Pennycress (Thlaspi arvense L.) seeds from the SPRING32 germline were used in this study. These seeds were obtained from the Nottingham Arabidopsis Stock Centre (NASC), UK. Seeds were germinated in plates on wet Whatman paper without addition of any other supplement. For germination, seeds were vernalized for 3 days at 4°C and then moved to a growth chamber for additional 10-14 days. No vernalization treatment was required for fully development of flowers and seeds in this germline. Once germinated, seeds were transferred to pots containing a 75:25 mixture of substrate (peat moss, Kekkilä White 420W: vermiculite) and grown in a bioclimatic chamber under a light intensity of 120-150 μmol m^-2 s^-1, with a 16h/8h light/dark photoperiod at 22°C and a relative humidity of 45%. For seed maturation studies, seeds from five developmental stages corresponding to GREEN (G, 12 days after flowering, DAF), GREENYELLOW (GY, 19 DAF), YELLOWGREEN (YG, 26 DAF), YELLOW (Y, 33DAF) and MATURE (M, 45 DAF), were chosen for analysis, similarly as described in Claver et al. (2017). Seeds separated from the pods, corresponding to these different maturation stages ( Figure 1A ), were harvested, frozen in liquid nitrogen, and stored at -80°C for further analysis unless indicated otherwise.

Total RNA was isolated from 0.1 g of Thlaspi arvense seeds from the five maturation stages analyzed using the Cethyl Trimethyl Ammonium Bromide (CTAB)-LiCl extraction method of Gasic et al. (2004). RNA concentration and integrity were measured in a Nanodrop 2000 UV-Vis Spectrophotometer (Thermo Scientific). cDNAs were synthesized from 3 µg of total RNA using SuperScript III Reverse Transcriptase (Fischer) and oligo dT primer, according to the manufacturer’s instructions. Quantitative PCR (qRT-PCR) of target genes was performed using a 7500 Real Time PCR System (Applied Biosystems), SYBR Green Master Mix (Applied Biosystems) and specific primers ( Supplementary Table S1 ). The Ct values were calculated relative to ACT2 and EF1α reference genes using 2^-ΔΔCt method (Livak and Schmittgen, 2001). Data were obtained from the analysis of at least three biological samples with three independent technical repeats for each sample.

RNA-Seq libraries were prepared and sequenced on an Illumina NovaSeq6000 at Novogene Ltd (www.novogene.uk). Ten libraries, corresponding to two biological replicates of the five different seed developmental stages, were constructed in this work. Messenger RNA was purified from total RNA using poly-T oligo-attached magnetic beads. The first strand cDNA was synthesized using random hexamer primers, followed by the second strand cDNA synthesis using either dUTP for directional library or dTTP for non-directional library. The library was checked with Qubit and real-time PCR for quantification and bioanalyzer for size distribution detection. Quantified libraries were pooled and sequenced. Original image data file from high-throughput sequencing was transformed to sequenced reads by CASAVA. Raw data were stored in FASTQ(fq) files, containing sequences of reads and corresponding base quality. For each library, raw reads, clean reads, quality parameters as Q20 (%), Q30 (%) and QC (%), as well as the mapped percentage were first monitored. The results are available in Supplementary Table S2 . Once raw reads were cleaned, alignments were performed with HISAT2 (Mortazavi et al., 2008). Mapped regions were classified as exons, introns, or intergenic regions, and annotated with respect to the Pennycress reference genome (www.ncbi.nl,.gov/assembly/GCA_91186555.2; Nunn et al., 2022). A 79.97% of the clean reads were detected in exonic regions, while 2.71% and 17.30% were detected in intronic and intergenic regions, respectively. The quality of the data was tested through a Pearson correlation analysis, showing that all libraries from the biological replicates were highly related and, therefore, good for the gene expression analysis. Gene expression level was estimated by FPKM values (short for the expected number of Fragments Per Kilobase of transcript sequence per Millions base pairs sequenced; Mortazavi et al., 2008). Correlation of the gene expression levels between samples was estimated by Pearson coefficient greater than 0.92 and the R² greater than 0.8. Up-regulated and down-regulated genes were identified for each seed maturation stage comparison. The screening criteria used for differential expressed genes was log₂(FoldChange) ≥ 1, and padj ≤ 0.05. Similar expression patterns were clustered together using the FPKM values of genes. The overall results of FPKM cluster analysis, clustered using the log₂(FPKM + 1) value, were generated. When required, heatmaps were generated using the online tool https://bar.utoronto.ca/ntools/cgi-bin/ntools_heatmapper_plus.cgi, using the log₂ ratio of (FPKM + 1) in each sample pair. We used the clusterProfiler (Yu et al., 2012) software for enrichment analysis, including GO Enrichment, DO Enrichment, KEGG and Reactome database Enrichment. GO terms with padj < 0.05 were regarded as significant enrichment. In the results of the GO enrichment analysis, the most significant 30 Terms were selected for display. The different colors represent the three GO subclasses of biological process (BP), cellular component (CC), and molecular function (MF). KEGG pathways with padj < 0.05 were regarded as significant enrichment. In the KEGG enrichment results, the most significant 20 KEGG pathways were selected for display.

Total lipids were extracted from Pennycress seeds (0.1 g) with chloroform:methanol (2:1, v:v) as described by Bligh and Dyer (1959). For total fatty acid quantification, we followed the method from Li et al. (2006) through direct whole seed transmethylation using triheptadecanoin (30-35 μg) as internal standard. Fatty acid methyl esters of total lipids were analyzed by GC-FID as described in Claver et al. (2017).

Quantification of each lipid species was carried out on the LIPANG platform by liquid chromatography-MS/MS as previously described (Jouhet et al., 2017). The lipid extracts corresponding to 25 nmol of total fatty acids were dissolved in 100 µL of chloroform/methanol [2/1, (v/v)] containing 125 pmol of each internal standard. Internal standards used were Phosphatidylethanolamine (PE) 18:0-18:0 and DAG 18:0-22:6 from Avanti Polar Lipid and Sulfoquinovosyl diacylglycerol (SQDG) 16:0-18:0 extracted from spinach thylakoid (Demé et al., 2014) and hydrogenated as described in Buseman et al. (2006). Lipids were then separated by high Performance Liquid Chromatography (HPLC) and quantified by MS/MS.

The HPLC separation method was adapted from Rainteau et al. (2012). Lipid classes were separated using an Agilent 1260 Infinity II HPLC system using a 150 mm×3 mm (length × internal diameter) 5 µm diol column (Macherey-Nagel), at 40°C. The mobile phases consisted of hexane/isopropanol/water/ammonium acetate 1M, pH5.3 [625/350/24/1, (v/v/v/v)] (A) and isopropanol/water/ammonium acetate 1M, pH5.3 [850/149/1, (v/v/v)] (B). The injection volume was 20 µL. After 5 min, the percentage of B was increased linearly from 0% to 100% in 30 min and stayed at 100% for 15 min. This elution sequence was followed by a return to 100% A in 5 min and equilibration for 20 min with 100% A before the next injection, leading to a total runtime of 70 min. The flow rate of the mobile phase was 200 µL/min. The distinct glycerolipid classes were eluted successively as a function of the polar head group.

Mass spectrometric analysis was done on a 6470 triple quadrupole mass spectrometer (Agilent) equipped with a Jet stream electrospray ion source under following settings: Drying gas heater: 260°C, Drying gas flow 13 L/min, Sheath gas heater: 300°C, Sheath gas flow: 11L/min, Nebulizer pressure: 25 psi, Capillary voltage: ± 5000 V, Nozzle voltage ± 1000. Nitrogen was used as collision gas. The quadrupoles Q1 and Q3 were operated at widest and unit resolution respectively. PC analysis was carried out in positive ion mode by scanning for precursors of m/z 184 at a collision energy (CE) of 35 eV. SQDG analysis was carried out in negative ion mode by scanning for precursors of m/z -225 at a CE of -55V. PE, phosphatidylinositol (PI), phosphatidylserine (PS), phosphatidylglycerol (PG), Phosphatidic acid (PA), monogalactosil diacylglycerol (MGDG), and digalactosil diacylglycerol DGDG measurements were performed in positive ion mode by scanning for neutral losses of 141 Da, 277 Da, 185 Da, 189 Da, 115 Da, 179 Da, and 341 Da at CEs of 29 eV, 21eV, 21 eV, 25 eV, 25 eV, 8 eV and 11 eV, respectively. Quantification was done by multiple reaction monitoring (MRM) with 30 ms dwell time. DAG and TAG species were identified and quantified by MRM as singly charged ions [M+NH₄]⁺ at a CE of 19 and 26 eV respectively with 30 ms dwell time. CL species were quantified by MRM as singly charged ions [M-H]- at a CE of -45 eV with 50 ms dwell time. The list of MRM transition was adapted from Romsdahl et al., 2022, and presented in Supplementary Table S3 . Mass spectra were processed by MassHunter Workstation software (Agilent) for identification and quantification of lipids. Lipid amounts (pmol) were corrected for response differences between internal standards and endogenous lipids and by comparison with a quality control (QC). QC extract corresponds to a known lipid extract from arabidopsis cell culture qualified and quantified by TLC and GC-FID as described by Jouhet et al. (2017).

Instruments for sample application, chromatographic development, densitometry and HPTLC-MS coupling were from CAMAG (Muttenz, Switzerland).

Before being used, plates were immersed in tetrahydofuran (THF) for cleaning by diffusion. Subsequently, they were dried at 70° C and vacuum (50 mbar) for 15 min. Clean plates provided stable baselines, monitored by UV at 190 nm. To avoid possible impurities from the solvent itself being deposited uniformly on the plate, an additional pre-development with the chosen mobile phase was carried out, in the absence of sample, up to 90 mm migration distance (m.d.).

Standards and chemicals used in the analysis are listed in Supplementary Data Information . Solutions of each above individual standards were also applied in triplicate on the same plate (concentration: 0.33-2 mg/ml per standard in DCM : MeOH, (1:1 v:v); application of effective mass: 3 μg/band). In order to optimize the applied sample volume and thus save sample, the ATS4 filling quality method was used. In a given plate, minimal distance between tracks was 6 mm and distances from the lateral and lower plate edges were 10 mm. One or more tracks were left empty, as blanks. The five seed maturation stages were studied: Two samples (lipid extracts) per maturation stage, corresponding to two different batches, were analyzed by HPTLC-densitometry-tandem MS. Each sample was applied on three different plates (9 measurements per sample). Samples were dissolved (3-4 mg ml^-1) in DCM : MeOH (1:1 v:v). 4 μl/band were applied on the corresponding HPTLC silica gel plate, at least in triplicate, as 4-mm bands, by using the Automatic TLC Sampler (ATS4) system.

Isocratic chromatographic development up to 70 mm-migration distance was performed in a horizontal developing chamber (20 x 10 cm) using an acidic medium: n-heptane (C7), methyl t-butyl ether (MTBE) and acetic acid (AcH) (70:30:1, v:v:v). The selected isocratic development applied to a standard mixture allowed to separate at baseline most of neutral lipid families in samples, e.g. mono-(MAG), di- (DAG), tri-acylglycerides (TAG), fatty acids (FA), fatty acyl-(FAE) and cholesteryl esters (ChOE), and phosphatidylcholine (PC, at the application point), over a total m.d. of 70 mm. Three plates per lipid extract sample of each maturation stage were developed on different days. Detection was carried out using a TLC Scanner 3 densitometer in mode UV at 190 nm. Baseline of chromatograms was corrected manually. WinCATS software (v 1.4.3.6336) was used to control and process data from sample application, chromatography and densitometry. HPTLC separation was first tested on silica gel plates using the standards mentioned in Materials and Methods. Chromatograms related to standards are shown in Supplementary Figure S1 . HPTLC chromatograms corresponding to samples at the different maturation stages, detected at UV 190 nm. Development conditions were selected to clearly separate TAG (55 mm, m.d.) from the other neutral lipid families ( Supplementary Figure S2 ).

TLC-MS Interface 2 was used from an extraction of each TAG peak directly from the plate. A detail of elution-based interface description and operation can be found elsewhere (Sancho-Albero et al., 2022). It was equipped with an oval, 4 x 2-mm extraction head that was positioned on the corresponding TAG-band maximum, whose the x,y coordinates were provided by WinCats software, using a laser crosshair. Then the interface head was lowered. MeOH was delivered for band extraction at 0.2 mL/min by using a PU-2080 HPLC pump (Jasco, Tokyo, Japan). The eluate was directed through a 2-μm stainless steel frit to remove silica gel and then sent to the mass spectrometer. Electrospray ESI-MS in positive mode (ESI⁺) was selected and mass spectra were registered on an Ion trap Amazon Speed Spectrometer (Brüker Daltonics, Bremen, Germany). ESI⁺-MS was conducted with capillary and endplate offset voltages of -4500 and -500 V, 36 psi as pressure of the nebulizer gas (N₂), 6.0 L/min as flow rate of the drying gas (N₂) and 120°C as drying gas temperature. Spectra were acquired in the m/z 70–1500 range at the ultra-scan mode. Bruker Daltonics Trap Control software packages v 8.0 and Data Analysis v 5.2 were used to control the mass spectrometer and process data. For each TAG peak, several HPTLC- ESI⁺-MS experiments were performed from replicate bands and confirmation of identity was carried out by MS². These experiments were performed from different plates. The HPTLC-ESI⁺-MS operating conditions are specified for each case in the Results and Discussion section.

MS acquisition was performed by a Quadrupole Time-of-Flight (QTOF) mass spectrometer equipped with an Electrospray Ionization Source (ESI) (MicrOTOF-Q, Bruker Daltonics, Bremen, Germany). High Resolution (HR)-MS experiments were carried out in positive ion mode. The nebulizer gas (N2) pressure, the drying gas (N2) flow rate and the drying gas temperature were 1.6 bar, 8.0 L/min, and 190°C, respectively. Spectra were acquired in the m/z 50–2000 range. The mass axis was calibrated by using Na-formate adducts [10 mmol/l NaOH, 2.5% (v/v) formic acid and 50% (v/v) 2-propanol] that were introduced through a divert valve at the beginning of each direct injection. Bruker Daltonik software packages micrOTOF Control v.3.4 and HyStar v.3.2 were used to control the system. Data Analysis v.4.2 was used to process the data.

Data are expressed as means ± SD, with at least three replicates in each experimental group. The statistical comparisons among the different developmental stages during seed maturation of Pennycress were made using one-way analysis of variance (ANOVA) and means were compared with the Duncan’s multiple range test (P < 0.05). When data showed non-normality, log or reciprocal transformations were made and ANOVA conducted with the transformed data.

Seeds from five different developmental stages of Pennycress SPRING32 germline, from the youngest GREEN (G) stage to the final MATURE (M) stage, were analyzed in this study ( Figure 1A ). These five stages covered the whole seed maturation process. Temporal changes in fatty acid composition were analyzed in total lipid fractions from each developmental stage. Erucic acid was highly abundant in all stages ( Figure 1B ). 22:1 represented a 20% of total fatty acids at the G initial stage, increasing from the G to the GY and YG stages, reaching values higher to 35%, to then slowly decrease in the latter Y and M stages ( Figure 1B ). The lower 22:1 values at the younger G stage were concomitant with higher 18:2 levels, that later decreased upon Pennycress seed maturation ( Figure 1B ). Other VLCFAs like 20:1 and 24:1 were also detected in all seed maturation stages, increasing their levels during the whole seed maturation process ( Figure 1B ).

To monitor the expression of genes encoding enzymes involved in seed oil biosynthesis during maturation, we first performed an RNA-Seq analysis to analyze transcriptome changes during maturation at the five stages described above. In general, and taking into account the latest annotation of the Pennycress genome available (García Navarrete et al., 2022), which estimated 28,034 genes from which 27,213 corresponded to protein coding genes, our RNA-Seq analysis identified 20,015 protein coding genes that covers a 73.54% of the total Pennycress genome. When pairwise comparisons were analyzed between seed maturation stages, the results indicated that 3,443 differentially expressed genes (DEGs) were identified in the GY vs G comparison, 6,406 in the YG vs G comparison, 9,214 in the Y vs G comparison, and 10,994 in the M vs G comparison ( Figure 2 ). Interestingly, the number of identified DEGs increased during seed maturation and were not clustered to the initial maturation stages ( Figure 2 ), indicating specific gene expression dynamics all-through the seed maturation process. The ratio of up-regulated to down-regulated genes also changed during seed maturation. While at the initial stages (G or GY) the up-regulated genes were slightly higher or similar to the down-regulated ones, at the late maturation stages (Y or M), the number of down-regulated genes was higher than that of the up-regulated genes ( Figure 2 ).

Genes encoding enzymes involved in fatty acid biosynthesis or modification were analyzed into more detail. This included the two condensing enzymes, TaKAS1 and TaKAS2, the two fatty acid ACP thioesterases, TaFATA and TaFATB, responsible of hydrolysing 16:0-ACP and 18:0-ACP substrates for export to the ER, acyl carrier proteins (ACPs) or Long Acyl Chain Synthetases (LACS). With the exception of TaFATB, that showed similar expression values, most, if not all these genes, showed higher expression values at the G, GY or YG stages and then decreased in the M stage ( Figure 6 ). Similarly, several genes encoding 3-ketoacyl-CoA synthase family members, involved in the biosynthesis of VLCFAs, like TaKCS8, TaKCS16 or TaKCS18, showed an expression pattern, higher at the early stages of maturation, similar to that of the FA biosynthetic genes ( Figure 6 ). Expression of these genes, more concretely that of TaKCS18 (FAE1), was consistent with 20:1 and 22:1 fatty acid accumulation during Pennycress seed maturation ( Figure 1B ; Claver et al., 2017). Several enzymes involved in fatty acid modification like the Δ9 acyl-lipid desaturases TaADS1 and TaADS2, involved in the desaturation of VLCFAs; the ER desaturases TaFAD2 and TaFAD3, responsible of the biosynthesis of 18:2 and 18:3 fatty acids, or the plastidial TaFAD4 desaturase, responsible of 16:1 synthesis in the plastid were also analyzed. TaADS1 showed maximum expression at the G stage, decreasing with seed maturation, while TaADS2 increased its expression from G to YG to decrease at the Y and M later maturation stages ( Figure 6 ). The ER TaFAD2 and TaFAD3 desaturases also increased their expression from G to GY (TaFAD2) or to YG (TaFAD3) to decrease in the latter maturation stages ( Figure 6 ). TaFAD4 also showed maximum expression at the G stage decreasing with seed maturation ( Figure 6 ).

The glycerol 3-phosphate acyltransferase (GPAT) is the enzyme that catalyzes the first step in TAG biosynthesis through the Kennedy pathway. GPAT is capable of transferring an acyl group to the sn-1 position of glycerol 3-P to generate lysophosphatidic acid (LPA). Plant GPATs are a multigenic family with different sub-cellular localizations and roles in lipid biosynthesis. In Arabidopsis, AtGPAT1 was located in the mitochondria playing a central role in the differentiation of the tapetum, male fertility and pollen development (Zheng et al., 2003). AtGAPT4 and AtGAPT8 are involved in extracellular lipid barriers (Yang et al., 2010), while AtGAPT5 was involved in suberin formation in seed coats (Men et al., 2017). AtGAPT9 has been directly involved in the synthesis of storage lipids (Shockey et al., 2016). Our RNA-Seq analysis showed a complex pattern of regulation of the TaGPAT genes. Thus, TaGPAT1 and TaGPAT2 showed fluctuations of their expression without great changes in the different maturation stages ( Figure 6 ). Interestingly, TaGPAT4 and particularly, TaGPAT8, showed an increase of their expression values from G to the GY or YG stages to the decrease dramatically at the latter Y and M ones ( Figure 6 ). On the contrary, TaGPAT9 and particularly TaGPAT5, showed a specific and important increase of their mRNA levels at the Y and M stages ( Figure 6 ). Genes involved in lipid transport, like TaLTP4, TaLTP5, TaLTP6 and TaLTP12, also showed a similar higher expression at the earlier stages of seed maturation and then decreasing in the later ones ( Figure 6 ). Particularly, TaLTP4 and TaLTP6 showed an important modification of their expression values suggesting a relevant role in the transport of acyl lipids in the seed.

Glycerol-3-phosphate and acetyl-CoA are the carbon sources necessary for fatty acid and TAG biosynthesis in the seed. Several genes involved in carbon assimilation like glycerol-3-phosphate dehydrogenases (G3PDHs), pyruvate dehydrogenases (PyrDHs), phosphoenolpyruvate carboxylases (PEPC) or phosphoenolpyruvate carboxylases:kinases (PEPCK) were monitored. All of them showed a similar expression pattern to those involved in fatty acid biosynthesis or modification, with higher expression values at the earlier stages of seed maturation (G, GY) to then decrease in the later ones (Y, M), ( Figure 6 ).

We focused our analysis on those genes involved in TAG biosynthesis as well as those involved in the biosynthesis and modification of fatty acids incorporated to TAG like erucic acid. On one hand, we used the RNA-Seq data (FPKM values) to monitor the expression of several selected genes. On the other hand, we performed a qPCR analysis on samples from each maturation stage of the same selected genes, comparing both expression data and validating the RNA-Seq results. Pennycress accumulates high levels of 22:1 during seed maturation (Claver et al., 2017; 2020). Expression of the TaFAE1 gene, encoding the elongase responsible of 22:1 production, increased 3 to 5 fold between the initial G stage to the GY and YG maturation stages declining thereafter either in the RNA-Seq data and in the qPCR analysis ( Figure 7 ). This induction of TaFAE1 at the initial stages of Pennycress seed maturation could be consistent with the rapid availability of 22:1-CoA in the acyl-CoA pool for its early incorporation to total lipids as shown in Figure 1B and reported previously by our group in Pennycress accessions of European origin (Claver et al., 2017). A very similar gene expression profile was obtained for the TaFAD2 desaturase, with a two-fold increase of mRNA levels obtained both at the RNA-Seq data and qPCR analysis at the GY stage to then decrease its mRNA levels upon Pennycress seed maturation ( Figure 7 ). This again was consistent with the high 18:2 levels in total lipids at the early stages of seed maturation ( Figure 1B ).

TaDGAT1 and TaDGAT2 genes encode two diacylglycerol acyltransferases with high homology, 90.6 and 81.3% with respect to the Arabidopsis AtDGAT1 and AtDGAT2 genes, respectively (Routaboul et al., 1999; Shockey et al., 2006). TaDGAT1 expression levels increased gradually during Pennycress seed maturation, particularly between the YG and Y stages, showing a 2,5-3 fold maximum increase at the Y stage ( Figure 7 ). On the contrary, TaDGAT2 expression results from both RNA-Seq and qPCR data showed higher mRNA levels at the initial maturation stages, G and GY, further declining upon seed maturation ( Figure 7 ). It is worth mentioning that the FPKM values of both TaDGAT1 and TaDGAT2 genes indicated that TaDGAT2 mRNA was more abundant than that of TaDGAT1 at the G stage and similar at the GY one, suggesting a specific role of TaDGAT2 in TAG biosynthesis, particularly at the initial stages of Pennycress seed maturation.

In higher plants, lysophosphatidic acid-acyltransferases (LPATs) are a multigenic family involved in DAG biosynthesis in the Kennedy pathway (Kim and Huang, 2004; Kim et al., 2005). Two LPAT genes, TaLPAT1 and TaLPAT2, with high homology to their Arabidopsis orthologues, were detected in the Pennycress genome. TaLPAT1 expression followed a similar pattern to that of TaDGAT1, with a mRNA increase from the YG stage to the Y and M stages between 2-3.5 fold, both in the RNA-Seq and qPCR data ( Figure 7 ). On the contrary, TaLPAT2 showed very subtle changes in its expression.

The expression of genes encoding enzymes of the acyl-editing pathway was also monitored. As occurred with the two DGAT genes, two TaPDAT genes, encoding the phospholipid-diacylglycerol acyltransferases responsible of the biosynthesis of TAG through the acyl-editing pathway (Zhang et al., 2009) were detected in the Pennycress genome with high homology, 87 and 88% with respect to the Arabidopsis AtPDAT1 and AtPDAT2 genes, respectively. Both PDAT genes showed completely different expression patterns. It is worth mentioning that the TaPDAT1 gene was not detected in the RNA-Seq data and only the TaPDAT2 gene was found. qPCR data indicated that TaPDAT1 mRNA levels were high at the initial stages of seed development and then rapidly declined ( Figure 7 ). On the contrary, the TaPDAT2 gene showed a continuous increase in mRNA levels upon seed maturation reaching maximum expression levels at the Y stage both in the qPCR and in RNA-Seq data. These results might suggest different roles for both PDAT enzymes at the early (PDAT1) or late (PDAT2) stages of seed maturation. The other enzyme of the acyl editing pathway, TaLPCAT, responsible of the reincorporation of an acyl group to PC, maintained its expression levels between the G to the YG stage to then slowly decreased at the Y and M stages, suggesting that LPCAT expression and/or activity might not be limiting for seed oil accumulation. Interestingly, the expression of the TaPDCT gene, involved in PC-derived DAG interconversion, showed a similar expression pattern to that of TaPDAT1, with high mRNA accumulation at the early stages of seed maturation ( Figure 7 ).

Expression of WRINKLED1, the TF involved in the control of many genes of the lipid biosynthetic pathway in the seed (Cernac and Benning, 2004; Baud et al., 2007), showed higher expression at the early stages of seed maturation, consistent with the upregulation of seed oil biosynthetic genes ( Figure 7 ).

Finally, we monitored the expression of genes involved in lipid droplet formation. These lipid droplets increase their number and size upon seed maturation (Farese and Walther, 2009). Genes encoding two oleosins, TaOLE1 and TaOLE2, showed a similar increase in mRNA levels both in the RNA-Seq data and in the qPCR analysis, with maximum values between the YG and Y stages ( Figure 7 ), consistent with higher TAG accumulation ( Figure 8 ). Interestingly, TaOBAP1a showed a similar increase during seed maturation although the extent of these changes seemed to be much higher than Oleosins at least at the transcript levels (15-30 fold at the Y stage; Figure 7 ).

Glycerolipid analysis of the different lipid species present in the Pennycress seeds was carried out by LC-MS (Jouhet et al., 2017). This analysis showed some significant changes in some lipid classes during Pennycress seed maturation. Thus, at the initial G stage, several lipid classes showed different relative abundances like TAG (29.3%), PA (28.5%), PE (3.4%) or the plastidial lipids MGDG (7.1%) or DGDG (8.4%), ( Figure 8 ). Other lipid classes like PG (2.7%), PC (10.1%) or DAG (3.3%) were also detected in this initial maturation stage ( Figure 8 ). Upon seed maturation, TAG levels showed the highest increases among the different lipid classes, with values ranging from 79.5% at the GY stage to 85.6% at the Y stage or 90.8% at the M stage ( Figure 8 ). Conversely, other lipid species like MGDG or DGDG showed a decrease during seed maturation, decreasing to 0.8% at the Y stage or to 0.09% at the M stage for MGDG ( Figure 8 ). Other lipid classes like PC or DAG did not show such relevant changes. Thus, PC levels decreased during Pennycress seed maturation, with relative levels ranging from 10.1% at the G stage to 5.4% at the M stage ( Figure 8 ). Similarly, DAG levels kept close to the 3% range at the G, GY, and YG stages, showing a decrease to lower values at the late M stages (0.8%), ( Figure 8 ). It is worth mentioning that PC levels were always higher than those from DAG in all seed maturation stages.

Quantitative analysis of the acyl composition of the different lipid classes revealed changes in their distribution with Pennycress seed maturation. Figure 9A shows the acyl composition of TAG, the major lipid fraction as well as those from DAG and PC as intermediate species during TAG biosynthesis. Fatty acid distribution in TAG showed that at the G stage, 58:4 (18:1/18:2/22:1) and 58:5 (18:2/18:2/22:1) together with 62:4 (22:1/18:2/22:1) were the most abundant TAG species detected in the LC-MS analysis ( Figure 9A ). Other TAG species like 56:3 (16:0/18:2/22:1), 54:4 (18:1/18:1/18:2) or 54:5 (18:1/18:2/18:2) were also very abundant at this initial G stage ( Figure 9A ). TAG 64:4 (22:1/18:2/24:1), as well as 64:3 or 64:5, was also present at this initial maturation stage although in very low amounts (less than 1%), ( Figure 9A ). It is worth mentioning that although TAG species containing 16 and 18 carbon fatty acids were highly abundant at this initial stage, many of them already contained 22:1, indicating a rapid incorporation of 22:1 to TAG even at the younger stages.

TAG species containing VLCFAs, particularly 22:1, increased upon seed maturation. Thus, 62:4 levels doubled from the G to the GY or YG stages ( Figure 9A ). The same occurred with 62:5 (18:3/22:1/22:1), being now 62:4 and 62:5 the most abundant TAG species in the rest of the stages upon Pennycress seed maturation. Similarly, TAG species containing 22:1 and 24:1 like 64:4 or 64:5, although much less abundant than 62:4 or 62:5, also increased with seed maturation ( Figure 9A ). Levels of 58:4 and 58:5 remained similar at the rest of the maturation stages. On the contrary, levels of TAG species like 54:4 or 54:5, containing 16:0, 18:1 and 18:2 fatty acids, that were very abundant at the G stage decreased rapidly with seed maturation ( Figure 9A ). The fatty acid composition of DAG was also analyzed. 34:2 (16:0/18:2) was the major DAG species detected at the initial G stage ( Figure 9B ), representing a 22.21% of the total DAG detected in this stage. Other abundant DAG species were 36:3 (18:1/18:2), 36:4 (18:2/18:2), 36:5 (18:2/18:3) or 36:6 (18:3/18:3), ( Figure 9B ). DAG species containing VLCFAs like 38:4 (18:3/20:1), 40:3 (18:2/22:1) or 40:4 (18:3/22:1) were also present at the G stage although their relative abundance was lower than those species containing C16 or C18 fatty acids ( Figure 9B ). This distribution changed dramatically upon seed maturation. Thus, 40:3 (18:2/22:1) levels increased up to 3-fold from 2.6% to 9.0 and 8,8% at the GY and YG stages, respectively, when compared to the G stage, reaching even higher 12.7% and 17.8% relative levels at the Y and mature M stages, respectively, ( Figure 9B ). DAG 38:3 (18:2/20:1) and 40:4 (18:3/22:1) also increased dramatically with Pennycress seed maturation with values of 3.2% at the G stage to 8.0% (GY), 7.2% (YG), 13.6% (Y) and 23.8% (M) in the case of 38:3 or 7.5%(GY), 6.4% (YG), 9.5% (Y) and 21.3% (M) for 40:4, indicating a higher accumulation of VLCFAs, particularly 22:1 in DAG during Pennycress seed maturation. In fact, 38:3, 40:3 and 40:4 were the major DAG species in mature seeds ( Figure 9B ). This increase in VLCFAs containing species was concomitant with the decrease of DAG 34:2, which was the most abundant one at the G stage ( Figure 9B ). Other DAG species like 36:4, remained almost constant with little variations at the different seed maturation stages ( Figure 9B ).

Acyl group distribution in phospholipids was also analyzed. 34:1 (16:0/18:1), 34:2 (16:0/18:2), 36:3 (18:1/18:2) and 36:4 (18:2/18:2) were the most abundant PC species detected in the initial G stage, with relative amounts of 14.4%, 14.6%, 19.3% and 17.4%, respectively ( Figure 9C ). PC species containing 20:1 like 38:2 (18:1/20:1), 38:3 (18:2/20:1), or 38:4 (18:3/20:1) were also detected although in much lower amounts (1.6-2.4%) when compared to PC species containing C16 and C18 fatty acids ( Figure 9C ). PC species containing 22:1 like 40:2, 40:3 or 40:4 were also detected in the G initial stage although in very low amounts (0.8%; Figure 9C ). Upon seed maturation, PC species like 34:1, 36:3 and 36:4 showed a decrease in their relative levels while other like 36:5 (18:2/18:3) or 36:2 (18:1/18:1) increased their levels with seed maturation. Interestingly, PC species containing VLCFAs increased their relative abundance with seed maturation. Thus, PC species like 38:2, 38:3, 38:4 (containing 20:1) and 40:2, 40:3 and 40:4 (containing 22.1) increased their levels, particularly from the G to the GY and YG stages ( Figure 9C ). It is worth mentioning that our data of the presence and distribution changes upon seed maturation of PC species containing VLCFAs differed from those previously reported by Romsdahl et al. (2022) that detected PC species containing VLCFAs at very low levels (if any) in their analysis. Acyl group distribution in other phospholipids showed a similar fatty acid composition with respect to that from PC, with species containing C16 and C18 fatty acids as major species, presence of VLCFAs in lower amounts at the initial stages of seed maturation that increases upon seed maturation ( Supplementary Figure S3 ). Particularly interesting were the results obtained with PS that showed a high accumulation of 40:3 (18:2/22:1) and 40:4 (18:3/22:1) species upon seed maturation.

We decided to monitor how the erucic acid, as well as other acyl groups, were incorporated to TAG, with particular interest in the positional analysis of the different acyl groups esterified to TAG. To that end, we used a lipidomics approach based in the application of HPTLC - UV densitometry - MS to analyze the chemical composition of the different TAG species at each seed developmental stage and then couple this analysis with tandem mass spectrometry to study the fragmentation pattern of these TAG species and obtain positional information. HPTLC-ESI-MS has proven to be useful for lipidomic analysis in complex lipid mixtures (Jarne et al., 2018, 2021; Sancho-Albero et al., 2022). This analysis focused on TAGs as it constitutes the major fraction of the total seed lipids in Pennycress ( Figure 8 ; Claver et al., 2017). Two different lipid extract samples per stage (G, GY, YG, Y and M), which corresponded to two different extraction batches, were analyzed by HPTLC-densitometry-MS. Percentages of TAGs in samples, as well as intra- and inter-plate HPTLC repeatability results (expressed in Area counts) for the separated TAG peaks are presented in Supplementary Table S4 , including the average, the relative standard deviation (RSD%), and the coefficient of variation for a confidence interval of 95% (CV).

Figure 10 shows the HPTLC-ESI⁺-MS spectra of TAG zones corresponding to each Pennycress seed maturation stage. For all samples, mass spectra were recorded at the same ionization time and conditions, in order to compare relative ion intensities of different TAG species in each maturation stage. In general, eight major TAG species were identified in all seed maturation stages with maxima intensities at m/z 853.8; 879.8; 907.8; 935.9; 959.9; 989.9; 1017.9 and 1046.0, ( Figure 10 ). These molecular species corresponded to 54:3, 56:3, 58:5, 60:4, 62:4 and 64:4, respectively. Although a quantitative analysis is excluded, there is an ion intensity-concentration relationship for each sample that allows the comparison between ion intensities from the ESI⁺-MS spectra of each sample. Accordingly, at the early G and GY stages, TAG species with m/z 959.9 (58:5) was the major TAG species ( Figure 10 ). Other TAG species identified in the analysis were 56:3 (m/z 935.9), 60:4 (m/z 989.9) and 62:4 (m/z 1017.9), ( Figure 10 ). This distribution was similar to that obtained by LC-MS ( Figure 9A ). Upon seed maturation, the distribution of TAG species at the YG and Y stages showed a change with respect to the G and GY initial stage: an increase of TAG 60:4 (m/z 989.9) and 62:4 (m/z 1017.9) corresponding to the species containing 20:1/18:2/22:1 and 22:1/18:2/22:1, respectively was observed ( Figure 10 ). Finally, at mature stage (M), TAG species like 58:5 (m/z 959.9), 60:4 (m/z 989.9) and 62:4 (m/z 1017.9) were the most abundant ones ( Figure 10 ). 24:1 in the TAG 64:4 (ion at 1046.0 m/z) species was also detected in the analysis. Results from HPTLC-MS using an ion trap were in good agreement with those from LC-MS using a triple quadrupole ( Figure 9A ) indicating that, upon maturation, an increase in TAG species containing VLCFAs like 22:1 or 20:1 occurred. These results validate the use of HPTLC-ESI-MS technology for the analysis of TAG species in complex lipid mixtures like Pennycress seed lipid fractions.

TAGs are weakly basic esters and, under the HPTLC and ESI⁺ conditions readily lead to the formation of [TAG+Na]⁺, which in our case can be fragmented to yield structural information (Jarne et al., 2018, 2021; Sancho-Albero et al., 2022). According to Ramaley et al. (2015), [M+Na]⁺ adducts from ion trap are the most suitable for TAG regioisomer analysis as they produced the most consistent level of positional sensitivity for the fragmentation. The preferential loss of the fatty acid at positions sn-1/3 seems to be general for TAG molecules regardless of energy and instrumentation employed, leading to the formation of two ions of similar abundance corresponding to the losses of the fatty acids substituents at sn-1 and at sn-3 and those are significantly more abundant than the ion corresponding to the loss of the fatty acid substituent at sn-2 (Hsu and Turk, 2010). Thus, the intensities of the resulting fragment ions reflect the FA distribution in the glycerol backbone.

TAG regioisomers can be identified with MSⁿ methods, but analysis of TAG enantiomers is not possible because the fragmentation methods cannot distinguish between sn-1 and sn-3 fatty acids due to the identical fragmentation efficiencies of fatty acids from these positions. In natural products, many isobaric TAG species, including isomers may produce shared isobaric fragment ions, making the analysis of TAG regioisomers even more challenging. However, on several occasions where the most abundant ion comes from a single triad of fatty acid composition, we have been able to identify the fatty acid in the sn-2 position. Ions m/z corresponding to TAG species are reported in Table 1 along with their fragmentation patterns (MS²). Therefore, ion at m/z 1017.9 corresponds to the sodium adduct of TAG (62:4) [C₆₅H₁₁₈O₆Na]⁺. As the stability of [M+Na]⁺ was high, a consecutive fragmentation was achieved in the ion-trap MS to have verification of identity. Hence, the respective HPTLC-ESI⁺-MS/MS spectrum of the precursor ion at m/z 1017.9 showed two ion products corresponding to losses of fatty acyl substituents as fatty acids: at m/z 679.6 (most intense) which corresponds to [M+Na−R_1,3COOH]⁺, R_1,3= C(22:1) fatty acids, and at m/z 737.62 which corresponds to [M+Na−R₂COOH]⁺, R₂= C(18:2) fatty acids; and two much less abundant ions products corresponding to losses of fatty acyl substituents as their sodium salts: 657.6 m/z which corresponds to [M+Na−R_1,3COONa]⁺, R_1,3 = C(22:1) fatty acids, and 715.7 m/z which corresponds to [M+Na−R₂COONa]⁺, R₂= C(18:2) fatty acid ( Table 1 , Figure 11 ). ESI MS/MS spectra were carried out using He as the collision gas, an optimal amplitude voltage of 0.6 V and an isolation width for the precursor ion of 1 m/z units ( Table 1 , Figure 11 ). Results are consistent with a TAG structure of 22:1/18:2/22:1 with linoleic acid at the sn-2 position. This TAG is already present at the youngest state, in all stages, and becoming the most important in GY, YG, Y and M. It was also possible to identify sn-2 position in the ion at m/z 989.8 and in the ion at m/z 1046.0. In the other ions at m/z 907,8, 935,9 and 959, 9, although fragments compatible with 18:2 at sn-2 were obtained, it was more difficult to identify the sn-2 position since several TAG species can contribute to the same ion. In these cases, this analysis should not be used as the basis for excluding the presence of other isomers.