Evaluation of oil accumulation and biodiesel property of Lindera glauca fruits among different germplasms and revelation of high oil producing mechanism for developing biodiesel

Background Lindera glauca with rich resource and fruit oil has emerged as novel source of biodiesel in China, but different germplasms show a variation for fruit oil content and FA profile. To develop L. glauca fruit oils as biodiesel, a concurrent exploration of oil content, FA composition, biodiesel yield, fuel property and prediction model construction was conducted on the fruits from 8 plus germplasms to select superior genotype for ideal biodiesel production. Another vital focus was to highlight mechanism that govern the differences in oil content and FA profile of different germplasms. The cross-accessions comparisons associated with oil-synthesized gene transcriptional level and oil accumulative amount led to the identification of potential determinants (enzymes, transporters or transcription factors) and regulatory mechanisms responsible for high-quality oil accumulation. Results To select superior germplasm and unravel regulatory mechanism of high oil production for developing L. glauca fruit oils as biodiesel, 8 plus trees (accession LG01/02/03/04/05/06/07/08) with high-yield fruits were selected to evaluate the differences in oil content, FA profile, biodiesel yield and fuel property, and to construct fuel property prediction model, revealing a variation in the levels of fruit oil (45.12–60.95%), monounsaturated FA (52.43–78.46%) and polyunsaturated FA (17.69–38.73%), and biodiesel yield (80.12–98.71%) across different accessions. Of note, LG06 had a maximum yield of oil (60.95%) and biodiesel (98.71%), and ideal proportions of C18:1 (77.89%), C18:2 (14.16%) and C18:3 (1.55%), indicating that fruit oils from accession LG06 was the most suitable for high-quality biodiesel production. To highlight molecular mechanism that govern such differences in oil content and FA composition of different accessions, the quantitative relationship between oil-synthesized gene transcription and oil accumulative amount were conducted on different accessions to identify some vital determinants (enzymes, transporters or transcription factors) with a model of carbon metabolic regulatory for high-quality oil accumulation by an integrated analysis of our recent transcriptome data and qRT-PCR detection. Our findings may present strategies for developing L. glauca fruit oils as biodiesel feedstock and engineering its oil accumulation. Conclusions This is the first report on the cross-accessions evaluations of L. glauca fruit oils to determine ideal accession for producing ideal biodiesel, and the associations of oil accumulative amount with oil-synthesized gene transcription was performed to identify some crucial determinants (enzymes, transporters or transcription factors) with metabolic regulation model established for governing high oil production. Our finding may provide molecular basis for new strategies of developing biodiesel resource and engineering oil accumulation. Supplementary Information The online version contains supplementary material available at 10.1186/s13068-023-02265-8.

biofuel, and industrial application [16]. In general, carbon source for oil synthesis of oil plants mostly as sucrose from photosynthetic tissues, which in non-green tissues (developing fruits and seeds) is mainly converted to pyruvate (PYR) via glycolysis or to glyceraldehyde 3-phosphate (GAP) via pentose phosphate pathway (PPP) in both cytosol and plastid, as crucial precursor for acetyl-CoA formation destined to de novo FA synthesis [12,17]. It is known that oil biosynthetic process is composed of FA synthesis and triacylglycerol (TAG) assembly involved in several regulatory enzymes [16], of which de novo FA synthesis is initiated by carboxylation of acetyl-CoA to malonyl-CoA in plastid, and transferred to malonyl-ACP as two-carbon unit for FA elongation via a condensation cycle to elongate acyl chain by two-carbon units. The newly synthesized free FAs are exported from plastid into cytosolic acyl-CoA and incorporated into phosphatidylcholine (PC) for desaturation in endoplasmic reticulum (ER). The traditional pathway to assemble FAs into triacylglycerol (TAG) is via three sequential acylation of G3P with acyl-CoA to produce diacylglycerol (DAG) and subsequent TAG (known as Kennedy pathway) [18]. It is also noted that in oil plants, three potential mechanisms have shown to control FA flux via PC for TAG assembly, including esterification of nascent FA to PC, conversion of FA from PC to DAG for TAG assembly by PDAT, and direct utilization of PC-derived DAG as substrate for TAG biosynthesis by DGAT [16,18]. All these revealed one complex metabolic regulatory network for oil synthesis at subcellular level. Yet, the mechanism of how carbon is channeled to FA synthesis and how they are assembled into TAG destined for oil accumulation in oil plants is still unclear. Recently, we performed transcriptome sequencing of developing L. glauca fruits to identify some enzymes and transcription factors relevant for oil biosynthesis [12], which could help us to gain new insight into the mechanism for controlling differences in oil content and FA composition among different germplasms.
The aim of this sequential study was to select ideal germplasm and to unravel regulatory mechanism of high oil production for better development of L. glauca fruit Page 3 of 20 Lin et al. Biotechnology for Biofuels and Bioproducts (2023) 16:14 oil as potential feedstock for biodiesel. To this end , 8  selected plus trees (accessions LG01, LG02, LG03, LG04, LG05, LG06, LG07 and LG08) with high-yield fruits was used as material to assess the differences in oil accumulation (content and FA composition), biodiesel yield and fuel properties of fruit oils from all accessions. Also, a triangular model was constructed for biodiesel property prediction of raw fruit oils from different accessions. Such evaluation could help to select superior accession for developing biodiesel. Another focus of this work was to highlight mechanism that govern differences in fruit oil content and FA profile of different accessions, and to gain valuable information for engineering oil accumulation or molecular-assisted selection. In this regard, based on our recent transcriptomic result of developing L. glauca fruits, the comparative analysis of cross-accessions correlation of gene transcriptional level by qRT-PCR with oil accumulative amount was performed as an important attempt to identify some crucial regulators (transporters, enzymes and transcription factors) involved specifically in carbon source supply (glycolysis, PPP and acetyl-CoA formation) and oil synthetic process (FA biosynthesis and TAG assembly), with the aim of deriving a central metabolic regulation model for high-quality oil accumulation in L. glauca fruits. This study presents for the first time the application of an integrated evaluation of oil content and biodiesel fuel property and analysis of oil accumulative pattern and oil-synthesized gene transcription cross different accessions of L. glauca to highlight high oil producing mechanism for developing biodiesel, which could facilitate the development of L. glauca fruit oils as biodiesel feedstock, and help to gain new insight into molecular regulatory mechanism of high-quality oil production for engineering oil accumulation in oilseed plants.

Variability in fruit oil content and biodiesel yield of 8 selected accessions
To develop L. glauca fruit oils as potential material for biodiesel, it was vital to determine ideal accession with high quantity and quality of fruit oils for gaining maximum economic benefits. Based on our recent studies of different germplasms [9,12], eight plus trees (germplasm accessions LG01, LG02, LG03, LG04, LG05, LG06, LG07 and LG08) with high fruit yield were selected to investigate the variability in fruit oil content and biodiesel yield of different accessions. Here, the fruit oil contents varied from 44.12% (LG01) to 60.95% (LG06), followed by LG03 (46.48%), LG02 (47.39%), LG08 (48.52%), LG04 (49.09%), LG07 (51.81%) and LG05 (53.16%), of which the fruits of LG05, LG06 and LG07 had oil content more than 51.5% (Fig. 1a). This allowed us to explore the differences in biodiesel yields from fruit oils across all accessions. The biodiesel yield varied among different accessions, ranging from 85.12% (LG01) to 98.71% (LG06) with an average value of 92.17% (Fig. 1b), of which biodiesel yield of LG05 (97.26%), LG06 (98.71%) and LG07 (96.81%) was in the standard of EN 14214 (96.5%). These revealed a difference in fruit oil or biodiesel yield across different accessions, and three high oil-bearing accessions (LG05/06/07) could be valuable as source for developing biodiesel. Yet, the fact of no correlation of oil content with biodiesel yield among different accessions (Fig. 1a, b) indicated that the biodiesel yield may depend on the trans-esterification effectiveness, but not on the oil content of the fruits.
These outcomes, together with the highest yield of oil and biodiesel for accession LG06, revealed that its fruit oil was the most suitable for biodiesel production.

Evaluation of biodiesel fuel properties of fruit oils from different accessions
Effective evaluation biodiesel fuel properties of iodine value (IV), cetane number (CN), oxidation stability (OS), cloud point (CP), cold filter plugging point (CFPP), density (D), kinematic viscosity (KV), degree of unsaturation (DU) and chain length saturated factor (LCSF) would provide a vital basis for exploiting biodiesel plants. Given a notable variation on oil content and total proportion of C18:1 and C18:2 in the fruits of different accessions LG01 0.29 ± 0.03 0. 25 Table 2), implying that the biodiesels from fruit oils of all accessions had good ignition quality. Also, IV is a crucial index for assessing the unsaturated degree (DU) of FA and OS of biodiesel [20]. The IV values of all accessions varied from 102.15 to 115.09 ( Table 2), all of which was less than the specified maximum limit (120) of EN 14214 standard.
KV is one key parameter for defining flow capability of biodiesel, used for estimating spray penetration and atomization of fuel [20]. The range of KV value (4.31-4.68 kg/m 3 ) of biodiesels from all accessions could meet with the standards of ASTM D6751 (1.9 < KV < 6.0), EN 14214 (3.5 < KV < 5.0) and GB/T20828 (1.9 < KV < 6.0) ( Table 2), revealing a good flow or spray capability of biodiesel fuel from L. glauca fruit oils of all accessions. Also, density (D) is one of vital fuel properties for assessing fuel transferred quantity by injection system for combustion [21]. The D value ranged from 871.54 to 887.14 mm/s of the biodiesels (Table 2), which satisfied with the standards of EN 14214 (860 < D < 900) and GB/T20828 (820 < D < 900), and thus concluded that the biodiesels from L. glauca fruit oils of all accessions possessed ideal combustion efficiency.
OS, as a crucial parameter, involves in the level and stability of biodiesel reaction with air [20]. In this study, the OS values of biodiesels from fruit oils of all accessions varied from 2.59 to 3.34 h ( Table 2), all of which did not reach the minimum limit (6 h) specified in the standards of EN 14214 and GB/T 20828, but only accession LG06 (3.34 h) could meet the ASTM D6751 standard (OS > 3.0 h), which was likely attributed to low content of PUFA (17.69%) in the fruit oils of LG06 compared with other accessions (29.01-38.73%) (Fig. 1c).
CFPP and CP, two vital low-temperature parameters, are used to describe the maximum of filterability, but not limited by the standards of US and European. The CFPP valve ranged from − 12.59 to − 9.74 °C for biodiesels across all accessions (Table 2), less than the maximum limit (0 °C) of Germany standard (DIN V51606) in summer, and this value for most accessions (except LG01/03) was lower than the minimum limit (− 10.0 °C) for spring and autumn, implying a good cold flow performance of biodiesel. Also, CP is one indicator for controlling fuel at low temperature [21], and its value ranged from − 9.91 to − 7.11 °C across all accessions (Table 2), which could meet ASTM D6751 standard (− 12 °C < CP < − 3 °C). This implied a better cold flow property of biodiesel from all accessions, especially LG06 with the minimum value of CFPP (− 12.59 °C) and CP (− 9.91 °C), which may be mostly attributed to high content of unsaturated FA (Fig. 1c) and small amount of length chain saturated factor (LCSF) ( Table 2).
DU and LCSF are known as two parameters based on the type of FAs. Of note, the values of CN, IV and OS are determined greatly by the DU [15], but both CFPP and CP mostly depend on LCSF [15,19] (Table 2). It seems therefore that both DU and LCSF may be as critical indicators for evaluation of biodiesel fuel properties of fruit oils of L. glauca from different accessions.
Another was concerned about the content of C18:3 and the FAs with four double bonds in the FAMEs. Here, low amount (1.55-4.86%) of C18:3 in fruit oils of all accessions and no four double-bond FAs (C18:4 and C20:4) in the FAMEs (Table 2) all satisfied the EN14214-2008 specification (< 12% and 1%, respectively).

Construction of prediction model for biodiesel properties of raw fruit oils from different accessions
Determining biodiesel fuel property is very difficult because it takes a lot of time and cost and thus several attempts have been made to use effective methods to calculate or predict fuel property of biodiesel [15,22]. Recently, triangular predict model was constructed to effectively evaluate fuel properties of biodiesel based on FA compositions of raw oils from developing P. sibirica and L. glauca fruits [7,12]. Such prediction as an attempt was performed here. To this end, the percentages of SFA, MUFA and PUFA in L. glauca fruit oils of all accessions ( Fig. 1c) were used as three angular points to establish a triangular graph (Fig. 2), in which one specific region (marked in gray) was delineated to predict biodiesel fuel property for fruit oils from all accessions, taking into account the satisfactions of key fuel properties (CN, IN, CFPP, OS and CP). All accessions were presented in the gray area of our constructed triangular graph (Fig. 2), of which accession LG06 was located at the far end of PMFA angular point (lower left vertex) and SFA angular point (lower right vertex), indicating that the fruit oils from all accessions (especially LG06) may be as potential material for biodiesel, coincided with our evaluated results of    fuel properties for different accessions (Table 2). Thus, triangular predict model for fuel properties by FA compositions of raw oils could provide simple and effective selection of ideal plant resource biodiesel. Together, the accession (LG05/06/07) with high fruit oil content and biodiesel yield, and ideal fuel property, could be suitable for high-quality biodiesel production. Yet, another question was the mechanism that governed such difference in oil content and FA composition of L. glauca fruits across all accessions ( Fig. 1a and Table 1). Effective increases in oil content and ideal FA profile require to unravel complex metabolic regulation network. Thus, in the following, our work focused on the identification of key regulators (enzymes, transcription factors and transporters) specific for carbon source supply and oil synthetic process (FA synthesis and TAG assembly) by the comparative analysis of cross-accessions association of fruit oil content with gene transcript level. Such exploration could gain a better insight into molecular regulation mechanism governing high oil production and provide new biotechnological targets to improve oil yield.

Transcript differences of carbon allocation specific for acetyl-CoA generation in fruits of different accessions
In oil plants, acetyl-CoA, one vital precursor for FA synthesis, is mainly derived from PYR via glycolysis or GAP via PPP in both cytosol and plastid by a series of regulatory enzymes [17,23,24]. Recently, our transcriptomic assay showed that oil accumulation of developing L. glauca fruits was regulated by differential transcripts of enzymes between plastidic and cytosolic glycolysis [12], including ATP-dependent phosphofructokinase (PFK), hexokinase (HXK), fructose-bisphosphate aldolase (FBA), phosphoglycerate kinase (PGK), triosephosphate isomerase (TPI), enolase (ENO), GAP dehydrogenase (GAPC) and pyruvate kinase (PK). To determine the relative flux of PYR from plastid or cytosol glycolysis destined to oil production in the fruits of different accessions, transcript differences of all enzymes of two glycolytic pathways were analyzed in the fruits of all accessions by qRT-PCR. The transcript levels of plastid glycolytic enzymes (HXK, PFK, FBA, TPI, GPI, GAPC, PGK, PGM, ENO1 and PK) increased with the increasing amount of fruit oils across all accessions, of which the richest transcripts were all noted for accession LG06 with the highest oil content (Figs. 1a, 3a). Yet, all cytosolic isoforms showed less transcript across all accessions (Fig. 3b). All these revealed a major role of plastid glycolysis in supply PYR for FA synthesis in the fruits of all accessions. Another note with regard to glycolysis was about transporter in interchange of glycolytic intermediate between cytosol and plastid [7,25]. Given that the orthologs for triose phosphate transporter (TPT), G6P transporter (GPT1/2), phosphoenolpyruvate transporter (PPT1/2), xylulose 5-phosphate transporter (XPT), glycolipid transporter (GLT), and bile acid/sodium symporter (BASS2) was marked in developing L. glauca fruits by recent transcriptome analysis [12], it was crucial to explore which of them may contribute to transport glycolytic metabolite from cytosol into plastid for fruit FA synthesis of different accessions. Only GPT1, PPT1 and BASS2 showed high transcript and a high correlation to the amount of fruit oils across all accessions (Figs. 1a, 3c), indicating that GPT1/PPT1/BASS2 may contribute to allocate cytosolic glycolytic metabolites (G6P, PEP or PYR) into plastid for FA synthesis destined to the eventual oil production in L. glauca fruits of all accessions.
Aside from glycolysis, FA biosynthesis was fed by GAP via PPP [7,26]. Our recent annotation of a complete plastidic PPP with differential transcripts in developing L. glauca fruits by transcriptome assay [12], allowed to address whether the fruit oil contents of different accessions were correlated to the increasing number of their transcripts. All enzymes of plastidic PPP, including 6-phosphogluconate dehydrogenase (6PGDH), transaldolase (TA), 6-phosphogluconolactonase (PGLS), G6P dehydrogenase (G6PDH), ribulose-5-phosphate (RP) epimerase (RPE), RP isomerase (RPI) and transketolase (TK), were detected with the abundantly coordinated transcripts of all accessions by qRT-PCR (Fig. 3d), and their transcript levels were associated with the amount of fruit oils across all accessions (Figs. 1a, 3d), pointing  to a role of plastidic PPP in provision of GAP for FA synthesis.
Another was concerned about acetyl-CoA generation from PYR via acetyl-CoA synthetase (ACS), ATPcitrate lyase (ACL), or PYR dehydrogenase complex (PDC) [24], all of which were detected with differential transcripts in developing L. glauca fruits by our recent transcriptome assay. Hence, we performed the crossaccessions comparisons of transcripts to explore which of them may specifically devote to allocate PYR flux for acetyl-CoA formation destined for FA synthesis in all accession fruits. High transcript of plastidic PDC was closely consistent with the increase of fruit oil content across the accessions (Figs. 1a, 4), while transcript level of cytosolic ACLB subunits (ACLB-1/-2) showed no notable up-regulation, and less transcript was detected for mitochondrial PDC, cytosolic ACLA and plastidic ACS (Fig. 4), implying that plastid PDC may mostly contribute to acetyl-CoA formation for FA synthesis in all accession fruits. Also of note was the roles of cytosolic ACLB and mitochondrial PYR carrier (MPC), citrate synthase (CS) and dicarboxylate/tricarboxylate carrier (DTC) in cytosolic acetyl-CoA generation for FA elongation [7]. Here, low transcript of mitochondrial MPC, CS4 and DTC was similar to that of cytosolic ACLB and mitochondrial PDC (Figs. 3c, 4), corresponding to small amount of C20:0 and C20:1 in fruit oils (Table 1), pointing to a role of them in providing cytosolic acetyl-CoA for FA elongation.

Transcript differences of enzymes and transporters for FA and TAG synthesis in fruits of different accessions
Increasing FA and TAG biosynthesis and storage oil yield would expand economic value for oil plants [7]. Given differential transcripts of oil synthesis enzymes noted in developing L. glauca fruits by our recent transcriptome assay [12], it was necessary to explore quantitative relationship between the transcript levels of oil-synthesized genes and the amount of fruit oils from all accessions. By qRT-PCR, the abundantly coordinated transcripts were detected for FA synthetic enzymes, including acetyl-CoA carboxylase (ACC), malonyl-CoA-ACP transferase (MAT), fatty acyl-ACP thioesterase A/B (FATA/B), 3-ketoacyl ACP synthase I/II/III (KAS I/II/ III), 3-ketoacyl ACP reductase (KAR), hydroxyacyl-ACP dehydrase (HAD), enoyl-ACP reductase (EAR) and 18:0-ACP desaturase 6 (SAD6) in the fruits of all accessions (Fig. 5a), which was the case for the enzymes for de novo TAG assembly [acyl-CoA:G3P acyltransferase 9 (GPAT9), acyl-CoA:LPA acyltransferase 2 (LPAAT2), acyl-CoA:DAG acyltransferase 1 (DGAT1) and PA phosphatase 2 (PAP2)] (Fig. 5b). A strong correlation of their transcript levels with fruit oil accumulation across all accessions (Figs. 1a, 5) emphasized their importance for FA synthesis and TAG assembly. Also of note was the role of FA exporter (FAX) in FA import into ER for TAG synthesis and long-chain acyl-CoA synthetase (LACS) in activating free FA to produce acyl-CoA pool [7,[27][28][29]. High transcript of FAX1/2 and LACS4 (Figs. 3c, 5b) was comparable to that of FA-synthesized enzymes (Figs. 5a), and showed a pattern that correlated with fruit oil content of all accessions (Fig. 1a), and thus referred that both FAX1/2 and LACS4 may be pivotal for FA export from plastid into ER for acyl-CoA pool generation destined to TAG assembly.
ER-located FAD2 and 3 is known for desaturation of C18:1 to produce C18:2 and then C18:3, respectively. Here, FAD2 showed a relative high transcript, but less transcript was marked for FAD3 (Fig. 5b), both of which were matched the level of C18:2 (high) and C18:3 (less) in fruit oils of all accessions (Table 1). High transcript similar to FAD2 was noted for acyl-CoA: lysophosphatidylcholine acyltransferase 2 (LPCAT2) (Fig. 5b), involved in the transfer of C18:1 into PC for desaturation and release of PUFA into acyl-CoA pool for TAG assembly [30], all of which displayed a strong correlation to the amount of C18:2 in fruit oils across all accessions (Fig. 1a), implying a specific contribution of FAD2 or LPCAT2 to PUFA production. Also, lower transcript was identified for PC: DAG cholinephosphotransferase (PDCT) and CDP-choline: DAG cholinephosphotransferase (CPT) (Fig. 5b), two enzymes for interconversion of DAG and PC to produce PUFA-rich TAG [30,31], indicating its unimportance for TAG synthesis. Another was concerned for DAG acyltransferase 1 (PDAT1) for the last assembly of TAG [32], and its low transcript showed no substantial difference across different accessions (Fig. 5b), reflecting a very limited contribution of PDAT1 to TAG synthesis in the fruits of all accessions.

Transcript differences in transcription factors for regulating fruit oil accumulation of different accessions
Transcription factors (TFs) have shown to regulate oil biosynthesis of oil plants. Recently, differential transcripts were annotated for the TFs (ABI3, LEC1, WRI1, FUS3, AP2, GL2, HSI2, TT2, PKL and VAL2) in developing L. glauca fruits by our transcriptome analysis [12], which allowed us to explore the possible association of their transcript levels with accumulative amount of fruit oils across all accessions. Here, the transcripts of ABI3, LEC1 and WRI1 increased with the increase in fruit oil content of all accessions, while the transcripts of AP2 and GL2 showed a downtrend with the increase in fruit oil content, and less transcript was detected for TT2, FUS3, HSI2, VAL2 and PKL across all accessions (Figs. 1a, 6), reflecting a complex transcriptional regulation for fruit oil synthesis of different accessions. Also, we performed the protein interaction analysis for the above TFs and oil-synthesis enzymes as an attempt to highlight TF-mediated regulation mechanism for fruit oil biosynthesis, and found that the enzymes for oil accumulation (carbon source supply, FA synthesis and TAG assembly) were highly associated with WRI1, and LEC1 showed a strong interaction with WRI1 (Fig. 7), and thus considered that LEC1 and its targeted WRI1, located in the center position of interaction network, may contribute to regulate transcriptional expressions of enzymes relevant for fruit oil accumulation of all accessions.
It was estimated that the total planting area of L. glauca was about 14,400 ha in China, and the average production of ripened fruits was about 11.5 ton /ha (equated to 5.0 kg of fruits per plant yield) [11,12,43]. Based on energy conservation principle and law of material invariance (including planting, fruit and biodiesel transport, oil and biodiesel production, and biodiesel combustion) for P. chinensis [44], the annual amount of biodiesel from L. glauca fruit oils was about 21,600 tons in China, and the resulting annual output value created by biodiesel was 178.84 million Yuan. Thus, it was concluded that our identified 3 accessions (LG05/LG06 /LG07) may be of interest for potential industrial application, specifically LG06 with the maximum yield of oil and biodiesel, and superior fuel properties.

Strong funneling of carbon flux toward plastid acetyl-CoA for high FA production in fruits of all accessions
Several studies have shown that oil accumulation and FA composition vary considerably among different plants, provenances, tissues or developing stages [6,7,16,17,23,24,30,33,[45][46][47], but the mechanism for controlling these differences still remains enigmatic. Thus, it is vital to unravel regulators for high-quality oil accumulation.
In oil plants, the source of PYR for high oil production is largely controlled by glycolysis in cytosol or plastid [12,17,23,24,[46][47][48][49][50]. Here, transcript levels of glycolytic enzymes were much higher in plastid than in cytosol of all accessions (Fig. 3a, b), and notably, transcript abundance of plastid glycolytic enzymes was positively correlated with the increase of fruit oil content across all accessions (Figs. 1a, 3a), revealing a critical role of plastid glycolysis in providing PYR for FA synthesis, as also noted for several oil plants [7,17,23,46,50,51]. In support of this finding, plastidic GPT1 (G6P transporter), PPT1 (PEP transporter) and BASS2 (PYR carrier) were transcriptionally coordinated with the increase of oil content in all accession fruits (Figs. 1a, 3c), implying a role of transporter-mediated import of G6P, PEP or PYR from cytosol into plastid glycolysis for FA synthesis in L. glauca fruits of all accessions, as could be evidenced by the fact that PPT1 or BASS2 overexpression promoted Arabidopsis seed oil increase [52,53]. Yet, PPT1 transcript was on average 1.1-and 1.3-fold higher than that of GPT1 and BASS2, respectively (Fig. 3c), implying a major glycolytic carbon flux into plastid from cytosol at the level of PEP in L. glauca fruits, similar to other oil plants [17,23,26]. All these also clearly highlighted a strong coordinated funneling of glycolytic intermediate toward PYR in plastid, as could be further reflected by less transcript for mitochondrial MPC (PYR carrier) across all accessions (Fig. 3c). Aside from glycolysis, PPP as alternative pathway has shown to supply GAP for FA synthesis in oil plants [26,51,54]. Such pathway could be shown by our finding that the abundantly coordinated transcripts of plastid PPP and GPT1 displayed a close relation to the amount of fruit oils of all accessions (Figs. 1a, 3c, d), implying a portion of GPT1-mediated G6P import into plastid destined for FA synthesis. Yet, the fact that transcript level of plastid glycolytic enzymes was on average twofold higher than that of plastid PPP in all accessions (Fig. 3a, d) revealed that the necessity of carbon source for de novo FA synthesis in L. glauca fruits was mainly derived from plastidic glycolysis. Acetyl-CoA, one vital precursor for de novo FA synthesis in plastid and its elongation in cytosol, was mostly derived from PYR via four alternative enzymes of PDC, ACL, ACS or CA [23,55], all of which were shown with differential transcript in developing L. glauca fruits by our recent transcriptomic analysis [12]. Here, only plastidic PDC and cytosolic ACLB subunits (ACLB-1/-2) were highly expressed in all accessions (Fig. 4), while transcript level of plastidic PDC was on average fourfold higher than that of cytosolic ACLB across all accessions, and showed a positive association with the increased fruit oil of all accessions (Fig. 1a), emphasizing that the availability of acetyl-CoA for FA synthesis in L. glauca fruits of all accessions was derived from glycolytic PYR via the action of PDC in plastid, as also noted for other oilseed plants [24,46]. Also noteworthy was the roles of cytosolic ACL and mitochondrial PDC in producing cytosolic acetyl-CoA for FA elongation, in which cytosolic PYR was imported into mitochondria by MPC (PYR transporter) for acetyl-CoA formation via mitochondrial PDC, as substrate for citrate formation by CS and then export by DTC transporter for cytosolic acetyl-CoA formation via ACL cleavage [7,12,23,56]. Our observation that cytosolic ACLB-1/-2, and mitochondrial MPC, PDC and CS4 all showed low transcript (Figs. 3c, 4), coincided with small amount of C20:0 and C20:1 in fruit oils ( Table 1), indicated that the production of cytosolic acetyl-CoA for FA elongation was coordinated at the transcript level.
Together, strongly increased plastid carbon supply via effective transport and glycolysis, and high action of plastid PDC, may provide great funneling of acetyl-CoA for high FA synthesis in L. glauca fruits of all accessions.

FA flux channeled into de novo TAG assembly for high-quality oil accumulation in fruits of all accessions
The PUFA (C18:2 and C18:3) content is a key factor affecting the quality of plant oils for human health, biofuel and other purposes [31]. In oil plants, C18:1, which was produced from C18:0 by SAD6 in plastid, is desaturated to C18:2 and then C18:3 by ER-localized FAD2 and FAD3, respectively. Here, transcript level of SAD6 in the fruits across all accessions was on average 1.3-and 3.5fold higher than that of FAD2 and FAD3, respectively, which closely matched the amount of C18:1, C18:2 and C18:3 in fruit oils of all accessions (Table 1 and Fig. 5b), and thus considered that C18:1 as the richest component in fruit oils may be mostly attributed to their differential coordinate transcripts. This was in line with previous results in different tissues of oil palm [57] and developing seeds of P. sibirica [58], and also be supported by high amount of C18:1 in fad2/3 double mutant soybean [59].
The carbon chain length of FA and relative proportion of SFA and MUFA in the oils may be determined by de novo FA synthesis from acetyl-CoA in plastid [16]. Given a positive correlation between abundant transcript of FAsynthesized enzymes (ACC, MAT, KAR, HAD, KAS I/ II/ III, EAR, FATA, FATB and SAD6) and accumulative amount of fruit oils across all accessions (Table 1 and Fig. 5a), it seems certain that the differences in fruit oil contents of different accessions may highly depend on the efficiency of FA synthesis in plastid, as also noted for oil palm [17]. Another difference across all accessions was transcript of ER-localized LACS4 crucial for activation of free FA to generate acyl-CoA pool for TAG assembly [16], and notably, its high transcript pattern (Fig. 5b) was closely related to fruit oil accumulative amount across all accessions (Table 1), indicating its importance in controlling plastid FA flux into acyl-CoA pool for TAG synthesis, as was evidenced by the increase of oil content in yeast mutant by overexpression of B. napus LACS4 [29]. Considering that FAX1/2 as plastid transporter for export FA into ER responsible for TAG synthesis [7,27,28] was highly expressed in all accession fruits (Fig. 3c), it was believed that the abundantly coordinated transcripts of FA-synthesized enzymes, LACS4 and FAX1/2 may provide a strong channeling of plastid FA flux into ER to generate acyl-CoA pool destined for TAG synthesis.
Plant oil synthesis involves a complex metabolic network for TAG assembly with distinct FA profiles, and de novo assembly of TAG from G3P and acyl-CoA can produce oils with rich C18:1 [18,30,60]. Yet, how acyl flux is channeled to produce ideal oil composition in oil plants remains largely unclear. In this work, the accumulative amount of fruit oils with rich C18:1 was positively associated with high and increasing number of transcripts of crucial enzymes (GPAT9, LPAAT2 and PAP2) for de novo DAG synthesis via Kennedy pathway across all accessions (Table 1 and Fig. 5a), reflecting that direct utilization of de novo DAG for TAG synthesis may be as dominant route for L. glauca fruit oils with rich C18:1 of all accessions, as also noted in other oil plants [30,61]. In general, C18:1 from plastid can be used directly for de novo DAG assembly by GPAT and LPAAT via Kennedy pathway, or incorporated into PC by LPCAT for desaturation by FAD2/3 [31]. Given that transcript level of LPCAT2 was on average 1.5-fold lower than that both GPAT9 and LPAAT2 for the first two acylation of TAG  16:14 assembly in fruits of all accessions (Fig. 5), integrated with the fact of C18:1 as the richest component in fruit oils (Table 1), it seems certain that a major flux of C18:1 from acyl-CoA pool may be directly channeled into de novo DAG synthesis, and thus a small flux of C18:1 was used for desaturation. This conclusion could support earlier idea that LPCAT2 may transfer PC-derived PUFA directly into acyl-CoA pool for TAG synthesis via Kennedy pathway [62], and also was evidenced by the fact that lpcat1/lpcat2 mutant reduced PUFA content in Arabidopsis seed oils [60], but gpat9 mutant increased PUFA content and decreased seed oil content [63]. It was also noted that PUFA-rich TAG may be mostly attributed to PDCT or CPT [30,31]. The fact that transcript levels of PDCT and CPT were on average 6.5-fold below that of both LPAAT and GPAT (Fig. 5), and showed no relation to variations for PUFA amount in fruit oils of all accessions ( Fig. 5 and Table1), indicated a negligible role of PDCT or CPT in PUFA production in fruit oils of all accessions, also noted in developing P. sibirica seeds [8]. This was in line with previous notion of low PUFA production in oil palm caused by less transcript of PDCT and CPT [57], but contrasted with a major contribution of PC-derived DAG to TAG synthesis in developing seeds of Arabidopsis and soybean [31,60,64]. Also of note was DGAT and PDAT vital for the final acylation of TAG synthesis [32], but it remains unknown what extent both enzymes devote to the last TAG assembly in oil plants. Here, PDAT1, similar to both PDCT and CPT, showed less transcript and no difference across all accessions (Fig. 5b), it appeared that PDAT1 was not a key contributor for TAG synthesis in L. glauca fruits of all accessions. This corresponded to earlier results in oil plants from transcriptomic or metabolic analysis [17,57,65], and also confirmed by no effect of PDAT1 mutant on seed oil content and FA profile [32]. In support of this conclusion, transcript level of DGAT1 was on average fivefold higher than that of PDAT1 across all accessions (Fig. 5b), and positively correlated with the amount of fruit oils (Table 1), revealing that DGAT1 may be a crucial regulator for oil synthesis in L. glauca fruits, coincided with metabolic and transcriptomic analyses of some oil plants [17,24,32,46,66], and also evidenced by the overexpression or mutant result of DGAT1 in several oil plants [67][68][69][70][71][72][73][74].
Overall, FAX1/2-mediated plastidic FA export, LACS4regulated effective generation of acyl-CoA pool in ER, and LPCAT2-mediated C18:1 flux into PC for desaturation and PUFA back into acyl-CoA pool, integrated with the abundantly coordinated transcripts for the enzymes of FA synthesis and de novo TAG assembly via Kennedy pathway, was crucial for the eventual high-quality oil accumulation in L. glauca fruits of all accessions (Fig. 8).

LEC1/WRI1-mediated regulatory network for oil accumulation in L. glauca fruits of all accessions
Plant FA and TAG biosynthesis, involving a series of gene expression, is highly regulated by several TFs [75][76][77][78][79][80][81][82]. Our findings that high transcript patterns of ABI3, LEC1 and WRI1 were positively correlated with the amount of fruit oils across all accessions, but the down-regulated profiles of GL2 and AP2 showed an inverse relationship with oil content (Figs. 1a, 6), indicated that ABI3, WRI1 and LEC1 may act as positive regulators for gene expression relevant for fruit oil synthesis of all accessions, but AP2 and GL2 may play negative regulatory factors, as was noted for developing P. sibirica seeds [8], and also compatible with our Fig. 8 Characterization of central metabolic model for controlling carbon and FA flux for high oil synthesis in L. glauca fruits. The constructed metabolism regulatory network of carbon flux allocation for high-quality oil production in L. glauca fruits represents the most relevant pathways (glycolysis, PPP, acetyl-CoA formation, de novo FA synthesis and its elongation, plastidic FA export and desaturation, and TAG assembly) and includes all identified regulators (enzymes, transcription factors and transporters) with strong correlation with carbon flux partitioning into FA synthesis and TAG assembly, all of which are characterized by the comparative analysis of cross-accessions association of gene transcription level (qRT-PCR detection) with oil accumulative amount in fruits across all accessions, together with our recent transcriptome sequencing data from developing L. glauca fruits. All enzymes presented here are shown in blue. The values generated from qRT-PCR indicate relative transcript levels in the fruits from high-oil accession LG06 (green values) and low-oil accession LG01 (red values), all of which are calculated by the standardized set of expression level for inner reference gene to 1.00. The background color distinguishes different subcellular locations or pathways for FA and TAG synthesis as follows: gray and bright blue signify cytosolic and plastidic glycolysis, respectively; light red refers to plastidic PPP; light green signifies de novo FA synthesis; light blue signifies TCA cycle in mitochondria; yellow signifies TAG assembly in ER. Black arrows represent all reactions for major carbon allocation into FA synthesis and TAG assembly, and brown arrows represent minor carbon allocation for FA and TAG synthesis. recent transcriptome result of developing L. glauca fruits [12]. Of note, WRI1, has been shown to control the expressions of its targeted genes involved in glycolysis, and FA and TAG biosynthesis during oilseed development [82][83][84]. This could be directly shown by the fact that the enzymes for carbon allocation (plastid glycolysis and acetyl-CoA formation), de novo FA synthesis and TAG assembly were transcriptionally coordinated with WRI1 across all accession fruits (Figs. 5, 6), which in turn was closely related to the fruit oil content (Fig. 1a). Such strong relationship implied that these genes may be as the targets were activated by WRI1, in which carbon flux could be efficiently channeled into oil synthetic pathway for oil accumulation  Biofuels and Bioproducts (2023) 16:14 in the fruits of all accessions by collaborative manner, similar to the results of other oilseed plants [57,[84][85][86], and also evidenced by the increase in FA synthesis and oil accumulation by ectopic expression of WRI1 from many oilseed plants [75-77, 80, 81, 84-86]. WRI1 is known to be up-regulated by LEC1/2, ABI3 or FUS3 [85]. A similar profile of high transcript for ABI3, LEC1 and WRI1 in all accessions (Fig. 5) revealed WRI1 as a target gene of LEC1 or ABI3, as was also reported in A. thaliana and P. sibirica seeds [24,87], but contrasted with the idea that WRI1 function might be independent of upstream TFs in controlling oil biosynthesis in oil palm [17]. LEC1 has been identified as a central regulator of oil synthesis in developing seeds [83,88], and overexpression of LEC1 increased oil content by upregulating several genes relevant for FA and oil synthesis in oilseed plants [78,83,89]. Overall, the abundantly coordinated transcripts were identified for the TFs (LEC1 and WRI1) and the enzymes of carbon flux allocation (plastidic glycolysis and acetyl-CoA generation) and FA synthesis as well as TAG assembly (Figs. 4-6), showing a positive correlation with the accumulative amount of fruits oils across all accessions (Table 1), and thus revealed that LEC1/WRI1-mediated transcription regulatory network may play a major role in FA synthesis and oil accumulation in L. glauca fruits of different accessions (Fig. 7).

Conclusions
In this work, the cross-accessions comparisons were conducted on oil content, FA profile, biodiesel yield, fuel property and prediction model construction to determine superior accession (LG06) as promising feedstock for ideal biodiesel production. An integrated assay for the associations between oil-synthesized gene transcription level and oil accumulative amount was used to identify some key regulators (enzymes, transcription factors and transporters) required for highquality oil accumulation of L. glauca fruits, including carbon allocation (plastidic glycolysis and PPP), metabolite transporter (plastid GPT1/PPT1/BASS2/FAX1/2), acetyl-CoA formation (plastidic PDH), transcription factor (LEC1/WRI1), FA synthesis (ACC, MAT, FATA/B, KAS I/II/III, KAR, HAD, EAR and SAD6), and TAG assembly (LACS4, LPCAT2, FAD2/3, GPAT9, LPAAT2, PAP and DGAT1). Such exploration has led to central metabolic model established for controlling carbon and FA flux for high oil synthesis in L. glauca fruits (Fig. 8). Together, our findings could facilitate development of L. glauca fruit oils as biodiesel feedstock, and provide new insight into regulatory mechanism of high oil production destined for further metabolic engineering of oil accumulation in L. glauca and other oil plants.

Oil extraction and trans-esterification
About 50 g of fresh fruit (3 samples per accession) were crushed into power, and then the oils were extracted with petroleum ether using Soxhlet apparatus at 45-50 °C [12]. After extraction for 6-8 h, the oil was separated from organic mixture by rotary evaporator (LABORTA 4000-efficient, Germany), and then dried at 105 °C in the ventilated drying oven. The content of extracted oil from each accession fruit was calculated as the difference between the weights of fruit sample before and after extraction, and expressed as the percentage of the extracted oil weight to fresh fruit weight (%, g/g). To analyze FA compositions, the fruit oils from each accession were trans-esterified as previously described [90]. All analysis was conducted on triplicate.

Analyses of FAMEs and biodiesel yield
The FA methyl esters (FAMEs) obtained from each accession fruit was used to determine FA compositions using Agilent 6890 (California, USA) gas chromatograph equipped with flame ionization detector (GC-FID) [12]. The HP-INNOWax capillary column (inner diameter 0.32 mm, film thickness 0.5 μm, split 1:20) was used, and the temperature was programmed at 60 °C, with a rise of 4 °C min −1 to 220 °C and heated to 240 °C for 10 min. The carrier gas was helium with a flow rate of 1.0 mL min −1 .
The peaks of FAMEs were identified by comparing their retention time with that of the known standards, and peak integration was performed by applying HP3398A software. Each FAME assay was performed in triplicate, and the data were present as the mean. The biodiesel yield was calculated by the previous method [7,12], where the yield was expressed as the percentage (%, g/g) of the obtained total amount of FAMEs (g) to the used amount of raw oils (g).

Quality of biodiesel fuel property
The fuel properties (IV, CN, CP, CFPP, OS, D and KV) were calculated from the FAME compositions of fruit oils by our previous methods [7,12]. Also, two additional vital parameters based on the type of FAs were defined for biodiesel feedstock: degree of unsaturation (DU) of FAs and chain length saturated factor (LCSF), of which LCSF is calculated taking into account the compositions of saturated FAs and their corresponding melting point, and DU value is calculated from the amount of monounsaturated and polyunsaturated FAs (%) in fruit oils according to the following equations [7,12,14,15]. All assay was performed for triplicate, and biodiesel fuel properties of fruit oils were compared with the relevant standards of EN 14214 (European), ASTM D6751 (USA) and GB/T 20828 (China): where Cn is percentage of the nth saturated FAs and MPn is the melting point (MP) of nth saturated FAs.

Construction of prediction model for biodiesel properties of raw fruit oils
Triangular prediction model of raw fruit oils from different accessions was constructed based on the influence of FA compositions [7,12]. To predict biodiesel properties of fruit oils from each accession, the percentages of SFA, MUFA and PUFA from each accession fruit were calculated to outline triangular prediction graph, in which three angular points of the triangle meant the 100% of SFA, MUFA and PUFA, respectively. In triangular graph, the region existed at the far end of the polyunsaturated angular point (lower left vertex) and the saturated angular point (lower right vertex) was delineated to predict the biodiesel fuel properties, taking into account the CN, IV, CFPP and OS [7,12].

qRT-PCR analysis
Total RNA was extracted using RNeasy Plant Mini Kits (Qiagen, Inc., USA), and the obtained RNA was qualified and quantified using Nanodrop ND-1000 Spectrophotometer. All the samples showed a 260/280 nm ratio from 1.9 to 2.1, and then was reverse transcribed using the Reverse Transcription System (Promega). All the qRT-PCR primers (Additional file 1:  [12,90]. The qRT-PCR was conducted on 7500 Real-Time PCR System by SYBR Premix Ex Taq Kit (TaKaRa). The negative controls consisting of nuclease-free water instead of template, and reverse transcriptase controls prepared by substituting reverse transcriptase for nuclease-free water in cDNA synthesis step were included in all analyses for each primer pair. Three biological replicates with three technical repetitions each were performed for qRT-PCR.

Protein interaction network analysis
The Search Tool for the Retrieval of Interacting Genes/ Proteins (STRING version 9.0, http:// strin g90. embl. de/) was employed to analyze the potential interactions between all our identified TFs and oil-synthesized enzymes or specific transporters. STRING analysis was conducted using high confidence (score 0.7), and cluster analysis was performed by using k-means with a value of k = 3.