In this study, the total lipid accumulation patterns in the avocado mesocarp and seed tended to increase and fluctuate slightly, respectively, during the fruit developmental period (Fig. 1a). These findings were consistent with the results of previous avocado studies [6, 7]. However, in contrast to these earlier investigations, several studies involving oil palm have confirmed that the total lipid contents in three oil-storing tissues (mesocarp, embryo, and endosperm) exhibit similar increasing trends in developing fruits [11, 12, 25]. Our histological results suggest that the total lipid content in avocado is generally directly proportional to the volume of the lipid droplets in the mesocarp and seed (Fig. 2). Several previous studies proved that the lipid droplets of the oil palm mesocarp and endosperm are relatively large (approximately 5–20 µm in diameter) and occupy most of the cellular volume at harvest, whereas those of the oil palm embryo are smaller (approximately 2 µm in diameter) and are located at the cell periphery at harvest [11, 12, 25]. Similarly, the total lipid contents of three oil-rich tissues of oil palm are consistent with lipid droplet volumes [11, 12, 25].
In this study, the average expression levels of 32 unigenes associated with FA synthesis were 4.56- and 4.00-fold higher than those of 24 unigenes contributing to TAG assembly in the developing mesocarp and seed, respectively (Additional file 4 Table S3). These results suggest that the genes related to FA synthesis may be more important than those related to TAG biosynthesis for the substantial accumulation of oil in avocado fruit tissues. Similar results were reported in previous studies that examined oil-rich tissues (e.g., mesocarp, seed, and tuber) in diverse oil-bearing crops [2, 7, 11, 20, 52]. Our data further confirm that the temporal transcriptional pattern of lipid-related genes is conserved in various oil-rich tissues of oil-bearing crops.
The conversion of pyruvate to fatty acids in plastids involves at least 13 enzymes and/or protein complexes (Fig. 4). Regarding the four subunits of the pyruvate dehydrogenase complex, each one is encoded by one unigene, of which PaPDH-E1α represented an average of 40% of the transcripts of these four unigenes in the developing mesocarp (Additional file 4: Table S3). This was similar to the findings of an earlier avocado mesocarp study [7]. In contrast, an average of 43% of the transcripts of these four unigenes corresponded to PDH-E1β in the developing seed (Additional file 4: Table S3). However, PDH-E1β is the predominant transcript in three oil-rich tissues of oil palm [25]. Two unigenes encoding three subunits of acetyl-CoA carboxylase were expressed in the avocado mesocarp and seed (Fig. 4), with PaACC-BC expression undetectable in the avocado mesocarp [7]. A comparison of the expression of these six acetyl-CoA carboxylase unigenes revealed that the average PaACC-BC-2 transcript levels, which were 259.92 and 55.74 FPKM/stage in the developing avocado mesocarp and seed, respectively, were greater than the transcript levels of the other five unigenes (Additional file 4: Table S3). In contrast to avocado, ACC-Ctβ is more highly transcribed in oil palm than three other unigenes in the mesocarp, embryo, and endosperm [25]. However, in oilseed crops, BCCP2 is predominantly expressed in Ricinus communis, Brassica napus, and Tropaeolum majus seeds [52].
Stearoyl-acyl carrier protein desaturase (SAD) is a precursor for the biosynthesis of polyunsaturated fatty acids and catalyzes the conversion of stearoyl-acyl carrier protein (ACP) to oleoyl-ACP [53]. Seven homologous AtSAD genes were identified in the Arabidopsis genome (AtFAB2, AtDES1, AtDES2, AtDES3, AtDES4, AtDES5, and AtDES6) [54]. In the current study, two homologous PaSAD unigenes, PaFAB2 and PaDES6, were expressed in the developing avocado mesocarp and seed, with PaFAB2 more abundantly transcribed (Fig. 4). The orthologs of FAB2 have often been detected in some oleic acid-rich tissues from nonseed and seed oil crops, in which they are generally more highly expressed than other SAD homologs, implying FAB2 is important for oleic acid synthesis [7, 20, 25, 53]. In contrast to PaFAB2, PaDES6 was mainly transcribed in the developing seed rather than in the developing mesocarp (Fig. 4). Similar studies revealed that SAD homologs usually exhibit tissue-specific expression patterns in oil palm and olive [25, 55]. The EgFAB2 paralogs are expressed in three oil-storing tissues (mesocarp, embryo, and endosperm), whereas EgDES5 is only transcribed in the embryo and endosperm [25]. Seven OeuSAD homologs in the wild olive genome are also differentially expressed among four tissues [55].
The PaSAD and PaACP4 transcript levels were greater than the transcript levels of the other unigenes contributing to the FA synthesis in the plastids of the developing avocado mesocarp and seed, and were generally consistent with the observed oil accumulation, indicative of the key roles for these genes related to avocado oil composition (Fig. 1a, Fig. 4). Similar results were obtained for various oil-storing tissues of other crops [2, 7, 20].
Acyl-ACP thioesterase catalyzes the hydrolysis of the acyl-ACP intermediates during the final step of FA biosynthesis to release free FA [56]. This enzyme has been divided into two families according to distinguishing substrate specificities, namely FatA and FatB, which determine, to some extent, the FA composition of storage lipids in plants [57]. Specifically, FatA exhibits high substrate specificity toward monounsaturated oleoyl-ACP, whereas FatB subfamily members exhibit specificity for palmitoyl- and stearoyl-ACPs [58]. In this study, we detected transcripts for one PaFatA and two PaFatB paralogs in the transcriptomes of the developing mesocarp and seed (Fig. 4); however, all PaFatA and PaFatB genes were expressed in the avocado mesocarp in an earlier investigation [7]. The PaFatA expression level was more than 4-fold (75 DAFB) and 45-fold (215 DAFB) higher in the developing mesocarp than in the developing seed (Additional file 4: Table S3). Similarly, C18:1 contents were more than 6- to 40-times greater in the developing mesocarp than in the developing seed from 110 to 215 DAFB (Additional file 1: Table S1). These results suggest that PaFatA expression is generally positively correlated with C18:1 contents, which is consistent with the results of previous studies [25, 58]. In contrast, the expression levels of two PaFatB paralogs were unrelated to the C16:0 and C18:0 contents in the avocado mesocarp and seed in the present study. The two PaFatB paralogs were more highly expressed in the avocado seed than in the mesocarp during most of the five fruit developmental stages, whereas C16:0 and C18:0 were less abundant in the developing avocado seed than in the developing mesocarp throughout the examined period (Fig. 4; Additional file 1: Table S1). This observation may be explained by the lower substrate specificity of FatB than of FatA [59]. Additionally, the enzymes encoded by two PaFatB paralogs were unable to efficiently hydrolyze 16:0-ACP and 18:0-ACP in the avocado mesocarp and seed, but may have complementary roles that enhance the PaFatA-mediated hydrolysis of 18:1-ACP. The EgFatB in the oil palm embryo and endosperm catalyzes the hydrolysis of 18:1-ACP very efficiently because of a lack of EgFatA [25].
Long-chain acyl-CoA synthetase (LACS) is an enzyme that exports newly synthesized free FAs in the plastid to the ER and converts them to acyl-coenzyme A (acyl-CoA) [60]. In Arabidopsis, LACS is encoded by a small gene family consisting of nine members [61]. In the current study, four homologous LACS genes were expressed in the avocado mesocarp and seed (PaLACS1, PaLACS4, PaLACS8, and PaLACS9) (Fig. 4). Both PaLACS8 and PaLACS9 were relatively abundantly transcribed in the developing avocado mesocarp, with average transcript levels of 119.99 and 87.47 FPKM/stage, respectively, whereas PaLACS4 might mainly contribute to the acyl activation in the avocado seed, with an average transcript level of 105.22 FPKM/stage (Additional file 4: Table S3). Accordingly, ER-associated LACS (PaLACS8) and plastidial LACS (PaLACS9) appear to codetermine the acyl activation from the plastid to the ER in the avocado mesocarp, but ER-associated PaLACS4 is important for the export of FAs between the plastid and ER in the avocado seed. However, a study by Kilaru et al. [7] indicated that LACS9 is most abundantly expressed in the avocado mesocarp. Regarding oil palm, EgLACS9 and two EgLACS4 paralogs are expressed in the oil palm mesocarp, embryo, and endosperm, but the expression of the LACS8 ortholog is undetectable in the oil palm [25]. These three LACS unigenes are expressed in the developing oil palm embryo and endosperm, but EgLACS9 and an EgLACS4 paralog are reportedly also expressed in the developing oil palm mesocarp [25]. Another study determined that CeLACS4 and CeLACS8 account for approximately 70% of the transcripts of seven LACS unigenes in the oil-rich tuber tissue of C. esculentus [20]. In oilseed crops and Arabidopsis, LACS9 orthologs are the most abundantly expressed unigenes in the seed [52, 62]. Mutational studies in Arabidopsis suggested the enzymes encoded by LACS4 and LACS9 exhibit overlapping functions [62]. Both LACS8 and LACS9 may be similarly associated with plastidial fatty acid export in sunflower seeds [58].
The TAG assembly in the ER requires a series of acylations of glycerol-3-phosphate with acyl-CoA and a subsequent dephosphorylation. In this study, we uncovered at least 12 enzymes involved in TAG assembly. For example, GPAT catalyzes the acylation of glycerol-3-phosphate to yield lysophosphatidic acid during the first step of membrane and storage glycerolipid assembly [63]. To date, nine GPAT gene family members have been identified in Arabidopsis. Three GPAT unigenes (one GPAT1 and two GPAT9 paralogs) were expressed in the avocado seed, whereas only two GPAT9 paralogs were transcribed in the avocado mesocarp (Fig. 4; Additional file 4: Table S3). Kilaru et al. [7] also suggested that only GPAT9 is expressed in the avocado mesocarp. Similarly, GPAT9 is reportedly more highly transcribed than the other family members, with GPAT1–8 either unexpressed or expressed at low levels in diverse oil crops [2, 20, 25, 52]. Unlike in other oil crops, the GPAT1 expression level was high in the developing avocado seed, and higher than the expression levels of two GPAT9 paralogs at 215 DAFB (Additional file 4: Table S3).
Among the various unigenes encoding 1-acylglycerol-3-phosphate acyltransferase (LPAAT), LPAAT2 was more highly expressed than LPAAT1 and two LPAAT5 paralogs in the avocado mesocarp and seed (Fig. 4). Similarly, LPAAT2 was identified as the predominantly expressed gene in the avocado mesocarp in a previous study [7]. The LPAAT2 expression levels were similar between the avocado mesocarp and seed; however, LPAAT1 and two LPAAT5 paralogs were transcribed at higher levels in the developing mesocarp than in the developing seed (Additional file 4: Table S3). Two kinds of PP, namely phosphatidate phosphohydrolase (PAH) and lipid phosphate phosphatase (LPP), have functions related to de novo DAG synthesis [64]. In the current study, PaPAH1, two PaPAH2 paralogs, two PaLPP2 paralogs, and two PaLPP3 paralogs were expressed in the avocado mesocarp and seed (Fig. 4). Four of these unigenes, PaPAH1, PaPAH2-1, PaLPP2-1, and PaLPP2-2, were more highly transcribed in the developing mesocarp than in the developing seed, whereas the other three unigenes exhibited the opposite expression pattern (Additional file 4: Table S3). Previous studies implied that LPP rather than PAH is the more probable candidate enzyme responsible for generating DAG [65, 66]. In our study, the average expression levels of four LPP unigenes were 29.40 and 25.23 FPKM/stage in the developing mesocarp and seed, respectively. These expression levels were higher than the average transcript levels of three other LPP unigenes, which were 11.61 and 8.97 FPKM/stage in the developing mesocarp and seed, respectively (Additional file 4: Table S3).
Earlier research confirmed that DGAT, which transfers acyl-CoA to the sn-3 position of diacylglycerol (DAG), is a key rate-limiting enzyme for TAG assembly in the ER [11]. The PaDGAT1 and PaDGAT2 transcript levels were on average 15.71- and 2.32-fold higher in the developing mesocarp than in the developing seed, respectively, which was consistent with the oil content differences observed between the two oil-rich tissues in this study (Fig. 1a; Additional file 4: Table S3). Additionally, PaDGAT2 was transcribed more abundantly than PaDGAT1 by an average of 1.64- and 6.79-fold in the developing mesocarp and seed, respectively (Additional file 4: Table S3), suggesting DGAT2 is the predominant enzyme synthesizing TAG in the avocado mesocarp and seed. Similarly, the DGAT2 transcript level is more than 3-times greater than that of DGAT1 in the C. esculentus tuber [20], but DGAT1 is more highly transcribed in oil seeds and fruits [7, 11, 25].
In addition to the de novo DAG synthesis via the Kennedy pathway, DAG precursors may be transferred from PC to DAG by PDAT to form lyso-PC and TAG [11]. Our data revealed two PaPDAT paralogs and two LPCAT paralogs were expressed in the avocado mesocarp and seed. The two PaPDAT paralogs were more highly transcribed in the avocado seed than in the mesocarp, whereas there were no differences in the transcription of the PaLPCAT paralogs between the avocado mesocarp and seed (Additional file 4: Table S3). Another possible route involves the PC-derived production of DAG as the substrate for TAG synthesis via the reversible activities of two enzymes, phosphatidylcholine:DAG cholinephosphotransferase (PDCT) and/or cytidine-5′-diphosphocholine:DAG cholinephosphotransferase (CPT), with the reversibility of the catalyzed reactions enabling the enrichment of DAG in PUFAs [7]. In our study, the PaCPT transcript level in the developing mesocarp was on average 57.61-fold higher than that in the developing seed. In contrast, the PaPDCT expression level in the developing seed was on average 259-fold higher than that in the developing mesocarp (Additional file 4: Table S3). Similarly, in an earlier investigation, the PaCPT expression levels were on average 6-fold higher than the PaPDCT expression levels in the avocado mesocarp [7]. Additionally, CPT orthologs are more highly expressed than PDCT orthologs in most oil-rich seeds and fruits [20, 25, 52]. Notably, PaCPT and PaPDCT exhibited tissue-specific transcriptional levels, yet PaCPT and PaPDCT do not contribute to major changes in the channeling of FAs from the PC pool for TAG synthesis in the avocado mesocarp and seed because of their relatively low expression levels (< 5 FPKM/stage in the mesocarp and seed) (Additional file 4: Table S3). Collectively, these results suggest that the flux through PC may contribute to the considerable accumulation of TAG in the avocado mesocarp and seed.
The enzyme encoded by FAD2 catalyzes the introduction of a second double bond in oleic acid, which generates polyunsaturated linoleic acid [67]. In our study, two PaFAD2 paralogs were expressed in the avocado mesocarp and seed, with higher transcription levels in the developing mesocarp than in the developing seed, consistent with the differences in the C18:2 contents between these two tissues (Fig. 4; Additional file 1: Table S1). Additionally, the PaFAD2-1 expression patterns in the developing mesocarp and seed were also in accordance with the changes in the C18:2 contents in the developing mesocarp and seed.
Lipid droplets accumulating TAG are encircled by a phospholipid monolayer and abundant amphipathic proteins, and they function as the hub for metabolic processes [68]. An earlier study proved that OBO, STERO, and CALO are TAG storage-related genes [20]. Recently, new lipid droplet-associated genes have been identified, namely LDAP1 and LDAP2 [69, 70]. In our study, PaLDAP2-1 and two PaCALO paralogs, with average transcript levels greater than 120 FPKM/stage, were the main genes involved in the formation of lipid droplets in the mesocarp, whereas four unigenes (PaLDAP2-1, PaOBO, and two PaCALO paralogs), with average transcript levels greater than 120 FPKM/stage, were the primary genes contributing to the formation of lipid droplets in the seed (Fig. 4; Additional file 4: Table S3). In contrast to avocado, earlier transcriptomic studies revealed that the substantial expression of OBO, STERO, and CALO in the oil-bearing seed and tuber tissues of oil crops is likely important for stabilizing TAG in developing seeds and tubers [2, 20, 52].
In the current study, co-expression analyses identified potential TFs regulating the expression of lipid-related genes affecting oil accumulation. Previous investigations proved that WRI1, LEC1, LEC2, FUS3, and ABI3 encode key TFs regulating oil biosynthesis [18]. Both WRI1 and ABI3 were expressed in the avocado mesocarp and seed, respectively, in the current study, but they were not classified as hub genes in the co-expression network. A total of 143 and 141 TFs were uniquely correlated with the oil content in the avocado mesocarp and seed, respectively, implying the oil biosynthesis regulatory network varies between these two tissues. Similarly, different ABA-responsive TF genes (EgNF-YA3, EgNF-YC2, and EgABI5) were identified as regulators of oil accumulation in the oil palm mesocarp instead of the TF genes (WRI1, LEC1, LEC2, FUS3, and ABI3) identified in other oilseed crops, although all of these TF genes belong to the nuclear factor Y (NFY) and basic leucine zipper (bZIP) families [18, 23]. Our analysis of the avocado mesocarp and seed transcriptomes uncovered 10 hub TF genes in the avocado mesocarp and seed, of which PaPBS1-1 and PaRAP2-3 are highly expressed and encode enzymes that interact with other lipid-related TFs in the developing avocado mesocarp and seed, respectively (Fig. 5a, b, Additional file 8: Table S6). We identified PaPBS1-1 as a serine/threonine-protein kinase, and a GO term enrichment analysis indicated that lipid-related TFs are associated with protein serine/threonine kinase activities in the avocado mesocarp and seed. Thus, serine/threonine-protein kinases may affect oil accumulation. Serine/threonine-protein kinase participants in histone modifications, and histone modifications play essential roles in chromatin remodeling and gene expression regulation [71]. Recently, increasing evidence has demonstrated that histone modifications provide a key switch for oil accumulation in A. thaliana [71]. Additionally, PaRAP2-3 was identified as an ethylene-responsive TF, and the subsequent KEGG analysis implied that the lipid-related TFs are involved in plant hormone signal transduction pathways in the avocado mesocarp and seed. Similarly, previous studies suggested that hormone-responsive TFs (such as WRI1) regulate the expression of several fatty acid biosynthetic genes [11, 18, 25]. Moreover, many TF genes, such as PaRAP2-1, PaCOL4-1, PaGRP-2, PaMBF1B, PaEIN3-1, PaCDL1, and PaASIL2, were highly expressed in the developing avocado mesocarp and seed (Additional file 8: Table S6). Future analyses of these TFs with forward and backward genetic methods may clarify their relationships in avocado.
Recent studies confirmed the importance of lncRNAs for regulating gene expression in eukaryotic cells, especially during some key biological processes [46]. However, the number of lncRNAs encoded in genomes as well as their characteristics remain largely unknown [72]. In the current study, 5,299 lncRNAs were identified as expressed in the developing avocado mesocarp and seed, after which 11 trans-acting lncRNAs and 79 cis-acting lncRNAs corresponding to 43 lipid-related unigenes were detected in the developing avocado mesocarp and seed (Additional file 8: Table S7; Additional file 10: Table S8). Our data also revealed that nine cis-acting lncRNAs were specifically expressed in the avocado mesocarp or seed (Additional file 10: Table S8). Further analyses indicated that lncRNAs PB.19359.1, PB.6205.1, PB.4443.3, PB.12340.3, and PB.19743.2 are more highly expressed (average transcript levels > 20 FPKM/stage) in the developing avocado mesocarp than in the developing seed (Additional file 8: Table S7). Previous studies also demonstrated that lncRNAs often exhibit tissue-specific expression patterns [73]. Furthermore, lncRNAs PB.19359.1 and PB.6205.1 were positively co-expressed with PDH (E3), whereas lncRNAs PB.4443.3, PB.12340.3, and PB.19743.2 expression levels were positively related with FAD2-2 expression (Additional file 4: Table S3; Additional file 8: Table S7). Therefore, these five lncRNAs may play substantial roles in accelerating oil accumulation, especially in the developing avocado mesocarp.