In vitro digestion of galactolipids from chloroplast-rich fraction (CRF) of postharvest, pea vine field residue (haulm) and spinach leaves

26 The removal of intact chloroplasts from their cell wall confinement offers a novel way to obtain 27 lipophilic nutrients from green biomass, especially carotenoids and galactolipids. These latter are the 28 main membrane lipids in plants and they represent a major source of the essential a -linolenic acid 29 (18:3; ALA). Nevertheless, knowledge on their digestion is still limited. We have developed a physical 30 method of recovering a chloroplast-rich fraction (CRF) from green biomass and tested its digestibility 31 in vitro under simulated gastrointestinal conditions. Using a two-step static model, CRF from both 32 spinach leaves and postharvest, pea vine field residue (haulm) were first exposed to enzymes from 33 rabbit gastric extracts and then either to pancreatic enzymes from human pancreatic juice (HPJ) or to 34 porcine pancreatic extracts (PPE). The lipolysis of monogalactosyldiacylglycerol (MGDG) and 35 digalactosyl diacylglycerol (DGDG) was monitored by thin layer chromatography and gas 36 chromatography of fatty acid methyl esters. For both CRF preparations, MGDG and DGDG were 37 converted to monogalactosylmonoacylglycerol (MGMG) and digalactosylmonoacylglycerol (DGMG), 38 respectively, during the intestinal phase and ALA was the main fatty acid released. Galactolipids were 39 more effectively hydrolysed by HPJ than by PPE, and PPE showed a higher activity on MGDG than 40 on DGDG. These findings may be explained by the higher levels of galactolipase activity in HPJ 41 compared to PPE, which mainly results from pancreatic lipase-related protein 2. Thus, we showed 42 that CRF galactolipids are well digested by pancreatic enzymes and represent an interesting vehicle 43 for ALA supplementation in human diet. digestibility (CRF) and simulated group chloroplasts from leaves using grinding sucrose solution and from PVH using a slow-screw twin juicer without water. gastric enzymes from rabbit gastric pancreatic enzymes HPJ Porcine Pancreatic Extract means


Introduction 50
The United Kingdom is the largest producer of frozen peas in Europe with large areas given 51 over for pea growing; at present, around 31,707 hectares. In 2015, the yield of pea production was 52 steam condition (section 2.4.1). Another batch of haulm (5 kg) was juiced immediately using a twin 137 gear juicer (Angel 7500) which separated the fibrous pulp from the nutrient rich juice. The juice was 138 filtered through a 75 µm stainless steel mesh sieve and processed (with or without pasteurisation) to 139 isolate the CRF (section 2.6). 140 2.4.1. Steam sterilisation of pea vine haulm 141 The pea vine biomass was packed into a vacuum sealed, clear, perforated bag (500 g of 142 pea vine per bag). These bags were placed in the rack of a Retort (Lagarde RP362). The chamber 143 was sealed, vented and heated over 5.30 min to reach a temperature of 100 o C and 1 bar. These 144 sterilisation conditions were held for 4 min before cooling and depressurising for 5.45 min. The sealed 145 bag of pea vine was plunged into ice-water bath to rapidly cool. The steam treated haulm was 146 immediately juiced using a twin gear juicer (Angel 7500) and the juice was filtered through a 75 µm 147 stainless steel mesh sieve and processed to isolate the CRF (section 2.6). 148

Pasteurisation of juice extracted from pea vine haulm 149
Pea vine juice (500 mL) was placed in a capped amber Duran bottle in a heated water bath 150 with a magnetic stirrer (800 rpm). The temperature of the juice was raised from room temperature 151 (20 o C) to 85 o C in 15 min. It was then held at this temperature for 1 min. The pasteurised juice was 152 immediately immersed in an ice-water bath to rapidly cool the juice down to room temperature before 153 further processing to isolate the CRF (section 2.6). 154

Hot-water blanching of spinach leaves 155
Spinach leaves (100 g) were blanched in hot water at 85 o C for 3 min and then immediately 156 immersed in an ice-water bath to rapidly cool to room temperature. Blanched spinach leaves were 157 homogenised in a blender (Waring TM ) for 30 s with 0.3 M sucrose solution 1:6 (w/v). The homogenate 158 was then filtered through a double-layered cheese cloth and processed to isolate the CRF (section 159 2.6). 160

Isolation of CRF 161
The PVH juice was centrifuged at 17,700 RCF or 10,000 rpm (Beckman Coulter JS-21M with 162 JA-10 rotor) for 10 min at 4 o C. The CRF-containing pellet was retained while the supernatant was 163 centrifuged again under the same conditions to obtain a further pellet. , equivalent to 21 µg lipase/mg) was added so that the final 180 concentration of gastric lipase was 17 µg /mL (20 U/mL). The pH was adjusted at 5 and the solution 181 was then incubated for 30 mins under gentle magnetic stirring. For the intestinal phase, the solution 182 from the gastric phase was diluted by half using a pancreatic enzyme-bile salts solution, pH was 183 adjusted at 6 and the incubation was continued for 60 min. The final concentration of pancreatic 184 lipase was set at 250 µg/mL or 2000 lipase U/mL and that of bile salt (NaTDC) was 4 mM. Two 185 sources of pancreatic enzymes were tested: (1) freeze-dried HPJ, that contained 228 lipase U per mg 186 of powder or (2) Sigma PPE, that contained 67 lipase U per mg of powder. In both cases, lipase units 187 (U) for pancreatic lipase refer to the assay using tributyrin as substrate.

Lipid extraction 195
Lipid extraction was performed using the method of Folch et al. (1957), 35 modified by Bligh 196 andDyer (1959). 36 Sample (25 mg CRF or 1 mL of digestion sample) was mixed with 1 mL of 150 mM 197 NaCl solution and 1.5 mL of a 2:1 v/v mixture of chloroform and methanol, then vortexed for 1 min. 198 The mixture was then centrifuged (using Thermo Electron Corporation, Jounan CR3i multifunction) at 199 3,000 rpm or 1,750 RCF for 10 min at 4 o C, which allowed the phase separation. The lowest organic 200 phase, which contains lipids, was collected using a Pasteur pipette and transferred into a fresh tube. 201 A further 1.5 mL of 2:1 v/v chloroform: methanol was added to the remaining aqueous phase and the 202 mixture was vortexed and centrifuged again according to the same procedure. The lipid extracts were 203 pooled and their volume was measured before the organic phase was dried using magnesium 204 sulphate. After centrifugation, the lipid extract was kept at -20°C until analysis. 205

Quantitative analysis of galactolipids and their lipolysis products by thin layer 206 chromatography 207
In order to separate and quantify galactolipids and lipolysis products, 10 to 50 µL of lipid 208 extracts and known amounts (2, 4, 6, 8, and 10 µg) of lipid standards (MGDG, DGDG and oleic acid) 209 were spotted as a 5 mm band onto a thin-layer silica plate using a Limonat IV (Camag) equipped with 210 a 100 µL Hamilton syringe. The separation of polar lipids was performed with a 211 chloroform/methanol/water (47.5:10:1.25, v/v/v) elution mixture. The separation of free fatty acids 212 (FFA) on a second silica plate was performed with a mobile phase consisting of heptane: diethyl 213 ether: formic acid (55:45:1, v/v/v) solvent mixture. Following chromatography, the plates for polar 214 lipid/galactolipid analysis were dried at room temperature under a fume hood for 15 min and then 215 dipped in a thymol solution prepared by dissolving 1g of thymol in 190 mL ethanol and then addition 216 of 10 mL of 96% sulphuric acid. Since the mixing reaction is highly exothermic, the ethanolic solution 217 has to be placed first in a cold water bath before sulphuric acid is added dropwise. The thymol 218 solution allows the staining of galactolipids while avoiding the interference of pigments, especially 219 chlorophylls, during the densitometric analysis of the plates. After staining with thymol, the plates 220 were dried again in the fume hood for 10 min and then placed in an oven at 110 o C for 10 min. The 221 plate for FFA analysis was dipped in a copper acetate-phosphoric acid solution prepared by mixing a 222 saturated copper acetate solution with 85% phosphoric acid in a 1 to 1 volume ratio. The plates were 223 dried for 10 min in the fume hood and then placed in an oven at 180 o C for 15 min. Densitometry 224 analysis of the stained lipids on the TLC plate was carried out using a Camag TLC scanner II and a 225 D2000+ chromato-Integrator (Merck). Lipid bands were scanned at 366 nm for thymol staining and at 226 500 nm for copper acetate-phosphoric acid staining, with a 0.5 x 7 mm slit and a speed of 2.5 227 cm/min. Slit conditions were selected accordingly to band size. The slit should always cover the whole 228 band size. The densitograms of all tracks were integrated using D2000+Chromator-Integrator. 229 Quantities of the lipids on the TLC plates were estimated from the linear standard curves established 230 with the pure lipid standards (MGDG, DGDG, and oleic acid (18:1)). 231

Fatty acid analysis by GC-MS 232
The fatty acids contained within the lipid extracts were esterified to fatty acid methyl esters 233 (FAMEs) and analysed using gas chromatography coupled to mass spectrometry detection (GC-MS) 234 (Thermo Scientific, DSQ) using a modified method based on Bahrami et al. (2014). 37 The solvent from 235 lipid extracts (2.1 mL) was first evaporated under nitrogen and the resulting dry material was re-236 dissolved in 1 mL of chloroform. Methylpentadecanoate (internal standard) and trimethyl 237 sulfoniumhydroxide (TMSH) were added to lipid extract in chloroform to convert both the FFA and 238 esterified fatty acids into FAMEs. The reaction was performed for at least 10 min to ensure a 239 completed conversion. The mixture (1 mL) was then filtered through a 0.45 µm PTFE filter membrane 240 into an amber glass vial. 10 µL of the sample was injected into a Phenomenex Zebron ZB-FFAP (30 241 m x 0.25 mm) column using a vaporising injector with a split flow of 50 mL/min of helium. The oven 242 temperature was maintained at 120 o C for 1 min and then increased to 250 o C at ramp 5 o C/min and 243 held for 2 min. Detection was conducted using a mass spectrophotometer and the identification of 244 individual fatty acids was achieved using a mass spectrum library by means of comparison of 245 retention time and molecular mass to FAME standards. 246

Statistical analysis 247
All experiment were performed in triplicate. The statistical analysis was carried out using IBM 248 SPSS statistic 25 using post hoc analysis of variance (ANOVA) and according to the Tukey test with statistically significant at p<0.05. The data were expressed as mean ± standard deviation. Differences 250 of means at p<0.05 were considered significant. 251

Effect of endogenous enzymes and heat-treatment on the CRF galactolipids 254
It is well known that nutrient concentrations in the plants start to decrease after harvesting 255 due to their degradation by endogenous enzymes. 38 Thermal processing can be used to inactivate 256 enzyme reactions after harvesting, extending the shelf life of the nutrients and the stabilisation of 257 texture, flavour and nutrients. 39 The action of endogenous enzymes on the lipids from PVH CRF was 258 studied (Figure 2). TLC separations of polar and neutral lipids were analysed first without staining 259 which allowed a qualitative visualisation of carotenoids and pigments, including chlorophylls ( Figures  260   2A and 2C). No major changes were observed whatever the treatment of CRF. After lipid staining and 261 from the comparison of non-heat treated and steam sterilised or pasteurised CRF, it was clear that 262 galactolipid levels, especially those of DGDG, were lower in non-heat treated samples than in steam 263 sterilised samples and after 1-hr incubation of these samples at 37 o C, the band of DGDG has 264 disappeared ( Figure 2B). Without heat treatment, and in contrast to DGDG, the band of MGDG was 265 apparent both before and after incubation. The low levels of galactolipids were associated with high 266 levels of free fatty acids (FFA), which suggests the action of endogenous galactolipases being 267 present in the material ( Figure 2D). Thermal treatments did inactivate endogenous enzymes as 268 shown in the Figure 2B where the bands of MGDG and DGDG are visible and remain at similar levels 269 both before and after 1-h incubation at 37°C, for each treatment. The higher levels of galactolipids 270 were associated with reduced levels of FFA, which indicates that endogenous galactolipases can be 271 heat-inactivated ( Figure 2D). In addition, CRF were also incubated for 1 hr at 37 o C after addition of 272 GPLRP2, an enzyme known to display galactolipase activity. 12 The galactolipids of both steam 273 sterilised and pasteurised CRF from PVH were hydrolysed to lysogalactolipids (MGMG and DGMG; 274 galactolipids and this phenomenon was also observed with spinach galactolipids (data not shown). 281 Therefore, heat treatment was applied to all CRF preparations used for in vitro digestion experiments.
While CRF from PVH was steam sterilised, the spinach leaves were first blanched in hot water at 283 85 o C for 3 min to knock out the endogenous enzymes before preparing CRF. It was checked that this 284 treatment allowed the inactivation of the endogenous galactolipase activity of CRF from spinach 285 leaves. 286

Galactolipid and free fatty acid content in CRF before digestion 287
The amounts of galactolipids and free fatty acids in the CRF of blanched spinach leaves and 288 steam sterilised or non-heat treated PVH were measured ( Table 1). The CRF prepared from heat-289 treated materials showed a higher MGDG content than DGDG. This is in agreement with what has 290 been reported in photosynthetic tissues, especially in the inner envelope membrane and thylakoid 291 membrane of chloroplast where the amount of MGDG is higher than DGDG, while DGDG amounts 292 exceed MGDG in the non-photosynthetic tissues.
9,40-41 The CRF of spinach leaves had more 293 galactolipids per dry mass of CRF than PVH CRF due to a higher lipid content in CRF of spinach 294 leaves, but galactolipids were represented at similar levels in total lipids from both CRF (around 100 295 mg/g of total lipid extracts). The other lipids, including carotenoids, tocopherols, chlorophyll esters, 296 sterols and phospholipids were not quantified. The lower relative levels of lipids in pea vine CRF may 297 be due to some dilution by components like starch. The level of free fatty acids in CRF spinach was 298 quite low 1.30 ± 0.35 mg/g CRF or 5.08 ± 1.35 mg/g lipid extract, in line with the fact that spinach 299 leaves were blanched to avoid lipolysis by endogenous enzymes. The CRF of steam sterilised pea 300 vine showed a greater amount of FFA (11.71 ± 0.72 mg/g CRF or 71.95 ± 4.43 mg/g lipid extract), 301 which indicates some significant lipolysis occurring from harvesting and before heat treatment of PVH. 302 However, the determination of the amount of FFA in haulm directly after harvesting was not possible 303 due to logistical limitation. Nevertheless, the amounts of FFA in the CRF from steam sterilised pea 304 vine are 3-fold lower than those found in CRF from non-heat treated pea vine as shown in the Table  305 1. In agreement, MGDG and DGDG levels in CRF from steam sterilised PHV are 3.4-fold and 1.8-fold 306 higher than in CRF from non-heat treated PVH. It is worth noting that the total masses of MGDG, 307 DGDG and FFA are similar in both CRF preparations, which confirms that most FFA are generated by 308 the endogenous hydrolysis of galactolipids. 309

In-vitro digestion of galactolipids in CRF
The experimental conditions to simulate lipid digestion were based on in vivo studies and 311 parameters measured at 50% meal gastric emptying, 27,33 such as the lipase concentrations (17 µg/mL 312 of gastric lipase in the stomach and 250 µg/mL of pancreatic lipase in the small intestine) and the pH 313 values (5 for the gastric phase and 6 for the intestinal phase). In these experiments, rabbit gastric 314 extract (RGE) was chosen as the source of gastric enzymes because it is composed of pepsin and 315 gastric lipase, and the activity of rabbit gastric lipase has a similar range of activity as the human 316 gastric lipase (HGL). 23  According to the TLC analysis of 327 galactolipid digestion (Figures 3, 4, S1 and S2), both the MGDG and DGDG of CRF from blanched 328 spinach leaves and steam sterilised PVH were hydrolysed and converted to lysogalactolipid 329 (monogalactosylmonoglycerol, MGMG, and digalactosylmono glycerol, DGMG) during the whole 330 digestion process. During the 30-min gastric phase, galactolipids decreased slightly from 25-26 to 22-331 23 mg MGDG /g CRF and 18-22 to 17-18 mg DGDG /g CRF for spinach leaves CRF ( Figure 4A and 332 4B) and from 15-16 to 14-15 mg MGDG /g CRF and 9-11 to 8-9 mg DGDG /g CRF for PVH CRF 333 ( Figure 4C and 4D). The weak increase in FFA during the gastric phase ( Figure 5B) Figures 4A, 4C and S1), and high amounts of free 339 fatty acids were released (Figure 5 and S2), especially during the first 5 min of the intestinal phase.
Because the monogalactosyl galactolipids (MGDG and MGMG) are revealed similarly upon thymol 341 staining (Figures 2B), and in the absence of a pure MGMG reference standard, the generation of 342 MGMG was tentatively quantitated by TLC using the calibration curve established with MGDG as 343 reference standard. A good correlation was observed between the appearance of MGMG and the 344 disappearance of MGDG (Figure 4). The band corresponding to DGMG on TLC plates (see Figure 3  345 and Figure S1 in Supplementary Data) was however too faint and not enough resolved to apply the 346 same method. These results are supported by those of Andersson et al. (1995) 11 , who demonstrated 347 that pure galactolipids were hydrolysed by Human Pancreatic Juice. This previous study had led to 348 the characterisation of the galactolipase activity of HPJ and its association with PLRP2 and to a lower 349 extent to CEH/BSSL. 12,14 Galactolipids from both CRF preparations were 350 hydrolysed at a slower rate when RGE and PPE were combined for in vitro digestion (Figures 4B and  351   4D). It confirms that PPE contains a lower galactolipase activity. 46 Additionally, it was observed that 352 PPE prefers to hydrolyse MGDG over DGDG, which is not the case for HPJ. 353 TLC analysis of the fatty acids released during 354 digestion of both CRF from spinach leaves and pea vine (See Figures 5 and S2

, in Supplementary 355
Data) showed that the FFA bands obtained using the mixture of RGE and PPE, had a higher intensity 356 than those obtained with the mixture of RGE and HPJ. This was not consistent with the lower 357 hydrolysis of galactolipids by PPE. We then performed control experiments without the CRF substrate 358 and found that FFA could be released during the intestinal phase when PPE was used but not HPJ 359 ( Figure 5A). This finding is probably due to the fact that PPE contains some lipids 47 and these lipids 360 can be hydrolysed during the digestion experiment. The FFA analysis by TLC were therefore 361 corrected accordingly by subtracting the FFA levels measured in the controls. This point was never 362 raised however in previous in vitro digestion studies, probably because the levels of FFA released 363 from PPE are much lower than those released from dietary triglycerides. But one has to be cautious 364 when low levels of lipids are concerned, as here with CRF galactolipids. 365

Composition and release upon digestion of the CRF fatty acids 366
The compositions of total fatty acids in spinach leaves and pea vine CRF were analysed 367 using GC-MS and the results are shown in the Figure 6. The main fatty acid in both spinach leave 368 and pea vine CRF was α-linolenic acid (ALA; 18:3) but its amounts in spinach CRF (35.56 ± 2.56 369 mg/g CRF DW) were 2.5-fold higher than in PVH CRF (14.29 ± 2.06 mg/g CRF DW). ALA represented 55.38 ± 1.33 % w/w of the total fatty acids in spinach leave CRF, which is in agreement 371 with the fatty acid composition of spinach leaves and isolated galactolipids. 10 In pea vine CRF, ALA 372 represented only 37.82 ± 0.31 % w/w of the total fatty acids. The second most abundant fatty acid 373 was palmitic acid (PA; 16:0), which was found at similar levels in spinach CRF (13.31 ± 0.24 mg/g 374 CRF; 20.77 ± 0.76 % w/w of total FA) and pea vine CRF (12.35 ± 1.86 mg/g CRF; 32.65 ± 0.12 % w/w 375 of total FA). Hexadecatrienoic acid (16:3) was found at significant levels in spinach CRF (5.98 ± 0.12 376 mg/g CRF; 9.34 ± 0.29 % w/w of total FA) but was not found in pea vine CRF. Therefore, the levels of 377 polyunsaturated omega 3 fatty acids were globally reduced in PVH CRF compared to spinach CRF. 378 This may results from a preferential oxidation of these fatty acids by endogenous lipoxygenases from 379 harvesting to steam sterilisation of PVH. Moreover, this oxidation is known to be coupled to and 380 favoured by the release of FFA by endogenous galactolipase activity as often seen in plant response 381 to stress. 48 The absence of 16:3 fatty acid in PVH CRF may also be explained by the classification of 382 After 90 min of digestion by RGE and HPJ, the composition of FFA generated from CRF 388 hydrolysis was similar to the composition of total FA in both spinach and PVH CRF (Figure 6), with 389 ALA representing the most abundant FFA, followed by PA. With the combination of RGE and PPE, 390 some changes were observed in the relative distribution of fatty acids, which certainly reflects the 391 contribution of fatty acids present in PPE (see Figure S3 and Table S1 in Supplementary data). 392 Therefore, we only discuss here the data obtained with RGE and HPJ. We observed that the amounts 393 of total FFA released during the CRF digestion were higher than those expected from the complete 394 lipolysis of MGDG and DGDG, both for spinach leave and PVH CRF (Table 2 and Figure 5B). 395 Moreover, we have seen that galactolipid lipolysis was not complete, with some accumulation of 396 MGMG (Figure 4) and DGMG (Figure 3 and S1), in line with the preferential hydrolysis by PLRP2 of 397 the ester bond at the sn-1 position of galactolipids.
10 Therefore, some FFA were certainly released 398 from other acyl lipids present in CRF preparations. This hypothesis is supported by the levels of total 399 fatty acids present in CRF and quantified by GC-MS ( Table 2). Figure 3D on TLC separation of neutral lipids shows that some apolar lipids present in PVH CRF at time 0 and migrating like 401 triglycerides or sterol esters are no more present at time 90 min. Phospholipids, mainly PG, are also 402 present in chloroplast membranes and can be degraded by the phospholipase activity of HPJ. During 403 stress or senescence, the degradation of chlorophyll and galactolipid from thylakoid membranes in 404 chloroplasts can also lead to the conversion of a large proportion of phytol and fatty acids into fatty 405 acid phytyl esters (PFAE) and triacylglycerol. 51 We did not search nor analyse these various lipids 406 here but it would be worth identifying these other sources of fatty acids in future studies in order to 407 better describe the lipid composition of CRF. At this stage, it is important to keep in mind that 87. pea wine haulm CRFs (panel A) and corresponding free fatty acids released after 90-min digestion by 669 RGE and HPJ (panel B). Data are expressed in mg per g of CRF dry weight and presented as mean ± 670 SD (n=3). 671 Table 2: Total fatty acids in PVH and spinach leave CRF and their release upon digestion. Total fatty 681 acids (FA) in CRF were estimated from GC-MS analysis. FA presents in galactolipids (MGDG and 682 DGDG) and free fatty acids (FFA) initially present in CRF were estimated from TLC analysis and data 683 in Table 1. Mass amounts (mg/g CRF, DW) of MGDG, DGDG and FFA were converted in µmoles per 684 g of CRF (DW) using average molar masses of 760, 922 and 271 g/mole, respectively, which were 685 estimated from the fatty acid composition of CRF ( Figure 6 and Table S1). FA mole equivalents 686 present in MGDG and DGDG were then estimated. FFA released after 90-min digestion by the 687 combination of RGE and HPJ were estimated from TLC analysis. Values, expressed either in mg/g 688 CRF (DW) or µmole/g CRF (DW), are means ± SD (n=3 Data were presented as a mean ± SD of 3 separated in-vitro digestion and analysed using an independent-691 sample t-test with statistically significance at p<0.0.5, a>b.