Transcriptome analysis at mid-stage development seeds in litchi with contrasting seed size

Litchi is a sub-tropical fruit crop with contrasting genotypes bearing fruits with variable seed size. Small seed size is a desirable trait in litchi, as it improves consumer preference and facilitates fruit processing. Seed specic transcriptome analysis was performed in two litchi genotypes with contrasting seed size to identify the genes associated with seed development in litchi fruits. The transcriptomic data from seeds at mid- development stages (16 to 28 days after anthesis) were de-novo assembled into 1,39,608 Trinity transcripts. Out of these, 6,325 transcripts expressed differentially between the two contrasting genotypes. Several putative genes for salicylic acid, jasmonic acid and brassinosteriod pathways were down-regulated in the small-seeded litchi. The putative regulators of seed maturation and seed storage were down-regulated in the small-seeded genotype. Embryogenesis, cell expansion, seed size and stress related Trinity transcripts exhibited differential expression. Further studies can lead to identication and characterization of early regulators of seed size in litchi.


Introduction
Litchi (Litchi chinensis Sonn.) is an important fruit crop of Sapindaceae family, widely cultivated in tropical and sub-tropical regions. The taste, aroma, nutritional value and medicinal properties of juicy aril have made it a popular fruit worldwide (Ibrahim and Mohamed 2015;Huang et al. 2016;Septembre-Malaterre et al. 2016;Kilari and Putta 2017). The development of aril in litchi is favored by disruption in its seed development. As a result, the fruits with small or rudimentary seed develop larger esh in fruit.
Litchi fruits with large pulp and small seed are liked by the consumers and fruit processing industries.
In dicots, generally seed development is initiated after double fertilization of the egg cell and the central cell in ovule, which grows into seed. Inside the seed, the fertilized egg cell (diploid) develops into embryo and the central cell into triploid endosperm. During early development stages, proliferation of endosperm happens, followed by consumption of the endosperm by growing embryo and subsequently, latter's development into cotyledons. The protective covering of ovule (integuments) develops into seed coat.
Numerous physiological events and biochemical reactions have been reported to determine seed mass and composition, and nally the size of seed in fruit (Ohto et al. 2005).
Seed development is in uenced by two main mechanisms, parthenocarpy, where fertilization fails, and stenospermy, where fertilization occurs but embryo development is disrupted (Varoquaux et al. 2000).
Incidences of both parthenocarpy and stenospermy (post-fertilization embryo abortion) have been observed in litchi cultivars. Embryo abortion in litchi happens at different embryo development stages e.g. globular, heart shaped, torpedo and cotyledon stage (Yu-Shen Liang 2012). Hence, litchi plant is an excellent material to understand molecular mechanism of seed development through differential transcriptomics at different developmental stages of seeds. Generally, the embryo aborts at or after globular stage and develops large aril with small seeded-normal size fruit (Yu-Shen Liang 2012).
We have earlier (Pathak et al. 2016) reported early-stage seed speci c transcriptional pro ling in two litchi genotypes with contrasting seed size. In this report, we take the earlier study forward. Global genomic expression pro les are now compared for mid-stage seeds of the two contrasting litchi genotypes. A number of genes and pathways whose expression is differentially regulated in small-seeded litchi at 16 to 28 days post anthesis were identi ed. The genome expression pro les at early-and mid-stage seeds were compared.

Material And Methods
Plant material, DNA and RNA isolation Plant materials (developing fruits) were collected from ICAR -National Research Centre on Litchi, Mushahari farm, Mushahari, Muzaffarpur, Bihar, India. In this study, Bedana (B) was chosen as the smallseeded litchi genotype and China (C) as the bold seeded genotype. The two genotypes with contrasting seed size are genetically fairly close (Nei genetic distance of 0.46) (Pathak et al. 2014) as compared to the other popular cultivars in India and grouped in the same cluster as the previously reported litchi transcriptome genotypes. Developing fruits at 16, 20, 24 and 28 days after anthesis (DAA) were harvested, dipped in liquid N 2 , and stored in -80 o C (Pathak et al. 2016). Seeds were excised from the stored fruits and total RNA was isolated, using Spectrum Plant Total RNA kit (Sigma Aldrich St. Louis, MO, USA), following manufacturer's protocol. DNase I (Sigma Aldrich St. Louis, MO, USA) was used to remove DNA. The integrity of total RNA was con rmed by using gel electrophoresis using BioAnalyzer. TrueSeq libraries were generated and sequenced on Illumina HiSeq 1000 (Illumina, San Diego, CA) platform at Centre for Cellular and Molecular Platforms C-CAMP, Bangalore, India. RNAseq datasets were submitted to NCBI for both genotypes at different seed developmental stages with accession number SRP076766.
De-novo reference assembly and prediction of transcript expression De-novo transcriptome assembly was performed using combined reads obtained from transcriptome sequencing of developing seeds (16, 20, 24 and 28 DAA) in the two contrasting genotypes (B & C).
Adapter and low quality sequences were removed, employing Trinity pipeline at default parameters, and the high quality reads were used for assembly. Putative functions were assigned to the assembled Trinity transcripts employing WImpiBLAST tool (Sharma and Mantri 2014) to perform BLASTx homology search against several public databases-NCBI NR, Swissprot, protein databases of the selected plants e.g. Arabidopsis thaliana, Citrus sinensis, Ricinus communis, Populus trichocarpa, Fragaria vesca, Carica papaya and Glycine max.
Relative expression of Trinity transcripts among RNAseq libraries from two litchi genotypes at four seed developmental stages was obtained as TPM (transcripts per million) and FPKM (fragments per kilobase of transcripts per million mapped reads), using RSEM version 1.2.9 at default parameters (Pathak et al. 2016). Expression levels of different Trinity transcripts were compared among the samples by Edger Bioconductor, using script run_DE_analysis.pl as default parameter. The differentially regulated transcripts were retrieved at the cutoff-Log 2 fold ≥ 2; P-value ≤ 0.001.

Quantitative real-time PCR analysis
A few representative, differentially expressed transcripts obtained by high throughput sequencing were validated by quantitative real-time PCR. Total RNA was isolated from seeds at 20 and 24 DAA, and cDNA synthesized from total RNA (500 ng) using oligo (dT) primers and M-MLV reverse transcriptase, according to the manufacturer's instructions (Invitrogen, USA) in a 20 μl volume. Transcript levels were analyzed by quantitative real-time PCR using the fast SYBR green master mix (Applied Biosystems, USA) and an ABI 7500 Real-Time PCR System (Applied Biosystems, USA) according to the manufacturers' instructions. All biological replicates were analyzed in duplicate. Real-time PCR reactions were normalized to the Ct values for litchi Actin (HQ615689). The relative expression levels of the target genes were calculated using the formula 2-△△CT.

Results
Transcriptome in developing seeds of bold-and small-seeded litchi cultivars Transcriptome was sequenced from developing seeds of large-(C) and small (B)-seeded litchi genotypes at 16, 20, 24 and 28 DAA using Illumina paired end sequencing. After stringent quality assessment and data ltering, 334,574,104 clean reads were assembled into 1,39,608 Trinity transcripts (Table. S1). These were matched (BLASTx) with ten different publically available databases (E-value ≤ 10 -5 ). A total of 56,305 transcripts were annotated, out which 36,099 transcripts matched with the protein coding genes.
In BLASTx analysis after NCBI-NR database, the litchi transcripts exhibited highest homology with Fragaria vesca. (Table.S2). BLASTn analysis (E-value of 10 -5 and query coverage ≥ 60 %) revealed transcription information of a total of 45,612 Trinity transcripts, which have not been reported previously.

Differential expression analysis
A total of 6,325 Trinity transcripts were identi ed as signi cantly differentially expressed (log 2 fold change ≥ 2; P value ≤ 0.001) in the developing seeds of the two contrasting litchi genotypes. Out of these, 866 transcripts were differentially expressed at all the four developmental stages (16, 20, 24 and 28 DAA). The highest deviation in expression pattern was observed between the bold-and small-seeded seeds at 16 DAA in mid-seed developmental stages (Fig. S1). Highest percentage of differentially expressed genes was observed at 0 DAA ( Fig S2). It supports dominant role of maternal factors in seed size regulation of litchi. The principal component analysis (Fig. S3) of the differentially expressed transcripts revealed 84.34 % variability by two components. The rst component explains 44.44% variability among the developing seeds, whereas the second component accounts for 39.90% variability and grouped the genomic pro les of the bold-and small-seeded litchi genotype separately.
On the basis of homology with Arabidopsis thaliana, the genes that expressed differentially were identi ed at each developmental stage. GO enrichment patterns showed higher representation of transcripts related to biological processes like regulation of primary metabolic process, cellular biosynthetic process, cellular metabolic process etc. in developing seeds of the small-seeded cultivar. This suggests that the seeds in small-seeded litchi are metabolically more active than in the bold-seeded varieties. The down-regulated transcripts in small-seeded genotype showed association with the biological processes, such as post-embryonic development, organ development, anatomical structure development, multicellular organismal development, defense response, protein amino acid phosphorylation etc.
Expression pro le of putative hormone related genes Putative hormone related transcripts were predicted in litchi on the basis of homology with Arabidopsis thaliana. The BLASTx search revealed the expression of a total of 1,357 hormone related transcripts in the developing seeds. A total of 86, 43, 55 and 72 transcripts were differentially expressed (bold-seeded vs small-seeded) in seeds at 16, 20, 24 and 28 DAA, respectively. Majority of the putative genes for brassinosteroid, jasmonic acid and salicylic acid pathways were down regulated in developing seeds of the small-seeded litchi (Fig. 1, 2 & 3). The putative genes involved in brassinosteroid signal transduction, BSK3 and BAK1 (c2787_g1_i1 _AT4G00710 & c66146_g1_i2 _AT4G33430), were down regulated in the seeds of small-seeded genotype at all the four developmental stages (Fig. 1). De ciency of BAK1 leads to induction of necrosis upon infection (Kemmerling et al. 2007). At 24 DAA, the putative NAC 100 (c56916_g1_i1 _AT5G61430) and EXO ( AT4G08950; c49843_g1_i1, c49843_g2_i1& c49843_g3_i1) genes, reported to regulate cell expansion (Schröder et al. 2009;Pei et al. 2013), showed higher expression in seed of small-seeded litchi (Fig. 1). The putative SAG 29 (c51255_g1_i1 _AT5G13170), which may induce cell death under stress (Seo et al. 2011) and organic cation/carnitine transporter 3 (OCT-3) (c54181_g1_i1 _AT1G16390) a stress induced gene (Küfner and Koch 2008), were up-regulated at 20 and 28 DAA, respectively, in small-seeded litchi (Fig. 1).

Expression pro le of putative transcription factors
Transcription factors (TFs) control various developmental aspects having direct relation with seed speci c traits, such as embryo and endosperm development, maturation etc (McElver et al. 2001;Braybrook and Harada 2008). BLASTx search revealed a total of 2,324 putative TFs involved in advanced stages of seed development in litchi. A total of 118, 45, 76 and 120 putative TF transcripts were differentially expressed (bold-seeded vs small-seeded) at 16, 20, 24 and 28 DAA, respectively (Fig. 4).
At 16 DAA most of the putative TFs showed relatively suppressed transcriptional pattern in small-seeded litchi (Fig. 4). The transcripts related to the TF families, WRKY, WOX, NF-YC, MYB, HSF, bHLH and bZIP, exhibited lower expression in small-seeded seeds at 16 and 28 DAA (Fig. 4). The TF families, RAV, SRS, TALE and TCP exhibited enhanced expression at 28 DAA, while TALE and TCP were down-regulated at 16 DAA in small-seeded litchi (Fig.4).
At 20 DAA, expression of transcription factor families possibly involved in seed development, such as AP2, E2F/DP, G2-like, GRF, MYB, NF-YC and M-type was found repressed in seeds of small-seeded litchi (Fig. 4). The TF families C2H2, NAC, Trihelix and NF-YB were up-regulated in seeds of small-seeded litchi (Fig. 4). The TFs, which are involved in regulation of cell proliferation, cell-wall synthesis, and seed lipid content, TOE1 (c54459_g2_i2 _AT2G28550) and WRI (c63744_g1_i1 _AT3G54320) (Cernac and Benning 2004), were down-regulated in the small-seeded genotype (Fig. 4). Single mutant of toe1 (Arabidopsis) is reported to cause early owering. Mutant of WRI1 is reported to fail seed oil content in Arabidopsis seeds ).

Embryogenesis and seed maturation related genes in litchi
The asymmetric cell division in zygote and differentiation of cells results into embryo development. In the litchi seed, transcripts orthologous to the genes essential for embryo development in Arabidopsis. thaliana were identi ed. Altered expression of these genes has been reported to develop defective embryos (Meinke et al. 2008) at pre-globular (AT2G32590, AT5G14760 & AT5G13690), globular (AT3G04340, AT5G16715 & AT1G08840), and cotyledon (AT1G21970, AT3G19700, AT4G30580, AT4G33090, AT1G20200, AT3G06350, AT3G54320, AT3G06350 & AT4G02570) stages. These were downregulated in the seeds of small-seeded litchi (Fig. 5). The aminopeptidase M1 (APM1) (AT4G33090) was down-regulated at all the mid-developmental stages (16 to 28 DAA) in small-seeded litchi genotype. AT1G08840, AT2G32590, AT3G06350, AT3G20070, AT3G54320, AT4G02570, AT5G13690, AT2G32950 and AT5G16715 were down-regulated at both early (0 to 14 DAA) and mid stage (16 to 28 DAA) stages of litchi embryo development (Fig. 5). The results suggest APM1 (AT4G33090) as a potential gene critical to embryo development in litchi and that leads to seed abortion, without affecting fruit size. It is known to cause embryo abortion at cotyledon stage in Arabidopsis thaliana.
Embryogenesis is tightly controlled for the development of different cell identities like the formation of outer (protoderm) versus inner layer, vascular tissues and determination of root and shoots domains. The putative PXL2 gene (c80364_g1_i1 _AT4G28650), which regulates the ordered cell division in undifferentiated cells (Fisher and Turner 2007), exhibited low level of expression in small-seeded genotype at 16 DAA (Table.S3). Suppressed transcription was recorded for the seed size regulator, CYP78A6 (24DAA) (c59232_g1_i1_AT2G46660) and IKU2 (16 & 28 DAA) (c62197_g2_i1_ AT3G19700) (Garcia et al. 2003;Fang et al. 2012), in the early-stages seeds of small-seeded litchi (Table.S3). Expression of CYP78A6 was less in small seeded genotype at 20, 24 and 28 DAA in this study and at all three developmental stages in our previous report (Pathak et al. 2016). We speculate CYP78A6 as a potential target for seed size determination of litchi.

Validation of representative results by quantitative PCR
To validate differential expression pro les analyzed from the digital data obtained by RNA-seq, real time PCR was performed on 4 representative transcripts selected for different levels of expression at 20 DAA and 24 DAA. The selected transcripts were AT1G31150, AT4G38970, AT1G71140 and AT2G38000. Out of these, AT1G31150 and AT4G38970 were up-regulated in bold-seeded litchi genotype at mid-and early (Pathak et al. 2016) seed developmental stages. AT1G71140 (transmembrane transport) and AT2G38000 (chaperone protein) are the most up-regulated genes between small and bold seeded genotypes. The expression patterns obtained in real time assay were in agreement with those analyzed from RNA-seq data (Fig. 6). Hence, the RNA-seq data can be used reliably for relative expression analysis of genes in seed development pathway in litchi.

Discussion
The litchi fruits containing eshy biomass of aril with small seeds being more desirable. A few litchi genotypes have been identi ed that bear fruits with small seed size (Pathak et al. 2014). Development of small seed in litchi is often due to embryo abortion (Yu-Shen Liang 2012) at different developmental stages. The embryo abortion episodes after globular stage (nearby 20 DAA) lead to signi cant reduction in seed size without compromising aril growth, suggesting mid-stage seeds as a crucial phase of seed and fruit development in litchi (Xiang et al. 2001). Application of maleic hydrazide at 2 weeks after anthesis, affects embryo development and results in fruits with shriveled seeds and 10 % more aril production in the fruit (Menzel and Waite 2005). We examined the transcriptome in the mid-stage developing seeds of contrasting genotypes to obtain molecular insight into seed development in litchi ( Fig S3).
The transcriptomic pro les in this work (Fig. S1) as well as the earlier report (Pathak et al. 2016) suggest that seeds of small-seeded litchi are metabolically more active than that of the bold-seeded genotypes. The embryogenesis and defense pathways are compromised in small seeded seeds. Maternal tissue in ovule, under the in uence of environmental factors is reported to, play crucial role in determining the seed fate (Sun et al. 2004). Under stress, the developing embryo gets reduced supply of nutrients, eventually resulting into retardation of seed development (BARNABÁS et al. 2008;Lemoine et al. 2013). The suppressed expression of genes associated with brassinosteroid, jasmonic acid and salicylic acid pathways in small-seeded litchi seeds is suggestive of a cellular environment with higher vulnerability to stress, in fruits with reduced seed size in litchi (Figs. 1, 2 & 3).
Hormones regulate exo-and endogenous signals in cellular and developmental processes, including seed development (Gray 2004). An imbalance in hormone signaling results into altered seed development (Goldberg et al. 1994;Chaudhury and Berger 2001;Pignocchi et al. 2009). For example, brassinosteroid concentration in seed development in uences integument and endosperm development, affecting seed size and shape in Arabidopsis Jiang and Lin 2013). It is known to positively regulate biomass accumulation under stressed (Bajguz and Hayat 2009) and unstressed (Bishop and Yokota 2001) conditions. Lower expression of BSK3 (AT4G00710) and BAK1 (AT4G33430) in litchi indicates suppressed brassinosteriod signaling in small-seeded seeds that may induce necrosis. BSK3 interacts with BRI1 and positively regulates brassinosteroid signal transduction (Tang et al. 2008). Mutants of BRI1 are reported to develop small seed phenotype in Arabidopsis, which is rescued by over expression of BSK3. Null mutants of BAK1 are less sensitive to brassinosteroids in Arabidopsis (Li et al. 2002). Higher expression of NAC100 (AT5G61430) which is a negative regulator of cell expansion in rose (Pei et al. 2013), is in agreement with down-regulation of CYP78A6 (c59232_g1_i1_AT2G46660) and IKU2 (c62197_g2_i1_ AT3G19700) in small seeded litchi. The higher expression of the positive regulators of cell death SAG 29 (AT5G13170) (Seo et al. 2011) and stress induced OCT 3 (AT1G16390) (Küfner and Koch 2008), in small-seeded litchi seed could contribute to seed abortion. Down regulation of APM1 (AT4G33090) at mid stage development suggests that aminopeptidase M1 may play an important role in seed development. Lower expression results in the seed abortion after globular stage, resulting in the reduction in seed size without affecting the fruit size. Jasmonic and salicylic acid hormones are involved in various biotic and abiotic stress responses (Rivas-San Vicente and Plasencia 2011; Wasternack et al. 2013). Jasmonate in young tissues activates vegetative storage proteins and pathogen resistance (Creelman and Mullet 1997). Differential transcription of jasmonic acid or salicylic acid metabolism related genes {TPC1 (AT4G03560, WRKY 70 (AT3G56400), JMT (AT1G19640), DAD1 (AT2G44810), MYB23 (AT5G38830), cystein-rich RLK6 (AT4G38830) and DREB1 (AT4G25480)}, suggests compromised defense response in small-seeded litchi. Differential expression of various TFs during seed development between bold-and small-seeded fruits indicates distinctive pathways involved in deciding the seed size in litchi. The role of the TFs such as WRKY, NAC, bZIP, AP2, and MYB in regulating stress responses in plant tissues has been reported (Wang et al. 2016). These exhibited disparate transcription in the seeds of contrasting litchi genotypes. WOX regulates tissue proliferation during Arabidopsis embryonic development (Wu et al. 2007). It was suppressed in small-seeded litchi seed.
The LEA transcripts, storage proteins and important regulators of seed maturation (FUS5, ABI3 & LEC1) were suppressed in small-seeded litchi; suggesting disturbed seed maturation. The basic body plan of a plant is acquired by the developing embryo which reserves food during embryogenesis, followed by seed maturation (Goldberg et al. 1989). Trinity transcripts homologuous to Arabidopsis thaliana genes for seed development were down-regulated in small-seeded litchi genotype (Fig. 5). These are possible targets for embryo arrest at different developmental stages in litchi. Some of the putative litchi genes regulating defense (galactinol synthase 1 & terpene synthase 21), brassinosteroid signaling (BR-signaling kinase 3) and biomass accumulation (fructose bisphosphate aldolase 2) were differentially expressed in the same pattern at all the studied developmental stages in this study and in our earlier report (Pathak et al. 2016). Further analysis will identify the earliest regulatory genes in these networks.
The transcriptional behavior of brassinosteroid related genes in mid-stage developing seed was similar to that at the early developmental stages (Pathak et al. 2016). The down-regulated transcription of auxin transport related genes in early-stage seeds agrees with the higher expression of senescence related transcripts (Ellis et al. 2005) during mid-stages of seed development in small-seeded litchi. In conclusion, the comprehensive analysis of seed speci c transcriptome shows dynamic changes that lead to arrest of seed development of litchi fruits. This investigation furthers our understanding of the molecular mechanism of seed development with contrasting seed size in litchi. The seed speci c transcriptome repository will serve as a foundation for pursuing functional genomic studies in litchi.