, and 20% glycerol. The samples were boiled at 100 • C for 15 min and then centrifuged for 10 min at 12,000 rpm. Protein extracts were separated on a 10% SDS-PAGE and transferred to PVDF blotting membranes as described previously [53]. The 35S::AlNAP1-FL-His and 35S::AlNAP1-AS1-His protein amounts were determined by protein gel blotting using anti-His antibodies.

Statistical Analysis
Epidermal samples were photographed under a microscope (E200MV; Nikon, Nanjing, China), and the length and density of epidermal fur were measured by ImageJ-Win64 software. The data were obtained from three biological replicates. SPSS v21 (Statistic Package for Social Science) software was used to analyze the significance of the differences. The mean values ± SD from three replicates are presented, and significant differ-ences compared with the control were determined by Student's t-test (* indicates p < 0.05; ** indicates p < 0.001).

Conclusions
In this study, we observed that Ae has long, straight, and high-density trichomes, while Al has short, distorted, and low-density trichomes. The second-and third-generation transcriptomic analysis indicated that the expression of the NAP1 gene was suppressed in Al compared with that in Ae. Additionally, alternative splicing of AlNAP1 existed in both leaves and fruits of Al but not in that of Ae. The defects of trichome development in Arabidopsis nap1 mutant were rescued by full-length AlNAP1 but not by the spliced transcript of AlNAP1. AlNAP1 gene overexpression does not affect trichome density in nap1 mutant or wild-type plants. The alternative splicing further decreased the level of functional transcript AlNAP1-FL, which might cause the short and distorted trichomes in Al. Overall, we revealed that AlNAP1 mediates trichome development and is a good candidate target for genetic modification of trichome length in kiwifruit.

Data Availability Statement:
The raw data of our transcriptomic analysis were uploaded to Bio-Project Library (https://www.ncbi.nlm.nih.gov/bioproject/, accessed on 18 October 2022), and the accession number is PRJNA889934.

Conflicts of Interest:
The authors declare no conflict of interest.